Fernando Zúñiga and Denis Creissels (Eds.

)
Applicative Constructions in the World’s Languages
Comparative Handbooks
of Linguistics

Edited by
Andrej Malchukov and Edith Moravcsik

Volume 7
Applicative
Constructions in the
World’s Languages

Edited by
Fernando Zúñiga and Denis Creissels
ISBN 978-3-11-073548-2
e-ISBN (PDF) 978-3-11-073095-1
e-ISBN (EPUB) 978-3-11-073109-5
ISSN 2364-4354

Library of Congress Control Number: 2023941755

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;
Detailed bibliographic data are available in the Internet at http://dnb.dnb.de

© 2024 Walter de Gruyter GmbH, Berlin/Boston

Cover image: Jupiterimages/PHOTOS.com/thinkstock
Typesetting: Integra Software Services Pvt. Ltd.
Printing and binding: CPI books GmbH, Leck

www.degruyter.com
Purpose and Aim of the Series
The purpose of the series Comparative Handbooks of Linguistics is to present collec-
tions of specific studies on particular topics in a format that facilitates cross-linguistic
and cross-theoretical comparison. Each handbook takes up a single theme and includes
a questionnaire; both the individual studies and the summary paper synthesizing the
results of the comparison reflect the structure of the questionnaire.
Unlike other handbooks that consist of chapters organized by theoretical topics,
the handbooks of this series present empirical studies on various languages uniformly
structured to facilitate comparison. A typical volume examines a grammatical feature
or construction across languages (such as interrogative structures or numeral construc-
tions), offers a survey of that feature through in-depth analyses cast in roughly the same
theoretical framework, and formulates cross-linguistic generalizations. The handbooks
also include accounts of constructions cast in different theoretical frameworks so that
the empirical differences among the theories can be assessed and their similarities
revealed.
The volumes are accessible to the general academic community of linguists, meet
the needs of students on the undergraduate and graduate level, and are of interest to
specialists as well. Each volume is self-contained. Some of the volumes have associated
websites to provide additional materials.

https://doi.org/10.1515/9783110730951-202
Volumes already published in this series
1. VALENCY CLASSES IN THE WORLD’S LANGUAGES. 2015.
Edited by Andrej Malchukov and Bernard Comrie

1.1 Volume 1. Introducing the framework and case studies from Africa and Eurasia
1.2 Volume 2. Case studies from Austronesia and the Pacific, the Americas, and the-
oretical framework

2. LEVELS IN CLAUSE LINKAGE. A crosslinguistic survey. 2018.

Edited by Tasaku Tsunoda

3. CURRENT APPROACHES TO SYNTAX. A comparative handbook. 2019.

Edited by András Kertész, Edith Moravcsik, and Csilla Rákosi

4. GRAMMATICALIZATION SCENARIOS: CROSSLINGUISTIC VARIATION AND UNIVER-

SAL TENDENCIES. 2020.
Edited by Walter Bisang and Andrej Malchukov

4.1. Volume 1. Grammaticalization scenarios from Europe and Asia

4.2. Volume 2. Grammaticalization scenarios from Africa, the Americas, and the
Pacific

5. NUMBER IN THE WORLD’S LANGUAGES. A comparative handbook. 2022.

Edited by Paolo Acquaviva and Michael Daniel

6. MERMAID CONSTRUCTION. A compound-predicate construction with biclausal

appearance. 2020.
Edited by Tasaku Tsunoda

Forthcoming volumes can be found on our website: www.degruyter.com

Contents
Purpose and Aim of the Series V

Part I: General chapters

Fernando Zúñiga and Denis Creissels

1 Applicative constructions: An introductory overview 3

Fernando Zúñiga
2 Questionnaire on applicative constructions 57

Denis Creissels and Fernando Zúñiga, with George Moroz

3 Languages examined or referred to in the present book 61

Part II: Case studies

Individual languages

Donna B. Gerdts
4 Hul’q’umi’num’ Salish applicative constructions 79

David Beck
5 Applicatives in Upper Necaxa Totonac 115

Marisa Censabella
6 Applicatives in Toba/Qom (Guaykuruan) 143

Fernando Zúñiga
7 The applicative constructions of Mapudungun 179

Denis Creissels
8 Applicative constructions and non-applicative uses of applicative morphology
in Tswana (Bantu) 211

Mengistu Amberber
9 Applicativization in Amharic 243

Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 279
VIII Contents

Areal overviews

Rik van Gijn

11 Contact-induced diffusion of applicatives in northwestern Amazonia? 307

William A. Foley
12 Applicatives in Papuan languages 347

Peter K. Austin
13 Applicative constructions in Australian Aboriginal languages 391

Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

14 Applicativizing preverbs in selected European languages 419

Genealogical overviews

Tim Thornes
15 Applicatives in Northern Uto-Aztecan languages 475

Albert Álvarez González and Zarina Estrada-Fernández

16 Applicative constructions in Uto-Aztecan languages from Northwestern
Mexico 509

Marianne Mithun
17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut)
languages 557

Hunter Thompson Lockwood and Monica Macaulay

18 Applicatives across Algonquian 601

Brad Montgomery-Anderson
19 Applicative constructions in Mayan languages: An overview with special focus
on Chontal 645

Néstor Hernández-Green and Oscar López Nicolás

20 Applicative constructions in two Otomanguean families: Otomi
and Zapotec 679

Sara Pacchiarotti
21 The polyfunctional applicative ✶-ɪd in Bantu languages 719
 Contents IX

Sylvie Voisin and Denis Creissels

22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 749

Doris L. Payne
23 Nilotic applicatives 783

Martine Vanhove
24 Applicative constructions in Cushitic 835

Peter Arkadiev, Yury Lander, and Irina Bagirokova

25 Applicative constructions in the Northwest Caucasian languages 869

Kevin Tuite
26 Applicative constructions in Kartvelian 913

Guillaume Jacques, Aimée Lahaussois

27 Applicative derivations in Kiranti 943

Bradley McDonnell, and Christina L. Truong

28 Applicative constructions in languages of western Indonesia 971

Part III: Theoretical/Comparative outlook

Maria Polinsky
29 Understanding applicatives 1007

George Moroz and Maria Polinsky

30 Applicatives cross-linguistically: Features and distribution 1033

Denis Creissels and Fernando Zúñiga

31 Applicative and related constructions: Results and perspectives 1045

Language index 1077

Subject index 1083

Part I: General chapters
Fernando Zúñiga and Denis Creissels
1 Applicative constructions: An introductory
overview
Abstract: This chapter presents and contextualizes the definition of applicative con-
structions used in the present book. (Although our definition is conservative in several
respects, it is broader than some alternative definitions and narrower than others; in
particular, we argue that a number of similar constructions are best regarded as looka-
likes rather than as types of applicatives.) The chapter also surveys the main morpho-
logical, syntactic, and semantic parameters of variation of applicatives identified in
the previous literature, and outlines the structural characteristics of the applicatives of
Classical Nahuatl, the language for which the term was first used. Lastly, we provide a
roadmap for the book and comment on its coverage and coherence.

1 Introduction
This chapter introduces the terminological and analytical prerequisites for the study of
the phenomena at the center of attention in the present book. After this brief roadmap,
Section 1 presents not only the definition of applicative constructions employed here
but also the characteristics of several constructions that show both important similar-
ities and crucial differences to them. Sections 2 through 4 survey the known variation
of applicative constructions regarding morphology, syntax, and semantics, respectively.
Section 5 outlines the applicatives of Classical Nahuatl, the language for whose descrip-
tion the term was first used. We have included this not only to show how multi-faceted
applicativization can be, even in one single language, but also to prepare the reader
for the kind of content and structure they will encounter in most of the chapters of
the book. Section 6 presents the historical background of the term applicative and dis-
cusses some relevant aspects of its present-day usage. Lastly, Section 7 comments on the
structure, the scope, and the coherence of the book as a whole.

Acknowledgments: We are indebted to Peter Arkadiev, Donna Gerdts, and Monica Macaulay for their
comments on the position paper that evolved into this chapter, as well as to the series editors for their
comments on a previous version of the chapter. Many thanks also go to Drew Hancock-Teed, Jean Rohleder
and especially Kevin Negele for their valuable help with the formatting of many chapters of the book.

https://doi.org/10.1515/9783110730951-001
4 Fernando Zúñiga and Denis Creissels

1.1 Applicatives defined

An applicative construction (henceforth: AC) is defined here in opposition to a base

construction (henceforth: BC) with respect to several formal features, including those
of one particular constituent of the former that bears a non-agentive semantic role,
namely the applied phrase (henceforth: AppP; see also Creissels, forthcoming). Example
(1) from a Tlacolula Valley Zapotec variety illustrates such an opposition with a clause
pair that most present-day linguists consider an instance of applicativization. In (1a),
the co-Agent Jwaany ‘John’ is the AppP, an argument in P role of the clause headed by
the applicativized verb ‘sing with’. The same participant is expressed as an optional
adjunct (cëhnn Jwaany ‘with John’) in (1b), where the verb is in its base, non-derived,
form and simply means ‘sing’:

(1) San Lucas Quiaviní Zapotec (Zapotecan; Munro 2000: 285–286, glosses adapted)
a. B-ìi’lly-nèe Gye’eihlly [Jwaany].
perf-sing-appl M. J.
‘Mike sang with John.’ (AC)
b. B-ìi’lly Gye’eihlly [cëhnn Jwaany].
perf-sing M. with J.
‘Mike sang with John.’ (BC)

This conforms to the following definition, which we use in the present chapter and in
the rest of the book:
The base construction (BC) and the applicative construction (AC) are related as
follows:
i) The predicates in both constructions are built upon the same root, but the one in
the AC bears additional overt marking that distinguishes it from the one in the BC.
ii) The participant encoded as S or A in the BC appears as S or A in the AC.
iii) The AC includes a noun phrase in a role other than S or A, the applied phrase
(AppP), which refers to a participant that either requires a non-core coding in the
BC different from its coding in the AC or cannot be expressed at all in the BC.

We follow mainstream usage whenever there is no potential confusion and employ the
term applicative as shorthand for either an applicative construction (e.g., the whole
clause in [1a]) or an applicative marker (e.g., the verbal suffix -nèe in [1a]).
One morphological feature of ACs, namely the marking asymmetry between the
predicates, is treated disparately in the literature: it is an almost constant hallmark of
ACs in functional-typological studies but not a criterial feature in some strands of for-
malist studies (see § 6.2). Other features of ACs, like their productivity, morphological
regularity, and semantic predictability, are typically assumed to be present but rarely
characterize the construction per se. Periphrastic constructions can be subsumed under
our definition insofar as they show evidence of monoclausality. Nevertheless, as with
 1 Applicative constructions: An introductory overview 5

other periphrastic voices (see, for instance, the abundant literature on periphrastic
causatives), it may be difficult to draw the line between a bona fide periphrastic applica-
tive and a biclausal construction that is merely the translational equivalent of an AC.
One syntactic feature of ACs, namely the syntactic status of the constituents cor-
responding to the same referent in the different constructions, shows variation that
is entwined with the defining traits of the construction and is addressed in Sections
1.2 and 3. Our definition uses the syntactic roles S and A; they correspond to the single
essential argument of (a major subclass of) monovalent predicates and to the agentive
argument of prototypical bivalent predicates, respectively. The other three roles used
in our model of syntax are A’s patientive counterpart P (i.e., a direct or primary object),
a “dative role” D found with trivalent predicates and indirective alignment (i.e., an indi-
rect object), and an ordinary oblique role X; these non-S/A roles are addressed in the
context of applicativization in Section 3.1. All these comparative concepts used in align-
ment typology are understood here as in what Haspelmath (2011) calls the “Comrian
approach” (Comrie 1989; Lazard 1994; Creissels 2006); specifically, A and P are “syn-
tactic terms whose prototypes are defined in semantic terms” (Comrie 1989: 111). We
use terms like subject and object, which refer to language-and-construction-particular
alignment patterns, only language-specifically.
Applicativization is a subtype of verb-coded valency alternation: more specifically,
several of its subtypes are special cases of nucleativization, that is, an operation that
allows participants not encoded as core terms in the base construction to be encoded
as such in the derived construction. We elaborate on this contextualization in Section 6,
after reviewing the variation of applicatives in Sections 2 through 4.

1.2 What applicatives are not: Lookalikes

Applicative lookalike constructions (henceforth: ALC) are either syntactically or mor-

phologically similar to true ACs. We discuss them in turn in what follows.

1.2.1 Syntactic lookalikes

Pairs of constructions whose syntactic make-up conforms to the above definition of an

AC-BC pair, but whose morphology does not, come in several types, depending on the
formal relationship holding between the predicates of the two constructions involved
(Comrie et al. 2015; Creissels, forthcoming). The alternation between the predicates of
both clauses can be coded or uncoded, and asymmetrical or symmetrical. Table 1 sum-
marizes the possibilities; the numbered examples provided further down are cross-ref-
erenced in the table. Applicativization appears in the upper left-hand cell; the alterna-
tions listed in the other cells correspond to different syntactic lookalike types.
6 Fernando Zúñiga and Denis Creissels

Table 1: Morphological marking of ACs and syntactic lookalikes.

Coded (“voice”) Uncoded

Asymmetrical Privative marking: —
Applicativization (1)–(2)

Symmetrical – Suppletivism (3) Flexivalency (5)–(9)

– Equipollent marking (4)

The applicatives discussed in the functional-typological literature typically cor-

respond to the privative-marking alternation type, in which the AC predicate shows
higher morphological complexity than the BC predicate—more precisely, an element
analyzable as an applicative marker, which in turn leads to a voice alternation.1 Exam-
ples (1) above and (2) below illustrate such an opposition:2

(2) Boumaa Fijian (Oceanic; Dixon 1988: 45)

a. E-la’o-va a suka a marama.
3sg.sbj-go-appl def sugar def woman
‘The woman is going for sugar.’ (AC)
b. E-la’o a marama.
3sg.sbj-go def woman
‘The woman is going.’ (BC)

By contrast, the suppletive alternation type consists in the predicates showing a

formal difference that cannot be analyzed as a particular instance of some regular pattern.
Instead of arbitrarily postulating a transitivizing (= applicativizing) or detransitivizing
(= antipassivizing) “suppletive derivation”, it seems best to treat such pairs as simply
alternating. (The abbreviation in inverted commas “BC” here and elsewhere signifies
that that construction is the equivalent of the base construction if its counterpart were
an applicative.) The following example illustrates this case:

(3) Akhvakh (Northeast Caucasian; p.k.)

a. Riλ̄’i q̄’am-a!
meat eat-imp
‘Eat some/the meat!’ (ALC)

1 Since it is not a syntactic lookalike of applicativization, we are glossing over the latter’s mirror image
here, namely antipassivization (where the undirected predicate is morphologically more complex than
its directed counterpart, as in Mandinka dómó-rì ‘eat [intr]’ vs. dómò ‘eat [tr]’). See Creissels (forthcom-
ing: Ch. 10) for more details.
2 The marker -va can have other functions in Fijian, but we gloss polysemous markers in this chapter
according to their function in the example under discussion. The same applies to Indonesian -kan in
(40)/(43)/(44), for instance, which can also function as a causative.
 1 Applicative constructions: An introductory overview 7

b. Ũk-a!
eat-imp
‘Eat!’ (“BC”)

In other cases, the predicates in both clauses also show the same number of
morphemes, but neither predicate is zero-marked for voice; rather, they are equipol-
lently marked; the predicates themselves are not unmarked, but the alternation is not
morphologically oriented (i.e., the marking of the predicates provides no evidence that
one of the constructions is basic and the other one derived). For instance, in (4), both
predicates take a so-called final suffix: the “benefactive” transitive animate -o in (4a)
and the animate intransitive -aa in (4b).3 (Note that the argument to analyze the verb
stem ending in -o as somehow derived from any of the related verb stems is semantic,
or possibly syntactic, rather than morphological.4 Compare this formal opposition to the
Blackfoot applicative -omo in [31].)

(4) Blackfoot (Algonquian; Frantz 2009: 102, glosses adapted)

a. Iihpómm-o-yii-wa-áyi ónnikiS-i.
pst.buy-ta-dir-3sg.prox-3sg milk-nspec
‘S/he bought (some) milk for him/her.’ (ALC)
b. Iihpómm-aa-wa ónnikiS-i.
pst.buy-ai-3sg.prox milk-nspec
‘S/he bought (some) milk.’ (“BC”)

With the last valency alternation type, both predicates show the same degree of
morphological complexity (i.e., neither predicate is marked for voice), and instead of an
applicative voice alternation there is a flexivalency alternation (Martin Haspelmath,
p.c.; Creissels, forthcoming: Ch. 15; Heidinger 2019 uses the related term “polyvalency”
instead). Predicate lability—more precisely: ambitransitivity—can be either weak or
strong (Creissels 2014). With the former, the alternating clauses differ formally only
with respect to the presence vs. absence of an argument NP (e.g., Evan drinks tea vs.
Evan drinks). With the latter, the alternating clauses differ formally in other respects as
well (most notably, in the coding of the shared argument NP in both clauses; e.g., Gareth

3 Ónnikii ‘(some) milk’ is a secondary object in both the AC and the BC—a grammatical relation that
does not trigger indexing on the verb; the suffix -wa indexes the subject in both clauses. Primary objects
appear indexed on the verb when non-3rd-person (but the 3rd-person AppP in [4a], an instance of such a
primary object, is visible only via the direct marker -(y)ii, which implicitly signals syntactic transitivity).
The 3sg index -áyi refers to the primary object, but the conditions governing its appearance are uncon-
nected to applicativization (see Frantz 2009: Ch. 9).
4 There are four related verb stems meaning ‘buy’ in Blackfoot, namely animate intransitive (aio)
ohpomm-aa- ‘buy (things)’, transitive inanimate (ti) ohpomm-atoo- ‘buy (inanimate)’, and two transi-
tive animate (ta) stems, namely ohpomm-at- ‘buy (animate)’ and ohpomm-o- ‘buy (things/inanimate/
animate) for (animate)’.
8 Fernando Zúñiga and Denis Creissels

broke the vase vs. the vase broke). In the case of weak lability, as in the English conative
alternation (5) and benefactive alternation (6), the only formal difference between both
clauses is the syntactic asymmetry regarding the phrase expressing the non-agentive
participant (and its flagging).

(5) English (Germanic; Levin 2015: 64)

a. Kelly kicked the intruder. (ALC)
b. Kelly kicked (at the intruder). (“BC”)

(6) English (p.k.)

a. Alun baked Gwen cookies. (ALC)
b. Alun baked cookies (for Gwen). (“BC”)

Something analogous is found in the German dative alternation in (7), where the
2nd-person indirect object dir ‘(to) you’ (7a) contrasts with the oblique an dich ‘to you’
(7b), and in the English alternation in (8), where the non-agentive participant is an
oblique in both clauses, albeit with different flagging (viz. for vs. instead of). In both
instances, the verb remains unaltered:

(7) German (Germanic; p.k.)

a. Ich schrieb dir den Brief.
1sg.nom wrote[1sg] 2sg.dat art.m.sg.acc letter
‘I wrote you the letter.’ (ALC)
b. Ich schrieb den Brief an dich.
1sg.nom wrote[1sg] art.m.sg.acc letter on/to 2sg.acc
‘I wrote the letter to you.’ (“BC”)

(8) English (p.k.)

a. Caitlín spoke up for the deputy. (ALC)
b. Caitlín spoke up instead of the deputy. (“BC”)

In the case of strong lability, as in the Mandinka alternation in (9), there is also some
marking difference between the clauses (here: the aspectual markers yè and -tá), but
that difference does not relate the predicates derivationally to each other:

(9) Mandinka (Mande; p.k.)

a. Díndíŋò yè ŋ́ jélè.
child.def compl.tr 1sg laugh
‘The child made fun of me.’ (ALC)
b. Díndíŋò jélè-tá.
child.def laugh-compl.intr
‘The child laughed.’ (“BC”)
 1 Applicative constructions: An introductory overview 9

1.2.2 Morphological lookalikes

Constructions whose morphological make-up corresponds to that of an AC, but some

other features of which (e.g., systematicity or syntax) do not, also come in several types.
Consider the overview in Table 2; the numbered examples provided further down are
cross-referenced in the table.

Table 2: Main morphological lookalike subtypes.

Characteristic Label
Non-compositional meaning Lexicalized applicatives (10)
No unmarked predicate Applicative deponents (11)
No valency change Semantic effect Syntax-neutral intensification (12b)
Pragmatic effect Oblique registration (14b/d), (17), (18b)

First, verbal markers that applicativize some predicates can also irregularly and/
or unproductively appear in so-called lexicalized applicatives.5 Compare the semantic
roles of the German direct object in the regular pair steigen ‘climb (intr)’ vs. besteigen
‘climb (tr)’ with their counterparts in the irregular pairs in (10):

(10) German (p.k.)

dürfen ‘may, be allowed to’ bedürfen ‘need, require’ (formal)
kennen ‘know (sb.)’ bekennen ‘confess, admit’
kommen ‘come’ bekommen ‘receive’
schreiben ‘write’ beschreiben ‘describe’

Despite the suggestive morphology of the derived predicates, and perhaps even a syntax
comparable to that of bona fide ACs, the semantics of such pairs is unpredictable. (See
Zúñiga, Arkadiev, and Hegedűs, this volume, for more data and references regarding
such German phenomena.)
Second, it is not uncommon to find applicative deponents (also called applicativa
tantum), that is, verbs that only occur with the marker that distinguishes base and
applicativized versions of other verbs. For example, the following German be-prefixing
verbs lack counterparts without that prefix:

(11) German (p.k.)

beabsichtigen ‘intend’ ✶
absichtigen
beanspruchen ‘claim, require, demand’ ✶
anspruchen
bejahen ‘affirm, approve’ ✶
jahen
bewerkstelligen ‘accomplish, put into execution’ ✶
werkstelligen

5 See Pacchiarotti (2020: 104f) on the lexicalized “pseudo-applicatives” of Tswana in this context.
10 Fernando Zúñiga and Denis Creissels

Third, the applicative-like marker can distinguish two related clauses without there
being an applied phrase, that is, predicate derivation may not bring about valency
modulation. Consider that it is cross-linguistically common for a given morphologi-
cal element to fulfill distinct functions in a particular language. On the one hand, the
element in question may be an applicative marker in constructions that fully meet the
definition of applicative construction given above. On the other hand, a homophonous
element may be found in constructions where its presence marks a term of the clause
as prominent in one way or another, without affecting its coding, or the coding of any
other term. Such constructions are like applicative constructions as conceptualized
here in that both operate on non-nuclear participants (i.e., participants other than those
coded as A, S, or P in the base construction), or on circumstantials (i.e., participants
bearing peripheral semantic roles). They are unlike applicative constructions, however,
in that they do not involve a valency alternation.
This phenomenon appears to come in at least two distinct guises. In the first, the
contrast between the marked and unmarked clauses is semantic and orbits notions
like aspect (e.g., completion, repetition, iteration, or continuativity) and manner (e.g.,
intensity, persistence, or excess). Such constructions might be called instances of syn-
tax-neutral intensification. The pair in (12) illustrates such a case, where the applica-
tive marker -ir appears in both (12a) and (12b) accommodating the AppP ‘the children’,
but the second instance of -ir in (12b) does not introduce another AppP:

(12) Nyole (Bantu JE.35; Wicks 2006: 107)

a. Ba-hayuh-ir-a aba-ana.
3.sbj.II-shout-appl-fv II-child
‘They shout at the children.’ (AC)
b. Ba-hayuh-ir-ir-a aba-ana.
3.sbj.II-shout-appl-asp-fv II-child
‘They are always shouting at the children.’ (ALC on AC)

Compare this with (13), where the applicative -ñma introduces an AppP every time it is
suffixed to the verb; there is one AppP in (9b) (Kuan ‘Juan’) and two AppPs in (9c) (Kuan
‘Juan’ and tañi fotüm ‘his son’):

(13) Mapudungun (isolate; based on Salas 2006: 122)

a. Weñe-i waka.
steal-ind[3.sbj] cow
‘S/he stole a/the cow.’ (BC)
b. Weñe-ñma-fi-i waka Kuan.
steal-appl-3.obj-ind[3.sbj] cow J.
‘S/he stole a/the cow from Juan.’ (AC)
 1 Applicative constructions: An introductory overview 11

c. Weñe-ñma-ñma-fi-i waka Kuan tañi fotüm.

steal-appl-appl-3.obj-ind[3.sbj] cow J. 3.psr son.of.man
‘S/he stole a/the cow from Juan’s son.’ (AC on AC)

The second kind of valency-neutral applicative-like marking serves a chiefly prag-

matic purpose. For instance, in the following examples, mó sɪ ̀fápáánòŋ̀ ‘on a cross’ is
an adjunct in both (14a) and (14b); in the latter, however, the adjunct is focused and
actually obligatory. Note that the same verb can occur with the applicative marker in
an AC (14c), and even with both the applicative and the applicative-like focus marker
stacked (14d):

(14) Tswana (p.k.)

a. Dʒɪ́sʊ́ ʊ́ -nè à-sw-á (mó sɪ ̀-fápáánò-ŋ̀).
I.J. 3.sbj.I-aux 3.sbj.I-die-fv loc VII-cross-loc
‘Jesus died (on a cross).’ (BC)
b. Dʒɪ́sʊ́ ʊ́ -nè à-sw-ɛ́l-à mó sɪ ̀-fápáánò-ŋ̀.
I.J. 3.sbj.I-aux 3.sbj.I-die-foc-fv loc VII-cross-loc
‘Jesus died on a cross.’ (ALC)
c. Dʒɪ́sʊ́ ʊ́ -nè à-rɪ́-sw-ɛ́l-à.
I.J. 3.sbj.I-aux 3.sbj.I-1pl.obj-die-appl-fv
‘Jesus died for us.’ (AC)
d. Dʒɪ́sʊ́ ʊ́ -nè à-rɪ́-sw-ɛ́l-ɛ́l-à mó sɪ ̀-fápáánò-ŋ̀.
I.J. 3.sbj.I-aux 3.sbj.I-1pl.obj-die-appl-foc-fv loc VII-cross-loc
‘Jesus died for us on a cross.’ (ALC on AC)

In order to avoid confusion in the discussion of markers that act as applicative

markers in some of their uses only—as we did for the Nyole semantically-motivated
clause alternation shown in (12)—we use the term oblique-registration construction
for constructions involving verbal coding of the pragmatic prominence of an oblique
without any change in the structure of the clause, like the one illustrated in (14). The
term registered oblique can be used for the phrase whose pragmatic status is high-
lighted. (The alternative terms “oblique-highlighting construction” and “highlighted
oblique”, respectively, could also be used.)6

6 An alternative terminology distinguishes “promotional applicativization proper” from “non-promo-

tional registration applicativization” (see Hernández-Green 2016 for some references). Nevertheless,
Nichols (1992) and many later studies use registration in a different sense, according to which most
default applicative markers are instances of registration (i.e., they do not express any features of the
dependent—like person, gender, or number of the AppP—on the verbal head), as opposed to indexation
(i.e., they do express some features of the dependent on the head). See also Pacchiarotti’s (2020) Type-D
applicatives for Bantu in the context of valency neutrality.
12 Fernando Zúñiga and Denis Creissels

Further consider Examples (15)–(17) below from three Mayan languages, where
clauses with a fronted and focused constituent expressing a machete (bearing the
semantic role of Instrument) and whose predicate takes the marker -b’e contrast with
simple clauses with the machete as a postverbal adjunct (omitted here). In K’iche’ (15),
the result of the b’e-operation is an AC with the Instrument ch’iich’ ‘machete’ promoted
to primary object. In Tz’utujil (16), the resulting AC looks just like the one in (15), but the
Instrument machat ‘machete’ is actually a secondary object (among other things, it is
not cross-referenced on the verb). Lastly, in Kaqchikel (17), the result of b’e-suffixation
is not applicativization, since the fronted participant is still an adjunct with instrumen-
tal marking (rik’in), despite the applicative-like verbal marking (see Mora-Marín 2003
for the individual analyses):

(15) K’iche’ (Mayan; Kaufman 1990: 78)

Ch’iich’ k-Ø-in-rami-b’e-j lee chee’.
machete incompl-3sg.P-1sg.A-cut-appl-tr det wood
‘A machete is what I cut the wood with.’ (AC)

(16) Tz’utujil (Mayan; Dayley 1981: 27)

Machat x-in-re-ch’oy-b’e-ej.
machete compl-1sg.P-3sg.A-cut-appl-tr
‘It was a machete that he cut me with.’ (AC)

(17) Kaqchikel (Mayan; Dayley 1981: 27)

R-ik’in jun machät x-i-ru-sok-b’e-j.
3sg-ins a machete compl-1sg.P-3sg.A-wound-oblreg-tr
‘With a machete he wounded me.’ (ALC)

Finally, note that Examples (15)–(17) above from Quichean Mayan show a situation
where applicativization and oblique registration appear in different languages. Some
Oaxaca Mixean languages are remarkable because applicativization and oblique regis-
tration coexist there in the same language; Zavala (2015) convincingly argues that this
is the case in the Mixean varieties spoken in Totontepec and Tamazulápam. In (18) from
the latter, for instance, a monotransitive base verb ja:p ‘shovel’ occurs in three alternat-
ing constructions. In the first two, we find the Instrument as a non-core participant and
the element më:t ‘with’, occurring as a preposition in (18a) and as a so-called preverb in
(18b). In (18c), however, the Instrument is a primary object, that is, it is in P role. There-
fore, the verbal prefix të- has two related but distinct functions in the language: it is an
oblique-registration marker in (18b) and an applicative marker in (18c):

(18) Tamazulápam Mixe (Mixe-Zoquean; Zavala 2015: 214–215)

a. N-ja:p-py më:t päl.
1.A-shovel-incompl.indep with shovel
 1 Applicative constructions: An introductory overview 13

b. Päl më:t n-të-ja:p-y.

shovel with 1.A-oblreg:ins-shovel-incompl.dep
c. Päl n-të-ja:p-py.
shovel 1.A-appl-shovel-incompl.indep
All three: ‘I shoveled it.’ (Sp. lo levanté con la pala in the original)

1.2.3 Summary

Table 3 below schematically summarizes an important part of the constructional space

covered by applicatives and lookalikes. (Lexicalized applicatives are not included here;
the numbered examples provided in §§ 1.2.1–1.2.2 are cross-referenced in the table.)
The vantage point / reference construction is represented by cell (a) in the table
and consists of an [S/A V (X)] clause, which features an argument in S/A role, a pred-
icate—represented as V for convenience here—and either an X argument in oblique
role (in the case of an optional applicative) or nothing (in the case of an obligatory
applicative).7 The prime symbol (ʹ) represents formal differences in the predicate (V vs.
Vʹ; these differences must be privative) and the oblique (X vs. Xʹ; these differences can
be of any kind). This constructional space is structured through four binary distinc-
tions. The first two are morphological and consist of the formal opposition between
unmarked and marked predicates (viz. the two columns) and the difference in formal
marking between the two obliques X and Xʹ (viz. the first and second rows). The other
two distinctions are syntactic and concern the status of the non-S/A argument, that is,
the distinction between oblique and non-oblique on the one hand and the one between
non-core and core on the other (viz. the second, third, and fourth rows).

Table 3: Selected applicatives and applicative lookalikes.

Non-S/A argument Predicate morphology

Unmarked Marked
Oblique (a) S/A V (X) (e) S/A Vʹ (X)
(12b) and (14b)
(b) S/A V Xʹ (f) S/A Vʹ Xʹ
(8a)
Non-oblique Non-core (c) S/A V D (g) S/A Vʹ D
(7a)
Core (d) S/A V P (h) S/A Vʹ P
(5a), (6a), and (9a)

7 For reasons of readability, Table 3 ignores the following detail: the reference construction may also
feature an argument in P role that undergoes no change, unless another argument is installed or pro-
moted in P-role, in which case the initial P may be demoted depending on language-specific rules.
14 Fernando Zúñiga and Denis Creissels

The non-shaded cells (f)–(h) represent different syntactic types of applicatives (viz.
those promoting or installing either a core argument in P role, a non-core, non-oblique,
argument in D role, or an oblique argument in X role). Cell (h) actually represents one
of the narrowest definitions of the phenomenon: having both overt predicate marking
and P status for the non-S/A argument.
The shaded cells (b)–(e) represent different kinds of lookalikes. Cells (b)–(d) rep-
resent syntactic lookalikes that participate in uncoded alternations with (a); the mor-
phosyntactic status of the argument in non-S/A role in these clauses shows variation,
but the predicate is invariably unmarked. One kind of morphological lookalike is only
indirectly represented in the table: applicative deponents are found in constructions of
the types (f)–(h) when there is no (a)-construction because the unmarked verb is not
in use. Cell (e) represents a morphological lookalike of a different kind: the predicate
bears applicative-like marking, but the syntactic status of the non-S/A argument in the
clause in question and in the vantage clause in (a) are the same. That clause structure
is the product of what we have labeled “syntax-neutral intensification” and “oblique
registration” here (distinguishing the two subtypes we identified in § 1.2.2).

2 Morphological variation in applicative

constructions
The main parameters of morphological variation in ACs are the wordhood status (§ 2.1),
the grammaticalization status (§ 2.2), the allomorphy (§ 2.3), and the specificity (§ 2.4)
of applicative markers.

2.1 Wordhood

The grammatical/morphosyntactic and phonological/prosodic autonomy of applicative

markers shows variation. As a rough first approximation, markers can be either com-
pletely dependent (i.e., affixes) or completely autonomous (e.g., verbs or particles). They
can also fall in between: they can be grammatically autonomous but phonologically
dependent (i.e., clitics), or grammatically dependent but phonologically autonomous
(i.e., “unclitics”, as per Woodbury 2011, or “anticlitics”, as per Zúñiga 2014). All exam-
ples of applicatives provided hitherto illustrate the affixal type; some exceptions are
given in Section 2.2.
 1 Applicative constructions: An introductory overview 15

2.2 Grammaticalization status of applicative markers

Incipient applicative markers may be virtually indistinguishable from their lexical

etymons, which, if they are verbs, may head their own clause. Markers that have
evolved further along Hopper and Traugott’s (2003) “cline of grammaticality”, however,
may retain part of their semantic content and morphosyntactic features but occur as
functional, rather than as fully lexical, elements in monoclausal constructions. A special
case of this consists in applicative periphrases (Creissels 2010). These may be either
serial verb constructions, as in Baule (19), or converbal constructions. In one subtype
of the latter, the functional verb (e.g., a verb of giving) appears in a finite form while
the lexical verb appears as a converb, as in Marathi (20), Japanese, and many other
languages. In the other subtype, the lexical verb appears in a finite form while the func-
tional element is a converb, as in Mankon (21); this seems to be a much rarer pattern.

(19) Baule (Kwa; Creissels and Kouadio 2010: 34)

Ákísí à-tɔ ̀n duô à-màn Kòfí.
A. perf-cook yam perf-give K.
‘Akissi has cooked yam for Kofi.’

(20) Marathi (Indo-Aryan; Pardeshi 1998: 147–148)

Rām-ne Sitā-lā bāg zāḍ-ūn di-l-ī.
R.-erg S.-dat garden(f) sweep-cvb give-pst-sg.f
‘Ram swept the garden for Sita.’

(21) Mankon (Grassfields Bantu; Leroy 2003: 459)

Mà m↑ɨ́ fàʔá γ↑á mbó zɯ́ ə́.
1sg fut work give.nfin to 3sg.enunc
‘I will work for him.’

Note that the coverage of the present book is skewed toward grammaticalized
affixal applicatives. Only few chapters address applicativizing particles (in European
languages), compound applicatives (in Papuan and Kiranti languages), or applicativizing
particles and auxiliaries (in Cushitic languages).

2.3 Allomorphy

Some applicative markers are invariable, like English out- in outgrow and Mapudungun
-tu in illkutun ‘get angry with’. Others show phonologically conditioned allomorphy, like
Mapudungun -ñma in (13) above: roughly, the marker appears as -ñma after vowels, as
16 Fernando Zúñiga and Denis Creissels

-ma after glides and some other sonorant non-vocoids, and as -üñma elsewhere.8 Like
other grammatical markers, applicatives sometimes fuse with adjacent formatives in
some languages.
Yet other markers show grammatically conditioned allomorphy. For instance, while
the primary applicative -omo of Blackfoot in (31) below is invariable, several of the
secondary applicatives are not. The locative applicative is ist- in the imperative and it-
elsewhere. The associative applicative is iihp- word-initially, omohp- immediately after
personal prefixes, and ohp- elsewhere (as in [29] below). Lastly, the allomorphs iiht-,
omoht-, and oht- of a marker targeting several different semantic roles follow the same
allomorphy rules as the associative applicative (Frantz 2009: 92–94); Example (31b)
illustrates the second of these allomorphs, whereas the following examples illustrate
the first (22a) and the third (22b):

(22) Blackfoot (Frantz 2009: 92, glosses adapted)

a. Iiht-waawayáki-aa-wa miistsís-i.
pst\sec.appl-hit.ta-dir-3sg.prox stick-nspec
‘S/he was hit with/by a stick.’
b. Nit-yáak-oht-waahkayi áípottaa-wa.
1-fut-sec.appl-go.home.ai plane-prox.sg
‘I’ll go home by plane.’

Finally, some applicatives show lexically conditioned allomorphy. The benefactive/

malefactive applicative of Aguaruna, for instance, is -tu for some verbs (23a) and -hu
for others (23b):

(23) Aguaruna (Chicham; Overall 2017: 303, 320)

a. Ami=na bala=na ɨŋkɨ-tu-hama-ka-mɨ ̃.
2sg=acc bullet=acc put.in-appl-2.obj-pfv-3.rec.pst.decl
‘He’s loaded his gun (to shoot) you.’
b. Wi numi=na tsupi-hu-ka-ta-himɨ-i.
1sg wood=acc cut-appl-pfv-imm.fut-1→2pl-decl
‘I will cut wood for you (pl).’

2.4 Specificity of applicative markers

Applicative markers can be dedicated or polysemous. Well-known cases in which the

markers may also have other, arguably related, functions in the same language include
those that applicativize some verbs but passivize, antipassivize, causativize, or decaus-

8 See Zúñiga (2009, this volume) for more details.

 1 Applicative constructions: An introductory overview 17

ativize others. See Zúñiga and Kittilä (2019: § 8.2.1) for a brief overview and some
references. See also Malchukov (2015, 2016) for more on such “voice ambivalence” /
“ambivalent voice”, and Bahrt (2021) for an in-depth study of co-expression patterns of
voice markers / valency operators (for applicatives, see §§ 4.3.1, 4.4, 5.2.3, 5.3.3, and 7.6
therein, as well as Malchukov 2017).
For instance, -aʔam can either applicativize or antipassivize the same verb q’ətxw-
‘burn’ in Sliammon (24):

(24) Sliammon (Coast Salishan; Watanabe 2015: 1331, 1327)

a. Q’ətxw-əxw-Ø=č šə=θ=pipa-h-uɬ.
burn-ntr-3.obj=1sg.sbj.ind det=2sg.psr=paper-ep-pst
‘I burned your paper.’
b. Q’ətxw-aʔam-θi=tθəm ʔə=tə=pəp-pipa.
burn-appl-ctr:2sg.obj=1sg.sbj.ind.fut obl=det=pl-paper
‘I will burn the paper for you (sg).’
c. Q’ətxw-aʔam=č ʔə=tə=pipa.
burn-antip=1sg.sbj.ind obl=det=paper
‘I burn paper.’

Likewise, -esh can either applicativize or causativize the same verb -men ‘break’ in Kin-
yarwanda (25):

(25) Kinyarwanda (Bantu JD.61; Jerro 2017: 753)

a. Habimana y-a-men-a igi-kombe.
H.(I) 3.sbj.I-pst-break-ipfv VII-cup
‘Habimana broke the cup.’
b. Habimana y-a-men-esh-eje igi-kombe in-koni.
H.(I) 3.sbj.I-pst-break-appl-pfv VII-cup IX-stick
‘Habimana broke the cup with a stick.’
c. Habimana y-a-men-esh-eje umw-ana igi-kombe.
H.(I) 3.sbj.I-pst-break-caus-pfv I-child VII-cup
‘Habimana made the child break the cup.’

The use of the same marker for applicative and causative constructions is particularly
frequent cross-linguistically. Labels usually employed in the literature in order to refer
to the phenomenon include “causative/applicative syncretism” (Shibatani and Pardeshi
2002: 116–122), “applicative/causative isomorphism” (Peterson 2007: 64–68, 133–140),
“causative-applicative polysemy” (Malchukov 2017: 403; Creissels, forthcoming: Ch.
14.4.1), and “causative-applicative syncretism” (Bahrt 2021: 93–96); a slightly different
term has been proposed by Croft (2022: 285): “causative-applicative co-expression”.
It is common to find lexical restrictions on such co-expression patterns. For
instance, while many Kinyarwanda verbs from several semantic verb classes allow
18 Fernando Zúñiga and Denis Creissels

both the instrumental-applicative and the causative interpretation of -esh, patientive

intransitives (e.g., rumbur- ‘bloom’) usually exclude the former, and transitives that
entail the use of an Instrument (e.g., kat- ‘cut’) typically exclude the latter (Jerro 2017:
757). Similarly, the Walapai verbal suffix -(w)o applicativizes agentive intransitives like
‘sing’ (26a–b) and transitives like ‘make’ (26c–d) but causativizes patientive intransi-
tives like ‘cry’ (26e–f):

(26) Walapai (Yuman; Ichihashi-Nakayama 1996: 228–231)

a. Jean-ch swa:d-k-i.
J.-sbj sing-3-aux
‘Jean is singing.’
b. Jean-ch ba ma-swa:d-o-y-k-i.
J.-sbj pl.obj 3→2-sing-appl-fut-3-aux
‘Jean will sing for you all.’
c. Nya-ch he’ yo:v-wi-ny.
1sg-sbj dress make-aux-pst
‘I made a dress.’
d. Nya-ch he’ nyi-yo:v-ò-wi-ny.
1sg-sbj dress 1→2-make-appl-aux-pst
‘I made you a dress.’
e. Nya-ch mi:-yi.
1sg-sbj cry-aux
‘I am crying.’
f. Ma:-ch nya mi-mi:-wo-ng-wi-ny.
2sg-sbj 1sg 2→1-cry-caus-2-aux-pst
‘You made me cry.’

See Section 3.4 for instances in which it is the intransitive-transitive divide that plays a
central role in the distribution of such co-expression patterns.

3 Syntactic variation in applicative constructions

The main syntactic parameters along which applicative constructions vary are the
status of the applied phrase (§ 3.1) and the status of the participant bearing the same
semantic role as the applied phrase in the base construction (§ 3.2). Other important
parameters are the status of different objects in the applicative construction (§ 3.3) and
the sensitivity of applicatives to predicate valency (§ 3.4).
 1 Applicative constructions: An introductory overview 19

3.1 Status of the AppP in the AC

In most of the examples given hitherto, the AppP is encoded as P. ACs with AppPs that
are syntactically less prominent include arguments in D role (i.e., indirect objects or
“datives” with the Recipient of verbs of giving as semantic prototype, and whose coding
differs from the one of the direct object while sharing with it some properties that make
it more core-like than oblique-like; see Creissels, forthcoming), as well as those in an
oblique X role. For comparative purposes, then, we can distinguish between P-, D-, and
X-applicatives.9 For descriptive, language-specific purposes, we may use traditional ter-
minology and distinguish between direct, primary, indirect, secondary, and oblique
applicatives.
Examples of D-applicatives are found in Kartvelian and Northwest Caucasian. In
the following sentence pair, for instance, the Beneficiary is encoded differently from the
argument in P role in (27a)—st’at’ia ‘(an) article’—and is outside the clausal core (i.e., it
is an adjunct) in (27b):10

(27) Georgian (Kartvelian; Manana Topadze, p.c.)

a. Da-v-u-ts’er-e bavšv-s st’at’ia.
pvb-1sg-appl-write-1sg.aor child-dat article[nom]
‘I wrote an article for the child.’ (AC)
b. Da-v-ts’er-e st’at’ia (bavšv-isa-tvis).
pvb-1sg-write-1sg.aor article[nom] child-gen-for
‘I wrote an article (for the child).’ (BC)

Examples of X-applicatives are found in Atlantic languages. In Seereer (28), for

instance, the base verb -ret ‘go’ takes an argument expressing Destination (optionally
marked by the semantically underspecified preposition na in [28a]) and cannot take
a Source argument unless applicativized. Note, however, that the coding of the latter
argument is even less P-like in (28b) than in (28a), na now being obligatory:

(28) Seereer (Atlantic; Renaudier 2012: 183)

a. A-ret-a (na) marse.
3sg.sbj-go-compl prep market
‘S/he went to the market.’ (BC)
b. A-ret-it-a na marse.
3sg.sbj-go-appl-compl prep market
‘S/he left the market.’ (AC)

9 Creissels (forthcoming) also uses P-, D-, and X-applicatives, but the latter type specifically refers to
constructions expressing a participant that cannot appear in the BC.
10 The markers u- (3rd person) and i- (1st/2nd person) are called “version” markers in Georgian studies;
see Tuite (this volume).
20 Fernando Zúñiga and Denis Creissels

A different brand of X-applicatives can be found in Algonquian languages. In

Examples (22) above and (29) below from Blackfoot, as well as in Example (30) below
from Ojibwe, for instance, the introduced participant can be expressed in a monoclausal
construction only as an AppP; the latter is a Comitative (‘with that book’) in (29) and bears
a less usual semantic role (‘in the form of a man’) in (30). This constituent is arguably
a secondary object in Blackfoot, which seems to lack bona fide obliques; Rhodes (2010)
shows that the Ojibwe AppP is less syntactically prominent than a secondary object but
more like a core argument than an adjunct:11

(29) Blackfoot (Algonquian; Frantz 2009: 92; underlying form and glosses adapted)
Yáak-ohp-innisi’yi-yi-aaw om-yi sináákia’tsis-yi.
fut-sec.appl-fall.ai-3pl-3pl dem-inan.sg book-inan.sg
‘They will fall with that book.’

(30) Ojibwe (Algonquian; Rhodes 2010: 306)

Aniniw-ing=sh go naa o-gii#iN-aabam-aa-an
man-loc=ptcl ptcl ptcl 3.A-pst#appl-see.ta-3.P.anim-obv
aniwi manidoo-an.
this.obv spirit-obv
‘He saw the spirit in the form of a man.’ (AC)

Some languages have distinct applicatives that allow different kinds of AppPs,
as illustrated in (31) from Blackfoot. In (31a), the P-applicative / primary applicative
marker -omo follows the verb stem and introduces a Beneficiary as primary object;12 in
(31b), the X-applicative / secondary applicative marker omoht- precedes the verb stem
and introduces a Source as secondary object:

(31) Blackfoot (Frantz 2009: 103, 93, glosses adapted)

a. Nit-ii-yIssksipist-omo-aa-wa n-itákkaa-wa w-óta’s-iksi.
1-pst-tie.ta-pri.appl-dir-3sg.prox 1-partner-prox.sg 3-mount-pl.anim
‘I tied up his horses for my partner.’
b. N-omoht-o’too Lethbridge.
1-sec.appl-arrive.ai L.
‘I came from Lethbridge.’

11 Rhodes (2010) calls this grammatical relation “relative root complement” and the oblique applicative
marker “relative root”. Frantz (2009) calls Blackfoot secondary applicative markers like ohp- in (29) and
omoht- in (31b) “linkers” and does not address the syntax of the arguments they introduce.
12 Blackfoot has both an equipollent benefactive marker -o (4a) and a benefactive applicative -omo
(31a). The former is the only so-called final suffix on the stem while the latter is always added to another
final; all the related stems meaning ‘tie’ (viz. transitive animate yIssksip-ist-, transitive inanimate yIssk-
sip-i-, and animate intransitive yIssksip-ist-aki-) feature final suffixes of their own.
 1 Applicative constructions: An introductory overview 21

Finally, consider the following example from a variety of Southern Quechua:

(32) Ayacucho Quechua (Quechuan; Parker 1969: 71)

Ñuqa-paq rima(-pu-wa)-nqa.
1sg-for speak-appl-1.obj-3.sbj.fut
‘S/he will speak for me (in my stead or on my behalf).’

The sentence is also grammatical without the applicative -pu and the 1st-person-object
marker -wa on the verb, in which case the constituent ñuqapaq ‘for me’ is simply an
optional adjunct. With those two suffixes, however, the status of the non-agentive con-
stituent in the clause cannot be unambiguously characterized; it is neither a run-of the-
mill object (since such arguments take either accusative -ta, allative -man or perhaps
even sociative -wan instead of benefactive -paq) nor a simple oblique (since such con-
stituents do not trigger object marking on the verb). The status and the distribution of
such apparent morphology-syntax mismatches are under-researched topics.
More generally, little is known about the distribution and the evolution of different
syntactic types of applicatives. Further research should also explore the extent to which
ACs conform to or deviate from the rest of the morphosyntactic patterns of any given
language, that is, the extent to which some ACs are best seen as subtypes of clause types
already present, as novel clause types complementing the existent ones, or as some-
thing else entirely.

3.2 Status of the semantic equivalent of the AppP in the BC

In Example (1) from Zapotec, the co-Agent is an adjunct in the base construction. Never-
theless, the participant corresponding to the applied phrase may be obligatorily absent
from the base construction, as in (33) from Tswana. In the latter language, the only way
to express a participant with the role of Beneficiary in a monoclausal construction is
via the AC:

(33) Tswana (p.k.)

a. Kì-tɬàà-bɛ́rɛ́k-ɛ́l-à Kítsɔ́ màítsɪ́bʊ̀ːá.
1sg-fut-work-appl-fv K.(I) evening(VI)
‘I’ll work for Kitso this evening.’ (AC)
b. Kì-tɬàà-bɛ́rɛ́k-á màítsɪ́bʊ̀ːá.
1sg-fut-work-fv evening(VI)
‘I’ll work this evening.’ (BC)
22 Fernando Zúñiga and Denis Creissels

Terminologically, we capture the difference between cases like those like Zapotec in (1)
and those like Tswana in (33) by distinguishing optional from obligatory applicatives.13
The participant bearing the same semantic role as the AppP in the BC may also
occur as a low-prominence object, as in (34) below. Rather than being a double-object
construction (henceforth: DOC),14 the BC in (34b) has a P argument (Schokolade ‘choco-
late’, a direct object in the accusative) and a D argument (den Kindern ‘to the children’,
an indirect object in the dative):

(34) German (p.k.)

a. Wir be-schenk-en die Kind-er mit Schokolade.
1pl.nom appl-give.as.gift-1pl art.acc.pl child-pl[acc] with chocolate
‘We present the children with chocolate as a gift.’ (AC)
b. Wir schenk-en den Kind-er-n Schokolade.
1pl.nom give.as.gift-1pl art.dat.pl child-pl-dat chocolate
‘We give the children chocolate as a gift.’ (BC)

Table 4 below summarizes the different applicative types discussed based on the
syntactic status of the participant corresponding to the applied phrase.

Table 4: Status of the AppP participant.

Status in BC
X —
(optional) (obligatory)
P P-appl
Status
in AC

D D-appl
X (?) X-appl

The two cells at the bottom of the table merit special attention. The one on the right-
hand side is what Creissels (forthcoming) calls “X-applicative” (see Footnote 9) and is
illustrated in (35); the pot in which the porridge is to be cooked is expressed as an X in
the AC but cannot be expressed in the BC:

13 Alternative terminologies distinguish “dynamic applicatives” from “non-dynamic applicatives”

(Donohue 2003) and “applicatives” from “objectives” (Cysouw 2023). See also Pacchiarotti’s (2020)
Type-B and Type-A applicatives.
14 The term usually refers to clauses with two morphologically indistinguishable objects. Further note
that the comparative term “double-P construction” is narrower, because it refers to instances in which
both non-S/A arguments show the coding properties that characterize P’s (not D’s) in monotransitive
clauses.
 1 Applicative constructions: An introductory overview 23

(35) Tswana (p.k.)

a. Lʊ̀rátɔ́ !ʊ́ -tɬáá-àpɛ ̀-ɛ ̀l-à mʊ̀-tɔ ̀χɔ́ mó pìtsé-ŋ̀
L.(I) 3sbj.I-fut-cook-appl-fv sg-porridge(III) loc pot(IX)-loc
é !
tʊ̂:nà.
lk.IX big.IX
‘Lorato will cook the porridge in the big pot.’ (AC)
b. Lʊ̀rátɔ́ !ʊ́ -tɬáá-àpàj-à mʊ̀-tɔ ̀:χɔ́.
L.(I) 3sbj.I-fut-cook-fv sg-porridge(III)
‘Lorato will cook the porridge.’ (BC)

The cell on the left-hand side could logically exist, but we are not aware of any concrete
instances in which the participant in question can appear as oblique in both BC and AC,
specifically with a different coding.
Further research should explore whether present-day obligatory applicatives may
have originated in erstwhile optional constructions (and the BC has ceased to be used
with a counterpart of the applied phrase of the AC, for instance). Alternatively, ACs may
begin as obligatory and become optional later, perhaps constrained by lexical or gram-
matical factors. Some data from the languages of Senegal suggest that language contact
may play a role, with sporadic occurrences of prepositional coding of Beneficiaries
modeled after another language in more or less pidginized varieties of languages that
normally code beneficiaries via an obligatory P-applicative construction.

3.3 Status of different objects in the AC

Some P-applicative constructions are DOCs; in such cases, the literature has captured
similarities and differences in the status of those two objects (i.e., two P’s) via the ter-
minological distinction between symmetry and asymmetry. Some authors distinguish
between symmetric and asymmetric applicatives (e.g., Pylkkänen 2008) while others
distinguish more generally between symmetrical-object and asymmetrical-object
languages (e.g., Bresnan and Moshi 1993). The following examples from Kikuyu (36)
and Chingoni (37) illustrate the difference between the two types. Despite the superfi-
cial similarity of the postverbal constituents in both sentences, diagnostics regarding
constituent order rigidity, passivization, and indexing show that the two objects in (36)
are equally prominent, whereas the object closest to the verb in (37) (vandu ‘people’)
outranks the rightmost constituent (ugimbi ‘beer’):

(36) Kikuyu (Bantu E51; Ngonyani and Githinji 2006: 35)

Mũ-geni a-ra-gũr-ĩ-ire ci-ana mũ-bira.
I-guest 3.sbj.I-prog-buy-appl-asp VIII-child III-ball
‘The guest bought children a ball.’
24 Fernando Zúñiga and Denis Creissels

(37) Chingoni (Bantu N12; Ngonyani and Githinji 2006: 35)

M-geni i-gul-i va-ndu u-gimbi.
I-guest prs-buy-appl II-person XIV-beer
‘The guest is buying beer for people.’

DOCs may be restricted or even absent in a given language. In such cases, ACs
may be almost valency-neutral, as in (34) above, where the BC has P and D arguments
(a direct and an indirect object), and the AC has a P and an X instead (a direct and an
oblique object). Alternatively, they may be redirecting, as in (38) below, where the P
and the X argument (the direct and the oblique objects) merely become rearranged
semantically and the operation is valency-neutral (see also § 3.4):

(38) German (p.k.)

a. Wir be-sprüh-te-n die Pflanzen mit Pestizid.
1pl.nom appl-spray-pst-1pl art.acc.pl plant-pl with pesticide
‘We sprayed the plants with pesticide.’ (AC)
b. Wir sprüh-te-n Pestizid auf die Pflanz-en.
1pl.nom spray-pst-1pl pesticide on art.acc.pl plant-pl
‘We sprayed pesticide on the plants.’ (BC)

3.4 Sensitivity to syntactic valency

This parameter refers to the number of core syntactic arguments in the base construc-
tion. In (39) below, for instance, the base predicate is syntactically monovalent and has
no objects; since the psych verb illku- ‘get angry’ has only two semantic arguments (viz.
Experiencer and Stimulus), the BC has no other non-agentive participants that might be
core syntactic arguments; ñi chaw mew ‘with my father’ in (39b) is an optional adjunct.
(The clause can also include adjuncts expressing spatiotemporal Locations, but these
are not relevant in the present context.)

(39) Mapudungun (p.k.)

a. Illku-tu-fi-n ñi chaw
get.angry-appl-3.obj-1sg.ind 1sg.psr father
‘I got angry with my father.’ (AC)
b. Illku-n (ñi chaw mew).
get.angry-1sg.ind 1sg.psr father postp
‘I got angry (with my father).’ (BC)

ACs with syntactically bivalent BC counterparts are by no means rare, as shown in

Polinsky (2013) and Moroz and Polinsky (this volume). Example (40) illustrates an instance
where the AC has two objects (teman saya ‘my friend’ and nasi gorengnya ‘the fried
 1 Applicative constructions: An introductory overview 25

rice’) while the BC has an object (nasi gorengnya ‘the fried rice’) and an adjunct (untuk
teman saya ‘for my friend’):

(40) Indonesian (Malayic; Vamarasi 1999: 73)

a. Ibu memasak-kan teman saya nasi goreng-nya.
mother tr.av.cook-appl friend 1sg rice fried-def
‘Mother cooked my friend the fried rice.’ (AC)
b. Ibu memasak nasi goreng-nya untuk teman saya.
mother tr.av.cook rice fried-def for friend 1sg
‘Mother cooked the fried rice for my friend.’ (BC)

In some languages, only oppositions like the one with Mapudungun ‘get angry
(with)’ in (39) exist, namely between a bivalent AC and a monovalent BC. (Mapudungun
is not one of those languages; see Zúñiga, this volume.) In others, only oppositions like
the one with Indonesian ‘cook’ in (40) exist, namely between a trivalent AC and a biva-
lent BC; even though ACs with trivalent BCs are less widespread, they do exist as well.
In yet other languages, different kinds of opposition coexist and sometimes require
specialized markers. In Salishan, for instance, some applicatives derive bivalent predi-
cates from base monovalent ones (in which case the AppP is typically a Stimulus, Goal,
or Topic of Speech) while others derive trivalent or bivalent predicates from base biva-
lent ones (in which case the AppP is a Recipient, Beneficiary, Source, or Possessor).
Examples (41)–(42) illustrate such a division of labor:

(41) Okanagan (Interior Salishan; Kiyosawa and Gerdts 2010: 2)

a. N-x̌íł-mə-nt-s-ən.
loc-afraid-appl-tr-2sg.obj-1sg.sbj
‘I got scared of you (sg).’ (AC)
b. Kən n-x̌íł.
1sg.sbj loc-afraid
‘I got scared.’ (BC)

(42) Shuswap (Interior Salishan; Kiyosawa and Gerdts 2010: 3)

a. M-k̕úl-x-t-s γ núx̌ʷənx̌ʷ tə mim̓x.
perf-make-appl-tr-3.sbj det woman obl basket
‘She made a basket for the woman.’ (AC)
b. M-k̕úl-n-s γ mim̓x.
perf-make-tr-3.sbj det basket
‘She made the basket.’ (BC)

Base predicate valency may constrain applicativization in principle, as a tendency,

or not at all. AC may begin as a monotransitivization device (i.e., ‘X get scared’ → ‘X get
scared of Y’), a ditransitivization device (i.e., ‘X make Y’ → ‘Y make Y for Z’), a general
26 Fernando Zúñiga and Denis Creissels

transitivization device (i.e., applicable to both), etc., and later become either more freely
applicable or more limited. Further research may unearth clear cases of ACs extending
their applicability range due to contact.
Table 5 below summarizes the different applicative types discussed above based
on the valency of both related clauses. Redirecting applicatives are valency-neutral
and occupy the top-left-to-bottom-right, shaded, diagonal (the 1-1 cell is logically empty,
since applicatives do not introduce arguments in S/A role). Transitivizing applicatives
are valency-increasing and occupy the cells immediately below the previous diagonal.
The cell corresponding to v3 → v2 operations is a special case of detransitivization (e.g.,
with German schenken/beschenken ‘give as a gift’ in [34]).

Table 5: Valency correspondences.

BC valency
1 2 3
1
AC valency

2 monotr redir detr

3 ditr redir
4 tritr

4 Semantic variation in applicative constructions

The parameter of semantic variation that has received most attention in the literature is
whether the semantic role of the AppP has to be inferred from the context or is denoted
by the particular construction (§ 4.1). A parameter that has received disparate treat-
ment in the literature is the kind of semantic role borne by the AppP (§ 4.2).

4.1 Semantic role specificity

Broad applicatives are not difficult to find; Indonesian -kan, for instance, derives not
only benefactive (as in [40] above) but also instrumental applicatives:

(43) Indonesian (Sneddon 1996: 79)

a. Dia meng-ikat-kan tali ke anjing.
3sg av-tie-appl rope to dog
‘He tied the rope to the dog.’ (AC)
b. Dia meng-ikat anjing dengan tali.
3sg av-tie dog with rope
‘He tied up the dog with a rope.’ (BC)
 1 Applicative constructions: An introductory overview 27

Peterson (2007: 39) calls such constructions—“when a single applicative marker refers
to multiple thematic participant types (regardless of whether or not these participants
are treated identically in terms of morphosyntax)”—generalized applicatives. Note
that the term covers not only instances where a particular marker can applicativize
the same predicate with different semantics but also those where the marker behaves
differently with predicates from different classes (e.g., valency classes).15
Further note that Indonesian -kan can function as a benefactive/instrumental
applicative with monotransitives, as in (43) above, but as an (optional) “antibenefac-
tive” marker with ditransitives, as in (44) below. The construction in (44b) promotes the
Theme and demotes the Recipient vis-à-vis the DOC in (44a):

(44) Indonesian (Chung 1983: 234, cited in Malchukov 2017: 18)

a. Ali beri [wanita itu] [surat].
A. give woman the letter
‘Ali gave the woman a letter.’
b. Ali beri(-kan) [surat] [kepada wanita itu].
A. give-appl letter to woman the
‘Ali gave a letter to the woman.’

Malchukov (2017: 17–21) calls this phenomenon “applicative reversal” (and considers it
a special case of his “markedness reversal”).
Semantically specific applicatives are easy to find as well, especially those that
introduce Beneficiaries, Instruments, accompanying Agents/Themes, and Locations
(Peterson 2007). Hakha Lai is particularly interesting in that it has not only the cross-lin-
guistically common benefactive, instrumental, and comitative applicatives but also a
series of idiosyncratic and rare applicatives, namely additional-benefactive -tseʔm (45a),
malefactive-allative -hnoʔ (45b), prioritive -kaʔn (45c), and relinquitive -taak (45d):

(45) Hakha Lai (Tibeto-Burman; Peterson 2007: 41, 19)

a. Thiŋ ʔa-ka-laak-tseʔm.
wood 3sg.sbj-1sg.obj-carry-appl1
‘He carried wood for me (in addition to carrying wood for himself).’
b. Kheeŋ ʔa-ka-hloʔn-hnoʔ.
dish 3sg.sbj-1sg.obj-throw-appl2
‘She threw the dish at me.’

15 Note Gil’s even broader notion: “A generalized voice marker is a marker M which, when applied to
form X, marks the argument of X bearing the thematic role T as having a set of properties P” (2002:
276, emphasis in the original). This author sees such voice markers as ranging from “weak” to “strong”,
depending on how many properties—for instance, morphosyntactic subject properties—the target argu-
ment shows. The term is also used differently by some theoretical studies, where additional aspects of the
grammar-and-lexicon architecture are relevant (see, e.g., Svenonius 2014; but cf. Polinsky, this volume).
28 Fernando Zúñiga and Denis Creissels

c. Booy ʔa-ka-toon-kaʔn.
chief 3sg.sbj-1sg.obj-meet-appl3
‘He met the chief ahead of / before me.’
d. ʔa-law ʔa-ka-thloʔ-taak.
3sg.psr-field 3sg.sbj-1sg.obj-weed-appl4
‘He left me and weeded his field.’

Just as benefactive and malefactive applicatives are sometimes distinguished

overtly (as in Amharic; see Amberber, this volume), the comitative type cannot be simply
lumped together with what Messerschmidt (2022), following Melnar (1998), labels
“portative”. With the (rather broad) comitative type, the AppP is “understood to be a
co-participant with the subject in the performance of the action described by the verb”
(Peterson 2007: 18); see Example (1) above and Example (46) below from Hakha Lai:

(46) Hakha Lai (Peterson 2007: 18)

Ka-law ʔan-ka-thloʔ-pii.
1sg.psr-field 3pl.sbj-1sg.obj-weed-appl
‘They weeded my field (together) with me.’

By contrast, the (narrower) portative type found in Caddoan languages and Osage
occurs with motion verbs and allows the derived predicate to take an argument in P
role that is portrayed as being carried by an Agent that is also a Theme (i.e., an Agent-
cum-Theme argument) in S/A role, that is, as a Patient that is also a Theme (i.e., a Patient-
cum-Theme).16 See the following illustration of the use of portative-applicative ni- in
Caddo with the verb wid(i)- ‘arrive’:

(47) Caddo (Caddoan; Melnar 1998: 170)

a. Kac’ikán=ʔawi-wid-ah.
prior.quot=abs.sg-arrive-perf
‘They say he’s (already) here.’
b. Ni-wid-ah.
appl-arrive-perf
‘He brought it.’ (Lit. ‘He arrived carrying it.’)

16 The semantics of ni- and its allomorphs is probably broader in Caddoan, as seen from instances
where the applicative marker co-occurs with a prefix denoting an animate Patient: ‘arrive’, ‘come’, and
‘go’ all become ‘follow’ then (see Rood 1976:72 for Wichita and Melnar 1998: 171 for Caddo). Messer-
schmidt herself cites a similar example from Sierra Popoluca, where the addressee is told to ‘take (=
appl+go) a dog’ and ‘bring (= appl+come) a man’, and “it can be really difficult to tell [. . .] whether it is a
sociative causative, a portative or a comitative construction” (2022: 176).
 1 Applicative constructions: An introductory overview 29

Among the many cross-linguistic generalizations discussed in Peterson (2007:

202–230), one that has become received wisdom in studies of valency and voice is the
implicational relationship between different semantic roles. AppPs denoting Locations
and those expressing Causes presuppose the existence of other applicatives, while
instrumental, comitative, and benefactive/malefactive applicatives occur alone much
more freely, “as anchors, as it were, for the development of distinct morphology (via
extension of already existing applicative morphology or by grammaticalization of
independent applicative constructions)” (p. 229). From this perspective, the exact
synchronic distributions and diachronic connections between different applicatives in
the same language and linguistic group are particularly interesting.

4.2 Semantic role kind

Semantic roles can be seen as central or peripheral, either in binary terms or along
a continuum of involvement (Lehmann 2006). Central, maximally involved, roles are
constitutive of the state of affairs and include Agents, Forces, Themes, and Patients.
Peripheral, rather loosely involved, roles presuppose the presence of central ones and
include Comitatives, Instruments, and Beneficiaries. Other roles, like Sources, Goals,
Recipients, and Experiencers, occupy an intermediate zone.
Lehmann (2006) and many other studies have noted the cross-linguistic tendency
for central roles to be granted core argument status in the syntax and for peripheral
roles to be routinely expressed as oblique objects or adjuncts. Accordingly, there seems
to be a cross-linguistic tendency for ACs to work on peripheral roles; Comitatives,
Instruments, and Beneficiaries are indeed the most common roles with applicatives
worldwide (Peterson 2007). ACs that are supposedly less common work on central roles
like Themes and Patients, as illustrated by the following Tolai example:17

(48) Tolai (Oceanic; Mosel 1991: 248)

a. A vavina i mim=e ra tava.
art woman 3.sbj drink=appl art water
‘The woman drank the water.’ (AC)
b. A vavina i momo.
art woman 3.sbj drink
‘The woman drank (something).’ (BC)

17 Messerschmidt (2022) reanalyzes some causatives and other applicatives as portative applicatives,
which leads her to regard the latter construction—which has a Patient-cum-Theme in P role—as fre-
quent: she finds it in 22 out of 49 languages in her convenience sample (second only to benefactive
applicatives, with 27 languages, and significantly outranking her third type, viz. locative-directional
applicatives, with 10).
30 Fernando Zúñiga and Denis Creissels

Finally, consider the question of the exact interpretation of the semantic role of the
AppP. The literature has often noted that German be-ACs are systematically close, but
not identical, semantic counterparts of their BCs. In the following pair, for instance, the
old house is a Location in (49a) but rather a (non-prototypical) Patient in (49b). Some
studies have claimed, however, that this holism effect may well be the result of direct
object status and/or accusative coding, rather than of applicativization per se (Wechsler
2015: 308–309):

(49) German (p.k.)

a. Wir wohn-te-n damals in ein-em alt-en Haus.
1pl.nom dwell-pst-1pl at.the.time in one-dat old-dat.sg house
‘We lived at the time in an old house.’ (BC)
b. Wir be-wohn-te-n damals ein alt-es Haus.
1pl.nom appl-dwell-pst-1pl at.the.time one[acc] old-acc.sg house
‘We occupied at the time an old house.’ (AC)

That the exact interpretation of the semantic role of the AppP can be related to the
applicative marker, however, is seen in (50), where ihr Auto ‘their car’ is an accusa-
tive-marked direct object in both versions. In (50a), with packen ‘pack’, the Millers pack
the inside of their car; in (50b), with bepacken ‘pack’, the object is interpreted as cover-
ing both the trunk and the top of their car:

(50) German (based on Brinkmann 1997: 58–59)

a. Müllers packen ihr Auto immer als blieben sie ein halbes Jahr lang weg.
b. Müllers bepacken ihr Auto immer als blieben sie ein halbes Jahr lang weg.
Both: ‘The Millers pack their car as if they will be away for half a year.’

This effect seems to be a special case of a more general topological restriction placed on
German be-verbs: the be-AC “resists interpretations in which the event denotes move-
ment into the interior of an object”—an “exteriority constraint” (Wechsler 2015: 309). If
the outlet in (51a) and the glass in (51b) are interpreted as (atypically) referring to the
exterior of the corresponding entities, the clauses are grammatical; if they are intended
to refer to their interior, they are not:

(51) German (Brinkmann 1997: 58)

a. Ted be-wirft [die Wand] / (✶)[den Abfluss] (mit Dreck).
T. appl-throws art wall art outlet with dirt
Lit. ‘Ted throws the wall / the outlet with dirt.’
b. Sue be-gießt [den Braten] / (✶)[das Glas] (mit Wasser).
S. appl-pours art roast art glass with water
Lit. ‘Sue pours the roast / the glass with water.’
 1 Applicative constructions: An introductory overview 31

5 Applicatives and applicative lookalikes

in Classical Nahuatl
This section outlines applicativization in the language for whose description the term
was first introduced (see § 6.1 for an outline of the term’s subsequent history). We do not
provide an introduction to Classical Nahuatl as a languoid here, but we otherwise follow
the same blueprint as many chapters of the present book: we first provide the basics of
Nahuatl clause and verb structure required for the discussion of applicatives (§ 5.1); then
we list morphological (§ 5.2), syntactic (§ 5.3), and semantic hallmarks of applicatives in the
language (§ 5.4); lastly, we make some remarks on the most prominent lookalikes found
(§ 5.5). All examples given here are from Launey (1979: 192–209); translations are from
the English version of that work by Christopher Mackay (Launey 2011: 202–221). For the
description and analysis of applicatives, we consulted both that source and Launey (1994:
195–199). The reader is referred to Newman (1967), Sullivan (1988), and Andrews (2003)
for other sources that present a very similar view of applicativization in the language.

5.1 Basics of Nahuatl morphosyntax

5.1.1 Core argument realization

In Classical Nahuatl, the existence of a grammatical relation “subject” conflating the

single argument of monovalent verbs and the most Agent-like argument of bivalent verbs
is uncontroversial. Neither subjects nor objects are flagged. 1st- and 2nd-person subjects
are obligatorily indexed by means of prefixes (which are also used for non-verbal pred-
icates), and there is no overt 3rd-person subject prefix; irrespective of person, subject
plurality is marked by a suffix. Objects of monotransitive verbs and primary objects of
ditransitive verbs are obligatorily indexed by means of markers from a paradigm that
differs from the one used for subjects and includes a non-zero form for 3rd person. More
generally, primary objects of ditransitive verbs behave identically to objects of monotran-
sitive verbs. Secondary objects of ditransitive verbs share a number of properties with
primary objects (including accessibility to antipassivization) but differ from them in
other respects. The basic constituent order in verbal clauses is VSO, but discursively
salient subjects or objects (either topicalized or focalized) can move to preverbal position.

5.1.2 Morphological structure of verbal predicates

The structure of Classical Nahuatl verb forms can be schematized as a sequence of mor-
phological slots. Tables 6 and 7 below give an overview of the main fillers found in the
slots that precede and follow the root, respectively.
32 Fernando Zúñiga and Denis Creissels

Table 6: Pre-root slots of Nahuatl verb forms.

–8 –7 –6 –5 –4 –3 –2 –1 0

sbj obj pl direc mid antip1 antip2 IN Root

The first three preradical slots correspond to argument indexes. Slot –8 can be filled
by a subject index of first and second person; no overt marker occurs in this slot with a
third-person subject. The imperative marker x(i)- (which implies a second-person subject)
can also occupy this slot. Slot –7 can be filled by an object index. With transitive verbs, an
overt object index obligatorily occurs in this slot with all persons. With ditransitive verbs,
only the primary object is indexed. Slot –6 can be occupied by im-, which marks 3pl
objects (either primary or secondary).18 Slot –5 can be occupied by a directional marker.
Slots –4 through –2 correspond to voice categories. Slot –4 can be filled by the
middle voice marker (viz. no- with a 1sg subject, to- with a 1pl subject, mo- with a 2nd-
or 3rd-person subject, and ne- in the “impersonal middle”, i.e., if the passive marker is
also included in the verb form). Slots –3 and –2 can be filled by the human antipassive
marker tē- and the non-human antipassive marker tla-, respectively.
Lastly, the slot immediately preceding the root can be filled by an incorporated
nominal lexeme.

Table 7: Post-root slots of Nahuatl verb forms.

0 1 2 3 4 5

Root caus appl pass tam pl

The first three postradical slots also correspond to voice categories. Slot 1 can be
filled by the causative marker -(l)tia.19 Slot 2 can be filled by the applicative marker,
which has the allomorphs -lia, -ia, -huia, or -lhuia. Slot 3 can be filled by the impersonal
marker -hua or the passive marker -lo, the uses of which overlap to some extent.
Slot 4 can be filled by a variety of tam suffixes, including the so-called particip-
ial suffix (which by itself expresses a value that can be labeled “narrative past”),20 the
sequence [z- + participial suffix] (which expresses the value “future”), the incompletive
past marker -ya, the potential marker -ni, and others (Launey 1994: 29).

18 The marking of object plurality by means of im- is not governed by the plurality of referents, but
by plurality marking on the corresponding object NP. The general Nahuatl rule (which allows for some
exceptions) is that only animate NPs can be marked for plural.
19 -tia is generally used for causatives of intransitives, and -ltia for causatives of transitives, but there
are exceptions.
20 The “participial suffix” selects a special form of the verb stem and has four distinct allomorphs (in-
cluding a zero allomorph).
 1 Applicative constructions: An introductory overview 33

The final slot can be occupied by a formative marking subject plurality. Subject
plurality is obligatorily marked in this slot, whereas subject singularity is not overtly
marked.21
In addition to the morphological slots listed above, the verb form may be preceded
by particles that are not attached to the verb morphologically, but have limited mobil-
ity and contribute to the expression of grammaticalized tam values. These particles
include ō (conventionally written as if it were a prefix), which, in combination with the
“participial suffix” in the tam slot, expresses a perfect meaning; the optative particle mā
and its negative counterpart mācamo; ye ‘already’; and oc ‘still’.

5.1.3 Some remarks on ditransitive verbs and causativization

Two successive object indexes in the same verb form are not allowed, but a prefix im-
marking 3pl objects may encode the plurality of two distinct participants. This justifies
the recognition of ditransitive verbs—such as maca ‘give (sth.) to (sb.)’ or ilhuia ‘tell (sth.)
to (sb.)’—which have a primary object (indexed) and a secondary object (not indexed).
The primary object represents an animate Goal. The secondary object (the Theme)
shares with the primary object the lack of flagging, plurality marking on the verb, and
accessibility to antipassivization. Unlike the primary object, the secondary object is not
accessible to passivization.
Causatives of intransitive verbs behave like regular underived monotransitive
verbs, with the causee in the role of object. Causatives of transitives behave like regular
ditransitive verbs, with the causee in the role of primary object and the initial object in
the role of secondary object.

5.2 The morphology of Nahuatl applicatives

The applicative suffixes are -lia, -ia, -huia and -lhuia; the former is the most common
one. They do not differ semantically; their distribution is partly determined by the pho-
nological nature of the ending of the stem to which they attach, but there is some free
variation between them. These markers may not only trigger modifications of the stem
but also undergo morphophonological modifications triggered by the formative that
follows them.
Exceptionally, applicative constructions can be marked by a suffix identical to the
causative suffix -tia, as in cuīca-tia ‘sing for (sb.)’ < cuīca ‘sing’ and nāmac-tia ‘sell (sth.)
to (sb.)’ < nāmaca ‘sell (sth.)’.

21 As with object plurality marking, the marking of subject plurality by means of this suffix is not gov-
erned by the plurality of referents, but by plurality marking on the corresponding subject NP.
34 Fernando Zúñiga and Denis Creissels

Applicative markers seem to freely combine with markers expressing other valency
operations, provided the combination is semantically felicitous. Nevertheless, Launey
provides no example of the combination [caus + appl + pass].22
Example (52) illustrates the possibility of converting the applied object into the
subject of a passive construction. Note that the initial object remains unindexed in the
[appl + pass] construction.

(52) Classical Nahuatl (Uto-Aztecan; Launey 2011: 208)

Ō-ni-cuī-lī-lō-c in no-mīl.
perf-1sg.sbj-take-appl-pass-pst def 1sg-field
‘I’ve had my fields, my land taken from me.’ /
‘My fields and my lands have been taken from me.’

In (53), the applicative marker combines with the two antipassive markers (human and
non-human): cui (transitive) ‘take’ > cuī-lia (ditransitive) ‘take for’ > tē-tla-cuī-lia (intran-
sitive).

(53) Classical Nahuatl (Launey 2011: 203)

Ni-tē-tla-cuī-lia.
1sg.sbj-antip1-antip2-take-appl
‘I take (unspecified) things from (unspecified) people.’

Example (54) illustrates the possibility of impersonalizing the intransitive construction

of (53) by the addition of the passive marker.

(54) Classical Nahuatl (Launey 2011: 208)

Tē-tla-cuī-lī-lo.
antip1-antip2-take-appl-pass
‘(Unspecified) things are taken from (unspecified) people.’

The examples in (55) illustrate the combination of applicativization and reflexivization.

(55) Classical Nahuatl

a. Ni-c-ne-tlātī-lia in yāōtl.
1sg.sbj-3.obj-mid-hide-appl def enemy
‘I hide (myself) from the enemy.’ (Launey 2011: 207)

22 Apart from the case where homophonous markers co-occur on a predicate with different functions
(see § 1.2.2), the previous sections have not addressed the issue of multiple marking, either involving
several applicative markers or an applicative and other voice markers / valency operators. Nevertheless,
we asked our authors via the questionnaire to explore this issue when examining the morphology of
applicatives in their languages.
 1 Applicative constructions: An introductory overview 35

b. Cāmpa ni-c-no-cuī-lī-z in no-tlàtlacōl?

where 1sg.sbj-3.obj-mid-take-appl-fut def 1sg-food
‘Where shall I get food for myself?’ (Launey 1979: 196)

Example (56) illustrates the applicativization of a causative construction.

(56) Classical Nahuatl (Launey 2011: 209)

Ti-nēch-in-cua-ltī-lia nacatl in no-pil-huān-totōn.
2sg.sbj-1sg.obj-pl-eat-caus-appl meat def 1sg-child-cstr.pl-dim.pl
‘You are making my children eat meat.’

5.3 The syntax of Nahuatl applicatives

Classical Nahuatl applicatives are P-applicatives. In general (but not always, see Exam-
ples [63]–[64] below), they can be characterized as obligatory applicatives that increase
the valency of the verb. Applicativization is possible for both intransitive and transitive
verbs, but it is more productive with the latter. The possibility of applicativization of
ditransitive verbs is not mentioned in the sources.
Applicativization converts intransitive BCs into monotransitive ACs with the
applied phrase in the role of object (57); see also Example (67) in Section 6.1.

(57) Classical Nahuatl (Launey 2011: 204)

a. Ni-tlaxtlāhua.
1sg.sbj-pay
‘I pay.’
b. Ni-mitz-tlaxtlāhu-ia.
1sg.sbj-2sg.obj-pay-appl
‘I pay you.’

Applicativization also turns transitive BCs into ditransitive ACs with the applied phrase
in the role of primary object and the initial object encoded as the secondary object, as
in (58)–(59).

(58) Classical Nahuatl (Launey 2011: 205)

a. Ni-c-cui in tomin.
1sg.sbj-3.obj-take def money
‘I take the money.’
b. Ni-mitz-cuī-lia in tomin.
1sg.sbj-2sg.obj-take-appl def money
‘I take the money from you.’
36 Fernando Zúñiga and Denis Creissels

(59) Classical Nahuatl (Launey 2011: 204)

a. Ni-qu-ixca tōtoltetl.
1sg.sbj-3.obj-fry egg
‘I am frying eggs.’
b. Ni-mitz-ixqu-ia tōtoltetl.
1sg.sbj-2sg.obj-fry-appl egg
‘I am frying eggs for you.’

Applicatives from intransitives behave like underived monotransitive verbs, and

applicatives from transitives behave like underived ditransitive verbs.

5.4 The semantics of Nahuatl applicatives

Applied objects mostly express the semantic roles of Beneficiary/Maleficiary, and more
generally animate Goals. These roles include those implied by the lexical meaning of
the verb, and consequently cannot be deemed “non-essential”, as in Example (57) above
with ‘pay’. Launey (1994) provides additional examples with ‘declare (sth. to sb.)’, ‘send
(sth. to sb.)’ and ‘throw (sth. to sb.)’.
A comitative meaning is possible with some verbs, for example ‘share’ > ‘share with’:

(60) Classical Nahuatl (Launey 2011: 206)

Ō-ni-c-xelô in nacatl. —
perf-1sg.sbj-3.obj-share.pst def meat
Ākin ō-ti-c-xel-huî?
who perf-2sg.sbj-3.obj-share-appl.pst
‘I shared out the meat. — Who did you share it out with?’

Inanimate applied objects expressing ‘up to’ or ‘beyond’ are found with motion verbs:

(61) Classical Nahuatl (Launey 2011: 207)

Ō-ni-c-chol-huî in ātoyatl.
perf-2sg.sbj-3.obj-flee-appl.pst def river
‘I fled to the other side of the river.’

There are also isolated cases of inanimate applied objects that do not seem to lend them-
selves to semantic generalizations:

(62) Classical Nahuatl

a. Ni-c-chōqui-lia in no-tlàtlacōl.
2sg.sbj-3.obj-cry-appl def 1sg-sin
‘I cry for (with reference to) my sins.’ (Launey 2011: 207)
 1 Applicative constructions: An introductory overview 37

b. Xi-c-tlāli-li iztatl in ātl.

imp-3.obj-put-appl.imp salt def water
‘Put salt in the water.’ (Launey 1979: 200, 394)

A particularity of Classical Nahuatl applicativization is that objects modified by an

adnominal possessor show a strong tendency to occur in an applicative construction in
which the applied object is coreferential with the adnominal possessor of the secondary
object. Rather than adding a participant, such ACs emphasize the fact that the referent
of the adnominal possessor can be viewed as concerned by the participation of the ref-
erent of the possessee in the event. The non-applicative monotransitive constructions
(63a) and (64a) are possible, but the same meanings are more commonly expressed as
(63b) and (64b), respectively.23

(63) Classical Nahuatl (Launey 2011: 206)

a. Ni-qu-in-tlazòtla in mo-pil-huān.
1sg.sbj-3.obj-pl-like def 2sg-child-cstr.pl
‘I love your children.’
b. Ni-mitz-in-tlazòti-lia in mo-pil-huān.
1sg.sbj-2sg.obj-pl-like-appl def 2sg-child-cstr.pl
‘I love your children (for you).’

(64) Classical Nahuatl (Launey 2011: 206)

a. Ni-c-mati in mo-tlàtlacōl.
1sg.sbj-3.obj-know def 2sg-sin
‘I know your sins.’
b. Ni-mitz-machi-lia in mo-tlàtlacōl.
1sg.sbj-2sg.obj-know-appl def 2sg-sin
‘I know your sins (regarding you).’

By contrast, if the object is incorporated, the possessor is treated as the object of a

non-applicative monotransitive construction; compare (65a) and (65b). Applicative
marking does occur in the more complex configuration illustrated by (65c):

(65) Classical Nahuatl (Launey 2011: 210)

a. Ni-mitz-pāqui-lia in mo-cuā.
1sg.sbj-2sg.obj-wash-appl def 2sg-head
‘I wash your head for you.’

23 The nominal suffix encoding that the noun is modified by an adnominal possessor is labeled “con-
struct form marker”.
38 Fernando Zúñiga and Denis Creissels

b. Ni-mitz-cuā-pāca.
1sg.sbj-2sg.obj-head-wash
Lit. ‘I head-wash you.’ / ‘I wash you in terms of the head.’
c. Ni-mitz-cuā-pāqui-lia in mo-pil-tzin.
1sg.sbj-2sg.obj-head-wash-appl def 2sg-child-dim
Lit. ‘I head-wash your son (for you).’

5.5 Nahuatl applicative lookalikes

The language has lexicalized applicatives. For instance, pōhua ‘count, read’ is compatible
with two applicative suffixes, -lia and -ia, but pōhui-lia has the regular meaning ‘count/
read (sth.) to (sb.)’, whereas pōhu-ia unpredictably expresses ‘hex, cast a spell on’.24
Interestingly enough, verb forms combining middle-voice marking and causative
or applicative marking can be used with a conventionalized meaning distinct from
their literal meaning ‘make oneself V’ or ‘V for oneself’. The combination [mid + caus]
can be used as the honorific form of intransitive verbs (66a), whereas the combination
[mid + appl] can be used as the honorific form of transitive verbs (66b), without any
change in valency:

(66) Classical Nahuatl (Launey 2011: 213, 215)

a. Ti-mo-cochī-tia.
2sg.sbj-mid-sleep-hon
‘You are sleeping.’ (honorific) (Lit. ‘You are making yourself sleep.’)
b. Qui-mo-chīhui-lia.
3.obj-mid-do-hon
‘He makes it.’ (honorific) (Lit. ‘He makes it for himself.’)

6 The notion of applicativization in context

6.1 Historical background

To the best of our knowledge, the term verbo aplicativo ‘applicative verb’ was first used
in Rincón (1595), one of the earliest descriptions of Classical Nahuatl, and was later
adopted by Carochi (1645) for his own description and by Uto-Aztecan studies ever
since. As already presented in some detail in Section 5, Nahuatl intransitive and transi-

24 Launey’s French original gives “je lui jette un sort” ‘I cast a spell on him’ (1979: 194), but Mackay’s
English translation gives “I read his fortune (through divination with kernels of corn)” (2011: 204).
 1 Applicative constructions: An introductory overview 39

tive predicates can take the applicative suffix -lia or one of its allomorphs and thereby
accommodate an (additional) object in the clause; (67) illustrates this with intransitive
tzàtzi ‘shout’:

(67) Classical Nahuatl (Launey 2011: 42, 203, glosses added)

a. Ni-tzàtzi.
1sg.sbj-shout
‘I shout.’
b. Ni-mitz-tzàtzi-lia.
1sg.sbj-2sg.obj-shout-appl
‘I shout after you.’ / ‘I call you by shouting.’

A comparable phenomenon is found in Bantu languages, and the term is regularly

used in that descriptive tradition as well. In Kinyarwanda, for instance, many verbs can
take the applicative suffix -iish and thereby accommodate another object in the clause
they head; (68) illustrates this with transitive -andik- ‘write’:

(68) Kinyarwanda (Kimenyi 1980: 32)

a. Umukoôbwa a-ra-andik-a íbárúwa n’-ííkárámu.
girl 3sg.sbj-prs-write-fv letter with-pen
b. Umukoôbwa a-ra-andik-iish-a íbárúwa íkárámu.
girl 3sg.sbj-prs-write-appl-fv letter pen
Both: ‘The girl is writing a letter with a pen.’

Examples (67) and (68) look quite similar, but they differ in a number of important
ways—which partly explains the fact that different descriptive traditions have gravi-
tated towards different ideas about what “canonical” applicatives look like (see § 6.2 for
more on this issue). Regarding syntax, the base verbs applicativized by the markers -lia
and -iish in the two languages belong to different valency classes, at least in the examples
chosen here: the new object is alone in the Nahuatl example but co-occurs with another
object in the Kinyarwanda clause. Regarding semantics, the new object can have various
semantic roles in Nahuatl, namely Beneficiary, Maleficiary, Goal, etc.; it can only be an
Instrument with Kinyarwanda iish-applicatives. Regarding morphology, the applicative
markers occupy quite different niches in their respective semiotic ecologies. In Nahuatl,
a homophonous suffix -lia also occurs on honorific verbs, but the causative -tia has a
different shape. Kinyarwanda does not have honorific verbs, but its causative is also
marked by -iish, and applicative constructions with an applied phrase expressing roles
other than Instrument require a different marker. Lastly, the broad applicative illus-
trated in (67) is typically the only way to accommodate that particular non-agentive par-
ticipant in the clause in Nahuatl, whereas the instrumental applicative normally allows
Kinyarwanda speakers to choose from the two roughly synonymous expressions given
in (68).
40 Fernando Zúñiga and Denis Creissels

Unlike Uto-Aztecan studies, early descriptions of Bantu languages show significant

terminological disparity regarding the treatment of applicatives. In his descriptive
sketch of Kongo, Brugiotti (1659: 40) used the term verbum respectivum ‘respective verb’
(opposed to the unmarked verbum absolutum ‘absolute verb’); Bleek (1873: 8) used “rel-
ative form of the verb” for southern languages, and Endemann (1876: 64) used the term
direktiv ‘directive’ for Sotho. Bantuist terminology seems to have unified in the late
19th century: Bentley (1887: 627) and Steere (1884: 158) used the term “applied form”
in their studies of Kongo and Swahili, respectively, and “applicative” is found—occa-
sionally alongside other terms—in Torrend (1891), Stapleton (1903), Meinhof and Van
Warmelo (1932), Watkins (1937), Guthrie (1967–1971), Meeussen (1967), as well as in
later studies. (See Dammann 1961: 160–161 for a review of the labeling of applicatives
in Bantu studies in particular and Pacchiarotti 2020: 28–31 for a more detailed termino-
logical survey in general.)
Authors working in other descriptive traditions have sometimes used alternative
terms, e.g. “rule of Dative” (e.g., Chung 1976 for Indonesian) and “object advancement”
(e.g., Norman 1978 for Eastern Mayan, as well as Aissen 1983, 1987 for Tzotzil). Modern
Mayanists not working in the Relational-Grammar framework have customarily used
the term applicative (e.g., Grinevald and Peake 2012: 38), and some scholars working on
Philippine languages have employed it as well (e.g., Bell 1983 for Cebuano). Other terms
have not spread beyond descriptive, often language-specific, studies. For instance, work
on Kartvelian uses “version”, a coinage derived from Georgian kceva ‘turning, change,
behavior’; Salas (2006: 119) labels the Mapudungun applicative markers sufijos indirec-
tizantes “indirecting suffixes”; Ainu descriptions have used the terms “appropriative”
and “demonstrative” (Bugaeva 2010: 752); and several dictionaries of linguistics from
the 1950s and 1960s use the term “accommodative”, possibly under the influence of the
traditional term dativus commodi (Pacchiarotti 2020: 30).
Despite this variation, the applicative has led until recently a comparatively quiet
life as a technical label, even by the somewhat inauspicious standards of late-20th-cen-
tury theoretical linguistics, where crucially related notions like transitivity, valency, and
voice have aroused some analytical controversy and spawned a non-negligible variety
of interpretations and sub-classifications. Unlike notorious terms like subject, ergative,
and even passive, which have not only attracted considerable interest but also proven
to be lasting sources of disagreement, applicative and the phenomena it covers seem to
have drawn just the right amount of attention for the label not to become contentious.

6.2 Present-day usage

Even though applicatives are less problematic than several other linguistic terms,
modern studies do show some variation regarding both the place the applicative occu-
pies in a theory of grammar and its exact definition. This is hardly surprising: even
though classical languages have similar constructions—consider Latin pairs like ridere
 1 Applicative constructions: An introductory overview 41

‘laugh (at)’, which can be intransitive or transitive, vs. deridere ‘laugh at’, which is
transitive—the traditional language descriptions of antiquity provided neither the ter-
minology nor the analytical apparatus to study such occurrences as the conspicuous
grammatical phenomenon they are in many languages. Moreover, as prominent and
pervasive as applicatives are in Uto-Aztecan, the descriptions of those languages have
never been at center stage in Western linguistics; it is the description of familiar Euro-
pean languages and the theorizing they have spawned that have informed the accounts
of non-European linguistic structures.

6.2.1 Applicatives and their counterparts

With respect to the company applicatives keep in linguistic theory, the general consen-
sus among present-day scholars is that such constructions are best regarded as a kind of
voice.25 Functional-typological studies of transitivity, valency, and voice customarily dis-
tinguish between operations that increase the valency of the predicate and those that
decrease it; applicatives are routinely placed in the first group. Another customary dis-
tinction made is whether the operation preserves the base inventory of semantic argu-
ments or alters it. Interestingly enough, applicatives are placed either in the former or
in the latter group, depending on how they are interpreted. Kroeger (2005), for instance,
sees applicatives as “increase[ing] the syntactic valence of a verb by introducing a new
primary object, [. . .] [t]ypically [. . .] ‘promot[ing]’ an oblique argument [. . .], and so [the
applicative] does not affect the argument structure of the verb” (p. 273, our emphasis).
By contrast, Kulikov (2011) sees applicatives as “introduc[ing] a Direct Object (lacking
in the initial structure), [. . .] [which] may denote an entirely new participant in the
situation, or it can be promoted from the periphery of the syntactic structure, where it
surfaced as an Oblique Object in the non-derived diathesis” (p. 389), and as an example
of the operations that “do not preserve the inventory of semantic roles” (p. 385, our
emphasis).
Applicatives in which the applied phrase is a core argument (i.e., P-applicatives)
constitute a special case of a morphosyntactic operation some recent studies call
nucleativization. Nucleatives allow a participant not encoded as a core argument in
the initial construction to be encoded as a core argument in the derived construction.
(Their mirror images, denucleatives, need not concern us here.) We employ Creissels’s

25 Mel’čuk (2006) is a notable recent example in the functional-typological tradition that proposes a
rather narrow notion of voice. Not only does that study distinguish between diathesis and voice based
on the predicate-marking parameter (as others do, e.g., Kulikov 2011 and Zúñiga and Kittilä 2019), but it
also distinguishes categories like “active”, “passive”, and “reflexive” (i.e., Mel’čuk’s “voices”) from those
like “causative”, “decausative”, and “applicative” (i.e., his “[de-]transitivizers”, which “are similar to
voices in that they also entail a modification of the basis diathesis” but “express some [essentially addi-
tional] propositional meaning”, p. 194).
42 Fernando Zúñiga and Denis Creissels

(forthcoming) use of the term, according to which nucleatives distinguish an S/A and a
P version, and P-applicatives are a special case of the latter. Note from Table 8 that, by
definition, applicatives targeting lower-prominence (i.e., non-core) grammatical rela-
tions are not nucleatives.26

Table 8: Applicatives and syntactic types of nucleatives.

Status of new argument

S/A P D X
S/A-nucleatives P-nucleatives = D-applicatives X-applicatives
P-applicatives

Incidentally note that portative constructions mentioned in Section 4.1 are problematic
neither for the definition of applicative employed in the present book nor for how we
propose to capture different types of applicatives. In our view, the Caddo portative as
described by Melnar (1998) is simply a P-applicative.
Zooming in on different types of nucleatives, and also following Creissels (forth-
coming), we distinguish between agentive and non-agentive S/A-nucleatives here; we
can additionally distinguish the same semantic types of P-nucleatives (see Table 9).27
Some examples follow.

Table 9: Semantic and syntactic types of nucleatives.

S/A-nucleatives P-nucleatives
Agent non-Agent Agent non-Agent
causative non-causative non-applicative applicative
S/A-nucleatives S/A-nucleatives P-nucleatives P-nucleatives
(25c) and (26f) (69) and (70a) (70b)

This chapter has already provided numerous examples of applicative (P-)nucle-

atives, which we neither repeat nor cross-reference here; for examples of causative
S/A-nucleatives, see Examples (25c) and (26f). Example (69) below from Ilocano shows a

26 Zúñiga and Kittilä (2019) use the term nucleativization for the installment of arguments in semantic
structure irrespective of their realization, whereas Creissels (forthcoming) frames the concept—as we
do here—in terms of the (semantic-)syntactic roles S, A, P, etc. In addition, Zúñiga and Kittilä’s (2019)
employ labels for language-specific grammatical relations and consequently distinguish between sub-
jective and objective nucleatives.
27 Zúñiga and Kittilä (2019) distinguish between subjective and objective nucleatives. Note that these
authors’ taxonomy regards causatives as prototypically subjective and applicatives as prototypically
objective.
 1 Applicative constructions: An introductory overview 43

non-causative S/A-nucleative; the clause corresponds to a construction combining the

effects of applicativization and passivization. Beneficiaries can simply appear in an
oblique role in Ilocano clauses, but, thanks to the circumfix i-. . .-an (i.e., the so-called
benefactive voice marker), the predicate ‘buy’ can also accommodate such participants
in S/A role:

(69) Ilocano (Philippine Austronesian; Rubino 2005: 337)

I-gatáng-an=n=ak man iti bagás.
bv-buy-bv=2/3sg=1sg.abs please obl rice
‘Please buy some rice for me.’

Example (70b) illustrates a non-applicative P-nucleative; the clause corresponds to a

construction combining the effects of causativization and the reversal of grammatical
relations without detransitivization. Mapudungun l-causatives often behave in a famil-
iar way: they install a causer-Agent in the clause, the erstwhile argument in S/A role
appears in P role, and the verb is in the (unmarked) Agent Voice (70a). Nevertheless,
when an obviative/non-topical 3rd person acts on a proximate/topical 3rd person, as in
(70b), the syntax of the language requires that the former be in P role and the latter in
S/A role; the verb is in the inverse-marked Patient Voice instead:

(70) Mapudungun (p.k. and Golluscio 2007: 211; orthography and glosses adapted)
a. Ti wentru aye-l-fi-i ta malen.
det man laugh-caus-3.P-ind[3] det girl
‘The man (prox) made the girl (obv) laugh.’
b. Tañi chaw aye-l-e-i-mew ta malen.
3.psr father laugh-caus-inv-ind-3.A[3] det girl
‘The girl’s father (obv) made her (prox) laugh.’
(Lit. ‘Her father made the girl laugh.’)

6.2.2 Applicatives and their types

With respect to how broadly applicatives are defined in the recent literature, there is
more variation than regarding their place in theories of grammar. Our term is broader
than Kulikov’s (2011), for instance, whose applicative is limited to our P-applicatives;
this author suggests using the label “benefactive” for D-applicatives and does not
address X-applicatives. Further note that our terminology is not explicitly framed in
terms of prototype vs. periphery—unlike Zúñiga and Kittilä’s (2019), which distin-
guishes prototypical applicatives, which target direct/primary objects, from non-pro-
totypical applicatives, which target grammatical relations of lower prominence, e.g.,
indirect/secondary objects and obliques. (We will return to the prototype vs. periphery
issue further down.)
44 Fernando Zúñiga and Denis Creissels

Our applicatives have the syntactic status of the applied phrase in the applicative
construction as an explicit parameter of variation, and are therefore subclassified into
P-, D-, and X-applicatives; the status of the companion non-agentive argument in the
applicative clause (if any) is an additional parameter of variation, captured by the dis-
tinction between symmetrical and asymmetrical. By contrast, Beck’s (2009: 539–540)
taxonomy conflates these two parameters. He first distinguishes direct applicatives,
which “realize the applied object as a direct object”, from non-direct applicatives, which
“add an additional indirect or oblique object”. Then he says that a third type might be
worth distinguishing, but instead of subclassifying his non-direct applicatives (as we
do), he splits up his first type: while one subtype of direct applicatives of transitive
stems “add a direct applied object, displacing the original direct object”, this additional
type “creates ditransitives with equally ranked (Upper Necaxa Totonac [. . .]) or symmet-
rical objects (Kichaga [. . .])” (p. 540).
Table 10 below compares these three different terminologies with the one we use it
in this book (which is the same as in Creissels, forthcoming).

Table 10: Some syntactic types of applicatives.

P D X
Kulikov (2011) applicatives benefactives —
Zúñiga and Kittilä (2019) prototypical non-prototypical
applicatives applicatives
Beck (2009) direct direct non-direct applicatives
applicatives 1 applicatives 2
This book P-applicatives D-applicatives X-applicatives
asymmetrical symmetrical

At their most exclusive, definitions of applicative constructions use the term

only for a clause type that grants direct object status to a non-agentive adjunct of a
related, base, construction. Until relatively recently and on both sides of the function-
alist-formalist divide—at least before Baker’s proposed extension, see further down—
overt morphological marking on the predicate was normally seen as a requisite as well.
Such an arguably rather narrow understanding of applicatives owes an important debt
to studies conducted in formalist traditions (viz. Relational Grammar in the 1980s, Gov-
ernment and Binding in the 1980s and 1990s, Lexical Functional Grammar in the 1990s,
and Minimalism in the 2000s and 2010s). As a consequence, even less restrictive notions
found in the current mainstream tend to see applicativization as essentially syntactic
(more specifically: valency-increasing and promotional).
Functionally oriented studies of voice and related phenomena have not fundamen-
tally departed from this tenet (see, e.g., Peterson 2007; Kulikov 2010; Givón 2001). A
case in point is Shibatani’s (1996) proposed terminology, which is superficially similar
 1 Applicative constructions: An introductory overview 45

to Kulikov’s (2011). Shibatani explicitly excludes X-applicatives from his definitions but
argues in favor of distinguishing his “applicatives” (= locative/instrumental P/D-applic-
atives) from his “benefactives” (= benefactive P/D-applicatives) rather strictly, based
both on their respective semantics and on the following cross-linguistic observation:
the former type “generally allow[s] intransitive bases, while [the latter] seldom admit[s]
intransitive bases” (p. 160). Notice that we are not aware of many authors that follow
this proposal; not even this author himself distinguished these notions as sharply in
later studies (e.g., Shibatani 2006).
At their most inclusive, characterizations of applicative constructions delineate a
family of clause types employed to give a selected non-agency-related element of the clause
some kind of syntactic, semantic, and/or pragmatic prominence vis-à-vis the (lower) status
that constituent has in the base construction. A subset of such characterizations concen-
trates on syntactic prominence and regards some syntactic applicative lookalikes as applic-
atives—but it is worth noting that such approaches have not usually made it into the func-
tional-typological mainstream. For example, Shibatani (2006) restricts the focus of attention
to constructions installing or promoting to P or D status (thereby excluding X-applicatives),
but it abstracts from morphological parameters, thus conflating diathesis and voice: his
“benefactive/malefactive/applicative” category subsumes Germanic and Romance exter-
nal-possession constructions with unmarked verbs, the Japanese benefactive converbal
applicative periphrasis with yaru ‘give (to non-human or inferior)’, and the Guarijío ben-
efactive affixal applicative with -ke. Similarly, Croft (2022) distinguishes between “overtly
verb-coded voice strategies” and “zero verb-coded voice strategies”; he consequently labels
oppositions like the English dative and locative alternations “zero coded simple predicate
[i.e., non-periphrastic, F.Z. & D.C.] strategy for applicative constructions” (pp. 280–287).
Studies in the Chomskyan tradition take a similar tack (although they usually
exclude both D- and X-applicatives). Baker’s (1988: 280–288) influential study of incor-
poration first distinguishes alternations involving Recipient-Beneficiaries and verbs
marked with some (non-null) reflex of ✶-ɪd in Bantu from the English so-called dative
shift by calling only the former applicatives. Then, he argues that the syntax of those two
kinds of alternations is comparable, later claiming that the valency-increasing mecha-
nism is the same in both kinds, namely preposition incorporation. Crucially, Baker ends
up analyzing English verbs occurring in double-object constructions as follows: “for
the relevant small and semi-idiosyncratic set of verbs, the applied affix is syntactically
present but is simply not seen morphologically” (p. 284). The same reasoning and con-
clusions are found in Marantz (1993: 114–115), and are later applied to wider samples of
languages and phenomena, for example in Jeong (2007: 3) and Pylkkänen (2008: 11–12).
It is not uncommon to find descriptive studies that work with notions that are not
overly restrictive, presenting applicatives as coming in “prototypical” and “peripheral”
versions—occasionally framed in terms of “canonical” and “non-canonical” versions.
First, applicative constructions are seen as clauses typically headed by predicates that
are overtly derived from those heading base constructions; “unmarked applicatives”
are regarded as non-canonical. Second, the syntactic status of the applied phrase in the
46 Fernando Zúñiga and Denis Creissels

AC is seen as prototypically being the P role, that is, that of a direct or primary object;
applicatives featuring less prominent syntactic roles are regarded as non-canonical.
Third, the canonical applicative construction is an optional variant, equivalent in
terms of its truth value to the base construction; ACs that do not have a monoclausal
BC counterpart (i.e., “non-promotional applicatives”) are seen as non-canonical. Lastly,
the semantic role borne by the applied phrase is required to be not only non-agentive
but also peripheral; applicatives targeting central non-agentive roles like Theme and
Patient, but also arguably intermediate roles like Stimulus and some Location-related
notions are considered non-canonical.28
Our definition treats “unmarked applicatives” as an oxymoron—more accurately:
applicativized predicates are morphologically marked by definition; constructions
that are analogous but lack morphological marking are called “syntactic lookalikes”
here. Regarding the other three domains, however, our definition simply includes the
non-canonical cases: the exact grammatical relation borne by the applied phrase, the
optionality or obligatoriness of the AC to express a given state of affairs, and the exact
(kind of) semantic role and are left open; rather than definitional stipulations, they are
parameters of variation. Therefore, our terminology is simultaneously narrower and
broader than many, if not most, canon-based terminologies. It is narrower because we
do not treat the boundary between applicatives and their syntactic and morphological
lookalikes as fuzzy: it may be that different languages show different frequencies of
tokens or types of particular applicatives and lookalikes, but such quantitative assess-
ment does not inform the notional distinction between the phenomena. Our terminol-
ogy is also broader, because we explicitly take different syntactic, morphological, and
semantic subtypes into account, but especially because we abstain from stipulating that
one particular bundle of features is the prototype and others are deviations thereof.
Irrespective of the terminological treatment given to “non-canonical applicatives”,
there are good reasons to have a closer look at such phenomena, and the reader is
referred to Pacchiarotti and Zúñiga (2022) for recent contributions to our knowledge
of some of them. The phenomena specifically showcased in that book include co-ex-
pression patterns / instances of polysemy that are not restricted to other voice cate-
gories (e.g., denominal and deadjectival verbalization, nominalization, and relativiza-
tion), functions of applicative-like morphology that are not semantically neutral (e.g.,
covering aspectuality, pluractionality, and intensity), and functions that are related to
information-structure categories (e.g., focalization and topic continuity). It was not the
editors’ intention to use the results of such explorations to justify broadening or nar-
rowing some received definition, however; rather, they thought it was important to use

28 See in this context Lehmann and Verhoeven (2006) for a terminological proposal that has not been
widely accepted. These authors argue in favor of recognizing a whole range of transitivization processes
targeting non-agentive participants in Yucatec Maya. This range can be seen as having two poles: one for
lexical derivation that targets central semantic roles with syntactic consequences (“extraversion”) and
another for syntactically regular promotion that targets peripheral semantic roles (“applicativization”).
 1 Applicative constructions: An introductory overview 47

them to deepen our understanding of applicatives and lookalikes in particular, as well

as of phenomena related to transitivity, valency, and voice in general.
Since the turn of the century, mainstream functional-typological linguistics has
tended to favor grammatical-category labels that are not excessively narrow on the
one hand and, at least partly, to move away from prototype-vs.-periphery character-
izations on the other (especially those that do not address the radial structure of the
notions explicitly, as well as those with fuzzy boundaries).29 It is in this spirit that we
propose the definition of applicative constructions used for the present book. Taking
into account predicate morphology and syntactic status of the applied phrase here only,
our applicative covers the uncontroversial case: the structure represented in cell (h)
vis-à-vis cell (a) in Table 11 (adapted from Table 3 in § 1.2.3). Following several other
studies, we have broadened the applicative to include the structure depicted in cell (g).
Following Creissels (forthcoming), we have added the structure of cell (f) to the cate-
gory, and we have made explicit that the oblique X in cell (a) need not be expressible.

Table 11: Selected applicatives and applicative lookalikes.

Status of Predicate morphology

non-S/A argument
Unmarked Marked
Oblique (a) S/A V (X) (e) S/A V′ (X)
(b) S/A V X′ (f) S/A V′ X′
Non-oblique Non-core (c) S/A V D (g) S/A V′ D
Core (d) S/A V P (h) S/A V′ P

We have stopped short of including in our definition the two most prominent
related structures of the table, that is, those represented in cells (d) and (e). The former
is a syntactic lookalike; including it would equate applicativization with a syntactic
operation, namely X-to-P promotion, and blur the distinction between diathesis and
voice, thereby additionally contradicting the spirit of the term as originally employed in
the description of Classical Nahuatl (there, it refers to a derived verb form). The struc-
ture represented in cell (e) is a morphological lookalike; including it would equate the
applicative with the overt predicate marker, which would in turn presumably correlate
with some broad notion of “functional” (i.e., semantic and/or pragmatic) prominence of
the argument in non-S/A role. Our notion is, we think, reasonably conservative (since we
do not abstract from either the morphology or the syntax) while also reasonably non-

29 An example of an explicit (if uninfluential) canon-based approach is Dixon and Aikhenvald (1997,
2000) and especially Dixon (2012: Ch. 25), with its rather restrictive definition of the applicative pro-
totype and several “extensions” (e.g., he regards our obligatory applicative as peripheral and labels it
“quasi-applicative”).
48 Fernando Zúñiga and Denis Creissels

restrictive (since we do not require the applied phrase to be in P role). Our goal has not
been to define the “correct” applicative but, rather, to make the analysis of the variation
found with applicatives and applicative-like constructions across and within languages
as consistent and principled as possible.

7 Structure, coverage, and coherence

of the present book
The first part of this book contains this introductory overview; a questionnaire designed
to be used as a brief checklist for exploring applicative constructions in individual lan-
guages or language groups (Ch. 2); and a very short chapter that includes maps depict-
ing the approximate geographical locations of the languages addressed in the book,
as well as those languages’ genealogical affiliations (Ch. 3). The second part includes
twenty-five case studies, grouped in three sections according to whether they deal with
individual languages or with areal or genealogical overviews. The third part includes a
chapter that comments on several terminological and theoretical issues raised by this
introductory chapter (Polinsky); a chapter that provides an update of Polinsky’s (2013)
typological survey of applicatives (Moroz and Polinsky); and a chapter summarizing
what we believe to be the main results of the case studies (Creissels and Zúñiga).
The chapters in the second part of the book deal with a sample of individual languages
or language groups that we thought were necessary, important, or especially interesting
in a comparative survey of applicative constructions. Chapters covering phenomena
found in the Americas include four that address three groups identified by Moroz and
Polinsky. Gerdts—who has already co-authored a comprehensive survey of applicatives
in Salishan in general (Kiyosawa and Gerdts 2010)—describes in detail the applicatives
found in the variety of Halkomelem spoken in Vancouver Island. Montgomery-Anderson
describes the applicatives of the Mayan family, with a special focus on the Chontal lan-
guage of southern Mexico. Reflecting both the historical importance and the comparative
significance of Uto-Aztecan applicatives, two chapters cover considerable part of the phe-
nomena found in those languages: Thornes covers languages from the Northern branch,
and Álvarez González and Estrada Fernández cover languages spoken in Northwest-
ern Mexico. Several chapters cover families and languages beyond those mentioned by
Moroz and Polinsky’s survey: Mithun gives an overview of the applicatives found in Eski-
mo-Aleutian, Lockwood and Macaulay survey those found in Algonquian, and Hernán-
dez-Green and López Nicolás present applicatives and lookalikes in Otomi and Zapotec.
Beck outlines applicativization in a Totonacan language spoken in south-central Mexico;
Censabella covers Toba/Qom, a Guaicuruan language spoken in northern Argentina; and
Zúñiga covers Mapudungun, an isolate of southern Chile and south-central Argentina.
Regarding Africa, Moroz and Polinsky say that applicatives are found “mostly in
Bantu”, and two chapters address them here: Pacchiarotti gives an overview for the
 1 Applicative constructions: An introductory overview 49

family, and Creissels covers in detail the applicatives of Tswana, a language spoken
in Botswana and South Africa. Also belonging to the Niger-Congo group, Atlantic lan-
guages have been comparatively neglected in the literature dealing with voice(-related)
phenomena; Voisin and Creissels’s chapter contributes to rectifying this neglect. Payne
surveys the applicatives of Nilotic (Nilo-Saharan), Vanhove those of Cushitic (Afro-Asi-
atic), and Amberber zooms in on the applicatives of Amharic (Ethio-Semitic).
Moroz and Polinsky also mention Austronesian in the western Pacific region. Two
chapters deal with those languages, namely McDonnell and Truong, who cover applic-
atives in languages of western Indonesia, and Musgrave, Wayan Arka, and Rajeg, who
cover them in Standard Indonesian. Two further chapters complete the picture for that
part of the world: Foley and Austin survey applicatives in Papuan and Australian lan-
guages, respectively.
Lastly, Polinsky (2013) briefly commented on “the dearth of applicatives in Eurasia”
and mentioned some Northwest Caucasian languages as an exception to the rule. Ark-
adiev, Lander, and Bagirokova cover precisely that family; Tuite surveys the applic-
atives of Kartvelian in the southern Caucasus. The two final chapters address groups
not singled out by Polinsky’s 2013 piece: Jacques and Lahaussois cover Kiranti (Sino-Ti-
betan), and Zúñiga, Arkadiev, and Hegedűs cover applicativizing preverbs in a number
of European languages (Germanic, Balto-Slavic, and Hungarian).
The titles in the Comparative Handbooks of Linguistics series face the challenge of
striking a balance between comparability across chapters and state-of-the-art analytic
coverage. In order to ensure comparability, we asked our authors to work with both a
significantly shorter version of this introductory chapter (in the form of an 8,500-word
position paper) and the two-page questionnaire mentioned above, as well as to follow a
thematic blueprint “morphology-syntax-semantics-lookalikes” rather closely in the con-
clusions section of their own chapters. The purpose of the position paper and the ques-
tionnaire was to provide the individual contributors with a suitable descriptive meta-
language—which is, ideally, both theoretically neutral and descriptively effective—and
a frame of reference. In order to ensure adequate analytic coverage, we explicitly gave
our authors relatively free rein not only as to how they may want to structure their
chapters—the thematic blueprint worked better in some instances than in others—but
also as to which phenomena they may want to include in their discussion in addition to
those explicitly mentioned in the questionnaire.
Consequently, chapters dealing with individual languages or groups first give an
introduction to the language(s) covered and their basic morphosyntactic structures,
and then survey morphological, syntactic, and semantic features of the applicatives and
lookalikes found in them, but they do not necessarily follow this structural template.
For example, Zúñiga’s chapter on Mapudungun follows it rather closely, but Creissels’s
chapter on Tswana does not, favoring instead a structure largely based on syntactic
aspects of the phenomena it surveys. (We sent early versions of these two chapters to
our authors alongside the position paper and the questionnaire.)
50 Fernando Zúñiga and Denis Creissels

Even though our contributors had considerable leeway when it came to structuring
their chapters, many did follow the thematic template, thereby giving a distinct formal
coherence to the book. We probably erred on the side of restrictiveness by imposing
some terminological coherence as well—hopefully in a reasonable way. For instance,
some Mesoamerican studies have employed the term registration applicative to refer
to some constructions we prefer to treat as lookalikes, and we asked Néstor Hernán-
dez-Green and Oscar López Nicolás to exert special care when using potentially con-
flicting terms and to make the correspondences between the labels used by Mesoameri-
canists and ours explicit. By the same token, Kartvelian studies traditionally employ the
term version to refer to the relevant part of those languages’ applicative morphology,
and we asked Kevin Tuite to make sure that both Kartvelologists and typologists could
follow his presentation and analysis comfortably. We proceeded in the same manner
regarding the traditional Algonquianist terms relative preverbs and relative roots in
Hunter Lockwood and Monica Macaulay’s contribution.

Abbreviations
A Agent-like argument of bivalent/trivalent predicate
AC applicative construction
acc accusative
ai animate intransitive
aio animate intransitive with an object
ALC applicative lookalike construction
anim animate
antip antipassive
aor aorist
appl applicative
AppP applied phrase
art article
asp aspect
av Agent voice
BC base construction
caus causative
compl completive
ctr control transitivizer
cstr construct form marker
cvb converb
D Recipient-like argument of trivalent predicate
dat dative
decl declarative
def definite
dem demonstrative
det determiner
detr detransitive
 1 Applicative constructions: An introductory overview 51

dim diminutive
dir direct
direc directional
ditr ditransitive
enunc enunciative
ep epenthesis
erg ergative
f feminine
foc focus
fut future
fv final vowel
gen genitive
hon honorific
imm immediate
imp imperative
IN incorporated nominal
inan inanimate
incompl incompletive
ind indicative
indir indirect
ins instrumental
intr intransitive
ipfv imperfective
lk linker
loc locative
mid middle
monotr monotransitive
nfin nonfinite
nom nominative
nspec nonspecific
ntr noncontrol transitivizer
obj object
obl oblique
oblreg oblique registration
obv obviative
P Patient-like argument of bivalent predicate
pass passive
perf perfect
pfv perfective
pl plural
postp postposition
pri primary
prior prioritive
prog progressive
prox proximate
psr possessor
pst past
ptcl particle
pvb preverb
52 Fernando Zúñiga and Denis Creissels

quot quotative
rec recent
S single argument of monovalent predicate
sbj subject
sec secondary
sg singular
SR semantic role
ta transitive animate
tam tense-aspect-mood
ti transitive inanimate
tr transitive
tritr tritransitive
Vs verbs
1, 2, 3 grammatical persons
I, II, . . . noun classes
- default morpheme boundary
= clitic morpheme boundary
# phonological word boundary

References
Aissen, Judith. 1983. Indirect object advancement in Tzotzil. In David Perlmutter (ed.), Studies in relational
grammar, Vol. 1, 272–302. Chicago: University of Chicago Press.
Aissen, Judith. 1987. Clause structure in Tzotzil. Dordrecht: Reidel.
Andrews, J. Richard. 2003. Introduction to Classical Nahuatl. Rev. edn. Norman: University of Oklahoma Press.
Bahrt, Nicklas N. 2021. Voice syncretism. Berlin: Language Science Press.
Baker, Mark C. 1998. Incorporation: A theory of grammatical function changing. Chicago: The University of
Chicago Press.
Beck, David. 2009. A taxonomy and typology of Lushootseed valency-increasing suffixes. International
Journal of American Linguistics 75(4). 533–569.
Bell, Sarah. 1983. Advancements and ascensions in Cebuano. In David Perlmutter (ed.), Studies in relational
grammar, Vol. 1, 143–218. Chicago: University of Chicago Press.
Bentley, William Holman. 1887. Dictionary and grammar of the Kongo language. London: Trübner.
Bleek, Wilhelm Heinrich Immanuel. 1873. Report of Dr. Bleek concerning his researches into the Bushman
language, presented to the Honorable House of Assembly by command of His Excellency the Governor (Cape
of Good Hope). Cambridge, UK: Published as a micro-fiche by Chadwyck-Healey in association with the
British Library and Avero Publications.
Bresnan, Joan & Lioba Moshi. 1993. Object asymmetries in comparative Bantu syntax. In Sam Mchombo
(ed.), Theoretical aspects of Bantu grammar, 47–91. Stanford: CSLI Publications.
Brinkmann, Ursula. 1997. The locative alternation in German: Its structure and acquisition. Amsterdam: John
Benjamins.
Brugiotti da Vetralla, Giacinto. 1659. Regulae quaedam pro difficillimi Congensium idiomatis faciliori captu,
ad grammaticae normam redactae, a F. Hyacintho Brusciotto, a Vetralla [Some rules for the more easy
understanding of the most difficult idiom of the people of the Congo, brought into the form of grammar
by Fr. Hyacinth Brusciotto from Vetralla]. Rome: Tipografia della Congregazione di Propaganda Fide.
Bugaeva, Anna. 2010. Ainu applicatives in typological perspective. Studies in Language 34(4). 749–801.
 1 Applicative constructions: An introductory overview 53

Carochi, Horacio. 1645. Arte de la lengva mexicana con la declaracion de los adverbios della. Mexico City: Juan
Ruyz.
Chung, Sandra. 1976. An object-creating rule in Bahasa Indonesia. Linguistic Inquiry 7(1). 41–87.
Chung, Sandra. 1983. An object-creating rule in Bahasa Indonesia. In David Perlmutter (ed.), Studies in
Relational Grammar I, 219–271. Chicago: University of Chicago Press.
Comrie, Bernard. 1989. Language universals and linguistic typology: Syntax and morphology. 2nd edn. Oxford:
Blackwell.
Comrie, Bernard, Iren Hartmann, Martin Haspelmath, Andrej Malchukov & Søren Wichmann. 2015.
Introduction. In Andrej Malchukov & Bernard Comrie (eds.), Valency classes in the world’s languages,
Vol. 1, 3–26. Berlin: De Gruyter Mouton.
Creissels, Denis. 2006. Syntaxe générale, une introduction typologique. 2 vols. Paris: Lavoisier-Hermès Science.
Creissels, Denis. 2010. Benefactive applicative periphrases. In Fernando Zúñiga & Seppo Kittilä (eds.),
Benefactives and malefactives: Typological perspectives and case studies, 29–69. Amsterdam: John
Benjamins.
Creissels, Denis. 2014. P-lability and radical P-alignment. Linguistics 52(4). 911–944.
Creissels, Denis. Forthcoming. Transitivity, valency, and voice. Oxford: Oxford University Press.
Creissels, Denis & Jérémie Kouadio. 2010. Ditransitive constructions in Baule. In Andrej Malchukov, Martin
Haspelmath & Bernard Comrie (eds.) Studies in ditransitive constructions: A comparative handbook,
166–189. Berlin: De Gruyter Mouton.
Croft, William. 2022. Morphosyntax: Constructions of the world’s languages. Cambridge: Cambridge University
Press.
Cysouw, Michael. 2023. Encyclopaedia of German diatheses. Berlin: Language Science Press.
Dammann, Ernst. 1961. Das Applikativum in den Bantusprachen. Zeitschrift der Deutschen Morgenländischen
Gesellschaft 111(1). 160–169.
Dayley, Jon P. 1981. Voice and ergativity in Mayan languages. Journal of Mayan Linguistics 2. 3–82.
Dixon, R. M. W. 1988. A grammar of Boumaa Fijian. Chicago: The University of Chicago Press.
Dixon, R. M. W. 2012. Basic Linguistic Theory, Volume 3: Further grammatical topics. Oxford: Oxford University
Press.
Dixon, R. M. W. & Alexandra Aikhenvald. 1997. A typology of argument-determined constructions. In Joan
Bybee, John Haiman & Sandra Thompson (eds.), Essays on language function and language type, 71–113.
Amsterdam: John Benjamins.
Dixon, R. M. W. & Alexandra Aikhenvald. 2000. Introduction. In R. M. W. Dixon & Alexandra Aikhenvald
(eds.), Changing valency: Case studies in transitivity, 1–29. Cambridge: Cambridge University Press.
Donohue, Mark. 2003. Morphological templates, headedness, and applicatives in Barupu. Oceanic
Linguistics 42(1). 111–143.
Endemann, Karl. 1876. Versuch einer Grammatik des Sotho. Berlin: Wilhelm Hertz.
Frantz, Donald. 2009. Blackfoot grammar. 2nd edn. Toronto: Toronto University Press.
Gil, David. 2002. The prefixes di- and N- in Malay/Indonesian dialects. In Fay Wouk & Malcolm Ross (eds.),
The history and typology of western Austronesian voice systems, 241–283. Canberra: Pacific Linguistics.
Givón, Talmy. 2001. Syntax. 2 vols. Amsterdam: John Benjamins.
Golluscio, Lucía. 2007. Morphological causatives and split intransitivity in Mapudungun. International
Journal of American Linguistics 43(2). 209–238.
Grinevald, Colette & Mark Peake. 2012. Ergativity and voice in Mayan languages: A functional-typological
approach. In Gilles Authier & Katharina Haude (eds.), Ergativity, valency and voice, 51–110. Berlin: De
Gruyter Mouton.
Guthrie, Malcolm. 1967–1971. Comparative Bantu: An introduction to the comparative linguistics and prehistory
of the Bantu languages, Vols. I–IV. Farnborough: Gregg.
Haspelmath, Martin. 2011. On S, A, P, T, and R as comparative concepts for alignment typology. Linguistic
Typology 15(3). 535–567.
54 Fernando Zúñiga and Denis Creissels

Heidinger, Steffen. 2019. Reflexive and unmarked anticausatives in French and Spanish: Frequency of
transitive use and undergoer overlap. Languages 216(4). 53–69.
Hernández-Green, Néstor. 2016. Registration versus applicative constructions in Acazulco Otomi.
International Journal of American Linguistics 82(3). 353–383.
Hopper, Paul & Elisabeth Closs Traugott. 2003. Grammaticalization. Cambridge: Cambridge University
Press.
Ichihashi-Nakayama, Kumiko. 1996. The “applicative” in Hualapai: Its functions and meanings. Cognitive
Linguistics 7(2). 227–239.
Jeong, Youngmi. 2007. Applicatives: Structure and interpretation from a minimalist perspective. Amsterdam:
John Benjamins.
Jerro, Kyle. 2017. The causative-instrumental syncretism. Journal of Linguistics 53(4). 751–788.
Kaufman, Terrence. 1990. Algunos rasgos estructurales de los idiomas Mayances con referencia especial
al K’iche’. In Nora England & Stephen Elliott (eds.), Lecturas sobre la lingüística maya, 59–114. Antigua
Guatemala: Centro de Investigaciones Regionales de Mesoamérica.
Kimenyi, Alexandre. 1980. A relational grammar of Kinyarwanda. Berkeley: University of California Press.
Kiyosawa, Kaoru & Donna Gerdts. 2010. Salish applicatives. Leiden: Brill.
Kroeger, Paul. 2005. Analyzing grammar: An introduction. Cambridge: Cambridge University Press.
Kulikov, Leonid. 2010. Voice typology. In Jae Jung Song (ed.), The Oxford handbook of linguistic typology,
368–398. Oxford: Oxford University Press.
Launey, Michel. 1979. Introduction à la langue et à la littérature aztèques, tome 1: grammaire. Paris:
L’Harmattan.
Launey, Michel. 1994. Une grammaire omniprédicative, essai sur la morphosyntaxe du nahuatl classique. Paris:
CNRS éditions.
Launey, Michel. 2011. An introduction to Classical Nahuatl. Translated & adapted by Christopher Mackay.
Cambridge: Cambridge University Press.
Lazard, Gilbert. 1994. L’actance. Paris: Presses universitaires de France.
Lehmann, Christian. 2006. Participant roles, thematic roles and syntactic functions. In Tasaku Tsunoda &
Taro Kageyama (eds.), Grammatical relations: Festschrift for Masayoshi Shibatani, 153–174. Amsterdam:
John Benjamins.
Lehmann, Christian & Elisabeth Verhoeven. 2006. Extraversive transitivization in Yucatec Maya and the
nature of the applicative. In Leonid Kulikov, Andrej Malchukov & Peter de Swart (eds.), Case, valency
and transitivity, 465–493. Amsterdam: John Benjamins.
Leroy, Jacqueline. 2003. Grammaire du mankon: Langue du Bantou des Grassfields parlée dans la province
Nord-Ouest du Cameroun. Paris: University of Paris 3 dissertation.
Levin, Beth. 2015. Semantics and pragmatics of argument alternations. Annual Review of Linguistics 1.
63–83.
Malchukov, Andrej. 2015. Valency classes and alternations: Parameters of variation. In Andrej Malchukov &
Bernard Comrie (eds.), Valency classes in the world’s languages, Vol. 1: Introducing the framework, and
case studies from Africa and Eurasia, 73–130. Berlin: De Gruyter Mouton.
Malchukov, Andrej. 2016. “Ambivalent voice”: Markedness effects in valency change. In Taro Kageyama &
Wesley M. Jacobsen (eds.), Transitivity and valency alternations, 389–422. Berlin: De Gruyter Mouton.
Malchukov, Andrej. 2017. Markedness effects in applicative formation. In Albert Álvarez González & Ía
Navarro (2017), Verb valency changes: Theoretical and typological perspectives, 3–29. Amsterdam: John
Benjamins.
Marantz, Alec. 1993. Implications of asymmetries in double object constructions. In Sam A. Mchombo (ed.),
Theoretical aspects of Bantu grammar, 113–150. Stanford: CSLI Publications.
Meeussen, Achille Emile. 1967. Bantu grammatical reconstructions. Africana Linguistica 3(1). 79–121.
Meinhof, Carl & Nicolaas J. van Warmelo. 1932. Introduction to the phonology of the Bantu languages. Being
the English version of “Grundriss einer Lautlehre der Bantusprachen” by Carl Meinhof, translated, revised
 1 Applicative constructions: An introductory overview 55

and enlarged in collaboration with the author and Dr. Alice Werner by N. J. van Warmelo. Berlin: Dietrich
Reimer Verlag.
Melnar, Lynette Renee. 1998. Caddo verb morphology. Chicago: University of Chicago dissertation.
Mel’čuk, Igor. 2006. Aspects of the theory of morphology. Edited by David Beck. Berlin: Mouton de Gruyter.
Messerschmidt, Maria. 2022. Increasing the valency of motion verbs: The case of the portative construction.
STUF – Language Typology and Universals 75(1). 161–197.
Mora-Marín, David. 2003. Historical reconstruction of Mayan applicative and antidative constructions.
International Journal of American Linguistics 69(2). 186–228.
Mosel, Ulrike. 1991. Towards a typology of valence. In Hansjakob Seiler & Waldfried Premper (eds.),
Partizipation. Das sprachliche Erfassen von Sachverhalten, 240–251. Tübingen: Narr.
Munro, Pamela. 2000. The leaky grammar of the Chickasaw applicatives. In Arika Okrent & John Boyle
(eds.), The proceedings from the main session of the Chicago Linguistic Society’s thirty-sixth meeting,
285–310. Chicago: Chicago Linguistic Society.
Newman, Stanley. 1967 [2021]. Classical Nahuatl. In Norman McQuown (eds.), Handbook of Middle American
Indians, Vol. 5: Linguistics, 179–199. New York: University of Texas Press.
Ngonyani, Deo & Peter Githinji. 2006. The asymmetrical nature of Bantu applicative constructions. Lingua
116(1). 31–63.
Nichols, Johanna. 1992. Linguistic diversity in space and time. Chicago: University of Chicago Press.
Norman, William. 1978. Advancement rules and syntactic change: The loss of instrumental voice in Mayan.
In Jeri Jaeger, Anthony Woodbury, Farrell Ackerman, Christine Chiarello, Orin Gensler, John Kingston,
Eve Sweetser, Henry Thompson & Kenneth Whistler (eds.), Proceedings of the 4th Annual Meeting of the
Berkeley Linguistics Society (General Session), 458–476. Berkeley: Berkeley Linguistics Society.
Overall, Simon. 2017. A grammar of Aguaruna (Iiniá Chicham). Berlin: De Gruyter Mouton.
Pacchiarotti, Sara. 2020. Bantu applicative constructions. Stanford: CSLI Publications.
Pacchiarotti, Sara & Fernando Zúñiga (eds.). 2022. Applicative morphology: Neglected syntactic and
non-syntactic functions. Berlin: Mouton de Gruyter.
Pardeshi, Prashant. 1998. A contrastive study of benefactive constructions in Japanese and Marathi. Sekai
no nihongo kyôiku 8. 141–165.
Parker, Gary. 1969. Ayacucho Quechua grammar and dictionary. The Hague: Mouton.
Peterson, David. 2007. Applicative constructions. Oxford: Oxford University Press.
Polinsky, Maria. 2013. Applicative constructions. In Matthew Dryer & Martin Haspelmath (eds.), The world
atlas of language structures online, Ch. 109. Leipzig: Max Planck Institute for Evolutionary Anthropology.
http://wals.info/chapter/109 (accessed 24 April 2020).
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Renaudier, Marie. 2012. Dérivation et valence en Seereer. Lyon: Lumière University Lyon 2 dissertation.
Rhodes, Richard. 2010. Relative root complement: a unique grammatical relation in Algonquian syntax.
In Jan Wohlgemuth & Michael Cysouw (eds.), Rara and rarissima: Documenting the fringes of linguistic
diversity, 305–324. Berlin: De Gruyter Mouton.
Rincón, Antonio del. 1595 [1895]. Arte mexicana / Gramática y vocabulario mexicanos por el padre Antonio del
Rincon. México: Oficina Tipográfica de la Secretaría de Fomento.
Rood, David. 1976. Wichita grammar. New York: Garland.
Rubino, Carl. 2005. Iloko. In Alexander Adelaar & Nikolaus Himmelmann (eds.), The Austronesian languages
of Asia and Madagascar, 326–349. London: Routledge.
Salas, Adalberto. 2006. El mapuche o araucano: Fonología, gramática y antología de cuentos. 2nd, revised edn.
Santiago de Chile: Centro de Estudios Públicos.
Shibatani, Masayoshi. 1996. Applicatives and benefactives: A cognitive account. In Masayoshi Shibatani &
Sandra Thompson (eds.), Grammatical constructions, 157–194. Oxford: Clarendon Press.
Shibatani, Masayoshi. 2006. On the conceptual framework for voice phenomena. Linguistics 44(2). 217–269.
56 Fernando Zúñiga and Denis Creissels

Shibatani, Masayoshi & Prashant Pardeshi. 2002. The causative continuum. In Masayoshi Shibatani (ed.),
The grammar of causation and interpersonal manipulation, 85–126. Amsterdam: John Benjamins.
Sneddon, James Neil. 1996. Indonesian: A comprehensive grammar. London: Routledge.
Stapleton, Walter Henry. 1903. Comparative handbook of Congo languages. Bolobo, Congo Independent
State: Hannah Wade Printing Press of the Baptist Missionary Society.
Steere, Edward. 1884. A handbook of the Swahili language as spoken at Zanzibar. London: Society for
Promoting Christian Knowledge.
Sullivan, Thelma D. 1988. Compendium of Nahuatl grammar. Translated from the Spanish by Thelma D.
Sullivan & Neville Stiles. Edited by Wick R. Miller & Karen Dakin. Salt Lake City: University of Utah Press.
Svenonius, Peter. 2014. Generalized applicatives: Reassessing the lexical-functional divide. Theoretical
Linguistics 40(3/4). 439–446.
Torrend, Julius. 1891. A comparative grammar of the South-African Bantu languages comprising those of
Zanzibar, Mozambique, the Zambezi, Kafirland, Benguela, Angola, the Congo, the Ogowe, the Cameroons,
the Lake Region, etc. London: Kegan Paul, Trench, Trübner & Co.
Vamarasi, Marit Kana. 1999. Grammatical relations in Bahasa Indonesia. Canberra: Pacific Linguistics.
Watanabe, Honoré. 2015. Valency classes in Sliammon Salish. In Andrej Malchukov & Bernard Comrie (eds.),
Valency classes in the world’s languages, Vol. 2: Case studies from Austronesia, the Pacific, the Americas, and
theoretical outlook, 1313–1358. Berlin: de Gruyter.
Watkins, Mark Hanna. 1937. A grammar of Chichewa: A Bantu language of British Central Africa. Philadelphia:
Linguistic Society of America.
Wechsler, Stephen. 2015. Word meaning and syntax: Approaches to the interface. Oxford: Oxford University
Press.
Wicks, Douglas Allen. 2006. A partial grammar sketch of Lunyole with emphasis on the applicative
construction(s). La Mirada, CA: Biola University MA thesis.
Woodbury, Anthony. 2011. Atkan Aleut “unclitic” pronouns and definiteness: A multimodular analysis. In
Etsuyo Yuasa, Tista Bagchi & Katharine Beals (eds.), Pragmatics and Autolexical Grammar: In honor of
Jerry Sadock, 125–142. Amsterdam: John Benjamins.
Zavala, Roberto. 2015. Instrumentos y comitativos en las lenguas de la familia mixezoque. Una exploración
desde la gramaticalización y la tipología. Amerindia 37(2). 189–228.
Zúñiga, Fernando. 2009. An exploration of the diachrony of Mapudungun valency-changing operations.
Ms., University of Zurich.
Zúñiga, Fernando. 2014. (Anti-)cliticization in Mapudungun. Morphology 24. 161–175.
Zúñiga, Fernando & Seppo Kittilä. 2019. Grammatical voice. Cambridge: Cambridge University Press.
Fernando Zúñiga
2 Questionnaire on applicative constructions
Please note that the structure of the contributions need not reflect the order of the ques-
tions in the questionnaire, but all the points listed in the questionnaire are expected to
be addressed in one way or another and should be briefly recapitulated in the conclud-
ing section of the article.

The language(s)
Please give (i) a general characterization of the language(s) in terms of geography, gene-
alogy, contact, sociolinguistic status, dialectal variation, etc. In case you are dealing with
an individual language, please also provide the ISO and Glottolog codes that identify
it. Either as an additional subsection or as a separate article section, please give (ii) a
short survey of the basic elements of clause structure necessary to understand your
contribution. In language-family contributions, the general characterization part will
probably need to be somewhat longer than a mere couple of sentences and might even
come close to 300 or 400 words. With languages whose morphology and/or syntax is
particularly complex, the morphosyntactic essentials part will probably take up most
of the space. Whatever kind of balance you end up striking between both introductory
subsections, please make sure these two parts do not exceed 1,500 words in total length.

1 Morphology
1.1 How are the main ACs of the language(s) marked on the predicate (e.g., affixally,
prosodically, with an auxiliary, a functional verb, a [clitic?] particle. . .)?
1.2 If there are applicative serial verb constructions or converbal constructions,
which verbs are involved, how can they be described structurally and function-
ally? How grammaticalized are the functional verbs?
1.3 If there is allomorphy for the individual markers, what conditions it (phonology,
morphology, lexical predicate class [e.g., valency-based])?
1.4 Do applicativized verbs show comparable inflectional paradigms to those of their
base counterparts, or are they defective/restricted/regularized. . .?

Acknowledgments: I am indebted to Denis Creissels and Sara Pacchiarotti for their comments on earlier
versions of this questionnaire.

https://doi.org/10.1515/9783110730951-002
58 Fernando Zúñiga

2 Syntax
2.1 What is the syntactic status of the applied phrase in the different ACs in the lan-
guage(s)?
2.2 Does the syntactic status of the applied phrase’s companion arguments/adjuncts
change between the BC and the AC?
2.3 Are there restrictions on the stacking/combination of voice operations? Is applica-
tivization required, allowed, dispreferred, blocked, etc., in the context of nominal
incorporation, (anti-)passivization, (anti-)causativization, reflexives, reciprocals. . .?
2.4 Are some ACs a special subset of ditransitive/double-object constructions in terms
of argument-realization specifics, or are they only morphologically different from
underived predicates belonging to the same valency class?
2.5 How rigid/flexible is the assignment of case/agreement frames to ACs when com-
pared to BCs? See Exx. (19–20) in the position paper, where AppPs in Mapudungun
are arguably not simply default primary objects.
2.6 Does applicativization condition the access of non-core syntactic arguments to
operations such as relativization or focalization?

3 Semantics
3.1 Are different markers specialized for different semantics (e.g., valency and seman-
tic role)?
3.2 Which semantic roles can be expressed by the applied phrase? Are there semantic
roles that can be expressed only by means of an applicative construction (in either
monoclausal or biclausal constructions)? For instance, the only way to express
the location ‘at my place’ in Blackfoot (1) is as the secondary object of an applica-
tivized verb (i.e., there is no adjunct equivalent to its English translation with an
adposition):

(1) Blackfoot (Frantz 2009: 94)

It-á-ooyi-wa n-ookóówa-yi.
appl-dur-eat.ai-prox.sg 1-home-inan.sg
‘S/he eats at my place.’

3.3 If a given semantic role can be expressed both in a BC and an AC, is there a seman-
tic difference between the Applied phrase and its counterpart in the BC (e.g.,
affectedness, individuation, specificity, . . .)? Are there other semantic differences
between the AC and the BC (e.g., aspect and manner)?
3.4 Is there a pragmatic difference between the applied phrase and its counterpart in
the BC (e.g., topicalization, [contrastive] focalization)? Are there other pragmatic
differences between the AC and the BC (e.g., theticity)?
3.5 If both a BC and an AC are available with a given verb (irrespective of whether
there is a semantic opposition or not), is there a discourse-sensitive specificity to
their distribution?
 2 Questionnaire on applicative constructions 59

4 Lookalikes
4.1 If there is no bona fide applicative marking on the predicate, are there (quasi-)
productive uncoded alternations (e.g., English I bought flowers for you vs. I bought
you flowers)? How can they be described structurally and functionally? How are
they conditioned/licensed?
4.2 Are there markers that may have turned from erstwhile applicatives into strictly
valency-neutral markers (e.g., spatial/directional markers, politeness markers. . .)?

Abbreviations
ai animate intransitive
appl applicative
dur durative
inan inanimate
prox proximate
sg singular

References
Frantz, Donald. 2009. Blackfoot grammar. 2nd end. Toronto: University of Toronto Press.
Denis Creissels and Fernando Zúñiga, with George Moroz
3 Languages examined or referred
to in the present book
1 The maps
Maps 1–8 were created by George Moroz with the help of the R (R Core Team 2023)
package lingytpology (Moroz 2017).

2 The classification
Languages are listed alphabetically within each of the following five areas: North
America, South America, Africa, Pacific, and Eurasia.
The genetic affiliation of the languages listed in this index is provided along the
lines of the classification of languages adopted in WALS Online (Dryer and Haspelmath
2013), with only some minor modifications that are commented in footnotes. For lan-
guages that are not explicitly listed in WALS, the classification copies that of closely
related languages that feature in WALS and for which it can be taken for granted that
they belong to the same genus.
Apart from pidgins and creoles, for which no genetic affiliation is proposed in
WALS, the general principle of this classification is that, as a rule, it does not mention
the intermediate groupings that appear in other classifications, restricting to only two
levels: genus and family. Family refers to the highest level widely accepted by special-
ists, while genus is commented on in WALS Online (https://wals.info/languoid/geneal-
ogy) as follows:

The notion genus is explained in Dryer (1989). It is intended as a level of classification which is
comparable across the world, so that a genus in one family is intended to be comparable in time
depth to genera in other parts of the world. The choice of term is intended to match the general
idea of genus in biological classification, where a genus is a set of species that are clearly closely
related to each other (and where words in everyday language often correspond to genera rather
than species). In the genealogical classification of languages, a genus is a group of languages whose
relatedness is fairly obvious without systematic comparative analysis, and which even the most
conservative “splitter” would accept. Genealogical groups deeper than a genus are often less
obvious and in the absence of detailed comparative work are often not universally accepted. If
there is evidence of time depth of groups, the genus would not have a time depth greater than 3500
or 4000 years. A genus may have a time depth much less than this, but if the time of the split of one
group of languages from other languages in the family appears to be greater than 4000 years, then
this constitutes a reason to say that this group of languages is a separate genus.

Here, as in WALS, levels of classification lower than that of genus are not taken into
account. For example, the classification of Somali refers to the genus “Lowland East

https://doi.org/10.1515/9783110730951-003
62 Denis Creissels and Fernando Zúñiga, with George Moroz

Cushitic”, ignoring the lower-level groupings “Nuclear Lowland East Cushitic” and
“Omo-Tana”. Similarly, Northern Paiute is classified as a member of the Northern
Uto-Aztecan genus, ignoring the lower-level grouping “Numic”.
An intermediate level between genus and family, that of subfamily, is only provided
when the genetic affiliation of the language in question is more commonly character-
ized with reference to the subfamily than to the genus to which it belongs. For example,
Baule is classified here as “Baule, Tano, Kwa, Niger-Congo”. The reason is that, in the lit-
erature, whatever the validity of the Kwa subfamily within the Niger-Congo family may
be, “Kwa”, rather than “Tano”, is the label commonly used to characterize the position
of Baule within the Niger-Congo family. More generally, the subfamilies of Niger-Congo
(Atlantic, Kwa, Benue-Congo), Afroasiatic (Cushitic) and Eastern Sudanic (Nilotic) that
do not meet the definition of genera (and whose very validity as genetic groupings is
sometimes questionable) but are commonly mentioned in the classification of Niger-
Congo, Afroasiatic, and Eastern Sudanic languages have been systematically added.
This said, some particular cases must be considered. First, some groups of lects
commonly treated as constituting a single (macro-)language, like the dialectal varieties
of Wolof, are commonly regarded as isolates within a genetic grouping—in the case of
Wolof, North Atlantic ⸦ Atlantic ⸦ Niger-Congo—whose time depth exceeds that admit-
ted for genera. In such cases, the name of the language is repeated as that of the genus
that coincides with the (macro-)language in question. For example, Wolof is classified
here as “Wolof (language), Wolof (genus), Atlantic (subfamily), Niger-Congo (family)”.
A second particular case is that of languages belonging to a genus that is not com-
monly regarded as included into a higher-level family (and consequently, also has the
status of family in the sense given here to this term). In such cases, the name of the lan-
guage is followed by a single label referring to the genus. Mayan languages are a case in
point. For example, K’iche’ is classified here as “K’iche’ (language), Mayan (genus and
family)”.
The third particular case concerns language with no known relatives, such as Cofán
(Colombia). In such cases, the name of the language or language variety if followed by
a single label that simply reproduces the name of the language, for example “Cofán
(language), Cofán (genus and family)”.

References
Dryer, Matthew S. 1989. Large linguistic areas and language sampling. Studies in Language 13. 257–292.
Dryer, Matthew S. & Martin Haspelmath (eds.). 2013. WALS Online (v2020.3) [Data set]. Zenodo. https://doi.
org/10.5281/zenodo.7385533 (Available online at https://wals.info, Accessed on 2023–03–08.)
Moroz, George. 2017. lingtypology: easy mapping for Linguistic Typology. <https://CRAN.R-project.org/
package=lingtypology>.
R Core Team. 2023. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
3 Languages examined or referred to in the present book 63

Map 1: World.
64 Denis Creissels and Fernando Zúñiga, with George Moroz

NORTH AMERICA
language genetic affiliation number
Acjachemem Northern Uto-Aztecan, Uto-Aztecan 52
Arapaho Algonquian, Algic 1
Blackfoot Algonquian, Algic 2
Caddo Caddo, Caddoan 17
Cahuilla Northern Uto-Aztecan, Uto-Aztecan 53
Central Alaskan Yup’ik Eskimo, Eskimo-Aleut 19
Chalcatongo Mixtec Mixtec, Oto-Mangean 43
Cheyenne Algonquian, Algic 3
Chickasaw Muskogean 37
Chontal Mayan 24
Classical Nahuatl Aztecan, Uto-Aztecan 54
Cupeño Northern Uto-Aztecan, Uto-Aztecan 55
Ch’ol Mayan 23
Ch’orti’ Mayan 25
Eastern Highlands Otomi Otomian, Oto-Manguean 38
Gosiute Shoshoni Northern Uto-Aztecan, Uto-Aztecan 56
Guarijío Tarahumaran, Uto-Aztecan 57
Hopi Northern Uto-Aztecan, Uto-Aztecan 58
Hul’q’umi’num Central Salish, Saolishan 47
Innu Algonquian, Algic 4
Itzaj Mayan 26
Ixil Mayan 27
Kalaallisut Eskimo, Eskimo-Aleut 20
Kaqchikel Mayan 28
K’iche’ Mayan 29
Ktunaxa Kutenai 22
Menominee Algonquian, Algic 5
Meskwaki Algonquian, Algic 6
Moose Cree Algonquian, Algic 7
Myaamia Algonquian, Algic 8
Nishnaabemwin Algonquian, Algic 9
Northern East Cree Algonquian, Algic 101
Northern Paiute Northern Uto-Aztecan, Uto-Aztecan 59
Northern Tepehuan Tepiman, Uto-Aztecan 60
Northern Zapotec Zapotecan, Oto-Manguean 39
Ojicree Algonquian, Algic 11
Okanagan Interior Salish, Salishan 48
Otomi Otomian, Oto-Manguean 40
Pahka’anil Northern Uto-Aztecan, Uto-Aztecan 61
Pima Bajo Tepiman, Uto-Aztecan 62
Plains Cree Algonquian, Algic 12
Poqomam Mayan 30
Q’anjob’al Mayan 31
Quiegolani Zapotec Zapotecan, Oto-Manguean 44
San Felipe Otomi Otomian, Oto-Manguean 41

1 Southern East Cree (mentioned in Chapter 18) does not appear on the map but is spoken south of
Northern East Cree.
 3 Languages examined or referred to in the present book 65

(continued)

NORTH AMERICA
language genetic affiliation number
San Lucas Quiaviní Zapotec Zapotecan, Oto-Manguean 42
Serrano Northern Uto-Aztecan, Uto-Aztecan 63
Shuswap Interior Salish, Salishan 49
Sliammon Central Salish, Salishan 50
Southeastern Huastec Mayan 32
Southwestern Ojibwe Algonquian, Algic 13
Swampy Cree Algonquian, Algic 14
Tamazulápam Mixe Mixe-Zoque 45
Tarahumara Tarahumaran, Uto-Aztecan 64
Teotitlán Zapotec Zapotecan, Oto-Manguean 46
Tojolabal Mayan 33
Tseltal Mayan 34
Tsotsil Mayan 35
Tümpisa Shoshoni Northern Uto-Aztecan, Uto-Aztecan 65
Tz’utujil Mayan 36
Unami Algonquian, Algic 15
Unangan Aleut, Eskimo-Aleut 21
Upper Necaxa Totonac Totonacan 51
Ute Northern Uto-Aztecan, Uto-Aztecan 66
Walapai Yuman, Hokan 18
Western Naskapi Algonquian, Algic 16
Western Shoshoni Northern Uto-Aztecan, Uto-Aztecan 67
Wixárika / Huichol Corachol, Uto-Aztecan 68
Yaqui Cahita, Uto-Aztecan 69

Map 2: Southern North America.

66 Denis Creissels and Fernando Zúñiga, with George Moroz

SOUTH AMERICA
language genetic affiliation number
Achagua Japura-Colombia, Arawakan 71
Aguaruna Jivaroan 83
Arabela Zaparoan 116
Ayacucho Quechua Quechuan 96
Barasano Tucanoan 107
Calderón Highland Quichua Quechuan 97
Camsá Camsá 87
Cañar Highland Quichua Quechuan 98
Carapana Tucanoan 108
Chimborazo Highland Quichua Quechuan 99
Cofán Cofán 88
Dâw Nadahup 92
Desano Tucanoan 109
Hup Nadahup 93
Imbabura Highland Quichua Quechuan 100
Iquito Zaparoan 119
Kakua Cacua-Nukak 91
Koreguaje Tucanoan 110
Kotiria Tucanoan 111
Kubeo Tucanoan 112
Loja Highland Quichua Quechuan 101
Mapudungun Araucanian 70
Miraña Boran 77
Muinane Witoto, Witotoan 78
Muniche Muniche 89
Murui Witoto, Witotoan 79
Napo Lowland Quechua Quechuan 102
Ocaina Witoto, Witotoan 80
Pastaza Quechua Quechuan 103
Piapoco Japura-Colombia, Arawakan 72
Puinave Puinave 117
Resígaro Japura-Colombia, Arawakan 73
Salasaca Highland Quichua Quechuan 104
Secoya Tucanoan 113
Shawi Cahuapanan 812
Shiwiar Jivaroan 84
Shiwilu Cahuapanan 82
Siona Tucanoan 114
Tanimuka Tucanoan 115
Tariana Japura-Colombian, Arawakan 74
Tena Lowland Quichua Quechuan 105
Tikuna Tikuna 106
Toba / Qom Qom, Guaycuruan 86

2 San Martín Quechua (mentioned in Chapter 11) does not appear on the map but is spoken south of
Shawi.
 3 Languages examined or referred to in the present book 67

(continued)

SOUTH AMERICA
language genetic affiliation number
Tukano Tucanoan 116
Urarina Urarina 80
Wampis Jivaroan 85
Warekena Alto Orinoco, Arawakan 75
Yagua Peba-Yaguan 95
Yuhup Nadahup 94
Yukuna Japura-Colombian, Arawakan 76

Map 3: Northwestern South America.

68 Denis Creissels and Fernando Zúñiga, with George Moroz

AFRICA
language genetic affiliation number
Agar Dinka Western Nilotic, Nilotic, Eastern Sudanic 171
Akie Southern Nilotic, Nilotic, Eastern Sudanic 172
Alagwa Southern Cushitic, Cushitic, Afroasiatic 154
Amharic Semitic, Afroasiatic 191
Arbore Lowland East Cushitic, Cushitic, Afroasiatic 155
Asimjeeg Datooga Southern Nilotic, Nilotic, Eastern Sudanic 173
Ateso Eastern Nilotic, Nilotic, Eastern Sudanic 174
Barbayiiga Datooga Southern Nilotic, Nilotic, Eastern Sudanic 175
Bari Eastern Nilotic, Nilotic, Eastern Sudanic 176
Baule Tano, Kwa, Niger-Congo 169
Beja Beja, Cushitic, Afroasiatic 156
Bijogo Bijogo, Atlantic, Niger-Congo 120
Boni Lowland East Cushitic, Cushitic, Afroasiatic 157
Burunge Southern Cushitic, Cushitic, Afroasiatic 158
Cherang’any Southern Nilotic, Nilotic, Eastern Sudanic 177
Chewa Bantu, Benue-Congo, Niger-Congo 130
Chingoni Bantu, Benue-Congo, Niger-Congo 131
Dhaasanac Lowland East Cushitic, Cushitic, Afroasiatic 159
Dholuo Western Nilotic, Nilotic, Eastern Sudanic 178
Ding Bantu, Benue-Congo, Niger-Congo 132
Elmolo Lowland East Cushitic, Cushitic, Afroasiatic 160
Gedeo Highland East Cushitic, Cushitic, Afroasiatic 161
Gisamjanga Datooga Southern Nilotic, Nilotic, Eastern Sudanic 179
Gombe Fula Fula-Serer, Atlantic, Niger-Congo3 121
Iraqw Southern Cushitic, Cushitic, Afroasiatic 162
Jóola Fóoñi Joola, Atlantic, Niger-Congo 122
Jumjum Western Nilotic, Nilotic, Eastern Sudanic 180
Kikuyu Bantu, Benue-Congo, Niger-Congo 133
Kongo ya Leta Bantu, Benue-Congo, Niger-Congo 134
Laalaa Cangin, Atlantic, Niger-Congo 123
Lango Western Nilotic, Nilotic, Eastern Sudanic 181
Lengola Bantu, Benue-Congo, Niger-Congo 135
Lomwe Bantu, Benue-Congo, Niger-Congo 136
Londo Bantu, Benue-Congo, Niger-Congo 137
Luba-Kasai Bantu, Benue-Congo, Niger-Congo 138
Lunda Bantu, Benue-Congo, Niger-Congo 139
Maasai Eastern Nilotic, Nilotic, Eastern Sudanic 182
Mabaan Western Nilotic, Nilotic, Eastern Sudanic 183
Mandinka Central Mande, Mande4 170
Mankanya Manjaku-Mankanya, Atlantic, Niger-Congo5 124

3 Peul is the French name of the language called Fula in English.

4 In the Mande family, West Mande does not meet the definition of a genus, and is best regarded as a
subfamily.
5 Manjaku and Mankanya belong to the same genus, but there is no reason to select “Manjaku” rather
than “Mankanya” as a label for the genus that includes Manjaku and Mankanya.
 3 Languages examined or referred to in the present book 69

(continued)

AFRICA
language genetic affiliation number
Mankon Wide Grassfields, Benue-Congo, Niger-Congo 168
Mbuun Bantu, Benue-Congo, Niger-Congo 140
Mongo-Nkundo Bantu, Benue-Congo, Niger-Congo 141
Nandi Eastern Nilotic, Nilotic, Eastern Sudanic 184
North Boma Bantu, Benue-Congo, Niger-Congo 142
Ndut Cangin, Atlantic, Niger-Congo 125
Ngwi Bantu, Benue-Congo, Niger-Congo 143
Noon Cangin, Atlantic, Niger-Congo 126
Northern Luo Western Nilotic, Nilotic, Eastern Sudanic 185
Nsong Bantu, Benue-Congo, Niger-Congo 144
Ntandu Bantu, Benue-Congo, Niger-Congo 145
Nuer Western Nilotic, Nilotic, Eastern Sudanic 186
Nyole Bantu, Benue-Congo, Niger-Congo 146
Nzadi Bantu, Benue-Congo, Niger-Congo 147
Oromo Lowland East Cushitic, Cushitic, Afroasiatic 163
Palor Cangin, Atlantic, Niger-Congo 127
Reel Western Nilotic, Nilotic, Eastern Sudanic 187
Rendille Lowland East Cushitic, Cushitic, Afroasiatic 164
Rundi Bantu, Benue-Congo, Niger-Congo 148
Rwanda Bantu, Benue-Congo, Niger-Congo 149
Seereer Fula-Serer, Atlantic, Niger-Congo 128
Shilluk Western Nilotic, Nilotic, Eastern Sudanic 188
Sikongo Bantu, Benue-Congo, Niger-Congo 150
Somali Lowland East Cushitic, Cushitic, Afroasiatic 165
Tiene Bantu, Benue-Congo, Niger-Congo 151
Turkana Eastern Nilotic, Nilotic, Eastern Sudanic 189
Tswana Bantu, Benue-Congo, Niger-Congo 152
Tunni Lowland East Cushitic, Cushitic, Afroasiatic 166
Wolof Wolof, Atlantic, Niger-Congo 129
Xamtanga Central Cushitic, Cushitic, Afroasiatic 167
Yao Bantu, Benue-Congo, Niger-Congo 153
70 Denis Creissels and Fernando Zúñiga, with George Moroz

Map 4: Western Africa. Map 5: Eastern and southern Africa.

PACIFIC
language genetic affiliation number
Abui Alor-Pantar, Greater West Bomberai 246
Alamblak Sepik Hill, Sepik 244
Amele Mabuso, Trans-New Guinea 242
Ampenan Sasak Malayo-Sumbawan, Austronesian 217
Arabana-Wangkanguru Central Pama-Nyungan, Pama-Nyungan 195
Aruamu / Mikarew Ruboni, Ramu-Lower Sepik 238
Bardi Nyulnyulan 194
Barupu Warupu, Skou 245
Boumaa Fijian Oceanic, Austronesian 218
Coastal Marind Marind-Yaqai, Trans-New Guinea 213
Central Asmat Asmat-Kamrau Bay 214
Dalabon Marne, Gunwinyguan 191
Diyari Central Pama-Nyungan, Pama-Nyungan 196
Dyirbal Northern Pama-Nyungan, Pama-Nyungan 197
Fore Fore-Gimi, Trans-New Guinea 237
Goreng-Goreng Southeastern Pama-Nyungan, Pama-Nyungan 198
Gungabula Northern Pama-Nyungan, Pama-Nyungan 199
Hua Siane-Yagaria, Trans-New Guinea 236
Ilocano Northern Luzon, Autronesian 219
Imonda Border 231
Kalkatungu Northern Pama-Nyungan, Pama-Nyungan 200
Kopar Lower Sepik, Ramu-Lower Sepik 239
Lower Grand Valley Dani Dani, Trans-New Guinea 232
Makasar South Sulawesi, Austronesian 220
 3 Languages examined or referred to in the present book 71

(continued)

PACIFIC
language genetic affiliation number
Margany Northern Pama-Nyungan, Pama-Nyungan 201
Marrithiyel Bringen, Western Daly 212
Mian Ok, Trans-New Guinea 215
Mountain Arapesh Kombio-Arapesh, Torricelli 248
Murrinhpatha Murrinhpatha, Southern Daly 210
Nasal Nasal, Austronesian 221
Nen Nambu, Yam 251
Ngamini Central Pama-Nyungan, Pama-Nyungan 202
Ngan’gityemerri Ngankikurungkurr, Southern Daly 211
Ngiyambaa Southern Pama-Nyungan, Pama-Nyungan 203
Ngkolmpu Kanum, Yam 252
Pendau Celebic, Autronesian 222
Pitta-Pitta Central Pama-Nyungan, Pama-Nyungan 204
Salako Land Dayak, Autronesian 223
Selepet Huon, Trans-New Guinea 233
Standard Indonesian Malayo-Sumbawan, Austronesian 224
Sundanese Malayo-Sumbawan, Austronesian 225
Tairora Tairora, Trans-New Guinea 235
Tauya Rai Coast, Trans-New Guinea 243
Teiwa Alor-Pantar, Greater West Bomberai 247
Telefol Ok, Trans-New Guinea 216
Toba Batak Northwest Sumatra-Barrier Islands, Austronesian 226
Tolai Oceanic, Austronesian 227
Tongan Oceanic, Austronesian 228
Tukang Besi Celebic, Austronesian 229
Urim Urim, Torricelli 249
Waka-Waka Southern Pama-Nyungan, Pama-Nyungan 205
Wambaya Wambayan, Mirndi 193
Warrgamay Northern Pama-Nyungan, Pama-Nyungan 206
Watam Lower Ramu, Ramu-Lower Sepik 240
West Coast Bajau Sama-Bajaw, Austronesian 230
Wubuy Nunggubuyu, Gunwinyguan 192
Yalarnnga Northern Pama-Nyungan, Pama-Nyungan 207
Yale Mek, Trans-New Guinea 234
Yaluyandi Central Pama-Nyungan, Pama-Nyungan 208
Yeri Wapei, Torricelli 250
Yidiny Northern Pama-Nyungan, Pama-Nyungan 209
Yimas Lower Sepik, Ramu-Lower Sepik 241
72 Denis Creissels and Fernando Zúñiga, with George Moroz

Map 6: Southwestern Pacific and northern Australia.

EURASIA
language genetic affiliation number
Abaza Northwest Caucasian 268
Abkhaz Northwest Caucasian 269
Akhvakh Avar-Andic-Tsezic, Nakh-Daghestanian 267
Bantawa Kiranti, Sino-Tibetan6 255
Bulgarian Slavic, Indo-European 279
Czech Slavic, Indo-European 280
Dutch Germanic, Indo-European 275
English Germanic, Indo-European 276
Georgian Kartvelian 2637
German Germanic, Indo-European 277
Hakha Lai Kuki-Chin, Sino-Tibetan 256
Hayu Kiranti, Sino-Tibetan 257
Hungarian Ugric, Uralic 287
Kabardian Northwest Caucasian 270
Khaling Kiranti, Sino-Tibetan 258

6 The genus labeled “Himalayish” in WALS is more commonly labeled “Kiranti”. Moreover, “Himalay-
ish” is ambiguous, since not all authors use it with the same extension.
7 Number 263 also applies to Old Georgian, which was spoken on roughly the same territory as Modern
Georgian.
 3 Languages examined or referred to in the present book 73

(continued)

EURASIA
language genetic affiliation number
Korean Koreanic 8
254
Latvian Baltic, Indo-European 273
Laz Kartvelian 264
Limbu Kiranti, Sino-Tibetan 259
Lithuanian Baltic, Indo-European 274
Macedonian Slavic, Indo-European 281
Marathi Indic, Indo-European 253
Mingrelian Kartvelian 265
Polish Slavic, Indo-European 282
Puma Kiranti, Sino-Tibetan 260
Romanian Italic, Indo-European9 278
Russian Slavic, Indo-European 283
Serbian Slavic, Indo-European 284
Slovene Slavic, Indo-European 285
Svan Kartvelian 266
Thulung Kiranti, Sino-Tibetan 261
Ubykh Northwest Caucasian 271
West Circassian Northwest Caucasian 272
Yakkha Kiranti, Sino-Tibetan 262

Map 7: Caucasus.

8 The classification of Korean depends on whether Jeju is considered a Korean dialect, or a closely relat-
ed language constituting alongside Korean a genus that can be labeled “Koreanic”.
9 A strict application of Dryer’s (1989) definitions leads to the conclusion that the Romance languages do
not constitute a genus, and that the genus to which they belong is rather the Italic branch of Indo-European.
Part II: Case studies
Individual languages
Donna B. Gerdts
4 Hul’q’umi’num’ Salish applicative
constructions
Abstract: Based on original fieldwork and data from texts, this paper details applica-
tives in Hul’q’umi’num’, spoken along the western shores of the Salish Sea in British
Columbia, Canada. Hul’q’umi’num’ has four applicative constructions marked by suf-
fixes that allow the expression of objects with such semantic roles as goal, beneficiary,
direction, and cause/stimulus. Applicatives are divided into two types: relationals,
which are formed on intransitive bases, and redirectives, formed on transitive bases.
(Transitivity is easily ascertained in Hul’q’umi’num’ due to transitive morphology and
ergative inflection.) Hul’q’umi’num’ is a direct/oblique object language: only two NPs
per verb can be direct arguments and other NPs are introduced by oblique marking.
In semantically ditransitive constructions, the applied object is always the direct object
and thus there are no non-applicative counterparts for redirective applicatives. As a
polysynthetic language, Hul’q’umi’num’ exhibits voice and valence marked by suf-
fixes—limited control, passive, causative, reflexive, reciprocal, and antipassive—as
well as lexical suffixes with the semantic meaning of nominals. These all co-occur with
the applicative suffixes. Applicative constructions are an important device for express-
ing topic-worthy NPs as direct objects or, if they are also passivized, as subjects.

1 Introduction
Hul’q’umi’num’ is the dialect of the Halkomelem language (Central Salish, Salish ISO
639-2 / 5 sal, Glottolog sali1255) spoken on Vancouver Island and neighboring islands in
British Columbia, Canada. Today there are only around twenty first language speakers
of Hul’q’umi’num’, but more than three hundred fluent and semi-fluent second lan-
guage speakers. The data in this chapter come from the author’s fieldwork (1975 to
present) and also from a corpus of texts.1 This chapter describes the applicative con-

1 Research on Hul’q’umi’num’ linguistic structure and the transcription and compilation of texts was
funded by grants from the Jacobs Research Fund and the Social Sciences and Humanities Research
Council. My thanks to the elders whose recordings make up the 5,000-line text corpus. These legacy

Acknowledgments: Thank you to Ruby Peter | Sti’tum’at and Arnold Guerin | Xwunuthut and other
speakers who shared their knowledge of Hul’q’umi’num’ on which this paper is based and who inspired and
guided our research mission. Thanks to my applicative research partners Mercedes Hinkson, Tom Hukari,
and Kaoru Kiyosawa. My heartfelt thanks to Lauren Schneider and Charles Ulrich for transliterating, editing,
and proofreading, and thanks as well as to the editors of this volume and to David Beck for comments and
corrections.

https://doi.org/10.1515/9783110730951-004
80 Donna B. Gerdts

structions found in Hul’q’umi’num’, which have been a topic of much interest to the
author over the years (see Gerdts 1984, 1988b/2016, 2010a, 2010b; Gerdts and Kiyosawa
2005; Kiyosawa and Gerdts 2010a). I refer you to Kiyosawa and Gerdts (2010b) and ref-
erences therein for a survey of applicative constructions in Salish languages.
Hul’q’umi’num’ has two types of applicative constructions, and it has two applic-
atives of each type, for a total of four applicatives (Gerdts 1988b). The first type of
applicative is the “relational” applicative, in which an applicative suffix is added to an
intransitive base to derive a transitive. The two relational applicatives are the general
relational applicative, formed with the suffix -meʔ, labelled rel, as illustrated in (1),
and the directional applicative, formed with the suffix -nəs, labelled dir, as illustrated
in (2).2

(1) c̓əq̓-meʔ-t č ceʔ kʷθə nəc̓əwməxʷ ʔi ceʔ tecəl.

surprise-rel-tr 2sg.sbj fut det visitor aux fut arrive
‘You will be surprised at the visitors when they arrive.’

(2) niʔ nəm-nəs-əs kʷθə swiw̓ləs kʷθə swəy̓qeʔ.

aux go-dir-3sbj det boy det man
‘The boy went up to the man.’

The second type of applicative is the “redirective” in which an applicative suffix is

added to a transitive base to derive a ditransitive clause. The two redirective applica-
tives are the dative applicative, formed with the suffix -as, labelled dat, as illustrated
in (3), and the benefactive applicative formed, with the suffix -əɫc, labelled ben, as illus-
trated in (4).3

(3) nem̓ č ʔam-əs-t tᶿə sqʷəmey̓ ʔə tᶿə st̓ᶿam̓!

go 2sg.sbj give-dat-tr det dog obl det bone
‘Go give the dog the bone!’

(4) nem̓ č ʔiləq-əɫc-t tᶿən̓ sil̓ə ʔə tᶿə səplil!

go 2sg.sbj buy-ben-tr det.2poss grandparent obl det bread
‘Go buy your grandfather some bread!’

stories were recorded by Tom Hukari, Wayne Suttles, and me, and transcribed by Arnold Guerin, Ruby
Peter, Theresa Thorne, and edited by Zack Gilkison, Tom Hukari, Sarah Kell, Kaoru Kiyosawa, Zoey Pe-
terson, and myself.
2 Data are presented in a modified Americanist Phonetic Alphabet; x represents the velar fricative and
x̌ the uvular fricative.
3 The dative suffix appears as -əs in an unstressed syllable.
 4 Hul’q’umi’num’ Salish applicative constructions 81

This description of Hul’q’umi’num’ applicatives is structured as follows. Section 2 gives

a brief introduction to Hul’q’umi’num’ morphosyntax. Section 3 gives details about the
morphology of applicative verbs, including their inflection for transitivity and person.
Section 4 details the syntactic status of the noun phrases in applicative constructions
and in their non-applicative counterparts, and it also discusses combinations with
other constructions, such as passives, antipassives, reflexives, reciprocals, causatives,
and lexical suffixes, all of which are formed with suffixes in Hul’q’umi’num’. Section 5
comments on the semantic range of each type of applicative and their use in discourse.
The conclusion in Section 6 gives a summary of the findings.

2 Basics of Hul’q’umi’num’ morphosyntax

Hul’q’umi’num’, a mildly polysynthetic language, has robust morphology that registers
valence and person marking. Hul’q’umi’num’ is a split ergative language (Gerdts 1988b):
third-person subjects of intransitive main clauses (5) and objects of transitive clauses
(6) are unmarked, but third-person subjects of transitive main clauses are marked by
the third-person subject suffix -əs (7).

(5) niʔ ʔiməš.

aux walk
‘He/she/it walked.’

(6) niʔ cən q̓ʷaqʷ-ət ʔə kʷθən̓ šapəl-əł.

aux 1sg.sbj club-tr obl det.2poss shovel-pst
‘I hit him with your shovel.’

(7) niʔ q̓ʷaqʷ-ət-əs tᶿə swəy̓qeʔ tᶿə speʔəθ.

aux club-tr-3sbj det man det bear
‘The man clubbed the bear (in view).’

In contrast, first- and second-person indexing works on a nominative/accusative basis.

Main clause subjects are marked with second-position clitics in both intransitive (8)
and transitive (9) clauses, while objects appear as verb suffixes (10), often fused with a
transitive marker:

(8) niʔ cən ʔiməš.

aux 1sg.sbj walk
‘I walked.’
82 Donna B. Gerdts

(9) niʔ cən q̓ʷaqʷ-ət tᶿə speʔəθ.

aux 1sg.sbj club-tr det bear
‘I clubbed the bear.’

(10) niʔ q̓ʷaqʷ-θam̓š-əs tᶿə swəy̓qeʔ.

aux club-tr.1sg.obj-3sbj det man
‘The man clubbed me.’

Compared to person marking, marking on noun phrases provides little information

about the syntactic role of a noun phrase. All argument noun phrases (including proper
nouns, plurals, and generics) are preceded by a determiner that registers semantic fea-
tures such as gender, viewpoint deixis, number, and spatial deixis. There are over fifty
determiners; the most frequently occurring are listed in Table-1 and illustrated in (11).

Table 1: Some Hul’q’umi’num’ determiners.

plain feminine (sg)

proximal tᶿə θə
non-proximal kʷθə łə
non-deictic kʷ̓

(11) tᶿə swəy̓qeʔ ‘the man’ (in view)

kʷθə słənłeniʔ ‘the women’ (out of view)
θə słeniʔ ‘the woman’ (in view)
łə sɫeniʔ ‘the woman’ (out of view)
k̓ʷ šukʷə ‘some sugar’

NP subjects and objects of active, stative, and transitive verbs in all tenses and aspects
are preceded by determiners.

(12) niʔ ʔəšəl tᶿə swəy̓qeʔ.

aux paddle det man
‘The man (in view) paddled.’

(13) niʔ q̓ʷəl tᶿə səplil.

aux cook det bread
‘The bread (in view) baked.’

(14) niʔ q̓ʷaqʷ-ət-əs tᶿə swəy̓qeʔ tᶿə speʔəθ.

aux club-tr-3sbj det man det bear
‘The man (in view) clubbed the bear (in view).’
 4 Hul’q’umi’num’ Salish applicative constructions 83

In contrast, semantically oblique noun phrases are oblique marked by the catch-all
preposition ʔə, which precedes the determiner. This preposition is used to mark a
variety of semantic roles, including goals (15) and instruments (16).

(15) nem̓ č ceʔ łe:l nem̓ ʔə-ƛ̓ ƛ̓əlpaləs!

go 2sg.sbj fut go.ashore go obl-det Cowichan.Bay
‘Go ashore at Cowichan Bay!’

(16) niʔ cən q̓ʷaqʷ-ət kʷθə speʔəθ ʔə kʷθən̓ šapəl-əł.

aux 1sg.sbj club-tr det bear obl det.2poss shovel-pst
‘I hit the bear with your shovel.’

As seen in the above examples, the verb complex (the verb and its surrounding auxil-
iaries and clitics) is in clause-initial order, while the noun phrases follow—word order
among the noun phrases is free, so the word order in (17) is also allowed.

(17) niʔ cən q̓ʷaqʷ-ət ʔə kʷθən̓ šapəl-əł kʷθə speʔəθ.

aux 1sg.sbj club-tr obl det.2poss shovel-pst det bear
‘I hit the bear with your shovel.’

Overall, we see that much of the weight of identifying the syntactic roles of the entities
related to an event is borne by the verbal morphology.

3 Morphology
As mentioned above, each of the four applicative constructions in Hul’q’umi’num’ is
associated with an applicative suffix. The general relational suffix -meʔ is suffixed to an
intransitive base to form a transitive verb and is inflected with the general transitive
suffix -(ə)t as in -meʔt.

(18) łciws ‘be tired’ łciwsmeʔt ‘be tired of him/her’

siʔsiʔ ‘be afraid’ siʔsiʔmeʔt ‘be afraid of him/her’
x̌iʔx̌eʔ ‘be ashamed’ x̌iʔx̌eʔmeʔt ‘be ashamed of him/her’
c̓əq̓ ‘be astonished’ c̓əq̓meʔt ‘be astonished, surprised at him/her’
ʔiyəs ‘be happy ʔiyəsmeʔt ‘be happy for him/her’

The directional suffix -nəs is also added to intransitive bases, but it is not followed by
the general transitive suffix.4

4 Since all other transitive verb forms have overt transitive morphology in Hul’q’umi’num’, we should
84 Donna B. Gerdts

(19) nem̓ ‘go’ nəmnəs ‘go toward him/her’

ʔewə ‘come’ ʔewənəs ‘come toward him/her’
x̌ʷčenəm ‘run’ x̌ʷčenəmnəs ‘run toward him/her’
c̓tem ‘crawl’ c̓temnəs ‘crawl toward him/her’
ʔəšəl ‘paddle’ ʔəšəlnəs ‘paddle toward him/her’

Redirective applicatives are formed from transitive bases, as seen by comparing the
mono-transitive verbs, suffixed with transitive morphology, with the benefactive verbs,
formed with the applicative suffix -əłc followed by the transitive suffix.

(20) kʷənət ‘take it’ kʷənəłcət ‘take it for him/her’

p̓et̓ᶿət ‘sew it’ p̓et̕ᶿəłcət ‘sew it for him/her’
θəyt ‘fix it’ θəyəłcət ‘fix it for him/her’
ʔiləqət ‘buy it’ ʔiləqəłcət ‘buy it for him/her’
t̕ᶿx̌ʷat ‘wash it’ t̕ᶿx̌ʷəłcət ‘wash it for him/her’

The benefactive suffix is productive in Hul’q’umi’num’: any transitive verb can be aug-
mented with this suffix so long as the meaning of benefaction is compatible with the
event. In contrast, there are only five applicative verbs formed with the dative suffix
-as, all followed by the transitive suffix.5

(21) √sem̓ ‘sell’ sam̓əst ‘sell him/her it’

√ʔeʔəm ‘give’ ʔaməst ‘give him/her it’
√ʔiw̓ ‘instruct’ ʔiw̓əst ‘show him/her it’
√xʷayəm ‘sell’ xʷayəməst ‘sell him/her it’
√yəθ ‘tell’ yəθəst ‘tell him/her about it’

Only one of the forms √sem̓ has a mono-transitive form sem̓ət ‘sell it’; the other forms
require the dative suffix in order to appear as transitive verbs. In sum, three of the four
applicative suffixes are quite productive, but the dative suffix -as appears on a few,
mostly frozen forms.
As noted above, applicatives formed with three of the four suffixes are inflected
with the general transitive suffix -t. Hul’q’umi’num’ pronominal object suffixes follow
and sometimes merge with the transitive suffix, producing paradigms as follows:

assume that either the applicative itself is a transitivizer or perhaps the form can be segmented into -n
and -s, with the later related to the causative suffix.
5 As previously noted by Hukari and Peter (1995: 371ff.), several, perhaps six, suffixes in the language
trigger vowel harmony of a preceding e vowel to a. See Gerdts and Hinkson (2004) for more discussion.
 4 Hul’q’umi’num’ Salish applicative constructions 85

(22) Object paradigm with basic transitive verb

leməθam̓š ‘look at me’
łeməθamə ‘look at you (sg)’
lemətal̓xʷ ‘look at us’
lemətalə ‘look at you (pl)’
lemət ‘look at him/her/it/them’

(23) Object paradigm with relational suffix -meʔ

łciwsmeʔθam̓š ‘be tired of me’
łciwsmeʔθamə ‘be tired of you (sg)’
łciwsmeʔtal̓xʷ ‘be tired of us’
łciwsmeʔtalə ‘be tired of you (pl)’
łciwsmeʔt ‘be tired of him/her/it/them’

(24) Object paradigm with benefactive suffix -əłc

kʷənəłcəθam̓š ‘take it for me’
kʷənəłcəθamə ‘take it for you (sg)’
kʷənəłcətal̓xʷ ‘take it for us’
kʷənəłcətalə ‘take it for you (pl)’
kʷənəłcət ‘take it for him/her/it/them’

(25) Object paradigm with dative suffix -as

ʔaməsθam̓š ‘give me it’
ʔaməsθamə ‘give you (sg) it’
ʔaməstal̓xʷ ‘give us it’
ʔaməstalə ‘give you (pl) it’
ʔaməst ‘give it to him/her/it/them’

The directional applicative suffix -nəs, which is not followed by a transitive suffix, takes
object suffixes from a slightly different (vowel-initial) paradigm.

(26) Object paradigm with the directional suffix -nəs

ʔew̓ənəsam̓š ‘come to me’
ʔew̓ənəsamə ‘come to you (sg)’
ʔew̓ənəsal̓xʷ ‘come to us’
̓
ʔewənəsalə ‘come to you (pl)’
ʔew̓ənəs ‘come to him/her/it/them’

We see this paradigm with other transitive suffixes, for example the causative suffix
-stəxʷ.
86 Donna B. Gerdts

(27) Object paradigm with causative suffix -stəxʷ

ʔəłtənəstam̓š ‘feed me’
ʔəłtənəstamə ‘feed you (sg)’
ʔəłtənəstal̓xʷ ‘feed us’
ʔəłtənəstalə ‘feed you (pl)’
ʔəłtənəstəxʷ ‘feed him/her/it/them’

Thus, applicative constructions have comparable inflectional paradigms to those of

other transitive constructions in the language.

4 Syntax
4.1 General comments

Hul’q’umi’num’ applicatives straightforwardly match the comparative notion of

applicative constructions advance in Zúñiga and Creissels (this volume).6 Relational
applicative constructions have corresponding intransitive basic clauses built on the
same root (28) and the applicative clause is syntactically transitive, marked with an
applicative suffix, with a non-theme nominal as the direct object, as seen in the direc-
tional applicative construction in (29).

(28) niʔ nem̓ kʷθə swiw̓ləs.

aux go det boy
‘The boy went.’ (Gerdts 2010b)

(29) niʔ nəm-nəs-əs kʷθə swiw̓ləs kʷθə čan.

aux go-dir-3sbj det boy det John
‘The boy went up to John.’ (Gerdts 2010b)

One way to clearly see the difference in transitivity is to compare the psychological
clause in (30) with its relational applicative counterpart in (31).

(30) niʔ siʔsiʔ kʷθə sƛ̓iʔƛ̓qəł ʔə kʷθə snəxʷəł.

aux afraid det child obl det canoe
‘The child was frightened of the car.’

6 Peterson (2007) references our research on Salish applicatives and places it in typological perspective.
 4 Hul’q’umi’num’ Salish applicative constructions 87

(31) niʔ siʔsiʔ-meʔ-t-əs kʷθə sƛ̓iʔƛ̓qəł kʷθə sqʷəmey̓.

aux afraid-rel-tr-3sbj det child det dog
‘The child was frightened of the dog.’

In (30) the stimulus of the psychological event is an adjunct, expressed as an oblique

noun phrase, while in (31) it is the direct object, expressed without oblique marking.
The verb in (30) is intransitive, while the verb in (31) is transitive, as evidenced by
the presence of overt transitive morphology. Also, as mentioned above, Hul’q’umi’num’
is a split ergative language (Gerdts 1988b: 49): third-person subjects of intransitive
main clauses are unmarked while third-person subjects of transitive main clauses are
marked by the third-person subject suffix. Third-person marking thus shows that the
basic clauses in (28) and (30) are intransitive, but the applicative clauses in (29) and (31)
are transitive.
In discussing redirective applicatives, we should note that they show all the hall-
marks of ditransitive constructions, with the sole difference being the suffixation of
applicative morphology. Ditransitive constructions are discussed at length in Gerdts
(2010a), but notably there are no indirect objects in Hul’q’umi’num’, rather the recipi-
ent/goal appears as the direct object and the patient/theme appears as an oblique object.

(32) nem̓ č ʔexʷeʔ-t tᶿə čač ʔə tᶿə ləpat!

go 2sg.sbj give-tr det George obl det cup
‘Go give George the cup!’

The theme appears as an oblique-marked noun phrase. The applied objects in redirec-
tive applicatives are also direct objects and thus appear as direct arguments with no
preposition, for example ‘grandfather’ in the dative applicative in (33) and the benefac-
tive applicative in (34):

(33) niʔ ʔam-əs-t-əs tᶿə-nə sil̓ə ʔə kʷθə pukʷ.

aux give-dat-tr-3sbj det-1sg.poss grandparent obl det book
‘He gave my grandfather the book.’

(34) niʔ ʔiləq-əɫc-t-əs tᶿə-nə sil̓ə ʔə kʷθə snəxʷəł.

aux buy-ben-tr-3sbj det-1poss grandparent obl det canoe
‘He brought my grandfather a car.’

Redirective constructions are obligatory in the sense that there is no non-applicative

equivalent in which the theme occurs as an object and the applied object occurs as an
oblique noun phrase. So, for example, the recipient in (35a) cannot be expressed as an
oblique-marked noun phrase in a non-applicative construction, as in (35b):
88 Donna B. Gerdts

(35) a. nem̓ cən sam̓-əs-t łə słeniʔ ʔə θə-nə snəxʷəł.

aux 1sg.sbj sell-dat-tr det woman obl det-1sg.poss canoe
‘I’m going to sell my car to the woman.’
b. ✶nem̓ cən sam̓-ət θə-nə snəxʷəł ʔə łə słeniʔ.
aux 1sg.sbj sell-tr det-1sg.poss canoe/car obl det woman

Thus, Hul’q’umi’num’ can be considered a primary/secondary object language (Dryer

1986) or, more accurately, a direct object/oblique object language (Gerdts 2010a).
However, it is possible to separate two aspects of the event—the effect on the theme
and the transfer of possession or benefit—and express each as a separate predicate.
This can be accomplished by means of a serial verb construction, as in (36) and (37), or
conjoined clauses, as in (38).

(36) niʔ cən wəł sem̓-ət nem̓-əstəxʷ ʔə-ƛ̓ čan θə-nə swetə.
aux 1sg.sbj prf sell-tr go-caus obl-det John det-1sg.poss sweater
‘I sold John my sweater.’

(37) q̓ʷəl-ət cən ceʔ k̓ʷ sce:łtən x̌ʷteʔ ʔə-ƛ̓ nəwə.

cook-tr 1sg.sbj fut det salmon go.toward obl-det 2sg.pro
‘I will barbecue some salmon for you.’

(38) q̓ʷəl-ət cən ceʔ k̓ʷ sce:łtən ʔiʔ nił s-weʔ-stamə ceʔ.
cook-tr 1sg.sbj fut det salmon cnj 3pro nmz-own-caus.2sg.obj fut
‘I will cook some salmon and it will be for you.’

Applied objects in all four applicative constructions can be expressed by object suffixes,
an additional reason for positing that they are direct objects.

(39) niʔ q̓əl̓-meʔ-θam̓š-əs kʷθə xʷələnitəm̓.

aux believe-rel-tr.1sg.obj-3sbj det White.man(pl)
‘The White men believed me.’

(40) ʔi ʔew̓ə-nəs-am̓š-əs łə słeniʔ.

aux come-dir-1sg.obj-3sbj det woman
‘The woman comes to me.’ (Gerdts 1988b: 141)

(41) niʔ ʔam-əs-θam̓š-əs łə słeniʔ ʔə kʷθə pukʷ.

aux give-dat-tr.1sg.obj-3sbj det woman obl det book
‘The woman gave me the book.’
 4 Hul’q’umi’num’ Salish applicative constructions 89

(42) niʔ θəy-əłc-θam̓š-əs ʔə kʷθə-nə snəxʷəł.

aux fix-ben-tr.1sg.obj-3sbj obl det-1sg.poss canoe
‘He fixed my canoe for me.’

Only subjects and objects are indexed in the verb complex. The theme noun phrase in
a redirective applicative appears as an oblique-marked noun phrase, as in the dative
applicative in (43a) and the benefactive applicative in (44a); omitting the oblique prep-
osition results in ungrammaticality, as in (43b) and (44b).

(43) a. niʔ ʔam-əs-t-əs kʷθə swiw̓ləs ʔə kʷθə pukʷ.

aux give-dat-tr-3sbj det boy obl det book
‘He gave the boy the book.’ (Gerdts 1988b: 101)
b. ✶niʔ ʔam-əs-t-əs kʷθə swiw̓ləs kʷθə pukʷ.
aux give-dat-tr-3sbj det boy det book
Intended: ‘He gave the boy the book.’

(44) a. niʔ x̌əl̓-əłc-ət-əs kʷθən̓ men ʔə kʷθə pipə-s.

aux write-ben-tr-3sbj det.2poss father obl det letter-3poss
‘He wrote the letter to/for your father.’ (Gerdts 1988b: 101)
b. ✶niʔ x̌əl̓-əłc-ət-əs kʷθən̓ men kʷθə pipə-s.
aux write-ben-tr-3sbj det.2poss father det letter-3poss
Intended: ‘He wrote the letter to/for your father.’

4.2 Relativization and similar operations

As detailed in Gerdts (1988b), noun phrases appear in pre-verbal position in a variety

of extraction constructions in Hul’q’umi’num’, including relative clauses, wh-questions,
and clefts (it-clefts, NP-clefts, and wh-clefts). The extracted noun phrase appears before
the clause that it is extracted from, as seen by comparing the monotransitive clause in
(45a) with its cleft counterpart in (45b):

(45) a. niʔ č lem-ət kʷθə swəy̓qeʔ.

aux 2sg.sbj look.at-tr det man
‘You looked at the man.’
b. nił kʷθə swəy̓qeʔ [niʔ lem-ət-əxʷ].
3pro det man aux look.at-tr-2sg.sbj
‘It’s the man that you looked at.’

The clause marked in square brackets in (45b) is a dependent clause, as seen by subject
indexing; first- and second-person subject markers appear as second-position clitics in
main clauses (45a) but as verbal suffixes in dependent clauses (45b). Applied objects
90 Donna B. Gerdts

in both relational and redirective constructions are similarly extracted: (46b) shows
extraction of an applied object in a directional applicative, (47b) in a relational applica-
tive, (48b) in a dative applicative, and (49b) in a benefactive applicative.

(46) a. niʔ nəm-nəs-əs kʷθə swiw̓ləs kʷθə swəy̓qeʔ.

aux go-dir-3sbj det boy det man
‘The boy went up to the man.’
b. łwet k̓ʷə niʔ nəm-nəs-əs kʷθə swiw̓ləs?
who det aux go-dir-3sbj det boy
‘Who did the boy go up to?’

(47) a. niʔ č q̓el̓-meʔ-t kʷθə ləplit.

aux 2sg.sbj believe-rel-tr det priest
‘You believed the priest.’
b. łwet niʔ q̓el̓-meʔ-t-əxʷ?
who aux believe-rel-tr-2sg.sbj
‘Who did you believe?’

(48) (Gerdts 1988b: 101, 103)

a. niʔ ʔam-əs-t-əs kʷθə swiw̓ləs ʔə kʷθə pukʷ.
aux give-dat-tr-3sbj det boy obl det book
‘He gave the boy the book.’
b. nił kʷθə swiw̓ləs niʔ ʔam-əs-t-əs ʔə kʷθə pukʷ.
3pro det book aux give-dat-tr-3sbj obl det book
‘It’s a boy that he gave the book to.’

(49) a. niʔ č qʷəl-əłc-t tᶿən̓ sil̓ə ʔə kʷθə səplil.

aux 2sg.sbj cook-ben-tr det.2poss grandparent obl det bread
‘You baked bread for your grandfather.’
b. łwet k̓ʷə niʔ qʷəl-əłc-t-əxʷ ʔə kʷθə səplil?
who det aux cook-ben-tr-2sg.sbj obl det bread
‘Who did you bake the bread for?’

In some languages of the world, the applicative construction serves to allow an NP

access to rules such as extraction if direct objects but not obliques undergo such rules.
However, In Hul’q’umi’num’ a secondary strategy of extraction via nominalization is
available for noun phrases that are not core arguments. Extractions are constructed
in different ways, depending upon the noun phrase’s grammatical relation in the
corresponding basic clause (Gerdts 1988b). As we have seen above, direct objects are
extracted with no special morphology. However, oblique noun phrases, as in (50a), are
extracted through a process of nominalization with the oblique nominalizer š(xʷ)-, as
in (50b).
 4 Hul’q’umi’num’ Salish applicative constructions 91

(50) a. niʔ č q̓el̓ ʔə kʷθə s-qʷaqʷəl̓-s kʷθə ləplit.

aux 2sg.sbj believe obl det nmz-talk(ipfv)-3poss det priest
‘You believed the priest’s words.’
b. stem k̓ʷə niʔ ʔən̓-š-q̓el̓?
what det aux 2poss-obl.nmz-believe
‘What did you believe?’

The subject of the relative clause is expressed as a possessor, e.g. ʔən̓ ‘your’.
Extraction reveals a difference between oblique noun phrases and the oblique-
marked noun phrase in redirective applicatives. A “true” oblique is extracted via nom-
inalization with the prefix š(xʷ)-, as noted above, but when the theme in a redirective
applicative is extracted, the predicate is nominalized with the prefix s-: the theme in
(48a) is extracted as in (51).7

(51) nił kʷθə pukʷ niʔ s-ʔam-əs-t-s kʷθə swiw̓ləs.

3pro det book aux nmz-give-dat-tr-3poss det boy
‘It’s a book that he gave the boy.’ (Gerdts 1988b: 103)

Gerdts (1988b) uses this fact as evidence against an analysis for Hul’q’umi’num’ ditran-
sitive constructions that would paraphrase an example like (48a) as “he gifted the boy
with the book” since “book” does not extract like a true oblique.8

4.3 Combinatory properties of applicatives

This section discusses combinations of applicatives with other voice/valence construc-

tions, such as passives, antipassives, reflexives, reciprocals, causatives, and lexical suffix-
ation, all of which are known to relate to the concept of direct object in Hul’q’umi’num’.9
All of these constructions are formed with suffixes in Hul’q’umi’num’ and, as previously
discussed in Gerdts (1988b, 2004), the ordering of Hul’q’umi’num’ morphology straight-
forwardly reflects the syntactic structure of a complex predicate. If construction A
serves as a base for construction B, then the morphology associated with A will appear

7 Gerdts (1988b, 2010c) shows other oblique objects [NPs that are semantically the patient of a transitive
event but grammatically an oblique-marked NP in an intransitive construction] also undergo extraction
in the s-nominalization construction. These include patients in antipassives, free-standing NPs doubling
a lexical suffix or denominal verb, and cognate objects.
8 Other evidence against this analysis presented in Gerdts (1988b) comes from the fact that the applied
object has some but not all of the properties of the objects of mono-transitive clauses. For example, ap-
plied objects do not undergo antipassive.
9 Hul’q’umi’num’, unlike some of the other Salish languages, does not allow multiple applicatives (Ki-
yosawa and Gerdts 2010b: Ch. 7).
92 Donna B. Gerdts

closer to the root than the morphology associated with B (as we will see in examples
below). Research reveals four types of patterns of ordering of morphology with respect
to applicative suffixes: (i) some voice/valence morphology can only appear before applic-
atives, (ii) some can only appear after applicatives, (iii) some can appear either before
or after, and (iv) some do not co-occur with applicatives at all. Sometimes relational and
redirective applicatives behave differently with respect to allowable combinations of
constructions.

4.3.1 Passives

Passives in Hul’q’umi’num’ differ from their active counterparts in several ways. In

a passive, for example (52b), the agent, if it appears, is expressed as an oblique noun
phrase:

(52) a. niʔ c̓ew-ət-əs θə słeniʔ tᶿə swəy̓qeʔ.

aux help-tr-3sbj det woman det man
‘The woman helped the man.’
b. niʔ c̓ew-ət-əm tᶿə swəy̓qeʔ ʔə θə słeniʔ.
aux help-tr-pass det man obl det woman
‘The man was helped by the woman.’

Because passives are intransitive, they do not take ergative agreement. Instead, the
verb in a passive adds intransitive morphology, labeled pass, to the transitive suffix;
in main clauses this is the suffix -əm, which is historically related to the middle suffix
(Gerdts and Hukari 2006b). First- or second-person subjects in passives are indexed by
a set of special passive suffixes that are historically related to the object suffixes (Gerdts
1989), as can be seen by comparing an active clause with a second-person plural object
to its passive counterpart:

(53) a. c̓ew-ətalə ct ceʔ.

help-tr.2pl.obj 1pl.sbj fut
‘We will help you (pl).’
b. c̓ew-ətaləm ceʔ.
help-tr.1/2pl.pass fut
‘You (pl) will be helped.’ [also ‘We will be helped.’]

Thus, indexing for the sole argument in a passive is a portmanteau morpheme

combining the general transitive suffix -t, a person suffix, and the passive suffix. This
yields a paradigm such as that for the verb ‘kill’:
 4 Hul’q’umi’num’ Salish applicative constructions 93

(54) q̓ay-θeləm ‘I was killed’

q̓ay-θa:m ‘you were killed’
q̓ay-taləm ‘we were killed’
q̓ay-taləm ‘you (pl) were killed’
q̓ay-təm ‘he/she/it/they were killed’

All four applicative constructions have passive counterparts:

(55) niʔ siʔsiʔ-meʔ-θeləm ʔə-ƛ̓ čan.

aux frighten-rel-tr.1sg.pass obl-det John
‘John was frightened of me.’
(Lit. ‘I am the source of John’s being frightened.’)

(56) niʔ nəm-nəs-əm kʷθə speʔəθ.

aux go-dir-pass det bear
‘They approached the bear.’ (Lit. ‘The bear was gone up to.’)

(57) ʔi ʔam̓-əs-t-əm tᶿə John ʔə-ƛ̓ meli. ʔə kʷθə šcki:ks.

aux give-dat-tr-pass det John obl-det Mary obl det vanilla.extract
‘John is being given vanilla extract by Mary.’ (Gerdts 1988b: 233)

(58) niʔ θəy-əłc-θeləm ʔə θə-nə snəxʷəł.

aux fix-ben-tr.1sg.pass obl det-1sg.poss canoe
‘Someone fixed my canoe for me.’
(Lit. ‘I was fixed my canoe for.’)

All show a full range of person inflection in the passive, e.g. the following paradigm for
benefactive applicatives.

(59) θəy-əłc-θeləm ‘I was fixed something for’

θəy-əłc-θa:m ‘you were fixed something for’
θəy-əłc-taləm ‘we were fixed something for’
θəy-əłc-taləm ‘you (pl) were fixed something for’
θəy-əłc-təm ‘he/she/it/they were fixed something for’

Hul’q’umi’num’ has a complex set of restrictions on when to use active versus passive
clauses (Gerdts 1988b; Gerdts and Hukari 2008), and these pertain to applicatives as
well. For example, when the agent is a proper noun, a passive rather than an active
clause is used (see [57] above). There is a ban on the combination of third-person subject
and second-person object, and passive is often used as a repair strategy.
94 Donna B. Gerdts

(60) niʔ ʔə č qəl-stəxʷ θə kəpu niʔ s-ʔiləq-əłc-θamət.

aux q 2sg.sbj bad-caus dt coat aux nmz-buy-ben-2sg.spass
‘Do you dislike the coat that was bought for you?’

Since passive morphology is conflated with person inflection, it is not surprising that it
is restricted to the very end of the verb complex, as derivational morphology appears
closer to the root than passive morphology. Passives cannot be further derived into
transitives by means of applicatives or causatives. Therefore, the pattern of combina-
tion is that applicatives can serve as a base for passives but not vice versa.

4.3.2 Antipassives

Most monotransitive clauses have antipassive counterparts (Gerdts and Hukari 2005):

(61) a. niʔ q̓ʷəl-ət-əs tᶿə sce:łtən.

aux cook-tr-3sbj det salmon
‘He cooked the salmon.’
b. niʔ q̓ʷəl-əm ʔə tᶿə sce:łtən.
aux cook-mid obl det salmon
‘He cooked the salmon.’

(62) a. naʔət qʷəs-t-əs tᶿə ƛ̓ełəm̓ sce:łtən.

aux go.in.water-tr-3sbj det salted salmon
‘She put the salted fish in water.’
b. naʔət qʷs-els ʔə tᶿə ƛ̓ełəm̓ sce:łtən.
aux go.in.water-act obl det salted salmon
‘She soaked the salted fish.’

Antipassives are formed with the middle suffix -əm (61b), or the activity suffix -els
(62b). The patient in the antipassive is expressed as an oblique object. A wide variety
of transitive verbs have antipassive counterparts. However, as Gerdts (1988b) notes,
applicative verbs do not form antipassives, as seen in the relational applicative in (63b)
and the benefactive applicative in (64b).

(63) a. niʔ cən q̓el̓-meʔ-t kʷθə ləplit.

aux 1sg.sbj believe-rel-tr det priest
‘I believed the priest.’
b. ✶niʔ cən q̓el̓-meʔ-əm/əls ʔə kʷθə ləplit.
aux 1sg.sbj believe-rel-mid/act obl det priest
Intended: ‘I believed the priest.’
 4 Hul’q’umi’num’ Salish applicative constructions 95

(64) a. nem̓ ʔə č θəy-əłc-t kʷθə-nə mən̓ə ʔə kʷθə

go q 2sg.sbj fix-ben-tr det-1sg.poss child obl det
snəxʷəł-s?
canoe-3poss
‘Are you going to fix his canoe for your son?’
b. ✶nem̓ ʔə č θəy-əłc-əm ʔə kʷθə-nə mən̓ə ʔə kʷθə
go q 2sg.sbj fix-ben-mid obl det-1sg.poss child obl det
snəxʷəł-s?
canoe-3poss
Intended: ‘Are you going to fix his canoe for your son?’

This follows from a general restriction in Hul’q’umi’num’ that antipassives are not
formed on derived transitive verbs, so, for example, causative constructions also do
not form antipassives. Antipassive is thus one construction that distinguishes between
objects in simple transitive clauses, which can be oblique objects in antipassives, and
applied objects, which cannot.

4.3.3 Reflexives and reciprocals

Hul’q’umi’num’ reflexives and reciprocals are formed by suffixing the reflexive -θət
or the reciprocal -təl to a wide variety of both intransitive and transitive verbs (Gerdts
2000). Table 2 gives some examples of reflexives and reciprocals formed on transitive
verb roots.

Table 2: Reflexives and reciprocals.

transitive verb reflexive reciprocal

a. q̓ ʷaqʷət q̓ ʷaqʷəθət q̓ ʷaqʷətəl
‘club it’ ‘club self’ ‘club each other’
b. ̓
ʔakʷət ̓
ʔakʷəθət ̓
ʔakʷtəl
‘hook it’ ‘hook self’ ‘get hung up with each other’
c. c̓ ewət c̓ awəθət c̓ awətəl
‘help him/her’ ‘help self’ ‘help each other’
d. x̌iq̓ ət x̌iq̓ əθət x̌iq̓ ətəl
‘scratch him/her’ ‘scratch self’ ‘scratch each other’

Both reflexives, e.g. (65) and (66), and reciprocal, e.g. (67) and (68), can be formed on
relational applicatives.
96 Donna B. Gerdts

(65) niʔ cən siʔsiʔ-meʔ-θət ʔə kʷθə nə qiʔx̌əneʔtən niʔ ʔə

aux 1sg.sbj frighten-rel-refl obl det 1sg.poss reflection aux obl
kʷθə šk̓ʷcastən̓.
det mirror
‘I frightened myself with my reflection in the mirror.’

(66) ʔi cən wəł łciws-maʔ-θət10 k̓ʷə-nə-s ʔi q̓aq̓iʔ.

aux 1sg.sbj already tired-rel-refl det-1poss-nmz aux sick
‘I’m tired of myself being sick.’ (Gerdts and Kiyosawa 2005: 336)

(67) ʔeʔət x̌i:ʔx̌eʔ-meʔ-təl̓ tᶿə sƛ̓əłiqəł kʷ-s qʷəl̓qʷəl̓-təl̓-s.

aux shy(ipfv)-rel-rec det children det-nmz speak(ipfv)-rec-3poss
‘The children are shy about speaking to each other.’

(68) ʔi yə-hən̓əm̓-nəs-təl̓ tᶿə sqʷəmqʷəmey̓.

aux dyn-go(ipfv)-dir-recp det dog(pl)
‘The dogs are going up to each other.’

In contrast, Gerdts (1988b) claims that the redirective suffixes -as and -əłc in Hul’q’umi’num’
cannot be followed by the reflexive suffix:

(69) ✶
niʔ cən ʔam-əs-θət.
aux 1sg.sbj give-dat-refl
Intended: ‘I gave it to myself.’ (Gerdts 1988b: 113)

(70) ✶
niʔ q̓ʷəl-əłc-θət ʔə kʷθə səplil.
aux cook-ben-refl obl det bread
Intended: ‘He baked the bread for himself.’ (Gerdts 1988b: 113)

Syntactically, there is no reason to expect that reflexive forms of redirective applicatives

should not be possible, 11 though Gerdts (2010a) shows that non-applicative ditransitive
constructions also disallow reflexives: for example, ʔexʷeʔt ‘give it to him/her’ does not
form a reflexive ✶ʔexʷeʔθət ‘give it to oneself’ and cset ‘tell him or her to do something’
does not form the reflexive cəsəθət ‘tell oneself to do something’. Semantically it is
somewhat awkward to direct an action toward the self using a construction that
purposely directs the action to another person.12

10 The vowel e in the relational suffix -meʔ changes to a before the reflexive suffix (Gerdts and Hinkson
2004).
11 See Kiyosawa and Gerdts (2010b: Ch. 7) for examples of reflexives formed on applicatives in other
Salish languages.
12 See the periphrastic construction in (38) above that can be used to express that meaning.
 4 Hul’q’umi’num’ Salish applicative constructions 97

In contrast, reciprocal forms of ditransitives were easier for speakers to construct, as

the meaning can be construed as an outwardly directed action doing something to
“each other”, and likewise reciprocal forms of redirective applicatives are also possible:

(71) niʔ ct nəwən-təl ʔiʔ θə-nə sqeʔəq ʔə kʷθə leləm̓ ct.

aux 1pl.sbj will-recp cnj det-1sg.poss sister obl det house 1pl.poss
‘My little sister and I willed each other our house.’

(72) ʔam-əs-tal
give-dat-recp
‘give it to each other’ (Gerdts 2000: 146)

(73) niʔ ct q̓ʷəl-əłc-təl.

aux 1pl cook-ben-recp
‘We cooked for each other.’ (Gerdts 2000: 146)

We see then that both reflexives and reciprocals can follow relational applicatives but
only reciprocals can follow redirective applicatives in Hul’q’umi’num’.13
Research on the opposite order has uncovered another asymmetry between applica-
tive types. Reflexives and reciprocals are detransitivizing constructions and thus they
are suitable as bases for relational but not redirective applicatives. One use of reflexive
morphology is on manner-of-motion verbs, e.g. qix̌ ‘slide, slip’ as a non-agentive action
versus qix̌əθət ‘slide (as in sledding)’ as a verb of controlled motion. Such reflexives
allow directional applicatives.

(74) nem̓ qix̌-əθət ʔə kʷθə sθimaʔ.

go slide-refl obl det ice
‘Go slide on the ice.’

(75) nem̓ č p̓eʔ qiqəx̌-əθət-nəs tᶿən̓ men.

go 2sg.sbj cert slide-refl-dir det.2poss father
‘Go and skate/slide to your father.’

The reciprocal suffix can be added to intransitive verbs to express the meaning that the
action was done “together”, and in the following example we see a motion verb suffixed
with the reciprocal followed by a directional applicative suffix.

13 The difference between the range of occurrence between reciprocals and reflexives is not unexpect-
ed from a cross-linguistic viewpoint. For example, in English, reciprocal pronouns, but not reflexive pro-
nouns, can function as possessives: compare ‘they looked at each other’s pictures’ with the unacceptable
‘✶he looked at himself’s picture’.
98 Donna B. Gerdts

(76) nem̓ ct ceʔ p̓eʔ ʔəw̓ šaqʷəl ʔəšəl-təl-nəs-amə.

go 1pl.sbj fut cert lnk go.across paddle-recp-dir-2sg.obj
‘We will all paddle across together toward you.’

To summarize, the two types of applicatives have different combinatory properties with
respect to reflexives and reciprocals. Only relational applicatives can follow reflexives and
reciprocals, and only relationals serve as bases for reflexives. Both relational and redirec-
tive applicatives serve as bases for reciprocals. The lack of reflexives built on redirective
applicatives is one difference between mono-transitive and ditransitive constructions.

4.3.4 Causatives

Hul’q’umi’num’ causatives (Gerdts 1988b; Gerdts and Hukari 2006a) are formed with
the suffix -stəxʷ. When the base is an active intransitive verb (77a), the causative (77b)
forms a transitive clause in which the causer is the subject and the causee is the direct
object, and when the base is a transitive verb (78a), the causative (78b) forms a ditransi-
tive clause in which the causee is the direct object, and the object in the corresponding
transitive is an oblique object.

(77) a. niʔ ʔəłtən θə qeq

aux eat det baby
‘The baby ate.’
b. niʔ cən ʔəłtən-stəxʷ θə qeq
aux 1sg.sbj eat-caus det baby
‘I fed the baby.’

(78) a. niʔ ʔiləq-ət-əs tᶿə sƛ̓iʔƛ̓qəł k̓ʷ sk̓ʷaw̓əs.

aux buy-tr-3sbj det child det bucket
‘The boy bought a bucket.’
b. ʔiləq-stəxʷ č tᶿə sƛ̓iʔƛ̓qəł ʔə k̓ʷ sk̓ʷaw̓əs.
buy-caus 2sg.sbj det child obl det bucket
‘Have the boy buy a bucket.’

Relational applicatives cannot be formed on causatives, as causatives are transitive

constructions, and relational applicatives are formed only on intransitive bases. In con-
trast, benefactive applicatives, which are formed on transitive bases, can be formed on
causatives.

(79) niʔ cən ʔəłtən-əst-əłc-ət ʔə θə qeq.

aux 1sg.sbj feed-caus-ben-tr obl det baby
‘I fed the baby for her.’
 4 Hul’q’umi’num’ Salish applicative constructions 99

(80) nem̓ ʔənəxʷ-st-əłc-θam̓š ʔə θə sti:č!

go stop-caus-ben-tr.1sg.obj obl det bus
‘Stop the bus for me!’

(81) nem̓ xʷəʔaləm̓-st-əłc-ət ʔə θə ʔe:y̓x̌al̓ł.

go return-caus-ben-tr obl det crab.young(dim)
‘Bring the little crabs back for him.’

Next, we consider the possibility of a causative being formed on an applicative construction.

Speakers easily constructed examples of directional applicatives followed by causatives:

(82) nəm̓-nəs-stəxʷ tᶿən̓ sil̓ə ʔə tᶿə ʔi sq̓əq̓ip

go-dir-caus det.2poss grandparent obl det aux gathered
qʷəli:lqʷəl̓-təl̓.
talk(pl.ipfv)-recp
‘Have your grandfather go to the people in discussion.’

(83) niʔ č ʔewə-nəs-stəxʷ tᶿə xʷəlməxʷ ʔi tecəl.

aux 2sg.sbj come.here-dir-caus det First.Nation.people aux arrive
‘Have the First Nation people that arrived come this way.’

Work with speakers did not reveal examples of causatives being formed on other
applicatives.14

4.3.5 Lexical suffixes

Hul’q’umi’num’ has over one hundred lexical suffixes, which are bound roots that have
meanings analogous to free-standing nominals expressing body parts, flora and fauna,
people, and cultural artifacts, such as houses, garments, and instruments. The lexical
suffix usually bears little resemblance to the free-standing noun of similar meaning.

Table 3: Some lexical suffixes.

Noun Meaning Lexical suffix Meaning

sʔaθəs ‘face’ -as ‘face’, ‘round object’
qələm̓ ‘eye’ -aləs ‘eye’, ‘loop’
θaθən ‘mouth’ -aθən ‘mouth’, ‘edge’
leləm̓ ‘house’ -ew̓txʷ ‘house’, ‘building’, ‘room’
qeq ‘baby’ -eyəł ‘baby’, ‘younger generation’

14 See Gerdts and Hukari (2006a) for more discussion of causatives formed on transitives.
100 Donna B. Gerdts

The syntax and semantics of lexical suffixes have been discussed elsewhere (Gerdts
2003, 2010c; Gerdts and Hinkson 1996; Hinkson 1999), but suffice it to say that the way
lexical suffixes stack with applicatives rests crucially on the type of lexical suffix con-
struction.
One type of lexical suffixes behaves as an adjunct to specify the instrument,
manner, or location of the verb; the suffix attaches to an intransitive base and yields an
intransitive verb.

(84) q̓t-aθən
go.along-mouth
‘walk along (a shore, etc.)’ (Gerdts 2003: 346)

Many cognitive/psychological predicates are formed with an adjective or intransitive

verb and a lexical suffix, which embodies the experience, and such forms can be tran-
sitivized with the relational applicative suffix -meʔ. Examples are provided in Table 4:

Table 4: Lexical suffixes preceding relational applicative.

Base Parse Base + -meʔ meaning

qil-əs bad-face ‘sad’ qiləsmeʔt ‘sad for him/her/it’
xʷ-θt-iwən loc-say-inside ‘think’ xʷθtiwənmeʔt ‘think, decide about him/her/it’
xʷ-qʷəl-əwən loc-talk-inside ‘think’ xʷqʷələwənmeʔt ‘think about him/her/it’
łc-iws cut-body ‘tired’ łciwsmeʔt ‘tired of him/her/it’

Another use of lexical suffixes is as a classifier relating to the direct object (whether
or not the object is actually expressed). So, for example, the suffix refers to qeq ‘baby’
in (85).

(85) nem̓ cən šk̓ʷ-əyəł-t tᶿə-nə qeq.

go 1sg.sbj bathe-child-tr det-1poss baby
‘I’m going to bathe my baby.’

Transitive clauses like (85) can serve as the base for benefactive applicatives, in which
the applied object is the direct object and the theme corresponding to the direct object
in the base form is expressed as an oblique object.

(86) šk̓ʷ-əyəł-əłc-θam̓š ʔə tᶿə-nə qeq.

bathe-child-ben-tr.1sg.obj obl det-1sg.poss baby
‘Bathe my baby for me.’
 4 Hul’q’umi’num’ Salish applicative constructions 101

Table 5 provides some additional examples.

Table 5: Lexical suffixes followed by applicatives.

Base Parse Base + -meʔ Meaning

θəy-eʔł-t fix-fabric-tr θəy-eʔł-əłc -t ‘make a bed for someone’
‘make a bed’
łəlq-ət̓ᶿeʔ-t wash-fibre-tr łəlq-ət̓ᶿeʔ-əłc -t ‘wash wool for someone’
‘wash wool’
xʷ-kʷaʔ-qə-t loc-open-container-tr xʷ-kʷaʔ-qə-əłc-t ‘open the container for someone’
‘open the container’
xʷ-t̓ᶿəx̌-wil-t loc-wash-vessel-tr xʷ-t̓ᶿəx̌-wil-əłc-t ‘washes dishes for someone
‘wash the dishes’

As with other cases of stacking with applicatives, relational and redirective applicatives
behave differently, as they have different conditions on transitivity. Lexical suffix con-
structions that are intransitive serve as bases for relational applicatives and ones that
are transitive serve as bases for redirective applicatives.
We have also found examples in which the lexical suffixes -ənəq ‘people’ and –eył ~
eyəł ‘child/children’ appear after the benefactive suffix -əłc. Compare the applicatives in
(87) and (88)—the latter uses the lexical suffix for people to refer to the applied object.

(87) nem̓ ʔa:ł-əłc-ət tᶿən̓ sil̓ə ʔə tᶿə x̌θəm.

go load-ben-tr dt.2poss grandparent obl det box
‘Go and load the box for your grandfather.’

(88) ʔi tecəl kʷθə swaw̓ləs ʔa:ł-əłc-ənəq ʔə kʷθə ləpla:š.

aux arrive det young.man.pl load-ben-people obl det board
‘The young men arrived who will load the lumber on the community’s behalf.’

This lexical suffix always detransitivizes the clause and thus obviates the need for
transitive marking.15 Additional examples of the benefactive suffix followed by human
lexical suffixes follow:

15 As discussed in Gerdts (2003, 2010c) lexical suffix constructions can be shifted from transitive to
intransitive simply by deleting the transitive morphology. The construction with -ənəq is unique in that
it does not have a transitive counterpart.
102 Donna B. Gerdts

(89) qəx̌ kʷθə səw̓al̓əm̓ ʔi nə-s-ʔiləq-əłc-eył.

much det toy aux 1sg.poss-nmz-buy-ben-child
‘I bought a lot of toys for the children.’

(90) nem̓ č θəy-əɫc-eyɫ ʔə-k̓ʷ šxʷʔiʔtət-s.

go 2sg.sbj make-ben-child obl-det bed-3poss
‘Go make up beds for the children.’

(91) nił ceʔ tᶿə yey̓səl̓ə p̓et̓ᶿ-əłc-ənəq ʔə-k̓ʷ ləx̌ʷtən-s tᶿə

3pro fut det two.people sew-ben-people obl-det blanket-3poss det
məstiməxʷ.
people
‘These two people will be the ones to sew their blankets for the people.’

(92) yey̓səl̓ə ceʔ k̓ʷə peθ-əłc-ənəq ʔə θə ləx̌ʷtən.

two.people fut det spread-ben-people obl det blanket
‘Two people will spread the blanket for the people.’

Instructions of this sort are often heard during longhouse ceremonies where the
ceremonial speakers are directing the collective work being done on behalf of a family.

4.3.6 Summary of combinations

I first summarize the combinations where the applicatives precede other construc-
tions. We find applicatives combine with passives but not antipassives. Applicatives
can form reciprocals, but only relational applicatives form reflexives. We see then that
there is one difference between objects in monotransitive clauses and applied objects:
the former but not the latter can be antipassivized. We also see that applicative con-
structions differ as to their allowable combinations: only directional applicatives form
causatives and only benefactive applicatives are known to be followed by person lexical
suffixes.
Next, to summarize examples where the applicatives follow other constructions,
we find that the allowable combinations are predictable according to the type of
applicative. Relational applicatives are formed on intransitive bases and thus they can
combine with reflexives and reciprocals, which are intransitive in Hul’q’umi’num’, but
not with causatives, which are transitive constructions. In contrast, redirective applica-
tives are formed on transitive bases and thus they can combine with causatives, but not
with reflexives or reciprocals. In the case of lexical suffixes, Hul’q’umi’num’ has both
 4 Hul’q’umi’num’ Salish applicative constructions 103

intransitive and transitive lexical suffix constructions, and the former combine with
relational applicatives and the latter combine with benefactive applicatives.

5 Semantics
5.1 Meanings associated with each applicative suffix

The relational suffix -meʔ appears on a wide variety of verbs; relational applicatives
are used when the applied object is the stimulus of a psychological predicate (the most
common use), the source of a verb of motion, the goal of a speech act, the sufferer of an
adversative, or the beneficiary of an intransitive verb.

(93) -meʔ general relational applicative

a. stimulus of psychological or cognitive predicate
łciws ‘tired’ łciws-meʔ-t ‘tired of him/her’
̓qel̓ ‘believe’ q̓el̓-meʔ-t ‘believe him/her’
siʔsiʔ ‘be afraid’ siʔsiʔ-meʔ-t ‘afraid of him/her/it’
x̌iʔx̌eʔ ‘ashamed’ x̌iʔx̌eʔ-meʔ-t ‘ashamed of him/her’
siwəl ‘sense’ siwəl-meʔ-t ‘sense him/her/it’
b. source of verb of motion
łəw̓ ‘run away’ łəw̓-mə-t ‘run away from him/her’
kʷe:l ‘hide’ kʷe:l-meʔ-t ‘hide from him/her’
c. goal of speech or expressive act
xʷəyxʷəyasəm ‘brag’ xʷəyxʷəyas-meʔ-t ‘bragging to him/her’
x̌e:m ‘cry’ x̌e:x̌əm̓-mə-t ‘crying over him/her’
qʷal ‘speak’ qʷəl-mə-t ‘lecture to, bawl out him/her’
d. adversative (often in passive)16
θeʔc ‘get dark’ θeʔc-meʔ-t ‘get dark on him/her’
łəməxʷ ‘rain’ θəməxʷ-meʔ-t-əm ‘(he/she/it) get rained on’
yəq̓ ‘snow’ yəq̓-meʔ-t-əm ‘(he/she/it) get snowed on’
sqʷəlqʷalx̌ʷ ‘hail’ sqʷəlqʷalx̌ʷ-meʔ-t-əm ‘(he/she/it) get hailed on’
e. beneficiary of intransitive verb
kʷukʷ ‘cook’ kʷukʷ-meʔ-t ‘cook for him/her’
ya:ys ‘work’ ya:ys-meʔ-t ‘work for him/her’17

16 See Gerdts (2012) and Kiyosawa and Gerdts (2010a) for discussion of adversatives in Hul’q’umi’num’.
17 This verb also means ‘to work on’ a person in a spiritual sense.
104 Donna B. Gerdts

The directional -nəs appears on a wide variety of motion verbs:

(94) nem̓ ‘go’ nəmnəs18 ‘go toward him/her/it/them’

ʔewə ‘come’ ʔewənəs ‘come toward him/her/it/them’
x̌ʷčenəm ‘run’ x̌ʷčenəmnəs ‘run toward him/her/it/them’
xʷəniʔ ‘get there’ xʷənin̓s ‘get there to him/her/it/them’

The dative applicative suffix -as appears in only a half dozen verb forms:

(95) (Gerdts and Hinkson 2004)

ʔeʔəm ‘give’ ʔam-əs-t ‘give it to him/her’
̓
sem-ət ‘sell it’ sam̓-əs-t ‘sell it to him/her’
xʷayəm ‘sell’ xʷayəm-əs-t ‘sell it to him/her’
√ʔiw̓ ‘show’ ʔiw̓-əs-t ‘show it to him/her’
√yəθ ‘tell’ yəθ-əs-t ‘tell him/her about it’

Gerdts and Hinkson (1996, 2004) claim that the dative applicative suffix is grammatical-
ized from the lexical suffix ‘face’.19 In many examples, this suffix has a concrete body
part meaning, e.g. š-t̕ᶿx̌ʷ-as ‘washed face’ (√t̕ᶿx̌ʷ ‘wash’), xʷ-łaq̓ʷ-əs-t ‘slap him/her on the
face’, xʷ-p̓aƛ̓-əs-t ‘feel his/her face’ (p̓eƛ̓ ‘feel’). It extends semantically to various loca-
tional and directional meanings, e.g. nəʔ-as ‘facing away’ (niʔ ‘be there’), qəl̓-as ‘back-
wards’ (qəl ‘bad’), qp̓-as-t, ‘turn it upside down’ (√qp̓ə ‘down’). There are also exam-
ples of metonymy where the lexical suffix -as face refers to the entire person or entity:
k̓ʷł-as-t ‘throw liquid on him’, ƛ̓əʔ-as-t ‘go pick him/her up and bring back’, xʷ-θq̓ʷ-as-t
‘meet, to go towards’. These two extended uses of lexical suffixes set the stage for the
further development of the lexical suffix into the dative applicative morpheme, which
adds to the verbal semantics the meaning that an action is directed toward a person.
The verbs in dative applicatives include verbs of transaction ‘give’ and ‘sell’, a verb of
perception ‘show’ (96), and the speech act verb ‘tell’ (97).

(96) niʔ ʔiw̓-əs-θam̓š-əs ʔə kʷθə qeq-s.

aux show-dat-tr.1sg.obj-3sbj obl det baby-3poss
‘She showed me her baby.’ (Gerdts and Hinkson 2004: 66)

18 When suffixed with the directional applicative suffix, the verb nem̓ ‘go’ frequently shows vowel
reduction. Also, some speakers lose the glottalization of the final m altogether, or they restructure it as
an intervocalic glottal stop: nəʔəmnəs.
19 Gerdts and Hinkson (2004) note that forms for ‘face’ have developed into grammatical markers else-
where in the world. For example, in Chalcatongo Mixtec (Brugman 1983, Macaulay 1996) ‘face’ is used
as a locative or dative preposition and in Ayoquesco Zapotec (MacLaury 1989) it is used as a dative
preposition with verbs of speaking. Hollenbach (1995) discusses the extensions of ‘face’ in nine Mixtec
dialects as well as Trique and Cuicatec.
 4 Hul’q’umi’num’ Salish applicative constructions 105

(97) niʔ yəθ-əs-t-əs łə Mary ʔə kʷθən̓ sya:ys.

aux tell-dat-tr-3sbj det Mary obl det.2poss work
‘He told Mary about your job.’ (Gerdts 1988b: 92)

In comparison, benefactive -əłc is productively added to a wide variety of appropriate

verbs (see Table 5 above for some examples).20 Benefactive applicatives are translated
with the prototypical benefactive meaning of doing something for someone’s benefit,
with the exception of one verb x̌əl̓ət ‘write’, which as an applicative can be translated as
‘write to’ or ‘write for’ (98):

(98) niʔ x̌əl̓-əłc-ət-əs kʷθən̓ men ʔə kʷθə pipə-s.

aux write-ben-tr-3sbj det.2poss father obl det letter-3poss
‘He wrote the letter to/for your father.’ (Gerdts 1988b: 101)

In addition, as Kiyosawa and Gerdts (2010a) note, an example such as the following can
also be used in the sense of delegation (99):

(99) q̓ʷəl-əłc-θamə cən ceʔ ʔə k̓ʷ sce:łtən.

cook-ben-tr.2sg.obj 1sg.sbj fut obl det salmon
‘I will bake some salmon for you.’

Our colleague the late Dr. Ruby Peter explained, “You can use this for your benefit in
whatever way: for you to eat, because you are unable to do it for whatever reason,
because you are too busy to do it and it needs to be done, because I am being substituted
to do your job, and so on.” The precise meaning is determined by the context. However,
the most normal or neutral reading would be that the salmon is being cooked for the
referent of the object to eat themselves rather than for the salmon to be cooked to give
it to someone else to eat.

5.2 Applicatives in discourse context

In the case of redirective applicatives, there are no corresponding basic constructions

that are regularly used, so the purpose of the applicative is to allow the expression of
the theme nominal (the oblique object) and the recipient/beneficiary (direct object) in
a ditransitive clause. In the case of relational applicatives, there is always an equiva-
lent intransitive clause in which the noun phrase corresponding to the applied object

20 The Hul’q’umi’num’ dictionary (Hukari and Peter 1995) lists 55 examples of words with benefactive
applicatives, many with sentential examples.
106 Donna B. Gerdts

is expressed as an oblique noun phrase.21 This raises the question: when is the basic
construction versus the applicative construction used? This section attempts to answer
this question by examining the noun phrases that appear in each type of clause in elic-
itations and in corpus data.
Hul’q’umi’num’ has previously been described as having an animacy restriction
on applied objects (Gerdts 1988a, 1988b). There is indeed a strong tendency for noun
phrases high on the person/animacy hierarchy to occur as applied objects rather than
as obliques (100); furthermore, noun phrases low on the person/animacy hierarchy
dis-prefer applicative constructions (101).

(100) niʔ cən siʔsiʔ-meʔ-t kʷθə sqʷəmey̓.

aux 1sg.sbj frighten-rel-tr det dog
‘I was frightened at the dog.’ (Gerdts and Kiyosawa 2005: 339)

(101) niʔ cən siʔsiʔ ʔə kʷθə snəxʷəł.

aux 1sg.sbj frighten obl det canoe
‘I was frightened at the car.’ (Gerdts and Kiyosawa 2005: 339)

Discussing person/animacy effects, Gerdts (1988b) notes speaker judgments that

animate noun phrases like ‘the priest’ in (102) are best expressed as applied objects
rather than obliques, in contrast to an inanimate noun phrase such as ‘the words of the
priest’ in (103).

(102) niʔ cən q̓el̓-meʔ-t kʷθə ləplit.

aux 1sg.sbj believe-rel-tr det priest
‘I believed the priest.’ (Gerdts and Kiyosawa 2005: 338)

(103) ??niʔ cən q̓el̓-meʔ-t kʷθə s-qʷaqʷəl̓-s kʷθə ləplit.

aux 1sg.sbj believe-rel-tr det nmz-talk(ipfv)-3poss det priest
‘I believed the words of the priest.’ (Gerdts and Kiyosawa 2005: 338)

Likewise, inanimate noun phrases (104) are better obliques than animate noun phrases
(105).

(104) niʔ cən q̓el̓ ʔə kʷθə s-qʷaqʷəl̓-s kʷθə ləplit.

aux 1sg.sbj believe obl det nmz-talk(ipfv)-3poss det priest
‘I believed the priest’s words.’ (Gerdts and Kiyosawa 2005: 341)

21 The endpoint of a motion verb is sometimes expressed as an oblique phrase in a serial verb construc-
tion using motion verbs such as nem̓ ‘go’ and x̌ʷteʔ ‘go toward’ (Gerdts 2010b).
 4 Hul’q’umi’num’ Salish applicative constructions 107

(105) ?✶niʔ cən q̓el̓ ʔə kʷθə ləplit.

aux 1sg.sbj believe obl det priest
Intended: ‘I believed the priest.’ (Gerdts and Kiyosawa 2005: 341)

However, as Gerdts and Kiyosawa (2005b) show, in certain contexts the acceptability of
an inanimate applied object improves greatly. For example, the fog is a force of nature
in (106).

(106) ʔeʔət xʷiʔ siʔsiʔ-meʔ-t-əs tᶿə speʔxʷəm kʷs nem̓-s

aux inch frightened-rel-3sbj det fog det.n go-3poss
ƛ̓əl’im̓-t-əs tᶿə snəxʷəł-s.
steer-tr-3sbj det canoe-3poss
‘He’s scared of the fog when he drives his car.’ (Gerdts and Kiyosawa 2005: 343)

Similarly, when an animate stimulus is expressed as an oblique (107), there is a down-

playing of the participation of the stimulus.

(107) niʔ ʔə č wəł kʷiłəm̓ ʔə kʷθə ʔi hiw̓al’əm̓ sƛ̓əl’iqəł?

aux q 2sg.sbj prf fed.up obl det aux play(ipfv) children
‘Are you fed up with the playing children?’ (Gerdts and Kiyosawa 2005: 343)

After all it is the disturbance made by the children that is annoying and not the children
themselves.
To try to enumerate the effect of person and animacy, we constructed a randomized
list of English sentences based on psych predicates known to take the relational suffix
-meʔ with a variety of potential applied objects and then asked for translations from one
speaker (the late Dr. Ruby Peter) over a period of several days. The results, summarized
in Table 6, show the higher the person/animacy of a noun phrase, the more likely that it
will appear as an applied object rather than as an oblique noun phrase.

Table 6: Applied object vs. oblique NP.

Applied object Oblique

1st/2nd person 40 100% 0 0%
proper noun 20 95% 1 5%
other human 57 90% 6 10%
animal 10 63% 6 37%
inanimate 19 46% 22 54%
total 146 81% 35 19%

Next, we turned to our corpus of texts to make a comparison of person/animacy effects

in applied objects compared to their oblique counterparts. We used a 5,000-line corpus
108 Donna B. Gerdts

of Hul’q’umi’num’ texts. We counted relational applicatives—relational applicatives

formed with the suffix -meʔ and directional applicatives formed with the suffix -nəs—
and also any intransitive clauses that contained an oblique phrase with the appropriate
semantics (stimulus, goal, etc.) and a verb that is known to take these suffixes. We sum-
marize the results in Table 7.

Table 7: Applied object vs. oblique NP.

Applied object Oblique

1st/2nd person 1 0
other human 8 8
animal 1 4
inanimate 3 4
location, clause 6 77
total 19 93

Comparing the text data in Table 7 with the elicited data in Table 6, we see some inter-
esting results. First, it is noticeable that the use of relational applicatives is fairly rare
in texts. There is only one example involving a first or second person, and this is an
applicative. But noticeably, almost half of the noun phrases referring to humans and all
but one of the noun phrases referring to animals were expressed as obliques. Why did 8
out of 16 human noun phrases appear as obliques rather than applied objects, given the
propensity of humans as applied objects in the elicited data? We found various factors
at work. For example, many of the humans expressed as oblique noun phrases did not
refer to individualized persons, but rather to institutionalized positions, such as Indian
agent, or to generics such as “elders”, “white man”, or “people”, see for example this line
from the story ‘Hunting with Flares’ by Samuel Tom.

(108) səw̓ nem̓-s ʔə tᶿə ʔičənt ʔiʔ qʷal, “nə sƛ̓iʔ

nmz.lnk go-3poss obl det agent cnj say 1sg.poss want
kʷən̓s ʔam-əs-θ-əxʷ ʔə kʷə-nə sʔəłtən.”
det.2poss.nmz give-dat-1sg.obj-2sg.sbj obl det-1sg.poss food
‘They go to the Indian agent and say, “I want you to give me my food.”

Oblique phrases are indeed used to express the majority of inanimate noun phrases and
locations and clauses (81 of 90, or 90%). We found only 9 examples where they were
expressed as applied objects. One observation is that when an inanimate item is the
central topic of the text, then it will tend to appear as an applied object, especially if it
has already been established. In the following excerpt from the story Syaləc̓aʔ by Basil
Alphonse, the smoke, expressed as an applied object in (112), is important because it is
leading them to the house of the title character.
 4 Hul’q’umi’num’ Salish applicative constructions 109

(109) ni.i.iʔ wəc̓eʔ ʔə k̓ʷəʔinəł ʔiʔ niʔ wəł wil̓ tᶿə

aux get.to.top obl over.there cnj aux then appear det
sƛ̓eyəq̓əm.
smoke
‘When they got to the mountain top they could see smoke.’

(110) səw̓ təl̓-nəxʷ-əs θəw̓nił, “wəł nił tᶿey̓ niʔ ƛ̓ey̓q̓əm̓.”

nmz.lnk think-nc-3sbj that.one now 3pro det aux smoke(ipfv)
‘She thought, “That is the place where the smoke is coming from.”’

(111) hay sis ʔəw̓ wəł nem̓.

only nmz.aux.3poss lnk then go
‘They started again.’

(112) m̓i.i.i ʔewə-nəs-əs tᶿə ƛ̓ey̓q̓əm̓.

come come-dir-3sbj det smoke(ipfv)
‘They walked towards the smoke.’

In example (113), from a Hul’q’umi’num’ story about the Elhwa people by Manson
George, the river is cast as an applied object; it is not only the home of the people being
discussed, but it is also the main place where the story is set:

(113) sis m̓iw ʔewə-nəs-əm tᶿə stal̓əw̓-s tᶿəw̓-neʔəlł

nmz.aux.3poss come. lnk come-dir-pass det river-3poss det-3pro(pl)
ʔiʔłx̌ʷa . . .
Elhwa
‘And they came to the river of the Elhwa people. . .’

Examples such as these lead Gerdts and Kiyosawa (2005b) to the conclusion that the
person/animacy effects are simply an artifact of other properties. What we see in the
data overall is that it is not the person or animacy of the noun phrase that determines
whether it appears as an applied object or an oblique, but rather its topic-worthiness.
Higher animates are inherently more topical, and things and places of interest to the
storyline or to the main character are also topical and thus can appear as applied
objects. First and second persons are universally more central to the discourse and thus
are topic-worthy. Animates generally outrank inanimates in their degree of importance
in a conversation. Thus, the person/animacy effects are a by-product of the salience of
the noun phrases to the discourse. Most research on topics in Salish language focuses on
subjecthood and the use of passive voice (see Gerdts and Hukari 2008, and references
therein). But our result here shows that more research is needed on the discourse prop-
erties of objects.
110 Donna B. Gerdts

6 Conclusion
This chapter surveys the basic morphological, syntactic, and semantic properties of
Hul’q’umi’num’ applicatives. The main conclusions are as follows:

Morphology
– Hul’q’umi’num’ has four applicative suffixes: -meʔ ‘relational’, -nəs ‘directional’,
-as ‘dative’, and -əłc ‘benefactive’.
– The dative suffix arose from the lexical suffix for ‘face’, suggesting a metonymic
construction as the path of grammaticization. Sources for the other applicative suf-
fixes are unclear.
– Applicative morphemes combine with a wide-variety of other morphemes marking
valence/voice phenomena, and the combinatory order is transparent from the
ordering of the morpheme associated with each construction. The applicative suf-
fixes can appear after the antipassive suffixes, as well as various aspectual suf-
fixes relating to verb classes, and they appear before transitive, object, subject, and
passive inflection. Applicative suffixes can appear both before and after reflexive,
reciprocal, causative, and lexical suffixes so long as restrictions on the transitivity
of the base form are followed and a suitable meaning can be found.

Syntax
– We can distinguish two types of applicative constructions in Hul’q’umi’num’:
relational applicatives are transitive clauses whose corresponding basic clause
is intransitive, and redirective applicatives are ditransitive clauses whose corre-
sponding basic clause (if it has one) is monotransitive.
– The two relational applicatives, formed with meʔ ‘relational’ and -nəs ‘direc-
tional’, have intransitive counterparts in which the relevant noun phrase may
appear as an oblique noun phrase (an optional adjunct) in the clause or in a serial
verb construction. All relational applicatives constructions have intransitive coun-
terparts. Relational applicatives have a valence-increasing effect and the applied
object is the direct object.
– Benefactive applicatives are ditransitives built on transitive constructions. Dative
applicatives (with the exception of the verb pair šem̓ət ‘sell it’ / šam̓əst ‘sell it to him/
her’) do not have non-applicative counterparts. The applied object in a redirective
applicative is always cast as the direct object and the theme noun phrase is cast as
an oblique-marked object. This is characteristic of ditransitives in Hul’q’umi’num’,
a primary/secondary object language. Redirective applicatives show all the hall-
marks of ditransitive constructions, with the sole difference being the presence of
applicative morphology.
– Applied objects have some but not all the properties of direct objects in simple
transitive sentences. The inflect with object person markers, they combine with
 4 Hul’q’umi’num’ Salish applicative constructions 111

passive, reflexive, and reciprocal constructions, and undergo a range of extraction

processes. However, they cannot be demoted in an antipassive construction.

Semantics
– The relational suffix -meʔ appears on a wide variety of verbs; relational applica-
tives are used when the applied object is the stimulus of a psychological predicate
(the most common use), the source of a verb of motion, the goal of a speech act, the
sufferer of an adversative, or the beneficiary of an intransitive verb. By contrast,
the semantic role associated with the relational suffix -nəs is limited to the goal of
a motion verb.
– The redirective suffix -əłc is productively added to transitive verbs to license
applied objects that are beneficiaries. By contrast, the redirective suffix -as appears
on only a half dozen verbs to express applied objects that are recipients or goals.
– Applicative constructions are useful devices for expressing topic-worthy noun
phrases as direct objects.

Abbreviations
act activity
aux auxiliary
ben benefactive applicative
caus causative
cert certainty
cnj conjunction
dat dative applicative
dem demonstrative
dim diminutive
dlm delimiter
det determiner
dir directional applicative
dyn dynamic
fut future
imp imperative
inch inchoative
ipfv imperfective
lnk linker used for connective and complementizer
loc locative prefix
mid middle
nmz nominalizer
nc non-control
obj object
obl oblique
pass passive
112 Donna B. Gerdts

pl plural
poss possessive
prf perfect
pro pronoun
pst past
q interrogative
rdr redirective applicative
recp reciprocal
refl reflexive
rel relational applicative
sg singular
spass subordinate clause passive
sbj subject
tr transitive

References
Brugman, Claudia. 1983. The use of body-part terms as locatives in Chalcatongo Mixtec. Report No. 4 of the
Survey of California and other Languages 4. 235–290.
Dryer, Matthew S. 1986. Primary objects, secondary objects, and antidative. Language 62(4). 808–845.
Gerdts, Donna B. 1984. A relational analysis of Halkomelem causals. In Eung-Do Cook & Donna B. Gerdts
(eds.), Syntax and Semantics, Vol. 16: The syntax of Native American languages, 169–204. New York:
Academic Press.
Gerdts, Donna B. 1988a. A nominal hierarchy in Halkomelem clausal organization. Anthropological Linguistics
30(1). 20–36.
Gerdts, Donna B. 1988b. Object and absolutive in Halkomelem Salish. New York: Garland Publishing.
Gerdts, Donna B.1989. Relational parameters of reflexives: The Halkomelem evidence. In Donna B. Gerdts &
Karin Michelson (eds.), Theoretical perspectives on Native American languages, 259–280. New York: State
University of New York Press.
Gerdts, Donna B. 2000. Combinatory restrictions on Halkomelem reflexives and reciprocals. In Zygmunt
Frajzyngier & Traci S. Curl (eds.), Reciprocals: Forms and functions, 133–160. Amsterdam: John
Benjamins.
Gerdts, Donna B. 2003. The morphosyntax of Halkomelem lexical suffixes, International Journal of American
Linguistics 69(4). 345–356.
Gerdts, Donna B. 2004. Combinatory conditions on Halkomelem causatives. Linguistics 42(4). 767–789.
Gerdts, Donna B. 2010a. Ditransitive constructions in Halkomelem Salish: A direct object/oblique object
language. In Andrej Malchukov, Martin Haspelmath & Bernard Comrie (eds.), Studies in ditransitive
constructions: A comparative handbook, 563–610. Berlin: De Gruyter Mouton.
Gerdts, Donna B. 2010b. Semantic effects in Halkomelem directional applicatives. Northwest Journal of
Linguistics 4(3). 1–17.
Gerdts, Donna B. 2010c. Three doubling constructions in Halkomelem Salish. In Donna B. Gerdts, John C.
Moore & Maria Polinsky (eds.), Hypothesis A / Hypothesis B: Linguistic explorations in honor of David M.
Perlmutter, 183–201. Cambridge, Massachusetts: The MIT Press.
Gerdts, Donna B. 2012. Ghosts, mirrors, and adversative passives in Hul’q’umi’num’, Northwest Journal of
Linguistics 6(4). 1–11.
 4 Hul’q’umi’num’ Salish applicative constructions 113

Gerdts, Donna B. 2016. Object and absolutive in Halkomelem Salish. London: Routledge. (Republication of
Gerdts 1988b)
Gerdts, Donna B. & Mercedes Hinkson. 1996. Salish lexical suffixes: A case of decategorialization. In Adele
Goldberg (ed.), Proceedings of the Conference on Conceptual Structure, Discourse, and Language, 163–176.
Stanford, California: CSLI Pubications.
Gerdts, Donna B. & Mercedes Hinkson. 2004. The grammaticalization of Halkomelem ‘face’ into a dative
applicative suffix, International Journal of American Linguistics 70(3). 227–250.
Gerdts, Donna B. & Thomas E. Hukari. 2005. Multiple antipassives in Halkomelem Salish. Proceedings of the
Twenty-sixth Annual Meeting of the Berkeley Linguistics Society, Special Session, 51–62. Berkeley: University
of California.
Gerdts, Donna B. & Thomas E. Hukari. 2006a. Classifying Hul’q’umi’num’ causatives. Papers for the 41st
International Conference on Salish and Neighbouring Languages, University of British Columbia Working
Papers in Linguistics 18. 129–145.
Gerdts, Donna B. & Thomas E. Hukari. 2006b. The Halkomelem middle: A complex network of
constructions. Anthropological Linguistics 48(1). 44–81.
Gerdts, Donna B. & Thomas E. Hukari. 2008. The expression of noun phrases in Halkomelem texts. Anthro-
pological Linguistics 50(3/4). 1–41.
Gerdts, Donna B. & Kaoru Kiyosawa. 2005. Halkomelem psych applicatives. Studies in Language 29(2).
329–362.
Hollenbach, Barbara E. 1995. Semantic and syntactic extensions of body-part terms in Mixtecan: The case of
‘face’ and ‘foot’. International Journal of American Linguistics 61(2). 168–190.
Hinkson, Mercedes Quesney. 1999. Salishan lexical suffixes: A study in the conceptualization of space. Burnaby,
British Columbia: Simon Fraser University dissertation.
Hukari, Thomas E. (editor) & Ruby Peter (associate editor). 1995. Hul’q’umi’num’ dictionary. Duncan, British
Columbia: Cowichan Tribes.
Kiyosawa, Kaoru & Donna B. Gerdts. 2010a. Benefactive and malefactive uses of Salish applicatives. In
Fernando Zúñiga & Seppo Kittilä (eds.), Benefactives and malefactives: Typological perspectives and case
studies, 147–184. Amsterdam: John Benjamins,
Kiyosawa, Kaoru & Donna B. Gerdts. 2010b. Salish applicatives. Leiden: Brill.
MacLaury, Robert. 1989. Zapotec body part locatives: Prototypes and metaphoric extensions. International
Journal of American Linguistics 55(2). 119–154.
Macaulay, Monica. 1996. A Grammar of Chalcontongo Mixtec. UCPL 127. Berkeley: University of California
Press.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
David Beck
5 Applicatives in Upper Necaxa Totonac
Abstract: Like other members of the Totonac family, Upper Necaxa Totonac (UNT) lacks
prepositions and morphological case. Monomorphemic verbs are either mono- or biva-
lent except for a handful of trivalent stems, and a complex derivational system of caus-
atives and applicatives is used to augment verbal valency. Applicatives in this respect
play a functional role in UNT grammar analogous to that of prepositions and oblique
“semantic” cases in other languages. As derivational elements used in word formation,
applicatives affect the meaning of their bases in a variety of ways: many applicative
forms are entirely compositional, while others are non-compositional but transparent
(i.e., psychologically plausible), and still others are fossilized and idiomatic—although
even fossilized forms continue to be associated with fairly consistent semantic domains.

1 Introduction
Upper Necaxa Totonac (UNT; iso-639 tku) is a member of the Totonacan (a.k.a.
Totonac-Tepehua) language family, spoken by approximately 3,000 people in the north-
eastern part of Puebla State, Mexico. Like other members of the family, UNT lacks prep-
ositions and morphological case. Monomorphemic verbs are either mono- or bivalent
except for a handful of trivalent stems. A complex derivational system of causatives and
applicatives is used to augment verbal valency, creating verbs with three, four, or even
five arguments. Consider the example in (1):1

1 Examples used in this paper are drawn from the Upper Necaxa Totonac database and the author’s
field notes, ultimately having their sources in texts, conversations, and interviews with speakers. The
initials of consultants who provided particular examples are given following the free translation. Exam-
ples use a practical orthography in which most symbols have the values they have in the IPA/APA, with
the following exceptions: <x> = /ʃ/, <tz> = /ts/, <ch> = /tʃ/, <lh> = /ɬ/, <h> = /ʔ/, <j> = /x/, <uj> = /ʍ/, and <y> =
/j/. A colon following a vowel indicates phonemic length, and a straight apostrophe, laryngealization; a
raised comma following a fricative indicates weak ejectivization. The acute accent marks lexical stress.
Semantic roles in lexicographic definitions of verbs are indicated by variables; roles added by applica-
tives are given using an abbreviation for their most typical semantic domain.

Acknowledgements: This paper has benefitted directly from comments by Paulette Levy, Igor Mel’čuk, and
Jim Watters, as well as suggestions from the volume editors. None of this work would have been possible
without the assistance of friends and consultants in Chicontla and Patla. Research for this paper was sup-
ported by the Social Sciences and Humanities Research Council of Canada.

https://doi.org/10.1515/9783110730951-005
116 David Beck

(1) tsa'má i'xhawá'cha' nakila'hmakapiní'ya' puská:t ti: ta'jatatláɬ

tsa'má i'x-hawá'cha' na-kin-∅-la'h-makapin-ni'-ya'
that 3po-boy fut-1obj-sg.obj-all.appl-send:2sbj-ben.appl-ipfv:2sg.sbj
puská:t ti: ∅-ta'jatatla-ɬ
woman hrel 3sg.sbj-sick-pfv
‘On behalf of heri son, you will send me to the womani who is sick .’ (LB)

The verb la'hmaka:'ní' ‘X sends Y to G on behalf of B’ in (1) is based on a bivalent stem,

maká:'n ‘X sends Y’, which is combined with the allative applicative la'h- and the bene-
factive -ni' to add two new objects expressing a Goal and a Beneficiary, respectively. In
total, the language has five applicatives—the benefactive -ni', the instrumental li:-, the
comitative ta:'-, the allative la'h-, and the containing instrumental pu:-. These will be
described in detail in Section 3, following a brief tour of UNT morphosyntax.

2 Morphosyntax
Like all Totonacan languages, UNT is polysynthetic with nominative-accusative align-
ment and flexible (unmarked VO/VS) constituent order. Verbs are inflected for tense,
aspect, mood, and person/number agreement with subject and objects. Person and
number of subjects is expressed cumulatively, while person and number of objects have
separate exponents:

(2) pus chuwá: ka:tama'hni:yá:n ka:talhkuyuyá:n

pus chuwá: ka:-ta-ma'hni:-ya:-n ka:-ta-lhku-yu-ya:-n
intj now pl.obj-3pl.sbj-kill-ipfv-2obj pl.obj-3pl.sbj-burn-trz-ipfv-2obj
‘Well, now they are going to kill you guys, they are going to burn you guys.’ (MR)

Both verbs in (2) use the prefix ta- to show agreement with an understood 3pl subject.
The person of the 2sg object is expressed by the suffix -n while its number is expressed
by the plural object prefix ka:-. The same prefix ka:- also expresses the plurality of first-
and third-person objects (see (13) below). Third-singular subjects and objects trigger
no overt agreement, although I will include ∅s in glosses to help keep track of verbal
valency:

(3) cha:'lho'ho'ho'tzá' ya:wá:lh kí'wi' cha:'ká:m

cha:'-lho'hó'ho'=tzá' ∅-∅-∅-ya:wá:-lh kí'wi' cha:'ká:m
shin-perforated=now 3obj-sg.obj-3sg.sbj-strike-pfv tree woodpecker
‘The woodpecker strikes the tree, making holes in its trunk.’ (PS)
 5 Applicatives in Upper Necaxa Totonac 117

While plural non-human and lower-animate subjects do not always trigger agreement,
third person animates and first- and second-persons always do, with some restrictions
on certain combinations of person and number (Beck 2004, 2016).
All arguments in the clause are bare noun phrases. UNT lacks morphological cases
and the closest thing it has to a preposition is the locative clitic nak=, which forms
optional (adjunct) locative adverbial phrases:

(4) he:po'hó:' wi:lh ha'wá'j naklha'mám

he:po'hó:' ∅-wi:lh ha'wá'j nak=lha'mám
piled 3sg.sbj-sit nixtamal loc=pot
‘The nixtamal is piled up in the pot.’ (SC)

The verb in (4) is wiːlh ‘X sits’ and the nak= phrase specifies the location where X is
sitting. The location, however, is not an argument of wi:lh: it is not part of the verb’s
meaning and does not correspond to an event-participant or variable in its lexico-
graphic definition.
Underived verbs in UNT are either monovalent (5), bivalent (6), or trivalent (7):

(5) xa'kha'chí: ktama'hawásli'

i'x-i'k-ha'chi: i'k-tama'hawas-li'
pst-1sg.sbj-be.drunk 1sg.sbj-fall.down-pfv
‘I was drunk and I fell down.’ (RM)

(6) wi'x nakine'héya'

wi'x na-kin-ne'he-ya'
you fut-1obj-sg.obj-fan-ipfv:2sg.sbj
‘You’re going to fan me (in the sweatlodge).’ (BC)

(7) wan tzu'ma'já:t, kinta:tá' kistaː'maxkí:n

wan tzu'ma'já:t kin-ta:tá' kin-∅-∅-maxki:-n
say girl 1po-father 1obj-sg.obj-3sg.sbj-give-2obj
‘Says the girl, my father gave me to you (in marriage).’ (LB)

The single argument of monovalent verbs like those in (5) is, naturally, a subject, and
the non-subject argument in underived bivalent verbs like (6) is a primary object
(Beck 2016). While the first- and second-person objects both control agreement in the
underived trivalent verb in (7), the primary object is the Recipient/Affected and the
secondary object is the Theme. The main evidence for this is the effect of the suppressive
antipassive suffix -nin/-nun/-nan, which targets the Recipient rather than the Theme, as
shown in (8):
118 David Beck

(8) na'kmaxkiːnín kistánku'

na-i'k-∅-∅-maxki:-nin kin-stánku'
fut-1sg.sbj-3obj-sg.obj-give-antipsuppr 1po-younger.sibling
‘I’m going to give my younger sister away (in marriage).’ (LB)
(Not: ‘I’m going to make gifts / a gift to my younger sister.’)

As we saw in (7), maxkí: ‘X gives Y to Z’ is normally trivalent, with both a Theme and a
Recipient; however, when the antipassive -nin is added, the verb becomes bivalent and
the Recipient is suppressed—it can no longer be expressed in the clause. As argued in
Beck (2016), this suggests that the Recipient argument is more “privileged” in syntactic
terms and so is a better candidate for primary object than the Theme.
There are no other underived valency classes, nor are there any other grammatical
relations in UNT syntax than subject, primary object, and secondary object. UNT nev-
ertheless has large numbers of multivalent stems with three, four, or even five objects,
derived through an extensive set of causative and applicative morphemes. The more
productive of the two causatives is a circumfix that can be applied to both monovalent
and bivalent stems:

(9) ja: ma:púpu:', xku'tánli'

ja: ∅-∅-ma:-pupu-u:' ∅-xku'tán-li'
neg 3obj-sg.obj-caus-bubble-caus:2sg.sbj:pfv 3sg.sbj-sour-pfv
‘You don’t boil it, it goes sour.’ (DR)

(10) tama:ku'kí:lh tza'má pú:ru

∅-∅-ta-ma:-ku'ka:-i:-lh tza'má pú:ru
3obj-sg.obj-3pl.sbj-caus-carry-caus-pfv that burro
‘They make the burro carry the load.’ (MR)

Following the antipassive criterion, the Causee in the causatives of bivalent bases is the
primary object. The second causative, ma'ha- (‘stimulus’), adds an indirect cause to certain
kinds of intransitive bases (e.g., pu'n ‘X blossoms’ < ma'hapú'n ‘A causes X to blossom’).
UNT also has an indefinite voice which suppresses the expression of the Actor:

(11) ja: kimaxki:kán kintapálh

ja: kin-∅-maxki:-kan kin-tapálh
neg 1obj-sg.obj-give-idf 1po-price
‘I haven’t been given my salary.’ (LC)

(12) wi'x ma:ma'hta'ha'lhni:pá:'ka'

wi'x ∅-∅-ma:-ma'hta'ha'lh-ni:-pa:'-ka'
you 3obj-sg.obj-caus-guard-caus-prog:2sbj-idf:2sg.sbj
‘You have been made to guard it.’ (MR)
 5 Applicatives in Upper Necaxa Totonac 119

In the first and third persons (11), this voice simply suppresses the subject and leaves
the object(s) intact; with second persons (12), the erstwhile object triggers second-per-
son subject agreement on the verb. The indefinite voice can also be used with monova-
lent verbs to express indefinite or generic Actors (e.g., ni:kán ‘(people) die [ni:]’). In this
respect, the indefinite voice resembles the impersonal voices of European languages
in that it affects subject of both transitive and intransitive verbs, although there is no
option for expressing the agent in an oblique phrase as there is in, say, Dutch.
With some verbs, -kan forms can have reflexive interpretations—the overlap
between agent suppression and reflexivity being reminiscent of the multiple functions
of the Spanish se. The indefinite voice can also be combined with the antipassive to
create verbs with zero syntactic valency (tzo'hnunkán ‘(people) write [tzo'h-]’).

3 Applicatives in UNT
UNT has five applicatives—four prefixes (li:- ‘instrumental’, ta:'- ‘comitative’, la'h-
‘allative’, and pu:- ‘containing instrumental’) and one suffix (-ni' ‘benefactive’).
These applicatives are freely combinable both which each other and with the causa-
tives, creating complex multivalent verbs like that in (13):2

(13) a'htú' chiwíx i'ka:ta:'pu:la'hmakamílh tza'ká:t kistánku' tza'má chi'xkú'

a'h-tu' chiwíx i'k-ka:-ta:'-pu:-la'h-makamin-lh tza'ká:t
clf-two rock 1sg.sbj-pl.obj-cinst.appl-ctnr-all.appl-throw-pfv sling
kin-stánku' tza'má chi'xkú'
1po-younger.sibling that man
‘I and my brother threw two rocks at that man with a sling.’ (LB)

The verb in (13) is ta:'pu:la'hmakamín ‘X throws Y at G using N along with C’, formed
from makamín ‘X throws Y’ in combination with the comitative applicative ta:'-, which
adds a Co-Actor (C) role to the verb, the allative applicative la'h-, which adds a Goal
(G), and the containing instrumental puː-, which adds a Containing Instrument (N).
The applied objects added by applicatives are secondary objects (Beck 2016), and so in
that sense resemble the prepositional objects required by the English gloss to express
the non-Patient/Theme semantic roles. Note that in (13) the controller of agreement is
the plural primary object (a'htú' chiwíx ‘two rocks’); however, secondary objects can
also control agreement, as seen in (7) above, and in (15) and (42) below. First- and sec-
ond-person objects always control agreement; competition for control of agreement by
third-persons is resolved on the basis of topicality and other discourse factors.

2 The four prefixal applicatives have a “preferred” order liː- >> ta̰ː- >> puː- >> la̰ʔ-; however, there are
verbs where the order is different, reflecting distinct derivational histories.
120 David Beck

Applicatives play a functional role in UNT grammar analogous to that of prepo-

sitions and oblique “semantic” cases in other languages, and are promiscuous in the
sense that they combine with almost any verbal base. Applicative forms account for
1,080/4,754 verbal headwords in the UNT lexical database, and speakers readily accept
novel forms when these are suggested in an appropriate context. As derivational ele-
ments used in word formation, applicatives affect the meaning of their bases in a variety
of ways: many uses of all five applicatives are entirely compositional, while others are
non-compositional but transparent (i.e., psychologically plausible), and still others are
fossilized and idiomatic—although, as I will show in the following sections, even these
are by and large associated with fairly consistent semantic domains.

3.1 -ni' ‘benefactive’

The benefactive applicative, -ni', appears in 293 lexical entries representing 251 inde-
pendent derivations.3 More so than the other applicatives discussed below, -ni' has a
broad range of semantic effects on its base, adding participants in a variety of non-Pa-
tient roles resembling those marked with dative case in Indo-European languages (par-
ticularly, and likely not coincidentally, indirect objects in Spanish). The overall effect of
the benefactive applicative is to add an event-participant that is lower than a Patient on
the scale of typical semantic features of transitive objects (Hopper and Thompson 1980);
given that these semantic features are somewhat heterogeneous, it is not surprising
that the specific semantic roles associated with -ni' are as well.
Over half of the benefactive forms in the lexical database add a semantic role that
we can characterize as Affected (159 entries, 152 independent derivations)—a partic-
ipant less directly involved in the action than a Patient but which nevertheless expe-
riences some effect from it. Several of these forms are given in Table 1. In all of these
examples, -ni' adds an event participant (B) whose interests are impacted either posi-
tively (la'hka:'naní' ‘X weeds a crop for B’), negatively (ha'lha:ní' ‘X steals Y from B’), or in
an indeterminate or context-specific way (la'htziní' ‘X looks at Y belonging to B’).

Table 1: -ni' Affected.

chu'laní' ‘X rinses out Y for B’ (< chu'lá ‘X rinses out Y’)

ha'lha:ní' ‘X steals Y from B’ (< ha'lhá:n ‘X steals Y’)
heːma:'kti:ní' ‘X removes Y from Z’s back for B’ (< heːma:'ktí: ‘X removes Y from Z’s back’)
la'hka:'naní' ‘X weeds a crop for B’ (< la'hka:'nán ‘X weeds a crop’)
laksakní' ‘X chooses Y for B’ (< laksák- ‘X chooses Y’)

3 “Independent” derivations are those based on different roots or derived stems, excluding antipas-
sives, anticausatives, and causatives of previously counted applicative forms, as well as combinations of
previously counted applicative bases with an additional applicative.
 5 Applicatives in Upper Necaxa Totonac 121

Table 1 (continued)

la'htziní' ‘X looks at Y belonging to B’ (< la'htzín ‘X sees Y’)

la'hwaní' ‘X disintegrates affecting B’ (< la'hwán ‘X disintegrates’)
lhta'nhní' ‘X pulls Y belonging to B’ (< lhta'nk- ‘X pulls Y’)
makwaní' ‘X suffices for B’ (< makwán ‘X suffices’)
ma:sputu:ní' ‘X uses up Y belonging to B’ (< ma:sputú: ‘X uses up Y’)
pu:la'hwaxtuní' ‘X eats the insides of Y belonging B’ (< pu:la'hwaxtú ‘X eats the insides of Y’)
tatu'kxní' ‘X breaks affecting B’ (< tatú'kx- ‘X breaks’)
tu'ksní' ‘X hits Y affecting B’ (< tu'ks- ‘X hits Y’)

Benefactives can be formed from monovalent (14), bivalent (15), and multivalent (16)
bases:

(14) kilaktzu'nuni'má:lh kila:xáx

kin-∅-∅-lak-tzu'nu-ni'-ma:lh kin-la:xáx
1obj-sg.obj-3sg.sbj-intns-shrivel-ben.appl-prog 1po-orange
‘My oranges are shrivelling up on me.’ (PS)

(15) kinha'lha:ní'ka' kinkawa:yúj

kin-∅-ha'lha:n-ni'-ka' kin-kawa:yúj
1obj-sg.obj-steal-ben.appl-idf 1po-horse
‘My horse was stolen from me.’ (LB)

(16) kakinhe:ma:'kti:ní'chi' tza'má tzu'ma'já:t kis’á'ta', wan chi'xkú'

ka-kin-he:-ma:'kti:-ni'-chi' tza'má tzu'ma'já:t kin-s’á'ta'
opt-1obj-sg.obj-back-remove-ben.appl-dist:2sg.sbj that girl 1po-child
wan chi'xkú'
say man
‘Take my child off that girl’s back there for me! says the man.’ (PS)

The sentence in (14) is based on laktzu'nuni' ‘X shrivels up on B’, from the monovalent
laktzu'nú ‘X shrivels up’, while ha'lha:ní' ‘X steals Y from B’ in (15) is derived from biva-
lent ha'lhá:n ‘X steals Y’. The verb in (16) is he:ma:'kti:ní' ‘X removes Y from Z’s back for
B’, which is based on the trivalent verb heːma:'ktí: ‘X removes Y from Z’s back’.
As with all applied objects in UNT, the benefactive object is a secondary object, as
we can see in the following examples illustrating the usage of the antipassive form of
the verb ha'lha:ní' ‘X steals Y from B’ (seen in [15] above):

(17) kit i'kha'lha:naní'n wi'x

kit i'k-∅-ha'lhá:n-nan-ni'-n wḭš
I 1sg.sbj-sg.obj-steal-antipsuppr-ben.appl-2obj you
‘I stole from you.’ (LB)
122 David Beck

As shown in Beck (2016), primary objects are distinguished from secondary objects in
that the former, and not the latter, are suppressed by the antipassive suffix; in (17) it is
the basic object, not the applied object expressing the Affected, that is removed from
the clause. This is different from the pattern reported for Tlachichilco Tepehua (Watters
1987) and Papantla Totonac (Levy 2002), where benefactive applied objects (but not
other applied objects) are primary objects.
Not unexpectedly, the benefactive applicative can also add a Recipient (28 lexical
entries, 20 independent derivations). Several examples are shown in Table 2. The
Recipient role in these verbs corresponds not only to the endpoint in verbs of transfer
(maka:'ní' ‘X sends Y to B’), but also to the recipient in verbs of sharing (kilhche'hení' ‘X
shares a chunk of Y with B’) and paying/owing (lakle:ní' ‘X owes Y to B’).

Table 2: -ni' Recipient.

kilhche'hení' ‘X shares a chunk of Y with B’ (< kilhche'hé ‘X breaks off a chunk of Y’)
lakle:ní' ‘X owes Y to B’ (< laklé:n ‘X owes Y’)
le:ní' ‘X takes Y to B’ (< le:n ‘X takes Y’)
li:maka:'ní' ‘X holds Y out to B’ (< li:maká:'n ‘X holds Y out’)
li:miní' ‘X brings Y to B’ (< li:mín ‘X brings Y’)
maka:'ní' ‘X sends Y to B’ (< maká:'n ‘X sends Y’)
makaminí' ‘X throws Y to B near speaker’ (< makamín ‘X throws Y towards speaker’)
tama'hxte'hní' ‘X is left for B’ (< tama'hxté'h- ‘X is left there’)
xo'honí' ‘X gives Y in payment to B’ (< xo'hó ‘X gives Y in payment’)
xte'hní' ‘X leaves Y for B’ (< xte'h- ‘X leaves Y behind’)

The benefactive also adds Addressees to expressions of verbal and non-verbal commu-
nication (32 entries, 24 independent derivations). A few examples are given in Table 3.
Several forms in this group are trivalent and include the message as an argument (waní'
‘X says Y to B’); however, UNT also has a number of monovalent verbs of vocalization
(pixlí:‘X sings’, ta'sá ‘X vocalizes’) and gesture (makawán ‘X makes a sound or gesture
with hands’) that, when affixed with -ni', create verbs of communication that do not
include the message as part of their valency.

Table 3: -ni' Addressee.

helhaski'ní' ‘X asks B about Y’ (< helha- ‘mouth’ + ski'n ‘X asks for Y’)
helhpanhní' ‘X betrays a confidence to B’ (< helhpánh- ‘X betrays a confidence’)
helhs’oliní' ‘X whistles at B’ (< helhs’olí ‘X whistles’)
makawaní' ‘X waves to B’ (< makawán ‘X makes a sound or gesture with hands’)
pixli:ní' ‘X sings to B’ (< pixlí:‘X sings’)
ta'saní' ‘X calls out to B’ (< ta'sá ‘X vocalizes’)
tzo'hní' ‘X writes Y to B’ (< tzo'h- ‘X writes Y’)
waní' ‘X says Y to B’ (< wan ‘X says Y’)
 5 Applicatives in Upper Necaxa Totonac 123

In addition to these three semantic roles, all of which are indirectly affected by the
action of the Actor, there are roles associated with the benefactive where the partici-
pant is essentially unaffected by or peripheral to the event. For instance, the database
contains 11 entries (9 independent derivations) where the applied object expresses a
stimulus for some kind of emotional state (Table 4). Another set of 20 entries (14 inde-
pendent derivations) add a participant that acts as a spatial reference point for the
event. Several of these are shown in Table 5. Note that in those cases where the action
of the verb is directed towards the new participant, that participant is not the actual
endpoint or physical target of the action, which distinguishes these uses of -ni' from the
allative la'h- (§ 3.4). For some of the forms in this group (e.g., tza:'laní' ‘X runs away from
B’) the new semantic role may in fact be more a Motive—or at least an entity that moti-
vates the action—than a Direction, and so these might be more akin to those in Table 4.

Table 4: -ni' Emotional stimulus.

jikwaní' ‘X is afraid of B’ (< jikwán ‘X feels fear’)

lha'hwaní' ‘X gets tired of B’ (< lha'hwán ‘X gets tired’)
ma:xananí' ‘X is ashamed in front of B’ (< ma:xanán ‘X is ashamed’)
si:'tzi:ní' ‘X gets angry with B’ (< si:'tzíː ‘X gets angry’)

Table 5: -ni' Direction.

he:miní' ‘X turns X’s back on B’ (< he:mín ‘X stands with back (heː-) to speaker’)
kilhwaní' ‘X opens one’s mouth at B’ (< kilhwán ‘X opens X’s mouth’)
la'hatze'hní' ‘X hides X’s face from B’ (< la'hatzé'h- ‘X hides X’s face’)
pu:'laní' ‘X leads B’ (< pu:'lá ‘X goes first’)
sta:laní' ‘X follows B’ (< sta:lá ‘X comes behind’)
tza:'laní' ‘X runs away from B’ (< tza:'lá ‘X flees’)

In another group of -ni' forms (31 entries, 20 independent derivations), the applicative
adds what seems to be a Ground—that is, an entity that frames or defines the locus of
the event (Table 6). Many of these forms are derived from verbs built on stative bases—
either bound roots like -nuː ‘inside, contained’ and -xtu ‘outside, projecting’, or stative
posture verbs like wilá ‘be seated’. The semantic role of Ground is the only one associ-
ated with the benefactive that is typically inanimate and non-human. There are also 12
independent derivations in the database that have idiosyncratic meanings.

Table 6: -ni' Ground.

la'hnu:ní' ‘X is stuck in B’ (< la'hnúː ‘X is stuck’)

ma:'hs’oní' ‘X illuminates B’ (< ma:'hs’ó ‘X casts light’)
ma:nu:ní' ‘X puts Y into B’ (< ma:núː ‘X puts Y in’)
124 David Beck

Table 6 (continued)

ma:xtuní' ‘X takes Y out of B’ (< ma:xtú ‘X takes Y out’)

pu'ní' ‘X buds on B’ (< pu'n ‘X buds’)
taxtuní' ‘X comes out of B’ (< taxtú ‘X leaves’)
wilaní' ‘X is placed on B’ (< wilá ‘X is seated’)
xtu'tuní' ‘X sucks Y from B’ (< xtu'tú ‘X sucks on Y’)
yujní' ‘X falls off of B’ (< yuj- ‘X descends’)

In addition to verbs where -ni' acts as an applicative, adding an object, there are 14
verbs (12 independent derivations) in which -ni' does not increase the valency of its
base (Table 7), but instead simply changes the semantic role of a non-Actor participant
from a Patient- or Theme-like role to something lower on Hopper and Thompson’s
(1980) scale of semantic transitivity. Consider, for example, the sentences in (18) and
(19) based on the verb la'hamilíː ‘X covers Y’s face with a cloth’ and its -ni' form, la'ha-
miliːní' ‘X shelters B’s face with a cloth’:

(18) kila'hamilí:lh
kin-∅-∅-la'ha-mili:-lh
1obj-sg.obj-3sg.sbj-face-cover-pfv
‘It covered my face.’ (RM)

(19) na'kla'hamili:ní kis’á'ta' ja: kasnó'hli' ú:'ni'

na-i'k-∅-∅-la'ha-mili:-ni' kin-s’á'ta' ja:
fut-1sg.sbj-3obj-sg.obj-face-cover-ben.appl 1po-child neg
ka-∅-∅-∅-sno'h-li' ú:'ni'
opt-3obj-sg.obj-3sg.sbj-whip-pfv air
‘I’m going to cover my child so that the wind doesn’t blow on her.’ (RM)

The base form here in (18) is used when the cloth is in direct contact with the person
or object being covered (the same form could be used, for example, when talking about
covering a plate of food to keep it warm), whereas in (19) what is being covered, the child,
is not necessarily in contact with the cloth—for instance, the child could be in a cradle
with a blanket draped over it or carried in someone’s arms under a shawl. The use of
-ni' here is semantically related to the benefactive applicative in that the action affects
the child, but the child is not directly/physically involved. Because the applied object
added by -ni' is a secondary object, the lower semantic transitivity is accompanied by
lower syntactic transitivity: the primary object is exchanged for a secondary object (an
object inaccessible to suppression by the antipassive), making la'hamiliːní' syntactically
intransitive.
 5 Applicatives in Upper Necaxa Totonac 125

Table 7: Detransitivizing -ni'.

a'hahi:ní' ‘X believes B’s words’ (< a'hahí: ‘X believes idea Y’)

la'halhtalaní' ‘X locks B in by jamming the door’ (< la'halhtalá ‘X covers opening Y with a board’)
la'hamili:ní' ‘X shelters B with a cloth’ (< la'hamilíː ‘X covers Y’s face with a cloth’)
lakpuwaní' ‘X desires B’ (< lakpuwán ‘X thinks about Y’)
lhajaní' ‘X makes a profit on B’ (< lhajá ‘X earns wage Y’)
ma'hla'htziní' ‘X serve as a midwife to B’ (< ma'hla'htzín ‘X sees another person’s Y’)
pi'taní' ‘X presses B (e.g., button)’ (< pi'tá ‘X pokes Y causing discomfort’)
wili:ní' ‘X puts B up against object; X strikes B’ (< wilíː ‘X places Y’)
xki'yuju:ní' ‘X scrubs B off a surface’ (< xki'tyujú: ‘X scrubs Y’)
x’a'ha:yuju:ní' ‘X wipes B off a surface’ (< x’a'ha:yujú: ‘X wipes Y clean’)

Similarly, the last two examples in Table 7 are based on a bound stem -yuju: ‘X takes
Y down’ (< yuj- ‘X descends’ + -uː ‘transitivizer’) and describe the action of wiping or
scrubbing a substance off of something. Without -ni', what is being cleaned is expressed
as the primary object (20), but in the -ni' form, the applied object expresses the sub-
stance being removed (21):

(20) i'kli:a'kpu:x’a'hayujú: mesa kili:x’a'hán

i'k-∅-∅-li:-a'kpu:-x’a'hayuju: mesa kin-li:-x’a'há-n
1sg.sbj-3obj-sg.obj-inst.appl-wipe.clean table 1obj-inst.appl-scrape-nmz
‘I’m wiping the tabletop with my brush’ (PS)

(21) x’a'ha:yuju:ni'má:lh pintura nakmesa

∅-∅-∅-x’a'ha:yuju:-ni'-ma:lh pintura nak=mesa
3obj-sg.obj-3sg.sbj-wipe.clean-ben.appl-prog paint loc=table
‘She wipes away the paint from the table.’ (RM)

The verb in (21) is not transitive because the endpoint of the action is not a discrete
or individuated object but a substance. Substances are low on the scale of semantic
transitivity, which motivates the verb’s syntactic intransitivity (i.e., its lack of a primary
object). While -ni' does not count as an applicative in forms in Table 7, the association of
both applicative and non-applicative -ni' with non-Actor participants on the lower end
of the scale of semantic transitivity points to an etymological connection between them.

3.2 liː- ‘instrumental’

The most prolific of the UNT applicatives is the instrumental li:-, which appears in 551
lexical entries representing 490 independent derivations in the lexical database. The
instrumental applicative has two functions: one is to add the semantic role of Instru-
ment/Means (I) to the valency of its base, the other to add the Reason (R) for which an
126 David Beck

event occurs. In either function, the instrumental applicative is near-freely combinable

with any verb expressing an action that can be performed with some sort of implement
or for any kind of motive.
In the database, there are 292 lexical entries (261 independent derivations) where
li:- adds an Instrument or a Means to the valency of its base—that is, it adds an applied
object expressing either some implement or tool deliberately used by the Actor, or some
entity that serves as a means of realizing the action expressed by the verb. As we can
see in Table 8, li:- combines with monovalent (kuxtú- ‘X weeds’), bivalent (nik- ‘X clubs
Y’), and multivalent bases (ma:wí: ‘X feeds Y to Z’).

Table 8: li:- Instrument/Means.

li:ha'pí ‘X locks Y with I’ (< ha'pí ‘X locks Y’)

li:ho'nún ‘X drinks I’ (< ho'nún ‘X drinks’)
li:lhawá ‘X makes Y out of I’ (< lhawá ‘X makes Y’)
li:kuxtú ‘X weeds with I’ (< kuxtú ‘X weeds’)
li:ku'chú: ‘X heals Y with I’ (< ku'chú: ‘X heals Y’)
li:lhká: ‘X measures the length of Y with I’ (< lhka: ‘X measures the length of Y’)
li:ma:wí: ‘X feeds Y to Z using I’ (< ma:wí: ‘X feeds Y to Z’)
li:ník- ‘X clubs Y with I’ (< nik- ‘X clubs Y’)
li:slantá ‘X glues Y with I’ (< slantá ‘X glues Y’)
li:sná't- ‘X winds Y around I’ (< sna't- ‘X twists Y’)

There are also 96 lexical entries (69 independent derivations) where li:- adds an Inte-
grant—an entity that is involved in or makes possible the event but is not deliberately
wielded by the Actor (Table 9). The applied objects in these forms are, broadly-speaking,
inanimate non-Patient participants or abstractions unaffected by the event or action of
the Actor. This set includes 16 forms derived from one of 4 motion verbs that express
events of taking/bringing (e.g., le:n ‘X takes I’ < a'n ‘X goes’; li:chín ‘X arrives here with I’
< chin ‘X arrives here’).

Table 9: li:- Integrant.

leːn ‘X takes I’ (< a'n ‘X goes’)

li:a'hachuyá: ‘X hears I that isn’t there’ (< a'hachuyá: ‘X has auditory hallucinations’)
li:a'hlhche'hxlá ‘X trips on I’ (< a'hlhche'hxlá ‘X stumbles’)
li:chá̰ːn ‘X arrives there with I’ (< cha̰ːn ‘X arrives there’)
li:chín ‘X arrives here with I’ (< chin ‘X arrives here’)
li:chiwi:nán ‘X speaks about I’ (< chiwi:nán ‘X speaks’)
li:kinkalá ‘X reeks of I’ (< kinkalá ‘X reeks’)
li:lhti'pí'n ‘X (water) is made murky by I’ (< lhti'pí'n ‘X (water) is murky’)
li:pa:stá'k- ‘X is reminded of Y by I’ (< pa:stá'k- ‘X remembers Y’)
li:pa'lha:nán ‘X vomits up I’ (< pa'lha:nán ‘X vomits’)
li:pixkú'n ‘X gets mumps from I’ (< pixkú'n ‘X has mumps’)
 5 Applicatives in Upper Necaxa Totonac 127

Table 9 (continued)

li:pixlí: ‘X sings I’ (< pixlí: ‘X sings’)

li:smaní: ‘X is accustomed to I’ (< smaní: ‘X feels at ease’)
liːmín ‘X brings I’ (< min ‘X comes’)

The li:- applicative is also used to add a Reason for an event occurring, as in (22):

(22) xa'ka:namá:lh tza'má chi'xkú', katu:wálh li:xa'ká:

∅-xa'ka:nan-ma:lh tza'má chi'xkú' katu:wálh
3sg.sbj-get.angry-prog that man anything
∅-∅-∅-li:-xa'ká:
3obj-sg.obj-3sg.sbj-inst.appl-get.angry
‘That man is always scolding, he gets mad about anything.’ (CF)

While there are 164 lexical entries where li:- has exclusively this effect on the meaning
of its base (Table 10), it is difficult to meaningfully quantify words of this type. This is
both because li:- Reason seems to be almost universally applicable (and so the number
of entries in the database is at best a convenience sample), and because li:- Reason
forms can be derived from the same bases as li:- Instrument forms, creating homopho-
nous words with different senses:

(23) kili:lhtu'kúlh cha:'tín hó'ni' a'htín kuchí:lu

kin-∅-∅-li:-lhtu'ku-lh cha:'-tin hó'tni' a'h-tin kuchí:lu
1obj-sg.obj-3sg.sbj-inst.appl-stab-pfv clf-one drunk.person clf-one knife
‘The drunk stabbed me with the knife.’ (PS)

(24) kili:lhtu'kúlh wa:má wa:káx i's’á'ta'

kin-∅-∅-li:-lhtu'ku-lh wa:má wa:káx i'x-s’á'ta'
1obj-sg.obj-3sg.sbj-inst.appl-stab-pfv this cow 3po-child
‘The cow gored me because of its calf.’ (RM)

Here, we seem to have two different senses of li:lhtu'kú < lhtu'kú ‘X stabs Y’)—‘X stabs
Y with I’ (23) and ‘X stabs I because of R’ (24). This might lead us posit two homopho-
nous prefixes, one which adds Instruments and the other Reasons; however, there are
a couple of arguments against this. One is simply distributional: the two hypothetical
affixes never seem to co-occur and speakers do not accept suggested forms with two
instance of li:- or with meanings along the lines of ‘X Vs with I because of R’. The other
argument is that Reasons and Instruments occupy adjacent semantic domains, and some
of the verbs formed with li:- add participants that could also be thought of as reasons
for an event taking place (e.g., li:la'hawi'tí ‘X is made dizzy by I’, li:taku'xatzí: ‘X feels
128 David Beck

suffocated by I’ < taku'xatzí: ‘X feels suffocated’). It seems more likely that Instrument
vs. Reason is a context-driven interpretation of a single element with a vaguer meaning.

Table 10: liː- Reason.

li:a'htuyún ‘X worries about R’ (< a'htuyún ‘X worries’)

li:a'kchuyá: ‘X acts crazy because of R’ (< a'kchuyá: ‘X acts crazy’)
li:ka'tzán ‘X feels pain because of R’ (< ka'tzán ‘X feels pain’)
li:kú'n ‘X swells up because of R’ (< ku'n ‘X is swollen’)
li:la'hlh’ó'p- ‘X melts because of R’ (< la'hlh’ó'p- ‘X melts’)
li:lhtatá ‘X feels sleepy because of R’ (< lhtatá ‘X sleeps’)
li:ma:xanán ‘X is ashamed because of R’ (< ma:xanán ‘X is ashamed’)
li:ní: ‘X dies because of R’ (< ni: ‘X dies’)
li:ská:k- ‘X dries out because of R’ (< ska:k- ‘X dries out’)
li:taxtú ‘X leaves because of R’ (< taxtú ‘X leaves’)

An interesting feature of the Reason use of li:-, one certainly tied to its promiscuity and
textual frequency, is that it has a constructional use in expressions that are equivalent
to English that’s why and because clauses:

(25) lhú:wa' wi:lh xka:n, u:tzá: li:lhu:wán xuj

lhú:wa' ∅-wi:lh xka:n u:tzá: ∅-li:-lhu:wán xuj
many 3sg.sbj-sit water that 3sg.sbj-inst.appl-be.many mosquito
‘There is a lot of water, that’s why there are so many mosquitos.’ (LA)

(26) li:taxtúlh i tza'má hawá'cha' xlhawasti:kán

∅-∅-∅-li:-taxtu-lh i tza'má hawá'cha'
3obj-sg.obj-3sg.sbj-inst.appl-leave-pfv jnct that boy
i'x-∅-∅-lhawasti:-kan
pst-3obj-sg.obj-mistreat-idf
‘The boy left because he was treated badly.’ (RM)

Rather than using a conjunction to relate an event and its reason for occurring, Totona-
can languages add an instrumental applicative to the main verb and express the reason
in a separate clause that functions as the applied object. Sentences like (26) are used
most frequently by older speakers, while younger speakers tend to use the borrowed
Spanish conjunction porque instead.
 5 Applicatives in Upper Necaxa Totonac 129

3.3 ta:'- ‘comitative’

The comitative applicative, ta:'-, appears in 139 entries (84 independent derivations). In
the most general sense, ta:'- adds a Co-Actor (C) or some other entity that co-performs or
assists the Agent in carrying out the event described by the verb.
In 39 independent forms, including those shown in Table 11, ta:'- adds an animate
Co-Actor, and speakers readily accept novel suggested forms with this meaning. The
comitative can be added to monovalent (ta:'á'n ‘X goes with C’ < a'n ‘X goes’) and bivalent
verbs (ta:'x’á: ‘X shucks Y with C’ < x’a: ‘X shucks corn’), as well as to more complex
stems formed with causatives (27) and other applicatives (28):

(27) na'kta:'ma:tanhapu:yá:n minkí'wi'

na-i'k-∅-ta:'-ma:-tanhapu:-ya:-n min-kí'wi'
fut-1sg.sbj-sg.obj-cmt.appl-caus-at.hill.bottom-ipfv-2obj 2po-tree
‘I’m going to get your wood down the slope with you.’ (SC)

(28) nai'ka:ta:'li:tanká: pu:laktín kí'wi' chi'xkuwín kimachi:tká'n

na-i'k-∅-ka:-ta:'-li:-tan-ka: pu:lak-tin kí'wi'
fut-1sg.sbj-3obj-pl.obj-cmt.appl-inst.appl-buttocks-chop clf-one tree
chi'xkú'-win kin-machí:t-ka'n
man-pl 1po-machete-pl.po
‘I and the men will cut down a tree with our machetes.’ (LB)

The verb in (27) is ta:'ma:tanhapú: ‘X helps C get Y down a hill’, derived from the causa-
tive stem ma:tanhapú: ‘A takes X down a hill’ (< tanhapú: ‘X is at the bottom of a slope’).
In (28), ta:'li:tanká: ‘X with C chops Y down using I’ is based on li:tanká: ‘X chops Y down
using I’ (< liː- ‘instrumental’ + tanká: ‘X chops Y down’). In both examples, the Co-Actor
controls object agreement.

Table 11: taː'- Co-Actor.

ta:'á'n ‘X goes with C’ (< a'n ‘X goes’)

ta:'chiwi:nán ‘X converses with C’ (< chiwi:nán ‘X speaks’)
ta:'hama:nán ‘X plays a game with C’ (< hama:nán ‘X plays a game’)
ta:'la'hslá'h- ‘X helps C stir Y’ (< la'hslá'h- ‘X stirs Y’)
ta:'lá ‘X performs an activity with C’ (< la ‘X performs an activity’)
ta:'li:tanká: ‘X with C chops Y down using I’ (< li:tanká: ‘X chops Y down using I’)
ta:'ma:tanhapú: ‘X helps C get Y down a hill’ (< ma:tanhapú: ‘X takes Y down a hill’)
ta:'pu:la'hmakamín ‘X with C shoots Y at G using I’ (< pu:la'hmakamín ‘X shoots Y at G using I’)
ta:'pu'tzá ‘X searches for Y with C’ (< pu'tzá ‘X searches for Y’)
ta:'x’á: ‘X shucks Y (corn) with C’ (< x’á: ‘X shucks Y [corn]’)
130 David Beck

Eleven independent comitative forms express an event where the Actor shares a mental
state—either literally (ta:'li:ka'tzí: ‘X agrees with C with respect to Y’ < li:ka'tzí: ‘X agrees
to/about Y’) or vicariously (ta:'pa:tí: ‘X feels C’s suffering’ < pa:tí: ‘X suffers’)—with a
Co-Experiencer (Table 12). The forms here are divided between those with ta:'- alone
and those forms that have both ta:'- and the reciprocal la:-. In the latter cases, the com-
itative has no effect on the valency of the basic verb, but instead “retransitivizes” a
detransitivized reciprocal verb form:

(29) ja: lakahí:, ja: ta:'to:laku'tún

ja: ∅-∅-∅-lakahi: ja: ∅-∅-∅-ta:'-tawilá-ku'tún
neg 3obj-sg.obj-3sg.sbj-like neg 3obj-sg.obj-3sg.sbj-cmt.appl-sit-dsd
‘She doesn’t like him, she doesn’t want to marry (lit. sit with) him.’ (CF)

(30) xlaká'n xtala:s’o'ha:lha:'wán porque tala:lakahí:

xlaká'n i'x-ta-la:-s’o'há-lha:'wán porque ta-la:-lakahi:
they pst-3pl.sbj-rcp-hug-walk because 3pl.sbj-rcp-like
‘They walk around embracing each other because they like each other.’ (NS)

(31) kata:'la:lakáhi:'
ka-∅-∅-ta:'-la:-lakáhi:'
opt-3obj-sg.obj-cmt.appl-rcp-like:2sg.sbj:pfv
‘May you and she like each other!’ (LC)

The transitive verb lakahí: ‘X likes Y’ is shown in (29). Its reciprocal form in (30)
(laːlakahí: ‘X and Y like each other’) is intransitive and obligatorily has a plural subject,
whereas the comitative form in (31) (ta:'laːlakahí: ‘X is mutually fond of Y’) has a sin-
gular subject and is once again transitive. In this form, laː- allows for the expression of
mutuality, but the comitative allows the point of view of one of the two participants to
be taken, that participant being expressed as the subject.

Table 12: taː'- Co-Experiencer.

ta:'ka'tzanajwán ‘X feels C’s pain’ (< ka'tzanajwán ‘X feels pain’)

ta:'la'halhu:má:n ‘X sympathizes with C’ (< la'halhu:má:n ‘X sympathizes with Y’)
ta:'la:lakahí: ‘X and C are attracted to each other’ (< lakahí: ‘X likes Y’)
ta:'la:pa:xkí: ‘X loves C who reciprocates’ (< pa:xkí: ‘X loves Y)
ta:'la:la'htzín ‘X watches out for C’s interests’ (< la'htzín ‘X sees Y’)
ta:'la:lé:n ‘X gets along with C’ (< le:n ‘X takes Y’)
ta:'li:ka'tzí: ‘X agrees with C with respect to Y’ (< li:ka'tzí: ‘X is in agreement about Y’)
ta:'li:puwán ‘X feels sadness for C’ (< li:puwán ‘X feels sadness’)
ta:'ma'hka'tzí: ‘X feels C’s physical discomfort’ (< ma'hka'tzí: ‘X feels a physical sensation’)
ta:'pa:tí: ‘X feels C’s suffering’ (< pa:tí: ‘X suffers’)
ta:'tala'hapu'tzí: ‘X feels compassion for C’s loss’ (< tala'hapu'tzí: ‘X suffers a loss’)
 5 Applicatives in Upper Necaxa Totonac 131

The comitatives in Tables 11 and 12 add either Co-Actors or Co-Experiencers to the

valency of their bases, and so conform to our general expectations of the semantic effect
of a comitative. There are, however, 8 independent forms where the applied phrase
expresses what might be called an Accessory—an event-participant that accompanies
the Actor in some sense but does not necessarily perform the same action or experience
the same mental state (Table 13). The applied object in these verbs expresses an addi-
tional participant that is in close proximity to or joined with the Actor, as seen in these
examples based on ta:'tamá: ‘X lies down with C’ (< tamá: ‘X lies down’):

Table 13: taː'- Accessory.

ta:'la:chi'pá ‘X and C are stuck to each other’ (< chi'pá ‘X is stuck to Y’)
ta:'la:li:tapi'tzí ‘X and C have neighbouring land’ (< la:li:tapi'tzí ‘X has neighbouring land’ [pl. sbj.])
ta:'la:pe'hxtó'h- ‘X and C are close to each other’ (< pe'hxtó'h- ‘X is close to Y’)
ta:'stúk- ‘X is joined with C end to end’ (< ✶-stuk ‘joined’)
ta:'tala'hxtó'h- ‘X and C get together’ (< tala'hxtó'h- ‘X is joined together’)
ta:'talakxtimí: ‘X and C get together’ (< talakxtimí: ‘X gathers, X comes together’)
ta:'tamá: ‘X lies down with C’ (< tamá: ‘X lies down’)
ta:'tape'hxtimí: ‘X and C are shoulder to shoulder’ (< tape'hxtimí: ‘X lines up at the shoulder’)
ta:'yá:lh ‘X stands holding C’ (< yaːlh ‘X stands’)

(32) lakxtím kinta:'tamá:lh kinchichí'

lakxtím kin-∅-∅-ta:'-tama:-lh kin-chichí'
lined.up 1obj-sg.obj-3sg.sbj-cmt.appl-lie-pfv 1po-dog
‘My dog lay down at my side.’ (PS)

(33) ta:'tamá:lh i'xtakú'ka:' minkawa:yúj katanlha'háki:'

∅-∅-∅-ta:'-tama:-lh i'x-takú'ka:' min-kawa:yúj
3obj-sg.obj-3sg.sbj-cmt.appl-lie-pfv 3po-load 2po-horse
ka-∅-∅-tanlha'háki:'
opt-3obj-sg.obj-slap.on.behind:2sg.sbj:pfv
‘Your horse lay down with its load, get it up!’ (PS)

In (32), there is no necessity that the added participant (‘I’) also be lying down, only that
it be in close proximity to the Actor (the dog). In (33), the added participant is inanimate
and so cannot have co-performed the action of lying down along with the horse. Addi-
tionally, there is a third sense of ta:'tamá:, seen in (34), where the verb does imply a sort
of co-action by the new participant:
132 David Beck

(34) nata:'tapá:ya' kimpuská:t, mat wani'ta:'kí:lh

na-∅-∅-ta:'-tapa:-ya' kin-puská:t mat
fut-3obj-sg.obj-cmt.appl-lie:2sbj-ipfv:2sg.sbj 1po-woman quot
wan-ní-ta:'kí:-lh
say-ben.appl-get.up-pfv
‘You will go to bed with my wife! he said and took off.’ (RM)

However, even here there is more to it than simply lying down with the Actor, so we
cannot say in any of these cases that the comitative adds a Co-Actor. Note also that three
of the forms in Table 13 include the reciprocal prefix, indicating the mutuality of the
spatial relation being expressed.
Eleven comitative forms in the database add the expression of an Associate, some-
one who participates in a particular social relationship with the Actor (Table 14).
Included in this category are a number of verbs of marriage and kinship, as well as
verbs of sharing and economic transaction. Beyond these, there are a few uses of ta:'-
where the new participant is inanimate and is more of a supplement or ingredient (e.g.,
ta:'lhawá ‘X makes Y [lhawá] with ingredient C’), as well as a dozen or so more where
the effect of ta:'- is even more idiosyncratic.

Table 14: ta:'- Associate.

ta:'ma'haxtó'h- ‘X is married to C’ (< ma'haxtó'h- ‘X is married’)

ta:'la:la'hxta:'palí: ‘X exchanges Y with C’ (< la'hxta:'palí: ‘X exchanges Y’)
ta:'la:ka:xlá ‘X resolves problem Y with C’ (< laːka:xlá ‘X resolves problem Y’)
ta:'pí'ks- ‘X breaks off a piece of Y to share with C’ (< pi'ks- ‘X breaks off a piece of Y’)
ta:'ski'ni:nín ‘X asks for a girl in marriage for C’ (< ski'n ‘X asks for Y’ + -ni: ‘subst’ + -nin ‘antipsuppr’)
ta:'skúj- ‘X works for C’ (< skuj- ‘X works’)
ta:'wá ‘X shares Y (food or drink) with C’ (< wa ‘X eats or drinks Y’)
ta:'wí:lh ‘X is married to C’ (< wiːlh ‘X is seated’)

3.4 la'h- ‘allative’

The allative applicative la'h- appears in 65 lexical entries, forming part of 35 independ-
ent derivations. Semantically more regular than the comitative, it adds a Goal (G) or
some other entity towards which the action is directed.
The bulk of the la'h- forms in the database (27 entries, 14 independent stems) are
based on motion verbs and add a Goal of Motion (Table 15). The Goal in these forms is
almost invariably human, as in (35) and (36):
 5 Applicatives in Upper Necaxa Totonac 133

(35) la'hyújli' tza'má tzu'ma'já:t

∅-∅-∅-la'h-yuj-li' tza'má tzu'ma'já:t
3obj-sg.obj-3sg.sbj-all.appl-go.down-pfv that girl
‘He came down to the girl.’ (MR)

(36) nakila'ha'nkán nakinchík nalhu:wa:wa'yankán

na-kin-∅-la'h-a'n-kan nak=kin-chik na-lhu:wá'-wa'yán-kan
fut-1obj-sg.obj-all.appl-go-idf loc=1po-house fut-much-eat-idf
‘They are going to come to me at my house and they will feast.’ (PS)

The first example is from a story in which a man is working on the roof of a house and
comes down to speak with a pretty girl. In the second example, the Goal participant
is expressed by the first-person object prefix kin-. Note that the clause also contains a
locative adverbial nakinchík ‘at my house’, which is the preferred way of expressing
inanimate destinations with verbs of motion, as in (37):

(37) ka'ná:j nakXico

i'k-a'n-ya:-uj nak=Xico
1sg.sbj-go-ipfv-1pl.sbj loc=Xicotepec
‘Weexcl are going to Xicotepec.’ (BC)

The verb in (37) is a'n ‘X goes’. The destination is expressed as an adverbial, but does
not constitute a syntactic or semantic argument of the verb either here or in (36) above.

Table 15: la'h- Goal of Motion.

la'hchá:'n ‘X arrives at G over there’ (< cha:'n ‘X arrives there’)

la'hchín ‘X arrives at G here’ (< chín ‘X arrives here’)
la'hmín ‘X comes to G’ (< min ‘X comes’)
la'htaa'hcho'hó: ‘X goes for a walk to G’ (< taa'hcho'hó: ‘X goes for a walk’)
la'htakút- ‘X crosses the river to get G’ (< takút- ‘X crosses the river’)
la'hxta'ya'hnán ‘X glides down to G’ (< ta'ya'hnán ‘X glides’)
la'hyúj- ‘X comes down to G’ (< yuj- ‘X descends’)
la'há'n ‘X goes to G’ (< a'n ‘X goes’)

The motion verbs a'n ‘X goes’ and min ‘X comes’ also give rise to 11 forms where the
applied object expresses a Target towards which the Actor directs the action but which
is not actually reached or affected by it. These verbs are all based on a combination
of a'n or min with a part prefix, as seen in Table 16. As with the forms in Table 15, the
applied objects found with these verbs are generally animate.
134 David Beck

Table 16: la'h- Target.

la'ha'hamín ‘X cocks X’s ear to listen to G’ (< a'hamín ‘X cocks X’s ear to listen’)
la'hlakamín ‘X faces G over here’ (< lakamín ‘X faces this way’)
la'hlakaá'n ‘X faces G over there’ (< lakaá'n ‘X faces that way’)
la'hmakamín ‘X throws Y to G’ (< makamín ‘X throws Y’)
la'hmaká:'n ‘X sends Y to G’ (< maká:'n ‘X sends Y’)
la'hxaka:lakamín ‘X chews G out’ (< xaká: ‘X is angry’ + lakamín ‘X faces this way’)

There are eight verbs (4 independent stems) where the allative applies the notion of
target in the metaphorical sense of Target of Emotion (Table 17): the applicative in these
forms has the effect of adding a participant towards which the feelings of the Actor are
directed.

Table 17: la'h- Target of Emotion.

la'hka'tzán ‘X is jealous over G’ (< ka'tzán ‘X feels pain’)

la'hkalhwán ‘X weeps for G’ (< kalhwán ‘X weeps’)
la'hpuwán ‘X wants G’ (< puwán ‘X thinks’)
la'hsi:'tzí: ‘X feel resentment or envy over G’ (< si:'tzí: ‘X feels anger’)

Another group of 15 verbs (7 independent stems) is based on a set of postural and

configurational roots. With these, the allative generally adds a Terminal Point for the
motion leading up to the stated configuration (Table 18). Unlike the Goals of Motion,
Terminal Points are rarely expressed in the clause, although there are a few examples
with overt objects like those in (38) and (39):

(38) i'kla'htakilhpu:tálh kintumí:n, i'ksáka

i'k-∅-∅-la'h-takilhpu:ta-lh kin-tumí:n
1sg.sbj-3obj-sg.obj-all.appl-bend.over-pfv 1po-money
i'k-∅-∅-sak-a
1sg.sbj-3obj-sg.obj-pick-ipfv
‘I bent over for my money, I pick it up’ (PS)

(39) kala'htalakapá'jtzo:' milakasmú:', palh já:xku: nala'has’awikána'

ka-∅-∅-la'h-talakapá'jtzu:' min-lakasmú:' palh ja:-xku:
opt-3obj-sg.obj-all.appl-get.closer:2sg.sbj:pfv 2po-girlfriend if neg-conj
na-la'ha-s’awi-kan-a'
fut-face-defeat-idf-ipfv:2sg.sbj
‘Get close to your girlfriend! If not, then they will win her away from you.’ (LB)
 5 Applicatives in Upper Necaxa Totonac 135

More commonly, the Terminal Point, being third person, is ∅ and is understood from
context, as in (40):

(40) na'kla'htama'hajú: nakincazuela

na-i'k-∅-∅-la'h-tama'haju: ∅ nak=kin-cazuela
fut-1sg.sbj-3obj-sg.obj-all.appl-immerse.hand 3 loc=1po-cooking.pot
‘I’m going to put my hand in the pot to reach it.’ (RM)

In (40) there is an understood ‘it’ the speaker is reaching for; the nak= phrase defines a
wider location into which the speaker is reaching, but does not express the Goal itself.
The contrast is clearer if we compare this sentence to the same verbal base without the
applicative:

(41) katama'háju:' namibolsa

ka-tama'haju:' nak=min-bolsa
opt-immerse.hand:2sg.sbj:pfv loc=2po-pocket
‘Put your hands in your pockets!’ (PS)

The verb in (41), tama'hajú:' ‘X immerses X’s hands’ (< ta- ‘inchoative’ + ma'ha- ‘hand’ +
-ju: ‘be down in’), is intransitive and the nak= phrase simply defines the locus of the
immersion rather than a Goal of motion.

Table 18: la'h- Terminal Point.

la'htakilhpu:tá ‘X bends over to reach G’ (< takilhpu:tá ‘X bends over’)

la'htalakapa'jtzú ‘X moves closer to G’ (< talakapa'jtzú ‘X moves closer’)
la'htalakatzunají: ‘X gets closer to G’ (< talakatzunají: ‘X gets closer’)
la'htama'hajú: ‘X reaches in for G with one’s hand’ (< tama'hajú: ‘X reaches into a container’)
la'htanú: ‘X is inside G’ (< tanú: ‘X goes inside’)
la'htawaká'lh ‘X climbs up to G’ (< tawaká'lh ‘X goes up high’)

3.5 pu:- ‘containing instrument’

The containing instrumental applicative, puː-, appears in 79 entries in the database

forming part of 73 independent derivations. The applied object of puː- (N) expresses
a container, enclosing object, or container-like device that surrounds either the Actor
or the Undergoer/Patient. Like the applied object of li:-, the object of puː- is used as an
Instrument/Means of realizing the action expressed by the verb.
There are 37 lexical entries (35 independent derivations) where pu:- adds a Con-
taining Instrument (Table 19). The necessity that the applied object be container-like
imposes restrictions on the verb—for instance, pu:chi'pá ‘X traps Y in N’ can be used only
136 David Beck

with container-like traps such as nets or baskets (but not snares), while pu:cha'panán
‘X grinds in N’ (< cha'panán ‘X grinds’) only applies to grinding in instruments such as
mortars, molcajetes, and metates, and not to grinding, say, on the ground with a rock.

Table 19: pu:- Containing Instrument.

pu:cha'panán ‘X grinds in N (e.g., mortar)’ (< cha'pa ‘X grinds Y’ + -nan ‘antipsuppr’)

pu:chi'pá ‘X traps Y in N (e.g., net)’ (< chi'pá ‘X traps Y’)
pu:kukta'lá ‘X hits Y with a stone thrown in N (e.g., sling)’ (< kukta'lá ‘X hits Y with a stone’)
pu:la'hchulú:t- ‘X cools Y by pouring it out of N’ (< la'hchulú:t- ‘X cools liquid Y by pouring it’)
pu:la'hs’á't- ‘X skims froth off boiling cane syrup with N’ (< la'hs’á't- ‘X skims froth off boiling cane syrup’)
pu:li:mín ‘X brings Y inside N’ (< li:mín ‘X brings Y’)
pu:lé:n ‘X takes Y inside N’ (< le:n ‘X takes Y’)
pu:ma:wí: ‘X feeds Z to Y in N (e.g., bowl)’ (< ma:wí: ‘X feeds Z to Y’)
pu:skúj- ‘X works in N (work clothes)’ (< skuj- ‘X works’)
pu:wa'yán ‘X eats from N (dish)’ (< wa'yán ‘X eats’)

Because this type of instrument is generally third person and inanimate, and is rarely
plural, there are very few sentences like (42) that show overt agreement with the applied
object:

(42) ah'tú' tsa'ká:t i'ka:ta:'pu:la'hmakamílh chiwíx kistánku' tza'má chi'xkú'

ah'-tu' tsa'ká:t i'k-∅-ka:-ta:'-pu:-la'h-makamin-lh
clf-two sling 1sg.sbj-3obj-pl.obj-cmt.appl-cinst.appl-all.appl-direct-pfv
chiwíx kin-stánku' tza'má chi'xkú'
stone 1po-brother that man
‘I and my brother threw stones at that man with two slings.’ (LB)

Examples like (43), which show agreement with some other object, are much more
common:

(43) kintapu:lé:lh i'xkuxta:lhká'n

kin-∅-ta-pu:-le:n-lh i'x-kuxtá:lh-ka'n
1obj-sg.obj-3pl.sbj-cinst.appl-take-pfv 3po-sack-pl.po
‘They carried me in their sack.’ (PS)

Here, the controller of agreement is the primary object, expressing what is carried,
rather than the secondary applied object, the container.
Another semantic role (18 forms, 18 independent derivations) associated with pu:-
is that of Conveyance (Table 20). In these forms, the applicative combines with a verb of
motion to express a Vehicle (44), Conduit (45), or Path (46):
 5 Applicatives in Upper Necaxa Totonac 137

(44) pu:mílh pu:lháuj

∅-∅-∅-pu:-min-lh pu:lháuj
3obj-sg.obj-3sg.sbj-cinst.appl-come-pfv vehicle
‘He came in a car.’ (RM)

(45) tubo pu:mimá:lh xka:n

tubo ∅-∅-∅-pu:-min-ma:lh xka:n
pipe 3obj-sg.obj-3sg.sbj-cinst.appl-come-prog water
‘The water comes through pipes.’ (CF)

(46) xi:wán ja: pu:mílh tej naka:'xa:wátna' milh

xi:wán ja: ∅-∅-∅-pu:-min-lh tej nak=ka:'-xa:wát-na'
Juan neg 3obj-sg.obj-3sg.sbj-cinst.appl-come-pfv path loc=plc-corn-pl
∅-min-lh
3sg.sbj-come-pfv
‘Juan didn’t come on the road, he came through the cornfield.’ (CF)

Note that all three of these examples are based on the same verb, which shows us that
each of the three specific roles (Vehicle, Conduit, Path) played by the new participant in
the event is a context-specific interpretation of a single, vaguer, semantic role (Contain-
ing Instrument).

Table 20: pu:- Conveyance.

pu:á'n ‘X rides in N’ (< a'n ‘X goes’)

pu:mín ‘X comes inside N’ (< min ‘X comes’)
pu:paxyalhnán ‘X goes for a ride on N’ (< paxyalhnán ‘X goes for a stroll’)
pu:taa'hapú: ‘X goes downhill in N’ (< taa'hapú: ‘X goes downhill’)
pu:taa'kxtú ‘X goes upstream in N (boat)’ (< taa'kxtú ‘X goes uphill’)
pu:takút- ‘X crosses river in N’ (< takút- ‘X crosses river’)
pu:xki'wa'hnán ‘X swims using N as a float’ (< xki'wa'hnán ‘X swims’)
pu:yúj- ‘X comes down using N (e.g., ladder, rope)’ (< yuj- ‘X descends’)

In several forms (18 lexical entries, 15 independent derivations), pu:- increases the
valency of the base by adding a Container that is not an Instrument, but instead is
simply an object within which the action of the verb is accomplished (Table 21). In some
cases, like the instance of pu:ma:ska:kí: ‘X puts Y into N to dry it’ (< ma:ska:kí: ‘X dries
Y’) in (47), the applied object might be construed as a Containing Instrument, but this is
strictly a contextual interpretation of the general semantic role of Container:
138 David Beck

(47) pu:ma:ska:ki:kán pi'n pu:skuyún

∅-∅-pu:-ma:-ska:k-i:-kan pi'n pu:skuyún
3obj-sg.obj-cinst.appl-caus-dry-caus-idf chili smoking.rack
‘They dry the chilies on the smoking rack.’ (CF)

In (47), N is a rack used specifically for smoking chiles and so can be understood as an
Instrument for drying; however, the same verb can be used with other types of objects
(tarps, bowls) which would simply be locations. Thus, pu:ma:ska:kí: is less selective than
verbs like pu:cha'panán ‘X grinds in N (e.g., mortar)’ from Table 19 above, and the new
semantic role specified by the applicative is simply a Container.

Table 21: pu:- Container.

pu:cha'hanán ‘X washes nixtamal in N’ (< cha'hanán ‘X washes nixtamal’)

pu:helhtawahá: ‘X studies in N (school)’ (< helhtawahá: ‘X studies’)
pu:ma:ska:kí: ‘X puts Y into N to dry it’ (< ma:ska:kí: ‘X dries Y’)
pu:mojó: ‘X puts Y inside N’ (< mojó: ‘X puts Y in’)
pu:pa:tí: ‘X suffers inside of N’ (< pa:tí: ‘X suffers’)
pu:pú'x- ‘X picks Y and puts it in N’ (< pu'x- ‘X picks Y’)
pu:tastó'h- ‘X gets together inside of N’ (< tastó'h- ‘X gathers’)
pu:tojó: ‘X gets inside of N’ (< tojó: ‘X is immersed or contained’)
pu:waká'lh ‘X is hanging inside N’ (< waká'lh ‘X is up high’)
pu:wilí: ‘X puts Y inside of N’ (< wilí: ‘X places Y’)

pu:- is unique among the UNT applicatives in that it has an obvious cognate in the part
prefix pu:- ‘vagina; container’. Like all part prefixes in Totonacan (Levy 1999), the part
prefix pu:- acts as a limitative (Mel’čuk 1994), serving to delimit what Langacker (1991)
refers to as the “active zone” of one of the event participants—that is, the subpart of
that participant most affected by the action. We can see the typical use of part prefixes
in (48):

(48) pu:masli'tzá' wa:má ha:'x

pu:-mas-li'=tzá' wa:má ha:'x
ctnr-rot-pfv=now this gourd
‘This gourd is rotten on the inside.’ (CF)

In the verb pu:mas- ‘X is rotten on the inside’ (< mas- ‘X rots’), puː- merely delimits the
active zone (the interior) of the entity that is rotting (the gourd), but does not add a new
semantic role to the event. The verb with and without pu:- is monovalent.
Because of these differences in meaning and syntactic effect, the two pu:- prefixes
have to be treated as separate morphemes. This predicts that the two can appear in the
same verb form. Consider the examples in (49)–(51):
 5 Applicatives in Upper Necaxa Totonac 139

(49) nai'klhká: wa:má kí'wi'

na-i'k-∅-∅-lhka: wa:má kí'wi'
fut-1sg.sbj-3obj-sg.obj-measure this tree
‘I’m going to measure (the length of) this wood.’ (RM)

(50) i'kpu:lhká:lh kinkú'xi'

i'k-∅-∅-pu:-lhka:-lh kin-kú'xi'
1sg.sbj-3obj-sg.obj-ctnr-measure-pfv 1po-corn
‘I weigh my corn.’ (RM)

(51) na'kpu:pu:lhká: pu:pu:lhká:n i kinkú'xi'

na-i'k-∅-∅-pu:-pu:-lhka: pu:pu:lhká:n i kin-kú'xi'
fut-1sg.sbj-3obj-sg.obj-cinst.appl-ctnr-measure scale jnct 1po-corn
‘I’m going to weigh my corn in a scale.’ (RM)

Example (49) shows the bivalent verb lhkaː ‘X measures Y’; in (50) we have pu:lhkaː ‘X
weighs Y’, which is also bivalent and bears the part prefix pu:- in recognition of the fact
that weighing something usually entails placing it in a bag or container of some kind (in
the time before mechanical scales).4 The verb in (51), pu:pu:lhkaː ‘X weighs Y in N’, on
the other hand, has both the partonym pu:- and the pu:- applicative. The addition of the
applicative pu:- makes it trivalent, and only with this form can the instrument used to
do the weighing, the scale, be expressed.
The relationship between the pu:- part prefix, the pu:- applicative, and the non-in-
strumental valency-increasing pu:- is not only an interesting example of grammat-
icalization, but is also significant for family-internal reconstruction. While li:- is the
principal instrumental applicative in the Totonac branch of the family, pu:- is the more
frequent instrumental in the Tepehua branch (Beck 2012). In one of the three Tepehua
languages, Huehuetla, the general applicative is pu:- and the cognate of liː-, ɬiː-, is more
limited (Smythe Kung 2007)—as it is in Tlachichilco Tepehua, which has both pu:- and
pa:- (‘belly’) instrumentals in addition to ɬi:- (Watters 1987). The third Tepehua language,
Pisaflores, uses both pu:- and ɬi:- as general instrumentals (J. Watters, p.c.), suggesting a
complex history for instrumental applicatives in the family.

4 This use of the part prefix is not, strictly speaking, limitative in the sense of defining an active zone
on the object being measured, so much as it defines a type of spatial domain in which the action takes
place, also a common function of part prefixes.
140 David Beck

4 Conclusion
Based on the questionnaire provided for this volume, UNT presents the following profile:

Morphology
1.1 The main AC is marked by affixation.
1.2 not relevant
1.3 There is virtually no allomorphy affecting any of the applicatives.
1.4 Applicativized verbs are inflectionally identical to basic verbs.

Syntax
2.1 The applied phrase is realized as a secondary object.
2.2 The syntactic status of the applied phrase’s companion arguments does not change
between the BC and the AC.
2.3 There are no restrictions on the stacking of applicatives or their combination with
voices.
2.4 ACs do not form a special verb class.
2.5 Applied objects are potential controllers of agreement, subject to conditions based
on person and discourse status.
2.6 Applied objects are accessible to relativization and are accessible to linearization
operations used to express Information Structure.

Semantics
3.1 Each applicative assigns a specific semantic role or one of a set of semantically
related roles to the applied phrase.
3.2 Each of the four applicatives is the only way the semantic role they are associated
with can be expressed.
3.3 not relevant
3.4 not relevant
3.5 The role of applicatives is to create lexical items expressing events involving
a certain set of participants, so the choice between the AC and the BC is mean-
ing-driven rather than discourse-sensitive. Applied objects are accessible to line-
arization operations used to express Information Structure and topic continuity.

Lookalikes
not relevant
 5 Applicatives in Upper Necaxa Totonac 141

Abbreviations
all.appl allative applicative
antipsuppr suppressive antipassive
ben.appl benefactive applicative
cinst.appl containing instrumental applicative
cmt.appl comitative applicative
conj conjunction
caus causative
ctnr container
dcs decausative
deb debitative
dsd desiderative
excl exclusive
fut future
hrel human relative
idf indefinite voice
ipfv imperfective
inst.appl instrumental applicative
intj interjection
intns intensive
loc locative
neg negative
nmz nominalizer
obj object
opt optative
pfv perfective
pl plural
po possessive
prog progressive
pst past
quot quotative
rcp reciprocal
rpt repetitive
sbj subject
sg singular
subst substitutive
trz transitivizer
- affix boundary
= clitic boundary

References
Beck, David. 2004. Upper Necaxa Totonac. Munich: LINCOM.
Beck, David. 2012. Apéndice: Tablas de morfología comparativa. In Paulette Levy & David Beck (eds.), Las
lenguas totonacas y tepehuas. Textos y materiales para su estudio, 587–596. Mexico City: UNAM Press.
142 David Beck

Beck, David. 2016. Uniqueness and grammatical relations in Upper Necaxa Totonac. Linguistics 54(1). 59–118.
Hopper, Paul J. & Sandra A. Thompson. 1980. Transitivity in grammar and discourse. Language 56(2).
251–299.
Langacker, Ronald W. 1991. Foundations of Cognitive Grammar, Volume II: Descriptive application. Stanford:
Stanford University Press.
Levy, Paulette. 1999. From “part” to “shape”: Incorporation in Totonac and the issue of classification by
verbs. International Journal of American Linguistics 65(2). 127–175.
Levy, Paulette. 2002. El aplicativo dativo/benefactivo en totonaco de Papantla. In Zarina Estrada Fernández
& Rosa María Ortiz Ciscomani (eds.), Memorias del VI Encuentro Internacional de Lingüística en el
Noroeste, 175–194. Hermosillo: Universidad de Sonora.
Mel’čuk, Igor A. 1994. Cours de morphologie générale, vol. 2. Montreal: Les Presses de l’Université de
Montréal.
Smythe Kung, Susan. 2007. A descriptive grammar of Huehuetla Tepehua. Austin: University of Texas
dissertation.
Watters, James K. 1988. Topics in Tepehua grammar. Berkeley: University of California dissertation.
Marisa Censabella
6 Applicatives in Toba/Qom (Guaykuruan)
Abstract: Toba or Qom l’aqtaqa (T/Q) is a polysynthetic Guaykuruan language spoken
mainly in the Argentinean Chaco region. It lacks morphological case marking and adpo-
sitions; instead, the role of the unmarked NPs is either specified by the lexical meaning of
the verb or encoded via applicativization. There are twelve applicative verbal suffixes:
five with a locative meaning, four with a goal/directional one, and three that encode
benefaction, reception, and accompaniment, respectively. Whether valency-increasing
or valency-re-arranging, applicativization introduces applied arguments as core P argu-
ments (as seen in number agreement and constituent order patterns, as well as in topi-
calization and its consequences for coordination and subordination pivots). When used
with transitive roots/bases, applicatives do not add a third argument but change the
meaning expressed by the verb, allowing a different semantic role from the one spec-
ified in the verbal root/base. T/Q is a beneficiary-prominent language: whenever the
speaker wants to highlight the benefit that a non-subject participant gets through the
verbal event, that participant will be encoded as a beneficiary. Serial verb constructions
seem to be the origin of applicatives, and the grammaticalization processes undergone
by the latter—in coexisting stages of evolution—show plausible routes for pragmatic
morphology.

1 Introduction
1.1 The language

Toba or Qom l’aqtaqa1 is a Guaykuruan language spoken mainly in the Argentinean

Chaco region (Chaco, Formosa, and Salta provinces) and in some important southern
cities, such as Rosario, Santa Fe and the surroundings of Buenos Aires. The latest pop-
ulation census estimates that Toba/Qom (T/Q) people in Argentina approaches 126,967
(INDEC 2012), whereas 2,057 live in Paraguay (DGEEC 2013).2 In rural settings, T/Q is
often the first language that children learn at home, while in large cities, a process of
displacement in favor of Spanish is observed. This process involves different commu-

1 Qom/qoml’ek/qomlashe and qom l’aqtaqa are the endogenous ethnonyms (‘people/person.m/person.f’)

and glottonym (‘the language of the People’) known in the ethnographic and linguistic literature as Toba
language and Toba people, respectively.
2 Since national population censuses are carried out every 10 years in Argentina, the figures mentioned
here are outdated, because the 2020 National Census was suspended due to the COVID-19 pandemic
disease.

https://doi.org/10.1515/9783110730951-006
144 Marisa Censabella

nicative competences within the indigenous language, according to biographical trajec-

tories, even among speakers of the same age living in the same community (Censabella
2009a: 167).
The glottonyms Toba, Qom, or Qom l'aqtaqa3 are used in linguistic studies to refer
to a set of related varieties that belong to the Guaykuruan linguistic family (together
with Mocoví, Pilagá, Caduveo, and the extinct Abipón and Mbayá; Tovar and Larrucea
Tovar 1984). The language is catalogued as tob (ISO 639–3) and toba1269 (Glottolog 4.5).
T/Q presents the largest number of linguistic and sociolinguistic studies within the men-
tioned linguistic family.4 T/Q speakers—as well as linguists (Klein 1981)—identify four
main dialects in Province of the Chaco in Argentina: lañaɢashek, noɁolɢranaq, rapigem-
lɁek, and takshek, all mutually intelligible. In this chapter, clauses from the three first
mentioned dialects are used.5

3 The phoneme inventory of T/Q is shown below. The bracketed graphemes are the ones used in this
paper.

labial alveolar pre-palatal palatal velar uvular glottal

plosives p t ʧ <ch> k q Ɂ
voiceless frict. s ʃ <sh> h
voiced frict. ʒ ɣ <g> ʁ <ɢ>
laterals l ʎ <ll>
semi-con. w j <y>
tap ɾ <r>
nasals m n ɲ <ñ>

front back
close i o
open e a

4 For T/Q reference grammars, Buckwalter (1980), Klein (1981), Censabella (2002), Messineo (2003),
Carpio (2012), Cúneo (2013), González (2015), and Zurlo (2016a–b). For sociolinguistic aspects, see Bigot
(2007); Censabella (2009a), Hecht (2010), and Medina (2017). There are few studies on grammatical fea-
tures—Avellana (2012, 2013); Avellana and Dante (2013); Censabella (2015); Zurlo (2013, 2016a)—and
discourse issues—Zurlo and Censabella (2014)—related to Spanish in contact with T/Q, and even fewer
about the influence of Spanish on T/Q—González and Censabella (2009); González (2013a) (the list is
not exhaustive)—. For comparative studies of Guaykuruan languages in some grammatical domains, see
Carpio and Mendoza (2018) and Carpio (2018); for comparative studies of Guaykuruan and Mataguayan
languages, see Censabella and Terraza (2009), Vidal (2010), Messineo, Carol and Klein (2016); and for com-
parative studies of Guaykuruan and Tupi-Guaraní languages, see Carpio, González, and Mendoza (2021).
5 Current T/Q linguistic varieties’ names are related to a non-linguistic denomination system composed
of traditional demonyms, cardinal points, and names of ecological regions. According to Braunstein and
Miller (1999), T/Q dialectal varieties relate to local groups of extended families or bands, and to tribes
(conceived as groups of bands); following Mendoza (1999), the bands were identified by other groups
with a proper name. In the ethnographic literature written in Spanish, these demonyms are called par-
 6 Applicatives in Toba/Qom (Guaykuruan) 145

1.2 Brief typological sketch

T/Q is a polysynthetic language with non-rigid word order (mainly VS, AVO, and OVA
when O is pronominal), head marking, a verb-noun opposition, and an alienable-inal-
ienable possession distinction. T/Q lacks an adjectival class of words, and also lacks both
morphological case marking and adpositions; instead, it has unflagged NPs whose role
is specified either by the lexical meaning of the verb or encoded via applicativization.
T/Q displays four types of nominal categorization: nominal classes or “gender”, nominal
classifiers (i.e., derivational morphemes that indicate the shape of objects encoded as
deverbal nouns), possessive classifiers, and demonstrative or deictic classifiers. The
expression of number in nouns combines a category of morphological number (sin-
gular and plural) with one of collectives and another of distributives; dual number is
marked syntactically.
The verb system is organized in an active-middle opposition for all persons (Table 1).
The verbal lexicon is accordingly divided into three main groups:
(i) active tantum verbs, which occur only with the active voice marker (e.g., rkeɁe ‘s/he
eats’, roɁoche ‘s/he sleeps’, hek ‘s/he goes’);
(ii) media tantum verbs, which occur only with the middle voice marker (e.g., nachel ‘s/
he bathes’, niyin ‘s/he cries’), and

cialidades and nacionalidades—nationalities—by T/Q people. For a semasiological analysis of the most
used demonyms see Censabella (2002, 2009); Gordillo (1992), de la Cruz (1995) and Mendoza (2002) list
and locate the bands at the end of 19th century as they were mentioned in the historical records and
contemporary T/Q oral narratives. lañaɢashek (‘sth./sb. related to lañaɢa [= dry land]’) designates an
ecological region with that specific characteristic, not a particular region in the map. NoɁolɢranaq refers
the band of the Roosters (cf. oleɢra ‘rooster’), usually located in the north-center of the Chaco province,
and rapigemlɁek ‘the man from the region where the land touches the sky’ (cf. pigem ‘sky’) refers to the
western region of the province where the land smoothly rises to the western mountains. Finally, takshek
(‘sth./sb. related to tageñi [= east]’) refers to a huge region towards the east of the Chaco region, along the
Paraguay and Paraná rivers. Migration to bigger cities in Chaco Province and settlement in suburban
areas since the 1960’s, in different and constant waves, have weakened the differences within the vari-
eties, although some older speakers still know the demonyms system of ancient neighbors and are able
to identify the phonological and lexical differences between some varieties. Some syntactic differences
between the varieties show different stages of grammaticalization processes and different contact drifts
in the verbal alignment and number systems, causative constructions, and uses of pragmatic particles.
No in-depth dialectological studies have been made; nevertheless, Censabella’s (2002) reference gram-
mar—mainly focusing on the rapigemlɁek syntax—presents phonological differences between a small
sample of noɁolɢranaq speakers born in Pampa del Indio (Chaco), lañaɢashek speakers born in Colonia
Aborigen (Chaco), and takshek speakers living in Misión Tacaaglé (Formosa). González (2015) specifical-
ly studies the phonology, morphology, and syntax of the takshek variety. Carpio (2012) studies the variety
spoken by Western Tobas or ñachilamoleɁk, which is closer to Pilagá. Messineo (2014: 19, 22) provides
maps with regions inhabited by Qom people in Argentina, as well as the areas covered by the four main
dialects in Chaco Province.
146 Marisa Censabella

(iii) a large group of verb roots and bases that can take either marker (e.g., iyo ‘s/he
washes sth./sb.’ vs. nyo ‘s/he bathes’; parenaɢan ‘s/he jumps’ vs. nparenaɢan
‘s/he jumps willingly’; yawan ‘s/he sees, knows sth.’ vs. nawan ‘s/he watches sth.’,
yapaɢagen ‘teaches sb.’ vs. napaɢagen ‘s/he studies’, imen ‘s/he sells sth.’ vs. n-men
‘s/he buys sth.’; iro ‘s/he takes sth., herds [animals]’ vs. nro ‘s/he brings sth., herds
[animals] for him, he is the owner of the herd’.

The middle voice participates in the same syntactic processes as the active voice, that is,
transitivity alternations or adjustments which increase syntactic valence as applicativi-
zation and causativization, as well as valency decrease adjustments such as impersonal
passive, reflexive, and reciprocal constructions.6 Further, the active voice shows a split
for the third person, used to distinguish actions, events, different kinds of processes,
and states. With non-derived verbs, the third personal marker i- appears with canoni-
cal transitive verbs and indexes the A argument; the complete clause requires plain or
pronominal A and P arguments,7 with or without applicatives. The personal markers
r-, ∅-, t-, and w- indexes the S argument, and they can appear in monovalent intran-
sitive clauses, but also with an intransitive verb plus an applicative. Thus, the active
voice shows a split alignment system based on person hierarchy: nominative-accusative
(A=S≠P) between speech act participants (1st and 2nd person singular and plural), and a
tripartite system (A≠S≠P) with splits in the S argument with the 3rd person singular and
plural. Furthermore, a small number of verbal roots show an active-inactive alignment
system, not verified in all varieties, and mainly used for 1st person singular and plural.

6 In previous studies, personal indexes shown in Table 1 were interpreted as morphemes encoding two
grammatical notions: a personal index plus the orientation of the event. The latter was interpreted as:
towards outside vs. towards inside (Buckwalter 1980), non-ad-corporeal vs. ad-corporeal action (Klein
1981), and centrifugal vs. centripetal movement (Bigot 1994). Following Benveniste (1966[1950]) and
Kemmer (1993), Censabella (1998, 2002) proposes an active-middle voice opposition in T/Q; the ad-cor-
poreal or centripetal movement meanings are subsumed in more general semantic features allowed
by the middle voice, in which the S/A arguments are semantically interpreted as affected, receiving a
benefit, or emotionally engaged while performing the event encoded by the verb. Zurlo (2016a), follow-
ing Klaiman (1991), Kemmer (1993) and Creissels (2006, 2007), proposes that T/Q shows a “basic voice”
system (viz., having active and middle voices) characterized by the following: (i) the verbal lexicon is
organized into three verb classes (invariant active, invariant middle, and alternate); (ii) within the alter-
nate verb class, which has the highest number of items, the middle markers are related to intransitive
meanings, but not exclusively, because T/Q uses a significant number of alternate transitive verbal roots
and bases and full P arguments with middle voice markers; (iii) the middle markers also encode differ-
ent classes of atelic, non-punctual, irrealis or future meanings; and (iv) middle markers encode specific
meanings that vary from one language to another but are similar in considering the logical subject as
the principal locus of the denoted event.
7 S and A personal index prefixes (Table 1 and Figure 1) are obligatory in well-constructed verbal syn-
tagms. Plain (lexical or full pronominal) S or A are not obligatory; they are used with pragmatic func-
tions.
 6 Applicatives in Toba/Qom (Guaykuruan) 147

Table 1: Active and middle voice personal markers.

Active voice Middle voice

1sg s- ñ-
2sg aw- an-
3sg i-; r-; ∅-; t-; w- n-
1pl s. . .q ñ. . .q
2pl qaw. . .i qan. . .i
3pl i-. . .Ɂ; r-. . .Ɂ; -. . .Ɂ; t-. . .Ɂ; w-. . .Ɂ n-. . .Ɂ

T/Q does not have tense; rather, it shows an aspectual system consisting of an obligatory
perfective-imperfective opposition for all verbs, with other aspectual markers distrib-
uted in different semantic fields. There is verbal number, and the language does not
have converbs. T/Q also possesses a rich locative- and directional-marking system with
different functional slots in the verb complex. In these two paradigms, morphemes of
one set operate as applicatives that increase or re-arrange transitivity, while morphemes
of the other set have semantic functions different from valency and voice change. Both
paradigms can co-occur in the verbal syntagm (cf. § 2). Valency changes are expressed
by derivational and syntactic mechanisms, the former in causative and antipassive con-
structions and the latter in applicative and non-promotional passive constructions.8
The gray columns in Figure 1 show the minimal obligatory templatic slots that a
well-formed verb form must contain: a portmanteau morpheme encoding the S/A per-
sonal index and the category of voice (cf. Table 1), the verbal root, a marker specifying
person and number of the S/A, and an aspectual marker (unmarked perfective, imper-
fective progressive -tak, and imperfective continuous -ta).
Verbal number
person index

(plural only)
Voice + S/A

CAUS/ANTIP/
Derivational

derivational
morphemes

Applicative
Verbal root

directional
RECP/REFL
S/A index

Locative /

P number
prefixes

Aspect
IMPRS

FACT

Verbal prefixes Lexical base Verbal suffixes

Figure 1: Verbal slots in T/Q.

To better understand the functions of applicatives and some lexicalization processes, it

is convenient to distinguish the lexical from the syntactic areas in T/Q verbs. The second
segment of 1st- and 2nd-plural discontinuous indexes clearly separate the lexical base

8 This typological sketch and the rest of the chapter is based on Censabella (2002, 2006a, 2006b, 2007,
2008, 2010, 2018), Carpio (2007, 2012), Carpio and Censabella (2010), González (2010a, 2010b, 2013b,
2016a, 2016b), Zurlo (2011, 2016b, 2019), and Zurlo and Censabella (2013).
148 Marisa Censabella

from the verbal suffixes; this boundary does not exist for the 1st-, 2nd- or 3rd-singular
indexes. The lexical base area contains, apart from the verbal root, unfrequently a der-
ivational prefix, and usually derivational morphemes that increase transitivity (viz.,
different types of causatives) or reduce it (e.g., the antipassive). One of these deriva-
tional morphemes, -aɢan, is a sort of transitivity operator, which increases or reduces
the number of core arguments that the base lexical root specifies. The causative deri-
vational morphemes used in this slot encode different types of causatives with instru-
mental, intentional direct and intentional indirect causation, and a non-intentional or
non-human direct causation; these morphemes can co-occur.9 Examples (1)–(3) show
the boundaries between derivational and syntactic areas that 1st- and 2nd-plural dis-
continuous morphemes distinguish inside the verb form.

(1) a. i-nan-tak na l-Ɂaɢanaɢat10

3tr-roll.up-prog dem 3poss-rope
‘S/he is rolling up her/his rope.’
b. se-nan-aq-tak na qar-Ɂaɢanaɢat
1pl-roll.up-1pl-prog dem 1pl.poss-rope
‘We are rolling up our rope.’

(2) a. se-men-aq-tak a-so npaqta-pi

1pl-exchange-1pl-prog f-dem basket-coll
‘We are selling baskets.’
b. se-men-aɢan-aq-tak qoɁollaɢa
1pl-exchange-antip-1pl-prog long.time.ago
‘We were selling in those days (we used to sell for a living).’

9 Examples (i) and (ii) below show two and three derivational morphemes in the verb form, respective-
ly. A double and triple gloss line is added to better understand the derivational process.

(i) se-kag-aqt-aɢan-aq-tak
1pl-[be.broken-instr]-antip-1pl-prog
1pl-break-antip-1pl-prog
‘We are breaking (intr).’

(ii) se-kag-aqt-aɢan-aɢan-aq-tak a-so Ɂalo

1pl-[[be.broken-instr]-antip]-caus-1pl-prog f-dem woman
1pl-[break-antip]-caus-1pl-prog
‘We are making/ordering (tr) that woman to break (intr).’
10 The examples presented in this chapter were mainly obtained by elicitation techniques and from
narrative texts, collected and translated by the author, except where indicated otherwise. For expository
reasons, some sentences have been simplified, testing their acceptability with two T/Q consultants by
elicitation techniques.
 6 Applicatives in Toba/Qom (Guaykuruan) 149

(3) qoɁomi s-alekten-aɢan-aɢa-Ɂa so towe

pro.1pl 1pl-mix-antip-1pl-appl:all dem salt
‘We mix things with salt.’

2 Applicative morphology
T/Q distinguishes two paradigms of verbal morphemes; one encodes a) locative or
directional notions and the other encodes b) locative, directional and other func-
tional notions, as shown in Tables 2 and 3 and discussed in 2.1 and 2.2, respectively.
The latter behave as applicatives, increasing valency or re-arranging transitivity, while
the former modify the verbal semantics without an applicative function. Morphemes
from both paradigms can co-occur in the verbal syntagm; the functional slots occupied
by each one are represented in Figure 1 above.

2.1 Locative-directional non-applicative verbal suffixes

Table 2: Locative-directional non-applicative

morphemes in T/Q.

-ñi ‘down(wards)’
-shegem ‘up(wards)’
-som ‘to the water’
-wo ‘in(side)’
-wek ‘out(side), to an open space’

With movement verbs, these morphemes indicate the direction of movement (4b, c, d).

(4) a. se-parenaɢan b. se-parenaɢaa-ñi

1-jump 1-jump-l/d:down
‘I jump.’ ‘I jump down.’
c. se-parenaɢaa-shegem d. se-parenaɢaa-som
1-jump-l/d:up 1-jump-l/d:to.water
‘I jump up.’ ‘I jump into the water.’

With stative constructions, these morphemes indicate location, as seen in (5)–(6).

150 Marisa Censabella

(5) we-ta-wo ʒe noɢotole-k

3intr-go11-l/d:in dem child-m
‘The boy is inside.’ (inside his mother’s belly)

(6) we-ta-wek ra shiyaɢawa

3intr-go-l/d:out dem person
‘This person is naked.’

With intransitive or transitive verbal roots or bases, l/d morphemes do not increase
valency and are therefore not applicatives, at least in the studied varieties of T/Q lan-
guage.12 In (7), the verb root -maq ‘to push’ is transitive and the l/d marker neither
demands the incorporation of a new argument nor re-arranges the function of the
previous arguments; instead, the l/d marker—via its own semantics or with the co-
occurrence of other morphological categories—modifies the meaning of the VP, as is
clearly shown in Example (8b), or in Example (9a) when contrasted with (9c). Notice
that the process of lexicalization between the verb root and the l/d marker is still weak,
because the S/A 1pl index is located between the verb root and the l/d marker, as seen
in examples (9b) and (9d).13 Most l/d morphemes are grammaticalized nouns, and some
probable clines could be the following: l-awo ‘his/her family’ > -wo ‘into’; pigem ‘sky’
> -shegem ‘upwards’; ñi is a deictic demonstrative meaning ‘sitting, lean on’14 and the
likely origin of -ñi ‘down’.

(7) a. Ɂam y-amaq

pro.2 3tr-push
‘S/he pushes you.’
b. Ɂam y-amaɢ-awo
pro.2 3tr-push-l/d:in
‘S/he pushes you inside (the house, a place).’
c. Ɂam y-amaɢ-awek
pro.2 3tr-push-l/d:out
‘S/he pushes you outside (the house, a place).’

11 In Examples (5) and (6), stativity arises from the combination of the 3rd-person index w- and the
movement verb -ta ‘to go’ (cf. the 3rd-person index split in Table 1).
12 Calqued clauses from Spanish in elicitation sessions could give the impression of l/d argument incor-
poration, but this behavior is not corroborated in narrative or conversational T/Q texts.
13 Some fully lexicalized l/d markers are found in nouns or attributive constructions: ʒoɁoɢoñi ‘dawn’,
naloñi ‘frost’, rapaqñi ‘s/he has a fever’ / ‘it is hot’.
14 This demonstrative is used prototypically with animals in stand-up position (leaned on the four legs),
but also with objects which lay on the ground, such as houses or cooking pots with supports.
 6 Applicatives in Toba/Qom (Guaykuruan) 151

(8) a. ashek b. sheg-ewo

1.go 1.go-l/d:in
‘I go.’ ‘I go home/to my place.’

(9) a. se-lar-iñi a-na ltañi

1-move.the.soil-l/d:down f-dem pumpkin
‘I plant (inhume) the pumpkin.’
b. se-lar-aɢa-ñi. . .
1pl-move.the.soil-1pl-l/d:down
‘We plant . . .’
c. se-lar-shegem ʒe lewaɢyk
1mid-move.the.soil-l/d:up dem dead.body
‘I exhume the dead body.’
d. ñi-lar-aɢa-shegem. . .
1plm-move.the.soil-1pl-l/d:up
‘We exhume. . .’

Finally, l/d morphemes can co-occur with morphemes of the applicative set, as seen
in (10).

(10) ramaʒe r-awi-shegem-lek a-ñi qomaɁ

pro.3 3intr-arrive-l/d:up-appl:over f-dem mountain
‘He arrives upwards over the mountain.’

2.2 Applicative verbal suffixes

The morphemes in Table 3 behave syntactically as applicatives: when used, “[they

allow] participants not encoded as core terms in a base construction to be encoded
as such in the derived construction. More precisely, applicativization is an instance
of non-S/A nucleativization” (Zúñiga and Creissels, this volume). Examples in this
section illustrate the less grammaticalized uses of the locative and directional applica-
tive morphemes with intransitive and transitive verbs. Section 3 addresses the syntax
of applicative constructions (ACs); Section 4 outlines the semantic consequences of
applicativization on selected verbal roots/bases. Finally, Section 5 explains grammati-
calization processes observed for the ‘under’ locative and the ‘approaching’ goal/direc-
tional applicatives.
152 Marisa Censabella

Table 3: Applicative morphemes in T/Q.

Subtype basic other appl meaning P.pl315

semantic semantic morp allo
role roles heme morp
hs
locative location animate goal -lek -leg ‘over’ -te
(stative) /others -lge
-ek
animate goal/ -ʔot -ot ‘under’ -oɁ
beneficiary
animate goal -gi ‘against / (on the) -lo
side’
— -igi ‘inside’ -lo
directional inanimate — -ta ‘to the other side’ -lo
goal/source animate -ge ‘moving away’ -lo
moving goal
human goal -get ‘approaching’ -oɁ
/source
animate goal -sop ‘around’ -oɁ
animate goal -ʔa -a ‘to’ -lo
/others -ya
benefactive/ beneficiary involuntary -em -tem ‘for’ -a
malefactive /maleficiary causer -am
-om
transfer recipient -i ‘to’ -lo
comitative co-agent -eʔ ‘with’ -oɁ

Unlike the examples in the previous sub-section, where the addition of l/d morphemes
to a base clause (BC) does not require an obligatory P argument, the main syntactic
function of applicativization is shown in example (11b), where a locative argument is
required in the applicative clause (AC), and in (11c), which is ungrammatical.

(11) a. ramaʒe ∅-kewo-tak

pro.3 3intr-walk-prog
‘He is walking.’ (BC)
b. ramaʒe ∅-kewo-tag-elek ra noɁonaɢa
pro.3 3intr-walk-prog-appl:over dem field
‘He is walking over the field.’ (AC)
c. ✶ramaʒe kewotagelek

15 When the applied P is plural, number agreement is marked after the applicative, as shown in (45c).
 6 Applicatives in Toba/Qom (Guaykuruan) 153

The twelve applicative markers stand in complementary distribution in all studied T/Q
varieties: five with a locative meaning, four with a goal/directional meaning, and three
that encode benefaction, reception, and accompaniment, respectively. All applicatives
co-occur with active and middle personal indexes, but the following sub-sections show
all 3rd-person indexes in active and middle voice only for the locative applicative -lek,
due to space limitations. Other alternations will be presented in Section 4.

2.2.1 Locative applicative -lek (‘over’)

(12) a. we-ta-lek ñi ñigellaq

3intr-go-appl:over dem roof
‘S/he is over the roof.’
b. t-alek a-ñi qomaɁ
3intr-go.appl:over f-dem mountain
‘S/he climbs the mountain.’
c. somaʒe ∅-pe-lek a-na Ɂalwa
pro.3 3intr-step-appl:over f-dem land
‘He steps over the land.’
d. r-an-aɢaa-lge-te na-wa noɁonaɢa-ʒe
3intr-give-antip-appl:over-P.pl dem-dl field-pl
‘S/he plants the fields.’ (Buckwalter 1980: 19)

(13) a. a-so Nsogoj Ɂam y-amaq

f-dem cannibal[f] pro.2 3tr-throw/push
‘The cannibal woman pushes you.’ (BC)
b. a-so Nsogoj Ɂam y-amaɢ-alek ʒe leɁ
f-dem cannibal[f] pro.2 3tr-throw/push-appl:over dem ember
‘The cannibal woman pushes you over the ember.’ (AC)

(14) no-wir-añi ñi Dios,

3mid-arrive-l/d:down dem God
n-ahañi-ta-lek ra Resistencia
3mid-fell.down-cont-appl:over dem R.
‘God comes down, he is tilting over Resistencia (city).’

2.2.2 Locative applicative -Ɂot (‘under’)

(15) a. ñi qagesaq we-to-ʔot a-ra ipaq

dem ant 3intr-go-appl:under f-dem tree
‘The ant is under the tree.’
154 Marisa Censabella

b. r-ewaɢay-ot seʔeso qomaʔ

3intr-shelter-appl:under dem.pron stone/mountain
‘S/he shelters under that stone.’

2.2.3 Locative applicative -gi (‘[in]side / against / lean on’)

(16) Ferocho ∅-keta-gi na awyaq

F. 3intr-walk-appl:side dem forest
‘Ferocho walks in/by the forest.’

The examples in (17) show that the applicatives are frequently used to adapt the seman-
tic specification of the verbal base to allow for the application of a different semantic
role of the P argument. In (17a), the verbal base -waɢan selects a human or animate
entity to fulfill the P argument while in (17b) the applicative allows a prototypical inan-
imate patient to fulfill that function.

(17) a. i-waɢ-an so l-qaya

3tr-hand/fist-vblz dem 3poss-brother
‘S/he hits his brother.’ (BC)
b. i-waɢ-an-gi a-na lacampana
3tr-hand/fist-vblz-appl:side f-dem bell
‘S/he hits (on) the bell.’ (AC)

(18) y-aa-ta-gi ra l-qayk a-ka l-asogeɁ

3tr-hold-cont-appl:side dem 3poss-head f-dem 3poss-bag
‘She holds on her head the bag.’

2.2.4 Locative applicative -igi (‘inside’)

This applicative is found with locative meaning, without a secondary animate-goal

meaning.

(19) ∅-chek-tak neɁena w-aygi na awyaq

3intr-eat-prog p.pron 3intr-go.appl:inside dem forest
‘S/he is eating this which it is inside the forest.’

(20) a. Juan ro-Ɂon-aɢaa-tak

J. 3intr-sing-antip-prog
‘Juan is singing.’ (BC)
 6 Applicatives in Toba/Qom (Guaykuruan) 155

b. Juan ro-Ɂon-aɢaa-ta-ygi ñi l-maɁ

J. 3intr-sing-antip-cont-appl:inside dem 3poss-home
‘Juan is singing inside his home.’ (AC)

2.2.5 Goal/directional applicative -ta (‘to the other side’)

Probably due to different paths of grammaticalization of the verbal root -ta ‘to go’ in
serial verb constructions (§ 5), this locative applicative, or trajectory applicative—used
to express the notion of crossing a creek, river, or lagoon—shows a relatively low type
frequency but an extremely low token frequency: it is the less frequent of all applica-
tives in the analyzed corpora. The fact that it frequently co-occurs with the expression
ra leɁego qaɁim/lapel/talaɁ ‘this side (of) the estero / lagoon / big river’ (21b–c) possi-
bly indicates the development of a process of desemantization, at least in the varieties
studied so far.

(21) a. saɢaa-ta na qaɁim

1pl.go.1pl-appl:other.side dem shallow.lagoon
‘We go to the other side of the estero.’
b. ño-wir-aɢa-ta na leɁego na tala
1pl.mid-arrive-1pl-appl:other.side dem side dem Paraná.river
‘We arrive to this side of the Paraná river.’
(Buckwalter 1980: 80)
c. y-ashike-ta ra leɁego na Ɂetaɢat
3tr-go.straight-appl:other.side dem side dem water
‘S/he goes straight across the water.’
(Buckwalter 1980: 192)

2.2.6 Goal/directional applicative -ge (‘moving away’)

This applicative—as well as its opposite -get (§ 2.2.7)—shows traces of serial verb con-
struction and grammaticalization processes in progress. In (22a), the applied P simply
refers to the locative goal; in (22b), with two animate arguments, the directional
meaning must be interpreted as A going behind P, which is moving away from A. In
(22c), P is the source of the event coded by the predicate.

(22) a. se-wir-age ra Salta

1-arrive-appl:away dem S.
‘I (will) arrive to Salta [Province].’ (S is far away from Salta)
156 Marisa Censabella

b. ayim sa-ta-te-ge a-na potay

pro.1 1-go-cont-appl:away f-dem antbear
‘I’m going towards (chasing) the antbear (which is running away from me).’
c. na laɁat i-chigoɁo-ge na tohlek Ɂalwa
dem wind 3tr-come-appl:away dem desert
‘The wind comes from the desert.’

2.2.7 Goal/directional applicative -get (‘approaching’)

Apart from the directional use with the ‘approaching’ meaning (23b), this applicative
indicates the source with ‘come’-type verbs, as in (24).

(23) a. r-alako ñi pioq

3intr-bark dem dog
‘The dog barks.’ (BC)
b. r-alakoi-get na shiyaɢawa-pi so pioq
3intr-bark-appl:approach dem person-coll dem dog
‘It barks—while approaching—to the people, that dog.’ (AC)

(24) Rosa Merino i-chigoɁo-get na rokshe

R. M. 3tr-come-appl:approach dem non.indigenous.person
‘Rosa Merino comes from the criollos.’ (Her family is not indigenous.)

2.2.8 Goal/directional applicative -sop (‘around’)

This applicative introduces an argument that represents an object, a person, or a geo-

graphical space around which an animate S/A argument circles. Example (25b) shows
the lexical productivity of this type of applicative.

(25) a. we-tai-sop ñi noyik

3intr-go-appl:around dem house
‘It [an animal] is/lives around the house.’
b. ramaʒe ya-koo-sop so noɢotolek
pro.3 3intr-grab-appl:around dem boy
‘He embraces the boy.’
c. [so qañaɢaʒe-Ɂ]. . . ∅-qolee-tege-sop-oɁ
dem scavenger.bird-pl 3intr-spin-prog-appl:around-P.pl
kenaɢa so-m Nsoɢoy-k
adv dem-top cannibal-m
‘He always circles around the two scavenger birds, that cannibal man. . .’
 6 Applicatives in Toba/Qom (Guaykuruan) 157

2.2.9 Allative applicative -Ɂa

This applicative can introduce an argument with the role of a goal, either inanimate
(26b) or animate/human (27b):

(26) a. so Pedro i-ro-wo so l-yalek,

dem P. 3tr-take-l/d:in dem 3poss-son
‘Pedro takes inside (the house/a place) his son.’ (BC)
b. i-ro-w-Ɂa ñi lohpital so l-yalek
3tr-take-l/d:in-appl:all dem hospital dem 3poss-son
‘He takes his son to the hospital.’ (AC)

(27) a. so jale i-wir-ewo ñi noyik

dem man 3tr-arrive-l/d:in dem house
‘The man arrives inside the house.’ (BC)
b. so jale i-wir-ew-Ɂa a-ra Ɂalo
dem man 3tr-arrive-l/d:in-appl:all f-dem woman
‘The man arrives to(wards) the woman.’ (AC) (Sp. ‘llega a/hasta la mujer’)

In some cases, there is no clear distinction in semantic roles that could explain the
use of the applicative marker, and the relevant factor seems to be rather the inherent
nature of the participant fulfilling a given role, but additional data would be necessary
before discussing a possible generalization. In (28), for instance, the verb -wen ‘to need’
takes, when applicativized, an object that is apparently to be used as an instrument or
raw material in an implicit future event:

(28) a. i-wen ka l-maɁ

3tr-need dem 3poss-home
‘S/he needs a place to live’ (BC)
b. i-wen-Ɂa a-ka laɢaray l-awe
3tr-need-appl:all f-dem palm tree 3poss-leaf
‘S/he needs a palm leaf (to make baskets).’ (AC)

By a related token, the verb -pot means ‘to touch or to step on (with the sole of the foot
or the palm of the hand and the nails)’; its expected P arguments are ‘soil’, ‘land’, ‘water’
and ‘sand’, as in (29a). Nevertheless, non-expected Ps, such as someone’s neck in (29b),
require the use of the applicative:16

16 Although a possessive construction is involved in (29b), it would not be correct to consider it an

example of external possession.
158 Marisa Censabella

(29) a. i-pot a-na Ɂalwa

3tr-touch f-dem land/soil
‘S/he touches the soil.’ (BC)
b. i-pot-ew-Ɂa a-so Ana l-qosot
3tr-touch-l/d:in-appl:all f-dem A. 3poss-neck
‘S/he touches Ana’s neck.’ (AC)

2.2.10 Benefactive/malefactive applicative -Ɂem

The beneficiary (30) or maleficiary (31) interpretation of the P introduced by this

applicative results from the semantics of the verbal root/base. This marker can also
introduce an involuntary cause, as in (32b) and (33b).

(30) a. ramaʒe do-Ɂon-aɢan

pro.3 3intr-sing17-antip
‘He sings.’ (BC)
b. ayim do-Ɂon-aɢan-em
pro.1 3intr-sing-antip-appl:ben
‘He sings for me.’ (AC)

(31) ayim i-lla-peg-em ra y-aqtak

pro.1 3tr-forbid-rep-appl:ben dem 1poss-word
‘S/he always forbids me to talk / my speech.’

(32) a. ramaʒe ∅-keɁemaq

pro.3 3intr-be.wounded
‘He is wounded.’ (BC)
b. aw-keɁemaɢ-am so l-kat
2-be.wounded-appl:ben dem 3poss-word
‘You are wounded by that piece of iron.’ (AC)

(33) a. a-ramaʒe i-lew18

f-pro.3 3tr-die
‘She dies.’ (BC)

17 In T/Q, the verbal root -Ɂon ‘to sing’ is transitive.

18 In T/Q, the verb ‘die’ (intr) co-occurs with the 3rd-person index i- used with transitive verbs, because
this index encodes telicity. Thus, it is also used with active accomplishments, which are usually lexical-
ized as transitive verbs (Zurlo and Censabella 2013).
 6 Applicatives in Toba/Qom (Guaykuruan) 159

b. a-ka Ɂalo i-lem ra l-koɁok

f-dem woman 3tr-die.appl:ben dem 3poss-giving.birth
‘That woman dies because of giving birth.’

2.2.11 Recipient applicative -Ɂi

Since the recipient applicative -Ɂi mainly occurs with the transitive verb -an ‘to give’, it
has an extremely low type frequency but a high token frequency. It does not necessarily
yield a ditransitive clause (cf. § 3).

(34) a. so yaɢaykyolek y-an so l-apoɁ

dem old.man 3tr-give dem 3poss-poncho
‘The old man gives his poncho away.’ (BC)
b. so yaɢaykyolek y-añ-i so l-qaya
dem old.man 3tr-give-appl:rec dem 3poss-brother
so l-apoɁ
dem 3poss-poncho
‘The old man gives his brother the poncho.’ (AC)

2.2.12 Comitative applicative -eɁ

The comitative applicative -eɁ could be the result of the further grammaticalization of
the goal/directional applicative -get (§ 2.2.7). In (36b), with a plural P, the comitative
applicative morpheme recovers the original -t- phoneme of its putative source. Other
examples are presented in Section 5.2.

(35) a. n-yom-eɁ ñi l-qaya

3mid-drink-appl:com dem 3poss-brother
‘S/he drinks with his brother.’ (González 2011: 147)
b. so yale r-apet-tag-eɁ ra l-aworewa
dem man 3intr-argue-prog-appl:com dem 3poss-neighbor
‘The man is arguing with his neighbor.’

(36) a. somaʒe i-kelola-peg-eɁ na l-qayañik-pi

pro.3 3tr-help-rep-appl:com dem 3poss-brother-coll
‘He always helps with his brothers.’
b. somaʒe i-kelola-peg-et-oɁ na-wa Ɂalo-l
pro.3 3tr-help-rep-appl:com-P.pl dem-dl woman-pl
‘He always helps with the two women.’
160 Marisa Censabella

3 Applicative syntax
Zúñiga and Creissels (this volume) define the applicative construction as follows:
i) The predicates in both constructions are built upon the same root, but the one in
the AC bears additional overt marking that distinguishes it from the one in the BC.
ii) The participant encoded as S or A in the BC appears as S or A in the AC.
iii) The AC includes a noun phrase in a role other than S or A, the applied phrase (AppP),
which refers to a participant that either requires a non-core coding different from
its coding in the AC or cannot be expressed at all in the BC.

All T/Q examples presented above show properties i) and ii): the AC shares the same
root with the BC and involves a morphological expression in the verb. As regards prop-
erty iii), T/Q applied phrases cannot be expressed at all in the BC; when applied, the
introduced argument shows all the properties of a transitive object. T/Q applicatives are
obligatory: the only way of expressing a new non-agentive participant in the clause is by
means of applicativization; we do not find the category of oblique argument in the BC
that could alternate with a promoted applied argument in the AC.
Applicatives in T/Q co-occur with either active or middle voice without restrictions,
and they can also co-occur freely within each semantic verb class, whose classification
is encoded by the 3sg and 3pl split active personal indexes. Each verbal root contains
a specification in terms of number of obligatory arguments; thus, in T/Q there is no
P-lability or ambitransitivity: verbal roots/bases are either intransitive or transitive.
Any transitivity clause alternation must resort to syntactic or derivational procedures.
Different types of causatives (instrumental, direct intentional causative, direct unin-
tentional causative and factitive) and the antipassive rely on derivational procedures
to create new verbal bases that increase or decrease transitivity, respectively. Syntac-
tic procedures such as non-promotional passives and applicatives, on the other hand,
respectively decrease and increase/re-arrange transitivity.
To denote states with locative and directional information, T/Q has applicative
deponents or applicativa tantum. For instance, the highly grammaticalized root -ta ‘go’
can only appear with an applicative, showing some degree of lexicalization and regres-
sive vowel assimilation, as in (37).

(37) a. t-aja na menaɢanaɢaki

3intr-go.appl:all dem market
‘S/he goes to the market.’
b. t-alek a-na qomaɁ
3intr-go.appl:over f-dem stone/mountain
‘S/he climbs the mountain.’
c. we-to-Ɂot a-ra ipaq
3intr-go-appl:under f-dem tree
‘S/he is under the tree.’
 6 Applicatives in Toba/Qom (Guaykuruan) 161

T/Q does not have double applicatives, and each VP only has one applicative at a time,
as seen in all examples above. Only lexicalized applicatives in the verbal base could
co-occur with a syntactic applicative, as in (38d), but here the lexicalized unit no longer
functions as an applicative; rather, it is fully integrated into the verb base, to the left of
the second segment of the 1pl index. Nevertheless, the existence of (38b) shows that
grammaticalization of the allative is still in progress with this verbal predicate.

(38) a. ramaʒe r-taqa

pro.3 3intr-talk
‘He talks.’ (BC)
b. ramaʒe r-taɢa-ya so l-awo
pro.3 3intr-talk-appl:all dem 3poss-family
‘He talks about his family.’ (AC)
c. qoɁomi se-taɢaya-qa-pek
pro.1pl 1pl-talk.about-1pl-rep
‘We always talk about sth./sb.’ (BC)
d. ramaʒe se-taɢaya-qa-peg-eɁ
pro.3 1pl-talk.about-1pl-rep-appl:com
‘We always talk about sth./sb. with him.’ (AC)

T/Q does not show ditransitive verbal roots/bases but has ditransitive constructions
based on coordinated clauses. Ditransitive constructions are allowed only with the verb
-an ‘give’ plus the recipient applicative -i, as shown in (34b) above. Nevertheless, these
ditransitive clauses could be the result of the elicitation work and mostly a replica of
ditransitive constructions in Spanish. Narrative texts in T/Q, however, are frequently
expressed as in (39a–b), where the same verb root/base appears twice: first in a BC and
then in an AC, both clauses being coordinated and sharing—by anaphoric persistence—
the non-applied argument, instead of constructing one clause with three arguments. In
these examples, the speakers’ strategy is to decompose the event into two sub-events,
each one associated with a two-argument clause. Thus, with transitive verbs, T/Q applic-
atives do not increase valency but redirects P-hood.

(39) a. i-ro so l-yalek, i-ra-Ɂa ñi Resistencia

3tr-herd dem 3poss-son 3tr-herd-appl:all dem R.
‘S/he takes his/her son, s/he takes [him] to Resistencia city.’
b. y-asaq a-so pelota, y-asaɢ-a ñi qar-qaya
3tr-throw f-dem ball 3tr-throw-appl:all dem 1pl.poss-brother
‘S/he throws the ball, s/he throws [it] to our brother.’

This strategy, based on the paratactic juxtaposition of clauses with reference tracking
and tight restrictions on the nominal arguments associated with each verb, could be
explained by the inexistence of inherently ditransitive verbs and oblique arguments
162 Marisa Censabella

(as Mithun 2004 suggests for some Native American languages). Narrative texts also
show examples like (40a–b), where an argument is anaphorically shared by the two
coordinated clauses; in these examples, the verb root is first detransitivized by the anti-
passive, -shin-aɢan in (40a) and -Ɂaqt-aɢan in (40b), and the 3rd-person index is r-, for
intransitive active verbs. Then, the applicative derives a verb with a non-agentive argu-
ment semantically different from the originally P argument specified in the lexical root.

(40) a. r-shin-aɢan-a a-so lawoɢo, y-añi a-so María

3tr-get-antip-appl:all f-dem flower 3-give-appl:rec f-dem M.
‘S/he gets to that flower, s/he gives [it] to María.’
b. qaq Ɂera ra la-Ɂaqtaɢanaɢak so i-taɁa,
coord ph.dem dem 3poss-story dem 1poss-father
qoɁollaq ayim ra-Ɂaqt-aɢan-em
adv pro.1 3intr-talk-antip-appl:ben
‘And this is my father’s story, long time ago he told (it) to me.’

There is a less frequent property where the applicative is related to antipassivized

verbal bases: the verbal base and the benefactive/malefactive applicative -em allow a
3rd person active voice index alternation: i- (for transitive verbs) vs. r- (for intransitive
ones). This feature is rare, because usually i- vs. r- 3rd-person index alternations require
valency adjustments involving the derivational or syntactic procedures explained
above. More research is needed to confirm whether such examples represent elicita-
tion deformations or are, presumably, a clue to a sort of converb construction, which
would arise via grammaticalization. In (41a) we find a recipient-beneficiary interpreta-
tion, while (41b) has a substitutive-benefaction interpretation only; the latter reading is
encoded only by choosing intransitive r- instead of transitive i-.

(41) a. a-ra Ɂalo ayim i-wosh-aɢan-em

f-dem woman pro.1 3tr-cook-antip-appl:ben
‘This woman cooks for me.’
b. a-ramaʒe ayim r-wosh-aɢan-em
f-pro.3 pro.1 3intr-cook-antip-appl:ben
‘She cooks for me.’ (she works as a cook in my restaurant)

Usually, when the semantic specification of a transitive verb root/base introduces

an inanimate P, the predicate accepts a human goal, recipient or beneficiary P when
applicativized, as in example (39a–b), above. Antipassivized transitive verbs change the
semantic specification of the verbal base, from an active accomplishment into an activ-
ity; thus, the latter is the allowed construction used in order to derive a new applicativ-
ized one, as in examples (42).
 6 Applicatives in Toba/Qom (Guaykuruan) 163

(42) a. a-ramaʒe y-alekten na l-leɁe

f-pro.3 3tr-stir dem 3poss-soup
‘She stirs the soup.’
b. a-ramaʒe r-alekten-aɢan
f-pro.3 3intr-stir-antip
‘She stirs (sth.).’ (BC)
c. a-ramaʒe y-alekten-aɢan-a so towe
f-pro.3 3tr-stir-antip-appl:all dem salt
‘She stirs (sth.) with salt.’ (AC)

Finally, when the semantic specification of the verb demands a human goal as P, the
applicative construction allows the introduction of a locative P argument, as in (43).

(43) a. qa-y-kotaq a-saɁaso Nsoɢoj

imprs-3tr-push f.dem.pron cannibal[f]
‘(Someone) pushes that Nsogoi.’
b. qa-y-kotaq-an-gi ʒeɁeʒe leɁ,
imprs-3tr-push-l/d:down-appl:side dem.pron ember
a-saɁaso Nsoɢoj
f.dem.pron cannibal[f]
‘(Someone) pushes downwards to the ember, that Nsogoi.’ (The Nsogoi is pushed
towards the ember.)

Whether increasing or re-arranging valency, all applied arguments behave as core P

arguments in three respects: a) P number agreement, b) constituent order, and c) topi-
calization and its consequences in coordination and subordination pivots.
Applied arguments behave in the same way as non-applied P arguments regarding
nominal number encoded in the verb, as shown in (44)–(45).

(44) a. so-wa noɢotol-qa y-asoɁt ñi pioq

dem-dl child-pl 3tr-kick.pl dem dog
‘The children kick the dog.’
b. so noɢotole-k y-asoʔt ñi-wa pioɢo-ʒe
dem child-m 3tr-kick.pl dem-dl dog-pl
‘The child kicks the dogs.’

(45) a. y-an so pan

3tr-give dem bread
‘S/he gives bread away.’
b. y-an-em so l-qaya, so pan
3tr-give-appl:ben dem 3poss-brother dem bread
‘S/he gives her/his brother bread.’
164 Marisa Censabella

c. y-an-em-a so l-qaya-qa, so pan

3tr-give-appl:ben-P.pl dem 3poss-brother-pl dem bread
‘S/he gives her/his brothers bread.’

Applied arguments have the same position in the clause as base P arguments. Both
follow the verb when lexical but precede it when pronominal, as seen in (46) and (47),
respectively.

(46) a. ayim s-alawat so shigyaq

pro.1 1-kill dem animal
‘I kill the animal.’
b. Ɂam s-alawat
pro.2 1-kill
‘I kill you.’

(47) a. so yaɢaykyolek y-añ-i so

dem old man 3tr-give-appl:rec dem
l-qaya so l-apoɁ
3poss-brother dem 3poss-poncho
‘The old man gives his brother the poncho.’
b. so yaɢaykyolek Ɂam y-añ-i so l-apoɁ
dem old man pro.2 3tr-give-appl:rec dem 3poss-poncho
‘The old man gives you the poncho.’

In T/Q, non-promotional (impersonal) passives demote the A argument without any

other modification in the basic clause (48).

(48) a. ramaʒe y-amaq a-so qomaɁ

pro.3 3tr-throw f-dem stone
‘He throws the stone.’
b. qa-y-amaq a-so qomaɁ
imprs-3tr-throw f-dem stone
‘They (imprs) throw the stone.’

When used in ACs, non-promotional passives have the same pragmatic and syntactic
consequences as in any BC: the (applied) P argument is topicalized and, because of this,
coordination and subordination pivots can relate the (applied) P with a co-referential
A or S argument in the next coordinated or subordinated clause, as seen in (49) and
(50), respectively. In (49), the one who says “Yes, thank you, my grandson” is Nsogoj, the
cannibal woman, an argument introduced by the allative applicative in the precedent
clause. In (50), the allative applicative introduces a complement clause.
 6 Applicatives in Toba/Qom (Guaykuruan) 165

(49) nache qa-y-wir-ew-Ɂa a-so-m Nsogoj

coord imprs-3tr-arrive-l/d:in-appl:all f-dem-top cannibal[f]
nache e-nak-oɁ ahaɁ ñachik i-wal
coord 3tr-say-foc adv thanks 1poss-grandson
‘And they arrive towards the Nsogoi and she says: “Yes, thank you, my grandson.”’

(50) qayka qa-y-aten-Ɂa n-eta-lek a-naɁana

neg.ex.pres imprs-3tr-know-appl:all 3mid-go-appl:over f-dem.pron
kor-ereɁ
1pl-book
‘It is not known what is in our book.’

4 Applicative semantics
This section presents semantic nuances that applied P arguments can encode accord-
ing to the combination of high frequency applicatives with certain verbal roots/bases.
Examples (51a, b, c) show that applicatives based on transitive roots/bases do not add
a third argument; instead, they change the meaning expressed by the verb. The lexical
base -waɢan ‘hit’ allows only two human/animate core arguments, A and P, with or
without applicatives; thus, only some applicatives can be used without altering the
basic semantic template of this lexical base. On the contrary, if the speaker wants to
use -waɢan with the intention of applying a locative P argument, the applicative used is
typically a locative one (51d–e).

(51) a. so Pedro i-waɢan so l-qaya

dem P. 3tr-hit dem 3-poss-brother
‘Pedro hits his brother (with the fist).’
b. so Pedro i-waɢan-lek so l-qaya
dem P. 3tr-hit-appl:over dem 3-poss-brother
‘Pedro gives his brother a beating.’
c. so Pedro i-waɢan-Ɂa so l-qaya
dem P. 3tr-hit-appl:all dem 3-poss-brother
‘Pedro accidentally hits his brother.’
d. so Pedro i-waɢan-gi l-qayk so l-qaya
dem P. 3tr-hit-appl:side 3-poss-head dem 3-poss-brother
‘Pedro accidentally hits (lit. hits on) his brother’s head.’
e. i-waɢan-gi a-na lacampana
3tr-hit-appl:side F-DEM bell
‘S/he hits the bell.’
166 Marisa Censabella

When applicatives used with transitive verbal roots/bases do not change the P’s seman-
tic role, they contribute to encode an adverbial meaning, adding a notion of intensity/
volition to the lexical content of the verbal root/base. The verbal base specifies an
animate P in (52a) and the directional applicative -ge contributes to modify the seman-
tic meaning of the original lexical root (52b):

(52) a. Juan i-lo so l-yalek

J. 3tr-look dem 3poss-son
‘Juan looks at his son.’
b. Juan i-lo-ta-ge so l-yalek
J. 3tr-look-cont-appl:away dem 3poss-son
‘Juan watches/supervises his son.’

A scale of saliency governs the selection of one of the four morphemes related to human
locative/directional goals (from less to more salient): -lek > -Ɂa > -eɁ > -Ɂot. Many verbs
can allow the alternation of at least two of these morphemes. Despite this saliency scale
and its overlap with the recipient-beneficiary encoding, whenever the speaker wants
to highlight the benefit that a non-subject participant receives from the verbal event,
that participant will be encoded as a beneficiary, which means that the same applica-
tive morpheme -em is also used in intransitive clauses with a substitutive benefaction
reading. Only with the verb ‘give’ is it possible to make a clear distinction between a
recipient and a recipient-beneficiary in T/Q; with other verbs, the benefactive applica-
tive marker -em is used to express both semantic notions. Thus, following Kittilä’s classi-
fication (2005: 277, 295), T/Q is a beneficiary-prominent language, since “languages with
few formally ditransitive verbs are more prone to beneficiary prominence”.

5 Grammaticalization
Except for the origin of the allative applicative marker, there is no specific research on
the grammaticalization clines of applicatives in T/Q yet. Censabella (2006b, 2008) pro-
poses that the verb -ta ‘go’ may have become a continuous imperfective aspect marker
-ta, and, alternatively, the allative applicative marker -Ɂa, following in both cases uni-
versal paths of grammaticalization. Regarding other grammaticalization routes of
applicatives, Censabella (2018) studied the different contexts that gave rise to a focus
marker from the locative applicative -Ɂot (‘under’).
In his study on T/Q serial verb constructions, González (2009) presents examples
that could be interpreted as contexts of grammaticalization in which a serial verb
construction gives rise to an aspectual marker plus an applicative one. Examples like
(53a–b) could be interpreted as bridging contexts in grammaticalization processes
because the morpheme -ta could be interpreted as the 3rd-person index tV- (cf. Table 1)
 6 Applicatives in Toba/Qom (Guaykuruan) 167

but also as the continuous imperfective aspect marker -ta (opposed to the progressive
-tak and to the unmarked perfective). Furthermore, the verb -ta ‘go’ has suffered pho-
nological erosion; the examples in (37) above probably show a more grammaticalized
stage of these constructions, where the verb ‘go’ evolved as follows: -ta > -Ɂa > ∅.

(53) a. ro-Ɂon-aɢa-ta-Ɂa-get na y-aqaya

3intr-sing-antip-cont-go-appl:approach dem 1poss-brother
‘S/he sings approaching my brother.’ (González 2009, gloss adapted)
b. ayim se-kaɁa-ta-Ɂa-ge ñi y-aqaya
pro.1 1-follow-cont-go-appl:away dem 1poss-brother
‘I am following my brother (who walks away).’ (González 2009)

Applicatives in T/Q show ongoing grammaticalization processes in different stages of

evolution. Non-canonical semantic roles introduced by each applicative are listed in
Table 3. Below, we focus on the grammaticalization processes which show syntactic
change and, even in an advanced stage, phonological erosion, namely the locative
applicative -oɁt (‘under’) (§ 5.1) and the directional applicative -get (‘approaching’)
(§ 5.2). We observe processes of extension and desemanticization; the applicative is used
in new contexts where it loses locative or directional referential content and acquires a
more abstract grammatical meaning. Heine (2002) classifies different types of contexts
in which the grammaticalization processes take place. Between the source meaning
and the target meaning, the author proposes intermediate ones: a bridging context,
a switch context, and a conventionalization context. The bridging context triggers an
inference that allows an AC interpretation rather in the sense of the target meaning
than the source one. In the switch context, the meaning of the AC is incompatible with
the source meaning. Finally, the conventionalization context allows the use of AC in
contexts different from the bridging and switch ones, even contradicting the semantics
of the source unit. Due to their graduality, the dividing line between these processes is
not always easy to draw.

5.1 Coexisting grammaticalization contexts of the locative

applicative -Ɂot (‘under’)

5.1.1 The putative source-meaning context

With intransitive roots/bases (stative, movement, or translational movement), -Ɂot

‘under’ applies a locative P, usually a non-human entity (54a–b).19 While providing

19 For this explanation, the locative applicative -Ɂot (‘under’) is taken as the less grammaticalized form
in the cline, even though the full verbal or nominal origin has not been identified yet.
168 Marisa Censabella

stative locational information, this applicative can relate to human arguments, provid-
ing the locus or spatial reference of P as related to A (55a–b).

(54) a. ñi qagesaq we-to-ʔot a-ra ipaq

dem ant 3intr-go-appl:under f-dem tree
‘The ant is under the tree.’
b. we-to-ʔot seɁeso qomaɁ
3intr-go-appl:undr dem stone/mountain
‘S/he shelters under that stone.’

(55) a. a-ñi l-yale ra-tato-ʔot so l-taɁa

f-dem 3poss-daughter 3intr-go-cont-appl:under dem 3poss-father
‘The daughter is under her father.’ (they are seated on a tribune/platform)
b. ñi l-taɁa ra-tato-ʔot a-so l-yale
dem 3poss-father 3intr-go-cont-appl:under f-dem 3poss-daughter
‘The father is under his daughter.’ (they are seated on a tribune/platform)

5.1.2 The bridging context

In this context, the speaker infers a new semantic nuance from the AC, namely that the
P has more saliency than the A. Both arguments refer to human entities, and with verbs
like ‘ask’, ‘beg’, ‘surrender’, the A argument is in an inferior hierarchical scale than
P (56). When P refers to a divinity, the spatial metaphor—coding A under P—is even
clearer (57a); by contrast, the comitative applicative is unacceptable for speakers (57b).

(56) ñ-añan-q-ot ñi qar-taɁa

1pl.mid-surrender-1pl-appl:under dem 1pl.poss-father
‘We surrender to our father.’ (Buckwalter 1980:149)

(57) a. se-taq-tek-ot ñi Dios

1-talk-prog-appl:under dem G.
‘I am talking to God.’
b. ✶se-taq-tek-geɁ ñi Dios
1-talk-prog-appl:com dem G.
‘I am talking with God.’20

20 The speaker explains as follows: “we can’t talk to God as if he were a friend”.
 6 Applicatives in Toba/Qom (Guaykuruan) 169

5.1.3 The switch context

Here, the locative interpretation fades away and allows a hierarchical one, not only in
relation with divinities—via lexical conditioning—but extending the coverage to any
other human being (58a–b). In these examples, the speaker freely chooses whether the
human P has more or less saliency or authority in relation to A.

(58) a. se-taq-tek-ot so Intendente

1-talk-prog-appl:under dem mayor
‘I am talking with the mayor.’ (talking to an authority)
b. se-taq-teg-eɁ so Intendente
1-talk-prog-appl:com dem mayor
‘I am talking with the mayor.’ (he is my friend)

5.1.4 Conventionalization context

In this context, the notion of saliency/hierarchy remains but another one appears too,
namely one of focus, as seen in (59a) and (60a). The alternation with the benefactive
applicative shows how the speaker can choose semantic nuances in different types
of benefaction situations, as shown in (59b) and (60b). Explanatory comments were
obtained via elicitation.

(59) a. a-so qomlashe y-anek a-so l-yale,

f-dem T/Q.woman 3tr-give f-dem 3poss-daughter
y-an-ot a-so rokshelashe
3tr-give-appl:under f-dem non.T/Q.woman
‘That Qom woman gives her daughter, she gives [her] to the white woman.’
(The daughter is given to work for that patroness white woman) (hierarchical
applied P)
b. a-so qomlashe y-anek a-so l-yale,
f-dem T/Q.woman 3tr-give f-dem 3poss-daughter
y-an-em a-so rokshelashe
3tr-give-appl:ben f-dem non.T/Q.woman
‘That Qom woman gives her daughter, she gives [her] to the white woman.’
(The daughter is given to work for that white woman) (substitutive benefactive
applied P)

(60) a. Ɂam se-wosh-aɢan-ot

pro.2 1-cook-antip-appl:under
‘I cook for you.’ (you are the patron/patroness)
170 Marisa Censabella

b. Ɂam se-wosh-aɢan-em
pro.2 1-cook-antip-appl:ben
‘I cook for you.’ (I’m the cook in your restaurant)

5.1.5 Context of advanced grammaticalization

In narrative texts, when using verba dicendi, speakers can optionally use the mor-
pheme -oɁ for introducing direct (49)—repeated here as (61)—or indirect speech (62).
We propose to interpret this morpheme as a focus marker. The clause that follows is
highlighted and its S/A or P arguments—the latter in a non-promotional passive—will
show the same coordination and subordination pivots as those presented in Section 3.

(61) nache qa-y-wir-ew-Ɂa a-so-m Nsogoj

coord imprs-3tr-arrive-l/d:in-appl:all f-dem-top cannibal[f]
nache e-nak-oɁ ahaɁ ñachik i-wal
coord 3tr-say-foc adv thanks 1poss-grandson
‘And they arrive towards the Nsogoi and she says: “Yes, thank you, my grandson.”’

(62) qalaɢaʒi so-m shiyaɢawa. . . e-nak-oɁ ʒi y-alawat

coord dem-top person 3tr-say-foc dem 3tr-kill
qome na rokshi-pi
adv dem non.t/q.person-coll
‘. . . but that person. . . s/he says that s/he kills, afterwards, the white people.’

5.2 Coexisting grammaticalization contexts of the goal/

directional applicative -get (‘approaching’)

Less studied in terms of Heine’s changing contexts, but displaying a clear path of gram-
maticalization nonetheless, is the cline -get (‘approaching’) > -eɁ ‘comitative’ (63). Exam-
ples (63a) and (63b) show the source-meaning context; notice that (63a) still retains the
semantics of the putative serial construction. In (63c) and (63d) we can identify the
bridging context, because a second interpretation of the clause is possible, while in
(63e) the comitative sense indicates the switch context. Finally, (63f) shows the conven-
tionalization context, alongside phonological erosion, with a clear comitative meaning.

(63) a. r-alakoi-get na shiyaɢawa-pi so pioq

3intr-bark-appl:approach dem person-coll dem dog
‘The dog barks to the approaching people.’
 6 Applicatives in Toba/Qom (Guaykuruan) 171

b. a-so Ɂaltemoy i-laɁat-get

f-dem car 3intr-destroy-appl:approach
so l-alo kaayo
dem 3poss-domestic.animal horse
‘The car crashes the horse.’
c. Ɂam se-wat-ta-get
pro.2 1-wait-cont-appl:approach
‘I’m waiting for you.’ or ‘I’m waiting with you.’
d. Rosa Merino i-chigoɁo-get na rokshe
R. M. 3tr-come-appl:approach dem non.indigenous.person
‘Rosa Merino comes from the criollos.’ or ‘R. M. comes with the criollos.’
e. qoɁomi s-ow-aq-ta-get a-ñi Rosita
pro.1pl 1pl-be.inside-1pl-cont-appl:approach f-dem R.
‘We are inside [the house] with Rosita.’
f. so yale r-asot-tag-eɁ a-ra Ɂalo
dem man 3intr-dance-prog-appl:com f-dem woman
‘The man is dancing with the woman.’

6 Conclusions
This chapter surveyed the applicative constructions attested in Toba/Qom language.
These can be characterized as follows:

Morphology
– T/Q presents a paradigm of twelve morphemes which behave like applicatives. This
paradigm occupies the last functional slot in the verb form, followed by a number
agreement marker if the applied P is plural. The locative applicative -lek (‘over’) and
-Ɂot (‘under’), the allative applicative -Ɂa and the benefactive/malefactive applica-
tive -em are the only markers in the paradigm that show morphophonologically
conditioned allomorphy.
– Applicativization is highly productive in T/Q: almost all verbal roots can accommo-
date additional arguments via an applicative morpheme.

Syntax
– T/Q applicatives are obligatory, meaning that the only way to express the seman-
tic roles expressed by applied phrases is by means of applicativization. Moreover,
applicativization is the only available strategy to encode semantic roles other than
those licensed as subjects or objects of non-applicative verb forms.
172 Marisa Censabella

– The applied phrase shows all the syntactic properties of a transitive object or P
argument.
– Applicatives in T/Q co-occur without restrictions in both active- and middle-voice
verbs, and can co-occur freely within each semantic verb class.
– Each verbal root specifies the number of its obligatory arguments. Thus, in T/Q
there is no P-lability or ambitransitivity: verbal roots/bases are either intransitive
or transitive.
– T/Q does not have double applicatives; each predicate has only one applicative at a
time. Some lexicalized applicatives could be identified, but they no longer behave
as applicatives.
– In free conversation or narrative texts, T/Q speakers usually do not accept ditransi-
tive clauses. Rather, speakers decompose complex events into sub-events.
– Whether increasing or re-arranging transitivity, in T/Q all applied arguments
behave as base P arguments in three main respects: P number agreement, constit-
uent order, and topicalization (and its consequences in coordination and subordi-
nation pivots).

Semantics
– When used with transitive roots/bases, applicatives do not add a third argument;
instead, they change the meaning expressed by the verb, allowing a different
semantic role from the one specified in the verbal root/base.
– Applicatives whose meaning relate to human locative/directional goals, show a
scale of saliency that governs arguments (from less to more salient): -lek > -Ɂa > -eɁ
> -Ɂot. Many verbs can take at least two of these morphemes.
– T/Q is a beneficiary-prominent language: whenever the speaker wants to highlight
the benefit that a non-subject participant gets through the verbal event, it will be
encoded as a beneficiary. Only with the verb ‘give’ is it possible to make a clear
distinction between a recipient and a recipient-beneficiary; with other verbs, the
benefactive/malefactive applicative marker -em is used to express both semantic
notions.

Grammaticalization
– Applicatives in T/Q seem to derive from serial verbs constructions, although more
research is needed in order to confirm this statement.
– Applicatives in T/Q show undergoing grammaticalization processes in different
stages of evolution, as shown by the non-canonical semantic roles introduced by
each applicative.
 6 Applicatives in Toba/Qom (Guaykuruan) 173

Abbreviations
A agent-like argument in a transitive clause
all allative
antip antipassive
appl applicative
ben benefactive
caus causative
coll collective
cont continuous
coord coordinator
dem demonstrative
dem.pron demonstrative pronoun
dl dual
emph emphatic
ex.pres existential presentative
f feminine
fact factitive
foc focus
h.goal human goal
imprs impersonal
instr instrumental
intr intransitive
l/d locative-directional
m masculine
mid middle voice
neg negative
P patient-like argument in a transitive clause
pro personal pronoun
ph.dem phrasal demonstrative
pl plural
poss possessive
prog progressive aspect
rec recipient
recp reciprocal
refl reflexive
rep repetitive (verbal number)
S single argument in an intransitive clause
sub subordinator
top topicalizer
tr transitive
unkw.ag unknown agent
vblz verbalizer
174 Marisa Censabella

References
Avellana, Alicia. 2012. El español en contacto con la lengua toba (qom): fenómenos de transferencia y
adquisición de segundas lenguas. Forma y Función 25(1). 83–111.
Avellana, Alicia. 2013. Tiempo y aspecto en el español en contacto con la lengua toba (qom). In Ángela
Di Tullio (Coord.), El español de la Argentina: estudios gramaticales, 153–176. Buenos Aires:
EUDEBA.
Avellana, Alicia & Patricia Dante. 2013. Algunas observaciones sobre el contacto lingüístico entre el toba
(qom) y el español en una comunidad del Gran Buenos Aires. In Ana Fernández Garay, Marisa
Censabella & Marisa Malvestitti (eds.), Lenguas amerindias: contribuciones y perspectivas, 153–162.
(Colección Nuestra América). Buenos Aires: Editorial de la Facultad de Filosofía y Letras, Universidad
de Buenos Aires.
Benveniste, Émile. 1966 [1950]. Actif et moyen dans le verbe. Problèmes de linguistique générale, 168–175.
Paris: Gallimard.
Bigot, Margot. 1994. La lengua Qom (toba) del Chaco, Argentina (expresión del espacio en los lexemas
verbales). In Ana Fernández Garay & Pedro Viegas Barros (eds.), Actas de las Segundas Jornadas de
Lingüística Aborigen, 177–189. Buenos Aires: Instituto de Lingüística, Facultad de Filosofía y Letras,
Universidad de Buenos Aires.
Bigot, Margot. 2007. Los aborígenes qom en Rosario. Contacto lingüístico-cultural, bilingüismo, diglosia y
vitalidad etnolingüística en grupos aborígenes “qom” (tobas) asentados en Rosario. Rosario: Universidad
Nacional de Rosario.
Braunstein, José & Elmer S. Miller. 1999. Ethnohistorical introduction. In Elmer S. Miller (ed.), Peoples of the
Gran Chaco, 1–22. Westpoint, Connecticut: Bergin & Garvin.
Buckwalter, Alberto. 1980. Vocabulario toba – seguido de algunos apuntes sobre la gramática del idioma toba.
Roque Saénz Peña, Chaco (Argentina): Author’s edition.
Carpio, María Belén. 2007. Sistemas de alineación en toba (familia Guaycurú, Argentina). Hermosillo:
Universidad de Sonora MA thesis.
Carpio, María Belén. 2012. Fonología y morfosintaxis de la lengua hablada por grupos tobas en el oeste de
Formosa (Argentina). Munich: LINCOM.
Carpio, María Belén. 2018. Variación paradigmática en los pronombres libres de primera persona
no-singular en lenguas de la familia guaycurú. LIAMES 18(1). 45–60.
Carpio, María Belén. 2019. Relaciones topológicas en toba del oeste (guaycurú, Formosa, Argentina). Signo
& Seña 36. 44–68.
Carpio, María Belén & Marisa Censabella. 2010. Género y número en toba (flia. guaycurú, Argentina).
In Zarina Estrada Fernández & Ramón Arzápalo Marín (eds.), Estudios de lenguas amerindias 2:
contribuciones al estudio de las lenguas originarias de América, 57–84. Hermosillo: Universidad de
Sonora.
Carpio, María Belén & Marcela Mendoza. 2018. Tobas occidentales del Chaco boreal: evidencia de contactos
etnohistóricos y lingüísticos. Indiana 35(1). 165–189.
Carpio, María Belén, Raúl González & Marcela Mendoza. 2021. Grammatical replication: First-person
nonsingular verbal indexes in Eastern Toba, Western Toba (Guaicuruan) and Tapiete (Tupi-Guaraní).
Lenguaje 49(1). 47–75.
Censabella, Marisa. 1998. Axiología de la voz media en toba. In Pedro Viegas Barros & Marisa Censabella
(eds.), Actas III Jornadas de Lingüística Aborigen, 91–101. Buenos Aires: Instituto de Lingüística, Facultad
de Filosofía y Letras, Universidad de Buenos Aires.
Censabella, Marisa. 2002. Descripción funcional de un corpus en lengua toba (familia Guaycurú, Argentina).
Sistema fonológico, clases sintácticas y derivación. Aspectos de sincronía dinámica. Córdoba:
Universidad Nacional de Córdoba dissertation.
 6 Applicatives in Toba/Qom (Guaykuruan) 175

Censabella, Marisa. 2006a. Relaciones gramaticales en lengua toba (flia. Guaycurú, Argentina). In Zarina
Estrada Fernández, Actas VIII Encuentro Internacional de Lingüística en el Noroeste, 81–103. Hermosillo:
Universidad de Sonora.
Censabella, Marisa. 2006b. Plurifuncionalidad del valor aspectual continuo en toba. Paper presented at the
52° Congreso Internacional de Americanistas, Universidad de Sevilla, 17–21 August.
Censabella, Marisa. 2007. Los aplicativos alativo y locativo en toba. In Ana Fernández Garay & Marisa
Malvestitti (eds.), Lenguas aborígenes de la Argentina. Aspectos lingüísticos y sociolingüísticos, 29–50.
Santa Rosa: Universidad Nacional de La Pampa.
Censabella, Marisa. 2008. Roles semánticos y funciones del aplicativo alativo en toba. In Roberto Bein,
Cristina Messineo & Marisa Malvestitti (eds.), Estudios en lingüística y antropología. Homenaje a la Dra.
Ana Gerzenstein, 81–90. Buenos Aires: Facultad de Filosofía y Letras, Universidad de Buenos Aires.
Censabella, Marisa. 2009a. IV Chaco ampliado. Section Argentina en el Chaco. In Inge Sichra (ed.), Atlas
sociolingüístico de los pueblos indígenas de América Latina, Tomo I, 143–169. Cochabamba: UNICEF/
FUNDPROEIB Andes.
Censabella, Marisa. 2009b. Denominaciones etnonímicas y toponímicas tobas. Introducción a la
problemática y análisis lingüístico. In José Braunstein & Cristina Messineo (eds.), Hacia una nueva carta
étnica del Gran Chaco, Vol. 8, 213–236. Las Lomitas (Formosa, Argentina): Centro del Hombre Antiguo
Chaqueño.
Censabella, Marisa. 2010. Beneficiaries and recipients in Toba (Guaycurú). In Fernando Zúñiga & Seppo
Kittilä (eds.), Benefactives and malefactives: Typological perspectives and case studies, 185–201.
Amsterdam: John Benjamins.
Censabella, Marisa. 2015. Codificación del nombre colectivo en el español hablado por tobas: un análisis
exploratorio desde la teoría de la gramaticalización inducida por contacto. RILI XIII. 25: 155–183.
Censabella, Marisa. 2018. ¿Proceso de pragmaticalización o de gramaticalización? Contextos de uso del
aplicativo “debajo” en qom/toba. Cuadernos de ALFAL 10. 103–116.
Censabella, Marisa & Jimena Terraza. 2009. Aplicativos en toba y en wichí: forma y función. In Marisa
Censabella & Raúl González (eds.), II Encuentro de Lenguas Indígenas Americanas & II Simposio
Internacional de Lingüística Amerindia (ALFAL). Buenos Aires: CONICET. CD ROM.
Creissels, Denis. 2006. Syntaxe générale, une introduction typologique. Paris: Lavoisier.
Creissels, Denis. 2007. Réflexivisation, transitivité et agent affecté. In André Rousseau, Didier Bottineau &
Daniel Roulland (éds.), L’énoncé réfléchi, 83–106. Rennes: Presses Universitaires de Rennes.
Cúneo, Paola. 2013. Léxico y clasificación nominal. El dominio etnobiológico en toba. Munich: LINCOM.
De la Cruz, Luis María. 1995. Qomlajépi naleua, nuestra tierra. Los sitios que contienen tierra que da vida
a los tobas de Sombrero Negro de la provincia de Formosa. In José Braunstein (ed.), Hacia una carta
étnica del Gran Chaco VI, 69–114. Las Lomitas (Formosa, Argentina): Centro del Hombre Antiguo
Chaqueño.
DGEEC. 2013. III Censo Nacional de Población y Viviendas para Pueblos Indígenas. Pueblos indígenas en el
Paraguay. Resultados preliminares 2012. Asunción: Dirección General de Estadísticas, Encuestas y
Censos.
González, Raúl. 2010a. Análisis sintáctico y semántico de dos aplicativos locativos en toba (familia
guaycurú). Lingüística 24. 123–140.
González, Raúl. 2010b. Verbos seriales en toba (familia guaycurú). In Marisa Censabella & Raúl González
(ed.), Actas del II Encuentro de Lenguas Indígenas Americanas y II Simposio Internacional de Lingüística
Amerindia. Buenos Aires: CONICET. CD ROM.
González, Raúl. 2011. El comitativo y el recíproco en toba. In Ana Fernández Garay & Antonio Díaz Fernández
(eds.), Investigaciones sobre lenguas indígenas sudamericanas, 143–168. Santa Rosa: Universidad
Nacional de La Pampa.
González, Raúl. 2013a. Reorganización del sistema de número nominal en toba (familia guaycurú).
UniverSOS 10. 127–139.
176 Marisa Censabella

González, Raúl. 2013b. Índices pronominales en cláusulas con argumentos aplicados en toba (familia
guaycurú). In Marisa Censabella & Cristina Messineo (eds.), Lenguas indígenas de América del Sur II.
Morfosintaxis y contacto de lenguas, 59–72. Mendoza: SAEL.
González, Raúl. 2015. Estudio fonológico y morfosintáctico de la lengua toba hablada en el este de la
provincia de Formosa (Argentina). Munich: LINCOM.
González, Raúl. 2016a. Indexación de argumentos no-sujeto promovidos por aplicativos en toba del este de
Formosa (Argentina). Forma y Función 29(1). 29–61.
González, Raúl. 2016b. Construcciones seriales simétricas y asimétricas en toba del este de Formosa
(Argentina). Signo y Seña 30. 53–72.
González, Raúl & Marisa Censabella. 2009. El colectivo -pi y la concordancia aliterativa en toba desde la
teoría de la replicación gramatical. Paper presented at the 53° Congreso Internacional de Americanistas.
México DF, 19–24 July.
Gordillo, Gastón. 1992. Cazadores, recolectores y cosecheros. Subordinación al capital y reproducción social
entre los tobas del oeste de Formosa. In Héctor Trinchero, Daniel Piccinini & Gastón Gordillo (eds.),
Capitalismo y grupos indígenas en el Chaco Centro-Occidental, 13–91. Buenos Aires: Centro Editor de
América Latina.
Heine, Bernd. 2002. On the role of context in grammaticalization. In Ilse Wischer & Gabriele Diewald (eds.),
New reflections on grammaticalization, 83–101. Amsterdam: John Benjamins.
Hecht, Ana Carolina. 2010. “Todavía no se hallaron hablar en idioma”. Procesos de socialización lingüística en el
barrio toba de Derqui (Argentina). Munich: LINCOM.
INDEC. 2012. Censo Nacional de Población, Hogares y Viviendas 2010: Censo del Bicentenario: Resultados
definitivos, Serie B, Nº 2. Buenos Aires: Instituto Nacional de Estadísticas y Censos.
Kemmer, Suzzane. 1993. The middle voice. Amsterdam: John Benjamins.
Kittilä, Seppo. 2005. Recipient-prominence vs. beneficiary-prominence. Linguistic Typology 9(2). 269–297.
Klaiman, Miriam H. 1991. Grammatical voice. Cambridge: Cambridge University Press.
Klein, Harriet E. Manelis. 1981. Una gramática de la lengua Toba: Morfología verbal y nominal. Montevideo:
Departamento de Lingüística, Universidad de la República.
Medina, Mónica. 2017. Descripción etnográfica de la funcionalidad de la lengua toba en la comunidad de
habla qom del barrio Mapic (Resistencia, Chaco). Resistencia: Universidad Nacional del Nordeste
dissertation.
Mendoza, Marcela. 1999. The Western Toba: Family life and subsistence of a former hunter-gatherer. In
Elmer S. Miller (ed.), Peoples of the Gran Chaco, 81–108. Westpoint, Connecticut: Bergin & Garvin.
Mendoza, Marcela. 2002. Band mobility and leadership among the Western Toba hunter-gatherers of Gran
Chaco in Argentina. New York: Edwin Mellen Press.
Messineo, Cristina. 2003. Lengua toba (guaycurú). Aspectos gramaticales y discursivos. Munich: LINCOM.
Messineo, Cristina. 2014. Arte verbal qom. Consejos, rogativas y relatos de El Espinillo, Chaco. Buenos Aires:
Asociación Civil Rumbo Sur.
Messineo, Cristina, Javier Carol & Harriet E. M. Klein. 2016. Deixis y contacto en la región del Gran Chaco:
los demostrativos en las lenguas guaycurúes y mataguayas. International Journal of the Sociology of
Language 240. 119–157.
Mithun, Marianne. 2004. The non-universality of obliques. Paper presented at the Syntax of the World’s
Languages Conference (SWL 1). Leipzig: Max Planck Institute for Evolutionary Anthropology, 5–8 August.
Tovar, Antonio & Consuelo Larrucea Tovar. 1984. Catálogo de las lenguas de América del Sur. Nueva edición
refundida. Madrid: Gredos.
Van Valin, Robert & Randy LaPolla. 1997. Syntax: Structure, meaning and function. Cambridge: Cambridge
University Press.
Vidal, Alejandra. 2010. Descripción y explicación en dos lenguas chaqueñas: pilagá (guaycurú) y wichí
(mataguaya). RASAL 2010 (1/2). 135–155.
 6 Applicatives in Toba/Qom (Guaykuruan) 177

Zurlo, Adriana. 2011. La expresión del número verbal en toba. In Ana Fernández Garay & Antonio
Díaz-Fernández (eds.), Investigaciones sobre lenguas indígenas sudamericanas, 5–34. Santa Rosa:
Universidad Nacional de La Pampa.
Zurlo, Adriana. 2013. Cuando me voy en la picada. . . Análisis semántico de construcciones de movimiento
(ir/irse. . .a/en) en corpus oral de español (L2) hablado por tobas de Resistencia (Chaco). In Marisa
Censabella & Cristina Messineo (eds.), Lenguas indígenas de América del Sur II. Morfosintaxis y contacto
de lenguas, 131–146. Mendoza: Sociedad Argentina de Estudios Lingüísticos.
Zurlo, Adriana. 2016a. Sistema medio en toba y español (Chaco, Argentina). Estudio tipológico-funcional.
Resistencia: Universidad Nacional del Nordeste dissertation.
Zurlo, Adriana. 2016b. Voz media en Toba (Guaycurú)”. LIAMES 16 (2). 285–306.
Zurlo, Adriana. 2019. Construcciones de voz media y alternancias de la transitividad en lengua toba
(flia. Guaycurú). Documentação e Estudos em Linguística Teórica e Aplicada 35(2). 1–29.
Zurlo, Adriana & Marisa Censabella. 2013. Identificación de las propiedades aspectuales inherentes
en raíces verbales co-ocurrentes con el índice pronominal de 3ra persona r- en toba. Trabajo
exploratorio. In Marisa Censabella & Cristina Messineo (eds.), Lenguas indígenas de América del Sur II.
Morfosintaxis y contacto de lenguas, 73–91. Mendoza: Sociedad Argentina de Estudios Lingüísticos.
Zurlo, Adriana & Marisa Censabella. 2014. Usos de PERO en español (L2) hablado por tobas de Resistencia
(Chaco). In María Marta García Negroni (ed.), Marcadores del discurso: perspectivas y contrastes,
277–292. Buenos Aires: Santiago Arcos Editor.
Fernando Zúñiga
7 The applicative constructions
of Mapudungun
Abstract: This chapter surveys the morphology, syntax, and semantics of the applicative
constructions of Mapudungun (isolate; Chile/Argentina). The four markers addressed
are the verbal suffixes -l, -ñma, -tu, and -ye. With respect to syntax, the Mapudungun con-
structions introduce to the clausal core an argument in P role (which, given the symmet-
rical-voice make-up of clausal syntax in the language, appears as either a subject or an
object, depending on factors unrelated to applicativization). With respect to the semantic
role of the applied phrase, the markers are relatively underspecified, albeit with some
tendencies: l- and ñma-applicatives center around the notions of approach/benefaction
and separation/malefaction, respectively, but they can also express a broader Concern-
ee-Concern relation; tu-applicatives typically introduce Stimuli, Goals, or Patients; and
ye-applicatives introduce Comitatives or Speech/Thought Topics. The marker -l also caus-
ativizes (with some complicating aspects in the allomorphy found with both processes),
-ñma appears on verbal/adverbial roots expressing spatial, temporal, and manner
notions, and -tu and -ye also function as denominal verbalizers. The suffix -tu also has
a number of other functions, including telicization and the derivation of reversionary/
repetitive forms (both of which are frequent), as well as antipassivization (which is
severely restricted and infrequent).

1 Introduction
The present chapter describes the applicative constructions of Mapudungun (a.k.a.
Mapuche language or, in older studies, Araucanian; ISO 639-3 arn, Glottolog mapu1245).
This indigenous language is an isolate spoken in south-central Chile and west-central
Argentina by 150,000–250,000 speakers, to different degrees of fluency (Zúñiga and
Olate 2017) and has been in contact with Spanish since the 16th century. There are a
handful of very similar regional varieties (including obsolescent Huilliche, occasionally
considered a separate language). Grammatical applicative-related facts seem to show
negligible variation across the Chilean dialects; lexical and discourse-related aspects
may show important inter-dialectal variation. This chapter focuses on the conservative
registers of Central Mapudungun as spoken in the Araucanía Region in Chile.

Acknowledgments: I am grateful to Denis Creissels for his valuable comments on previous versions of
this chapter. His contribution was especially important for improving the presentation of analytical issues
in Section 2.

https://doi.org/10.1515/9783110730951-007
180 Fernando Zúñiga

The phenomena discussed here center around the opposition illustrated in (1).1 Suf-
fixing -ñma to the verb allows the clause to accommodate as a primary object a partici-
pant not expressible with the underived verb:

(1) a. Düngu-nge juez mew!

speak-2sg.imp judge postp
‘Speak to/with the judge!’ (base construction)
b. Düngu-ñma-fi-nge juez mew!
speak-appl-3.obj-2sg.imp judge postp
‘Speak with the judge in his/her favor!’ (applicative construction)

To judge from Polinsky’s (2013) classification of Mapudungun as not having applic-

atives, the topic of the present chapter may be considered controversial. Nevertheless,
Peterson’s (2007: 1) description of applicatives as “means some languages have for
structuring clauses which allow the coding of a thematically peripheral argument or
adjunct as a core-object argument” covers (1) rather easily. In fact, Polinsky’s own defi-
nition straightforwardly applies to (1): “[i]n an applicative construction, the number
of object arguments selected by the predicate is increased by one with respect to the
basic construction”. (Her source, Moesbach 1962, addresses several of the phenomena
treated here, but without using the label applicative.) Contemporary studies of Mapudu-
ngun morphosyntactic topics have employed the notion without discussing or review-
ing it (e.g., Baker, Aranovich, and Golluscio 2005; Golluscio 2007; Zúñiga 2010a).
I employ here the comparative notion of applicative constructions advanced in
Zúñiga and Creissels (this volume). An applicative construction (AC) contrasts in three
important respects with a base construction (BC) headed by a predicate built upon the
same root. First, the predicate in the AC shows additional overt marking that distin-
guishes it from the unmarked predicate in the BC. Second, the participant encoded as A
or S in the BC appears as A or S in the AC. Third, the applied phrase (AppP) of the AC, in
a role other than A or S, cannot be expressed in the BC at all, or it appears in the latter
as a non-core constituent with a different coding. This comparative notion, however, is
based on a Comrian model that uses the notions of A, P, and S understood as “syntactic
terms whose prototypes are defined in semantic terms” (Comrie 1989: 111), which is
insufficient to adequately capture Mapudungun verbal clause structure due to what I
analyze as a symmetrical voice opposition in Section 2.
The chapter is structured as follows. Section 2 sketches the necessary basic infor-
mation about Mapudungun morphosyntax and outlines what I think is the necessary

1 Unless otherwise specified, numbered examples come from elicitation sessions I conducted with
speakers in Chile in the early 2000s. The orthographic convention used here is the Alfabeto Mapuche
Unificado, I have unified the spelling, and all interlinear glosses are my own. Examples provide phono-
logically underlying forms; to arrive at surface forms, some minor elision, assimilation, resyllabifica-
tion, and epenthesis rules apply (Zúñiga 2006: Ch. VII.1.1).
 7 The applicative constructions of Mapudungun 181

adaptation to the basic Comrian A/P/S model in order to account for the structure of
transitive clauses in the language. Sections 3, 4 and 5 survey the morphological, syn-
tactic, and semantic characteristics of the applicatives of the language, respectively.
Section 6 outlines other uses of the applicative markers, and Section 7 summarizes the
main findings.

2 Basics of Mapudungun morphosyntax

2.1 Morphology

Mapudungun morphology is predominantly suffixing and agglutinative; it shows vir-

tually no flexivity or fusion of formatives, and most grammaticalized elements show
separative exponence and occur in concatenative strings. Nominal morphology is rather
limited: the only process that leads to some structural complexity there is productive
compounding. By contrast, verbs allow for verb-noun and verb-verb compounding and
can take a high number of derivational and inflectional affixes, only very few of which
need to be present for words to be well formed (Zúñiga 2017). Voice(-related) suffixes
include reflexive -w, Agent-less passive -nge, some causatives, and several applicatives
(see Golluscio 2010 and Zúñiga 2015 for more on transitivity, valency, and voice in
Mapudungun).
Extant descriptions of Mapudungun morphology propose a templatic, rather than
a layered or hybrid, structure of the verb complex. Nevertheless, accounts disagree
regarding both the template’s rigidity and the number of slots and fillers to be distin-
guished. Despite bearing some superficial similarities to Smeets’s (2008) account, the
model espoused here is closer to the partial proposals advanced in Augusta (1903), Har-
melink (1996), Salas (2006), and Golluscio (2010).

2.2 Basic syntax

In this study, the terms avalent, monovalent, bivalent, and trivalent refer to the syntactic
valency of predicates (and are abbreviated as “v0”, “v1”, etc.). Zero-argument, one-argu-
ment, two-argument, and three-argument describe clause types.
In Mapudungun, heads mark the relationships between themselves and some of
their dependents at the clause level, and participants can be subjects, primary and sec-
ondary objects, oblique objects, or adjuncts.2 Subjects and primary objects can bear a
wide range of semantic roles, not only due to the specific predicates that assign them

2 Instead of subject and object, other accounts have proposed focal and satellite person (Salas 2006),
and primary and secondary argument (Zúñiga 2015).
182 Fernando Zúñiga

to their arguments but also because of the alternation between actor and undergoer
voices, as will become apparent shortly. Secondary objects are usually Theme-like argu-
ments of trivalent predicates. The distinction between oblique objects and adjuncts is
semantic; the former express participants more involved in the predicate semantics
while the latter bear peripheral semantic roles. See Zúñiga (2019) for more details on
grammatical relations in Mapudungun.
The valency classes of Mapudungun underived predicates are summarized in
Table 1 below (cf. Zúñiga 2015). Examples of the predicate and clause types follow.

Table 1: Valency classes of Mapudungun underived predicates and associated clause types.

Predicate class Clause type Arguments

I. Avalent (e.g. mawün- ‘rain’) Zero-argument —

IIa. Monovalent (e.g. la- ‘die’) One-argument sbj
IIb. Extended monovalent (e.g. kon- ‘enter’) sbj + oobj
Two-argument
IIIa. Bivalent (e.g. mütrüm- ‘call’) sbj + pobj
IIIb. Extended bivalent (e.g. tüku- ‘put’) sbj + pobj + oobj
Three-argument
IV. Trivalent (e.g. elu- ‘give’) sbj + pobj + sobj

(2) a. Mawün-i.
rain-ind[3]
‘It rained.’ (I)
b. La-i Rayen.
die-ind[3] R.
‘Rayén died.’ (IIa)
c. Kon-i Nawel (Rayen mew).
enter-ind[3] N. R. postp
‘Nahuel entered (Rayén’s house).’ (IIb)
d. Mütrüm(-fi)-n Rayen. — Mütrüm-e-n-mew Rayen.
call-3.obj-1sg.ind R. call-inv-1sg.ind-3.obj R.
‘I called Rayén.’ (av, direct; IIIa) ‘Rayén called me.’ (uv, inverse; IIIa)
e. Tüku(-fi)-n ilo challa mew.
put-3.obj-1sg.ind meat pot postp
‘I put meat in the pot.’ (av, direct; IIIb)
f. Elu-fi-n makuñ. —
give-3.obj-1sg.ind blanket
‘I gave him/her a blanket.’ (av, direct; IV)
Elu-e-n-mew makuñ.
give-inv-1sg.ind-3.obj blanket
‘S/he gave me a blanket.’ (uv, inverse; IV)
 7 The applicative constructions of Mapudungun 183

There are two kinds of clauses headed by bivalent or trivalent predicates: actor-
voice (av) and undergoer-voice (uv) clauses, which feature direct and inverse verb
forms, respectively. The choice between these clause types is based on person and
animacy of the subject and the primary object, as well as on their relative topicality
(Zúñiga 2006: Ch. VII). A simplified nominal hierarchy like [1/2 > 3 topical > 3 non-top-
ical] will suffice for our present purposes. In interactions between two animate refer-
ents, when the agentive core argument outranks the non-agentive core argument on
the nominal hierarchy, av clauses are used; uv clauses are used when the converse sit-
uation holds. The referential status of secondary objects (e.g., Theme-like arguments
of trivalent predicates) and oblique objects is irrelevant for the choice between clause
types. In both clauses in (d), the 1st person is the subject and the woman is the primary
object; Agent and Patient are the subject and the primary object in the av clause on the
left-hand side, and the primary object and the subject in the uv clause on the right-hand
side, respectively. In (f), Agent, Recipient, and Theme are the subject, the primary object,
and the secondary object in the av clause on the left; they are the primary object, the
subject, and the secondary object in the uv clause on the right. The agentive 3rd-person
object in uv clauses is invariably head-marked as -mew. The conditions governing the
appearance of the non-agentive 3rd-person object marker -fi in av clauses are complex;
-fi invariably occurs with trivalent predicates like elu- ‘give’ in (f) and variably occurs
with bivalent predicates like mütrüm- ‘call’ in (d), depending on animacy, definiteness,
and topicality of the object, as well as the proper-common distinction (Zúñiga 2010b).
Thus, derived and underived bivalent and trivalent predicates occur in the actor-
voice direct form, the undergoer-voice inverse form, or in either, depending on the core
arguments’ inherent semantic and relative pragmatic feature values.3 Instead of a one-
to-one correspondence between the syntactic roles and the coding characteristics of
core arguments, we find a correspondence mediated by a system of symmetrical voices,
as depicted in Table 2. A (the argument of transitive verbs whose morphosyntactic
behavior is that of the Agent of prototypical transitive verbs) is the subject in av clauses
and the object in uv clauses; P (the argument of transitive verbs whose morphosyntac-
tic behavior is that of the Patient of prototypical transitive verbs) is the subject in uv
clauses and the object in av clauses.4

3 This holds not only for high-transitivity predicates like langüm- ‘kill’ and i- ‘eat’ (cf. the “prototypical
transitive verbs” in Creissels, forthcoming) but also for low-transitivity ones like pe- ‘see’ and ayü- ‘love’.
Some high-transitivity verbs (e.g., trafo- ‘break’) never occur in inverse forms because their Patient is
inanimate, and two-argument constructions with inanimate Agents and animate Patients are banned
in Mapudungun.
4 This symmetrical-voice account refers to syntax only. Both the morphology and the discourse fre-
quency of direct and inverse forms are asymmetrical; the formal direct-inverse opposition is privative,
rather than equipollent (see [2d] and [2f]), and inverse forms are seldom found in narrative texts (de-
spite some being obligatory to encode those specific states of affairs, viz. those involving interactions
between 1st/2nd and 3rd persons).
184 Fernando Zúñiga

Table 2: Grammatical relations in transitive clauses.

A P

Actor voice (direct) subject primary object

Undergoer voice (inverse) primary object subject

Constituent-order regularities have not been explored systematically yet, but what
we know so far about the flexible and (possibly) dominant VS+VO order of Mapudun-
gun can be summarized as follows. Typical verbal clauses have no overt argument NP,
or only one. Typical one-argument verbal clauses have their subject NP immediately
after the predicate (2b). Lastly, typical two-/three-argument verbal clauses have their
object NPs immediately after the predicate (2c–f) (if both objects are present, the sec-
ondary object usually precedes the primary object).

2.3 Valency/voice operations

Passivization with -nge applies only to bivalent and trivalent predicates; it suppresses
the Agent, promotes the Patient/Recipient from primary object to subject, and leaves
the Theme secondary object (if any) unaltered. The passive counterparts of (2d/f) are
given in (3). Note the new two-argument clause type IIIc illustrated in (b), which has a
non-agentive subject and a secondary object; this plays a role in the discussion of an
applicative lookalike subtype in Section 6.2.1.

(3) a. Mütrüm-nge-n.
call-pass-1sg.ind
‘I was called / someone called me.’ (IIa)
b. Elu-nge-n makuñ.
give-pass-1sg.ind blanket
‘I was given a blanket.’ (IIIc)

Nominal incorporation (NI) is productive with bivalent verbs, possible but lexically
constrained with trivalent verbs, and restricted with monovalent verbs. At first sight,
bivalent verbs seem to become monovalent when they incorporate their Patient/Theme
(4); we will see in Section 4.3, however, that they can actually become labile.

(4) a. Weñe-ke-n sañchu.

steal-hab-1sg.ind pig
b. Weñe-sañchu-ke-n.
steal-pig-hab-1sg.ind
Both: ‘I usually steal pigs.’
 7 The applicative constructions of Mapudungun 185

Non-agentive monovalent verbs allow NI, but only if the new subject and the incorpo-
rate are in a possessive relation (5) (Baker, Aranovich, and Golluscio 2005). This is nor-
mally found in spontaneous discourse with body parts, but examples with owned items
are not difficult to find:

(5) a. Lüf-i (iñche) ñi ruka.

burn-ind[3] 1sg.pro 1sg.psr house
b. (Iñche) lüf-ruka-n.
1sg.pro burn-house-ind-1sg.ind
Both: ‘My house burned down.’

The voice operators addressed in this chapter are nucleativizers: they allow par-
ticipants not encoded as core terms in base constructions (BCs) to be encoded as such in
derived monoclausal constructions (see Zúñiga and Creissels, this volume). Causatives
introduce a causer-Agent in A role, and applicatives a non-Agent in P role, to the clause.
Unlike in many other languages, rather than invariably introducing these arguments
as subjects and objects, respectively, in Mapudungun the new arguments’ realization is
contingent on the choice between actor and undergoer voices; when both highest-rank-
ing participants are animate, these nucleativizers lead to constructions that systemati-
cally alternate. With agentive nucleativizers (ans), the causer in A role is the subject
with the actor-voice direct form (6b) and the primary object with the undergoer-voice
inverse form (6c):

(6) a. Aye-n.
laugh-1sg.ind
‘I laughed.’
b. Aye-l-fi-n ñi wenüy.
laugh-caus-3.obj-1sg.ind 1sg.psr friend
‘I made my friend laugh.’ (direct, av)
c. Aye-l-e-n-mew ñi wenüy.
laugh-caus-inv-1sg.ind-3.obj 1sg.psr friend
‘My friend made me laugh.’ (inverse, uv)

With non-agentive nucleativizers (nans), the new argument in P role is the primary
object with the actor-voice direct form (7a) and the subject with the undergoer-voice
inverse form (7b):

(7) a. Aye-tu-fi-n ñi wenüy.

laugh-appl-3.obj-1sg.ind 1sg.psr friend
‘I laughed at my friend.’ (direct, av)
186 Fernando Zúñiga

b. Aye-tu-e-n-mew ñi wenüy.
laugh-appl-inv-1sg.ind-3.obj 1sg.psr friend
‘My friend laughed at me.’ (inverse, uv)

Thus, with such Mapudungun nucleatives, there is a disconnect between the intro-
duction or promotion of a core argument and the realization of that argument (see
Figure 1 below). The direct-verb, actor-voice, Examples (6b) and (7a) instantiate causa-
tives and applicatives, respectively, in the familiar sense; the inverse, undergoer-voice,
Examples (6c) and (7b) are their syntactic mirror images:

Default Mapudungun
AN NAN AN NAN
SBJ CAUS SBJ (6b) (7b)
OBJ APPL OBJ (6c) (7a)
CAUS APPL

Figure 1: Two selected nucleative opposition types.

3 Morphology
The four applicative markers in the language are the verbal suffixes -tu, -ye, -l, and -ñma.
No serial verb constructions or converbal constructions in the language qualify as ACs.5
The markers -tu and -ye are invariable; -l and -ñma show phonologically conditioned
allomorphy. The allomorphs of -l are: -l after vowels, -ül after w, and -el elsewhere. Note
that the suffix commonly appears as -lel with some vowel-final roots (viz. i- ‘eat’, pe- ‘see’,
pi- ‘say’, ina- ‘follow’, kullkü- ‘reserve [domestic animals]’, and kintu- ‘look for’), and as
-yel (with epenthetic y) with la- ‘die’ (see [36]). The allomorphs of -ñma are: -ñma after
vowels and -ma elsewhere. Nevertheless, the suffix consistently appears as -üñma after
the m-causative and some m-final roots—viz. kim- ‘know’ and rungüm- ‘grind’. Other
exceptions include roots like wew-üñma- ‘win on (sbdy.)’.6

5 I do not address here two markers that occur only with few verbs, namely -mpe ~ -ñpe (e.g., illku- ‘get
angry’ > illku-ñpe- ‘call names’; cf. Smeets 2008: 310) and -kütuye (e.g., rüpu- ‘sever’ > rüpu-kütuye- ‘carve
[with an axe]’, Augusta 1916: 199; cf. Smeets 2008: 288f). Verb compounding with künu- ‘leave’, nie- ‘have’,
and tüku- ‘put’ also lies beyond the scope of the present chapter; see Augusta (1903: 261–268) and Smeets
(2008: 293f, 316).
6 Augusta (1916: 57) records the anomalous minimal pair ngüf-ma- ‘have (a body part) blocked’ vs.
ngüf-üñma- ‘be surprised by dusk’. Upon closer inspection, the latter meaning appears to be regularly
expressed by ngüfke-ñma- (or other verbs built on a non-cognate root); the unexpected form ngüf-üñma-
seems to occur only sporadically and in unsystematic variation nowadays. (I am grateful to Aldo Olate
for his help with this issue.)
 7 The applicative constructions of Mapudungun 187

All four markers are fully grammaticalized elements. The suffix -tu—actually, all
tu-suffixes, see further down—probably originated from the verbal root tu- ‘get, take’.
Strictly verbal -ye is likely a grammaticalized version of ye1- ‘carry, bring’, and denom-
inal -ye probably originated in ye2- ‘regard as’. The etymon of -l may be el- ‘set, put’,
elu- ‘give’, or wül- ‘give away, hand’. The suffix -ñma might be related to the verbal root
man- ‘be lucky’, recorded in Augusta (1916: 131) for the Huapi dialect, but it might also
have been originally polymorphemic (Zúñiga 2009b).7
The four markers show variation regarding their specificity. First consider applica-
tive -l vis-à-vis causative -l. Mapudungun has four morphological causatives, two of
which can co-occur with the applicatives and are particularly common. The m-caus-
ative is restricted to a closed but non-negligible class of monovalent verbs where the
causee is low in animacy/control, for instance, wadkü-m-i ko (boil-caus-ind[3] water)
‘s/he boiled the water’. By contrast, the l-causative is productive and used frequently
with monovalent verbs where the causee is high in animacy/control, for instance, aye-
l-e-n-mew (laugh-caus-inv-1sg.ind-3.obj) ‘s/he made me laugh’. A number of verbs can
take either suffix (e.g., tran- ‘fall’ > tran-üm- ‘fell’ vs. tran-el- ‘knock down’); see Golluscio
(2007) for details. The allomorphy facts of causative -l mirror those of applicative -l,
with the caveat that -lel is only an applicative morph. I therefore see this as an erstwhile
instance of the much-cited causative-applicative polysemy that, in this case, has given
rise to two different templatic slots—the causative appears closer to the root (see [8b]),
and several other derivational markers can come in between both positions—and two
functionally distinct fillers with very similar allomorphs.8 In addition, -l is a denominal
verbalizer.
Second, applicative -ye has three homonyms: a denominal verbalizer, an unclear
and obsolescent aspectual marker denoting completion or duration, and a marker
denoting distributivity or multiplicity.
Third, some ambiguous elements that can be either verbal roots or adverbs and
express spatial, temporal, and manner notions can appear extended by -ñma. Those
elements can occur either on their own (e.g., llekü- ‘[get] close’, fentre- ‘[be] much’, and
welu- ‘but; cross [v1]’), with -ñma in derived verbs (e.g., llekü-ñma- ‘approach [v2]’, fen-
tre-ñma- ‘last a long time’, welu-ñma- ‘mistake [v2]’), or with -ñma in adverbs (e.g., lle-
kü-ñma ‘near’, fentre-ñma ‘for a long time’, welu-ñma ‘upside down, backwards’). Except
for these “relationalizers”, -ñma does not seem to have any homonyms.9

7 Evidence for this comes from the existence of autobenefactive -ñmu (e.g. in küdaw-ñmu-n ‘I worked
for my own benefit’, Smeets 2008: 274). This element may be etymologically related to -ñma—e.g., via
an erstwhile composition along the lines of ✶-ñ(-)m(-a) ‘appl’ + ✶-u ‘refl’—but it is not an applicative.
8 Smeets (2008) postulates three different morphemes -(e)l, -(ü)l and -(l)el in three separate templatic
slots. In her account, the suffix -(ü)l denotes higher affectedness of the primary object (pp. 287–288).
9 Smeets (2008: 276f, 301f) postulates two different morphemes -(ñ)ma (“experience”) and -(ü)ñma (“in-
direct object”) in two separate templatic slots.
188 Fernando Zúñiga

Lastly, it is still unclear how to best analyze the instances of grammaticalized tu’s
in the verbal complex. At least three homophonous suffixes need to be distinguished
synchronically: the denominal verbalizer (e.g., kofke ‘bread’ > kofke-tu1- ‘eat bread’),
the telicizer (e.g., nge- ‘be’ > nge-tu2- ‘become’), and the reversionary/repetitive (e.g.,
aku- ‘arrive’ > aku-tu3- ‘arrive back’). (One of the ways to build iterative stems might be
a special case of the reversionary/repetitive, e.g. rüngkü- ‘jump’ > rüngkü~rüngkü-tu-
‘bounce’; see Zúñiga and Díaz-Fernández 2014). The applicative and the reversionary/
repetitive could be originally the same as the antipassive -tu4 (e.g., imül- ‘roll’ [v2] >
imül-tu4- ‘roll for fun [v1]’). The valency-neutral suffixes (e.g., kedi[-tu]- ‘shear [v2]’ and
kewa- ‘fight [v1]’ > kewa-tu- ‘fight for fun [v1]’) might also be instances of a bleached
erstwhile applicative, and I regard them as such here. In sum, I tentatively distinguish
deverbal -tu1, aspectual -tu2 and -tu3, and a broad polysemous -tu4 that has applicativi-
zation as one of its functions.
Applicatives can co-occur with causatives on the one hand and passives on the
other; the former co-occurrence seems to be subject to semantic/pragmatic restrictions
only; I have not found any restrictions for the latter. Regarding the combinability of
the applicative suffixes among themselves, valency-altering -tu and -ye do not appear
to occur twice on the same verb. They can occasionally combine with -l and -ñma, for
instance: tüku- ‘put’ > tüku-tu- ‘put at/on’ > tüku-tu-l- ‘put at/on for’. How freely -l and
-ñma can combine with themselves or with each other is controversial; Smeets (2008:
280) says that the combinations -l+-ma and -ñma+-ñma are infrequent and unacceptable,
respectively. Salas (2006:122) found that ñma-applicativization of an applicative form
is admissible, at least with passivized verbs (8). My own work confirms Salas’s findings
for active clauses as well, albeit with some variation and uncertainty on the part of the
speakers.

(8) a. Weñe-ñma-ñma-nge-i-m-i waka tami fotüm.

steal-appl-appl-pass-ind-2-sg cow 2sg.psr son.of.man
‘Someone/people stole your son’s cow.’
b. Küpa-l-el-ma-nge-i-m-i kuram tami ñuke.
come-caus-appl-appl-pass-ind-2-sg egg 2sg.psr mother
‘Someone/people brought your mother eggs.’

Augusta (1903: 62) reports the following instance of l-stacking that does not involve
causatives, and I have been able to find analogous forms rather easily in elicitation
with other bivalent verbs. In such examples, each -l simply accommodates an AppP of
its own into the clause (see § 4):

(9) a. Nentu-en!
remove-2sg→1sg.ind
‘Get me out (e.g., of jail)!’
 7 The applicative constructions of Mapudungun 189

b. Nentu-l-en wayun!
remove-appl-2sg→1sg.ind thorn
‘Remove the thorn from me!’
c. Nentu-l-el-en ñi wayun tañi fotüm!
remove-appl-appl-2sg→1sg.ind 3.psr thorn 1sg.psr son.of.man
‘Remove the thorn from my son!’

Instances of ñma-stacking are rare in published sources but can be found, as illus-
trated in (10). In (10a), anü- ‘sit down’ first accommodates via ñma-suffixation the
location on which the interlocutor sits as core argument, and then further accommo-
dates the location’s possessor, that is, the speaker. In (10b), with üfülü- ‘sip, suck up’, the
first -ñma is syntactically neutral—it merely changes the verb’s meaning from ‘sip’ to
‘swallow’—and the second accommodates the possessor of the ship, who appears as
subject of an uv clause because it is animate (unlike the inanimate but potent waves):

(10) a. Anü-ñma-ñma-ki-eli ñi lifro!

sit-appl-appl-neg-2sg→1sg.subj 1sg.psr book
‘Don’t you (sg) sit on my book!’ (Augusta 1916: 9)
b. Foche üfülü-ñma-ñma-e-i-mew ñi nafiu.
wave suck.up-ñma-appl-inv-ind[3]-3.obj 3.psr ship
‘The waves swallowed his ship.’ (Augusta 1916: 271)

Mapudungun applicatives are not systematically restricted by either tense-aspect-

mood values or the direct-inverse opposition, nor do they occur less often on main-
clause finite verb forms than on subordinate nonfinite forms. Verbs with more than
two applicatives, which would normally have more than three objects in the clause, are
absent in extant texts and are virtually always rejected in elicitation. There seems to be
a processing-related tendency not to applicativize verbs whose morphological make-up
is already considerably complex, which might explain why applicativization is seldom
found together with nominal incorporation, even though there is no hard constraint on
their co-occurrence.

4 Syntax
4.1 General comments

Mapudungun applicatives are P-applicatives (Creissels, forthcoming; Zúñiga and Creis-

sels, this volume). Most are transitivizing, but some are redirecting; the latter are valen-
cy-neutral but not syntax-neutral (i.e., they affect argument realization).
190 Fernando Zúñiga

Regarding the status of the applied phrase in the derived construction, Mapudu-
ngun applicatives are either primary-objective or subjective; the correspondences
between predicate classes, clause argument inventories, and applicative types are sum-
marized in Table 3 below.

Table 3: Mapudungun transitivizing applicative types.

base predicate/clause derived predicate/clause

A IIa2/IIb sbj (+oobj) IIIa sbj + pobj

B IIIa/IIIb sbj + pobj (+oobj) IV sbj + pobj + sobj
C IV sbj + pobj + sobj V sbj + pobj + 2x sobj

With trivalent, bivalent, and most monovalent base predicates, primary-objective and
subjective applicatives regularly alternate, as mentioned in Section 2.3. Argument reali-
zation is determined by the rules that govern morphosyntactic inversion and the oppo-
sition between actor-voice and undergoer-voice clauses: the AppP is a primary object
in an av clause and a subject in an uv clause. Note that one clause type instantiated
in derived clauses is not available with underived predicates, namely four-argument
clauses (Clause Type V, with Applicative Type C).
Regarding the status of any companion objects to the AppP in the clause, base primary
objects appear as secondary objects in the derived construction; there are no double-
or triple-object constructions in Mapudungun in the sense of syntactically equivalent
primary objects.
Regarding the status of the semantic equivalent of the AppP in the base construc-
tion, two cases must be distinguished. If the referent expressed by the AppP is a Con-
cernee (or “External Possessor”; see Van de Velde 2020), it can be expressed as a non-ar-
gumental, NP-internal, constituent in the base clause, even though speakers consider
such constructions stilted or unidiomatic, especially with kinship terms and part-whole
relations (optional applicatives). By contrast, non-Possessors could be expressed in an
adpositional phrase in the base clause in some very few instances in principle (e.g., with
the semantically void postposition mew), but such clauses are virtually never found
outside of elicitation, and even there they are often semantically opaque and strongly
dispreferred (obligatory applicatives).

4.2 Individual markers/constructions

The suffix -tu can have a syntactic effect, either transitivizing (e.g., Type A, ad- ‘be beau-
tiful’ > ad-tu- ‘find beautiful’, and Type B, ütrüf- ‘throw’ > ütrüf-tu- ‘throw at’) or, rather
exceptionally, redirecting (e.g. ingka- ‘help’ > ingka-tu- ‘ask for help’, Augusta 1916: 65). The
marker can also have a syntactically neutral and even a valency-decreasing effect (see § 6).
 7 The applicative constructions of Mapudungun 191

Ye-suffixation can also be syntax-neutral (see § 6), but it often has a syntactic effect,
either transitivizing (e.g., Type A, ngüma- ‘weep’ > ngüma-ye- ‘weep about, mourn’) or
redirecting (with n[g]ütram- ‘tell, narrate’, pi- ‘tell, say’, and wifül- ‘throw [liquid]’). See in
(11) how bivalent/trivalent n(g)ütram- ‘tell (sthg.), tell (sthg.) to (sbdy.)’ becomes strictly
trivalent with -l (11b) and bivalent while changing the semantic role of the object with
-ye (11c):

(11) a. Nütram-n tüfachi epew. / Nütram-fi-n tüfachi epew.

tell-1sg.ind this story tell-3.obj-1sg.ind this story
‘I told this story.’ ‘I told this story to him/her/them.’
b. Fey fey nütram-el-e-n-mew.
dem dem tell-appl-inv-1sg.ind-3.obj
‘S/he told me that.’ (Augusta 1916: 153)
c. Nütram-ye-i chi weychan.
tell-appl-ind[3] art war
‘S/he talked about the war.’ (Augusta 1916: 153)

The statement-question pair in (12) illustrates Type-B l-applicativization of the

underived bivalent base predicate dewma- ‘make’ in a sentence pair (data from Augusta
1903: 75). Kiñe makuñ ‘a blanket’, the primary object in (12a), is a secondary object in
(12b); the third person asked about is the primary object in (12b), which has access to
fi-indexing on the verb and is the subject of a corresponding passive.

(12) a. Tañi ñuke dewma-ke-i kiñe makuñ.

1sg.psr mother make-hab-ind[3] one blanket
‘My mother is making a blanket.’ (BC)
b. Iney dewma-l-ke-fi-i kiñe makuñ tami ñuke?
who make-appl-hab-3.obj-ind[3] one blanket 1sg.psr mother
‘For whom is your (sg) mother making a blanket?’ (AC)

In such av clauses, the subject remains unaltered by applicativization; this holds for all
markers except for those cases where -ñma behaves like an applicative and a passive
combined (see § 6.2.1). If the AppP is a highly topical 3rd person or a speech-act partic-
ipant, however, an uv clause must be used, in which the non-agentive new argument
is the subject; both the Theme (kiñe makuñ ‘a blanket’) and the Agent (tami ñuke ‘your
mother’, the base subject) are objects (13):

(13) Dewma-l-ke-e-n-mew kiñe makuñ tami ñuke.

make-appl-hab-inv-1sg.ind-3.obj one blanket 2sg.psr mother
‘Your mother made a blanket for me.’
192 Fernando Zúñiga

L-applicativization is not productive with underived monovalent predicates (i.e., Type

A); cases like wirar- ‘shout (v1)’ > wirar-el- ‘shout at’ amount only to a handful of verbs
and show significant variation in how speakers use them and in how confident the
latter are about their usage.
The pair in (14) illustrates Type-B ñma-applicativization of bivalent nütu- ‘take’:

(14) a. Nütu-n kiñe kawello.

take-1sg.ind one horse
‘I took the horse.’ (BC)
b. Nütu-ñma-fi-n ñi kawello.
take-appl-3.obj-1sg.ind 3.psr horse
‘I took his/her horse.’ (AC)

The following examples show how an underived monovalent verb in (15a) becomes
bivalent in (15b) (Type A) and, after further applicativization, trivalent in (15c) (Type B).
Unlike with l-applicativization, instances like (15b) are rather frequent; those like (15c)
seem to be less common (see §§ 3 and 5).10

(15) a. Feyti küpa-lu anü-a-i.

dem come-ptcp sit.down-fut-ind[3]
‘Those who have come will have to sit down.’ (Smeets 2008: 89)
b. Anü-ñma-ki-lnge chumpiru!
sit.down-appl-neg-2sg.subj hat
‘Do not sit on the hat!’ (Augusta 1916: 9)
c. Anü-ñma-nie-ñma-en ñi makuñ.
sit.down-appl-asp-appl-2sg→1sg.ind 1sg.psr blanket
‘You (sg) sat on my coat.’ (Smeets 2008: 303)

Underived trivalent predicates are not numerous, but they allow Type-C applicativi-
zation with -l and -ñma. Labile pi- ‘tell/say (to)’—the verb can take one, two, or three core
arguments—behaves as expected when trivalent, both syntactically and semantically. In
an av applicative construction, the Theme appears as secondary object and the primary
object is a Beneficiary or Maleficiary, often construed via kinship (in uv applicative con-
structions, subjects and primary objects swap places): pi-lel- means ‘tell (sbdy.) (sthg.) on
behalf of’ and pi-ñma- ‘to tell (sbdy.) (sthg.) to the detriment of’. Trivalent müntu- ‘take
away’ behaves alike. By contrast, trivalent elu- ‘give’ in (16) behaves as expected only
with -l (16a); with -ñma (16b), there is no syntactic effect and the reading is one of per-
missive causation (and the suffix is actually optional or “emphatic” in that construction:

10 Golluscio (2010: 738) says that ñma-suffixation can be valency-neutral with bivalent base predicates,
which I have found to be the case only exceptionally.
 7 The applicative constructions of Mapudungun 193

(16) a. Elu-l-en kofke tañi witran!

give-appl-2sg→1sg.ind bread 1sg.psr foreigner
‘Give bread to my foreigner!’ (Augusta 1916: 39)
b. Ñi chaw elu(-ñma)-e-n-mew ñi fem-a-el.
1sg.psr father give-appl-inv-1sg.ind-3.obj 1sg.psr do-fut-nfin
‘My father has allowed me to do it.’ (Augusta 1916: 39)

Morphosyntactic inversion operates exactly as with bivalent base predicates (see the
uv clause in [16b]); with nontopical 3rd-person AppPs, av clauses are used, in which the
(new) non-agentive participant is the primary object (17):

(17) a. Fey pi-lel-fi-n tañi chaw.

dem tell-appl-3.obj-1sg.ind 3.psr father
‘I told that to her/his father for her/him.’
b. Ñi chaw elu(-ñma)-fi-i ñi fem-a-el.
3.psr father give-appl-3.obj-ind[3] 3.psr do-fut-nfin
‘Her/his father has allowed her/him to do it.’

4.3 Summary and further points of interest

The syntactic effects of different applicatives are summarized in Table 4 below.

Table 4: Mapudungun transitivizing applicative types.

-tu -ye -l -ñma

Valency-increasing/
Transitivizing
Type A: IIa/IIb>IIIa ( )
Type B: IIIa/IIIb>IV
Type C: IV>V ( )

Valency-neutral/
Redirecting
IIa>IIa ( )

Note how applicativization interacts with nominal incorporation. Example (4)

above illustrated the latter process with the bivalent verb weñe- ‘steal’, which is also
used in (18) below (from Augusta 1903: 293). Applicativization works as expected with
the default clause (18a), turning the bivalent verb trivalent, but the verb-noun com-
pound in (18b) does not take -ñma (or any other applicative marker) in order to license
the third argument; as mentioned in Section 2 above, weñe-sañchu- is labile (bivalent/
trivalent) and means ‘steal a pig / pigs (from)’:
194 Fernando Zúñiga

(18) a. Weñe-ñma-e-n-mew ñi sañchu

steal-appl-inv-1sg.ind-3.obj 1sg.psr pig
b. Weñe-sañchu-e-n-mew.
steal-pig-inv-1sg.ind.3.obj
Both: ‘S/he stole my pig from me.’

Something analogous happens with incorporating monovalent verbs (19) (cf. Example
[5]). The simple stem can be applicativized and become bivalent (19a), but the strictly
monovalent compound verb lüf-ruka- cannot (19b):

(19) a. Lüf-ma-n ñi ruka.

burn-appl-1sg.ind 1sg.psr house
b. Lüf-ruka(✶-ma)-n.
burn-house-appl-1sg.ind
Both: ‘My house burned down on me.’

In fact, this alternation between applicativized and incorporating versions of non-agen-

tive monovalent base predicates with possessors appears to be systematic with body
parts, as in (20):

(20) a. Ngüf-ma-n ñi yüw.

get.blocked-appl-1sg.ind 1sg.psr nose
b. Ngüf-yüw-n.
get.blocked-nose-1sg.ind
Both: ‘I got my nose blocked.’

Now note that transitivizing applicativization with -l and -ñma seems to make the
marking of the new argument on the predicate obligatory, not unlike the situation found
in three-argument clauses.11 1st and 2nd person core arguments are always overtly
marked in some way in two- and three-argument BCs,12 but 3rd-person primary object
show variation in direct verb forms (they are invariably marked on inverse forms). For
instance, the primary object of bivalent leli- ‘look at’ can, but need not, be marked, while
the primary object of trivalent elu- ‘give’ must be marked (21):13

11 I have found this to be the case with most older speakers, but the speech of some younger speakers
shows the pattern reported by Golluscio (2010: 721–722): fi-marking of the primary objects of derived
verbs is as fluid as the one found with underived bivalent verbs, rather than as rigid as found with tri-
valent verbs. More research is needed here.
12 Finite verbs expressing 2sg→1sg interactions constitute an exception: they take the endings -e-n
(-inv-1sg.ind) and -e-li (inv-1sg.subj); see Zúñiga (2006: Ch. VII) for details. This, however, is unrelated to
applicativization.
13 This is related to the conditions governing differential object marking mentioned in Section 2.2.
 7 The applicative constructions of Mapudungun 195

(21) a. Feyti ngürü leli-nie(-fi)-i tañi malle.

art fox look.at-asp-3.obj-ind[3] 3.psr paternal.uncle
‘The fox was watching his paternal uncle.’ (based on Salas 2006: 274)
b. Fey elu-✶(fi)-i chi pu trewa.
then give-3.obj-ind[3] art pl dog
‘Then he gave [them] to the dogs.’ (Salas 2006: 262)

Both applicativized trivalent verbs (21a) and applicativized bivalent verbs (21b) require
the 3rd-person primary object marker -fi to appear on the verb. This is illustrated with
-ñma here, but is independent of the exact applicative used (22):

(22) a. Leli-ñma-✶(fi-)n ñi ñawe.

look.at-appl-3.obj-1sg.ind 3.psr daughter.of.man
‘I looked at his daughter (with bad intentions).’ (Salas 2006: 120)
b. Tofkü-ñma-✶(fi-)n.
spit-appl-3.obj-1sg.ind
‘I spit at him/her.’ (cf. tofkü-n spit-1sg.ind ‘I spat’)

Applicativization is naturally connected to access to relativization, because attrib-

utive constructions are headed by core arguments only. In the following example, the
head noun is the subject in (23a) and the secondary object in (23b); the latter is the most
frequent kind of occurrence in narratives:

(23) a. Chi wentru küme-künu-tu-fi-lu chi domo

art man be.good-leave-appl-3.obj-ptcp art woman
ka amutu-i.
also go.away-ind[3]
‘The man who had cured the woman also went away.’ (Salas 2006: 265)
b. Xosé ñi ngilla-ñma-fiel-chi libru Xuan ñam-küle-i.
J. 3.psr buy-appl-nfin-attr book J. get.lost-res-ind[3]
‘The book which José bought from Juan is lost.’ (Smeets 2008: 354)

Applicativization does not seem connected to specific restrictions on the access to dis-
course-related operations like focalization, but more research is needed here.

5 Semantics
With -tu and -ye, the interpretation of the AppP normally relies on the marker and the
semantics of the base predicate. With -l and -ñma, the exact reading depends on an
interplay between the applicative marker, the specific semantics and syntactic valency
196 Fernando Zúñiga

of the base predicate, and contextual features. (Zúñiga 2010a provides additional details
of the l-ñma opposition from a semantic perspective.)
The semantic role of the AppP with valency-increasing tu-verbs is often predicta-
ble; such arguments are usually Stimuli of experiencer verbs (e.g., nümu- ‘smell [v1]’ >
nümu-tu- ‘smell [v2]’), Goals of motion verbs (e.g., nag- ‘descend’ > nag-tu- ‘descend
towards’), or Patients of change-of-state verbs (e.g., pütre ‘burn [v1]’ > pütre-tu- ‘burn
[v2]’). When syntactically neutral but semantically relevant, some alternations involv-
ing -tu are otherwise transitivizing, with tu-marked verbs denoting more intensity (e.g.,
ngeyku- ‘rock’ > ngeyku-tu- ‘rock strongly’) or more affectedness (e.g., ñidüf- ‘sew’ >
ñidüf-tu- ‘mend’). Many verbs show semantic idiosyncrasies not clearly related to such
effects.
Verbal ye-suffixation as illustrated in (24) introduces AppPs as motion Comitatives
(a) or Topics of Speech/Thought (b). With denominal ye-verbs, the AppP expresses a
social relationship (‘regard as N’, [24c]):

(24) a. Amu-ye-fi-n ñi wenüy.

go-appl-3.obj-1sg.sbj.ind 1sg.psr friend
‘I went with my friend.’
b. Ngüma-ye-fi-n ñi ñawe.
weep-appl-3.obj-1sg.sbj.ind 1sg.psr daughter.of.man
‘I wept for my daughter.’
c. Patron-ye-nge-n.
boss-vblz-pass-1sg.sbj.ind
‘I am considered the boss.’

Syntactically neutral ye- is rather unpredictable semantically, as are some instances of

redirecting ye- (25) (data from Augusta 1916: 260):

(25) a. Wifül-nge-i mollfüñ.

spatter-pass-ind[3] blood
‘They were spattered with blood.’
b. Wifül-ye-i ñi külen chi trewa.
spatter-appl-ind[3] 3.psr tail art dog
‘The dog spatters (around with) its (wet) tail.’

All instances of -l or -ñma with trivalent base predicates seem to introduce a Con-
cernee (or “External Possessor”), usually via a kinship or part-whole relation, as in (26).
This holds not only with underived bases (26a) (except for elu-ñma- ‘allow’ in [16b]),
but also with derived ones (26b). Although near-equivalent states of affairs can be con-
veyed via other constructions (viz. with NP-Internal Possessors as non-arguments),
External-Possession applicatives are the preferred, idiomatic, mode of expression (also
with those derived from bivalent derived bases, like [26c]):
 7 The applicative constructions of Mapudungun 197

(26) a. Elu-l-fi-n sañchu tañi wenüy tañi fotüm.

give-appl-3.obj-1sg.ind pig 3.psr friend 1sg.psr son.of.man
‘I gave my son’s friend a pig.’ (Golluscio 2010: 737)
b. Dewma-l-el-e-n-mew kiñe makuñ ñi ñuke.
make-appl-appl-inv-1sg.ind-3.obj one blanket 1sg.psr mother
‘S/he made a blanket for my mother.’
c. Illku-tu-l-fi-n Xwan tañi fotüm.
get.angry-appl-appl-3.obj-1sg.ind J. 3.psr son.of.man
‘I scolded Juan’s son.’ (Golluscio 2010: 733)

By contrast, -l or -ñma that applicativize bivalent base predicates introduce a par-

ticipant that can bear some of the other roles borne associated with the Ancient Greek
and Latin extra-thematic dative. Verbs allowing both markers often show an opposition
between motion towards or away from a referent when applicativized (27) (data from
Salas 2006: 120):

(27) a. Ngilla-lel-fi-n Antonio ñi kawello.

buy-appl-3.obj-1sg.ind A. 3.psr horse
‘I bought Antonio’s horse for him.’ (i.e., from someone else)
b. Ngilla-ñma-fi-n Antonio ñi kawello.
buy-appl-3.obj-1sg.ind A. 3.psr horse
‘I bought Antonio’s horse from him.’

The opposition is conventionalized with some verbs, but it often leads to an interpreta-
tion in terms of benefaction and malefaction, even with motion verbs (28) (data from
Augusta 1903: 75):

(28) a. Ye-l-fi-n ñi kuchillo.

carry-appl-3.obj-1sg.ind 3.psr knife
‘I brought him his knife (in order to help him).’
b. Ye-ñma-fi-n ñi kuchillo.
carry-appl-3.obj-1sg.ind 3.psr knife
‘I brought his knife away from/to him (in order to harm him).’

Other predicates show no such semantic opposition, for instance: ruka- ‘build a house’ >
ruka-lel- ~ ruka-ñma- ‘build a house as protection for’; see also denominal chadi-l- ~
chadi-ñma- ‘salt (v2)’.
Interestingly enough, the AppP of ñma-marked predicates is sometimes interpreted
as broadly affected irrespective of the base verb’s syntactic valency (albeit without
granting the AppP subject status; see § 6.2.1). Many instances of monovalent base pred-
icates introduce such an argument when applicativized with -ñma; quite often, the
198 Fernando Zúñiga

actions performed are customary, as with labile pütu- ‘drink’ in (29), or even ritual, as
with monovalent ngilla-tu- ‘perform rogations’ in (30):

(29) a. Pichi pütu-rke-i.

a.little drink-rep-ind[3]
‘S/he drank a little, I am told.’ (Smeets 2008: 465)
b. Ka pichi pütu-ñma-pu-fi-i-i-n.
and a.little drink-appl-trans-3.obj-ind-1-pl
‘And there we (pl) drank to him (the deceased).’ (Smeets 2008: 277)

(30) a. Feychi che itro rumel ngillatu-ke-i

art person upright always perform.rogation-hab-ind[3]
kiñe füta kura mew.
one big stone postp
‘This Mapuche always performs a rogation on a large boulder.’ (Guevara
1911: 110)
b. Ngillatu-ñma-nge-ke-la-n.
perform.rogation-appl-pass-hab-neg-1sg.ind
‘No rogations are performed in my honor.’ (Augusta 1916: 62)

Some applicativized verbs have AppPs of unpredictable semantics, especially with

-ñma. Cases in point include monovalent lladkü- ‘grow sad, upset’; lladkü-tu- means,
as expected, ‘get upset with, scold’ but lladkü-ñma- means ‘offer one’s condolences’. A
similar example is bivalent dapill- ‘clean (potato/legume plant before earthing it up)’;
dapill-ma- means ‘earth up (potato/legume plant)’. Monovalent anü- ‘sit down’ can take
-ñma to introduce a locative argument (see Example [10a]), but this derivation also has
a different, probably metaphorically related, reading:

(31) Anü-ñma-e-i-mew wekufü.

sit.down-appl-inv-ind[3]-3.obj demon
‘S/he was possessed by a demon.’

It is hard to find Mapudungun verbs that occur with all four applicatives; moreo-
ver, even in those cases where a verb takes three of them, the potential marker-specific
semantic differences often appear neutralized. Examples showing some semantic oppo-
sition include monovalent tofkü- ‘spit’ and bivalent ütrüf- ‘throw’. The former invari-
ably adds the person spat at/on when applicativized; -l and -ñma denote the familiar
benefactive and malefactive meanings, and -tu encodes an additional, unpredictable,
nuance (‘with contempt’). Similarly, the latter introduces a Goal/Source with -l and -ñma
and a Goal with added meaning with -tu (‘in order to scare/hurt him/her’).
The semantic roles borne by the AppP of different ACs are summarized in Table 5
below.
 7 The applicative constructions of Mapudungun 199

Table 5: Semantics roles of AppPs in Mapudungun.

-tu Stimulus, Motion Goal, Change-of-state Patient

-ye Motion Comitative, Topic of Speech/Thought
-l affected participant; implied Motion Goal, Beneficiary, Concernee
-ñma affected participant; implied Motion Source, Maleficiary, Concernee

There are no systematic explorations of applicativization in the context of reflex-

ives and reciprocals. Reflexive -w appears to usually work compositionally with -tu,
does not normally co-occur with -ñma (see Footnote 7), and often shows subtle meaning
idiosyncrasies when co-occurring with -l. Consider the pairs in (32); the reflexive has
the expected yield with -tu in (a–b), is a detransivizer or has the expected yield with
-ñma (c–d), and shows unpredictable yields with -l (e–f):14

(32) a. ayfiñ- ‘become decorated’ > ayfiñ-tu- ‘decorate’ > ayfiñ-tu-w- ‘decorate oneself’
b. tün- ‘look for lice’ > tün-tu- ‘look for lice on (sbdy.)’ > tün-tu-w- ‘look for lice
on oneself’
c. ngilla-tu- ‘conduct rogations (v1); ask (v2)’ > ngilla-tu-ñma- ‘conduct rogations
for (v2)’
> ngilla-tu-ñma-w- ‘conduct rogations (v1)’
d. wütrü- ‘water, irrigate’ > wütrü-ñma- ‘have (liquid) spilled on one; spill
(liquid) on’
> wütrü-ñma-w- ‘spill (liquids) on each other’
e. düngu- ‘speak’ > düngu-l- ‘make speak, read, play (caus)’ > düngu-l-w- ‘speak
to oneself’
f. üwe- ‘become solitary/empty’ > üwe-l- ‘take to a deserted area’ (appl)
> üwe-l-w- ‘get lost, lose one’s way’

The reflexive freely occurs with denominal -ye, but I have not found robust exam-
ples with deverbal -ye. An interesting example of deadjectival -ye is found with afma
‘faithful, loyal’: labile afma-tu- means ‘be amazed; admire’, bivalent afma-ye- means ‘be
careful of/with’, and monovalent afma-ye-w- means ‘incur expenses’ (Augusta 1916: 3).

14 The latter is probably related to a more general pattern involving -l (whether causative or applica-
tive), namely its tendency to combine with some other morphemes. A particularly frequent instance
of this is found with the bimorphemic deadjectival verbalizer -l-ka (e.g., pichi- ‘be little’ > pichi-l- ‘give
a little piece’ > pichi-l-ka- ‘make smaller, abbreviate’, Augusta 1916: 178; see also Golluscio 2007: 223).
200 Fernando Zúñiga

6 Lookalikes
6.1 Syntactic lookalikes

Mapudungun constructions whose syntactic make-up conforms to the definition of ACs

but whose morphology does not are either coded and symmetrical (suppletivism) or
uncoded and symmetrical (flexivalency). Suppletivism is easy to find with causatives;
with applicatives, it seems to be found only with bivalent-trivalent pairs (e.g., wül- ‘give
[v2]’ vs. elu- ‘give [v3]’), not with monovalent-bivalent pairs. Flexivalency is not difficult
to find with v1-v2 pairs (e.g., küdaw- ‘work [on]’) and also occurs with v2-v3 pairs (e.g.,
pi- ‘say, tell’).

6.2 Morphological lookalikes

6.2.1 Non-alternating non-agentive nucleatives

With avalent predicates and a subclass of non-agentive monovalent predicates,

ñma-derivation causes non-agentive nucleativization, but invariably of the subjective
kind; this phenomenon seems to be unrelated to the existence of symmetrical voices,
and quite rare (Zúñiga 2020). As with applicatives (see § 4.1), such derived predicates
head not only clauses of a type available to underived ones (IIa1, nan Type D) but also
another clause type (IIIc, nan Type E); the latter is like the passive of a three-argument
clause mentioned in Section 2.3 (see Example [3b]). The distinction between Operations
D-E on the one hand and Applicatives A-C surveyed in Section 4 on the other has a
semantic correlate: predicate classes I and IIa1 have no agentive argument. This is sche-
matically summarized in Table 6 below.

Table 6: Selected transitivizing Mapudungun ñma-nucleatives.

base predicate/clause derived predicate/clause

D I — IIa1 sbj
E IIa1 sbj IIIc sbj + sobj

The examples in (33) illustrate this phenomenon. In (a), the marker increases the
valency of avalent maw(ün)- ‘rain’ so as to make it monovalent (Type D); the same
happens with other meteorological predicates, as well as with astronomical predicates
(see [35a] below). In (b), the marker increases the valency of non-agentive monovalent
nag- ‘descend’ in the presence of an inanimate companion argument, making the predi-
cate bivalent (Type E); the same happens with other comparable verbs and nouns:
 7 The applicative constructions of Mapudungun 201

(33) a. Mawün-i. — Mawün-ma-n.

rain-ind[3] rain-nan-1sg.ind
‘It rained.’ ‘It rained on me / I got rained on.’
b. Nag-i mawün. — Nag-ma-n mawün.
descend-ind[3] rain descend-nan-1sg.ind rain
‘Rain fell.’ ‘Rain fell on me.’

Unsurprisingly, the 1st-person new argument is the subject, rather than the primary
object, in the derived construction. Two features of such constructions are unexpected,
however. First, the new argument being the subject is not due to the person-related
rules governing morphosyntactic inversion. In (a), those rules cannot apply at all, since
there is only one argument on the right-hand side; 3rd-person arguments are also
installed as subjects in those instances, for example: mawünmai ‘s/he got rained on’,
mawünmayengu ‘they (du) got rained on’, and nagmayengün mawün ‘rain fell on them
(pl)’. (Such examples are extremely rare in published texts but unproblematic in elicita-
tion.) Second, while in (a) it may seem natural that the only semantic argument appears
as subjective syntactic argument in the derived construction, in (b) one could have
expected the derived bivalent predicate nag-ma- to be morphologically bipersonal, that
is, to have inverse morphology (e.g., ✶nag-ma-e-n-mew), or at least passive morphology
(e.g., nag-ma-nge-n).
In fact, to judge from the relevant dictionary entries in Valdivia (1606), passive mor-
phology was found with such forms in earlier stages of the history of Mapudungun.
Consider (34), an instance of -ñma deriving a bivalent verb from a noun, where passive
marking is apparently still optional:

(34) Witran-ma(-nge)-n.
visitor-vblz-pass-1sg.ind
‘I’ve got a visitor.’ (Smeets 2008: 303)

In the original form of this kind of ñma-nucleativization, in order to accommodate

the additional (usually human) argument in the absence of an animate Agent, the
clause had to be overtly passivized. This constraint was later relaxed, and nowadays
such verbs no longer require the passivizing morpheme -nge, but the AppP is the
subject nonetheless. In (33), -ñma resembles a Philippine-like obligatory applica-
tive-cum-passive that simultaneously introduces a new argument to the clause and
grants it subject status.
Regarding the semantics of such ñma-operators applied to low-transitivity Agent-
less base predicates like those in (35), the new argument appears to be a participant
broadly construed as affected. It is not systematically interpreted in terms of percep-
tion, motion, possession, or even benefaction/malefaction:
202 Fernando Zúñiga

(35) a. Kiñekemew pun-ma-i-i-n.

sometimes night-nan-ind-1-pl
‘Sometimes night fell on us.’ (Smeets 2008: 302)
b. Iñche aku-ñma-n kiñe küme dungu.
1sg arrive-nan-1sg.ind one good message
‘I received a nice message.’ (Smeets 2008: 379)

Some few verbs seem to optionally or even obligatorily take -l instead of -ñma in
Type-E operations, but more research is needed here. The most robust example of this
I have found is la- ‘die’ (36), which expectedly takes -üñma when the base verb is an
(irregular) m-causative (36a) but appears as la-yel- when there is no causativization
(36b), without any difference regarding the new argument’s semantic role:

(36) a. Langüm-üñma-nge-n ñi fotüm.

kill-appl-pass-1sg.ind 1sg.psr son.of.man
‘They killed my son on me.’ (Salas 2006: 123)
b. La-yel-n tañi küme chaw.
die-nan-1sg.ind 1sg.psr good father
‘My good father has died on me.’ (Augusta 2016: 111)

6.2.2 Syntax-neutral markers

Constructions whose syntax is identical with derived and underived verbs can be found
in the language, especially with the markers -tu and -ye. More research is needed here,
but the available facts suggest that the syntactically neutral markers are not focaliza-
tion devices; instead, they seem to be either semantic explicators or simply semantic
differentiators. Lists of verbs that make syntactically neutral use of the applicative
markers can be found in Zúñiga (2009a).
As mentioned in Section 3, tu-suffixation is syntactically neutral with a consider-
able number of verbs, either with an unpredictable semantic effect (e.g., allkü- ‘hear
[v1/v2]’ > allkü-tu- ‘listen [to]’) or, somewhat less frequently, without any recognizable
semantic effect (e.g., illam- ~ illam-tu- ‘despise’). With other verbs, however, it is valen-
cy-decreasing, that is, -tu can also reflexivize a verb by suppressing the expression of its
non-agentive argument (ñiküm- ‘protect from the wind’ > ñiküm-tu- ‘protect oneself from
the wind’). Antipassivization is found only with few verbs that do not seem to belong
to any particular formal or semantic class, some of which are actually labile (e.g., ira-
‘split [v2]’ > ira-tu- ‘split [wood] [v2/v1]’). Labile nütram- ‘tell, narrate’ is a case in point;
compare the examples given in (11) above with the ones in (37) below. Note that even
the tu-marked verb is labile:
 7 The applicative constructions of Mapudungun 203

(37) Nütram-tu-n. / Nütram-tu-fi-n.

tell-antip-1sg.ind tell-antip-3.obj-1sg.ind
‘I talked.’ ‘I talked with him/her/them.’

The exact semantic role of the non-Agent is rather varied with these verbs, usually
showing some idiosyncrasy, for instance: lawen- ‘use as medicine (v2)’ > lawen-tu- ‘take
medicine (v1); treat/heal (v2)’ vis-à-vis kafkü- ‘whisper to’ > kafkü-tu- ‘whisper (v1)’.
Ye-suffixation can also be syntax-neutral, either with (e.g., traf- ‘meet [v2]’ > traf-
ye- ‘bump into on the road’) or without a semantic effect (e.g., duam- ~ duam-ye- ‘need’).
Syntax-neutral ñma-suffixation is rather rare (see [10b] for an example), but syn-
tax-neutral -l seems to occur less seldom. The following examples with wül- ‘give away,
hand’ from Augusta (1916: 256) show valency-increasing -l and -ñma (38) alongside syn-
tax-neutral -l (39); the particular reading ‘give blows, hit’ seems to require -l for most
speakers:

(38) a. Fey wül-el-a-e-n-mew ñi makuñ.

dem give.away-appl-fut-inv-1sg.ind-3.obj 1sr.psr blanket
‘S/he will sell my blanket for me, instead of me.’
b. Wül-ma-e-n-mew ñi ofisha.
give.away-appl-inv-1sg.ind-3.obj 1sg.psr sheep
‘S/he sold my sheep (against my will).’

(39) Wül-el-fi-i kura mew.

give.away-l-3.obj-ind[3] stone postp
‘S/he hit him/her with a stone.’

6.2.3 Applicative deponents

Applicative deponents or applicativa tantum—i.e., verbs that appear with one of the
applicative markers but do not stand in opposition to an unmarked verb of the same
root—are common in Mapudungun only as denominals. Some verbs with fossilized
markers seem to have shown productive oppositions until relatively recently; for instance,
present-day añel-tu- ‘threaten’ used to occur alongside añel- a century ago, apparently
without any difference in meaning (Augusta 1916: 10).15 Denominal verbs with -tu and
-ye are numerous (e.g., wekufü ‘devil’ > wekufü-tu- ‘bedevil, bewitch’ and patron ‘boss’ >
patron-ye- ‘consider [sbdy.] one’s boss’); those with -l and -ñma are less frequent but not
rare (e.g., piwke ‘heart’ > piwke-l- ‘put in the middle of’ and [34]).

15 The original root seems to have been ane- ~ añe-, and Erize (1960: 54) lists ane-l-, ane-tu- and ane-l-
tu- as synonyms.
204 Fernando Zúñiga

The effects of applicative-like marking are summarized in Table 7 below.

Table 7: Syntactic and semantic effects of selected Mapudungun markers.

-tu -ye -l -ñma

Valency-increasing:
– denominal verbalization
– non-agentive nucleativization ( )

Valency-decreasing:
v2>v1(/v2)

Valency-neutral:
– strictly neutral, some some
semantic effect
– strictly neutral, ( ) some some
no semantic effect

7 Conclusions
This chapter surveyed the applicative constructions attested in Mapudungun and other
uses of applicative morphology. These can be characterized as follows:

Morphology
– Mapudungun does not have constructions that could be analyzed as applicative
periphrases or analytical applicative constructions.
– The different subtypes of applicative constructions use one of four verbal suffixes
occupying slots in the verbal template near the root, in close vicinity to other valen-
cy-changing markers. Combining two applicative markers in one verb complex is
possible but subject to some restrictions.
– The applicative suffixes -tu and -ye do not show allomorphy. The allomorphy of -l
and -ñma is conditioned phonologically and, albeit less importantly, lexically.
– All markers have homonyms and some (viz. -tu and -l) are arguably polysemous.

Syntax
– Mapudungun applicatives are P-applicatives. The new participant could appear in
the base construction in some cases in principle, either as an optional adjunct or as
a non-argumental Possessor.
– Depending on argument-realization rules unrelated to applicativization, the
applied phrase may occur either as a primary object in actor-voice clauses, with
direct verbs, or as a subject in undergoer-voice clauses, with inverse verbs. With
 7 The applicative constructions of Mapudungun 205

verbs belonging to two clearly delimited valency classes, ñma-derivation installs a

new participant in a non-P role as the subject, and is therefore a lookalike.
– Mapudungun applicatives differ regarding how they interact with valency classes.
Ye-verbs are bivalent (whether via valency increase from a monovalent verb or
via valency redirecting from a bivalent verb). Tu-verbs add a core argument,
either making a monovalent verb bivalent or making a bivalent verb trivalent. The
markers -l and -ñma have the same effect as -tu, but also apply to base predicates
belonging to otherwise excluded valency classes, namely trivalent verbs (-l) or
avalent verbs (-ñma).
– The use of Mapudungun applicatives is not conditioned by the limited access of
adjuncts to some structural operations. Rather, it is related either to the impossi-
bility to clearly express some semantic roles not related to possession otherwise
than with this operation (in the obligatory cases), or to an idiomaticity stipulation
that requires kinship- or part-whole-based Possessors of patientive arguments to
be expressed as indirectly affected and therefore as applied core arguments (in the
optional cases).
– In applicative constructions, the applied phrase is an overt/covert NP showing all
the syntactic properties that characterize primary objects or subjects in non-ap-
plicative constructions. When 3rd-person primary objects, they seem to always, or
almost always, co-occur with the verbal suffix -fi on direct verbs (unlike compara-
ble objects in non-applicative constructions, which trigger the marking much less
often, taking into account considerations of animacy, person, and relative topical-
ity). There is no other difference between applicative constructions and construc-
tions of underived verbs involving the same number of objects, but one clause type
occurs only as applicative construction (viz. four-argument clauses).
– There are no particular restrictions on the combination of applicatives with caus-
atives and passives, and verb-noun compounds can take applicative marking as
well, apparently as long as the resulting word is not too difficult to parse. Reflexives
and reciprocals of applicatives need to be explored in more detail; some markers
behave as expected and others show syntactic and semantic idiosyncrasies. Verb
forms including more than two occurrences of the applicative marker, applicativ-
ized verb-noun compounds, and applicative constructions including more than two
new argument phrases seem to be ruled out.

Semantics
– All applicative suffixes are semantically underspecified markers that license several
semantic roles.
– In the case of -tu and -ye, the interpretation of the new argument usually depends
on the marker itself and the semantics of the base predicate. With -tu, the new
participant is typically either the Stimulus of experiencer verbs, the Goal of motion
206 Fernando Zúñiga

verbs, or the Patient of change-of-state verbs. With -ye, it is a motion Comitative or

the Topic of Speech/Thought.
– With -l and -ñma, the exact interpretation of the new argument depends on the
marker itself, the syntax and semantics of the base predicate, and contextual fea-
tures. When applicativized, trivalent predicates accommodate a Concernee (“Exter-
nal Possessor”). Monovalent predicates accommodate a participant broadly con-
strued as affected, with lexical semantics of the elements in the clause and context
in charge of the specifics. Bivalent predicates show lexically conditioned variation:
some verbs use the markers to encode approach and separation, respectively; others
encode benefaction and malefaction; yet others neutralize the distinction in favor
of broadly construed affectedness or Concernee-Concern relationship and either
take any marker interchangeably or lexically requires one of them irrespective
of meaning. Several ñma-applicatives license semantically unpredictable applied
phrases.
– Mapudungun applicative constructions seem to have no pragmatic or discursive
implications.

Lookalikes
– Valency-neutral instances of tu- and ye-marking—either with or without a differ-
ence in semantics—are numerous; strict syntactic neutrality is exceptional with -l
and -ñma.
– Non-applicative tu-marking is frequent and can, on a lexical basis, antipassivize,
telicize, or form a reversionary/repetitive; the latter is possibly related to one of the
iterative stem formations. With some verbs, tu-marking denotes higher intensity
or higher Patient affectedness. Non-applicative ye-marking is relatively infrequent
and appears to be obsolescent; it can encode distributivity/multiplicity of partici-
pants and, possibly, completion or duration.
– With -l and -ñma, three non-applicative functions are prominent. First, -l is also a
(high-control) causative that increases the valency of (mostly) monovalent predi-
cates. Second, a small closed class of ambiguous verbal/adverbial roots expressing
spatial, temporal, and manner notions can form “relational” adverbs with -ñma.
Third, ñma-derivation of avalent and some non-agentive monovalent verbs is tran-
sitivizing/valency-increasing: it installs the new participant (which is interpreted as
broadly affected) as the subject and is therefore reminiscent of a combined applica-
tive-cum-passive operation.
– Lexicalized applicatives with verbal roots do not seem to be particularly numerous.
By contrast, the use of -tu and -ye to verbalize nouns is an important lexicon-ex-
panding device in the language.
 7 The applicative constructions of Mapudungun 207

Abbreviations
AC applicative construction
an agentive nucleativizer
antip antipassive
appl applicative
art article
asp aspect
attr attributive
av actor voice
BC base construction
caus causative
dem demonstrative
fut future
hab habitual
imp imperative
ind indicative
inv inverse
nan non-agentive nucleativizer
neg negative
nfin nonfinite
NI nominal incorporation
obj object
oobj oblique object
pass passive
pl plural
pobj primary object
postp postposition
pro pronoun
psr possessor
ptcp participle
refl reflexive
rep reportative
res resultative
sbj subject
sg singular
sobj secondary object
subj subjunctive
trans translocative
uv undergoer voice
vblz verbalizer
v0, v1, v2. . . syntactic valency of predicates
x→y x acts on y
208 Fernando Zúñiga

References
Augusta, Félix José de. 1903. Gramática araucana. Valdivia: Imprenta Central J. Lampert.
Augusta, Félix José de. 1916. Diccionario araucano. Santiago: Imprenta Universitaria.
Baker, Mark, Roberto Aranovich & Lucía Golluscio. 2005. Two types of syntactic noun incorporation: Noun
incorporation in Mapudungun and its typological implications. Language 81(1). 138–176.
Comrie, Bernard. 1989. Language universals and linguistic typology: Syntax and morphology. 2nd edn. Oxford:
Blackwell.
Creissels, Denis. Forthcoming. Transitivity, valency, and voice. Oxford: Oxford University Press.
Erize, Esteban. 1960. Diccionario comentado mapuche-español. Araucano, pehuenche, pampa, picunche,
ranculche, huilliche. Buenos Aires: Universidad Nacional del Sur.
Golluscio, Lucía. 2007. Morphological causatives and split intransitivity in Mapudungun. International
Journal of American Linguistics 73(2). 209–238.
Golluscio, Lucía. 2010. Ditransitive constructions in Mapudungun. In Andrej Malchukov, Martin Haspelmath
& Bernard Comrie (eds.), Studies in ditransitive constructions: A comparative handbook, 710–756. Berlin:
Mouton De Gruyter.
Guevara, Tomás. 1911. Folklore araucano. Santiago: Imprenta Cervantes.
Harmelink, Bryan. 1996. Manual de aprendizaje del idioma mapuche. Temuco: Ediciones Universidad de la
Frontera.
Moesbach, Ernesto Wilhelm de. 1962. Idioma mapuche. Padre Las Casas: Editorial San Francisco.
Peterson, David. 2007. Applicative constructions. Oxford: Oxford University Press.
Polinsky, Maria. 2013. Applicative constructions. In Matthew Dryer & Martin Haspelmath (eds.), The
World Atlas of Language Structures Online, Ch. 109. Leipzig: Max Planck Institute for Evolutionary
Anthropology. Available online at http://wals.info/chapter/109, accessed on 2021–06–01.
Salas, Adalberto. 2006. El mapuche o araucano. 2nd, revised edn. Santiago: Centro de Estudios Públicos.
Smeets, Ineke. 2008. A grammar of Mapuche. Berlin: Mouton de Gruyter.
Valdivia, Luis de. 1606. Arte y gramatica general de la lengua que corre en todo el Reyno de Chile. Lima:
Francisco del Canto.
Van de Velde, Mark. 2020. Concernee-Concern constructions: A comparative study of external possession in
the Bantu languages. Studies in Language 44(1). 70–94.
Zúñiga, Fernando. 2006. Deixis and alignment: Inverse systems in indigenous languages of the Americas.
Amsterdam: John Benjamins.
Zúñiga, Fernando. 2009a. The applicatives of Mapudungun. Paper presented at the SSILA Annual Meeting,
San Francisco, 8–11 January.
Zúñiga, Fernando. 2009b. An exploration of the diachrony of Mapudungun valency-changing operations.
Unpublished manuscript, University of Zurich.
Zúñiga, Fernando. 2010a. Benefactive and malefactive applicativization in Mapudungun. In Fernando
Zúñiga & Seppo Kittilä (eds.), Benefactives and malefactives: Typological perspectives and case studies,
203–218. Amsterdam: John Benjamins.
Zúñiga, Fernando. 2010b. La marca diferencial del objeto en mapudungún. Lingüística 24. 141–164.
Zúñiga, Fernando. 2015. Valency classes in Mapudungun. In Andrej Malchukov & Bernard Comrie (eds.),
Valency classes in the world’s languages, Vol. 2, 1515–1544. Berlin: De Gruyter Mouton.
Zúñiga, Fernando. 2017. Mapudungun. In Michael Fortescue, Marianne Mithun & Nicholas Evans (eds.), The
Oxford handbook of polysynthesis, 696–712. Oxford: Oxford University Press.
Zúñiga, Fernando. 2019. Grammatical relations in Mapudungun. In Alena Witzlack-Makarevich & Balthasar
Bickel (eds.), Argument selectors: A new perspective on grammatical relations, 39–67. Amsterdam: John
Benjamins.
Zúñiga, Fernando. 2020. Subjective applicatives in Even, Yupik, and Mapudungun. Paper read at the 53rd
Annual Meeting of the Societas Linguistica Europaea, 28 August.
 7 The applicative constructions of Mapudungun 209

Zúñiga, Fernando & Antonio Díaz-Fernández. 2014. Reduplication in Mapuzungun: Form and function. In
Gale Gómez & Hein van der Voort (eds.), Reduplication in indigenous languages of South America, 17–37.
Leiden: Brill.
Zúñiga, Fernando & Aldo Olate. 2017. El estado de la lengua mapuche, diez años después. In Isabel Aninat,
Verónica Figueroa & Ricardo González (eds.), El pueblo mapuche en el siglo XXI: propuestas para un nuevo
entendimiento entre culturas en Chile, 343–374. Santiago: Centro de Estudios Públicos.
Denis Creissels
8 Applicative constructions and non-applicative
uses of applicative morphology in Tswana
(Bantu)
Abstract: This chapter provides a detailed description of applicative constructions in
Tswana, and of constructions involving the same morphological marking as applicative
constructions, although they do not meet the definition of applicative constructions.
Leaving aside an applicative-like use of the causative suffix, Tswana can be analyzed
as having a single verbal suffix involved in the marking of applicative constructions.
In P-applicative constructions, this suffix licenses applied objects expressing various
semantic roles, such as beneficiary, whose only common feature is that they cannot
be expressed as oblique phrases in clauses projected by the non-applicative form of
the same verb. In X-applicative constructions, the same suffix licenses locative expres-
sions that show no evidence of a change in their morphosyntactic status but express
roles other than those they would express with the non-applicative form of the same
verb. The same suffix is also found in a construction in which it marks the promotion
of instrumental adjuncts to subject role. It also has several uses not related to valency
operations: focalization of locative adjuncts, expression of habituality of action at some
place and expression of intensity of action. Finally, lexicalized applicatives are common
in Tswana.

1 Introduction
This chapter describes applicative constructions and non-applicative uses of applicative
morphology in Tswana (aka Setswana, ISO 639-3 tsn, Glottolog tswa 1253), a southern
Bantu language spoken in Botswana and South Africa by more than 6 million speakers,
whose closest relatives are Pedi and Southern Sotho.1
Tswana has very productive applicative constructions marked by a verbal suffix
-ɛl, reflex of the Proto-Bantu reconstructed applicative suffix ✶-ɪd. However, the con-
structions in question do not meet some of the narrow definitions of applicativization

1 In Botswana, ethnic Batswana constitute 80% of the population, estimated at 2,3 million (2020). In
South Africa, Tswana is dominant in the Northwest Province and in some districts of the Free State
Province, and the number of its speakers is estimated at 5 million. Tswana, Pedi and Southern Sotho are
so close to each other that, from a strictly linguistic point of view, they should be considered as three
varieties of a single language. Pedi is commonly designated as Northern Sotho, but this term is ambig-
uous, since it is also used with reference to lects (Lobedu, Tswapong, etc.) that, linguistically, are better
considered languages distinct from Sotho-Tswana proper.

https://doi.org/10.1515/9783110730951-008
212 Denis Creissels

that may be found in the literature, and can only be identified as applicative if the latter
term is taken in the broad sense suggested in the position paper. Moreover, the suffix
-ɛl is also found in constructions involving valency operations that cannot be viewed
as varieties of applicativization, even if this term is taken in a very broad sense, and in
constructions in which its function is not related to valency.2
The chapter is structured as follows. Section 2 provides the necessary information
about the basics of Tswana morphosyntax. Section 3 is a general introduction to the
various types of applicative constructions that can be found in Tswana, and a general
presentation of the applicative marker -ɛl. Section 4 discusses the applicative construc-
tions in which the applied phrase fulfills the syntactic role of object. Section 5 discusses
the applicative constructions in which the applied phrase cannot be analyzed as an
object. Section 6 is about a valency-related use of -ɛl which, however, does not meet the
definition of applicativization. Section 7 is devoted to the uses of -ɛl that at not related
to valency operations. Section 8 is about lexicalized applicatives. Section 9 discusses the
question of whether some uses of the causative marker -is could be analyzed as applica-
tive-like. Section 10 summarizes the main conclusions.

2 The basics of Tswana morphosyntax

Typologically, Tswana is in almost every respect a typical Bantu language. For an overall
presentation of Tswana, see Cole’s (1955) reference grammar or Creissels’ (2003) sketch
grammar.

2.1 Nouns, noun phrases, and locatives

Tswana common nouns consist of a stem and an obligatory prefix expressing number,
and number morphology is closely related to the division of nominal lexemes into
genders. The gender system of Tswana relies on an inventory of 12 agreement patterns,
traditionally referred to as “classes”, each of them triggered by a particular subset of
noun forms. As a rule, noun forms that trigger the same agreement pattern also share
the same prefix, and vice-versa, but this correlation is far from perfect. Some classes

2 Throughout this chapter, Tswana words and sentences are given in broad phonetic transcription. The
reason for not using current Tswana orthography is that it is quite misleading in linguistic analysis. The
point is that current orthography distinguishes only 5 vowels and does not note tones at all, whereas
Tswana has 9 vowel phonemes, and tones are crucial for morphological analyses. Moreover, many mor-
phemes that are unquestionably prefixes (in particular, subject indexes and object indexes) are written
as if they were separate words. The tones indicated in this transcription are surface tones (which may
be very different from underlying tones, given the complexity of tone realization rules), and the length-
ening that affects the penultimate syllable of words in immediate prepausal position is noted explicitly.
 8 Applicative constructions in Tswana 213

(in the sense of agreement patterns) are associated with sets of singular noun forms,
others with sets of plural forms, and nominal lexemes can be grouped into genders on
the basis of correspondences such as mʊ̀-sádí (cl. 1) ‘woman’ / bà-sádí (cl. 2) ‘women’:
mʊ̀-sádí ‘woman’ triggers class 1 agreement, the corresponding plural bà-sádí ‘women’
triggers class 2 agreement, and consequently, ‘woman’ as a nominal lexeme belongs to
a gender that can be labeled gender 1–2.
In Tswana noun phrases, noun dependents follow their head, and express gen-
der-number agreement with their head.
The attachment of a locative marker to the first word of noun phrases converts
them into locative phrases. Locatives optionally combine with one of the locative prep-
ositions kó (relative remoteness), fá (relative proximity), or mó (interiority, contact).
Neither locative affixes nor locative prepositions specify the distinction between static
location, source of motion, or direction of motion. In contrast to the situation observed
in Central Bantu languages, locatives do not have access to the subject function.

2.2 Clause structure

The basic constituent order is Subject-Verb-Objects-Obliques. There is no flagging of

either subjects or objects, but in independent assertive or interrogative clauses, the verb
includes an obligatory subject index representing the single core argument of intransi-
tive verbs and the agent of prototypical transitive verbs (1a). Objects are not obligato-
rily indexed, but topical objects whose precise description is considered superfluous by
the speaker are represented by object indexes inserted between the subject index and
the verb stem (1b). Objects can be converted into subjects of passive constructions (1c).3

(1) a. ŋwàná ʊ́ -tʰùb-íl-é màːɪ́.

child(1) sI:cl1-break-prf-fv eggs(6)
‘The child broke the eggs.’
b. ʊ́ -à-tʰúb-îːl-è.
sI:cl1-oI:cl6-break-prf-fv
‘He broke them.’
c. màɪ́ á-tʰùb-íl-w-è kɪ́ ŋwàːná.
eggs(6) sI:cl1-break-prf-pass-fv by child(1)
‘The eggs were broken by the child.’

3 Since the details of nominal morphology are irrelevant in this chapter, in the Tswana examples, noun
forms are not segmented, and are just given with a translation reflecting their number value and the
indication of the agreement pattern they trigger.
214 Denis Creissels

Tswana is an unproblematic “accusative” language in which the single core argument

of intransitive verbs and the agent of prototypical transitive verbs not only share the
same coding characteristics, but also consistently align with respect to their behavior
in operations such as focalization, questioning, equi-NP deletion, raising, or control.
Consequently, there is no inconvenient in discussing Tswana morphosyntax in terms of
subjects and objects as grammatical relations identified in Tswana by their indexation
properties, rather than in terms of A, S and P.
Tswana does not have a grammatical relation “indirect object” in the sense of a
grammatical relation distinct from the grammatical relation “direct object”, whose
semantic prototype would be the recipient of verbs of giving, but has very productive
multiple-object constructions (used in particular with the verbs of giving) in which the
linear order of object NPs is rigidly determined by the Animacy Hierarchy (2a), seman-
tic roles being taken into consideration only if two objects are of equal rank according
to the Animacy Hierarchy (for example, beneficiaries precede recipients or causees,
causees precede patients, etc.). Each object can be represented by an object index, and
all the objects can be indexed simultaneously in the verb form. The order of object
indexes is the mirror image of the order of their conominals (2b). In multiple-object
constructions, each object can be converted into the subject of a passive construction in
which the other objects are maintained in object role (2c–d), and reflexive derivation
may encode coreference between the subject and any of the objects (2e–f).

(2) a. ŋwàná !ʊ́ -f-íl-é !

kátsɪ́ dìːdʒɔ́.
child(1) sI:cl1-give-prf-fv cat(9) food(10)
‘The child gave food to the cat.’
b. ŋwàná !ʊ́ -dí-ɪ ̀-f-îːl-è.
child(1) sI:cl1-oI:cl10-oI:cl9-give-prf-fv
‘The child gave it (the food) to it (the cat).’
c. kátsɪ́ !ɪ́-f-íl-w-é dìːdʒɔ́.
cat(9) sI:cl9-give-prf-fv food(10)
‘The cat was given food.’
d. dìdʒɔ́ !
dí-f-íl-w-é !
kâːtsɪ ̀.
food(10) sI:cl10-give-prf-fv cat(9)
‘The food was given to the cat.’
e. bàtʰʊ̀ bá-ì-pʰ-íl-é nákɔ ̀ j-á-χʊ̀-ɪ́-ákáɲêːts-à.
people(2) sI:cl2-REFL-give-prf-fv time(9) cl9-gen-inf-oI:cl9-think.about-fv
‘The people took the time (lit. gave themselves the time) to think about it.’
f. bàtʰʊ̀ bá-ì-pʰ-íl-é nâːχà.
people(2) sI:cl2-REFL-give-prf-fv bush(9)
‘The people ran away (lit. ‘gave themselves to the bush).’

As illustrated in (3), valency-increasing derivations may create derived verbs whose

coding frame is a two- or even three-object construction.
 8 Applicative constructions in Tswana 215

(3) a. lʊ̀sɪ́á !
lʊ́ -nʊ́ -l-é mâːʃì.
baby(11) sI:cl11-drink-prf-fv milk(6)
‘The baby drank milk.’
b. kɪ ̀-nʊ́ -s-íts-é lʊ̀sɪ́á mâːʃì.
sI:1sg-drink-caus-prf-fv baby(11) milk(6)
‘I made the baby drink milk.’
c. kɪ ̀-nʊ́ -s-éd-íts-é dím̀ pʰɔ́ lʊ̀sɪ́á mâːʃì.
sI:1sg-drink-caus-appl-prf-fv Dimpho(1) baby(11) milk(6)
‘I made the baby drink milk for Dimpho.’
d. kɪ ̀-á-lʊ́ -mʊ̀-nʊ́ -s-éd-îːts-è.
sI:1sg-oI:cl6-oI:cl11-oIcl1-drink-caus-appl-prf-fv
‘I made it drink it for her.’

Combinations of valency-changing derivations resulting in constructions with less than

three objects are freely allowed. By contrast, those that would result in four-object con-
structions are completely ruled out, and there are also apparently arbitrary restrictions
with three-object constructions. For example, it is difficult to explain why fá ‘give’ lends
itself to applicative derivation, but not to causative derivation.
On the use of prepositions to mark various semantic types of obliques, see Creis-
sels (2013). For a general presentation of valency-changing mechanisms, see Creissels
(2002).

2.3 Verb morphology

A Tswana verb form consists of a root (irreducible lexical element), an obligatory suffix
(the final vowel, or simply final), and a variable number of other affixes whose pres-
ence depends on a variety of factors, each affix having a fixed position in the template.
The root may be immediately followed by derivational suffixes that modify its meaning
without altering its valency. Together with the root they constitute the extended root.4
Starting from the extended root as the zero point, the order in which the affixes
appear can be described as a sequence of positions numbered from –4 (the leftmost pos-
sible position, before subject indexes) to +5 (the rightmost possible position, after the
so-called final vowel). Some of the positions may host up to three successive affixes of
the same category. Given the topic of this chapter, it is sufficient to indicate the positions
in which valency operators can be found:
– The reflexive marker (or middle voice marker) í occupies position –1 (immediately
before the root), which is also the position occupied by object indexes.

4 For a more detailed account of Tswana verb morphology, see Creissels et al. (1997) and Creissels (2006,
2017).
216 Denis Creissels

– Causative (i)s or J,5 applicative ɛl, anticausative ɛχ, al, afal, aχal, ɛsɛχ, or Jɛχ, and
reciprocal an occupy position +1 (immediately after the extended root).
– The passive marker (i)w ocupies position +3 (immediately before the final vowel).

Position –1 may host up to three successive affixes (three object indexes, or the reflex-
ive marker plus two object indexes), and the same applies to position +1.

3 Introductory remarks on Tswana applicatives

3.1 Tswana applicative constructions and the definition
of applicativization

Applicative constructions are sometimes defined as derived constructions in which a

participant encoded as an oblique in the base construction is encoded as a direct object
(or as P, depending on the theoretical framework). According to this definition, Tswana
simply doesn’t have applicative constructions, since in all the valency-related uses of
the applicative marker -ɛl, the phrases it licenses refer to a participant that cannot be
encoded at all in a mononverbal construction involving the non-derived form of the
verb.
All the applicative constructions found in Tswana are obligatory applicatives
whose use is just conditioned by the fact that they constitute the only available option
for the encoding of a given semantic role in the construction of a given verb, as opposed
to optional applicatives whose use may be motivated by information structure or
restrictions on the access of obliques to operations such as focalization or relativization.
For example, none of the Tswana prepositions can assign the role of beneficiary to
its complement, and in contrast to languages such as Eton (Bantu; Van de Velde 2008) or
Jóola Fóoñi (Atlantic; Creissels and Voisin, this volume), beneficiaries cannot be encoded
as objects without using a special form of the verb. In Tswana, the role of beneficiary
expressed in (4b) by the applied object kítsɔ́ cannot be expressed within the frame of a
monoverbal construction whose nucleus would be -bɛ́rɛ́ká in its underived form.

(4) a. kɪ ̀-tɬàà-bɛ́rɛ́k-á màítsɪ́bʊ̀ːá.

sI:1sg-FUT-work-fv evening(6)
‘I’ll work this evening.’
b. kɪ ̀-tɬàà-bɛ́rɛ́k-ɛ́l-à kítsɔ́ màítsɪ́bʊ̀ːá.
sI:1sg-fut-work-appl-fv Kitso(1) evening(6)
‘I’ll work for Kitso this evening.’

5 J is an abstract morphological element that can be posited in order to account for consonant alterna-
tions analyzable as originating historically from a phonological process of palatalization.
 8 Applicative constructions in Tswana 217

According to a less restrictive definition of applicatives, which is also found in the litera-
ture, an applicative construction is a construction in which a derived verb form assigns
the syntactic role of direct object (or P) to a noun phrase (the applied object) referring
to a non-core participant (i.e., to a participant in the event that cannot be coded as a
core syntactic term in the construction of the base verb). Some of the constructions
discussed in this chapter (for example the construction illustrated in (4b) above) meet
this definition, which doesn’t put any condition about the possibility of expressing the
semantic role expressed by the applied object in clauses whose nucleus is the underived
form of the same verb.
However, not all the constructions discussed in this chapter as applicative con-
structions meet the definition of applicative constructions as constructions including
an applied object. Some of them only meet a broader definition such as that put forward
in the position paper, encompassing any type of construction in which a derived verb
form assigns a grammatical relation other than subject to an NP (the applied phrase)
representing a non-core participant that could not be coded in the same way (or could
not be coded at all) in the construction of the base verb.
Another important property of the constructions analyzed in this chapter is that
they may involve applied phrases referring to essential participants in the event
denoted by the verb. Such a possibility is left open by the definition put forward in
the position paper and adopted in this chapter, but is explicitly excluded by some of
the more restrictive definitions of applicativization that can be found in the literature.
However, Tswana is typically a language whose description would be considerably
(and unnecessarily) complicated if a definition of applicativization excluding applied
phrases representing essential participants were adopted.

3.2 The applicative marker -ɛl

All the Tswana constructions that meet the definition of applicative constructions
adopted in this chapter involve a verbal suffix whose underlying form can be posited as
-ɛl. Like all the formatives that constitute the verb stem with the exception of the verb
root, it is underlyingly toneless, and its tone varies depending on tone spreading rules
whose input is the tone of the root and the tone patterns associated to each individual
tam-polarity form. Depending on the adjacent formatives, -ɛl may undergo regular mor-
phophonological processes giving rise to the following allomorphs: -el (5a), -ed (5b), or
-ets (5c).6

6 The vowel lengthening observed in (5a), (5c), and (6) is due to the general rule of prepausal lengthen-
ing mentioned in Footnote 2 above.
218 Denis Creissels

(5) a. χà-kɪ́-bà-bérék-êːl-ɪ ̀.
neg-sI:1sg-oI:cl2-fut-work-appl-fv
‘I do not work for them.’
b. kɪ ̀-bíl-éd-íts-é bàná ŋâːkà.
sI:1sg-call-appl-prf-fv children(2) doctor(9)
‘I’ve called the doctor for the children.’
c. kɪ ̀-tɬàà-bá-χʊ̀-bérék-ís-êːts-à.
sI:1sg-fut-oI:cl2-oI:2sg-work-caus-appl-fv
‘I will make them work for you.’

Moreover, depending on the context, -ets may be not only an allomorph of -ɛl, as in (5c),
but also the result of the fusion of -ɛl with the perfect marker J,7 as in (6).

(6) kɪ ̀-bá-bérék-êːts-ɪ ̀.
sI:1sg-oI:cl2-work-appl.prf-fv
‘I’ve worked for them.’

As illustrated by example (7), Tswana has double-applicative constructions with two

successive occurrences of -ɛl in the verb form and two applied phrases expressing two
distinct roles (in (7), ɪ ̀ŋ́ ‘what’ expressing the role of cause/purpose, and kítsɔ́ expressing
the role of beneficiary).

(7) ʊ̀-lɪ ̀m-ɛ ̀l-ɛ ̀l-à ɪ ̀ŋ́ !

kítsɔ́ !
tsʰîːmʊ̀?
sI:2sg-plough-appl-appl-fv what Kitso(1) field(9)
‘Why are you ploughing the field for Kitso?’

By contrast, constructions with more than two applied phrases and more than two
occurrences of -ɛl in the verb form are impossible, although the variety of semantic
roles that can be expressed via applicative derivation is such that it would not be very
difficult to imagine the possibility of such constructions.

3.3 Non-applicative uses of the applicative marker -ɛl

The suffix -ɛl is the only possible marker of applicative constructions in Tswana, but it
also has uses that cannot be analyzed as instances of applicativization, even if this term
is taken in a relatively broad sense. For example, in (8b), -ɛl licenses an applied object
referring to a beneficiary, but in (8c), the same suffix marks the focalization of a locative

7 The morphophoneme J (see Footnote 5 above) occurs not only as one of the two allomorphs of the
causative marker -is ~ -J, but also as one of the two allomorphs of the perfect marker -il ~ -J.
 8 Applicative constructions in Tswana 219

phrase expressing the localization of the event, without any evidence of a change in
valency. (8d) shows that both uses of -ɛl may coexist in a construction whose nucleus is
a verb form including two successive occurrences of -ɛl.

(8) a. dʒɪ́sʊ́ ʊ́ -nè à-sw-á (mó sɪ ̀fápáánòː-ŋ̀).

Jesus(1) sI:cl1-aux sI:cl1-die-fv loc cross(7)-loc
‘Jesus died (on a cross).’
b. dʒɪ́sʊ́ ʊ́ -nè à-rɪ́-sw-ɛ̂ːl-à.
Jesus(1) sI:cl1-aux sI:cl1-oI:1pl-die-appl-fv
‘Jesus died for us.’
c. dʒɪ́sʊ́ ʊ́ -nè à-sw-ɛ́l-à mó sɪ ̀fápáánòː-ŋ̀.
Jesus(1) sI:cl1-aux sI:cl1-die-foc-fv loc cross(7)-loc
‘Jesus died ON A CROSS.’
d. dʒɪ́sʊ́ ʊ́ -nè à-rɪ́-sw-ɛ́l-ɛ́l-à mó sɪ ̀fápáánòː-ŋ̀.
Jesus(1) sI:cl1-aux sI:cl1-oI:1pl-die-appl-foc-fv loc cross(7)-loc
‘Jesus died for us ON A CROSS.’

The non-applicative uses of -ɛl will be described in Sections 6–7.

4 Applied-object constructions
Tswana applicative constructions divide into two subtypes according to the syntactic
role of the applied phrase. This section is devoted to the applicative constructions in
which the applied phrase fulfills the syntactic role of object, henceforth designated as
applied-object constructions (as opposed to applied-oblique constructions, in which
the applied phrase is syntactically an oblique).

4.1 Applied-object constructions as obligatory applicatives

In Tswana, applied-object constructions are very productive. As already commented in

Section 3.1, and further illustrated by example (9), they are obligatory applicative con-
structions, in the sense that the participant encoded as the applied object cannot feature
in a monoclausal construction whose nucleus would be the base verb.

(9) a. lʊ̀rátɔ́ !
ʊ́ -tɬáà-rʊ́ k-á mʊ̀sɪ̂ ːsɪ ̀.
Lorato(1) sI:cl1-fut-sew-fv dress(3)
‘Lorato will sew a dress.’
220 Denis Creissels

b. lʊ̀rátɔ́ !
ʊ́ -tɬáà-rʊ́ k-ɛ́l-á dím̀ pʰɔ́ mʊ̀sɪ̂ ːsɪ ̀.
Lorato(1) sI:cl1-fut-sew-appl-fv Dimpho(1) dress(3)
‘Lorato will sew a dress for Dimpho.’
c. ✶
lʊ̀rátɔ́ !
ʊ́ -tɬáà-rʊ́ k-á dím̀ pʰɔ́ mʊ̀sɪ̂ ːsɪ ̀.
Lorato(1) sI:cl1-fut-sew-fv Dimpho(1) dress(3)
d. ✶lʊ̀rátɔ́ !
ʊ́ -tɬáà-rʊ́ k-á mʊ̀sɪ́sɪ́ prep dím̀ ːpʰɔ́.
Lorato(1) sI:cl1-fut-sew-fv dress(3) Dimpho(1)

In some Bantu languages, an applicative suffix cognate with -ɛl can be found not only
in obligatory applicative constructions, but also in optional applicative constructions
in which the applied object expresses the role of instrument, also expressible by means
of a prepositional phrase in a non-applicative construction (Pacchiarotti, this volume).
However, this possibility does not exist in Tswana, where instrumental adjuncts can
only be expressed as prepositional phrases.

4.2 The treatment of the initial object in applied-object

constructions

Consistently with the use of double-object constructions as the coding frame of triva-
lent verbs, in Tswana, the presence of an applied object in applicative constructions of
transitive verbs does not necessitate the demotion of the initial object: if the base verb
is transitive, the construction of the applicative verb is a double-object construction
similar to that of trivalent verbs, showing the same symmetries, and subject to the same
animacy-driven rule as regards the order of the object phrases and object indexes. For
example, in an applied-object construction such as (10a), both the initial object and
the applied object can be indexed at the same time and can equally be converted into
the subject of a passive construction. The only constraint is that, in the construction of
rʊ́ kέlá, the promotion of the initial object in a passive construction blocks the possibility
of indexing the applied object, but a similar constraint operates in the double-object
construction of underived verbs such as ‘give’.

(10) a. lʊ̀rátɔ́ !
ʊ́ -tɬáà-rʊ́ k-ɛ́l-á dím̀ pʰɔ́ mʊ̀sɪ̂ ːsɪ ̀.
Lorato(1) sI:cl1-fut-write-appl-fv Dimpho(1) dress(3)
‘Lorato will sew a dress for Dimpho.’
b. lʊ̀rátɔ́ !
ʊ́ -tɬáà-ʊ́ -mʊ̀-rʊ́ k-ɛ̂ːl-à.
Lorato(1) sI:cl1-fut-oI:cl3-oI:cl1-sew-appl-fv
‘Lorato will sew it for her.’
c. dìm̀ pʰɔ́ !
ʊ́ -tɬáà-rʊ́ k-ɛ́l-w-á mʊ̀sɪ̂ ːsɪ ̀.
Dimpho(1) sI:cl1-fut-sew-appl-pass-fv dress(3)
Lit. ‘Dimpho will be sewn.for a dress.’
 8 Applicative constructions in Tswana 221

d. mʊ̀sɪ́sɪ́ !ʊ́ -tɬáà-rʊ́ k-ɛ́l-w-á dím̀ ːpʰɔ́.

dress(3) sI:cl3-fut-sew-appl-pass-fv Dimpho(1)
Lit. ‘The dress will be sewn.for Dimpho.’
e. dìm̀ pʰɔ́ !
ʊ́ -tɬáà-ʊ́ -rʊ́ k-ɛ̂ːl-w-à.
Dimpho(1) sI:cl1-fut-oI:cl3-sew-appl-pass-fv
Lit. ‘Dimpho will be sewn.for it.’
f. ✶mʊ̀sɪ́sɪ́ !ʊ́ -tɬáá-mʊ̀-rʊ́ k-ɛ̂ːl-w-à.
dress(3) sI:cl3-fut-oI:cl1-sew-appl-pass-fv

Depending on the valency of the base verb, it is possible to have three-object construc-
tions in which one of the objects is an applied object, whereas the other objects either
express arguments of the underived verb, or have been introduced by causative deriva-
tion. The latter possibility, already illustrated in example (3) above, is further illustrated
in (11).

(11) nɛ́ɔ́ !
ʊ́ -tɬáà-rʊ́ k-ís-éts-à dìm̀ pʰɔ́ lʊ̀rátɔ́ mʊ̀sɪ̂ ːsɪ ̀.
Neo(1) sI:cl1-fut-write-caus-appl-fv Dimpho(1) Lorato(1) dress(3)
‘Neo will make Lorato sew a dress for Dimpho.’

Examples (12) and (13) illustrate three-object constructions resulting from applicativi-
zation of a verb whose basic coding frame is a double-object construction.

(12) a. kɪ ̀-f-íl-é ŋwánàkɛ́ màːdí.

sI:1sg-give-prf-fv my.child(1) money(6)
‘I gave money to my son.’
b. kɪ ̀-f-éts-ɪ́ ŋwánàkɛ́ báɪ ̀sɪ́kɪ ̀lɪ́ màːdí.
sI:1sg-give-appl.prf-fv my.child(1) bicycle(9) money(6)
‘I gave money to my son for a bicycle.’
c. kɪ ̀-á-ɪ́-mʊ̀-f-êːts-ɪ ̀.
sI:1sg-oIːcl6-oI:cl9-oIːcl1-give-appl.prf-fv
‘I gave it to him for it.’

(13) a. kɪ ̀-f-íl-é díqʰòmʊ́ lɪ ̀tswâːɪ ̀.

sI:1sg-give-prf-fv cows(10) salt(5)
‘I gave salt to the cows.’
b. kɪ ̀-f-éts-ɪ́ màlʊ́ mɛ́ díqʰòmʊ́ lɪ ̀tswâːɪ ̀.
sI:1sg-give-appl.prf-fv my.uncle(1) cows(10) salt(5)
‘I gave salt to the cows for my uncle.’
c. kɪ ̀-lɪ́-dí-mʊ̀-f-êːts-ɪ ̀.
sI:1sg-oIːcl5-oI:cl10-oIːcl1-give-appl.prf-fv
‘I gave it to them for him.’
222 Denis Creissels

Comparison of examples (12) and (13) shows that the order of the objects in such con-
structions is not determined by their syntactic status as base objects or applied objects,
but simply by the general animacy-based rule that also determines the linear order of
objects with verbs whose basic coding-frame is a double-object construction.8

4.3 The semantic roles of applied objects

Some languages have applicative markers that specify the semantic role expressed
by the applied object, or at least limit the semantic roles that can be expressed by the
applied phrase. This is clearly not the case in Tswana, which can be characterized as
having a semantically unspecified applicative marker available to license the expres-
sion of semantic roles that cannot be expressed as objects in a non-applicative con-
struction, or by means of a preposition. In applied-object constructions, three semantic
types of applied objects can be distinguished; they are examined in turn in Sections
4.3.1–4.3.3.

4.3.1 Applied objects representing non-essential participants

Examples (12) and (13) above show that, in the applied-object construction of ‘give’, the
applied object may express the roles of beneficiary or purpose. Applied objects inter-
preted as beneficiaries are particularly common, but applied objects with a meaning of
cause or purpose (these two meanings being often difficult to distinguish) are also quite
common. Example (14) illustrates the possibility of a benefactive or purposive interpre-
tation of the applied object of bílétsá (applicative form of bítsá ‘call’).

(14) a. mʊ̀sádí !
ʊ́ -bíl-éts-á bàná dìːdʒɔ́.
woman(1) sI:cl1-call-appl-fv children(2) food(10)
‘The woman is calling the children to eat.’
b. mʊ̀sádí !
ʊ́ -bíl-éts-á bàná ŋâːkà.
woman(1) sI:cl1-call-appl-fv children(2) doctor(9)
‘The woman is calling the doctor for the children.’

Examples (15) to (20) further illustrate the possibility of a causal or purposive reading of
applied objects. Note that, in examples (18) and (20), the applied phrase is an infinitive.

8 An exception to this rule will be mentioned in Section 4.3.1.

 8 Applicative constructions in Tswana 223

(15) kítsɔ́ !
ʊ́ -bɛ́rɛ́k-ɛ́l-à tíɛ̂ːχɔ ̀.
Kitso(1) sI:cl1-work-appl-fv delay(9)
Lit. ‘Kitso is working for the delay.’ (> in order to make up for lost time)

(16) mà-χòdù á-bʊ́ lá-éts-ɪ́ mʊ̀ńná màːdí.

thieves(6) sI:cl6-kill-appl.prf-fv man(1) money(6)
‘The thieves killed the man for money.’

(17) kɪ ̀-lèbʊ̀χ-ɛ ̀l-à kítsɔ́ màːdí.

sI:1sg-thank-appl-fv Kitso(1) money(6)
‘I am thanking Kitso for the money.’

(18) lʊ̀sɪ́á !
lʊ́ -lɪ́l-ɛ́l-à χʊ̀-âːɲà.
baby(11) sI:cl11-cry-appl-fv inf-suck
‘The baby is crying [because he wants] to suck.’

(19) kɪ ̀-tɬàà-bɔ́n-ɛ́l-á ńtɬʊ̀ káɪ́ màːdí?

sI:1sg-fut-see-appl-fv house(9) where money(6)
‘Where shall I find money for the house?’

(20) màpòdísí á-mʊ̀-tsʰwár-éts-ɪ́ χʊ̀-χát-ís-á mʊ́ tʰʊ̀

policemen(6) sI:cl6-oI:cl1-arrest-appl.prf-fv inf-step.on-caus-fv person(1)
m̀ mótʊ́ rʊ̀kâːrà.
car(3)
‘The policemen arrested him for driving over a person with his car.’

(21) mʊ̀sɪ́sɪ ̀ ó kɪ́-tɬàà-ʊ́ -ápár-ɛ ̀l-à mʊ̀dìːrɔ ̀.

dress(3) cl3.dem sI:1sg-fut-oI:cl3-wear-appl-fv ceremony(3)
‘This dress, I’ll wear it for the ceremony.’

In particular, constructions with the interrogative pronoun ɪ ̀ŋ́ in the role of applied
object are a very common strategy to question about the cause or purpose of the event,
as in (22), to be compared with (18) above, or (23).

(22) lʊ̀sɪ́á !
lʊ́ -lɪ́l-ɛ́l-à ɪ ̀ŋ́ ?
baby(11) sI:cl11-cry-appl-fv what
‘Why is the baby crying?’

(23) ʊ̀-rɪ́-tɬʰɔ́dí-ɛ́l-à ɪ ̀ŋ́ mó χàrɪ́ χ-á-bʊ̀sîːχʊ̀.

sI:2sg-oI:1pl-disturb-appl-fv what loc middle(17) cl.17-gen-night(14)
‘Why do you disturb us in the middle of the night?’
224 Denis Creissels

In this particular use of applied-object constructions, a special rule according to which

ɪ ̀ŋ́ referring to the cause or purpose of the event must immediately follow the verb
supersedes the general rule determining the linear order of objects in multiple-object
constructions (Cole 1955: 432), as can be observed by comparing (24a) (where ɪ ̀ŋ́ repre-
sents the initial object of ‘buy’) with (24b) (where ɪ ̀ŋ́ in the role of applied object is the
equivalent of English why).

(24) a. ʊ̀-rɛ́k-ɛ́l-á bàná ɪ ̀ːŋ́ ?

sI:2sg-buy-appl-fv children(2) what
‘What are you buying for the children?’
b. ʊ̀-rɛ́k-ɛ́l-ɛ́l-à ɪ ̀ŋ́ bàná díàpàːrɔ ̀?
sI:2sg-buy-appl-appl-fv what children(2) clothes(10)
‘Why are you buying clothes for the children?’

4.3.2 Applied objects expressing the role of concernee

One of the possible functions of applicative derivation in Tswana (as in many other
languages) is to license a concernee-concern (or “external possession”) construction
with the concernee (“external possessor”) in the role of applied object, the concern
fulfilling the subject or object role in accordance with its role in the event denoted by
the verb.9
In Tswana, non-applicative constructions with an object phrase expressing the
role of concernee are possible if the relationship that motivates the use of a concern-
ee-concern construction is a whole-part relationship, as in (25) and (26). Note that
sɪ ̀tɬʰàrɪ ̀ in (25) and ŋwàná in (26) cannot be analyzed as adnominal possessors, since
the corresponding adnominal possession constructions would be dɪ ̀kàlà ts-á-sɪ́tɬʰàrɪ ̀‘the
branches of the tree’ and sɪ ̀àtɬà s-á-ŋwàná ‘the child’s hand’.

(25) kɪ ̀-rém-íl-é sɪ́tɬʰàrɪ ̀ dìkàːlà.

sI:1sg-chop-prf-fv tree(7) branches(10)
Lit. ‘I chopped the tree the branches.’ > ‘I chopped off the branches of the tree.’

(26) mʊ̀ńná !ʊ́ -tsʰwér-ɪ́ ŋwàná sɪ́àːtɬà.

man(1) sI:cl1-seize.prf-fv child(1) hand(7)
Lit. ‘The man seized the child the hand.’ > ‘The man seized the child by the hand.’

9 For a discussion of the notion of concernee-concern construction, and a general survey of concern-
ee-concern constructions in Bantu, see Van de Velde (2020).
 8 Applicative constructions in Tswana 225

However, if the semantic relationship between the concernee and the concern is other
than a whole-part relationship, the concernee must be encoded as an applied object, as
in (27) and (28).

(27) a. m̀ pʰɔ́ !
ʊ́ -dʒ-ɪ́l-é dínàːwá.
Mpho(1) sI:cl1-eat-prf-fv beans(10)
‘Mpho ate the beans.’
b. m̀ pʰɔ́ !
ʊ́ -dʒ-éts-ɪ́ !
kítsɔ́ dínàːwá.
Mpho(1) sI:cl1-eat-appl.prf-fv Kitso(1) beans(10)
Lit. ‘Mpho ate.appl Kitso the beans.’ > ‘Mpho ate Kitso’s beans.’ (i.e., the beans
that had been prepared for Kitso)

(28) a. mʊ̀χɔ́tɛ́ ʊ́ -χʊ̀d-îːl-è.

fever(3) sI:cl3-grow-prf-fv
‘The fever has gone up.’
b. mʊ̀χɔ́tɛ́ !ʊ́ -mʊ̀-χʊ́ l-êːts-ɪ ̀.
fever(3) sI:cl3-oI:cl1-grow-appl.prf-fv
Lit. ‘The fever has gone.up.appl him.’ > ‘His fever has gone up.’

Examples (29) and (30) illustrate the conversion of an applied object representing a con-
cernee into the subject of a passive construction. Such a combination of applicativiza-
tion and passivization results in constructions functionally similar to the “adversative
passives” found in Japanese and other languages.

(29) rɪ ̀-ʃ-ɛ́l-w-à kɪ́ máǹːtɬʊ̀.

sI:1pl-burn-appl-pass-fv by houses(6)
Lit. ‘We are burnt.appl by houses.’ > ‘Our houses are burning.’

(30) kítsɔ́ !
ʊ́ -sw-éts-w-ɪ ̀ kɪ́ r ̀ráàːχwɛ́.
Kitso(1) sI:cl1-die-appl.prf-pass-fv by his.father(1)
Lit. ‘Kitso has been died.appl by his father.’ > ‘Kitso’s father has died.’

Semantically, there is an obvious affinity between applied objects expressing the seman-
tic role of concernee and applied objects expressing the semantic role of beneficiary.
The difference is that the notion of concernee (in contrast to the notion of beneficiary,
which carries no such implication) implies the existence of a previously established
relationship between the concernee and the concern.
226 Denis Creissels

4.3.3 Applied objects referring to essential participants

In Tswana, applicative derivation does not only license applied objects expressing
semantic roles independent from the lexical meaning of the verb. In many cases, the
applied object is in fact best analyzed as a semantic argument of the verb that can
only be expressed as an applied objet, since its semantic role is implied by the lexical
meaning of the verb.
For example, the lexical meaning of dúɛ́lá ‘pay’ implies three participants: the
payer, the recipient, and the thing being paid. In Tswana, dúɛ́lá in its underived form
can only be used in a single-object construction whose object represents the recipient
(31a-b), and there is no possibility of referring to the thing being paid by means of a
prepositional phrase. Consequently, the thing being paid can only be mentioned as the
applied object of the derived verb dúɛ́lɛ́là, as in (31c).

(31) a. kɪ ̀-tɬàà-χʊ̀-dúɛ́l-à ká !
tʃʰɛ̂ːkɛ ̀.
sI:1sg-fut-oI:2sg-pay-fv with check(9)
‘I’ll pay you by check.’
b. ✶kɪ ̀-tɬàà-dúɛ́l-á !
páákâːɲɔ ̀.
sI:1sg-fut-pay-fv repair(9)
intended: ‘I’ll pay the repair fee.’
c. kɪ ̀-tɬàà-dúɛ́l-ɛ́l-à páákâːɲɔ ̀.
sI:1sg-fut-pay-appl-fv repair(9)
‘I’ll pay the repair fee.’

Examples (32) to (36) provide further illustrations of applied objects referring to partic-
ipants for which an analysis as essential participants in the event denoted by the verb
can be considered.

(32) a. qʰósí !
ɪ́-átɬʰʊ́ l-éts-ɪ́ mʊ̀ńná bʊ́ χòːdù.
king(9) sI:cl9-condemn-appl.prf-fv man(1) theft(14)
‘The king condemned the man for theft.’
b. qʰósí !
ɪ́-átɬʰʊ́ l-éts-ɪ́ mʊ̀ńná lʊ̀ːsʊ́ .
king(9) sI:cl9-condemn-appl.prf-fv man(1) death(11)
‘The king condemned the man to death.’

(33) mʊ̀sádì jó !
ʊ́ -ák-ɛ́l-à rálɪ ̀bɪ́ńtɬɪ̂ ːlɪ ̀.
woman(1) cl1.dem sI:cl1-tell.lies-appl-fv shopkeeper(1)
‘This woman is telling lies about the shopkeeper.’

(34) màbɛ ̀lɛ́ !

á-áláf-ɛ ̀l-w-à tsʰùːpà.
sorghum(6) sI:cl6-treat-appl-pass-fv sg.tshupa(9)
‘The sorghum is treated against tshupa (a kind of worm).’
 8 Applicative constructions in Tswana 227

(35) mʊ̀sétsánà jó !
ʊ́ -fós-éts-à sɪ ̀tswâːnà.
girl(1) cl1.dem sI:cl1-miss-appl-fv Tswana.customs(7)
‘This girl contravenes Tswana customs.’

(36) kɪ ̀-χáqʰámál-ɛ ̀l-à bʊ̀pɪ ̀lʊ́ qʰálɪ ̀ dʒw-á-ŋwànáː jò.

sI:1sg-be.impressed-appl-fv courage(14) cl14-gen-child(1) cl1.dem
‘I am impressed by the courage of this child.’

In this connection, the behavior of kwálá ‘write’ is particularly interesting to analyze.

kwálá is basically a bivalent verb, but the selection of lɔ ̀kwálɔ́ ‘letter’ as its object implies
a third essential participant with the semantic role of recipient, since a letter is intended
to be sent to someone. However, the recipient of kwálá lʊ̀kwálɔ́ ‘write a letter’, contrary
to the recipient of the verbs that are inherently verbs of giving, must be encoded as an
applied object, with the consequence that, in a clause such as (37b), the applied object
can be understood as referring to the recipient (essential participant) or to a beneficiary
(non-essential participant). Note that the repetition of the applicative suffix makes it pos-
sible to express both roles simultaneously, and then the applied object that immediately
follows the verb is unambiguously interpreted as a beneficiary, whereas the applied
object in second position is interpreted as the recipient, exactly as in the three-object
construction expressing ‘give s.th. to s.o. in behalf of s.o.’ (see example (2) above).

(37) a. kɪ ̀-tɬàà-kwál-á lʊ̀kwâːlɔ ̀.

sI:1sg-fut-write-fv letter(11)
‘I’ll write a letter.’
b. kɪ ̀-tɬàà-kwál-ɛ́l-á m̀ pʰɔ́ lʊ̀kwâːlɔ ̀.
sI:1sg-fut-write-appl-fv Mpho(1) letter(11)
‘I’ll write a letter to Mpho.’
OR ‘I’ll write a letter on behalf of Mpho.’
c. kɪ ̀-tɬàà-kwál-ɛ́l-ɛ́l-à r ̀rɛ́ m̀ pʰɔ́ lʊ̀kwâːlɔ ̀.
sI:1sg-fut-write-appl-appl-fv my.father(cl1) Mpho(1) letter(11)
‘I’ll write a letter to Mpho on behalf of my father.’

5 Applied-oblique constructions
5.1 Introductory remarks

The definition of applicative constructions adopted in this chapter includes construc-

tions in which a derived form of the verb assigns an oblique grammatical relation to an
NP (the applied phrase) representing a non-core participant that could not be coded in
the same way (or could not be coded at all) in the construction of the base verb.
228 Denis Creissels

Bantu languages in general, and Tswana in particular, provide many interesting

data about constructions meeting the broad definition of applicative constructions that
involve the same derived form of verbs as applied-object constructions, but in which
the applied phrase is a locative phrase that does not have the syntactic status of object.
In the uses of -ɛl examined in this section, as in its use in applied-object con-
structions, -ɛl licenses a term expressing a particular semantic role that could not be
expressed in the construction of the base verb. However, the term in question is not
encoded as an object NP, but as a locative phrase showing no evidence of a syntac-
tic status different from that of ordinary obliques: it cannot be cross-referenced by an
object index, or converted into the subject of a passive construction, and more gener-
ally, apart from the fact that its deletion results in ungrammaticality, it behaves like
locative phrases accompanying non-derived verbs.
For a proper understanding of the data commented in this section, it is crucial
to keep in mind that, in Tswana, as in the vast majority of Bantu languages, locative
phrases are not specified for the location vs. source vs. destination distinction.10 In
Tswana, the precise semantic role of locatives in the construction of non-applicative
verbs is regulated in the following way:
(a) any Tswana verb can combine with a locative phrase expressing the location of the
event, or of a participant in the event, as in example (38);
(b) in combination with some motion verbs, locative phrases are assigned the seman-
tic role of destination, as in example (39);
(c) with some other motion verbs, locative phrases are assigned the role of source, as
in example (40).

(38) kítsɔ́ !
ʊ́ -bɛ́rɛ́k-à kó kàːɲɛ́.
Kitso(1) sI:cl1-work-fv loc Kanye
‘Kitso is working in Kanye.’

(39) kítsɔ́ ʊ́ -ìl-é kó kàːɲɛ́.

Kitso(1) sI:cl1-go.prf-fv loc Kanye
‘Kitso went to Kanye.’

(40) kítsɔ́ !
ʊ́ -húdúχ-íl-è kó kàːɲɛ́.
Kitso(1) sI:cl1-move-prf-fv loc Kanye
‘Kitso moved from Kanye.’

10 This particularity in the encoding of spatial relationships is found in many language families of
sub-Saharan Africa, in particular (but not only) among those belonging to the Niger-Congo phylum.
 8 Applicative constructions in Tswana 229

Interestingly, applicative derivation may license locative phrases expressing roles other
than the role they would express in combination with the underived form of the same
verb. Three cases must be distinguished.

5.2 Motion verbs that cannot assign the role of source

or destination

tábʊ́ χá ‘run’ is semantically a motion verb, but in its underived form, it has no seman-
tic role to assign to a locative phrase, which means that the only available interpre-
tation for a locative term in the construction of tábʊ́ χá in its underived form is the
default interpretation of location of the event. By contrast, a locative in the construc-
tion of the applicative form tábʊ́ χɛ́là can be interpreted as referring to destination of
motion—Example (41b). The same behavior is observed with àkʊ̀fà ‘hurry’, fʊ̀fà ‘fly’,
fɪ ̀tà ‘pass’, etc.

(41) a. kɪ ̀-tɬàà-tábʊ́ χ-à kó tsɪ ̀lêː-ŋ̀.

sI:1sg-fut-run-fv loc road(9)-loc
‘I will run on the road.’
b. kɪ ̀-tɬàà-tábʊ́ χ-ɛ́l-à kó tsɪ ̀lêː-ŋ̀.
sI:1sg-fut-run-appl-fv loc road(9)-loc
‘I will run to the road.’11

In this particular case (but not in those examined in the remainder of this section), an
applied-object construction, as in (42), would be possible with the same meaning.

(42) kɪ ̀-tɬàà-tábʊ́ χ-ɛ́l-à tsɪ ̀ːlà.

sI:1sg-fut-run-appl-fv road(9)
‘I will run to the road.’

There is an obvious relationship with the fact that, in Tswana, non-derived verbs of
motion that assign the role of destination (such as jà ‘go’) have an alternative construc-
tion in which the destination is encoded as the object of a transitive construction.

11 See Section 6.1 for another possible interpretation of this sentence.

230 Denis Creissels

5.3 Motion verbs that assign the role of source in their

underived form

With motion verbs whose underived form assigns the role of source to locative comple-
ments, the applicative form has the same formal valency as the non-derived form, but
assigns to its locative complement the role of destination, as illustrated in example (43)
by húdúχá ‘change one’s residence’.

(43) a. kɪ ̀-tɬàà-húdúχ-à kó Kàːɲɛ́.

sI:1sg-fut-move-fv loc Kanye
‘I will move from Kanye.’
b. kɪ ̀-tɬàà-húdúχ-ɛ́l-à kó χàbʊ́ rôːnɪ ̀.
sI:1sg-fut-move-appl-fv loc Gaborone
‘I will move to Gaborone.’

Crucially, in contrast to applied-object constructions, the applicative constructions

of motion verbs with a locative as the applied phrase are not valency-increasing, but
valency-rearranging. The addition of the suffix -ɛl to verbs such as húdúχá gives them
the capacity to take a locative complement referring to destination of motion, but at the
same time precludes them from taking a locative complement referring to source of
motion. In order to express ‘move from A to B’, Tswana must use a clause chain with a
first clause in which the non-derived form of húdúχá is followed by a locative comple-
ment specifying the source of motion, and a second clause in which the applicative form
of the same verb is followed by a locative complement specifying the destination of
motion, as in (44). More generally, Tswana, like many languages of sub-Saharan Africa,
cannot specify the source and destination of motion within the frame of monoverbal
constructions.

(44) kɪ ̀-tɬàà-húdúχ-à kó Kàɲɛ́ kɪ ̀-húdúχ-él-ɪ ̀ kó χàbʊ́ rôːnɪ ̀.

sI:1sg-fut-move-fv loc Kanye sI:1sg-move-appl-fv loc Gaborone
‘I will move from Kanye [and will move.appl] to Gaborone.’

Here again, it is interesting to observe that the applied phrase cannot be analyzed as
referring to a marginal participant or a circumstance of the event, since the lexical
meaning of húdúχá ‘change one’s residence’ cannot be defined without mentioning
both a source and a destination of motion, and there is no a priori reason why special
marking should be required to express the destination but not the source, rather than
the other way round.
 8 Applicative constructions in Tswana 231

5.4 Verbs that do not express motion

Verbs that do not express motion freely combine with locatives expressing the location
of the event or of a participant, as already illustrated by example (38) above, but the use
of the applicative form is obligatory to license the presence of a locative whose semantic
role departs more or less from the mere indication of a location.
For example, Tswana syntax is sensitive to the difference in the semantic role of in
the yard and in the big pot in She is cooking porridge in the yard / She is cooking porridge
in the big pot. In the first sentence, in the yard expresses nothing more than the location
of the event, whereas in the event represented by the second sentence, the pot contains
the porridge, which justifies coding it as a locative, but it is also an essential element of
the porridge cooking event, in which it plays the role of an indispensable instrument. In
other words, the spatial relationship between the pot and the porridge is not accidental;
it follows from the role they play in the cooking event, and this may explain why, in the
Tswana equivalent of She is cooking porridge in the yard, the verb cook can remain in
its underived form, whereas in the equivalent of She is cooking the porridge in the big
pot, the verb cook must be in the same applicative form as when, for example, a noun
phrase referring to a beneficiary is added to the construction of this verb. Interestingly,
the applicative derivation must be reiterated in order to make it possible to mention
both the vessel used to cook the porridge and the beneficiary of the cooking event—
Example (45).

(45) a. lʊ̀rátɔ́ !
ʊ́ -tɬáá-àpàj-à mʊ̀tɔ ̀ːχɔ́.
Lorato(1) sI:cl1-fut-cook-fv porridge(3)
‘Lorato will cook the porridge.’
b. lʊ̀rátɔ́ !
ʊ́ -tɬáá-àpɛ ̀-ɛ ̀l-à bàná mʊ́ tɔ ̀ːχɔ́.
Lorato(1) sI:cl1-fut-cook-appl-fv children(2) porridge(3)
‘Lorato will cook the porridge for the children.’
c. lʊ̀rátɔ́ !
ʊ́ -tɬáá-àpɛ ̀-ɛ ̀l-à mʊ̀tɔ ̀χɔ́ mó pìtsé-ŋ̀ é
Lorato(1) sI:cl1-fut-cook-appl-fv porridge(3) loc pot(9)-loc cl9.lk
!
tʊ̂ːnà.
cl9.big
‘Lorato will cook the porridge in the big pot.’
d. lʊ̀rátɔ́ !
ʊ́ -tɬáá-àpɛ ̀-ɛ ̀l-ɛ ̀l-à bàná mʊ́ tɔ ̀χɔ́
Lorato(1) sI:cl1-fut-cook-appl-appl-fv children(2) porridge(3)
mó pìtsé-ŋ̀ é !
tʊ̂ːnà.
loc pot(9)-loc cl9.lk cl9.big
‘Lorato will cook the porridge for the children in the big pot.’

Examples (46) and (47) provide additional illustrations of the obligatory use of applica-
tive forms of verbs that do not express motion to license a locative phrase whose seman-
232 Denis Creissels

tic role is not limited to the mere expression of location, since it refers to an essential
element in the situation denoted by the verb.

(46) dìqʰòmʊ́ !dí-nw-ɛ́l-à mó mʊ̀kórôː-ŋ̀.

cows(10) sI:cl10-drink-appl-fv loc mokoro(3)-loc
‘Cows drink from a mokoro.’ (a tree trunk carved in the shape of a canoe)

(47) rɪ ̀-kwál-ɛ́l-à mó pám̀ pírîː-ŋ̀.

sI:1pl-write-appl-fv loc paper(9)-loc
‘We write on paper.’

5.5 The particular case of rʊ́ká ‘sew’ and bɔ́fá ‘bind’

rʊ́ ká ‘sew’ and bɔ́fá ‘bind’ can be found in applicative constructions of the type dis-
cussed in Section 5.4, in which the applied phrase is a locative whose semantic role does
not boil down to the expression of location, since it refers to a participant playing an
essential role in the event referred to: the surface on which something is being fixed
(48), or a stationary object to which something is being tied (49). What is, however,
difficult to explain, is that, with these two verbs (and not with others), the applicative
marker must be repeated in a construction including a single applied phrase.

(48) lʊ̀rátɔ́ !
ʊ́ -rʊ́ k-ɛ́l-ɛ́l-à dìkʊ́ nʊ́ pɔ́ !
mó sɪ́àpàrò-ŋ̀ s-âː-mɪ ̀.
Lorato(1) sI:cl1-sew-appl-appl-fv buttons(10) loc dress(7)-loc cl7-gen-1sg
‘Lorato is sewing buttons on my dress.’

(49) kɪ ̀-tɬàà-bɔ́f-ɛ́l-ɛ́l-à pìtsɪ́ !

fá sɪ́tɬʰàrɪ ̀ː-ŋ̀.
sI:1sg-fut-tie-appl-appl-fv horse(9) loc tree(7)-loc
‘I’ll tie the horse to the tree.’

5.6 Applicative constructions triggered by the adverb rúrí

rúrí is an adverb that can be found in two distinct constructions. In clause-initial posi-
tion, it expresses a meaning that can be glossed as ‘really’, ‘truly’, ‘surely’. It does not
modify the propositional content of the clause and has no incidence on the verb form.
In post-verbal position, it expresses a meaning that can be glossed as ‘for a long time’,
‘for ever’, ‘for real’, ‘irrevocably’. In this construction, in which rúrí contributes to the
propositional content of the clause, the verb must be in the applicative form, and the
deletion of the applicative marker results in ungrammaticality.
 8 Applicative constructions in Tswana 233

(50) bá-tɬáá-χʊ̀-bʊ́ lá-ɛ́l-à rûːrì.

sI:cl2-fut-oI:2sg-kill-appl-fv for.real
‘They will kill you for real.’

(51) bá-nè bà-χópʊ́ l-á χʊ́ rɪ ̀ tóŋ́ kí !

ɪ́-látɬʰéχ-éts-ɪ ̀ rûːrì.
sI:cl2-aux sI:cl2-think-fv that donkey(9) sI:cl9-get.lost-appl.prf-fv definitely
‘They thought that the donkey had definitely been lost.’

This is, to the best of my knowledge, the only case of a Tswana applicative construction
in which the applied phrase is not a nominal or a locative.

6 -ɛl in constructions with instrumental subject

In Tswana, participants usually treated as instrumental adjuncts in the construction of
the underived form of a verb, i.e. encoded as complements of the instrumental preposi-
tion ká, as in (52a), cannot be encoded as applied phrases. However, if no agent is men-
tioned, instruments can take the role of subject in constructions such as (52b), where the
verb is in the same form as in applicative constructions, and the referent of the subject
of the base verb (the agent) is left unexpressed (and is interpreted as non-specific). Even
in a very broad conception of applicative constructions, a crucial property of applica-
tivization is that it triggers no change in the semantic role of the subject. Consequently,
in spite of the fact that the verb form in (52b) obligatorily includes the same marker -ɛl
as the verb forms found in applicative constructions, (52b) cannot be analyzed as an
applicative construction, and rather meets the definition of oblique passivization (i.e.,
subject demotion compensated by the promotion of an oblique to subject role).

(52) a. ʊ́ -nè à-ʃàbà bʊ̀-χɔ́bɛ́ !

ká námà
sI:cl1-aux sI:cl1 -flavor-fv porridge(14) with flesh(9)
j-á-qʰâːkà.
cl9-gen-guinea.fowl(9)
‘He flavored the porridge with the flesh of the guinea-lowl.’
b. nàmà ɪ́-ʃáb-ɛ́l-à bʊ̀χɔ̂ːbɛ ̀.
flesh(9) sI:cl9-flavor-?-fv porridge(14)
‘Meat gives flavor to the porridge.’

As already mentioned above, in Tswana, instruments usually encoded as locatives by

virtue of the spatial relationship they necessarily have with the referent of the object
require applicative marking on the verb. Interestingly, with such instruments, as illus-
trated in (53), promotion to subject does not necessitate the insertion of an additional
applicative marker.
234 Denis Creissels

(53) a. mʊ̀sádí ʊ́ -nè à-tsʰʊ̀l-ɛ ̀l-à bʊ̀χɔ́bɛ́ mó

woman(1) sI:cl1-aux sI:cl1-dish.out-appl-fv porridge(14) loc
mɪ ̀χʊ́ pʊ̂ː-ŋ̀.
wooden.bowls(4)-loc
‘The woman dished out the porridge into the wooden bowls.’
b. mʊ̀χʊ́ pʊ́ !ʊ́ -tsʰʊ́ l-ɛ́l-à bʊ̀χɔ̂ːbɛ ̀.
wooden.bowl(3) sI:cl3-dish.out-?-fv porridge(14)
‘The wooden bowl is used to dish out porridge.’

7 Uses of -ɛl not related to valency operations

7.1 -ɛl and the focalization of locative adjuncts

In Tswana clauses including a locative phrase expressing the location of the event, the
suffix -ɛl may be added to the verb form without any other change in the form of the
clause and without any change in the semantic role of the locative phrase, which rules
out analyzing -ɛl as marking a valency operation (applicativization or other). In this
case, the function of -ɛl is to mark focalization of the locative expressing the location
of the event, as in (54b). This use of -ɛl constitutes an alternative to cleft constructions,
which are in Tswana the standard way to express focalization. However, it is only avail-
able if the term to be focalized is a locative expressing the location of the event.

(54) a. mʊ̀ńnà w-á-mɪ́ ʊ́ -nè à-sw-á kó mʊ̀ráfôː-ŋ̀.

man(1) cl1-gen-1sg sI:cl1-aux sI:cl1-die-fv loc mine(3)-loc
‘My husband died in the mine.’
b. mʊ̀ńnà w-á-mɪ́ ʊ́ -nè à-sw-ɛ́l-à kó mʊ̀ráfôː-ŋ̀.
man(1) cl1-gen-1sg sI:cl1-aux sI:cl1-die-foc-fv loc mine(3)-loc
‘My husband died IN THE MINE.’

Examples (55) and (56) provide further illustration of the focalizing use of -ɛl.

(55) lʊ̀rátɔ́ !
ʊ́ -ápɛ́-ɛ ̀l-à mó dʒáràtêː-ŋ̀.
Lorato(1) sI:cl1-cook-foc-fv loc yard(9)-loc
‘Lorato is doing the cooking IN THE YARD.’

(56) kɪ ̀-tsál-éts-w-ɪ ̀ kó kàːɲɛ́.

sI:1sg-give.birth-foc.prf-foc-fv loc Kanye
‘I was born IN KANYE.’
 8 Applicative constructions in Tswana 235

Interestingly, this use of the suffix -ɛl results in ambiguity in the case of motion verbs
that cannot assign the role of source of motion or destination of motion to a locative
complement, since with such verbs, -ɛl may also mark an applicative construction in
which the locative expresses the role of destination of motion. For example, in (41)
above, repeated here as (57), the second sentence is in fact ambiguous between an
interpretation according to which -ɛl marks a change in the role-assigning properties of
tábʊ́ χá (‘I will run to the road [not on the road]’), and another interpretation according
to which the applicative suffix marks the focalization of a locative phrase without
modifying its semantic role of location.

(57) a. kɪ ̀-tɬàà-tábʊ́ χ-à kó tsɪ ̀lêː-ŋ̀.

sI:1sg-fut-run-fv loc road(9)-loc
‘I will run on the road.’
b. kɪ ̀-tɬàà-tábʊ́ χ-ɛ́l-à kó tsɪ ̀lêː-ŋ̀.
sI:1sg-fut-run-appl/foc-fv loc road(9)-loc
‘I will run to the road.’ OR ‘I will run ON THE ROAD (and nowhere else).’

7.2 -ɛl and the expression of habituality of action at some place

In Tswana clauses including a locative phrase expressing the location of the event, in
addition to its use to mark the focalization of the locative phrase without any change in
the assignment of semantic roles, the applicative marker -ɛl can also be used to express
the aspectual notion of habituality of action at the place referred to by the locative
phrase, without any change in the construction or in the assignment of semantic roles.
Example (55) above, repeated here as (58b), is in fact ambiguous between a focalizing
and a habitual reading of the verbal suffix -ɛl also used in applicative function.

(58) a. lʊ̀rátɔ́ !
ʊ́ -ápáj-à mó dʒáràtêː-ŋ̀.
Lorato(1) sI:cl1-cook-fv loc yard(9)-loc
‘Lorato is doing the cooking in the yard.’
b. lʊ̀rátɔ́ !
ʊ́ -ápɛ́-ɛ ̀l-à mó dʒáràtêː-ŋ̀.
Lorato(1) sI:cl1-cook-foc/hab-fv loc yard(9)-loc
‘Lorato does the cooking IN THE YARD.’
OR ‘Lorato habitually cooks in the yard.’

7.3 -ɛl and the expression of intensity

In many Bantu languages, applicative markers also have a use in which they imply
neither a formal change in the construction nor a change in the semantic roles, and are
interpreted as expressing intensity of the action (completeness, persistency, effort, iter-
236 Denis Creissels

ativity). Most of the time, this use of applicative markers implies reduplication, and this
the case in Tswana, as illustrated in (59) by lɪ ̀b-à ‘look at’ > lɪ ̀b-ɛ ̀lɛ ̀l-à ‘watch carefully’.

(59) a. lɪ́b-á sɪ́tɬʰàrɪ ̀ː sé!

look.at-fv tree(7) cl7.dem
‘Look at this tree!’
b. bàná bá-ì-tʰút-á tírɔ ̀ j-á-χʊ́ -lʊ́ χ-à
children(2) sI:cl2-REFL-teach-fv work(9) cl9-gen-inf-plait-fv
ká χʊ̀-lɪ́b-ɛ́lɛ́l-à bàχʊ̂ːlʊ̀.
by inf-look.at-emph-fv adults(2)
‘Children learn to plait by observing adults.’

The following verb pairs illustrate the intensive use of -ɛlɛl:

χán-á ‘refuse’ χán-ɛ́lɛ́l-à ‘refuse completely, be adamant’

ɔ́m-á ‘become dry’ ɔ́m-ɛ́lɛ́l-à ‘become completely dried out’
fɪ ̀t-à ‘pass’ fɪ ̀t-ɛ ̀lɛ ̀l-à ‘go too far, exaggerate’
dùp-à ‘scent’ dùp-ɛ ̀lɛl-à ‘scent’ (more emphatic than dùp-à)
χát-á ‘tramp, ‘tread’ χát-ɛ́lɛ́l-à ‘press down, oppress’
qʰábétɬ-á ‘chop into pieces’ qʰábétɬ-ɛ́lɛ ̀l-à ‘chop into very small pieces’
lɪ́k-á ‘try, attempt’ lɪ́k-ɛ́l-ɛ́là ‘try out, test, fathom’
tsɛ́n-á ‘enter’ tsɛ́n-ɛ́lɛ́l-à ‘go deep into’
útɬw-á ‘hear’ útɬw-ɛ́lɛ́l-à ‘listen’
bàl-à ‘count’ bàl-ɛ ̀lɛ ̀l-à ‘calculate with precision’

There are also pairs such as χ-á / χ-ɛ́lɛ́l-á ‘draw (water)’, or lètɬ-à / lètɬ-ɛ ̀lɛ ̀l-à ‘allow,
permit’, described as fully synonymous in the dictionaries.
Interestingly, with some verbs, the same meaning of intensity without any change
in the valency is expressed by the reduplication of the causative suffix.

8 Lexicalized applicatives
The verbal lexicon of Tswana includes a non-negligible proportion of verbs whose stem
ends with -ɛl, and for which a semantically plausible source of derivation can be iden-
tified, but with a meaning and a construction that preclude a synchronic analysis in
terms of applicativization. Such verbs can be designated as lexicalized applicatives, or
pseudo-applicatives.
For example, lálɛ́lá ‘have dinner’ is probably cognate with lálá ‘lie down, go to
bed, spend the night’. Having dinner is precisely what one normally does before going
to bed, and consequently, there is no difficulty in analyzing a semantic shift from ‘lie
 8 Applicative constructions in Tswana 237

down’ to ‘have dinner’ as a case of metonymy. However, the details of the evolution that
led to the present situation are unclear, particularly regarding the possible involvement
of an applicative marker. The only sure thing is that, in present-day Tswana, lálɛ́lá and
lálá are equally intransitive, and ká díqʰɔ ̀bɛ ̀ in (60b) shows no evidence of being any-
thing else than an ordinary adjunct.

(60) a. rɪ ̀-tɬàà-lál-à mó náχèː-ŋ̀.

sI:1pl-fut-lie.down-fv loc bush(9)-loc
‘We will lie down / spend the night in the bush.’
b. rɪ ̀-tɬàà-lálɛ́l-à ká díqʰɔ ̀ːbɛ ̀.
sI:1pl-fut-have.dinner-fv with maize.and.beans(10)
‘We will have maize-and-beans for dinner.’
Lit. ‘We will have dinner with maize-and-beans.’

Similarly, Tswana ìlɛ ̀là ‘revere’ is certainly a reflex of the same Proto-Bantu root ✶gid
‘abstain from, avoid’ as ìlà ‘hate’, since it is easy to imagine how ‘revere’ and ‘hate’
may have developed as two diverging specializations of the meaning ‘abstain from,
avoid’ reconstructed for this root. What is much less clear is the role that an applicative
marker might have played in this process, since synchronically, as illustrated in (61),
both ìlɛ ̀là ‘revere’ and ìlà ‘hate’ are plain transitive verbs.

(61) a. kɪ́ ɪ ̀ŋ́ !

ʊ́ -mʊ́ -îːl-à?
it.is what sI:2sg-sI:cl1-hate-fv
‘Why do you hate him/her?’
b. bá-ílɛ́l-à mʊ̀dîːmʊ̀.
sI:cl2-revere-fv god(3)
‘They revere God.’

Tswana also has many verbs ending with -ɛlɛla that may be analyzed as resulting from
the lexicalization of the intensive use of the reduplication of the applicative suffix, such
as ɛ́mɛ́lɛ́là ‘move off’, probably related etymologically to ɛ́má ‘stand up’.
Pacchiarotti (2020) provides an in-depth analysis of the lexicalized applicatives of
Tswana, to which readers are referred for a comprehensive account.

9 The question of applicative-like uses

of the causative marker
Some of the Bantu languages that have a semantically under-specified applicative
marker cognate with -ɛl productively use it in applicative constructions in which the
applied object expresses the role of instrument, whereas others (for example, Kinyar-
238 Denis Creissels

wanda) encode instruments as applied objects in applicative constructions in which the

verb is not in the form used for other semantic types of applicative constructions, but
in the form typically used in causative constructions. This strategy can be viewed as
expressing a conceptualization ‘Agent makes Instrument act on Patient’, made possible
by the involvement of instruments in the causal chain.
In Tswana, instruments are standardly encoded by means of the preposition ká,
without any marking on the verb, as in (62).

(62) ŋwàná ʊ́ -ìté-íl-é ɲ̀tʃá !

ká tʰûːpà.
child(1) sI:cl1-hit-prf-fv dog(9) with stick(9)
‘The child hit the dog with a stick.’

However, one may wonder whether sentences such as (63) could not be analyzed as
involving a marginal applicative-like use of the causative marker -is.

(63) kítsɔ́ !
ʊ́ -χát-ís-íts-é ɲ̀tʃá m̀ mótʊ́ rʊ̀kâːrà.
Kitso(1) sI:cl1-step.on-caus-prf-fv dog(9) car(3)
‘Kitso drove over a dog with his car.’

The analysis of (63) as an applicative construction with an applied object expressing

the role of instrument is suggested by the English translation. However, ‘car’ in (63) is
clearly not a typical instrument, nor is ‘Kitso’ (normally) interpreted as a typical agent,
but rather as an “involuntary agent”. Moreover, the event referred to as (63) includes a
sub-event that can be referred to as (64).

(64) m̀ mótʊ́ rʊ̀kárá !ʊ́ -χát-íl-é ɲ̀ːtʃá.

car(3) sI:cl3-step.on-prf-fv dog(9)
‘The car ran over a dog.’

Consequently, there is no reason not to analyze ‘the car’ in (63) as the causee in a permis-
sive causative construction: ‘Kitso (inadvertently) let the car run over a dog’. In other
words, such constructions do not put into question the statement that all the types of
applicative constructions that are possible in Tswana make use of the same applicative
marker -ɛl.

10 Conclusion
In this chapter, based on a relatively broad definition of applicative constructions, I
have surveyed the various types of applicative constructions attested in Tswana, and
the non-applicative uses of applicative morphology. According to the questionnaire pro-
 8 Applicative constructions in Tswana 239

posed as a guideline for the contributions to this volume, Tswana applicative construc-
tions can be characterized as follows:

Morphology
– All subtypes of applicative constructions attested in Tswana make use of the same
marker, a verbal suffix occupying the same slot in the verbal template as several
other valency-changing suffixes (causative, anticausative, and reciprocal).
– Tswana does not have constructions that could be analyzed as more or less gram-
maticalized applicative periphrases or analytical applicative constructions.
– The allomorphs of the applicative suffix involve no lexical conditioning, they can
be exhaustively described as resulting from regular morpho-phonological pro-
cesses operating on an underlying form -ɛl.
– The presence of the applicative marker -ɛl has no incidence on the other aspects of
verb morphology.

Syntax
– Applicativization is not conditioned by the transitivity properties of the base verb;
some uses of applicative constructions are conditioned by the participant frame
of the base verb, but, for example, applied-object constructions with the applied
object representing a beneficiary are equally possible with intransitive, transitive,
and ditransitive verbs.
– Tswana has both applicative constructions in which the applied phrase is a noun
phrase showing all the properties that characterize objects in non-applicative con-
structions, and applicative constructions in which the applied phrase is a locative
showing no evidence of a syntactic status distinct from that of ordinary obliques.
– With just one exception, discussed in Section 4.3, Tswana applicatives are valen-
cy-increasing (as opposed to valency-rearranging) applicative constructions, in
which the status of the applied phrase’s companion arguments/adjuncts does not
change between the base construction and the applicative construction.
– There are no particular restrictions on the combination of applicativization
with the other types of valency-changing operations that are grammaticalized in
Tswana (causativization, anticausativization, reflexivization, reciprocalization,
passivization), apart from the fact that combinations that would result in construc-
tions including more than three objects are ruled out. Moreover, repetition of the
applicative marker may license constructions including two applied phrases with
distinct semantic roles. However, verb forms including more than two occurrences
of the applicative marker, or applicative constructions including more than two
applied phrases, are not allowed.
– There is no difference between applicative constructions and constructions of
underived verbs involving the same number of object NPs.
240 Denis Creissels

– Tswana applicatives are obligatory applicatives whose use is not conditioned by

limited access of obliques to some operations, but simply by the impossibility to
express some semantic roles otherwise than via applicativization.

Semantics
– The applicative marker is a semantically under-specified marker available to
license a wide variety of semantic roles.
– The only possible generalization about the semantic roles expressed by applied
phrases is that they refer to participants or circumstances that cannot be encoded
as objects or as complements of prepositions: beneficiaries, cause, purpose, con-
tainers playing an essential role in the event denoted by the verb, etc. Applied
phrases do not necessarily refer to peripheral participants or circumstances of the
event: applied phrases referring to essential participants (i.e., participants implied
by the lexical meaning of the verb) are also widely attested; for example, with some
motion verbs, destination of motion can only be expressed via applicativization.
– Since Tswana applicatives are obligatory applicatives, Tswana applicative con-
structions have no pragmatic or discursive implication.

Lookalikes and others

– Non-applicative functions of the applicative marker include a passive-like use in
which instrumental adjuncts are promoted to subject role, the focalization of loca-
tives expressing the location of the event, the expression of habituality of action at
some place, and the expression of intensity of action.
– Lexicalized applicatives constitute a non-negligible proportion of the verbal lexicon
of Tswana.

Abbreviations
appl applicative
aux auxiliary
caus causative
clX class (agreement pattern) X
dem demonstrative
emph emphatic
foc focalization marker
fut future
fv final vowel (a vowel analyzable as the inflectional ending of verbs which however is not
necessarily analyzable as carrying a particular tam value by itself)
gen genitive
hab habitual
inf infinitive
 8 Applicative constructions in Tswana 241

lk linker
loc locative
neg negation marker
oI object index
pass passive
pl plural
prf perfect
sg singular
sI subject index

Numbers between parentheses immediately after the gloss of noun forms refer to the agreement pattern
(“class”) triggered by the noun form.

References
Cole, Desmond T. 1955. An introduction to Tswana grammar. Cape Town: Longman.
Creissels, Denis. 2002. Valence verbale et voix en tswana. Bulletin de la Société de Linguistique de Paris 97(1).
371–426.
Creissels, Denis. 2003. Présentation du tswana. Lalies 23. 5–128.
Creissels, Denis. 2006. Tswana verb morphology and the Lexical Integrity Principle. Lingue e linguaggio 5(1).
49–66.
Creissels, Denis. 2011. Tswana locatives and their status in the inversion construction. Africana Linguistica 17.
33–52.
Creissels, Denis. 2013. Les prépositions simples du tswana. In Jesse Tseng (ed.) Prépositions et postpositions:
approches typologiques et formelles, 17–56. Paris: Lavoisier.
Creissels, Denis. 2017. The conjoint/disjoint distinction in the tonal morphology of Tswana. In Jenneke van
der Wal & Larry M. Hyman (eds.), The conjoint/disjoint alternation in Bantu, 200–238. Berlin: De Gruyter
Mouton.
Creissels, Denis, Anderson M. Chebanne & Heather W. Nkhwa. 1997. Tonal morphology of the Setswana verb.
Munich: LINCOM.
Creissels, Denis & Sylvie Voisin. This volume. B-applicatives and I-applicatives in Atlantic languages
(Niger-Congo).
Pacchiarotti, Sara. 2020. Bantu applicative constructions. Stanford: CSLI Publications.
Van de Velde, Mark. 2008. A grammar of Eton. Berlin: Mouton de Gruyter.
Van de Velde, Mark. 2020. Concernee-Concern constructions: A comparative study of external possession in
the Bantu languages. Studies in Language 44(1). 70–94.
Mengistu Amberber
9 Applicativization in Amharic
Abstract: This chapter provides a detailed description of the applicative construction
and of a related non-applicative construction involving the same verbal marking in
Amharic (Ethiosemitic, Ethiopia). The applicative suffixes -ll- and -bb-—benefactive
and malefactive, respectively—always co-occur with object agreement suffixes which
cross-reference the applied phrase. These two suffixes can be used with virtually any
verb irrespective of transitivity or the lexical semantics of the verb root, presumably
because any event can be potentially cast as benefiting or harming some entity. The
same suffixes can also occur in a non-applicative construction, where the oblique
phrase is marked with a preposition but cross-referenced on the verb as if it were an
object.

1 Introduction
This chapter provides a detailed description of the applicative construction and of a
related but non-applicative construction involving the same verbal marking in Amharic
(Ethiosemitic, Ethiopia). Amharic, self-name [amarɨɲɲa] (አማርኛ), belongs to the Trans-
versal South Ethiosemitic language group (see Figure 1) with Harari, the East Gurage
languages and Argobba (Hetzron 1972: 119; Faber 1997: 11–13; Rubin 2008: 80).1 It is
spoken by approximately 31.8 million people as mother tongue in Ethiopia, and by
approximately 25.1 million people as a second language (Eberhard, Simons, and Fennig
2021).2 It is the working language of the federal government of Ethiopia. There are five
regional dialects of Modern Amharic, namely, that of Addis Ababa, Gonder, Gojjam,
Menz, and Wello (Meyer 2011; see also Zelealem Leyew 2007).

1 Figure 1 is identical to Figure 5 in Rubin (2008: 92), which is based on Hetzron (1972). The numbers in
parenthesis indicate the number of languages within the subgroup. According to the traditional classifi-
cation of Ethiosemitic, there are 23 languages. Ge‘ez is no longer spoken and survives only as a liturgical
language of the Orthodox Christian churches of Ethiopia and Eritrea (Weninger 2011b: 1125). Argobba is
severely endangered (Eberhard, Simons, and Fennig 2021) and Gafat is believed to be extinct (Weninger
2011a: 1114).
2 According to Eberhard, Simons, and Fennig (2021), 14.8 million speakers are monolingual.

Acknowledgments: I am very grateful to Fernando Zúñiga and Denis Creissels for inviting me to be part of
this project. I thank them for their guidance and meticulous attention to detail in shaping the final version of
this chapter. Many thanks also to an anonymous external reviewer who provided me with detailed feedback
that improved the quality of the contribution. All remaining errors are of course my own.

https://doi.org/10.1515/9783110730951-009
244 Mengistu Amberber

Figure 1: The Ethiosemitic language family.

The data used in this chapter is based on different sources including the author’s native
speaker knowledge, examples from published sources as well as corpora of naturally
occurring discourse.
The chapter is organised as follows. In § 2 we provide a brief background on the
morphosyntax of Amharic, followed by a discussion of the morphology (§ 3), syntax
(§ 4), semantics (§ 5) of the applicative construction and of the non-applicative construc-
tion that involves the same verbal marking as the applicative construction. In § 6 we
conclude with a summary of the main findings.

2 Basics of Amharic morphosyntax

2.1 Morphology

As a Semitic language, Amharic employs the word formation strategy known as root-
and-pattern morphology particularly prevalent in the domain of verb morphology.
The core semantic content of a word is signalled by the consonants, also known as
‘root radicals’, whereas grammatical meaning is encoded by the consonant-vowel (CV)
template.
There are six major templates of the verb: perfective, imperfective, jussive, imper-
ative, converb/gerund, and verbal noun. The most common type of roots has three
consonants (triradicals). Traditionally, triradical verbs are classified into three conju-
gational classes known as Type A, Type B, and Type C. This classification is based on
whether or not the penultimate radical of the root is geminated. In Type B verbs the
radical is geminated in all templates. In Type A verbs, it is geminated in the perfec-
 9 Applicativization in Amharic 245

tive template only. In Type C verbs it is geminated in the perfective and imperfective
templates. This can be seen in Table 1 for the perfective, imperfective, jussive, and
converb/gerund templates using the verbs səbbər- ‘break’, fəlləg- ‘search/want’ and
marrək- ‘cause to surrender’.

Table 1: Verb templates.

Type Perfective Imperfective Jussive Converb

A C1əCC2əC3 C1əC2C3 C1C2əC3 C1əC2C3ə

səbbər- -səbr- -sbər səbrə-
B C1əCC2əC3 C1əCC2ɨC3 C1əCC2ɨC3 C1əCC2ɨC3ə
fəlləg- -fəllɨg -fəllɨg- fəllɨgə-
C C1aCC2əC3 C1aCC2ɨC3 C1aC2C3 C1aC2C3ə
marrək- -marrɨk- -mark markə-

Nouns can be marked for definiteness, gender, number, and case. The definite is
marked by the suffix -u/-w (masculine) -wa (feminine) but the indefinite is unmarked.
The plural is marked by the suffix -ottʃ, but the singular/transnumeral is unmarked.
There is differential object marking (DOM) whereby the accusative is marked (by the
suffix -n) typically if the NP is definite/specific (see Amberber 2009). The nominative is
unmarked.
Verbs are marked obligatorily for subject agreement but often optionally for object/
indirect object agreement. In the perfective and converb conjugations, subject agree-
ment consists of suffixes only, whereas in the imperfective and jussive a combination of
prefixes and suffixes is used. Intransitive verbs take S NPs, transitive verbs take A and O
NPs, and extended transitive verbs take A, O, and IO NPs. The basic temporal distinction
is between past (perfect) and non-past (imperfect).
As the applicative construction requires object marking on the verb, it is important
to be familiar with the full range of object marking (see also Haile 1970). Table 2 pre-
sents the object agreement markers in Amharic using the verb form səbbərə (perfective)
jɨsəbr- (imperfective), ‘break’, where the subject is the third person masculine singular,
but the object expresses different persons, numbers and genders (e.g., səbbərə-w ‘he
broke him/it’, səbbərə-h ‘he broke you’, etc.):3

3 This is not the full set of object pronominal suffixes. Thus, when the verb takes the first person singu-
lar subject marker (-hu), the 3rd person masculine object suffix is -t (as in səbbər-hu-t ‘I broke it’).
246 Mengistu Amberber

Table 2: Object agreement markers (with a

3sg.m subject).

Perfective Imperfective
3sg.m səbbərə-w jɨsəbr-əw
3sg.f səbbər(✶ə)4-at jɨsəbr-at
2sg.m səbbərə-h jɨsəbr-ɨh
2sg.f səbbərə-ʃ jɨsəbr-ɨʃ
1sg səbbərə-ɲɲ jɨsəbr-əɲɲ
3pl səbbər(✶ə)-atʧəw jɨsəbr-atʧəw
2pl səbbər(✶ə)-atʧhʷ jɨsəbr-atʧuh
1pl səbbərə-n jɨsəbr-ən

In the above paradigm, the subject is the 3rd person masculine singular which is realised
by the suffix -ə in the perfective stem (səbbər-ə) and by the prefix jɨ- in the imperfective
stem (jɨ-səbr).

2.2 Basic syntax

Amharic is a nominative-accusative language, that is to say that the subject of an intran-

sitive clause and the subject of a transitive clause are in the nominative distinct from
the object of a transitive clause which is in the accusative. As already pointed out, the
nominative is not morphologically marked, whereas the accusative is marked by the
suffix -n (-ɨn after consonant final stems) typically only if the noun is definite or specif-
ic.5 Consider the following examples:

(1) a. lɨdʒ-u wədə bet hed-ə

boy-def.m6 toward home go\pfv-sbj.3sg.m
‘The boy went home.’
b. aster lɨdʒ-u-n ajjə-tʧ-(ɨw)
Aster boy-def.m-acc see\pfv-sbj.3sg.f-obj.3sg.m
‘Aster saw the boy.’

4 When a vowel-initial suffix attaches to a vowel-final stem, the vowel of the latter is deleted in accord-
ance with the vowel hiatus resolution rule of Amharic.
5 By the terms “definite” and “specific” we include personal pronouns, demonstratives, noun phrases
with a possessive modifier, the possessive suffix, and quantifiers.
6 The stem glossed as ‘boy’ here actually means ‘child’, and it is the definite suffix -u that leads to the
meaning ‘the boy’ (i.e., a more accurate gloss would be: child-def.m).
 9 Applicativization in Amharic 247

In (1a), the subject NP (‘the boy’) is unmarked for the nominative. The object NP in (1b)
is marked for the accusative case by the suffix -n. Notice also that in (1b) the verb can
optionally take object agreement marking. If we replace the definite NP in (1b) with an
indefinite one, not only will the accusative case marking be unacceptable, but the verb
cannot be marked for object agreement, as shown in (2):

(2) aster (and) lɨdʒ(✶-ɨn) ajjə-tʧ(✶-ɨw)

Aster one boy-acc see\pfv-sbj.3sg.f-obj.3sg.m
‘Aster saw a/one boy.’

In addition to the direct object as shown in (2), Amharic also has what might be called
“indirect” and/or “oblique” objects, typically marked by adpositions:

(3) aster lə-lɨdʒ-u məs’haf sət’t’ə-tʧ(-ɨw)

Aster to-boy-def.m book give\pfv-sbj.3sg.f-obj.3sg.m
‘Aster gave a book to the boy.’

While the unmarked constituent order in Amharic is clearly [S/A O V], it is possible to
depart from this through topicalization and focus. Furthermore, overt NPs are often
omitted once introduced in discourse, and this is particularly true with the subject NP
as the verb is obligatorily marked for subject agreement.
Amharic has adpositions which include both prepositions (which, in most cases
are formally prefixes) and postpositions. The presence of prepositions is typologically
unusual as the language is otherwise strictly head-final. The postpositions are often
independent words which are historically derived from nominals. There are about 8
to 10 prepositions and about the same number of postpositions. As we will see shortly,
the prepositions bə- and lə-7 are particularly relevant to the discussion of the applica-
tive construction. The main adpositional meanings encoded by these prefixes are listed
below.8
The prefix bə- can be used to encode a range of different meanings including Instru-
mental, Locational, Time, and Malefactive:

(4) a. Instrumental: bə-billawa ‘by a knife’

b. Locational: bə-gəbəya ‘at a market’
c. Time (starting/finishing point of an event): bə-hulət səʔat ‘at 2 o’clock’

The prefix lə- encodes three core meanings: Beneficiary/Goal, Reason, and Time:

7 With allomorphs b- and l- respectively before vowel-initial stems.

8 The other common prepositions with multiple meanings are: kə- ‘from, out of, at, in, on, to, by, with’,
wədə- ‘toward, into, to, about’, ɨskə- ‘as far as, until, to’, sɨlə- ‘about, because of’ and ɨndə- ‘like, as’. Most
of these prepositions are traditionally (Leslau 1995: 597–616) analysed as prefixes.
248 Mengistu Amberber

(5) a. Beneficiary/Goal
lə-lɨdʒ-u sət’t’ə-hu-t
to-boy-def.m give\pfv-sbj.1sg-obj.3sg.m
‘I gave it to the boy.
b. Reason
wətadər-ottʃ-u l-agər-attʃəw
soldier-pl-def for-country-poss.p
jɨ-mot-all-u
sbj.3p-die\ipfv-aux.npst-sbj.3pl
‘The soldiers (will) die for the sake of their country.’
c. Time
lə-mɨsa asa bəll-attʃ
for-lunch fish eat\pfv-sbj.3sg.f
‘She ate fish for lunch.’

In terms of clausal syntax more broadly, there are three types of subordinate clauses:
complement, relative, and adverbial (see also Demeke 2003). Complement clauses are
typically found with verbs of perception and attention (e.g., ajjə ‘see’, səmma ‘hear’),
cognition (ammənə ‘believe’, awwəkʼə ‘know’), desire (fəlləgə ‘want’, təməɲɲə ‘wish’),
speaking (nəggərə ‘tell’, t’əjjək’ə ‘ask’). Such clauses are marked by the complementizer
ɨnd(ə)- which is attached to the verb of the complement clause as shown below:

(6) [aster [ləmma ɨndə-mət’tʼa] səmma-ttʃ]

Aster Lemma comp-arrive\pfv.sbj.3sg.m hear\pfv-sbj.3sg.f
‘Aster heard that Lemma has arrived.’

Another productive complementizer is bɨl- (equivalent to the English complementizer

‘that’), which is historically derived from the converb form of the verb ‘to say’ (alə in
the perfective).9 This complementizer is inflected for person, number, and gender and
is often found with verbs of cognition (assəbə ‘think’) and speech (nəggərə ‘tell’). It is
commonly used to express direct and indirect speech/thought.10

(7) a. [aster [pro wədǝ-bet ɨ-hed-allə-hʷ bɨl-a]

Aster to-house sbj.1sg-go\ipfv-aux.npst-sbj.1sg say\cnv-sbj.3sg.f
assəb-ətʧ]
think\pfv-sbj.3sg.f
‘Aster thought that she would go home.’
(Lit. ‘Aster thought, saying: “I will go home.”’)

9 The grammaticalization of a quotative verb ‘to say’ into a complementizer is cross-linguistically well
known (see Kuteva et al. 2019: 357–358).
10 Null arguments are marked as pro here for expository purposes.
 9 Applicativization in Amharic 249

b. [pro [pro wədǝ-bet ɨ-hed-allə-hʷ bɨl-a]

to-house sbj.1sg-go\ipfv-aux.npst-sbj.1sg say\cnv-sbj.3sg.f
assəb-ətʧ]
think\pfv-sbj.3sg.f
‘She thought that she would go home.’
(Lit. ‘She thought, saying: “I will go home”.’)

Adverbial clauses can be formed by attaching various conjunctions to the verb: sɨlə
‘because’, sɨ ‘when’, ɨndə ‘as soon as’. Here is one example:

(8) [pro [pro sɨlə-səddəb-ə-ɲɲ] mətta-hu-t]

because-insult\pfv-sbj.3sg.m-obj.1sg hit\pfv-sbj.1sg-obj.3sg.m
‘Because he insulted me, I hit him.’

The relative clause involves the use of the markers jə- in the perfective and jəmm- in
the imperfective. The relative clause verb is marked for agreement with the head noun,
which it precedes. Consider the following:

(9) a. lɨdʒ-u tɨlantɨnna mət’tʼa

boy-def.m yesterday come\pfv.sbj.3sg.m
‘The boy came yesterday.’
b. lɨdʒ-u-n ajjə-hu-t
boy-def.m-acc see\pfv-sbj.1sg-obj.3sg.m
‘I saw the boy.’

Now consider (10), which is based on the combination of the above two clauses:

(10) tɨlantɨnna jə-mət’tʼa-w-ɨn lɨdʒ ajjə-hu-t

yesterday rel-come\pfv.sbj.3sg.m-def.m-acc boy see\pfv-sbj.1sg-obj.3sg.m
‘I saw the boy who came yesterday.’

Notice that the relativised verb has nominal properties in that it is marked for definite-
ness and accusative case. The head noun modified by the relative clause must occur
without definiteness and case marking. We will examine the interaction of the relative
clause with the applicative construction in § 4.4.

2.3 Valency and voice operations

There are productive valency-decreasing and valency-increasing derivations. The

valency-decreasing derivations include the passive, anticausative, reflexive and recip-
rocal. These involve the use of the prefix tə-, which attaches to the transitive verb. Due
250 Mengistu Amberber

to this multifunctionality, the term medio-passive may be more appropriate (than terms
such as “reflexive/passive”) to refer to the prefix tə- (see also Meyer 2011). Here are
some examples:

(11) a. aster dabbo-w-ɨn kʼorrətʼ-ətʧ

Aster bread-def.m-acc cut\pfv-sbj.3sg.f
‘Aster cut the bread.’
b. dabbo-w (bə-aster) tə-kʼorrətʼ-ə
bread-def.m by-Aster mp-cut\pfv-sbj.3sg.m
‘The bread was cut (by Aster).’

The term anticausative refers here to an intransitive verb that is morphosyntacti-

cally distinct from the passive and reflexive. Thus, the anticausatives tə-səbbərə ‘break
(intr)’, tə-kəffətə ‘open (intr)’, and tə-bəttənə ‘scatter (intr)’ correspond to the transitive
verbs səbbərə ‘break’, kəffətə ‘open’, and bəttənə ‘scatter’, respectively.
While the anticausative, the passive, and the reflexive are distinct types of valen-
cy-changing alternation, they are marked by the same form. A verb so marked can be
ambiguous between the anticausative, the passive, and the reflexive reading though
one can force a passive reading by expressing the Agent in a by-phrase (Demoz 1964:
13–27; Amberber 2000, Amberber 2002: 14–25). When the prefix tə- is attached to the
perfective verb səkkʼələ ‘he hanged’, the reflexive form tə-səkkʼələ ‘he hanged himself’
is derived, which can be ambiguous between the reflexive meaning ‘he hanged
himself’ or the passive reading ‘he was hanged’. The reflexive reading is not univer-
sally available for all transitive verbs. Thus, the verb kʼorrətʼə ‘he cut’ has the tə-stem
tə-kʼorrətʼə ‘it was cut’, but the reading is passive or anticausative rather than reflex-
ive. On the other hand, while verbs that express activities related to parts of the body,
so-called ‘self-grooming’ verbs, are typically reflexive, e.g., tat’tʼəbə (< ✶tǝ-at’t’ǝbǝ) ‘he
washed himself’, tə-latʧʼə ‘he shaved himself’, it is also possible to interpret them as
passive.
Reciprocity is also expressed by the prefix tə- but with a special reduplicative stem
of the verb. For a triradical verb, the reduplicative stem is tə-C1əC2aCC2əC3, e.g., nəkkəsə
‘bite’ → tə-nəkakkəs-u (mp-bite.recp\pfv-3pl) ‘they bit each other’.
The valency-increasing derivations are the causative and applicative. The causative
is formed by attaching the prefix a- (direct causative) or as- (indirect causative).

(12) a. kʼɨbe-w kʼəllətʼ-ə

butter-def.m melt\pfv-sbj.3sg.m
‘The butter melted.’
b. sǝwɨjjə-w kʼɨbe-w-ɨn a-kʼǝllətʼ-ə-w
man-def.m butter-def.m-acc csd-melt\pfv-sbj.3sg.m-obj.3sg.m
‘The man melted the butter.’
 9 Applicativization in Amharic 251

Some intransitive verbs such as tʃʼəffərə ‘dance’, zəmmərə ‘sing’, təɲɲa ‘sleep’, azzənə
‘be sad’, fərra ‘be afraid’, which can be characterized as unergative (as opposed to unac-
cusative), cannot take the prefix a-. Such intransitive verbs take the other causative
prefix, namely as- marking the indirect causative, e.g., tʃʼəffərə ‘dance’ → as-tʃʼəffərə
‘make someone dance’ (not ✶a-tʃʼəffərə).
The causative of transitive verbs is formed by attaching the prefix as- (indirect
causative). Here is an example:

(13) a. almaz dabbo kʼorrətʼ-əttʃ

Almaz bread cut\pfv-sbj.3sg.f
‘Almaz cut some bread.’
b. aster almaz-ɨn dabbo as-kʼorrətʼ-əttʃ-at
Aster Almaz-acc bread csi-cut\pfv-sbj.3sg.f-obj.3sg.f
‘Aster made Almaz cut some bread.’

Note that the causee argument in (13b), ‘Almaz’, is marked by the accusative case suffix
and is cross-referenced by the object agreement marker. If the initial object is defi-
nite/specific (dabbo-w ‘the bread’), it will be marked by the accusative case suffix (-n).
However, the object agreement marker still cross-references the causee—not the initial
object, as can be seen below.11

(14) aster almaz-ɨn dabbo-w-ɨn as-kʼorrətʼ-əttʃ-at (✶-ɨw)

Aster Almaz-acc bread-def.m-acc csi-cut\pfv-sbj.3sg.f-obj.3sg.f obj.3sg.m
‘Aster made Almaz cut the bread.’

The other valency changing derivation is the applicative, which is the focus of this
chapter. It should be pointed out here that the term ‘applicative’ is not commonly used
in the grammatical description of Amharic or other Ethiopian languages. As far as we
are aware, the term was used for the first time in Amberber (1996). This chapter will
show that based on the definition provided by Zúñiga and Creissels (this volume), it is
possible to identify an applicative construction in Amharic (see also Amberber 2000,
2002: 55–60; Yabe 2007: 76–85). In the following sections of the chapter, the morphology,
syntax, and semantics of the applicative will be discussed in detail.

11 It should be noted here that the occurrence of two objects both marked by the accusative suffix -n as
the example shows is less preferred in actual discourse and there is a tendency for the initial object to
appear as non-definite/non-specific which makes it ineligible for the accusative suffix -n.
252 Mengistu Amberber

3 Morphology
There are two applicative morphemes, -ll- and -bb-, which we refer to as benefactive
and malefactive respectively. The following examples provide a minimal pair contrast
between these two morphemes:

(15) daɲɲa-w səwɨjjə-w-ɨn12 fərrəd-u-ll-ət

judge-def.m man-def.m-acc judge\pfv-sbj.3pol-appl-obj.3sg.m
‘The judge passed judgment in favour of the man.’ (i.e., the man was acquitted)

(16) daɲɲa-w səwɨjjə-w-ɨn fərrəd-u-bb-ət

judge-def.m man-def.m-acc judge\pfv-sbj.3pol-appl-obj.3sg.m
‘The judge passed judgment against the man.’ (i.e., the man was convicted)

The applicative suffixes always co-occur with object agreement suffixes which cross-ref-
erence the applied phrase. Thus, in the above examples, -ll- (benefactive) and -bb- (male-
factive) are followed by the object agreement suffix which cross-references the applied
phrase (‘the man’).
Notice that the applied phrase takes the accusative case marking -n consistent with
the differential object marking phenomenon noted earlier (§ 2.1).
It is important to note here that the suffixes -ll- and -bb- can also occur in a non-ap-
plicative construction where the oblique phrase is marked with a preposition but
cross-referenced as if it were an object. To distinguish this usage of the suffixes from
their applicative function, we use the term “oblique cross-reference” (ocr)13. Thus, con-
sider the following:

(17) daɲɲa-w lə-səwɨjjə-w(✶-ɨn) fərrəd-u-ll-ət

judge-def.m for-man-def.m-acc judge\pfv-sbj.3pol-ocr-obj.3sg.m
‘The judge passed judgment in favour of the man.’ (= ‘the man was acquitted.’)

(18) daɲɲa-w bə-səwɨjjə-w(✶-ɨn) fərrəd-u-bb-ət

judge-def.m on-man-def.m-acc judge\pfv-sbj.3pol-ocr-obj.3sg.m
‘The judge passed judgment against the man.’ (= ‘the man was convicted.’)

Notice the formal similarity between the prepositions lə- ‘for’ and bə- ‘on’, on the one
hand and the applicative/oblique cross-reference markers -ll- and -bb- with identical
consonants in each pair on the other. The proper analysis of this similarity is controver-

12 The form səwɨjjəw consists of the “singulative” morpheme -jjə, which is used with some nouns such as
ethnic names and the generic nouns referring to people, səw ‘man/person’, set ‘woman’ (Meyer 2011: 1192).
13 I thank the editors of this volume for suggesting I use the term “oblique cross-reference” (ocr) here
as distinct from “applicative” (appl).
 9 Applicativization in Amharic 253

sial. As Kramer (2014) pointed out, there are several possible hypotheses, including: that
these markers are part of a complex agreement morpheme (Mullen 1986; Amberber
1996), that they are incorporated prepositions (Yabe 2007), or that they are applicative
heads (Demeke 2003). The arguments for or against these hypotheses are often theo-
ry-internal. For the present purposes, we analyse -ll- and -bb- as applicative markers
when they occur in constructions such as (13) and (14), and will be largely agnostic as to
the proper characterisation of their formal similarity with the prepositions lə- and bə-.
While the term “malefactive” is appropriate for the meaning in (16)/(18), the applica-
tive suffix -bb- can also encode a range of other meanings, including locational and instru-
mental, similar to what is found in the Gumer language (Völlmin 2010).
There are no serial verbs in Amharic, but there are converbs which are used in
clause chaining constructions, as already mentioned above. Converbs can take the
applicative morphemes as well as the full range of derivational and inflectional suffixes
except for tense/aspect inflection. Thus, consider the following example:

(19) ɨndʒəra lə-lɨdʒ-e gaggər-hu

injera for-child-1.poss bake\pfv-sbj.1sg
‘I baked injera for my child.’

In (19) the Beneficiary of the event (of baking) is expressed with the preposition lə- ‘for’.
In (20) below, the event of ‘baking’ occurs as a sub-event in a clause chain that involves
the use of the converb:

(20) ɨndʒəra lə-lɨdʒ-e gagɨrr-e ɨ-hed-allə-hu

injera for-child-1.poss bake\cnv-sbj.1sg sbj.1sg-go\ipfv-aux.npst-sbj.1sg
‘Having baked injera for my child, I (will) leave.’

In (21) below, the converb occurs with the oblique cross-reference suffix cross-referenc-
ing the Beneficiary argument ‘my child’:

(21) ɨndʒəra lə-lɨdʒ-e gagɨrr-e-ll-ət

injera for-child-1.poss bake\cnv-sbj.1sg-ocr-obj.3sg.m
ɨ-hed-allə-hu
sbj.1sg-go\ipfv-aux.npst-sbj.1sg
‘Having baked injera for my child, I (will) leave.’

The converb is productively used to represent a series of events in a clause chain (see
Meyer 2012; Amberber, in preparation). Here is another example:

(22) məsob kəft-o dabbo wəssəd-ə

basket open\cnv-sbj.3sg.m bread take\pfv-sbj.3sg.m
‘Having uncovered the basket, he took bread.’
254 Mengistu Amberber

The converb kəft-o ‘having opened’ is used here as a subordinate clause. A character-
istic property of the converb in Amharic is that it cannot be marked for tense/aspect
and thus it is arguably not fully verbal. Now consider (23) where the converb takes the
applicative suffix:

(23) məsob kəft-o-ll-ət dabbo wəssəd-ə

basket open\cnv-sbj.3sg.m-appl-obj.3sg.m bread take\pfv-sbj.3sg.m
‘Having uncovered the basket for him, he took bread.’

Unlike in (22), in (23) the Agent that opens the basket is not the same as the Agent that
takes the bread. Someone X opened the basket for the benefit of someone Y where the
latter took the bread. To make the distinction clearer, we can use overt NPs as follows:

(24) ləmma məsob kəft-o dabbo wəssəd-ə

Lemma basket open\cnv-sbj.3sg.m bread take\pfv-sbj.3sg.m
‘Lemma, having uncovered the basket, he took (some) bread.’

(25) ləmma məsob kəft-o-ll-ət ali dabbo

Lemma basket open\cnv-sbj.3sg.m-appl-obj.3sg.m Ali bread
wəssəd-ə
take\pfv-sbj.3sg.m
‘Lemma having uncovered the basket for him, Ali took (some) bread.’

Interestingly, the applicative / oblique cross-reference suffix is also used to form modal
constructions of necessity and obligation. In such cases, the composite of the verb allə
‘to exist’ and the suffix -bb- and object suffixes are used as shown below:

(26) mədan kə-fəlləg-ə hakim_bet məhed

cure\vn if-want\pfv-sbj.3sg.m hospital go\vn
allə-bb-ət
exist\pfv.sbj.3sg.m-appl-obj.3sg.m
‘If he wants to be cured/be well, he must go to the hospital.’

This grammaticalized use of the appl/ocr morphemes with the verb allə ‘to exist’ will
not be examined further in this chapter (but see Meyer 2012; see also Ahland 2009 for
the grammaticalization of the possessive construction).
We will now investigate the syntactic structure of the constructions involving the
appl/ocr markers in detail.
 9 Applicativization in Amharic 255

4 Syntax
4.1 The syntactic behaviour of applied phrases
and cross-referenced obliques

Before examining the syntactic status of the applied phrase or cross-referenced oblique,
let us first look at the base construction (BC). Consider the following example:

(27) aster bə-bɨrtʃ’ɨk’k’o wətət t’ət’t’a-ttʃ

Aster with-glass milk drink\pfv-sbj.3sg.f
‘Aster drank milk from a glass.’14

The verb t’ət’t’a ‘to drink’ takes two arguments – the Agent (‘Aster’) and the Theme
(‘milk’). The phrase ‘with a glass’ is clearly an adjunct as it is not required by the argu-
ment structure of the verb. Now, consider the applicative construction below:

(28) aster bɨrtʃ’ɨk’k’o-w-ɨn wətət t’ət’t’a-ttʃ-ɨbb-ət

Aster glass-def.m-acc milk drink\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster drank milk from the glass.’

Notice that the applied phrase is definite and is marked by the accusative case. The verb
is marked by the applicative suffix and object agreement that agrees with the applied
phrase. All these formal properties of the construction are obligatory. For example, if
the applied phrase is indefinite the resulting construction will be ungrammatical, as
shown below:

(29) ✶
aster bɨrtʃ’ɨk’k’o wətət t’ət’t’a-ttʃ-ɨbb-ət
Aster glass milk drink\pfv-sbj.3sg.f-appl-obj.3sg.m

The applicative marker and the object agreement suffix are both obligatory. If they are
missing, the resulting construction will be ungrammatical:

(30) ✶
aster bɨrtʃ’ɨk’k’o-w-ɨn wətət t’ət’t’a-ttʃ
Aster glass-def.m-acc milk drink\pfv-sbj.3sg.f

Thus, in Amharic the construction illustrated by example (28) satisfies the definition
provided in Zúñiga and Creissels (this volume) regarding the AC and its relationship
with the BC.

14 While the Amharic preposition bə- means ‘with’, ‘by’, here it is translated into idiomatic English as
‘from’. The preposition kə- is equivalent to ‘from’ but cannot be used in this context.
256 Mengistu Amberber

(31) (i) The predicates in both constructions are built upon the same root, but the
one in the AC bears additional overt marking that distinguishes it from the
one in the BC.
(ii) The participant encoded as S or A in the BC appears as S or A in the AC.
(iii) The AC includes a noun phrase in a role other than S or A, the applied phrase
(AppP), which refers to a participant that either requires a non-core coding
different from its coding in the AC or cannot be expressed at all in the BC.

The applicative so defined can be derived from virtually any verb irrespective of tran-
sitivity. In the following examples, we see the (benefactive) applicative derived from
intransitive and transitive verbs:

(32) a. aster rot’-əttʃ

Aster run\pfv-sbj.3sg.f
‘Aster ran.’
b. aster rot’-əttʃ-ɨll-ət
Aster run\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster ran for him.’

(33) a. aster dabbo k’orrət’-əttʃ

Aster bread cut\pfv-sbj.3sg.f
‘Aster cut some bread.’
b. aster dabbo k’orrət’-əttʃ-ɨll-ət
Aster bread cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster cut some bread for him.’

Now consider the following example with the ditransitive verb ‘to give’:

(34) a. aster lə-lɨdʒ-u məs’haf sət’t’ə-ttʃ(-ɨw)

Aster to-boy-def.m book give\pfv-3sg.f-obj.3sg.m
‘Aster gave a book to the boy.’
b. aster lə-lɨdʒ-u məs’haf sət’t’ə-ttʃ-ɨll-ət
Aster to-boy-def.m book give\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster gave a book (to someone) for the benefit of the boy.’

The verb ‘to give’ takes three arguments, and in (34a) we see the relevant arguments
occurring as subject (the Giver), indirect object (the Recipient), and direct object (the
Gift). Optional object agreement in (34a) is with the Recipient. In (34b), on the other
hand, where the verb takes the applicative suffix, the Recipient is implicit (someone
 9 Applicativization in Amharic 257

unspecified). The phrase ‘the boy’ is now a non-core argument (Beneficiary) and is
cross-referenced by the appl suffix and object agreement.15
It is very important to keep in mind that in Amharic natural discourse the use of
overt NPs to express arguments is very limited. Most often the subject, but also other
grammatical functions are left implicit. Sometimes this is because these implicit argu-
ments are cross-referenced by agreement morphology on the verb. This can be seen in
the following examples based on (15) and (16) but without the NP arguments:

(35) fərrəd-u-ll-ət
judge\pfv-sbj.3.pol-appl-obj.3sg.m
‘(They) passed judgment in favour of the him.’ (i.e., he was acquitted)

(36) fərrəd-u-bb-ət
judge\pfv-sbj.3.pol-appl-obj.3sg.m
‘(They) passed judgment against him.’ (i.e., he was convicted)

None of the arguments of the verb ‘to judge’ are expressed by overt NPs.
It is important to note that there can only be one applicative and/or oblique cross-ref-
erence marker in a verb (see also Kramer 2014). Consider the following example:

(37) aster lə-ləmma dabbo-w-ɨn bə-billawa-w

Aster for-Lemma bread-def.m-acc with-knife-def.m
k’orrət’-əttʃ-ɨll-(✶ɨbb)-ət
cut\pfv-sbj.3sg.f-ocr-ocr-obj.3sg.m
‘Aster cut the bread with the knife for Lemma.’

Thus, while there are two peripheral arguments which can potentially be encoded as a
cross-referenced oblique or as an applied object, only one of them can do so at a time.
Note also that when there are two peripheral arguments, the Beneficiary and the Instru-
ment, the former should occur first (or higher) in the structure as the contrast in (38a)
and (38b) shows:16

15 I’m grateful to Denis Creissels for pointing out that the lack of ambiguity in (34b), that is, the fact that
the verbal suffix ɨll- can only refer to the Beneficiary, has crucial implications in that the construction
should be regarded as an instance of a D-applicative (“dative” applicative) distinct from a P-applica-
tive’. See Zúñiga and Creissels (this volume) for discussion of these terms, which are based on Creissels
(forthcoming).
16 See McGinnis (2008) for one formal account of the applicative construction, where the author argues
for the structurally higher position of the benefactive applied phrase.
258 Mengistu Amberber

(38) a. ✶
aster bə-billawa-w lə-ləmma dabbo-w-ɨn
Aster with-knife-def.m for-Lemma bread-def.m-acc
k’orrət’-əttʃ-ɨll-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
b. aster lə-ləmma bə-billawa-w dabbo-w-ɨn
Aster for-Lemma with-knife-def.m bread-def.m-acc
k’orrət’-əttʃ-ɨll-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut the bread with the knife for Lemma.’

Now, if we replace the Beneficiary in (37) with the Instrument and use -bb- to refer to
the instrumental phrase, the resulting construction is ungrammatical:

(39) ✶
aster lə-ləmma dabbo-w-ɨn bə-billawa-w
Aster for-Lemma bread-def.m-acc with-knife-def.m
k’orrət’-əttʃ-ɨbb-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m

The instrumental phrase should precede the Beneficiary if the oblique cross-reference
suffix -bb- refers to the resulting structure. Thus, (40) is better than (39):

(40) ?aster bə-billawa-w lə-ləmma dabbo-w-ɨn

Aster with-knife-def.m for-Lemma bread-def.m-acc
k’orrət’-əttʃ-ɨbb-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut the bread with the knife for Lemma.’
(Lit. ‘Aster, with the knife, she cut the bread with it for Lemma.’)

This sentence, while not ungrammatical, sounds awkward. In the more felicitous
version of the sentence, the instrumental phrase occurs without the Beneficiary:

(41) aster bə-billawa-w dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət

Aster with-knife-def.m bread-def.m-acc cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut the bread with the knife.’
(Lit. ‘Aster, with the knife, she cut the bread with it’)

This seems to be due to information structure, where the appl/ocr suffix is partially
used to focus on a constituent and to achieve maximum prominence, the instrumental
phrase is not only fronted but also, ideally, occurs without another peripheral phrase
(see Zúñiga and Creissels, this volume, for discussion on fronted and focused constitu-
ents in Western Mayan languages). In fact, we can also find the cross-referenced oblique
at a clause-initial position displacing the subject:
 9 Applicativization in Amharic 259

(42) bə-billawa-w aster dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət

with-knife-def.m Aster bread-def.m-acc cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut the bread with the knife.’
(Lit. ‘With the knife, Aster cut the bread with it.’)

Note that if the speaker wants to focus on any of the non-subject arguments, they can
omit the subject and place the focussed argument in clause-initial position as can be
seen below:

(43) bə-billawa-w dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət

with-knife-def.m bread-def.m-acc cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘She cut the bread with the knife.’
(Lit. ‘With the knife, she cut the bread with it.’)

The motivation for focus is also implicated in the interaction between the relative clause
and the applicative construction, as we will see in § 4.4.

4.2 On the stacking/combination of voice operations

in the applicative construction

Applicativisation is allowed in the context of valency-encoding derivations including

passivisation, anticausativisation, causativisation, and reflexives/reciprocals. For the
sake of space, we discuss two such derivations: passivisation and causativisation.

4.2.1 Passivisation

In the following examples, (44a) shows the simple passive whereas (44b) and (44c) show
the applicativised passive with the malefactive and benefactive suffixes respectively:

(44) a. dabbo-w (bə-aster) tə-kʼorrətʼ-ə

bread-def.m by-Aster mp-cut\pfv-sbj.3sg.m
‘The bread was cut (by Aster).’
b. dabbo-w tə-kʼorrətʼ-ə-bb-at
bread-def.m mp-cut\pfv-sbj.3sg.m-appl-obj.3sg.f
‘The bread was cut to her detriment (by someone).’
(or ‘it was cut accidentally by her’)
c. dabbo-w tə-kʼorrətʼ-ə-ll-at
bread-def.m mp-cut\pfv-sbj.3sg.m-appl-obj.3sg.f
‘The bread was cut for her benefit (by someone).’
(or ‘it was possible for her to cut it’)
260 Mengistu Amberber

In (44b) the malefactive applicative can have a reading whereby the bread was cut by
accident and as such it is not necessarily the case that a different Agent is involved.
Thus, Aster didn’t mean to cut the bread, but she did so accidentally. Likewise, the ben-
efactive use doesn’t necessarily implicate the involvement of a different Agent who
cut the bread for the benefit of someone. It could be the case that the thing to be cut
(‘the bread’ in the example) was particularly hard or difficult and the Agent eventually
managed to cut it.

4.2.2 Causativisation

In (45a) we see the simple causative and in (45b) and (45c) we show the applicativised
causatives:

(45) a. aster dabbo-w-ɨn as-kʼorrətʼ-ətʧ

Aster bread-def.m-acc csi-cut\ocr-sbj.3sg.f
‘Aster made the bread cut (by someone).’
b. aster dabbo-w-ɨn as-kʼorrətʼ-ətʧ-ɨll-ət
Aster bread-def.m-acc csi-cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster made the bread cut for him (by someone).’
c. aster dabbo-w-ɨn as-kʼorrətʼ-ətʧ-ɨbb-ət
Aster bread-def.m-acc csi-cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster made the bread cut against him (by someone).’

Notice that neither the causee nor the Beneficiary/Maleficiary are expressed by overt
NPs. In (45a), the verb is more appropriately understood as the causative of the passive:
‘the bread was made cut’. The original agent (‘causee’) can be expressed by a preposi-
tional phrase: bə-səw ‘by someone’. In (45b) and (45c) the grammatical features of the
Beneficiary/Maleficiary (person, number, gender) are indicated by the agreement mor-
phology on the verb. These arguments can be overtly expressed as can be seen in (45’)
for the Beneficiary:

(45’) aster dabbo-w-ɨn lə-lɨdʒ-u as-kʼorrətʼ-ətʧ-ɨll-ət

Aster bread-def.m-acc for-child-def.m csi-cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster made the bread cut for the boy (by someone).’
 9 Applicativization in Amharic 261

4.3 On case and agreement in the applicative construction

We have seen that the applicative derivation advances an oblique argument into the
core, placing it as the direct object of the clause. It thus appears that, in their applicative
use, the appl/ocr markers enable a non-core terms to bear the accusative case marker
-n. Consider the following examples, adapted from Hetzron (1970: 309):17

(46) a. səw-ottʃ-u bə-gzijabher j-amn-all-u

men-pl-def in-God sbj.3pl-believe\ipfv-aux-sbj.3pl
‘The people believe in God.’
b. səw-ottʃ-u bə-gzijabher j-amn-u-bb-ət-all
men-pl-def in-God sbj.3pl-believe\ipfv-sbj.3pl-ocr-obj.3m-aux
‘The people believe in God.’
c. səw-ottʃ-u ɨgzijabher-ɨn j-amn-u-bb-ət-all
men-pl-def God-acc sbj.3pl-believe\ipfv-sbj.3pl-appl-obj.3m-aux
‘The people believe in God.’

We assume that the verb ‘to believe’ takes two arguments: the believer (Experiencer)
and the entity believed in (Theme). The Theme occurs as a prepositional phrase, ‘in
God’, in both (46a) and (46b). The only difference between the two is that in the latter
the verb is marked by the oblique cross-reference suffix, -bb- and object agreement -ət .
In (46c), on the other hand, the Theme is marked by the accusative case -n like any other
direct object and -bb- here functions as an applicative marker.
Interestingly, Hetzron (1970: 309) suggests that the construction in (46c) has “a
clearly stylistic euphonic function.” The idea is that (46c) is preferred to (46b) to avoid
too many [b] sounds, “the abundance of [b]’s in bə- and later in -bb- is thus avoided by
replacing the first one with -n” (p. 309). Hetzron (1970: 309–310) further argues that the
construction in (46c) is possible only with certain types of ‘complements’; “the comple-
ment must be an organic, not only incidental, part of the content of the verb” (Hetzron
1970: 309). Thus, consider the following example:

(47) bə-məkina-w mət’t’a-ttʃ-ɨbb-ət

with-car-def.m come\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘She came with the car.’

According to Hetzron (1970: 309), məkinaw ‘the car’ in the prepositional phrase cannot
be marked by the accusative suffix -n because “the presence of an instrument does not

17 The interlinear glosses have been added and the transcription is also modified here to be consistent
with the IPA.
262 Mengistu Amberber

necessarily follow from the content of the verb ‘come’.” Thus, according to Hetzron
(1970), (48) below, which is the applicative version of (45), would be ungrammatical:

(48) məkina-w-ɨn mət’t’a-ttʃ-ɨbb-ət

car-def.m-acc come\pfv-sbj.3sg.f-appl-obj.3sg.m
‘She came with the car.’

While it is true that not all types of non-core terms can be advanced by the applicative
derivation in all contexts, I have found no evidence that (48) is ungrammatical. Perhaps
there is more tolerance for constructions such as (48) than was possible at the time of
Hetzron’s (1970), a topic for a detailed diachronic investigation.
So far in our examination of the constructions involving an appl/ocr marker, we
have shown the conditions under which the suffixes, -ll-, -bb- and the following agree-
ment suffixes, are obligatory on the verb. Thus, consider the following:

(49) aster bə-billawa-w dabbo k’orrət’-əttʃ(-ɨbb-ət)

Aster with-knife-def.m bread cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut some bread with the knife.’

(50) aster billawa-w-ɨn dabbo k’orrət’-əttʃ✶(-ɨbb-ət)

Aster knife-def.m-acc bread cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster cut some bread with the knife.’

In (49) we observe that when the Instrument occurs in a prepositional phrase headed
by bə- ‘with’, the oblique cross-reference marker is optional. On the other hand, in (50)
we see that when the Instrument is marked by the accusative case and occurs without
the preposition bə-, the applicative suffix is required.

4.4 The constructions involving an appl/ocr marker

in the context of relativisation

Another area of the grammar where the applicative suffix is obligatory is relativisation.
Consider once again (50) above. Any of the arguments can be modified by a relative
clause. The interaction between the relative clause and the applicative shows that rel-
ativisation is used to focus on an argument (cf. Schachter 1973). In (51) below we see
that the Theme argument ‘bread’ is modified by the relative clause, whereas in (52) the
Instrument argument ‘the knife’ is modified by the relative clause:
 9 Applicativization in Amharic 263

(51) aster bə-billawa18 jə-k’orrət’-əttʃ✶(-ɨw) dabbo

Aster with-knife rel-cut\pfv-sbj.3sg.f-obj.3sg.m bread
‘The bread which Aster cut with a knife.’

(52) aster dabbo jə-k’orrət’-əttʃ✶(-ɨbb-ət) billawa

Aster bread rel-cut\pfv-sbj.3sg.f-appl-obj.3sg.m knife
‘The knife with which Aster cut some bread.’

The crucial observation is that with the relative clause, the applicative suffix and the
associated object agreement are obligatory. Interestingly, the applicative suffix will
be required even when the associated phrase does not have the preposition bə- ‘with’.
Thus, in (53a) the goal/destination of the going event occurs as the prepositional phrase
with the preposition wədə ‘to’, ‘towards’. Likewise, in (54a) the source/location of the
event occurs as a prepositional phrase with kə- ‘from’. In both (53b) and (54b) the verb
is relativised to modify the noun kətəma ‘town’:

(53) a. lɨdʒ-u wədə-kətəma hed-ə

boy-def.m to-town go\pfv-sbj.3sg.m
‘The boy went to town.’
b. lɨdʒ-u jə-hed-ə-bb-ət kətəma
boy-def.m rel-go\pfv-sbj.3sg.m-appl-obj.3sg.m town
‘The town where the boy went to.’

(54) a. lɨdʒ-u kə-kətəma mət’t’-a

boy-def.m from-town come\pfv-sbj.3sg.m
‘The boy came from (the) town.’
b. lɨdʒ-u jə-mət’t’-a-bb-ət kətəma
boy-def.m rel-come\pfv-sbj.3sg.m-appl-obj.3sg.m town
‘The town from which the boy came.’

Again, the suffix -bb- and the associated object agreement are obligatory on the relativ-
ised verbs. Notice, however, that the suffix -bb- has no formal affinity with the preposi-
tional elements wədə ‘to’, kə – ‘from’.

18 There is a slight preference for the noun in the instrumental phrase to occur as indefinite/non-spe-
cific (‘a knife’ rather than ‘the knife’).
264 Mengistu Amberber

5 Semantics
5.1 The lexical semantics of the verb root

It is important to note that virtually any verb can take the benefactive/malefactive
applicative morphemes. From a broad semantic and pragmatic perspective this is not
surprising, because any event can be cast as either benefiting someone/something or
harming someone/something. Consider for example the event depicted by the verb
zənnəbə ‘to rain’ in (55):

(55) zənnəb-ə
rain\pfv-sbj.3sg.m
‘It rained.’

The event encoded by the verb zənnəbə ‘to rain’ can be cast not as a neutral event but
rather as an event with either positive or negative consequences for someone. While
these consequences can be left unmarked and inferred contextually, it is also possible
to encode them explicitly through the applicative construction as follows:

(56) a. zənnəb-ə-ll-ət
rain\pfv-sbj.3sg.m-appl-obj.3sg.m
‘It rained for him.’ (i.e., to his benefit)
b. zənnəb-ə-bb-ət
rain\pfv-sbj.3sg.m-appl-obj.3sg.m
‘It rained on him.’ (i.e., to his detriment)

There is some evidence to support this semantic motivation for the applicative. As we
saw, the applicative construction normally involves the presence of one of the two
applicative suffixes -bb- or -ll- plus the object suffixes. However, interestingly, in some
cases the verb may occur without the applicative suffix. Consider the following exam-
ples with the verb motə ‘to die’:

(57) a. zəməd mot-ə

relative die/pfv-sbj.3sg.m
‘A relative died.’
b. aster-(ɨn) zəmed mot-ə-bb-at
Aster-acc relative die\pfv-sbj.3sg.m-appl-obj.3sg.f
Lit. ‘A relative (of hers) died on Aster.’
c. aster-(ɨn) zəməd mot-at
Aster-acc relative die\pfv.sbj.3sg.m-obj.3sg.f
Lit. ‘A relative (of hers) died on Aster.’
 9 Applicativization in Amharic 265

In (57a) the verb is used in a typical intransitive frame with one argument. However,
a peripheral argument can be introduced as in (57b) with the applicative suffix -bb-
and object agreement that cross-references the peripheral argument. Interestingly, the
applicative suffix -bb- can be omitted without affecting the meaning as can be seen in
(57c). This is curious because, as we have seen so far, a phrase that expresses a periph-
eral semantic role is cross-referenced by the applicative suffix and the associated object
agreement marker. In (57c) we see that it is possible to use the object agreement suffix
only to derive the same meaning as the applicative marked verb. What might be the
reason for this? One possibility is that when the event in question is construed as obvi-
ously adversative, such as the death of a relative, the requirement that the -bb- suffix
should be present is relaxed. Thus, (55c) may be regarded as a syntactic lookalike of the
applicative (Zúñiga & Creissels, this volume).
Many other verbs behave in a similar way. Consider for example the verb t’əffa ‘be
lost, disappear’ in (58):

(58) a. gənzəb t’əffa

money disappear\pfv.sbj.3sg.m
‘Money is lost.’
b. aster-(ɨn) gənzəb t’əffa-bb-at
Aster-(acc) money disappear\pfv.sbj.3sg.m-appl-obj.3sg.f
‘Aster lost some money.’ (Lit. ‘Some money disappeared against Aster.’)
c. aster-(ɨn) gənzəb t’əffa-t
Aster-acc money disappear\pfv.sbj.3sg.m-obj.3sg.f
‘Aster lost some money.’ (Lit. ‘Some money disappeared Aster.’)

Again in (58c) the construction does not have the applicative suffix, presumably because
the loss of property is construed as obviously adversative and thus obviating the need
to mark the verb with the applicative affix -bb- which would otherwise be required to
mark the malefactive applicative. This semantic motivation can be seen clearly when
we compare the verb sərrək’ə ‘to steal’ with the verb wəssədə ‘to take’ (examples from
Hetzron 1970: 315):

(59) and leba gənzəb sərrək’-ə-w

one/a thief money steal\pfv-sbj.3sg.m-obj.3sg.m
‘One/a thief stole money from him.’

(60) and leba gənzəb wəssəd-ə-bb-ət

one/a thief money take\pfv-sbj.3sg.m-appl-obj.3sg.m
‘One/a thief took money from him.’

Again, the intuition is that with the verb sərrək’ə ‘to steal’ the verb is not marked by the
malefactive applicative since being a victim of a theft is clearly not a good thing. On the
266 Mengistu Amberber

other hand, the verb wəssədə ‘to take’ is neutral as to intent, thus is explicitly marked for
the malefactive. In fact, if we replace the applicative with the simple object agreement
in (60), the result is an ungrammatical sentence:

(61) ✶
and leba gənzəb wəssəd-ə-w
one/a thief money take\pfv-sbj.3sg.m-obj.3sg.m

The sentence improves in grammaticality if we also omit the object marker, but then
the resulting structure simply means ‘the thief took money’ without any reference to
the source of the money:

(62) and leba gənzəb wəssəd-ə

one/a thief money take\pfv-sbj.3sg.m
‘One/a thief took money.’

As Hetzron (1970: 315) insightfully observed, the sentence with the verb ‘to take’ while
pragmatically implying an act of theft, this is not necessarily the case as “[t]he thief
might have taken the money quite ‘unprofessionally’, as a legitimate act by a man who
also happens to be a thief.”
Note that the meaning of the applicative suffix (whether it is clearly benefactive
or malefactive) may depend on real world knowledge. It is not hard to imagine that
even an obviously adversative event such as dying can, in certain contexts, be cast as
a positive act. Thus, we find many examples in religious as well as secular texts where
the verb motə ‘to die’ takes the benefactive applicative. Here is one example (from the
HaBiT database):19

(63) ክርስቶስ ቤተክርስቲያንን ስለወደዳት

kɨristos betəkɨrɨstijan-ɨn sɨlə-wəddəd-at
Christ church-acc because-love\pfv.sbj.3sg.m-obj.3sg.f
ሞተላት።
mot-ə-ll-at
die\pfv-sbj.3sg.m-appl-obj.3sg.f
‘Because Christ loved the church, he died for her [the church].’

The event of dying here is cast as a positive event and makes sense in the context of a
specific doctrine. Similarly, the benefactive interpretation of a soldier dying in the act
of defending their country derives its meaning from a particular patriotic discourse.

19 The Amharic corpus in the HaBiT database is made up of text collected from the internet and has
approximately 26 million words. For details, see Rychlý and Suchomel (2016). Here and elsewhere the
original data is in Amharic orthography (Fidel) and is reproduced here for authenticity. I have provided
the IPA transcription, the interlinear glosses, and the English translation.
 9 Applicativization in Amharic 267

5.2 Semantic roles that can be expressed by applied phrases

or cross-referenced obliques

For the present purposes, following Zúñiga and Creissels (this volume) we distinguish
between three types of semantic roles: central or maximally involved (Agents, Forces,
Themes, Patients), peripheral (Comitative, Instruments, Beneficiaries), and intermedi-
ate (Sources, Goals, Recipients, Experiencers). As Zúñiga and Creissels point out, there
is a cross-linguistic tendency for maximally involved or central roles to be expressed
as core arguments in the syntax, whereas peripheral roles are often expressed as
obliques or adjuncts. Interestingly, there is also “a cross-linguistic tendency for ACs to
work on peripheral roles; Comitatives, Instruments, and Beneficiaries are indeed the
most common roles with applicatives worldwide” (Zúñiga and Creissels, this volume).
With this background, let us examine closely the range of semantic roles involved in the
applicative construction in Amharic. Consider a typical transitive verb k’orrətə ‘to cut’:

(64) aster dabbo-w-ɨn k’orrət’-əttʃ(-ɨw)

Aster bread-def.m-acc cut\pfv-sbj.3sg.f-obj.3sg.m
‘Aster cut the bread.’

In (64) all the required arguments of the verb ‘to cut’ are realised in the syntax: the
Agent as the subject and the Theme as the direct object. As the object is definite it is
marked by the accusative case morpheme -n. Neither of these core arguments can be
applicativised consistent with the cross-linguistic profile of the AC mentioned above
(but see the discussion on passives in § 4.2.1).
Now, a peripheral argument such as the instrument used to carry out the event of
cutting, can be introduced as an adjunct phrase as in (65):

(65) aster dabbo-w-ɨn bə-billawa k’orrət’-əttʃ(-ɨw)

Aster bread-def.m-acc with-knife cut\pfv-sbj.3sg.f-obj.3sg.m
‘Aster cut the bread with a knife.’

While the Theme object typically precedes the adjunct, the order is flexible, as can be
seen below:

(66) aster bə-billawa dabbo-w-ɨn k’orrət’-əttʃ(-ɨw)

Aster with-knife bread-def.m-acc cut\pfv-sbj.3sg.f-obj.3sg.m
‘Aster cut the bread with a knife.’
(Lit. ‘Aster, with a knife, cut the bread.’)

Now consider the same sentence, but with the verb taking the appl/ocr suffix and the
object agreement cross-referencing the Instrument:
268 Mengistu Amberber

(67) aster bə-billawa-w dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət

Aster with-knife-def.m bread-def.m-acc cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut the bread with the knife.’
(Lit. ‘Aster, with the knife, she cut the bread with it.’)

The instrumental phrase in (67) has two key properties: (a) it must be definite and (b) it
must precede the theme argument. If either of these requirements is not met, the con-
struction becomes ill-formed, namely when the instrumental phrase is indefinite (68):

(68) ✶
aster bə-billawa dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət
Aster with-knife bread-def.m-acc cut\pfv-sbj.3sg.f-ocr-obj.3sg.m

and when the Theme precedes the instrumental phrase (69):

(69) ✶
aster dabbo-w-ɨn bə-billawa-w k’orrət’-əttʃ-ɨbb-ət
Aster bread-def.m-acc with-knife-def.m cut\pfv-sbj.3sg.f-ocr-obj.3sg.m

Now consider the following construction with the instrument coded as an applied
phrase:

(70) ?aster billawa-w-ɨn dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət

Aster knife-def.m-acc bread-def.m-acc cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster cut the bread with the knife.’
(Lit. ‘Aster, with the knife, she cut the bread with it.’)

While (70) is not ungrammatical, it sounds awkward, presumably because both the
applied phrase (‘the knife’) and the Theme argument (‘the bread’) are marked by the
accusative case. This awkwardness disappears when the Theme argument is expressed
as an indefinite noun phrase thus making it ineligible for the accusative case:

(71) aster billawa-w-ɨn dabbo k’orrət’-əttʃ-ɨbb-ət

Aster knife-def.m-acc bread cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster cut a bread with the knife.’
(Lit. ‘Aster, the knife, she cut a bread with it.’)

Note, crucially, that it is only the Theme argument that can occur as an indefinite noun
phrase. The applied phrase cannot be indefinite, as the ungrammatical structure in (72)
shows:

(72) ✶
aster billawa dabbo-w-ɨn k’orrət’-əttʃ-ɨbb-ət
Aster knife bread-def.m-acc cut\pfv-sbj.3sg.f-appl-obj.3sg.m
 9 Applicativization in Amharic 269

One question is how to account for the contrast between the awkward (70), where both
the applied phrase and the Theme argument are marked as accusative, and (71), where
only the applied phrase is marked accusative. This may be due to information structure,
in that the applied phrase is the main topic of conversation and thus it displaces the
object in terms of definiteness and specificity. This may also account for the fact that
the verb can only be marked either for object agreement (agreeing with the theme argu-
ment of the verb), (73a) or oblique agreement (agreeing with the Beneficiary/Malefi-
ciary phrase), (73b), but never for both object and oblique agreement, (73c):

(73) a. k’orrət’-əttʃ-ɨw
cut\pfv-sbj.3sg.f-obj.3sg.m
‘She cut it.’
b. k’orrət’-əttʃ-ɨll-ət
cut\pfv-sbj.3sg.f-appl-obj.3sg.m
‘She cut [it] for him.’
c. ✶k’orrət’-əttʃ-ɨw-ɨll-ət
cut\pfv-sbj.3sg.f-obj.3sg.m -appl-obj.3sg.m
(Intended: ‘She cut it for him.’)

In discussing the semantics of the applicative morphemes in Gumer, Völlmin (2010:

321) points out that “[t]he semantics of a verb may restrict the use of some of these
[applicative] suffixes to very specific contexts or meanings, but it does not necessarily
preclude them from appearing.” Völlmin (2010) shows how the verb meaning ‘to taste’
in Gumer can be used in a number of different contexts. As the facts are nearly identical
in Amharic, we demonstrate this using the equivalent verb t’affət’ə ‘to be tasty’:

(74) a. mɨgb-u t’affət’-ə

food-def.m become_tasty\pfv-sbj.3sg.m
‘The food became tasty.’
b. ləmma-(n) mɨgb-u t’affət’-ə-w
Lemma-acc food-def.m become_tasty\pfv-sbj.3sg.m-obj.3sg.m
‘Lemma found the food tasty.’
c. ləmma-(n) mɨgb-u t’affət’-ə-ll-ət
Lemma-acc food-def.m become_tasty\pfv-sbj.3sg.m-appl-obj.3sg.m
‘The food became tasty for Lemma.’
(He cooked it well.)
d. ləmma-(n) mɨgb-u t’affət’-ə-bb-ətt
Lemma-acc food-def.m become_tasty\pfv-sbj.3sg.m-appl-obj.3sg.m
‘The food became tasty to Lemma’s detriment.’ (i.e., the food became too sweet)

In (74a) the verb t’affət’ə ‘become tasty’ is used with its single obligatory argument – the
thing that undergoes the change of state, i.e., ‘the food’. In (74b), the Experiencer argu-
270 Mengistu Amberber

ment (‘Lemma’) is explicitly expressed, but notice that this argument is cross-referenced
by the object agreement suffix on the verb. In (74c) and (74d) the verb is marked by the
applicative suffix and there is object agreement with the applied phrase (‘Lemma’) with
the Experiencer role. Interestingly, the applied phrase in both (74c) and (74d) is not
necessarily an Experiencer. Lemma may not have actually tasted the food himself, but
rather someone may have tasted it and determined that the food is tasty with either a
favourable (=benefactive) or a not favourable (=malefactive) judgment about Lemma’s
culinary skills.
As already mentioned above (see Examples [34a] and [34b]), with inherently dit-
ransitive verbs such as ‘to give’ the benefactive applicative marks an argument that
benefits from the event but is not a Recipient. Thus consider (34a) repeated below
as (75):

(75) aster lə-lɨdʒ-u məs’haf sət’t’ə-tʧ(-ɨw)

Aster to-boy-def.m book give\pfv-sbj.3sg.f-obj.3sg.m
‘Aster gave a book to the boy.’

In (75) the Agent argument occurs as a subject (‘Aster’), the Recipient occurs as an indi-
rect object (‘to the boy’), and the Theme argument is realised as the direct object (‘a
book’). If we keep the same three noun phrases as they appear in (75) but mark the verb
for the benefactive applicative, we get (34b), repeated below as (76):

(76) aster lə-lɨdʒ-u məs’haf sət’t’ə-tʧ-ɨll-ət

Aster to-boy-def.m book give\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster gave a book (to someone) for the benefit of the boy.’

Notice that, while in both (75) and (76) the phrase ‘the boy’ is formally identical,20 the
meanings are subtly different: in (76) the Recipient of the book is not and cannot be
‘the boy’, but rather someone not specified explicitly in the sentence. This is similar to
the so-called deputative applicative (see Van Valin and LaPolla 1997: 384), where the
applicative is depicting an act carried out on behalf of someone (expressed in the applied
phrase).
When we examine other ditransitive verbs such as ‘to send’ we get a different
picture. Consider the following:

20 It is due to this ambiguity of the phrase headed by the preposition lə- between the Recipient and
Beneficiary senses that some Amharic speakers of English as a second language usually confuse the two
meanings encoded by the English prepositions ‘to’ and ‘for’, saying I gave the book for him, instead of I
gave the book to him.
 9 Applicativization in Amharic 271

(77) aster lə-lɨdʒ-u məs’haf lak-ətʧ-ɨll-ət

Aster to-boy-def.m book send\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster sent a book to the boy.’
Also: ‘Aster sent a book to someone for the boy’s benefit.’

Interestingly, (77) is ambiguous between the Recipient interpretation (the boy is the
Recipient of the book) or the Beneficiary interpretation (the boy is the Beneficiary of the
book being sent to someone unspecified). Also note that unlike the case with the verb
sət’t’ə ‘to give’ in (75), without the applicative marker the construction in (77) with the
verb lakkə ‘to send’ becomes ungrammatical:

(78) ✶
aster lə-lɨdʒ-u məs’haf lakə-tʧ-ɨw
Aster to-boy-def.m book send\pfv-sbj.3sg.f-obj.3sg.m

However, if the Theme argument is individuated and fronted, the structure becomes
grammatical without the applicative suffix:

(79) aster məs’haf-u-n lə-lɨdʒ-u lakə-tʧ-ɨw

Aster book-def.m-acc to-boy-def.m send\pfv-sbj.3sg.f-obj.3sg.m
‘Aster sent the book to the boy.’

Notice that the object suffix now agrees with the Theme argument (‘the book’) and not
the Recipient (‘the boy’). How do other ditransitive verbs behave with respect to the
applicative construction? Let us look at the verb ʃət’t’ə ‘to sell’:

(80) aster lə-lɨdʒ-u məs’haf ʃət’ə-tʧ-ɨll-ət

Aster to-boy-def.m book sell\pfv-sbj.3sg.f-appl-obj.3sg.m
‘Aster sold a book to the boy.’
‘Aster sold a book to someone (unspecified) for the benefit/on behalf of the boy.’

Again, like the verb ‘to give’, the applicative of the verb ʃət’ə ‘to sell’ is ambiguous
between the recipient reading and the benefactive reading. But like with the verb
‘to send’, and unlike with the verb ‘to give’, removing the applicative suffix from (80)
renders the structure ungrammatical as the object pronominal suffix does not cross-ref-
erence the Recipient argument:

(81) ✶
aster lə-lɨdʒ-u məs’haf ʃət’ə-tʧ-ɨw
Aster to-boy-def.m book sell\pfv-sbj.3sg.f-obj.3sg.m

Now turning to the meaning of the malefactive benefactive, as already mentioned, it is

important to note that while it is typically employed to derive the meaning ‘something
272 Mengistu Amberber

is negative/bad for someone’, it is also used to mark the locative and instrumental.21 To
see this, let’s start with a sentence that has instrumental and locational adjuncts:

(82) aster bə-t’ərəp’p’eza-w laj bə-billawa sɨga k’orrət’-ətʧ

Aster on-table-def.m on with-knife meat cut\pfv-sbj.3sg.f
‘Aster cut some meat with a knife on the table.’

The first adjunct expresses the location, that is, where the event of cutting took place. It is
marked by the preposition bə- ‘on’ and the adposition laj ‘on’. The second adjunct expresses
the instrument by which the event was carried out, and it is also marked by the preposi-
tional prefix bə- ‘with’. Now, either adjunct can in principle be cross-referenced or become
an applied phrase. Consider (83) where the verb is cross-referencing the locational adjunct:

(83) ?aster bə-t’ərəp’p’eza-w laj bə-billawa sɨga

Aster on-table-def.m on with-knife meat
k’orrət’-ətʧ-bb-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut some meat on the table with a knife.’
(‘Aster, on the table, she cut some bread on it with a knife.’)

This sentence sounds awkward when the cross-referenced locational adjunct occurs
with the instrumental adjunct. When only the cross-referenced locational adjunct is
present without the instrumental adjunct, the sentence becomes perfectly acceptable:

(84) aster bə-t’ərəp’p’eza-w laj sɨga k’orrət’-ətʧ-bb-ət

Aster on-table-def.m on meat cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut some bread on the table.’
(= ‘Aster, on the table, she cut some bread on it.’)

Likewise, the instrumental adjunct can be cross-referenced as shown in (85a), but it is

more felicitous if it occurs without the locational adjunct as in (85b):

(85) a. ?aster bə-billawa-w bə-t’ərəp’p’eza-w laj sɨga

Aster with-knife-def.m on-table-def.m on meat
k’orrət’-ətʧ-bb-ət
cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut some meat with the knife on the table.’
(= ‘Aster, with the knife, on the table, she cut some meat with it.’)

21 This section has benefited from consulting Völlmin (2010: 322ff), who discusses the same phenome-
non in Gumer. See also Völlmin (2017).
 9 Applicativization in Amharic 273

b. aster bə-billawa-w sɨga k’orrət’-ətʧ-bb-ət

Aster with-knife-def.m meat cut\pfv-sbj.3sg.f-ocr-obj.3sg.m
‘Aster cut some meat with the knife.
(= ‘Aster, with the knife, she cut some meat with it.’)

This state of affairs, i.e., why a single cross-referenced adjunct is more felicitous is likely
to be due to information structure. The speaker chooses one of the multiple adjuncts
and foregrounds it in the conversation by the use of the appl/ocr marker.
When we talk about the locational applicative, it is important to keep in mind that
typically it corresponds to the meaning ‘at’ or ‘on’. It cannot correspond to the goal
location (‘towards a place’) nor to the source location (‘from a place’). Thus, consider
the following:

(86) aster wədə kətəma-w hed-ətʧ

Aster to city-def.m go\pfv-sbj.3sg.f
‘Aster went to the city.’

The goal location (‘to the city’) cannot be cross-referenced in main clauses (but see the
discussion above, § 4.4, on relative clauses):

(87) ✶
aster wədə kətəma-w hed-ətʧ-ɨbb-ət
Aster to city-def.m go\pfv-sbj.3sg.f-appl-obj.3sg.m

Similarly, a source location (‘from a place’) cannot be cross-referenced as the contrast in

the pair of examples in (88)–(89) shows:

(88) aster kə-kətəma-w wət’t’a-tʧ

Aster from-city-def.m leave\pfv-sbj.3sg.f
‘Aster left from the city.’

(89) ✶
aster kə-kətəma-w wət’t’a-ətʧ-ɨbb-ət
Aster from-city-def.m leave\pfv-sbj.3sg.f-appl-obj.3sg.m

It is obvious by now, but it is worth mentioning again, that with the appropriate context,
the applicativised verb can also be used to mark the malefactive if the event is cast as
negative for someone. Thus, when the verb is used in isolation, i.e., without the overt
NP arguments, which as we noted is possible in Amharic, it can be ambiguous between
three possible interpretations, namely, instrumental, locational, and malefactive:
274 Mengistu Amberber

(90) k’orrət’-ətʧ-bb-ət
cut\pfv-sbj.3sg.f-appl-obj.3sg.m
(i) ‘She cut [it] with it.’ = Instrumental
(ii) ‘She cut [it] on it.’ = Locational
(iii) ‘She cut [it] to his detriment.’ = Malefactive

Note that the object agreement suffix on the verb agrees with the applied phrase rather
than with the original object. Thus, when the verb is used in isolation, the grammatical
features of person, number and gender of the original object are not known. In the
above example where the verb k’orrət’ə ‘to cut’ is used by itself, the English translation
presents the original object (Theme argument) as ‘it’, but this is not necessarily the case
in Amharic as the object could be 3rd person singular feminine, 3rd person singular mas-
culine, or 3rd person plural.
We mentioned earlier that only certain arguments and peripheral roles can occur
as an applied phrase / cross-referenced oblique. Thus, we saw that while the locational
role depicted in English by the prepositions ‘on’, ‘at’ can be cross-referenced or become
an applied phrase, source or goal locations cannot. Likewise, the comitative cannot be
cross-referenced or become an applied phrase. First consider the comitative in the BC:

(91) aster kə-lɨdʒ-u gar hed-ətʧ

Aster with-boy-def.m together go\pfv-sbj.3sg.f
‘Aster went with the boy.’

The comitative role cannot become an applied phrase, whether in its adpositional form,
(92), or as an accusative marked phrase, (93):

(92) ✶
aster kə-lɨdʒ-u gar hed-ətʧ-ɨbb-ət
Aster with-boy-def.m together go\pfv-sbj.3sg.f-appl-obj.3sg.m

(93) ✶
aster lɨdʒ-u-n gar hed-ətʧ-ɨbb-ət
Aster boy-def.m-acc together go\pfv-sbj.3sg.f-appl-obj.3sg.m

Before concluding this section, we briefly look at one productive benefactive construc-
tion which involves the verb bɨl- ‘having said’ which is the converb form of the verb alə
‘to say’. Following Völlmin’s (2010: 327) description of a similar phenomenon in Gumer,
we can see below that the verb ‘to say’ can be used to form a benefactive construction
in Amharic. Thus, consider the following example:

(94) aster lə-lɨdʒ-u bɨl-a məs’haf gəzza-tʧ

Aster for-boy-def.m say\cnv-3sg.f book buy\pfv-sbj.3sg.f
‘Aster bought a book for the boy.’
(Lit. ‘Aster, having said for the boy, she bought a book.’)
 9 Applicativization in Amharic 275

Notice that in (94) with the verb ‘to say’, the Agent of the buying event (‘Aster’) has
thought of the Beneficiary when carrying out the event and this is consistent with other
grammaticalized uses of the verb alə ‘to say’ (for example, as a complementizer with
verbs of cognition and speaking). Interestingly, this construction is available only for
a Beneficiary in a peripheral role rather than the Recipient argument of a ditransitive
verb such as sət’t’ə ‘give’, as can be seen in (95).

(95) ✶
aster lə-lɨdʒ-u bɨl-a məs’haf sət’t’ə-tʧ
Aster for-boy-def.m say\cnv-3sg.f book give\pfv-sbj.3sg.f
(Intended: ‘Aster gave a book to the boy.’)

Thus, if the intended meaning is ‘Aster gave a book to the boy’, where ‘the boy’ is the
Recipient argument, the sentence is unacceptable. If the intended meaning is that ‘the
boy’ is a Beneficiary of the giving event whose Recipient is not specified, the construc-
tion is perfectly grammatical.

6 Conclusion
In this chapter we examined the applicative construction in Amharic, as well as a related
construction that does not meet the definition of an applicative construction provided by
Zúñiga and Creissels (this volume) but involves the same verbal marking.
We identified two distinct (although quite obviously related) constructions involv-
ing the suffixes -ll-/-bb-: a construction that fully meets the definition provided by Zúñiga
and Creissels (this volume), and a construction in which an adjunct expressed as a prep-
ositional phrase is cross-referenced as if it were an object.
We established that the Amharic applicative construction overall satisfies the
definition provided by Zúñiga & Creissels (this volume) and is consistent with a more
inclusive characterisation of the applicative whereby a non-Agent element is given a
more prominent morpho-syntactic and/or semantic-pragmatic prominence relative to
its status in the basic construction.
A key morphosyntactic feature of the applicative in Amharic is the use of the two
verbal suffixes -ll- and -bb-, which co-occur with object agreement suffixes on the verb.
While subject agreement is obligatory in the language, agreement with the object and/
or other (peripheral) arguments is governed by factors such as definiteness/specificity
or topicality/focus. The suffix -ll- is used when the applied phrase is a Beneficiary of the
event encoded by the verb, whereas the suffix -bb- is used when the applied phrase is
malefactive, locative, or instrumental. We showed that these two suffixes can be used
with virtually any verb irrespective of transitivity or the lexical semantics of the verb
root because any event can be potentially cast as benefiting or harming some entity. We
identified a syntactic lookalike to the applicative whereby a potential peripheral argu-
276 Mengistu Amberber

ment can be cross-referenced by an object agreement suffix alone, without the need
for the applicative suffixes -ll-/-bb-. This syntactic lookalike appears to be restricted to
certain verbs which can be independently construed as implicating a malefactive inter-
pretation, for example with verbs such as motə ‘to die’, sərrək’ə ‘to steal’.
We discussed the formal similarity between the suffixes -ll-/-bb- with the prepositions
lə-/bə- respectively. We observed that when the phrase cross-referenced by the agreement
suffix that follows -ll-/-bb- occurs with one of these prepositions, the suffixes -ll-/-bb- may be
optional. On the other hand, when the phrase cross-referenced by the agreement suffix that
follows -ll-/-bb- occurs without one of these prepositions (i.e., in the applicative construc-
tion), it must be marked for the accusative case and the suffixes -ll-/-bb- become obligatory.
It is plausible that the preposition marked applied phrase may be an intermedi-
ate construction before the emergence of the accusative marked applicative. To what
extent this may be true diachronically is a topic we leave for future work.

Abbreviations
acc accusative
appl applicative
aux auxiliary
cnv converb
comp complementizer
csd direct causative
csi indirect causative
def definite
f feminine
imp imperative
intr intransitive
ipfv imperfective
jus jussive
m masculine
mp medio-passive
neg negation
npst non-past
obj object
ocr oblique cross-reference
pfv perfective
pl plural
pol polite
poss possessive
recp reciprocal
rel relative
s singular
sbj subject
vn verbal noun
 9 Applicativization in Amharic 277

References
Ahland, M. 2009. From topic to subject: Grammatical change in the Amharic possessive construction.
Studies in Language. 33(3). 685–717.
Amberber, Mengistu. 1996. Transitivity alternations, event-types and light verbs. Montréal: McGill University
dissertation.
Amberber, Mengistu. 2000. Valency-changing and valency-encoding devices in Amharic. In R. M. W. Dixon &
Alexandra Y. Aikhenvald (eds.), Changing valency: case studies in transitivity, 312–332. Cambridge:
Cambridge University Press.
Amberber, Mengistu. 2002. Verb classes and transitivity in Amharic. Munich: LINCOM.
Amberber, Mengistu. 2009. Differential case marking of arguments in Amharic. In Andrej Malchukov &
Andrew Spencer (eds.), The Oxford handbook of case, 742–755. Oxford: Oxford University Press.
Amberber, Mengistu. In preparation. The Amharic converb in clause chaining. In Hannah S. Sarvasy and
Alexandra Y. Aikhenvald (eds.), Clause chaining in the world’s languages. Oxford: Oxford University
Press.
Baker, Mark & Ruth Kramer. 2014. Rethinking Amharic prepositions as case markers inserted at PF. Lingua
145. 141–172.
Creissels, Denis. Forthcoming. Transitivity, valency, and voice. Oxford: Oxford University Press.
Demeke, Girma A. 2003. The clausal syntax of Ethio-Semitic. Tromsø: University of Tromsø dissertation.
Demoz, Abraham. 1964. The meaning of some derived verbal stems in Amharic. Los Angeles: University of
California dissertation.
Eberhard, David M., Gary F. Simons & Charles D. Fennig (eds.). 2021. Ethnologue: Languages of the world.
24th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com.
Faber, Alice. 1997. Genetic subgrouping of the Semitic languages. In Robert Hetzron (ed.), The Semitic
languages, 3–15. London: Routledge.
HaBiT (n.d.) [Harvesting Big Text Data for Under-resourced Languages] Amharic Web 2013–17 (amWaC17).
https://corpora.fi.muni.cz/habit/run.cgi/corp_info?corpname=amwac17&struct_attr_
stats=1&subcorpora=1 (4 November, 2021).
Haile, Getachew. 1970. The suffix pronouns in Amharic. In Chin-Wu Kim & H. Stahlke (eds.), Papers in African
linguistics 3, 101–111. Edmonton: Linguistic Research.
Hetzron, Robert. 1970. Toward an Amharic case-grammar. Studies in African Linguistics 1. 301–354.
Hetzron, Robert. 1972. Ethiopian Semitic: Studies in classification. Manchester: Manchester University Press.
Kramer, Ruth. 2014. Clitic doubling or object agreement: The view from Amharic. Natural Language &
Linguistic Theory 32(2). 593–634.
Kuteva, Tania, Bernd Heine, Bo Hong, Haiping Long, Heiko Narrog & Seongha Rhee. 2019. World lexicon of
grammaticalization. 2nd edn. Cambridge: Cambridge University Press.
Leslau, Wolf. 1995. Reference grammar of Amharic. Wiesbaden: Harrassowitz.
McGinnis, Martha. 2008. Applicatives. Language and Linguistics Compass 2(6). 1225–1245.
Meyer, Ronny. 2011. Amharic. In Stefan Weninger (ed.), The Semitic languages: An international handbook,
1178–1212. Berlin: Mouton de Gruyter.
Meyer, Ronny 2012. The converb in Amharic. In Domenyk Eades (ed.), Grammaticalization in Semitic, 165–191.
London: Oxford University Press.
Mullen, Dana. 1986. Issues in the morphology and phonology of Amharic: The lexical generation of pronominal
clitics. Ottawa: University of Ottawa dissertation.
Rubin, Aaron D. 2008. The subgrouping of the Semitic languages. Language and Linguistics Compass 2(1).
79–102.
Rychlý, Pavel & Vít Suchomel. 2016. Annotated Amharic corpora. In Petr Sojka, Aleš Horák, Ivan Kopeček &
Karel Pala (eds.), Text, speech, and dialogue, 295–302. Cham: Springer.
Schachter, Paul. 1973. Focus and relativization. Language 49(1). 19–46.
278 Mengistu Amberber

Van Valin, Robert & Randy LaPolla. 1997. Syntax: Structure, meaning, and function. Cambridge: Cambridge
University Press.
Völlmin, Sascha. 2010. Benefactives and malefactives in Gumer (Gurage). In Fernando Zúñiga & Seppo
Kittilä (eds.), Benefactives and malefactives: Typological perspectives and case studies, 317–330.
Amsterdam: John Benjamins.
Völlmin, Sascha. 2017. Towards a grammar of Gumer—Phonology and morphology of a Western Gurage variety.
Zurich: University of Zurich dissertation.
Weninger, Stefan. 2011a. Ethiosemitic in general. In Stefan Weninger (ed.), The Semitic languages: An
international handbook, 1114–1123. Berlin: De Gruyter Mouton.
Weninger, Stefan. 2011b. Old Ethiopic. In Stefan Weninger (ed.), The Semitic languages: An international
handbook, 1124–1142. Berlin: De Gruyter Mouton.
Yabe, Tomoyuki. 2007. The morphosyntax of complex verbal expressions in the Horn of Africa. New York: City
University of New York dissertation.
Zelealem Leyew. 2007. Amharic varieties revisited. In Rainer Voigt (ed.), From beyond the Mediterranean:
Akten des 7. Internationalen Semitohamitistenkongresses (VII. ISHaK), Berlin, 13. bis 15. September 2004
(Semitica et Semitohamitica Berolinensia 5), 449–480. Aachen: Shaker Verlag.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg
10 Applicative constructions in Standard
Indonesian (Bahasa Indonesia)
Abstract: Indonesian (Bahasa Indonesia) is a standardised variety of Malay. The lan-
guage has two suffixes, -i and -kan, which can attach to verbs and function as applica-
tive morphology: in each case, the suffix causes the argument array of the verb to be
modified and it is the non-subject arguments which are affected. Both suffixes also have
other functions; in one case the suffix is extensively used also as a causative morpheme.
Indonesian has some features of a symmetrical voice system, and undergoer subject
constructions are more common than passives in English, for example. Applicative con-
structions interact with the voice system; in particular, in some cases the applicative
possibility is preferred in undergoer voice. For one suffix (-kan), it is not uncommon for
the morpheme to appear but for the preposition introducing what would be expected
to be an applied argument to be retained. In light of these various complications, we
suggest that applicatives in Indonesian are best understood as constructions with char-
acteristics, some more prototypical than others, which can be manipulated to accom-
modate syntactic and pragmatic factors.

1 Introduction
This chapter describes applicative constructions in Indonesian (ISO639-3: ind, Glot-
tolog: indo1316, Glottolog lists Bahasa Indonesia as a dialect of Standard Indonesian
with separate code stin1234), an Austronesian language. According to the census of
2010, the language is a first language for 43 million speakers with an additional 156
million speakers who have it as an additional language (Badan Pusat Statistik 2013: 421,
427). Standard Indonesian (Bahasa Indonesia) is the national language of the Republic
of Indonesia and is a codified version of (High) Malay. Its status as a future national
language was decided in 1928 and became official with the establishment of the nation
in 1948. The formal standard is described in an official grammar (Moeliono et al. 2017)
and there is a reliable description in English (Sneddon et al. 2010). This variety is not
used consistently in everyday life and, within the nation, there is a continuum of related
language varieties with the standard at the acrolectal end and localised Malay varieties
at the basolectal end (e.g. Ternate Malay, Litamahuputty 2012); some of these basilectal
varieties can even be considered creoles (e.g. Ambon Malay, Minde 1997). The variety

Acknowledgments: We are grateful to the editors of this volume and to an anonymous reviewer for helpful
comments.

https://doi.org/10.1515/9783110730951-010
280 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

spoken in Jakarta has influence beyond the capital city and can be considered an infor-
mal standard (Sneddon 2006). We concentrate here on the standard variety but will
occasionally provide relevant information about colloquial varieties.
Indonesian has productive applicative constructions marked by the verbal suffixes -i
and -kan which rearrange the argument array of a verb to make a peripheral participant
a non-subject core argument, an Applied Phrase (AppP) in the terminology of this volume.
The two suffixes are semantically differentiated: -i is used typically with locations and
related semantic roles, while -kan is typically used with beneficiaries and instruments.
Many verb roots can form derived verbs using either suffix and both suffixes can be used
to derive transitive verbs from intransitive roots and to derive ditransitive verbs from
transitive ones. Both of the suffixes are polyfunctional. The suffix -kan also functions as
a causative morpheme,1 as can the -i suffix (although this is less common). The suffix -i
also functions to encode extension of the action denoted by a verb, encompassing dura-
tion, iteration or intensity. There is also a construction in which the suffix -kan attaches
to a verb but there is no restructuring of the clause; that is, the oblique which would be
expected to become an AppP remains an oblique expressed as a prepositional phrase.
This chapter is organised as follows. Section 2 introduces the basics of Indonesian
morphosyntax necessary to understand the applicative construction. Section 3 describes
the two applicative suffixes and gives examples of their various functions. Section 4 dis-
cusses the semantic contrast between the two applicatives, suggesting that the distinction
is less clear cut than is traditionally claimed. Section 5 examines various possibilities where
some aspect of the prototypical applicative construction with -kan is optional, resulting in
a range of patterns which do not have all the properties of a typical applicative. Sections
4 and 5 are therefore the main basis for our overall argument that applicativisation in
Indonesian is best viewed from the perspective of prototype theory: the language has a
constructional template for applicatives which provides a prototype, but specific realisa-
tions of the construction may not have all the prototypical features. We suggest that these
less typical patterns should still be considered as falling within the overall construction.

2 Basics of morphosyntax of Indonesian

Indonesian is a SVO language with relatively rigid word order. The default linear order
is given in (1), with grammatical subject (subj) (S/A) coming by default in the clause-in-
itial position. The subject NP is also the default topic.2 Word-order variation is possible,

1 There is another homophone in the spoken language, a discourse particle marking shared knowledge
and requesting agreement or acknowledgement (Wouk 1998).
2 In the interests of simplicity, we do not show the internal structure of words other than verbs, except
in cases where glossing would be confusing without such information. The Actor Voice prefix, meN-,
has complex allomorphy which involves fusion of segments at the morpheme boundary in some cases.
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 281

and is pragmatically motivated. For example, a unit bearing a marked pragmatic func-
tion such as (contrastive) topic/focus appears at the left-most periphery position. This
is exemplified by Example (2) showing a topicalised/left-dislocated non-subject NP.

(1) [NP:S/A]SUBJ – Verb – [NP:P/G]OBJ – [NP:T]OBJ2 – [PP]OBL

(2) [Lagu itu]i barangkali [saudara akan menyukai(=nyai)]IP

song that perhaps 2sg fut av.like.appl=3sg
‘That song, perhaps you will like (it).’ (Arka 2021: 200)

subj selection in Indonesian is regulated by a voice system which is marked on the verb.
For example, Actor Voice (av), which selects the A3 argument (saudara in Example [2])
as subj is indicated by the presence of a prefix meN-, where the last element is a nasal
stop with various realisations depending on the following segment (Blust 2004: 81–84).
The Indonesian voice system retains some aspects of a symmetrical voice system as
seen in the Philippine-type languages. That is, Indonesian has the so-called Undergoer
Voice (uv), which appears in two constructional types depending on the formal struc-
tural properties of the verb and the A argument, exemplified in (3a–b). Traditionally in
Indonesian linguistics (Chung 1976; Sneddon et al. 2010), the structures in (3a–b) are
lumped together with (3c) as passive (pass). However, recent studies in Indonesian (e.g.
Arka and Manning 2008; Arka 2021) show that it is more accurate to distinguish the uv
structures in (3a–b) from the (real) pass structure in (3c) as they have different gram-
matical properties associated with the syntactic status of A. The A argument of the uv
structure—ku= ‘1sg’ in (3b), or =nya ‘3sg’ in (3c)—is a core argument as its properties
meet formal and behavioural properties of core arguments. These include its expres-
sion as a bare pronominal clitic. It occupies a core argument position immediately adja-
cent to the verb. It is therefore hosted by or affixed to the verb.4 This A core argument
is in contrast to the expression of the A oblique, which is flagged by the preposition oleh
as in (3c).

It is therefore confusing to indicate the break between morphemes, and we gloss Actor Voice forms
thus: av.[verb]. Many of our examples are drawn from the 300k sentence sample of the ind_mixed_2012
corpus (available from: https://wortschatz.uni-leipzig.de/en/download/Indonesian#ind_mixed_2012,
see also Goldhahn, Eckart, and Quasthoff 2012). Examples are referenced by sentence number in that
source. Examples for which no source is specified are provided by the second and third authors who are
both speakers of Indonesian.
3 We use Actor (A) and Undergoer (U) to refer to the semantic macroroles associated with a transitive
verb.
4 This possibility is restricted to first and second person actors although address terms can be used as
pronoun substitutes and Musgrave (2003) argues that the pronominal clitic in this construction should
be considered to be a verbal affix even though it is written as a separate word.
282 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

(3) a. Saya membaca buku itu

1sg av.read book that
‘I read the book.’
b. Buku itu ku=baca
book that 1sg=[uv]read
‘The book, I read (it).’
c. Buku itu di-baca=nya
book that diUV-read=3sg
‘The book, (s)he read (it).’
d. Buku itu di-baca oleh Amir
book that diPASS-read by Amir
‘The book was read by Amir.’ (Arka and Manning 2009: 47)

Given that an A can be a core argument but not subj, we use the term “non-subject
argument” in preference to “object” throughout. It should be noted that pass is
constructional in nature with verbal morphology (di-) being only an element of the
construction. Thus, di- simply marks that the U is selected as subj, which means that
the status of the A argument is unspecified, and it can be unexpressed. When it is
expressed, its status is determined by the syntactic flagging of oleh for pass, or its bare
clitic realisation hosted by the verb for uv (Examples [3b–c]). At the morphological level,
di- is ambiguous between pass and uv; hence our glossing of diPASS and diUV. The con-
structions marked by the prefix di- are only possible where the subject (the undergoer)
is third person, whereas the uv construction with a proclitic pronoun is not possible
with a third person subj. Intransitive verbs occur with various prefixes such as ber- and
including meN- but they may also be unprefixed. As noted previously, subjects precede
the verb in unmarked word order for all voice constructions.
Obliques are structurally peripheral. Obliques and core arguments are clearly
marked differently: obliques are realised as prepositional phrases whereas core argu-
ments appear as bare noun phrases.5 As is common in Austronesian languages, pro-
nouns in Indonesian canonically only refer to humans; this applies both to free pro-
nouns and to clitic pronouns. This stricture applies in principle to the clitic =nya, but
even in formal contexts, current usage allows this pronoun to have non-human refer-
ence. Additionally, =nya is normally treated as a 3sg form, but as there is no 3pl reduced
pronoun, some flexibility in this regard is also observed.
The applicative morphemes which are our focus are suffixes, and as mentioned
above, both have more than one function. -kan occurs commonly with both applicative
and causative function; -i can also function as a causative (see Footnote 9), but this is
much less common than the applicative use, and also functions to code extension of the

5 In Section 5, we discuss one construction where this distinction may be less clearcut.
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 283

action denoted by a verb. Indonesian has a constraint which allows only one suffix in
a complete word and therefore it is not possible to form a causative derived from an
already applicative verb; one of our examples (Example [11]) will illustrate this point.6

3 Applicative constructions in Indonesian

Both applicatives -i and -kan give rise to mono-/di-transitive argument arrays as cap-
tured in Table 1. The main difference is the semantic role of the applied argument, real-
ised grammatically as the postverbal (first) non-subject argument of the (av) sentence
schematised in (1). The applicative suffix -i selects a locative-related role (goal/source/
location), whereas -kan selects a non-locative (i.e. beneficiary/recipient or theme) role.
Beneficiaries and recipients can be thought of as locations, but in these cases some-
thing moves to the location. Kroeger (2007), amongst others, argues that this distinc-
tion is important. The third argument is thematically theme for both -i/-kan when the
derived structure is ditransitive. When the third argument is obl, it is thematically
instrument for -i and goal/locative for -kan. Each will be discussed in turn in the sub-
sections below.

Table 1: Sub-categorisation frames for derived -i/-kan structures (Arka et al. 2009).

(a) Monotransitive (b) Ditransitive

NPSUBJ NPNSUBJ NPOBL NPSUBJ NPNSUBJ1 NPNSUBJ2

-i A goal/ instrument -i A goal/ theme

source/ source/
location location
-kan A patient/ goal/ -kan A beneficiary/ theme
theme/ source/ recipient
instrument location

6 This constraint is also relevant to the question of whether Indonesian has any lexicalised applicatives.
There are no plausible candidates for words derived with -kan; the process is transparent and produc-
tive. However, there is a handful of verbs whose citation form ends with the vowel i and for these the
question does arise. Double vowels do not appear in Indonesian, therefore suffixation with -i in such
cases would presumably be invisible. Two of the verbs are beri ‘give’ and beli ‘buy’, both denoting situa-
tions where themes are transferred. Both can form derivatives with -kan, which suggests that the words
are synchronically monomorphemic, but does not tell us anything about their history, a topic which is
beyond the scope of this chapter.
284 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

3.1 Applicative -i and other uses of the suffix -i

The suffix -i is a derivational transitivising suffix, possibly affixed to stems of different

categories including a noun, an adjective or a verb; e.g., air (N) ‘water’ air-i ‘water’
(VTR), basah (A) ‘wet’ basah-i ‘dampen’ (VTR), lompat ‘jump’ (VINTR) lompat-i ‘jump
over’ (VTR). As seen from these examples, -i is a polysemous affix whose range of uses
includes, but is not limited to, applicativisation and causativisation. Its effect depends
on the semantics of the stem, but in both applicative and causative functions, it is often
(but not always) characterised by the addition of affectedness involving locative-related
meaning.
The interpretation of -i as causative or applicative typically depends on the seman-
tics of the stems. Broadly speaking, a causative interpretation arises when the stem is
an adjective expressing a stative event (e.g. panas ‘hot’ panasi ‘to heat’) or a noun con-
ceptually related to an event involving a displaced entity/theme depicted by the stem
(e.g. kulit ‘skin’ kuliti ‘to peel, to remove the skin of X’ (Example 4). An applicative
interpretation arises when the stem is a verb which expresses a dynamic event, typ-
ically involving agentivity (e.g. duduk ‘sit’ duduki ‘sit on X’) or directed motion (e.g.
datang ‘come’ datangi ‘come to X’, jatuh ‘fall’ jatuhi ‘fall onto X’).

(4) Dia mengulit-i pisang itu

3sg av.skin-caus banana that
‘S/he peeled (lit. removed the skin from) the bananas.’
(Arka et al. 2009: 90)

As a transitiviser, -i changes the valence or transitivity of the stem. For example, the
applicativisation of the intransitive verb duduk ‘sit’ (Example [5a]) results in a transitive
structure (Example [5b]). In (5a), the locative ‘chair’ is oblique, prepositionally flagged
by a locative preposition di, but in the applicative structure in (5b), it is a U non-subject
argument.

(5) a. Ia duduk (di kursi itu) (intr tr)

3sg sit loc chair that
‘S/he sat (on the chair).’
b. Ia menduduk-i kursi itu
3sg av.sit-appl chair that
‘S/he was sitting on the chair.’
(Arka et al. 2009: 88)

When the stem is transitive, the outcome of applicativisation varies. It depends on

whether or not the AppP can be understood as a goal/recipient-like argument that fits
with the mono- or di-transitive structure shown Table 1. For example, the applicativi-
sation of kirim ‘send’ results in a ditransitive structure as seen in (6b). This is because
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 285

the oblique dia ‘3sg’ (6a) is a human participant, high in animacy hierarchy, which
makes it easily understood as a recipient in the applicative structure (6b). The stem’s
U argument, uang ‘money’, is naturally understood as a displaced theme. It meets the
role requirement of the second non-subject argument of the ditransitive structure (6b).
Example (6c) shows that the AppP can become subject in an Undergoer Voice clause.

(6) a. Ayah mengirim uang kepada dia (goal) (intr tr)

father av.send money to 3sg
b. Ayah mengirim-i dia uang
father av.send-appl 3sg money
‘Father sent money to him/her.’
c. Dia yang di-kirim-i uang oleh Ayah
3sg rel diPASS-send-appl money by father
‘He who was sent money by father’.

However, if the applied argument is inanimate (e.g. sawah ‘rice field’ in Example [7]),
it typically only receives a locative role not a goal/recipient role interpretation. The
derived structure is not typically a ditransitive structure. Instead, as seen in (7b), the
result is a monotransitive structure with the theme argument interpreted to bear an
instrument role flagged by dengan, grammatically obl (cf. the subcategorisation frame
[a] in Table 1). Attempting a ditransitive structure downgrades acceptability as seen in
(7c) and ditransitive examples of this type do not occur in our corpus data.

(7) a. Mereka menanam padi di sawah

3pl av.plant paddy loc rice.field
‘They planted paddy in the field.’
b. Mereka menanam-i sawah dengan padi
3pl av.plant-appl rice.field with paddy
‘They planted paddy in the field.’
c. ??Mereka menanami sawah padi

The suffix -i may encode progressive or durative aspect, possibly with intensity, without
altering the valence of the stem. This is exemplified in Examples (8) and (9). Saya is
non-subject U argument in the non-applicative structure without the suffix (8a) and
also in the structure with a suffixed verb in (8b). As seen from the contrast in the free
translation, the suffix -i here signals repetition/intensity (or durative aspect), absent in
(8a). The same contrast is observed in (9a–b). It is explainable in terms of the extended
affectedness of U, due to the locative meaning imposed by -i. That is, in the case of
Example (8b), the U (saya ‘1sg’) is understood to be affected by the action of hitting
in an extended manner, e.g. on different parts of the body, while in Example (9b), the
affectedness is extended in intensity. This property of suffixes which also function as
applicatives is seen in a number of Austronesian languages. Balinese has a verb nyagur
286 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

‘av.hit’ and a derived verb nyagur-in ‘av.hit-appl’ which can have a progressive aspect
meaning, and the Javanese applicative morpheme -i also codes iterative action and/or
progressive aspect (Hemmings 2013).

(8) a. Ia memukul saya b. Ia memukul-i saya

3sg av.hit 1sg 3sg av.hit-appl 1sg
‘S/he hit me.’ ‘S/he was hitting me.’

(9) a. Ia memegang pencuri itu b. Ia memegang-i pencuri itu

3sg av.hold thief that 3sg av.hold-appl thief that
‘S/he held the thief.’ ‘S/he was holding the thief tightly.’

3.2 Applicative -kan and other uses of the suffix -kan

The -kan suffix has four main functions. Two of these functions result in applicative
constructions where participants otherwise expressed as obliques become part of the
syntactic core of the clause, that is, AppPs. The semantic roles in question in these con-
structions are beneficiary (§ 3.2.1) and instrument (§ 3.2.2). The third function is an
applicative lookalike where the suffix function is to make clear that the non-subject
argument of a verb is the patient of the action (§ 3.2.3). Finally, -kan functions as a caus-
ative (§ 3.2.4).

3.2.1 Beneficiaries

There are numerous verbs for which a beneficiary can be a participant in the action,
and that participant can be encoded either as an oblique in a prepositional phrase (typ-
ically with untuk ‘for’) with an unsuffixed verb (Example [10a]), or as an AppP with a
suffixed verb. A typical example is the verb bawa ‘carry’. Example (10b) shows that the
derived verb in this case is ditransitive, and Example (10c) shows that the beneficiary
can become subject in an Undergoer Voice construction:

(10) a. Tae Kyu pulang membawa bunga untuk Pal Kang

Tae Kyu return.home av.carry flower for Pal Kang
‘Tae Kyu came home carrying flowers for Pal Kang.’
b. orang~orang Majus membawa-kan Yesus mur serta emas
person~pl magus av.carry-appl Jesus myrrh with gold
‘The Magi brought Jesus myrrh and gold.’
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 287

c. Seorang mukmin cukup di-bawa-kan satu dalil

someone resident enough diPASS-carry-appl one proposition
saja dari sekian banyak dalil
only from so.much many proposition
‘It is enough if a community member is brought one idea from the many ideas.’
(Leipzig Corpora ind_mixed_2012_300K-sentences.txt: 158190, 121749, 277948)

We have referred to the semantic role involved here as “beneficiary”, reflecting stand-
ard descriptions, but in fact the AppP can be affected by the action in either a positive
or a negative way. That is, this applicative with -kan can be a malefactive construction:

(11) a. Ternyata pembantu=nya membawa-kan dia kopi beracun

it.appears servant=3sg av.bring-appl 3sg coffee poisonous
‘It turned out that his servant brought him poisoned coffee.’
b. kita hanya memberi-kan mereka ancaman jika mereka
1pl.incl only av.give-appl 3pl threat if 3PL
melaku-kan sesuatu hal yang negatif
av.behaviour-caus a.certain thing rel negative
‘We will only threaten them if they do something negative.’7

There is a small number of verbs which have derivatives with both applicative and
causative -kan suffixes, for example, jahit ‘sew’:

(12) a. tetapi saya harus menjahit celana ini!

but 1sg must av.sew pants this
‘. . .but I must sew these pants!’
(Leipzig Corpora ind_mixed_2012_1M-sentences.txt: 344166)
b. orang~orang (. . .) ikut menjahit-kan baju mereka kepada Anteh
person~pl join av.sew-caus shirt 3pl.poss to Anteh
‘. . .people (from village afar) came to have their shirt sewn by Anteh.’
(Leipzig Corpora ind_mixed_2012_1M-sentences.txt: 378654)
c. Tukang jahit itu menjahit-kan saya kemeja baru
craftsman sew that av.sew-appl 1sg shirt new
‘That tailor sewed me a new shirt.’
(Sneddon et al. 2010: 87)

Note that Example (12c) has two non-subject core arguments, that is, the derived verb
here is ditransitive. In contrast, Example (12b) is a monotransitive clause; the agent of

7 https://www.kompasiana.com/linistianah/5a0544565a676f79ee7047b2/sering-mengancaman-anak-
lihatlah-dampak-negatifnya
288 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

the base construction is coded as an oblique with the derived verb (see also discussion
of Example [14] below).

3.2.2 Instruments

With some verbs, the AppP is the entity used to carry out an action, the instrument.
Examples of this type are the verbs pukul ‘strike’ (Example [13a]) and tutup ‘cover;
close’ (Example [13c]):

(13) a. Hye Sung memukul kursi dengan tangan=nya

Hye Sung av.strike chair with hand=3sg
b. Hye Sung memukul-kan tangan=nya ke kursi
Hye Sung av.strike-appl hand=3sg to chair
Both: ‘Hye Sung struck the chair with his hand.’
c. aku memang menutup wajah dengan telapak tangan
1sg certainly av.cover face with palm hand
d. aku memang menutup-kan telapak tangan di wajah
1sg certainly av.cover-appl palm hand loc face
Both: ‘I would certainly cover my face with the palm of my hand.’
(Leipzig Corpora ind_mixed_2012_300K-sentences.txt: 267508, 269340)

There are a few verbs, for example tulis ‘write’, which have -kan forms both for an
instrument (Example [14b]) and for a beneficiary (Example [14c]):

(14) a. Dia banyak menulis naskah drama.

3sg many av.write manuscript drama
‘(S)he writes many drama/play manuscripts.’
(Leipzig Corpora ind_mixed_2012_1M-sentences.txt: 388787)
b. Dia menulis-kan spidol=nya ke baju saya
3sg av.write-appl marker=3sg to shirt 1sg
‘(S)he wrote with a marker on my shirt.’
c. Dia menulis-kan ayah=nya surat
3sg av.write-appl father=3sg letter
‘(S)he wrote a letter for their father.’
(Sneddon et al. 2010: 87–88)

Similar to the facts discussed in relation to Example (12) above, there is a contrast here
in the valency of the two derived verbs. The example with a benefactive applicative
(Example [14c]) is ditransitive, while the example with the instrumental applicative
(Example [14b]) is monotransitive.
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 289

3.2.3 Transitive clauses with -kan: valency increase and morphological lookalikes

In Section 2 above, we introduced the Indonesian voice system, including the use of
the prefixes meN- and di-. The possibility of both prefixes occurring before a stem is
a diagnostic for transitive verbs. Some verbs do not have a prefixed (i.e. transitive)
form without the suffix -kan. An example of this type is the root kerja ‘work’ which
derives an intransitive verb with the prefix be- (Example [15a]), and a transitive
verb with the suffix -kan, then also allowing prefixation with meN- or di- (Examples
[15b–c]):

(15) a. Cuma kini, sejak jatuh sakit, Julia tidak lagi be-kerja
only now since fall ill Julia neg again be-work
‘But now, since she fell ill, Julia has not worked again.’
b. ia mengerja-kan terjemahan baru keempat Kitab Injil
3sg av.work-appl translation new fourth book Injil
‘He is working on a new translation of the fourth gospel of Injil.’
c. Masih banyak hal yang perlu di-kerja-kan dan
still many thing rel necessary diPASS-work-appl and
di-persiap-kan
diPASS-ready-caus
‘There are still many things which need to be carried out and prepared.’
(Leipzig Corpora ind_mixed_2012_300K-sentences.txt: 87349, 145728, 218944)

Another verb of this type is tinggal ‘leave’. Example (16a) shows that a prefixed form
of this verb is possible without the -kan suffix, but this is only a transitive verb in a
small number of fixed idioms such as meninggal dunia ‘die’ (lit. ‘leave the world’). Suf-
fixation with -kan derives a transitive verb which can be used with any appropriate
non-subject argument and shows the expected range of voice possibilities (Examples
[16b–c]):

(16) a. Banyak orang bilang, orang baik biasanya meninggal cepat

many person say person good commonly av.leave fast
‘Many people say, good people usually pass away quickly.’
b. Para tamu itu memaksa Aidit meninggal-kan rumah
coll friend that av.force Aidit av.leave-appl house
‘Several friends made Aidit leave his home.’8

8 Note that this example uses a periphrastic causative construction; the constraint on multiple suffixa-
tion means that a derived causative verb is not possible.
290 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

c. Elemen traumatis yang di-tinggal-kan oleh semua tragedi

element traumatic rel diPASS-leave-appl by all tragedy
kemanusiaan abad ini. . .
humanity era this
‘The traumatic elements which have been left by all the human tragedies of
this era. . .’
(Leipzig Corpora ind_mixed_2012_300K-sentences.txt: 81349, 43695, 24484)

The following section (Section 4) has further discussion of another set of examples
involving emotion and cognition predicates, where both applicative suffixes can be
used to increase valency.
With other verbs, -kan can be added with no valency change. That is, such verbs
have prefixed forms both with and without the suffix and both are transitive, for
example sebut ‘mention’:

(17) a. Aristoteles menyebut tiga cara untuk mempengaruh-i manusia

Aristotle av.mention three way for av.influence-appl mankind
‘Aristotle mentions three methods to influence people.’
b. Kita akan menyebut-kan dua hal saja
1pl.incl fut av.mention-appl two matter only
‘We will mention only two issues.’
c. Bacalah dosa~dosa yang di-sebut-kan Kristus
read.emph sin~pl rel diUV -mention-appl Christ
‘Just read the sins which are mentioned by Christ.’
(Leipzig Corpora ind_mixed_2012_300K-sentences.txt : 12737, 180106, 152941)

Such constructions are examples of a morphological lookalike: the morphology of an

applicative construction is present, but there is no restructuring of the argument array
of the clause. Traditional accounts (e.g. Sneddon et al. 2010: 73) subsume these uses of
-kan under the description “marking the object as patient”.

3.2.4 Causative -kan

The suffix -kan can be used to add an additional agent-like participant to a clause. This is
possible both with intransitive verbs, resulting in a transitive clause (Example [18]), and
with transitive verbs, where the result is another transitive clause in which the agent of
the base construction is coded as an oblique (Example [19]).

(18) a. Siti bangun

Siti wake
‘Siti woke up.’
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 291

b. Ibu membangun-kan Siti

mother av.wake-caus Siti
‘Mother woke Siti up.’
(Sneddon et al. 2010: 78)

(19) a. Wanita itu mencuci pakaian saya

woman that av.launder clothes 1sg
‘That woman washes my clothes.’
b. Saya mencuci-kan pakaian pada wanita itu
1sg av.launder-caus clothes at woman that
‘I have my clothes washed by that woman.’
(Sneddon et al. 2010: 79)

Table 2 shows that causative -kan also functions to derive transitive verbs from nouns
and adjectives:

Table 2: Verbalising –kan.

Base Gloss Derived verb Gloss

kabar ‘news’ mengabar-kan ‘report’
sekolah ‘school’ menyekolah-kan ‘send to school’
bersih ‘clean’ (adjective) membersih-kan’ ‘clean’ (verb)
bebas ‘free’ membebas-kan ‘release’

4 The semantic contrast between the two

applicative morphemes
As mentioned initially and exemplified in the previous section, the two applicative suf-
fixes differ in terms of the semantic roles preferred for the AppP: locative/goal-related
roles by -i and other roles (patient/recipient/instrument) by -kan. The suffixes -i/-kan
can be affixed to the same stem, and as noted by Kaswanti Purwo (1995), Sneddon et al.
(2010) and Kroeger (2007), we therefore have linking alternations in Indonesian similar
to the spray-load alternation in English. This is exemplified in (20) and (21). The base
verb in Example (20), muat ‘hold, contain’, has an unsuffixed transitive form, seen in
Example (20c), while the base verbs in Examples (21) and (22) are intransitive.

(20) a. Buruh itu memuat-kan beras ke kapal

worker that av.load-appl rice to ship
‘The workers loaded the rice onto the ship.’
292 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

b. Buruh itu memuat-i kapal dengan beras

worker that av.load-appl ship with rice
‘The workers loaded the rice onto the ship.’
(Arka et al. 2009 : 89)
c. Kapal itu sedang memuat batu bara
ship that cont av.hold stone ember
‘The ship is loading coal.’
(Echols and Shadily 1994: 377)

(21) a. . . .air sudah menempel pada dek

water already av.stick at deck
‘. . .water already stuck at the deck.’
(Leipzig Corpora ind_mixed_2012_1M-sentences.txt: 372288)
b. Anak~anak menempel-kan poster ke tembok
child~pl av.stick-appl poster to wall
‘The children stuck a picture to the wall.’
c. Anak~anak menempel-i tembok dengan poster
child~pl av.stick-appl wall with poster
‘The children stuck a picture to the wall.’

Given the different roles selected by -i/-kan, the different effects as seen in (22) are
expected:

(22) a. Dia datang lagi dalam kehidupan=mu

3sg come again within life=2sg
‘(S)he came into your life again.’
b. Mereka mendatang-kan polisi
3pl av.come-kan police
‘They arrived with the police.’ (comitative-applicative -kan)
‘They called for the police / made the police come.’ (causative -kan)
c. Mereka mendatang-i polisi
3pl av.come-appl police
‘They came to / approached the police.’ (applicative -i)

There is an intriguing overlap with the goal-related role of -i and the recipient/bene-
factive role of -kan, which supports the idea that applicativisation is constructional in
nature. This is related to the ditransitive construction, exemplified in (23), where -i or
-kan is affixed to the same verb kirim, and the AppP is human, Tini. The differences in
meaning appear to be tenuous, especially between mengirimi (reading [i]) and men-
girimkan (reading [ii]). Reading (ii) might also imply (i) (i.e. a direct goal/recipient and
beneficiary role). Reading (iii) is only available for -kan (where Tini is the owner/source
of the book, benefiting from the assistance in sending the book).
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 293

(23) Saya [mengirimi](i)/[mengirimkan](ii)/(iii) Tini buku

(i) ‘I sent Tini a book.’ (Tini=the goal/recipient; the book was directly sent to Tini.)
(ii) ‘I sent a book for Tini.’
(Tini is the (intended) beneficiary; the book might have been sent via somebody else.)
(iii) ‘I sent a book (to somebody) for the benefit of Tini.’

Furthermore, evidence that the applicative is constructional and that the suffix -i/-kan
is only one element in the construction comes from the fact exemplified in (24). Here,
the applicative suffix is optional because the larger constructional template already
encodes the structure [NPSUBJ V NPNSUBJ:theme PPOBL:goal] (cf. subcategorisation [a] in
Table 1). In short, when the constructional template is satisfied, and it is consistent with
the stem’s (non-applicative) subcategorisation frame, then the suffix -kan is optional.

(24) Ayah mengirim(-kan) uang kepada saya (monotransitive)

father av.send-appl money to 1sg
‘Father sent money to me.’

Voskuil (1996) argues that the two applicatives in Indonesian reflect semantic distinc-
tions which are also visible in other languages. He claims that the -i applicative is similar
to the prefix be- in Dutch and the suffix -an in Tagalog in that all of these morphemes
derive change of state verbs along the lines of the zero-derived verb butter in English.
-kan, on the other hand, is similar to the Tagalog prefix i- in that both of them derive
change of location verbs like the English pocket class. Voskuil also sees this semantic dis-
tinction reflected in the fact that dative type relations are encoded with -i applicatives,
while benefactives are encoded with -kan.
This approach, in which the semantic properties of derived verbs are the result
of the interaction of the root and the rather general properties of the applicative mor-
phemes is taken even further by several scholars who attempt to include causative -kan
in the analysis. Vamarasi (1999) argues that, at least in the case of intransitive verbs, the
effect of the suffix -kan depends on the class of the base verb: for actor-oriented verbs
(unergatives), -kan behaves as an applicative adding a non-subject argument. And for
undergoer-oriented verbs (unaccusatives), -kan behaves as a causative, adding an agent
subject. Kroeger (2007) suggests that -kan has three functions: a morphosyntactic one
which is the benefactive applicative; a morphosemantic one, which covers the instru-
mental applicative and the causative with transitive bases, as well as constructions with
an optional suffix (see Section 6); and a category changing function which covers caus-
atives with intransitive bases. Which function applies depends on semantic properties
of the base. Cole and Son (2004) take this approach even further, claiming that -kan has
a single function: “licensing of a new argument in the argument structure that is not
licensed syntactically by the base verb” (p. 339).
All of these approaches give important insights into the functions and behaviour of
the two applicative morphemes in Indonesian, but they all encounter problems when
294 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

we include the class of emotion and cognition verbs.9 The basic pattern for this class of
words in Indonesian is for the experiencer to be the subject in the clause and for the
stimulus to be encoded in a prepositional phrase (some verbs allow a bare NP stimulus):

(25) Saya takut dengan anjing

1sg afraid with dog
‘I am afraid of dogs.’

To form a prefixed verb, that is to be part of the full transitive system, such verbs almost
all require an applicative derivation. Some verbs have -i derivatives, some have -kan
derivatives, some allow both possibilities and some allow both applicative and caus-
ative derivations with -kan. The verb introduced above, takut illustrates the range of
possibilities:10

(26) a. Hanya Tuhan yang sewajarnya kita takut-i

only lord rel truly 1pl.incl uv.fear-appl
‘We need fear only the Lord.’ (-i applicative)
(Echols and Shadily 1994: 544)
b. Tapi yang di-takut-kan oleh Mustika bukanlah wanita yang
but rel dipass-afraid-appl by Mustika neg.emph woman rel
di-kejar~kejar oleh Sandy
diPASS-chase~cont by Sandy
‘But what was feared by Mustika was not the woman who was being chased
by Sandy.’ (-kan applicative)
(Fredy S 1991: 30)
c. Berarti ia melangkah ke dalam bahaya yang menakut-kan
meaning 3sg av.step to inside danger rel av.afraid-caus
‘Meaning that he stepped into a terrifying danger.’
(Christie 1986: 274)

These examples show that, at least when associated with this class of verbs, either
applicative suffix can be used to derive a transitive verb. The semantic role of the par-
ticipant which becomes a core argument is the same; it is always a stimulus. While

9 See Musgrave (2003: Section 4.1.3) for discussion of whether these words should be considered verbs
or adjectives.
10 A causative verb can also be derived from this base with the suffix -i, that is, the full set of possibili-
ties can be seen with one verb:

Ia menakut-i dan membentak orang bawah=nya

3sg av.afraid-caus and av.snap.at person below=3sg
‘She intimidated her subordinates and spoke harshly to them.’ (Echols and Shadily 1994: 544)
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 295

the suffixes are typically associated with certain semantic roles when they function as
applicatives, these examples suggest that they can also function equally as transitivisers
in this context.
There is also a handful of adjectives which can be the base for derived verbs with
both i and -kan and for which there is little difference in meaning between the two
derivations. Examples of this type are habis ‘finished’ (Examples [27a–b]) and lengkap
‘complete’:

(27) a. Ia menghabis-i cerita=nya dengan nasihat

3sg av.finished-appl story=3sg with advice
‘(S)he finished their story with a warning.’
b. Ia menghabis-kan uang jaja=nya untuk membeli mainan
3sg av.finished-appl money snack=3sg for av.buy toy
‘(S)he used up their snack money buying toys.’
(Moeliono et al. 2017: 143)

Usage-based research (i.e. using corpora) also suggests that the semantic differentiation
of the two applicative suffixes is not as clear as has traditionally been assumed, but
is also dependent on voice alternations. A corpus-based study on the Indonesian verb
kena ‘be hit’ (Rajeg, Rajeg, and Arka 2020) shows that the -i and -kan forms of this verbal
base exhibit statistically distinct semantic preference in active (meN-) and passive (di-).
The predominant sense of mengena-kan (av) is the physical ‘to wear cloth/body-acces-
sories’ (Example [28a]) while the predominant sense of the passive di-kena-kan (pass) is
the metaphorical ‘to be imposed/be subject to (a regulation, obligation, etc.)’ (Example
[28b]). This ‘impose/subject to’ sense of kena-kan in passive is also the predominant one
for kena-i in pass (dikenai) (Example [28d]), but the active form of kenai (mengenai [av])
predominantly expresses ‘to hit’ (Example [28c]). Example (26e) illustrates the use of
the verb kena without a suffix.

(28) a. Dia masih mengena-kan celana seragam warna abu~abu

3sg still av.kena-appl trousers uniform colour ash~like
‘(S)he is still wearing the grey uniform pants.’
b. Kalau terlambat se-hari saja, Anda langsung dikena-kan
if late nsp-day only 2sg direct diPASS.kena-appl
denda atau bunga tambahan
fine or interest extra
‘If you are even one day late, you will be fined or charged interest.’
c. . . .panah=ku tepat mengenai badan=nya
arrow=1sg exact av.kena-appl body=3sg
‘. . .my arrow will precisely strike their body.’
296 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

d. saya akan di-kena-i biaya sebesar 60 persen

1sg fut diPASS-kena-appl fee as.large.as 60 percent
‘I will be charged a fee up to 60 per cent.’
(ind_mixed_2012_1M: 938247, 975688, 955107, 302404)
e. Akibatnya Putri kena tegur manajer kafe
as.a.result Putri be.hit reprimand manager cafe
‘As a result, Putri gets reprimanded by the cafe manager.’
(www.indosiar.com, collected 26/04/2012)

Two points are worthy of comment here. First, active and passive forms of the same
verb may have distinct semantic preference. Second, for kena specifically, the passive di-
forms of the applicative forms kena-i and kena-kan are semantically synonymous, and
felicitously interchangeable, (given their strong association with the same meaning),
but the applicative forms are not semantically interchangeable in active because the
active forms have distinct semantic preference. This also shows that verbs with applica-
tive morphemes may have different semantic preference in certain voice morphology
(cf. Rajeg, Rajeg, and Arka 2022 for another study on the interaction of verb semantics
and Indonesian voice morphology with motion verbs).
We also note a study (Rajeg, Denistia, and Musgrave 2019) which looked at the distri-
bution of Indonesian verbs derived from nouns in a semantic vector space model, a way
of quantifying relationships between lexemes based on their usage in a body of text. All
the nouns allowed derivation of prefixed verbs without any suffix, with the -i suffix and
with the -kan suffix. In almost all cases, the three verbs clustered together closely in the
semantic space; where this did not happen, one of the derivatives had acquired a special-
ised meaning (not an unusual phenomenon in the realm of derivational morphology). The
study shows no consistent pattern in the difference between each of the derived verbs
and the base verb; that is, -i applicatives are not consistently closer to the base verb than
-kan applicatives or vice versa. The study also shows that the contexts of use of the base
verb and the derivatives are generally similar and this is further evidence that semantic
differentiation between the different applicatives and a base verb is not always strong.

5 Optionality with -kan

There are two ways in which the -kan applicative construction has optional aspects.
First, there are some verbs which are monotransitive with or without the suffix and
for which the U argument does not change. This contrasts with the valency remapping
possibilities exemplified previously for the verb muat ‘contain’ (Example [20]). Second,
in some cases it is possible to include the suffix but without having one of its prototypi-
cal applicative functions. That is, in such cases, the oblique which would be expected to
become an AppP remains an oblique.
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 297

5.1 Optional suffix

Section 3.2.1 discussed -kan suffixation where the primary function of the morpheme
was to emphasise that the non-subject argument of the verb was a patient. For some
verbs where this happens, the suffix is optional; there may be no difference in meaning
between the two possibilities or, as in the following example, the suffix intensifies the
meaning. Echols and Shadily (1994: 470) gloss merusak as ‘damage, ruin’, while mer-
usakkan is glossed as ‘destroy, devastate, break something entirely’:

(29) a. hal~hal yang bisa merusak pikiran manusia

matter~pl rel can av.damage thought humanity
‘. . .matters which can hurt people’s thinking. . .’
b. hal~hal yang najis seperti itu yang merusak-kan
matter~pl rel dirty like that rel av.damage-appl
pikiran kita
thought 1pl
‘. . .matters which are similarly dirty that can destroy our thinking. . .’
(ind_mixed_2012_1M: 94890, 95351)

There are also verbs for which the suffix is optional in Actor Voice (with the prefix
meN-), but obligatory in passive clauses:

(30) a. Dia mengajar(-kan) Bahasa Indonesia

3sg av.teach-appl language Indonesia
‘(S)he teaches Indonesian.’
b. Bahasa Indonesia di-ajar-kan di sini
language Indonesia diPASS-teach-appl loc here
‘Indonesian is taught here.’
(Sneddon et al. 2010: 261)

Again, these are verbs for which -kan emphasises the role of the non-subject argument;
that is, they are a subset of those for which -kan is optional.
For one of the verbs exemplified here, corpus data suggests that the unsuffixed and
suffixed forms occur at very similar rates.11

11 Data collated from datasets made available by Wortschatz Leipzig (https://wortschatz.uni-leipzig.de/

en/download/Indonesian). We are grateful to Karlina Denistia for sharing her morphologically parsed
frequency list, allowing us to extract these data.
298 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

Table 3: Rates of occurrence of unsuffixed and suffixed forms for two verbs.

Unsuffixed Suffixed with -kan

Verb Gloss Raw count Percent Raw count Percent

rusak ‘damage’ 13,646 97.9 295 2.1

ajar ‘teach’ 13,715 51.6 12,865 48.4

The comparative frequencies reported in Table 3 vary considerably but they do give
a quantitative indication of what optionality can mean here although we have not
separated causative and applicative uses of -kan for those verbs and these figures do
not take account of voice. Further analysis along these lines will be time-consuming.
It requires making judgments for each verb as to whether one or both functions of the
suffix are possible, and then for cases where both functions are possible, judgments
have to be made about individual examples. Comprehensively parsed and tagged data
would reduce some of this work, but achieving that would also involve a considerable
effort.

5.2 -kan with retained oblique

Section 3.2.2 showed -kan as an applicative which changed the syntactic status of a
beneficiary. With a base verb, the beneficiary is encoded as a prepositional phrase, an
oblique, but with the derived applicative verb, the beneficiary is a core argument, an
AppP. In some case, however, a derived verb with -kan can occur in a clause in which
the beneficiary is still an oblique:

(31) Pelayan mengambil-kan se-gelas air untuk tamu

waiter av.take-appl nsp-glass water for guest
‘The waiter fetched a glass of water for the guest.’
(Sneddon et al. 2010: 86)

Sneddon et al. point out that if the beneficiary is clearly understood in context, then a
clause without any overt expression of that role is grammatical, and they suggest that
this may explain the apparently redundant construction seen in Example (31). The con-
struction has been stigmatised by some (e.g., Johns 1978: 232), or attributed to Javanese
influence (Verhaar 1984). A construction of this kind is typologically unusual, but there
is an even more surprising possibility here. As Kroeger (2007) notes, -kan in this con-
struction can reduce the valence of a verb. The verb beri ‘give’ is ditransitive in its base
form (Example [32a]), but it can be suffixed with -kan and then have the beneficiary
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 299

expressed as an oblique (Example [32b]), making the overall construction monotran-

sitive:12

(32) a. John memberi Mary buku itu

John av.give Mary book that
‘John gave Mary the book.’
b. John memberi-kan buku itu kepada Mary
John av.give-appl book that to Mary
‘John gave that book to Mary.’
(Kroeger 2007)

The preceding discussion has set up distinctions between different situations where
parts of the applicative pattern with -kan are optional. It is also possible to analyse these
possibilities in a more unified way. Such an analysis would suggest that the presence of
-kan sometimes (but not always, see e.g. Example [30]) has the function of emphasising
that the semantic status of the non-subject argument of the action as being an affected
participant. The affectedness can be prototypically evaluated negatively (giving rise
to the patient-marking -kan), or positively (giving rise to beneficiary-marking -kan).
These evaluations are also constructional in the sense that the proper interpretation
depends on a specific constructional context. Prototypically, the monotransitive frame
(cf. Table [1a]) selects the patient-making function of -kan, and the ditransitive frame
selects the beneficiary interpretation. When the base verb is already monotransitive
(Example [30a]), then it is expected that the patient -kan appears to be redundant, as it
does not add a non-subject argument. Its presence in this example is arguably seman-
tically motivated: it encodes emphasis to express a higher degree of affectedness, pro-
totypically linked to the first non-subject argument. This prototypical linking of first
non-subject argument with high affectedness is critical when the same argument must
be selected as subject for pragmatic or syntactic reason. Therefore, it is not surprising
to see why -kan is obligatory in the passive (Example [30b]) as (meng)ajar is polysemous
having a monotransitive frame, exemplified in (30a), and a ditransitive frame [menga-
jar NPrecipient/beneficiary NPtheme], not exemplified here. The obligatory use of -kan with the
patient selected as subject in (30b) serves as evidence for the complex semantic-syntac-
tic functional basis of -kan in the wider grammatical (voice) system in Indonesian. That
is, -kan is required as it functionally disambiguates different frames associated with
(meng)ajar.
The advantage of adopting a unified prototype analysis within the construction
theory is that we can account for a surprising yet attested demotion effect of the applica-
tive -kan, as seen in (32b). In this line of analysis, -kan is a polysemous suffix with a set of
semantic-constructional properties that include three critical features: (a) high degree

12 Denis Creissels (p.c.) suggests that this construction could be analysed as an antipassive.
300 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

of affectedness (syntactically related to transitivity; cf. Hopper and Thompson 1980), (b)
patient/recipient/beneficiary semantic role selection (in contrast to the locative selec-
tion of -i), and (c) constructional positive/negative evaluation (allowing specific role(s)
selection of (b). Critically, role selection (b)–(c) by -kan is not a selection of a unique or
exclusive role in the sense that the participant made salient by -kan may simultaneously
carry more than one of these roles. Thus, Yesus in (10b) is a participant bearing recipi-
ent and beneficiary roles simultaneously.
In a prototype theory analysis, forms can vary from a highly representative one
exhibiting all the typical features to a less typical or atypical one. In addition, there
is no one-to-one form-meaning association associated with a feature realisation. For
example, the beneficiary role can be expressed via the applicative -kan construction, in
which typically it is first non-subject argument, but it can also be prepositionally flagged
by untuk ‘for’ or kepada ‘to’. Empirical evidence of the type seen in (32b) is expected on
our prototypical analysis: the beneficiary is associated with more than one exponent or
coding (-kan and kepada), and the same coding (-kan) encodes more than one semantic
role (theme and beneficiary). From a derivational point of view, the structure of the type
exemplified in (32b) would look like involving a demotion. From a non-derivational
constructional analysis in prototype theory, however, it is an expected structure, albeit
non-typical, as certain meanings are simultaneously given emphasis—in this case,
affectedness of the patient buku itu ‘the book’ and positive/benefactive evaluation of the
recipient/goal ‘Mary’.
The prototype theory analysis would also suggest that the presence of -kan points
to a reading in which some party benefits from the action, even if that entity is not
encoded as a core argument. This account would therefore allow for some flexibility
of encoding of participants in a clause with an Actor Voice verb, but would predict that
passive clauses would require explicit indication of the syntactic status of the partici-
pant which is subject.
These possibilities suggest that taking a prototype theory approach to applicatives
is useful (cf. Michaelis and Ruppenhofer 2000). From a derivational point of view, the
prototypical construction involves a restructuring of the argument structure of a pred-
icate, with valence increasing if it does change, and taking the AppP from a range of
semantic possibilities. The type of construction seen in Example (32) is not prototypical
in this way, but it has some flavour of the prototype (in this case, the semantic specificity
of the construction).

6 Summary
Indonesian has two suffixes, -i and -kan, which can function as applicatives, but both
also have other functions. Both suffixes are used to derive verbs from base words,
which may be verbs or may belong to another word class in some cases. As applicatives,
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 301

the suffixes are semantically specific to some extent: -i typically makes a location the
AppP, while -kan has three specific functions: to emphasise that the AppP is a patient,
to make a beneficiary the AppP or to make an instrument the AppP. However, there are
examples, particularly when the suffixes are used with emotion and cognition verbs,
which suggest that the valence changing function of the applicative suffixes is not tied
closely to the typical semantics. This property, along with various possibilities where
the suffix -kan is optional and where its presence does not result in a restructuring of
the argument array, suggest that applicativisation in Indonesian can be considered to
be a constructional template with some prototypical realisations and some realisations
which lack some prototypical properties.
A number of our examples are drawn from corpus data and we believe that working
with such data is an important way forward to better understanding some of the issues
we have raised in this chapter. These include the possible similarities or differences
between -kan and -i dependent on voice morphology of the applicative verb, and the pat-
terns of co-occurrence of applicative -kan with a retained oblique argument. However,
such research is challenging because there is no way to separate the different functions
of the applicative morphemes, particularly -kan, using simple searches. Future research
on applicatives in Indonesian should be part of more extensive research on the complex
interface of morphosyntax and pragmatics. The discussion in this chapter has been
mainly focused on the morphosyntactic properties of -kan/-i applicatives, with some dis-
cussion on the pragmatic-syntactic motivation in relation to the passivisation respon-
sible for the tenacity of the applicative -kan (Example [30b]). The discourse pragmatics
of applicatives is an understudied area in Indonesian linguistics. An in-depth study of
the pragmatics of applicatives must be ideally based on a corpus of natural language
use, which covers different text types (written and spoken) and registers (standard/
formal and colloquial). Challenges to such a study include critical issues in producing
a large corpus, properly annotated with rich tags allowing us to retrieve information,
have breakthroughs in delineating complex variables, and answer questions regarding
the complex interface of the morphosyntax-discourse-pragmatics of Indonesian applic-
atives.

Abbreviations
A Actor
appl applicative
av Actor Voice
caus causative
coll collective
cont continuous aspect
emph emphasis
fut future
302 Simon Musgrave, I Wayan Arka, and Gede Primahadi Wijaya Rajeg

incl inclusive
intr intransitive
like having some quality of
loc locative
neg negation
NP noun phrase
nsp non-specific
nsubj non-subject
obl oblique
pass passive
PP prepositional phrase
pl plural
rel relativiser
sg singular
subj subject
tr transitive
U Undergoer
uv Undergoer Voice

References
Arka, I Wayan. 2021. Pivot and puzzling relativisation in Indonesian. In I Wayan Arka, Ash Asudeh & Tracy
Holloway King (eds.), Modular design of grammar: Linguistics on the edge, 181–202. New York: Oxford
University Press.
Arka, I Wayan, Mary Dalrymple, Meladel Mistica, Suriel Mofu, Avery Andrews & Jane Simpson. 2009.
A linguistic and computational morphosyntactic analysis for the applicative -i in Indonesian. In
Miriam Butt & Tracy Holloway King (eds.), Proceeedings of the LFG09 Conference, 85–105. Stanford: CSLI
Publications.
Arka, I Wayan & Christopher Manning. 2008. Voice and grammatical relations in Indonesian: A new
perspective. In Peter Austin & Simon Musgrave (eds.), Voice and grammatical relations in Austronesian
languages, 45–69. Stanford: CSLI Publications.
Badan Pusat Statistik. 2013. Penduduk Indonesia Hasil Sensus Penduduk 2010 (Result of the Indonesian
Population Census 2010). Badan Pusat Statistik. https://web.archive.org/web/20150402221528/http://
www.bps.go.id/index.php/publikasi/14.
Blust, Robert A. 2004. Austronesian nasal substitution: A survey. Oceanic Linguistics 43(1). 73–148. https://
doi.org/10.1353/ol.2004.0004.
Christie, Agatha. 1986. Satu, Dua, Pasang Gesper Sepatunya. (Trans.) Alex Tri Kantjono. Jakarta: Penerbit PT
Gramedia.
Chung, Sandra. 1976. On the subject of two passives in Indonesian. In Charles N. Li (ed.), Subject and topic,
57–98. New York: Academic Press.
Cole, Peter & Min-Jeong Son. 2004. The argument structure of verbs with the suffix -kan in Indonesian.
Oceanic Linguistics 43(2). 339–364. https://doi.org/10.1353/ol.2005.0003.
Echols, John M. & Hassan Shadily. 1994. Kamus Indonesia Inggris: An Indonesian-English dictionary (revised
and edited by John U. Wolff and James T. Collins with the assistance of Hassan Shadily). 3rd edn. Jakarta:
Penerbit PT Gramedia.
Fredy S. 1991. Perasaan yang ditinggalkan. Jakarta: Gultom Agency.
 10 Applicative constructions in Standard Indonesian (Bahasa Indonesia) 303

Goldhahn, Dirk, Thomas Eckart & Uwe Quasthoff. 2012. Building large monolingual dictionaries at the Leipzig
corpora collection: From 100 to 200 languages. In Nicoletta Calzolari, Khalid Choukri, Thierry Declerck,
Mehmet Uğur Doğan, Bente Maegaard, Joseph Mariani, Asunción Moreno, Jan Odijk & Stelios Piperidis
(eds.), Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC’12),
759–765. Istanbul: European Language Resources Association. https://aclanthology.org/L12-1154/
Hemmings, Charlotte. 2013. Causatives and applicatives: The case for polysemy in Javanese. In Working
Papers in Linguistics, Vol. 16, 167–194. London: SOAS.
Hopper, Paul & Sandra Thompson. 1980. Transitivity in grammar and discourse. Language 56(2). 251–299.
Johns, Yohanni. 1978. Bahasa Indonesia: Langkah Baru, a new approach. Canberra: Australian National
University.
Kaswanti Purwo, Bambang. 1995. The two prototypes of ditransitive verbs: The Indonesian evidence.
In Werner Abraham, Talmy Givón & Sandra A. Thompson (eds.), Discourse, grammar and typology:
papers in honor of John W. M. Verhaar, 77–99. Amsterdam: John Benjamins.
Kroeger, Paul. 2007. Morphosyntactic vs. morphosemantic functions of Indonesian -kan. In Annie E. Zaenen,
Jane Simpson, Tracy Holloway King, Jane Grimshaw, Joan Maling & Chris Manning (eds.), Architectures,
rules, and preferences: variations on themes by Joan W. Bresnan, 229–251. Stanford: CSLI Publications.
Litamahuputty, Bathseba Helena Johanna. 2012. Ternate Malay: Grammar and texts. Leiden: University of
Leiden dissertation. https://openaccess.leidenuniv.nl/handle/1887/19945 (28 August, 2023).
Michaelis, Laura & Josef Ruppenhofer. 2000. Valence creation and the German applicative: The inherent
semantics of linking patterns. Journal of Semantics 17(4). 335–395. https://doi.org/10.1093/jos/17.4.335.
Minde, Don van. 1997. Malayu Ambong : Phonology, morphology, syntax. Leiden: CNWS.
Moeliono, Anton M., Lapoliwa, Hans, Hasan Alwi, Sry Sattya Tjatur, Wisnu Sasangka & Sugiyono. 2017. Tata
bahasa baku bahasa Indonesia (Standard Grammar of Indonesian). Edisi keempat. Jakarta, Indonesia:
Badan Pengembangan dan Pembinaan Bahasa, Kementerian Pendidikan dan Kebudayaan.
Musgrave, Simon. 2003. Non-subject arguments in Indonesian. Melbourne: University of Melbourne
dissertation.
Rajeg, Gede Primahadi Wijaya, Karlina Denistia & Simon Musgrave. 2019. Vector Space Models and the
usage patterns of Indonesian denominal verbs: A case study of verbs with meN-, meN-/-kan, and
meN-/-i affixes. NUSA: Linguistic Studies of Languages in and around Indonesia 67. 35–75.
Rajeg, Gede Primahadi Wijaya, I Made Rajeg & I Wayan Arka. 2020. Corpus-based approach meets LFG: The
puzzling case of voice alternations of kena-verbs in Indonesian. In Miriam Butt & Tracy Holloway King
(eds.), Proceedings of the LFG’20 conference, on-line, 307–327. Stanford: CSLI Publications. https://doi.
org/10.6084/M9.FIGSHARE.12423788 (2 April, 2022).
Rajeg, I Made, Gede Primahadi Wijaya Rajeg & I Wayan Arka. 2022. Corpus linguistic and experimental
studies on the meaning-preserving hypothesis in Indonesian voice alternations. Linguistics Vanguard
8(1). 367–382. https://doi.org/10.1515/lingvan-2020-0104.
Sneddon, James Neil. 2006. Colloquial Jakartan Indonesian (Pacific Linguistics 581). Canberra: Pacific
Linguistics.
Sneddon, James N., Alexander Adelaar, Dwi Noverini Djenar & Michael C. Ewing. 2010. Indonesian reference
grammar. 2nd edn. Crows Nest, N.S.W.: Allen & Unwin.
Vamarasi, Marit Kana. 1999. Grammatical relations in Bahasa Indonesia (Pacific Linguistics, Series D, Vol. 93).
Canberra : Pacific Linguistics.
Verhaar, John W. M. 1984. Affixation in contemporary Indonesian. In Bambang Kaswanti Purwo (ed.),
Towards a description of Contemporary Indonesian: Preliminary studies, Part 1 / NUSA Linguistics Studies of
Indonesian and Other Languages in Indonesia, Vol. 18, 1–26. Jakarta: Universitas Atma Jaya.
Voskuil, Jan Eckhard. 1996. Comparative morphology: Verb taxonomy in Indonesian, Tagalog and Dutch. Leiden:
University of Leiden dissertation.
Wouk, Fay. 1998. Solidarity in Indonesian conversation: The discourse marker kan. Multilingua / Journal of
Cross-Cultural and Interlanguage Communication 17(4). 379–406.
Areal overviews
Rik van Gijn
11 Contact-induced diffusion of applicatives
in northwestern Amazonia?
Abstract: This paper surveys applicative constructions in the northwest Amazon and
adjacent Andean slopes. Previous research suggests that this area may feature contact-
induced diffusion of valency markers, specifically applicatives, across family bounda-
ries. On closer scrutiny, however, there seems to be no firm basis for this conclusion. First
of all, applicatives do not seem to be overly common in the area: many languages do not
have applicatives, and those that do often have no more than one (with a few notable
exceptions). Second, although some commonalities can be observed between the applica-
tive constructions across the area, they involve common features of applicative construc-
tions anywhere, like suffixed applicative markers, and a preponderance of benefactive
or sometimes malefactive semantics. To a lesser extent, comitative-related semantics are
found. There is some overlap in form, but these seem to be largely coincidental. A pos-
sible exception is a connection between Arabela (Zaparoan) and Yagua (Peba-Yaguan),
which do show signs of non-accidental similarities. In addition, there are some intriguing
but inconclusive functional similarities between Shiwilu (Kawapanan) and some of the
Arawakan languages of central Peru, outside the northwest Amazon.

1 Introduction
South America, in particular in western Amazonia, displays dazzling linguistic diver-
sity, especially in terms of the abundance of genealogical units. At the same time, gram-
matical traits are shared across genealogical boundaries (see e.g. Derbyshire 1987;
Doris L. Payne 1990; David L. Payne 1990; Dixon and Aikhenvald 1999; Campbell 2012;
Aikhenvald 2012 for overviews based on increasing amounts of information). Some of
these lists of shared features include valency-changing morphology, including applica-
tive morphology. For instance, Doris L. Payne (1990) discusses applicatives as one of the
salient features in her chapter on widespread morphological characteristics of Ama-
zonian languages. David L. Payne (1990), an overview of widespread forms in Amazo-
nian languages, mentions valency-changing markers: causative affix mV, a causative
verbal prefix V-, and a form ka, which is associated with different functions, among
them valency-changing (mainly causative, also including denominal verbs in Arawa-

Acknowledgments: I would like to thank the editors for their very useful comments, remaining errors are
mine. This publication is part of the ERC Consolidator project South American Population History Revisited,
funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and
innovation programme (grant agreement No. 818854 - SAPPHIRE). This support is gratefully acknowledged.

https://doi.org/10.1515/9783110730951-011
308 Rik van Gijn

kan). Aikhenvald (2012) also highlights that applicatives are common in Amazonia,
and that the commonality of comitative and instrumental applicatives is a specifically
Amazonian trait. She also mentions that applicatives and causatives are intertwined,
and connected through the notion of togetherness, which links one common feature of
Amazonia, sociative causatives (Guillaume and Rose 2010) and comitative applicatives.
All of this suggests that valency-increasing constructions are prone to contact-induced
diffusion in the Amazon, and that the distinction between applicatives and causatives
may be weak.
These conclusions seem to be echoed in some of the literature on the languages of
the north-west Amazon (henceforth NWA), a highly diverse area, with much evidence
to suggest intense (historical) intercultural contact (see e.g. Sorensen 1967; Aikhenvald
2002; Epps and Stenzel 2013; Seifart 2015; Chacon 2017). Payne (1985b) discusses verbal
markers in Peba-Yaguan and Zaparoan languages, which indicate, among other func-
tions, that the direct object should be interpreted as an instrument or accompanying
object. The case for a connection between these families was made in Payne (1984),
and assessed as possibly genetic in nature. Although Payne (1985b) does not commit
fully to interpreting the nature of the shared -ta forms, she regards a shared retention
as plausible. To date, there is no consensus for a genetic link between Peba-Yaguan
and Zaparoan, so contact-induced diffusion should still be considered a distinct pos-
sibility (especially in the light that it was only tentatively discarded by Payne). Wise
(1999, 2002), discussing a number of northern Peruvian languages, mentions that valen-
cy-changing suffixes that include the consonant t are common, and that, in addition to
Zaparoan and Peba-Yaguan languages, similar markers are found in Cahuapanan and
Witotoan languages, although these are less clearly applicative in nature, and more
towards the causative end of the continuum. Wise (2002) also connects a causative
marker present in some central Peruvian Arawakan languages of a similar form -ta/-da
and a transitivizing and verbalizing suffix which includes -t-. She tentatively concludes
that this is an areal feature, spread through contact. She does not commit to stating the
extent of this areal feature, whether it is northwestern Amazonian, western Amazo-
nian, or maybe even a more widespread distribution. Picking up the open questions
with respect to the areal extent of -ta marked valency-increasing suffixes, Crevels and
Voort (2020) identify two epicenters of valency markers with or including the form -ta,
one in the southwestern Amazon and one in the northwestern Amazon. In the latter
area, they identify applicative suffixes of the form -tV in Sikuani (Guahiban), Yagua,
as well as in languages from the Chicham, Kawapanan, and Zaparoan families, and
causative suffixes of similar forms in some Arawakan languages, as well as in Muniche
(isolate), Kokama-Kokamilla (Tupí-Guaraní), and Witoto (Witotoan).
The spread of forms is somewhat unexpected for the area. Most of what we know
about the NWA complies with the idea of function (pattern) borrowing, combined with
low levels of form (in particular lexical) borrowing. In many cases, it is the functional or
meaning component that is shared across languages rather than the form. This particu-
lar pattern, found in several areas across Amazonia (Epps and Michael 2017) is argued
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 309

by Epps (2020), based on earlier work by Londoño Sulkin (2012, 2017), to be related to
a “the Amazonian package”—a loosely shared system in which identity preservation in
the form of language maintenance and in some cases purism in a historical context of
widespread bilingualism and inter-group dependency—and plays an important role in
shaping the patterns of grammatical convergence in combination with relatively low
degrees of formal diffusion we find today. Uncharacteristically for this area, then, it is
the form that seems to have spread, while the functions of these ta-markers, although
they center around the concept of valency increase (and sometimes decrease), seem
somewhat diffuse.
As mentioned, several explanations for this distribution have been put forward:
while Payne (1984, 1985a) tentatively favors inheritance as an explanatory principle for
similar forms found in Zaparoan and Peba-Yaguan, Wise (2002) and Crevels and Voort
(2020) seem to favor contact-induced diffusion by way of form borrowing. However, as
discussed by Crevels and Voort (2020), a potential problem for the suggestion of areal
spread is the minimal form of the marker, containing a very common CV sequence,
which increases the likelihood that these similarities are due to chance. This then opens
up a third possibility, which has not been pursued systematically, that these similar
markers represent independent historical developments.
In this chapter, I will give an overview of the applicative constructions of the lan-
guages of the NWA, assessing to what extent the patterns found show functional and
formal commonalities within and across language families. As far as possible, I will
try to interpret whether these commonalities are best explained by common or inde-
pendent historical developments in the different genealogical units. The paper is built
up as follows: in Section 2, I briefly introduce the NWA and the sample for this paper.
Section 3, which makes up the bulk of the paper, discusses the applicative patterns
found in the languages of the area, first in terms of their presence or absence (§ 3.1),
then in terms of their morphology (§ 3.2), syntax (§ 3.3), and semantics (§ 3.4). Section 4
provides a discussion of the patterns found in terms of historical processes that may be
responsible for them.

2 The northwest Amazon

There are no clear natural geographical boundaries of the NWA, and so it is hard to
avoid arbitrariness in defining this area. In this paper, I will consider the area roughly
between the Upper Rio Negro and the Marañon River. This corresponds to the areas of
present-day Ecuador, southern Colombia, northern Peru, and the westernmost adjacent
strip of land in Brazil. The leading principle for the areal definition for the present
paper has been to include a number of known or presumed contact situations. The
most famous of these is probably the Vaupés, or more generally the Upper Rio Negro
area in the northeastern corner of the NWA (see e.g. Aikhenvald 2002; Epps 2006; Epps
310 Rik van Gijn

and Stenzel 2013), where in particular Arawakan and Tukanoan-speaking groups have
been in intensive contact with each other. This is part of a larger pattern of long-term
interactions between Tukano and Arawakan languages in this area (Chacon 2017), but
it also includes smaller language families like the Kakua-Nukak and Nadahup fami-
lies. Towards the south, there have been interactions between Arawakan, Boran, and
Witotoan languages (Seifart 2015), Other contact scenarios have been proposed for the
southern Marañon valley, which has been considered to be an exchange route (Rojas
Berscia & Eloranta 2019), also with connections to Andean societies (see e.g. Valen-
zuela 2015). The western portion of the area has received less attention, but there are
certainly suggestions of different kinds of contact situations, involving Chicham, Bar-
bacoan, West-Tukanoan, northern Quechuan, and Kawapanan languages, as well as
possibly languages further to the south along the Andes (Wise 2011; Valenzuela 2015;
Kohlberger 2020; Muysken 2021).
These considerations have led to the following sample of languages to be discussed
here, which maximizes the inclusion of these contact situations. Map 1 gives the approx-
imate locations of the languages in the sample, Table 1 lists them with their affiliations.

Map 1: The sample languages and their approximate locations.

 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 311

Table 1: The sample languages with their affiliations.

Language affiliation Language affiliation

Desano Tukanoan Shiwiar Chicham

Kotiria Tukanoan Wampis Chicham
Tukano Tukanoan Aguaruna Chicham
Barasano Tukanoan Shiwilu Kawapanan
Carapana Tukanoan Shawi Kawapanan
Tanimuka Tukanoan Ocaina Witotoan
Kubeo Tukanoan Murui Witotoan
Siona Tukanoan Miraña Boran
Secoya Tukanoan Muinane Boran
Koreguaje Tukanoan Arabela Zaparoan
Achagua Arawakan Iquito Zaparoan
Tariana Arawakan Dâw Nadahup
Piapoco Arawakan Hup Nadahup
Resígaro Arawakan Yuhup Nadahup
Warekena Arawakan Kakua Kakua-Nukak
Yukuna Arawakan Yagua Peba-Yagua
Imbabura Quechuan Tikuna Tikuna-Yuri
Chimborazo Quechuan Camsá Isolate
Calderón Quechuan Puinave Isolate
Salasaca Quechuan Urarina Isolate
Cañar Quechuan Cofán Isolate
Napo Quechuan Muniche Isolate
Pastaza Quechuan
Tena Quechuan
Loja Quechuan

The Tukanoan language family is entirely spoken in the NWA as it is defined here.
There is a major, high-level split between east and west Tukano (Chacon 2014), which, as
the name suggests, has a clear geographical correlate as well. The Arawakan languages
spoken in the NWA mostly belong to the Japura-Colombian (JC) branch of the family (the
majority of Arawakan languages are spoken outside the NWA). Warekena is the only non
JC language in the sample, as it is classified as Alto Orinoco (Hammarström et al. 2021).
The Quechuan family is mostly found in the Andean mountain range (stretching from
northern Chile and Argentina to southern Colombia), but in the NWA, the family extends
deep into the lowlands. In this study I have considered both lowland and highland varie-
ties, to achieve a more complete picture for Quechua. All of the languages in the sample
belong to the northern Quechua IIc branch, probably a late (Colonial) arrival in the area,
certainly in the lowlands (Muysken 2000; Ciucci and Muysken 2011). The Quechuan lan-
guages in the area likely came about through processes of language shift (Muysken 2021).
In addition, a number of smaller families and isolates are represented in the sample.
The Chicham family is concentrated in the Peru-Ecuador border area, and has five known
languages, of which three (Aguaruna, Wampis, and Shiwiar) are included in this study.
The Zaparoan family (de Carvalho 2013) has 6 known members in Ecuador and northeast-
312 Rik van Gijn

ern Peru, but three of these are extinct, and the others nearly extinct (Hammarström et
al. 2021). Given the lack of data, only Iquitos and Arabela could be included in this study.
The Kawapanan family of northeastern Peru has 2 known languages, Shiwilu and Shawi,
both represented in the sample. Witotoan has 7 known languages, but most of these are
poorly described, so that only Murui and Ocaina are included in the sample. Boran is a
small, 2-member language family, for which the state of description allows for the inclu-
sion of both Miraña and Muinane. Naduhup and nearby Kakua-Nukak are small families
spoken in the Vaupés area, next to Arawakan and Tukanoan languages. Peba-Yaguan, and
Ticuna-Yuri are small families found in the southeast of the NWA, which are represented
by a single language in the sample for lack of sufficient data on the other. In addition,
the sample has a few isolates: Camsá (also Kamsá) and Cofán (A’ingae) in the northeast,
Puinave (Wänsohot) in the northeast, and Urarina and Muniche in the southeast.

3 Applicatives in the NWA

3.1 The distribution of applicative constructions in the sample

Before zooming in on the specific characteristics of the applicative constructions found

in the area, I briefly consider the distribution of the languages that have and those that
do not have an applicative construction.

Map 2: Presence versus absence of applicative constructions.

The first observation to make is that applicative constructions are not particularly wide-
spread in the area, although the distribution may suggest some areal spread (see Map 2). As
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 313

we will see in more detail below, however, much of the observed patterns can be explained
by making reference to the language family.
Some families, or branches of families, are characterized by the absence of applica-
tive constructions. A case in point is Quechuan. Most Quechuan languages spoken further
to the south have an applicative-like marker -pu, but the Ecuadorian and Colombian
varieties of Quechua seem to have lost it. The grammars of Chimborazo (Beukema 1975)
and Imbabura Quechua (Cole 1985) do not describe an applicative marker, nor does the
dialectal overview study by Carpenter (1982), which includes data on most varieties of
Ecuadorian Quechua. Quechuan languages belonging to the same branch (Quechua IIc)
as the Ecuadorian Quechuan varieties surveyed here, but which are geographically dis-
connected have retained the applicative, as exemplified for Ayacucho Quechua in (1).
Note that this is not an applicative construction in the strict sense, in that the verbal
marker is not obligatory, and the benefactive is still marked with an oblique case, so that
there is no actual promotion of oblique objects to direct object status.

(1) Ayacucho Quechua (Quechua IIc; Sola and Parker 1964: 96)
ñuqa-paq rima-pu-wa-nqa
I-ben speak-ben-1obj-3fut
‘He will speak for me.’

Geographically nearby varieties from the northern Peruvian Upper Amazon (San
Martín Quechua), that belong to a different branch have also retained -pu.

(2) San Martín Quechua (Quechua IIb; Claassen 2018: 20)

pay wañu-chi-pu-n obeha-ta ñuka-pa
he die-caus-ben-3 sheep-acc 1-ben
‘He kills the sheep for me.’

In a comparative study of manuscripts, Muysken (2009: 98) concludes that the loss of the
benefactive is part of a set of changes that can be dated around 1700 AD, some time after
the arrival of Quechuan varieties in the area, and Muysken suggests (with some hesi-
tation) that these later changes may be the result of substrate influences. Possibly, Bar-
bacoan languages may have played a role, as they seem to lack morphological applica-
tives (see e.g. Curnow 1997 for Awa Pit, and Dickinson 2002 for Tsáfiki). For Ecuadorian
Quechua, then, if anything, contact-induced changes led to the loss of applicative-like
structures.
Although slightly less absolutely than Quechuan, the Arawakan languages of the
sample can be characterized by a general absence of applicative constructions, with
only one sample language, Yukuna, with an applicative. This lack of applicatives in the
Arawakan languages of the Japura-Colombian branch is in contrast to Arawakan lan-
guages spoken to the south of the NWA, in south and central Peru, which tend to have
several applicative morphemes. For instance, Michael (2008: 285–289) describes four
314 Rik van Gijn

applicative suffixes in Nanti (instrumental -aNt, “presencial” -imo ‘in the presence
of’, separative -apitsa, and indirective -ako). Mihas (2015: 275) also lists a number of
different morphological applicative constructions for Asháninka Perené, including a
sociative-causative -aka and three applicative morphemes that seem to be formally
related: benefactive -vint/-vent, benefactive/recipient -ront/-nont and instrumental/
reason -ant.
The Witotoan language family, of seven known members, is mostly scantily described,
with the exception of Murui and Ocaina, which have full grammars devoted to them (Woj-
tylak 2017 and Fagua 2013, respectively). Wojtylak (2017: 380–390) describes two valen-
cy-increasing constructions, both causative. Addition of causative -ta to intransitives or
transitives results in the demotion of the original subject to object (resulting in double
object constructions for underlying transitives), and the introduction of a causer subject.
It can also apply to non-verbal elements with a verbalizing effect. The second construc-
tion is a double causative, with the marker -ta applied twice, and introducing two causers.
The grammar describes no applicative constructions. This seems to be true for sister lan-
guage Ocaina as well. Fagua (2013: 273) reports three lexically determined causative allo-
morphs: -(ʔ)ta, -ha, and -(ʔ)sa, but no applicatives.
The Boran language family has two known members, of which Miraña is best
described. The descriptions by Seifart (2005) and Thiesen and Weber (2012) discuss
causation and reflexivity, but no applicative among the valency-changing operations.
The causative morpheme is -tsho, which again is formally similar to some of the other
markers of the area, as observed by Crevels and Voort (2020). Sister language Muinane
has received less descriptive attention, but it does not seem to have an applicative
construction either, and the causative suffix -su seems to be related to Miraña’s caus-
ative marker (see de Vengoechea 2012: 172–174 for a description).
Finally, some of the (near-)isolates also lack applicative constructions. Tikuna has
three antipassive morphemes (-ètà, -ē, -tàē) as well as a causative suffix (-’ẽ’́ e). There are
also passive and factitive operations that are not associated to specific dedicated morphol-
ogy (Bertet 2020: 365–385), but no applicatives. There does not seem to be any applicative
construction in Camsá (isolate) either. Arguments (case marked) can simply be added to
the core (O’Brien 2018: 206) There are also double-object constructions, but without any
marking on the verb (ibid.: 207). Girón (2008: 360–363) discusses a morphological causa-
tive in Puinave (isolate) that can increase the valency by one, adding a causer participant,
but sometimes it also just changes the role of the agent. In addition, there are valen-
cy-decreasing operations (ibid.: 363). There seem to be no applicatives, however. Urarina
(isolate) has valency-decreasing possibilities (reflexive, passive, detransitivizer), and also
a number of causative operations: -erate which attaches to either intransitive or transitive
verbs, the less productive -a attaches to intransitives only. Additional causative strategies
involve syntactic constructions with the verb letoaa ‘send’ (Olawsky 2006: 609–622), but
again, there is no description of an applicative construction. Arguments beyond the core
are marked with the oblique marker -ke (ibid.: 622). Although the description of Cofán
(isolate) is patchy, an overview paper (Fischer and Hengeveld 2023) discusses several
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 315

causative morphemes or allomorphs (-an, -en, -ña) but no applicative constructions. Like
Cofán, Muniche (isolate) is not very thoroughly described. Both Gibson (1996: 65) and
Proyecto de documentación del idioma Muniche (2009: 20–21) describe only causative
-cha/-chi in terms of valency-increasing operations.

3.2 The morphology of NWA applicatives

For the sample languages that do have applicative constructions, Table 2 lists the forms.

Table 2: The Tukanoan languages of the sample and their applicative markers.

Language Form Source

Desano ( Tukanoan) -basa Miller (1999: 117–118)

Kotiria (Tukanoan) -bosa Stenzel (2013 344–345)
Tukano (Tukanoan) bosa Ramirez (1997: 178–180)
Barasano (Tukanoan) -bosa Jones and Jones (1991: 68)
Carapana (Tukanoan) -boja Metzger (1981: 81–82)
Kubeo (Tukanoan) -ka Morse and Maxwell (1999: 60–61)
Siona (Tukanoan) -ka Bruil (2014: 213, 251)
Secoya (Tukanoan) -cai Johnson and Levinsohn (1990: 64)
Koreguaje (Tukanoan) -khaj Cook and Criswell (1993: 72–73)
Yukuna (Arawakan) -ñaa/-ñai/-ña Lemus Serrano (2020: 75–76)
Aguaruna (Chicham) -hu/-tu Overall (2007: 306–307)
Wampis (Chicham) -hu/-tu Peña (2015: 584–594)
Shiwiar (Chicham) -hu/-tu Kohlberger (2020: 297–302)
Shiwilu (Kawapanan) ek- Valenzuela (2016: 516)
Shiwilu (Kawapanan) -lapi Valenzuela (2016: 516)
Shiwilu (Kawapanan) -pa Valenzuela (2016: 516)
Shiwilu (Kawapanan) -tu Valenzuela (2016: 516)
Shiwilu (Kawapanan) -i Valenzuela (2016: 516)
Shiwilu (Kawapanan) -wa Valenzuela (2016: 516)
Shiwilu (Kawapanan) -nan Valenzuela (2016: 516)
Shawi (Kawapanan) ichi- Rojas Berscia (2013: 92)
Shawi (Kawapanan) -të Hart (1988: 269); Rojas Berscia, 2013: 50)
Arabela (Zaparoan) -t(i)a Rich (1999: 55–57)
Iquito (Zaparoan) -nii Lai (2009: 299, 301); Michael (2009: 155)
Hup (Naduhup) -ʔṹh Epps (2008: 500)
Kakua (Kakua-Nukak) -áʔbuhú Bolaños (2016: 338)
Yagua (Peba-Yaguan) -ta Doris L. Payne (1985a: 271–278)

On the basis of Table 2, a number of generalizations can be made. All applicative markers
in the sample languages appear to the right of the lexical verb, mostly in the form of a
suffix, although in East Tukanoan languages the markers, which are clearly cognates,
show different degrees of morphophonological integration with the stem. In the Colom-
316 Rik van Gijn

bian East-Tukanoan language Tukano, bosa is regarded as a dependent verb in a serial

verb construction. Although dependent verbs in this language are underlyingly toneless
and thus prosodically dependent, they do not partake in nasal spread. In addition, many
dependent verbs can also be used as independent verbs. They also appear closer to the
root than bona fide suffixes, and they are bimoraic, unlike suffixes (Ramirez 1997: 173).
A serial verb construction involving the root bosa is exemplified in (3).

(3) Tukano (East Tukanoan; Ramirez 1997: 185)

Pédudu-de da’da bosa-bĩ
Pedro-obj work ben-pres.vis
‘He works for Pedro.’

Likewise, in Kotiria, -bosa is part of a set of noninitial verb roots, which appear contigu-
ous to an initial verb root, and can be contrasted with nonroot stem morphemes, which
appear further from the root, and represent a layer of verbal morphology that is more
grammaticalized (Stenzel 2013: 244–245).

(4) Kotiria (East Tukanoan; Stenzel 2013: 209)

yu’ú wa’í-ré do’á-bósá-í-tá mu’ú-ré
1sg fish-obj cook-ben-1/2.m-int 2sg-obj
‘I’m going to cook the fish for you.’

In Barasano, -bosa is analyzed as a suffix, and elements can come in between it and the
verb root (although the reverse order is also possible).

(5) Barasano (East Tukanoan; Jones and Jones 1991: 68)

yu karta yu-re ãbi-a-bosa-ya bũ
1sg letter 1sg-obj pick.up-mot-ben-pres.ipfv 2sg
‘Take my letter for me.’

In Carapana, the benefactive marker -boja is regarded to be a second order suffix. The
difference between second order and first order suffixes in Carapana is that the latter
may occur on the initial member of a serial verb construction, the latter cannot (Metzger
1981: 60). The analysis furthermore suggests that the second order suffixes are further
removed from the root than the first order suffixes if both types occur, but unfortu-
nately, Metzger does not provide examples to corroborate this directly.1 Example (6)
gives an instance of the use of the suffix -boja.

1 He does, however, give examples of the second order suffix -nucũ, which shares the same slot as -boja,
in combination with first order suffixes, which indeed precede it.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 317

(6) Carapana (East Tukanoan; Metzger 1981: 82)

yu áti-boja-ya
I do-ben-imp
‘Please do (it) for me

Flexivity is relatively uncommon for the applicative markers in the sample, but it does
occur. Yukuna (Lemus Serrano 2020), for instance, has a morpheme -ñaa with allo-
morphs -ñai, and -ña. The variant -ñai is morphologically conditioned by the presence
of the past tense marker -cha; -ñaa and -ña are free variants.
In all three Chicham languages, there is lexically determined allomorphy, as well
as homophony with the first person singular object marker, which shows the same allo-
morphy with the same distribution. The first person object in combination with the
applicative, yields deviant, but systematic patterns, as shown for Aguaruna in Table 3.

Table 3: Morphological interaction of noun class, applicative marking,

and first person object marking in Aguaruna (Overall 2007: 319).

class applicative 1sg.obj applicative+1sg.obj

1 -hu -hu -hu-tu

2 -tu -tu -tu-hu

Whereas the more abstract formal characteristics discussed above show a number of
similarities across languages, with a tendency towards suffixation, or at least the order
lexical verb – applicative marker, the forms themselves show considerable variation,
which seems to be mainly constrained by genealogical units.
The Tukanoan languages in the sample feature two recurring forms to mark applic-
atives: -ka and (-)bosa/basa. These two markers follow an almost perfect East-West dis-
tinction, the former being mostly restricted to West Tukano, the latter to East Tukanoan
languages. The only exceptions are the East Tukanoan language Kubeo, which follows a
western pattern, and Tanimuka (also East Tukanoan), for which no applicative marker is
described. As discussed above, the element bosa and related forms in the East Tukanoan
languages seems to have its origin in a serial verb construction, and presently is in
different stages of grammaticalization towards becoming an affix in the different lan-
guages. This suggests that the form -k(h)a(j) and its reflexes may represent an older form.
That Kubeo, as the only East Tukanoan language has retained the probably older form
is in line with the assessment that Kubeo has relatively more retentions and probably
developed in relative isolation from the other East Tukanoan languages (Chacon 2013:
414–418).
The Chicham languages also show a clearly coherent pattern. The forms in the three
languages are clearly cognate (-hu/-tu in Aguaruna, -ru/-tu in Wampis and Shiwiar), the
allomorphy follows the same pattern, including the pattern given in Table 3, and, as we
318 Rik van Gijn

will see below, the syntax (§ 3.3) and semantics (§ 3.4) are also very similar. All of these
very specific commonalities strongly suggest that the applicative, with its idiosyncra-
sies, can be traced back to Proto-Chicham. This is relevant in the discussion of ta-like
applicative markers, of which Chicham languages are hypothesized to be part (Crevels
and Voort 2020: 189). If so, then areal influence must have taken place prior to the diver-
sification of the family, or else the Chicham languages must be hypothesized to be among
the donor languages for the areal spread of this form.
The remainder of the families in the sample show a less coherent picture, unless it
is in terms of the absence of applicative constructions (see § 3.1 above). Of the Arawa-
kan languages in the sample, only Yukuna2 has an applicative construction, marked
with -ñaa. Hanson (2010: 272–280) describes two applicative morphemes for Yine, an
Arawakan language spoken outside the NWA region in southern Peru, one of which is
-ya, marking a number of different roles, including source and malefactive, two mean-
ings also found for the applicative construction in Yukuna. This marker may therefore
be cognate with Yukuna -ñaa, but this requires further research. Lemus Serrano (2020)
mentions that the lexical source for the applicative marker in Yukuna is probably the
verb ñáa/ñái ‘escape from’.
Kawapanan Shiwilu and Shawi are the only languages in the sample with more
than one applicative construction. Although they differ considerably from each other
in their inventories, the prefixes ek- (Shiwilu) and ichi- (Shawi), as well as the suffixes
-tu3 (Shiwilu) and Shawi -të (pronounced /tɨ/ or possibly /tɯ/, with allomorph -ta in com-
bination with progressive verb semantics), can be considered cognates, but the remain-
der of Shiwilu applicatives do not seem to be present in Shawi. The two Kawapanan
languages, then, show both stability and change regarding morphological applicative
marking, as either Shawi has lost a number of applicative and applicative-like construc-
tions, or Shiwilu has acquired them through contact. The functional overlap with some
pre-Andine Peruvian Arawakan languages (presential, relinquitive) would suggest a
potential connection with Shiwilu (see Section 4).
Zaparoan languages Iquito and Arabela also show disparity in their applicative
markers. Iquito -nii is possibly related to one of the causative allomorphs of Arabela,
although at this point this is only a conjecture. Arabela -ta/-tia introducing an accompa-
nying person or thing for the patient of a transitive or the S of an intransitive verb, with
a range of interpretations (Rich 1999). The allomorphy is also present in the form of the

2 Tariana has syntactic (labile verbs) and morphological lookalikes. The morpheme surfacing in the
latter clearly has a different etymology than the marker in Yukuna.
3 Valenzuela (2016) distinguishes two markers -tu in Shiwilu, which are historically related. One is ana-
lyzed as an applicative (with locative semantics), the other, termed valency modifier, is not analyzed as
an applicative, but its presence is required with some of the applicatives. In addition, -ta can appear on
its own with differential effects on the valency of the base verb (decreasing or increasing). The Shawi
marker -të is semantically closer to this latter use of Shiwilu -tu, suggesting that the locative applicative
use was a later development in Shiwilu.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 319

homophonous instrument marker -ta/-tia, as noted by Payne (1985a) and Rich (1999),
who assumes that the applicative marker stems from the instrument postposition (p. 76).
Given that the instrument markers in Arabela and Iquito may well be related, and
given that only Arabela seems to have extended its use to an applicative, it may well
be that this grammaticalization process was influenced by Yagua, as suggested by Wise
(2002). For Yagua, Payne (1985a: 271) discusses the marker -ta, which is most frequently
used to license semantic comitatives or instruments.4 The same marker is also found
with nouns, where it marks the comitative/instrument role. These facts seem to point at
a shared history between Arabela and Yagua, although the nature of this shared history
is subject to debate (see Section 1 above). Interestingly, Yagua has a lexically restricted
causative morpheme -niiy which may be related to Arabela causative -ni. Although it is
unclear what the distribution of -ni in Arabela is, the fact that it is presented as one of
many causative morphemes leaves open the possibility that -ni in Arabela is also lexi-
cally restricted. In addition, Yagua has the productive causative marker -taniiy, possibly
a combination of the -ta applicative and causative -niiy. In short, although all of this
requires further research, there is a very likely connection between Yagua and Arabela
-ta, as well as a possible connection between Yagua -niiy, Arabela -ni (both causative),
and Iquito -nii (applicative).
The final languages to be discussed for this section are Hup and Kakua. The suffix
-ʔũh in Hup adds an animate participant (Epps 2008: 500). This marker (in slightly differ-
ent guises) has a number of seemingly unrelated functions. In preverbal or independent
position, it marks reciprocal and a kin term (sibling of opposite sex). As a postverbal
marker it can indicate applicative, jussive, and epistemic modality. The uses of prever-
bal and postverbal each seem to be related, and it may even be that all of these are
historically related (see Epps 2008: 504–505 for discussion).
For the sister languages Yuhup and Dâw, no applicative construction is described,
but Yuhup has a possibly related element ~ɟah, which Ospina Bozzi (2002: 400) analyzes
as a bleached and eroded verb root of the verb ‘let, allow’, which has among its functions
(next to e.g. causative and permissive uses, but also reflexive, intensity and repetition)
the ability to introduce an experiencer argument, affected by the situation, to an origi-
nally intransitive clause (ibid.: 402–407). The situation in Dâw is less clear, but Martins
(2004: 266) mentions a grammaticalization path of the verb dóʔ from a movement verb
to verb of causation or permission in verbal compounds. However, there seems to be no
extension towards an applicative-like use. In addition, Dâw has a transitivizing oper-
ation (marked by a descending tone). Although it is sometimes associated with partly
unpredictable semantics, it mostly seems to have a causative effect (Martins 2004: 180).
All in all, there seems to be a family-internal grammaticalization process, from serial
verb construction to affix, which has led to an unusual set of functions of historically

4 Some of the examples discussed in Payne (1985a, e.g. p. 275) suggest that the marker -ta in Yagua also
has an allomorph -tya.
320 Rik van Gijn

related markers, including both valency-increasing and valency-decreasing uses in Hup

and Yuhup.
Bolaños (2016) discusses no valency-increasing operations for Kakua (Kakua-Nukak).5
In fact, she mentions that the reflexive/reciprocal prefix mǐk- is the only valency-changing
morpheme of the language (Bolaños 2016: 247). There is, however, a benefactive impera-
tive, though, with the marker -áʔbuhú, which licenses a benefactive participant.

(7) Kakua (Kakua-Nukak; Bolaños 2016: 338)

kăn=diʔ ʔɨbʔ-áʔbuhú
3sg.m=obj take.out-imp.ben
‘Take (it) out for him.’

Given the length of the morpheme, Bolaños assumes that it is diachronically complex. A
possible element of the suffix is the directional enclitic =buh, which in turn comes from
a verb root meaning ‘to perform an action from a distance’, which may be connected to
performing an action on someone’s behalf (Bolaños 2016: 338–339).
In conclusion, although the very general formal aspects of the applicative markers
coincide to some degree (they tend towards morphological, suffixal expression), the
forms themselves suggest more family-internal developments.

3.3 The syntax of applicative constructions in the sample

In this section, I highlight the more syntactic aspects of the applicative constructions
found in the area, focusing mainly on the expression and morphosyntactic marking of
arguments, particularly in relation to clauses with underived transitive and ditransi-
tive verbs. Unfortunately, the available information in the grammars does not allow for
a systematic treatment of the potential of applicative constructions to form input for
further valency-changing operations, or for relative clauses. I will therefore have less
to say about how applicative constructions relate to other constructions, and restrict
myself to making occasional remarks where information was available. Given that the
languages of the area employ a range of techniques to mark grammatical relations, and
that these techniques tend to be relatively homogeneous within language families, the
discussion in this section will be organized around families.
Tukanoan languages are nominative-accusative in that S and A arguments are treated
the same in terms of case marking and indexing, and differently from P arguments. Most

5 There is some debate with respect to the genealogical affiliation of Kakua and Nukak, as Martins
(2005) proposes they are part of the Nadahup family. Jolkesky (2016) also includes Puinave in this gene-
alogical unit. Epps and Bolaños (2017) consider a distant connection between Nadahup, Kakua-Nukak,
and Puinave plausible, but at this point inconclusive. In this paper I follow the most conservative classi-
fication of Epps and Bolaños (2017).
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 321

Tukanoan languages have a system of differential object marking, where the presence
of the object marker interacts with definiteness and specificity. The same marker is also
used to code a wider range of roles, but without the differential aspect. In terms of index-
ing on the verb, Tukanoan languages mark the subject (S/A) but not the object.
The presence of the applicative marker in Tukanoan languages licenses an argu-
ment that is marked with the case marker -re. The construction does not promote an oth-
erwise oblique participant in that there seems to be no alternative way to express bene-
ficiaries (on the interpretation of the applied object in Tukanoan languages, see Section
3.4). With respect to the flagging behavior (there is no object indexing in Tukanoan lan-
guages, so the discussion will be restricted to flagging), the applied object does not always
behave entirely similar to direct objects, however. This can be illustrated with data from
Kotiria. Kotiria objects can be marked with the suffix -re. However, as mentioned above
for Tukanoan more broadly, direct objects are not always flagged in Kotiria: non-flagged
direct objects are less specific and definite than flagged objects. Zero-marked direct
objects in Kotiria are only possible in immediately preverbal position.

(8) Kotiria (East Tukanoan; Stenzel 2013: 331)

a. hí-phiti-ro chʉ́ adá-ta-ra
cop-coll-sg food get-come-vis.ipfv.2/3
‘Everyone brings a lot of food.’
b. tí-da dá-sá chʉ́ a-re chʉ́ yoá-ra
anph-pl get-mot.inside food-obj eat do/make-vis.impfv.2/3
‘They take the food inside and eat.’

The marker -re is also used for flagging R participants in ditransitive constructions,
for temporal adjuncts, and spatial complements of motion predicates that have been
established in the preceding discourse. Example (9) illustrates the inherently ditransi-
tive verb wa- ‘give’, which takes two objects, both marked with -re.

(9) Kotiria (East Tukanoan; Stenzel 2013: 161)

mu’u=yahíripho’na-re yu’u-re wa-ga
2sg.poss=heart-obj 1sg-obj give-imp
‘Give me your heart.’

The recipient in these constructions with inherently ditransitive verbs is always

marked with -re. In other words: there is no pattern of differential case marking for
indirect objects. The same pattern can be observed with applied objects. Applied objects
therefore behave as R participants in ditransitive constructions. Inherently intransi-
tive verbs with the applicative marker -bosa are not discussed in Stenzel (2013), whose
description of -bosa as “increas[ing] the valency of the independent verb from one to
two object arguments” suggests that -bosa only combines with transitive bases.
322 Rik van Gijn

To the extent that I have been able to ascertain, this non-differential marking pattern
of the applied object as well as indirect objects of underived ditransitives is found almost
throughout the Tukanoan languages of the sample, independently of whether they
follow the -k(h)a(i) pattern or the -bosa/basa pattern. Tukano marginally allows differen-
tial object marking for indirect objects. In Example (10), the morpheme dɨ ̃kɨ ‘each’ makes
the indirect object participant generic, according to Ramirez (1997: 227).

(10) Tukano (Tukanoan; Ramirez 1997: 227)

dãâ-dɨ ̃kɨ o’ô-ya teé-de
they-each give-imp those.things-obj
‘Give those things to each of them.’

In addition, Example (6) above suggests that in Carapana, differential case marking is
also available for applied objects. It is not clear from the description in Metzger (1981)
whether this is true more generally for R participants of monomorphemic ditransitives.
The Arawakan languages generally have no core case marking, but Yukuna marks
R participants in ditransitive constructions with the oblique enclitic =jló, illustrated
in (11).

(11) Yukuna (Arawakan; Lemus Serrano 2020: 68)

na=a'-chá píño [ri=jló]OBL [kujnú]O.
3pl=give-pst again 3sg.nf=to cassava
‘They gave cassava to him again.’

The applied objects pattern with direct objects in being unmarked. The applicative
seems to combine with intransitives only, forming transitives, so in that sense, a regular
transitive syntactic frame is associated with the applicative.

(12) Yukuna (Arawakan; Lemus Serrano 2020: 68)

Chúwa tâ kája wa=i'jna-ñáa pi=ikhá.
now emph already 1pl=go-appl 2sg=pro
‘We’re leaving you now.’

The Chicham languages are strictly nominative-accusative, and employ both case
marking (including an accusative marking) and indexing (subject and object) to code
grammatical relations. Object indexing is only available for speech act participants
in all three Chicham languages investigated here, third person objects are unmarked.
Shiwiar and Wampis have portmanteau markers for certain scenarios.6 Both direct and

6 Shiwiar uses portmanteau suffixes for all scenarios involving SAP objects except first person singular,
Wampis has portmanteau markers for local scenarios involving a second person object.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 323

indirect objects can control object indexing. With ditransitive verbs, the verb will mark
an SAP object if available. This generally is the R participant.

(13) Aguaruna (Chicham; Overall 2007: 316)

mi-na su-hu-sa-ta
1sg-acc give-1sg.obj-att-imp
‘Give it to me!’

The presence of an applicative marker makes the object indexing slot available for the
applied object, as can be seen in (14) from Shiwiar.

(14) Shiwiar (Chicham; Kohlberger 2020: 298)

papá-r=ka hɨ ̃ã́=n naha-t-ru-á-mia-ji
father-1sg.poss=top house=acc make-appl-1sg.obj-pfv-rem.pst-3.decl
‘My father made me a house.’

Example (15) illustrates that overtly expressed applicative participants are marked
with the accusative.

(15) Aguaruna (Chicham; Overall 2007: 306)

amitʃi mi-na ataʃu-na yu-hu-tu-a-ɨ ̃
fox 1sg-acc chicken-acc eat-appl-1sg.obj-hiaf-3:pfv
‘A fox ate my chicken.’

On the other hand, Peña (2015) reports that the applied object optionally carries a ben-
efactive marker in addition to the accusative marker:

(16) Wampis (Chicham; Peña 2015: 591)

[mama=na] [ui-nau=na] hii-tu-ma-ji
manioc=acc 1sg-ben=acc pull.out.pfv-appl-rec.pst-3.pst+decl
‘He harvested manioc for me.’

There are also cases where the applicative does not trigger object marking on the verb.
This is the case in semi-reflexive interpretations in Wampis, where the applicative par-
ticipant is identical to the subject.

(17) Wampis (Chicham; Peña 2015: 590)

napi=na karama-ru-u a-sa-nu ʃiitra-ʧau
boa=acc dream.of-appl-nmlz cop-sub-1sg.ss good-neg.nmlz
nikapɨ-a-ha-i
feel-ipfv-1sg-decl
‘Having dreamed of the boa (to my detriment), I feel bad.’
324 Rik van Gijn

As can be seen, the applicative here does not license a 1sg object marker on the verb,
and so it does not seem to increase the valency of the (already transitive) verb karama
‘dream of’. It is not clear to what extent this type of construction exists in the other
Chicham languages. Kohlberger (2020: 299) discusses similar examples for Shiwiar (i.e.
without object indexing), but they include a reflexive marker (a construction that also
exists in Wampis, see e.g. Peña 2015: 890).
Like the Chicham languages, the Kawapanan languages of the sample display both
person indexing on the verb and case marking on nominal arguments, although the
families differ in the more detailed aspects of their systems. In Shiwilu, both arguments
are obligatorily marked on the verb by means of a portmanteau indexing suffix, with
the exception of scenarios with a third person object, which take the unipersonal subject
markers also found on intransitive verbs. Ditransitive verbs take the portmanteau
morphemes, which refer to the A and R arguments, leaving the T argument unmarked
(Valenzuela 2016: 520). Shawi indexing follows a system of separable subject and object
markers. Subject markers also differ according to tense and mood (Rojas Berscia 2019).
The object index in transitive clauses seems to refer to the R argument.

(18) Shawi (Kawapanan; Bourdeau 2015: 26)

kiya-ri-n-su pei keta-ra-i-nkema.
1pl.exclm-erg-?-def house give-ind-1.pl.excl-2.pl.o
‘We will offer you a house.’

In terms of case marking, the Kawapanan languages have an optional marker of A par-
ticipants, =ler in Shiwilu, -ri in Shawi, whose appearance is conditioned by disambigua-
tion requirements and pragmatic status (Valenzuela 2015: 42–45).
Applicative constructions in Shiwilu follow transitive or ditransitive patterns in
that the applied object is indexed on the verb:

(19) Shiwilu (Kawapanan; Valenzuela 2016: 528, 533)

a. ñiñi’wa=pen ek-pekla-llun. /✶pekla-llun
dog=poss.2sg soc.caus-produce.noise-nfut.3sg>1sg
‘Your dog barked at me.’
b. Papa=wek tek-susu-pa’-llun wila=wek.
father=poss.1sg caus-grow-appl-nfut.3sg>1sg child-poss.1sg
‘My father helped me raise my child.’

Applicatives that combine with ditransitive roots mark the applied (fourth) argument
on the verb:
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 325

(20) Shiwilu (Kawapanan; Valenzuela 2016: 530)

Pidir uku’la-lapi-llun kusher=wek.
Fidel sell-appl-nfut.3sg>1sg pig=poss.1sg
‘Fidel sold my pig (to sb. else) and left leaving me behind.’
(for example, acted without my consent and left with the money)

At least some of the applicatives also allow for passivization, confirming the object
status of the applicative:

(21) Shiwilu (Kawapanan; Valenzuela 2016: 528, 535)

a. Kwa ek-pekla-pi=ku nuka’-ka
1sg soc.caus-produce.noise-nmlz=pred.1sg cop-1sg
(ñini’wa=pen=lek).
dog=poss.2sg=instr
‘I was barked at (by your dog).’
b. Imicha anu’-tu-pi nuka’-a tunla=lek.
Emérita fall-loc-nmlz cop-3sg worm=instr
‘Emérita was fallen on by the worm.’

Some markers (-i, -wa, -nan, -willi) require the presence of the valency modifier -tu, a
construction termed double applicative marking by Valenzuela (2016).7 The marker -tu
can either increase or decrease valency, depending on the verbal8 host.
Example (22) illustrates such a double applicative construction with affective -i.
One way to interpret this construction is that the applicative morpheme that precedes
-tu functions as its specifier, not only forcing a valency increase interpretation, but
guiding the interpretation of the role of the applied object.

(22) Shiwilu (Kawapanan; Valenzuela 2016: 536)

a. wila=wek=sha ipa’ wichi’-lli
child=poss.1sg=dim already fall.asleep-nfut.3sg
‘My little grandson already fell asleep’.
b. Wila=wek=sha ipa’ wichi’-i-tu-llun. /✶wichi’-i-llun
child=poss.1sg=dim already fall.asleep-appl-vm-nfut.3sg>1sg
‘My little grandson already fell asleep to my benefit.’

The situation for Shawi is less clear. Example (23) suggests that the sociative causative
adds an object argument that can control person indexing on the verb. Semantically

7 Valenzuela regards the valency modifier -tu to be synchronically different from the locative marker
-tu, although they are historically related,
8 The marker also attaches to adjectives or nouns to form verbs.
326 Rik van Gijn

speaking, these sociative causatives can be argued to be either causatives or applic-

atives, depending on what is regarded as the underlying (in the sense of the event
without the valency-increasing marker): ‘I work’ or ‘you work’ in Example (23) below.

(23) Shawi (Kawapanan; Bourdeau 2015: 25)

Ka-ri=nke ichi-saka-te-ra-(w)-nke.
1-erg=2.obj soc.caus-work-val-ind-1-2.obj
‘I make you work (and work with you).’

Barraza de García (2005), focusing on the Sillay variety of Shawi, shows that the mul-
ti-functional marker -tɨ in its applicative use licenses object indexing of the applied
argument on the verb.

(24) Shawi (Kawapanan; Barraza de García 2005: 175–176)

mašu-ri imu-tɨ-r-in-nkɨn
old.man-erg/top vomit-appl-ind-3-2
‘The old man has vomited on you.’

Zaparoan languages Arabela and Iquito do not to mark objects, either on the verb or
on the noun, although Arabela has some different pronominal forms, depending on the
function. For peripheral functions, postpositions or case suffixes are used, which form a
large group. The lack of morphosyntactic marking of core arguments makes it harder to
show the morphosyntactic effects of the applicative. The main characteristic of interest
is that in applicative constructions, the applied object bears no case marker or adpo-
sition, suggesting it is part of the core, as can be seen for the applied objects morejaca
‘yuca’ in (25) and nuú ‘him’ in (26).

(25) Arabela (Zaparoan; Rich 1999: 55–57)

cua morejaca tiurii-tia-ree-nijia.
1sg yuca trip-appl-pfv-1sg
‘I tripped with my load (yuca).’

(26) Iquito (Zaparoan; Michael 2009: 155)

pɨyɨ́ɨni yahuɨ́ɨni=jina nu=árɨɨ-yaáriquɨ tií
all day=loc 3sg=pass.by-rem.pst.ipfv there
nu=puhuaajɨɨ-nii-yaáriquɨ=na nuú.
3sg=whistle-appl-rem.pst.ipfv=rep 3sg
‘Whenever he passed by there, it (a forest spirit) whistled at him.’

Yagua indexes the gender or person/number information of subject arguments on

the verb when no subject NP is expressed, or if the subject NP is expressed post-verbally
(and there are different sets for animate and inanimate subjects). Objects are generally
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 327

marked by enclitics,9 which immediately precede the object NP, but prosodically attach
to any constituent that precedes them. When no object NP is expressed, the object clitics
tend to be clause-final. Yagua does not have any core case markers. Payne (1985a: 271)
discusses the marker -ta, which is most frequently used to license semantic comitatives
or instruments. The same marker is also found with nouns, where it marks the comita-
tive/instrument role. The following pair shows the uses of -ta as a postposition (27a) and
as an applicative marker (27b).

(27) Yagua (Peba-Yaguan; Payne 1985: 273–274)

a. sa-jᶖchitiy jᶙmurutᶏᶏ-ta-rà tíͅ íͅ staͅaͅsuuy
3sg-poke machete-com/instr-inan ball
‘He pokes the ball with a machete.’
b. sa-jᶖchitiy-ta-rà jᶙmurutᶏᶏ-rà tíͅ íͅ staͅaͅsuuy
3sg-poke-appl-inan machete-inan ball
‘He pokes the ball with a machete.’

As can be seen in this example, the presence of the applicative marker on the verb
in (27b) licenses the presence of a non-case-marked instrumental NP, as well as the
presence of the inanimate marker -rà, immediately preceding its referent jᶙmurutᶏᶏ,
effectively creating a double-object construction. This is the same type of construction
as an underived ditransitive (Doris L. Payne, 1985a: 274).
Kakua indexes subjects on the verb by means of proclitics, and marks objects (both
direct and indirect) on the NP by means of a case clitic =diʔ. This case marker is com-
patible with a wide range of semantic interpretations, including patient, benefactive,
malefactive. As is common in the area, Kakua marks objects differentially, although
the exact parameters governing the presence or absence of the object case marker are
not entirely clear (see Bolaños 2016: 191–205 for discussion). The applicative in Kakua,
restricted to imperatives, does allow for marking of the applied object with the object
marker, but since benefactive is part of the semantics of the object marker, and since
there seems to be no alternative base construction, it is unclear to what extent this con-
stitutes a bona fide applicative construction.

(28) Kakua (Kakua-Nukak; Bolaños 2016: 338)

kăn=diʔ ʔɨbʔ-áʔbuhú
3sg.m=obj take.out-imp.ben
‘Take (it) out for him.’

9 It is not clear what exactly governs or licenses the absence of object enclitics. Doris L. Payne (1985a:
46–48) suggests that it may be determined by specificity and individuation of the object argument, but
acknowledges that there are situations with specific, individuated object participants, which neverthe-
less do not trigger object clitics.
328 Rik van Gijn

However, if we compare (28) to an imperative of a three-place verb like ‘give’, the recipi-
ent can also be marked with the object case, but no special imperative form is required,
suggesting that -áʔbuhú does extend the valency of two-place verbs to three, which then
behaves in the same way as an underived ditransitive.

(29) Kakua (Kakua-Nukak; Bolaños 2016: 189)

Hãmu=diʔ ma=nɨm=diʔ ma=wɨ̌ ʔ-ɨ́
̃
Hamu=obj 2sg-daughter=obj 2sg=give-imp
‘Give Hamu your daughter!’

Like Kakua and other languages of the area, Hup also has a system of differential
object marking. The case marker for objects is -ăn, which obligatorily marks human
objects, while non-human animates are optionally marked, and inanimates remain
unmarked. Case marker -ăn is also used for beneficiaries and recipients, but in that
role there is no differential marking (Epps 2008: 166).10 The suffix -ʔũh in Hup adds an
animate participant. The added participant is most commonly a recipient, benefactive
or malefactive (Epps 2008: 500). The participant introduced in this way is marked as an
object if overtly expressed.

(30) Hup (Nadahup; Epps 2008: 501)

hɨd (tɨ́h-ăn) g’ãʔ-ʔṹh-ṹy
3pl 3sg-obj be.suspended-appl-dyn
‘They’re lying in (someone’s) hammock.’ (i.e., without his knowledge or approval)

Summarizing the patterns discussed above, applied objects generally follow

marking patterns that exist for direct and indirect objects of underived verbs, with
a tendency for increased morphological markedness of applied objects in some lan-
guages, and decreased markedness in others:
– In systems with differential object marking, the applied objects tend to always be
marked (Tukanoan, Hup, possibly Kakua).
– In languages with object indexing on the verb with a single slot available, the
applied object gets preference over other potential controllers (Chicham, Kawap-
anan).
– In languages without core case marking, applied objects are zero-marked, unlike
oblique arguments (Yagua, Zaparoan, Yukuna).

10 There is also a historically related oblique marker -an, which marks several kinds of locational rela-
tions. This marker shows a close resemblance to the object marker, the only difference being the rising
tonal pattern in the object marker. It is not always clear which formative is used (Epps 2008: 184).
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 329

There is some variation in terms of the transitivity of the host, which is summarized
in Table 4—to the extent that this could be determined on the basis of the written
sources (these are the same as the sources given in Table 2 above, of which Table 4 is
an extension).

Table 4: Overview of applicatives in the sample in terms of form, possible hosts, and morphosyntactic marking.

Language Form host morphosyntactic marking

Desano ( Tukanoan) -basa transitives object case

Kotiria (Tukanoan) -bosa transitives object case
Tukano (Tukanoan) bosa transitives object case
Barasano (Tukanoan) -bosa transitives object case
Carapana (Tukanoan) -boja transitives object case
Kubeo (Tukanoan) -ka transitives object case
Siona (Tukanoan) -ka transitives object case
Secoya (Tukanoan) -cai transitives object case
Koreguaje (Tukanoan) -khaj transitives object case
Yukuna (Arawakan) -ñaa/-ñai/-ña intransitives zero
Aguaruna (Chicham) -hu/-tu transitives, intransitives object case, object index
Wampis (Chicham) -hu/-tu transitives, intransitives object case, object index
Shiwiar (Chicham) -hu/-tu transitives, intransitives object case, object index
Shiwilu (Kawapanan) ek- transitives, intransitives object index
Shiwilu (Kawapanan) -lapi intransitives object index
Shiwilu (Kawapanan) -pa transitives object index
Shiwilu (Kawapanan) -tu intransitives object index
Shiwilu (Kawapanan) -tu11 intransitives, transitives object index
Shiwilu (Kawapanan) -i transitives object index
Shiwilu (Kawapanan) -wa intransitives object index
Shiwilu (Kawapanan) -nan transitives, intransitives object index
Shawi (Kawapanan) ichi- transitives, intransitives object index
Shawi (Kawapanan) -të transitives, intransitives object index
Arabela (Zaparoan) -t(i)a intransitives zero
Iquito (Zaparoan) -nii intransitives zero
Hup (Naduhup) -ʔṹh transitives (intransitives) object case
Kakua (Kakua-Nukak) -áʔbuhú transitives object case
Yagua (Peba-Yaguan) -ta transitives, intransitives zero

11 There are two homophonous and diachronically connected markers -tu in Shiwilu. The second is
classified as a valency modifier by (Valenzuela 2016), but this marker has, among its uses, applicative
(see also Section 3.4).
330 Rik van Gijn

3.4 The semantics of the applicative constructions in the sample

The most common semantic interpretation of the applicatives in the area is benefactive/
recipient, which is also the most common semantic interpretation world-wide (Polinsky
2013), a further common interpretation is malefactive. Benefactive seems to be the only
possible interpretation in most Tukanoan languages. An exception is West-Tukanoan
Koreguaje, whose marker -khaj adds an argument role to the verb semantics that can
be interpreted as beneficiary (31a) or maleficiary (31b).

(31) Koreguaje (West Tukanoan; Cook and Criswell 1993: 72–73)

a. ǰɨʔɨ mɨʔɨ-re ǰoʔo-kha-sa
I you-obj work-ben-des.sg
‘I will work for you.’
b. khani-sõ-ɨ-na hamu-ǰai-pi mɨo-ra’-ǰo-rɨ
sleep-ints-seq.m.sg-ds armadillo-jaguar-src finger-dim-clf-dim
ãni-sõ-khai-si-na-‘-me ǰɨ’ɨ-re
eat-ints-ben-pfv-nmlz.pl-vblz-pl 1sg-obj
‘While I was sleeping, the dogs ate my little finger (to my detriment).’

The situation as in Koreguaje, with both benefactive and malefactive as possible inter-
pretations is common in the area. The applicative in Chicham languages, for instance,
can have both readings, depending on the context, as can be seen in e.g. Examples (14)
and (15) above, repeated here for convenience as (32) for benefactive, and (33) for mal-
efactive.

(32) Shiwiar (Chicham; Kohlberger 2020: 298)

papá-r=ka hɨ ̃ã́=n naha-t-ru-á-mia-ji
father-1sg.poss=top house=acc make-appl-1sg.obj-pfv-rem.pst-3.decl
‘My father made me a house.’

(33) Aguaruna (Chicham; Overall 2007: 306)

amitʃi mi-na ataʃu-na yu-hu-tu-a-ɨ ̃
fox 1sg-acc chicken-acc eat-appl-1sg.obj-hiaf-3:pfv
‘A fox ate my chicken.’

The Shiwilu applicative -i, in combination with -tu, yields a benefactive or malefactive
interpretation:12

12 These are considered separate morphemes because -tu also combines with other applicative mor-
phemes.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 331

(34) Shiwilu (Kawapanan; Valenzuela 2016: 536–537)

a. wila=wek=sha ipa’ wichi’-i-tu-llun.
child=poss.1sg=dim already fall.asleep-appl-vm-nfut.3sg>1sg
‘My little grandson already fell asleep to my benefit.
b. Idu chimin-i-tu-llun.
Eleuterio die-appl-vm-nfut.3sg>1sg
‘Eleuterio died to my detriment.’

The applicative marker -lapi has a default interpretation of a relinquitive (leaving

behind), but can have benefactive or malefactive overtones. For some verbs, such as
lun- ‘speak’, the only available interpretations of -lapi are malefactive or benefactive.

(35) Shiwilu (Kawapanan; Valenzuela 2016: 531)

lun-lapi-llun.
speak-appl-nfut.3sg>1sg
‘S/he spoke for me (in my favor, instead of me).’ / ‘S/he accused me.’
✶
‘S/he spoke and left me behind.’ / ✶‘S/he spoke having left me behind.

The imperative applicative in Kakua seems to have only a benefactive interpretation

(28 above). A dative (benefactive/recipient) and/or malefactive interpretation is found
for the only applicative marker in Hup. Compare (30) above with (36).

(36) Hup (Nadahup; Epps 2008: 501)

hɨd-ăn hɔ̃ṕ tɨh nɔʔ-ʔṹh-ṹy
3pl-obj fish 3sg give-appl-dyn
‘He’s giving them fish (as a service to someone else, probably the owner of the
fish).’

For Arawakan Yukuna, malefactive (and connected to that relinquitive) are the
only possible interpretations.

(37) Yukuna (Arawakan; Lemus Serrano 2020: 76)

ri=éjẽ ́ kujnú anó'-ña-cha na=i'michá.
3sg.nf=toward cassava wet-appl-pst 3pl=rem.pst
‘Then their cassava got wet on them (i.e. to their detriment).’

The applicative marker in Iquito can also possibly be connected to this group. Since
there is no dedicated description of this applicative marker, the interpretation of its
semantics must be based on example sentences, of which there are few. The semantic
interpretation seems to hinge around the notion of benefactive/goal (see [26] above),
but allows for some freedom of interpretation, as shown in (38), where the combination
carii ‘look’ with -nii yields different (though relatable) interpretations.
332 Rik van Gijn

(38) Iquito (Zaparoan; Lai 2009: 299, 301)

a. canɨɨca pɨ-mayasiini carii-nii-rɨɨ?
who 1pl.incl-play.inf look-appl-mom.pfv
‘Who will direct our game?
b. carii-nii nuu!
look-appl 3s
‘Take care of him!’

A second, smaller group of markers hinge around (sociative) comitative semantics. For
Arabela, Rich (1975: 2) describes a suffix -ta, which marks a sociative causative (but
termed applicative in Rich 1999), introducing an accompanying person or thing for the
patient of a transitive or the S of an intransitive verb, with a range of interpretations.
Rich (1999: 55–57) lists three types of interpretation, which he terms passive accom-
paniment (39a), temporary condition for the subject participant (39b), and contents of
subject or object participant (39c).

(39) Arabela (Zaparoan; Rich 1999: 55–57)

a. cua morejaca tiurii-tia-ree-nijia.
1sg yuca trip-appl-pfv-1sg
‘I tripped with my load (yuca).’
b. Najaaca cuso-ta-a-nijia.
flu be.sick-appl-cnt-1sg
‘I am ill with the flu.’
c. Cana-qui-ti-jia canaa
who-exclm-inter-exclm 1pl.excl
caashoque ti-ni-jio-ta-ree-jaj?
bag fall-caus-vpl-appl-pfv-emph
‘Who knocked over our bags (with content)?’

This is semantically rather close to some of the readings of -ta in Yagua:

(40) Yagua (Peba-Yaguan; Doris L. Payne 1985ª: 272)

a. sa-ya-ta jį́įt́ a-níí juvąda
3sg-go-appl disc.prt-3sg kill[noun]
‘He goes with his kill.’
b. sa-tiryǫ́ ǫ́-ta-rà
3sg-lie:down-appl-inan
‘He lies down with it (e.g. a book).’

Shawi has a sociative causative construction, marked with the prefix ichi-, which serves
a similar function as Shiwilu ek-:
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 333

(41) Shawi (Kawapanan; Rojas Berscia 2013: 92)

ka ichi-nansa-r-awe ina
I soc.caus-dance-assist-1sg him
‘I make him dance (but I also participate in the dancing).’

(42) Shiwilu (Kawapanan; Valenzuela 2016: 527)

Rubisha ek-lansa’-llun (kwa)
Robertina soc.caus-dance-nfut.3sg>1sg 1sg
‘Robertina asked me to dance and we both danced.’

A more idiosyncratic interpretation of ek- in Shiwilu can be translated as ‘wearing’, the

applied object being a piece of clothing associated with one of the participants of the
action. This is reminiscent of the examples in Arabela (39) and Yagua (40) above.

(43) Shiwilu (Kawapanan; Valenzuela 2016: 528)

a. Atushupi chimin-lli
Augusto die-nfut.3sg
‘Augusto died.’
b. Atushupi ek-chimin-lli kutun=nen.
Augusto soc.caus-die-nfut.3sg>3sg clothes=poss.3sg
‘Augusto died in his clothes (i.e., in the clothes that he is wearing).’

A further applicative marker of Shiwilu, -pa, which can be used to indicate a participant
next to whom an action is performed, can also take on more comitative and assistive
meanings, as shown in (44).

(44) Shiwilu (Kawapanan; Valenzuela 2016: 533)

a. Malallina nerku’-lli nun=kekla.
Magdalena row-nfut.3sg canoe=abl
‘Magdalena rowed from the canoe.’
b. Malallina nerku’-pa’-llun.
Magdalena row-appl-nfut.3sg>1sg
‘Magdalena helped me row (i.e., we both rowed).’

Semantic extensions of Shiwilu ek- and Yagua -ta are closer to a dative-like interpreta-
tion, akin to the semantics of Iquito -nii.

(45) Shiwilu (Kawapanan; Valenzuela 2016: 528)

ñiñi’wa=pen ek-pekla-llun
dog=poss.2sg soc.caus-produce.noise-nfut.3sg>1sg
‘Your dog barked at me.’
334 Rik van Gijn

(46) Yagua (Peba-Yaguan; Doris L. Payne 1985a: 278)

sa-jųnúúy-ta-níí dee-un
3sg-look-appl-3sg dim-clf:anim:sg
‘She watches/takes care of the boy.’

The Yagua marker in addition can yield instrument interpretations, as illustrated by

(27) above.
A third group that occurs in more than one family are locatives. Shiwilu has an
applicative marker -tu, which promotes a locative argument to core argument status:

(47) Shiwilu (Kawapanan; Valenzuela 2016: 534)

a. Peksek wichi’-lli.
peksa’=ek wichi’-lli
bed=loc sleep-nfut.3sg
‘He slept in the bed.’
b. Peksa’ wichi’-tu-lli.
bed sleep-appl-nfut.3sg>3sg
‘He slept in the bed.’

A further applicative marker in Shiwilu, -wa, in combination with -tu, marks the goal
of a movement:

(48) Shiwilu (Kawapanan; Valenzuela 2016: 540)

napi’=ima ala’=sa’ shaya’ Kukama’-lun=ima
long.ago=rep one=just man’s.older.sister Cocama-f=rep
Tandek=k ñi-apa-sik=ima
Marañón.river=loc exist-cnt-dur.3sg=rep
pa’-wa-tu-lli ala’=sa’ wapur=ler
go-appl-vm-nfut.3sg>3sg one=lim boat=erg
‘Long ago a boat came towards a woman, a Cocama woman, who was living on
the (banks of the) Marañón (river).

A locative interpretation is also possible in Chicham languages, although this particu-

lar instance of the applicative marker is best classified as a lookalike. In Shiwiar, with
verbs of motion, the applicative marker can introduce a locative argument, as exem-
plified in (49).

(49) Shiwiar (Chicham; Kohlberger 2020: 299)

ukú-r-ki-ár-mia-ji
leave-appl-pfv-pl-rem.pst-3.decl
‘They left her there.’
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 335

The verb root ukú ‘leave’ in (49) is transitive; the applicative shifts the focus to a loca-
tion. Kohlberger stresses that it is unclear that the applicative increases valency in these
cases, since there is never an explicit NP referring to the location, and third person
objects are not formally marked on the verb.
This is reminiscent of the situation in the Arawakan language Tariana. Strictly speak-
ing, Tariana does not have an applicative construction, but certain uses of the causative
-ita can be classified as applicative lookalikes. When applied to transitive verbs (except
a small subset), the marker combination -ita indicates that a peripheral participant must
be present in the clause (although it does not seem to be treated as a core argument
morphosyntactically). The precise interpretation of this peripheral participant depends
on the verb semantics, but one of the possible interpretations is a locative (50), other
interpretations are dative (addressee/benefactive), instrument, or purpose.

(50) Tariana (Arawakan; Aikhenvald 2000: 166)

a. diwhida na-pisa na-pala-pidana
3sg.nf+head 3pl-cut 3pl-put-rem.pst.inf
‘They cut his head and put (it) somewhere.’
b. ita-whya hi-nuku pi-pale-ta
canoe-clf dem:an-top.non.a/s 2sg-put:caus-caus
‘Put the canoe here.’

In Example (50a), the verb stem pala- without the causative means ‘to get’ or ‘to put’;
with the causative marker13 the stem means to put something in a particular location
(50b).
Shiwilu has a number of further applicatives or further readings of the applicative
markers discussed above. The first marker (-tu, homophonous with the locative applica-
tive) to be discussed is not classified as an applicative marker by Valenzuela (2016),
because it clearly has uses that are not applicative. Depending on the verb14 it attaches
to, it may have valency-decreasing or -increasing effects. The latter can be interpreted
as applicative constructions that add a (direct or indirect) object. Example (51a) shows
the valency-increasing effect of -tu, adding an addressee to the verb lamapu’ ‘scream’,
changing the interpretation to ‘scream at’. In (51b), the same marker descreases the
valency of the transitive verb apu’- ‘abandon’, to derive an intransitive interpretation
‘disappear, go away’, where the former direct object is expressed as an oblique.

13 The causative marker i-ta is bimorphemic, with the -i (here merged with the stem) representing the
causation, and -ta an increased affectedness of the P participant (Aikhenvald 2000: 158–159)
14 The marker also attaches to adjectives or nouns to form verbs.
336 Rik van Gijn

(51) Shiwilu (Kawapanan; Valenzuela 2016: 524–525)

a. nana shaya’ sudinen lamapu’-tu-lli.
that man’s.sister husband:poss.3sg scream-vm-nfut.3sg>3sg
‘The woman screamed at her husband.’
b. Iñika apu’-tu-lli pidek=ñiklan.
Inés abandon-vm-nfut.3sg house=poss.3sg:abl
sudin apu’-lli.
husband:poss.3sg abandon-nfut.3sg>3sg
‘Inés disappeared from her house. She abandoned her husband.’

Shawi has a cognate marker -të (pronounced /tɨ/ or possibly /tɯ/, with allomorph -ta in
combination with progressive verb semantics). Hart (1988) describes three functions of
the suffix -të: valency-increasing (52a–b), deriving a personal from an impersonal verb
(52c), and valency-decreasing (52d). Example (52b) can be classified as an applicative
use of -të.

(52) Shawi (Kawapanan; Hart 1988: 269)

a. amarin ‘He bathes.’ vs. amatërin ‘He bathes someone.’
b. a’parin ‘He sends it.’ vs. a’patërin ‘He sends it to someone.’
c. të’narin ‘It gets cold.’ vs. të’natërin ‘He gets cold.’
d. nohuirin ‘He is angry with someone.’ vs. nohuitërin ‘He is angry.’

Interestingly, Rojas Berscia (2013: 50) indicates that, for the Balsapuerto dialect he was
studying, most of the examples mentioned by Hart (1988) could not be corroborated.
Instead, he found that the suffixes of the same form were used as tense markers.
The relinquitive applicative marker -lapi in Shiwilu is also found in constructions
that are interpreted as comparative (surpassing) or prioritive (beating someone to it),
which are possibly semantic extensions of the relinquitive. There is a further applica-
tive marker -nan, which marks applied objects in the presence of whom an action was
performed, sometimes with malefactive overtones. The same marker can also be inter-
preted as referring to a participant who is spared, in contrast to other participants, who
are affected by the action. Valenzuela (2016: 542) notes that this applicative interpreta-
tion is unique.
Summarizing, there are a number of recurring semantic roles assigned to the applied
object in the area. The most common ones are benefactive and malefactive, with comita-
tive-like interpretations and locations as two further recurring, but clearly less common
groups.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 337

4 The areality of applicative constructions

in the northwest Amazon
After having reviewed the morphology, syntax, and semantics of the applicative con-
structions found in the NWA, we can return to the question to what extent the distribu-
tions found can be explained by making reference to contact.
A first point to be made here is that the area, in some ways, seems to be the opposite
of an “applicative area”. Only about half of the languages in the sample under investi-
gation in this paper have applicative constructions. In addition, of the languages that
do have applicatives, the most common situation is that there is only one applicative
marker, Shiwilu being the exception. The Quechuan and Arawakan languages of the
area, moreover, mostly or entirely lack applicatives, unlike other languages belonging
to the same family, but spoken in other areas. For Quechuan languages, as mentioned,
this may be a contact effect (substrate, possibly from Barbacoan), for Arawakan this is
less clear. Two Arawakan languages in the area do have applicative or applicative-like
markers, Tariana and Yukuna.
In terms of their morphology, the applicative markers in the NWA are overwhelm-
ingly suffixal, with a few exceptions (prefixes in Kawapanan); with respect to the spe-
cific forms, the most conspicuous commonalities are family-internal, in particular for
Tukanoan and Chicham, suggesting established patterns prior to diversification. There
are a few other similar forms, which may be discussed in the light of the idea that a
marker -ta spread through the area (Crevels and Voort 2020). A few markers discussed
above expressing applicative or applicative-like constructions have a form that can
broadly be connected to the form [ta].

Table 5: Forms similar to [ta] in the sample.

form language meaning

-tu Aguaruna benefactive, malefactive, locative
-tu Wampis benefactive, malefactive, locative
-tu Shiwiar benefactive, malefactive, locative
-tu Shiwilu locative
-t(i)a Arabela comitative
-ta Yagua comitative, instrument
-ita Tariana benefactive, malefactive, locative,
instrument, purpose (lookalike)

At first sight, the form-meaning pairs in Table 5 look promising: there is certainly formal
similarity, and this seems to be coupled to some degree of functional overlap as well.
However, although there certainly may be contact-induced aspects to be discerned, the
formal similarities seem to be mostly accidental. The Chicham form -tu is an allomorph
338 Rik van Gijn

of -hu (Aguaruna) or -ru (Wampis, Shiwiar). This is the same allomorphy as the first
person singular object, lexically determined by the same group of verbs. Although the
diachrony of this allomorphy is unclear (Overall 2007: 322–324), it is obvious that there
is a diachronic connection, which points to an internal development of the form -tu, at
least since the time of diversification. Of course, it is possible that the allomorphy (an
uncommon feature in Chicham languages) is contact-induced, caused by the borrowing
of a form at a stage before the diversification of the Chicham languages, which, given the
similarities between the Chicham languages, does not lie too far back in the past.
The locative applicative -tu in Shiwilu has a diachronic connection to the valency-
adjusting suffix -tu (-të in Shawi). This suffix can manipulate the valency of a verbal root,
with either decreasing or increasing effects. The interpretation of the marker depends on
the verb root it attaches to, and it is therefore difficult to ascribe any specific semantics
to it. The presence of a cognate in Shawi suggests the presence of the marker in a stage
before the split between Shawi and Shiwilu.
Given its form, it is enticing to regard the Tariana applicative in the light of the
spread of -ta forms, discussed in Crevels and Voort (2020). It must be said, however, that
the applicative reading is only one of the functions of -ita, and probably a secondary one,
as its main function is causative. As a causative marker, forms related to -ita are found
throughout the Arawakan family, and seems to be inherited from proto-Arawakan, see
Aikhenvald 1999: 90). For the forms in the sample of this paper, see Table 6.

Table 6: Causative morphology in northwestern Arawakan languages.

Language Causative morpheme Source

Achagua -da Wilson and Levinsohn (1992)

Tariana -ita Aikhenvald (2003)
Piapoco -ida Klumpp (2019), Reinoso Galindo (2002)
Resígaro -tú Allin (1976)
Warekena -ta Aikhenvald (1998)
Yukuna -ta/-chi Lemus Serrano (2020)

That leaves us, as far as formal borrowing is concerned, with the case of Yagua and
Arabela, which was discussed in Section 3.2, based on the observations made in Payne
(1985b) and Wise (2002).
It is, however, possible that some of the language-internal developments were influ-
enced by language contact, and that formal similarities may have helped these pro-
cesses along. For instance, the development of a locative interpretation of the valency
modifier in Shiwilu may have happened under the influence of the presence of a hom-
ophonic applicative suffix in the Chicham languages, and the extension of the causative
in Tariana may have been influenced by the presence of similar sounding applicative
markers in neighboring languages. Likewise, the extension of the Koreguaje applicative
to include malefactive may have been contact-induced (with perhaps the Chicham lan-
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 339

guages as the most likely source). This is, however, hard to prove or disprove. We know
that Tariana was influenced particularly by East Tukanoan languages, which have no
applicative marker that is formally similar. A connection between Kawapanan and
Chicham has been suggested (Valenzuela 2015) and therefore seems more promising.
The Kawapanan-Chicham connection also surfaces when looking at the syntactic
patterns, where both languages have a system in which applicative arguments have
prioritized access to the object indexing slot on the verb over direct objects. The other
shared syntactic pattern worth mentioning here is the fact that in both Hup (and pos-
sibly Kakua) and Tukanoan languages, the applicative argument is in most cases not
subject to differential argument marking, unlike direct objects. Given the known shared
history between Hup and Eastern Tukanoan languages (Epps 2006) both the system of
differential object marking, and the exception to this of applied objects, may be con-
tact-induced influence of Tukanoan on Hup.
The Yagua-Arabela connection is reinforced by semantics, both languages having
readings that include the extension of the effects of the event on objects in the posses-
sion of, or physically associated with the S or P participant. This may be extended to
some of the uses of Shiwilu ek-, which can promote clothing worn by the S or P partici-
pant. A further interesting venue to look into on the basis of semantic correspondences
(but outside the scope of this paper) is historical connections between Shiwilu and the
languages of central and southern Peru. The Arawakan language Nanti, for instance,
has a presential (53a) and relinquitive (53b) applicative, semantically similar to the cor-
responding Shiwilu applicatives:

(53) Nanti (Arawakan; Michael 2008: 286–287)

a. Birari tog-imo-ak-i=npi
Birari fell.tree-appl-pfv-real=2obj
‘Birari felled (it) in your presence.’
b. i=shig-apitsa-ak-i=ri
3mS=run-appl-pfv-real=3m.obj
‘He ran away from him.’

In sum, there seems to be little (and at best tentative) evidence for contact-induced
diffusion of forms in the applicative systems of the NWA languages of the sample
beyond the already recognized connection between Yagua and Arabela -ta, which is
also supported by overlapping semantics. There are some indications of contact-in-
duced influence concerning Kawapanan and Chicham languages, as well as between
Hup and Tukanoan languages, but these—with the possible exception of Kawapanan
and Chicham—involve semantics or syntax rather than form. An interesting avenue
for further research is the connection between Shiwilu and Nanti (as well as perhaps
further languages of central and southern Peru).
340 Rik van Gijn

5 Conclusions
The use of applicatives in the NWA is limited. Many languages of the area have no
applicative construction, even if related languages spoken elsewhere do. The languages
that do have an applicative construction, mostly have only one. The exceptions are
Kawapanan languages Shawi, and in particular Shiwilu, which have a richer inventory
of applicative constructions.
Beyond this, a few generalizations can be made about applicatives in the area on
the basis of the overview given above:
– Almost all applicative markers are prosodically dependent, postposed markers,
mostly suffixes (exceptions are prefixed markers in the Kawapanan languages
which occupy a middle ground between causatives and applicatives).
– There is some overlap in form, but on closer inspection, these overlaps are largely
coincidental (the main exception seems to be the applicative marker -ta in Yagua
and Arabela).
– In terms of morphosyntactic marking of the applied object, languages tend to follow
the internal logic of marking direct or indirect objects of non-derived verbs. Two
patterns are worth highlighting:
– In Hup and the Tukanoan languages, applied objects are generally not subject
to differential case marking, unlike direct objects (but like indirect objects).
– In Chicham and Kawapanan languages, applied objects are given priority
access to the object indexing slot on the verb over direct objects.
– In terms of semantics, affective (benefactive-malefactive) is clearly the most
common interpretation of the role of the applicative object, followed by a group
of interpretations that can be loosely connected to the notion of comitativity, and
finally locations.

Abbreviations
abl ablative
acc accusative
an animate
anph anaphor
appl applicative
assist assistance
att attenuative
ben benefactive
caus causative
clf classifier
cnt continuative
coll collective
com comitative
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 341

cop copula
decl declarative
def definite
dem demonstrative
des desiderative
dim diminutive
disc discourse
ds different subject
dur durative
dyn dynamic
emph emphasis
erg ergative
excl exclusive
exclm exclamative
f feminine
fut future
hiaf high affectedness
imp imperative
inan inanimate
incl inclusive
ind indicative
inf inference
instr instrument
int intentional
inter interrogative
ints intensifier
ipfv imperfective
lim limitative
loc locative
m masculine
mom momentary
mot motion
neg negation
nf non-feminine
nfut non-future
nmlz nominalizer
non.a/s non-subject (A/S) argument
obj object
pfv perfective
pl plural
poss possessive
pred predicative
pres present
pro pronominal
prt particle
pst past
real realis
rec recent
rel relativizer
342 Rik van Gijn

rem remote
rep reportative
seq sequential
sg singular
soc sociative
src source (of action)
ss same subject
sub subordinator
top topic
val valenciator
vblz verbalizer
vis visual
vm valency modifier
vpl verbal plural
x>y x acts on y

References
Aikhenvald, Alexandra Y. 1998. Warekena. In Desmond C. Derbyshire & Geoffrey K. Pullum (eds.), Handbook
of Amazonian languages, Vol. IV, 225–440. Berlin: Mouton De Gruyter.
Aikhenvald, Alexandra Y. 1999. Arawak. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.), The Amazonian
languages, 65–106. Cambridge: Cambridge University Press.
Aikhenvald, Alexandra Y. 2000. Transitivity in Tariana. In R. M. W. Dixon & Alexandra Y. Aikhenvald (eds.),
Changing valency: case studies in transitivity, 145–172. Cambridge: Cambridge University Press.
Aikhenvald, Alexandra Y. 2002. Language contact in Amazonia. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. 2003. A grammar of Tariana, from northwest Amazonia. Cambridge New York:
Cambridge University Press.
Aikhenvald, Alexandra Y. 2012. The languages of the Amazon. Oxford & New York: Oxford University Press.
Allin, Trevor R. 1976. A grammar of Resígaro. Buckinghamshire: Summer Institute of Linguistics.
Barraza de García, Yris Julia. 2005. El sistema verbal en la lengua Shawi. Recife: Universidad Federal de
Pernambuco.
Bertet, Denis. 2020. Aspects of Tikuna grammar (San Martín de Amacayacu variety, Colombia): phonology,
nominal phrase, predicative phrase. Lyon: Université Lumière Lyon 2 dissertation.
Beukema, Ronald William. 1975. A grammatical sketch of Chimborazo Quichua. New Haven, CN: Yale University.
Bolaños, Katherine. 2016. A grammar of Kakua. Utrecht: LOT.
Bourdeau, Corentin. 2015. Ergativity in Shawi (Chayahuita). Nijmegen: Radboud University Nijmegen MA
Thesis.
Bruil, Martine. 2014. Clause-typing and evidentiality in Ecuadorian Siona. Utrecht: LOT.
Campbell, Lyle. 2012. Typological characteristics of South American indigenous languages. In Lyle Campbell
& Verónica Grondona (eds.), The Indigenous languages of South America: a comprehensive guide,
259–330. De Gruyter Mouton. https://doi.org/10.1515/9783110258035.259.
Carpenter, Lawrence Kidd. 1982. Ecuadorian Quichua: Sescriptive sketch and variation. Gainesville: University
of Florida dissertation.
Chacon, Thiago. 2013. Kubeo: Linguistic and cultural interactions in the Upper Rio Negro. In Kristine Stenzel
& Patience Epps (eds.), Upper Rio Negro: cultural and linguistic interactions in northwestern Amazonia,
129–162. Rio de Janeiro: Museu Nacional / Museu do Índio (FUNAI).
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 343

Chacon, Thiago. 2014. A revised proposal of Proto-Tukanoan consonants and Tukanoan family classification.
International Journal of American Linguistics 80(3). 275–322. https://doi.org/10.1086/676393.
Chacon, Thiago. 2017. Arawakan and Tukanoan contacts in Northwest Amazonian prehistory. PAPIA: Revista
Brasileira de Estudos Crioulos e Similares 27. 237–265.
Ciucci, Luca & Pieter Muysken. 2011. Hernando de Alcocer y la Breve declaraciуn del Arte y Bocabulario de la
lengua del Ynga conforme al estilo y vso de la provincia de Quito. Indiana 28. 359–393.
Claassen, Simon. 2018. Verbal derivation in Upper Amazonian Quechua: A preliminary Semantic Syntax
perspective. Nijmegen: Radboud University Nijmegen MA thesis.
Cole, Peter. 1985. Imbabura Quechua. London & Dover, NH: Croom Helm.
Cook, Dorothy M. & Linda Criswell. 1993. El idioma koreguaje (tucano occidental). 1st edn. Santafé de Bogotá,
Colombia: Asociación Instituto Lingüístico de Verano.
Crevels, Mily & Hein van der Voort. 2020. Areal diffusion of applicatives in the Amazon. In Norval Smith,
Tonjes Veenstra & Enoch Oladé Aboh (eds.), Advances in contact linguistics: In honour of Pieter Muysken,
180–216. Amsterdam: John Benjamins. https://doi.org/10.1075/coll.57.06cre.
Curnow, Timothy. 1997. A grammar of Awa Pit (Cuaiquer): An indigenous language of south-western Colombia.
Canberra: Australian National University dissertation.
Derbyshire, Desmond C. 1987. Morphosyntactic areal characteristics of Amazonian languages. International
Journal of American Linguistics 53(3). 311–326.
Dickinson, Connie S. 2002. Complex predicates in Tsafiki. Eugene: University of Oregon dissertation.
Dixon, R.M.W. & Alexandra Y. Aikhenvald. 1999. Introduction. In R.M.W. Dixon & Alexandra Y. Aikhenvald
(eds.), The Amazonian languages, 1–22. Cambridge: Cambridge University Press.
Epps, Patience. 2006. The Vaupés melting pot: Tukanoan influence on Hup. In R. M. W. Dixon & Alexandra Y.
Aikhenvald (eds.), Grammars in contact: A cross-linguistic typology, 267–289. Oxford: Oxford University
Press.
Epps, Patience. 2008. A grammar of Hup. Berlin: Mouton de Gruyter.
Epps, Patience. 2020. Amazonian linguistic diversity and its sociocultural correlates. In Mily Crevels & Pieter
Muysken (eds.), Language dispersal, diversification, and contact: a global perception, 275–290. Oxford:
Oxford University Press. http://dx.doi.org/10.1093/OSO/9780198723813.003.0016.
Epps, Patience & Katherine Bolaños. 2017. Reconsidering the “Makú” language family of northwest
Amazonia. International Journal of American Linguistics 83(3). 467–507. https://doi.org/10.1086/691586.
Epps, Patience & Lev Michael. 2017. The Areal Linguistics of Amazonia. In Raymond Hickey (ed.), The
Cambridge handbook of areal linguistics, 934–963. Cambridge: Cambridge University Press. https://doi.
org/10.1017/9781107279872.033.
Epps, Patience & Kristine Stenzel (eds.). 2013. Upper Rio Negro: cultural and linguistic interaction in northwest
Amazonia. Rio de Janeiro: Museu Nacional/Museo do Índio — FUNAI.
Fischer, Rafael & Kees Hengeveld. 2023. A’ingae (Cofán/Kofán). In Patience Epps & Lev Michael (eds.),
Amazonian languages: An international handbook, Language isolates 1: Aikanã to Kandozi-Shapra, 65–123.
Berlin: De Gruyter Mouton.
Gibson, Michael Luke. 1996. El Munichi: un idioma que se extingue. Pucallpa: Instituto Lingüístico
de Verano.
Girón, Jesús Mario. 2008. Una gramática del wãńsöjöt (puinave). Amsterdam: LOT.
Guillaume, Antoine & Françoise Rose. 2010. Sociative causative markers in South-American languages: a
possible areal feature. In Franck Floricic (ed.), Essais de typologie et de linguistique générale, Mélanges
offerts à Denis Creissels, 383–402. Lyon: ENS Editions.
Hammarström, Harald, Robert Forkel, Martin Haspelmath & Sebastian Bank. 2021. Glottolog 4.5. Zenodo.
https://doi.org/10.5281/ZENODO.5772642.
Hanson, Rebecca. 2010. A grammar of Yine (Piro). Bundoora: La Trobe University dissertation.
Hart, Helen. 1988. Diccionario chayahuita-castellano (Canponanquë nisha nisha nonacaso’). Yarinacocha:
Ministerio de Educación / Instituto Lingüístico de Verano.
344 Rik van Gijn

Johnson, Orville E. & Stephen H. Levinsohn. 1990. Gramática secoya (Cuadernos Etnolingüísticos 11). Quito:
Instituto Lingüístico de Verano.
Jolkesky, Marcelo. 2016. Estudo arqueo-ecolinguístico das terras tropicais sul-americanas. Brasilia:
Universidade de Brasilia dissertation.
Jones, Wendell & Paula Jones. 1991. Barasano syntax. Dallas & Arlington: Summer Institute of Linguistics &
University of Texas at Arlington.
Klumpp, Deloris A. 2019. A grammar of Piapoco (SIL E-Books 71). Dallas: SIL International.
Kohlberger, Martin. 2020. A grammatical description of Shiwiar. Utrecht: LOT.
Lai, I-Wen. 2009. Time in the Iquito Language. Austin: University of Texas at Austin dissertation.
Lemus Serrano, Magdalena. 2020. Pervasive nominalization in Yukuna: an Arawak language of Colombian
Amazonia. Lyon: Université Lumière Lyon 2 dissertation.
Londoño Sulkin, Carlos David. 2012. People of substance: An ethnography of morality in the Colombian Amazon.
Toronto: University of Toronto Press.
Londoño Sulkin, Carlos David. 2017. Moral sources and the reproduction of the Amazonian package. Current
Anthropology 58(4). 477–501. https://doi.org/10.1086/692783.
Martins, Silvana Andrade. 2004. Fonologia e gramática Dãw. Utrecht: LOT. http://books.google.com/
books?id=ofEaAQAAIAAJ. (15 February, 2022).
Martins, Valteir. 2005. Reconstrução fonológica do Protomaku oriental (LOT International Series).
Utrecht: LOT.
Metzger, Ronald. 1981. Carapana. Bogotá: Instituto Lingüístico de Verano.
Michael, Lev. 2008. Nanti evidential practice: Language, knowledge, and social action in an Amazonian society.
Austin: University of Texas at Austin dissertation.
Michael, Lev. 2009. The semantics of clause linking in Iquito. In R. M. W. Dixon & Alexandra Y. Aikhenvald
(eds.), The semantics of clause linking: a cross-linguistic typology, 145–166. Oxford: Oxford University
Press.
Mihas, Elena. 2015. A grammar of Alto Perené (Arawak). Berlin: De Gruyter Mouton.
Miller, Marion. 1999. Desano grammar (Studies in the Languages of Colombia 6). Arlington: Summer
Institute of Linguistics & University of Texas at Arlington.
Morse, Nancy L. & Michael B. Maxwell. 1999. Cubeo grammar. Arlington: Summer Institute of Linguistics &
University of Texas at Arlington.
Muysken, Pieter. 2000. Semantic transparency in Lowland Ecuadorian Quechua morphosyntax. Linguistics
38(5). https://doi.org/10.1515/ling.2000.018.
Muysken, Pieter. 2009. Gradual restructuring in Ecuadorian Quechua. In Rachel Selbach, Hugo C. Cardoso
& Margot van den Berg (eds.), Gradual Creolization: Studies celebrating Jacques Arends, 77–100.
Amsterdam: John Benjamins. https://doi.org/10.1075/cll.34.09muy.
Muysken, Pieter. 2021. Substrate influence in Northern Quechua languages. In Enoch Oladé Aboh &
Cécile B. Vigouroux (eds.), Variation rolls the dice: A worldwide collage in honour of Salikoko S. Mufwene,
133–160. Amsterdam: John Benjamins. https://doi.org/10.1075/coll.59.06muy.
O’Brien, Colleen. 2018. A grammatical description of Kamsá, a language isolate of Colombia. Manoa: University
of Hawai’i at Manoa dissertation.
Olawsky, Knut J. 2006. A grammar of Urarina. Berlin & New York: Mouton de Gruyter.
Overall, Simon E. 2007. A grammar of Aguaruna. Bundoora: La Trobe University dissertation.
Payne, David L. 1990. Some widespread grammatical forms in South American languages. In Doris L. Payne
(ed.), Amazonian linguistics: Studies in lowland South American languages, 75–87. Austin: University of
Texas Press.
Payne, Doris L. 1984. Evidence for a Yaguan-Zaparoan connection. Work Papers of the SIL, University of North
Dakota 28. 131–156.
Payne, Doris L. 1985a. -ta in Zaparoan and Peba-Yaguan. International Journal of American Linguistics 51(4).
529–531.
 11 Contact-induced diffusion of applicatives in northwestern Amazonia? 345

Payne, Doris L. 1985b. Aspects of the grammar of Yagua: a typological perspective (Peru). Los Angeles:
University of California dissertation.
Payne, Doris L. 1990. Morphological characteristics of lowland South American languages. In Doris L. Payne
(ed.), Amazonian linguistics: studies in lowland South American languages, 214–241. Austin: University of
Texas Press.
Peña, Jaime. 2015. A grammar of Wampis. Eugene: University of Oregon dissertation.
Polinsky, Maria. 2013. Applicative constructions. In Matthew S. Dryer & Martin Haspelmath (eds.), The World
Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://
wals.info/chapter/109. (21 April, 2022).
Proyecto de documentación del idioma Muniche. 2009. Una breve descripción del idioma muniche. Iquitos:
Cabeceras Aid Project.
Ramirez, Henri. 1997. A fala tukano dos Ye’pâ-masa. Tomo 1: Gramática. Guajará-Mirim: UNIR.
Reinoso Galindo, Andrés Eduardo. 2002. Elementos para una gramática de la lengua piapoco. Bogotá:
Ministerio de Cultura.
Rich, Rolland. 1975. Sufijos verbales y apuntes gramaticales sobre el idioma arabela. Lima: Instituto Lingüístico
de Verano.
Rich, Rolland. 1999. Diccionario arabela-castellano. Lima: Instituto Lingüístico de Verano.
Rojas Berscia, Luis Miguel. 2013. La sintaxis y semántica de las construcciones causativas en el chayahuita de
Balsapuerto. Lima: Pontificia Universidad Católica del Perú BA thesis.
Rojas Berscia, Luis Miguel. 2019. From Kawapanan to Shawi: Topics in language variation and change.
Nijmegen: Radboud University Nijmegen dissertation.
Rojas Berscia, Luis Miguel & Rita Eloranta. 2019. The Marañón-Huallaga exchange route: “stones” and
“grains” as counting devices. Línguas Indígenas Americanas 19. e019011. https://doi.org/10.20396/
liames.v19i0.8655449.
Seifart, Frank. 2005. The structure and use of shape-based noun classes in Miraña (North West Amazon).
Nijmegen: Max Planck Institute for Psycholinguistics dissertation.
Seifart, Frank. 2015. Tracing social history from synchronic linguistic and ethnographic data: The prehistory
of Resígaro contact with Bora. Mundo Amazónico 6(1). 97–110. https://doi.org/10.15446/ma.v6n1.48222.
Sola, Donald F. & Gary J. Parker. 1964. The structure of Ayacucho Quechua. Ithaca, NY: Cornell University.
Sorensen, Arthur P. 1967. Multilingualism in the Northwest Amazon. American Anthropologist 69(6). 670–684.
Stenzel, Kristine. 2013. A reference grammar of Kotiria (Wanano). Lincoln: University of Nebraska Press.
Thiesen, Wesley & David Weber. 2012. A grammar of Bora: with special attention to tone. Dallas: SIL International.
Valenzuela, Pilar M. 2015. ¿Qué tan “amazónicas” son las lenguas kawapana? Contacto con las lenguas
centro-andinas y elementos para un área lingüística intermedia. Lexis 39(1). 5–56.
Valenzuela, Pilar M. 2016. “Simple” and “double” applicatives in Shiwilu (Kawapanan). Studies in Language
40(3). 513–550. https://doi.org/10.1075/sl.40.3.02val.
Vengoechea, Consuelo de. 2012. Categorisation lexicale en muinane (Amazonie Colombienne). Toulouse:
Université de Toulouse dissertation.
Wilson, Peter J. & Stephen H. Levinsohn. 1992. Una descripción preliminar de la gramática del achagua
(arawak). 1st edn. Bogotá: Asociación Instituto Lingüístico de Verano.
Wise, Mary Ruth. 1999. Small language families and isolates in Peru. In R. M. W. Dixon & Alexandra Y.
Aikhenvald (eds.), The Amazonian languages, 307–340. Cambridge: Cambridge University Press.
Wise, Mary Ruth. 2002. Applicative affixes in Peruvian Amazonian languages. In Sérgio Meira, Mily Crevels,
Simon van de Kerke & Hein van der Voort (eds.), Current studies on South American languages, Vol. 3,
329–344. Leiden: CNWS Publications.
Wise, Mary Ruth. 2011. Rastros desconcertantes de contactos entre idiomas y culturas a lo largo de los
contrafuertes orientales de los Andes del Perú. In Willem F. H. Adelaar, Pilar M. Valenzuela & Roberto
Zariquiey Biondi (eds.), Estudios sobre lenguas andinas y amazónicas: homenaje a Rodolfo Cerrón-
Palomino, 305–326. Lima: Fondo Editorial Pontificia Universidad Católica del Perú.
William A. Foley
12 Applicatives in Papuan languages
Abstract: The term Papuan languages is a purely negative characterization, covering
the languages of more than thirty language families plus over a dozen isolates, spoken
on and around the island of New Guinea. In spite of their enormous genetic and struc-
tural diversity, Papuan languages can on the whole be categorized as head marking and,
commensurate with that classification, as possessing applicative constructions. The
number of applicative constructions in Papuan languages ranges from one (in typical
Trans New Guinea languages of the central highlands like Fore) to thirteen (in the Mac-
ro-Sko language Barupu), but in all cases identified they historically have arisen from
reanalysis of verb roots in verb compounds or serial verb constructions. The crosslin-
guistically most widespread applicative construction marks beneficiary participants,
but languages with richer inventories go well beyond that to indicate a very wide range
of erstwhile peripheral semantic roles like locations, goals, and comitatives.

1 The nature of Papuan languages

Unlike most other chapters in this volume which treat single languages or language
families, this chapter necessarily has a very large purview, essentially a geographical
area centered around the island of New Guinea and adjacent smaller islands, within
which are spoken some 700–800 languages belonging to between 30 and 60 distinct
languages families, excluding the languages of the Austronesian family in this region,
which number a few hundred themselves. Papuan languages are defined purely nega-
tively, that is, they are the languages of the New Guinea region which do not belong to
the Austronesian language family, a fact made clear in their alternative name, Non-Aus-
tronesian languages: the name does not denote a coherent genetic grouping. Hence
Papuan language is a cover term to cover all the many distinct language families of
this region which are not Austronesian, and the exact number of these is still rather
undetermined (see Palmer 2018), but clearly cannot be less than thirty, and in addition
to these there are some two dozen linguistic isolates.
Papuan languages are not only highly genetically diverse, they are also very struc-
turally diverse. They range from analytical isolating languages similar to those of South-
east Asia, like Kimaghama (Kolopom sub-family, Trans New Guinea family; Drabbe
1949) to mildly inflecting languages like Watam (Lower Ramu sub-family, Lower Sepik-
Ramu family) or Moskona (East Bird’s Head family; Gravelle 2011) to the richly inflect-
ing often fusional languages of the many sub-families of the vast Trans New Guinea
family like Coastal Marind (Anim sub-family; Olsson 2021) and finally to the aggluti-
native incorporating polysynthetic languages of the Sepik region like Yimas (Lower

https://doi.org/10.1515/9783110730951-012
348 William A. Foley

Sepik sub-family, Lower Sepik-Ramu family). Syntactically most Papuan languages are
of an OV left branching typology with postpositions, with the morphologically richer
languages commonly allowing flexibility in clausal constituent order, but not all are; in
particular the languages of the Torricelli and West and East Bird’s Head families are VO
right branching in structure, with prepositions instead of postpositions. For the latter
two families this is probably an innovation due to influence from neighboring Aus-
tronesian languages, which share this typology, but this is unlikely in the case of Tor-
ricelli family languages. The morphologically richer languages are also head marking
languages (Nichols 1986), often extremely so like Coastal Marind, Yimas and Central
Asmat (Asmat-Kamoro sub-family, Trans New Guinea family), expressing all grammat-
ical relations simply by bound pronominal affixes with no nominal case marking, as in
this Yimas example (Roman numerals indicate noun classes throughout this chapter):

(1) Yimas
ŋaykum makaw wa-mpu-ŋa-r-akn panmal
woman.II.pl fish.sp.X.sg X.sg.nom-3pl.erg-give-pfv-3sg.dat man.I.sg
‘the women gave makaw to the man’

though some of the Trans New Guinea languages are double marking, having both
bound pronominal affixes and some case marking for nominal arguments, like Fore
(Gorokan sub-family, Trans New Guinea family):

(2) Fore (Scott 1978)

mási wá-má a-ka-i-e
boy man-nom 3sg.obj-see-3sg.subj-decl
‘the man sees the boy’

Dependent marking languages are largely restricted to more morphologically depau-

perate languages such as those of the Lower Ramu, Lakes Plain or East Cenderawasih
Bay families. Some Sepik family languages like Abau (Locke 2011) are particularly inter-
esting and typologically rare in combining a polysynthetic incorporating typological
profile with dependent marking.

2 Verbal sources of applicative morphemes

in Papuan languages
Applicative constructions in which an erstwhile oblique argument has been promoted
to a core object grammatical relation and hence indicated as such by an applicative mor-
pheme in the verb are mostly restricted in Papuan languages to the morphologically richer,
head marking languages in which grammatical relations are indicated by bound pronom-
 12 Applicatives in Papuan languages 349

inal affixes. Languages which lack these constructions typically perform similar func-
tions through the use of serial verb constructions, and indeed there is clear evidence that
applicative constructions in Papuan languages in many cases have evolved from re-anal-
ysis of prior serial verb constructions and in some cases synchronically alternate with
them. Consider these examples from Teiwa (Alor-Pantar sub-family, Trans New Guinea
family), in which various serial verb constructions perform the functions of applicative
morphemes of verbal origin in other Papuan languages to be discussed in detail below:

(3) Teiwa (Klamer 2010)

a. iman gon quan pin te
3pl gong drum hold walk
‘they walk away with the gong and drum’
b. ped mat ma man taxar
machete take come grass cut
‘cut the grass with a machete’
c. Sangubal ma bir-an daa
pn come run-r go_up
‘running up from Sangubal’
d. in nuk yivar ga-taax wan hor
thing one dog 3sg-throat be_at hang
‘something hangs around the dog’s throat’
e. uy ga’an u sen ma n-oma’ g-an
person 3sg dist money come 1sg-father 3sg-give
‘that person gave money to my father’

Note the roles of the arguments of the bolded verbs in series in the above examples:
comitative in (a) with pin ‘hold’, instrument in (b) with mat ‘take’, source in (c) with ma
‘come’, location in (d) with wan ‘be at, exist’, and exchanged theme in (e) again with ma
‘come’. In a morphologically rich Papuan language like Barupu (Piore River sub-fam-
ily, Macro-Sko family) some of these semantic roles would be expressed by applicative
affixes on verbs, albeit ones quite transparently derived from verbs grammaticalized as
affixes from prior serial verb constructions:

(4) Barupu (Corris 2005)

a. -ě from
k-en-úte-n-ě-mú
r-1sg.f.subj-walk-1sg.subj-applfrom-2sg.f.obj
‘I walked away from you (sg)’
b. -ô cause
k-en-ké<n>í-n-ô-wa
r-1sg.f.subj-sit<1sg.subj>-1sg.subj-applcause-3sg.m.obj
‘I’m staying behind because of him’
350 William A. Foley

c. -o for
k-en-úte-n-o-mu
r-1sg.f.subj-walk-1sg.subj-applfor-2sg.f.obj
‘I walked for you (sg)’
d. -î with
n-en-úte-n-î-mu
irr-1sg.f.subj-walk-1sg.subj-applwith-2sg.f.obj
‘I will walk to be with you (sg)’

Barupu verbs are always inflected for the person, number and gender of their subjects
by prefixes (and occasionally by infixes), and, if transitive, also for the person, number
and gender of their objects by suffixes. Both verb roots in the examples of (4), úte- and
kéi- ‘sit’ are intransitive, and as such occur with subject marking en- 1sg.f.subj, but
they also take direct object suffixes here, -mu 2sg.f.obj or -wa 3sg.m.obj, because they
have been transitivized by applicative suffixes. But note that each of the applicative
suffixes also takes subject marking n- 1.sg.subj, agreeing with the subject marking of
the verb root. This is because each of these originally goes back to a separate verb with
its own verbal inflection in a serial verb construction which has now been reanalyzed
as a bound affix functioning as an applicative morpheme, but one which still bears
its older inflectional patterns. Only -o ‘to, for’ is synchronically a verb root in Barupu,
specifically the verb root for ‘give’; the others now only function as bound applicative
affixes with no link to current verb roots. As we shall see in what follows, re-analysis of
verb roots in verb compounds or serial verb constructions is a very common diachronic
source for applicative morphemes in Papuan languages, often transparently so in that
the form still functions as a verb root elsewhere in the language.
But it is not universally the case in Papuan languages that applicative morphemes
arise from re-analyzed serial verb constructions; it seems restricted to those with an
OV left branching typological profile. The minority of Papuan languages which have a
VO right branching profile sometimes grammaticalize applicative morphemes by incor-
porating a preposition. Consider the case of Mountain Arapesh (Arapeshan sub-family,
Torricelli family) which has a preposition umu:

(5) Mountain Arapesh (Conrad and Wogiga 1991)

a. u-a-salik stoakipa umu mabeguh
1sg.subj-r-ask clerk prep marbles
‘I asked the store clerk for marbles’
b. h-a-dúk-ech umu katres
3pl.m.subj-r-kill-3pl.obj prep bullet
‘they (pl) killed them with bullets’
 12 Applicatives in Papuan languages 351

There is a general applicative suffix in Mountain Arapesh which is clearly an incorpo-

rated and phonologically reduced form of this preposition):

(6) Mountain Arapesh (Conrad and Wogiga 1991)

a. m-u-bani-m-ech bi-ech yawihas
1pl.subj-irr-plant-appl-3pl.obj two-3pl garden.pl
‘we will plant two gardens for them (pl)’
b. i-tal-um-ona-li yeguh
1sg.subj.irr-buy-appl-3.sg.m.obj-toward fish
‘I will buy the fish for him and bring it’
c. ch-a-na-mu bulguh
3pl.subj-r-go-appl pigs.pl
‘they (pl) went for pigs’
d. yék i-na-m-enyú
1sg 1sg.subj.irr-go-appl-2sg.obj
‘I will go with you (sg)” or “I will
go for you (sg)’
e. élgei-no-mu nobat
fear-3sg.m.obj-appl dog
‘he is afraid of the dog’
f. i-chu-sah-um-ona-li
1sg.subj.irr-VIII.pl.obj-carry-appl-3sg.m.obj-toward
‘I will carry the things on my shoulder for him and come’
g. ch-a-núk-as-um-ech-i
3pl.subj-r-pull-IX.pl-obj-appl-3pl.obj-toward
‘they (pl) pulled them (slit drums) for them (pl) toward (speaker)’

Note that the initial and final vowels of the incorporated preposition umu delete fol-
lowing (6a, c, d, e) and preceding (6a, b, d, f, g) a vowel respectively. Note also, as (6f,
g) demonstrate, that the introduced applied participant does not usurp the function of
the original direct object of the verb, which can still be marked by a bound pronominal
affix (prefix [6f] or suffix [6g] depending on verb class). The applicative suffix along
with any bound pronominal suffix for the introduced applied participant then simply
follows.

3 Benefactive marking in Papuan languages

By far the most widespread applicative constructions in Papuan languages are those
which express the beneficiaries of an action. This usually involves use of the verb roots
meaning ‘give’ in a verb compounding or serial verb construction, though in some cases
352 William A. Foley

the etymological source of the benefactive applicative is no longer recoverable. Con-

sider these examples from Central Asmat and Imonda:

(7) a. Central Asmat (Voorhoeve 1965)

enám nes jimin akat jik-tam-por-m-í-n
fish flesh sago.stick good bind-applben-try-pres-1sg.subj-2.obj
‘I make tasty sticks of sago and fish for you’
b. Imonda (Seiler 1985)
ka-m fi-n-fin
1-dat do.pl-applben.nsg-pfv
‘they did it for us’

The verb root for ‘give’ in Central Asmat is tetam-, clearly related to the benefactive
applicative suffix and diachronically derived from it (the Proto-Asmat-Kamoro verb
root for ‘give’ is ✶tam-[Usher and Suter 2020]), but no longer transparently equivalent
to it, while the root for ‘give’ in Imonda is ai-, with no relationship whatsoever to the
benefactive applicative suffix.
In the vast Trans New Guinea family across numerous sub-families, benefactive
applicative constructions are formed by compounding with the main verb an auxil-
iary verb, usually etymologically ‘give’ or ‘put’, though not necessarily (Telefol, example
[8e]), which commonly takes person and number inflection as direct object for the ben-
eficiary. Hence these like most Papuan languages, are secundative marking languages
for ditransitive verbs in Dryer’s (1986) typology. Often, this is the only way to express
a beneficiary. For some languages there is no basic construction in which the benefi-
ciary can be realized as an oblique and the verb lacks the benefactive marker, while for
others like Telefol and Hua there are basic construction alternatives (see Hua examples
in [18]). Consider these examples drawn from a number of sub-families:

(8) a. Fore (Scott 1978)

pu-na-ʔta-i-e
do-1sg.obj-applben/put-3sg.subj-decl
‘he did it for me’

The benefactive suffix and agreement is obligatory in Fore, though it does allow an
overt noun phrase functioning as the beneficiary to be case marked as an oblique with
the allative case suffix -ti:

(8) b. Fore (Scott 1978)

nae-ti mae-wae-na-ʔta-i-e
1sg-all get-all-1sg.obj-applben/put-3sg.subj-decl
‘he gets it all for me’
 12 Applicatives in Papuan languages 353

c. Tairora (Vincent 1973)

rumpa-ti-mi-te-ro
tie-1sg.obj-applben/give-pfv-3sg.subj
‘he tied it for me’
d. Amele (Roberts 1987)
uqa ahul gel-i-te-i-a
3sg coconut scrape-applben/give-1sg.obj-3sg.subj-nr.past
‘she scraped a coconut for me’
e. Telefol (Healey 1965)
am díne-ʔk-ee-m-antém-íb
house build-2sg.obj-applben-dur-fut-3pl.subj
‘they will build a house for you (sg)’
f. Lower Grand Valley Dani (Bromley 1981)
wam pa-n-et-h-e
pig cut-1sg.obj-applben/give-r-3sg.subj
‘he selected a pig for me’
g. Lower Grand Valley Dani (Bromley 1981)
hakki pa-n-akeik-h-e
banana cut-1sg.obj-applben/put-r-3sg.subj
‘he cut (and put aside) some bananas for me’

Note that the beneficiary is marked for person and number by prefixes to the benefac-
tive applicative suffix in all the languages except Amele. These are the bound pronomi-
nal markers for direct objects, which would occur as prefixes to normal transitive verbs,
Fore na-ka-i-e 1sg.obj-see-3sg.subj-decl ‘he sees me’ (Scott 1978), but here are prefixed
to the benefactive suffix instead, again demonstrating its erstwhile status as a verb. In
Amele, direct object markers are suffixed for simple transitive verbs, jab-ade-i-a chase-
3pl.obj-3sg.subj-past ‘he chased them’, as is the benefactive agreement marker in (8d),
so the same pattern still holds, but here via suffixation. Only in the Telefol example (8e)
is the benefactive affix not identical to a synchronic verb root meaning ‘give’ or ‘put’,
though the verbal structure is identical to that of Fore, Tairora and Lower Grand Valley
Dani and almost certainly goes back to Proto-Trans New Guinea. Interestingly, in Lower
Grand Valley Dani, there are two benefactive suffixes, each equivalent to a different
verb root in the language, ‘give’ (8f) versus ‘put’ (8g), with a corresponding difference
in meaning as captured in the translations of (8e, f): the benefactive with ‘put’ denotes
that objects are deposited for the beneficiary who will then claim them, but this is not
entailed for benefactives with ‘give’.
Multiple benefactive affixes similar to those of Lower Grand Valley Dani are not
uncommon in Papuan languages. Consider these examples from Alamblak (Sepik
Hill sub-family, Sepik family) and Yimas (Lower Sepik sub-family, Lower Sepik-Ramu
family):
354 William A. Foley

(9) Alamblak (Bruce 1984)

a. kuɲ-t hingna-nho-më-m-r
house-f work-applben-rem.past-3pl.subj-3sg.m.obj
‘they (pl) helped him build a house’
b. këfra-t tu-hay-më-r-r
spear-f throw-applBEN/give-rem.past-3sg.m.subj-3sg.m.subj
‘he threw a spear (to him) for his benefit or to his detriment’

The second benefactive marker, the serial verb construction with hay- ‘give’ in (9b)
is labeled by Bruce as the indirect benefactive and simply means “something good
happens to the actor or he does something which has a good or bad effect on the benefi-
ciary” (Bruce 1984:159). But the form in (9a) with -nho, called by Bruce the “direct ben-
efactive”, is much more restricted: 1. the event and the benefactive effect must happen
at the same time and in the same place; 2. the same event is experienced by both the
benefactor and the beneficiary; 3. the benefactor acts intentionally; and 4. the effect
can only be benefactive, never malefactive. None of these conditions needs to hold for
the serial verb benefactive construction with -hay ‘give’. Now consider these examples
from Yimas:

(10) Yimas (Foley 1991)

a. Mitchell kat ya-ka-taŋ-wayk-r-akn
pn card.V.pl V.pl.nom-1sg.erg-appl-buy-pfv-3sg.dat
‘I bought cards for Mitchell’
b. wampunŋ k-mpu-kra-taŋ-mntk-ntut
sago.flour.VI.sg VI.sg.nom-3pl.erg-1pl.dat-appl-finish-rem.past
‘they (pl) used up the sago flour on us (pl)’
c. awt ŋa-kra-yawra-mpi-waraca-ŋa-n
fire imp-1pl.dat-gather-seq-return-applBEN/give-imp
‘bring back fire for us (pl)’

The simple benefactive applicative prefix taŋ- in (10a, b), also used as a comitative and
external possessor applicative (see Section 4.2), can have either a benefactive (10a)
or malefactive (10b) meaning, but crucial to its usage like Alamblak -nho ben is the
requirement that the actor and the beneficiary or harmed participant be physically
present together during the event. The use of ŋa- ‘give’ in (10c) has no such requirement
and in fact typically denotes the converse. This is brought into sharp relief in the fol-
lowing two examples:

(11) Yimas (Foley 1991)

a. narm p-ka-taŋ-warapak-r-akn
skin.VII.sg VII.sg.nom-1sg.erg-appl-cut-pfv-3sg.dat
‘I cut his skin for him’ (as in a male initiation)
 12 Applicatives in Papuan languages 355

b. narm p-ka-kan-ŋa-r-akn
skin.VIII.sg VII.sg.nom-1sg.erg-pierce-applBEN/give-pfv-3sg.dat
‘I pierce the skin for him’ (skin of some other creature, likely an animal)

(11a) with taŋ- expresses physical and temporal proximity between the actor and the
beneficiary during the act of skin cutting. It is the skin of the beneficiary which is being
cut as part of a formal ritual of male initiations. Example (9b) with ŋa- ‘give’ does not
require the beneficiary to be present at the time of the piercing, merely that he benefit
from the act; hence the entailment that the skin being pierced must belong to some
other being, probably an animal.
Mian (Ok sub-family, Trans New Guinea family) has a further fascinating twist on
this theme of two benefactive applicative affixes, but here the split is determined by
aspect. Rather like Slavic languages, Ok languages typically have a division of many
of their verb stems according to aspect, imperfective versus perfective, and Mian is
typical:

(12) Mian (Fedden 2011)

pfv ipfv
fa- faka- ‘make fire’
têm- temê- ‘see’
ngele- ngen- ‘beg’
ge- ga- ‘say’

The verb ‘give’ exhibits suppletion for aspect:

(13) Mian (Fedden 2011)

a. nē naka=e éil=o
1sg man=m.sg pig=f.sg
om-ub’-a-∅-i-bio=be
3sg.f.obj-give.pfv-3sg.m.dat-r-1sg.subj-past=decl
‘I gave the sow to the man’
b. ī blatik=o
3pl.an plastic_bag=n.pl
do-ka-ye-bina-b-io=be
3pl.obj-give.ipfv-pl.an.dat-hab-ipfv-2/3.pl.an.subj=decl
‘they usually give a few (vomit) bags to us’

Interestingly the verb stem for ‘give’ in the imperfective is clearly the homophonous
transitive stem ka- meaning ‘put’. So, what we find in Mian is a recapitulation of the
contrast in Lower Grand Valley Dani (8f, g), but here semantically bleached and deter-
mined by an aspectual contrast. Like other Trans New Guinea languages as exemplified
in (8), Mian forms benefactive applicative constructions by compounding ‘give’ with the
356 William A. Foley

main verb in the perfective, but not in the imperfective, which simply uses a benefac-
tive applicative suffix cognate with that in the Telefol example of (8e), hence preserving
the aspectual contrast:

(14) Mian (Fedden 2011)

a. as=e fe-ˆb’-o-n-e=be
fire=m.sg make_fire.pfv-applBEN/give.pfv-3sg.f.dat-r-3sg.m.subj=decl
‘he made a fire for her’
b. imen=o fu-k-e-be-i=be
taro=n.pl cook-2sg.dat-appl-ipfv-1sg.subj=decl
‘I’m cooking taro for you (sg)’

As seen with examples from Alamblak and Yimas, the benefactive suffix is often
extended to cover a range of related roles like maleficiary, recipient, etc.; Amele exem-
plifies well these extended uses of the benefactive affix:

(15) Amele (Roberts 1987)

a. maleficiary
age ege na mala j-i-g-eig-a
3pl 1pl poss chicken eat-appl/give-1pl.obj-3pl.subj-nr.past
‘they ate our chicken on us’
b. recipient
ija cabi ihac-i-h-ig-en
1sg work show-appl/give-2sg.obj-1sg.subj-fut
‘I will show the work to you (sg)’
c. allative
h-u-t-ag-a
come-appl/give-1sg.obj-2sg.subj-imp
‘come to me’
d. ablative
uqa sigin ebe-ni na dec hu-i-te-i-a
3sg knife hand-sg.poss of from grab-appl/give-1sg.obj-3sg.subj-nr.past
‘he grabbed the knife out of my hand’

Interestingly, like the Fore example (8b), the ablative participant is marked as an oblique
phrase with the postposition dec ‘from’, yet an applicative construction is employed
here nonetheless. Further, the meaning of ‘give’ as the source of the applicative suffix
is semantically bleached here, as the action denoted is one of removal, a fact indicating
complete grammaticalization of ‘give’ as an applicative in Amele.
As mentioned above, the applicative construction is commonly the obligatory con-
struction in the Papuan languages to express beneficiaries, there being no basic con-
struction alternative in which beneficiaries can be expressed solely as obliques. But
 12 Applicatives in Papuan languages 357

there are languages which do have such alternatives. Consider Urim (Urim-Kombio
sub-family, Torricelli family) and Hua (Gorokan sub-family, Trans New Guinea family).
Urim has a single applicative suffix -(e)n that covers a range of meanings like benefi-
ciary, recipient, and cause, but the applicative construction always alternates with a
basic construction in which the erstwhile applied object is realized as an oblique noun
phrase headed by a preposition:

(16) Urim (Wood 2012)

a. men kark ekng tu tungkoren
1pl afraid prep 3pl white
‘we (pl) were afraid of the white people’
b. kupm ti pike kark-en tungkoren yat
1sg here before afraid-appl white also
‘I too was afraid of the white people before’

(17) Urim (Wood 2012)

a. mentekng kor kha ekng kitn
1dl seek grasshopper prep 2sg
‘we (dl) seek grasshoppers for you (sg)’
b. mentekng kor-n=teitn kha
1dl seek-appl=2sg.obj grasshopper
‘we (dl) seek grasshoppers for you (sg)’

The examples in (16) involve an intransitive verb kark ‘be afraid’. The cause of that fear
can be realized either as an oblique constituent marked with the preposition ekng (16a)
or as the direct object of the verb immediately following the verb without the prepo-
sition and with the verb suffixed with the applicative suffix -n (16b). In the examples
of (17) we have the transitive verb kor ‘seek’. A beneficiary again can be realized as an
oblique constituent (17a) or as the direct object of a verb suffixed with the applicative
marker (17b); pronominal direct objects are realized as enclitics to the verb, here =(t)
eitn 2sg.obj.
Hua, although in the same Gorokan sub-family as Fore in (8a, b), behaves like Urim
in exhibiting an alternation between a basic construction in which a beneficiary is
realized as a case marked oblique constituent and an applicative construction where it
functions as the direct object:

(18) Hua (Haiman 1980)

a. dgai-suʔ zu ki-e
1sg-ben house build.3sg.subj-decl
‘he built a house for me’
358 William A. Foley

b. zu ki-na d-t-e
house build-3sg 1sg.obj-applben/put.3sg-decl
‘he built a house for me’

ki-na build-3sg in (18b) is a same subject dependent form of the verb with third person
singular anticipatory subject marking. Example (18b) is a biclausal chaining construc-
tion, so less fusional than the single word structure found in Fore.
Yimas is particularly interesting in this regard. As a strongly head marking polysyn-
thetic language, all core grammatical relations of a verb are indicated by pronominal
affixes on the verb; there is no case marking on core argument nominals. When a ben-
eficiary occurs with either an intransitive or transitive verb, it must be expressed with
either of the applicative affixes illustrated in (10) and (11) above. However, ditransitive
verbs present a problem: all argument positions are saturated, as Yimas does not permit
verbs to have more than three core arguments. Hence when a beneficiary is added to
a clause with a ditransitive verb, applicative constructions like those of (10) and (11)
are prohibited and instead the benefactive is expressed with the postposition nampan
‘toward’:

(19) Yimas
anti i-ka-pul-c-akn mpu-nampan
ground.VIII.sg VIII.nom-1sg.erg-rub-pfv-3sg.dat 3pl-toward
‘I rubbed dirt on him for them (pl)’

4 Papuan languages with multiple applicative

constructions
A number of Papuan languages have multiple applicative constructions; I present three
brief case studies here.

4.1 Coastal Marind (Anim sub-family, Trans New Guinea family;

Olsson 2021)

Coastal Marind has a rich system of applicative constructions, but an unusual fact about
it is that, unlike the languages discussed in Section 3, it does not use an applicative
construction to express benefaction. It uses the same basic ditransitive construction for
beneficiaries (and maleficiaries) as it does for recipients, namely a set of bound dative
pronominals:
 12 Applicatives in Papuan languages 359

(20) Coastal Marind (Olsson 2021)

a. recipient
surat mak-o-ikalen Simon
letter fut.1.A-3sg.dat-send:III.U pn
‘I’m going to send a letter to Simon’
b. beneficiary
mesiwag mak-o-kahos-e
old.woman fut.1.A-3sg.dat-chew_betel-ipfv
‘I will chew betelnut for grandma’ (she has no teeth)
c. maleficiary
nggat tamuɣ mak-a-na-ɣi
dog food nafut-3sg.A-1.dat-eat
‘the dog might eat the food on me’

Coastal Marind does have four distinct applicative constructions: 1. a comitative-instru-

mental marked by k- ~ ka-; 2. an accompaniment marked by e-, expressing a co-par-
ticipant in a motion event, brought along or being chased by the actor; 3. an allative
‘toward’ indicated by ind-; and 4. a separative ‘away from’, denoted by is-. Examples
follow:

(21) Coastal Marind (Olsson 2021)

a. comitative
kak Wobi da menda-b-∅-ka-man
aunt pn sago pfv-A-3sg.A-applcom-come
‘Aunt Wobi already brought sago’
b. instrumental
alib ∅-no-d-ka-w-as
arrow_type nvo-1.A-dur-applinstr-3sg.U-shoot
‘I was shooting with alib arrows’
c. accompaniment
maɣaɣ anem Poce ∅-∅-e-umuh
first man pn nvo-3sg.A-applaccomp-go:3sg.U
‘at first it was that man Poce who brought him’
d. allative
ɣap m-a-n-ind-a-y-lolaw-em
night ovo-3sg.A-1.dat-applall-1.dat-1pl-crawl:3sg.U-toward
‘at night he came sneaking toward us’
e. separative
eham m-∅-is-ihon
husband:3sg ovo-3sg.A-applsep-run:3sg.U
‘she ran away from her husband’
360 William A. Foley

4.2 Yimas (Lower Sepik sub-family, Lower Sepik-Ramu family;

Foley 1991, 1997)

Yimas has six contrasting applicative constructions, two of which, taŋ- and ŋa-, were
illustrated in (10) and (11) above. There I illustrated the benefactive applicative use
of the prefix taŋ-. However, this is not its only usage; rather it commonly functions as
an applicative for comitative participants. When marking beneficiaries, applicative
constructions with taŋ- are obligatory, there being no alternative basic construction
with the beneficiary as an oblique phrase, except in the situation of a ditransitive verb
with argument positions already saturated, as mentioned in Section 3. But with comita-
tive participants, there is an alternative between a basic construction with an oblique
phrase and an applicative with taŋ-:

(22) Yimas
a. ama kantk na-mpu-tar-kwalca-t
1sg with 3sg.nom-3pl.erg-caus-rise-pfv
‘they (pl) woke him up along with me’
b. na-mpu-ŋa-taŋ-tar-kwalca-t
3sg.nom-3pl.erg-1sg.dat-applcom-caus-rise-pfv
‘they (pl) woke him up along with me’

Both constructions are possible, though (22b) is definitely preferred with pronominal
comitative participants. Possession is also indicated with the comitative postposition
kantk ‘with’, though in this usage it is suffixed to mark the number of the possessor.
Again, these alternate with an applicative construction with taŋ-, in this usage collo-
cated with the verb root taw- sit, be at’:

(23) Yimas
a. tawra kantk-mampan aymbak
money.IX.sg with-dl cop.3dl
‘they (dl) have money’
b. tawra impa-na-taŋ-taw-n
money.IX.sg 3dl.nom-pres-applcom-sit-pres
‘they (dl) have money’

Both of these constructions seem equally common.

Another applicative construction which exhibits alternations with a basic construc-
tion is that involving the prefix ira-. It typically marks allative, a place or person toward
which someone moves and alternates with a basic construction involving the postposi-
tion nampan ‘toward, for’:
 12 Applicatives in Papuan languages 361

(24) Yimas
a. na-nampan na-way-mpi-ya-ntut
3sg-toward 3sg.nom-turn-seq-come-rem.past
‘he turned around and came back to her’
b. na-n-way-mpi-ira-ya-ntut
3sg.nom-3sg.erg-turn-seq-applall-come-rem.past
‘he turned around and came back to her’

There is also what might be termed a metaphorical use of ira-, generally in association
with verbs of emotional or cognitive states, in which the applied object is not only the
cause of the state, but also the person or thing toward which the state is directed:

(25) Yimas
a. na-n-pay-ira-wampuŋ-kra-ntut
3sg.nom-3sg.erg-now-APPLALL-heart-cut-rem.past
‘he worried about her now’
b. yaŋkuraŋ k-mp-ira-aykapiŋa-k-nakn
thought.VI.sg VI.sg.nom-3dl.erg-applall-know-irr-3sg.dat
‘they (dl) thought about her’

The remaining three applicatives in Yimas, like the benefactive usage of taŋ- lack alter-
native basic constructions with oblique phrases marked by postpositions. The first is
visual taŋkway-, which indicates that the actor performs an action while carefully vis-
ually monitoring the applied, necessarily animate, participant. Consider the following
contrastive examples:

(26) Yimas
a. na-n-ira-wampaki-kia-k-nakn
3sg.nom-3sg.erg-applall-throw-night-irr-3sg.dat
‘he threw it toward him’ (in his direction)
b. na-n-taŋkway-wampaki-kia-k-nakn
3sg.nom-3sg.erg-applvis-throw-night-irr-3sg.dat
‘he threw it at him’ (looking at him)

A couple of other examples of taŋkway- vis:

(27) Yimas
a. na-mpu-taŋkway-iranta-irm-kia-ntut
3sg.nom-3pl.erg-applvis-dance-stand-night-rem.past
‘they (pl) danced for her’ (in her honor, watching for responses)
362 William A. Foley

b. na-n-taŋkway-iray-ɲcut
3sg.nom-32sg.erg-applvis-cry-rem.past
‘he cried over her’ (looking at her body lying in the canoe)

The next non-alternating applicative is the kinetic pampay-, derived from an irregular
reduplication of the verb root pay- ‘carry, lie’. It is used whenever the core argument
introduced by the applicative is involved as the passive partner in act of carrying; it par-
allels some of the usages of the Coastal Marind associative applicative. Compare again
the following contrastive examples:

(28) Yimas
a. na-mpu-taŋ-wapal-kia-k
3sg.nom-3pl.erg-applcom-climb-night-irr
‘they (pl) came up with her’ (comitative: she walked along too)
b. na-mpu-pampay-wapal-kia-k
3sg.nom-3pl.erg-applkin-climb-night-irr
‘they (pl) came up with her’ (carrying her)

A couple of other examples of pampay- kin:

(29) Yimas
a. na-mp-pampay-arm-kia-k
3sg.nom-3dl.erg-applkin-board-night-irr
‘they (dl) boarded with him’ (carried him in the same canoe as them)
b. tpwi i-mp-awkura-pampay-wapal-kia-k
sago.X.pl X.pl.nom-3dl.erg-gather-applkin-climb-night-irr
‘they (dl) were gathering sago and carrying it up’

Finally, the last Yimas applicative is a rare tur-, which again expresses an idea related
to the Coastal Marind associative, namely that the actor participant is pursuing an
animate applied argument driving it to a place; ‘chase’ is a good translation:

(30) Yimas
pu-n-tur-awramuŋ-k ma-nan
3pl.nom-3sg.erg-applCHASE-enter-irr male.cult.house-obl
‘he drove them into the male cult house’
 12 Applicatives in Papuan languages 363

4.3 Barupu (Piore River sub-family, Macro-Sko family;

Corris 2005; Donohue 2003)

Barupu has far and away the most complex inventory of applicatives known in any
Papuan language and one of the richest of any language anywhere. It has no less than
thirteen distinct applicative forms, which are divided into two large groups: those which
take subject agreement prefixes or infixes like main verbs and hence clearly derive
from earlier serial verb constructions, partially illustrated in the examples of (4) above,
and those which lack such agreement. Those which lack agreement are further divided
into two sub-groups: those which denote locational notions and those which do not.
The most basic and general non-locational applicative affix is -nâ. This suffix is
reconstructable as an applicative in Proto-Macro Sko (Donohue 2004) and has a wide
range of meanings, but generally clustered around a notion of desired object, as in these
examples (I will simply gloss it as appl):

(31) Barupu
a. rua k-a-ko k-a-rói-nâ kamo
bow r-3sg.m.subj-get r-3sg.m.subj-stand-appl door
‘he got the bow and stood with it at the doorway’
b. k-en-tova-nâ-re bǐyó
r-1sg.f.subj-walk_around-appl-3pl.f.obj cassowary
‘I’m hunting cassowary’
c. era k-ama-yôyó-nâ-ni?
q r-2sg.m.subj-dream-appl-1sg.f.obj
‘did you (sg) dream about me?’

The other two non-locational applicatives are more transparent semantically:

(32) Barupu
a. -kê adversative (negative affect on applied participant)
kua Betty á k-u-ai-kê-u
d pn rain r-3sg.f.subj-rain-appladvs-3sg.f.obj
‘it’s raining on Betty’ (bad for her)
b. -bo without
mônrai n-opu-tîtí-bo-na
dance irr-2pl.m.subj-dance-applwithout-1sg.m.obj
‘you (pl) keep dancing without me’

The locational applicatives number four. In Barupu only inherently locational nouns
like ôro ‘house’ can be used as bare nouns without applicativization; using any other
type of noun in a locational usage requires one of these locational applicative suffixes:
364 William A. Foley

(33) Barupu
a. k-e-ké<m>í ôro
r-1pl.f.subj-sit<1pl.f.subj> house
‘we sit in the house’
b. ✶k-e-ké<m>í aka
r-1pl.f.subj-sit<1pl.f.subj> father
‘we are sitting on father’
c. k-e-ké<m>í-tá-ka aka
r-1pl.f.subj-sit<1pl.f.subj>-applON-3sg.m.obj father
‘we are sitting on father’

Here are examples of the four locative applicatives:

(34) Barupu
a. -tâ on
k-a-kéi-tá âi niau
r-3sg.m.subj-sit-applon tree log
‘he is sitting on a log’
b. -para under
k-rói-para-i aniania ku
r-stand-applunder-3pl.m.obj tree.sp root
‘he stood under them at the roots of the aniania tree’
c. -ya near
k-en-úte k-en-no<n>i-ya-mu
r-1sg.f.subj-walk r-1sg.f.subj-go_along<1sg.f.subj>-applnear-2sg.f.obj
‘I walked past you (sg)’
d. -rǒmó amid
n-e-ké<n>í-romǒ-ré
irr-1sg.f.subj-sit<1sg.f.subj>-applamid-3pl.f.obj
‘I would sit among them (pl)’

Interestingly when these locative applicatives are used with transitive verbs which
already have overt pronominal object suffixes, the applicative suffix follows the object
suffix and any object suffix for the introduced applied participant then follows that.
Hence, like in Mountain Arapesh (6f, g), the applicative suffix adds a core participant
but otherwise does not affect the argument structure nor the basic inflectional pattern
of the main verb:

(35) Barupu
n-en-ere-ma-tá-ka
irr-1sg.f.subj-put-2sg.m.obj-applon-3sg.m.obj
‘I will put you (sg) on him’
 12 Applicatives in Papuan languages 365

This pattern of the applicative suffix after object inflection suggests that they were
originally independent words, possibly verbs, like the next set of applicative suffixes.
(they are unlikely to have been prepositions, as the language has none and is of OV, left
branching typology). Their current lack of subject inflection probably is due to phono-
logical factors (Corris 2005: 255).
There are six subject-agreeing applicatives in Barupu. Four of them were exempli-
fied in (4), repeated here as (36) to which a fifth is added:

(36) Barupu
a. -ě from
k-en-úte-n-ě-mú
r-1sg.f.subj-walk-1sg.subj-applfrom-2sg.f.obj
‘I walked away from you (sg)’
b. -ô cause
k-en-ké<n>í-n-ô-wa
r-1sg.f.subj-sit<1sg.subj>-1sg.subj-applcause-3sg.m.obj
‘I’m staying behind because of him’
c. -o for
k-en-úte-n-o-mu
r-1sg.f.subj-walk-1sg.subj-applfor-2sg.f.obj
‘I walked for you (sg)’
d. -î with
n-en-úte-n-î-mu
irr-1sg.f.subj-walk-1sg.subj-applwith-2sg.f.obj
‘I will walk to be with you (sg)’
e. -ái surround
á k-u-ai-r-a<r>í-ni
rain r-3sg.f.subj-rain-3sg.subj-applsur<3sg.subj>-1sg.f.obj
‘the rain is blocking me’ (surrounding me so I can’t go out)

Although these applicative suffixes very much look like verbs due to their inflections, and
undoubtedly their use as applicatives derives from earlier serial verb constructions, only
one of them, -o for, is synchronically a verb root, not surprisingly, our familiar verb in
this benefactive usage -o ‘give’. Note that it is possible for multiple applicative suffixes to
appear on the same verb, including multiple subject agreeing applicative suffixes (37c):

(37) Barupu
a. aro n-en-râivi-tá-u-n-o-a
greens irr-1sg.f.subj-cook-applON-3sg.f.obj-1sg.subj-applfor-3sg.m.obj
ám něni
husband 1sg.f.poss
‘I’ll cook greens on it for my husband’
366 William A. Foley

b. k-en-úte-nâ-ka-n-î-mu
r-1sg.f.subj-walk-appl-3sg.m.obj-1sg.subj-applwith-2sg.f.obj
‘I’m bringing him back to you’
c. k-e-ké<n>í-n-ě-
r-1sg.f.subj-sit<1sg.f.subj>-1sg.subj-applfrom
-mú-n-î-ya
-2sg.f.obj-1sg.subj-applwith-3sg.m.obj
‘I stayed away from you (sg), I stayed with him’

There is one further subject-agreeing applicative: the benefactive. Unlike all the others,
it takes a prefix, not a suffix. Like Alamblak (9) and Yimas (10, 11), Barupu has two
applicatives which cover the semantic range of benefactive, -o for in (36c) and (37a)
above and the benefactive prefix e-. Consider the following contrastive examples:

(38) Barupu
a. n-an-aro-n-o-ma
irr-1sg.m.subj-bring_down-1sg.subj-applfor-2sg.m.obj
‘I will bring it down to you (sg)’
b. n-em-e-na-kô<m>e
irr-2sg.m.subj-applben-1sg.m.obj-bring_up<2sg.m.subj>
‘bring it up for me’

The difference between these is subtle, but seems to revolve around prior possession
of the object being transferred to the beneficiary. The benefactive prefix entails that
the actor already has possession of the object that he will give to the beneficiary, while
-o for suggests that the actor will have to go and acquire the object beforehand. This
would also explain the usage of -o for in (37a), although the semantics of (36c) seems
entirely different. In light of its anomalous character, here are a few more examples of
the benefactive applicative prefix in Barupu:

(39) Barupu
a. n-em-e-nǎ-m-á
irr-2sg.m.subj-applben-1sg.m.obj-2sg.m.subj-eat
‘eat (it) for me’
b. k-er-e-a-r-ere-tá âi
r-3sg.f.subj-applben-3sg.m.obj-3sg.f.subj-put-applon tree
‘she put (them) for him on the tree’
 12 Applicatives in Papuan languages 367

5 Polysemy, homophony and extensions

of applicative morphemes

5.1 Adverbial uses

As we have seen thus far, applicative morphemes in Papuan languages commonly

derive from older compound or serial verb constructions which have undergone gram-
maticalization and re-analysis to a greater or lesser extent. Adverbials also commonly
arise from the same source, for example in Yimas:

(40) Yimas
a. source verb root: makɲc- ‘move quietly/stealthily’
wurmpl pla-mpu-makc-mpi-wuntampwi-k
flute.VI.dl VI.dl.nom-3pl.erg-quietly-adv-blow_on-irr
‘they (pl) quietly played the flutes’
b. source verb root: pramuŋ- ‘sleep’
impa-n-taŋ-praŋka-mpi-aypu-kia-k
3dl.nom-3sg.erg-com-sleeping-adv-recline-night-irr
‘he slept with them (dl)’

Given these facts, it is not surprising that some of the applicative affixes in Yimas have
adverbial uses in which their basic meaning holds, but they do not add arguments:

(41) Yimas
a. ya-n-taŋkway-wampaki-pra-k
V.pl.nom-3sg.erg-applvis-throw-toward-irr
‘he threw them down carefully’
b. aympanuŋ ku-mp-ira-yawra-k
pestle.X.sg X.sg.nom-3dl.erg-applall-pick.up-irr
‘they (dl) fetched a stick’

The prefix taŋkway- preserves its meaning of visually monitoring an event in (41a), but
here fails to introduce a recipient participant who is watched during the performance
of the event, so simply best translates as ‘carefully’ doing the action. Similarly, the alla-
tive meaning of ira- still obtains in (41b), but again in the absence of a goal participant
being introduced, the intended goal meaning is taken to refer to the actor, so that ‘pick
up’ becomes ‘fetch’, i.e. what is collected is intended to be in the possession of the actor.
368 William A. Foley

5.2 Applicative/causative homophony

The homophony of applicative and causative morphemes has been noted in a number
of other languages, perhaps the best known case being Indonesian (Musgrave, Arka and
Rajeg, this volume), and this is true also of some Papuan languages. We have already
seen above (9b) that Alamblak has two benefactive applicative suffixes, one of which is
homophonous with the verb root hay- ‘give’. Alamblak has no less than four causative
prefixes, contrasting direct versus indirect causation and spatiotemporal contiguity, but
one of these is none other than hay-:

(42) Alamblak (Bruce 1984)

hɨnu-t doh-t hay-ni-më-t-t
high.water-f canoe-f caus-go-rem.past-3sg.f.subj-3sg.f.obj
‘the high water took the canoe away’

Note that when hay functions as a benefactive applicative marker it occurs as a suffix,
but when a causative, it is a prefix. Also note that hay- is clearly a causative prefix in
(42); the semantics of ‘give’ is completely bleached out, unlike in the corresponding
usage of the benefactive suffix. Still (42) is clearly the result of a re-analysis of an earlier
serial verb construction involving the verb ‘give’ in causative relationship with a fol-
lowing verb as in (43):

(43) Alamblak
yima-r hay-noh-më-r-a
person-m give-unconscious-rem.past-3sg.m.subj-1sg.obj
‘a man gave me (something) (causing) me (to become) unconscious)’

Kopar (Lower Sepik sub-family, Lower Sepik-Ramu family) is another language in which
the causative and applicative morphemes are homophonous. The causative prefix in
this language is t-, possibly derived historically from a verb tu- ‘hit, kill’, now lost in
Kopar but preserved in its sister Yimas:

(44) Kopar
a. akən naŋgun ma-na pet t-mbu-t-kam-a
sun skin 1sg-poss dark pfv-3.erg-caus-become-pfv
‘the sun darkened my skin’
b. indan mbu-t-riker-ana-k
house 3.erg-caus-get_up-3sg.dat-rem.past
‘she erected a house for her’
c. ku-t-rərəja-bi-duku paret ŋgari
tr.imp-caus-shake-im.fut-2pc outside dat
‘you (pc) shake (it) until (it comes) out!’
 12 Applicatives in Papuan languages 369

A prefix of the same form functions as a comitative applicative:

(45) Kopar
a. ŋga-t-ra-(a)r-aŋg-naya
inv-applcom-stay-prog-pres-1sg
‘she looks after me’ (literally ‘stays with me’)
b. mayndəpak mbu-t-ra-(a)r-oro-k-ududu
husband.pl 3.erg-applcom-stay-prog-ext-rem.past-3pc
‘they (pc) remained with the husbands for a while’
c. Wak yo mbu-t-kar-ar-oro-k-ondu
pn d 3.erg-applcom-walk-prog-ext-rem.past-3pl
‘they (pl) walked around with Wak for a while’

The connection between the causative and comitative uses of this prefix can be gleaned
from examples like the following:

(46) Kopar
kiŋgep mbu-t-rapo~rapo-sa-(a)r-oro-k-ondu
ladder 3.erg-applcom/caus-run~iter-in-prog-ext-rem.past-3pl
‘they (pl) kept running around in with a ladder’

Obviously, a ladder, being inanimate, cannot run around under its own power; it needs
to be caused to run. At the same time, it cannot be run around without someone or
some people holding it while they run around with it; hence the ladder accompanies
them while they are running with it. It appears likely that it was this use of the causa-
tive prefix with such motion verbs that is the source of the comitative meaning of this
prefix.

5.3 Applicative/detransitivizer homophony

A few languages of the southern border region between Papua New Guinea and the
Indonesian province of South Papua exhibit a quite striking and unexpected homoph-
ony between an applicative affix and a detransitivizing affix, treated here simply as a
valence marker and glossed val. This is a feature of the languages of the Tonda sub-fam-
ily of the Yam family such as Ngkolmpu (Carroll 2017) and Komnzo (Döhler 2018). Con-
sider these examples from Ngkolmpu:

(47) Ngkolmpu (Carroll 2017)

a. Markus-w pr pi s-wance-y
pn-sg.erg tree 3.abs 3.U-fall-sg.A.nr.past
‘Markus felled the tree’
370 William A. Foley

b. Markus t-a-wance-y
pn mid.pfv-val-fall-sg.A.nr.past
‘Markus fell’
c. Markus-w pr pi nson b-a-wance-y
pn-sg.erg tree 3.abs 1sg.dat 1sg.U-val-fall-sg.A.nr.past
‘Markus felled the tree for me’

(47a) is a clause with a transitive verb: the subject is case marked with the ergative suffix
and the object with the absolutive postposition. In addition, the verb is inflected transi-
tively, with a prefix agreeing in person with the absolutive object and a suffix agreeing
in number with the ergative subject. This clause can be detransivitized into an intran-
sitive middle construction by the prefix a- in (47b), indicating that the action affects the
actor, not the direct object, as in the transitive clause of (47a). The now intransitive verb
only has subject agreement for number by a suffix and in addition takes an intransitive
middle prefix for perfective aspect t-. Finally, (47c) takes a prefix a- of exactly the same
form and distribution, but now functioning as an applicative prefix in order to add a
beneficiary as a core argument. This participant appears as the dative case marked
pronominal for first person singular, but also as a prefix to the verb as its direct object,
exactly as did the absolutive case marked direct object in (47a), although both person
and number are now indicated instead of just person as in (47a).
This is a very strange pattern: the same form both reduces and increases transitiv-
ity in Ngkolmpu. But an explanation may be forthcoming from the languages of another
sub-family of the Yam family, those of the Nambu sub-family like Nen. In this language
there is a reflexive-reciprocal prefix which detransitivizes verbs and has several allo-
morphs, but one of these is a-:

(48) Nen (Evans 2015)

bm k-a-waka-ta-∅
2.abs mid-refl-look.at-ndl.ipfv-2sg.imp
‘look at yourself’

And a prefix of the form (w)a-, which functions as an applicative to promote beneficiar-
ies to core argument and direct object status:

(49) Nen (Evans 2015)

ynd nu n-a-w-az-na-n be-gta
1sg.erg water 2sg.U-ben-caus-pour-ipfv.npast-1sg.A 2sg-dat
‘I’m pouring water for you (sg)’

A quite plausible scenario for the origin of the strange homophony in Ngkolmpu is leve-
ling of the allomorphs of the two prefixes we find in Nen, resulting in homophony for
what were earlier two distinct prefixes.
 12 Applicatives in Papuan languages 371

5.4 External Possession

In a number of Papuan languages, the benefactive applicative marker is also used to

indicate external possession. External possession constructions occur when a human
possessor of a nominal argument occurs as a pronominal agreement affix on the verb
or otherwise as its core argument, instead of being realized as a possessor constitu-
ent internally in a noun phrase. Externalization of possessors crosslinguistically occurs
most commonly from direct objects, but some languages extend this to subjects of
intransitive unaccusative verbs, and sometimes even beyond, as Kopar, for instance,
allows externalization of possessors from oblique nominals. The types of nouns from
which possessors can be externalized is also commonly restricted to those which are
inalienably possessed, typically those which denote body parts, but again not always.
Barupu is a language which uses the benefactive applicative affix to mark external pos-
session and then expresses the externalized possessor through pronominal agreement
for direct object:

(50) Barupu (Corris 2005)

a. bo k-en-e-ma-yǎrá
bum r-1sg.m.subj-applben-2sg.m.obj-see
‘I can see your (sg) bum’
b. tó mú n-ep-e-n-tón
breast milk irr-3pl.m.subj-applben-1sg.f.obj-drink
‘they will drink my breast milk’
c. anoku k-er-e-nâ-irai-r-o-re
story r-3sg.m.subj-applben-1sg.m.obj-say-3sg.m.subj-applfor-3pl.f.obj
‘he told my stories to them’

Yimas has an interesting twist in its external possession constructions. Externalized

possessors of inalienably possessed nouns like body parts are expressed in basic tran-
sitive or ditransitive constructions, with a bound pronominal from the dative series to
express the possessor:

(51) Yimas
a. ŋarwa wa-ŋa-kwalca-t
penis.IX.sg IX.sg.nom-1sg.dat-get_up-pfv
‘I have an erection’
b. yampaŋ k-mpu-ŋa-kra-t
head.VI.sg VI.sg.nom-3pl.erg-1sg.dat-cut-pfv
‘they (pl) cut my hair’

But when the possessed nouns are alienably possessed, the applicative prefix taŋ-, nor-
mally beneficiary/comitative, is used, again with a bound pronominal for the external-
372 William A. Foley

ized possessor from the dative series; however, the meaning here is typically malefac-
tive for the externalized possessor:

(52) Yimas
a. impram p-ŋa-na-taŋ-tat-n
basket.VII.sg VII.sg.nom-1sg.dat-pres-appl-hold-pres
‘(they) seize my basket’
b. manpa na-kay-taŋ-awkura-kr-mpun
crocodile.sg 3sg.nom-1pl.erg-appl-get-rem.fut-3pl.dat
‘we (pl) will steal their (pl) crocodile’

6 Lookalike constructions

6.1 Promiscuous promotion to core from oblique without

applicative marking

Languages sometimes permit promotion of an oblique participant to core argument

status without any overt morphological marking, especially when that participant is
a beneficiary. English, for example, is one such language: John baked a cake for Mary,
John baked Mary a cake. Such promotion of beneficiaries to core without any applica-
tive marking does not appear to be common among Papuan languages, but Coastal
Marind is one such language, as in (20a). Yeri (West Palei sub-family, Torricelli family)
is another, but permits this construction to extend far beyond beneficiaries to oblique
participants with a range of semantic roles. Like Torricelli family languages generally,
Yeri inflects verbs with bound pronominals for both subjects and objects. In Torricelli
languages bound subject pronominals are prefixed, but bound object pronominals can
be prefixes or suffixes. In some Torricelli languages like Urim, bound object pronomi-
nals are always suffixes or enclitics (see [17]), but in others like Mountain Arapesh, they
can be either prefixes or suffixes depending on verb class (see [6f, g]). Yeri is particu-
larly interesting in that the split between prefix or suffix for bound object pronominals
is determined by person; first and second person bound object pronominals are always
prefixed:

(53) Yeri (Wilson 2017)

a. y-b-ogera
2pl.subj-1sg.obj-chase.r
‘you (pl) chased me’
b. wo n-w-osia
sun 3sg.m.subj-1pl.obj-heat.r
‘the sun heats us’
 12 Applicatives in Papuan languages 373

c. lawiaki m-y-asɨkera
long_ago 1sg.subj-2.obj-carry_on_hip.r
‘long ago I carried you’

Like English pronouns, bound object pronominals in Yeri distinguish number in first
person, but not in second. Third person bound pronominals are suffixes (sometimes
infixed for a minority of the most common verb roots, just over 10%). When suffixed,
they cannot be directly added to the verb root. Rather before suffixation of the bound
object pronominal, the verb root must first be augmented by another suffix with a
number of allomorphs morphologically determined by the verb, here simply glossed
as aug:

(54) Yeri (Wilson 2017)

a. y-ogera-we-i
2pl.subj-chase.r-aug-3pl.obj
‘you (pl) chased them’ (compare [53a])
b. w-okɨ hilian w-gare-wa-n
3sg.f.subj-use.r sand 3sg.f.subj-dig.r-aug-3sg.m.obj
‘she took the sand and dug it’

These bound object pronominals can be used to denote participants functioning as

direct objects for a number of semantic roles that are typically realized as oblique con-
stituents in other languages, but with no other morphological marking:

(55) Yeri (Wilson 2017)

a. recipient
te-n ta n-b-nobia hem
3-m.sg fut 3sg.m.subj-1sg.obj-tell.r 1sg
‘he will tell me’
b. beneficiary
ye ta n-w-a<ne>gutɨ nanu-la hebi
2sg fut 2sg.subj-1pl.obj-burn.r<3sg.m.obj> fish-sg 1pl
‘you (sg) will cook a fish for us’

Note the triple agreement in this example. Promotion of the beneficiary to core with
this transitive verb agutɨ- ‘burn, cook’ derives a ditransitive verb. The original direct
object nanu-la fish-sg continues to agree via a bound object pronominal as it would in
an underived verb, i.e. by infixation, while the new added object agrees via the normal
prefix for first person direct objects. There is no demotion of the original direct object
when the oblique role is promoted to object function. With a third person benefactive,
suffixation with augmentation occurs, again along with infixation for the original direct
object:
374 William A. Foley

(55) Yeri (Wilson 2017)

c. beneficiary
hem m-a<ne>gutɨ-da-n nanu-la te-n
1sg 1sg.subj-burn.r<3sg.m.obj>-aug-3sg.m.obj fish-sg 3-m.sg
‘I cooked a fish for him’
d. external possessor
huba n-y-ilkial yewal
hawk 3sg.m.subj-2.obj-pull.irr eye
‘may the hawk pull out your eyes’
e. cause
yem tomal y-ana y-w-arwal hebi
2pl proh 2pl.subj-come 2pl.subj-1pl.obj-cry.r 1pl
‘don’t cry because of us’
f. goal
ta n-gei-ka-n nebal yewal-tɨ
fut 2sg.subj-leave.r-aug-3sg.m.obj tree eye-sg
n-anor-e-n yot-ua-n
3sg.m.subj-descend-aug-3sg.m.obj dem.fr.dist-m.sg
‘you (sg) will put the cocoa beans into it there’
g. location
hasieki-l ∅-awɨl-a-i woli woli
fire-pl 3pl.subj-hang.r-aug-3pl.obj side side
‘they have many lights on them’
h. accompaniment
puetɨ yot-u-n n-b-giekɨr
betelnut dem-nr.dist-m.sg 3sg.m.subj-1sg.obj-bend.r
‘that betelnut palm would bend with me’

In addition to this morphologically unmarked promotion to object, Yeri also has a

canonical applicative marked by the suffix -ki; this can occur with (56b, c) or without
(56a, d, e) a bound pronominal suffix. Most commonly it introduces a benefactive core
participant, but can indicate other roles as well (56c, d, e):

(56) Yeri (Wilson 2017)

a. beneficiary
la magɨl ∅-y-ati-ki hasieki-l yem
past who 3pl.subj-2pl.obj-blow_on-appl fire-pl 2pl
‘who blew on the fires for you (pl)?’
b. beneficiary
te-n n-ga-<∅>kua-ki-da-m motɨ
3-m.sg 3sg.m.subj-wash.r<3sg.f.obj>-appl-aug-pl.obj pot
‘he washed pots for them’
 12 Applicatives in Papuan languages 375

c. cause
∅-b-ie<m>kewa-ki-da-∅
3sg.f.subj-1sg.obj-be-angry.r<ipfv>-appl-aug-3sg.f.obj
‘she got angry with me because of it’
d. cause
te-i la ∅-ogera-we-i-ki maŋa-∅
3-pl past 3pl.subj-chase-aug-pl.obj-appl what-f.sg
‘what did they chase them for?’
e. goal/purpose
hebi la ∅-aro nania-ki nanu-bia
1pl past 1pl.subj-go.r go_in.r-appl fish-pl
‘we went in for the fish’

6.2 Re-analysis of applicative morphemes into transitivity

markers

The languages of the Finisterre-Huon sub-family of the Trans New Guinea family have
re-analyzed what were applicative markers into conjugation markers for verb classes.
Consider again the kind of applicatives through verb compounding with ‘give’ or ‘put’
for marking benefactives in Trans New Guinea languages by exemplified in (8). Already
in Lower Grand Valley Dani this pattern of verb compounding to express pronominal
direct objects is extended beyond only marking benefactive to indicating nearly all
pronominal direct objects with human referents. In this language, very few transitive
verbs can be inflected underived with bound pronominal prefixes for direct objects
with human referents; in fact, the only verbs that can are (w)at- ‘hit, kill’, hei- ‘put’
and ha- ‘see’, the latter two of which have complex irregular and sometimes suppletive
conjugations:

(57) Lower Grand Valley Dani (Bromley 1981)

a. n-at-h-e
1sg.obj-hit-r-3sg.subj
‘he hit me’
b. h-akeik-h-e
2sg.obj-put-r-3sg.subj
‘they put you (sg)’

All other verbs require one of the four supporting verbs, hei- ‘put’, ha- ‘see’, et- ‘give’
or ap- ‘do for’, to be compounded with them in order to co-occur with a bound object
pronominal, which are prefixed to the supporting verb:
376 William A. Foley

(58) Lower Grand Valley Dani (Bromley 1981)

a. hetamoʔ-n-et-h-e
scold-1sg.obj-give-r-3sg.subj
‘he scolded me’
b. woʔ-n-ap-h-e
take-1sg.obj-do_for-r-3sg.subj
‘he received me’
c. wam woʔ-n-et-h-e
pig take-1sg.obj-give-r-3sg.subj
‘he gave me a pig’

Note that et- ‘give’ and ap- ‘do for’ do not occur as independent verbs, only as support-
ing verbs which must be compounded with another main verb, as in (58). A given main
verb can often be compounded with multiple supporting verbs, for example pa- ‘cut,
sever, divide’:

(59) Lower Grand Valley Dani (Bromley 1981)

a. wvt pa-n-ap-he-e
initiation cut-1sg.obj-do_for-r-3sg.subj
‘he initiated me’
b. wam pa-n-et-h-e
pig cut-1sg.obj-give-r-3sg.subj
‘he selected a pig for me’ (compare [58c])
c. n-esi pa-n-eeik-h-e
1sg.poss-hair cut-1sg.obj-see-r-3sg.subj
‘she cut my hair’ (external possession)
d. hakki pa-n-akeik-h-e
bananas cut-1sg.obj-put-r-3sg.subj
‘he cut some bananas and put them aside for me’

Languages of the Finisterre-Huon sub-group of the Trans New Guinea family also
require supporting verbs to carry bound object prefixes for transitive verbs. However,
in these languages the semantic contrasts we find between them in Lower Grand
Valley Dani is lost, so that each transitive verb occurs with only one supporting verb. In
Selepet, for example, the supporting verbs are ek- ‘see’, oho- ‘hit’ and ihi- ‘give’, but the
semantic contrasts motivating their usage in Lower Grand Valley Dani no longer holds,
so the selection of supporting verb by any main verb seems synchronically arbitrary:

(60) Selepet (McElhanon 1970)

a. gɔi-n-ek-sa-p
cut-1sg.obj-see-im.past-3sg.subj
‘he cut me’
 12 Applicatives in Papuan languages 377

b. pene-n-ihi-a-p
join-1sg.obj-give-im.past-3sg.subj
‘he joined me’
c. tɔn-n-oho-a-p
help-1sg.obj-hit-im.past-3sg.subj
‘he helped me’

With 1. the requirement that all transitive verbs save ‘see’, ‘hit’ and ‘give’ co-occur with
a supporting verb, 2. the collapse of the semantic distinctions motivating the choice of
supporting verb, and finally 3. the restriction of one and only one supporting verb spec-
ified for each main verb, the synchronic system of supporting verbs in Selepet simply
reduces to transitivity marking morphology determined by verb class, fundamentally
a set of conjugation classes. In some languages of the Madang sub-family of Trans New
Guinea, the system collapses further to a single transitivizing suffix, as in Tauya:

(61) Tauya (MacDonald 1990)

a. ʔumu-a-ʔa
die-3.sg.subj-ind
‘he died’
b. ʔumu-fei-fe-a-ʔa
die-3sg.obj-tr-3sg.subj-ind
‘he killed him’

With transitive verb roots, the transitivizing suffix -fe derives ditransitive verbs, typi-
cally with beneficiaries, so here preserves something of its older applicative function so
well attested elsewhere in Trans New Guinea languages:

(62) Tauya (MacDonald 1990)

a. wate eʔi-i-ʔa
house make-3pl.subj-ind
‘they built a house’
b. wate eʔi-ya-fe-i-ʔa
house make-1sg.obj-tr-3pl.subj-ind
‘they built me a house’

The language isolate Yale also exemplifies a system of verbal inflection that goes back
to verb compounding with earlier supporting verbs now grammaticalized into conjuga-
tion markers. Verb roots in Yale belong to one of four conjugation classes. The primary
distinction is between those verb roots which take prefixal and suffixal agreement for
subjects and those which only take suffixes. The very few verb roots which take pre-
fixal subject agreement belong to one class; all other verb roots take exclusively suffixal
378 William A. Foley

agreement and in turn break down into three conjugation classes depending on the
consonant that occurs between the subject and object bound pronominal suffixes:

(63) Yale (Aannestad, Campbell, and Campbell 2020)

a. hui-no-d-∅-ë-o
see-1sg.subj-d-3sg.f.obj-1sg.subj-decl
‘I see her’
b. swa-no-to-m-ë
wash-1sg.subj-t-3pl.obj-1sg.subj
‘I wash them’

The three consonantal conjugation markers are as follows:

-d the largest class, with no obvious semantic or syntactic grounds for the
grouping; includes both transitive and intransitive verbs
-t found with transitive verbs whose subjects cause a change of state in their
objects, e. g. ‘slice’, ‘cut’, ‘sharpen’, ‘wash’, ‘cause something’
-b a small class of intransitive verbs denoting involuntary events, e. g. ‘die’, ‘sleep’

Individual verb roots can shift between these conjugation classes with a corresponding
change in semantics:

(64) Yale (Aannestad, Campbell, and Campbell 2020)

a. tëbo-dë-de
be.ill-3sg.m.subj-d
‘he is ill’
b. tëbo-do-te-d-o
be.ill-3sg.m.subj-t-3sg.m.obj-decl
‘he made him ill’

This system is highly reminiscent of the patterns of conjugation by supporting verbs

that we found in Trans New Guinea languages like Lower Grand Valley Dani and
Selepet. These conjugation markers were originally verb roots in their own right, and in
an earlier period of the language, all verbs took prefixes for subject agreement, though
only three now preserve this older system. As in Selepet, over time the supporting verbs
have become morphologically bound to the main verb resulting in the synchronic
system of verb inflection:

(65) Yale (Aannestad, Campbell, and Campbell 2020)

✶
hui no-d-∅-ë > hui-no-d-∅-ë
see 1sg.subj-give-3sg.f.obj-1sg.subj see-1sg.subj-d-3sg.f.obj-1sg.subj
‘I see her’
 12 Applicatives in Papuan languages 379

Indeed, it is obvious that the -d of the d-conjugation class is sourced in none other than
the verb root ‘give’ (compare with the Selepet form in [60b]):

(66) Yale (Aannestad, Campbell, and Campbell 2020)

a. kide-ne-d-∅-ë-o
abandon-1sg.subj-d-3sg.f.obj-1sg.subj-decl
‘I left it behind’
b. në-d-∅-ë-o
1sg.subj-give-3sg.f.obj-1sg.subj-decl
‘I gave (it) to her’

This re-analysis of supporting verbs which originally functioned like applicatives into
licensors for bound object pronominals for human referents, especially speech act par-
ticipants like first and second person, and ultimately into morphologically determined
transitivity markers is attested among Papuan languages of various families, but is
especially widespread among sub-families of the vast Trans New Guinea family.

6.3 Re-analysis of applicatives into pronominals

Abui (Alor-Pantar sub-family, Trans New Guinea family) (Kratochvíl 2007, 2011) illus-
trates another pathway of re-analysis for applicative morphemes, but this one seems
much less common outside of languages spoken on Alor Island. Abui and some other
languages of Alor such as Adang (Robinson and Haan 2014) and Kamang (Schapper
2014) have multiple sets of bound pronominals, each associated with a different seman-
tics. Here is the basic system in Abui:

Table 1: Abui pronominals (Kratochvíl 2011: 591).

A U
pat rec loc goal ben
sg 1 na na- no- ne- noo- nee-
2 a a- o- e- oo- ee-
pl 1excl ni ni- nu- ni- nuu- nii-
1incl pi pi- pu-/po- pi- puu-/poo- pii-
2 ri ri- ri-/ro- ri- ruu-roo- rii-
3 ∅ ha- ho- he- hoo- hii-

The actor pronouns are free forms, but all five of the undergoer series of pronouns
are bound prefixes. The basic opposition of actor versus undergoer is correct here,
as Abui clausal structure is organized along rough semantic lines that contrast these
notions, not subject and object: schematically, the actors of both transitive and unerga-
380 William A. Foley

tive intransitive verbs occur with the actor set of pronouns, while undergoers of both
transitive and unaccusative intransitive verbs select one of the undergoer pronominal
prefixes. Formally it is clear that all of the undergoer prefixes except those for patients
are derived from the patient form plus a vowel: rec o, loc e, goal oo and ben ee. These
vowels look like applicative prefixes and certainly their syntactic behavior with transi-
tive verbs suggests that:

(67) Abui
a. na a-ruidi
1sg.A 2sg.pat-wake_up.pfv
‘I woke you (sg) up’
b. Fanmalei no-k yai
pn 1sg.rec-throw laugh.pfv
‘Fanmalei laughed at me’
c. a palootang ne-l bol
2sg.A rattan 1sg.loc-give hit
‘you (sg) hit me with a stick’
d. a noo-dik
2sg.A 1sg.goal-prick
‘you (sg) are poking me’
e. ma na ee-bol
be 1sg.A 2sg.ben-hit
‘let me hit (it) for you (sg)’

The same verb root can commonly occur with multiple undergoer prefixes with differ-
ent meanings:

(68) Abui
a. wik ha-wik no-wik
‘carry’ 3.pat-carry 1sg.rec-carry
‘carry him’ ‘carry for myself’
he-wik noo-wik nee-wik
3sg.loc-carry 1sg.goal 1sg.ben
‘carry it’ ‘let me carry’ ‘carry for me’
b. rumai ha-rumai no-rumai
‘strong’ 3sg.pat-strong 1sg.rec-strong
‘strengthen it’ ‘I feel strong’
he-rumai noo-rumai nee-rumai
3sg.loc-strong 1sg.goal-strong 1sg.ben-strong
‘it is strong’ ‘rely on me’ ‘strong for me’
 12 Applicatives in Papuan languages 381

All of this is remarkably reminiscent of the different semantics for the choice of sup-
porting verbs in Lower Grand Valley Dani exemplified in (59), and in fact Kratochvíl
(2007: 2005) argues that the contrastive vowels, plausibly applicatives in the examples
in (67), of at least some of the undergoer prefixes go back to older incorporated verb
roots, ✶a ‘be at’, ✶e ‘add, continue’, ✶o ‘point’:

(69) Abui (Kratochvíl 2007: 2005)

a. ✶na-a fanga > ✶n-a=fanga > na-fanga
1sg-be_at say 1sg-be_at=say 1sg.pat-say
‘tell/order me’
b. ha-e fanga > ✶h-e=fanga > he-fanga
3sg-add say 3-add=say 3.loc-say
‘say it’
c. ✶na-o fanga > ✶n-o=fanga > no-fanga
1sg-point say 1sg-point=say 1sg.rec-say
‘scold me’

This scenario of re-analysis would suggest that Abui is just another example of the
common process in Papuan languages of re-analyzing verbs in compound or serial
verb constructions into applicative morphemes. Like the use of supporting verbs in
other Trans New Guinea languages, the original lexical semantics of these verbs has
been bleached, so that they essentially signal the typical semantic roles of oblique par-
ticipants, like canonical applicative affixes crosslinguistically. But a purely applicative
analysis of these forms simply cannot be synchronically correct for Abui, for whatever
the plausible origin of these pronominal prefixes in verbs and whatever their appar-
ent applicative functions in (67), this analysis is no longer tenable, since they can now
be used to denote undergoers of unaccusative intransitive verbs. Appearing on single
argument intransitive verbs, an applicative function is not possible:

(70) Abui
a. na-kaai
1sg.pat-stumble
‘I stumbled’
b. no-bui
1sg.rec-short
‘I am short’
c. he-beeka
3sg.loc-bad
‘it is bad’
d. noo-lila
1sg.goal-hot
‘I feel hot’
382 William A. Foley

So, while the contrasting vowels of the sets of undergoer prefixes in Abui may have
started out as verbs, which were then re-analyzed into applicative prefixes in the
common pathway we have seen in Papuan languages, they can no longer be analyzed
as such. Rather, they are synchronically simply semantically contrastive sets of bound
pronominals. What seems to be responsible for this rare development in Abui is its
thoroughgoing actor-undergoer split. Due to the basic actor-undergoer contrast that is
pivotal to the clause level grammar of Abui and crosscuts transitivity, once undergo-
ers of transitive verbs exhibited these semantic contrasts in pronominals, this would
spread to undergoers of intransitive verbs, rendering any applicative function obsolete
and any such analysis of these forms untenable.

6.4 Information structure markers from applicatives

Some languages of the Lower Ramu family illustrate yet another type of re-analysis of
an earlier applicative, here into an information structure marker. Consider the case of
Watam. In this language all oblique arguments are case marked by postpositions, while
subjects and inanimate direct objects are unmarked (animate direct objects are usually,
though not obligatorily, marked with the dative postposition mo). Word order of core
arguments is free, but they must precede the verb; oblique arguments can follow the
verb, but crucially they cannot immediately precede the verb:

(71) Watam
a. was nakan padoŋ an mo panai-ri nimoŋ mba
wind big tree d do bend-past night loc
‘a big wind bent the tree at night’
b. nimoŋ mba padoŋ an was nakan mo panai-ri
night loc tree d wind big do bend-past
‘a big wind bent the tree at night’
c. ✶was nakan padoŋ an nimoŋ mba mo panai-ri
wind big tree d night loc do bend-past
d. namot an yak mo ruŋg-ri endau nik
man d 1sg dat hit-past house inside
‘the man hit me inside the house’
e. endau nik yak mo namot an ruŋg-ri
house inside 1sg dat man d hit-past
‘the man hit me inside the house’
f. ✶namot an yak mo endau nik ruŋg-ri
man d 1sg dat house inside hit-past

This permutability of constituents, however, is not the whole picture of Watam clausal
constituency. With the addition of the prefix ŋga- with various allomorphs, oblique con-
 12 Applicatives in Papuan languages 383

stituents can immediately precede the verb and be stripped of their case marking post-
position (72a), although semantically fuller postpositions remain (72b):

(72) Watam
a. was nakan padoŋ an nimoŋ ŋga-mo panai-ri
wind big tree d night ŋga-do bend-past
‘a big wind bent the tree at night’
b. namot an yak mo endau nik ŋga-ruŋg-ri
man d 1sg dat house inside ŋga-hit-past
‘the man hit me in the house’

The placement possibilities for oblique constituents with ŋga- marked verbs are exactly
the opposite of those without ŋga-. They can only immediately precede the verb; any
other order is ungrammatical:

(73) Watam
a. ✶endau nik namot an yak mo ŋga-ruŋg-ri
house inside man d 1sg dat ŋga-hit-past
‘the man hit me in the house’
b. ✶namot an yak mo ŋga-ruŋg-ri endau nik
house d 1sg dat ŋga-hit-past house inside
‘the man hit me in the house’

These data suggest that ŋga- is a type of applicative promoting an oblique constituent
to core status, although the postposition remaining in examples like (72b) is problem-
atic. While the origins of ŋga- were applicative in function, as indicated by Aruamu/
Mikarew examples in (78)–(80) below, problematic for a such an analysis in Watam
synchronically is the fact that core arguments also co-occur with ŋga- marked verbs and
are subject to the same absorption of case postposition and restriction on movement as
oblique arguments:

(74) Watam
a. namot an yak ŋga-ruŋg-ri
man d 1sg ŋga-hit-past
‘the man hit me’
b. ✶yak namot an ŋga-ruŋg-ri
1sg man d ŋga-hit-past
‘the man hit me’ (compare [71e])

The verb root ruŋg- ‘hit’ is transitive with two core arguments, a hitter and a hittee, so
in no sense can ŋga- be claimed to be functioning as an applicative in (74a) because no
argument is being added to this verb’s argument array and there is no promotion of
384 William A. Foley

an oblique argument to core status. Nor does it function adverbially without adding
arguments, as with the Yimas applicatives in (41); no such meaning is added in (74a) as
opposed to (71d). Also, adverbs can co-occur with ŋga- marked verbs; adverbs are not
arguments at all, so, again, there is no promotion of an oblique argument:

(75) Watam
a. ma ama ŋga-saŋg-ri
3sg again ŋga-go-past
‘he went again’
b. namot an yaoŋ ga-bop-ri
man d good ŋga-speak-past
‘the man spoke well’

So, while ŋga- originated as an applicative morpheme and still often functions like one,
in other cases like (74) and (75) it clearly does not. So, what is its function? It is a marker
of information structure, indicating the constituent immediately preceding the ŋga-
marked verb is focused. Consider these mini-dialogs:

(76) Watam
a. Q: u sumba?
2sg to_where
‘where are you going?’
A: a: yak maŋar ŋga-saŋ-ta
1sg beach ŋga-go-pres
b: ✶
yak maŋar saŋ-ta
1sg beach go-pres
‘I’m going to the beach’
b. Q: tai mo namot an rugu-r minik-ri?
who dat man d hit-r die-past
‘who did the man kill?’
A: a: namot an Matit ŋga-rugu-r minik-ri
man d pn ŋga-hit-r die-past
‘the man killed Matit’
b: ✶
Matit mo namot an ŋga-rugu-r minik-ri
pn dat man d ŋga-hit-r die-past

Only the (a) answers in (76) are acceptable responses to the posed questions. The ques-
tions set up the place I am going to or the person killed by the man as focused infor-
mation, highlighted as being sought by the questioner. When supplied in the answer, it
must be in the focused position selected by a ŋga- marked verb. The following contras-
tive examples further demonstrate this usage:
 12 Applicatives in Papuan languages 385

(77) Watam
a. ma ɲjinak an ore-r aŋgi-ri
3sg knife d find-r take-past
(looking for a knife) ‘he found the knife’
b. ma ɲjinak ŋg-ore-r aŋgi-ri
3sg knife ŋga-find-r take-past
(looking for something else) ‘he found a knife’

In (77a) ɲjinak ‘knife’ is already activated in discourse; what is under discussion is the
search for a particular knife. Consequently when found, the event is reported without
the use of ŋga. In (77b) the topic under discussion is the search for some object; ɲjinak
‘knife’ is unactivated. When it is reported that a knife is found, ɲjinak ‘knife’ is now
focused information, and as such must be presented in the focused immediately pre-
verbal position in combination with a ŋga- marked verb. Clearly, ŋga- now mainly
functions as a discourse based information structure marker for focused constituents.
Whatever applicative function it now has is clearly secondary, though that is its ulti-
mate origin, as the applicative function of its cognate prefix g- is more transparent in
another Lower Ramu language, Aruamu/Mikarew (g- has an allomorph, ba- with verb
roots with a following /o/ which then deletes if immediately following the /a/ of the
prefix) (Relyea 1992a, 1992b):

(78) Aruamu/Mikarew
a. me ikerar-i
3pl play-pres
‘they (pl) are playing’
b. me soka g-ikerar-i
3pl soccer appl-play-pres
‘they (pl) are playing soccer’

(79) Aruamu/Mikarew
a. Araka bogonaro-n ot-e
pn yesterday-obl give_birth-past
‘Araka gave birth yesterday’
b. Araka bogonaro-n guivi-m ba-(o)t-e
pn yesterday-obl daughter-sg appl-give_birth-past
‘Araka gave birth to a daughter yesterday’

(80) Aruamu/Mikarew
ko ofos-n John ba-to
1sg office-obl pn appl-arrive.past
‘I met John at the office’ (lit. ‘arrived with/at John’)
386 William A. Foley

7 Conclusion
In spite of the enormous genetic and typological diversity of Papuan languages, it is
possible to make a few generalizations about their patterns of applicativization. The
following broad claims hold true:

Morphology
– Applicative morphemes in Papuan languages usually arise historically from
re-analysis and grammaticalization of incorporated verbs. This seems invariably
true of those languages with an OV left branching typology. In languages with a VO
right branching typology, applicative morphemes developing from incorporated
prepositions is attested.
– The degree of lexicalization of applicative morphemes in Papuan languages is
unknown, but in the languages in which applicative morphemes have been most
studied it seems negligible.

Syntax
– Papuan languages do not normally allow verbs to have more than three core argu-
ments, so applicative constructions are typically restricted to intransitive and tran-
sitive verb roots.
– Applicative morphemes in Papuan languages are valency increasing; exceptions
are when they function as adverbial modifiers, a not surprising double function
given their origin in verbs.
– Most commonly, the applied participant assumes the grammatical relation of direct
object and usurps the bound pronominal of the original direct object when an
applicative construction occurs with a transitive verb. However, some languages
(Barupu, Coastal Marind, Mian, Mountain Arapesh, Yimas) are examples of the
crosslinguistic rarity of triple agreement languages and allow the original direct
object to preserve its pronominal position and either double up on bound pronom-
inal direct object marking or express the applied participant in a dative marked
pronominal.
– Applicative constructions are commonly obligatory for benefactive participants,
there being no basic construction with the beneficiary as an oblique constituent.
This constraint is less binding for other types of semantic roles, though some lan-
guages like Barupu do insist on applicative constructions in all such cases.
– Where available, applicative affixes freely combine with causative affixes, as long
as their combination does not exceed a derived ditransitive verb; hence such dual
derivation is restricted to intransitive verb roots. Papuan languages normally lack
voice oppositions like passives, so applicative morphemes cannot interact with
them.
 12 Applicatives in Papuan languages 387

Semantics
Applicative morphemes in Papuan languages range from a single morpheme seman-
tically unspecified (Amele, Mountain Arapesh) to a large inventory of them with rich
semantic specifications (Barupu, Yimas).

Lookalikes and others

Non-applicative functions of applicative markers include their use as transitivity
markers, their use as pronominals, and as a focalization device for preverbal constitu-
ents.

Abbreviations
A actor
abs absolutive
accomp accompaniment
adv adverbial
advs adversative
all allative
an animate
appl applicative
aug augment
ben benefactive
caus causative
com comitative
cop copula
d determiner
dat dative
decl declarative
dem demonstrative
dist distal
dl dual
dur durative
erg ergative
excl exclusive
ext extended aspect
f feminine
fr.past far past
fut future
hab habitual
imp imperative
im.past immediate past
incl inclusive
ipfv imperfective
ind indicative
388 William A. Foley

instr instrumental
inv inverse
irr irrealis
iter iterative
kin kinetic
loc locative
m masculine
mid middle
n neuter
nafut non-asserted future
ndl non-dual
nom nominative
npast non-past
nr.dist near distal
nr.past near past
nsg non-singular
nvo neutral verb orientation
obj object
obl oblique
ovo object verb orientation
past past
pat patient
pc paucal
pfv perfective
pl plural
pn proper name
poss possessive
prep preposition
pres present
prog progressive
proh prohibitive
q question
r realis
rec recipient
refl reflexive
rem.past remote past
sg singular
sep separative
seq sequence
subj subject
sur surrounding
tr transitive
U undergoer
val valence
vis visual
1, 2, 3 grammatical persons
I, II, . . . noun classes
 12 Applicatives in Papuan languages 389

References
Aannestad, Aidan, Carl Campbell & Jody Campbell. 2020. Towards a grammar of the Yale language: Taking
another look at archived field data.
https://pnglanguages.sil.org/resources/archives/85952 (accessed 4 September 2021).
Bromley, Myron H. 1981. A grammar of Lower Grand Valley Dani. Canberra: Pacific Linguistics.
Bruce, Leslie P. 1984. The Alamblak language of Papua New Guinea (East Sepik). Canberra: Pacific Linguistics.
Carroll, Matthew J. 2017. The Ngkolmpu language, with special reference to distributed exponence. Canberra:
Australian National University dissertation.
Conrad, Robert J. & Kepas Wogiga. 1991. An outline of Bukiyip grammar. Canberra: Pacific Linguistics.
Corris, Miriam. 2005. A grammar of Barupu, a language of Papua New Guinea. Sydney: University of Sydney
dissertation.
Döhler, Christian. 2018. A grammar of Komnzo. Berlin: Language Sciences Press.
Donohue, Mark. 2004. A grammar of the Skou language of New Guinea. http://www.papuaweb.org/dlib/
tema/bahasa/skou/Skoufull1.pdf (accessed 4 September 2021).
Drabbe, Petrus. 1949. Bijzonderheden uit de Talen van Frederik-Hendrik-Eiland: Kimaghama, Ndom en
Riantana. Bijdragen tot de Taal-, Land- en Volkenkunde / Journal of the Humanities and Social Sciences of
Southeast Asia 105(1). 1–24.
Dryer, Matthew. 1986. Primary objects, secondary objects and antidative. Language 62(4). 808–845.
Evan, Nicholas. 2015. Valence in Nen. In Andrej Malchukov & Bernard Comrie (eds.), Valency classes in the
world’s languages, Volume 2: Case studies from Austronesia and the Pacific, the Americas, and theoretical
outlook, 1069–1116. Berlin: De Gruyter Mouton.
Fedden, Sebastian. 2011. A grammar of Mian. Berlin: De Gruyter Mouton.
Foley, William A. 1991. The Yimas language of New Guinea. Stanford: Stanford University Press.
Foley, William A. 1997. Polysynthesis and complex verb formation: the case of applicatives in Yimas. In Alex
Alsina, Joan Bresnan & Peter Sells (eds.), Complex predicates, 355–395. Stanford: Center for the Study
of Language and Information.
Foley, William A. 2017. The polysynthetic profile of Yimas: a language of New Guinea. In Michael Fortescue,
Marianne Mithun & Nicholas Evans (eds.), The Oxford handbook of polysynthesis, 808–829. Oxford:
Oxford University Press.
Gravelle, Gloria J. 2011. A grammar of Moskona: An East Bird’s Head language of West Papua. Amsterdam: Vrije
Universiteit dissertation.
Haiman, John. 1980. Hua: A Papuan language of the Eastern Highlands of New Guinea. Amsterdam: John
Benjamins.
Healey, Phyllis M. 1965. Telefol clause structure. In Phyllis M. Healey (ed.), Papers in New Guinea Linguistics 3,
1–26. Canberra: Pacific Linguistics.
Klamer, Marian. 2010. A grammar of Teiwa. Berlin: De Gruyter Mouton.
Kratochvíl, František. 2007. A grammar of Abui: A Papuan language of Alor. Utrecht: LOT Publications.
Kratochvíl, František. 2011. Transitivity in Abui. Studies in Language 35(3). 588–635.
Locke, Arnold (Arjen) Hugo. 2011. Abau grammar sketch. Data Papers in Papua New Guinea Languages 57.
https://www.sil.org/system/files/reapdata/33/13/86/33138664671583783026160672477079448010/
DPv57_Abau_Grammar_final_%28pdfA%29.pdf (accessed 4 September 2021).
MacDonald, Lorna. 1990. A grammar of Tauya. Berlin: De Gruyter Mouton.
McElhanon, Kenneth A. 1970. Selepet grammar, Part I: From root to phrase. Canberra: Pacific Linguistics.
Nichols, Johanna. 1986. Head-marking and dependent-marking grammar. Language 62(1). 56–119.
Olsson, Bruno. 2021. A grammar of Coastal Marind. Berlin: De Gruyter Mouton.
Palmer, Bill (ed.). 2018. The languages and linguistics of the New Guinea area. Berlin: De Gruyter.
Relyea, Marsha. 1992a. A comparative grammar of Aruamu and Kire. Unpublished manuscript, Pioneer Bible
Translators, Madang, Papua New Guinea.
390 William A. Foley

Relyea, Marsha. 1992b. “The Story of the Death of my Friend”: Coherence and cohesion in Aruamu narrative.
Unpublished manuscript, Pioneer Bible Translators, Madang, Papua New Guinea.
Roberts, John R. 1987. Amele. London: Croom Helm.
Robinson, Laura C. & Johnson Welem Haan. 2014. Adang. In Antoinette Schapper (ed.), The Papuan
languages of Timor, Alor and Pantar: Volume 1, 221–284. Berlin: De Gruyter Mouton.
Schapper, Antoinette. 2014. Kamang. In Antoinette Schapper (ed.), The Papuan languages of Timor, Alor and
Pantar: Volume 1, 285–350. Berlin: De Gruyter Mouton.
Scott, Graham. 1978. The Fore language of Papua New Guinea. Canberra: Pacific Linguistics.
Seiler, Walter. 1985. Imonda, a Papuan language. Canberra: Pacific Linguistics.
Usher, Timothy & Edgar Suter. 2020. The Asmat-Muli languages of southwestern New Guinea. https://uploads.
documents.cimpress.io/v1/uploads/4a477a55-a673-44fe-b1a5-fade5893fe71~110/original?tenant=v-
bu-digital (accessed 4 September 2021).
Vincent, Alexander. 1973. Tairora verb structure. In Howard McKaughan (ed.), The languages of the Eastern
family of the East New Guinea Highlands stock. Seattle: University of Washington Press.
Voorhoeve, Clemens L. 1965. The Flamingo Bay dialect of the Asmat language. The Hague: Martinus Nijhoff.
Wilson, Jennifer. 2017. A grammar of Yeri, a Torricelli language of Papua New Guinea. Buffalo: State University
of New York dissertation.
Wood, Joyce Kathleen. 2012. Valence-increasing strategies in Urim syntax. Dallas: Graduate Institute of Applied
Linguistics / Arlington: University of Texas MA thesis.
Peter K. Austin
13 Applicative constructions in Australian
Aboriginal languages
Abstract: Applicative constructions are found in a number of the Indigenous languages
spoken in Australia, though they do not exist in the majority of languages for which
reliable morpho-syntactic data is available.1 Several typological patterns can be found,
suggesting a basic division between comitative-locative constructions, and benefactive-
malefactive constructions. These are encoded morphologically by prefixation in some
languages, and by suffixation in others. In a large number of languages with applica-
tives, there is a relationship with causative constructions; in some a single affix serves as
both applicative and causative depending on the semantic nature of the verb to which
it is affixed. In some languages, applicatives only occur with intransitive base verbs. For
languages where applicatives occur to transitive base verbs there are two types: (a) in
those with ditransitive base verbs, applicative constructions show the applied element as
a direct argument of the applied verb; (b) in those without ditransitive base verbs, anti-
passive constructions must be applied to the base verb before applicatives can be added,
with an argument of the base transitive verb appearing in a non-argument role in the
resulting applicative. Applicative lookalike constructions are found in a few languages.

1 Introduction
At the time of colonisation (beginning in the 18th century),2 Australia was occupied by
about 600 groups of Indigenous peoples;3 each group had their own territory, laws,
socio-cultural characteristics, and distinctive ways of speaking. Linguistically, these can
be grouped into up to 500 separate languages (Bowern 2022, Horton 1996).4

1 For an earlier survey see Austin (1997), which was written before most of the detailed research on
northern Australian languages was completed and published. This paper supersedes the analysis pre-
sented there.
2 Colonisation began in 1788 in Sydney, New South Wales, and gradually extended throughout the
south-east and south-west of the continent, plus Tasmania. Central Australia and the north-east of the
Northern Territory were settled in the 20th century.
3 Indigenous Australians are officially categorised as Aboriginal and Torres Strait Islanders (ATSI).
4 See https://aiatsis.gov.au/explore/map-indigenous-australia for an interactive version of the Horton
map (accessed 2023-11-20).

Acknowledgments: For valuable feedback and comments on an earlier draft I am grateful to Denis Creissels
and an anonymous reviewer.

https://doi.org/10.1515/9783110730951-013
392 Peter K. Austin

ATSI languages belong to four distinct groupings:

1. Tasmanian—the languages spoken in Tasmania are very poorly known, mostly
from badly-transcribed wordlists. Almost nothing about the morpho-syntax of the
languages can be gleaned from the existing sources (Crowley and Dixon 1981), so
they will be excluded from this survey;
2. Eastern Torres Strait—the Meriam Mer language spoken on the eastern islands in the
Torres Strait (between Australia and Papua New Guinea) is a Papuan language related
to those of the Fly River delta (see Piper 1989). It will be excluded from this survey;
3. Pama-Nyungan—languages spoken across the southern two-thirds of the continent,
from Western Torres Strait to the south-west of Western Australia, and including
the Yolngu group of north-east Arnhemland, form a single genetic family (named
for the word for ‘person’ in the far north-east, pama, and far south-west, nyungar).
Low-level subgroupings of Pama-Nyungan are well established but the detailed
higher-level classification of the family is ongoing and not yet finalised (see Bouck-
aert, Bowern, and Atkinson 2018; Miceli 2015); and
4. non-Pama-Nyungan—this is a cover term for around 25 other genetic families spoken
in the far north (‘Top End’) of Australia through the Kimberley, Daly and Arnhem-
land regions of Western Australia and the Northern Territory. It is unclear whether
there are deeper relationships between these families and/or with Pama-Nyungan
(Evans 2003a).

Groups 3 and 4 will be henceforth referred to as Australian Aboriginal Languages

(AALs) with data from them informing the typological, areal, and comparative survey
in this paper. Map 1 shows the genetic classification picture as it is currently known.
Traditionally, Australian Indigenous people lived up in highly multi-lingual envi-
ronments, with exogamous marriage systems promoting the use and learning of several
AALs by individual community members (Rumsey 2018). A consequence of this is the
existence of areal phenomena in phonology, morphology, and syntax that cut across the
genetic groupings outlined above, such as switch-reference (Austin 1981b), pronominal
agreement (Blake 1977), and nominative-accusative morpho-syntax (Dench 1982). We
will show that applicatives are also areally distributed.
Currently, only about 10 languages remain in daily use with vibrant speaker com-
munities and being learned by children as a home language. Of the remainder, around
150 are endangered to varying degrees, with small numbers of older speakers and not
being passed on to children. The remaining languages are categorised as sleeping and
have no-one who learned them as children, though some partial knowledge may remain.
A large number of revitalisation projects are under way, especially in south-eastern
Australia, some of which are recognised and supported by the State and Federal govern-
ments and employed in various ways in education (see Lee and Obata 2010; Marmion,
Obata and Troy 2014)
 13 Applicative constructions in Australian Aboriginal languages 393

Map 1: Pama-Nyungan (yellow) and non-Pama-Nyungan groups. (from Wikipedia)

2 Data sources
The available information about AALs varies widely, from detailed grammatical, lexical
and textual analyses for a small number of languages, to sketch studies, mostly covering
lexical and basic morphological material, for many others5. For very many languages
spoken in the areas of first colonisation (New South Wales, southern Queensland, south-
ern South Australia, and Victoria) very little is known about their morpho-syntax, with
the available descriptions often coming from semi-speakers or rememberers recorded
in the 1970s or 1980s. Researchers at that time also tended to pay more attention to
phenomena such as case-marking and cross-clausal coreference, and ignored or were
unaware of constructions impacting on argument coding such as causatives and applic-
atives. This means that a typological survey such as this one is necessarily limited by the
available descriptions and must be far from exhaustive or conclusive.

5 For sources see https://collection.aiatsis.gov.au/austlang/search (accessed 2022-10-31)

394 Peter K. Austin

3 Morpho-syntax
Most AALs make a fundamental distinction in their lexical categorisation between roots
which are nominal (covering what would be nouns and adjectives in other more famil-
iar languages) and those which are verbal (with a frequent distinction, particularly in
Pama-Nyungan, between independent and dependent verb word forms). Most languages
show fairly transparent agglutinative morphology (though fusional paradigms are often
found for pronouns, and some verbs), with case typically encoded on nominals and
tense-aspect-mood encoded on verbals. PN morphology is typically dependent-marking
or double-marking (in the sense of Nichols 1986), while NPN languages are generally
head-marking. Pama-Nyungan languages are entirely suffixing in their morphology,
while non-Pama-Nyungan languages show both suffixing and prefixing, with prefixes
on nominals often marking gender categorisation, and prefixes on verbals marking
pronominal arguments (and hence being head-marking). Some non-Pama-Nyungan
languages allow nominal incorporation of arguments into verbs (with or without the
form of the incorporated nominal being different from its free-standing citation form)
and are thus sometimes referred to as “polysynthetic”.
AALs typically make a strict distinction in verbal roots between those that are intran-
sitive and take a single argument, and those that are transitive and take two arguments
(Dixon 1980: 378; Blake 1987: 12). It is generally a simple matter to determine the tran-
sitivity of any lexical verbal. Typologists refer to the single argument of an intransitive
verbal as S, the agent-like argument of a transitive verbal as A, and the non-agent second
argument as P.
In most AALs that express nominal case, the morphological encoding operates on
a split-ergative basis, with some nominals distinguishing nominative-accusative (S/A
versus P) from ergative-absolutive (A versus S/P) and/or three-way (each of S, A, and P
having separate forms), typically determined by the category (pronoun versus noun) and
inherent lexical content of the inflected nominal (at least some pronouns being nomina-
tive-accusative, and some human or animate nominals being three-way or ergative-ab-
solutive). Syntactically, cross-clausal coreference in both co-ordination and dependent
clause constructions typically operates on an entirely nominative-accusative basis, with
the S/A in one clause necessarily being understood as coreferential with the S/A in the
linked clause (i.e. the cross-clausal pivot is S/A). In a very small number of syntactically
ergative languages (Dyirbal, Yidiny, Kalkatungu, Bandjalang) cross-clausal coreference
works in terms of an ergative-absolutive pivot (with sharing of S/P between the clauses).

4 Non-Pama-Nyungan applicatives
Applicative constructions are found in some non-Pama-Nyungan languages, but are
generally absent from most of the available descriptions:
 13 Applicative constructions in Australian Aboriginal languages 395

(i) languages without applicatives: Nakkara (Eather 1990), Mawng (Singer 2016: 28),
Jaminjung-Ngaliwurru (Schultze-Bernd 2000: 84),6 Wardaman (Merlan 1994: 206),
Alawa (Sharpe 1972), and Gaagudju (Harvey 2003);
(ii) languages with applicatives: Gunwinyguan (§ 4.1), Bardi (§ 4.2), Daly (§ 4.3), and
Wambaya (§ 4.4).

4.1 Gunwinyguan

The Gunwinyguan family is spoken in central Arnhemland and comprises 14 languages

(some with local dialect variation). Map 2 (from Horrack 2018) shows their approximate
locations.

Map 2: The Gunwinyguan family and neighbouring languages.

This family has the best descriptions of applicative constructions in Australia, with
available details on Bininj Gun-wok (Evans 2003b), Dalabon (Ponsonnet 2021), Wubuy

6 But note that “verbs of different valency, combined with the same coverb, often fulfil the same function
as applicative markers, causativisers, and other valency-changing morphology in other languages” (ibid.).
396 Peter K. Austin

(Horrack 2018), Rembarnga (McKay 2011, Saulwick 2003), Ngalakgan (Merlan 1983; she
calls the language Ngalakan), Ngandi (Heath 1978), and Mangarayi (Merlan 1982).
Gunwinyguan shows two types of applicatives, both marked by prefixes to the base
verb, which may be either intransitive or transitive:
(i) a benefactive-malefactive marked by marne- in Bininj Gunwok (Evans 2003b:
427–432), marnu- in Dalabon (Ponsonnet 2021: 122) and by pak- in Rembarnga
(McKay 2011: 261–282; Saulwick 2003: 208–226), with cognates in Ngalakgan
(Merlan 1983: 47) and Ngandi (Heath 1978: 81);
(ii) a comitative-locative marked by yi- in Bininj Gun-wok (Evans 2003b: 432–437),
and ye- in Dalabon (Ponsonnet 2021: 129). There are two different comitative pre-
fixes yi- and re- in Rembarnga (Saulwick 2003: 227–236; McKay 2011: 151–154).

Ponsonnet (2021: 128) explains that in Dalabon the benefactive-malefactive marnu-

increases the base verb arguments by one, adding an animate P (with a base transitive
P surfacing as a second object, not indexed on the verb):

Vintr S
Vintr+ben.appl A P [animate]

Vtr A P
Vtr+ben.appl A P [animate] P2

The applied P is understood as the recipient of benefactive transfer (‘give to’) as in (1), or
the source of malefactive transfer (‘remove from’) as in (2). However, depending on the
base verb, the applied argument “can also be an emotional stimuli (3), an addressee (4),
a goal or location (5), or a possessor (6)” (Ponsonnet 2021: 129). Its person and number is
encoded in the verb agreement prefix which references A and P in portmanteau.

(1) byunrul yila-h-marnu-yidjnja-ninj-wurd

funeral 1pl.excl>3-real-appl-have-pi-dim
‘We had a small funeral for her.’ (Ponsonnet 2021: ex 6)

(2) ngorr bula-h-marnu-ngu-yan

1pl.incl 3pl>1-real-appl-eat-fut
‘They will eat (it) on us.’ (Ponsonnet 2021: ex 7)

(3) kardu bula-h-marnu-djong-m-inj

maybe 3pl>3sg-real-appl-fear-inch-pp
‘Maybe they were afraid of them.’ (Ponsonnet 2021: ex 8)
 13 Applicative constructions in Australian Aboriginal languages 397

(4) bulu ka-h-marnu-yenjyenjdju-ng

3pl 3sg>3-real-appl-talk.rdp-prs
‘He talks to them.’ (Ponsonnet 2021: ex 4)

(5) buka-h-marnu-bo-ninj darnkih

3sg>3sg-real-appl-go-pi close
‘He was coming close to him.’ (Ponsonnet 2021: ex 9)

(6) dja-h-marnu-labbarl-n-iyan
1sg>2sg-real-appl-pond-see-fut
‘I will see your pond.’ (Ponsonnet 2021: ex 10)

For Wubuy, Horrack (2018: 7, 123) prefers to call the corresponding aG- ~ waaG- prefix an
“affectee applicative” because of the wide range of semantics associated with it, namely
“beneficiary, adverse beneficiary (i.e., who is hurt rather than helped), source (as in ‘take
from’, usually associated with adverse beneficiary), addressee (‘to shout at’), owner or
possessor (of implicit direct object), and object of emotion (‘be afraid of’, ‘be suspicious
of’, etc.)” (Horrack 2018: 135, quoting Heath 1984: 380). These more-or-less overlap with
the semantics of the applied argument for Dalabon marnu-. Unlike Dalabon, however,
the introduced argument (which triggers P verb agreement) bears an overt case marker,
typically allative-dative -wuy ~ -guy but also purposive -yungguyung, ablative -wala, or
genitive -yinyung. Consider these examples:

(7) ngarra-mani-nyung nguna-a-jaalibu-mana na-doctor-wuy

f-woman-hum.sg 3f.sg>3m.sg-appl-cough-prs m-doctor-dat
‘The woman coughs for the doctor.’ (Horrack 2018: ex 1-8)

(8) anubani ngayawi-wuy ngambi-i-gawanggii-na

that.n.anaph 1sg.pro-all 3pl>1sg-appl-listen-prs
‘They listen to me.’ (Horrack 2018: ex 4-28)

(9) nguna-a-gamaji na-walyi-nyung-gala

3f.sg>3m.sg-appl-thieve.pc m-man-hum.sg-abl
‘She was thieving from the man (3m.sg)’ (Horrack 2018: ex 4-31)

(10) ngarrani-i-ngawii-yn ngagurra-yungguyung

3m.sg>1pl-appl-die-pp 1incl.pl.obl-purp
‘He died for us.’ (Horrack 2018: ex 4-36)

Note that the applied P “with transitive and ditransitive inputs appears to have an inter-
pretation of alienable external possession where the affectedness of the introduced
argument is being emphasised” (Horrack 2018: 143).
398 Peter K. Austin

The comitative-locative applicative in Gunwinyguan introduces an animate or

inanimate P expressing a person or thing with which an action is carried out (including
movement). Ponsonnet (2021: 130) gives the following Dalabon examples, showing both
intransitive and transitive base verbs:

(11) Yila-h-ye-dudj-mu wulk-no

1pl.excl>3-real-appl-return-prs fat-fill
‘We bring back some fat.’ (Ponsonnet 2021: 130 ex 11)

(12) Nunda njel ka-h-ye-ba-ng

this 1pl.excl 3sg>1-real-appl-bite-prs
‘This is to bite us with.’ (Ponsonnet 2021: 130 ex 12)

For Wubuy, Horrack (2018: 216) shows that the aynji- comitative prefix requires contact
between the S or A of the base verb and the introduced applied argument, as in:

(13) ngi-iynji-ngama-a ngarra-ngarrugalij warra-mijburraayung

f-appl-swim-pc f.top-dugong pl-children
‘The dugong was swimming with the children.’
(i.e. the children were sitting on the dugong’s back) (Horrack 2018: ex 6-17)

Horrack (2018: 216) notes further that “instead of the subject [of the transitive] being
interpreted as carrying the comitative referent, the comitative referent is realised as
primary object and is itself interpreted as carrying the argument that would have been
the primary object of the underived verb”, as in (14):

(14) wunguna-aynji-yarrbu-mana ngarra-manum-baa warra-gujuju

3f.du>3m.sg-comit-wash-prs f-woman-du pl-baby
na-walyi-nyung
m-man-hum.sg
‘The two women wash the babies with the man.’
(i.e. the man is holding the babies and the two women wash the babies)
(Horrack 2018: ex 6-24)

4.2 Bardi

Bardi is spoken in the Kimberley region of Western Australia and has two applicatives
(both encoded by the verb suffix -na ~ -ng), described by Bowern (2012: 488–495). One
applicative can be added to a restricted set of intransitive verbs to create transitive
verbs with idiosyncratic semantic relationships to the base verb; Bowern (2012: 190)
 13 Applicative constructions in Australian Aboriginal languages 399

cites just five instances: ‘go’ ‘touch’, ‘visit’ ‘help’, ‘come’ ‘come out’, ‘collect’
‘collect’, ‘say’ ‘leave’.
The second Bardi applicative is more productive and may appear with either intran-
sitive or transitive roots, and promotes an adjunct goal, accompaniment, or instrument
to be a direct P argument of the applied verb. Examples with an intransitive base verb
are, firstly a goal:

(15) Wirr i-ny-jarrmi-ni-ng Iila-nim

rise 3-pst-rise-rem.pst-appl2 dog-erg
‘The dog jumped on him.’ (Bowern 2012: ex 12.66)

and, secondly, an accompaniment:7

(16) Ngayoo-nim bard roowil nga-n-nya-na-ng birrii

1min-erg off walk 1-trz-catch-rem.pst-appl2 mother.abs
‘I walked with my mother.’ (Bowern 2012: ex 12.67b)

Transitive examples of applied goals and accompaniments are:8

(17) I-na-ng-ga-na-nga=moord.
3-trz-pst-bring-rem.pst-appl2=1aug.dobj
‘He brought us it.’ (Bowern 2012: ex 12.65b)

(18) I-n-jala-ng=jarrngay.
3-trz-see-appl2=1min.dobj
‘He saw me with [it].’ / ‘He saw it with [me].’ (Bowern 2012: ex 12.73)

The instrumental applicative appears only to occur with transitive base verbs and to
result in a ditransive with two P-marked direct arguments:

(19) I-na-m-boo-ng-gal ginyinggin irrol.

3-trz-pst-hit-appl-rec.pst 3min-excl spear.abs
‘He hit me with that very spear.’ (Bowern 2012: ex 12.72)

7 Some Pama-Nyungan languages have a number contrast between MINimal and unit AUGmented
where first person dual inclusive (1/2 person) is treated as minimal (even though numerically it is not
singular). McKay (1979) originated this analysis for Australian languages.
8 Example (18) is ambiguous as to whether the first-person minimal agreement is with the base P, or
with the applied (unexpressed third person) P.
400 Peter K. Austin

Alternatively, the applied instrument can be marked with instrumental case and thus
appear to be an adjunct, not a P, even though the verb carries an applicative suffix, as
in (20):

(20) Baal-nga i-ng-orr-ondi-ni-ng=irr.

bark-ins 3-pst-aug-trz-cover-rem.pst-appl=3A.dobj
‘They covered them with bark.’ (Bowern 2012: ex 12.75)

This is thus an instance of an X-applicative construction.

4.3 Daly languages

The Murrinhpatha language spoken in the Daly region of the Northern Territory (see
Map 2, west of Gunwinyguan) has an applicative construction coded by the verb prefix
-ma-, that is highly unusual typologically in that it encodes a source or malefactive
semantics (and never a benefactive, or the cross-linguistically more usual locative/com-
itative or instrument). Nordlinger (2019) gives the following examples of source seman-
tics with an intransitive base verb:

(21) ngem-nhi-ma-nham
1sg.sbj.poke:rr(22).nfut-2sg.obj-appl-fear
‘I’m afraid of you.’ (Nordlinger 2019: ex 42c)

Nordlinger (2019: 416) notes that “the majority of applicative examples in the corpus
involve base transitive verbs which entail transfer of possession or location of a theme
argument, with the animate source represented as the applicative object”; examples are:

(22) nganam-nhi-ma-kut
1sg.sbj.be(4):nfut-2sg.obj-appl-collect
‘I collected (the money) from you.’ (Nordlinger 2019: ex 2)

(23) bim-pun-ma-yepup
1sg.sbj.hear(16):nfut-3pl.obj-appl-listen
‘I heard (the story) from them.’ (Nordlinger 2019: ex 41d)

Note that the applied verb can take a reflexive-reciprocal derivation (marked by prefix
-nu) which relates to the applied P, as in (cf. (21) above):

(24) them-nu-ma-nham
1incl.sbj.poke:nfut-rr-appl-fear
‘We’re (incl) frightened of each other.’ (Nordlinger 2019: ex 45)
 13 Applicative constructions in Australian Aboriginal languages 401

Other Daly languages have applicatives, but with different semantics. According to Reid
(1990, 2000) Ngan’gityemerri has a comitative applicative marked by the prefix -mi- that
promotes an adjunct to a P role, as in (25):

(25) gaganiny-nyi-mi-wap
1sg.sbj.go.pfv-2sg.obj-appl-sit
‘I sat down with you.’ (quoted in Nordlinger 2019: ex 54b)

Marrithiyel (Green 1989) has an applicative marked by the prefix -mu- that is similar to
Murrinhpatha in expressing an applied P with source semantics:

(26) wiyan ngurr-inj-mu-duk-wa

tobacco 1sg.sbj.irr.rr-2sg.obj-appl-remove-fut
‘I’ll take your tobacco off you.’ (Green 1989: ex 5-173b)

According to Nordlinger (2019), the Murrinhpatha prefix has arisen from grammatici-
sation of ma meaning ‘hand’, while Ngan’gityemerri -mi- derives diachronically from
‘eye’, and Marrithiyel -mu- comes from ‘hand’. These are the only instances of applica-
tives apparently arising from body parts in Australia.

4.4 Wambaya

Spoken in the Barkly region of the Northern Territory, this language has an applicative
suffix -(ba)bu which, according to Nordlinger (1998: 169), occurs with intransitive base
verbs and promotes a NP of accompaniment to object, deriving a transitive verb. This suffix
expresses an anti-benefactive sense, usually translated into English with ‘away’, as in:

(27) Dingbari-j-babu ngiy-a gayangga wardangarringa-ni.

fly.off-th-appl 3sg.nm.A-pst high.moon.husband-loc
‘The moon flew off with (the sun’s baby) up (into the sky).’ (Nordlinger 1998:
ex 6-50)

(28) Mawula-j-babu ngiy-a ganjimi.

play-th-appl 3sg.nm.A-pst finish.nfut
‘She played all her money away.’ (Nordlinger 1998: ex 6-51)

(29) Ngarri balamurru gin-a bard-babu.

1sg.poss.I.acc spear.IV.acc 3m.sg.A-pst run-appl
‘He ran off with my spear.’ (Nordlinger 1998: ex 6-52)
402 Peter K. Austin

5 Pama-Nyungan applicatives
A limited number of Pama-Nyungan languages spoken in central, northern, and eastern
Australia have applicative constructions marked by a verb suffix, typically expressing
comitative or locative semantics and typically only attached to a subset of intransi-
tive base verbs. Only a handful have applicatives of transitive base verbs. A couple of
instances of applicative lookalikes are found in this family.

5.1 Karnic

The Karnic group was traditionally spoken from northern South Australia and western
New South Wales to western Queensland and comprises several subgroups (Bowern 2001).
The Central Karnic subgroup comprises Diyari, Ngamini, Yaluyandi, Yandruwandha and
Yawarrawarka. They all show applicatives coded by a suffix (Diyari -lka-, Ngamini -ka-,
Yaluyandi -kalka-). In Diyari (Austin 1981a, 2021) intransitive motion and stance verbs
can be transitivised by the addition of the suffix -lka to introduce a P which is held by or
is located with the A argument (the S of the base verb), as in (30):

(30) Nhulu kanku-yali pirta thika-lka-yi

3nf.sg.erg boy-erg stick.acc return-appl-prs
‘The boy is taking a stick back’ (Austin 2021: ex 368)

When the P is animate then the A is understood as causing and directing the motion or
rest and is simultaneously moving or at rest, as in:

(31) Ngathu tyipi~tyipi thika-lka-rna wara-yi

1sg.erg rdp~sheep.acc return-appl-ptcp aux-prs
‘I drove the sheep back’ (Austin 2021: ex 370)

(32) Nhulu kinthala tharka-lka-yi

3nf.sg.erg dog.acc stand-appl-prs
‘He is standing with (his) dog (holding it)’ (Austin 2021: ex 371)

The verb ngama-lka- ‘sit-appl-’ has become conventionalised as a transitive verb of pos-
session and can be used when there is no physical contact between the A possessor and
the P possessee, as in:

(33) Yundru karna tharla ngama-lka-yi

2sg.erg person name.acc sit-appl-prs
‘Do you have an Aboriginal name?’ (Austin 2021: ex 327)
 13 Applicative constructions in Australian Aboriginal languages 403

(34) Yundru kaku ngama-lka-yi

2sg.erg older.sister.acc sit-appl-prs
‘Do you have an older sister?’ (Austin 2021: ex 328)

With a limited number of base verbs that do not express motion or stance, -lka- derives
a transitive verb with a more affective meaning, that is, the P is understood as affected
by and undergoing the action denoted by the verb, as in:

(35) Thalara-li ngalinha kurda-lka-yi

rain-erg 1du.excl.acc fall-appl-prs
‘The rain is pouring on us’ (Austin 2021: ex 373)

(36) Paya-li nhinha kuna-lka-rna wara-yi

bird-erg 3nf.sg.acc shit-appl-ptcp aux-prs
‘The birds shat on him’ (Austin 2021: Text 1 line 50)

Note that verbs of communication or sensation cannot take the applicative (unlike in
Gunwinyguan, discussed in § 4.1).
Central Karnic also has a D-applicative construction which is found with transitive
base verbs to express action done for the benefit of someone other than the A of the
base verb. This is encoded with the suffix -pa- (in Diyari,9 Ngamini, Yarluyandi) or -na-
(in Yandruwandha, Yawarrawarka) which also occurs as a causativiser of some intran-
sitive verbs. This is not a benefactive P-applicative, however, as the beneficiary is in the
dative case and is not a direct argument of the resulting verb (we gloss it as ALTruistic).
Some Diyari examples are:

(37) Ngathu kupa~kupa nhayi~nhayi-ipa-rna wanthi-yi walpala-ya

1sg.erg rdp~child.acc rdp~see-alt-ptcp aux-prs white.man-dat
‘I looked after the children for the white man.’ (Austin 2021: ex 58)

Note that -pa- can be used with an applicativised base verb, as in:

(38) Minha-nhi ngayani nyarnikuti thika-lka-ipa-rnanthu nhangkarni

what-loc 1pl.excl.erg goat.acc return-appl-alt-implds 3f.sg.dat
‘Why must we bring the goats back for her?’ (Austin 2021: ex 70)

Arabana-Wangkanguru belongs to another subgroup of Karnic and has been described

by Hercus (1990: 149–152) as possessing an affix -la- which reveals a split within intran-

9 In Diyari this affix neutralises the preceding base verb final vowel to i—in Austin (1981a) it is notated
as -ipa-. There is an alternative synonymous affix -iyirpa- found in Diyari only.
404 Peter K. Austin

sitive base verbs: for most it is a causativiser (where the P is inanimate or a non-con-
trolling animate), while for five volitional intransitive verbs it has an applied pattern.

thudni- ‘to cry’ thudni-la- ‘to cry over, mourn’

wiya- ‘to laugh’ wiya-la- ‘to mock, deride, laugh at’
pankipanki- ‘to be pleased’ pankipanki-la- ‘to be pleased with’
yanhi- ‘to talk’ yanhi-la- ‘to tell’
yirji- ‘to move’ yirji-la- ‘to work for (someone)’

An example is:

(39) Arluwa-kari-ri wiya~wiya-la-thira

child-pl-erg rdp~laugh-appl-punc
‘The children laugh at (him).’ (Hercus 1990: ex 351)

As we saw for Central Karnic, the same affix can be added to transitive verbs to indicate
action done for the benefit of someone other than the A; it is a D-applicative and does
not promote the beneficiary to direct argument status. An example is:

(40) Unkunha punga karra-l-ta.

2sg.dat hut.abs tie-appl-prs
‘(He) is fixing your hut for you.’ (Hercus 1990: ex 358)

Notice that, if the base verb has a negative effect reading, the dative argument will be
understood as malefactive, as in:

(41) Anthunha arluwa pirda-la-yirra

1sg.dat child.abs hit-appl-punc
‘They are hitting my child on me.’ (Hercus 1990: ex 359)

The Northern Karnic group comprises several languages, the best known of which is Pit-
ta-Pitta, once spoken in western Queensland, and described by Blake (1979a). Here there
is a -la- suffix which creates causatives of most intransitive verbs but has an applicative
function with just four intransitives in the corpus that have volitional agent-like subjects.

Applied -la- (S = A)
mirnti- ‘to play’ mirnti-la- ‘to play with’
tiwa- ‘to be jealous’ tiwa-la- ‘to be jealous of’
wapa- ‘to look for’ wapa-la- ‘to look for’
wiya- ‘to laugh’ wiya-la- ‘to laugh at’
 13 Applicative constructions in Australian Aboriginal languages 405

This -la- affix can be used with transitive verbs to signal the involvement of a benefac-
tive (or malefactive if the predicate is aversive); the beneficiary NP is advanced to P
and is case-marked like a regular transitive object (case-marking on the base transitive
object is unaffected, resulting in a ditransitive construction). Thus, contrast the follow-
ing pair:

(42) Nhan-pa-ka karnta-ka yanthurru-nha marri-linga nganyari-nha

3f.sg-nom-here go-pst food-acc get-purp 1sg.dat-acc
‘She went to get food for me.’ (Blake 1979a: ex 60)

(43) Nhan-pa-ka karnta-ka yanthurru-nha marri-la-linga nganya

3f.sg-nom-here go-pst food-acc get-appl-purp 1sg.acc
‘She went to get food for me.’ (Blake 1979a: ex 61)

A malefactive example is:

(44) Thithi-nha nganya pithi-la-ya

older.brother-acc 1sg.acc hit-appl-prs
‘(He) hit my older brother on me.’ (Blake 1979a: ex 64)

5.2 Kalkatungu and Yalarnnga

Kalkatungu, formerly spoken in Western Queensland north of Karnic had two applica-
tive constructions, one encoded by the suffix -(ny)tjama(yi)- and one with the suffix -nti-
(Blake 1979b). The first of these can be added to intransitive verbs or transitive verbs
to indicate that a goal is a P argument of the resulting verb. Thus from rlunga- ‘to cry’
we derive rlunga-nyjama- ‘to cry for’ which can then be detransitivised by the recipro-
cal-reflexive suffix, as in:

(45) Rlunga-nthithi-tjama-ti malthanha

cry-pl-appl-rr many
‘They are all crying for one another.’ (Blake 1979b: ex 5.44)

With transitive roots, -(ny)jama(yi)- indicates a beneficiary that is encoded as a P argu-

ment of the resulting ditransitive verb, as in:

(46) Ngai-tji ngalhu-yu ngai karri-nyjamayi kunti

1sg-dat daughter-erg 1sg.abs clean-appl house.abs
‘My daughter cleaned the house for me.’. (Blake 1979b: ex 5.47b)
406 Peter K. Austin

When the base verb involves a negative effect then the applied P argument is under-
stood in a malefactory sense, as in:

(47) Tjipa-yi ngai nhau-thu nhitha-nytjamayi maa nhungu

this-erg 1sg.abs child-erg steal-appl food.abs hence
‘This child stole the food from here on me.’ (Blake 1979b: ex 5.46)

The second applicative -nti- can be added to intransitive roots of stance or motion to
derive a transitive verb where S corresponds to A and a locative/comitative P is added.

kapani- ‘to go hunting’ kapaninti- ‘to go hunting (something)’

rna- ‘to stand’ rnanti- ‘to stand with/on (something)’
rnu- ‘to lie’ rnunti- ‘to lie with/on (something)’
thuna- ‘to run’ thunanti- ‘to run with (something)’
wani- ‘to play’ waninti- ‘to play with (something), play (a part
in a) corroboree’
wanti- ‘to follow’ wantinti- ‘to follow (something)’
yu- ‘to go up, climb’ yunti- ‘to climb (something), mount (a horse)’

Sentence examples where the P is a location are:

(48) Thuku-thu rnu-ntiyi kulapuru

dog-erg lie-appl blanket.abs
‘The dog lay on the blanket.’ (Blake 1979b: ex 5.35b)

(49) Nga-thu tjaa rnanya rntia nguu rna-nti

1sg-erg here see.pst stone rel stand-appl
‘I saw the stone he stood on.’ (Blake 1979b: ex 5.38)

With transitive verbs, the applicative indicates that an oblique instrument or cause
has been promoted to P function, creating a ditransitive construction. An instrumental
applicative example is:

(50) Ntia nga-thu maa mani-ntiyi

money.abs 1sg-erg food.abs get-appl
‘I got food with the money.’ (Blake 1979b: ex. 5.31b)

An example of a cause applicative is:

(51) Lhaji-manti tjaa marapai tjipa-yi iti-yi

hit-appl here woman.abs this-erg man-erg
‘This man hit (him) because of the woman here.’ (Blake 1979b: ex 5.39)
 13 Applicative constructions in Australian Aboriginal languages 407

Blake mentions the use of -nti- with transitive base verbs to add a locative argument as
P, but gives no relevant examples.
The neighbouring, but apparently unrelated, Yalarnnga language also has an applica-
tive -nti-, briefly mentioned in the sketch of Breen and Blake (2007: 47) that is added to
intransitive motion and location verbs deriving a transitive stem whose P has a locative
sense. An example is:

(52) Mangurru-yu tjala ngu-nti-ma tjala kulapurru

dog-erg this lie-appl-prs this.abs blanket.abs
‘The dog is lying on the blanket.’ (Breen and Blake 2007: ex 3-191b)

The description does not mention whether this affix can be added to transitive base
verbs.

5.3 Maric

East of Kalkatungu and Karnic we find the Maric group of languages, spoken through-
out central and southern Queensland where there are a limited number of applicatives
derived from intransitive bases with a volitional agent-like S subject (often verbs of motion
or location). The P of the resulting transitive is understood as a location, comitative or
goal. Breen (1981: 319) gives the following examples from Margany and Gungabula:

gambira- ‘to return’ gambiny-ma- ‘to bring back’

ngandhi- ‘to talk’ ngandhi-ma- ‘to talk to’
dharti- ‘to like’ dharti-ma- ‘to like’

Holmer (1983: 186-187) has a slightly longer list for the closely related Gunggari:

ngalga- ‘to speak’ ngalga-ma- ‘to speak to’

binda- ‘to sit’ binda-ma ‘to sit with, to nurse’
wari- ‘to think’ wari-l-ma- ‘to think about’
warda- ‘to go away’ warda-ma- ‘to go away from’
gadi- ‘to lie, tell falsehoods’ gadi-ma- ‘to lie to’
warra- ‘to play’ warra-ma- ‘to play with’

In the closely related Wiri and Biri the corresponding affix is -ri-, as in (items marked
[H] are from Holmer 1983: 303, those marked [B] are from Beale 1974: 24–25):
408 Peter K. Austin

wadja- ‘to go’ wadja-ri- ‘to go away with, take’ [B, H]

yanhi- ‘to go, come’ yanhi-ri- ‘to bring, send’ [B, H]
dana- ‘to sit’ dana-ri- ‘to sit with’ [H]
wuna- ‘to lie’ wuna-ri- ‘to sleep with’ [H]

A sentence example is:

(53) Gayurrba-nggu dana-ri-ngala yalu

woman-erg sit-appl-prs baby.abs
‘The woman is sitting with (nursing) the baby.’ (Holmer 1983: 304)

The related Gangulu has -ni- or -yi-, depending on dialect (Holmer 1983: 273) with the
following instances recorded:

burra- ‘to get away’ burra-ni- ‘to get away with’

yani- ‘to go’ yani-ni- ‘to go with’
wuba- ‘to come’ wuba-yi- ‘to come with, to bring’
gandi- ‘to come’ gandi-yi- ‘to come with, to bring’
duni- ‘to say’ duni-yi- ‘to say to, to tell’

Maric languages lack applicatives of transitive verbs.

5.4 Paman

Several members of the Paman language group, spoken north of Maric in far north
Queensland, have applicatives based on volitional intransitive verbs, mostly expressing
stance and location but also including ‘laugh’ and ‘cry’, where the resulting P expresses
a location or comitative semantic role. Descriptions illustrating this are Kuuk Thayore
(Gaby 2006: 402–409), Wik-Mungkan (Kilham et al. 1986:407), and Yir-Yoront (Alpher
1991: 48). No Paman language has applicatives of transitive base verbs.

5.5 Waka-Waka and Goreng-Goreng

According to Holmer (1983: 8, 22, 94) languages of the Waka-Waka and Goreng-Goreng
groups spoken on the south-east Queensland coast have an applicative -ndi- or -ri- added
only to a sub-set of intransitive verb bases, as in the following Goreng-Goreng examples
(sentence usage instances are missing):
 13 Applicative constructions in Australian Aboriginal languages 409

bi- ‘to go’ bi-ndi- ‘to take’

balba- ‘to stand’ belbe-ndi- ‘to stand with’
mai- ‘to run’ mai-ndi- ‘to run after’
ngina- ‘to sit’ ngine-ndi- ‘to sit with’
yunma- ‘to lie’ yunme-ndi- ‘to sleep with, cohabit with’

From Holmer’s fragmentary data, it seems that these affixes cannot be added to transi-
tive base verbs.

5.6 Ngiyambaa

Ngiyambaa (Donaldson 1980: 163), once spoken in central New South Wales, south of
Maric, also has two transitivisation patterns: most intransitive verbs take the causative
affix -ma-l, however there are just two verbs which take the affix -ba-l, namely ginda-y
‘to laugh’ and yunga-y ‘to cry’, which follow an applicative pattern, where the transitive
P is understood as the goal of the action:

ginda-y ‘to laugh’ ginda-y-ba-l ‘to laugh at’

yunga-y ‘to cry’ yunga-y-ba-l ‘to cry at’

An example of such an applicative is:

(54) Burraa-dhu-nuu yunga-y-ba-ra

child-erg-2sg.acc cry-cm-appl-prs
‘The child is crying at you.’ (Donaldson 1980: ex 6-19)

There is no applicative construction for transitive base verbs. Note that languages
adjacent to Ngiyambaa—such as Baagandji to the west (Hercus 1982), Wangkumara to
the north (Robertson 1985) and Yuwaalaraay-Gamilaraay to the north-east (Williams
1980)—show no sign of applicativisation and have just one transitivisation pattern,
namely the causative.

5.7 Yidiny

This language was spoken in northern Queensland north-east of Maric and has an affix
-nga- which Dixon (1977: 304) labels “comitative”. It subdivides intransitive verbs into
those which it causativises and those which it applicativises, taking a P with a locative
or comitative semantics. The reported sub-set of applicativising intransitives is:
410 Peter K. Austin

djana- ‘to stand’ djana-nga- ‘to stand with’

nyina- ‘to sit’ nyina-nga- ‘to sit with’
badi- ‘to cry’ badi-nga- ‘to cry for’
mangga- ‘to laugh’ mangga-nga- ‘to laugh at’

Compare the following examples:

(55) Waguudja bunyaa-y gali-ng

man.abs woman-comit go-prs
‘The man is going with the woman.’ (Dixon 1977: ex 502)

(56) Wagudja-nggu bunya galii-nga-l

man-erg woman.abs go-appl-prs
‘The man is taking the woman.’ (Dixon 1977: ex 503)

Yidiny also has constructions where -nga- is added to a transitive verb root—here it
introduces a locative or instrumental element as P. Importantly, however, the transi-
tive verb base must first be intransitivised by occurring in the antipassive construction
where the base verb P nominal is placed in dative or locative case, the A becomes an S,
and the verb takes an affix -dji-. So, from the following transitive clause:

(57) Bamaa-l djugi galbaan-da gundaa-l

man-erg tree.abs axe-ins cut-pst
‘The man cut the tree with an axe.’ (Dixon 1977: ex 509)

We have the corresponding antipassive intransitive clause:

(58) Bama galbaan-da gundaa-dji-nyu djugii-l

man.abs axe-ins cut-antip-pst tree-loc
‘The man cut the tree with an axe.’ (Dixon 1977: ex 510)

Only now may the applicative -nga- be added to express the instrument in P function:

(59) Bamaa-l galban gundaa-dji-ngaa-l djugii-l

man-erg axe.abs cut-antip-appl-pst tree-loc
‘The man cut the tree with an axe.’ (Dixon 1977: ex 511)

It is clear that transitive base verbs can take the applicative -nga- but only when they
have been first detransitivised and made into volitional intransitives. A variant of this
strategy is also found in Dyirbal and Warrgamay (see § 5.7).
The Djabugay language is spoken immediately north of Yidiny and is apparently
quite closely related. Here we find a single transitivising affix -rri- (Patz 1991: 283–284,
 13 Applicative constructions in Australian Aboriginal languages 411

297), with a split for intransitive verbs of the same type observed above, i.e., causative
for most verbs but applicative with volitional predicates with an agent-like S argument.
Patz’s description mentions:

jungga- ‘to run’ djungga-rri- ‘to run with’

burra- ‘to fly’ burra-rri- ‘to fly with’
mangga- ‘to laugh’ mangga-rri- ‘to laugh at’
yarrn.ga- ‘to be afraid, dislike’ yarrn.ga-rri- ‘to hate’

Note that -rri- cannot be added to transitive verb roots.

5.8 Dyirbal and Warrgamay

Dyirbal was spoken in north Queensland adjacent to Yidiny and Djabugay (see § 5.7)
and has a single transitivising affix -m(b)a- that creates applied transitive stems from
intransitive stance roots (plus miyanday- ‘to laugh’, Dixon, p.c.), as in:

(60) Balan djugumbil banggul yara-nggu nyinay-ma-n

she.abs woman.abs he.erg man-erg sit-appl-nfut
‘The man is sitting with the woman’, ‘The man is married to the woman’ (Dixon
1972: ex 258)

(61) Balay djana-nggu bayi miyanday-ma-n

there 3pl-erg he.abs laugh-appl-nfut
‘They are laughing at the man there’ (Dixon p.c.)

Dyirbal is like Yidiny (§ 5.7) in having an antipassive construction which detransitivises

regular transitive clauses, creating an intransitive whose S corresponds to the transi-
tive base verb A and whose dative corresponds to the base verb P, as in the following
contrast:

(62) Bayi bargan banggul yara-nggu djurga-nyu

he.abs wallaby.abs he.erg man-erg spear-nfut
‘The man is spearing the wallaby.’ (Dixon 1972: ex 64)

(63) Bayi yara bagul bargan-gu djurga-na-nyu

he.abs man.abs he.dat wallaby-dat spear-antip-nfut
‘The man is spearing the wallaby.’ (Dixon 1972: ex 68)

For transitive base verbs, -m(b)a- may be added to create an applicative construction
where an instrument or locative has P function, but the erstwhile P of the base transi-
412 Peter K. Austin

tive root must be inflected for dative case. Thus, contrast the following transitive con-
struction with an instrument:

(64) Balan djugumbil banggul yara-nggu banggu yugu-nggu balga-n

she.abs woman.abs he.erg man-erg it.ins stick-ins hit-nfut
‘The man is hitting the woman with a stick.’ (Dixon 1972: ex 242)

with its applicative counterpart:

(65) Bala yugu banggul yara-nggu balga-lma-n

it.abs stick.abs he.erg man-erg hit-appl-nfut
bagun djugumbil-gu
she.dat woman-dat
‘The man is hitting the woman with a stick.’ (Dixon 1972: ex 253)

In the Mamu dialect, the applicative verb form required here is balga-nay-mba-n (Dixon
1972: 97), containing the antipassive affix (as in Yidiny, cf. (59) above). Antipassivisation
is covert in other dialects, except for locative/comitative applicatives where the antipas-
sive affix is obligatory before the applicative, as in:

(66) Bayi nyalngga banggun djugumbi-ru nyuga-nay-mba-n

he.abs boy.abs she.erg woman-erg grind-antip-appl-nfut
bagum djububala-gu.
it.dat flour-dat
‘The woman is grinding the flour with a boy beside her.’ (Dixon 1972: ex 265)

Again, we see that the applied affix can only be added to transitive verbs which have
first been detransitivised by the antipassive to create volitional active intransitive verbs.
Warrgamay, spoken south of Dyirbal and not closely related to it, has applicatives
marked by -ma- for motion and stance intransitive verbs, plus ‘laugh’ and ‘cry’ (Dixon
1981). The -ma- suffix can be added to transitive base verbs to create applicatives whose
P expresses an instrument, however the base P must be placed in dative case, as we saw
for Dyirbal above. This suggests that this language also has a covert antipassive that
applies to create a derived intransitive before the applicative can be added.

6 Conclusions
This chapter surveys various types of applicative constructions found in Australian Abo-
riginal languages of the Pama-Nyungan and non-Pama-Nyungan groups. Applicatives
are found in a number of languages, but unfortunately the available data in descrip-
 13 Applicative constructions in Australian Aboriginal languages 413

tions tends to be scanty, and many of the relevant languages are no longer spoken.
Geographically, applicatives only occur in central and northern languages of South
Australia, Northern Territory and Western Australia, plus the whole of Queensland. No
language in Central or Western Australia below the Kimberley region (including those
which are well described) has applicative constructions. Only one example is found in
New South Wales (see § 5.6) and none in Victoria.10
The constructions observed can be characterized as follows:

Morphology
– Pama-Nyungan applicatives are encoded by word-building suffixes which occur
between the verb base and tense/aspect/mood/dependent suffixes. Non-Pama-Nyun-
gan applicatives can be suffixes or prefixes to the verb.
– A few languages have two applicative affixes, one for locative/comitative applica-
tives and a different one for benefactive/malefactive applicatives.
– No language allows double applicativisation with a single base verb, but some
allow causative to be followed by applicative.
– Applicativised verbs have the same morphological properties as basic transitive or
ditransitive verbs and can be further derived by, for example, reflexive-reciprocal
constructions.
– There are no reported periphrastic or analytical applicatives in any Australian lan-
guage.

Syntax
– Almost all the constructions surveyed are P-applicatives and involve an increase
in valence.
– A few languages (Diyari, Arabana-Wangkanguru) have D-applicatives where the
added participant carries a non-P case marker, such as dative or instrumental.
Bardi has an X-applicative.
– All applicatives are optional in the sense that semantic roles like locative, comita-
tive, instrumental, or benefactive can always be expressed as case-marked adjuncts
in regular non-applicative constructions.
– Most languages only have applicatives of active intransitive verbs with a volitional
S, typically ‘to laugh at’ and ‘to cry for’, plus, optionally verbs of stance or motion.
– In some languages the same affix that marks applicatives with such verbs also
attaches to other intransitive verbs (e.g., change of state or location) to express
a causative meaning. In some languages there is a separate causative morpheme
(which does not attach to transitive verbs).

10 The available sources for Victoria and much of southern New South Wales are fragmentary, so the
apparent geographical gap may rather be due to the lack of reliable descriptions.
414 Peter K. Austin

– A limited number of languages have applicatives of transitive base verbs, and for
some of those which do (such as Yidiny, Dyirbal, Warrgamay) the base verb must
first be detransitivised by an antipassive construction before the applicative can be
created (i.e. Vtr Vintr Vtr).

Semantics
– The applied phrase typically expresses locative or comitative semantic roles. Mur-
rinhpatha is exceptional in that only a source or malefactive role is associated with
the applicative construction;
– Some languages also extend applicatives to expressing an instrument role as the
P-applicative argument.
– In languages with two applicative constructions, the additional one typically
expresses a beneficiary or maleficiary (depending on verb semantics) as the P
argument.

Diachrony
The only potential historical sources for applicatives that have been identified are in
Daly languages where the origin appears to be grammaticisation of incorporated body
parts, such as ‘hand’ or ‘eye’ (see § 4.3).

Abbreviations
A transitive subject
abl ablative
abs absolutive
acc accusative
all allative
alt altruistic
anaph anaphoric
antip antipassive
aug augmented
aux auxiliary
ben benefactive
cm conjugation marker
comit comitative
dat dative
dim diminutive
dobj direct object
du dual
erg ergative
excl exclusive
 13 Applicative constructions in Australian Aboriginal languages 415

f feminine
fill morphological filler
fut future
hum human
implds implicated clause-different subject
incl inclusive
inch inchoative
ins instrumental
irr irrealis
loc locative
m masculine
min minimal
n neuter
nf non-feminine
nfut non-future
nm non-masculine
nom nominative
npst non-past
obl oblique
obj object
P transitive object
pc past continuous
pfv perfective
pi past imperfective
pl plural
poss possessive
pp past perfective
prs present
pro pronoun
pst past
ptcp participial
punc punctual
purp purposive
rdp reduplication real realis
rec recent
rel relativiser
rem remote
rr reflexive-reciprocal
S intransitive subject
sbj subject
sg singular
th thematic consonant
top topic
trz transitiviser
x>y x acts on y
416 Peter K. Austin

References
Alpher, Barry. 1991. Yir-Yoront lexicon. Berlin: Mouton de Gruyter.
Austin, Peter. 1981a. A grammar of Diyari, South Australia. Cambridge: Cambridge University Press.
Austin, Peter. 1981b. Switch-reference in Australia. Language 57(2). 309–333.
Austin, Peter. 1997. Causatives and applicatives in Australian Aboriginal languages. In Kazuto Matsumura &
Tooru Hayasi (eds.), The dative and related phenomena, 165–225. Tokyo: Hituzi Syobo.
Austin, Peter. 2021. A grammar of Diyari, South Australia. 2nd edn. London: School of Oriental and African
Studies, University of London.
Beale, Tony. 1974. A grammar of the Biri language of north Queensland. Unpubl. manuscr., Australian
National University.
Blake, Barry J. 1977. Case marking in Australian languages. Canberra: AIATSIS.
Blake, Barry. J. 1979a. Pitta-Pitta. In R. M. W. Dixon & Barry J. Blake (eds.), Handbook of Australian languages,
Vol. 1, 183–242. Canberra: Australian National University Press.
Blake, Barry. J. 1979b. A Kalkatungu grammar. Canberra: Pacific Linguistics
Blake, Barry J. 1987. Australian Aboriginal grammar. New York: Croom Helm.
Bouckaert, Remco, Claire Bowern, & Quentin Atkinson. 2018. The origin and expansion of Pama-Nyungan
languages across Australia. Nature Ecology & Evolution 2. 741–749.
Bowern, Claire. 2001. Karnic classification revisited. In Jane Simpson, David Nash, Mary Laughren, Peter
Austin, & Barry Alpher (eds.), Forty years on: Ken Hale and Australian languages, 245–261. Canberra:
Pacific Linguistics.
Bowern, Claire. 2012. A grammar of Bardi. Berlin: Mouton de Gruyter.
Bowern, Claire. 2022. How many languages are and were spoken in Australia? In Claire Bowern (ed.), The
Oxford guide to Australian languages. Oxford: Oxford University Press.
Breen, Gavan. 1981. Margany and Gunya. In R. M. W. Dixon & Barry J. Blake (eds.), The handbook of
Australian languages, Vol. 2, 275–393. Canberra: Australian National University.
Breen, Gavan & Barry J. Blake. 2007. The grammar of Yalarnnga: A language of western Queensland. Canberra:
Pacific Linguistics.
Crowley, Terry & R. M. W. Dixon. Tasmanian. In R. M. W. Dixon & Barry J. Blake (eds.), Handbook of Australian
languages, Vol. 2, 419–445. Amsterdam: John Benjamins.
Dench, Alan. 1982. The development of an accusative case marking pattern in the Ngayarda languages of
Western Australia. Australian Journal of Linguistics 21. 43–59.
Dixon, R. M. W. 1972. The Dyirbal language of North Queensland. Cambridge: Cambridge University Press.
Dixon, R. M. W. 1977. A grammar of Yidiny. Cambridge: Cambridge University Press.
Dixon, R. M. W. 1980. The languages of Australia. Cambridge: Cambridge Unversity Press.
Dixon, R. M. W. 1981. Warrgamay. In R. M. W. Dixon & Barry J. Blake (eds.), The handbook of Australian
languages, Vol. 2, 1–145. Canberra: Australian National University.
Donaldson, Tamsin. 1980. Ngiyambaa: the language of the Wangaaybuwan. Cambridge: Cambridge University
Press.
Eather, Bronwyn 1990. A grammar of Nakkara (central Arnhem Land coast). Canberra: Australian National
University dissertation.
Evans, Nicholas (ed.). 2003a. The non-Pama-Nyungan languages of northern Australia: comparative studies of
the continent’s most linguistically complex region. Canberra: Pacific Linguistics.
Evans, Nicholas. 2003b. Bininj Gun-Wok: A pan-dialectal grammar of Mayali, Kunwinjku and Kune. Canberra:
Pacific Linguistics.
Gaby, Alice R. 2006. A grammar of Kuuk Thaayorre. Melbourne: University of Melbourne dissertation.
Green, Ian. 1989. Marrithiyel: A language of the Daly River Region of Australia’s Northern Territory. Canberra:
Australian National University dissertation.
 13 Applicative constructions in Australian Aboriginal languages 417

Harvey, Mark. 2003. A Grammar of Gaagudju. Berlin: Mouton de Gruyter.

Heath, Jeffrey. 1978. Ngandi grammar, texts and dictionary. Canberra: Australian Institute of Aboriginal
Studies.
Heath, Jeffrey. 1984. Functional grammar of Nunggubuyu. Canberra: AIAS.
Hercus, Luise A. 1982. The Bagandji language. Canberra: Pacific Linguistics.
Hercus, Luise A. 1990. A grammar of the Arabana-Wangkangurru language, Lake Eyre Basin, South Australia.
Canberra: Pacific Linguistics.
Holmer, Nils M. 1983. Linguistic survey of South-east Queensland. Canberra: Pacific Linguistics.
Horrack, Kate. 2018. Argument realisation in Wubuy. Melbourne: University of Melbourne dissertation.
Horton, David. 1996. Map of Indigenous Australia. Canberra: AIATSIS.
Kilham, Christine, Mabel Pamulkan, Jennifer Pootchemunka & Topsy Wolmby. 1986. Dictionary and source
book of the Wik-Mungkan language. Darwin: Summer Institute of Linguistics / Australian Aborigines
Branch.
Lee, Jay & Kazuko Obata. 2010. Languages of Aboriginal and Torres Strait Islander People – A uniquely
Australian heritage. In Year Book Australia (Vols. 2009–10). Canberra: Australian Bureau of Statistics.
Marmion, Doug, Kazuko Obata & Jacky Troy. 2014. Community, identity, wellbeing: The report of the Second
National Indigenous Languages Survey. Canberra: Australian Institute of Aboriginal and Torres Strait
Islander Studies.
McKay, Graham R. 1979. Gender and the category unit augmented. Oceanic Linguistics 18(2). 203–210.
McKay, Graham R. 2011. Rembarrnga, a language of central Arnhem Land. Munich: LINCOM.
Merlan, Francesca. 1982. A grammar of Mangarayi. Berlin: Mouton de Gruyter.
Merlan, Francesca. 1983. Ngalakan grammar, texts and vocabulary. Canberra: Pacific linguistics.
Merlan, Francesca. 1994. A grammar of Wardaman: A language of the Northern Territory of Australia. Berlin:
Mouton de Gruyter.
Miceli, Luisa. 2015. Pama-Nyungan. In Claire Bowern & Bethwyn Evans (eds.), The Routledge handbook of
historical linguistics, 704–725. London: Routledge.
Nichols, Johanna. 1986. Head-marking and dependent-marking grammar. Language 62(1). 56–119.
Nordlinger, Rachel. 1998. A grammar of Wambaya, Northern Territory Australia. Canberra: Pacific Linguistics.
Nordlinger, Rachel. 2019. From body part to applicative: Encoding ‘source’ in Murrinhpatha. Linguistic
Typology 23(3). 401–433.
Patz, Elizabeth. 1991. Djabugay. In R. M. W. Dixon & Barry J. Blake (eds.), The handbook of Australian
languages, Vol. 4, 245–347. Melbourne: Oxford University Press
Pensalfini, Rob. 2003. A grammar of Jingulu, an Aboriginal Language of the Northern Territory. Canberra: Pacific
Linguistics.
Piper, Nick. 1989. A sketch grammar of Meryam Mer. Canberra: Australian National University MA thesis.
Ponsonnet, Maia. 2021. Transfer and applicative constructions in Gunwinyguan languages (non-Pama-
Nyungan, Australia). In Myriam Bouveret (ed.), Give Constructions across languages, 121–144.
Amsterdam: John Benjamins.
Reid, Nicholas. 1990. Ngan’gityemerri: a language of the Daly River region, Northern Territory of Australia.
Canberra: Australian National University dissertation.
Reid, Nicholas. 2000. Complex verb collocations in Ngan’gityemerri: A non-derivational mechanism for
encoding valency alternations. In R. M. W. Dixon & Alexandra Aikhenvald (eds.), Changing valency: Case
studies in transitivity, 333–359. Cambridge: Cambridge University Press.
Robertson, Carol. 1985. Ngalkagarla ngandrangu yawarranha nhanha Wangkumadanha: Let’s learn
Wangkumara! Based on the notes of Gavan Breen and the Wangkumara language tapes of Albert and
Cecil Ebsworth. Illustrated by Carol Robertson and Stephen Fernando. Armidale: Aboriginal Education
Unit, Department of Technical and Further Education.
Rumsey, Alan. 2018. The sociocultural dynamics of indigenous multilingualism in northwestern Australia.
Language and Communication 62. 91–101.
418 Peter K. Austin

Saulwick, Adam. 2003. Aspects of the verb in Rembarrnga, a polysynthetic language of northern Australia:
Grammatical description, texts and dictionary. Melbourne: University of Melbourne dissertation.
Schultze-Bernd, Eva. 2000. Simple and complex verbs in Jaminjung: A study of event categorisation in an
Australian language. Nijmegen: Katholieke Universiteit Nijmegen dissertation.
Sharpe, Margaret. 1972. Alawa phonology and grammar. Canberra: AIAS.
Singer, Ruth. 2016. The dynamics of nominal classification: Productive and lexicalised uses of gender agreement
in Mawng. Berlin: Walter de Gruyter.
Williams, Corinne J. 1980. A grammar of Yuwaalaraay. Canberra: Pacific Linguistics.
Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs
14 Applicativizing preverbs in selected
European languages
Abstract: This chapter surveys the morphology, syntax, and semantics of applicativiz-
ing preverbs in English, German, Hungarian, and the Slavic and Baltic languages, with
some comments on their non-applicativizing uses. Applicativizing preverbs may be
particles or affixes, are transparently related to adverbs/adpositions with spatial and/
or aspectual functions, and introduce a new participant to the clause (usually a direct
object, occasionally an oblique argument, rarely an indirect object). The chapter pays
special attention to English out-verbs and German be-verbs, which have semantics of
a kind hitherto unattested outside Germanic; it also gives a detailed overview of the
semantics of the relatively numerous Slavic and Baltic preverbs.

1 Introduction
The term preverb as used in Indo-European linguistics denotes a class of preverbal
particles or prefixes that form a close semantic unit with their host verb and frequently
appear as adverbs or adpositions as well (Booij and Van Kemenade 2003). Such pre-
verbs often have spatial and/or aspectual semantics; this has been extensively explored
in the literature on Romance, Germanic, and Slavic. (The authoritative sources on early
Indo-European preverbs are Kuryłowicz 1964 and Watkins 1964; see Rousseau 1995,
Booij and Van Marle 2003, and references therein for studies covering modern lan-
guages, also beyond Indo-European, and Arkadiev 2014, 2015 for a broader perspec-
tive.) What has attracted comparatively less attention is the fact that some preverbs
have a syntactic effect, either in addition to or instead of the spatial-aspectual meaning:
a number of preverbs can alter the verb’s argument structure to introduce or promote
non-subjects in the clause. (See Zúñiga and Creissels, this volume, for a discussion of
the notion of applicativization and an explanation of the terms P-applicative, D-applica-
tive, and X-applicative.) After this introduction, the present chapter surveys the mor-
phology, syntax, and semantics of such applicativizing preverbs in selected languages
of Europe, namely English (§ 2), German (§ 3), Slavic and Baltic languages (§ 4), and
Hungarian (§ 5).1 Section 6 closes the chapter summarizing the findings and tentatively

1 Close parallels are found across Germanic. Specialized studies focusing on preverbs in languages
other than the ones outlined here range from numerous (for Dutch; see, e.g., Van der Auwera 1999 and

Acknowledgments: We are grateful to Denis Creissels, Andrew McIntyre, Jurgis Pakerys, Guido Seiler, Sergej
Skorwid, Mladen Uhlik, and Björn Wiemer for their valuable comments on earlier versions of this chapter.

https://doi.org/10.1515/9783110730951-014
420 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

framing them in comparative terms, along the lines proposed in Zúñiga and Creissels
(this volume).
The Dutch example in (1) illustrates the kind of phenomena at the center of attention
here. The intransitive verb werken ‘work’ can occur with a direct object when applicativ-
ized by a preverb like the particle uit:

(1) Dutch2
a. Het systeem werk-t langzaam.
the system work-3sg slow(ly)
‘The system works slowly.’
(https://www.asus.com/be-nl/support/FAQ/1042398, 31.12.21)
b. Hier kun je ongestoord nieuwe plannen uit-werk-en.
here can you undisturbed new plans appl-work-inf
‘Here, you can elaborate new plans undisturbed.’
(https://www.beleefzaltbommel.nl/winkels/work-zaltbommel, 29.04.22)

We find close semantic equivalents of Dutch uitwerken ‘elaborate, effect’ in the lan-
guages of our sample that are also close morphosyntactic equivalents: English work out,
German ausarbeiten, Hungarian kidolgoz, and Czech vypracovat. (English is the excep-
tion here, featuring as it does a postverbal particle; all other languages have preverbal
particles or prefixes). The pattern is definitely old: even the classical languages have
equivalents (viz. Classical Greek exergázomai and Classical Latin ēlabōrō). Among the
many related phenomena that fall outside the scope of this chapter are those addressed
by Jackendoff (1997) in his study of English constructions like Bill slept the afternoon
away and Susan worked her head off last night.

2 English
2.1 Morphology

English particles are postverbal rather than preverbal and characteristically have a
spatial and/or aspectual meaning; the particles out and off occasionally applicativize as
well (e.g., in I worked the instructions out and he slept off the flu). There are cases like

Van Kemenade and Los 2003) through few (for Scandinavian; see, e.g., Gronemeyer 1995) to non-existent
(for Yiddish). Parallels are also found in Romance, for instance, French le coureur est passé devant son
concurrent ‘the runner passed in front of his competitor’ vs. le coureur a dépassé son concurrent ‘the
runner overtook his competitor’ (Denis Creissels, p.c.).
2 Unless otherwise specified, numbered examples come from the authors’ personal knowledge (in case
they speak the specific language) or personal documentation (in case they have worked on it but have
not yet published the data).
 14 Applicativizing preverbs in selected European languages 421

outride, which means both ‘surpass in riding’ and ‘(tackle and) survive, ride out’, but
they are infrequent; many out-verbs have only one of these kinds of meaning, namely
either the one that clearly differs from the meaning of the [V out] complex or the one
that is closer to it.3
The most productive applicative marker is the verbal prefix out-. It can also, albeit
less productively, verbalize adjectives (outsmart) and nouns (outnumber). The few
instances where deverbal out- is blocked include the verb have (e.g., we have cars vs.
✶
we outhave them), some change-of-state verbs (e.g., the mugs dried vs. ✶the mugs out-
dried the glassware), and some idiomatic expressions (e.g., we shot the breeze vs. ✶we
outshot them) (Ahn 2022: 446). The prefix be- applicativizes some verbs but is no longer
productive in present-day English; many be-verbs are also either archaic, formal, or
informal in standard varieties of the language. This prefix can also occasionally verbal-
ize adjectives (bedumb) and nouns (befriend).4 As in German, in some cases, non-finite
forms are in use even where finite verbs are not (beloved, benighted). The prefix out-
does not combine with the particle out or the prefix be-.5 Other prefixes, like down-,
over-, and under- are quite productive in general but applicativize only occasionally
(e.g., in downplay, overcome, undergo).
The prefix out- is a spatial or aspectual verbal marker in instances that are mostly
archaic or poetic, as in outburst ‘burst out’ and outgo ‘go out’. More often, however, it
functions as a dedicated applicativizer or verbalizer, much as be-.
The origin of out(-) is OE ūt ‘out’ / ūte ‘outside’ < PG ✶ūt / ✶ūtai < PIE ✶úd ‘upwards,
away’ (cf. its German cognate aus ‘out’ < OHG ūz, which can also applicativize in that
language; see § 3.2). The English prefix be- is cognate with German be-; see Section 3.1.

2.2 Syntax

English applicatives are normally optional P-applicatives. Out-verbs are invariably tran-
sitive, irrespective of their base’s transitivity.6 Applicative be-verbs can be monotransi-
tive or ditransitive.
The applied phrase (AppP) is a direct object in applicative constructions (ACs) with
either out-verbs (2) or be-verbs (3):

3 See Dehé et al. (2002), Haiden (2006), Dehé (2015), and McIntyre (2015) for overviews of so-called
particle/phrasal verbs in Germanic. See Thim (2012) for English preverb+verb constructions and their
diachrony.
4 Not only out-, be-, and other prefixes but also several particles appear with nominal and adjectival
bases that do not normally occur as verbs on their own, for instance, in tone down, clock in, tail off,
soldier on, pig out, ramp up and dumb down, single out, wise up. See McIntyre (2016) for more details.
5 See Schröder (2011) for an overview of the productivity of verbal prefixes in English.
6 See Ahn (2022) for an analysis of this particular feature in the Chomskyan tradition.
422 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(2) English
a. Stocks performed better than other investments. (BC)
b. Stocks outperformed other investments. (AC)

(3) English (b: Peter Russell, Nine, 1954)

a. Imagine Achilles barking at a plan of Agamemnon’s! (BC)
b. Imagine Achilles bebarking a plan of Agamemnon’s! (AC)

Passivized out-ACs grant their P’s subject status; this is illustrated for outdo and
outrun in (4):

(4) English
a. He himself was outdone by Jocelyne.
(J. Willis, The Irish Nation, 1875, p. 105)
b. Not all of the Cops were outrun.
(Tampa Bay Times, Florida, September 3, 2005)

Transitive applicative be-verbs in passive clauses behave like out-verbs: the new P
is granted subject status, as in the relative lack of female doctors was bemoaned.7 Other
verbs show some syntactically and lexically conditioned variation. Active ditransi-
tive begrudge, for instance, can occur either with the double-object construction (e.g.,
they begrudged him his wealth) or, albeit under more restricted circumstances, with
the to-construction (e.g. they begrudged its facilities to a city that. . .) (Wasow 1997: 84).
Note in (5) that either non-Agent-like (i.e., non-Experiencer) participant can be a passive
subject, but this is not a simple consequence of the corresponding active syntax; with
the to-construction, the subject is—as expected—the Theme-like argument (5a); with
the double-object construction, however, the subject can be either the Goal-like argu-
ment (5b) or the Theme-like argument (5c):

(5) English
a. They took satisfaction in abusing the advantages that were begrudged to them.
(Cumberland Evening Times, Maryland, January 28, 1952)
b. I was begrudged what I was paid.
(Archbold Buckeye, Ohio, August 8, 2007)
c. Even these precarious occupations were begrudged them.
(Joseph Strauss, ‘A modern Synhedrin’, Westminster Review, 1914, p. 304)

The prefix out- is normally transitivizing/valency-increasing, but it can also be redi-

recting/valency-neutral; (6) shows one example of each type. The marker transitivizes

7 Source: McKinstry, Brian. 2008. Are there too many female medical graduates? Yes. The BMJ 336(7647): 748.
 14 Applicativizing preverbs in selected European languages 423

either intransitive or ambitransitive verbs (outstay, outsit, outdrink) (6a); it also appears
on monotransitives and ditransitives when construed with generic/implicit objects in
the BC (outkill, outgive) (6b):

(6) English (b: Pinker 2012: 119, The better angels. . ., Penguin ed.)
a. The dictator outstayed several of other leaders.
b. Southerners do not outkill northerners in homicides carried out during
robberies. . .

When occurring on (di)transitive base verbs, out- suppresses any base direct objects in
the AC in order to accommodate the applied phrase (e.g., she killed more people than
you → she outkilled you). By contrast, valency-neutral be-—found more often in earlier
stages of the language—does not normally change clausal syntax (e.g., he cursed the
enemy → archaic he becursed the enemy). The prefix be- can also be valency-increasing,
as in the alternation illustrated in (3) above.
Out-verbs normally participate in straightforward BC-AC oppositions when they
are not deadjectival or denominal; consider (7), for instance, where the relevant con-
stituent is licensed by the expression longer than in the BC:

(7) English
a. The guests stayed longer than their hosts. (BC)
b. The guests outstayed their hosts. (AC)

By contrast, be-verbs seldom show alternations like the one in (8), where a PP in the BC
corresponds to an object in the AC:

(8) English
a. An unusual calm fell upon the wetlands. (BC)
b. An unusual calm befell the wetlands. (AC)

Even with deverbal be-, frequently there is no BC-AC alternation; consider verbs like
behave ‘act’, become ‘befit’, begive ‘endow’, for example.
Particles have a comparable effect when they applicativize; with out, for instance,
the applied phrase is a direct object and the AC (9b) can undergo passivization (9c):

(9) English
a. She is still working on those details. (BC)
b. She is still working out those details. (AC)
c. Those details haven’t been worked out yet.

Note that the valency-increasing effect of spatial/aspectual particles is a special case of a

broader phenomenon in Germanic (McIntyre 2007); intransitive vote (10a), for instance,
424 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

can accommodate a Figure object by adding out with a covert or overt Ground (10b), or
by adding other particles, like in and off (and even up or down) (10c):8

(10) English
a. She is too young to vote.
b. We voted her out (of office).
c. New Boston voted the measure in as well during their meeting Thursday.
(https://kkyr.com/four-day-school-weeks-begin-this-fall-in-four-east-texas-
districts, 29.04.22)

Table 1 below summarizes the syntactic effects of out-, be-, and out.

Table 1: Syntactic effects of three English applicative markers.

out- be- out

valency-increasing (1/2 → 2) ( ) ( )
valency-neutral (2 → 2)
valency-decreasing (3 → 2) ( )

2.3 Semantics

Like many other English compound verbs, those of the out-V and [V out] types often show
spatial and aspectual meaning components related to their prefix or particle and do not
alter argument structure. When some prefixes or particles applicativize, however, not
only do they introduce an argument to the clause, but they also have “a scalar or quanti-
tative reading, rather than a purely locative one” (Bauer et al. 2013: 353).
More precisely, the semantics of ACs headed by out-verbs seems to be meaning-
fully captured by an interpretational cline ranging from two related but distinct poles
(Kotowski 2020). With the “comparative” reading, the subject and the applied object
engage in an event of the same kind, and the latter is a threshold exceeded by the
former. With the “resultative” reading, the applied object is a participant of a sub-event
caused by an event in which the subject participates; here, the notion of competition
(and therefore of defeat) is prominent. Depending on semantic features of the partici-
pants and contextual clues, one of these readings is usually favored. In (11), for instance,
the applicative clause to outsit your neighbors can be given a comparative interpreta-
tion, according to which the implicit subject simply surpasses the object in sitting (i.e.,
someone sits longer than their neighbors). Alternatively, the resultative interpretation

8 See McIntyre (2007) for a discussion of applied Ground objects, as well as for a general discussion of
the argument structure of English and German verbs.
 14 Applicativizing preverbs in selected European languages 425

regards sitting as a competition in which the object loses out (i.e., some sits longer than
their neighbors and thereby the latter are outdone):

(11) English (Kotowski 2020: 63)

The trick is to outsit your neighbors. Lots of hunters get tired and antsy after
spending many hours in a stand, and start coming down to the ground by 10 A.M.
or so.

Further note that out-verbs are monotransitive irrespective of their base’s transitivity
(see § 2.3), and that the semantic roles of their subjects and objects are not mechanically
adopted from their base counterparts as with canonical applicatives. Consider Atlanta
also out-rained Seattle in 1922 and 1923 (Ahn 2022: 459), where both arguments are
arguably introduced into the clause by out-. By a similar token, both the subject and the
object in (12a) are agentive (they are Agents of transitive sell), just as both arguments
in (12b) are patientive (they are Patients of sell in its habitual/potential-passive use).9 In
both clauses, however, an asymmetrical threshold-/competition-related feature is com-
bined with base agentivity/patientivity in order to arrive at a composite semantic role:

(12) English (Ahn 2022: 457)

a. We out-sell all other fruit sellers.
b. Bananas out-sell plums.

Lastly, note that base verbs show clear actionality-related tendencies (Kotowski 2020):
activities (outrun) and semelfactives (outblink) seem to be much more common than
states (outweigh) and achievements (outwin).
As in German, the semantic effect of be-prefixation is heterogeneous. With many
verbs (like befall in (8) above), be- in the AC closely mirrors the spatial and related
notions expressed by prepositions like about, across, around, at, by, on, over, and to in
the BC. Interestingly enough, the semantics seem to have included something akin to the
“holistic” effect present with German be-verbs in the past, namely a feature of spatial
expansion, intense action, and/or higher affectedness (see § 3.3).10 Nowadays, this holds
only marginally and occasionally, for verbs like beset ‘attack, especially from all sides’,
and for some speakers with belabor ‘work (hard) on’ and bedazzle ‘disarm by dazzling’.

9 Interestingly enough, as mentioned in Section 2.1, some change-of-state verbs do not behave like
sell, which applicativizes in both its transitive and its patientive-intransitive (“middle”) version. For
instance, dry can be applicativized in its transitive version (Alun dried silverware better than Colin in
the dish-washing competition vs. Alun outdried Colin in the dish-washing competition), but not in its
patientive-intransitive (non-causative) version (the mugs dried vs. ✶the mugs outdried the glassware)
(Ahn 2022: 446).
10 See Beavers (2017) for a discussion of the holistic effect in the so-called locative alternation in
English.
426 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

2.4 Lookalikes

As mentioned in Section 2.1, English applicative preverbs exist in spaces teeming with
lookalikes. Since uncoded polyvalency alternations are quite common in the language,
syntactic lookalikes are the norm rather than the exception, and there is a vast litera-
ture that addresses them (see Levin and Rappaport Hovav 2005 and references therein).
The following examples illustrate three common alternations that often correspond to
BC-AC pairs in other languages, namely the benefactive alternation (13a–b), the locative
alternation (13c–d), and the conative alternation (13e–f):

(13) English
a. I bought flowers for Bridget.
b. I bought Bridget flowers.
c. You sprayed paint on the wall.
d. You sprayed the wall with paint.
e. She shot at the sheriff.
f. She shot the sheriff.

By contrast, the dative alternation (14a–b) does not normally correspond to BC-AC pairs
in languages with applicatives:

(14) English
a. He gave flowers to Claire.
b. He gave Claire flowers.

Morphologically, there is no English preverb, either particle or prefix, that is

dedicated for, or even predominantly used with, an applicativizing function; all such
markers have spatial/aspectual notions as their most usual functional domain. The
closest present-day English comes to having something like an applicative marker that
has other functions as well (rather than the other way round) is the prefix out- with
intransitive bases.

3 German
3.1 Morphology

Before discussing applicativizing preverbs, some general remarks on German preverbs

are in order. A number of these elements occur only as verb prefixes, like be- and ent-;
they are unstressed and cannot be detached from their host. Other preverbs occur as
either particles (sometimes called “separable prefixes” in the literature) or preposi-
 14 Applicativizing preverbs in selected European languages 427

tions, like ab(-) ‘off, from’ and aus(-) ‘out’. Yet other preverbs occur as either prefixes,
particles, or prepositions, like durch(-) ‘through’ and um(-) ‘around’. Verb particles are
stressed and appear detached from their host in some forms (e.g., finite synthetic forms,
like es laufen Gerüchte um ‘rumors are circulating’) but attached to them in others (e.g.,
nonfinite forms and finite analytic forms that used them, like es sind Gerüchte umge-
laufen ‘rumors have circulated’). The slot for prefixes is closest to the verb root, while
the slots for particles are more peripheral (e.g., an-zu-be-fehlen ‘to order, entrust’). Most
prefixes and particles are invariable and multifunctional (i.e., they applicativize some
verbs but have different functions with others; see § 3.4); the prefix ent- changes to emp-
before f in some verbs (empfangen, empfehlen, empfinden) but not in others (entfallen,
entfesseln, entführen).
German applicative preverbs can also be classified in three groups. Group I consists
of the prefixes be-, er-, ver-, zer-, and ent-, which only occur as prefixes; (15) illustrates
the first of these:

(15) German
a. Sie arbeit-et an ihr-em letzt-en Roman. (BC)
she.nom work-3sg at her-sg.m.dat last-sg.m.dat novel
b. Sie be-arbeit-et ihr-en letzt-en Roman. (AC)
she.nom appl-work-3sg her-sg.m.acc last-sg.m.acc novel
Both: ‘She is working on her last/latest novel.’

These markers differ regarding their semantics and often also their syntax. As to their
occurrence across the lexicon, many verbs can take several of them: schießen ‘shoot’,
for instance, can be applicativized with any of these markers (erschießen ‘shoot dead’,
verschießen ‘shoot off’, zerschießen ‘pierce/destroy through shooting’). Others are
slightly more restrictive: sprühen ‘spray’ is fine and predictable with be- and ver- (both
‘spray’, with some nuances; see Stiebels 1996: 98–102) but does not take er- or zer-. By
contrast, auxiliaries like sein ‘be’, werden ‘become’ and all modal verbs except dürfen
‘may’, as well as verbs like heißen ‘be called’ and wissen ‘know’, cannot take any of these
markers.
Group II consists of the prefixes durch- ‘through’, über- ‘over’, and um- ‘around’ (as
well as, albeit more marginally, unter- ‘under’). Example (16) illustrates um-:

(16) German
a. Der Satellit läuft um den Planeten. (BC)
art.sg.m.nom satellite run.3sg around art.sg.m.acc planet.acc
b. Der Satellit um-läuft den Planeten. (AC)
art.sg.m.nom satellite appl-run.3sg art.sg.m.acc planet.acc
Both: ‘The satellite orbits the planet.’
428 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

These elements can also occur as prepositions (as in [16a]) or as particles (as in the
example given at the beginning of this section, úmlaufen ‘circulate’, different from
umláufen ‘orbit’ in [16b]). Note that the productive applicativizing prefixes in Baltic-Slavic
and some applicativizing particles in Hungarian have roughly the same meanings; see
Sections 4.1 and 5.1, respectively. As with those in Group I, these markers appear in lex-
ically conditioned, unpredictable patterns. Laufen ‘run’, for instance, is applicativized
by durch- (‘run through’) and um- (‘run around’), but not by über- (‘overflow; defect’).
By contrast, denken ‘think’ is applicativized by durch- (‘think through’), über- (‘rethink,
reconsider’), and um- (ambitransitive ‘rethink’).
Group III consists of particles that have spatial-aspectual meanings with many
verbs in most contexts but can occasionally applicativize, like ab(-) ‘off, from’ and an(-)
‘at, in’ , as well as like aus(-) ‘out’ and auf(-) ‘up, on’ in (17).11 Note also the subtle seman-
tic variation in such cases:

(17) German
a. Wir arbeit-en an unser-em neu-en Plan. (BC)
we.nom work-1pl at our-sg.m.dat new-sg.m.dat plan
‘We are working on our new plan.’
b. Wir arbeit-en unser-en neu-en Plan aus. (AC)
we.nom work-1pl our-sg.m.acc new-sg.m.acc plan appl
‘We are working out / elaborating our new plan.’
c. Wir arbeit-en die Vergangenheit auf. (AC)
we.nom work-1pl art.sg.f.acc past appl
‘We are processing the past.’

For an overview of German applicativizing preverbs, see Table 2.

Table 2: Main German applicativizing preverbs.

Group I Group II Group III

be- durch- ‘through’ ab ‘off’

er- über- ‘over’ auf ‘up’
ver- um- ‘around’ aus ‘out’
zer- (unter- ‘under’) ...
ent-

The etymons of Group I markers are as follows: be- < OHG bi- < PG ✶bi- ‘on, at, by’; er-
is originally spatial in origin, namely OHG ar-/ir- < PG ✶uz- ‘out, up’; ver- < OHG. far-/fir- <

11 See Stiebels (1996) for a detailed discussion of such particles. See also Cysouw (2023: 362–264) for a
brief overview of applicativizing prefixes and particles.
 14 Applicativizing preverbs in selected European languages 429

PG ✶fra- ‘in front of’, ✶firi- ‘through, across’, ✶furi- ‘for, in front’; zer- possibly comes from
OHG. zar-/zur- < ze-ar-/ze-ir- < PG ✶twiz- ‘in two, apart’; ent- < OHG ant- < PG ✶anda-/andi-
‘against, un-’ (apparently originally an allative of the PIE noun ✶h2énts ‘face’). Group II
markers originate in Proto-Germanic prepositions/adverbs, namely durch < PG ✶þurhw
‘through’ (apparently originally a PIE verbal compound of ✶terh2 ‘cross’ and ✶h3ekw
‘see’), über < PG ✶ubiri < PIE ✶upér(i) ‘over, above’, and um < PG ✶umbi < PIE ✶h2mbhi
‘around’ (apparently originally an instrumental of ✶h2énts ‘face’) (see Koonen 2013 for
the PG and PIE forms). Group III markers also originate in Proto-Germanic prepositions/
adverbs.
Applicative prefixes cannot co-occur, either with each other or with themselves.
The markers or their lookalikes can occur after some verbal particles, for instance,
in mit-be-kommen ‘notice’, an-be-halten ‘keep (sthg.) on’, auf-er-legen ‘impose’, and
an-er-kennen ‘acknowledge’.12 Verbs that have prefixes before the particles are actu-
ally denominal (e.g., be-an-standen ‘complain about’ < Anstand ‘objection’, be-auf-tra-
gen ‘mandate’ < Auftrag ‘assignment’, ver-an-lagen ‘assess’ < Anlage ‘investment,
asset’).
German applicative prefixes are not restricted by specific tense-aspect-mood values,
but some markers often work as denominal verbalizers, in which case many lexemes
appear as past participles only (e.g., beblümt ‘flowery, flowered’; †beblumen ‘flower
[v.]’ is obsolete). In other such cases, finite forms do exist, but nominalized ones are
much more frequently used (e.g., behausen ‘house [v.]’ vs. Behausung ‘abode, dwelling’).
Unlike their Slavic, Baltic, and Hungarian counterparts, the aspectual yield of German
applicative particles seems to be typically telicizing, rather than perfectivizing (see also
§§ 4.1 and 5.1).

3.2 Syntax

3.2.1 Be-applicatives

The syntax of the most common be-applicatives is schematically summarized in

Table 3.13

12 In most German varieties, an- is a particle and er- is a prefix in anerkennen. Nevertheless, in some
southwestern varieties (e.g., Swiss High German, Swiss German, and Liechtensteiner German), the string
aner- is treated as a complex prefix (http://mediawiki.ids-mannheim.de/VarGra/index.php/Anerkennen).
13 We are glossing over some uncommon patterns here, like the one found with lehren ‘teach (sthg.
to sbdy.)’ (which takes two accusative-marked objects) vs. belehren ‘instruct, teach (sbdy.)’ (which only
takes one), or like those where the oblique constituent in the AC takes the genitive instead of a preposi-
tion, as in (28).
430 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

Table 3: Syntax of be-applicatives.

Type Valency BC AC

A 1→2 sbj V [prep npi] sbj be-V dobji

B 1+ → 2 sbj V iobji [prep npj] sbj be-V dobji [prep npj]
C 2→2 sbj V dobji [prep npj] sbj be-V dobjj [prep npi]
D 3→2 sbj V iobji dobjj sbj be-V dobji [prep npj]

Cases where the AppP is semantically predictable show syntactic variation. The
most common instances are those where intransitives are transitivized with be-, either
as in (15) and in (18), where the non-agentive participant is a prepositional phrase in
the BC (Type A; this also corresponds to the default Group-II cases), or as in (19), where
it is an indirect object (Type B).

(18) German
a. Die Regierung kämpf-te gegen das Parlament.
art.sg.f.nom government fight-pst[3sg] against art.sg.n.acc parliament
b. Die Regierung be-kämpf-te das Parlament.
art.sg.f.nom government appl-fight-pst[3sg] art.sg.n.acc parliament
Both: ‘The government fought against parliament.’

(19) German
a. Ein-e Katastrophe droh-te diesen Wäldern.
art-sg.f.nom catastrophe threaten-pst[3sg] these.dat forests.dat
b. Ein-e Katastrophe be-droh-te diese Wälder.
art-sg.f.nom catastrophe appl-threaten-pst[3sg] these[acc] forests[acc]
Both: ‘A catastrophe threatened these forests.’

Cases broadly corresponding to the so-called locative-alternation are also numerous

and come in three subtypes. First, some (few) verbs like stopfen ‘stuff’ can only participate
in an uncoded alternation:

(20) German (Brinkmann 1995: 76)

a. Sie (✶be-)stopf-te-n Wachs in das Loch.
they.nom appl-stuff-pst-3pl wax into art.sg.n.acc hole
‘They stuffed wax into the hole.’
b. Sie (✶be-)stopf-te-n das Loch mit Wachs.
they.nom appl-stuff-pst-3pl art.sg.n.acc hole with wax
‘They stuffed the hole with wax.’

Some (also relatively few) verbs like laden ‘load’ participate in either an uncoded or a
coded alternation:
 14 Applicativizing preverbs in selected European languages 431

(21) German
a. Sie lud das Heu
she.nom load.pst[3sg] art.sg.n.acc hay
(auf den Wagen).
on/onto art.sg.m.acc wagon
‘She loaded the hay onto the wagon.’
b. Sie (be-)lud den Wagen
she.nom appl-load.pst[3sg] art.sg.m.acc wagon
(mit dem Heu).
with art.sg.n.dat hay
‘She loaded the wagon with the hay.’

Nevertheless, it is more common for locative verbs to participate only in a coded alter-
nation (Type C in Table 3); consider the following example with sprühen ‘spray’:

(22) German
a. Er sprüh-te Wasser auf die Pflanzen. (BC)
he.nom spray-pst[3sg] water on art.pl.acc plants
‘He sprayed water on the plants.’
b. Er be-sprüh-te die Pflanzen mit Wasser. (AC)
he.nom appl-spray-pst[3sg] art.pl.acc plants with water
‘He sprayed the plants with water.’

Cases where a three-argument clause is rearranged by applicativization are less

common but do exist; in (23), the BC has a direct and an indirect object, the AC has a
direct and a prepositional object, and the non-agentive referents are swapped vis-à-vis
the BC (Type D). Due to a special reading of the holism effect (viz. the “many effect”; see
§ 3.3), (23a) is comparatively neutral while (23b) means that probably many flowers
were given:

(23) German
a. Er schenk-te ihr diese Blumen. (BC)
he.nom gift-pst[3sg] she.dat these[acc] flowers
b. Er be-schenk-te sie mit diesen Blumen. (AC)
he.nom appl-gift-pst[3sg] she.acc with these.dat flowers
Both: ‘He gave her these flowers as a gift.’

Cases where variation is more considerable (and that often concern semantic unpre-
dictability) include those with transitive verbs in the unmarked construction (see § 3.4).
432 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

3.2.2 Er-applicatives

The syntax of er-applicatives is summarized in Table 4.

Table 4: Syntax of er-applicatives.

Type Valency BC AC
A 1→2 sbj V [prep npi] sbj (sich) er-V dobji
B 2→2 sbj V dobji [prep npj] sbj er-V dobjj [prep npi]
C 3→2 sbj V iobji dobjj sbj er-V dobji

Er- transitivizes intransitive kämpfen ‘fight’ (24) (Type A) but rearranges the clause
eliminating any indirect objects with ditransitive geben ‘give’ (25) (Type C):

(24) German
a. Die Regierung kämpf-te um den Sieg. (BC)
art.sg.f.nom government fight-pst[3sg] about art.sg.m.acc victory
‘The government fought for victory.’
b. Die Regierung er-kämpf-te den Sieg. (AC)
art.sg.f.nom government appl-fight-pst[3sg] art.sg.m.acc victory
‘The government eked out victory.’

(25) German
a. Sie gab dem Polizist-en keinen Beweis
she.nom give.pst art.sg.m.dat policeman-dat no.acc proof
seine-r Schuld.
his-gen guilt
‘She gave no proof of his guilt to the policeman.’
b. Die Untersuchung er-gab keinen Beweis seine-r Schuld.
art.sg.f.nom investigation er-give.pst no.acc proof his-gen guilt
‘The investigation found no evidence of his guilt.’

With fragen ‘ask’, however, the syntactic effect of er-prefixation corresponds to that of
locative-alternation be-verbs, that is, the direct object and the prepositional object swap
places (Type B):

(26) German
a. Sie frag-te ihn nach dem Weg. (BC)
she.nom ask-pst[3sg] he.acc after art.sg.m.dat way
‘She asked him about the way.’
 14 Applicativizing preverbs in selected European languages 433

b. Sie er-frag-te den Weg von ihm. (AC)

she.nom appl-ask-pst[3sg] art.sg.m.acc way of he.dat
‘She inquired (information about) the way of him.’

3.2.3 Ent-applicatives

Finally consider one possible effect of ent-prefixation. Here, the AppP is an indirect,
rather than a direct, object;14 it is relatively easy to find intransitive verb bases (like
steigen ‘climb’) that show such a pattern:

(27) German (Stiebels 1996: 110)

a. Sie steig-t aus dem Auto.
she.nom climb-3sg out art.sg.n.dat car
b. Sie ent-steig-t dem Auto.
she.nom appl-climb-3sg art.sg.n.dat car
Both: ‘She gets out of the car.’

We have not found any instances of this D-applicative with transitive bases; consider,
however, Examples (33)–(34) in Section 3.4, both based on transitive base verbs.

3.2.4 Summary

Most German applicatives are optional P-applicatives; ent-derivation yields optional

D-applicatives. Table 5 below summarizes the syntactic effects of be-, er-, and ent-.

Table 5: Selected syntactic effects of three German preverbs.

be- er- ent-

valency-increasing (1 → 2) (iobj)
valency-neutral (2 → 2)
valency-decreasing (3 → 2)

14 Note in passing that the prefix unter-, which applicativizes only occasionally, can also introduce AppPs
as indirect objects / D’s (e.g., liegen ‘lie’ → unterliegen ‘be defeated by, be subject to’, stehen ‘stand’ → unter-
stehen ‘be subordinate to’).
434 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

3.3 Semantics

Cross-linguistically common applied objects are Instruments and Comitatives on the

one hand, and Beneficiaries and Locations on the other. In German ACs, only the latter
group is found, and the picture is much more complicated than what such a simple
semantic-role-based account can provide.
Semantically, the applied object found with Group-II prefixes and Group-III par-
ticles is usually transparently related to whatever spatial meaning the homophonous
prepositions or particles have. Applied objects of verbs with Group-I prefixes show a
more varied picture—except ent-, which has a separative meaning. Depending on the
verb it applicativizes, er- introduces notions like killing, as in schießen (auf) ‘shoot (at)’
vs. erschießen ‘shoot dead’, or intended possession, as in arbeiten ‘work’ vs. erarbeiten
‘work (sthg.) out’. Similarly, ver- conveys resultativity, usually with some intensification
(urteilen ‘deliver judgment’ vs. verurteilen ‘convict, condemn’), or damage/destruction
(pinkeln ‘pee’ vs. verpinkeln ‘ruin by peeing’). Zer- typically also denotes, among other
things, damage/destruction (treten ‘tread’ vs. zertreten ‘crush underfoot’).
Applied objects found with German be-verbs bear a fairly complex semantic rela-
tionship to their predicate. With locative base verbs—i.e., those that take Goal direct
objects—the be-applicative corresponds to auf ‘on’ or an ‘at’, as in werfen (auf) ‘throw
(on)’ vs. bewerfen ‘throw on/to’; with intransitive base verbs in particular, it can also
correspond to in ‘in’ (cf. treten ‘tread’ vs. betreten ‘enter’). Notably, while such verbs
denote a variety of spatial relationships in base clauses, depending on the specific
adposition used and on lexical context, be-verbs denote relationships of contact with
the applied phrase and usually exclude the interior of entities in applicative clauses,
that is, the so-called “topological restriction” between Figure and Ground (Brinkmann
1995: 78–82). Some, but not all, locative verbs have be-forms; Brinkmann also points
out (p. 86) that Goals must not be containers and subjects must be Agents in order for
applicative clauses to be admissible.15
Moreover, Brinkmann (1995: 84–86) distinguishes six semantic classes of be-verbs
in addition to the locative class just outlined, which are heterogeneous as to the seman-
tic role of the non-agentive participant in the base clause. Four of these classes are
rather transparent and systematic: material manipulation (e.g., arbeiten [an] ‘work
[on]’ vs. bearbeiten ‘work on, edit’), active perception (e.g., tasten [nach] ‘feel [for]’ vs.
betasten ‘examine by touching repeatedly’), emotion (e.g., staunen [über] ‘be amazed
[at]’ vs. bestaunen ‘contemplate in amazement’), and speech (e.g., sprechen [über] ‘talk
[about]’ vs. besprechen ‘talk about, discuss’). The other two classes are more idiosyn-
cratic: Recipient/Beneficiary-verbs include not only those like schenken ‘present, gift’

15 Chapter 6 of Brinkmann’s study deals with the details of the restrictions on which verbs participate
in the so-called locative alternation in German, which include, but are not limited to, the topological
restriction. See Stiebels (1996) for details on non-be-applicatives in German and their semantic charac-
teristics.
 14 Applicativizing preverbs in selected European languages 435

and erben ‘inherit’ but also those like kochen ‘cook’ (bekochen ranges from simple ‘cook
for’ to ‘provide/support via cooking’); the last class includes some privative verbs (e.g.,
rauben ‘steal’ vs. berauben ‘rob, steal from’). Example (28) illustrates a pair with these
latter verbs:

(28) German
a. Sie raub-te ihm sein-e Ersparnisse. (BC)
she.nom steal-pst [3sg] he.dat his-pl.acc savings
‘She stole his life savings from him.’
b. Sie be-raub-te ihn (sein-er Ersparnisse). (AC)
she.nom appl-steal-pst[3sg] he.acc his-pl.gen savings
‘She robbed him (of his life savings).’

Besides this considerable semantic range, the topological restriction with locative
verbs mentioned above shows an important but difficult-to-capture semantic parallel
with the other verb classes. Brinkmann states that “these verbs usually imply a holis-
tic interpretation of the goal, and the speaker can express the theme in an optional
with-phrase” (1995: 86). Nevertheless, Michaelis and Ruppenhofer (2001) question some
fundamental tenets of Brinkmann’s account and propose semantic constraints that
originate differently, that is, not in general principles and the verbal input, but in the
be-prefixation rule itself. This study advances an analysis in terms of a prototype and
several extensions thereof (which also receives support from the historical evidence, pp.
89–92). With the prototype, the base object is a Theme that covers a surface expressed
as the applied object (schmieren vs. beschmieren, both ‘smear’); with the extensions, this
Theme-covers-surface schema is extended and superimposed to the domains of per-
ception (riechen vs. beriechen, both ‘smell’), communication and discourse (sprechen
‘speak’ vs. besprechen ‘discuss’), and the others of Brinkmann’s non-locative classes.
Examples (29)–(31) illustrate the case frames of (be)schmieren, (be)riechen, and (be)
sprechen, respectively:

(29) German
a. Er schmier-te Butter auf-s Brot. (BC)
he.nom smear-pst[3sg] butter on-art.sg.n.acc bread
‘He smeared butter on the bread.’
b. Er be-schmier-te das Brot mit Butter. (AC)
he.nom appl-smear-pst[3sg] art.sg.n.acc bread with butter
‘He smeared the bread with butter.’

(30) German
a. Der Hund roch an mir. (BC)
art.sg.m.nom dog smell.pst on 1sg.dat
436 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

b. Der Hund be-roch mich. (AC)

art.sg.m.nom dog appl-smell.pst 1sg.acc
Both: ‘The dog sniffed at me.’16

(31) German
a. Wir sprach-en über die Lokalpolitik. (BC)
we.nom speak.pst-1pl about art.sg.f.acc local.politics
‘We spoke about local politics.’
b. Wir be-sprach-en die Lokalpolitik. (AC)
we.nom appl-speak.pst-1pl art.sg.f.acc local.politics
‘We discussed local politics.’

Note that, with extensions, the notion of coverage (i.e., the “holism effect”) typically
turns into saturation and affectedness (wohnen ‘dwell’ vs. bewohnen ‘inhabit’) or repe-
tition and intensification (siegen ‘win’ vs. besiegen ‘defeat’).17

3.4 Lookalikes

Regarding syntactic lookalikes, German differs markedly from English in that labile
verbs like laden ‘load’ (see Example [21] above for the relevant case frames) and stopfen
‘stuff’ do exist but constitute a small class, which makes polyvalency alternations like
the ones illustrated in (20)–(21) above certainly notable precisely because they are
not the typical case. German does have regular and frequent uncoded alternations,
however; as in Slavic-Baltic and Hungarian, dative-marked NPs can be accommodated
in clauses headed by identical predicates quite flexibly, in order to express a wide
variety of extra-thematic participants, including Beneficiaries/Maleficiaries, Viewpoint
Holders, etc. (cf. §§ 4.4 and 5.4).18
Regarding morphological lookalikes, German shows a plethora of important phe-
nomena. First, applicativa tantum are numerous, especially with denominal and deadjec-
tival verbs built with Group-I prefixes, for instance, verstauben ‘get dusty, dust (tr.)’ (from
Staub ‘dust [n.]’) and berichtigen ‘correct (v.), rectify’ (from richtig ‘correct [adj.], right’).

16 The underived labile verb riechen ‘smell’ is unusual in that it can be A- or P-labile; it also allows a
direct object (der Hund roch mich ‘the dog smelled me’ contrasts with (30b) only regarding semantics).
17 Along similar lines, Dewell (2015) proposes a Construction-Grammar account of German preverbs in
general. In his view, German applied direct objects denote a “route-path” specified by the corresponding
adposition (if any), either concrete (with locative verbs) or abstract (with other verbs). While applicative
particles would then lead to a sequential reading of that route-path, paying attention to its successive lo-
cations, applicative prefixes would lead to a synoptic/holistic reading, focusing on a whole stable setting
whose part that route-path is (Dewell 2015: 313).
18 See Cysouw (2023: 98–100) and references therein for an overview of such clauses.
 14 Applicativizing preverbs in selected European languages 437

Second, not only valency-neutrality but also broad syntax-neutrality with prefixes
from Group I is also quite common. Consider, for instance, fragen ‘ask’ vs. befragen ‘ques-
tion, interview’ in (32). The underived verb can occur with or without a direct object
(32a); the derived verb normally occurs with a direct object corresponding to the same
referent, but the main effect of prefixation is fundamentally semantic, not syntactic
(32b)—not unlike what occurs with some Bantu applicative lookalikes (see Pacchiarotti,
this volume):

(32) German
a. Er frag-te (sie) nach dem Weg.
he.nom ask-pst[3sg] she.acc after art.sg.m.dat way
‘He asked (her) about the way.’
b. Er be-frag-te sie nach dem Weg.
he.nom intens-ask-pst she.acc after art.sg.m.dat way
‘He asked her about the way.’

A particularly productive pattern consists in ver-verbs that take the middle marker
sich;19 such expressions denote a faulty action (laufen ‘walk’ vs. sich verlaufen ‘get lost’).
Lastly, there are many cases where be-, er-, ver-, and zer- have a purely semantic yield
(occasionally idiosyncratic and often also aspectual, i.e. broadly telicizing; e.g., schließen
‘shut’ vs. erschließen ‘deduce, unlock, open up’), as well as all instances where ent- is
simply a reversive prefix (decken ‘cover’ vs. entdecken ‘discover’).
Ent-verbs based on transitive predicates are not applicatives. Consider the follow-
ing example with ziehen ‘draw, pull’—here in a coded alternation of the equipollent
marking subtype (viz. wegziehen, with the particle weg, vs. entziehen) rather than bona
fide applicativization:

(33) German
a. Sie zog ihr-e Hand von ihm weg.
she.nom pull.pst[3sg] her-sg.f.acc hand of he.dat away
b. Sie ent-zog ihm ihr-e Hand.
she.nom ent-pull.pst[3sg] he.dat her-sg.f.acc hand
Both: ‘She withdrew her hand from him.’

Other “transitive” pairs like erben ‘inherit’ vs. enterben ‘disinherit’ show a direct object
in both clauses, but that object’s semantic role is altered (as in the English equivalents),
and the oblique participant disappears from the clause headed by the derived verb:

19 An erstwhile accusative-marked reflexive, sich is nowadays used in reflexive (sich fragen ‘ask one-
self’), reciprocal (sich treffen ‘meet’), anticausative (sich öffnen ‘open’), potential passive (sich lesen ‘[can]
be read’), and other functions.
438 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(34) German
a. Er erb-te ein Haus (von sein-em Großvater).
he.nom inherit-pst[3sg] art.sg.n.acc house of his-sg.m.dat grandfather
‘He inherited a house (from his grandfather).’
b. Sein Großvater ent-erb-te ihn.
his[sg.m.nom] grandfather ent-inherit-pst[3sg] he.acc
‘His grandfather disinherited him.’

Finally, there is at least one instance of the causative-applicative polysemy common

in other languages (Zúñiga and Kittilä 2019: Ch. 8.2.1; Bahrt 2021: Ch. 4.3.1): leben ‘live’
vs. beleben ‘quicken, stimulate, enliven’.

4 Slavic and Baltic

4.1 Morphology

There are about twenty verbal prefixes (preverbs) in Slavic languages and about a dozen
in Baltic. None of them has applicativization as its only or primary function, their ability
to modify the base verb’s argument structure and introduce arguments being a con-
sequence of their origin as spatial modifiers and further development into aktionsart
markers and perfectivizers. (For a general overview of Slavic, see Oertle 2016, Janda
2020 and references therein; for Baltic, the most comprehensive overview to date is
still Endzelīns 1971[1907].) Most prefixes in both branches have an obvious formal and
functional relation to prepositions and/or spatial adverbs (see again Oertle 2016 for a
general and comparative overview of Slavic and Petit 2011 for Baltic) and both originate
from adverbial or nominal roots with locational meanings, often with cognates across
Indo-European. Table 6 lists the most common preverbs of both branches with their
basic meanings.

Table 6: Common preverbs in Slavic and Baltic.

Slavic (represented by Russian) Baltic

do- ‘until’; to completion Latg. + Lith., Latv. dialectal da-20

iz- ‘out of’ Lith. iš-, Latv. iz- ‘out of’
na- ‘on top’; large quantity Lith. nu-, Latv. no- ‘from top’
o(b)- ‘around’ ap- ‘around’; partly
ot- ‘away’ at- ‘towards’; in response

20 Possibly a borrowing from Slavic.

 14 Applicativizing preverbs in selected European languages 439

Table 6 (continued)

Slavic (represented by Russian) Baltic

Lith. par- ‘home’, to the ground, Latv. pār- ‘across, home’

pre-/pere- ‘across’; excess, repetition, distributivity Lith. per- ‘across’; excess, repetition
po- surface; temporal boundedness pa- ‘under’; temporal boundedness
pod- ‘under’
pri- ‘near’ Lith. pri-, Latv. pie- ‘near’
pro- ‘through’ Lith. pra- ‘through’
raz- ‘apart’
s- ‘together’; from a surface Lith. su-, Latv. sa- ‘together’
u- ‘away’
v- ‘into’ Lith. į-, Latv. ie- ‘into’
vy- ‘out’
v(o)z- ‘up’ Lith. už- ‘behind; up’, Latv. uz- ‘up’
za- ‘behind’; inceptive Latv. aiz- ‘behind’

Although most of the Baltic and Slavic preverbs can function as applicatives, at
least occasionally, only for a subset of them are the applicative uses prominent and
productive. Those are, in particular, Slavic iz-, Baltic iš-/iz-, Slavic vy-, Slavic o(b)-, Baltic
ap-, Slavic pre-/pere-, Baltic per-, Baltic pri-/pie-, Slavic pro-, Lithuanian pra- and Slavic
za-, Baltic už-/uz-. (Semantically, these correspond to the Group-II applicativizing pre-
fixes in German and to some of the applicativizing particles in Hungarian; see §§ 3.1
and 5.1, respectively.) Each of these preverbs is fairly polysemous and introduces argu-
ments with a variety of semantic roles. Besides that, there is a number of more special-
ized applicative uses of preverbs occurring in individual languages or sub-branches
(e.g., nad- ‘surpass’ in South Slavic, see § 4.3).
Preverbs do not show much allomorphy and most of it is due to (morpho)phonolog-
ically conditioned sandhi. The general rule (apart from some lexicalized exceptions) is
that prefixed verbs inflect exactly like their simplex counterparts. A notable exception
to this consists in the reflexive verbs in Lithuanian, which attach the reflexive affix
at the right edge of the word when unprefixed (e.g., juokti-s ‘laugh’) and immediately
before the root when prefixed (e.g., pra-si-juokti ‘burst into laughter’); this rule is sen-
sitive to the presence of any prefix regardless of its function. Another and more impor-
tant morphological complication is due to the fact that prefixes typically perfectivize
verbs, which affects the range of contexts they occur in and partly also their paradigms
(on this, see e.g., Wiemer and Seržant 2017 and references therein). Thus, in North
Slavic, imperfective verbs (including the vast majority of simplex verbs) have syn-
thetic present and past tenses and a periphrastic future, while perfective verbs (mostly
formed via prefixation) have a synthetic future (formally identical to the present) and
past tenses and do not form periphrastic futures, see Table 7.
440 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

Table 7: Tense paradigms of simplex (imperfective) and prefixed (perfective) verbs in Polish
(Swan 2002: 270).

Imperfective Perfective

Past pisałem ‘I was writing, I wrote’ napisałem ‘I wrote (to the end)’
Present piszę ‘I am writing, I write’ —
Future będę pisał ‘I will write, I will be writing’ napiszę ‘I will write (to the end)’

In South Slavic and Baltic, the paradigms of simplex and prefixed verbs are more
symmetrical. Besides that, since prefixed perfective verbs are not allowed in durative/
progressive contexts, and, at least in the eastern Slavic languages, in most habitual con-
texts as well, in order to express the lexical content of a prefixed verb in imperfective
contexts, the so-called secondary imperfective is usually productively derived by suffix-
ation, see (35) and (36).21

(35) Russian
rabotat’ ‘work’ (ipfv) → za-rabotat’ ‘earn’ (pfv) → za-rabat-yva-t’ ‘earn’ (ipfv)

(36) Lithuanian
bėgti ‘run’ (ipfv) → per-bėgti ‘run across’ (pfv) → per-bėg-inė-ti ‘run across’ (ipfv)

4.2 Syntax

Preverbs in Slavic and Baltic mainly function as either X-applicatives or P-applicatives;

D-applicatives are rare. X-applicatives involve the spatial meanings of preverbs in com-
bination with base verbs not denoting displacement. Such preverbs introduce argu-
ments expressing the landmark of real or metaphorical motion, which are normally
excluded with simplex verbs, see (37) and (38).

(37) Lithuanian (CCL)

puslap-į į Latvij-os oper-os istorij-ą
page-sg.acc in Latvia-sg.gen opera-sg.gen history-sg.acc
į-raši-au.
appl-write.pst-1sg
‘I added (lit. wrote-in) a page into the history of Latvian opera.’

21 On the multiple functions of Slavic aspects and their inter-Slavic variation, see, for instance, Dickey
(2000) and Fortuin and Kamphuis (2015). For a broader picture including Baltic, see Arkadiev (2014,
2015).
 14 Applicativizing preverbs in selected European languages 441

(38) Russian (Biskup 2017: 21)

a. On mёrz (✶v ajsberg).
3[sg.m.nom] freeze[pst.sg.m] into iceberg
‘He was cold (✶in an iceberg).’
b. On v-mёrz v ajsberg.
3[sg.m.nom] appl-freeze[pst.sg.m] in iceberg
‘He froze into an iceberg.’

A case of X-applicative with a non-spatial meaning is the Slavic construction con-

sisting of the preverb do- and the reflexive/middle marker (which is a suffix in East
Slavic and an enclitic elsewhere), one of whose meanings is the acquisition of an object
or property (often metaphorical, see Oertle 2016: 143–144). In Russian the object is
usually expressed in a prepositional phrase (39), but in Polish and some other languages
(see, e.g., Richardson 2007: 81–83) the bare genitive is possible (40) (accusative is ruled
out for reflexive verbs).

(39) Russian (RNC)

Ja ne mog-u do n-eё do-kriča-t’-sja.
1sg.nom neg can-prs.1sg till 3-sg.f.gen appl-shout-inf-rfl
‘I can’t get her attention by shouting (because she cannot hear me).’

(40) Polish (Przybylska 2006: 62)

Do-służy-ł=się stopni-a pułkownik-a.
appl-serve-pst[sg.m]=rfl rank-sg.gen colonel-sg.gen
‘He rose (lit. served) to the rank of colonel.’

P-applicatives predominantly show various non-spatial meanings (to be discussed

in greater detail in § 4.3). With intransitive base verbs, such preverbs can either add a
new argument expressed as a direct object, (41)–(42), or promote an indirect or oblique
object to direct object status, (43)–(44).

(41) Bulgarian
a. Măž-ăt rabot-i.
man[sg]-def.sg.m work-prs.3sg
‘The man works.’
b. Njama măž, koj-to da ja nad-rabot-i.
neg.exist man[sg] what-rel sbjv 3sg.f.acc appl-work-prs.3
‘No man can overwork her.’22

22 Source: http://old.segabg.com/article.php?id=221116, accessed 7 November 2021.

442 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(42) Lithuanian
a. Aš aug-au.
1sg.nom grow-pst.1sg
‘I was growing.’
b. Aš iš-aug-au uniform-os keln-es.
1sg.nom appl-grow-pst.1sg uniform-sg.gen trousers-pl.acc
‘I outgrew the uniform’s trousers.’ (CCL)

(43) Czech (CNC)

a. Vlád-l-i <. . .> svět-ov-ým moř-ím.
rule-pst-pl.m world-adjz-pl.dat sea-pl.dat
‘They ruled . . . in the world’s seas.’
b. Snaž-ím se o-vlád.nou-t svoj-e vzrušen-í.
try-prs.1sg rfl.acc appl-rule-inf rfl.poss-sg.n.acc excitement-sg.acc
‘I am trying to take control over my excitement.’

(44) Lithuanian (CCL)

a. Labai daug galvoj-au apie jūs-ų žodži-us.
very much think-pst.1sg about 2pl-gen word-pl.acc
‘I thought about your words quite a lot.’
b. Ap-galvoj-au j-o žodži-us.
appl-think-pst.1sg 3-sg.m.gen word-pl.acc
‘I thought his words over.’

Applicatives can also attach to zero-place verbs, at least marginally (45).

(45) Lithuanian
a. Jau tem-o.
already grow_dark-pst.3
‘It already was growing dark.’ (CCL)
b. Mus ap-tem-o toli nuo nam-ų.
1pl.acc appl-grow_dark-pst.3 far from house-pl.gen
‘It grew dark over us when we were far from home.’ (Kozhanov 2015: 109)

With transitive base verbs, applicative preverbs may either add a new P-argument
with a concomitant demotion (46) or elimination (47) of the original P, or rearrange
the arguments promoting an oblique object of the BC to direct object in the AC and
demoting the original P to an oblique (48).
 14 Applicativizing preverbs in selected European languages 443

(46) Lithuanian (CCL)

a. J-ie sėj-o griki-us.
3-sg.m.nom sow-pst.3 buckwheat-pl.acc
‘They sowed buckwheat.’
b. Pernai aštuonet-ą hektar-ų griki-ais už-sėj-o.
last_year eight-sg.acc hectare-pl.gen buckwheat-pl.ins appl-sow-pst.3
‘Last year, they sowed eight hectares with buckwheat.’

(47) Russian (RNC)

a. On-a u-kra-l-a23 u menj-a
3-sg.f.nom pvb-steal-pst-sg.f at 1sg-gen
zolot-uju cep-očk-u!
golden-sg.f.acc chain-dim-sg.acc
‘She stole a golden necklace from me!’
b. Tjuring-a obo-kra-l odin iz
T.-sg.acc appl-steal-pst[sg.m] one[sg.m.nom] from
j-ego ljubovnik-ov
3-sg.m.gen lover-pl.gen
‘Turing was robbed by one of his lovers.’

(48) Russian (RNC)

a. Ego žen-a dari-l-a mam-e
3.sg.m.gen wife-sg.nom give_as_present-pst-sg.f mother-sg.dat
plat’j-a i tufl-i.
dress-pl.acc and shoe-pl.acc
‘His wife used to give (my) mother dresses and shoes as presents.’
b. Otec ščedro o-dari-l
father[sg.nom] generously appl-give_as_present-pst[sg.m]
ix zolot-om, serebr-om i drug-imi dar-ami.
3.pl.acc gold-sg.ins silver-sg.ins and other-pl.ins gift-pl.ins
‘Their father had given them many gifts of silver and gold and articles of
value.’24

Some of the preverbs in their spatial meanings can work alternatively as P-applic-
atives and as markers of oblique registration leaving the coding of the landmark intact,
often with subtle differences in meaning (see § 4.3). This is especially characteristic of
Slavic pre-/pere-, Lithuanian per-, Latvian pār- ‘across, through’ (49), Slavic pro-/pre-, Baltic

23 The preverb u- has a purely perfectivizing function here and does not affect argument structure.
24 Source: Old Testament, 2 Chronicles 21:3, English version quoted after https://www.biblegateway.
com/passage/?search=2%20Chronicles+21&version=NIV, accessed 19 February 2022.
444 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

pra- ‘passing by, through’ and Slavic o(b)-, Baltic ap(i)- ‘around’, but other preverbs may
behave in this way in individual languages as well, cf. Macedonian nad- ‘over’ (50),
Czech pod- ‘under’ (Oertle 2016: 57) or Lithuanian pri- ‘approaching’ (Kozhanov 2016:
372–374).

(49) Lithuanian (CCL)

a. J-i per-ėj-o gatv-ę.
3-sg.f.nom appl-go-pst.3 street-sg.acc
‘She crossed the street.’
b. J-is per-ėj-o per gatv-ę.
3-sg.m.nom oblreg-go-pst.3 over street-sg.acc
‘He crossed (lit. went across) the street.’

(50) Macedonian (Mitkovska and Bužarovska 2012: 138)

a. Eden helikopter ja=nad-let-a zgrad-a-t-a.
one[sg.m] helicopter[sg] 3sg.f.acc=appl-fly-aor.3sg building-sg-def-sg.f
b. Eden helikopter nad-let-a nad zgrad-a-t-a.
one[sg.m] helicopter[sg] oblreg-fly-aor.3sg over building-sg-def-sg.f
Both: ‘A helicopter flew over the building.’

When combined with transitive verbs of displacement, the aforementioned pre-

verbs normally do not affect the encoding of the base P-argument (51a–b). An exception
to this rule consists in metaphorical uses of the preverb ob- ‘around’ such as shown in
(51c), where the resulting verb denotes the trajectory of the person’s gaze over some
object or spatial region encoded as a direct object, the abstract entity undergoing meta-
phorical motion (‘eyes’, ‘gaze’) being expressed as an instrument.

(51) Russian (RNC)

a. Rodžer vё-l nas po svo-im polj-am.
R.[sg.nom] lead-pst[sg.m] 1pl.acc along rfl.poss-pl.dat field-pl.dat
‘Roger led us through his fields.’
b. Potom on ob-vё-l nas vokrug zamk-a.
afterwards 3[sg.m.nom] appl-lead-pst[sg.m] 1pl.acc around castle-sg.gen
‘Afterwards he showed us around the castle.’
c. Ver-a ob-ve-l-a glaz-ami komnat-u.
V.-sg.nom appl-lead-pst-sg.f eye-pl.ins room-sg.acc
‘Vera looked around the room.’

The same pattern is observed with non-motion base verbs, where the landmark is also
promoted to direct object (52).
 14 Applicativizing preverbs in selected European languages 445

(52) Russian
a. Ja po-sadi-l25 derev’j-a vokrug prud-a.
1sg.nom pvb-plant-pst[sg.m] tree-pl.acc around pond-sg.gen
‘I planted trees around the pond.’
b. Ja ob-sadi-l prud derev’j-ami.
1sg.nom appl-plant-pst[sg.m] pond[sg.acc] tree-pl.ins
‘I surrounded the pond with trees.’

As already mentioned, D-applicatives are rare. They are more frequently attested
in Latvian, where, for instance, the landmark introduced by the prefix pie- ‘approach-
ing’ can be marked either by the corresponding preposition or by the dative (53), but not
by the accusative (Holvoet and Nau 2016: 24).

(53) Latvian (based on Holvoet and Nau 2016: 24–25)

Viņ-š pie-gāj-a pie plaukt-a / plaukt-am.
3-sg.m.nom appl-go.pst-3 to shelf-sg.gen shelf-sg.dat
‘He walked to the shelf.’

Likewise, the landmark of motion verbs prefixed with ap- ‘around’ in Latvian can
be expressed not only by the prepositional phrase and by the bare accusative, but also
by the dative, with no difference in meaning (54).

(54) Latvian (Holvoet and Nau 2016: 24)

Viņ-a ap-gāj-a ap māj-u / māj-u / māj-ai.
3-sg.f.nom appl-go.pst-3 around house-sg.acc house-sg.acc house-sg.dat
‘She walked around the house.’

In Slavic, the rare D-applicatives are exemplified by the comitative use of the East
Slavic preverb pod- and West and South Slavic preverb pri-, attested with a very limited
class of verbs denoting sound emission (Oertle 2016: 236, 267–268), cf. parallel examples
from Russian and Bulgarian in (55).

(55) Russian / Bulgarian (RNC parallel corpus)

a. Pod-pe-va-ja orkestr-u, Varj-a tixon’ko
appl-sing-ipfv-cvb orchestra-sg.dat V.-sg.nom quietly
za-tjanu-l-a. . .
pvb-draw-pst-sg.f

25 The preverb po- is purely perfectivizing here.

446 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

b. Pri-glas-ja-jki na orkestăr-a, Varja tixičko

appl-sing-ipfv-cvb to orchestra-def.m.obl V. quietly
za-pri-pjav-a. . .
pvb-pvb-sing-aor.3sg
Both: ‘Joining the orchestra, Varja quietly started singing. . .’

In most cases the AppP introduced by the P-applicatives shows all the syntactic
properties of a regular direct object: the ability to be promoted to subject in passives
(56), conversion of the accusative to the genitive under negation in East Slavic, Polish,
Slovene and Lithuanian (on this see Pirnat 2015, Arkadiev and Kozhanov forthcoming)
(57), and cross-referencing by bound pronominals (known as “clitic doubling”, Kalluli
and Tasmowski eds. 2008) in Bulgarian and Macedonian, see (50b) above.

(56) Russian (RNC)

a. My pere-š-l-i most.
1pl.nom appl-go-pst-pl bridge[sg.acc]
‘We crossed (lit. over-went) the bridge.’
b. Most pere-jd-ёn.
bridge[sg.nom] appl-go-pst.pp[sg.m.nom]
‘The bridge has been crossed.’

(57) Lithuanian
a. Už-dirb-au t-uos pinig-us.
appl-work-pst.1sg dem-pl.m.acc money-pl.acc
‘I have earned that money.’
b. <. . .> men-o žmon-ės dideli-ų pinig-ėli-ų
art-sg.gen people-pl.nom big-pl.gen money-dim-pl.gen
ne-už-dirb-a.
neg-appl-work-prs.3
‘artists do not earn much money.’ (CCL)

Things are more complicated in the case of P-applicatives introducing phrases that
express distance and temporal duration (the so-called perdurative), mainly Slavic pro-/
pre- and Lithuanian pra-, see, for instance, Letučij (2012: 133–136) on Russian, Žaucer
(2009: 146–164, 2012) on Slovene and Kozhanov (2016: 376–380, 382–385) on Lithua-
nian. First, distance and temporal phrases in the bare accusative case freely combine
with simplex verbs (58a/59a), so the prefixes simply make these optional adjuncts oblig-
atory (58b/59b).
 14 Applicativizing preverbs in selected European languages 447

(58) Russian
a. Im priš-l-o-s’ beža-t’
3.pl.dat get.to-pst-sg.n-rfl run-inf
(cel-yj kilometr).
whole-sg.m.acc kilometer[sg.acc]
‘They had to run (a whole kilometer).’
b. Im priš-l-o-s’ pro-beža-t’
3.pl.dat get.to-pst-sg.n-rfl appl-run-inf
✶
(cel-yj kilometr).
whole-sg.m.acc kilometer[sg.acc]
‘They had to cover a whole kilometer running.’ (RNC)

(59) Lithuanian
a. Gyven-au Vilni-uje (dvidešimt dvej-us met-us).
live-pst.1sg Vilnius-sg.loc twenty two-pl.m.acc year-pl.acc
‘I lived in Vilnius (for twenty two years).’
b. ✶(Dvidešimt dvej-us met-us) pra-gyven-au Vilni-uje.
twenty two-pl.m.acc year-pl.acc appl-live-pst.1sg Vilnius-sg.loc
‘I have been living in Vilnius for twenty two years.’ (CCL)

Second, such applicatives freely attach to transitive base verbs, which retain their
original direct objects (60a/61a), as well as to intransitive reflexive verbs, which retain
their reflexive marking thus remaining intransitive (60b/61b).

(60) Russian (RNC)

a. A u nas čelovek možet
pro-nosi-t’ odn-u par-u vs-e
appl-wear-inf one-sg.f.acc pair-sg.acc all-pl.acc
četyr-e sezon-a.
four-pl.acc season-sg.gen
‘At our place one can wear a single pair [of shoes] during all four seasons.’
b. Ja tam pro-uči-l-a-s’ poltor-a god-a.
1sg.nom there appl-study-pst-sg.f-rfl one_and_a_half-acc.m year-sg.gen
‘I studied there for one and a half years.’

(61) Lithuanian
a. Srov-ė . . . nu-neš-ė j-į žemyn
current-sg.nom appl-carry-pst.3 3-sg.m.acc down
kel-is šimt-us metr-ų.
several-pl.m.acc hundred-pl.acc meter-pl.gen
‘The current . . . carried him down for several hundred meters.’ (CCL)
448 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

b. Kamuol-ys nu-si-rit-o du metr-us.

ball-sg.nom appl-rfl-roll-pst.3 two[m.acc] meter-pl.acc
‘A ball rolled down for two meters.’ (Kozhanov 2016: 378)

However, such distance and temporal phrases can undergo the genitive of negation
rule (62) and are at least marginally passivizable (63), testifying to their hybrid status
between obligatory adjuncts and full-fledged direct objects.

(62) Slovene (Žaucer 2012: 341)

Juš v ječ-i ni pre-sede-l tr-eh let.
J.[sg.nom] in jail-sg.loc neg appl-sit-pst[sg.m] three-gen year[pl.gen]
‘Juš did not spend three years in jail.’

(63) Lithuanian (CCL)

po pra-šok-t-os nakt-ies
after appl-dance-pst.pp-sg.f.gen night-sg.gen
‘after a night of dancing (lit. danced-through night)’

P-applicativization of transitive verbs, apart from the distance and temporal cases
just discussed, results in either the demotion of the original P to peripheral status or
its complete elimination, see (46)–(49) above. The fate of the P of the BC apparently
depends on the semantics of the AC, for instance, on whether the original argument can
be construed as an instrument, means or spatial landmark of the new event. However,
even in some cases where such construal would be possible, the original P still cannot
be expressed in the AC, being generic and backgrounded (64).

(64) Lithuanian
a. J-i skalbi-a drabuži-us.
3-sg.f.nom wash-prs.3 clothes-pl.acc
‘She washes clothes.’
b. J-i ap-skalbi-a savo vyr-ą (✶drabuži-ais).
3-sg.f.nom appl-wash-prs.3 rfl.poss husband-sg.acc clothes-pl.ins
‘She washes (clothes and stuff) for her husband.’ (CCL)

4.3 Semantics

Most X-applicatives introduce landmarks of the spatial relations roughly outlined in

Table 6 above. Some of the spatial preverbs are Goal-oriented, for instance: Slavic v-
‘into’, na- ‘on’, pod- ‘under; close to’, vz- ‘up’, za- ‘behind; up’, or Lithuanian į- ‘into’, už-
‘behind’. Others are Source-oriented, for instance: Slavic vy- ‘out’, ot- ‘from’, u- ‘from’,
s-/z- ‘from above, off’, or Lithuanian iš- ‘out’, nu- ‘from above, off’. Interestingly, in
 14 Applicativizing preverbs in selected European languages 449

accordance with the so-called “Goal bias” (e.g., Stefanowitsch and Rohde 2004), verbs
with Source-oriented preverbs can combine with expressions denoting Goals, see (65).

(65) Russian
a. vy-pisa-t’ slov-o iz slovarj-a
appl-write-inf word-sg.acc out dictionary-sg.gen
‘copy (lit. out-write) a word from a dictionary’
b. vy-pisa-t’ slov-o v tetrad’
appl-write-inf word-sg.acc in notebook[sg.acc]
‘copy (lit. out-write) a word into a notebook’

Those preverbs that can introduce the landmark both as a prepositional phrase
and as an accusative object in their spatial functions deserve a special discussion.
As mentioned above, the most common of these are the Slavic pre-/pere-, Lithuanian
per-, Latvian pār- ‘across, through’, Slavic pro-/pre-, Baltic pra- ‘passing by, through’
and Slavic o(b)-, Baltic ap(i)- ‘around’. While with the ‘across’-prefixes the two types
of encoding seem to be synonymous, see (50) above, the situation with the other two
prefixes is more intricate. With a prepositional complement, the verbs with ob-/ap- and
pro-/pra- express the purely spatial meaning of, respectively, motion around (66a) or
past (67a) the landmark, while with an accusative complement, in addition to the same
meaning (66b) and (67b), these verbs tend to denote motion covering the whole area of
the landmark (66c) or passing through it (67c).26

(66) Russian (RNC)

a. My medlenno obo-š-l-i vokrug dom-a.
1pl.nom slowly pvb-go-pst-pl around house-sg.gen
‘We slowly went around the house.’
b. My obo-š-l-i dom <. . .>
1pl.nom appl-go-pst-pl house[sg.acc]
v nёm okazalas’ eščё odna dver’.
‘We went around the house . . . and found another door [at the other side].’
c. My obo-š-l-i dom.
1pl.nom appl-go-pst-pl house[sg.acc]
Xozjain pokazal nam svoi masterskie.
‘We passed through the whole house. The landlord showed us his workshops.’

26 Cf. observations about the Croatian o(b)- in Šarić and Mikolić (2015: 260–261); on Latgalian ap-, cf.
Svilans-Dennis (1982: 43–45).
450 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(67) Lithuanian (CCL)

a. Mači-au, kaip j-is pra-važiav-o pro mano nam-us.
see.pst-1sg how 3-sg.m.nom pvb-drive-pst.3 by my house-pl.acc
‘I saw him drive past my house.’
b. Stot-į <. . .> pra-važiav-o <. . .> karin-is traukin-ys.
station-sg.acc appl-drive-pst.3 military-sg.m.nom train-sg.nom
‘A military train passed the station.’
c. Be sustojim-o pra-važiav-o-me Vilni-ų.
without stopping-sg.gen appl-drive-pst-1pl V.-sg.acc
‘We passed through Vilnius without stopping.’

Most of the non-spatial preverbs functioning as P-applicatives are fairly polyse-

mous; the overview below is based largely on Oertle (2016) for Slavic and Kozhanov
(2016) for Lithuanian.

(i) Temporal (perdurative): Slavic ot-, Lithuanian at- ‘spend a period of time V-ing’, often
with additional nuances such as ‘as an obligation’ or ‘as punishment’, for instance:
Russian rabotat’ ‘work’ ~ otrabotat’ ‘work for a period of time’, služit’ ‘serve’ ~ otslužit’
‘serve for a period of time (e.g., in the army)’, sidet’ ‘sit’ ~ otsidet’ ‘serve a term in prison’,
Polish czekać ‘wait’ ~ odczekać ‘wait for some time’, Serbian stajati ‘stand’ ~ odsta-
jati ‘spend time standing’, Lithuanian kalėti ‘stay in prison’ ~ atkalėti ‘serve a term in
prison’, also verkti ‘cry’ ~ išverkti ‘spend time crying’. The range of objects such verbs
combine with is not limited to temporal periods and by metonymy includes activities
that “fill” these periods (68) and even wages or prizes (69).

(68) Serbian (Šarić and Tchizmarova 2013: 25–26)

Mora-l-i=smo od-sjedi-ti još jedn-o
must-pst-pl.m=aux.prs.1pl appl-sit-inf yet one-sg.n.acc
dosadn-o predavanj-e.
boring-sg.n.acc lecture-sg.acc
‘We had to sit through another boring lecture.’

(69) Russian (RNC)

On=by ot-rabota-l svoj-u
3.sg.m.nom=sbjv appl-work-pst.sg.m rfl.poss-sg.f.acc
Nobel-evsk-uju premij-u
Nobel-adjz-sg.f.acc prize-sg.acc
‘He (Barack Obama) would work for his Nobel prize.’

The most productive perdurative preverbs are Russian pro-, West and South Slavic
pre- (see, e.g., Przybylska 2006: 158–161 on Polish; Oertle 2016: 251–252 on Slavic in
general), Lithuanian pra-, see above.
 14 Applicativizing preverbs in selected European languages 451

(ii) Covering and filling: Slavic o(b)- and za-, Lithuanian ap-, pri-, su- and už-, see (70)–(71):

(70) Russian
a. On-i stroi-l-i cerkv-i v gorod-e.
3-pl.nom build-pst-pl church-pl.acc in city-sg.loc
‘They built churches in the city.’
b. On-i <. . .> za-stroi-l-i cerkvj-ami ves’ gorod.
3-pl.nom appl-build-pst-pl church-pl.ins all[sg.m.acc] city[sg.acc]
‘They . . . built churches all over the city.’ (RNC)

(71) Lithuanian (CCL)

a. Lauke snig-o.
outside snow-pst.3
‘It was snowing outside.’
b. [Aš] buv-au su-ly-t-a, su-snig-t-a.
1sg.nom be-pst.1sg appl-rain-pst.pp-sg.f.nom appl-snow-pst.pp-sg.f.nom
‘I have been rained and snowed upon.’

(iii) Distributive, with the AppP denoting a mass or a set of objects: Slavic o(b)-, Lithua-
nian ap- (72).

(72) Russian
a. Po-zvon-i student-am.
pvb-ring-imp[2sg] student-pl.dat
‘Phone the students.’
b. Ob-zvon-i student-ov.
appl-ring-imp[2sg] student-pl.acc
‘Phone all the students.’

(iv) Overtaking, surpassing and outperforming (compare the English out-applicatives

in § 2.3): Slavic pre-/pere- and less productively o(b)-, in Czech and Slovak also pred-,
Lithuanian ap- and per-: Russian igrat’ ‘play’ ~ obygrat’ ‘beat in a game’, kričat’ ‘shout’ ~
perekričat’ ‘shout louder than’, Slovak predbehnúť ‘overtake while running’, Lithua-
nian lošti ‘play (cards)’ ~ aplošti ‘beat in a card game’, galėti ‘be able to’ ~ pergalėti ‘win,
overcome’. In Lithuanian this meaning can also be expressed by iš- (e.g., augti ‘grow’ ~
išaugti ‘outgrow’) and pra- (gyventi ‘live’ ~ pragyventi ‘outlive’). The most productive
‘surpass’-applicative is the South Slavic prefix nad-, see Mitkovska and Bužarovska
(2012: 145–146), Tchizmarova (2012: 242–244) and Oertle (2016: 179), for instance: Mac-
edonian nadpee ‘outsing’, Serbo-Croat nadglasavati ‘outvote’, Bulgarian nadlăža ‘lie
more than’.
452 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(v) Creation of an object: Slavic vy-, iz-, Lithuanian iš-, also su-, cf. Russian vydumat’,
Bulgarian ižmislja, Lithuanian sugalvoti ‘devise, invent’, from the base verbs meaning
‘think’; Russian rezat’ ‘cut’ ~ vyrezat’ ‘carve’; also Slavic pro-, Lithuanian pra- with the
created object being an aperture or way (Russian rubit’ ‘chop’ ~ prorubit’ dver’ ‘cut a
door’, Lithuanian minti ‘trod’ ~ praminti taką ‘trod a path’). Besides that, the Slavic
preverb na-, whose prominent function is cumulative (Filip 2000; Žaucer 2009; Oertle
2016: 169–171), derives creation verbs with the meaning of excessive activity (73).

(73) Russian (RNC)

Von on-a kak ščёk-i na-e-l-a.
there 3-sg.f.nom how cheek-pl.acc appl-eat-pst-sg.f
‘Look at the cheeks she’s got by eating too much.’

(vi) Acquisition of an object (compare the German er-applicatives in § 3.2): Slavic vy-, iz-,
za-, na-, Lithuanian iš-, su-, už-, pri-, for instance: Russian molit’ ‘pray’ ~ vymolit’ ‘obtain
by praying’, rabotat’ ‘work’ ~ zarabotat’ ‘earn’, Czech ženit ‘marry’ ~ vyženit ‘acquire
through marriage’, Macedonian prosi ‘ask’ ~ isprosi ‘get by (repeated) asking’, Upper
Sorbian nawajchtarich ‘I have earned (it) as a watchman’ (Oertle 2016: 169 quoting
Faßke and Michalk 1981: 116), Lithuanian kovoti ‘fight’ ~ iškovoti ‘conquer’, ieškoti ‘look
for’ ~ suieškoti ‘find’27, Lithuanian prakaituoti ‘sweat’ ~ užprakaituoti ‘earn by hard
work’ (Kozhanov 2015: 252–253), gyventi ‘live’ ~ prigyventi ‘obtain during one’s lifetime;
get a child’. In Slovene this meaning is also productively expressed by pri- (Oertle 2016:
46), cf. prikvartati ‘win in a card game’.

(vii) Elimination of an object: Slavic vy-, iz-, s-, za-, Lithuanian iš-, nu-, už-, for instance:
Russian trjasti ‘shake’ ~ vytrjasti ‘remove by shaking’, plakat’ ‘cry’ ~ vyplakat’ ‘relieve
by tears’, teret’ ‘rub’ ~ steret’ ‘wipe off’, dut’ ‘blow’ ~ zadut’ ‘blow out’, Lithuanian loti
‘bark’ ~ išloti ‘drive away by barking’, lyti ‘rain’ ~ nulyti ‘wash away (of rain)’, lieti ‘pour
liquid’ ~ užlieti ‘extinguish’. The prefixes Slavic pro-/pre- and Lithuanian pra- create
verbs with a meaning of losing something as a result of drinking or gambling, for
instance: Russian pit’ ‘drink’ ~ propit’ ‘spend on drinking’, Lithuanian kortuoti ‘play
cards’ ~ prakortuoti ‘lose by playing cards’. A meaning related to this is the one of
missing something, cf. Czech zaspat ‘miss by sleeping’, Russian progljadet’ ‘overlook’.

(viii) Damage to the object as a result of the activity (compare the German verbs with
the prefixes ver- and zer-, see § 3.2): various preverbs whose choice is often deter-
mined by the lexical semantics of the verb (see, e.g., Oertle 2016: 189–190, 204). The
AppP can be the subject’s own body part, cf. Lithuanian rėkti ‘shout’ ~ prarėkti balsą
‘shout one’s voice hoarse’, Russian ležat’ ‘lie’ ~ otležat’ bok ‘make one’s side numb by
lying on it’, igrat’ ‘play’ ~ pereigrat’ ruku ‘overplay one’s hand’; or artifacts and persons,

27 In this case the AppP is marked by the accusative, while in the BC it is in the genitive.
 14 Applicativizing preverbs in selected European languages 453

cf. Lithuanian sėdėti ‘sit’ ~ susėdėti suknelę ‘rumple dress by sitting’, vogti ‘steal (an
object)’ ~ apvogti ‘rob (a person or a place)’, Russian est’ ‘eat’ ~ ob”est’ ‘eat at some-
body’s expense’, govorit’ ‘speak’ ~ ogovorit’ ‘slander’, xlopat’ ‘clap’ ~ zaxlopat’ ‘clap off
(speech or orator)’, sčitat’ ‘count’ ~ obsčitat’ ‘shortchange’.

(ix) Exhaustion of surface or means: Slavic iz-, Lithuanian iš-, for instance: Russian
pisat’ ‘write’ ~ ispisat’ stranicu ‘to cover a whole page with writing’, ispisat’ ručku ‘to
exhaust a pen by writing’, Lithuanian piešti ‘draw’ ~ išpiešti ‘cover with drawings, deco-
rate’. See also Section 3.3 for German parallel be-and ver-verbs, as well as Example (96)
from Hungarian.

(x) Object that the activity is directed at without literally affecting it: mainly Slavic o(b)-,
Baltic ap-, for instance: Russian govorit’ ‘speak’ ~ obgovorit’ ‘discuss’, Polish płakać
‘cry’ ~ opłakać ‘mourn over’.

(xi) Object affected by speech or magic: Slavic za-, Lithuanian už-, cf. Russian boltat’
‘chatter’ ~ zaboltat’ ‘overwhelm with one’s chatter’, koldovat’ ‘perform or practice
magic’ ~ zakoldovat’ ‘cast spell over an object or person’, Lithuanian kalbėti ‘speak’ ~
užkalbėti ‘cast spell’, also juoktis ‘laugh’ (reflexive) ~ išjuokti ‘deride’.

(xii) Object or person defended: Slavic za-, Lithuanian už-, Lithuanian tarti ‘speak’ ~
užtarti ‘intercede, protect by speech’, stoti ‘stand up’ ~ užstoti ‘defend’ (74a), which is
related to the spatial meaning of ‘screening’ (74b).

(74) Lithuanian (CCL)

a. J-ie <. . .> už-stoj-a <. . .> t-uos, kuri-uos mėg-sta.
3-pl.m.nom appl-stand-prs.3 dem-pl.m.acc which-pl.m.acc like-prs.3
‘They defend those whom they like.’
b. nugar-a už-stoj-u lang-ą.
back-sg.ins appl-stand-prs.1sg window-sg.acc
‘I [stand up in front of her and] hide the window with my back.’

(xiii) Object whose consumption is facilitated by the activity: Slavic za-, Lithuanian
už-. The situations described by such verbs are, for instance, those of taking medi-
cines and drinking water immediately afterwards (75b), or eating something after
drinking alcohol (76a); metaphorically, such verbs describe eating or drinking some-
thing as a means of dealing with emotionally loaded events (76b). The object actually
consumed, which would be expressed as the P in the BC (75a), is marked by the instru-
mental case (75b).

(75) Lithuanian
a. Gėri-au vanden-į.
drink-pst.1sg water-sg.acc
‘I was drinking water.’
454 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

b. Tr-is piliul-es už-gėri-au vandeni-u.

three-pl.acc pill-pl.acc appl-drink-pst.1sg water-sg.ins
‘I washed down three pills with water.’ (CCL)

(76) Russian (RNC)

a. za-ed-a-ja vodk-u zelёn-ym luk-om
appl-eat-ipfv-cvb vodka-sg.acc green-sg.ins onion-sg.ins
‘having vodka with spring onions’
b. I vy poroj za-jed-aj-ete neudač-u konfet-k-oj. . .
and 2pl.nom sometimes appl-eat-ipfv-prs.2pl failure-sg.acc sweet-dim-sg.ins
‘You too sometimes soothe your failures by taking a sweet. . .’

Many preverb+verb combinations are polysemous, their exact interpretation being

determined by the type of the object, for instance: Russian igrat’ ‘play’ ~ razygrat’ obidu
‘simulate an offence’, razygrat’ kartu ‘play a card’, razygrat’ prijatelja ‘to trick a friend’.
As is evident from the list above, the cross-linguistically most common functions of
applicatives, namely benefactive, comitative and instrumental, are virtually unattested
in Slavic and Baltic (on the very restricted comitative-like use see [55] above).
In those cases where the AppP can be expressed in the BC either as an optional
adjunct or an oblique object, the AC always has meanings related to perfectivity and
telicity, that is, attainment of the result (77), total affectedness of the object, as in (72)
above, or higher intensity (78).

(77) Russian
a. My govori-l-i o srok-ax.
1pl.nom talk-pst-pl about deadline-pl.loc
‘We were talking about the deadlines.’
b. My ob-govori-l-i srok-i.
1pl.nom appl-talk-pst-pl deadline-pl.acc
‘We have discussed the deadlines (and settled them).’

(78) Russian
a. On-a sme-ёt-sja nado mnoj.
3-sg.f.nom laugh-prs.3sg-rfl over 1sg.ins
‘She is laughing at me.’
b. On-a vy-sme-iva-et menja.
3-sg.f.nom appl-laugh-ipfv-prs.3sg 1sg.acc
‘She derides me.’
 14 Applicativizing preverbs in selected European languages 455

4.4 Lookalikes

Since none of the Baltic and Slavic preverbs has applicativization as its primary let
alone only function, there is little reason to speak about morphological lookalikes or
“applicativa tantum” verbs. There are numerous verbs whose argument structure is not
affected by the addition of prefixes that in the same or very similar meanings applica-
tivize other verbs, for instance, ‘complete affectedness’ (Russian rezat’ ‘cut’ ~ zarezat’
‘stab to death’), ‘elimination’ (Russian gnat’ ‘drive’ ~ prognat’ ‘drive away’) or ‘creation’
(Lithuanian kepti ‘bake’ ~ prikepti ‘bake a lot’). In many cases this lack of applicativi-
zation is an effect of the so-called “subsumption” of the semantic contribution of the
prefix by the lexical meaning of the verb (see Nübler 1990, Janda et al. 2013 and refer-
ences therein). There are also cases where prefixation affects the argument structure
in different ways with different verbs even when the resulting semantics are similar,
cf. Russian verit’ ‘believe’ ~ uverit’ ‘persuade’ (causative) and govorit’ ‘talk’ ~ ugovorit’
‘persuade, talk into something’ (applicative).
Still, one instance of morphological lookalikes consists in spatial preverbs combin-
ing with verbs of displacement. In such cases, the optional landmark adjuncts, which
are most often expressed as prepositional phrases (the preposition frequently being
related to the preverb), are rendered obligatory oblique arguments by the preverbs
without affecting their coding properties, see (79)–(80).

(79) Russian
a. Mužčin-a nёs čemodan (v komnat-u).
man-sg.nom carry.pst[sg.m] suitcase[sg.acc] in room-sg.acc
‘The man was carrying the suitcase (into the room).’
b. Mužčin-a v-nёs čemodan v komnat-u.
man-sg.nom pvb-carry.pst[sg.m] suitcase[sg.acc] in room-sg.acc
‘The man carried the suitcase into the room.’

(80) Lithuanian (CCL)

a. Martyn-as lėtai lip-o laipt-ais.
M.-sg.nom slowly climb-pst.3 stairs-pl.ins
‘Martin slowly climbed the stairs [unclear up or down].’
b. Zuik-a nu-lip-o nuo dvirači-o.
Z.-sg.nom pvb-climb-pst.3 from bicycle-sg.gen
‘Zuika got off the bicycle.’

With prefixed verbs, such landmark phrases are obligatory and can be omitted
only if their referent is given in context, in which case they are understood as either
deictically anchored (81a) or definite (81b). By contrast, with unprefixed verbs the
absence of a locative phrase does not imply any definite Goal or Source of motion, see
(79a) and (80a) above.
456 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(81) Russian
a. V-xod-i-te.
pvb-walk-imp-2pl
‘Come in [here where the speaker is].’
b. Tramvaj dolgo vёz eё odn-u,
tram[sg.nom] long.time carry.pst.sg.m 3.sg.f.acc one-sg.f.acc
potom vo-š-l-a eščё požil-aja par-a.
then pvb-go-pst-sg.f more elderly-sg.f.nom couple-sg.nom
‘For a long time she was alone on the tram, then an elderly couple got on [the
tram].’ (RNC)

The semantic difference between the constructions with simplex and prefixed verbs in
these cases is purely aspectual: the former is normally interpreted as progressive and
the latter as completive.
Syntactic lookalikes include the productive addition of benefactive (in some lan-
guages also malefactive) phrases in the dative without any change to the verb’s mor-
phology, see (82)–(83); in the generative literature this phenomenon has been analyzed
as applicativization (see, e.g., Gogłoza 2020).

(82) Czech (based on Janda and Townsend 2000: 71)

a. Babičk-a u-pek-l-a dort.
grandmother-sg.nom pvb-bake-pst-sg.f cake[sg.acc]
‘Grandmother baked a cake.’
b. Babičk-a nám u-pek-l-a dort.
grandmother-sg.nom 1pl.dat pvb-bake-pst-sg.f cake[sg.acc]
‘Grandmother baked us a cake.’

(83) Czech (based on Janda and Townsend 2000: 71)

a. Záplav-a z-niči-l-a dům.
flood-sg.nom pvb-destroy-pst-sg.f house[sg.acc]
‘The flood destroyed the house.’
b. Záplav-a nám z-niči-l-a dům.
flood-sg.nom 1pl.dat pvb-destroy-pst-sg.f house[sg.acc]
‘The flood destroyed our house (on us).’

Another case of syntactic lookalikes include argument structure alternations of the

“spray/load” type, sometimes interacting with prefixation (see, e.g., Lenartaitė 2011 and
Sokolova 2012), as shown in (84), where the prefix na- in (84a–b) is a “pure perfec-
tivizer” compatible with the same two argument structures that are available to the
simplex verb, while the prefix po- admits only the location-as-object frame (84c–d).
 14 Applicativizing preverbs in selected European languages 457

(84) Russian
a. (na-)maza-t’ masl-o na xleb
pvb-smear-inf butter-sg.acc on bread[sg.acc]
‘to put butter on bread’
b. (na-)maza-t’ xleb masl-om
pvb-smear-inf bread[sg.acc] butter-sg.ins
‘to cover bread with butter’
c. po-maza-t’ xleb masl-om
pvb-smear-inf bread[sg.acc] butter-sg.ins
‘to cover bread with butter’
d. ✶po-maza-t’ masl-o na xleb
pvb-smear-inf butter-sg.acc on bread[sg.acc]
‘to put butter on bread’

5 Hungarian
5.1 Morphology

Hungarian uses verbal particles (sometimes also called preverbs in the literature) pro-
ductively, many of which can change the argument structure of verbs in various con-
structions, as the applicative constructions show in (85b) and (86b) as compared to (85a)
and (86a). There are no particles whose only function is applicativization; they all have
spatial origins and their directional meaning is often present while they also have a
role in determining the aktionsart and the aspectual properties of the verbal expres-
sion (Kiefer 1994; É. Kiss 2006). While the verbs in (85a) and (86a) describe unbounded
activities, their counterparts with the particles involve a bounded, completed event.28

(85) Hungarian
a. Péter nevet-ett János-on. (BC)
P. laugh-pst.3sg J.-sup
‘Peter was laughing at John.’
b. Péter ki-nevet-t-e János-t. (AC)
P. appl-laugh-pst-defobj.3sg J.-acc
‘Peter laughed at John.’

28 Hungarian verbs exhibit definiteness agreement with their object (i.e., definite objects trigger differ-
ent verbal agreement markers than indefinite or bare objects; see, e.g., É. Kiss 2002), but this is orthog-
onal to applicativization.
458 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(86) Hungarian
a. Péter úsz-ott a folyó-ban. (BC)
P. swim-pst.3sg the river-ine
‘Peter was swimming in the river.’
b. Péter át-úsz-t-a a folyó-t. (AC)
P. appl-swim-pst-defobj.3sg the river-acc
‘Peter swam across the river.

Most Hungarian particles are formally (and functionally) closely related to direc-
tional postpositions or case suffixes. Etymologically, they go back to Goal-denoting
postpositions or adverbs, which, in turn, originated as case-marked nouns (and some
of them have grammaticalized into case suffixes; see Dékány and Hegedűs 2021 for a
recent overview). Some particles have started grammaticalizing more recently and are
formally identical to postpositions: (86b) and (87a) involve one such particle, namely át
‘across, over, through’; (87b) shows át occurring as a postposition (the PP is a focused
adjunct).

(87) Hungarian
a. Péter át-úsz-ott a sziget-re.
P. appl-swim-pst.3sg the island-sublat
‘Peter swam over to the island.’
b. Péter London-on át fog New York-ba repül-ni.
P. L.-sup across fut[3sg] N.Y.-ill fly-inf
‘It is via London that Peter will fly to New York.’

There is no agreement on the exact number or list of particles in the language.

This is mostly due to the fact that some of them are less grammaticalized while others
have proceeded further along the grammaticalization; since this has an effect on their
productivity, some grammars may include more elements than others. Dékány and
Hegedűs (2021) provide a representative list and description. Here, two types of parti-
cles are worth pointing out. Group A includes the oldest particles of the language, which
are productive in various kinds of valency-changing patterns (see § 5.2); while they
are mostly directional, they are not used as postpositions in present-day Hungarian.
The particles from this group whose applicativizing function is most prominent are
be ‘into’, ki ‘out’, le ‘down’, and meg ‘orig. backdir, behinddir’. Group B includes particles
like át ‘over, through’ in (86)–(87) above), which are also postpositions in present-day
Hungarian, and as such they select a case-marked complement. Particles from Group
B started grammaticalizing later (from around the 16th century) and often co-occur
with obligatory spatial arguments (which usually bear the same case they select when
used as postpositions), but they are less prolific as particles, especially when it comes to
introducing object arguments (see Hegedűs 2020 for a more detailed discussion). Some
of such particles are: át ‘over, through’, túl ‘over, beyond’, and végig ‘to the end, along’.
 14 Applicativizing preverbs in selected European languages 459

All these common particles—whose semantics roughly corresponds to that of some

of the applicativizing prefixes in German and Baltic-Slavic, see §§ 3.1 and 4.1, respec-
tively—are summarized in Table 8.

Table 8: Main Hungarian applicativizing particles.

Group A Group B

be ‘into’ át ‘over, through’

ki ‘out’ túl ‘over, beyond’
le ‘down’ végig ‘to the end, along’
meg < ‘backdir, behinddir’

5.2 Syntax

Hungarian applicatives are optional P-applicatives; there is no obligatory marker on

the verb for introducing an argument. In neutral declarative-affirmative sentences,
particles immediately precede the verb, forming one phonological unit with it.29 By
contrast, they occur separated from the verb in negative and interrogative sentences,
or in clauses with a focused preverbal constituent.30 Particles can also undergo long
distance (contrastive) topicalization, whereby they can even cross clause boundaries.
Their syntactic independence is the main reason for treating them as heads of phrases
(i.e., they are projecting particles in the sense of Los et al. 2012). Applicative uses of
various particles are possible and productive to some extent, depending on lexical
factors.
Particles can transitivize verbs that are intransitive in the BC. Át ‘over, through’,
which we have seen in (86)–(87), is one of the directional particles that also appears
with verbs that do not involve (actual or metaphorical) motion. It can appear with
regular activity verbs that may optionally take a case-marked non-core argument, but
the AC has an obligatory direct object (which takes accusative marking when expressed
as an overt NP), see (88)–(89). With speech verbs like beszél ‘talk’, the particle meg is also
possible with the same arguments, with a very small difference in meaning (89).

29 Verbal particles are a subset of so-called verb modifiers (all of which appear immediately before the
verb in such neutral sentences). Particles are the most frequent and most lexicalized verb modifiers in
the language (Kálmán 1985; É. Kiss 2002, 2006).
30 Contrary to the received opinion, and not unlike some German elements (see § 3.1), preverbal el-
ements like be ‘into’, fel ‘up’ and ki ‘out’ can also occur like prefixes with a limited number of verbs
in a special derivational pattern, albeit without any connection to applicativization. See Hegedűs and
Dékány (2017) for details.
460 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

(88) Hungarian
a. Péter dolgoz-ott (a könyv-é-n). (BC)
P. work-pst.3sg the book-poss.3sg-sup
‘Peter worked (on his book).’
b. Péter át-dolgoz-t-a a könyv-é-t. (AC)
P. appl-work-pst-defobj.3sg the book-poss.3sg-acc
‘Peter revised his book.’

(89) Hungarian
a. Mari és Anna beszél-t-ek (a problémá-ról).
M. and A. talk-pst-3pl the problem-del
‘Mary and Anna talked (about the problem).’
b. Mari és Anna át-beszél-t-ék a problémá-t.
M. and A. appl-talk-pst-defobj.3pl the problem-acc
‘Mary and Anna discussed (all aspects of) the problem.’
c. Mari és Anna meg-beszél-t-ék a problémá-t.
M. and A. appl-talk-pst-defobj.3pl the problem-acc
‘Mary and Anna discussed the problem.’

Another such particle is végig ‘to the end, along’, which can appear with verbs such
as sír ‘cry’ and nevet ‘laugh’, which take an optional oblique; the former verb is illus-
trated in (90). It can also appear with verbs like gondol ‘think, consider’, which takes an
oblique argument when underived (91a) and promotes the argument to an obligatory
direct object when applicativized (91b).

(90) Hungarian
a. Sír-t-unk (a film-en).
cry-pst-1pl the film-sup
‘We cried (over the film).’
b. Végig-sír-t-uk a film-et.
appl-cry-pst-defobj.1pl the film-acc
‘We cried the whole time we watched the film.’

(91) Hungarian
a. Simon gondol-t egy jó feladat-ra.
S. think-pst.3sg a good task-sublat
‘Simon thought of a good task.’
b. Simon végig-gondol-t minden feladat-ot.
S. appl-think-pst.3sg every task-acc
‘Simon considered every task.’
 14 Applicativizing preverbs in selected European languages 461

The particle le ‘down’ can applicativize verbs like győz ‘win (against someone)’, which
can have an optional postpositional argument (92a), and also verbs such as jár ‘walk’ or
táncol ‘dance’, as well as other activity verbs with which the particle adds the meaning
of wearing down, depleting or using up (93).

(92) Hungarian
a. A csapatunk győz-ött (a másik csapat ellen).
the team-poss.1pl win-pst.3sg the other team against
‘Our team won (against the other team).’
b. A csapat-unk le-győz-t-e a másik csapat-ot.
the team-poss.1pl appl-win-pst-defobj.3sg the other team-acc
‘Our team defeated the other team.’

(93) Hungarian
a. Le-jár-t-am a cipő-m talp-á-t.
appl-walk-pst-1sg the shoe-poss.1sg sole-poss-acc
‘I got the sole of my shoes worn out by walking.’
b. Le-táncol-t-am néhány kiló-t.
appl-dance-pst-1sg some kilogram-acc
‘I danced off a few kilos.’
c. Le-hullámvasutaz-t-am minden jegy-et.
appl-ride_roller_coaster-pst-1sg every ticket-acc
‘I used up all the tickets riding the roller coaster.’

The particle ki ‘out’ can be used with motion verbs, where it merely functions as
a directional modifier. In (85) above, however, we saw that ki ‘out’ can also change the
valency of the verb and introduce an obligatory accusative-marked object with an
intransitive verb that can only appear with an optional superessive-marked argument
in the BC. Verbs such as könyörög ‘beg’, sír ‘cry’, harcol ‘fight’ can also be applicativized
with ki; the BC in (94a) features an optional argument in the causal-final case, and the
AC in (94b) is actually ditransitive, not only with the base P as direct object but also with
an optional dative-marked beneficiary (cf. also Example [24] from German, which could
take the reflexive in sich etwas erkämpfen ‘get oneself something by fighting’):

(94) Hungarian
a. Lili könyörg-ött egy új bicikli-ért.
L. beg-pst.3sg a new bicycle-cau
‘Lily begged for a new bicycle.’ (BC)
b. Lili ki-könyörg-ött (magá-nak / Anná-nak) egy új bicikli-t.
L. appl-beg-pst.3sg self-dat A.-dat a new bicycle-acc
‘Lily got (herself/Anna) a new bicycle by begging.’ (AC)
462 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

Example (95) illustrates the situation with nő ‘grow’, which can only take an object that
measures out the event when underived, not a regular referential object. Ki provides
the semantic content of measuring out the event, however, so the object can be a refer-
ential argument in (95b).

(95) Hungarian
a. A fa nő-tt (egy méter-t / sok-at).
the tree grow-pst.3sg one meter-acc lot-acc
‘The tree has grown (one meter / a lot).’
b. Anna ki-nő-tt-e a cipő-jé-t.
A. appl-grow-pst-defobj.3sg the shoe-poss.3sg-acc
‘Anna has grown out of her shoes.’

Similarly to what we saw in (46) for Lithuanian, the Hungarian particle be ‘into’ can
add a new object that refers to a location or area, and the base object of the BC becomes
an optional instrumental-marked argument in the AC (96). The particle also indicates
that the activity covers the whole area expressed by the new object (see § 3.3 on the
German Theme-covers-surface schema).

(96) Hungarian
a. János búzá-t vet-ett. (BC)
J. wheat-acc sow-pst.3sg
‘John sowed wheat.’
b. János be-vet-ett két hektár föld-et (búzá-val). (AC)
J. appl-sow-pst.3sg two hectare land-acc wheat-ins
‘John sowed two hectares of land with wheat.’

Hungarian has examples similar to the well-known English locative alternation (see
also Examples [21]–[22] from German). Most of the time, they involve the particle meg,
which used to mean ‘backdir, behinddir’ but has no productive directional use in pres-
ent-day Hungarian and is mostly used as a perfectivizer. The most frequent verbs found
in the locative alternation are rak and pakol ‘put, load’, tölt ‘fill’, and szór ‘sprinkle’.
The BC has a direct object (often an indefinite or bare nominal) and an argument in the
illative, while the AC has a direct object (generally a specific, often definite, nominal)
and an argument in the instrumental (97).

(97) Hungarian
a. Péter bor-t tölt az üveg-ek-be. (BC)
P. wine-acc fill[3sg] the bottle-pl-ill
‘Peter fills wine into the bottles.’
 14 Applicativizing preverbs in selected European languages 463

b. Péter meg-tölt-i az üveg-ek-et bor-ral. (AC)

P. appl-fill-defobj.3sg the bottle-pl-acc wine-ins
‘Peter fills the bottles with wine.’

The particle meg is also the one used in the few instances of ditransitive verbs. In the
BC with ajándékoz ‘present, gift’, for instance, the accusative-marked Theme is gener-
ally indefinite or a bare nominal and the Recipient is in the dative (98a); in the AC, the
Recipient is the object in the accusative and the Theme takes the instrumental (98b).31

(98) Hungarian
a. Anna könyve-t ajándékoz-ott Tomi-nak. (BC)
A. book-acc gift-pst.3sg T.-dat
b. Anna meg-ajándékoz-t-a Tomi-t egy könyv-vel. (AC)
A. appl-gift-pst-defobj.3sg T.-acc a book-ins
Both: ‘Anna gave Tommy a book as a gift.’

Finally, there are also instances in which the particle does not introduce an object
but a directional argument with a selected case. In (99a), the motion verb ugrik ‘jump’
occurs with various directional arguments, and the form of the PP is not selected; the
sentence is grammatical as long as the argument is directional. Nevertheless, when the
particle bele ‘into’ is added in (99b), the directional complement must take illative case,
which is formally and semantically very closely related to the particle itself.

(99) Hungarian
a. Pál a medencé-be / az ernyő alá / Anna mellé
P. the pool-ill the umbrella under A. next.to
ugr-ott.
jump-pst.3sg
‘Paul jumped into the pool / under the umbrella / next to Anna.’
b. Pál bele-ugr-ott a medencé-be.
P. into-jump-pst.3sg the pool-ill
‘Paul jumped into the pool.’

31 Hungarian has a number of “definiteness effect” verbs (É. Kiss 1995) that denote existence, becom-
ing, creating, or appearing in a certain place or in a certain manner (kap ‘receive, get’, talál ‘find’ also be-
long here). Such verbs do not allow definite NPs in one of their argument positions, a property Szabolcsi
(1986) attributes to the fact that their meaning includes ‘existence’ or ‘appearance (on the scene)’. This
effect disappears once a particle is added to the verb, or when there is a structural focus in the clause
(see É. Kiss 2021 for a descriptive overview).
464 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

There are several recently developed particles that show the same pattern: bele ‘into’
and illative -ba/-be, hozzá ‘towards’ and allative -hoz/-hez/-höz, neki ‘toward’ and dative
-nak/-nek (originally a general lative case), and rá ‘onto’ and sublative -ra/-re.

5.3 Semantics

Hungarian particles generally encode either a spatial meaning where the particle lexi-
calizes the Goal of the event (generally with motion verbs) or a perfectivizing meaning,
signaling that the event has reached its endpoint or telos (É. Kiss 2006; see also Dékány
and Hegedűs 2021). Unsurprisingly, since Hungarian particles have developed from
directional adverbs or adpositions, all the recently developed ones still have such a
semantic contribution. Among the oldest particles, only meg is no longer productive
with motion verbs and cannot express the endpoint of motion anymore; the others are
productive both in their original spatial meaning and in their more functional, perfec-
tivizing, function.
Whether the AppP is a prototypical or atypical Patient depends on the lexical
meaning of the verb. Whereas the non-subject is marked with a locative or directional
case in the BC with verbs of mental activities or speech, it becomes an AppP with the
introduction of the particle, and the meaning is—not unlike German with be-verbs, see
§ 3.3—that the totality of this object is involved in the mental activity or speech. When
át ‘over, through’ is used with an activity verb such as dolgozik ‘work’, the AC refers to
a complete transformation or thorough change caused by the subject and affecting the
AppP.
In ACs with le ‘down’, there is a semantic element of destruction or wearing down
that applies to the object (cf. also German ver- and zer-verbs in § 3). At the other end
of the scale, when we look at examples like (94), the AppP appears on the scene and
becomes the possession of the dative-marked argument as a result of the event.

5.4 Lookalikes

As mentioned earlier, verbal particles do not necessarily change verb valency: neither
do all transitive or ditransitive verbs feature a particle nor does a verbal particle auto-
matically transitivize its host verb.
Regarding syntactic lookalikes, Hungarian clauses can accommodate dative-
marked NPs without resorting to verbal derivation, as in German and Slavic-Baltic, in
order to express several extra-thematic participants, including Beneficiaries/Malefi-
ciaries, External Possessors, etc. (cf. §§ 3.4 and 4.4).
Regarding morphological lookalikes in terms of preverbs introducing no new argu-
ments, consider (100), which illustrates a frequent minimal pair in the language. In the
 14 Applicativizing preverbs in selected European languages 465

clause with ki ‘out’, there is no obligatory object introduced; the preverb only turns an
activity verb into an accomplishment:

(100) Hungarian
a. János takarít-ott.
J. clean-pst.3sg
‘John was cleaning.’
b. János ki-takarít-ott.
J. out-clean-pst.3sg
‘John did the cleaning.’

Other morphological lookalikes are similar to German denominal be-verbs but have a
slightly more complicated history and morphology (Hegedűs and Dékány 2017).

6 Conclusions
This chapter surveys various types of applicative constructions and some non-applica-
tive uses of applicative morphology attested in English, German, and Hungarian, as well
as in Slavic and Baltic languages. According to the questionnaire proposed as a guide-
line for this volume’s contributions, the constructions presented can be characterized
as follows:

Morphology
– Slavic-Baltic applicative markers are verbal prefixes. Hungarian applicatives are
preverbal particles. Germanic applicative markers can be either prefixes or parti-
cles (postverbal in English, preverbal in German).
– Most markers surveyed occupy the same slot as their spatial-aspectual counter-
parts in the verbal complex. German Group-I prefixes (e.g., be-) occur closer to the
verb root.
– German ent- and Slavic-Baltic prefixes show some morpho-phonological allomor-
phy; other applicativizing preverbs are invariable.
– Applicativized verbs do not show any morphological idiosyncrasies, with the
exception of the restrictions on the expression of present and/or future tense forms
in Slavic related to the perfectivizing function of prefixes and the interaction with
the reflexive marker in Lithuanian.
– There are no applicative periphrases or analytic applicative constructions in our
sample.
466 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

Syntax
– Most of the constructions surveyed are P-applicatives. Baltic and Slavic languages
also feature X-applicatives and, rarely, D-applicatives; some preverbs in Latvian
and ent- in German create D-applicatives as well.
– European preverbal applicativization is typically optional in that the participant
expressed by the AppP does not require applicativization to be expressed in a
clause. Nevertheless, that participant often has similar, but not identical, semantic
roles in the BC and the AC.
– The applicatives surveyed are normally valency-increasing. In some instances in
English, German, and Hungarian, they can be valency-neutral (or even valency-re-
ducing, like with some be-verbs in German).
– Preverbal applicatives do not appear to show any restrictions in combination with
other valency-changing operations (at least passivization and causativization). The
reflexivization and reciprocalization of applicatives does not seem to be subject
to structural restrictions in Germanic and Hungarian; these morphological pro-
cesses are lexically restricted in Baltic-Slavic, so there appear to be some important
restrictions here, but more research is needed on this issue.

Semantics
– The applied phrase bears a variety of space-related, landmark-like roles in the
applicatives surveyed. Notably, German, Slavic, and Baltic show P-applicatives with
the meanings ‘through’, ‘across’, and ‘around’. In many cases, these roles can be
metaphorically extended and cover an ever wider range of possible participants.
– Cases that arguably represent extensions of spatial roles include the exceeded
threshold / surpassed competitor with English out-applicatives or South Slavic
nad-applicatives, the “holistic” Patient of many German be-applicatives, and most
of the Hungarian applicatives.

Lookalikes
– Some syntactic lookalikes, particularly those due to uncoded polyvalency with A-la-
bile predicates, can be found in Slavic-Baltic and Hungarian. They appear to be less
common in German and much more common in English (cf. the locative, benefac-
tive, and dative alternations).
– The most prominent syntactic lookalike to a D-applicative in all the languages sur-
veyed except English consists of simply adding a dative-marked NP to a clause,
subject only to semantic and pragmatic restrictions (a so-called extra-thematic
dative with several possible semantic roles: Possessor, Beneficiary, Maleficiary,
Experiencer, Viewpoint Holder. . .).
– Morphological lookalikes are extremely common in all the languages surveyed. In
addition to the same markers having an applicativizing function with some verbs
and a spatial-aspectual yield with others (or, in the case of some English particles,
 14 Applicativizing preverbs in selected European languages 467

in different contexts with the same verbs), the English prefix out- appears to occur
comparatively often as an applicative marker. The picture in German is rather
varied, especially with Group-I prefixes (and particularly be-), which occur on
numerous verbs as lexicalized forms, as “applicativa tantum”, or as verbs whose
applied phrase in the applicative clause bears only an unpredictable and occasion-
ally tenuous semantic relationship to its counterpart in the base clause.
– A special case of morphological lookalike is found in Baltic-Slavic, with verbs of dis-
placement. Here, preverbs turn optional landmark adjuncts into obligatory oblique
arguments, a case of oblique registration.

Abbreviations
AC applicative construction
acc accusative
adjz adjectivizer
aor aorist
appl applicative
AppP applied phrase
art article
aux auxiliary verb
BC base construction
cau causal-final
cvb converb
dat dative
def definite
defobj definite object
del delative
dem demonstrative
dim diminutive
dobj direct object
f feminine
gen genitive
ill illative
imp imperative
ine inessive
inf infinitive
ins instrumental
iobj indirect object
ipfv imperfective
loc locative
m masculine
n neuter
neg negation
nom nominative
np noun phrase
468 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

obl oblique
oblreg oblique registration
OHG Old High German
pfv perfective
PG Proto-Germanic
PIE Proto-Indo-European
pl plural
poss possessive
pp passive participle
prep preposition
prs present
pst past
pvb preverb
rel relativizer
rfl reflexive
sbj subject
sbjv subjunctive
sg singular
sublat sublative
sup superessive

Online sources
CCL: Corpus of Contemporary Lithuanian, http://tekstynas.vdu.lt/tekstynas
CNC: Czech National Corpus, https://www.korpus.cz/
RNC: Russian National Corpus, https://ruscorpora.ru/new/index.html

References
Ahn, Byron. 2022. Mapping OUT- Argument Structure. Syntax 25(4). 417–465.
Arkadiev, Peter. 2014. Towards an areal typology of prefixal perfectivization. Scando-Slavica 60(2). 384–405.
Arkadiev, Peter. 2015. Areal’naja tipologija prefiksal’nogo perfektiva (na materiale jazykov Evropy i Kavkaza)
[Areal typology of prefixal perfective (on the data of the languages of Europe and the Caucasus].
Moscow: LRC.
Arkadiev, Peter & Kirill Kozhanov. Forthcoming. Object partitive of negation: An areal perspective. To appear
in Björn Wiemer, Peter Arkadiev, Rogier Blockland & Petar Kehayov (eds.), Convergence and divergence
in the eastern Circum-Baltic Area. A synthetic view, especially on Finnic, and case studies. Berlin: de Gruyter.
Bahrt, Nicklas. 2021. Voice syncretism. Berlin: Language Science Press.
Bauer, Laurie, Rochelle Lieber & Ingo Plag (eds.). 2013. The Oxford reference guide to English morphology.
Oxford: Oxford University Press.
Beavers, John. 2017. The spray/load alternation. In Martin Everaert & Henk van Riemsdijk (eds.), The Wiley
Blackwell companion to syntax, 2473–2503. DOI: 10.1002/9781118358733.wbsyncom066
 14 Applicativizing preverbs in selected European languages 469

Biskup, Petr. 2017. Prepositions, case and verbal prefixes: The case of Slavic. Leipzig: University of Leipzig
habilitation thesis.
Booij, Geert & Ans van Kemenade. 2003. Preverbs: An introduction. In Geert Booij & Jaap van Marle (eds.),
Yearbook of Morphology 2003, 1–11. Dordrecht: Kluwer.
Booij, Geert & Jaap van Marle (eds.). 2003. Yearbook of Morphology 2003. Dordrecht: Kluwer.
Brinkmann, Ursula. 1997. The locative alternation in German: Its structure and acquisition. Amsterdam: John
Benjamins.
Cysouw, Michael. 2023. Encyclopaedia of German diatheses. Berlin: Language Science Press.
Dehé, Nicole. 2015. Particle verbs in Germanic. In Peter Müller, Ingeborg Ohnheiser, Susan Olsen & Franz
Reiner (eds.), Word-formation: An international handbook of the languages of Europe, Vol. 1, 611–626.
Berlin: De Gruyter Mouton.
Dehé, Nicole, Ray Jackendoff, Andrew McIntyre & Silke Urban (eds.). 2002. Verb-particle explorations. Berlin:
Mouton de Gruyter.
Dékány, Éva & Veronika Hegedűs. 2021. Postpositions: formal and semantic classification. In Katalin É.
Kiss & Veronika Hegedűs (eds.), Syntax of Hungarian: Postpositions and postpositional phrases, 11–192.
Amsterdam: Amsterdam University Press.
Dewell, Robert. 2015. The semantics of German verb prefixes. Amsterdam: John Benjamins.
Dickey, Stephen M. 2000. Parameters of Slavic aspect. A cognitive approach. Stanford: CSLI Publications.
É. Kiss, Katalin. 1995. The definiteness effect revisited. In István Kenesei (ed.), Approaches to Hungarian 6,
63–88. Szeged: JATE.
É. Kiss, Katalin. 2002. The syntax of Hungarian. Cambridge: Cambridge University Press.
É. Kiss, Katalin. 2006. The function and the syntax of the verbal particle. In Katalin É. Kiss (ed.), Event
structure and the left periphery: Studies on Hungarian, 17–56. Dordrecht: Springer.
É. Kiss, Katalin. 2021. PPs used as obligatory adjuncts. In Katalin É. Kiss & Veronika Hegedűs (eds.), Syntax of
Hungarian: Postpositions and postpositional phrases, 355–370. Amsterdam: Amsterdam University Press.
Endzelīns, Jānis. 1971 [1907]. Latyšskie predlogi [Latvian Prepositions]. Part II. In: Jānis Endzelīns, Darbu izlase
[Selected Writings], Vol. I. Riga: Zinātne, 521–655.
Faßke, Helmut & Siegfried Michalk. 1981. Grammatik der obersorbischen Schriftsprache der Gegenwart:
Morphologie. Bautzen: Domovina.
Filip, Hana. 2000. The quantization puzzle. In James Pustejovsky & Carol Tenny (eds.), Events as grammatical
objects, 3–60. Stanford: CSLI Publications.
Gogłoza, Aleksandra. 2020. Polish datives—an applicative analysis. Berlin: Humbold-Universität dissertation.
Gronemeyer, Claire. 1995. Swedish applied verbs derived by the prefix be-. Lund University Dept. of Linguistics
Working Papers 44. 21–40.
Haiden, Martin. 2006. Verb particle constructions. In Martin Everaert & Henk van Riemsdijk (eds.), The
Blackwell companion to syntax, Vol. V, 344–375. Oxford: Blackwell.
Hegedűs, Veronika. 2020. A névutón át, közel az igekötőhöz: A grammatikalizálódó ragvonzó névutók
kategóriájáról [Through postpositions, close to verbal particles: On the category of case-assigning
postpositions undergoing grammaticalization]. In Éva Dékány, Tamás Halm & Balázs Surányi (eds.),
Általános Nyelvészeti Tanulmányok XXXII [Studies in General Linguistics XXXII], 83–99. Budapest:
Akadémiai Kiadó.
Hegedűs, Veronika & Éva Dékány. 2017. Two positions for verbal modifiers: Evidence from derived particle
verbs. In Harry van der Hulst & Anikó Lipták (eds.), Approaches to Hungarian 15, 65–94. Amsterdam:
John Benjamins.
Holvoet, Axel & Nau, Nicole. 2016. Variation in argument realization in Baltic: An overview. In Axel Holvoet &
Nicole Nau (eds.), Argument realization in Baltic, 1–34. Amsterdam: John Benjamins.
Jackendoff, Ray. 1997. Twistin’ the night away. Language 73(3). 534–559.
Janda, Laura. 2020. Prefixation. In Marc L. Greenberg (ed.), Encyclopedia of Slavic Languages and Linguistics
Online. http://dx.doi.org/10.1163/2589-6229_ESLO_COM_036310
470 Fernando Zúñiga, Peter Arkadiev, and Veronika Hegedűs

Janda, Laura, Anna Endresen, Julia Kuznetsova, Olga Lyashevskaya, Anastasia Makarova, Tore Nesset
& Svetlana Sokolova. 2013. Why Russian aspectual prefixes aren’t empty: Prefixes as verb classifiers.
Bloomington, Ind.: Slavica.
Janda, Laura & Charles Townsend. 2000. Czech. Munich: LINCOM Europa.
Kalluli, Dalina & Liliane Tasmowski (eds.). 2008. Clitic doubling in the Balkan languages. Amsterdam: John
Benjamins.
Kálmán, László. 1985. Word order in neutral sentences. In István Kenesei (ed.), Approaches to Hungarian 1,
13–23. Szeged: JATE.
Kiefer, Ferenc. 1994. Aspect and syntactic structure. In Ferenc Kiefer & Katalin É. Kiss (eds.), Syntax and
Semantics 27: The syntactic structure of Hungarian, 415–464. San Diego / New York: Academic Press.
Koonen, Guus. 2013. Etymological dictionary of Proto-Germanic. Leiden: Brill.
Kotowksi, Sven. 2020. The semantics of English out-prefixation: A corpus-based investigation. English
Language & Linguistics 25(1). 61–89.
Kozhanov, Kirill. 2015. Balto-slavjanskie areal’nye kontakty v oblasti glagol’noj prefiksacii [Balto-Slavic contacts
in the domain of verbal prefixation]. Moscow: Institute of Slavic Studies of the Russian Academy of
Sciences dissertation.
Kozhanov, Kirill. 2016. Verbal prefixation and argument structure in Lithuanian. In Axel Holvoet & Nicole
Nau (eds.), Argument realization in Baltic, 363–402. Amsterdam: John Benjamins.
Kuryłowicz, Jerzy. 1964. The inflectional categories of Indo-European. Heidelberg: Carl Winter.
Lenartaitė, Kristina. 2011. Argumentų raiškos alternavimas lietuvių kalboje [Alternations in argument
realization in Lithuanian]. Vilnius: Vytautas Magnus University dissertation.
Letučij, Aleksandr B. 2012. Vvodjat li pristavki prjamoe dopolnenie? [Do prefixes introduce direct objects?].
In Problemy jazyka. Sbornik naučnyx statej po materialam Pervoj konferencii-školy “Problemy jazyka: vzgljad
molodyx učenyx” (20-22 sentjabrja 2012 g.) [Problems of Language. Proceedings of the 2012 young
linguists’ conference], 124–139. Moscow: Institute of Linguistics of the Russian Academy of Sciences.
Levin, Beth & Malka Rappaport Hovav. 2005. Argument realization. Cambridge: Cambridge University Press.
Los, Bettelou, Corrien Blom, Geert Booij, Marion Elenbaas & Ans van Kemenade. 2012. Morphosyntactic
change: A comparative study of particles and prefixes. Cambridge: Cambridge University Press.
McIntyre, Andrew. 2007. Particle verbs and argument structure. Language and Linguistics Compass 1.
10.1111/j.1749-818X.2007.00013.x
McIntyre, Andrew. 2015. Particle verb formation. In Peter Müller, Ingeborg Ohnheiser, Susan Olsen & Franz
Rainer (eds.), Word-formation: An international handbook of the languages of Europe, Vol. 1, 435–449.
Berlin: Mouton de Gruyter.
McIntyre, Andrew. 2016. Denominal verbs. In Peter Müller, Ingeborg Ohnheiser, Susan Olsen & Franz Rainer
(eds.), Word-formation: An international handbook of the languages of Europe, Vol. 4, 1406–1423. Berlin:
De Gruyter Mouton.
Michaelis, Laura & Josef Ruppenhofer. 2001. Beyond alternations: A constructional model of the German
applicative pattern. Stanford: CSLI Publications.
Mitkovska, Liljana & Eleni Bužarovska. 2012. The preposition and prefix nad in South Slavic languages with
emphasis on Macedonian. Jezikoslovlje 13(1). 107–150.
Nübler Norbert. 1990. Zum Begriff der “Subsumptionspräfixe” in der Aspektforschung. Anzeiger für Slavische
Philologie 20. 123–134.
Oertle, Simon. 2016. Die slavischen Verbalpräfixe und Präpositionen: Polysemie und Grammatikalisierung.
Herne: Gabriele Schäfer Verlag.
Pacchiarotti, Sara. This volume. The polyfunctional applicative ✶-ɪd in Bantu languages.
Petit, Daniel. 2011. Préverbation et préfixation en baltique. In Daniel Petit, Claire Le Feuvre & Henri
Menantaud (éds.), Langues baltiques, langues slaves, 243–279. Paris: Éditions CNRS.
Pirnat, Žiga. 2015. Genesis of the genitive of negation in Balto-Slavic and its evidence in contemporary
Slovenian. Slovene Linguistic Studies 10. 3–52.
 14 Applicativizing preverbs in selected European languages 471

Przybylska, Renata. 2006. Schematy wyobrażeniowe a semantyka polskich prefiksów czasownikowych do-, od-,
prze-, roz-, u- [Image schemas and semantics of Polish verbal prefixes do-, od-, prze-, roz-, u-]. Cracow:
Universitas.
Richardson, Kylie. 2007. Case and aspect in Slavic. Oxford: Oxford University Press.
Rousseau André (ed.). 1995. Les préverbes dans les langues d’Europe. Introduction à l’étude de la préverbation.
Lille: Presse Universitaires de Septentrion.
Šarić, Ljiljana & Petra Mikolić. 2015. A semantic analysis of the verbal prefix o(b)- in Croatian. Croatica et
Slavica Iadertina 11(2). 249–283.
Šarić, Ljiljana & Ivelina Tchizmarova. 2013. Space and metaphor in verbs prefixed with ot- ‘from’ in Bosnian/
Croatian/Serbian and Bulgarian. In Ljiljana Šarić (ed.), Space in South Slavic, special issue of Oslo Studies
in Language 5(1). 7–33.
Schröder, Anne. 2011. On the productivity of verbal prefixation in English: Synchronic and diachronic perspectives.
Tübingen: Narr.
Sokolova, Svetlana. 2012. Asymmetries in linguistic construal: Russian prefixes and the locative alternation.
Tromsø: University of Tromsø dissertation.
Stefanowitsch, Anatol & Ada Rohde 2004. The goal bias in the encoding of motion events. In Günter Radden
& Klaus-Uwe Panther (eds.), Studies in linguistic motivation, 249–267. Berlin: Mouton de Gruyter.
Stiebels, Barbara. 1996. Lexikalische Argumente und Adjunkte: Zum semantischen Beitrag von verbalen Präfixen
und Partikeln. Berlin: Akademie Verlag.
Svilans-Dennis, Iréna. 1982. Verb prefixes in Latgalian: A discussion based on data found in a collection of
Latgalian folk tales. Canberra: Australian National University MA thesis.
Swan, Oscar E. 2002. A grammar of Contemporary Polish. Bloomington, Indiana: Slavica.
Szabolcsi, Anna. 1986. From the definiteness effect to lexical integrity. In Sjaak de Mei & Werner Abraham
(eds.), Topic, focus, and configurationality, 321–348. Amsterdam: John Benjamins.
Tchizmarova, Ivelina K. 2012. A cognitive analysis of the Bulgarian prepositions and verbal prefixes nad and
pod. Jezikoslovlje 31(1). 219–260.
Thim, Stefan. 2012. Phrasal verbs: The English verb-particle construction and its history. Berlin: De Gruyter
Mouton.
van der Auwera, Johan. 1999. Dutch verbal prefixes: meaning and form, grammaticalization and
lexicalization. In Lunella Mereu (ed.), Boundaries of morphology and syntax, 121–136. Amsterdam: John
Benjamins.
van Kemenade, Ans & Bettelou Los. 2003. Particles and prefixes in Dutch and English. In Geert Booij & Japp
van Marle (eds.), Yearbook of Morphology 2003, 79–117. Dordrecht: Kluwer.
Wasow, Thomas. 1997. Remarks on grammatical weight. Language Variation and Change 9. 81–105.
Watkins, Calvert. 1964. Preliminaries to the reconstruction of Indo-European sentence structure. In Horace
Lunt (ed.), Proceedings of the 9th International Congress of Linguists, 1035–1042. The Hague: Mouton.
Wiemer, Björn & Ilja Seržant. 2017. Diachrony and typology of Slavic aspect: What does morphology tell
us? In Walter Bisang & Andrej Malchukov (eds.), Unity and diversity in grammaticalization scenarios,
239–307. Berlin: Language Science Press.
Žaucer, Rok. 2009. A VP-internal/resultative analysis of 4 “VP-external” uses of Slavic verbal prefixes. Ottawa:
University of Ottawa dissertation.
Žaucer, Rok. 2012. The syntax of perdurative-prefixed verbs and the VP-internal/VP-external prefix
distinction. In Markéta Ziková & Mojmír Dočekal (eds.), Slavic languages in formal grammar, 339–355.
Frankfurt am Main: Peter Lang.
Zúñiga, Fernando & Seppo Kittilä. 2019. Grammatical voice. Cambridge: Cambridge University Press.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Genealogical overviews
Tim Thornes
15 Applicatives in Northern Uto-Aztecan
languages
Abstract: Applicative constructions in Northern Uto-Aztecan (NUA) languages, although
dissimilar in form, share many functional-semantic features and similar historical
developments. All NUA languages carry a verb suffix signaling the addition of a bene-
factive argument to the organic valence. In most cases, they represent the only means
for expressing a beneficiary as a core argument. Applicative suffixes in NUA appear
in most cases to originate from a ‘give’ verb. Although not always noted as such in the
available grammatical descriptions, NUA applicative constructions show causative-ben-
efactive syncretism associated with semantic verb type. The benefactive function is
quite consistent with transitive and active (agentive) intransitives, whereas with stative
(or patientive) intransitive verbs, the suffix functions as a causative. In languages that
have distinct applicative forms and morphological causatives, their frequency of use
still follows this pattern.

1 Introduction
The focus of this chapter is to present a typological overview of applicative construc-
tions (ACs) in the four genetic subunits that constitute the purported Northern branch of
the Uto-Aztecan language family (NUA). The languages of these subunits constitute two
genetic groupings, Numic and Takic, and two singletons within NUA, Pahka’anil (for-
merly known as Tübatulabal) and Hopi. Numic languages represented through exam-
ples in this paper include Northern Paiute, Shoshoni, Tümpisa, Southern Paiute, and
Ute, while Takic languages cited in what follows include Cupeño, Luiseño, Acjachemem,
Serrano, and Cahuilla. The unity of NUA itself as a genetic grouping remains in conten-
tion. Applying the rubric of “relative cognate density” Haugen, Everdell and Kuperman
(2020) find no clear evidence for NUA, in line with work by Miller (1984) using the more
traditional method of lexical correspondence. Others (for example, Manaster Ramer
1992) implicate NUA as a major branch of Uto-Aztecan.
Northern Uto-Aztecan languages occupy a mostly contiguous area that includes the
Great Basin region of the western United States, along the eastern slopes of the Sierra

Acknowledgments: I wish to express my appreciation to the editors, Fernando Zúñiga and Denis Creissels, for
their thoroughness in finding the errors and inconsistencies in this work, while offering brilliant suggestions
for improving it. I’d also like to acknowledge Albert Álvarez González, whose commentary helped immensely
to improve the work as well. Any remaining omissions or inconsistencies are my responsibility.

https://doi.org/10.1515/9783110730951-015
476 Tim Thornes

Nevada mountain range, and including southern California and northern Arizona to
western Colorado. Comanche, Shoshoni’s closest relative in Central Numic, is the only
language of the group found outside of this area, occupying portions of the southern
plains of Texas and Oklahoma.
Whether unified as a major branch of Uto-Aztecan, a set of four distinct subgroups
that parallel a southern branch (SUA), or something in between, it remains instructive
to make comparisons across NUA as contributions to our understanding of historical
developments within the family. There will be inevitable gaps in the discussion that
follows due to the nature and extent of the available grammatical descriptions and
searchable text data that pertain directly to properties of ACs across NUA. Beginning
in Section 2, we offer a brief overview of some of the key features of the languages in
question—verb structure (§ 2.1), participant coding (§ 2.2), voice (§ 2.3), and the sec-
ondary verb construction (§ 2.4)—in order to provide context for the description and
discussion of ACs. We will then proceed with a discussion of the morphological, syntac-
tic, and semantic properties of ACs (§ 3.1, § 3.2, and § 3.3, respectively) in each of the
four branches included under NUA. These discussions will necessarily include a look at
causatives, insofar as the two operations, causativization and applicativization, share
important formal, functional, and historical features.
Discourse-functional considerations are taken up in Section 3.4. A unique construc-
tion type in Northern Paiute that combines a denominalizing construction with applica-
tivization will be brought under consideration in Section 4. A summary overview of
the typologically significant properties of ACs in the NUA languages with questions for
further study is found in Section 5.
My own field experiences have informed an inevitable bias toward Numic (and
Northern Paiute, in particular). Fortunately, however, there are some excellent descrip-
tive materials for Takic (cf. Hill 2005; Hill and Hill 2019). Relevant materials drawn upon
for the present study for Pahka’anil and Hopi are limited mainly to Voegelin (1935) and
the Hopi Dictionary Project (HDP 1998) in addition to Jeanne (1971), yet even so, there is
enough material to draw upon for comparison both to contribute to our understanding
of developments within the family and to the interests of typology and grammaticaliza-
tion theory more generally.

2 Morphosyntactic character of the languages

Generally speaking, NUA languages have a rich morphology—mostly concatenating
and synthetic—particularly in the verb. They are strongly verb-final in constituent
order and associated features and are characterized by a clitic complex in second
position, consisting most typically of elements carrying pronominal information and
features of modality and evidentiality (Thornes 2018). The following sub-sections are
 15 Applicatives in Northern Uto-Aztecan languages 477

meant to illustrate some basic properties of NUA languages in order to shed light
on how these relate to the applicative constructions we encounter in this language
group.

2.1 Verb structure

The verb complex in the NUA languages includes a variety of morphologically marked
valence-changing operations, including those that increase valence, such as causatives
and applicatives. As a sample template, consider the following diagram of Northern
Paiute (Western Numic) verb structure (Thornes 2009):

(1) Object= [Phasal [[Valence [ip1 [Root]] Valence] dir/asp] sub/nmlz]

pro= aux Theme Stem Theme Prefinal Final

The Northern Paiute verb’s morphological inventory includes stem-forming affixes

known as “instrumental prefixes” (IPs) (a particularly salient derivational process in
the Numic subgroup2), valence-changing operators (both prefixes and suffixes) that
form the theme, an array of directional and aspectual (prefinal) suffixes, subordinat-
ing and nominalizing (final) suffixes, and a small set of phasal auxiliary elements that
appear between the direct object pronominal proclitic and the remainder of the verbal
complex. These elements express adverbial notions like ‘nearly’, ‘easily’, ‘only’, and
‘truly’.
Cupeño (Takic) shares much in common with Northern Paiute verb structure,
with some (mostly) minor differences, in particular the scattering of subject agree-
ment forms. Consider the following schema of the verb in Cupeño, adapted from Hill
(2005: 106):

(2) Object= (Subject') Root (Subject") Valence Tense-Aspect

pro= (sbj agr) Stem (sbj agr)3 Theme Final

By way of comparison, Northern Paiute object pronominals also appear as proclitics,

whereas there appears little trace of subject agreement as there is in Cupeño or in Hopi
(see below). The appearance of subject forms as both prefixes and suffixes in Cupeño is

1 The actual semantic contribution of instrumental prefixes to the resulting verb root can be quite
idiosyncratic.
2 Remnants of the instrumental prefix construction may be found in the other NUA languages, but
generally consist of lexicalizations.
3 By Hill’s analysis, there are two subject positions here, yet thus far I haven’t encountered a context
where both positions are filled in a single verb.
478 Tim Thornes

dependent upon formal verb class considerations described in Hill (2005). Within the
thematic and final zones of verb structure, in particular those relating to valence or
tense-aspect, there may appear one or more suffixes. The thematic suffixes include both
a morphological causative marker and a benefactive applicative. As elsewhere in NUA,
these always appear closest to the stem and in that order.
Verb structure for Pahka’anil is described in Voegelin (1935: 96–97) as consisting of
suffixes only, appearing in two groupings, medial and final. Included in the medial set
are the causative and benefactive suffixes in closest proximity to the verb stem, as with
Northern Paiute and Cupeño. As discussed below, these two valence-increasing devices
in Pahka’anil are near homophones and are favored with particular verb types. Final
suffixes include some tense and aspect forms, while, in common with Northern Paiute,
also are found subordinating and clausal nominalizing suffixes.
Bound pronominals in Pahka’anil, both nominative and accusative, are second posi-
tion enclitics, which in narrative (see, for example, the narratives presented in Marean
et. al. 2021) bind them to verb forms (following, of course, the final suffixes), and also to
adverbials. Where both are present, the nominative precedes the accusative pronoun.

2.2 Coding properties for grammatical relations

Considering word order, case-marking, and verb agreement, the NUA languages demon-
strate a robust nominative-accusative alignment for grammatical relations. The lan-
guages are verb-final (AOV/SV) in the majority of instances both in recorded, spontane-
ous speech and in responses to direct elicitation.
Case-marking shows some variation of expression. All NUA languages express three
case distinctions—nominative, accusative, and genitive, with some degree of formal
syncretism between the latter two. The nominative is most generally the unmarked
case. These distinctions are manifested in the form of pronouns and demonstratives
and, to a less predictable extent, through case suffixes on nominals and NP dependents.
Secondary cases in the form of bound postpositions express a range of locative, tempo-
ral, and instrumental semantic roles, among others.
The following examples demonstrate the variety of core case (nominative subject,
accusative object) formations in NUA. In Shoshoni (Central Numic), we find case marking
on both head nouns and their dependents. Consider:
 15 Applicatives in Northern Uto-Aztecan languages 479

(3) Gosiute Shoshoni (Miller 1996: 706)4

s-u-tɨn sɨmmɨn tu:pihtɨn punku nukkimia
prox5-that-nom one.nom black.nom horse[nom] running
‘That one black horse is running.’

(4) Gosiute Shoshoni (Miller 1996: 706)

nɨ u-kka sɨmmɨ-a punku-i puikka
I that-acc one-acc horse-acc see
‘I see that one horse.’

In (3) and (4), we see contrasting nominative and accusative case forms on the demon-
strative, numeral, and head noun, respectively. Nominative case is unmarked on nouns
as NP heads. Accusative object case suffixes on head nouns are varied and only partly
predictable on the basis of specific morphophonological properties. The reader is
referred to Miller (1996) and Crum and Dayley (1993) for further explication of allomor-
phic variation in the marking of syntactic objects in Shoshoni.
In Northern Paiute (Western Numic), we see a complete absence of distinct core
case suffixes on head nouns as one finds elsewhere in Numic. Instead, we find case-sen-
sitive determiner proclitics in a straightforward nominative-accusative alignment
pattern, as in the following examples.

(5) Northern Paiute: proclitics (Thornes 2003: 145)

su=panɨɨnadɨ tɨasɨ-pɨ
nom=lake freeze-perf
‘The lake is frozen.’

4 The Shoshoni examples from Miller (1996) and Dayley (1989) are represented phonemically to ac-
commodate four distinct consonant grades—simple unaspirated (lenis), geminate (fortis), prenasalized,
and preaspirated. Four grades of alveolar stop, for example, are written <t> (phonetically [d] or [ɾ]), <tt>
(phonetically [tː]), <nt> (phonetically [nd]), and <ht> (phonetically [ð]).
Northern Paiute has just two grades of obstruent—simple unaspirated (lenis) and voiceless gemi-
nate (fortis). “Lenis” forms are typically perceived as voiced (and lightly fricativized) in both languages,
especially word-medially, in which position they are in contrast with “fortis” in Northern Paiute. In this
paper, the lenis forms in Northern Paiute are represented as voiced and the fortis as voiceless obstru-
ents. Word-initially, consonant grades across Numic neutralize to simple voiceless.
Suffix-initial consonants are prone to variation across consonant grades depending upon a preced-
ing morpheme, whether affix or root, and have traditionally been termed “final features” whose effect
has scope only on the following consonant. Interestingly, the rare exception is the pan-Numic applicative
suffix, whose initial consonant grade is impervious to final feature affects.
5 Crum and Dayley (1993: 25) use the term proximate to refer to a referent functioning as a definite and
continuing topic.
480 Tim Thornes

(6) Northern Paiute: proclitics (Thornes 2003: 145)

su=tsɨaʔa ka=tɨpi mayɨ-u
nom=girl acc=rock find-pnc
‘The girl found the rock.’

The case-sensitive determiner proclitics, su= and ka=, do not attract stress, as do pre-
fixes, and are bound to the first element of a noun phrase.
Determiner proclitics most likely developed from contracted demonstrative forms
in Northern Paiute and may have arisen under the wide variation and erosion we witness
in core case-marking elsewhere in Numic. Case distinctions also appear on dependent
modifiers, however. The following examples illustrate the nominative-accusative align-
ment pattern instantiated in formal properties of noun phrase dependents.

(7) Northern Paiute: case suffixes (Thornes 2003: 149)

su=udɨ-ʔyu naatsi kima-u-gi-na
nom=tall-nom boy come-pnc-cisl-ptcp
‘The tall boy is coming this way.’

(8) Northern Paiute: case suffixes (Thornes 2003: 149)

nɨ u=punni ka=udɨ-u naatsi
1sg.sbj 3sg.obj=see acc=tall-acc boy
‘I see him, the tall boy.’

The conflation of determiner proclitics and modifier case suffixes results in the appar-
ent double-marking of nominative and accusative case in these examples. Subject pro-
nouns are independent words, whereas object pronominals usually appear as proclitics
attached to verbs, as in (8) above. These forms are identical to possessor pronominal
proclitics on possessed nouns as well as on nominalized verbs in some subordinate
clause constructions.
Recall from (2) above that the Takic language Cupeño has plural subject agreement
distributed over two (or three) distinct position classes in the verb depending upon verb
class distinctions in what appears to be a highly idiosyncratic system. As in Northern
Paiute, objects, indirect objects, and applied benefactive objects appear as verbal pro-
clitics in Cupeño. Consider:

(9) Cupeño: pronominal object proclitics (Hill 2005: 113)

Em-em=qwe=me chimi=mixaan me chimi=meqan-max
2pl.sbj-pl=can=2pl.erg 1pl.obj=do.hab and 1pl.obj=kill-ben.hab
hunwe-t pe’ aya chimi=tul-qa
bear-npn the now 1pl.obj=finish-prs
‘You all must do something for us, and kill for our sake the bear who is now
finishing us off.’
 15 Applicatives in Northern Uto-Aztecan languages 481

The first person plural proclitic appears in (9) to encode both direct and applied objects,
the latter licensed by the benefactive-applicative suffix, -max. As is also the case for
Northern Paiute, only one object proclitic can appear with a particular verb. In such a
case with two pronominal objects, the benefactive pronominal takes precedence.
In Hopi we find accusative case marking on both noun phrase heads and depend-
ents, like the demonstrative, in the following:

(10) Hopi (Hill 2013)

umu-mi pu-t maana-t peonakna
2pl-to that-acc girl-acc mention
‘(She) mentioned that girl to you.’

This is identical to the situation we saw in Shoshoni, where we find accusative case
affixes on both head nouns and modifiers in contrast to unmarked nominative forms.
Distinct case forms can also be illustrated for Pahka’anil, as in the following set of
examples.

(11) Pahka’anil (Voegelin 1935: 149–151)

yi’xpa’l leca’t
the.door.nom is.opening
‘The door is opening.’

(12) Pahka’anil (Voegelin 1935: 149–151)

ta’twa’l lecci’ina’t yipala’
the.man.nom is.opening.caus the.door.acc
‘The man is opening the door.’

(13) Pahka’anil (Voegelin 1935: 149–151)

ta’twa’l ha’maca’t
the.man.nom is.sad
‘The man is sad.’

These examples show distinct nominative and accusative case forms of nouns. Consider
the distinct forms for ‘door’ in (11) and (12), whereas the forms for ‘man’ in (12) and (13)
are identical as subjects of transitive and intransitive verbs, respectively.
Aside from object proclitics (recall the Northern Paiute and Cupeño examples
above), verb agreement is of limited distribution in NUA. In Pahka’anil, Voegelin (1935)
distinguishes free pronouns from those that operate as ‘conjunctive particles’, but
are perhaps best classed together with a variety of second position enclitics. Clitics in
second position carry a host of modal, evidential, and pronominal information in Pah-
ka’anil as they do in most other Uto-Aztecan languages.
482 Tim Thornes

In Hopi, only plural subject agreement marking is present. Note the following
examples from Hill (2013):

(14) Hopi (Hill 2013)

pu-t koyaanisqatsi-t peonakna-ya
that- acc corrupt.life-acc mention-pl
‘They brought up the matter of the corrupted life.’

(15) Hopi (Hill 2013)

soosovik sinom pas-va-ya
all.around people field-diff-pl
‘All the people around are farming (lit. are in the fields).’

According to Hill (2013), the -ya marker for plural subject is really an enclitic to the final
constituent of the clause. In (14), we see it following the verb, whereas in (15), it follows
a postpositional phrase serving as a verbless predicate, and so again it represents a
marginal case of verb agreement.

2.3 Valence and voice

As we have seen, grammatical relations in basic intransitive and transitive verbs show
nominative-accusative alignment through overt coding properties (word order, case
marking and verbal agreement). In ditransitive clauses, both non-agentive participants
are encoded as accusative case-marked noun phrases:

(16) Hopi (HDP 1998: 880)

taaqa wuuti-t sami-t maka
man woman-acc fresh.corn-acc gave
‘The man gave fresh corn to the woman.’

Derived ditransitives follow the same pattern of case marking on the two objects.
Compare the transitive clause in (17) with its derived causative counterpart in (18):

(17) Hopi (HDP 1998: 880)

nï’ hopilavayi-t tïtïqayi
1sg.sbj Hopi.language-acc learn
‘I am learning the Hopi language.’
 15 Applicatives in Northern Uto-Aztecan languages 483

(18) Hopi (HDP 1998: 880)

nï’ pahan-mï-y hopilavayi-t titiqay-na
1sg.sbj white.people-pl-acc Hopi.language-acc learn-caus
‘I am teaching (i.e. causing to learn) white people Hopi.’

Passive voice renders ditransitive verbs transitive as in the following:

(19) Hopi (HDP 1998: 881)

itam siiva-t mak-iw-ni
1pl.sbj money-acc give-pass-fut
‘We’re going to be given money.’

Throughout NUA, passive constructions operate equally well, whether with base objects
or applied objects as subject of passive:

(20) Northern Paiute (Thornes 2003: 307)

a. su=miidɨ nɨ-ka na-kuhani-kkɨ-wɨnɨ
nom=meat 1sg-acc pass-cook-appl-prog.sg
‘The meat is being cooked for me / (s.o.) is cooking the meat for me.’
b. ka=miidɨ nɨ na-kuhani-kkɨ
acc=meat 1sg.nom pass-cook-appl
‘I am being cooked the meat.’

Intransitive verbs are rendered impersonal, as in the following Hopi and Northern
Paiute examples:

(21) Hopi (HDP 1998: 881)

pay angqeʔ took-iwa
prt all.around go.to.sleep-pass
‘(They) have gone to sleep.’ / ‘Going to sleep took place.’

(22) Northern Paiute

unaʔyu na-tɨnikwɨhɨ
out.there pass-singing
‘(They) are singing out there.’ / ‘There is singing out there.’

Interestingly, an operational asymmetry appears with regard to the target of the anti-
passive construction, what has often been referred to as the unspecified object con-
struction in the literature on Uto-Aztecan languages.
484 Tim Thornes

(23) Northern Paiute

u-su i =tɨ-kuhani-kkɨ
3sg-nom 1sg.acc=apass-cook-appl
‘S/he’s cooking for me.’

(24) Northern Paiute

nɨ miidɨ kuhani-kkɨ-u-kwɨ
1sg.nom meat cook-appl-pnc-fut
‘I’ll cook meat for (someone).’

In (23), the applicative licenses the first person benefactive object and the patient object
is marked as unspecified by the antipassive prefix. On the other hand, when the bene-
factive applicative suffix in (24) is present the applied object can be unspecified without
being so marked.
As these examples illustrate, the antipassive construction appears to reveal syntac-
tic asymmetry with respect to which object is unspecified. An equally plausible analysis
would be that, in fact, the construction indicates that the patient role of a transitive
clause is unspecified. With organically ditransitive verbs like ‘give’ and ‘send’, we find
a similar asymmetry, whereby the target for under specification in the antipassive con-
struction appears to be the patient or theme (T), and not the recipient (R). We will revisit,
time and again, in this study what appears to be a pattern of operation that is controlled
more by verbal semantics than grammatical relations when it comes to valence change.

2.4 Secondary verbs

What are often referred to as “secondary verbs” in Uto-Aztecan languages actually

refers to a special complex predicate construction involving a subset of high frequency
motion and posture verbs (among others) that often develop broader aspectual and
directional properties in the construction. They appear somewhat scattered through-
out Uto-Aztecan (see especially Crapo 1970 for an overview of the phenomenon and
historical developments therein). Hill and Hill (2018) refer to the members of the sec-
ondary verb category as “non-conforming verbs”. Crum and Dayley (1993: 101–105)
refer to them in Western Shoshoni as “secondary auxiliary verbs” and describe them as
“forming loose compounds” with main lexical verbs. In at least one language, Northern
Paiute, secondary verbs can be readily distinguished from typical compounding struc-
tures (Thornes 2011), having more in common with Aikhenvald’s (2006) “asymmetrical
serial verb” construction. The importance of this construction in relation to the AC in
Numic is taken up below.
Secondary verbs form a unit with the main lexical verb by being incorporated
into the verbal complex. Several of these secondary verbs are transparently related
to various of the directional and aspectual morphology in these languages and are an
 15 Applicatives in Northern Uto-Aztecan languages 485

active zone for grammaticalization (Thornes 2009, 2011). Basic motion and posture
verbs fit into the secondary verb category in Numic (25) alongside the non-conforming
verbs (Hill and Hill 2019) in the Takic languages (26).

(25) Western Shoshoni (Crum and Dayley 1993: 102)

Teiten naiʔpi sukkuh yakai-katte
little girl there cry-sit.dur
‘The little girl was sitting there crying.’

(26) Cupeño (Hill and Hill 2019: 800)

kanaasta=kuʔut ajʔani-sh pym-jaw-nyq
basket=quot big-npn 3pl-carry-come.nfut
‘They brought a big basket.’

Important distinctions are made in Thornes (2011) between the secondary verb con-
struction (SVC) and typical verb + verb compounding in Northern Paiute (Western
Numic). These include the fact that, in traditional parlance, verb + verb compounding
is semantically right-headed, whereas in the SVC the main lexical verb precedes the
secondary verb (i.e. is left-headed). This analysis is supported by the fact that secondary
verbs always develop from a finite subset of verbs in the language and express such
grammaticalized functions as direction and aspect. Relevant for our discussion is the
observation that the position of the causative-applicative suffix in the language follows
the verb + verb compound in (27), while it comes between the main and secondary, as
in examples (28) and (29).

(27) Northern Paiute (Thornes 2011)

ɨ=sakwa ka=toissapui i=sami-tɨkɨ-kkɨ
2sg.nom=mod acc=chokecherry 1sg.obj=soak-put.sg-appl
‘You should put the chokecherries in to soak for me.’ (NK: ‘Chokecherries’)6

(28) Northern Paiute (Thornes 2011)

nɨ i=kɨkɨ ʔyui-kkɨ-katɨ
I 1sg.poss=feet be.warm-appl-sit.sg
‘I sit warming my feet (making my feet warm).’

6 The Northern Paiute data included in this paper, when possible, come from narrative text or however
they appear in the secondary sources I have consulted. Northern Paiute material was recorded, tran-
scribed, and analyzed by me with the help of native, first language speakers. The examples tagged as
narratives are coded with speaker initials and a short title. If not so tagged, the reader may assume the
example came from direct, usually text-based, elicitation.
486 Tim Thornes

(29) Northern Paiute (Thornes 2011)

yaisi ka=yuu “corner”-wai uuni-ku mɨ=wɨnɨ-kkɨ-tɨŋɨ-yakwi
then acc=this.way corner-loc that.kind-acc pl=stand.sg-appl-tell.to-hab
‘And this way, in the corner, that kind, would tell them to stand.’
(NK: ‘Boarding School Days’)

One can stipulate, based on this distribution, that the applicative suffix attaches to the
verbal stem portion of the verb structure (note the schema in [1] above), whereas the
secondary verb applies to a distinct layer of structure, the theme layer.

3 Applicative forms in Northern Uto-Aztecan

We turn now to a discussion and description at the morphological, syntactic, and seman-
tic levels of ACs as found in each of the four NUA subunits—Numic, Takic, Pahka’anil
(Tübatulabal), and Hopi. The ensuing discussion will naturally derive from a decidedly
Numic bias for the simple reason that some typological questions would otherwise
remain unanswered due to a lack of available data. Discussion of the phenomenon of
causative-benefactive syncretism and the historical connection between morphological
causatives and applicatives has been largely absent, for example, in most descriptions
of NUA languages.
The NUA languages each typically carry just one applicative suffix that signals
valence increase, mainly through the addition of an object argument whose featured
semantic role is benefactive (as well as malefactive). Some variation in semantic role
occurs as a natural consequence of distinct semantic properties of the verbs to which
they attach. The construction appears to be mostly obligatory across NUA, offering the
only means by which to express a benefactive argument. Minor exceptions are dis-
cussed in Section 3.2.
The ensuing sections summarize and exemplify the morphological, syntactic, and
semantic properties of the constructions, as well as discussion of some discourse-based
properties of the AC with evidence from Northern Paiute. We begin our comparative
discussion of NUA ACs with their most salient morphological properties.

3.1 Morphology

Canonical ACs in NUA appear exclusively to consist of a suffix closely attached to the
verb stem. There appears very little in the way of either lexically or phonologically
conditioned allomorphy. Indeed, the pan-Numic applicative suffix -ŋkɨ appears to be
 15 Applicatives in Northern Uto-Aztecan languages 487

resistant to the widespread lexically conditioned allomorphy (in the form of consonant
mutation) that affects most of the other suffixes.7
Applicative morphemes for each of the NUA languages sampled for this study are
listed in Table 1.

Table 1: Applicative forms in the four branches of Northern Uto-Aztecan and their origins.

Subgroup Language applicative give ✶

PUA
✶
Numic Northern Paiute -kkɨ8 ✶
kia
Tümpisa & Western Shoshoni -ŋkɨ(n)
Southern Paiute & Ute -(ŋ)kɨ
✶
Takic Cupeño -max ✶
maka
Luiseño & Acjachemem -max
Serrano -itʃuna ✶
(i)na
✶
Hopi Hopi -toyna ✶
(i)na
✶
Pahka’anil Pahka’anil -(a)na ✶
(i)na

Aside from the two GIVE verbs in Numic and Takic from which applicatives have clearly
grammaticalized in these languages, are three likely related forms in Serrano, Hopi, and
Pahka’anil. Note that all three carry a remnant of what has been proposed as a probable
Proto-Uto-Aztecan causative, ✶-(i)na (Langacker 1977: 145). The sporadic alternation of
[i] and [a] relates to transitivity, albeit inconsistently, across the Uto-Aztecan family.9
The denominalizing suffix –tu, similarly attested in Numic and Serrano as meaning,
broadly, ‘to make N’, could well account for the initial portions of the Hopi and Serrano
forms. These Serrano and Hopi forms can be related phonologically, with Hopi [t] corre-
sponding to the [tʃ] in Serrano by simple palatalization in the environment of high front
[i]. Further, Hopi [o] corresponds to [u] across the family.10
Applicative suffixes typically appear alongside morphological causatives in the
NUA languages which carry both. Where both are present as suffixes, the applicative

7 Consonant mutation is an important morphophonemic process in Numic, making reconstruction

difficult. There is both stem-internal contrast and lexically-induced gradation at most morpheme and
enclitic boundaries. Prenasalization in Central Numic (especially Shoshoni) corresponds with a some-
what unusual voiced geminate grade in the southernmost dialects of Northern Paiute, whereas in the
northern lects, illustrated here, prenasalization has merged with voiceless gemination, hence, -kkɨ. John
McLaughlin (i.p.c.) considers the prenasalised reflex to be the Proto-Numic form.
8 I present the Northern Paiute causative-applicative in all examples to follow in its underlying form to
emphasize the fact that its initial consonant does not mutate, but remains geminate (fortis) in all conditions.
9 Vocalic ablaut induced by certain suffixes on stems and its historical implications for Uto-Aztecan
historical study has long been recognized (cf. Heath 1977: 29–33 and Hill 2003: 42–46). In short, stem
final variation between [i] and [a] can sometimes have transitivizing effects.
10 I am grateful to Ken Hill (i.p.c.) for his insight and corroboration of the probability of the historical
relationships I am proposing here among these forms.
488 Tim Thornes

typically follows the causative morpheme, as is described by Voegelin (1935) for Pah-
ka’anil where it is stated of the benefactive suffix that it is used “after virtually all tran-
sitive verbal stems or verbal themes which are transitivized by means of the causative
suffix (104)”. Note:11

(30) Pahka’anil (Voegelin 1935: 101–103)

a. halai’- ‘to be wet (intr)’
b. halai’-ina-t ‘he is wetting him (making him wet)’
c. halai’-ina-ana-t ‘he is wetting him for her’

The following example from Cupeño illustrates this preference as well:

(31) Cupeño (Hill 2005: 264)

Ni=kwaw-nin-max=’em=pe ne-pulinma-y.
1.sg.acc=call-caus-appl=2.pl=irr 1.sg.poss-son-acc
‘You all will cause my son to be called for my sake.’ (Faye Creation 116)

The forms are varied across the four subgroups, with Pahka’anil and Hopi (and possi-
bly Serrano) appearing to preserve (at least in part) causative forms traceable to pro-
to-Uto-Aztecan.
The Northern Paiute applicative construction involves the reflex -kkɨ whose
appearance most typically signals the addition of a benefactive participant to the base
argument structure and bearing the grammatical relation of direct object. Compare:

(32) Northern Paiute

ɨ yadua-kwɨ
2.sg.nom talk.sg-fut
‘. . .you will talk/respond. . .’

(33) Northern Paiute

umɨ mi=yadua-kkɨ-kwɨ-u
2/3.pl.nom 1pl=talk.sg-appl-fut-pnc
‘. . .you all will talk/interpret for us. . .’ (NK: Boarding School Days)

The most likely source for the applicative suffix in the Numic sub-group can be found
in the still extant Northern Paiute verb kia meaning ‘give.’ According to Peterson (2007),
GIVE verbs are the most common source of benefactive applicatives, cross-linguistically.

11 It is unfortunate that Voegelin (1935) elected to display examples involving only third person singu-
lar participants—the unmarked form—thereby obscuring some of the key indicators of transitivity in
the examples.
 15 Applicatives in Northern Uto-Aztecan languages 489

The likely source construction for the grammaticalization of GIVE > APPLICATIVE
seems clearly, in Numic at least, but possibly across the Uto-Aztecan family, to be the
secondary verb construction, described above. Although not prototypical, it represents
an example of a one-word serial verb construction of Aikhenvald’s (2006) asymmetrical
type.12 My claim here is that the pan-Numic applicative suffix –(ŋ)kɨ has as a source a
still extant, trivalent verb kia, meaning ‘give’, which became grammaticalized as a par-
ticipant in the secondary verb construction. Consider:

(34) Northern Paiute (Thornes 2003)

yaɁa mii ɨni-na; u-su tɨ=sɨɨ-tsida u=kia
here quot say-ptcp 3-nom logo=willow-dish 3=give
‘“Here!” so replying; he gave her his willow cup.’ (ML: Bear and Deer)

That the historical source construction is likely as a secondary verb can be further illus-
trated with examples like the following:

(35) Northern Paiute (Thornes 2003)

i=yuŋa-kia
me=pick.up.a.container-give
‘Give that (container) to me!’ (alternatively, ‘Pick it up for me!’)

Note that in (35), the first person object proclitic is licensed by the secondary verb ‘give’,
in the same way as the applicative suffix would, by turning a bivalent verb ‘pick up’ into
a trivalent predicate complex.13 This represents a step in the direction of full grammat-
icalization—semantic bleaching—followed by phonological reduction and acategorial-
ity. Comparable examples in the more canonical serializing languages of west Africa
and southeast Asia can readily be found (cf. Lord, Yap, and Iwasaki 2002).
An applicative morpheme distinct from that found in the Numic languages is found
in Takic, but one which also appears to derive historically from a still extant lexical
verb meaning ‘give.’ In the case of the Takic subgroup, most of the languages, with the
exception of Serrano, have an applicative form -max transparently related to the extant
verb maxa ‘give’.14

12 Asymmetrical serial verb constructions are those whereby one verb in the series is from a restricted
(closed) class, or subset, of verbs, especially basic motion and posture verbs.
13 Note as well that there is only one participant coded overtly in this example, the first person singular
recipient (benefactive). The second person subject is understood, as is common in imperative construc-
tions, and the third person theme is unmarked, due to the recipient being prioritized in the proclitic
position.
14 Other Uto-Aztecan languages have maka, semantically narrowed to mean ‘give food; feed’, as in the
Hopi example (16) above. Note the main verb magha in the Ute examples (47–49) below.
490 Tim Thornes

(36) Cupeño (Hill and Hill 2019: 802)15

$aw~$aw-in-max=yn
plur~make.bread-tr-ben=1sg.acc
‘Make a few tortillas for me!’16

(37) Cahuilla (cited in Hill and Hill 2019: 831)

tuʔ tuʔ ne-ʔem-jaxe-max-am pe-n-paʔ-ka
grind grind 1sg.obj-2pl-do-ben-ipfv.pl 3pl.obj-1sg-drink-1sg.fut
‘Grind them for me so that I can drink them!’

Here, as elsewhere, the beneficiary is licensed by the applicative suffix and is the only
option for including a benefactive participant in the clause. This fact appears to be true
across NUA.
Serrano is an exception among Takic languages in that it is the suffix -ichun(a), that
signals the addition of a benefactive argument in an applicative construction (Hill and
Hill 2019: 712). Compare the following:

(38) Serrano (Hill and Hill 2019: 712)

a. icha-j ‘dip (water) b. ich-ichun ‘dip for’
c. kuuhan ‘call; invite’ d. kuuhan-ichun ‘call/invite for’

All three arguments of a derived ditransitive verb can appear as part of the second
position auxiliary complex in Serrano (Hill and Hill 2019: 534).

(39) Serrano (Hill and Hill 2019: 534)

amaʔ ymy-j=vy=chiʔ kuuhan-ichuna-qaʔ
dist 2sg.pro-acc=3sg.sbj=1pl.obj call-ben-ifut
‘He is going to call you for us.’

The applicative function in Pahka’anil, as we have been describing elsewhere in NUA,

is carried by a single verbal suffix -(a)na. This benefactive-applicative suffix occurs
with what Voegelin (1935) refers to as the “a-increment”, a vowel which may harmo-
nize under certain conditions. The only difference in form between the benefactive and
causative suffix is that the causative suffix -(i)na occurs with a distinct “i-increment”.
The following examples are meant to illustrate the transitivizing effect of both suffixes:

15 Hill and Hill (2019) use orthographic $ to represent what they refer to as a voiceless, apico-alveolar
fricative.
16 It is difficult to account for the -in transitivizer in this example other than to posit that it has lex-
icalized as part of the verb stem ‘to make bread.’ On its own it appears obviously related to the PUA
causative suffix (maintaining its pre-applicative position).
 15 Applicatives in Northern Uto-Aztecan languages 491

(40) Pahka’anil: applicative (Voegelin 1935: 102–103)

a. ha’ibi’ït ‘he is joking (intr)’
b. ha’ibi’-ana-t ‘he is teasing him’

(41) Pahka’anil: causative + applicative (Voegelin 1935: 102–103)

a. yu’udzat ‘it is fading (intr)’
b. yu’udz-ina-t ‘he is washing it (tr)’
c. yu’udz-ina-ana-t ‘he is washing it for him’

Interestingly, Voegelin (1935) dismisses the idea of a relationship between the two suffixes,
since “no generally discernable meaning can be otherwise assigned to the vowel incre-
ments (98)”. He places both of these suffixes in the positions closest to the verb root, as we
have seen in other NUA languages, with the causative appearing directly adjacent to it.

3.2 Syntax

Applied objects in NUA languages are basically symmetrical in character in compari-

son to their organic direct object counterparts. Applied objects can be promoted (i.e.
appear as subjects) in passive just as direct objects can. We find some constituent order
preferences, however, that seem clearly to privilege the applied object and speech act
participants as primary objects, appearing before secondary objects. In double object
constructions, where both objects marked with accusative (or oblique) case, we see this
constituent order preference as in the following examples from Hopi (HDP 1997: 880)
and Tümpisa Shoshoni (Central Numic; Dayley 1989: 184):

(42) Hopi (HDP 1997: 880)

nï’ pumu-y iqaʔo-y mokyaato-toyna
1sg 3pl-acc my.corn-acc wrap-appl17
‘I am wrapping up my corn for them.’

(43) Tümpisa Shoshoni (Dayley 1989: 184)

waʔippü tangumm-i tüpa-nna tukummahanni-ŋkü-nna
woman[nom] man-acc pinenut-acc cook-appl-ptcp
‘The woman is cooking pinenuts for the man.’

Note that in both examples, the applied benefactive object appears before the direct
object, although there remains some flexibility, at least in Tümpisa. Considering,
however, the fact that applied benefactive objects are typically human, there may be a

17 The parsing of the Hopi examples is my own (TT).

492 Tim Thornes

semantic or pragmatic motivation, as opposed to a syntactic one, underlying the word

order patterns we find.
Interestingly, Tümpisa allows applied objects to be postposed following the verb,
where they may optionally appear in the unmarked (nominative) case. Compare (44)
with (43) above:

(44) Tümpisa Shoshoni

waʔippü tüpa-nna tukummahanni- ŋkü-nna tangummü
woman[nom] pinenut-acc cook-appl-ptcp man[nom]
‘The woman is cooking pinenuts for the man.’

It would be helpful to have more prosodic information here to see whether the postpos-
ing of objects is common in discourse. Clauses don’t normally allow for two participants
to appear in the nominative case, suggesting that perhaps a postposed object should
rather be considered an adjunct to the main clause. Prosodic evidence might corrobo-
rate this hypothesis.
Both Ute (Southern Numic; Givón 2011) and Northern Paiute appear to be less flex-
ible when it comes to the coding of applied and direct objects. In the latter language,
however, any proposed object asymmetry can be attributed to a person hierarchy that
privileges speech act participants over third persons and humans over non-humans.
This is perhaps best illustrated by considering access to the object pronominal proclitic
slot in Northern Paiute. Consider:

(45) Northern Paiute

su=mokoʔni u-ka mi=kuhani-kkɨ
nom=woman 3sg-acc 1pl.excl=cook-appl
‘The woman cooked it/that for us.’

(46) Northern Paiute

ma-tu ɨmɨ ɨ=kia
3.prox 3pl 2=give
‘They’re giving you to him (as in marriage).’

As Givón (2011) demonstrates for Ute, when the benefactive applicative construction
operates on an organically transitive clause, the resulting ditransitive treats the applied
benefactive object differently, both syntactically and morphologically from the direct
object. Note:

(47) Ute (Givón 2011: 91)

ʔaa-ruachi-u magha-qa-amu
new-child.obj-pl feed-pst-3pl
‘(s/he) fed the babies’
 15 Applicatives in Northern Uto-Aztecan languages 493

(48) Ute (Givón 2011: 91)

mamachi ʔaa-ruachi-u magha-ku-qa-ʔu
woman new-child.obj-pl feed-appl-pst-3sg
‘(s/he) fed the babies for the woman’

Givón asserts here that the “word order is strict, with the benefactive invariably preced-
ing the patient object” (2011: 91). This is similar to what we have shown for Northern
Paiute, above. Note that object agreement also rests with the benefactive (third person
singular) rather than the patient (third person plural), which is shown to be ungram-
matical. Compare (48) with (49) below:

(49) Ute (Givón 2011: 91)

✶
mamachi ʔaa-ruachi-u magha-ku-qa-amu
woman new-child.obj-pl feed-appl-pst-3pl

These same restrictions apply to organically ditransitive clauses with verbs like ‘give’,
‘show’, and ‘tell’. Under the circumstances outlined here, therefore, there doesn’t appear
to be asymmetry between applied and direct objects.
Across NUA, the ACs appear to be obligatory—that is, there is no alternative,
non-applicative construction for encoding a benefactive argument. As Dayley (1989)
points out for Tümpisa, “[t]he process of forming benefactives from transitive verbs is
completely productive, and in fact it is obligatory whenever a benefactive participant is
involved in the action” (1989: 117).
The universality of this observation may not be entirely true for NUA languages,
however. Hill and Hill (2018: 328) provide the following pair of examples from Acjache-
mem, a Coastal Cupan language alongside Luiseño:

(50) Acjachemem (Hill and Hill 2018: 328)

maxannaʔ sopul punʔxanʔt-a kwamool-a
give.imp one peso-acc fisherman-acc
‘Give one peso to the fisherman!’

(51) Acjachemem (Hill and Hill 2018: 328)

ni-jk ngavvaʔ na-kchiijoʔ -ka-j
1sg-dat sharpen.imp 1sg-knife-poss-acc
‘Sharpen my knife for me!’

In (50), the organically trivalent verb ‘give’ licenses two accusative case-marked objects,
while the bivalent verb ‘sharpen’ in (51) marks the benefactive with the dative case.
This suggests that perhaps, in Acjachemem at least, the AC is not obligatory, as it is
in most other NUA languages. This merits further exploration. In elicitation Northern
Paiute speakers will sometimes offer examples like (52), where it is possible to find
494 Tim Thornes

beneficiaries marked with one of the locative cases, in this instance, the allative -tu,
without using the AC.

(52) Northern Paiute

ma-tu ɨ i =ma-tɨpuni
3.obv-loc you 1sg.obj=ip:by.hand-awake
‘You woke me up for him.’

This construction may in fact have been produced under the influence of English mainly
to accommodate the native English speaker present, me, in direct elicitation. It has not
to date been found anywhere in the natural speech corpus.

3.3 Semantics

The languages under consideration in this study each have just one marker functioning
as an applicative morpheme, as opposed to an array of markers specialized for bringing
distinct semantic participants into position as core arguments. The dominant semantic
role assigned to such status is that of benefactive across all the NUA languages, and ACs
are, by and large, the only means for expressing a benefactive participant. There is,
however, some inevitable variation in semantic role arising as a natural consequence of
the semantic properties of the host verbs, as in the ‘yell’ examples in (53) and (54) below.
Across the Numic subfamily are reflexes of the applicative suffix -(ŋ)kɨ, with
minimal variation and which behaves most like a benefactive-applicative suffix with
active-intransitive and most transitive verbs. Note the following examples from North-
ern Paiute and Tümpisa Shoshoni:

(53) Northern Paiute

wohi ‘yell’ wohi-kkɨ ‘call out to’
tɨmɨ ‘buy’ tɨmɨ-kkɨ ‘buy for’

(54) Tümpisa Shoshoni

tükümmüi ‘talk’ tükümmüi-ŋkün ‘interpret for’
paitsü ‘yell’ paitsü-ŋkün ‘yell at’

The most significant exception to this general exclusivity is apparent in the Numic
subgroup. Here we find causative-benefactive syncretism associated with the AC. With
what could be generally classified as stative and patientive intransitives, the suffix
signals the addition of an agent or cause to the derived valence—that is, -(ŋ)kɨ takes on
the function of a morphological causative.
 15 Applicatives in Northern Uto-Aztecan languages 495

(55) Northern Paiute

tɨʔoyai ‘be sick’ tɨʔoyai-kkɨ ‘make sick’
ɨwi ‘sleep’ ɨwi-kkɨ ‘make sleep’

(56) Tümpisa Shoshoni

kottoʔeh ‘boil (intr)’ kottoʔe-ŋkün ‘boil (tr)’
sünnünnüki ‘shiver’ sünnünnüki-ŋkün ‘shake’

Basic posture verbs (and patientive intransitives like wiʔiu ‘to fall’ and wiʔikɨu ‘to drop’)
are treated as stative intransitives, when it comes to the AC, in causative function.

(57) Northern Paiute

sɨsɨʔma u-kuba mi=wɨnɨ-kkɨ-u-ʔyai-na
sometimes 3sg.acc-upon 1pl.excl.acc=stand- appl-pnc-hab-ptcp
‘. . .sometimes (they) made us stand on that.’ (NK: Boarding School Days)

Agentive ambitransitives, like verbs of ingestion, follow the same pattern, but with
the flavor of a more mediated, or indirect, causative. The indirect reading could be
explained by the fact that such examples involve human causees.

(58) Northern Paiute

mi=tɨka-kkɨ-u-si nɨmmi tɨwau mia-si
1pl.excl=eat-appl-pnc-seq we.excl.nom again go.sg-seq
‘Having allowed us to eat, we went on again and. . .’(NK: Boarding School Days)

What is clear from the Northern Paiute data, but has not always been asserted in
descriptions of other Numic languages, is that causative-benefactive syncretism is not a
simple matter of the transitivity of the base verb. Rather, the semantics of the base verb
helps determine the outcome of the AC as either a morphological causative or benefac-
tive applicative. Tables 2 and 3 are a partial inventory of verbs that enter into the AC in
Northern Paiute, divided by semantic output.

Table 2: Verbs with which -kɨ has a benefactive-applicative function in Northern Paiute.

verb gloss applicative derivation

hani prepare; do hani-kɨ prepare for; give

hima take (in a container) hima-kɨ take/ get for
kuhani cook kuhani-kɨ cook for
kutsa split wood kutsa-kɨ split wood for
kwɨ[h]ɨ get; obtain kwɨ[h]ɨ-kɨ get for; acquire for
mabutuʔi roll (tr) mabutuʔi-kɨ roll for
matɨpuni wake up (tr) matɨpuni-kɨ wake up for
496 Tim Thornes

Table 2 (continued)

verb gloss applicative derivation

mɨda extend; stretch mɨda-kɨ extend for

nanisutɨhai pray nanisutɨhai-kɨ pray for
nisagwaiʔi curse nisagwaiʔi-kɨ be cruel to
noho roast under ashes noho-kɨ roast for
noo carry noo-kɨ carry for
pida build a fire pida-kɨ build a fire for
sɨta be bad; angry sɨta-kɨ be angry at
suaʔi laugh; smile suaʔi-kɨ laugh at; flirt with
tɨbo write (intr) tɨbo-kɨ write for
tɨmɨ buy tɨmɨ-kɨ buy for
tɨničui teach; tell stories tɨničui-kɨ tell stories to
tɨnikwɨhɨ sing tɨnikwɨhɨ-kɨ sing for
tsagwɨɨ pick up tsagwɨɨ-kɨ pick up for
tsikaʔa cut tsikaʔa-kɨ cut for
tsiʔwoŋi comb tsiʔwoŋi-kɨ comb for
wohi yell wohi-kɨ yell at; call out to
yadua talk.sg yadua-kɨ interpret
yuŋa dip into yuŋa-kɨ scoop up

Table 3: Verbs with which -kɨ has a causative function in Northern Paiute.

verb gloss causative derivation

čɨčɨkwi sit.distr (intr) čɨčɨkwi-kɨ put up.distr (tr)

kakia wear around neck kakia-kɨ have/ make wear
katɨ sit.sg (intr) katɨ-kɨ set/put up.sg (tr)
kwissi weave kwissi-kɨ lasso
mani do mani-kɨ cause to be; occur
maŋi cross water; wade maŋi-kɨ take across
nai burn (intr) nai-kɨ burn (tr)
nɨɨma feel; be injured nɨɨma-kɨ make hurt; injure
pakomamaʔi wash the face pakomamaʔi-kɨ make/ have wash
patakwitsia shine (intr) patakwitsia-kɨ shine (tr)
patawi blow up; stir up (intr) patawi-kɨ blow up (tr)
pisa be good pisa-kɨ cure; make well
pitɨ arrive pitɨ-kɨ bring
punni see punni-kɨ show
tiʔoyai be sick tiʔoyai-kɨ make sick
tɨka eat tɨka-kɨ make/have eat
tɨmanaga be paid tɨmanaga-kɨ pay
toki be correct toki-kɨ do correctly
wadzi hide wadzi-kɨ lose
wakwami stand.distr (intr) wakwami-kɨ stand.distr (tr)
wɨnɨ stand.sg (intr) wɨnɨ-kɨ stand.sg (tr)
 15 Applicatives in Northern Uto-Aztecan languages 497

Table 3 (continued)

verb gloss causative derivation

yuai stop (intr) yuai-kɨ stop (tr)

ʔɨdɨtɨ be hot ʔɨdɨtɨ-kɨ make hot
ʔɨwi sleep ʔɨwi-kɨ make sleep
ʔyui be warm ʔyui-kɨ make warm

Clearly, the semantics of the base verb drives the output of the AC in Northern Paiute.
Dayley (1989), in his grammar of Tümpisa Shoshoni, draws near to that conclusion by
dividing intransitives into “statives and actives”, but provides little evidence to demon-
strate that they behave either similarly or differently, despite the fact that nearly all in
his list of base verbs entering into the causative construction are stative or patientive
intransitives. His sample set of stative and patientive intransitive base verbs includes
what is in Table 4:

Table 4: Verbs with which -ŋkün has a Causative Function in Tümpisa Shoshoni.

hapi” ‘lie.sg’ hapi-ŋkün ‘to make fall’

Hipittain ‘to get drunk’ hipittai-ŋkün ‘to intoxicate’

hüttsawü ‘to be cool’ hüttsawü-ŋkün ‘to cool’
kammah ‘to be sick’ kamma-ŋkün ‘to make sick’
kuppä ‘to cook (intr) kuppä-ŋkün ‘to cook (tr)’
kotto’eh ‘to boil’ kotto’e-ŋkün ‘to boil (tr)’
kwitasu'u" ‘to fart’ kwitasu’u-ŋkün ‘to make fart’
mi’a ‘to go’ mi’a-ŋkün ‘to send’
pitsi ‘to suckle’ pitsi-ŋkün ‘to nurse’
potso’in ‘to be wet’ potso’i-ŋkün ‘to moisten’
sii” ‘to urinate’ sii”-ŋkün ‘to make urinate’
sünnünnüki ‘to shiver’ sünnünnüki-ŋkün ‘to shake’
tammayain ‘to be crazy’ tammayai-ŋkün ‘to drive crazy’
tamminoi ‘to be tired’ tamminoi-ŋkün ‘to make tired’
üitsü’in ‘to be cold’ üitsü’i-ŋkün ‘to make cold’
üppüih ‘to go to sleep’ üppüi-ŋkün ‘to make sleep(y)’
ütüin ‘to be hot’ ütüi-ŋkün ‘to heat’
watsi” ‘to be lost’ watsi-ŋkün ‘to lose’
wükkatüh ‘to be a pile’ wükkatü-ŋkün ‘to stack’
yuhupükkan ‘to get fat’ yuhupükka-ŋkün ‘to fatten’

Benefactive applicatives throughout NUA appear to be most productively associated

with transitive base verbs and fairly productive with agentive intransitives. Caus-
ative-benefactive syncretism is found in Numic and thoroughly attested in Northern
Paiute (Thornes 2003: 383–387) as well as many other languages of North America and
498 Tim Thornes

around the world—for example, Hualapai (Ichihashi-Nakayama 1996) and several lan-
guages of Australia (Austin 1997 and Austin, this volume).
Even in those NUA languages with distinct benefactive and causative forms, such as
Cupeño, Pahka’anil, and Hopi, there appears the same division of labor—benefactives
are productive with transitive verbs and causatives are rare with transitives but pro-
ductive (to the point of lexicalization) with stative intransitives, as we will see below.
The nature of causative-benefactive syncretism is key for understanding the impacts of
ACs in NUA languages, and so information on causation and causative morphology has
been included here.
I turn now to some interesting discourse-related properties of the AC. The focus
here will be on Northern Paiute, but can be seen as instructive for this and future
surveys of related and neighboring languages. More thorough studies are needed of
the other NUA languages, not only to seek confirmation (and clarity) of what has been
found in this one, but also to provide more clues as to the development of causatives
and applicatives across the family.

3.4 Discourse

Some of the Northern Paiute data provide evidence whereby core arguments are rear-
ranged in the AC, but without valence increase, in a way that could indicate a boost in
topicality of the participant(s) encoded in the applied phrase. Consider the following
examples with the transitive verb meaning ‘to tell stories to’. Example (59) illustrates it
as the main verb in a base transitive construction. Looking at example (60) from an eth-
nohistorical narrative, we have the speaker and her family meet with others to gather
wild roots. She and her family are central to the narrative.

(59) Northern Paiute

ɨmi i=tɨničui
2pl.sbj 1sg.obj=tell.story
‘You tell me stories.’

(60) Northern Paiute

nɨmmi pokwa-na, umɨ mi=tɨničui-kkɨ-u
1pl.excl.nom lie.pl-sim 3pl.sbj 1pl.excl.obj=tell.story-appl-pnc
‘While we lay down, they told us stories.’ (NK: Root-digging Time)

One can speculate that the AC in (60) is functioning as a kind of voice construction by
indicating the importance or centrality of the speaker and her family’s experience. Taken
together, these examples illustrate that the valence does not change, either with the
applicative suffix or without it. Alternatively, the benefactive semantics of the applica-
 15 Applicatives in Northern Uto-Aztecan languages 499

tive suffix may be on display here in conjunction with an ethical dative function we find
elsewhere in NUA.18
As far back as Sapir (1930) is suggested an “ethical dative” sense for the benefac-
tive-applicative function in Southern Paiute (Southern Numic)—a function in which the
“indirect (sic) object is not really affected by the action at all but is merely interested
in it. Such ethical datives with first person indirect object are frequently employed to
indicate an affectionate attitude on the part of the speaker” (1930: 145).
It turns out to be a simple matter, with certain verbs, to elicit more of Sapir’s “dative
of interest” scenarios that are acceptable to elder Northern Paiute speakers. Note:

(61) Northern Paiute

i=nodɨkwa nɨ mia-kkɨ-si
1sg.poss=wife 1sg.nom go.sg-appl-seq
‘My wife left me (~I let her go).’

Of interest here is the fact that the affected first person appears in its nominative case
form. By default, then, the notional subject of the base intransitive verb is accusative, as
there can only be one nominative form per clause in Northern Paiute. The parenthetical
alternative interpretation of (61) would cast the AC in its permissive causative function.
However, compare:

(62) Northern Paiute

i=hammaʔa nɨ yaʔi-kkɨ-si
1sg.poss=elder.sister 1sg.nom die.sg-appl-seq
‘My big sister died on me.’

In (62), however, a parallel alternative seems faulty, since ‘I’ did not cause my elder sis-
ter’s death, but was obviously affected by it. The following examples present a similar
conundrum for analysis. Consider example (63) from a conversation about a man that
had recently lost his wife. Here, the intransitive subject of the first clause (the woman)
is presented morphosyntactically as the applied phrase with accusative case-marking.
As an alternative, the accusative case-marking on ‘woman’ could be a reflection of the
dependent nature of the sequentially marked clause.19 Indeed, the accusative marker
here could be interpreted as marking the dependent clause as a whole, as though it
were itself an argument of the independent clause that follows it. At this point in the
conversation, the speaker could be seen as emphasizing the affectedness/interest of the
nominative argument (the man) in the second clause:

18 I would like to thank Denis Creissels for this observation.

19 Thanks to Albert Álvarez González for this observation, based upon similar phenomena in other
Uto-Aztecan languages.
500 Tim Thornes

(63) Northern Paiute

ka=mokoʔni yaʔi-kkɨ-si, su=nana pɨ=sɨʔmɨ nobi-ča-ʔyu
acc=woman die-appl-seq nom=man restr=only house-have-pred
‘When that woman died, the man lived alone.’ (JB: Driving to Drewsey)

When offered a grammatically and pragmatically neutral alternative—a simple

sequence of two intransitive (non-applicative) clauses with both subjects case-marked
nominative, it was deemed perfectly acceptable to speakers, but without the same
emphasis:

(64) Northern Paiute

su=mokoʔni yaʔi-si, su=nana pɨ=sɨʔmɨ nobi-ča-ʔyu
nom=woman die-seq nom=man restr=only house-have-pred
‘When the woman died, the man lived alone.’

Examples like these appear to support a discourse-centered function of the AC in North-

ern Paiute. That is to say, once again, that ACs do more than transitivize. Transitiviza-
tion is a side-effect of the basic function of ACs, namely, to assign discourse prominence
to otherwise peripheral (or, at least, affected) arguments. In NUA languages, and no
doubt elsewhere, the participant role being given such prominence is entirely depend-
ent upon the semantics of the verb of the base construction. More research is needed to
account for such discourse effects.
A more covert property of the AC in Northern Paiute is revealed in consideration
of its high frequency of use in some first person narratives, belying a stronger interest
on the part of the speaker. A narrative offered by one speaker described her residential
boarding school experience. Its high frequency of ACs (nearly two dozen) merits con-
sideration, since one would otherwise rank the AC as a rather infrequent construction
type. Clearly, its frequency in this narrative is related to the fact that the key players, the
speaker and her elder sister, were not in control of the events that unfolded for them
(or to them) in the boarding school context. Any further exploration of the discourse
context(s) that trigger the use of the AC lies beyond the scope of the present study.

4 Denominalization and applicativization

in Northern Paiute
One phenomenon that has to now not received much (if any) attention in the available
descriptions of NUA languages is the presence of an applicative-like structure associ-
ated exclusively with denominal verbs. This phenomenon is present in Northern Paiute.
Denominalization is found across the Uto-Aztecan language family (Haugen 2008) and
involves deriving a verb from a noun using one of several suffixes with meanings
 15 Applicatives in Northern Uto-Aztecan languages 501

ranging from MAKE, HAVE, WEAR, and HUNT/GATHER, to name a few. The following
examples provide a sample from NUA:

(65) Hopi: MAKE (Hill 2003: 225, cited in Haugen 2008)

kii-ta
house-make
‘build a house’

(66) Northern Paiute: HAVE (Thornes 2003: 131)

pisa miawo-gaʔyu
good knee-have
‘to have good knees (idiom. “to have a knack for showing up at mealtime”)’

(67) Southern Paiute: PUT ON (Sapir 1930: 135, cited in Haugen 2008)
maavï-ai
clothing-put.on
‘puts on (his) clothes’

(68) Pahka’anil: GATHER (Voegelin 1935: 132, cited in Haugen 2008)

mac-kay-ïlat
wild.oats-gather-go
‘He is going to gather wild oats.’

The MAKE form in Northern Paiute is -tu and can have a secondary derivation that
allows for the addition of a benefactive participant with the addition of -ʔi, as in the
following examples:

(69) Northern Paiute

a. nɨ kopiʔi-tu
1sg coffee-make
‘I made coffee.’
b. nɨ ɨ=kopiʔi-tu-ʔi
1sg you=coffee-make-ben.appl
‘I made coffee for you.’

(70) Northern Paiute

a. watta kutia-tu
pole fence-make
‘make a pole fence’
b. umɨ i=kutia-tu-ʔi
3pl me-fence-make-ben.appl
‘They’re making a fence for me.’
502 Tim Thornes

Note that the valence shift is just as it is with the more canonical benefactive-applica-
tive constructions we have been exploring up to now. Here we see the valence increase
through the addition of a human beneficiary of the action coded by the denominal verb.
The -ʔi suffix has no other known function elsewhere in the language, and in all exam-
ples in my corpus, I only find it in consort with the –tu denominalizer. That said, it
appears to be fully productive, but only in this particular denominalizing context.
As mentioned elsewhere in this paper, there are, throughout the Uto-Aztecan lan-
guage family, sporadic alternations between final vowels [a] and [i] patterning with
transitivity. As described in Thornes (2013: 254–255), one finds at least a couple of lexi-
calized examples of this alternation aligning [a] with intransitive and [i] with transitive.
Note:

(71) Northern Paiute (Thornes 2013: 254–255)

a. yadua b. yadui
‘talk’ ‘talk to X’

(72) Northern Paiute (Thornes 2013: 254–255)

a. timadza’a b. timadzai
‘help (intr)’ ‘help X (tr)’

The path from causative to general transitive is a well-trodden one. That it found its way
to an applicative function, and to a particular destination where it could flourish seems,
at least on the face of it, not far-fetched. Clearly, more work needs to be done in order to
explore constraints on the benefactive-applicative function of the -ʔi suffix in Northern
Paiute and its presence elsewhere in the family.
Applicativization restricted to verbs formed by means of a particular deriva-
tional operation is rare. A possible historical scenario20 harkens to a time when the
-tu denominalizer was still an independent verb and -Ɂi was a productive applicative
suffix. Replacement by another applicative suffix may have corresponded with the
persistence of -tu only in Noun + Verb compounds and the lexicalization of -Ɂi. As we
have already seen, the association of [i] with transitivity and transitivization is well
supported across the family. Further support can be reconstructed in our analysis of
the Hopi and Serrano forms from Table 1. Across NUA, we see the bits and pieces of
causative-applicative developments. An additional such development, briefly explored
here, may have persisted in this particular denominalizing context, whose historical
development would look something like: ✶[N + V-appl] > [N-denom-appl].

20 I thank Denis Creissels for this observation.

 15 Applicatives in Northern Uto-Aztecan languages 503

5 Summary and conclusions

This overview of applicatives and applicative constructions in Northern Uto-Aztecan
languages is necessarily incomplete, but is hoped to provide some needed direction for
undertaking further study. Here I summarize the following areas and features of var-
iation across the languages as well as the connections, both historical and typological,
between them.

Morphology
– Each NUA language carries only a single suffix, adjacent to the verb stem, that
is clearly identifiable as an applicative. Allomorphy is minimal, and in Numic in
particular, the applicative morpheme appears resistant to the consonant mutation
processes one expects with most verb suffixes. In languages with separate causa-
tive and applicative morphemes, the causative appears closer to the root than the
benefactive applicative.
– The distribution of the applicative in Numic distinguishes stem-compounding from
secondary verb constructions by following the former and preceding the latter.
The languages of NUA demonstrate quite transparent origins of applicatives from
verbs meaning ‘give’, at least for Numic and Takic. Some form of productive verb-
verb compounding served as a likely vehicle for the development of GIVE into an
applicative suffix.

Syntax
– NUA applicatives are typically obligatory applicatives.
– Applied objects appear to carry the same privileges of operation as organic, direct
objects. That is to say, there does not appear to be any restrictions on the appear-
ance of the applicative in combination with other voice or valence-altering opera-
tions like the passive and antipassive.
– In Northern Paiute, at least, it would appear that the applicative could be consid-
ered first in order of operation—that is, the passive and/or antipassive effect(s)
are on the applied object primarily. We have also seen that there are constituent
ordering patterns and access to registration on the verb that prefer speech act par-
ticipants or human objects—not driven by the syntax.

Semantics
– Applicatives in NUA languages appear mainly to signal the addition of a syntac-
tic object playing the semantic role of beneficiary with most transitive and active
intransitive verbs. In the Numic languages, the same morpheme operates like a
causative with mostly intransitive stative and patientive verbs, signaling an addi-
tional external cause or agent to the event frame.
504 Tim Thornes

– In other NUA languages, a morphological causative is distinct from the benefac-

tive applicative, but the two combine mainly along lines of the same division of
semantic verb classes and bearing the same functional split as we have seen with
the single form causative-benefactive morpheme we see throughout Numic. The
morphological causative in Pahka’anil, for example, is said to appear mainly with
stative intransitives and “only rarely” with transitive stems, whereas the benefac-
tive appears productively with transitive and active verbs.

Discourse
– Evidence from naturally occurring speech has been found that suggests a possi-
ble discourse basis for the development and distribution of the applicative and the
outcome of the AC. Such evidence has been found in Northern Paiute by looking at
the discourse prominence or topicality of applied objects in parallel constructions
and the pervasiveness of the AC in certain autobiographical narrative contexts.
– These observations, if corroborated elsewhere, could be seen as supporting the
idea that ACs encode topical participants that are not present organically. The role
of these participants is determined by the semantics of the main verb. Traces of the
historical relationship of the applicative suffix to a GIVE (secondary) verb, when
combined with a particular main (primary) verb can be felt in the output of the
construction.
– As a derivational process, applicatives combine to create new, if related, lexical
items. In NUA languages, the relationship between causative and applicative follows
two paths. In one, causative and applicative morphemes are formally distinct,
and are most productive with particular verb types, generally centering around
whether the verbs are stative (non-active) or active and only incidentally relating
to transitivity. In the other, causative and applicative functions are carried by the
same form and are fully productive, but their semantic output is also dependent
largely upon the same stative-inactive versus active semantic division of the verbs
with which it combines.

Abbreviations
acc accusative
agr agreement
appl applicative
apass antipassive
asp aspect
aux phasal auxiliary
ben benefactive
caus causative
cisl cislocative
 15 Applicatives in Northern Uto-Aztecan languages 505

dat dative
diff diffuse
dir directional
dist distal
distr distributive
dur durative
erg ergative
excl exclusive
fut future
hab habitual
imp imperative
intr intransitive
ip instrumental prefix
ipfv imperfective
loc locative
logo logophoric possession
mod modal
nfut non-future
nom nominative
nmlz nominalizer
npn non-possessed noun
obj object
obv obviative
pass passive
perf perfect
pfv perfective
pl plural
plur pluractional
pnc punctual
poss possessive
pred predicative
pro pronominal
prox proximate
prs present
prt particle
pst past
ptcp participle
quot quotative
restr restrictive pronominal
sbj subject
seq sequential
sg singular
sim simultaneous
sub subordinator
tr transitive
trnsl translocative
- default affix boundary
~ reduplicative boundary
= clitic boundary
506 Tim Thornes

References
Aikhenvald, Alexandra Y. 2006. Serial constructions in typological perspective. In Alexandra Y. Aikhenvald
& R. M. W. Dixon (eds.), Serial verb constructions: A cross-linguistic typology, 1–68. Oxford: Oxford
University Press.
Austin, Peter. 1997. Causatives and applicatives in Australian Aboriginal languages. In Kazuto Matsumura &
Tooru Hayasi (eds.), The dative and related phenomena, 165–225. Tokyo: Hituzi Syobo.
Crapo, Richley H. 1970. The origin of the directional adverbs in Uto-Aztecan languages. International Journal
of American Linguistics 36(1). 181–190.
Crum, Beverly & Jon Dayley. 1993. Western Shoshoni grammar. Boise: Boise State University.
Dayley, Jon P. 1989. Tümpisa (Panamint) Shoshone grammar. Berkeley: University of California Press.
Haugen, Jason D. 2008. Denominal verbs in Uto-Aztecan. International Journal of American Linguistics 74(4).
439–470.
Haugen, Jason, Michael Everdell & Benjamin A. Kuperman. 2020. Uto-Aztecan lexicostatistics 2.0.
International Journal of American Linguistics 86(1). 1–30.
Heath, Jeffery. 1977. Uto-Aztecan morphophonemics. International Journal of American Linguistics 43(1). 27–36
Hill, Jane H. 2005. A grammar of Cupeño. Berkeley: University of California Press.
Hill, Jane H. & Kenneth C. Hill. 2019. Comparative Takic grammar. Berkeley: University of California Press.
(https://escholarship.org/uc/item/6tr732gg)
Hill, Kenneth C. 2013. Verbless predicates in Hopi. Paper presented at the Friends of Uto-Aztecan
Conference, Boise, Idaho.
Hopi Dictionary Project. 1998. Hopi dictionary / Hopìikwa Lavàytutuveni: A Hopi-English dictionary of the Third
Mesa dialect. Tucson: University of Arizona Press.
Ichihashi-Nakayama, Kumiko. 1996. The “applicative” in Hualapai: its functions and meanings. Cognitive
Linguistics 7(2). 227–239.
Jeanne, Laverne Masayesva. 1971. Aspects of Hopi grammar. Cambridge, Massachusetts: Massachusetts
Institute of Technology dissertation.
Langacker, R. W. 1977. Studies in Uto-Aztecan grammar, Vol. 1: An overview of Uto-Aztecan grammar. Arlington,
Texas: SIL Publications.
Lord, Carol, Foong Ha Yap & Shoichi Iwasaki. 2002. Grammaticalization of “give:” African and Asian
perspectives. In Ilse Wischer & Gabriele Diewald (eds.), New reflections on grammaticalization, 217–235.
Amsterdam: John Benjamins.
Manaster Ramer, Alexis. 1992. A Northern Uto-Aztecan sound law: -c- > -y-. International Journal of American
Linguistics 58(3). 251–268.
Miller, Wick R. 1984. The classification of the Uto-Aztecan languages based on lexical evidence. International
Journal of American Linguistics 50(1). 1–24.
Miller, Wick R. 1996. Sketch of Shoshone, a Uto-Aztecan language. In Ives Goddard (ed.), Handbook of North
American Indians, Vol. 17: Languages, 693–720. Washington: Smithsonian Institution.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Sapir, Edward. 1930. Southern Paiute: a Shoshonean language. American Academy of Arts and Sciences 65.
1–296. [Reprinted in William Bright (ed.), The collected works of Edward Sapir X: Southern Paiute and Ute
linguistics and ethnography. New York: Mouton de Gruyter.]
Thornes, Tim. 2003. A Northern Paiute grammar with texts. Eugene: University of Oregon dissertation.
Thornes, Tim. 2009. Historical pathways in Northern Paiute verb formation. In Patience Epps & Alexandre
Arkhipov (eds.), New challenges in typology: Transcending the borders and refining the distinctions,
295–320. Berlin: Mouton de Gruyter.
Thornes, Tim. 2011. Dimensions of Northern Paiute multi-verb constructions. In Pieter Muysken &
Alexandra Y. Aikhenvald (eds.), Multi-verb constructions: A view from the Americas, 27–61. Leiden: Brill.
 15 Applicatives in Northern Uto-Aztecan languages 507

Thornes, Tim. 2013. Causation as “functional sink” in Northern Paiute. In Tim Thornes, Erik Andvik,
Gwendolyn Hyslop & Joana Jansen (eds.), Functional-historical approaches to explanation: In honor of
Scott DeLancey, 237–258. Amsterdam: John Benjamins.
Thornes, Tim. 2018. Evidentiality in the Uto-Aztecan languages. In Alexandra Y. Aikhenvald (ed.), The Oxford
handbook of evidentiality, 409–430. Oxford: Oxford University Press.
Voegelin, Charles F. 1935. Tübatulabal grammar. University of California Publications in American Archeology
and Ethnology 34(2). 55–190.
Albert Álvarez González and Zarina Estrada-Fernández
16 Applicative constructions in Uto-Aztecan
languages from Northwestern Mexico
Abstract: This chapter proposes a description of applicative constructions in five
Uto-Aztecan (UA) languages (Pima Bajo, Northern Tepehuan, Tarahumara, Guarijío
and Yaqui) spoken in Northwestern (NW) Mexico. Two main types of applicative can be
distinguished in each language. The first type is an optional peripheral applicative, in
which the applied phrase refers to a non-core participant (usually, a beneficiary) that
has been promoted to the object position comparing to the base construction, in which
it can be encoded as an oblique (usually, introduced by a benefactive postposition). The
second type is an obligatory central applicative, in which the applied phrase refers to a
central (essential) non-patient participant in the event denoted by the verb (recipient/
goal, stimulus, source, etc.); this participant cannot be encoded at all in the construction
with the underived form of the same verb. In both types, the applied object is usually a
human participant. After exploring some aspects of the syntax, the semantics and the
pragmatics of applicativization in these languages, the study also presents the causa-
tive/applicative syncretism found in the data, and discusses some cases of applicative
lookalikes (lexicalized applicatives, applicative deponents).

1 Introduction
This chapter proposes a description of applicative constructions in five Uto-Aztecan
(UA) languages (Pima Bajo, Northern Tepehuan, Tarahumara, Guarijío and Yaqui)
spoken in the states of Sonora and Chihuahua in Northwestern (NW) Mexico. These
languages belong to the Sonoran sub-group of Southern Uto-Aztecan languages (Miller
1984); two are from the Tepiman branch (Pima Bajo and Northern Tepehuan) and three
from the Taracahitan branch (the two closely related Tarahumara and Guarijío, and
Yaqui). According to INEGI (2020), there are approximately 1,037 speakers of Pima Bajo
(PB), 9,855 of Northern Tepehuan (NT), 91,554 of Tarahumara (TA), 2,139 of Guarijío
(GU), and 19,376 of Yaqui (YQ).
The data for this study come from reference grammars (Brambila 1953 for Noro-
gachi TA, Burguess 1984 for Western TA, Dedrick and Casad 1999 for YQ, Miller 1996
for Mountain GU, Félix 2005 for River GU, Bascom 1982, 2003 for NT, Estrada 2014 for
PB) and dictionaries (Brambila 1976 for Norogachi TA and Estrada et al. 2004 for YQ),
as well as from previous studies directly or indirectly related to the study of applicative
constructions in the UA languages considered here (Caballero 2008 for Choguita TA,
Félix 2007, Medina 2002, Ávila 2012, Casas 2018 for GU, Álvarez 2007, Guerrero 2007,

https://doi.org/10.1515/9783110730951-016
510 Albert Álvarez González and Zarina Estrada-Fernández

Harley et al. 2009, Estrada-Fernández et al. 2015, Álvarez 2019, and Guerrero 2022 for
YQ, Estrada 2007, Ramírez 2010 for PB, Ramos 2010, and Carrillo 2013 for NT).1
Following the definition proposed in the position paper (Zúñiga and Creissels, this
volume), applicative constructions are regarded here as constructions in which a mor-
phologically derived verb form assigns a syntactic role other than S and A—usually, a
role as direct object or P (DIRECT/PRIMARY APPLICATIVES or P-APPLICATIVES), but also,
although less frequently, as indirect object or D (INDIRECT/SECONDARY APPLICATIVES
or D-APPLICATIVES) and as oblique role or X (OBLIQUE APPLICATIVES or X-APPLICA-
TIVES)—to a participant (the APPLIED PHRASE), which in the base construction either
requires a non-core coding different from its coding in the applicative construction
(OPTIONAL APPLICATIVES) or cannot be expressed at all (OBLIGATORY APPLICATIVES).
Since the applied object is usually an additional participant towards which the
activity of the referent A is directed, P-applicativization is a way to encode a type of
functional alternation named the undirected/directed alternation (Creissels, forthcom-
ing), which is represented by an undirected/directed verb pair in which the subject of
the undirected verb form and the directed verb form is the same, but the directed form
has an additional object.
P-applicativization represents only one of the possible strategies for encoding the
undirected/directed alternation, namely a morphologically oriented way with the use
of a privative marking on the undirected base verb. The other possibilities are ANTIPAS-
SIVIZATION (another morphologically oriented alternation with a privative marking
this time on the directed base verb), the use of an EQUIPOLLENT marking (both undi-
rected/directed verb forms are formally related but they have the same degree of
complexity, which implies a non-oriented alternation), the use of LABILE verb forms
(another non-oriented alternation in which the same form is used for expressing the
undirected and the directed verbs), and the use of SUPPLETIVE verb forms (another
non-oriented alternation in which two different lexical verbs are used for expressing
the undirected and the directed verbs).2
Two main types of applicative constructions can be found in the five Uto-Aztecan
languages under study. The first type is an optional peripheral applicative, in which the
applied phrase is referring to a non-core participant (usually, a beneficiary) that has

1 As the data come from different sources and authors, some discrepancies in the glosses and transcrip-
tions may appear between examples of the same language. We have decided to homogenize the gloss
when strictly necessary and to keep the original transcription as long as it does not affect the interpre-
tation of the data.
2 The same formal possibilities are found for the non-causal/causal alternation (Haspelmath 1993; Cre-
issels, forthcoming), which is represented by a non-causal/causal verb pair in which the subject of the
non-causal verb form is the object of the causal verb form, and the causal form has a new agentive
(causer) subject. In addition to the equipollent, labile, and suppletive possibilities, the two morpholog-
ically-oriented alternations are, in this case, causativization (characterized by a privative marking on
the non-causal base verb) and decausativization (characterized by a privative marking on the causal
base verb).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 511

been promoted to the object position comparing to the base construction in which it
can be encoded as an oblique (usually, introduced by a benefactive postposition). The
second type is an obligatory central applicative, in which the applied phrase is referring
to a central (essential) non-patient participant in the event denoted by the verb (recipi-
ent/goal, stimulus, source, etc.) and this participant cannot be encoded at all in the con-
struction with the underived form of the same verb. One important feature shared by
both types is that the applied phrase has to usually refer to an animate (mostly human)
participant.
The five Uto-Aztecan languages considered in this study present both types of
applicative constructions but while some languages use the same marker for encod-
ing both types, other languages have dedicated markers for benefactive applicatives.
A pervasive overlap between transitivization, causativization and applicativization is
also present in these languages. In addition to applicativization, another productive
strategy is possible for encoding the undirected/directed alternation in TA and GU, since
the presence of a verbal pattern involving the change of the final vowel of the verb
stem implies the possibility to have an undirected/directed alternation encoded via an
equipollent marking.
The structure of this chapter is as follows. Section 2 presents the main aspects of
the morphosyntax in the UA languages of NW Mexico. Section 3 describes the different
constructions and suffixes used in these languages for applicativization, and also pre-
sents the final stem vowel alternation corresponding to the equipollent marking of the
undirected/directed alternation found in TA and GU. Section 4 highlights some aspects
of the syntax, the semantics and the pragmatics of applicativization in the languages
under study. Section 5 is focused on the causative-applicative syncretism found in the
data. Section 6 discusses some cases of applicative lookalikes (lexicalized applicatives,
applicative deponents). Lastly, Section 7 provides a summary of the main conclusions.

2 The basics of morphosyntax in UA languages

from NW Mexico
UA languages from NW Mexico are agglutinative languages, with an important use of
suffixes. The alignment system is nominative-accusative, but while YQ presents a case
marking distinction for core noun phrases (nom: zero, acc: -ta, with the exception of
plural object NPs that are not case-marked, as in [1b]), in the other four UA languages,
core NPs are not case-marked.
Regarding the word order in neutral declarative sentences, YQ is probably the only
language with a strong tendency for an SOV word order, PB and TA show a tendency for
free word order highly motivated by discourse factors, but still keep a preference for an
SOV order, and GU and NT can also be considered as free word order languages, with a
strong preference, however, for a VSO order.
512 Albert Álvarez González and Zarina Estrada-Fernández

The examples below illustrate these NP case-marking systems and these preferred
word orders in the 5 languages.

(1) YQ: SOV

a. U yoeme-∅ u-ka kari-ta jinu-k.
det man-nom det-acc house-acc buy-pfv
‘The man bought the house.’
b. U yoeme-∅ u-me kari-m jinu-k.
det man-nom det-pl house-pl buy-pfv
The man bought the houses.’
(Álvarez 2021: 316)

(2) PB: SOV

Huaan a-kompaal mua.
John 3nntr.nsbj-compadre kill.sg.obj.pfv
‘John killed his compadre.’
(Estrada 2014: 171)

(3) TA: SOV

Échi rejói saapá sikiré-ri.
det man meat cut-pfv
‘The man cut the meat.’
(Brambila 1976: 519)

(4) NT: VSO

Tií Mañííri imó suiimáli.
find.pfv Manuel det deer
‘Manuel saw a deer.’
(Bascom 2003: 6)

(5) GU: VSO

Paápiarí tetewá-ru hustína waní.
early.morning see-pfv.ev Agustina John
‘Agustina saw John this morning.’
(Félix 2005:124)

In contrast to YQ, nominal core arguments are not case-marked in TA, GU, PB, and NT,
thus showing a neutral alignment for core NPs. The nominative-accusative alignment
is, however, clear in the pronominal system, but while YQ distinguishes between the
accusative, the dative/oblique and the possessive marking, the other four UA languages
use the same pronominal markers for these non-subject functions.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 513

Examples in (6) illustrate that object pronouns in YQ are used for direct object (6a),
and for indirect object with a few underived ditransitive verbs (as with maka ‘give’ in
[6b], resulting in a double object construction). In other cases of trivalent verbs such
as nenka ‘sell’ (6c) or mabeta ‘receive’ (6d), the indirect object is marked by the oblique
pronoun plus the directional suffix -u or another postpositional marker, and these post-
positional complements can be omitted.

(6) a. Inepo Bikam-me-u am toja-k.

1sg.nom Vicam-pl-dir 3pl.acc take-pfv
‘I took them to Vicam.’
b. Inepo tomi-ta am maka-k.
1sg.nom money-acc 3pl.acc give-pfv
‘I gave them money.’
c. Inepo ame-u tajo'o-ta nenka-k.
1sg.nom 3pl.obl-dir clothes-acc sell-pfv
‘I sold clothes to them.’
d. Inepo tomi-ta ame-betana mabeta-k.
1sg.nom money-acc 3pl.obl-from receive-pfv
‘I received money from them.’
(Estrada et al. 2004)

Examples from GU in (7) and from PB in (8) serve to illustrate how, contrary to YQ, the
same non-subject pronouns are used in these languages (the same occurs in TA and NT)
for direct object, indirect object, possessor and postpositional object.

(7) a. no’ó no’nó no’ó wewé-ru.

1sg.nsbj father 1sg.nsbj hit-pfv.ev
‘My father hit me.’
b. waní no’ó itočá-re muní.
John 1sg.nsbj send-pfv beans
‘John sent me beans.’
(Félix 2007: 111)
c. tara-rú=ne muuní.
buy-pfv=1sg.sbj beans
‘I bought beans.’
(Félix 2005: 57)

(8) a. aan timitim am-niar.

1sg.sbj tortillas 2sg.nsbj-buy.pfv
‘I bought you tortillas.’
514 Albert Álvarez González and Zarina Estrada-Fernández

b. aan am-viin him tieend-am.

1sg.sbj 2sg.nsbj-com go.cont store-loc
‘I go with you to the store.’
(Estrada 2007: 88)
c. am-obi-ga-r
2sg.nsbj-needle-al-poss
‘your needle’
(Estrada 2014: 128)
d. aap tu’ut piees-tam.
2sg.sbj dance.pfv party-loc
‘You danced in the party.’
(Estrada 2014: 188)

Another important difference between YQ and the other four languages is that third
person object pronouns are not omitted in YQ, as can be observed in (6), contrary to what
happens in the other 4 languages, in which third person object pronouns are usually
dropped. In fact, in natural discourse, not only third person object pronouns in TA, GU,
PB, and NT are commonly not overtly expressed, but also first and second persons, as
exemplified in (9) with data from Norogachi TA, implying that much information has to
be activated from the context in order to correctly identify core arguments.3

(9) kepi ne muku-mea; éruka mi’ri-méa-we?

neg I die-fut who kill-fut-or
‘I won’t die; or who is going to kill (me)?’
(Brambila 1953: 571)

As for verbal morphology, tense/aspect/mood/valency information is usually marked by

verbal suffixes. Of special interest for applicativization, is the presence of verb classes
in TA and Tepiman languages, depending on some morphophonological changes in
verbal stems.
According to the presence or absence of stress and final stem vowel alternation,
three classes of verbal roots can be distinguished in TA (Caballero 2008), as shown in
Table 1. Class I is represented by stressed roots with no morphophonological changes,
class II by roots with morphophonological changes: class IIa corresponds to unstressed
roots with no final root vowel alternation and with stress shift one syllable to the right
when combined with stress-shifting suffixes, and class IIb to unstressed roots with final
root vowel alternation and stress shift one syllable to the right when combined with
stress-shifting suffixes.

3 As we will see below, this dropping can also be associated with the presence of the applicative mark-
ing on the verb (Brambila 1953: 53).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 515

Table 1: Verb classes in Choguita TA (adapted from Caballero 2008: 99).

Class I Class IIa Class IIb

bené ‘learn’ suku ‘scratch’ rara ‘buy’

Stress- pfv bené-ri sukú-ri rará-ri
neutral prog bené-a sukú-a rará-a
suffixes ipfv bené-i sukú-i rará-i

Stress- fut.sg bené-ma suku-méa rari-méa

shifting cond bené-sa suku-sá rari-sá
suffixes des bené-nare suku-náre rari-náre

As for Tepiman languages, they are known for having different verb stem forms asso-
ciated with the imperfective/perfective distinction, with a reduced form usually being
used for the perfective verb stem. For instance, in NT, the verb stem can present three
different alternating forms (Bascom 2003: 67). The basic verb stem is found with imper-
atives, a second stem form with the final vowel of the basic form changing to i is usually
found in present tense. Both verb stems are used in imperfective forms. The third verb
stem, which is used for perfective aspects, is a reduced form of the basic stem in which
the last syllable has usually been dropped. This truncated form ends with two identical
vowels or with a combination of the last vowel of the root and an i. According to this
stem alternation, three verb classes are distinguished as illustrated in Table 2: a) verbs
that do not present stem alternation, b) verbs that have only two alternating stems, and
c) verbs that present three stem alternants.

Table 2: Verb classes in Northern Tepehuan (adapted from Bascom 2003).

Verb class I: aráava- ‘repeat’

Verb class II: ááxta- ‘throw’ only for imperfective
ááxti- ‘throw’ only for perfective
Verb class III: isáá- ‘plant’ only for imperatives
ixií- ‘plant’ only for imperfective
ií- ‘plant’ only for perfective

In PB, the basic verb form is more unpredictable, although six different verb classes
have been proposed (Estrada 2014: 66–75). Again, the perfective verb form is usually a
truncated form without any suffixes.4

4 In Tepecano, another now extinct Tepiman language, the preterit stem of action verbs most frequent-
ly undergoes the dropping of the final consonant or syllable (Mason 1917: 351).
516 Albert Álvarez González and Zarina Estrada-Fernández

3 Applicativization in UA languages of NW Mexico

3.1 Pima Bajo

In PB, the suffix -di/-id5 can be attached to a base verb in order to promote to the object
position a benefactive participant (10b) that can be encoded in the base construction
as an oblique complement introduced by the benefactive postposition vuika ‘for’ (10a).
This participant is thus peripheral, since it is not required by the argument structure
of the base verb, and the applicative construction is optional, since the same semantic
content is alternatively expressed in the base construction through a benefactive post-
positional phrase.

(10) a. Marii ik mo’ovil soom Hosip vuika.

María det.obj blouse sew.pfv Josefina for
‘María sewed a blouse for Josefina.’
b. Marii Hosip som-di ik mo’ovil.
María Josefina sew.pfv-appl det.obj blouse
‘María sewed a blouse for Josefina.’
(Estrada 2007)

The applied object can receive a malefactive interpretation according to the negative
meaning of the base construction, as can be observed in (11b). This example also illus-
trates a construction in which the applied object can be interpreted as the possessor of
the original object participant (external possessor). Interestingly, the possessive rela-
tionship is here also overtly marked as an internal possession (in-iipar ‘my skirt’).

(11) a. irvin vainit-kar hikt-in!

rope knife-ins cut-imp
‘Cut the rope with the knife!’
(Estrada 2014: 132)
b. Marii in-hik-id-im in-iipar
María 1sg.nsbj-cut-appl-cont 1sg.nsbj-skirt
‘Maria cut off (me) my skirt.’
(Estrada 2014: 178)

This benefactive/malefactive applicativization via -di/-id suffixation is restricted in PB

to transitive base constructions (Estrada 2007: 91). However, this suffix can also be used
in this language for licensing a human recipient as object. This is possible not only with

5 In some cases, this applicative suffix appears as -ti, -ir, or -li (as, for example, in narrative texts found
in Ramírez 2010).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 517

active intransitive verbs like in (12), but also with some transitive verbs of movement,
like in (13).

(12) a. lii sii hink.

dim.sg wolf shout
‘The little wolf shouted.’
(Estrada 2014: 166)
b. higam tit-hink-id.
3pl.sbj 1pl.nsbj-shout-appl
‘They shouted at us.’
(Estrada 2014: 122)

(13) a. tomin hih-hodav.

money obj.indf-throw.pfv
‘The money was thrown away.’
(Estrada 2014: 231)
b. in-hodav-id-an.
1sg.nsbj-throw-appl-imp
‘Throw them to me!’
(Estrada 2014: 179)

In both examples, the applied object is a human recipient. The derived verb indicates
that the action denoted by the base verb is directed towards the human participant
encoded as an applied object. Other active intransitive verbs that function as applica-
tive verbs with the same suffix in PB are gig ‘gesticulate, wave hands’ (applicative: gig-id
‘greet someone’) and tii ‘talk’ (applicative: tii-di ‘say to someone’).
The suffix -di/-id can also be used in PB for causativization depending on the base
verb class (Estrada 2007). As can be observed in (14), when attached to inactive (stative
and inchoative) intransitives, the suffix -di/-id derives causative transitives, while it
derives applicatives when combined with active intransitives and transitives.

(14) Intransitives Causative transitives

onmag ‘be salty’ onama-d(i) ‘salt’
doa ‘be cured’ doa-li, doa-r < doa-di ‘cure’
tuuk ‘be dark’ tuk-id ‘darken’
tipilik ‘be flat’ tipilik-id ‘flatten’
toahk ‘be white’ toah-id ‘whiten’
i’ov ‘be sweet’ i’ov-id ‘sweeten’
tig ‘be named’ tig-di ‘name’
bid ‘turn around’ bib-id/bib-di ‘flip’
hoin ‘rock’ hoin-id ‘rock’
tuk ‘remember’ tuk-id ‘recall’
518 Albert Álvarez González and Zarina Estrada-Fernández

These data clearly illustrate a causative-applicative syncretism, in which the same

verbal suffixation corresponds to applicativization if the base verb is a transitive verb
or an active intransitive verb implying no change of state, and to causativization if the
base verb is a stative/inchoative intransitive.

3.2 Northern Tepehuan

The same type of transitivizing suffix is present in NT, as the suffix -d(y)a/-d(y)i (Bascom
1982: 359–360, 2003: 83–90). Again, attached to stative and inchoative intransitives,
it derives causative transitives (15), and applicative transitives (16) when attached to
active intransitives.

(15) Intransitives Causative transitives

dapáka ‘be smooth’ dapáka-da ‘make smooth’
totópikii ‘boil’ totópish-dya ‘boil’
mimííjii ‘burn’ mimíí-di ‘burn’
úkai ‘warm’ úka-da ‘warm’

(16) Intransitives Applicative transitives

didívia ‘come’ didíívi-di ‘visit’
niníínai ‘see’ niníí-di ‘look at’
axíógai ‘smile’ axíógi-di ‘smile at’
ásii ‘laugh’ axíí-di ‘laugh at’
aatágai ‘converse’ aatági-dya ‘converse’

Both uses as causativizer and applicativizer are exemplified with their base construc-
tions in (17) and (18), respectively. If the base construction is an inchoative intransi-
tive, the derived transitive construction includes a new subject participant (a causer)
as in (17), while it includes a new object participant if the base construction is an active
intransitive, as in (18).

(17) a. mi~mííji-i go-váso-i

rdp~burn-ipfv det-grass-abs
‘The grass is burning.’
b. mi~míi-di igai go-váso-i
rdp~burn-caus dem det-grass-abs
‘He is burning the grass.’
(Bascom 2003: 87)
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 519

(18) a. axíóga-i
smile-ipfv
‘He/she smiles.’
b. axíógi-di
smile-appl
‘He/she smiles at him/her.’
(Bascom 2003: 84)

The applicative transitive verbs resulting from the -di suffixation take as their applied
object a human participant interpretable not always as a recipient, but also, for instance,
as a stimulus.
With transitive base verbs, this suffix functions as an applicativizer, and the new
object participant can be a human recipient or a human beneficiary, depending on the
meaning of the transitive base verb.

(19) a. áága-i
tell-ipfv
‘‘He/she tells it.’
b. áági-di
tell-appl
‘He/she tells it to him/her.’
(Bascom 2003: 86)

(20) a. báí bííkai go-doad yí-ga-mi

this_way bring et-cure-adjr-nmlz
‘Bring the medicine here!’
b. báí im-bii-d ya go-so~sóna-karoi
this_way me-bring-appl det-rdp~tap-ins
‘Bring me the hammer!’
(Bascom 1982: 294)

(21) a. ga-isa-i aáni úúnu-i

part-plant-ipfv I corn-abs
‘I am planting corn.’
b. ga-ixíí-di aáni úúnu-i
part-plant-appl I corn-abs
‘I planted corn for him/her.’
(Bascom 2003: 87)

As can be observed in (19b) and (21b), the applied object can be left unexpressed in
the applicative construction. In (19b), both objects (the original object and the applied
object) are omitted and the only indication of the presence of a human recipient in the
520 Albert Álvarez González and Zarina Estrada-Fernández

situation denoted by the verb is the applicative suffix -di. The omission of the applied
object can also be observed in (18b).
Interestingly, the -di suffixation illustrated in the previous examples seems to
always be associated with morphophonological modifications in the base verb stem
implying the change of the final stem vowel to i.6
Besides the possible benefactive meaning conveyed by the suffix -d(y)a/-d(y)i with
transitive verbs (exemplified in [21b]), NT also has a dedicated construction for bene-
factive applicatives in which the transitivizing suffix is combined with the suffix -tul(i).
Example (22a) shows the benefactive participant encoded as the pronominal object of
the benefactive postposition viitari ‘for’. In the derived construction (22b), this same par-
ticipant has been promoted to a pronominal object prefixed to the applicativized verb.

(22) a. saví-da-ñi aápi go-vasááraga-i aáni iñ-viitári.

buy-tr-imp you det-shirt-abs you me-for
‘Buy this shirt for me.’
b. giñ-savil-tul-da-ñi aápi go-vasááraga-i.
me-buy-ben-tr-imp you det-shirt-abs
‘Buy me this shirt.’
(Bascom 2003: 88)

This benefactive applicative suffix is always combined with the transitivizing suffix -d(y)
a/-d(y)i,7 except in the truncated perfective form in which it appears as -tuli, as shown
in (23).

(23) a. iñ-savíl-tul-da=ñi!
me-buy-ben-tr=imp
‘Buy it for me!’
b. iñ-savíl-tul-di.
me-buy-ben-tr
‘He/she is buying it for me.’
c. iñ-savíl-tuli.
me-buy-ben
‘He/she bought it for me.’
(Bascom 2003: 69)

6 Interestingly, the forms of the applicative suffixes in historical varieties of Tepiman languages include
the change of the final stem vowel to i: -(i)d/-(i)t for Tepecano (Mason 1917), -(i)da for Nevome or Lower
Pima (Smith 1862, cited in Mason 1917: 365), -(i)de, -(i)di for NT (Rinaldini 1743, cited in Mason 1917:
365).
7 As shown in (23) and following the rule for verb class III in NT (see § 2), the form -d(y)a is used in the
imperative, -d(y)i in the imperfective, and -tuli in the perfective (Bascom 2003: 68). This distribution is not
so systematic when the transitivizing suffix d(y)a/-d(y)i is used without the benefactive suffix.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 521

Unlike the suffix -di/-id in PB, whose benefactive meaning seems to be exclusive to tran-
sitive verb bases (Estrada 2007: 91), the benefactive suffix can also be attached in NT to
active intransitive verbs with the same benefactive meaning. An example is given in (24).

(24) i-ñíóóki-tul-d ya-ñi áapi!

me-speak-ben-tr-imp you
You, speak for me!
(Ramos 2010: 141)

This example shows the use of two applicative markers for rendering the benefactive
meaning. The differences in the implication of a human participant associated with
the suffixes -d(y)a/-d(y)i and -tul(i) can be seen with this verb ñioókai ‘speak’, as shown in
(25). The -d(y)i suffixation to this intransitive base verb implies the presence of a human
recipient as object and the addition of the benefactive suffix -tul implies the presence
of a human beneficiary.

(25) ñiokai ‘speak.intr’

ñioóki-dyi ‘speak to someone’
ñioóki-tul-dyi ‘speak for someone’
(Carrillo 2013: 64)

3.3 Yaqui

In YQ, it is the suffix -ria that conveys the benefactive applicative function, as exempli-
fied in (26).

(26) a. Aurelia-∅ u-ka wakabak-ta joa-k Goyo-ta-betchi’ibo.

Aurelia-nom det-acc wakabaki-acc make-pfv Goyo-acc-for
‘Aurelia cooked wakabaki for Goyo.’
b. Aurelia-∅ Goyo-ta u-ka wakabak-ta joa-ria-k.
Aurelia-nom Goyo-acc det-acc wakabaki-acc make-appl-pfv
‘Aurelia cooked wakabaki for Goyo.’
(Guerrero 2007: 182–183)

This benefactive applicative suffix is very productive. It is usually attached to transitive

base verbs, but active intransitive verbs like ye’e ‘dance’ can also receive this marking
in order to promote a human beneficiary to object position.

(27) María-∅ Goyo-ta yi’i-ria-k.

María-nom Goyo-acc dance-appl-pfv
‘María danced for Goyo’.
522 Albert Álvarez González and Zarina Estrada-Fernández

In some cases, the applied object introduced by the suffix -ria is a human participant
that cannot really be interpreted as a beneficiary. This is the case for instance with the
stative intransitive verb alle’a ‘be happy’, in which the applied object seems to be more
an object of empathy.

(28) a. María-∅ alle’a-∅.

María-nom be_happy-prs
‘María is happy.’
(Estrada et al. 2004)
b. María-∅ Kajlo-ta alle’a-ria-∅.
María-nom Carlos-acc be_happy-appl-prs
‘María is happy for Carlos.’
(Guerrero 2007: 187)

Another case is represented by the intransitive base verb gojana ‘run away’, in which
the applied object corresponds more to a human source than to a beneficiary.

(29) a. Mario gojana.

Mario run_away
‘Mario runs away.’
(Estrada et al. 2004: 31)
b. U ili uusi mala-wa gojan-ria.
det little boy mother-poss run_away-appl
‘The child runs away from his mother.’
(Estrada et al. 2004: 31)

This possibility to have a non-benefactive applicative is also found with the verb etejo
‘tell, talk’, in which the suffix -ria is used to promote to object position a human partic-
ipant that can be encoded as a comitative (Example [30a]) or a directional (Example
[30b]) complement in the base construction. In this case, as can be observed in (30c), in
addition to the human participant marked as accusative, the derived applicative con-
struction also licenses the presence of the topic of conversation as object (here the nom-
inalization káa túa á’a alléa’u ‘his not truly happiness’), something that is not possible
in the base construction.

(30) a. in jala’i-mak né etéjo.

my friend-with I talk
‘I am talking with my friend.’
(Dedrick and Casad 1999: 190)
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 523

b. emo-u ne etejo-bae.
you.pl-dir I tell-des
‘I want to tell you.’
(Estrada et al .2004)
c. káa túa á’a alléa-’u née etejó-ria-k.
neg truly his be_happy-obj.nmlz me tell-appl-pfv
‘He told me he was a bit sad.’
(Lit. ‘He told me his not truly happiness.’)
(Dedrick and Casad 1999: 374)

In the case of the transitive verb be’a ‘set aside’ in (31), the human participant who is
benefited by the action is encoded as the object of the directional postposition (31a),
while its applicative derivate licenses the beneficiary as an accusative object (31b).

(31) a. Peo bwa'am-ta ne-u be'a-k be'eri-ta.

Peter food-acc 1sg.nom-dir set_aside-pfv leftover-acc
‘Peter left me the leftover food.’
b. Joana-ta ne paan-im be'e-ria-k.
Joana-acc 1sg.nom bread-pl set_aside-appl-pfv
‘I left bread for Juana.’
(Estrada et al. 2004)

The examples in (32) can be used to illustrate the applicativization of trivalent verbs, i.e.
verbs that require three semantic participants, regardless of whether these participants
are encoded as A, R and T according to the prototype verb GIVE in the language. Like
the verb be’a ‘set aside’, the verbs nenka ‘sell’ in (32) and teuwa ‘tell’ in (33) require the
suffix -ria in order to promote the human recipient encoded as a directional oblique in
the base construction, to the object position in which it receives the accusative marking,
without the postposition.

(32) a. bempo kowi-ta u-e jamut-ta-u nenki-ne.

3pl.nom pig-acc det-obl woman-acc-dir sell-pot
‘They will sell the pig to the woman.’
b. bempo kowi-ta u-ka jamut-ta nenka-ria-ne.
3pl.nom pig-acc det-acc woman-acc sell-appl-pot
‘They will sell the pig (to) the woman.’
(Guerrero 2022: 33)

(33) a. Maria-∅ Luisa-ta-u lutu’uria-ta teuwa-k.

Mary-nom Luisa-acc-dir truth-acc tell-pfv
‘Mary told the truth to Luisa.’
524 Albert Álvarez González and Zarina Estrada-Fernández

b. Maria-∅ Luisa-ta lutu’uria-ta teuwa-ria-k.

Mary-nom Luisa-acc truth-acc tell-appl-pfv
‘Mary told the truth (to) Luisa.’
(Guerrero 2022: 33)

With the verb mabeta ‘receive’, it is the human source that is promoted to object in the
applicative construction.

(34) a. inepo u-ka bwa’am-ta mabeta-k kobanao-ta-betana.

1sg.nom det-acc food-acc receive-pfv governor-acc-from
‘I received the food from the governor.’
b. u-∅ kobanao-∅ tomi-ta kaa ne mabeti-ria-k.
det-nom governor-nom money-acc neg 1sg.acc receive-appl-pfv
‘The governor didn’t receive the money (from) me.’
(Guerrero 2022: 34)

With the verb benta ‘smear, spread’, it is the human locus that is promoted to object in
the applicative construction.

(35) a. Inepo techoa-ta e-t benta-k pujba-chi.

1sg.nom mud-acc 2sg.obl-loc smear-pfv face-loc
‘I smeared mud on your face.’
b. Inepo enchi techoa-ta benta-ria-k pujba-chi.
1sg.nom 2sg.acc mud-acc smear-appl-pfv face-loc
‘I smeared you mud on your face.’
(Álvarez’s fieldnotes)

Interestingly, with this same verb, when the locus of spreading is inanimate, the applica-
tive suffix serves to introduce a human beneficiary.

(36) a. Goyo-∅ pann-im-met mantekia-ta benta-k.

Goyo-nom bread-pl-pl.loc butter-ACC spread-pfv
‘Goyo spread butter on the bread.’
b. Goyo-∅ Iban-ta pann-im-met mantekia-ta benta-ria-k.
Goyo-nom Ivan-acc bread-pl-pl.loc butter-ACC spread-appl-pfv
‘Goyo spread butter on the bread for Ivan.’
(Guerrero 2007: 191)

As can be observed in all these examples of applicativization of trivalent verbs, the

resulting applicative construction is a double object construction. A three-object con-
struction is also possible with the verb u’ura ‘take from someone’, in which the applica-
tivization adds a beneficiary to the situation denoted by the base verb.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 525

(37) a. Carmen-∅ Lupe-ta tomi-ta u’ura-k.

Carmen-nom Lupe-acc money-acc take-pfv
‘Carmen took the money (from) Lupe.’
b. Carmen-∅ Joana-ta Lupe-ta tomi-ta u’ura-ria-k.
Carmen-nom Juana-acc Lupe-acc money-acc take-appl-pfv
‘Carmen took the money (for) Juana (from) Lupe.’
(Guerrero 2022: 33)

This verb u’ura can be considered a ditransitive verb in YQ, since it licenses the same
double object construction as the prototypical verb of transfer miika ‘give’. Note that
this verb miika does not admit the applicative suffix, since the base construction
already contains a human participant acting as recipient/beneficiary and encoded as
an accusative object.

(38) a. Joana-∅ Lupe-ta soto’i-ta miika-k.

Juana-nom Lupe-acc pot-acc give-pfv
‘Juana gave Lupe the pot.’
b. ✶Joana-∅ Maria-ta soto’i-ta Lupe-ta miik-ria-k.
Juana-nom Mary-acc pot-acc Lupe-acc give-appl-pfv
Intended: ‘Juana gave Lupe the pot for Mary.’
(Guerrero 2022: 33)

Additionally, YQ has three verbs that seem to illustrate the grammaticalization of an old
verb iyaa ‘to pretend, to wish’ into an applicative marker. This verb is no longer used
nowadays in YQ as a lexical verb, but it has been documented for Tehueco, a Cahitan
variety from the beginning of the XVII century (Buelna 1890: 39). The use of this verb in
Tehueco is exemplified in (39). Interestingly, like in its applicative use, these examples
show a use of the form iyaa in which it does not head its own clause, but it is already
involved in a two-verb construction (39a) or in a converbal construction (39b).

(39) a. emchi ne sim-yia.

2sg.acc 1sg.nom go-pretend
‘I intend for you to leave.’
(Buelna 1890: 39)
b. emchi ne ieuatua emchi netz eria iyaa-cari.
2sg.acc 1sg.nom correct 2SG.acc 1SG.acc love pretend-ger
‘I correct you to love me.’
(Lit. ‘I correct you, pretending that you love me.’)
(Buelna 1890: 74)

Two mental intransitive verbs in YQ seem to show the presence of this old verb in order
to obtain their transitive counterparts. The first case is represented by the intransitive
526 Albert Álvarez González and Zarina Estrada-Fernández

verb ’ea ‘to think, to wish’ and its transitive related form ’eiyaa, which can both be
used to express the notion of trust in someone. While the source of trust is encoded as
a locative oblique with ’ea, it receives the object marking with ’eiyaa, as shown in (40).

(40) a. Hóan-ta-t ne yoem-’ea-n.

John-acc-on I person-think-pcn
‘I had been trusting in John.’
b. Lióh-ta ne yoem-'eiyaa.
God-acc I person-think
‘I am trusting God.’
(Dedrick and Casad 1999: 368)

In other uses of ’eiyaa, the introduction of a human stimulus as an object is accompa-

nied by a less predictable meaning change, as in its use with the meaning ‘to judge, to
think highly of, to esteem’ exemplified in (41b).

(41) a. née hunen ’ea.

I thus wish
‘That is how I think.’
b. tu’isi n=a’a ’eiyaa.
much I=him esteem
I esteem him very much.
(Dedrick and Casad 1999: 263)

Morphologically similar is the pair hu’unea ‘to know, to know about X’ with its transitive
counterpart hu’uneiyaa ‘to know X’ given in (42).

(42) a. kan=nee hu’unea-k.

not=I know.intr-pfv
‘I didn’t know.’
(Dedrick and Casad 1999: 282)
b. haibu=ne ’ae-t hu’unea.
now=I it-on know.intr
I already know about it.’
(Dedrick and Casad 1999: 368)
c. nehpo ket hunuen ’a’a hú’uneiyaa.
I also thus it know.tr
‘I also know it that way.’
(Dedrick and Casad 1999: 338)
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 527

The third case involves the motion verb weye ‘to go’ and its transitive counterpart
weiyaa ‘carry’. Here again an idiosyncratic change of meaning seems to be involved to
some extent.

(43) a. Aapo a'abo weye.

3sg.nom here go
‘He is coming here.’
b. U jamut ba'am soto‘o-po weiyaa.
det woman water pot-loc carry
‘The woman is carrying water in the pot.’
(Lit. ‘The woman is intending for the water to come in the pot.’)
(Estrada et al. 2004)

No other cases of iyaa applicative verbs have been identified. These few cases could
perhaps be analyzed as the vestiges of an applicative periphrasis, that is a biverbal
construction comparable to a monoverbal applicative. In this biverbal applicative con-
struction, one of the verbs (the lexical verb) determines the type of event encoded by
the applicative periphrasis, while the other verb (in our case, iyaa) acts as a valency
increasing marker that licenses the expression as object of an additional participant
in the event encoded by the lexical verb, without modifying the morphosyntactic treat-
ment of the other participants (Creissels, forthcoming).
However, the form iyaa does not correspond currently to a lexical verb, it is only
found attached to a few verbs, acting synchronically as an idiosyncratic unproductive
applicative suffix. Therefore, these transitive verb forms could only be cases of applica-
tive periphrases from a diachronic point of view. These cases also show that the original
meaning of iyaa has been lost, with the possible presence of some idiosyncratic seman-
tic change.

3.4 Tarahumara and Guarijío

The situation in the two closely related TA and GU is more complex, since in these two
languages several suffixes can be used for adding a new participant encoded as an
object to the situation denoted by the base construction, and applicative suffixation can
also be combined with vowel alternation and stress shift.
In TA and GU, different markers can be suffixed to the base verb for introducing in
the situation denoted by the base verb, a new participant encoded as an object. Accord-
ing to sources, these markers are: the suffixes -e, -ri/-ti/-ki/-gi, -ši, -ne, -i, -mi, -tze/-che
in Norogachi TA (Brambila 1953, 1976), the suffixes -i, -ni, -chi, -wi, -ki in Choguita TA
(Caballero 2008), and the suffixes -é/-íe, -ke/-ge/-kie/-gie, -ne/-nie/-ini, -che, -tze in Moun-
tain GU (Miller 1996: 160–163).
528 Albert Álvarez González and Zarina Estrada-Fernández

The first applicative suffixes to be discussed are three unproductive transitivizing

suffixes ending in e, which correspond to the applicative alternants of transitivizing
causative suffixes ending in a. In Norogachi TA, Brambila (1953: 193) identified three
different suffixes (-na, -tza, -wa) that can be combined with a few intransitive base
verbs in order to obtain the corresponding causative transitive verbs:

(44) Intransitives Causative transitives

a. ko' ‘eat’ kó-na ‘make eat’
go’so ‘choke’ go’so-na ‘choke’
baji ‘drink’ bají-na ‘make drink’
buwe ‘wait’ buwe-na ‘make wait’
b. guri ‘turn’ guri-tza ‘spin’
weka’ ‘get wrong’ weka-tza ‘mislead’
rishi ‘get tired’ reshe-tza ‘tire’
c. nayu ‘get sick’ nayú-wa ‘make sick’
muku ‘die’ muku-wa ‘make die’

A few intransitive verbs can take the three unproductive suffixes -ne, -tze, -we for
applicativization. As shown in (45), (46) and (47), the base verb is an active intransitive
verb and the applied object is a human participant that can be acting as a recipient, a
stimulus or a beneficiary.

(45) a. we nawaji-re rarámuri baji-sa.

a_lot sing-pfv tarahumara drink-ger
‘The Tarahumaras drank and sang a lot.’
(Lit. ‘The Tarahumaras sang a lot, drinking.’)
(Brambila 1976: 370)
b. má nawa-re tata pari picharo, má
already arrive-pfv father father Pichardo already
nawaji-ne-a kuchi.
sing-appl-ant girl
‘Father Pichardo has arrived, and the girls are already singing to/for him.’
(Brambila 1953: 183)

(46) a. ne che'reba-re nara-ga.

I wake_up-pfv crying-ger
‘I woke up crying.’
(Brambila 1976: 360)
b. nawika ta nara-tze-bo ke oneruame.
together only cry-appl-imp.1pl expl God
‘Together let us cry to God.’
(Brambila 1976: 361)
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 529

(47) a. ketza mu maja-re bowi-chi?

inter.neg you be_afraid-pfv road-loc
‘Weren’t you afraid on the road?’
(Brambila 1976: 295)
b. ké ne tasi Martin maja-we.
neg I neg Martin be_afraid-appl
‘I am not afraid of Martin.’
(Brambila 1976: 293)

In Choguita TA, the corresponding suffixes are -ni, -chi, -wi, since unstressed e changes
to i in pre-tonic and post-tonic syllables in this variety (Caballero 2008: 55).
In GU, the suffixes -na, -cha, -wa are also used as causative markers with a few
intransitive base verbs, but only the suffixes -ne and -che are found for applicativization
(Miller 1996; Ávila 2012; Casas 2018). As shown in (48), (49) and (50), the base verb can
also be transitive in GU. Attached to the verb of speaking ihta ‘to ask for’ or to the verb
of movement mahtó ‘to bring’, these suffixes license the presence of a human recipient
as an object argument.

(48) a. ihta-má.
ask-fut
‘He will ask for it.’
(Medina 2002:174)
b. maría noˀó ihta-né-na takári.
María 1sg.nsbj ask-appl-prs tortillas
ʽMaría asks me for tortillasʼ.
(Ávila 2012: 139)

(49) a. puu-á mahtó-ré kuhú.

3sg.s-emph bring-pfv firewood
ʽShe brought firewood.’
b. puu-á ahpo yeˀyé mahtó-ne-ré kuhú.
3sg.sbj-emph 3sg.nsbj mother bring-appl-pfv firewood
‘She brought firewood to her mother.’
(Ávila 2012: 137)

(50) muú tamó muké-če-mapo naaráso.

2sg.sbj 1pl.nsbj load-appl-fut orange
‘You are going to load us with oranges.’
(Ávila 2012: 140)
530 Albert Álvarez González and Zarina Estrada-Fernández

The applicative suffix can be reduced to -e (with the possibility to have an epenthetic y
or w). In Examples (51), (52) and (53) presented below, the base verb is an active intran-
sitive or a transitive verb, and the applied object is a recipient or a beneficiary.

(51) Choguita TA
ru-é, ru-yé, ru-wé ‘tell-appl’
bu-é, bu-yé, bu-wé ‘wait-appl’
(Caballero 2008: 78)

(52) Norogachi TA
a. jipe nepa-re erasmo
today call-pfv Erasmo
‘Erasmo called today.’
(Brambila 1976: 376)
b. ké ne ka nepá-e-re
neg 1sg.nom neg call-appl-pfv
‘I did not call him.’
(Brambila 1976: 376)

(53) GU
ča’pi ‘grab something’ ča’pí-e ‘grab for someone’
wi’ko ‘whistle’ wiko-é ‘whistle at someone’
(Medina 2002)

Although the benefactive meaning can be expressed in some cases by these different
suffixes, the most productive benefactive suffixes are, in fact, suffixes -ke in GU, -ki in
Choguita TA, and -i in Norogachi TA. These productive benefactive applicative suffixes
are exemplified in (54), (55) and (56), respectively.

(54) GU
a. hustína pasu-ré muní no’ó ičió.
Agustina cook-pfv beans 1sg.nsbj ben
‘Agustina cooked beans for me.’
b. hustína no’ó pasú-ke-re muní.
Agustina 1sg.nsbj cook-ben-pfv beans
‘Agustina cooked beans for me.’
(Félix 2007: 115)

(55) Choguita TA
a. ma=n rata-bá-či-ki ko’wá-ami.
already=1sg.nom heat-inch-tr-pt:1 eat-ptcp
‘I already heated up the food.’
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 531

b. ne mi ba’wí rata-bá-č-ki-ra?
1sg.nom 2sg.acc water heat-inch-tr-appl-pot
‘Shall I heat the water for you?’
(Caballero 2008: 417)

(56) Norogachi TA
a. mi ne mi’ri-mea ena ata wi-gá
2sg.acc 1sg.nom kill-fut this bow take-ger
‘I take this bow and I kill you.’
(Lit. I will kill you, taking this bow.’)
(Brambila 1976: 320)
b. owétza-ka ne ka bayé-nura.
healer-expl 1sg.nom emph call-send
‘I've already sent for the healer.
Píri be mu mi’rí-i-ma?
what and 2sg.nom kill-appl-fut
—And what are you going to kill for him?’
(Brambila 1953: 184)

Examples in (56) and (57) show how morphophonology can be involved in applicativi-
zation in Norogachi TA, with the case of the benefactive suffix -i that triggers the change
of the final verb stem vowel to stressed i. Additionally, the verbs listed in (57) indicate
that, although transitive base verbs are more frequent, some intransitive base verbs are
also possible (such as ripi ‘be left, remain’ and nawa ‘arrive’).

(57) Applicative verbs marked by the suffix -i:

ripi ‘be left, remain’ ripí-i ‘be left for someone’
mi'ri ‘kill’ mi’rí-i ‘kill for someone’
newa ‘do’ newí-i ‘do for someone’
nawa ‘arrive’ nawí-i ‘arrive for someone’
ora ‘do’ orí-i ‘do for someone’
cha’pi ‘catch’ cha’pí-i ‘catch for someone’
acha ‘put’ achí-i ‘put for someone’
(Brambila 1953: 184)

With some verbs belonging to class IIb (see § 2) and allowing two different types of
human participant as object, applicative suffixation is accompanied by vowel alter-
nation and stress shift in TA. As shown in (58) and (59), the verbal stems ending in a
stressed front vowel é (probably resulting from the fusion of the applicative suffix -é to
the verb base) are used for introducing the human source or recipient, while the stem
ending in a stressed high front vowel í is combined with -i suffixation in order to intro-
duce a human beneficiary.
532 Albert Álvarez González and Zarina Estrada-Fernández

(58) a. rari- ‘buy something’(stem form with stress shifting suffixes)

b. rará-‘buy something’ (stem form with stress neutral suffixes)
c. raré- ‘buy from someone’
d. rarí-i ‘buy for someone’
(Caballero 2008: 106–107)

(59) a. osi- ‘write something’ (stem form with stress shifting suffixes)
b. osá- ‘write something’ (stem form with stress neutral suffixes)
c. osé- ‘write to someone’
d. osí-i- ‘write for someone’
(Brambila 1953: 182–184)

The use of a final vowel stem alternation from a/i to e for encoding the undirected/
directed alternation is, in fact, frequent in TA and GU. Brambila (1953: 177–179) indi-
cates that the -e suffixation usually implies in Norogachi TA the loss of the final vowel of
the undirected verb stem (for instance, final i is usually replaced by e), thus resulting in
a final vowel stem alternation in which the verb stem ending in e is the directed stem.
In Choguita TA, Caballero (2008) identifies the possibility to have two different types of
“applicative stems”:8 the directed stem ending in é as in (58c) and (59c), and the directed
stem ending in í, which is usually followed by an applicative suffix as in (58d) and (59d).
In GU, Miller (1996: 160) also mentions that the suffix -é can, in some cases, replace the
final vowel of the verbal stem. This situation thus implies that, strictly speaking, we can
consider that we no longer have an undirected/directed alternation marked by a priv-
ative marking as in applicativization, but we have an equipollent marking, 9 in which
the directed stem is obtained by changing the final vowel of the undirected stem to e.
Like the applicative suffixes, the final stem vowel alternation from a/i to e is used in
TA for introducing a non-patient participant, usually human, as a syntactic object. This
strategy is available not only for some intransitive verbs but also for some transitive
verbs, as exemplified in (60), (61) and (62) from Norogachi TA, and (63) from Choguita
TA.

(60) a. notza-riwa echi.

work.intr-pass there
‘There is working over there.’
(Brambila 1976: 391)

8 “Applicative stem” is the term used by Caballero (2008). As the applicative construction by definition
implies the presence of an affix (the applicative morpheme attached to the undirected verb form), we
prefer to use the term “directed stem”.
9 The equipollent marking for the undirected/directed alternation is considered a typological rarity by
Creissels (forthcoming).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 533

b. antresi ne notze.
Andres 1sg.nom work.tr
‘I work for Andres.’
(Brambila 1976: 392)

(61) a. echi bera ne gite achi-é.

that for 1sg.nom for laugh.intr-pst.ipfv
‘That’s why I was laughing.’
(Brambila 1976: 4)
b. nejé nimí aché.
1sg.nom 2sg.acc laugh.tr
‘I am laughing at you’
(Brambila 1976: 4)

(62) a. kumí a’tará echi?

where use_to_buy.tr dem
‘Where do you usually buy these things?’
(Brambila 1976: 35)
b. Pe róbrika a’ataré ne sunú pa.
still Rodrigo use_to_buy.tr I corn expl
‘I still use to buy corn from Rodrigo.
(Brambila 1976: 35)

(63) a. naparí noká-ri ronochí=ni okó.

when move.intr-pst legs=1sg.nom hurt
‘When I moved, my legs hurt.’
b. noké-ri ré=n má-o.
move.tr-pst dub=1sg.nom maybe-emph
‘Maybe I moved him.’
(Caballero 2008: 166)
c. nihé mi troka noké-ri.
1sg.nom 2sg.acc truck move.tr-pst
‘I will move the truck for you.’
(Caballero 2008: 107)

As shown in these examples, the new object of the directed construction marked by the
é verbal stem is a human participant referring to a diversity of semantic roles, such as a
recipient, a beneficiary, a stimulus, a source or even a theme, as in (63b). The examples
in (63) from Choguita TA are interesting, since they show that the verb form ending in
é can be used as a causal transitive (63b) or as a directed ditransitive (63c) verb, but,
in both cases, we can notice the presence of a human participant as an object (as P in
[63b], as R in [63c]).
534 Albert Álvarez González and Zarina Estrada-Fernández

Interestingly, these equipollent verbs have the same distributional pattern as in the
applicative/causative uses of transitivizing suffixes in PB and NT (see §§ 3.1 and 3.2):
Final stem a/i to e alternation in TA serves to encode the undirected/directed alterna-
tion, if the a/i counterpart is a transitive verb (as in [62]) or an active intransitive verb
(as in [60]), and it serves to encode the non-causal/causal alternation, if the a/i counter-
part is inchoative intransitive as in (64).

(64) Intransitives ending in i Causal transitives ending in e

raji ‘burn’ raje ‘light (a fire)’’ cause to burn
noki ‘move’ noke ‘move’ cause to move
sa’wi ‘heal’ sa’we ‘heal’ cause to heal
lowi ‘go crazy’ lowe ‘drive crazy’ cause to go crazy
waki ‘dry’ wake ‘dry’ cause to dry
wasi ‘cook’ wase ‘cook’ cause to cook

The same is found in GU, although the alternation is almost always from a to e, since
most intransitive and transitive verbs tend to end in a in this language. Examples in (65)
show the use of a to e alternation for non-causal/causal alternation with both inchoa-
tive and stative intransitives. In these cases, the change of the final vowel to e serves to
create a transitive verb in which the object corresponds to the subject, usually animate
(mostly human), of the a verb form.

(65) Intransitives ending in a Causal transitives ending in e

i’óa- ‘cure’ i’óe- ‘cure’
kema-/kemi- ‘tuck in’ kemé- ’tuck’
taha-/tahi- ‘burn’ tahé- ‘burn’
upá- ‘bathe’ upé- ‘bathe’
yeloába- ‘be poisoned’ yeloé- ‘poison’
pi’wá ‘be cleaned’ pi’wé- ‘clean someone’

Verb pairs in (66) exemplify in GU the a to e alternation changing active intransitive

verbs to directed transitive verbs. In these cases, the object of the final e stem has been
added in comparison with the clause in which the predicative nucleus is the active
intransitive verb. Again, we can observe that the additional object is a human partici-
pant, usually acting as a recipient.

(66) Intransitives ending in a Directed transitives ending in e

tepóra- ‘greet’ tepóre- ‘greet’
wikóa- ‘whistle’ wikoé- ‘whistle at’
naósa- ‘talk’ naóse- ‘talk to’
pahkóra- ‘keep vigil’ pahkóre- ‘keep vigil over’
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 535

With transitive verbs, the final vowel change to e is used for the undirected/directed
alternation, and the animate (mostly human) participant added as object in the final e
verb form can be a beneficiary/maleficiary but also a recipient or a source, as can be
seen in (67).

(67) Transitives Directed transitives ending in e

olá- ‘do something’ oríe- ‘do something for someone’
werá-. ‘put something on stand’ weré- ‘put for someone’
tainía- ‘sell something’ tainié- ‘sell something to someone’
puha-/puhi- ‘remove something’ puhé- ‘remove from someone’ (e.g. the load
from an animal)
nulá- ‘ask something’ nuré- ‘ask to someone’
u’i- ‘bring something’ u’é- ‘bring to someone (sbj.sg)’
po’á ‘take something away’ po’é ‘take away from someone’

That the additional object associated with the final e verb form is an animate (mostly
human) participant, is obvious in (68), where two forms of the verb of believing are
provided. In this case, the vowel alternation is not associated with transitivization, but
with the (in)animacy of the object participant. Both verb forms are transitive, but, as
can be observed in (68), the final e verb form is for a human object, and the final a verb
form is for an inanimate object.

(68) a. pichiká- ‘believe in something’

b. pichiké- ‘believe in someone’

In this example, the a/e alternation does not cause a valency increase, it just changes
the type of object participant involved in the situation, being the e verb form the one
associated with human participant.
Regarding the optional or obligatory type of constructions encoded via final stem
vowel alternation, there is a strong tendency for these constructions to be obligatory.
However, some cases of optional constructions encoded via final stem vowel alternation
are possible. This possibility is found in TA and GU, only when the additional object is
a beneficiary. In this case, the verb stem ending in a/i can be used with the benefactive
human participant marked with a benefactive postposition. This possibility to have an
optional construction encoded via final stem vowel alternation is exemplified in (69)
from Norogachi TA.

(69) a. pe tami gite newa pa gema-ka.

only us for make.tr part.expl blanket-expl
‘He makes a blanket only for us.’
(Brambila 1976: 547)
536 Albert Álvarez González and Zarina Estrada-Fernández

b. raberi ne newe-ke.
violin I make.appl-pfv
‘I made a violin for him.’
(Brambila 1976: 379)

Double suffixation is also found in TA for encoding applicative constructions. In Chogu-

ita TA, the second suffix is always the benefactive -ki, while in Norogachi TA, this is the
suffix -ri.10 According to Caballero (2008: 245–247), this double applicative that implies
the loss of the final vowel of the first applicative suffix (-n-ki, -w-ki, -š-ki), is redundant
in Choguita TA, since there is no semantic change in comparison with the simple suffix-
ation, as illustrated in (70).

(70) a. ne mi pakó-ni-ra plato.

1sg.nom 2sg.acc wash-appl-pot plate
‘I’ll wash the plates for you.’
b. ne mi pakó-n-ki-ra plato.
1sg.nom 2sg.acc wash-appl-appl-pot plate
‘I’ll wash the plates for you.’
(Caballero 2008: 247)

In Norogachi TA, according to Brambila (1953: 187), the double applicative is mainly
used with transitive verbs in order to indicate the presence of both direct and indirect
objects, without the necessity to overtly mark these two complements outside the verb
as arguments. Recall that in TA, GU, PB, and NT, 3rd person direct and indirect object
pronouns are usually dropped. Two examples of this use are given in (71). Interestingly,
in this double applicativization, the first applicative suffix licenses as object an inan-
imate patient participant, and the second one, a human recipient/beneficiary partici-
pant.

(71) rikibu- ‘download’ natabu- ‘drill’

rikibú-ne ‘download (it)’ natabú-ne ‘drill (it)’
rikibú-ne-ri ‘download (it to someone)’ natabú-ne-ri ‘drill (it for someone)’

The double marking can also be the result of the combination of the directed verb stem
ending in é and an applicative suffix (-ki in Choguita TA, -ri in Norogachi TA), as exem-
plified in (72) from Choguita TA.

10 Brambila (1953: 180) regards the forms -ki, -gi, -ti, -ri as allomorphs, since in Norogachi TA, /r/ can
freely alternate with /k/, /g/, and /t/ in intervocalic contexts (Brambila 1953: 6). This allomorphy can
cause some potential ambiguities between both causative and applicative meanings, since in this variety
the suffix -ri is also a causative suffix.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 537

(72) a. ma=ni mi suwé-ri remé.

already=1sg.nom 2sg.acc finish.up.appl-pst tortillas
‘I already finished (ate) up your tortillas.’
b. ma=ni mi suwé-ki-ri remé.
already=1sg.nom 2sg.acc finish.up.appl-appl-pst tortillas
‘I already finished (ate) up your tortillas.’
(Caballero 2008: 239)

Again, these cases of double marking always involve the productive benefactive suffix
-ki, and they do not imply in Choguita TA a semantic difference in comparison with the
simple marking. Caballero (2008) explains this situation by the morphological opacity
caused by the fusion of the applicative marker -é to the verb base, which triggers the
double marking.
In Norogachi TA (Brambila 1953: 187), the same combination is also possible, as
shown in (73), with the same discursive function as the one mentioned for the double
applicative suffixation (see Example [71]).

(73) achá- ‘put’ bujá- ‘take away’

aché- ‘put (it)’ bujé- ‘take (it) away’
aché-ri ‘put (it to someone)’ bujé-ri ‘take (it) away (from someone)’

Probably triggered by this double marking, Norogachi TA shows some cases in which
the vowel alternation to é and the applicative suffixes -e, -ne are used to refer to an
inanimate patient, as shown in (74), not to a human non-patient participant as usual, as
well as some other cases in which the suffix -ri, the most productive causative marker
in the language, is used alone as an applicative marker (75b).

(74) ro’á ‘pour something’ ro’é ‘pour (it)’

ku ro’á ba’wi-ki! chokirá tsa ro’é-ma ne?
again pour water-expl near inter pour-fut 1sg.nom
‘Pour the water again!—Do I pour (it) near (the cross)?’
(Brambila 1953: 179)

(75) a. ne kobi-ka ariché-re ke rapako.

1sg.nom make_pinole-ger spend_the_day-pfv expl yesterday
‘Yesterday I spent the day making pinole.’
b. ketza tami kobí-ri-ma pe tá?
not_true 1sg.acc make_pinole-appl-fut neg little
‘Won't you make me some pinole?
(Brambila 1976: 257)
538 Albert Álvarez González and Zarina Estrada-Fernández

In GU, double applicatives are infrequent but still possible. An example is given in (76)
in which the two applicative suffixes are associated with the expected semantic values.
The first applicative suffix is used to introduce an animate recipient, the second one is
used to introduce a human beneficiary.

(76) čuh- ‘be hung’

čuh-čá ‘hang something’
čuh-čé ‘hang something on an animal’
čuh-čé-nie ‘hang something on an animal for someone’
(Medina 2002)

This example also serves to illustrate the allomorphy involved in the applicative suf-
fixes in GU. Miller (1996: 160) identifies different allomorphs for the applicative suf-
fixes (-ne/-nie/-íni; -é/-ie; -ké/-ge, kie/-gie), and this allomorphy seems to be lexically con-
ditioned.
Miller (1996: 161) points out that, with some verb bases, two different applicative
suffixes can be used with a semantic contrast: suffix -ne/-nie for benefactive, -ke/-kie/-gie
for surrogative.11 Some examples are presented in (77).

(77) četé-gie play a musical instrument instead of another person

četé-nie play a musical instrument for another person
toe-ké run with the ball instead of another person
toe-né run with the ball for helping another person
yahča-kie put the load (on a donkey) instead of another person
yahča-nie put the load (on a donkey) for helping another person

Other minor and unproductive applicative suffixes used for applicativization in Noro-
gachi TA are the suffix -ši exemplified in (78) (usually associated with a few motion verbs,
also present in Choguita TA (79) and in GU as -se12), and the suffix -mi, exemplified in (80),
which is only found with two verbs in Norogachi TA and with a benefactive meaning.

(78) e ‘remove’ é-ši ‘remove from someone’

pa ‘throw’ pá-ši ‘throw to someone’
ropa ‘outperform (sg)’ ropá-ši ‘outperform someone’
toba ‘outperform (pl)’ tobá-ši ‘outperform someone’
wichi ‘fall (sg)’ wichí-ši ‘fall to someone (sg)’
ruji ‘fall (pl)’ rují-ši ‘fall to someone (pl)’

11 Zúñiga (2014) proposes the notion of surrogation for a particular subtype of benefaction in which
the beneficiary “benefits from the fact that s/he does not have to perform a particular action thanks to
the intervention of the surrogate”.
12 Miller (1996: 162) mentions only one case: the verb ihpába-se ‘throw-appl’.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 539

(79) a. pá-ka!
throw-imp.sg
‘Throw it!’
b. tamí ku pá-ši-ri pelóta!
me rev throw-appl-imp.sg ball
‘Throw the ball back at me!’
(Caballero 2008: 415)

(80) a ‘search’ á-mi ‘search for someone’

nakare ‘cut off ears’ nakaré-mi ‘cut off ears for someone’

Finally, in River GU, Félix (2007: 126–128) has identified locative, instrumental and com-
itative applicativizations, in which the corresponding postposition of the base construc-
tion is attached to the base verb stem in the derived construction in order to license the
applied object. No mention of these applicative constructions is made by Miller (1996)
for Mountain GU and, according to subsequent studies on applicativization in River
GU (Ávila 2012; Casas 2018), these non-benefactive optional applicatives in GU are not
really accepted, at least for some native speakers of River GU (Casas 2018: 167). These
cases are presented below.

(81) a. waní simi-ré tiendá-čí.

John go-pfv store-loc
‘John went to the store.’
b. waní simi-ri-áči tiendá.
John go-pfv-appl store
‘Juan went-to the store.’

(82) a. hustína wičó-na wakirá haóni-e / ooná-e.

Agustina wash-prs shirt soap-ins salt-ins
‘Agustina washes the shirts with soap/salt.’
b. hustína wičo-ná-e wakirá haóni / ooná.
Agustina wash-prs-appl shirt soap salt
‘Agustina washes the shirts with soap/salt.’

(83) a. maría simi-ré obregón ahpó ye'yé-ma.

Mary go-pfv Obregón 3sg.nsbj mother-com
‘María went to Obregón with her mother.’
b. maría simi-ré-ma obregón ahpó ye'yé-ma.
Mary go-pfv-appl Obregón 3sg.nsbj mother-com
‘Did Mary go to Obregón with her mother?’
540 Albert Álvarez González and Zarina Estrada-Fernández

In the case of the comitative applicativization in (83b), the presence of the postposition
as a verbal suffix is not accompanied by the encoding of the original postpositional
complement as a direct object. This participant still has the postpositional marking.
No valency-increasing is present here. Additionally, as indicated by the translations in
(82) and (83), these constructions would imply a change in the communicative function
from indicative to interrogative.
In sum, two main mechanisms can be identified in TA and GU for the marking of
the undirected/directed alternation. The first one is morphological and it corresponds
to the applicative suffixes illustrated in this section. The second one is morphophono-
logical and it refers to the final vowel alternation in which the verbal stem ending in
e is used to encode the directed verb. It is clear that both marking strategies are his-
torically related. According to Brambila (1953), Miller (1996) and Caballero (2008), the
vowel alternation seems to be the consequence of the e- suffixation but it is also possible
that the final vowel alternation is the oldest system13 and that new markers have been
created by combining the final vowel alternation and different types of transitivizing
suffixes, as suggested by the examples of -na, -tza, -wa ‘caus’ vs. -ne, -tze, -we ‘appl’.14
Independently of the chronological relations between both strategies, they are clearly
interconnected and they illustrate how morphology and phonology can be related in
the encoding of the undirected/directed alternation.
Another example of how morphophonology is used for marking the undirected/
directed alternation is exemplified in (84) from Choguita TA. In (84b), the benefactive
applicative suffix -i seems to have fused to the verbal stem, with the consequence that
the encoding strategy for the undirected/directed alternation is now reduced to the
single stress shift of the final vowel i. The alternation is in this case between a final
unstressed i, which is associated with the transitive stem and a final stressed i associ-
ated with the directed transitive stem.

(84) a. nihé ba’arí iči-méa muní.

1sg.nom tomorrow plant-fut.sg beans
‘I will plant beans tomorrow.’
b. nihé ba’arí ne yé-ra ičí-ma.
1sg.nom tomorrow 1sg.nom mother-poss plant.appl-fut.sg
‘I will plant for my mom tomorrow.’
(Caballero 2008: 238–9)

13 The presence of an old verbal pattern involving the change of the final vowel of the stem for distin-
guishing between intransitive and transitive verbs has been proposed for the PUA by Langacker (1977:
132). In his reconstruction, this non-affixal variation corresponds to a general UA i/a distinction, with a
for transitives and i for intransitives.
14 Regarding the suffix -é, Miller (1996: 160) considers it to be the most basic applicative suffix and to be
probably included in the other GU applicative markers.
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 541

This possibility to encode the undirected/directed alternation via stress shift is also
found in GU with the verb nahté ‘pay’, which is the lexicalized directed version of nahte
‘cost’ (Miller 1996: 362).

4 Syntax, semantics and pragmatics

of applicativization

4.1 The status of the object in applicative constructions

from transitive verbs

Although applicative constructions from transitive base verbs in YQ, PB, NT, TA and GU
are double object constructions, the original object tends to become a low prominence
object, since the only object that can be passivized is usually the applied object, showing
cases of asymmetric applicatives (Pylkänen 2008). This asymmetry is exemplified in
(85) with data from YQ.

(85) a. María-∅ u-ka toto’i-ta jinu-ria-wa-k.

Mary-nom det-acc hen-acc buy-appl-pass-pfv
‘Mary was bought the hen.’
b. ✶ U toto’i-∅ María-ta jinu-ria-wa-k.
det hen-nom Mary-acc buy-appl-pass-pfv
Intended: ‘The hen was bought for Mary.’
(Guerrero 2007: 198)

The only language in which symmetric applicatives have been found is GU. Example
(86b) from River GU show the passivization of a benefactive applicative transitive
marked by the suffix -ke, in which the subject is the applied object (ne ‘1sg.nom’), while
the subject is the original object (muní ‘beans’) in (86c).

(86) a. Hustína no’ó pasú-ke-re muní.

Agustina 1sg.nsbj cook-appl-pfv beans
‘Agustina cooked beans for me.’
b. pasu-ke-ré-tu=ne muní (Hustína-e).
cook-appl-pfv-pass=1sg.s beans Agustina-ins
‘I was cooked beans (by Agustina).’
c. muní no’ó pasu-ke-ré-tu (Hustína-e).
beans 1sg.nsbj cook-appl-pfv-pass Agustina-ins
‘Beans were cooked for me (by Agustina).’
(Félix 2005: 259)
542 Albert Álvarez González and Zarina Estrada-Fernández

The same possibility is present in Mountain GU. In (87a), the verb wera ‘put’ that has
been applicativized by the suffix -ié, and passivized by the suffix -ru, has as subject the
original object (the theme sigori ‘pot’), while this subject is the applied object (the recip-
ient remé ‘1PL.SBJ’) in (87b).

(87) a. Pié sigori wer-ié-ba-ru=ra isuki wa’a.

one pot put-appl-inch-pass.pst=rpt tesgüino over_there
‘They put him (the coyote) a pot of tesguino over there.’
(Miller 1996: 97)
(Lit. They say that a pot was put (to him), tesgüino over there.’)
b. Pa=remé yomá to’-é-reru pa?wi.
already=1pl.sbj all put-appl-pass.pst water
‘We’ve already been watered.’
(Miller 1996: 98)

4.2 Applicativization, valency and discourse

Applicativization in UA languages of NW Mexico always implies a valency increase, in

which a non-agent and usually non-patient animate (mostly human) participant has
been added to the situation denoted by the base construction, encoded as object. The
only rare and dubious exception that has been found is the comitative applicativization
in GU exemplified in (83b), in which the expected valency increase is not present since
the comitative participant retains its oblique marking in the applicative version. As for
the undirected/directed alternation encoded by the equipollent marking, the absence of
valency increase is also observed in only one case, that is the final stem vowel alterna-
tion involved in the verb pair of GU pichiká ‘believe in something’ / pichiké ‘believe in
someone’ (see Example [68]).
In cases of optional applicatives, the applied participant is promoted from an
oblique to a core function, implying the acquisition of more topicality and more prag-
matic prominence in comparison with its original position in the base construction.
This explains why in discourse, applicativization can be used as a topicalization device
for referential discourse continuity. This is the case for instance in YQ with Example
(88), taken from Álvarez (2019).

(88) Es ké=nee a=tá’aru-ria-k

the_fact_is_that=1sg.acc 3sg.acc=lose-appl-pfv
úme ilí jaámuch-im áma’a.
det.pl little woman-pl there
‘The fact is that my daughters lost it.’
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 543

This example is extracted from a conversation in which the speaker is talking about
himself. After a few clauses in which the speaker is referring to himself as the subject
of the clauses, comes the exemplified clause in which the applicativization is accompa-
nied by the right-dislocation of a different subject (ume ilí jaámuchim ‘the girls’), which
allows the applied object (referring to the speaker: nee ‘1sg.acc’) to occupy the first posi-
tion after the focus structure borrowed from Spanish es ké ‘the fact/reason is that’. In
(88), the speaker is thus using simultaneously a focus structure, applicativization, and
right-dislocation to topicalize himself and to maintain referential discourse continuity.
Applicativization can also be used as a focalization device. This possibility is illus-
trated in (89) from Norogachi TA, with the verb rara ‘to buy’. In (89a), the subject is in
initial position, followed by the benefactive oblique complement, and by the predicate
with the base verb and the object complement in final position. The applicativization
in (89b) serves to focalize the benefactive participant, which appears in the initial posi-
tion and marked by the emphatic particle ka (the subject pronoun between the applied
object and the emphatic ka is a second position clitic pronoun).

(89) a. mujé ko pe kúrui gite rara-re bera kawa.

2sg.nom conj only children for buy-pfv conj eggs
‘But you bought eggs just for the kids.’
(Brambila 1976: 180)
b. tewé ne ka súkaro rarí-i-ma.
daughter 1sg.nom emph sugar buy-appl-fut
‘I will buy sugar for my daughter.’
(Lit. ‘For my daughter, I will buy sugar.’)
(Brambila 1953: 183)

The focalization of the directed participant can also be expressed by the equipollent
marking of the undirected/directed alternation. The examples in (90) from GU (Miller
1996: 101) illustrate this possibility with the grooming verb u’upa ‘to bathe’.

(90) a. u’upa-ma=ne.
bathe.intr-fut=1sg.nom
‘I am going to bathe.’
b. no’ó u’upé-ma=ne (wa’ábi).
1sg.nsbj bathe.appl-fut=1sg.nom self
‘I am going to bathe.’
(Lit. To me, I am going to bathe.’)

In (90a) and (90b), two different constructions (intransitive-reflexive in [90a] and transi-
tive-directed in [90b]) are used for denoting the same reflexive situation. The difference
lies of course in the discourse and pragmatic functions associated with these construc-
tions. While (90a) is the unmarked construction, (90b) is used to indicate information
544 Albert Álvarez González and Zarina Estrada-Fernández

that is contrary to the presupposition of the interlocutor, that is a contrastive focus. Note
again that, like in applicativization in (89b), the directed participant appears as the first
and most topical element of the clause, here as an example of focus fronting (Krifka
2008; Neeleman and Vermeulen 2012).

4.3 The semantic contrast in optional applicatives

In optional applicatives, the applicative construction with the applied object and the
base construction with the same participant encoded as a postpositional object can
present some subtle semantic differences. For instance, in YQ, the activity denoted by
the base verb bwika ‘sing’ is carried out on behalf of someone if this human participant
is encoded as a postpositional object (91a), and for the benefit of someone if the human
participant is an applied object (91b).

(91) a. aapo bwika-k e-betchi’ibo.

3sg.nom sing-pfv 2sg.obl-for
‘He/she sang on behalf of / instead of you.’
b. aapo enchi bwika-ria-k.
3sg.nom 2sg.acc sing-appl-pfv
‘He/she sang for you.’
(Guerrero 2022: 30)

Another example involves the verb jima ‘throw’. In the base construction in (92a), the
human participant marked by the directional postposition -u is a human goal/recipient
and the clause is understood as Peter and the child are playing ball together. In the
applicative construction in (92b) the reading is malefactive and the clause is understood
as Peter threw the ball at the child in order to hurt him or hit him.

(92) a. Peo-∅ ili usi-ta-u pelotam jima-k.

Peter-nom little child-acc-dir ball.pl throw-pfv
‘Peter threw a ball to the boy.’ (goal/recipient reading)
b. Peo-∅ ili usi-ta pelotam jima-ria-k.
Peter-nom little child-acc ball.pl throw-appl-pfv
‘Peter threw a ball (at) the boy.’ (malefactive reading)
(Guerrero 2022: 34)

With the verb etbwa ‘steal’, the applied object is associated with a source/malefactive
meaning (93a), while the postpositional object introduced by betchi’ibo triggers the ben-
efactive meaning (93b).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 545

(93) a. Goyo-∅ Aurelia-ta u-ka toto’i-ta etbwa-ria-k.

Goyo-nom Aurelia-acc det-acc hen-acc steal-appl-pfv
‘Goyo stole the hen from Aurelia.’
b. Goyo-∅ u-ka toto’i-ta etbwa-k Aurelia-ta-betchibo.
Goyo-nom det-acc hen-acc steal-pfv Aurelia-acc-for
‘Goyo stole the hen for Aurelia.’
(Guerrero 2007: 189)

In the cases of locative, instrumental and comitative applicatives in GU, Félix (2007)
points out the existence of a semantic contrast (not always very clear) between the
applied and the postpositional versions. The locative applicative implies that John
remained in the store for a longer period of time in (81b), while instrumental and comi-
tative applicatives in (82b) and (83b), respectively, provide unexpected information and
are therefore perceived by the native consultant as a question (Félix 2007: 130).

4.4 The use of applicativization for discourse coherence

As mentioned in Section 2, third person object pronouns in TA, GU, PB, and NT are
usually dropped in discourse, when context is clear enough to identify these object
arguments. This dropping is frequently associated with the applicative marking in these
languages.
As Brambila (1953: 53) pointed out, in Norogachi TA accusative and dative pro-
nouns of applicative verbs are often dropped in discourse because “they are included
in the verb”. The applicative suffixes express the relation between a verb form and its
directed complements, without the necessity to express this complement in the clause,15
and the presence of the applicative suffix (or the directed verb stem ending in e) is often
the unique indicator of the presence of this additional (most commonly, human) partic-
ipant, thus functioning as an important device for referential continuity.
In Norogachi TA discourse, it is thus frequent to have no (pro)nominal mention of
the applied object. The applied object is left implicit, being anaphorically present and,
therefore, contextually retrievable. In this case, only the verb form implies the presence
of the applied object. Although the semantic valency has been increased (a new partic-
ipant has been added to the situation denoted by the base construction), the syntactic
expression of this increasing is not always present.
Estrada (2007: 97–101, 2014: 213) also mentions the same phenomenon in PB, which
also allows the non-specification of applied arguments. The presence of the applicative
marker is sufficient to indicate that the situation denoted by the base verb is applied

15 The applicative verb is named by Brambila (1953: 176) a “relative verb”, since “it includes the rela-
tion to its complements” (our translation).
546 Albert Álvarez González and Zarina Estrada-Fernández

to an animate third person participant, even though this non-specification may cause
sometimes some ambiguities. For GU, Miller (1996: 97) also mentions that, in discourse, it
is common to omit the applied object when the context is clear. Félix (2005: 259–260) also
states that all third person participants can be omitted, like in the transitive construction
(94a), in the applicative construction (94b) or even in its passivized version (94c).

(94) a. pasu-ré.
cook-pfv
‘She/he cooked it/them.’
b. pasú-ke-re.
cook-appl-pfv
‘She/he cooked it/them for him/her/them.’
c. pasu-ké-re-tu.
cook-appl-pfv-pass
‘She /he was cooked something.’ /
‘Something was cooked for him/her/them.’

The use of the applicativized verb to indicate the presence of an object complement
without the need to overtly encode it outside the verb as a (pro)nominal argument,
is obviously favored by the fact that 3rd person object pronouns are dropped, but this
object omission is also possible with 1st and 2nd person pronouns, as exemplified in (95b)
from Norogachi TA. In this case, since the object is not expressed but only implied, the
function is alike a pragmatic antipassivization, when compared with the construction
without the applicative suffix in (95a).

(95) a. tasi chi ku'wi karlo.

neg me help Carlos
‘Carlos does not help me.’
(Brambila 1976: 280)
b. bineri tza mu gayena? képi-o, bonirá ku’wi-ri-re.
alone perhaps 2sg.nom do neg-emph brother help-appl-pfv
‘Did you do it alone?—No, my brother helped (me).’
(Brambila 1953: 182)

In this use, the implied object complement is not always a human participant, it can also
be an inanimate patient (this possibility is usually found with the directed verb stem
ending in e as in (74) or with the applicative suffixes -e or -ne).16 However, Brambila

16 The examples proposed by Brambila (1953) for exemplifying the uses of the different applicative
suffixes in Norogachi TA show that the suffixes -(C)i (-i, -ri, -ki, -gi, -mi) have a strong tendency to combine
with transitive base verbs in order to add an animate participant as a beneficiary, while the suffixes -(C)
e (-e, -ne, -tze, -we) are usually combined with transitive and intransitive verbs in order to add an object
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 547

(1953: 188) pointed out that, in this case, the relative form is more often substituted by
the verb form ending in i/a. In the same vein, when the directed verb form is used, the
absence of the object complement is obviously not mandatory, and the object comple-
ment can thus be overtly encoded.17
As mentioned above (Example [74]), this use in discourse could have favored
the need for a double suffixation in cases of ditransitive verbs, since two objects are
involved with these verbs. The use of the double marking of the directed relation could
also have opened up the possibility to have inanimate participants added as implicit
applied objects, something that can be found with the directed verb stems and some
applicative suffixes in Norogachi TA but that is very infrequent in applicative construc-
tions of the other UA languages of NW Mexico.

5 Causative-applicative syncretism
The causative-applicative syncretism is pervasive in UA languages from NW Mexico. It
appears with the suffix -di/-id in PB and the suffix -d (y)i/-d(y)a in NT, which act as causa-
tivizers if the base verb is an inactive intransitive verb, and as applicativizers if the base
verb is an active intransitive or a transitive verb.
Additionally, in PB, the use of the suffix -di/-id with some active intransitive verbs
triggers a sociative causative reading. This is the case for example with the verbs hi’a
‘urinate’ (96a) and tikpan ‘work’ (96b).

(96) a. Marii lii oob hiaa-di.

María dim.sg person urinate-appl
‘Mary took the child to urinate.’
(Estrada 2014: 215)
b. ig di’ir ilbaah tikpan-di-a.
det.sbj mother girl work-appl-pros
‘The mother will take the girl to work.’
(Estrada 2014: 215)

complement that can be a human participant with a diversity of semantic roles or even an inanimate
patient. The directed verb stem ending in e has a behavior similar to the suffixes -(C)e.
17 As Brambila (1953: 188) also acknowledged, the use of the relative verb forms is not always consist-
ent. The directed verb form can be used with the object complement expressed as a pronoun or as a
noun phrase, and some transitive verbs ending in i/a can be used intransitively, that is, without an object
complement, although in the context, it is clear that the situation includes an object participant, as in (i).

(i) má ne kašina. instead of má ne kašine.

already 1sg.nom break.intr already 1sg.nom break.appl
‘I have already broken.’ ‘I have already broken (it).’
548 Albert Álvarez González and Zarina Estrada-Fernández

In GU, the suffix -če, which has been exemplified in its applicative use in (50), can also
have a causative function with stative or inchoative intransitive base verbs.

(97) čeha-, cehi-, če- ‘be stung’ če-če- ‘sting, vaccinate’

tahtá- ‘be warm’ táhta-pa- ‘be hot-inch ‘get warm’ táhta-pa-če ‘heat’

The causative-applicative syncretism is also present in Norogachi TA with the causative

suffix -ri that acts as an applicative marker when used as the second element in double
applicatives (see Examples [71]). In some cases, it can even serve as an applicative
marker in simple suffixation (see Example [75b]). As mentioned before (see Footnote
10), this situation may cause cases of ambiguity between causativization and applica-
tivization (Brambila 1953: 180).
The derivational uses of applicative suffixes in UA languages from NW Mexico also
illustrate the causative-applicative syncretism, since these suffixes also serve to create
causative verbs usually out of adjectives and/or nominals. For instance, the suffix -di/-id
in PB can be used as a denominal factitive verbalizer, that is, a derivational marker cre-
ating active verbs out of nouns with the general meaning ‘to exert on a patient an action
related with N’ (Estrada 2007: 92–93).

(98) suusk ‘shoes’ suusuk-id ‘to shoe (horse)’

hivil ‘wind’ hivil-di ‘to fan, to air’
kubis ‘smoke’ kubis-di ‘to smoke’
si’i ‘breast’ si’i-di ‘to feed’

It can also be attached to adjectives/stative verbs in order to create a causative verb

with the general meaning ‘to cause the state denoted by the base’.

(99) onmag ‘be salty’ onama-d(i) ‘salt’

kaplik ‘be small’ kap-id-ir ’shrink’
doa ‘be cured’ doali, doar < doadi ‘cure’
tuuk ‘be dark’ tuk-id ‘darken’
tipilik ‘be flat’ tipilik-id ‘flatten’
toahk ‘be white’ toah-id ‘whiten’
i’ov ‘be sweet’ i’ov-id ‘sweeten’
u’uv ‘be odorous’ uuvag-id ‘sniff’

The suffix -d(y)i/d(y)a in NT can also be used with this same verbalizing effect (Bascom
1982: 299–300), as exemplified in (100) and (101).

(100) kukúruš-dya-ñi!
cross-vbzr-imp
‘Make the sign of the cross!’
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 549

(101) viiš tása-i vaamí-óma to~tóíñ-dya-ri-i.

all day-abs more-compar rdp~cont-hot-vbzr-become-prs
‘Every day it keeps getting hotter.’

The same type of verb formation is also found in YQ with the applicative suffix -ria
(Dedrick and Casad 1999; Álvarez and Estrada 2009; Guerrero 2007: 185), as shown in
(102) from nominal bases and in (103) from adjectival bases.

(102) bwichopia ‘smut’ bwichop-ria ‘smoke’

pajko ‘party’ papajko-ria ‘celebrate’

(103) suka ‘hot’ suka-ria ‘heat’

awi ‘fat’ awi-ria ‘fatten’
bali ‘cold’ bali-ria ‘cold, freeze’

This formation of active transitive verbs is also possible with at least one adverbial base
and one intransitive base verb.

(104) bu'u ‘too, a lot’ bu’u-ria ‘increase, accumulate’

yo’otu ‘grow’ yo’otu-ria ‘raise, cultivate, make grow’

Some examples of active verbs created out of nouns via applicative suffixation in (105),
and via equipollent marking in (106), can also be found in Norogachi TA.

(105) nakí ‘addendum’ naki-we ‘add’

wesó ‘mud’ weso-ge ‘mire’

(106) niwí ‘nostalgia’ newe ‘long for someone’

sawí ‘coal’ sawe ‘carbonize’

Additionally, the causative-applicative overlap can also be illustrated with the verb
weiyaa ‘carry’, which has been analyzed above as probably involving an applicative
periphrasis. However, the meaning of this verb can also be rendered by a causative
periphrasis from the base verb weye ‘go’, and Example (43b) could be thus interpreted
as a causative construction, with the meaning ‘The woman is making the water go in
the pot.’
Lastly, the same polyfunctionality found with the suffixes used in applicativiza-
tion and causativization can also be observed in the equipollent marking in which the
change of the final verb stem vowel from i/a to e in TA and GU, is associated with the
non-causal/causal alternation if the verb stem ending in i/a is an inactive intransitive,
and with the undirected/directed alternation if the stem ending in i/a is an active intran-
sitive or a transitive verb.
550 Albert Álvarez González and Zarina Estrada-Fernández

6 Applicative lookalikes

6.1 Lexicalized applicatives

Several verbs in the UA languages from NW Mexico end in the form found in applicative
constructions but, in a strictly synchronic perspective, they are not clearly analyzable
as applicative derivates, since the meaning of the base verb has changed in such a way
that the derived construction cannot be interpreted as the same situation denoted by
the base verb but with the addition of a new object participant. However, a seman-
tically plausible relationship between the meanings of the base construction and of
the derived construction can usually be identified, justifying the use of the applicative
marker at least in diachrony and via a semantic shift (usually, a metonymy). In these
cases, exemplified in (107) for GU, YQ and NT, it is possible to consider these verbs as
lexicalized applicatives, or pseudo-applicatives.

(107) GU ihpa- ‘throw something (sg.obj)’ ihpá-ge, ipa-ké ‘milk (the cow)’
tui- ‘say’ tu-ké ‘ask’
tari ‘buy’ tari-ké ‘sell something to someone’
YQ wike ‘drag, pull’ wiki-ria ‘owe’
siime ‘go (sg.subj)’ sim-ria ‘abandon someone’
NT gáágai ‘look for’ gaagí-dya ‘provide for’

As expected, the same lexicalization is possible with equipollently-marked verb pairs,

as shown in (108) from GU.

(108) GU po’táča- ‘tie’ po’tačé ‘close with lid’

yahčá- ‘seat, put’ yahčé- ‘serve someone (food)’
sipa ‘scrape’ sipé ‘do harm, bewitch’
peni ‘learn’ pené ‘know how to do, have the habit of’
u’natá ‘think about, remember’ u’naté ‘care’

6.2 Applicative deponents

Several verbs in the UA languages from NW Mexico only occur with an ending that
could be the marker that distinguishes base and applicativized versions of other verbs,
but only the seemingly applicativized verb is existing, and the expected base verb does
not exist in the language. These cases exemplified in (109) with data from YQ and GU
can be viewed as cases of applicativa tantum or applicative deponents (Zúñiga and Cre-
issels, this volume).
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 551

(109) YQ e’e-ria ‘keep’ ✶

e’e-
nanaj-ria ‘avoid someone’ ✶
nanaj-
koo-ria ‘rock someone’ ✶
koo-
ku-ria ‘turn, tangle’ ✶
ku-
jijja’ria ‘scare’ ✶
jijja’
GU inatu-ké ‘ask’ ✶
inatu-
ihkó-ke ‘offer something to one person’ ✶
ihkó-
ihkó-ge ‘distribute, offer’
ihkó-nie ‘offer something to several persons’
nate-ké ‘forget’ ✶
nate-

Estrada (2014: 178) points out that many ditransitive verbs in PB are diachronically
derived with the applicative suffix -id/-ir, although in synchrony, the base verb is no
longer present in the language. Some examples are:

(110) PB bid ‘serve something to someone’

ho’ir ‘offer something to someone’

7 Conclusions
In this chapter, the applicative constructions attested in UA languages of NW Mexico
have been described, as well as the pervasive causative-applicative syncretism found
in the data. Following the questionnaire proposed as a guideline for the contributions
to this volume, the main aspects of applicative constructions in UA languages of NW
Mexico are as follows:

Morphology
– Applicativization is made via suffixation in all these languages (§ 3), but some lan-
guages use only one suffix (PB, YQ), while others have several applicative suffixes
(NT, TA, GU). The marking of the undirected/directed alternation by means of equi-
pollent marking or stress shift, found in TA and GU, probably resulted historically
from the fusion of a suffix with the base (§ 3.4).
– The possible vestiges of an applicative periphrasis have been found in YQ (§ 3.3).
This unproductive strategy involves a functional verb iyaa ‘to pretend, to wish’.
– The only applicative suffix that does not show allomorphy is the suffix -ria in YQ.
In the other cases, the allomorphy seems to show different kinds of conditioning:
phonological (suffix -di/-id in PB), morphological (suffix -d(y)a is used for imperative,
-d(y)i for imperfective, and -tuli for perfective in NT, Footnote 7) or lexical (applica-
tive suffixes in TA and GU).
552 Albert Álvarez González and Zarina Estrada-Fernández

– Except for the above case of morphologically-conditioned allomorphy in NT, inflec-

tional verb morphology does not seem to be affected by applicative suffixation in
UA languages of NW Mexico.

Syntax
– Applicativization in UA languages of NW Mexico is generally not allowed for inac-
tive intransitive base verbs, and benefactive applicativization shows a strong pref-
erence for transitive base verbs.
– UA languages of NW Mexico only have P-applicativization, in which the applied
phrase is a noun phrase showing all the properties that characterize objects in
non-applicative constructions. The only case of X-applicativization exemplified in
(83b) is problematic.
– Applicatives in UA languages of NW Mexico are always valency-increasing con-
structions, in which the syntactic status of the applied phrase’s companion argu-
ments/adjuncts does not change in comparison with the base construction.
– Double marking is found in NT, TA and GU, and it can be the result of double
applicative suffixation or the combination of a directed verb stem and the bene-
factive applicative suffixation. This double marking is redundant in Choguita TA
(§ 3.4), it is the main form to express benefactive applicatives in NT (§ 3.2), while in
Norogachi TA, it is mainly used for referential discourse coherence (§§ 3.4 and 4.4).
– There is no difference between applicative constructions and constructions with
underived predicates belonging to the same valency class.
– Optional applicatives in UA languages of NW Mexico are not conditioned by limited
access of obliques to some syntactic operations.

Semantics
– The applicative suffixes found in UA languages of NW Mexico are usually markers
that can license, depending on the meaning of the verb base, all the different non-
agent and non-patient semantic roles that can usually be associated with animate
(mostly human) participants (recipient, beneficiary/maleficiary, stimulus, source,
concernee, etc.). Their semantic under-specification is thus restricted by animacy
(mostly humanness). However, TA, GU and NT have developed dedicated applica-
tive markers for the benefactive meaning.
– With the exception of Norogachi TA in which the applied object can refer to an
inanimate patient, the semantic roles expressed by the applied phrase in UA lan-
guages of NW Mexico are those that can be associated with non-agent and non-pa-
tient animate (mostly human) participants. These participants can be peripheral
(such as a beneficiary) but they can also be central participants required by the
lexical meaning of the base verb. In this case, the applicative construction tends to
be an obligatory applicative construction in those languages in which no marking
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 553

difference is made between direct and indirect object (TA, GU, PB, and NT). By con-
trast, benefactive applicatives are always optional applicatives.
– Some semantic contrasts between the base construction and the applicative con-
struction have been pointed out in Section 4.3. These differences are sometimes
quite subtle and no clear semantic pattern has been detected, except in YQ in which
the applied object in applicative constructions from active intransitive verbs tends
to be associated with benefaction, while its counterparts encoded as a benefactive
postpositional object in the base construction is associated with surrogation (see
Examples in [91])
– Some cases of topicalization and focalization via applicativization have been men-
tioned in Section 4.2. Interestingly, they tend to involve the promotion of a speech
act participant (the speaker) to the object function and to the first position of the
applicative clause, suggesting that discourse referential continuity, topicaliza-
tion/focalization and speech act participants are motivating factors for the use of
applicative constructions.
– The distribution of applicativized verbs in UA languages of NW Mexico with the
dropping of object pronouns (TA, GU, PB and NT) is strongly dependent on their
uses as referential coherence markers in discourse (§ 4.4).

Lookalikes and others

– Non-applicative functions of the applicative markers are mainly associated with
the pervasive causative-applicative syncretism found in UA languages of NW
Mexico (§ 6).
– Lexicalized applicatives and applicative deponents are not rare in UA languages of
NW Mexico (§ 7).

Abbreviations
abs absolute
acc accusative
adjr adjectivizer
al alienable
ant anterior
appl applicative
ben benefactive
caus causative
com comitative
compar comparative
compl completive
cond conditional
conj conjunction
cont continuative
554 Albert Álvarez González and Zarina Estrada-Fernández

dem demonstrative
des desiderative
det determiner
dim diminutive
dir directional
dub dubitative
emph emphatic
expl expletive
ev evidential
fut future
ger gerund
hab habitual
imp imperative
ipfv imperfective
inch inchoative
ins instrumental
inter interrogative
intr intransitive
it iterative
loc locative
neg negative
nmlz nominalizer
nntr neutral number
nsbj non-subject
nom nominative
obj object
indf indefinite
obl oblique
part particle
pass passive
pcn past continuative
pst past
pfv perfective
pl plural
poss possessive
pot potential
prog progressive
pros prospective
prs present
ptcp participle
rdp reduplication
rpt reportative
sg singular
sbj subject
tr transitive
vbzr verbalizer
 16 Applicative constructions in Uto-Aztecan languages from Northwestern Mexico 555

References
Álvarez González, Albert. 2007. Construcciones de aspecto resultativo en yaqui. In Zarina Estrada
Fernández, Albert Álvarez González, Lilián Guerrero & Belén Carpio (eds.), Mecanismos de voz y
formación de palabra. 17–43. México & Hermosillo: Plaza y Valdez & Universidad de Sonora.
Álvarez González, Albert & Estrada Fernández, Zarina. 2009. Mecanismos de creación de palabras en lengua
yaqui. In Arzápalo Marín, Ramón (ed). Lingüística amerindia: contribuciones recientes, 147–168. México
DF: UNAM-IIA.
Álvarez González, Albert. 2019. Discourse coherence in narratives and conversations: A case study in Yaqui
(Uto-Aztecan). Anthropological Linguistics. 61(2). 250–309.
Álvarez González, Albert. 2021. The rise of cause/reason adverbial markers in Yaqui. In Alexander Haselow
& Sylvie Hancil (eds.), Studies at the grammar-discourse interface: Discourse markers and discourse-related
grammatical phenomena, 313–351. Amsterdam: John Benjamins.
Ávila Enríquez, Diana Amelia. 2012. Construcciones aplicativas en guarijío de Sonora. Hermosillo: Universidad
de Sonora MA thesis.
Bascom, Burton. 1982. Northern Tepehuan. In Ronald W. Langacker (ed.), Studies in Uto-Aztecan grammar,
Vol. 3: Uto-Aztecan grammatical sketches, 267–393. Dallas, TX: Summer Institute of Linguistics and the
University of Texas at Arlington.
Bascom, Burton. 2003. Gramática del Tepehuán de Norte. México: SIL International
Brambila, David. 1953. Gramática Raramuri. México: Editorial Buena Prensa.
Brambila, David. 1976. Diccionario Raramuri-Castellano. México: La Obra de la Buena Press.
Buelna, Eustaquio. 1890. Arte de la lengua Cahita por un padre de la Compañía de Jesus, México: Imprenta del
Gobierno Federal
Burgess, Donald. 1984. Western Tarahumara. In Ronald W. Langacker (ed.), Studies in Uto-Aztecan grammar,
Vol. 4: Southern Uto-Aztecan grammatical sketches, 1–149. Dallas: Summer Institute of Linguistics.
Caballero, Gabriela. 2008. Choguita (Rarámuri) phonology and morphology. Berkeley: University of California
dissertation.
Carrillo Carrillo, Araceli. 2013. Formación de palabras en tepehuano del norte (odami). Hermosillo: Universidad
de Sonora MA thesis.
Casas Salido, Luis Ramiro. 2018. Mecanismos de aumento de valencia en guarijío. Hermosillo: Universidad de
Sonora MA thesis.
Creissels, Denis. Forthcoming. Transitivity, valency, and voice. Oxford: Oxford University Press.
Dedrick, John M. & Eugene H. Casad. 1999. Sonora Yaqui language structures. Tucson: University of Arizona
Press.
Estrada Fernández, Zarina, Crescencio Buitimea Valenzuela, Adriana E. Gurrola Camacho, María Elena
Castillo Celaya & Anabela Carlón Flores. 2004. Diccionario yaqui-español y textos: obra de preservación
lingüística. México: Plaza y Valdés.
Estrada Fernández, Zarina. 2007. Aspectos semántico-sintácticos de las construcciones aplicativas y su
manifestación en pima bajo. In Zarina Estrada Fernández, Albert Álvarez González, Lilián Guerrero &
Belén Carpio (eds.), Mecanismos de voz y formación de palabra, 85–105. México & Hermosillo: Plaza y
Valdez & Universidad de Sonora.
Estrada Fernández, Zarina. 2014. Gramática de referencia del pima bajo. Volumen I. Hermosillo: Universidad
de Sonora.
Estrada-Fernández, Zarina, Mercedes Tubino Blanco & Jesús Villalpando. 2015. Valency classes in Yaqui.
In Andrej Malchukov & Bernard Comrie (eds.), Valency classes in the world’s languages, Vol. 2: Case
studies from Austronesia, the Pacific, the Americas, and theoretical outlook, 1359–1390. Berlin: De Gruyter
Mouton.
Félix Armendáriz, Rolando Gpe. 2005. A grammar of River Warihío. Houston, TX: Rice University dissertation.
556 Albert Álvarez González and Zarina Estrada-Fernández

Félix Armendáriz, Rolando. 2007. Los aplicativos en warihío. In Zarina Estrada Fernández, Albert Álvarez
González, Lilián Guerrero & Belén Carpio (eds.), Mecanismos de voz y formación de palabra, 107–132.
México & Hermosillo: Plaza y Valdez & Universidad de Sonora.
Guerrero, Lilián. 2007. Estructuras argumentales alternativas: las cláusulas aplicativas en yaqui. In Zarina
Estrada Fernández, Albert Álvarez González, Lilián Guerrero & Belén Carpio (eds.), Mecanismos de voz y
formación de palabra, 177–204. México & Hermosillo: Plaza y Valdez & Universidad de Sonora.
Guerrero, Lilián. 2022. Typical and atypical applicative constructions in Yaqui. In Sara Pacchiarotti &
Fernando Zúñiga (eds.), Applicative morphology: Neglected syntactic and non-syntactic functions, 21–50.
Berlin: De Gruyter Mouton.
Harley, Heidi, Mercedes Tubino Blanco & Jason D. Haugen. 2009. Applicative constructions and suppletive
verbs in Hiaki (Yaqui). Rice Working Papers in Linguistics 1. 42–52.
Haspelmath, Martin. 1993. More on the typology of inchoative/causative verb alternations. In Bernard
Comrie & Maria Polinsky (eds.), Causatives and transitivity, 87–121. Amsterdam: John Benjamins.
Jelinek, Eloïse & Fernando Escalante. 2000. Unaccusative and unergative Verbs in Yaqui. In Eugene Casad &
Thomas Willett (eds.), Uto-Aztecan: Structural, temporal and geographic perspectives. Papers in memory of
Wick R. Miller by the friends of Uto-Aztecan, 171–182. México: Universidad de Sonora.
Krifka, Manfred. 2008. Basic notions of information structure. Acta Linguistica Hungarica, 55(3). 243–276.
Langacker, Ronald W. 1977. Studies in Uto-Aztecan grammar, Volume 1: An overview of Uto-Aztecan grammar.
Dallas: Summer Institute of Linguistics & University of Texas at Arlington.
Mason, J. Alden. 1917. Tepecano, a Piman language of Western Mexico. Annals of the New York Academy of
Sciences XXV. 309–416.
Medina Murillo, Ana Aurora. 2002. Diccionario morfológico: formación de palabras en el guarijío. Hermosillo:
Universidad de Sonora MA thesis.
Miller, Wick R. 1984. The classification of the Uto-Aztecan languages based on lexical evidence. International
Journal of American Linguistics 50(1). 1–24.
Miller, Wick R. 1996. Guarijío: Gramática, textos y vocabulario. México, D.F.: Universidad Nacional Autónoma
de México.
Neeleman, Ad, & Reiko Vermeulen. 2012. The syntactic expression of information structure. In Ad Neeleman
& Reiko Vermeulen (eds.), The syntax of topic, focus, and contrast, 1–38. Berlin & Boston: Mouton de
Gruyter.
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Ramírez Barceló, Ana María. 2010. Seguimiento de referencia: tópico discursivo en textos narrativos en pima
bajo. Hermosillo: Universidad de Sonora MA thesis.
Ramos Bierge, Stefanie. 2010. Tipos de cláusulas completivas en tepehuano del norte: un continuo de
complejidad. Hermosillo: Universidad de Sonora MA thesis.
Rinaldini, Benito. 1743. Arte de la lengua tepeguana con vocabulario, confessionario y catechismo. México:
Viuda de Hogal [Facsimile edition. 1994. México: Conaculta, Gobierno del Estado de Durango].
Smith Buckingham. 1862. Grammar of the Pima or Névome, a Language of Sonora, from a Manuscript of the
XVIII Century. [Facsimile edition. 1970. New York: AMS Press, Cramoisy Press].
Zúñiga, Fernando. 2014. Benefaction proper and surrogation. Studies in Language 38(3). 543–565.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Marianne Mithun
17 Applicative constructions in the Inuit-
Yupik-Unangan (Eskimo-Aleut) languages
Abstract: Six applicative constructions can be identified in languages of the Yupik-In-
uktitut-Unangan family. The languages show basic ergative/absolutive patterning in
nominal case marking, and only definite referents can be core arguments. The applic-
atives add an argument which is cast as the absolutive of a transitive clause. The base
clause may be intransitive or transitive, but if it was transitive, the original absolutive
is expressed as an oblique or not mentioned. One general applicative, with cognates
across the family, is quite productive and can add a recipient, a beneficiary, a referent
affected by natural phenomena, a companion, an instrument, or a reason. Within the
Yupik-Inuktitut branch, one additional applicative built on this can add a reason in
Yupik, and an instrument, means, cause, or reason in Kalaallisut (Greenlandic). Another
can add a spatial goal. Yup’ik also contains an adversative applicative and a replacive
applicative ‘in place of’. Kalaallisut also contains a comitative applicative. Applicative
clauses can be nominalized to form terms referring to the applied absolutive. Within
discourse, applicatives can function to bring topical referents into the core, but because
they are derivational, their use depends on the inventories of derived lexical items in
each language.

1 Introduction
Languages of the Inuit-Yupik-Unangan family are spoken over a wide area of the Arctic
from Siberia to Greenland. The family consists of two main branches, Inuit-Yupik and
Unangan (Aleut). Relations among the languages essentially as laid out by Fortescue,
Jacobson, and Kaplan are in Figure 1. Languages are listed in boldface and major dia-
lects in italics.
Views on the prehistory of the family, based on archaeological and linguistic evi-
dence, are summarized in Dorais (2010: 95–105). It is estimated that about 4500 years
ago, speakers of the common parent language were living in what is now Alaska, their
language having replaced those of earlier communities there. Around this time ances-
tors of the Unangan (Aleut) began migrating to the Aleutian Islands. Between 3000 and
2000 years ago the remaining group split. The Sirenikski crossed the Bering Strait to
Chukotka. (The place of Sirenikski, as a separate subbranch within Inuit-Yupik, or part
of the Yupik subbranch, remains under discussion.) The Yupik peoples then spread
out: the Sugpiat (Alutiit) made their way to south central Alaska, and the ancestors of
the Central Siberian Yupiget and the Naukanski followed the Sirenikski to Chukotka.
Between 1000 and 800 years ago the Iñupiat moved into Yupik territory on the Seward

https://doi.org/10.1515/9783110730951-017
558 Marianne Mithun

Inuit-Yupik (Eskimoan)
Inuit-Iñupiaq
Greenlandic Inuit
North Greenlandic = Polar Eskimo (Inuktun, Inughuit), West Greenlandic =
Kalaallisut, East Greenlandic (Tunumiisut, Iivin)
Eastern Canadian Inuit
Aivilik, South Baffin, Tarramiut, North Baffin-Iglulik, Itivimmiut, Labrador =
Nunatsiavummiutet
Western Canadian Inuit
Siglit, Copper, Caribou, Netsilik
North Alaskan Inupiaq
Malimiut, North Slope
Seward Peninsula Iñupiaq
Bering Strait, Qawiaraq
Yupik
Sirenikski
Central Siberian Yupik
Chaplinski = Ungazigmiistun, St. Lawrence Island = Sivuqaghmiistun
Naukanski = Nuvuqaghmiistun
Central Alaskan Yup’ik
General Central Alaskan Yup’ik = Yugtun, Hooper Bay/Chevak = Cup’ik,
Egegik, Nunivak = Cyp’ig, Norton Sound = Cugtun
Alutiiq = Pacific Gulf Yupik = Sugpiaq
Koniag = Alutit’stun, Chugach = Sugt’stun
Unangan (Aleut)
Unangan = Aleut
Eastern, Western

Figure 1: The Inuit-Yupik-Unangan family (Fortescue, Jacobson, and Kaplan 2010).

Peninsula in Alaska. The ancestors of the modern Inuit, the Thule, began moving east-
ward across the Arctic, reaching Greenland by around 1600 (McGhee 2015).
It should be noted that the term Eskimo is no longer used among some groups.
Yupiit is used by many for peoples in southwestern Alaska, Inuvialuit by those in the
Mackenzie region of the Northwest Territories, Inuktun in much of western Canada,
Inuit (‘human beings’) in eastern Canada, and Kalaallit in Greenland. Other groups do
not generally object to the term.
One applicative construction has cognates across the family, but additional construc-
tions have developed in the Yupik and Inuktitut languages. The Yupik branch is repre-
sented here by General Central Alaskan Yup’ik (Yup’ik; ISO esu), spoken in southwestern
Alaska, the Inuit-Iñupiaq group by Kalaallisut (ISO kal), spoken on the west coast of Green-
land, and Unangan (ISO ale) by the Eastern and Western dialects of the Aleutian Islands.
The discussion has benefited greatly from work with Yup’ik speakers Elena Charles,
George Charles, and Elizabeth Ali, and Kalallisuut linguist and first-language speaker
Carl Christian Olsen. Modern published works on each of these languages contain impor-
tant discussions of applicatives. For Central Alaskan Yup’ik there is a detailed pedagogi-
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 559

cal grammar (Jacobson 1995), a comprehensive dictionary (Jacobson 2012), and a mon-
umental reference grammar (Miyaoka 2012) among others. A grammar of the Chevak
dialect of Yup’ik is in Woodbury (1981). For Kalaallisut there is an extensive dictionary
(Schultz-Lorentzen 1927) and grammars by Fortescue (1984/1997), Sadock (2003), and
Kahn and Valijärvi (2021). For Unangan there is a substantial dictionary and a reference
grammar (Bergsland 1994, 1997). In what follows, Section 2 describes the basic structures
of the languages, Section 3 their applicative constructions, and Section 4 interactions with
morphology, syntax, and discourse. Section 5 provides a summary.

2 Basic grammatical structure

All of the languages are polysynthetic and nearly exclusively suffixing. There are three
lexical categories: verbs, nouns, and particles. Verbs and nouns consist of a single root
(termed a base in the literature), optionally followed by any number of suffixes (termed
postbases), and ending in an obligatory inflectional complex. The verb structure is sum-
marized in Figure 2.

Root (Suffixes) Inflectional Ending

= Base ( = Postbases)
Mood Pronominal suffix

Figure 2: Verb template.

The relative order of postbases is for the most part hierarchical, reflecting scope,
though frequently-recurring sequences of suffixes have fused. Inuit-Yupik languages
have four independent moods: Indicative, Interrogative, Imperative, and Optative, as
well as various dependent moods. Some moods, including the Indicative, distinguish
transitivity. The pronominal suffixes identify the core arguments, whether or not coref-
erential nominals are also present: one for intransitives and two for transitives. As can
be seen in the Yup’ik examples in (1) and (2), verbs can stand alone as complete clauses
in themselves.1

(1) Yup’ik intransitive (Elizabeth Ali, speaker)

Kipusvigtellinilria.
kipute-vig-te-llini-lria
buy-loc.nmlz-go.to-apparently-intr.part.3sg
‘He apparently went to the store.’

1 Examples are presented here in the current standard orthographies.

560 Marianne Mithun

(2) Yup’ik transitive (Elizabeth Ali, speaker)

Tangersuumiitamken.
tangerr-yuumir-ite-a-mken
see-want-neg-tr.ind-1sg>2sg
‘I do not want to see you.’

The ending on nouns specifies number, case, and optionally possession. Core cases
follow an ergative/absolutive pattern. Ergative forms (traditionally called relative in the
literature) match genitives. Absolutives are generally unmarked. Additional cases in
Yup’ik are Ablative, Allative, Locative, Vialis (for instruments and paths), and Aequalis.
Possessive suffixes identify both the possessor and possessed, which is normally third
person. Basic noun structure is in Figure 3.

Root (Suffixes) Inflectional Ending

= Base ( = Postbases)
Number, Case (Possession)

Figure 3: Noun template.

Uses of the various cases are for the most part as would be expected.

(3) Yup’ik Absolutive (Elizabeth Ali, speaker)

Anngacayaqa tamaantuq?
anngaq-cayaq-ka tamaante-u-q
older.brother-baby-1sg>3sg.abs.sg be.there-ind.intr-3sg
‘Is my older brother (abs) there?’

(4) Yup’ik Ergative (Susan Charles, speaker)

Ilumun yungcaristem tagesqaaten.
ilumun yungcari-ste-em tage-sqe-a-aten
indeed treat.medically-agt.nmlz-erg.sg go.up-request-tr.ind-3sg>3sg
‘Indeed the doctor (erg) has asked you to go up [to the hospital].’

(5) Yup’ik Ablative (Elizabeth Ali, speaker)

Nunamnek watua avaii ayallruunga.
nuna-mnek watua avai ayag-llru-u-nga
land-1sg>3sg.abl right.now yonder leave-past-intr.ind-1sg
‘I just traveled from my home (abl).’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 561

(6) Yup’ik Allative (Elena Charles, speaker)

Tacomamun alguta.
Tacoma-mun age-lu-ta
placename-all.sg go.over-subord-1pl
‘We went over to Tacoma (all).’

(7) Yup’ik Locative (Geroge Charles, speaker)

Utaqallruut-qaa misvigmi?
utaqa-llru-u-t=qaa mit’e-vik-mi
wait-past-intr.ind-3pl=q alight-loc.nmlz-loc.sg
‘Did they wait at the airport (loc)?’

(8) Yup’ik Vialis (George Charles, speaker)

Ayaumalriaci-qaa nunakuarcuutkun?
ayag-uma-lria-ci=qaa nuna-kuar-cuun-kun
go-for.long.time-intr.prtcp-2pl=q land-go.by.way.of-ins.nmlz-via.sg
‘Did you travel for a long time by car?’

(9) Yup’ik Aequalis (Elizabeth Ali, speaker)

Qanemcilria elli-gguq, yugtun.
qanemci-lria ellii=gguq yuk-tun
talk.about.things-intr.prtcp.3sg 3sg=hrs person-aeq.sg
‘He talks, in Yup’ik.’

The case inventories are similar across the Inuit-Yupik languages, though the termi-
nology used to describe them varies slightly. The Kalaallisut cases are Absolutive, Rel-
ative (=Ergative), Ablative, Allative, Locative, Instrumental, Perlative or Prolative, and
Equative.
In the Inuit-Yupik languages, Absolutives of transitives must be definite. Referents
cast as indefinite objects in other languages are expressed as obliques: Ablatives in
Yup’ik and the cognate Instrumentals in Kalaallisut. In the Yup’ik sentence in (10), the
newly introduced beaver was indefinite. The noun ‘beaver’ was Ablative and the clause
was grammatically intransitive, with the intransitive form of the Indicative mood
marker and reference only to the agent in the pronominal suffix.

(10) Yup’ik indefinite patient (George Charles, speaker)

Paluqtamek tangerrlinilria,
paluqtaq-mek tangerr-llini-lria
beaver-abl.sg see-apparently-intr.prtcp.3sg
‘Then he saw a beaver’
562 Marianne Mithun

kuimalriamek kuigmi.
kuimar-lria-mek kuik-mi
swim-nmlz-abl.sg river-loc.sg
‘swimming in the river.’

Inuit-Yupik verb roots vary according to their possible argument structures (without
further derivation). A good discussion is in Jacobson (1995, Chapter 8). Some are used only
as intransitives and some only as transitives, but many, termed ambitransitives, can be
inflected as either. For some of these, termed agentive, the intransitive absolutive corre-
sponds to the transitive Ergative. But for others, termed patientive, the intransitive Abso-
lutive corresponds to the transitive Absolutive. The categories are summarized in (11).

(11) Yup’ik verb transitivity categories (Jacobson 1995, 2012)

Intransitive only igt-u-q ‘it is falling’
Transitive only tegu-a ‘s/he is taking it’
Ambitransitive
Agentive kenir-tu-q ‘s/he is cooking’
kenir-a-a ‘s/he is cooking it’
Patientive alleg-tu-q ‘it tore’
allg-a-a ‘s/he tore it’

For some transitive verbs, the Absolutive is comparable to an English direct object:
kenir-a-a ‘s/he is cooking it’. For others, the Absolutive is a recipient, beneficiary, etc.:
qilug-a-a ‘it is barking at him/her/it’. Derivational processes can of course alter the
argument structures.
The Unangan inflectional system is slightly different, though cognate relationships
can be discerned. There are Ergative, Absolutive, Ablative, and Locative nominal cases,
but the Allative, Instrumental, Vialis, and Aequalis cases have been lost.
Unangan sentence structure is also slightly different. As throughout the family, a
verb may constitute a complete clause on its own or be accompanied by lexical nomi-
nals. In intransitives, the Absolutive is identified by a pronominal enclitic if it is first or
second person, and just a number suffix on the verb if it is third.

(12) Unangan intransitive (Bergsland 1997: 343)

Awa-ku-x̂=txin.
work-prs-sg=2sg
‘You are working.’

(13) Unangan intransitive (Bergsland 1997: 126)

Awa-ku-x̂.
work-prs-sg
‘S/he is working.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 563

If there is a lexical nominal, it is Absolutive.

(14) Unangan intransitive (Bergsland 1997: 126)

Tayaĝu-x̂ awak-ku-x̂.
man-abs.sg work-prs-sg
‘The man is working.’

For transitives, both core arguments are marked in the verb, but a lexical agent is
marked as Ergative only if there is no lexical Absolutive.

(15) Unangan transitive (Bergsland 1997: 138)

Hlam kidukuu.
hla-m kidu-ku-aa
boy-erg.sg help-prs-3sg>3sg
‘The boy is helping him/her.’

If both arguments are identified by lexical nominals, both are inflected as Absolutives,
and only the first is marked in the verb.

(16) Unangan transitivity (Bergsland 1997: 138)

Hla-x̂ asxinu-x̂ kidu-u-x̂.
boy-abs.sg girl-abs.sg help-prs-sg
‘The boy is helping the girl.’

Basic clause structure in all languages of the family is SOV, with alternatives for prag-
matic purposes. Because of potential ambiguities, order is somewhat less fluid in
Unangan.

3 Applicatives
Six applicative constructions can be identified among these languages. One has cognates
across the family. Two more have cognates across the Inuit-Yupik languages. A fourth
and fifth can be seen primarily in Yup’ik, and a sixth in Kalaallisut. All are marked
by derivational suffixes in verbs. All usually add a core argument, which is cast as a
full-fledged Absolutive and behaves like other transitive Absolutives syntactically. Since
clauses in these languages can have no more than two core arguments, the addition of
this argument results in the displacement of an original transitive Absolutive from the
core. The displaced participant may be expressed as an oblique or not mentioned at
all. There are no ‘double-object’ constructions. Verb bases formed with the applicative
suffixes show the same inflectional possibilities as other verb bases. All show some allo-
564 Marianne Mithun

morphy conditioned primarily by phonological context, but the conditions vary some-
what across the languages.

3.1 General applicative ✶-utə-

All of the languages in the family share an applicative construction formed with a der-
ivational suffix reconstructed by Fortescue, Jacobson, and Kaplan as ✶-utə- ‘do with or
for’ (2012: 475–478). It is quite productive in all of the languages and shows a range of
meanings.
The Yup’ik applicative suffix -ute- triggers certain changes in preceding sounds.
The initial u is lost after a full vowel i, a, or u, and the final e does not appear before any
other vowel. This applicative can add a recipient.

(17) Yup’ik -ute- with recipient (Elizabeth Ali, speaker)

Qanrucarturluki
qaner-ute-yartur-lu-ki
speak-appl-go.in.order.to-subord-r>3pl
‘He went to tell’

taukut kipusviliurtet,
tauku-t kipute-vik-liur-ta-t
that-abs.pl buy-place-work.with-agt.nmlz-abs.pl
‘the storekeepers’

assiitelriamek-gguq, tunellinikiit.
assiite-lria-mek=gguq tune-llini-ke-iit
be.bad-nmlz-abl.sg=hrs sell-apparently-tr.prtcp-3pl>3sg
‘that they had sold him a bad one.’

It can add a referent affected by natural phenomena.

(18) Yup’ik -ute- as malefactive (Jacobson 2012: 881)

Aniullugutai.
aniu-llug-ut-a-i
snow.on.ground-have.bad-appl-tr.ind-3sg>3pl
‘Soft melting snow came upon them.’
(They encountered soft snow conditions while traveling.)
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 565

It can add a beneficiary.

(19) Yup’ik -ute- as benefactive (Elena Charles, speaker)

Taukuk-llu,
tauku-k=llu
those-du=also
‘And those two’

Tommy-m aaniin,
Tommy-m aana-in
name-3sg>3sg.gen.sg mother-3sg>3sg.erg
‘and Tommy’s mother’

kalukaulluta.
kalukar-ute-lu-ta
have.feast-appl-subord-3r>1pl
‘made a feast for us.’

It can add a companion. This construction is frequently used where a co-causative

might be used in some other languages.

(20) Yup’ik -ute- as comitative (Elizabeth Ali, speaker)

Elliin ayaullua.
ellii-n ayag-ute-lu-a
3sg-erg.sg go-appl-subord-3r>1sg
‘He went with me.’ = ‘He took me along.’

It can add a reason.

(21) Yup’ik -ute- with reason (Jacobson 2012: 66)

Aaryuutaa.
aaryug-ute-a-a
be.wary-appl-tr.ind-3sg>3sg
‘S/he feels concerned on account of what might happen to him/her.’

This applicative is added to intransitive-only bases, transitive-only bases, agentive ambi-

transitive bases, and patientive ambitransitive bases.

(22) Yup’ik derivational possibilities (Jacobson 2012)

Intransitive
arenqig- ‘be satisfactory, feasible, comfortable
arenqig-tuq ‘it is satisfactory’
arenqi-ut-aa ‘it is satisfactory to him/her’
566 Marianne Mithun

Transitive
piqer- ‘strike, hit, whack, whip’
pigr-aa ‘s/he whacked it’
pigr-ut-aa ‘s/he hit it against something’
Agentive ambitransitive
ekrar- ‘cross over’
ekrar-tuq ‘s/he is crossing over’
ekrar-aa ‘s/he is crossing over it’
ekra-ut-aa ‘s/he going across with it’ = ‘taking it across’
Patientive ambitransitive
arulair- ‘stop moving’
arulair-tuq ‘it stopped’
arulairt-aa ‘s/he stopped it’
arulair-ut-aa ‘s/he stopped for him/her’

When the applicative is added to a transitive base, the added argument usually dis-
places the original Absolutive. (In the Subordinative mood, only the Absolutive is
overtly marked, but case marking on lexical nominals is as in other moods.) As noted,
the Absolutives of some basic transitives, like angussaag- ‘hunt’ correspond to direct
objects in languages like English. The Absolutives of other basic transitives, like igaq-
‘mark, write’, correspond to English indirect objects: recipients, goals, beneficiaries, etc.
The applicative -ute- affects these two types of transitive bases in opposite ways. To
the first it can add what would be a recipient, goal, beneficiary, companion, etc., and
eliminate the original Absolutive from the core: ‘s/he is hunting (it) for him/her’. To the
second, it can add a patient/theme and eliminate a recipient, goal, beneficiary, compan-
ion, etc.: ‘s/he is writing it’ (to him/her).’ The demoted referent can still be expressed as
an oblique, inflected for case according to its semantic role.

(23) Yup’ik argument configurations (Jacobson 2012)

Agentive ambitransitive
angussaag- ‘hunt, try to catch game’
angussaag-tuq ‘s/he is hunting’
angussaag-aa ‘s/he is hunting it’
angussaaq-ut-aa ‘s/he is hunting for him/her, providing for him/her’
Agentive ambitransitive
igaq- ‘mark, write’
igar-tuq ‘s/he is writing’
igar-aa ‘s/he is writing to him/her’
iga-ut-aa ‘s/he is writing it’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 567

Since applicatives add an argument, they form transitive verb bases. With the verb agu-
‘go over’, the general applicative -ute- adds comitative meaning ‘go over with’ = ‘take
over’. The added argument is the person or thing taken over, as in (24).

(24) Yup’ik comitative applicative (George Charles, speaker)

Kuigkun agukutelliniki.
kuik-kun agu-ute-llini-ke-ii
river-vialis go.over-appl-apparently-tr.prtcp-3sg>3sg
‘He took him across the river.’

The root agu- is an agentive ambitransitive. It can be inflected as an intransitive with

just an agent, as in (25b) ‘s/he went over’, or inflected as a transitive, as in (25c) with two
core arguments ‘s/he went over it’ (or more idiomatically ‘it covered it’, as snow cover-
ing a mountain). The applicative adds a new absolutive argument, the person or thing
taken, as in (24) above and (25d) below. But only definite referents can be transitive
Absolutives. If the person or thing taken over is indefinite, the verb is simply inflected
as intransitive as in (25e). The indefinite participant, if mentioned, is oblique.

(25) Yup’ik age- ‘go over’ (Jacobson 2012: 69)

a. age- ‘go over’ Basic agentive ambitransitive root
b. ag’-u-q ‘s/he or it went over’ Basic intransitive
c. ag-a-a ‘s/he went over it, ‘it covered it’ Basic transitive
d. ag-ut-a-a ‘s/he took him/her/it over’ Transitive applicative
e. ag-ut-u-q ‘s/he took something over’ Intransitive applicative

Constructions like that in (25e) are much like antipassives. They indicate that there is a
second participant, but they eliminate that participant from the core. They differ from
prototypical antipassives in that there is no antipassive suffix.
This construction has served as the foundation for an extension of sequences of
applicatives and intransitive inflection to antipassive markers in certain contexts.
One involves transitive-only bases. The root tegu- ‘take, pick up’ can be inflected as a
transitive, as in (26b), but it cannot be inflected directly as an intransitive. An antipas-
sive-like construction can be formed, however, by means of the applicative suffix -ute-
and intransitive inflection, as in (26c). The suffix does not actually add any applicative
meaning.

(26) Yup’ik basic transitive (Jacobson 2012: 627)

a. tegu- ‘take, pick up’ Basic transitive-only root
b. tegu-a ‘s/he took it’ Basic transitive verb
c. teg-ut-uq ‘s/he took something’ Antipassive-like intransitive
568 Marianne Mithun

A second context involves patientive ambitransitive bases, those which occur inflected
as both intransitive and transitives, but whose sole intransitive argument corresponds
to the absolutive of the transitive. An example is the root nalke- ‘find’.

(27) Yup’ik patientive ambitransitive: nalke- ‘find’

a. Nalkaa. Transitive
nalke-a-a
find-tr.ind-3sg>3sg
‘S/he found it.’
b. Nalkuq. Intransitive
nalke-u-q
be.found-intr.ind-3sg
‘It has been found.’

Simply inflecting the transitive base as intransitive will not produce the counterpart of
an antipassive ‘s/he found an X, found something’. Here, too, a sequence consisting of the
applicative -ute- and intransitive inflection is exploited for this function. The suffix adds no
applicative meaning. The verb ‘find’ in (28) is formally intransitive, with just one argument,
the agentive finder, but it is clear that something was found, here the indefinite rhubarb.

(28) Yup’ik antipassive (George Charles, speaker)

Tauna cali yuaryaaqelria
tauna cali yuar-yaaqe-lria
that.one again search-in.vain-intr.prtcp.3sg
‘He looked once again’

nalkuteksaunani tamakunek cuassaanek.

nalke-ute-ksaite-lu-ni tamaku-nek cuassaaq-nek
find-appl-not.yet-subord-3r.sg those-abl.pl rhubarb-abl.pl
‘but didn’t find any wild rhubarb.’

Applicatives formed with the suffix -ute- can also be inflected as intransitives with dual
or plural arguments to indicate reciprocal or joint activity, as in (29) and (30).

(29) Yup’ik intransitive reciprocal applicative -ute- (George Charles, speaker)

Tua-i-llu-gguq tuakenirnek, tauna tulukaruq,
tuai=llu=gguq tuakenir-nek tauna tulukaruq
so.then=and=hrs there-abl.pl that.abs.sg raven.abs.sg
‘And from that time on, that Raven,’

tauna-llu angun ciuqliq,

that=and man.abs.sg first.one.abs.sg
‘and that first man’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 569

qalarulutek pillinilriik.
qalarte-ute-lu-tek pi-llini-lri-ik
talk-appl-subord-2du do-apparently-intr.prtcp-3du
‘would talk to each other.’

(30) Yup’ik intransitive joint applicative -ute- (Jacobson 1995: 160)

Aquitut.
aqui-ute-tu-t
play-appl-intr.ind-3pl
‘They are playing together.’

The cognate applicative in Kalaallisut is -ut(i)-. The initial u shifts to a following another
a and unpredictably following certain consonants. Before an indicative mood suffix,
the final i disappears and the t assimilates to the following stop. It functions much like
its Yup’ik cognate.

(31) Kalaallisut applicative in -ut(i)- (Carl Christian Olsen, speaker)

Piniartup pisani qujaruppa.
piniartu-p pisa-ni qujar-ut-pa-a
hunter-erg catch-3r>3sg.abs thank-appl-tr.ind-3sg>3sg
‘The hunter is thankful for his catch.’

As in Yup’ik, this applicative is used as a comitative with verbs of motion where causa-
tives might be used in other languages.

(32) Kalaallisut applicative -ut(i)- (Carl Christian Olsen, speaker)

Aavartup tuttutani
aavartu-p tuttu-t-taq-ni
caribou.hunter-erg caribou-catch-pass.prtcp-3r>3sg.abs
kuukkut ikaaruppa.
kuu-kkut ikaar-ut-pa-a
river-via cross-appl-tr.ind-3sg>3sg
‘The caribou hunter crossed with his caught caribou through the river.’
= ‘brought his caribou over through the river.’

(33) Kalaallisut applicative -ut(i) (Carl Christian Olsen)

Natural phenomena
apivoq ‘there is snow on it, it is covered with snow’
appuppaa ‘there also falls snow on it’
Instrumental
eqqorpaa ‘s/he hits him or it’
eqquuppaa ‘s/he hits with it’
570 Marianne Mithun

Also as in Yup’ik, this suffix occurs with intransitives, transitives, agentive ambitransi-
tives, and patientive ambitransitive bases.

(34) Kalaallisut applicatives in -uti- (Schultz-Lorentzen 1927)

Intransitive
assorpoq ‘s/he rows, drives against the wind, has the wind against him’
assoqquppaa ‘s/he goes to windward of it’
Intransitive
imarorpoq ‘it becomes open water, the ice breaks up’
imaruuppaa ‘it becomes open water for him/her’
Transitive
akilerpaa ‘s/he pays him or it’
akili-up-paa ‘s/he pays for him or on his behalf, pays his debt’
Agentive ambitransitive
erlippoq ‘s/he is stingy, chary’
erligaa ‘s/he is stingy with it’
erliguuppaa ‘s/he is stingy towards him’
Patientive ambitransitive
immerpoq ‘it is filled’
immerpaa ‘s/he fills it’
immiuppaa ‘s/he fills or empties it into something’

Here, too, bases formed with this applicative may be inflected as intransitives with a
plural pronominal suffix and used to signal reciprocal or joint activity.

(35) Kalaallisut reciprocal (Schultz-Lorentzen 1927: 7)

a. Assortorpaa.
assortor-pa-a
fight-tr.ind-3sg>3sg
‘S/he contradicts him, resists him, quarrels with him.’
b. Assortuupput.
assortor-ut-pu-t
fight-appl-intr.ind-3pl
‘They fight with each other, they contradict each other, they quarrel.’

(36) Kalaallisut joint activity (Schultz-Lorentzen 1927: 209, Fortescue 1997: 165)
a. Kaffisorpoq.
kaffi-sor-pu-q
coffee-consume-intr.ind-3sg
‘S/he drank coffee.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 571

b. Kaffisuupput.
kaffi-sor-ut-pu-t
coffee-consume-appl-intr.ind-3pl
‘They drank coffee together.’

A detailed discussion of the Kalaallisut -ut- is in Woodbury (1977).

Cognates of this applicative suffix in other Inuit-Yupik languages listed by Fortes-
cue, Jacobson, and Kaplan (2010: 475–476) include Eastern Canadian Inuit -uti- ‘do with
or for, in a group, reciprocally’; Western Canadian Inuit -(yy)uti- ‘do with or for, recip-
rocally’; North Alaskan Inuit -utɨ- ‘do with, for, reciprocally’; Seward Peninsula Inuit
-uti- ‘do with, for’; Sirenikski -ətə- ‘do with’; Central Siberian Yupik -utə- ‘do with, for,
reciprocally’; Naukanski -utə- ‘do for’; and Alutiiq -utə- ‘do with, for, reciprocally’.
The Unangan cognate has the basic forms -usa-/-asa-. The alternation between the
two forms is not fully predictable. It functions as a goal, instrumental, comitative, and
reason applicative.

(37) Unangan -usa-/-asa- (Bergsland 1994: 488–489)

a. Goal
aqi- ‘to flee, run away, get away’
aqi-isa- ‘to flee toward’
b. Instrumental
chuhni- ‘to stab’
chuhn-usa- ‘to stab with’
c. Comitative
aygag- ‘to walk’
aygag-usa- ‘to walk with’
kim- ‘to descend’
kim-usa- ‘to go down with, take down’
d. Reason
qida- ‘to cry’
qida-asa- ‘to cry because of’
e. nana- ‘to ache’
nana-sa- ‘to cause pain’ (‘[body part] to be painful because of’)
f. susi- ‘to hurry up’
sus-usa- ‘to worry about’ (‘to be hurried because of’)

Like its cognates in other languages, it is added to both intransitive and transitive bases.

(38) Unangan applicative -usa-/-asa- (Bergsland 1994: 488–489)

Intransitive
tanaanu- ‘to approach land’
tanaan-usa- ‘to take ashore’ (‘approach land with’)
572 Marianne Mithun

Transitive
a. chx̂uuĝ- ‘to wash’
atxaĝ-asa ‘to wash with’
b. taya- ‘to visit the market, shop’, ‘to buy’
taya-asa- ‘to buy with’
c. aluĝ- ‘to write, write on’
aluĝ-asa ‘to write with’

As in the other languages, when it is added to an intransitive base, it produces a tran-

sitive.

(39) Unangan intransitive → transitive (Bergsland 1994: 488)

Tanax̂ aqiisanax̂.
tana-x̂ aqi-usa-na-x̂
land-abs flee-appl-intr.rem-3sg
‘He fled toward the shore.’

When it is added to most transitive bases, it displaces the second argument. If the dis-
placed argument is a lexical nominal, it is inflected as a dative.

(40) Unangan transitive → transitive (Schultz-Lorentzen 1997: 162)

Chahmamaan ulux̂ chx̂adusakuq.
chahma-maan ulu-x̂ chx̂at-usa-ku-q
dish-dat meat-abs fill-appl-prs-3sg
‘I filled the dish with meat.’

3.2 Applicative ✶-utəkə-

A second applicative occurs in just the Inuit-Yupik branch of the family, reconstructed
by Fortescue, Jacobson, and Kaplan (2010: 476) as ‘have as means or reason for doing’.
The Yup’ik form behaves phonologically like the -ute- applicative. It can trigger
changes in preceding sounds, the initial u is lost after a full vowel i, a, or u, and the final
e does not appear before any other vowel.

(41) Yup’ik applicative -uteke- ‘on account of’ (Jacobson 2012: 883)
Angniutekaa qetunrami kassuutellra.
angni-uteke-a-a qetunraq-mi kassuute-llreq-a
be.happy-appl-tr.ind-3sg>3sg son-3r.sg>3gen marry-nmlz-3sg>3sg.abs.sg
‘She is rejoicing over son’s wedding.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 573

(42) Yup’ik applicative -uteke- ‘on account of’ (Jacobson 2012)

a. quyauq ‘s/he is thankful, glad’
quya-tek-aa ‘s/he is thankful for it’
b. ngel’artuq ‘s/he is laughing’
ngela-utek-aa ‘s/he is laughing at him/her’
c. qalriaguq ‘it cried out’
qalria-tek-aa ‘s/he is begging for it’
d. aptuq ‘s/he asked’
aptaa ‘s/he asked him/her’
apy-utek-aa ‘s/he asked about it’

It is attached to both intransitive and transitive bases.

(43) Yup’ik applicative -uteke- ‘on account of’ (Jacobson 2012)

Intransitive
a. qia-tuq ‘s/he is crying’
qia-tek-aa ‘s/he is crying on account of it’
b. nag-tuq ‘it got snagged’
nag-utek-aa ‘s/he is being held back by it’
Transitive
qalart-uq ‘s/he is talking’
qalart-ua ‘s/he is talking to him/her’
qalar-utk-aa ‘s/he is talking about it’

Some of the same bases appear with both -ute- and -uteke- applicatives.

(44) Yup’ik applicatives based on qaner- ‘to speak, converse’ (Jacobson 2012)
qanert-uq ‘s/he is speaking’
qaner-aa ‘s/he said it’
qanr-ut-aa ‘s/he said to him/her, told him/her’
qanr-utk-aa ‘s/he is speaking about it’

The Kalaallisut cognate has the basic form -utig(i)-, with phonological adjustments
similar to those for -ut(i)-. In combination with a following indicative suffix there is con-
traction: -utigivaa > -utigaa. This applicative can add an instrument, means, cause, or
reason to the set of arguments.

(45) Kalaallisut applicative -utig(i)- (Carl Christian Olsen, speaker)

Illuliortup kiffiutissat nutaat
illu-lior-tu-p kiff-utiss-at nuta-at
house-build-agt.nmlz-erg.sg insulate-ins.nmlz-pl.abs new-pl.abs
574 Marianne Mithun

kiffiutigai.
kiff-utigi-va-i
insulate-appl-tr.ind-3sg>3pl
‘The builder insulated it with new insulation.’

(46) Kalaallisut applicative -utig(i)- (Carl Christian Olsen, speaker)

Angutip tusilarnini innarluutigaa.
anguti-p tusiliar-niq-ni innarlu-utigi-va-a
man-erg be.deaf-nmlz-3r>3sg.abs.sg be.handicapped-appl-tr.ind-3sg>3sg
‘The man is handicapped by his deafness.’

(47) Kalaallisut applicative -utig(i)- (Carl Christian Olsen, speaker)

Meeqqap innarluteqarnini ittoorutigaa.
meeqqa-p innarlu-t-qar-ni-ni ittoor-utigi-va-a
child-erg handicap-nmlz-have-nmlz-3r>3sg.abs.sg be.shy-appl-tr.ind.3sg>3sg
‘The child is shy because of his handicap.’

(48) Kalaallisut applicative -utig(i)- (Carl Christian Olsen, speaker)

Inuup sallusorineqarnini
innu-p sallu-sori-neqar-ni-ni
person-erg.sg lie-think.that-pass-nmlz-3r>3sg.abs.sg
mamiasuutigaa.
mamiasu-utigi-va-a
offended-appl-tr.indic-3sg>3sg
‘The person was insulted at his being thought a liar.’

(49) Kalaallisut applicative -utig(i)- (Carl Christian Olsen, speaker)

Arnap uimi toquneraa
arna-p ui-mi toqu-ner-a
woman-erg.sg husband-3r>3gen.sg die-nmlz-3sg>3sg.abs.sg
aliasuutigaa.
aliasu-utigi-va-a
mourn-appl-tr.ind-3sg.>3sg
‘The woman is grieving over her husband’s death.’

A form -ssutiki- has a similar range of meanings, most often adding a means, reason, or
cause.

(50) Kalaallisut applicative -ssutig(i)- (Schultz-Lorentzen 1927)

a. ilungersuavoq ‘s/he is anxious’
ilungersua-ssutig-aa ‘s/he is anxious about it’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 575

b. pakitsivoq ‘s/he is ashamed’

pakatsi-ssutig-aa ‘s/he is ashamed of it’
c. pisuuvoq ‘s/he is rich’
pisu-ussutig-aa ‘s/he has become rich by it’
d. puuppoq ‘s/he is confused, not quite right in the head’
puu-ssutig-aa ‘s/he gets confused by it’
e. nangaavoq ‘s/he is doubtful, does not know what to decide’
nanga-ssutig-aa ‘s/he is in doubt as to it’
f. siooravoq ‘s/he is fearful’
siioragaa ‘s/he is afraid of him or it’
sioora-ssutig-aa ‘s/he is afraid for that reason’

Schultz-Lorentzen lists alternative forms, with and without the ss, among them nakim-
matigaa / nakimassutigaa ‘s/he is in doubt about it’, and naqquutigaa / naqqissutigaa
‘s/he corrects, confirms something by it, gives evidence about him or it, corroborates
something for that reason, was revived by it’ (1927: 361). Fortescue (1997: 91) observes
that the form with ss is more productive, and that there are some lexicalized distinc-
tions, such as allaatigaa ‘he wrote about (or with) it’ and allassutigaa ‘he wrote for that
reason’.
This applicative, too, is added to bases of various transitivity types.

(51) Kalaallisut applicative -utig(i)- (Schultz-Lorentzen 1927)

Intransitive
isumalior-poq ‘s/he reflects’
isumaliu-utig-aa ‘s/he reflects on it’
Agentive ambitransitive
qujavoq ‘s/he says thanks’
qujavaa ‘s/he says thanks for him’
quja-ssutig-aa ‘s/he says thanks for it’
Patientive ambitransitive
kalerrippoq ‘s/he gets a hint of something, is scared’
kalerriipaa ‘s/he gives him a hint of something, scares it (an animal)’
kalerri-ssutig-aa ‘s/he gets a hint of something through it, is scared by it’

There are also multiple applicative constructions built on the same base.

(52) Kalaallisut multiple applicatives (Schultz-Lorentzen 1927)

isussorpoq ‘s/he whispers’
isussorpaa ‘s/he whispers it’
isussu-up-aa ‘s/he whispers something to him/her’
isussu-utig-aa ‘s/he whispers something about him/it’
576 Marianne Mithun

Fortescue, Jacobson, and Kaplan (2010: 476) list cognates in other Inuit-Yupik languages:
Eastern Canadian Inuit -utiɣi-; North Alaskan Inuit -utiɣɨ- ‘have as means or reason for
-ing’; Seward Peninsula Inuit -utiɣi- ‘have as means or reason for -ing’; Central Siberian
Yupik -utkə- ‘on account of’; and Alutiiq -utəkə- ‘concerning, on account of’.
The source of this suffix can still be discerned (Jacobson 1995: 297). A number of
roots in all of the languages have noun and verb counterparts, such as the Yup’ik qalleq
‘rust’, qaller- ‘to rust, be rusty’ and qanuk ‘snowflake’, qanug- ‘to snow’. (Velar and uvular
stops and fricatives alternate according to context. Only stops can occur word-finally.) In
some cases, it is clear that the noun came into the language first, in others it is clear that
the verb came first, and in still others, it is difficult to know. A count of bases in the Jacob-
son 2012 dictionary showed that 12% had doublets (Mithun 2017.) Similarly, a number
of suffixes have doublets, one deriving nouns, another deriving verbs. The applicative
✶
-utə- has a nominalizing counterpart which derives instrumental and reason nominals,
with varying productivity across the languages. Describing Yup’ik, Jacobson (1995: 297)
notes that this suffix is no longer productive as a nominalizer, but it is still evident in
many words. Its word-final form in Yup’ik is -un, as in the noun cav-un ‘oar’, derived
from the verb root cave- ‘to row’; in the noun ipu-un ‘ladle’ from the verb root ipug- ‘to
scoop’, in the noun anguar-un ‘paddle, propeller blade’ from the verb root anguar- ‘to
paddle’, and more. The fuller form appears word-medially, as in ciki-un ‘gift’, ciki-ute-
kaq give-nmlz-fut = ‘future gift’. Kalaallisut has a cognate instrumental nominalizer
-uti- with word-final form -ut, as in the noun angu-ut ‘paddle’ from the verb root anguar-
‘to paddle’, in atungi-ut ‘sewing needle, for soles’ from the verb atunngi- ‘prepare sole
skins’; and in the noun ilisarnaq-ut ‘mark, emblem’, from the verb ilisarnar- ‘be easily
known, recognizable’ (Schultz-Lorentzen 1927: 56, 83, 135). The Unangan cognate has
a corresponding instrumental/locative/manner nominalizer -usa-/-asa-, as in the noun
ayx̂a-asi- ‘boat’ derived from the verb ayx̂a- ‘to go by sea’; the noun taanga-asi- ‘place
for drinking’ from the verb taanga- ‘to drink’, and the noun hag-usi- ‘growth, stature’
from the verb hag- ‘to grow up’ (Bergsland 1997: 109). The Proto-Inuit-Yupik applicative
✶
-utəkə- is an amalgamation of this nominalizer and the Proto-Inuit-Yupik verbalizing
derivational suffix ✶-kə- ‘have as one’s’.

(53) Yup’ik suffix -ke- ‘have as’ (Elena Charles, speaker)

Ataakenritcaaqaat.
ataa-ke-nrite-yaaqe-a-at
father-have.as-neg-actually-tr.ind-3pl>3sg
‘They do not actually have him as their father.’
= ‘He is not actually their natural father.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 577

3.3 Proto-Inuit-Yupik applicative ✶-δvikə-

This third suffix functions primarily as a locative applicative, adding spatial specifica-
tion to the set of core arguments. The Yup’ik cognate -vike- can affect preceding sounds,
and the final e does not appear before vowels.

(54) Yup’ik applicative -vike- (Elena Charles, speaker)

Tamaai-llu, anguyautellratni, anguyiit,
tamaai=llu anguyag-ute-tur-llru-atni anguyak-it
back.then=too fight-appl-repeatedly-pst.contemp-3pl warrior-abs.pl
‘And at that time, when they used to war with each other, the warriors,’

iirvikaqluki aglenrraraat.
iir-vike-aqe-lu-ki aglenrraq-aat
hide-loc.appl-hab-subord-3pl one.having.first.menstruation-3pl>3sg.abs.pl
‘they would hide in the first-menstruating women’s huts.’

Some additional formations are in (55).

(55) Yup’ik applicative -vike- (Jacobson 2012: 900, 88, 131, 205, 256, 305, 433)
a. aqume- ‘to sit down’
aqum-vik-aa ‘s/he sat down on it’
b. kuve- ‘to spill’
kuve-vik-aa ‘s/he spilled something on him/her’
c. alair- ‘to appear, come into view, come on the scene’
alair-vik-aa ‘it came into his/her view’
d. angllur- ‘to dive under water, submerge, be baptized’
angllur-vik-aa ‘s/he dived after him/her’
e. ciktaar- ‘to bow repeatedly, worship by bowing, pay one’s respects’
ciktaar-vik-aa ‘s/he is worshipping it, bowing down before it’
f. ellngar- ‘to leak liquids from a container, to drip’
ellngar-vik-aa ‘it is leaking liquid out of it’
g. iqu- ‘to fall over from an upright position’
iqu-vik-aa ‘it fell on it’
h. nau- ‘to grow’
nau-vik-aa ‘s/he descends from him/her, it is growing on it (plant, cancer)’

The Kalaallisut cognate has the basic shape -figi-. It interacts with the preceding pho-
nological context, and in combination with certain inflectional endings, it is usually
contracted. With the Indicative mood suffix -va- and the transitive pronominal suffix
3sg>3sg -a the result is -figi-va-a > -figaa.
578 Marianne Mithun

(56) Kalaallisut applicative -figi- (Carl Christian Olsen, speaker)

Umiarsuup nunaqarfik akunniffigaa.
umiarsu-up nuna-qar-fik akunni-figi-va-a
ship-erg land-have-loc.nmlz.abs.sg spend.time-appl-tr.ind-3sg>3sg
‘The ship spent a short time in the village.’

(57) Kalaallisut applicative -figi- (Carl Christian Olsen, speaker)

Nunaqqatini perulluliormata
nuna-qati-ni perullulior-mata
village-fellow-3rsg>3pl.abs do.violence-conseq.3pl
‘Because his fellow villagers did something violent,’

ilaquttani qimaaffigai.
ilaqut-ani qimaa-figi-va-a
family-3sg>3sg.abs leave-appl-tr.ind-3sg>3sg
‘he fled to his family.’

(58) Kalaallisut applicative -figi- (Carl Christian Olsen, speaker)

Angutip arnaq ajortuliffigaa.
anguti-p arnaq ajortuli-figi-va-a
man-erg.sg woman.abs.sg do.harm-appl-tr.ind-3sg>3sg
‘The man did something sinful toward the woman.’

Fortescue, Jacobson, and Kaplan reconstruct a Proto-Inuit-Yupik form ✶δvikə- ‘have as

place or time of doing’ (2010: 440), with cognates Eastern Canadian Inuit +viɣi-; Western
Canadian Inuit +viɣi-; North Alaskan Inuit +viɣi-; Central Siberian Yupik +vikə-; and Alutiiq
+wikə- ‘have as place or time of -ing or person toward whom one is -ing’. The source of
this suffix is traceable to a development much like that of the previous one. There is a
pervasive locative nominalizer reconstructed by Fortescue, Jacobson, and Kaplan as Pro-
to-Inuit-Yupik ✶-əviɣ- or ✶-viɣ- (2010:440). This suffix can be seen in (59) and (60).

(59) Yup’ik locative nominalizer (Elizabeth Ali, speaker)

Ellvia muiraan, . . .
elli-vik-aa muir-a-an
put-loc.nmlz-3sg>3sg.abs.sg become.full-conseq-3sg
‘Because his container was full, . . .’

(60) Kalaallisut locative nominalizer (Carl Christian Olsen, speaker)

Akiminni qaloortarfeqartarput.
aki-minni qaloor-tar-fik-qartar-pu-t
place.on.other.side-3r>3.loc scoop-hab-loc.nmlz-have-intr.ind-3pl
‘They have scooping places on the other side of their village.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 579

The locative applicative was formed by fusion of this locative nominalizer with the same
suffix ✶-kə- ‘have as’ seen in the reason applicative ✶-utəkə- in the previous section.
This applicative also appears with some of the same bases as other applicatives.

(61) Yup’ik multiple applicatives

quyauq ‘s/he is thankful, glad’
quya-tek-aa ‘s/he is thankful or glad because of it’
quya-vik-aa ‘s/he is thankful to him/her’

(62) Kalaallisut multiple applicatives (Schultz-Lorentzen 1927: 263)

a. kanngusup-poq ‘s/he is ashamed’
kanngusu-uti-gaa ‘s/he is ashamed of it’
kanngusu-ffi-gaa ‘s/he is ashamed before him/her’
b. isumalluar-poq ‘s/he is confident’
isumallu-utig-aa ‘s/he feels confident because of it’
isumalluar-fig-aa ‘s/he places confidence in him/her’

3.4 Yup’ik adversative -(g)i-

Yup’ik contains what at first looks like a straightforward adversative applicative, a suffix
-(g)i- which derives transitive verbs whose Absolutive argument is a person adversely
affected. The suffix triggers certain changes in a preceding base, and the g appears after
bases ending in two vowels. The construction is described in rich detail by Miyaoka
(2012: 1096–1109).

(63) Yup’ik adversative -(g)i- (Miyaoka 2012: 1100)

Ner-i-a-nga neqe-m neqca-mnek.
eat-appl-tr.ind-3sg>1sg fish-erg.sg bait-1sg>3pl.abl
‘The fish ate my bait (on me).’

(64) Yup’ik adversative -(g)i- (Miyaoka 2012: 1103)

Nulia-ni tamar-i-lu-ku tau-m nukalpiarta-m.
wife-3sg>3sg lose-appl-subord-r>3sg that-erg.sg hunter-erg.sg
‘That great hunter had his (own) wife disappear (on him).’

A closer look at the full range of constructions involving this suffix reveals a more
complex picture. As noted, some verb bases in Yup’ik and related languages can be clas-
sified as intransitive only, transitive only, agentive ambitransitive, or patientive ambi-
transitive. These categories were laid out earlier in (11), repeated here.
580 Marianne Mithun

(11) Verb transitivity categories (Jacobson 1995, 2012)

Intransitive only igt-u-q ‘it is falling’
Transitive only tegu-a ‘s/he is taking it’
Ambitransitive
Agentive kenir-tu-q ‘s/he is cooking’
kenir-a-a ‘s/he is cooking it’
Patientive alleg-tu-q ‘it tore’
allg-a-a ‘s/he tore it’

The adversative construction in (63) is built on the agentive ambitransitive nere- ‘eat’,
and that in (64) is built on the patientive ambitransitive tamar- ‘lose, be lost’.
Adversative constructions are also built on what are termed ‘elemental’ bases
(Jacobson 2012: 27). These generally denote processes of nature and can be inflected as
both intransitives and transitives, with only a subtle difference in meaning. Discussing
the root -ciku- ‘freeze’, Jacobson describes the difference.

With the intransitive the emphasis is on result, while with the transitive the emphasis is on process.
Thus, one would say cikuuq (intransitive) to suggest that freezing had occurred and was now prob-
ably complete, while one would say cikua (transitive) to suggest that freezing had occurred and
perhaps was still occurring. There is no lexical subject. Since the difference is mainly one of empha-
sis and since various speakers might differ in their use of the intransitive or transitive of these
verbs to describe the same situation, we translate the intransitive and transitive of elemental verbs
the same way. (Jacobson 2012: 27)

Miyaoka provides the example in (65) of an adversative construction built on this verb
‘freeze’. The ‘net’ is oblique (Ablative).

(65) Yup’ik adversative -(g)i- (Miyaoka 2012: 1103)

Ciku-i-ga-anga kuvya-mnek.
freeze-appl-tr.ind-3sg>1sg net-1sg>3sg.abl.sg
‘My net froze on me.’ (Lit. ‘It froze my net on me.’)

Such constructions are sometimes termed “transimpersonal” (Malchukov 2008).

Adversative constructions can also be built on intransitive-only bases, but there is
a surprising shift in argument structure. The verbs are transitive and the experiencer,
the person adversely affected, is coded as the Ergative.

(66) Yup’ik adversative -(g)i- (Miyaoka 2012: 1097)

Kic-i-a-qa maklaar-t-a-qa.
sink-appl-tr.ind-1sg>3sg young.seal-catch-nmlz-1sg>3sg.abs.sg
‘The young bearded seal I caught sank on me (to my disadvantage).’
(Lit. ‘I had my caught seal sink.’)
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 581

This structure occurs both with intransitive-only bases whose single argument is a
semantic patient, like ‘sink’ in (66) above and those whose single argument is a semantic
agent, like ‘go out’ in (67) below.

(67) Yup’ik adversative -(g)i- (Miyaoka 2012: 1098)

An-i-a-qa kaviaq igti-nek.
go.out-appl-tr.ind-1sg>3sg fox-abs.sg den-3sg>3sg.abl
‘I let the fox out on me from his hole.’
= ‘The fox got out on (got away from) me from the hole.’

Furthermore, both of these applicative constructions can be inflected as intransitives, in

which the only core argument is the experiencer, and the stimulus is oblique (Ablative).

(68) Yup’ik adversative -(g)i- (Miyaoka 2012: 1098)

Kic-i-u-nga maklaar-i-mnek.
sink-appl-intr.ind-1sg young.seal-catch-nmlz-1sg>3sg.abl
‘The young bearded seal I caught sank on me (to my disadvantage).’

(69) Yup’ik adversative -(g)i- (Miyaoka 2012: 1099)

Kaviar-mek an-i-u-q angun.
fox-abl.sg go.appl-intr.ind-3sg man.abs.sg
‘A fox got out (escaped) on the man.’

These constructions have meanings similar to what are termed adversative passives
in other languages; the single core argument is the experiencer, and the stimulus is
oblique. There is, however, no passive morphology on the verb.
It is likely that there is considerable variation across the Yup’ik dialects. Anthony
Woodbury (p.c.) notes that this suffix is relatively rare in the Chevak dialect of Central
Alaskan Yup’ik. The suffix is discussed further in Section 4.4.

3.5 Yup’ik -ucite- ‘in place of’

An additional applicative suffix has developed in Yup’ik which adds the meaning ‘in
place of’ or ‘instead of’. The added Absolutive argument is the person replaced. It shows
the same phonological alternations as the Yup’ik applicatives -ute- and -uteke- described
earlier. Jacobson provides examples.
582 Marianne Mithun

(70) Yup’ik substitutive applicative (Jacobson 2012: 753)

Calicitaa.
cali-ucite-a-a
work-appl-tr.ind-3sg>3sg
‘S/he is working in his/her place.’

It is added to intransitive and transitive bases, both transitives with patient Absolutives
and those with recipient/beneficiary/etc. Absolutives.

(71) Yup’ik substitutive applicative on intransitive (Jacobson 2012: 380, 754)

a. Maantuq.
‘She, he, or it is here.’
b. Maanlucitaa.
maante-ucite-a-a
be.here-appl-tr.ind-3sg>3sg
‘S/he is here instead of him/her.’

(72) Yup’ik substitutive applicative on agentive ambitransitive (Jacobson 2012: 531.754)

a. Qanertuq.
‘S/he is speaking.’
b. Qaneraa.
‘S/he said it.’
c. Qanerucitaa.
qaner-ucite-a-a
speak-appl-tr.ind-3sg>3sg
‘S/he spoke in his/her place.’

(73) Yup’ik substitutive applicative on agentive ambitransitive (Jacobson 2012: 402, 754)
a. Mertartuq.
‘S/he is fetching water.’
b. Mertaraa.
‘S/he is fetching water for him/her/it.’
c. Mertaucitaa.
mertar-ucite-a-a
fetch.water-appl-tr.ind-3sg>3sg
‘S/he is packing water in place of him/her.’

(74) Yup’ik substitutive patientive ambitransitive (Jacobson 2012: 427, 754)

a. Elitnaurtuq.
‘S/he is attending school.’
b. Elitnauraa.
‘S/he is teaching him/her’ or ‘s/he is endeavoring to learn it.’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 583

c. Elitnauristeci naulluungan
elite-naur-i-sta-ci naulluu-nga-n
teach-hab-antip-agt.nmlz-2pl>3sg be.ill-conseq-2sg>3sg
‘Because your teacher is ill’

elitnauriciciiqaqa.
elite-naur-i-ucite-ciiqe-a-qa
teach-hab-antip-appl-fut-tr.ind-1sg>3sg
‘I shall teach in his place.’

The same verb can serve as the base for a variety of applicatives.

(75) Yup’ik applicatives (Jacobson 2012: 523, 753)

Qalartuq. ‘S/he is talking’
Qalarutaa. ‘S/he is talking to him/her’ -ute- ‘to’
Qalarutkaa. ‘S/he is talking about it’ -uteke- ‘about’
Qalarucitaa. ‘S/he spoke in his/her place.’ -ucite- ‘in place of’

Miyaoka (2012: 1094) provides an account of the development of this applicative suffix as
an amalgamation of the basic applicative suffix -ute-, the antipassive -(g)i-, and a second
occurrence of the applicative -ute-.

3.6 Kalaallisut -qatig(i)- comitative applicative

A Kalaallisut suffix -qatig(i)- functions as a comitative applicative, adding a companion

or associate.

(76) Kalaallisut comitative applicative -qatig(i)- (Carl Christian Olsen, speaker)

Aanaasup ernuttani asiaqatigigaa
aannas-up ermutaq-ni asia-qatigi-ga-a
grandmother-erg grandchild-3r>3sg.abs go.out-appl-tr.prtcp-3sg>3sg
ippassaq takuara.
ippassaq taku-a-ra
yesterday see-tr.ind-1sg>3sg
‘Yesterday I saw the grandmother walking with her grandchild through nature.’

Some additional examples are in (77).

(77) Kalaallisut comitative applicative -qatig(i)- (Schultz-Lorentzen 1927)

a. isummerpaa ‘s/he gets an idea, makes a resolution’
isumme-qatig-aa ‘s/he agrees with him/her’
584 Marianne Mithun

b. naguigaa ‘s/he descends from him’

naggue-qatig-aa ‘s/he has a common origin with him, is of the same race
as him/her’
c. sakkuarpaa ‘s/he carries weapons against him, attacks him/her’
sakkuaa-qatig-aa ‘s/he fights with him/her’

Such constructions, like applicatives seen earlier, can be inflected intransitively with a
plural argument for actions carried out jointly.

(78) Kalaallisut comitative applicative -qatig(i)- (Carl

Christian Olsen, speaker)
Sakkutuut umiarsuarmiut ilaat
sakkutu-ut umiarsuar-miu-t ila-at
soldier-pl ship-dweller-pl member-3pl>3pl
alloraqatigillutik ingerlasarput.
allorar-qatigi-llu-tik ingerlasar-pu-t
march-appl-intr.contemp-3r.pl move-intr.ind-3pl
‘The sailors marched together.’

This suffix also occurs in similative constructions.

(79) Kalaallisut similative (Schultz-Lorentzen 1927: 53)

Nukki angeqatigaa.
nuk-i angi-qatigi-va-a
younger.brother-3r.sg>3sg.abs.sg be.big-appl-tr.ind-3sg>3sg
‘His younger brother is as large as he.’

The source of this suffix is clear. It is an amalgamation of a nominalizer suffix -qat(i-)

‘fellow, companion’ and the verbalizer suffix -gi- ‘have as’. The first element is added to
both noun and verb bases: illu ‘house’, illo-qat ‘house-mate’; innuvoq ‘s/he lives, is born,
is a man’, inoo-qat ‘he who lives with him, his contemporary’.

4 Morphological, syntactic, and discourse

interactions
The applicatives described here are for the most part prototypical: they are derivational
suffixes added to intransitive verb bases or transitive verb bases with either patient/
theme or recipient/beneficiary/etc. Absolutives, to derive new transitive verb bases.
They generally add a recipient, beneficiary, patient, theme, goal, instrument, compan-
ion, location, or reason to the set of core arguments, coded as an Absolutive. If the tran-
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 585

sitive base had a semantic patient/theme Absolutive, the applicative can add a recip-
ient/beneficiary/etc. If the transitive base had a recipient/beneficiary/etc. Absolutive,
it can add a patient/theme. The addition of an applicative to a transitive base usually
results in the displacement of an original transitive Absolutive.
The fact that the primary function of applicatives is to add a core argument raises
questions concerning their relation to other patterns affecting argument structure. In
4.1 the issue of noun incorporation is discussed, in 4.2 the role of applicatives in nomi-
nalization, in 4.3 their interactions with passives, in 4.4 their interactions with antipas-
sives, and in 4.5 their functions within larger discourse contexts.

4.1 Noun incorporation?

None of the Inuit-Yupik-Unangan languages have noun incorporation in the strict sense,
that is, noun-verb compounding: verbs contain one and only one root. They do, however,
have likely descendants of noun incorporation. Each of the languages has a sizeable set
of derivational suffixes with relatively concrete meanings typical of verb roots in many
other languages. Jacobson’s 2012 Yup’ik dictionary includes such suffixes as -ci- ‘buy’,
-cugnite- ‘smell or taste like’, -cur-/-ssur- ‘hunt, check’, -(ng)icag- ‘need, lack’, -(ng)ir-
‘deprive or be deprived of’, remove from’, -(ng)ir(ar)te- ‘injure or be injured one’s’, -(ng)
ite- ‘lack’, -kegte- ‘be good X’, -kite- ‘give, supply with’, -ksagute- ‘acquire’, -li- ‘make’, -liqe-
‘catch a lot of’, -lir- ‘have a lot of’, -lliqe- ‘have poor’, -mirte- ‘act like’, miuyaar- ‘speak
the language or dialect of the residents of’, -nge- ‘acquire’, -ngqerr- ‘have’, -qu- ‘hunt
with’, -rpagnite- ‘smell or taste strongly of’, -rapu- ‘be or have a large X’, -te- ‘catch (game
animal)’, -te- ‘apply (liquid, etc.)’, and -tu- ‘be well endowed with’ and more. These ver-
balizing suffixes never serve as the foundation of words on their own, but must always
be attached to a nominal base.

(80) Yup’ik verbalizing suffix (Elena Charles, speaker)

Iciugg’ Frankiq angyangqellrul’.
iciugg Frankieq angyaq-ngqerr-llru-lria
remember name.abs.sg boat-have-past-subord-3sg
‘Remember Frankie had a boat.’

As in noun incorporation in other languages, the initial noun does not serve a specific
semantic or grammatical role: it is not a syntactic argument. It simply narrows the
scope of the verb in some useful way.
There is little significant interaction between the noun bases in such constructions
and the applicatives. Applicative suffixes are simply applied to verb bases built on
nouns plus verbalizing suffixes as they are to other verb bases, simplex or complex. The
Yup’ik verbs in (81)–(82) are based on the noun umugaq ‘mind’.
586 Marianne Mithun

(81) Yup’ik applicative (Jacobson 2012: 675)

Umyuarniurtuq.
umyaq-niur-tu-q
mind-endure.difficulty.pertaining.to-intr.ind-3sg
‘S/he is worried.’

(82) Yup’ik applicative (Jacobson 2012: 675)

Umyuarniurutkaa.
umyuaq-niur-uteke-a-a
mind-endure.difficulty.pertaining.to-appl-tr.ind-3sg>3sg
‘S/he is worried about it.’

A sentence may contain additional lexical nominals further identifying or describing

the item evoked by the initial noun base, but these are in the default oblique case, in
Yup’ik the Ablative. In (83) the verb is intransitive, and the question word ‘what’ is
Ablative.

(83) Yup’ik cooccurrence (Elizabeth Ali, speaker)

Camek neqengqercit?
ca-mek neqe-ngqerr-tsi-t
what-abl.sg food-have-interr-2sg
‘What do you have to eat?’

In (84) the verb ‘they bowl have’ is intransitive, and ‘big ones’ is oblique.

(84) Yup’ik cooccurrence (Elena Charles, speaker)

Qantangqelalriit angelrianek.
qantar-ngqerr-lar-lria-t ange-lria-nek
bowl-have-hab-intr.prtcpt-3pl be.big-nmlz-abl.pl
‘They have large bowls.’

4.2 Nominalization and relativization

The Inuit-Yupik-Unangan languages are rich in both event and participant nominalizing
constructions. A common Yup’ik nominalizer is -lria. It can form participant nominali-
zations designating the Absolutive of the base, either intransitive or transitive. In (85),
‘the ones coming from other villages’ was built on an intransitive verb base ‘approach
from a distance’, and ‘the ones they invited’, on a transitive verb base ‘invite’.
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 587

(85) Yup’ik nominalization with -lri- (Elena Charles, speaker)

[‘When you could see them approaching by dogsled across the lake’],
nunanek agiirtellriit,
nuna-mek agiirte-lri-it
village-abl.pl approach.from.distance-nmlz-pl
‘the ones that were coming from other villages,’

kelellriit,
keleg-lri-it
invite-nmlz-3pl>3pl
‘the ones they invited,’
[‘then my grandmother would light some Labrador tea leaves.’]

Both of these nominalizations designate the Absolutive argument of the verb base.
When the base is transitive, the ending is actually a transitive possessive suffix, with the
agent as possessor, here ‘their invited ones’. The nominalization built on the transitive
verb ‘hear’ with past nominalizer -lleq in (86) shows the same pattern.

(86) Yup’ik nominalization with -lleq (Elizabeth Ali, speaker)

Wanirpak qanemciqatartua,
wani-rpak qanemci-qatar-tu-a
here-present tell.story-fut-intr.ind-1sg
‘Now I am going to relate a story,’

niitellemnek, . . .
niite-lleq-mnek
hear-nmlz-1sg>3sg.abl.sg
‘about something I heard . . .’

There are no relative clause constructions built with relative pronouns, but comparable
ideas can be expressed by nominalizations which may be appositive to other nominals,
including demonstratives.

(87) Yup’ik appositive nominalization (George Charles, speaker)

Aataka imkut-llu allanret tekillriit
aata-la irnku-t=llu allaner-t tekite-lri-it
father-1sg>3sg.abs.sg aforementioned-pl=also guest-pl arrive-nmlz-3pl
‘My father and the others who arrived’

qulirinaurtut.
quliri-naur-tu-t
tell.legend-hab-intr.ind-3pl
‘would tell stories.’
588 Marianne Mithun

Nominalizations are formed on applicative constructions just as on other transitives, to

designate the applied object (transitive Absolutive).

(88) Yup’ik nominalized applicative (Elena Charles, speaker)

GC: [‘What would you like to talk about?’]

EC: Augkunek qanruteksaiteltellrenka.

angku-nek qaner-ute-ksaite-lrii-nka
that.extended-abl.pl tell-appl-not.yet-nmlz-1sg>3pl
‘Those things I have not talked about before.’

4.3 Passivization

The languages differ in their uses of passives. Passivization is rare in Yup’ik, but Kala-
allisut contains several passive constructions. The most basic involves a suffix -neqaq-.
The sentence in (89), from a text in Berge (1997), contains a passivized applicative. The
applicative brings the father into the core, then the passive eliminates an unidentified
agent, leaving the father as the sole argument of the intransitive clause.

(89) Kalaallisut passivized applicative (Berge 1997: 423)

Ataataga
ataata-ga
father-1sg>3sg.abs.sg
‘My father’

imaannaanngitsorsuartut oqaatigineqarsinnaavoq.
imaannaanngit-soq-suaq-tut uqa-utigi-neqaq-sinnaa-v-oq
not.without.importance-nmlz-big-eq say-appl-pass-can-intr.ind-3sg
‘was said to be an exceptional person.’ (‘like one not without importance’)

As observed by Berge, however, passivization is much more common in Unangan [Aleut].

The passive is used very frequently in Aleut, far more so than in any other Eskimo-Aleut language.
By way of comparison, in Greenlandic texts totaling over 600 clauses, there are fewer than 20
passive constructions, as opposed to over 100 passives in similar numbers of clauses in Aleut texts.
Some of the Aleut passives are the result of a distancing strategy whereby first plural is expressed
as third singular passive; in the Eastern dialect of Aleut, this has become grammaticized. More
generally, the passive is used to focus on the topic and de-emphasize other participants. (Berge
2003: 196)

Unangan contains basic passive suffixes -lga-/-sxa- (and late Atkan also -lĝa-/-sx̂a-, and
Attuan -lu-/-sxa-), which are applied to both transitive and intransitive bases (Bergsland
1997: 117).
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 589

(90) Unangan passive (Bergsland 1997: 117)

Hla-s kidu-lga-qa-s.
boy-pl help-pass-intr.remote-pl
‘The boys were helped.’

There are special forms for the passives of the applicatives -usa-/-asa- and -ula-/-ala-
(Attuan -Vlu-).

(91) Unangan applicative passive (Atkan 1860) (Bergsland 1994: 93, 100; 1997: 117)
a. haqa- ‘to come’
haqa-asa- ‘to come with, bring, hand over, give
haaqqa-ala- ‘to be come with, to be brought’
b. asx̂at- ‘to kill’
asx̂a-asa- ‘to die of’
asx̂ad-usa- ‘to kill with’
asx̂ad-ula- ‘to be killed with’
c. quganas ngiin asx̂adulazaax̂tas
qugana-s ngiin asx̂at-ula-zaa-x̂ta-s
stone-pl 3pl.dat kill-pass.appl-distr-opt-3pl
‘that they should be killed with stones, be stoned’

4.4 Antipassivization

As described in earlier sections, counterparts to antipassive constructions are often

formed by simply inflecting a transitive verb as an intransitive. The clause is formally
intransitive, with an intransitive mood suffix on the verb and just the agent is men-
tioned in the pronominal suffix, and the demoted argument implied but cast as oblique
or unexpressed. This is obligatory for indefinites but not restricted to them. Such a con-
struction can be seen with the Yup’ik verb ‘buy’ in (92).

(92) Yup’ik antipassive counterpart (Elizabeth Ali, speaker)

Tua-i-llu-gguq tauna angukara’urluq,
tuai=llu=gguq tauna angute-karaq-urluq
and.then=also=hrs that.abs.sg man-little-dear.abs.sg
‘And so that dear old man’

kiputellinilria im’umek levaamek.

kipute-llini-lria im’u-mek levaaq-mek
buy-apparently-intr.prtcp.3sg the.aforementioned-abl.sg motor-abl.sg
‘bought this motor.’
590 Marianne Mithun

Applicative constructions are agentive ambitransitives: the applicative suffix creates a

transitive verb with added Absolutive argument. This verb may be inflected as an intran-
sitive, however, whose single argument is the agent. As described in Section 3.1, sequences
consisting of the general applicative -ute- and intransitive inflection have been extended
to function as antipassives, added to transitive-only and patientive ambitransitive bases.
The suffix -(g)i-, which functions as an adversative applicative in Yup’ik, also serves
this antipassive function. Miyaoka notes that some constructions formed from patien-
tive ambitransitives with this suffix can in fact have two interpretations. The verb tegu-
means ‘take’ when inflected as a transitive and ‘lose’ when inflected as an intransitive.
With the suffix -(g)i-, it can be understood as either an adversative applicative or an
antipassive.

(93) Yup’ik patientive ambitransitive (Miyaoka 2012: 1106)

Tegu-i-gu-kut yug-mek.
take-gi-intr.ind-1pl person-abl.sg
‘We (a village) lost, were deprived of a person.’ Adversative applicative
‘We took a person.’ Antipassive

Miyaoka observes that the antipassive construction is highly productive, occurring with
the majority of patientive ambitransitives. The adversative applicative construction is
less so.

By contrast, use of the adversative -(g)i- varies widely among speakers, generations, and (possibly)
dialect areas, as well as depending on the verb stems concerned and according to whether the verb
is transitive or intransitive. It thus comes as no surprise if at least some (or many) of the adver-
sative verbs cited should be unheard of, taken as strange, or (almost or totally) unacceptable, by
many speakers, especially younger speakers. It happens, however, that even these speakers may
use some lexicalized remnants. (Miyaoka 2012: 1106)

Cognates of the suffix -(g)i- are pervasive throughout the family, but for the most part
with just the antipassive function. A few adversative formations can be found in some
of the languages, however: a relic in Central Siberian Yupik, a semi-productive con-
struction in Iñupiaq, and a number of lexicalized transitives in Kalaallisut. Miyaoka
sees a link between the two functions, noting that the implication that an event is acci-
dental or involuntary is not unrelated to an adverse effect on the only core argument.

4.5 Discourse uses

The arguments added by applicatives are in general topical within the discourse,
though there is not the extensive manipulation of roles seen in some other languages.
Their uses depend on the inventory of lexicalized derivations that speakers have to
choose from.
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 591

Uses of the various available argument structures of the Yup’ik verb qaner- ‘speak,
say’ can be compared in the following examples. This verb is an agentive ambitransi-
tive. It can be inflected as both an intransitive and a transitive, the single argument
of the intransitive matches the agent of the transitive, and the Absolutive of the basic
transitive is what is said, what would be a direct object in languages like English. An
applicative form can add a recipient, the hearer.

(94) Yup’ik ambitransitive, agentive, indirective qaner- ‘speak’

qaner- ‘speak, utter, converse’
qanertuq ‘s/he is speaking, spoke’ Intransitive
qaneraa ‘s/he said it’ Basic transitive
qanrutaa ‘s/he told him/her’ Applicative transitive

A basic intransitive use with only an Absolutive can be seen in (95).

(95) Yup’ik basic intransitive qaner- ‘speak’ (Elizabeth Ali, speaker)

Yuut qanyuunateng.
yug-t qaner-yu-ite-na-ten
person-abl.pl speak-customarily-neg-subord-3r.pl
‘The people do not speak.’

As an intransitive, this verb can appear with various additional oblique participants. In
(96) it appears with an Allative nominal naming the addressee ‘you’. The speaker was
introducing a new topic for discussion. This new topic ‘something’ was indefinite at this
point, so could not be cast as a core argument, but the addressee ‘you’ was inherently
definite, so could have been. The speaker chose to cast it as oblique, however, because
this was not about the addressee.

(96) Yup’ik intransitive qaner- ‘speak’ with Allative recipient (Elizabeth Ali, speaker)
Atam qanerqangssakua elpenun.
atam qaner-qar-ngssak-u-a elpe-nun
listen speak-briefly-non-vitally-intr.ind-1sg 2sg-all
‘Listen, I have a little something to talk to you about.’

The intransitive in (97) appears with an oblique naming the subject matter in the Abla-
tive case.

(97) Yup’ik intransitive qaner- ‘speak’ with Ablative content (Elena Charles, speaker)
Alerquutinek qaneryugyaaqua.
alerqur-ute-inek qaner-yug-yaaqe-u-a
instruct-device-3sg>3pl.abl talk-want.to-actually-intr.ind-1sg
‘I actually want to talk about their prescriptions.’
592 Marianne Mithun

Applicatives can be used to bring topical referents into the core. In (98), the recipient,
Raven, was mentioned immediately before the applicative clause ‘she spoke to him’.

(98) Yup’ik applicative qaner-ute- ‘speak to’ (Elizabeth Ali, speaker)

Tulukara’urluq caukii,
tulukaruk-urluq cau-ke-ii
raven-dear.abs.sg face-tr.prtcp-3sg>3sg
‘She faced Raven,’

waten-llu qanruskii, “. . .”
waten=llu qaner-ute-ke-ii
like.this=and speak-appl-tr.prtcp-3sg>3sg
‘and spoke to him thus, “ . . .”

Mrs. Ali and her brother had been discussing a group of students. When she referred
to them as recipients in (99), she brought them into the core with the applicative ‘speak
to’ = ‘tell’.

(99) Yup’ik applicative qaner-ute- ‘speak to’ (Elizabeth Ali, speaker)

[‘They ask me about you, ask how you’re doing.’]

Qanrutelaranka, assirluten-gguq.
qaner-ute-lar-a-nka assir-lu-ten=gguq
speak-appl-hab-tr.ind-1sg>3pl be.good-subord-2sg=hrs
‘I tell them you’re good.’

A question was posed about the woman Nayagaq. In the answer, she was cast as part of
the Absolutive by the applicative.

(100) Yup’ik applicative qaner-ute- (George Charles, Elena Charles, speakers)

GC: Nayagam-qaa tauna nallunritellrua?
Nayagaq-m=qaa tauna nallu-nrite-llru-a-a
name-erg=q that.abs not.know-neg-past-tr.indic-3sg>3sg
‘Did Nayagaq know that?’
EC: Aaniin-wa
aana-an=wa
mother-3sg>3sg.erg=probably
‘I guess her mother’

ilallrin qanrutelallrulliki,
ila-ller-in qaner-ute-lar-llru-li-ki
relative-past-3sg>3pl.gen talk-appl-hab-past-opt-3sg>3pl
‘used to tell them her deceased relatives’’
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 593

atritnek.
ater-itnek
name-3pl>3pl.abl
‘names.’

A similar pattern can be seen with verbs based on kenir- ‘cook, make a fire’. This verb is
similarly an agentive ambitransitive. An applicative form adds a beneficiary.

(101) Yup’ik kenir- ‘cook’

kenir- ‘cook, make a fire’
kenirtuq ‘s/he is cooking, making a fire’ Intransitive
keniraa ‘s/he is cooking it, making a fire under it’ Basic transitive
keniutaa ‘s/he is cooking for him/her’ Benefactive applicative

George Charles was discussing his father’s dogs. As the main topic of conversation, they
were cast as a core argument of the applicative ‘cook for’, rather than as oblique.

(102) Yup’ik kenir-ute- ‘cook for’ (George Charles, speaker)

[‘When I was small, when we lived in Kasigluk and in Bethel, my father used
to travel around with dogs. I used to watch my father taking good care of those
dogs. He used to feed the dogs every day in winter. He would feed them dried
fish. Sometimes he would have more than ten dogs. And during the summer, he
would take care of the dogs and try to fatten them up.’]

Aataqa, nerevavkarilukilu, cal’, keniulluki.

aata-qa nere-vkar-lu-ki=llu cali kenir-ute-lu-ki
father-1sg>3sg eat-caus-subord-r>3pl=too and cook-appl-subord-r>3pl
‘My father would feed them and cook for them.’

Similar patterns appear in the other languages. In Unangan, a passive applicative can
bring a topical participant into the core and eliminate an agent, leaving that topical
participant as the only argument.

(103) Unangan passive applicative (Eastern) (Bergsland 1997: 174)

Ulaan nuguun aaluulalix.
ula-an nu-guun aalu-ula-lix
house-loc come-cond.3r laugh-pass.appl-conj.3sg
‘When he got back to the house, he was laughed at.’

The same pattern can be seen in (104). The passive applicative in the first clause brings
the water into the core and leaves it as the only argument, and the applicative in the
594 Marianne Mithun

second ensures that it remains in the core. (Bergsland 1997: 97 notes that in anterior
clauses, the subject is marked by possessives in the Locative case.)

(104) Unangan passive applicative and applicative (Bergsland 1997: 174)

Taangam anĝaĝii igiim uulaangan,
taanga-m anĝaĝii igiim u-ula-angan
water.loc.poss live 3r.dat go-appl.pass-anterior.3sg
‘When the living water was brought to him,’

igiim ĝulaasaqalikux̂ awa.

igiim ĝula-asa-qali-ku-x̂ awa
3r.dat wash-appl-begin-ind-sg over.there
‘he began to wash himself with it.’

Applicatives do not appear to play a special role in focus constructions. Focused constit-
uents occur as obliques as often as Ergatives or Absolutives, like the foci of questions in
(105) and (106):

(105) Yup’ik oblique focus of question (George Charles, speaker)

Bobankuni-qaa uitaciquten?
Bob-aq-nkut-ni=qaa uita-ciqe-u-ten
name-lk-assoc.pl-loc.pl=q stay-fut-intr.ind-2sg
‘Will you stay with Bob and his family?’

(106) Yup’ik oblique focus of question (George Charles, speaker)

Aqvaluci nunakuarcuutkun-qaa?
aqva-lu-ci nuna-kuar-cuute-kun=qaa
fetch-subord-r>2pl land-go.by.way.of-device-via=q
‘Did they get you with a car?’

5 Conclusion
The languages of the Inuit-Yupik-Unangan family contain applicative constructions
marked by verbal suffixes. Six are described here, one of which can be reconstructed for
the common parent Proto-Inuit-Yupik-Unangan, two for the Proto-Inuit-Yupik branch of
the family, two more for Yup’ik, and one more for Kalaallisut.

Morphology
– A general applicative reconstructed as ✶-utə- for Proto-Inuit-Yupik-Unangan can
add a recipient, goal, beneficiary, companion, reason, or instrument.
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 595

– A reason applicative reconstructed as ✶-utəkə- for Proto-Inuit-Yupik was formed by

the amalgamation of an instrumental nominalizer of the same shape as the general
applicative, ✶-utə- and a verbalizer ✶-kə ‘have as’.
– A locative applicative reconstructed as ✶δviɣə- for Proto-Inuit-Yupik is an amalga-
mation of a locative nominalizer ✶-δvik- and the verbalizer ✶-kə- ‘have as’.
– A Yup’ik adversative -(g)i- adds an adversely affected participant. The same marker
also functions as an antipassive here and in related languages.
– A Yup’ik substitutive applicative -ucite- ‘in place of’ may have developed from an
amalgamation of the general applicative ✶-utə-, the antipassive ✶-(g)i-, and a repeti-
tion of the general applicative.
– A Kalaallisut comitative applicative -qatig(i)- was formed from a suffix -qati-’fellow’
and the verbalizer ✶-kə- ‘have as’.
– The suffixes show some allomorphy, but it is phonologically conditioned.

Syntax
– Applicative suffixes can be added to intransitive bases, transitive bases, and ambi-
transitive bases, those which can be inflected as either intransitive or transitive.
Agentive ambitransitives are those whose intransitive argument is a semantic
agent, and Patientive ambitransitives are those whose intransitive argument is a
semantic patient. Applicatives can be added to both.
– All applicatives add an argument to the clause, which usually assumes the role of
transitive Absolutive. In this capacity the added argument has all the same charac-
teristics as other transitive Absolutives.
– Applicative verb stems are inflected in the same ways as other verb stems.
– When applicatives are added to transitive bases, the original base Absolutive is
inflected as oblique or not mentioned at all.
– There are no double-object constructions.
– The languages have no noun incorporation, though all have a likely descendant
of noun incorporation, consisting of a noun base and following verbalizing suffix.
These derived bases function like other verb bases for the formation of applicatives.
– The languages all contain participant nominalizers which derive nominals desig-
nating the Absolutives of their bases, either intransitive or transitive. Nominaliza-
tions of transitive bases are inflected with possessive endings, with the base Erga-
tive cast as the possessor and the Absolutive as the possessed. What is expressed by
relative clause constructions in many other languages can be expressed simply by
nominalized clauses in these languages, appositive to nouns or other nominals, to
demonstratives, or on their own. Applicative constructions are nominalized in the
same ways as other transitives for these purposes.
– Passivization is rare in Yup’ik, somewhat more pervasive in Kalaallisut, and con-
siderably more frequent in Unangan. In Kalaallisut, applicative constructions can
be passivized. In Unangan, a special passive applicative suffix has developed. The
596 Marianne Mithun

most common way of detransitivizing transitive verbs is simply to inflect them as

intransitive.
– In these languages only definite referents can be transitive Absolutives. Indefinite
patients, themes, etc. are cast as obliques. Like other ambitransitives, applicative
constructions are obligatorily inflected as intransitives if the introduced argument
is indefinite. The result is similar to that of antipassives. The counterpart of the
transitive Ergative is cast as the intransitive Absolutive, and the counterpart of
the transitive Absolutive is oblique. Unlike prototypical antipassive constructions,
there is no antipassive marker in these applicatives. The functions of sequences
of applicative suffixes and intransitive inflection have, however, been extended to
serve as antipassive marking on transitive-only bases and patientive ambitransi-
tive bases, without adding applicative meaning.

Semantics
– Applicative suffixes are derivational: they are used to create new lexical items. The
meanings of these new lexemes are usually relatively transparent, but the precise
meanings added by the applicatives vary.
– Transitive verb bases can have a semantic patient/theme as Absolutive, or a recip-
ient/beneficiary, etc. as Absolutive, among other roles. If the base Absolutive was a
patient or theme, the applicative can introduce a recipient/ beneficiary/etc. as a core
argument. If the base Absolutive was a recipient/beneficiary/etc., the applicative
can bring a semantic patient/theme into the core. The cases of any other obliques in
the base construction remain unchanged.
– Like other ambitransitives, applied verb stems can be inflected as intransitives
with dual or plural pronominal suffixes for reciprocal or joint actions.
– Speaker choices between basic and applicative constructions, where those exist,
are motivated primarily by topicality. More topical participants are generally cast
as core arguments where the vocabulary permits, and as obliques otherwise.

Appendix
1. General Central Alaskan Yup’ik: Jacobson (2012: 47)

Consonants labial apical velar uvular labio-velar labio-uvular

Stops p tc k q u͡k u͡q
Voiced fricatives v l s/y g r u͡g͡ u͡r
Voiceless fricatives vv ll ss gg rr w’ u͡rr
Voiced nasals m n ng
Voiceless nasals m’ n’ n’g
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 597

Vowels front back

High i u
e
Low a

The symbol c represents an affricate. All of the vowels are lowered before uvulars. The
vowel e represents a schwa. Full vowels i, u, and a may occur in sequence. Numerous
phonological adjustments occur at morpheme boundaries.

2. Kalaallisut: Kahn and Valijarvi (2021: 17–22)

An older orthography established by Kleinschmidt in 1851 was revised in 1977 to the

mostly phonemic system below.

Consonants labial apical velar uvular

Stops p tc k q
Voiced fricatives v g r
Voiceless fricatives f ll s gg rr
Voiced nasals m n ng
Voiced lateral, glide lj

Vowels front back

High i u
e
Low a

All consonants except j and v may be geminated. Words end only in a vowel, stop, or
rarely n. As in Yup’ik, the mid vowels are allophones of the high vowels occurring
before uvulars q and r.

3. Unangan: Bergsland (1997: 16)

Sounds in parentheses occur only in Russian and English loanwords.

Consonants labial apical palatal velar uvular glottal
Stops (p) t ch k q
(Voiced stops) (b) (d) (g)
Voiced fricatives (v) d z g ĝ
Voiceless fricatives (f) hd s x x̂
Voiced nasals m n ng
Voiceless nasals hm hn hng
598 Marianne Mithun

Voiced approximants w l (r) y

Aspirated approximants hw hl hy h

Vowels front back

High i, ii u, uu
(e, ee) (o, oo)
Low (ää) a, aa

Abbreviations
abl ablative
abs absolutive
aeq aequalis
agt agentive
all allative
antip antipassive
appl applicative
assoc associative
caus causative mood
cond conditional mood
conj conjunctive mood
conseq consequential mood
contemp contemporative mood
dat dative
distr distributive
du dual
eq equative
erg ergative
fut future
gen genitive
hab habitual
hrs hearsay
ind indicative mood
ins instrumental
interr interrogative mood
intr intransitive
lk linker
loc locative
neg negative
nmlz nominalizer
opt optative mood
pass passive
pl plural
poss possessive
prs present
 17 Applicative constructions in the Inuit-Yupik-Unangan (Eskimo-Aleut) languages 599

prtcp participial mood

pst past
q question marker
r coreferential
rem remote
sg singular
subord subordinative mood
tr transitive
via vialis
x>y x acts on y

References
Berge, Anna. 1997. Topic and discourse structure in West Greenlandic agreement constructions. Berkeley:
University of California dissertation.
Berge, Anna. 2009. Tracking topics: A comparison of topic in Aleut and Greenlandic discourse.
In Marc-Antoine Mahieu & Nicole Tersis (eds.), Variations on polysynthesis: The Eskaleut languages,
185–200. Amsterdam: John Benjamins.
Bergsland, Knut. 1994. Aleut dictionary: Unangam tunudgusii. Fairbanks, AK: Alaska Native Language Center,
University of Alaska.
Bergsland, Knut. 1997. Aleut grammar: Unangam tunuganaan achixaasix̂ . Fairbanks, AK: Alaska Native
Language Center, University of Alaska.
Dorais, Louis-Jacques. 2010. The language of the Inuit: Syntax, semantics, and society in the Arctic. Montreal:
McGill-Queen’s University Press.
Fortescue, Michael. 1984/1997. West Greenlandic. London: Croom Helm.
Fortescue, Michael, Steven Jacobson, & Lawrence Kaplan. 2010. Comparative Eskimo dictionary with Aleut
cognates. 2nd edn. Fairbanks, AK: Alaska Native Language Center.
Jacobson, Steven A. 1995. A practical grammar of the Central Alaskan Yup’ik Eskimo language. Fairbanks, AK:
Alaska Native Language Center.
Jacobson, Steven A. 2012. Yup’ik Eskimo dictionary. 2nd edn. Fairbanks, AK: Alaska Native Language Center.
Kahn, Lily & Riitta-Liisa Valijärvi. 2021. West Greenlandic: An essential grammar. London: Routledge.
Malchukov, Andrej. 2008. Split intransitives, experiencer objects, and ‘transimpersonal; constructions: (re-)
establishing the connection. In Mark Donohue & Søren Wichmann (eds.), The typology of semantic
alignment, 76–100. Oxford: Oxford University Press.
Malchukov, Andrej. 2017. Markedness effects in applicative formation. In Albert Álvarez González &
Ía Navarro (eds.), Verb valency changes: Theoretical and typological perspectives, 3–29.
John Benjamins.
McGhee, Robert 2015. Thule culture. The Canadian Encyclopedia. https://www.thecanadianencyclopedia.ca/
en/article/thule-culture
Mithun, Marianne. 2017. Polycategoriality and zero derivation: Insights from Central Alaskan Yup’ik.
In Valentina Vapnarsky & Edy Venezian (eds.), Lexical polycategoriality: Cross-linguistic, cross-theoretical,
and language acquisition approaches, 156–174. Amsterdam: John Benjamins.
Mithun, Marianne. To appear. Word classes in Eskimo-Aleut languages. In Eva van Lier (ed.), Oxford
handbook of word classes. Oxford, UK: Oxford University Press.
Miyaoka, Osahito. 2012. A grammar of Central Alaskan Yupik (CAY). Berlin: De Gruyter Mouton.
600 Marianne Mithun

Sadock, Jerrold. 2003. A grammar of Kalaallisut (West Greenlandic Inuttut). Munich: LINCOM.
Woodbury, Anthony C. 1977. The Greenlandic verbal suffix -ut-: Interactions of linguistic form and
grammatical function. Proceedings of the Berkeley Linguistic Society 3: 251–269.
Woodbury, Anthony C. 1981. Study of the Chevak dialect of Central Yup’ik Eskimo. Berkeley: University of
California dissertation.
Hunter Thompson Lockwood and Monica Macaulay
18 Applicatives across Algonquian
Abstract: This paper surveys applicatives and related constructions across the Algon-
quian language family. Most of the languages have multiple affixes that license benefac-
tives, goals, recipients, and other participants as objects, and there are some constructions
which license oblique arguments as well. Examples with both an overt O1 and O2 are
relatively rare, but the data collected shows that word order in applicative constructions
is fairly free, with all possible orderings of verb and object(s) attested. We also consider a
number of other forms, including the “relative root” construction, in which a derivational
component of the stem licenses an oblique object, concluding that they are syntactic looka-
likes rather than true applicatives because there is no corresponding BC. The closely-re-
lated relative preverb construction, however, is a true applicative because a correspond-
ing BC can be identified. “Relational verbs” are also addressed; these are valence-neutral
morphological lookalike forms found primarily in the Cree language group.

1 Introduction
This chapter surveys applicatives and related constructions across Algonquian, a large
family of North American languages which extends eastward from the Canadian Rocky
Mountains, across the northern Great Plains and midwestern United States, and along
the east coast of both countries.
The Algonquian languages are rich in applicative and applicative-like constructions.
On the one hand, in most of the languages there are multiple affixes that license bene-
factives, goals, recipients, and other participants as objects. This reflects what Rhodes
describes as a conspiracy “to avoid the creation of oblique nominals” (2010: 428). At
the same time, there are also constructions which do license oblique arguments, and
even what Junker and Toivonen (2015) call “ghost participants” (participants which are
indexed on the verb, but cannot be instantiated as an overt argument).
In the next section we provide relevant background on Algonquian language struc-
ture. In sections 3 and 4 we discuss affixal applicatives, and in Section 5 we look at the

Acknowledgments: We are grateful to Julie Brittain, Dave Costa, Dianne Friesend, Inge Genee, Ryan Henke,
Will Oxford, Joe Salmons, Olivia Sammons, Natalie Weber, an anonymous reviewer, and the editors of this
volume for their comments and help with this chapter. Of course, they are not responsible for any errors. The
authors’ names appear in alphabetical order. This material is in part based upon work supported by the Uni-
versity of Wisconsin-Madison Graduate School and the National Science Foundation DLI-DEL program under
grant number BCS1953103. Any opinions, findings, and conclusions or recommendations expressed in this
material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

https://doi.org/10.1515/9783110730951-018
602 Hunter Thompson Lockwood and Monica Macaulay

syntax of such constructions. Sections 6 and 7 then present two constructions which
include what Algonquianists call “initials” (or roots) and “preverbs”, and Section 8 dis-
cusses the relational verb construction. Section 9 concludes.1

2 Background on Algonquian morphosyntax

Algonquian languages are polysynthetic, with complex inflectional and derivational
morphology. In this section we briefly introduce topics relevant to applicatives.

2.1 Verb types

Verbs in Algonquian languages fall into four basic types based on the (grammatical)
animacy of their absolutive arguments (subject of intransitive and object of transi-
tive), as shown in Table 1.2 Each type has a distinct set of agreement paradigms, called
“orders”. In most of the languages, the most common are independent (generally for
main clauses) and conjunct (generally for subordinate clauses).

Table 1: Basic verb types.

Type Abbreviation Subject Object

Inanimate Intransitive II inanimate

Animate Intransitive AI animate
Transitive Inanimate TI inanimate
Transitive Animate TA animate

Where the distinction is not relevant, we use the term “object” as a cover term for the
grammatical relations O1 and O2, which behave differently with respect to person,
animacy, and indexation (see Section 2.5). O1 and O2 coexist in ditransitive construc-
tions, and can then be characterized as primary and secondary objects, respectively.
The transitive verbs of Table 1 have an object of the O1 type.

1 Our sources are listed in Appendix A. Some Menominee examples are taken from Bloomfield (1928);
these are noted with a text title abbreviation and a line number. The title abbreviations appear in Ap-
pendix B.
2 Animacy is the standard term for the division of nouns into noun classes in the Algonquian languages.
Based on a study of stem contrasts, Goddard finds evidence that animate nouns typically denote “the
special or particular counterpart of the more ordinary, general, or inclusive inanimate, or the animate
is a part and the inanimate the whole” (2002: 216), and notes the influence of other factors like size. Ul-
timately, however, Goddard concludes that animacy is grammatical in Algonquian, and that “[t]he basic
meaning of the animate gender is a function of the contrast with the inanimate gender” (2002: 224).
 18 Applicatives across Algonquian 603

In addition to the verb categories in Table 1, there are a few mismatches between
syntax and morphology; most relevant for our purposes are the so-called “AI+O verbs”.
These are verbs which in some of the languages inflect exactly like AI verbs but allow an
object, and in other languages add special inflection to an AI verb reflecting an object,
without derivational morphology changing the stem category. As described further in
Section 2.5, this object is of the O2 type. AI+O verbs are illustrated in (1) and (2):3

(1) Menominee (Bloomfield 1962: 47)

Nepēw ne-kāta-mena͞ e-m.
water 1-fut-drink.AI-1
‘I want to drink some water.’

(2) Ojibwe [Odawa] (Rhodes and Valentine 2015: 1208)

Ngii-bwen wiiyaas
ni-gii=abwe-n wiiyaas
1-past=roast.AI-n meat
‘I roasted the meat’

In (1), the stem /mena͞ e-/ is animate intransitive and inflects as a normal AI verb, despite
the fact that it has an object. In (2), however, the AI stem /abwe-/ is marked with the
suffix -n, a special inflectional marker which appears in various environments in
Ojibwe, including on AI+O verbs. The details will not concern us here, but as we show
below, AI+O verbs play an important role in applicativization.

2.2 Derivation of verb stems

Most basic verb stems contain at least two derivational parts, the initial and the final.4
(3) provides a few examples of what is known among Algonquianists as primary deri-
vation:5.

3 We use the orthography of our sources in examples, except for Menominee, where we use the modern
orthography. Several sources use capital letters for morphophonemes which show particular alternations
(which are not important for present purposes). Vowel length may be indicated with double vowels, a
raised dot, a macron, or a circumflex. Some authors use a raised dot or colon before a suffix to indicate
that it lengthens a preceding short vowel. Abbreviations used in examples appear at the end of the chapter.
4 The maximal template for a simple word across Algonquian is tripartite: initial-medial-final. Me-
dials form a small class relative to the other types of element, and are optional. Initials are also called
‘roots’ by many Algonquianists, but we prefer to use ‘initial’ because of the potential confusion with
other uses of the term ‘root’ in linguistics.
5 The translations and glosses in our examples may be slightly simplified, for ease of exposition. In
this set of examples, the second and third lines show only the derivational morphemes in each word,
enclosed by slashes.
604 Hunter Thompson Lockwood and Monica Macaulay

(3) SW Ojibwe (OPD)

a. gwayakosin ‘it (inan.) lies straight’
/gwayakw-sin/
/straight-lie.II/
initial-final
b. gwayakoshin ‘s/he/it (an.) lies straight’
/gwayakw-shin/
/straight-lie.AI/
initial-final
c. bagamiwidoon ‘arrive carrying it (inan.)’
/bagam-wid/
/arrive-carry.TI/
initial-final
d. bagamiwizh ‘arrive carrying him/her/it (an.)’
/bagam-wiN/
/arrive-carry.TA/
initial-final

Each pair of examples contains an initial (gwayakw- or bagam-) followed by a final pro-
viding both lexical and grammatical information. The grammatical information in each
case is the verb type, and as these examples show, finals often come in pairs accord-
ing to the animacy of the relevant argument. We will see this again when considering
applicative suffixes.
Verb stems can be modified further by adding another final through secondary
derivation. (4a), below, shows a TI verb meaning ‘sip it (inanimate)’. In (4b) a secondary
final -a' has been added to the stem of that verb, causativizing it, and yielding the TA
verb ‘give (animate) a sip of it’.

(4) SW Ojibwe (OPD)

a. gwaabandan ‘sip it (inan.)’
/gwaab-and/
/scoop.up-act.on.by.mouth.TI/
initial-final
b. gwaabanda' ‘give him/her/it (an.) a sip of it’
/[gwaaband]-a'/
/[sip.it]-cause.TA/
[stem]-final

Many of the applicative suffixes that we describe below fall into the category of
secondary derivation; that is, they attach to a stem (the BC), creating a new stem
(the AC).
 18 Applicatives across Algonquian 605

2.3 Notes on verb inflection

A few inflectional categories and related phenomena are relevant to the discussion of
applicatives.
The marking of inflection on transitive verbs is conventionally treated as governed
by a set of prominence hierarchies (see Zúñiga 2008; Macaulay 2009). Consider the
examples in (5) (where stems are bracketed):

(5) Meskwaki (Dahlstrom 2021, Ch. 4: 14, 17–18)

a. ne-[wa·pam]-a·-w-a
I-[look.at]-TS-3-sg
‘I look at him’
b. newa·pamekwa
ne-[wa·pam]-ekw-w-a
I-[look.at]-TS-3-sg
‘He looks at me’
c. ke-[wa·pam]-i
2-[look.at]-TS
‘You (sg.) look at me’
d. ke-[wa·pam]-en-e
2-[look.at]-TS-ep
‘I look at you (sg.)’

Both (5a) and (5b) carry the first person prefix, despite the fact that the first person is
subject in the former and object in the latter. In (5c) and (5d) the second person prefix
appears, again marking subject in the former and object in the latter. The customary
analysis is that the prefixes are governed by a prominence hierarchy of the form 2 >
1 > 3. The suffixes glossed “TS” (theme sign) in the examples then distinguish subject
and object. Traditionally, -a·, as in (5a), is called the direct theme sign, meaning that
the subject outranks the object on the hierarchy, while -ekw, in (5b), is called inverse,
meaning that the object outranks the subject. Authors differ on whether the two theme
signs in (5c) and (5d) should likewise be treated as direct and inverse; we do not address
that here.
TI verbs also have theme signs, which can be said to agree with the inanimate
object. Many Algonquian languages have three classes of TI verbs, illustrated below for
Meskwaki:

(6) Meskwaki (Dahlstrom 2021, Ch. 4: 11–12)

a. [wa·pat]-am-w-a
[look.at]-TS-3-sg
‘S/he looks at it.’
606 Hunter Thompson Lockwood and Monica Macaulay

b. [pye·t]-o·-w-a
[bring]-TS-3-sg
‘S/he brings it’
c. [mi·či]-∅-w-a
[eat]-TS-3-sg
‘S/he eats it.’

(6a) illustrates a TI Class 1 verb, marked by theme sign -am, which alternates with -a· in
other person/number combinations. (6b) illustrates Class 2, marked by theme sign -o·.
Some languages also have a small and idiosyncratic class of TI3 verbs, which in Mesk-
waki has no theme sign (we have added a zero to make this explicit).
Many analyses treat some or all of the theme signs as object agreement (e.g. Pent-
land 1999; Oxford 2019); this puts them into the realm of more familiar morphologi-
cal functions. Theme signs are relevant in our description of several of the applicative
types below.

2.4 Preverbs

Most Algonquian languages have a class of preverbs, which attach to the beginning of
the verb stem and contribute tense, aspect, adverbial, and other meanings. The follow-
ing examples illustrate preverbs (underlined) in Cheyenne:

(7) Cheyenne (Leman 2014: 250–251)

a. Né-to'se-vá'ne-onésė-[héhpȯheh]-a
2-going.to-only-try-[scare]-inv
‘He is only going to try to scare you.’
b. Ná-ohkė-sáa-'-oné'seóme-pėhéve-[tsėhést-o'ané]-he
1-hab-not-ep-truly-well-[Cheyenne-pronounce]-neg
‘I truly do not pronounce Cheyenne well.’

The categories of preverb and initial overlap to a great extent in terms of content and
form; in some of the languages many are identical in form, but in others preverbs have
an extra derivational morpheme, often -i, as shown in Table 2.
The two are clearly distinguished in terms of relative position, however: the initial
is always the first morpheme in the stem, and preverbs appear to the left of the stem.
They are also distinguished by boundary; initials are part of derivational morphology,
forming stems, but in most Algonquian languages preverbs compound with stems, with
a word boundary intervening.
We show in sections 6 and 7 how preverbs participate in applicative formation in
some contexts.
 18 Applicatives across Algonquian 607

Table 2: Meskwaki initials and preverbs

(Dahlstrom 2021, Ch. 6: 12–13).

Initial Preverb Meaning

asa·m- asa·mi- ‘too much’
menw- menwi- ‘well’
nah- nahi- ‘know how’
ni·šw- ni·šwi- ‘two’
po·n- po·ni- ‘cease’

2.5 Grammatical relations

Grammatical relations in Algonquian languages have not been extensively studied,

with some exceptions such as work on Ojibwe by Rhodes (e.g. 1990) and Dryer (1986),
and work on Meskwaki by Dahlstrom (e.g. 2009). It is broadly accepted, though, that
Algonquian languages are of the PO (primary object) type, in Dryer’s (1986: 815) terms;
that is, that the relations subject, primary object (O1), and secondary object (O2) are
the relevant grammatical relations in these languages.6 Because (as we show below)
objects of monotransitive verbs and primary objects of ditransitive verbs are treated
alike, contrasting with secondary objects, the languages can be said to show secunda-
tive alignment (Malchukov et al. 2010: 3–5).
Subjects are morphologically marked on verbs as the only argument of true intran-
sitive verbs and as one argument of some forms of transitive verbs, as we have seen in
many of the examples above, including (7b). Primary objects are distinguished first and
foremost by animacy, as illustrated in Table 1, and in (8) below:

(8) Menominee (Bloomfield 1975: 261)

a. wanēhnetaw
[wan-e-hNEt]-ā-w
[lose-ep-lay.TI]-TS-3
‘S/he loses it’
b. wanēhsemaew
[wan-e-hsem]-a͞ e-w
[lose-ep-lay.TA]-TS-3
‘S/he loses him, her, it (an.)’

6 We remain neutral here on whether nominals or agreement satisfy argument structure. Both posi-
tions have been taken in approaches to Algonquian syntax; see especially LeSourd (2006).
608 Hunter Thompson Lockwood and Monica Macaulay

In these examples the object’s animacy is determined by the form of the final: -hNEt
derives a TI verb (with inanimate object), and -hsem derives a TA verb (with animate
object). Additionally, the form of the theme sign signals the animacy of the object: -ā in
(8a) is a TI theme sign and -a͞ e in (8b) is a TA theme sign.
Ditransitives in Algonquian languages take two objects, an O1 (usually recipient or
benefactive, but also a range of other thematic roles) and an O2 (almost always a patient
or theme). Dahlstrom (2009: 227) makes the point that a prepositional dative alterna-
tion of the sort found in English is not available in Meskwaki (nor in other Algonquian
languages): “The double object construction is the only possibility”.7 (9) provides an
example of a basic ditransitive in Plains Cree:

(9) Plains Cree (Wolvengrey 2011: 59)

Ni-kī-[miy]-ā-w ana awāsis maskisina
1-ipfv-[give]-TS-3 that child shoe.inan.pl
‘I gave that child shoes.’

Here, the primary object is the recipient, the child. The secondary object is maskisina
‘shoes’, which is inanimate. The verb has a TA theme sign (-ā), which shows that it agrees
with the (animate) primary object.
Various authors have pointed out that the two objects in a ditransitive construction
show a range of morphosyntactic differences. While primary objects are restricted only
by animacy (as shown in [8]), Rhodes (1990), for example, shows that secondary objects
are restricted to third person and cannot passivize. Dahlstrom (2009) provides a longer
list of tests for Meskwaki that show the same result. An illustration of the person restric-
tion is shown in (10):8

(10) Ojibwe (Lochbihler 2012: 118)9

a. ✶gi-gii-miin-aa niin
2-past-give-TS me
Intended: ‘You gave me to him/her.’
b. ✶ni-gii-miin-aa giin
1-past-give-TS you
Intended: ‘I gave you to him/her.’

7 In the Algonquianist tradition, ditransitives and applicatives are usually called “double object con-
structions”, “double goal constructions”, “benefactives”, and “TA+O verbs”.
8 Meadows (2010: 108) provides elicited examples from Blackfoot showing a first and second person O2,
so that language may not share this restriction.
9 Lochbihler (2012: 23) says that most of her data come from Eastern Ojibwe, but that she also includes
data from other dialects. She marks these two examples as just “Ojibwe”.
 18 Applicatives across Algonquian 609

These examples show that ditransitives are ungrammatical with a non-third person O2.
Furthermore, these authors (and others) have also shown that, by these criteria, the
object of an AI+O is likewise an O2.
There is less consensus on the status of oblique arguments in Algonquian. The term
“oblique” often goes undefined in the Algonquianist literature, but Dahlstrom defines
it as follows: “Oblique arguments . . . are ones in which a thematic role is explicitly
encoded, perhaps by choice of preposition, as in English, or by semantic case marking,
as in Finnish” (2014: 58).10 She goes on to say that the most common type of encoding
for obliques found in Algonquian languages is the appearance of what are known as
relative roots; they are the subject of Section 6. Briefly (and incompletely), relative roots
are derivational components (initials) which license oblique arguments, as in (11):

(11) Meskwaki (Dahlstrom 2015: 60)

Meneseki e·hočiwenekoči
menes-eki e·h-očiwen-ekoči
island-loc aor-carry.O.from.(there)-3′>3/aor
‘It (an eagle) carried him from the island.’

In (11), the verb ‘carry (animate) from there’ contains the relative root ot- ‘source of
motion’ and the final -iwen- ‘carry’. This adds an oblique argument to the valence of the
verb, which in this sentence is realized by the locative-marked noun meneseki ‘island’.
However, Dahlstrom points out that a variety of categories may fulfill this requirement
for an oblique argument – a noun (bare or locative), a demonstrative, or an adverbial
particle, for example. More examples can be found in Section 6.

2.6 Word order

The syntax of Algonquian languages is strongly influenced by discourse factors. As

Dahlstrom (2017) points out, many authors have argued that clauses in Algonquian
languages are basically verb-initial, but with preverbal positions for at least a topic
and a focus constituent.11 She also points out that attempts at determining basic word
order based on the relative ordering of S, O, and V inevitably find a lack of significant
patterns in the data. Costa (2017: 349) goes so far as to say “the concept of ‘basic word
order’ as it is applied to configurational languages is not useful in describing Miami-Il-
linois word order”. Sullivan (2012) notes the importance of considering the source of

10 Dahlstrom is writing here within the framework of Lexical Functional Grammar, but we find the
definition broad enough to generalize.
11 There is also a large literature on Algonquian language structure in the minimalist tradition, which
for the most part has a very different approach to word order. We do not address that kind of approach
here.
610 Hunter Thompson Lockwood and Monica Macaulay

data in such studies. He points out that elicitation bias towards an English-like SVO
pattern is a real possibility for languages that do allow that as one acceptable word
order. He resolves this with a picture-description task for Southwestern Ojibwe, the
results of which support the kind of discourse-driven model that Dahlstrom describes.
Dahlstrom (1995: 3 and elsewhere) argues for a relatively flat templatic structure
for Meskwaki word order, with slots for topic, negative, focus, and oblique before the
verb, and for other constituents after it. Other authors (e.g. Johnson et al. 2015; Costa
2017) have made similar claims for other Algonquian languages, although the relative
positions of specific elements differ across the languages.
At the same time, as is typical with polysynthetic languages, phrases corresponding
to arguments are often omitted. This plays a large role in our discussion of the syntax of
ditransitives, because of the difficulty of finding a large enough set of data for any given
language with both objects present as overt nominals.
One final factor that is relevant to word order is the possibility for discontinuous
constituents. Quantifiers and demonstratives, for example, may be separated from their
heads, and placed in initial position, as in (12):

(12) Swampy Cree (Reinholtz 1999: 201, 204)

a. Nîso kî-sipwêhtêwak awâsisak
two 3.pfv-leave.pl child.pl
‘The two children left.’
b. Awa kî-sipwêhtêw awâsis
this 3.pfv-leave child
‘This child left.’

Reinholtz argues that these modifiers appear preverbally because they are in focus
position; a further indication of the discourse factors that drive syntax in Algonquian
languages.

2.7 Our data and examples

We provide full sentence examples where possible, but our sources do not always
provide sentences containing the relevant forms. This is partly because many of our
examples come from dictionaries which only supply headwords, and partly because
many of the other works cited are more focused on morphology than syntax.
The data are also unavoidably skewed towards those languages for which there are
good descriptive grammars and dictionaries, as well as towards those we know best.
This chapter thus serves as a preliminary survey of applicatives in Algonquian, but
further empirical research is needed.
 18 Applicatives across Algonquian 611

3 The ubiquitous applicative suffixes

✶
-amaw and ✶-aw
Most Algonquian languages have an applicative affix that can be traced back to Proto-
Algonquian ✶-amaw (Bloomfield 1946: 115), illustrated in (13) with data from Menom-
inee.12

(13) Menominee (Bloomfield 1928 [MGV 106], 1975)

a. TI: ena͞ esenam ‘s/he reaches for it; moves it by hand’
Stem: /aenaesen-/
b. TA: Ena͞ esenamowaewen anenoh oma͞ etemōhseman eneh ohkānaeh.
he.hand.it.to.her.obv this.obv his.wife.obv that awl
‘He handed the awl to his wife.’

In (13a) we first provide the surface form of the verb, with the (inflectional) TI theme
sign -am, and below that we show its stem, /aenaesen-/. That stem becomes an applica-
tive TA verb in (13b) with the addition of (derivational) -amow.
In addition, many of the languages have an applicative that Bloomfield reconstructs
as PA ✶-aw (1946: 115), illustrated for Unami in (14):

(14) Unami (Goddard 2021: 165)13

TI: |səkàpəpat-| ‘wet it’
> TA: |səkàpəpataw-| ‘wet O2 for O1’

Both ✶-amaw and ✶-aw fit the definition of P-applicative adopted in this volume: there
is a morphological marking asymmetry, there is no change to the S/A argument, and
verbs in which the applicative appears license a primary object AppP (e.g. anenoh
oma͞ etemōhseman ‘his wife’ in [13b]).
We address three topics in the sections that follow: first (§ 3.1) we provide a thor-
ough look at ✶-amaw and ✶-aw, showing some of the many differences in their evolution
in the daughter languages. Second, in § 3.2 we look at the relationship between the two
suffixes, and third, in § 3.3 we look at the status of ✶-aw in Blackfoot, which differs from
its status in other languages.

12 We adopt the convention in this section of using the PA form when we want to refer to a category
across the family; thus ✶-amaw will stand for the various forms (-amaw, -amow, -omo, etc.) found in the
daughter languages.
13 Goddard uses the vertical bar to enclose underlying forms.
612 Hunter Thompson Lockwood and Monica Macaulay

3.1 Description of ✶-amaw and ✶-aw

(15)–(17) provide a larger set of examples containing reflexes of ✶-amaw from a variety
of Algonquian languages.14

(15) Blackfoot -omo (Frantz 1991: 105)

a. TA: nitsííyissksipistayaawa (stem /yIssksipist-/) ‘I tied up [your horses]’
> TA: nitsííyissksipistomoawa (stem /yIssksipist-omo-/) ‘I tied it up for him’

b. TA: Nitsííyissksipistomoawa nitákkaawa óta’siksi.

I.tied.them.up.for.him my.partner his.horses
‘I tied his horses up for my partner.’

(16) SW Ojibwe -amaw (OPD)

a. AI: ininan (stem /inin-/) ‘hold, handle it a certain way’
> TA: ininamaw (stem /ininamaw-/) ‘hand (it) to him or her a certain way’
b. TA: Ningii-ininamawaa akiwenzii asemaan.
I.handed.it.to.him old.man tobacco
‘I handed the old man some tobacco.’

(17) a. Innu -am(a)u (Drapeau 2014: 227–228)15

AI+O: /tshimuti-/ ‘voler’ (‘steal’)
> TA: /tshimutam(a)u-/ ‘voler quelque chose à quelqu'un’
(‘steal something from somebody’)
b. Meskwaki -amaw (Dahlstrom 2021, Ch. 7:16)
TI: /mešen-/ ‘catch it’
> TA: /mešenamaw-/ ‘catch O2 for O’
c. Myaamia -amaw (MPD)
TI: /mahsenaham-/ ‘owe it’
> TA: /mahsenahamaw-/ ‘owe it to him’
d. Plains Cree -amaw (Wolfart 1973: 75)
TI: /ātot-/ ‘tell of it’
> TA: ātotamawēw (stem /ātotamaw-/) ‘he tells of it for him’

14 Some authors do not treat ✶-amaw as a unitary affix; we address this in § 3.2. Also note that applica-
tives are not formed from II verbs, presumably for the pragmatic reason that inanimates are unlikely to
be subjects of verbs whose semantics would allow for applicativization.
15 Innu examples from Drapeau (2014) are given in their original orthography, which does not mark
vowel length. The symbol <ǹ> represents a sound which is [l] or [n] in different dialects of Innu. Transla-
tions have been supplied by the present authors. Drapeau writes the suffix under discussion as -am(a)u
to indicate contraction it undergoes in certain contexts.
 18 Applicatives across Algonquian 613

e. Unami -amaw (Goddard 2021: 165)

TI: |sōkah-| ‘pour (it)’
> TA: |sōkahamaw-| ‘pour O2 on O1’

In general, reflexes of ✶-amaw are described as creating “benefactives” (e.g. Bliss 2010,
Drapeau 2014) or “double-object verbs” (e.g. Wolfart 1973; Goddard 2021). The applied
object usually carries roles such as beneficiary, goal/recipient, or source, and sometimes
comitative or various locative roles.
Consider next (18) and (19), which provide more examples of ✶-aw applicatives:16

(18) Menominee -uw (Bloomfield 1928 [BOY 008], 1975)17

a. TI: osēhtaw (stem /oseht-/) ‘s/he makes it’
> TA: osīhtuwaew (stem /osehtuw-/) ‘s/he makes it for him or her’
b. TA: . . . ’s osīhtuwacen omāhkesenan onīcianaehsan.
when.she.made.them.for.him.obv his.moccasins her.child.obv
‘. . . when she made moccasins for her child.’

(19) a. Plains Cree -aw (Wolfart 1973: 75)

AI+O: /nahasta-/ ‘place it right, put it away’
> TA: nahastawēw (stem /nahastaw-/) ‘he places it right for him’
b. Meskwaki -aw (Dahlstrom 2021, Ch. 6:5)
AI: /pene·hke·-/ ‘hunt turkeys’
> TA: /pene·hkaw-/ ‘hunt turkeys for O’
c. Myaamia -aw (MPD)
TI: /mam-/ ‘take it, buy it’
> TA: /mamaw-/ ‘buy it from him, take it from him’
d. SW Ojibwe -aw (OPD)
AI: /manise-/ ‘s/he harvests, cuts firewood’
> TA: manisaw (stem /manisaw-/) ‘cut (it) as firewood for him/her’

The applied objects added by ✶-aw bear the same sorts of thematic roles as those added
by ✶-amaw, with beneficiary and goal among the most common.
Bloomfield’s (1946: 115) description of the distribution of ✶-amaw and ✶-aw in PA
is instructive for understanding their distribution in the modern languages: there he

16 When ✶-aw attaches to a vowel-final AI stem, the stem vowel drops. There are also often stem-final
vowel replacements before the various applicative suffixes; these will be evident in the examples below.
17 Bloomfield (1946: 115) describes the relationship between ✶-aw and Menominee -uw as “phonetical-
ly queer”, presumably meaning that the change of ✶a > u is unexplained. In this example -uw triggers
vowel harmony, resulting in the /ī/ in the second syllable of the TA form.
614 Hunter Thompson Lockwood and Monica Macaulay

says that ✶-amaw attached to TI1 stems, but ✶-aw attached to TI2 stems.18 This is to some
extent reflected in the modern data: ✶-aw appears with AI stems, and it also appears
with TI stems of classes 2 and 3 in the languages that have them. Table 3 provides a
closer look at the distribution of ✶-amaw and ✶-aw with respect to the stems to which
they attach.

Table 3: Base categories for ✶-amaw and ✶-aw.

✶
-amaw ✶
-aw
AI TI(1) TI2 TI3 TA AI TI(1) TI2 TI3 TA

Blackfoot
Plains Cree
SW Ojibwe
Menominee
Meskwaki
Potawatomi
Myaamia
Nishnaabemwin
N. East Cree
Naskapi
Innu
Unami

Over time, the pattern has become somewhat less clear, but the basics of the system
described by Bloomfield are still visible. First, only Blackfoot allows an applicative to be
derived from a TA verb. Second, languages with TI2 verbs never use ✶-aw for a TI1. At
the same time, Nishnaabemwin and SW Ojibwe have some TI2 verbs that take ✶-amaw,
and a few languages use ✶-amaw with AI or TI3 verbs.19
In some of the descriptions of applicatives formed from AIs, it is not clear whether
the stems are plain AI or AI+O.20 But in Plains Cree, Wolfart (1973: 75) is explicit that
both ✶-amaw and ✶-aw attach to “syntactically transitive AI stems” (AI+Os), although he
implies that it is less common with ✶-amaw. As he puts it, “This is obviously an area of
extreme productivity and considerable fluctuation”.
Since reflexes of the two suffixes are found across such a wide range of Algonquian
languages, they both appear to be of PA origin. But, as we have shown, their relative
attachment sites are no longer completely complementary for all of the languages, and
so in most cases they can no longer be considered allomorphs.

18 Bloomfield says that TI2 stems developed out of “pseudo-transitive”, or AI+O stems. Other authors
(e.g. Goddard 1979: 71–72) reconstruct ✶-am and ✶-aw for TI classes 1 and 2.
19 Unami has two subtypes of TI1; Goddard (2021: 68) notates them as TI(1a) and TI(1b).
20 We use “AI” as a cover term in Innu for a set of subcategories of “transitive” AI verbs which includes
AI+O; see Drapeau (2014: 227–228).
 18 Applicatives across Algonquian 615

At this point we turn to the relationship between ✶-amaw and ✶-aw.

3.2 The /am/ in ✶-amaw

In many of the languages, ✶-amaw attaches to all TI verbs, or primarily to TI verbs of

class 1 in languages that have subclasses of TI verbs. The shorter suffix ✶-aw attaches to
AI verbs, and in the languages that have them, to TI2 (and sometimes TI3) verbs.
At the same time, Bloomfield (1946: 99) reconstructs a theme sign ✶-am for TI1
verbs with third person subjects. It is natural to wonder whether this ✶-am accounts
for the /am/ in the applicative ✶-amaw, especially given the existence of the applicative
✶
-aw. That is, is ✶-amaw synchronically one suffix, ✶-amaw, or two, ✶-am-aw? Both anal-
yses are found in the literature: some authors (including Bloomfield 1962 and Valentine
2001) treat it as a unitary affix, while others (e.g. Wolfart 1973; Goddard 2021; Dahl-
strom 2021) treat it as consisting of the TI theme sign ✶-am plus a secondary final ✶-aw.
This difference of analysis is illustrated with Plains Cree data in (20):

(20) Plains Cree (PCD)

TI1: âtotam (/âtot-/) ‘s/he tells about something’
> TA: âtotamawêw ‘s/he tells about (it/him) for someone’
a. Unitary applicative suffix -amaw
[âtot]-amaw-êw
[tell.about.it]-appl-inflection
b. Theme sign -am + applicative suffix -aw
[âtot]-am-aw-êw
[tell.about.it]-TI1.TS-appl-inflection

At first glance, it might seem like the most parsimonious analysis of ✶-amaw is the one
shown in (20b), since it allows us to generalize ✶-aw across all TI subtypes. Dahlstrom’s
analysis of the Meskwaki applicative takes exactly this approach, as illustrated in (21):

(21) Meskwaki -aw (Dahlstrom 2021, Ch. 7:16)

a. AI: /ašike·-/ ‘build a house’
> TA: /ašikaw-/ ‘build a house for O’
b. TI1: /mešen-/ ‘catch O’
> TA: /mešenamaw-/ ‘catch O2 for O’
c. TI2: /pi·tikat-/ ‘bring O inside’
> TA: /pi·tikataw-/ ‘bring O2 inside for O’
d. TI3: /na·t-/ ‘go after O’
> TA: /na·taw-/ ‘go after O2 for O’
616 Hunter Thompson Lockwood and Monica Macaulay

In order to claim that there is a single applicative suffix, Dahlstrom must describe its
distribution as “attach[ing] to AI, TI2, and TI3 stems; with TI class 1 verbs the applicative
attaches to a theme, composed of the verb stem plus the TI class 1 theme sign -am-” (2021,
Ch. 7: 16).
There is an analytic trade-off here: on the one hand, if we claim that there is just
one applicative suffix, ✶-aw, this comes at the cost of complicating the statement of the
types of base it attaches to (as in Dahlstrom’s account just given). On the other hand, we
could simplify the statement about the type of base to which the applicative attaches
(the stem in all cases), but that comes at the cost of positing two suffixes, ✶-amaw for TI1
verbs and ✶-aw for the rest.
Our inclination is to take the latter approach. A major consideration for us is that
most Algonquianists consider theme signs to be inflectional, and the analysis repre-
sented by (20b) and (21b) places an inflectional morpheme (-am) inside a derivational
morpheme (-aw).21 Some linguists, however, would not see this as a problem (either not
embracing the distinction between inflection and derivation, or not accepting the strict
relative linear ordering of the two types); cf. Bochner (1984) and LeSourd (1995).
Hoffman and Oxford (2021) suggest a third approach: that the -am inside deriva-
tional suffixes (like -aw) is a distinct morpheme, but that it is not synchronically the same
morpheme as the theme sign -am, whatever its diachronic origin may be. We refer the
reader to their paper for further discussion, and here leave the issue an open question.

3.3 ✶-aw as a syntactic lookalike in Blackfoot

Although in most of the languages, ✶-aw appears to be a full-fledged applicative, in

Blackfoot it functions as an equipollent syntactic lookalike. Consider (22)–(24):

(22) Blackfoot (Frantz 1991: 104)

a. TI: nítsskiitatoo’piaawa ‘I baked them’
Root /ihkiit-/ + TI final /-watoo/
b. TA: Nítsskiitoawaistsi nitána.
nit-ihkiit-o-a:-wa-aistsi n-itán-wa
1-bake-ben-dir-3sg-pro 1-daughter-3sg
‘I baked them for my daughter’

21 Hoffman and Oxford (2021: 136) point out that there has also been some discussion in the literature
of whether theme signs should be treated as inflectional or derivational; see the references cited in their
paper.
 18 Applicatives across Algonquian 617

(23) Blackfoot (BD, Taylor 1969: 249)

AI: innootaa ‘butcher’
> TA: /-Innooto-/ ‘butcher (for someone)’

(24) Blackfoot (BD)

AI: ohpommaa ‘buy (something)’
> TA: ohpommo ‘buy for, buy from, trade to’

The first pair of examples above show the initial /ihkiit-/ ‘bake’; it is followed by a TI
final in (22a), and by the benefactive TA final -o (✶-aw) in (22b). In (23)–(24) the AI verbs
both contain the AI final -aa; this is replaced by -o in the TA forms. Thus the BC and the
AC are both marked, with a common initial in each case.

4 Other affixal applicatives

In addition to ✶-aw and ✶-amaw, there are numerous other, less common applicative suf-
fixes listed for many of the languages. These also satisfy the criteria for a true applica-
tive. Space does not permit full description of all of these affixes; instead, we provide an
overview of the more frequently noted ones.
We have organized these by the initial consonant(s) of the suffix; this is purely for
convenience, and no claims about cognacy are intended.

4.1 /-st/ applicatives

This set of applicative suffixes, illustrated in (25)–(28), occurs in at least four languages
across the Cree–Montagnais–Naskapi (CMN) continuum.

(25) Innu -sht (TI) / -shtu (TA) “circumstantial” (Drapeau 2014: 229–230)22
a. AI: Maǹi atimikapu
Marie she.stands.with.back.turned
‘Marie est debout de dos.’ (‘Marie is standing with her back turned.’)
b. > TI: nitatimikapushten katshitapatakanit
I.stand.with.back.to.it television
‘Je suis debout dos à la télévision.’
(‘I am standing with my back to the television.’)

22 In Drapeau (2011: 61) these are given as -štaw / -štam, and defined as ‘Locational Reference’.
618 Hunter Thompson Lockwood and Monica Macaulay

c. AI: Maǹi nipepu

Marie she.sits.up.watching
‘Marie passe la nuit assise à veiller.’
(‘Marie spends the night watching.’)
d. > TA: Maǹi nipepishtueu utauia
Marie she.sits.up.watching.over.him her.father
‘Marie passe la nuit à veiller son père.’
(‘Marie spends the night watching her father’.)

(26) Western Naskapi “on behalf of” (Brittain 2001: 4)

AI > TA: -stimuw

(27) Northern East Cree “benefactive” (Collette 2014: 90–91)

AI > TI: -stim
AI > TA: -staw
AI > TA: -stimaw

(28) Plains Cree “action on a general goal with a transitive animate beneficiary”
(Wolfart 1973: 75)
AI > TA: -stamaw

This particular set of applicatives may be composed of more than one suffix, either syn-
chronically or diachronically. Plains Cree, for example, has a suffix -st which derives a
TI verb from an AI verb, and it seems plausible that -stamaw is -st + -amaw. But Wolfart
argues that this is not the correct analysis since the specific inanimate object that would
be expected from the addition of -st is not present in the sense of the derived TA verb (note
his definition “action on a general goal”). Brittain (1993: 76) gives a suffix for Innu which
is parallel to the Plains Cree one, but different from the ones Drapeau gives, -shtamau. She
explicitly analyzes it as “the unidentified -sht-, the TI theme sign -am- and the applicative
-au-”. Brittain (2001: 4) likewise treats Western Naskapi -stimuw as composed of three suf-
fixes, concluding that -st is a secondary TI final which accounts for the distinctive seman-
tics, ‘on behalf of’. We leave the internal structure of these suffixes an open question.

4.2 /-ht/ applicatives

A few languages have applicatives in /-ht/, as follow:

(29) Unami |-ht| (TI[1a]), |-htaw| (TA), |-htam| (TA) “applicative” (Goddard 2021: 164)
a. AI: |apī-| ‘be (somewhere)’
> TI: |apīht-| ‘be in’
> TA: |apīhtaw-| ‘be in’
 18 Applicatives across Algonquian 619

b. AI: tkáwsu ‘he is gentle, good-natured, mild, well-minded’

> TA: |təkawəsīhtam-| ‘regard favorably’

(30) Meskwaki “applicative” (Dahlstrom 2021, Ch. 7: 17–18)

AI > TI: -ht
AI > TA: -·htaw
AI > TA: -h

These suffixes all attach to an AI stem, and form both TA and TI applicatives. In the
Meskwaki case, Dahlstrom pairs the suffixes -ht and -·htaw, which generally add a goal
argument, and -ht and -h, “associated with a range of thematic roles” (2021, Ch. 7: 18).

4.3 /-m/ applicatives

Many of the languages have an applicative of the form -m (or in Blackfoot, -:m), which
appears in the comitative construction discussed in § 7. But the suffix appears on its
own without comitative meaning in some cases as well:23

(31) Blackfoot -:m (TA) (BD)

AI: okska’si ‘run’
> TA: yiistapokska’siim ‘flee, run away from’

(32) Menominee -m (TA) “secondary suffix” (Bloomfield 1962:334)

AI: kemōtaew ‘s/he steals’
> TA: kemōtemaew ‘s/he steals from him/her’

(33) Unami |-m| (TA) “applicative” (Goddard 2021:164)

AI: |kīwīkē-| ‘go visiting’
> TA: |kīwīkam-| ‘visit’

The few examples found are formed from AI verbs and have animate goal and source
objects.

23 Thanks to Natalie Weber for bringing the Blackfoot example to our attention. In (31), yiistap- is a
preverb.
620 Hunter Thompson Lockwood and Monica Macaulay

4.4 /-t/ applicatives

A number of languages have applicatives that either consist solely of /-t/ or start with /t/,
as illustrated in (34)–(38).

(34) Menominee -t (TI) “action [. . .] upon an object” (Bloomfield 1928 [HMM:030–031],

1962: 342)
a. AI: mēcema͞ ehkaew ‘s/he prepares food for storage’
b. > TI: eneq-peh enoh meta͞ emoh māwaw
then that woman all
kew-mēcema͞ ehkatah eneh mēcemēhsaeh
hab-she.prepare.it.as.stored.food that meat
‘then the woman would preserve all that meat and store it away’

(35) Unami “applicative” (Goddard 2021: 163)

AI > TI(1a): |-t|

(36) Western Naskapi “benefactive” (Brittain 2001: 4)

AI > TA: -tuw

(37) Northern East Cree “applicative” (Collette 2014: 92)

AI > TI: -tut

(38) Innu “applicative” (Drapeau 2014: 230)24

AI > TI: -tut
AI > TA: -tutu

The Menominee and Unami suffixes each have a matching suffix which creates TA
verbs, but does not start with /-t/. In Menominee it is /-N/ and in Unami it is |-l|. These
both result from a merging of PA ✶-θ and ✶-l (Bloomfield 1946: 87; Goddard 1980) and
subsequent changes, so they are likely cognate.
For Western Naskapi, again, Brittain (2001: 4) treats the form given as composed of
multiple suffixes: -t-uw. She claims that the applicative -uw is barred from attaching to
an AI verb, so the TI final -t attaches to create a non-occurring TI stem, allowing -uw to
follow.

24 These are given as -ttam and -ttaw in Drapeau (2011) and called the “generalized applicative”. By
this, she means that they add an AppP with a “semantically unspecified” thematic role (2011: 58), in
contrast to other more specified applicative suffixes in the language.
 18 Applicatives across Algonquian 621

4.5. Additional applicative suffixes

In this section we list, by language, suffixes that do not fall into the classes above.

(39) Arapaho (Cowell and Moss 2008: 148–150)

AI > TI: -:t “applicative”
AI > TA: -:ton “applicative”

(40) Blackfoot (Taylor 1969: 253)25

AI > TI: -ISkiSt “transitive inanimate”
AI > TA: -ISkiSto “transitive animate”

(41) Menominee (Bloomfield 1962: 342, 355)

AI > TA: /-N/ “action [. . .] upon an object” (matches TI -t)
AI > TI1: -qt “secondary suffix”
AI > TA: -qtaw “secondary suffix”

(42) Meskwaki (Dahlstrom 2021, Ch. 7: 18)

AI > TI: -not “applicative”
AI > TA: -notaw “applicative”

(43) Myaamia (Costa p.c.)26

AI > TI: -ilotam “applicative”
AI > TA: -ilotaw “applicative”

(44) Unami (Goddard 2021: 163–166)

AI > TA: |-l| “applicative” (matches TI |-t|)
AI > TA: |-lax| “benefactive”
AI+O > TA: |-mal| “applicative” (only one example)

5 The syntax of affixal applicatives

Recall from Section 2.5 that there are a number of factors which distinguish an O1 and
an O2 in Algonquian languages. Monotransitive verbs (the TA and TI categories) select
for the animacy of their (primary) object, for example, while ditransitives restrict their

25 The author says, “It is likely that more than one morpheme is present, but if so, their identities are
unknown” (1969: 253).
26 The Meskwaki suffixes in (42) and the Myaamia in (43) are likely cognate.
622 Hunter Thompson Lockwood and Monica Macaulay

O2 to third person. Also recall that the object of an AI+O verb has been shown to be an
O2 by numerous authors (e.g. Rhodes 1990).
With that background in mind, consider the two primary monotransitive sources for
affixal ditransitives, TI and AI+O stems. TI verbs by definition have a third person inani-
mate base object, which serves as the O2 in the ditransitive. The object of an AI+O likewise
serves as the O2 if the verb is made ditransitive. An important point about the O2 of a
ditransitive, though, is that it is not limited in animacy, even when derived from a TI verb:

(45) Menominee (Bloomfield 1962: 48)

a. Wāwanon kepītuan.
egg.inan.pl I.bring.them.to.you
‘I bring you eggs.’
b. Anōhkanak kepītuan.
raspberry.an.pl I.bring.them.to.you
‘I bring you raspberries.’

The examples in (45) contain an applicative built off the TI2 verb pītāw ‘s/he brings it
(inan.)’ with the applicative suffix -uw (and subsequent contraction). (45a) shows an
inanimate O2, ‘eggs’, while (45b) shows an animate O2, ‘raspberries’.
The examples in (46), from Meskwaki, are parallel, but with an AI+O verb:

(46) Meskwaki (Dahlstrom 2009: 231)

a. ahpe·nemo-wa na·tawino·n-i
depend.on-3.ind medicine-inan.sg
‘He relies on the medicine.’
b. ahpe·nemo-wa o-si·me·h-ani
depend.on-3.ind his-younger.sibling-an.obv.sg
‘He relies on his younger brother.’

Thus, a monotransitive TI verb has an inanimate O1, necessarily third person, which
becomes the O2 of a derived applicative, losing the animacy restriction in the process.
An AI+O verb has an O2 object, restricted to third person but unrestricted in terms of
animacy, and it remains the O2 if the AI+O undergoes applicativization.
Turning to word order, we pointed out in Section 2.6 that discourse factors like topic
and focus have been shown to drive the syntax of simple clauses in Algonquian lan-
guages, and that many linguists have argued that there is no “basic” word order that can
be described in terms of subject, object, and verb. When it comes to ditransitives, claims
differ. Dahlstrom (2009: 226) and Bruening (2001: 59) both find that the most common
order is V O1 O2. As Dahlstrom points out, this is consistent with the unmarked position
for non-topic and non-focused arguments.
Large, tagged corpora are not available for most Algonquian languages, so a
thorough investigation of the syntax of applicatives requires further resources and
 18 Applicatives across Algonquian 623

research. Furthermore, as we noted earlier, most of the works that describe applicatives
in Algonquian languages focus on the morphology, and often do not give an example in
a sentence. But a survey of the data containing applicatives which was available to us
likewise shows a skew towards postverbal positioning of secondary objects, although
the full picture indicates that word order in these constructions is no more constrained
than word order in corresponding monotransitive constructions.
To this point we have seen the orders V O1 O2 (in [13b], [15b], and [16b]), V O2 O1
(in [18b]), V O1 (in [22b], [25b], [25d], etc.), and O2 V (in [45a–b]). In total, from the arti-
cles and grammars we surveyed, we were able to collect 99 sentences with one or more
overt objects in a construction with an applicative suffix.27 This is purely a convenience
sample, but even this small number of examples showed every possible ordering of
object and verb. Table 4 illustrates.

Table 4: Applicative construction word orders.

O1 < V O2 < V V < O1 V < O2

O1 O2 V 1 O2 O1 V 1 V O1 6 V O2 29
O1 V O2 8 O2 V 3 V O1 O2 20 V O2 O1 27
O2 V O1 4
9 8 26 56

Table 4 shows that it is most common for object(s) to appear postverbally, consistent with
unmarked word order. Clearly, among such examples, V O2 (O1) word order is the most
common, but we hesitate to draw any firm conclusions from the data. This is in part due
to the small size of our sample, but also due to the possibility of elicitation bias, since so
many of the examples come from articles containing what we believe to be sentences elic-
ited in isolation. What we can conclude is that word order in Algonquian affixal applica-
tive constructions is fairly free, with all possible orderings of verb and object(s) attested.
The combination of the employment of preverbal topic and focus positions and
the fact that overt nominals are often omitted account at least in part for the fact that
the applicative and lookalike constructions we have described above do not require a
particular position for an AppP.

6 Relative root constructions

Algonquian languages have a class of initials known as relative roots, which, as men-
tioned in Section 2.5, license an oblique argument in the clause (Dahlstrom 2014: 57–60;

27 36 of the 99 sentences are in Menominee, due to our easy access to that data.
624 Hunter Thompson Lockwood and Monica Macaulay

Costa 2017: 363–367). Bloomfield defined relative roots as a special class of initials “that
refer to an antecedent in the phrase” (1946: 120, emphasis in original). The relative roots
that he reconstructed are exemplified in Table 5, below, with the forms in a sample of
Algonquian languages.28

Table 5: Relative roots.

Proto-Algonquian PCree Men Nish Mesk Unami Gloss (after Dahlstrom 2014)
✶
eθ- is-/it- aeN- iN- in- əl- goal, manner
✶
went- oht- oht- ond- ot- wənt- source, cause, reason
✶
axk(w)- isko- ahkw- akw- ahkw- sahk- so long
✶
tahθ(w)- tahto- tahNw- daS(w)- tasw- (-ən)tax- so many, so much
✶
taθ- tat- taN- daN- tan- (-ən)tal- stationary location
✶
axpi·ht- ahpēht- apiit- ahpi·t- degree

Relative roots, like all initials, are the first morpheme of a primarily derived stem,
as described in Section 2.2. Like many initials, they also have corresponding preverb
forms, shown in Table 6 for Odawa (preverbs are conventionally joined with a hyphen
to the verb stem, and are in boldface here).29

Table 6: Odawa relative roots and corresponding preverbs (Rhodes 2006: 11).

Relative Root Corresponding Preverb

iN- ‘to, like’ zhi-niikmod ‘growl like so’
ond- ‘from, because of’ nji-googiid ‘dive from there’
akw- ‘length’ ko-gkenmaad ‘know s.o. for so long’
das(w)- ‘number’ dso-bboon’gizid ‘be so many years old’
daN- ‘at’ dzhi-siboodood ‘sharpen s.t. on it’
apiit- ‘extent’ piichi-ngadenmaad ‘know s.o. so much’

We argue that some relative root constructions are X-applicatives and others are looka-
likes. We begin by discussing relative preverbs, which we analyze as true applicatives.
For reasons of space, we omit discussion of the syntax of relative root constructions, but
see Rhodes (2010) for a thorough overview of the topic in Ojibwe.

28 Abbreviations in the table: PCree = Plains Cree, Men = Menominee, Nish = Nishnaabemwin, Mesk =
Meskwaki. See Appendix A for sources. Orthographic representations and glosses have been updated to
follow modern work.
29 Nishnaabemwin and Odawa are roughly the same varieties of Ojibwe; see Valentine (2001: 1–3) for
discussion.
 18 Applicatives across Algonquian 625

6.1 Relative preverbs as X-applicatives

In their preverb forms, relative roots serve an applicativizing function. In each set of
examples below, from Blackfoot, the verbs in (a) contrast with the verbs in (b), which
contain the relative preverb it- ‘then, there’ (underlined in the examples). In both
cases, the preverb is an additional overt marker that distinguishes it from the base
construction.

(47) Blackfoot (Bliss 2014: 2)

a. Anna Leo áíkskima.
ann-wa Leo a-ikskimaa
dem-prox Leo ipfv-hunt.AI
‘Leo hunts.’
b. Anna Leo itáíkskima omi itáó’tsstoyi.
ann-wa Leo it-a-ikskimaa om-yi itao’tsstoyi
dem-prox Leo then-ipfv-hunt.AI dem-inan November
‘Leo hunts in November.’

(48) a. Nitsooyi.
nit-ioyi
1-eat.AI
‘I ate.’
b. Nitsítsooyi anni itáísooyo’pi.
nit-it-ioyi ann-yi itaisooyo’p-yi
1-there-eat.AI dem-inan table-inan
‘I ate at the table.’

In the forms with relative preverbs, the S/A participant is unchanged, the verb is
marked with respect to the BC, and the AC requires an additional phrase. We treat this
additional phrase as an oblique, following Dahlstrom’s definition given in Section 2.5
above.30
The semantics of that additional phrase depends on the relative preverb used. The
example above contained Blackfoot it-, adding a time or location. The Ojicree examples

30 There is some disagreement in the literature over the status of this type of argument. We follow Dahl-
strom (2014), whose definition of ‘oblique’ was given in Section 2.5 above. Recall that for her, relative
roots (and by extension, relative preverbs) are in fact the most common encoding of oblique arguments
in Algonquian languages. However, see Rhodes (2006), who argues that relative root complements are
not only distinct from oblique arguments, but are a unique type of argument distinct from all others. See
also Kim (2020) for a formal treatment of oblique nominal constructions in Blackfoot.
626 Hunter Thompson Lockwood and Monica Macaulay

below illustrate the relative preverbs onci-, which adds a source or reason, and ahpiihci-,
which licenses a degree expression.31

(49) Ojicree (Slavin 2012: 73)

a. Waahsa onci-piishaawak.
far from-they.come.AI
‘They came from far away.’
b. Niishitana ta-ahpiihci-tahkaayaa.
twenty fut-such-be.cold.II
‘It will be twenty degrees below zero.’

Relative root constructions are still grammatical if the applicative phrase is not present
in the sentence, as long as the oblique argument is understood in context:32

(50) Nishnaabemwin (ND; speaker Angeline Williams)

Mii dash gaa-izhi-waabmigod
and.so ic.past-there-it.see.him/her.TA
‘So now it saw him there.’

When the applicative phrase is not overt, the interpretation is “one of definite pronom-
inal/deictic reference” (Rhodes 2006: 13).

6.2 Relative roots as lookalikes

As initials, relative root constructions are syntactic lookalikes rather than true applic-
atives. Verbs with relative roots are not built on fully-derived verbs, so no well-formed
construction can be identified as the BC. There are often verbs that are parallel in struc-
ture and meaning, differing only in the initial chosen, but since the initial component is
required for a well-formed stem, there is no alternation.
Consider the following Menominee example, which contains a relatively pragmati-
cally neutral TA verb meaning ‘chase, pursue’:

(51) Menominee (Bloomfield 1928 [LNX 079])

Wōh, nepa͞ emenaesehopah ayāpa͞ ew!
whew but.I.was.chasing.it.an stag
‘Whew, but I was chasing a stag!’

31 Ojicree is a dialect of Ojibwe (also known as Severn Ojibwe).

32 Initial change (ic in the gloss of [50]) is a grammatically conditioned morphological process of ablaut
(and in some Algonquian languages, other processes like infixation) that occurs at the left edge of some
verb forms; see Costa (1996).
 18 Applicatives across Algonquian 627

This verb consists of the initial /paem-/ ‘straight-line movement’ and the TA final
/-naesehw/ ‘chase’. Many verbs with the ‘chase’ final can be formed by substituting dif-
ferent initials:

(52) Menominee (Bloomfield 1975)

a. pītenaesehaew ‘s/he chases him/her/it (an.) here’; initial /pīt-/ ‘come’
b. onēneseha͞ ew ‘s/he gets, chases him or her out of bed’; initial /on-/ ‘rise,
lift, up’
c. matāpīneseha͞ ew ‘s/he chases him/her/it (an.) down to the water's edge’;
initial /matāpī-/ ‘down toward water’

Combining the relative root /aeN-/ ‘to there, in that way’ with this final creates ena͞ ese-
haew ‘s/he chases, drives him, her, it (animate) there, towards there’, exemplified in (53).33

(53) Menominee (Bloomfield 1928 [BOY 123])

Wahkītāhkyah eneq kew-ena͞ esehotituaq.
hilltop it.is.there hab-they.chase.each.other.there
‘It was there that they chased each other to the summit’

Here, the relative root /aeN-/ of the AC licenses an oblique expressing goal; in this case
‘hilltop’. Yet, as we have shown, there is no corresponding BC that lacks the relative root.
There are a small number of relative root verbs which look at first glance like true
applicatives. Consider (54):

(54) Menominee (Bloomfield 1975)

a. ācemow
/āt-e-mi/
/tell.of-ep-by.vocal.sound.AI/
‘s/he narrates, reports an event, tells a story’
b. enācemow
/aeN-ācemi/
/in.that.way-narrate.AI/
‘s/he tells a story, narrates in that way’

It might look from the surface forms like (54a) is the BC for (54b), but ācemow is made up
of an initial āt- plus a final -mi, while enācemow is made up of the relative root aeN- plus

33 The verb in (53) has undergone secondary derivation, adding a reciprocal final and becoming mor-
phologically intransitive; this has no bearing on the relative root construction being discussed.
628 Hunter Thompson Lockwood and Monica Macaulay

the final -ācemi. While -ācemi is diachronically related to the verb ācemow, we do not
see this as a synchronic relationship, and so (54a) cannot be the BC for (54b).34

7 Comitative constructions
Many Algonquian languages have a comitative construction, which adds another partic-
ipant to the action described by the verb. Most of these constructions are P-applicatives,
although at least one (described below) is not an applicative at all. As with the rela-
tive root construction, the comitative may contain either an initial or its corresponding
preverb, but in addition there may also be a final, -w or -m. Table 7 illustrates:

Table 7: Elements of the comitative.

Language Form(s) Language Form(s)

Final only Initial + final

Arapaho -:w Menominee wēt- + -m
Initial only Meskwaki wi·t- + -·m
Plains Cree wîcê(w)- Nishnaabemwin wiid- + -m
Swampy Cree wît- Unami wīt- + -m
Menominee wēt- Proto-Algonquian ✶
wiit- + ✶-m
Meskwaki wît-
Preverb + final
Preverb only Blackfoot ohpok- + -:m
Plains Cree wîci- Arapaho niit- + -:w
Menominee wēc- Swampy Cree wîci + -m
Unami wiči- Meskwaki wi·či – + –·m
Nishnaabemwin wiiji- + -m
Innu wîci- + -m
Unami wiči- + -m
Proto-Algonquian ✶
wiiči- + ✶-m

As the table shows, most of the languages require a bipartite construction to create a
comitative, although some can form it with just the preverb or the final.
Taking first the case with a final only, Arapaho adds -:w to an intransitive verb of
joint action such as that shown in (55a) to produce a transitive example like the one in
(55b):

34 There are differences of opinion among Algonquianists on what counts as synchrony and what as
diachrony; we follow Macaulay and Salmons (2017) here.
 18 Applicatives across Algonquian 629

(55) Arapaho (Cowell and Moss 2008: 148)35

a. ceehyóootí3i’
ceehyoooti-3i’
ic.quarrel.AI-3pl
‘They are quarreling’
b. ceehyóootiibé3en
cehyoootiiw-e3en
ic.quarrel.TA-1sg>2sg
‘I am quarreling with you’

It is not clear, however, if an overt AppP could appear with this construction.
Just like relative roots, when the comitative construction involves only an initial, no
BC can be identified, because any potential alternates show the same level of morpho-
logical complexity. These verbs tend to be translated as ‘do X together’, and we do not
consider them applicatives. But in some of the languages, such as Menominee, use of the
initial plus the final -m does create an applicative:

(56) Menominee (Bloomfield 1975)

a. wēhpaew
/wēt-pa͞ e/-w
/along.with-sleep.AI/-3
‘s/he sleeps with someone, with people’ (AI)
b. wēhpemaew
/[wēhpae]-m/-a͞ ew
sleep.with-appl-3>3obv
‘s/he sleeps with him, her, it (animate)’ (TA)

Here, the BC (56a) is intransitive with an indefinite object, and contains the relative
root wēt-. But the AC (56b) adds a suffix and the person slept with becomes the primary
object of an applicative transitive verb.
Use of the preverb alone might count as a true applicative in some of the languages,
but unfortunately the examples we have lack an AppP, so we cannot be sure of their
status. However, when the comitative involves a preverb and the relevant final, the
resultant construction is a true applicative, as seen in (57):

35 The symbol <3> in Arapaho stands for /θ/.

630 Hunter Thompson Lockwood and Monica Macaulay

(57) Arapaho (Cowell and Moss 2008: 370)

a. héétnii3kóohúno’
eti-niit-i-koohu-no’
ic.fut-with-ep-run.AI-1pl.incl
‘We will run together’ (AI)
b. héétnii3kóohúúwoot híbio
eti-niit-i-koohu-:w-oot i-bi[h]-o
ic.fut-with-ep-run-TA-3sg>3obv 3sg-older.sister-obv
‘S/he will run with his older sister.’

(57a) contains the preverb /niit-/ ‘with’ combined with the final ‘run’, forming a verb
of joint action. However, use of /niit-/ and the suffix -:w, as in (57b), makes the added
participant an O1, creating the comitative applicative construction.
Much more work is clearly required on the comitative construction across the
family.

8 Relational verb constructions

The relational is a valence-neutral morphological lookalike construction found primar-
ily in the CMN languages.36 That is, it is morphologically marked with respect to a base
construction, but it does not change the valence of the verb. Junker describes relational
forms as verbs “where the person affected by the event is in the background”, and
adds that “The relational construction does not increase the syntactic valency of verbs.
Rather it registers the presence, in the universe of discourse, of additional third-person
participants” (2003a: 319). Consider (58), where the stems are bracketed and the rela-
tional marker -w is underlined:37

(58) Plains Cree (Wolfart 1973: 60)

a. ēkwa ētokwē ē-[napatēstāwikē]-w-iht
then I.guess ic-[build.lean-to.AI]-rel-indef.act
‘Then, I guess, a lean-to was built for her.’

36 Goddard (1995: 141–146) describes a similar construction in Meskwaki. Additionally, Possessor Rais-
ing is found in e.g. Meskwaki (Dahlstrom 2021, Ch. 7: 20–22) and Mi’kmaq (Hamilton 2017; Denny et
al. 2021); the latter has been called a relational by Hewson (1991: 25–26), but we leave the relationship
between these constructions and the relational for future research.
37 Segmentation and analysis has been added to these examples and the ones that follow.
 18 Applicatives across Algonquian 631

b. tāpwe [matōtisānihkē]-w-ān
indeed [build.sweat-lodge.AI]-rel-indef.act
‘Accordingly, a sweat-lodge was built for him.’

Both of these examples are in the indefinite actor form, generally translated as an agen-
tless passive. But note that both also have a benefactive, the presence of which is only
indicated by the relational -w (and indeed, cannot be indicated by an AppP). These are
the “ghost participants” referred to in the introduction.
In the next section we explore the most commonly-described relational construc-
tion, based on AI and TI stems. In § 8.2 we show how, in Innu, a TA verb with an inan-
imate subject may also participate in the relational. In § 8.3, we discuss a similar con-
struction containing TA verbs with animate subjects.

8.1 Canonical relationals: Form, function, and characteristics

(59) and (60) exemplify non-relational and relational forms of independent order AI
and TI verbs, respectively, for purposes of comparison.

(59) Western Swampy Cree (Cenerini 2014: 35)

a. ni-[nipâ]-n
1-[sleep.AI]-lcl
‘I sleep’
b. ni-[nipâ]-w-â-n
1-[sleep.AI]-rel-TA.dir-lcl
‘I sleep in relation to him/her/them.’

(60) Western Swampy Cree (Cenerini 2014: 35)

a. ni-[wâpaht]-ê-n
1-[see.TI]-TI.TS-lcl
‘I see it.’
b. ni-[wâpaht]-am-w-â-n
1-[see.TI]-TI.TS-rel-TA.dir-lcl
‘I see it in relation to him/her/them.’

These examples illustrate the morphological marking asymmetry of the construction

with the absence vs. presence of -w. The rest of the morphology appears to be drawn
from multiple paradigms, and consequently it is difficult to fit relationals into any single
category in opposition to other verb forms. Figure 1 provides a template for AI and TI
relational verbs; below that, we walk through each position.
632 Hunter Thompson Lockwood and Monica Macaulay

1 2 3 4
AI(+O) stem relational -w TA TS -â/-ê person
TI stem TI TS -am relational -w TA TS -â/-ê person

Figure 1: Relational verb templates: AI and TI Verbs.

Position 1: Canonical relational verbs are built on an AI, AI+O, or TI base. In the
latter, the relational marker follows the TI theme sign -am. This theme sign has a dif-
ferent distribution with relational verbs in the independent order, though, than it does
with non-relational verbs in that order: it only appears in forms with third person sub-
jects in non-relational verbs, but appears in forms with any subject in relational verbs.
(Compare [60a] and [60b].)
Position 2: This suffix is generally considered the main marker of the relational par-
ticipant, always an animate third person. -w marks third person across many other par-
adigms in Algonquian languages as well (although it is not always restricted to animate
arguments).
Position 3: This suffix appears to be the TA independent order direct theme sign. In
TA verbs in most of the CNM group, this theme sign is -â in forms with first and second
person subjects acting on third person objects, and -ê in third on third obviative forms,
and it has the same distribution in relational verbs. Many authors treat this theme sign
in non-relational verbs as agreement with a third person object; here, it could be said
to agree with a third-person participant present in the discourse but not the clause. (We
address the transitivity of the relational construction directly below.)
Position 4: A marker for person of subject follows the TA theme sign. In the inde-
pendent relational it is the same as AI person marking, while in the conjunct relational
it parallels TA person marking.
Despite the presence of the TA theme sign in the construction, there is disagree-
ment in the literature over the transitivity of relational verbs. Drapeau (2014: 248–249)
concludes that they are transitive because of the (notional) presence of a third person
object. However, this analysis requires -w, an inflectional suffix, to change lexical cate-
gory, which inflectional morphemes are not generally believed to do. If we were to treat
relational -w as derivational, though, it would follow an inflectional morpheme (the TI
theme sign -am), which (as we have discussed in § 3.2) is at least an atypical relative
ordering.
Junker (2003a:318) argues that the addition of the relational suffix does not change
valence: an AI relational verb remains intransitive and a TI relational verb remains
monotransitive with an inanimate object. She supports this by showing that it is
ungrammatical to add an overt object corresponding to the participant marked by the
relational -w.
Cenerini takes a more nuanced view of the transitivity of the construction, saying
that “by acknowledging a second animate participant in the discourse, the relational
verb takes on some transitive animate (mono- or ditransitive) properties, without being
 18 Applicatives across Algonquian 633

fully transformed into one” (2014: 44). She suggests instead that transitivity forms a
continuum, as illustrated in Table 8:

Table 8: Cline of transitivity (Cenerini 2014: 42).

Intransitivity ↔ Transitivity ↔ Ditransitivity

AI non-relational AI relational TI non-relational TI relational Applicative -amaw

That is, in her view, the AI relational is not completely transitive, and the TI relational
is not completely ditransitive.
Despite the ungrammaticality of the presence of an overt AppP, relational verbs do
cooccur with the kinds of phrases expected of their base forms: a locative in the case of
an AI (61) and an inanimate object in the case of a TI (62). Significantly, though, these
nominals are possessed in each case, and the possessor is the participant singled out by
the relational.38

(61) East Cree (Junker 2003a: 32)

Nipâ-w-e-u u-nipewin-iyi-hch
sleep.AI-rel-TA.dir-3 3-bed-obv-loc
‘S/he sleeps in his/her (someone else’s) bed.’

(62) East Cree (Junker 2003a: 32)

Wâpahtam-w-e-u u-mûhkumân-iyû
see.TI-rel-dir-3 3-knife-obv
‘S/he sees his/her (someone else’s) knife.’

As these examples show, one feature of the construction is that it enforces disjoint ref-
erence between the subject and the relational participant.39 In (61), the possessor of the
bed must be distinct from the subject, and in (62), the knife must belong to someone
else. When there is coreference between a possessor in this position and the subject of
the verb, the relational is not grammatical.
Relational verbs are also distinct from TA verbs in other ways. First, the inflection
following the TA theme sign in the independent order is AI/TI inflection, but it is TA
inflection in the conjunct order. Second, an overt inanimate object may be present in a
TI relational construction (whether it is analyzed as the primary object or not), which is
not possible in a non-relational TA construction. Third, although the added participant

38 Junker discusses interactions between the obviative and the relational at length. We omit this due
to space considerations.
39 Junker (2003a: 319) says that for East Cree, the relational form must be used to express disjoint ref-
erence. Cenerini (2018: 96), however, shows that it is not obligatory in Swampy Cree when expressing
disjoint reference.
634 Hunter Thompson Lockwood and Monica Macaulay

is animate (as one would expect with a TA verb), it is restricted to third person animate.
This is reflected in the TA theme signs (-â/-ê) that appear in the construction. Unlike true
TA verbs, though, relationals based on AI or TI verbs do not allow the inverse theme
sign, nor inverse constructions (although see § 8.2 below).
Three contexts for use of the relational are reported in the literature. The first,
which adds reference to a possessor, has been demonstrated above in (61)–(62). In such
cases, the relational picks out an animate third person possessor of an inanimate object
as more salient than the object itself.40
A second context for relationals is cross-clausal, “in the matrix clause when the
actor of the subordinate verb is a third person, and in the subordinate clause when
the actor of the main clause is a third person” (Cenerini 2014: 53).41 (63)–(64) illustrate:

(63) Innu (Drapeau 2014: 245)

Nimishkamuan passikanǹu Pieǹ ka natuapatak
I.find.it.rel rifle Pierre sub.past he.look.for.conj
‘Je trouve le fusil que Pierre cherchait.’ (‘I find the rifle that Pierre was looking for.’)

(64) Innu (Drapeau 2014: 248)

Nitshisseǹimiku eka iaǹimishuk
she.knows.about.me neg I.am.strict.conj.rel
‘Elle sait de moi que je ne suis pas sévère.’ (‘She knows of me that I am not strict.’)

In (63) the third person subject of the subordinate clause is marked by the relational -u
on the verb in the main clause (‘find’), while in (64) the third person subject of the main
clause is highlighted by the relational in the subordinate clause (‘be strict’).
Finally, Junker reports a third context for the relational in East Cree, which she calls
the “presentative”, meaning that the action happens “in the presence of someone else”
(2003a: 324).42 Consider (65):

(65) East Cree (Junker 2003a: 325)

Ni-wâpahtam-w-â-n mistiku-yû.
1-see.TI-rel-dir-1 wood-obv
‘I see a stick (in the presence of him/her).’

40 When there is an overt possessor, it is part of the possessive phrase, rather than a separate argu-
ment. For discussion of the parallels between relational verbs and external possessor constructions, see
Cenerini (2018).
41 Cenerini (2014: 45) also notes that the relevant third person could appear in a different sentence than
the one containing the relational.
42 Cenerini was not able to find examples of this in Swampy Cree (2014: 89).
 18 Applicatives across Algonquian 635

Junker says that use of (65) might imply that the stick was located near some other
person or that another person was present who did not see the stick.

8.2 TA relationals with inanimate subjects

Drapeau (2014: 246–247) reports that in Innu, the relational can also occur with a TA
verb with an inanimate subject. Figure 2 shows the parallels in structure between such
TA relationals and the more widely-described types. (66) provides an example.43

1 2 3 4
AI(+O) stem relational -w TA TS -â/-ê person
TI stem TI TS -am relational -w TA TS -â/-ê person
TA stem TA TS -(i)ku relational -u TA TS -â person

Figure 2: Relational verb templates: AI and TI verbs; TA verbs with inanimate subject.

(66) Innu (Drapeau 2014: 246)

a. Nipishtaukun utapan
it.hits.me car
‘Une auto me frappe.’ (‘A car hit me.’)
b. Nipishtaukuan Pieǹ utapan
it.hits.me.rel Pierre car
‘L’auto de Pierre me frappe.’ (‘Pierre’s car hit me.’)

Figure 2 shows that this type of relational has the same basic structure as the other
types. The theme sign in position 1 is the inverse -(i)ku because the forms have inani-
mate subjects. The second theme sign, in position 3, is restricted to -â because the con-
struction does not permit third person undergoers of the action: a sentence like ‘Jean
est frappé par l’auto de Paul’ (‘Jean is hit by Paul’s car’) cannot be expressed with a
relational verb in Innu.
Person marking in these forms is the same as we have seen for the others: in
the independent, AI/TI person marking appears, while the conjunct takes TA person
marking.

43 (66b) contains a possessive relational; Drapeau (2014: 246) also provides a cross-clausal example
which we omit for brevity.
636 Hunter Thompson Lockwood and Monica Macaulay

8.3 TA relationals with animate subjects

Several Algonquian languages, including those in the CMN group, have a morpheme -im
which occurs with TA verbs and was treated as an obviative marker in earlier analyses
(e.g. Wolfart 1973: 47 for Plains Cree and Pentland 1999: 234 for Proto-Algonquian).44
However, a number of authors (e.g. Junker 2002; Cenerini 2014; Drapeau 2014) have noted
parallels between relational -w and the -im construction. Consider the following examples:

(67) Plains Cree (Wolvengrey 2011: 67)

Nikī-wīcihimāwa cān otānisa.
ni-kī-wīcih-im-ā-w-a cān o-tānis-a
1-past-help.TA-im-dir-1>3′ John 3-daughter-obv
‘I helped John’s daughter(s).’

(68) Plains Cree (Dahlstrom 1991: 48)

wa·pam-im-e·-w o-kosis-iyiw-a
see.TA-im-dir-3 3-son-obv-obv
‘Hei [prox.] saw hisj [obv.] sonk [obv].’

(69) Moose Cree (Cenerini 2014: 104, from Ellis 2004: 499)
kî-ayâwêw napakâhtikwa kâ-kî-natawêlimimatipan.
kî-ayâw-ê-w napakâhtikw-a kâ-kî-natawêlim-im-atipan
past-have.TA-TA.dir-1>3 plank-obv.an conj-past-want.TA-im-2past
‘He had the planks which you had been wanting.’

The morpheme -im appears in similar, although not identical, constructions as those used
with relational -w: (67) and (68) show possessives with -im, and (69) shows a cross-clausal use.
Figure 3 highlights the structural parallels between between verbs with relational
-w and one with -im:45

1 2 3 4 5 6
AI(+O) stem relational -w TA TS -â/-ê person
TI stem TI TS -am relational -w TA TS -â/-ê person
TA stem TA TS -(i)ku relational -u TA TS -â person
TA stem -im TA TS -â/-ê (obv.) person (plural/obv.)

Figure 3: Relational verb templates compared to TA verbs with –im (based on Cenerini 2014: 117, Table 5.12).

44 Cenerini (2014: 108–110) shows that this construction also occurs in Kickapoo and Northern Ojibwe.
45 This template does not include all inflectional positions for the TA verb. For a more thorough treat-
ment see e.g. Wolfart (1973: 47–49) or Collette (2014: 243).
 18 Applicatives across Algonquian 637

As Figure 3 shows, -im appears immediately after the TA stem, and before the direct
theme sign, followed by person marking. In this case -â/-ê function as a true theme sign,
marking direct interactions on a TA verb.
Dahlstrom (1991: 39, 48), Wolvengrey (2011: 77), and Cenerini (2014: 114–115) all
describe two contexts for use of -im, both direct: a local person acting on a third person
obviative (illustrated by [67] and [69]), and a third person proximate acting on a further
obviative (68).46 As both Wolvengrey and Cenerini point out, in each of these contexts,
the subject and the object are two steps away from each other on the person hierar-
chy (local > 3 > 3′ > 3′′), and the added participant is thus always a third person falling
between the other two. Mühlbauer (2008: 133), following Junker (2003b), argues that -im
should be viewed as a marker of disjoint reference between the argument and the other
participant invoked in the construction, and this highlights another parallel between
the -im construction and the relational.

9 Conclusion
This chapter has surveyed applicative and applicative-like constructions across the
Algonquian language family. In what follows, we summarize the characteristics of these
constructions.

Morphology
– Canonical applicatives are marked on the predicate with a suffix. The two most
common ones, ✶-aw and ✶-amaw, are reconstructible and were distinguished in
Proto-Algonquian by the verb category to which they attached; this distinction has
been obscured in many of the present-day languages.
– Relative preverbs and comitative preverbs (the latter usually in concert with an
applicative suffix) combine with verb stems to form true applicatives, although the
corresponding derivational constructions do not (see below).
– Suffixal applicativization derives TA and TI verbs, and comitative preverbs form TA
verbs. All of the derived transitive verbs inflect regularly.
– The inflection of relational verbs appears to be cobbled together from multiple par-
adigms.

46 The CNM languages and the others with the -im construction show various differences in how it is
realized; e.g. in Innu it can occur in the inverse with an obviative subject (Drapeau 2014: 251). We leave
these differences aside for present purposes.
638 Hunter Thompson Lockwood and Monica Macaulay

Syntax
– The applied phrase is a primary object with suffixal applicatives. This is also the case
for comitatives formed with both a preverb and a suffix, but evidence is lacking for
comitatives formed just with a preverb.
– Relative roots and preverbs add an oblique argument, which is most often filled by
an adverbial particle or a nominal. Nominal obliques may be optionally marked
with a locative suffix when semantically appropriate.
– The most salient feature of the relational construction is that it does not allow an
overt AppP.
– The object of an AI+O verb is an O2, and this remains the case when an O1 is added
by applicativization. The object of a TI verb is an O1, and this base object becomes
an O2 when the verb is applicativized.
– Research remains to be done on combinations of applicativization and other voice
operations, as well as on comparison of applicatives and underived ditransitives.
– We have little information on the word order of applicative constructions in Algon-
quian. The unmarked location of the AppP in the clause appears in most of the
languages to be postverbal, but AppPs are subject to the same discourse factors that
drive word order in non-applicative clauses, and so they may occur preverbally as
well. More detailed research across the family is needed on this.
– In many of the languages the position of the AppP in relative root constructions is
flexible, as with the affixally-licensed ACs. However, in Meskwaki (Dahlstrom 2014:
57) and Miami-Illinois (Costa 2017: 363), oblique arguments (including relative root
complements) typically immediately precede the verb.

Semantics
– Most of the affixal applicatives license arguments that can be of various types, with
benefactives and goals being the most common roles added. However, the examples
provided in our sources often seem to be somewhat arbitrary, with little attempt
at exhaustive coverage of semantics. A notable exception to this is Drapeau (2011),
who contrasts what she calls the “generalized applicative” of Innu (with a range of
meanings added) to several other, more specific constructions.
– Relative roots and relative preverbs each license specific types of obliques, e.g.
source, extent, manner, etc.
– There is also a dedicated construction for the comitative across the family, which is a
true applicative in some cases in some languages, but a syntactic lookalike in others.

Lookalikes
– In two cases, the relative root construction and the comitative construction, an
applicative lookalike is created with the use of a derivational morpheme (an initial)
that licenses an additional participant or role in the construction. These are syntac-
tic lookalikes because they lack corresponding BCs.
 18 Applicatives across Algonquian 639

– The relational construction is morphologically marked but does not change the
valence of the verb; thus, it is a valence-neutral morphological lookalike.
– All of the languages have a handful of syntactic lookalikes with weak predicate
lability, usually including the verb ‘give’.

Appendix A: Sources of data by language

Language Source Abbreviation (if used)
Arapaho Cowell and Moss (2008)
Blackfoot Blackfoot Dictionary online BD
Bliss (2010, 2014)
Frantz (1991)
Taylor (1969)
Innu Brittain (1993)
Drapeau (2011, 2014)
Menominee Bloomfield (1928, 1962, 1975)
Macaulay fieldwork
Meskwaki Dahlstrom (2009, 2014, 2021)
Mi’kmaq Inglis (1986)
Myaamia Myaamia-Peewaalia Dictionary MPD
Lockwood research
Nishnaabemwin Nishnaabemwin Dictionary online ND
Valentine (2001)
Northern East Cree Eastern James Bay Dictionary on the Web
Collette (2014)
Odawa Rhodes (2006)
Ojicree Slavin (2012)
Plains Cree Dahlstrom (1991)
Plains Cree Dictionary PCD
Itwêwina Dictionary ITW
Wolfart (1973)
Wolvengrey (2011)
Proto-Algonquian Bloomfield (1946)
Hewson (n.d.)
Southern East Cree Junker (2003a)
SW Ojibwe Ojibwe People’s Dictionary OPD
Swampy Cree Ellis (2000)
Western Naskapi Brittain (2001)
Western Swampy Cree Cenerini (2014)
Unami (Delaware) Goddard (2021)
640 Hunter Thompson Lockwood and Monica Macaulay

Appendix B: Abbreviations from Bloomfield (1928)

Text Title Pages Abbreviation

A Boy is Blessed by Mosquitos and by a Hairy Serpent 554–559 BOY

How the Menomini Married 2–5 HMM
Lynx Tries to Kill a Stag 364–367 LNX
Me’napus Goes A-Visiting 186–197 MGV

Abbreviations
AC applicative construction
AI animate intransitive verb
an animate
aor aorist
appl applicative
BC base construction
ben benefactive
CMN Cree-Montagnais-Naskapi
conj conjunct
dem demonstrative
dir direct
emph emphatic
ep epenthetic
fut future
hab habitual
ic initial change
II inanimate intransitive verb
ipfv imperfective
inan inanimate
ind independent indicative
indef.act indefinite actor
inv inverse
lcl local
loc locative
neg negative
O object
obv obviative
pfv perfective
pl plural
pro pronominal
PA Proto-Algonquian
prox proximate
rel relational
sg singular
s.o. someone
 18 Applicatives across Algonquian 641

s.t. something
sub subordinate
TA transitive animate verb
TI transitive inanimate verb
TS theme sign
3′ third person obviative
3′′ third person further obviative
x>y x outranks y in a participant hierarchy
x>y x acts on y

References
Blackfoot Dictionary. 2015–2021. https://dictionary.blackfoot.atlas-ling.ca. (Accessed 3 October 2021.)
Bliss, Heather. 2010. Argument structure, applicatives, and animacy in Blackfoot. In Heather Bliss & Raphael
Girard (eds.), Proceedings of WSCLA [Workshop on Structure and Constituency in the Languages of the
Americas] 13 and 14, 58–69. Vancouver: UBC Working Papers in Linguistics.
Bliss, Heather. 2014. Marking the boundaries: Blackfoot preverbs in narrative and elicitation. In Pat Littell,
Analía Gutiérrez, Raphael Girard & Natalie Weber (eds.), Proceedings of WSCLA [Workshop on Structure
and Constituency of the Languages of the Americas] 17, 1–14. Vancouver: UCB Working Papers in
Linguistics.
Bloomfield, Leonard. 1928. Menomini texts. New York: Publications of the American Ethnological Society.
Bloomfield, Leonard. 1946. Algonquian. In Cornelius Osgood & Harry Hoijer (eds.), Linguistic structures of
Native America, 85–129. (Viking Fund Publications in Anthropology 6). New York: Viking Fund.
Bloomfield, Leonard. 1962. Charles F. Hockett (ed.), The Menomini language. New Haven: Yale University
Press.
Bloomfield, Leonard. 1975. Charles F. Hockett (ed.), Menomini lexicon. Milwaukee: Milwaukee Public
Museum Publications in Anthropology and History.
Bochner, Harry. 1984. Inflection within derivation. The Linguistic Review 3(4). 411–421.
Brittain, Julie. 1993. Two valency-increasing processes in Sheshatshit Innu-Aimun: Applicative and causative
formation. St. John’s, Newfoundland: MUN MA thesis.
Brittain, Julie. 2001. The morphosyntax of the Algonquian conjunct verb. New York: Garland.
Bruening, Benjamin. 2001. Syntax at the edge: Cross-clausal phenomena and the syntax of Passamaquoddy.
Cambridge, Massachusetts: MIT dissertation.
Cenerini, Chantale. 2014. Relational verbs: Paradigm and practice in a Manitoba dialect of Swampy Cree.
Regina, Saskatchewan: University of Regina MA thesis.
Cenerini, Chantale. 2018. External possessor constructions and Cree relational inflection compared. In
Evelien Keizer & Hella Olbertz (eds.), Recent developments in Functional Discourse Grammar 205. 89–129.
Amsterdam: John Benjamins.
Collette, Vincent. 2014. Description de la morphologie grammaticale du cri de l’Est (dialecte du Nord, Whapma-
goostui). Quebec: Laval University dissertation.
Costa, David J. 1996. Reconstructing initial change in Algonquian. Anthropological Linguistics 38(1). 39–72.
Costa, David J. 2017. Miami-Illinois word order. Anthropological Linguistics 59(4). 349–389.
Cowell, Andrew & Alonzo Moss, Sr. 2008. The Arapaho language. Boulder: University Press of Colorado.
Dahlstrom, Amy. 1991. Plains Cree morphosyntax. New York: Garland.
Dahlstrom, Amy. 1995. Topic, focus, and other word order problems in Algonquian. The Belcourt Lecture.
Winnipeg: Voices of Rupert’s Land.
642 Hunter Thompson Lockwood and Monica Macaulay

Dahlstrom, Amy. 2009. OBJθ without OBJ: A typology of Meskwaki objects. In Miriam Butt & Tracy Holloway
King (eds.), Proceedings of the LFG09 Conference, 222–239. Stanford: CSLI Publications.
Dahlstrom, Amy. 2014. Multiple oblique arguments in Meskwaki. In J. Randolph Valentine & Monica Macaulay
(eds.), Papers of the 42nd Algonquian Conference, 56–68. Albany: State University of New York Press.
Dahlstrom, Amy. 2015. Meskwaki comparatives: A first look. In J. Randolph Valentine & Monica Macaulay
(eds.), Papers of the 43rd Algonquian Conference, 15–27. Albany: State University of New York Press.
Dahlstrom, Amy. 2017. Seeking consensus on the fundamentals of Algonquian word order. In Monica
Macaulay, Margaret Noodin & J. Randolph Valentine (eds.), Papers of the 45th Algonquian Conference,
59–72. East Lansing: Michigan State University Press.
Dahlstrom, Amy. 2021. Meskwaki syntax. https://lucian.uchicago.edu/blogs/adahlstrom/publications-2/
selected-manuscripts/meskwaki-syntax-book/. (Accessed 2 June 2021.)
Denny, Yvonne, Arlene Stevens, Elizabeth Paul, Barbara Sylliboy & Dianne Friesen. 2021. A ditransitive
analysis of possessor raising in Mi’kmaq. In Monica Macaulay & Margaret Noodin (eds.), Papers of the
50th Algonquian Conference, 83–98. East Lansing: Michigan State University Press.
Drapeau, Lynn. 2011. A generalized applicative in Innu. In J. Randolph Valentine & Monica Macaulay (eds.),
Papers of the 43rd Algonquian Conference, 58–71. Albany: Albany: State University of New York Press.
Drapeau, Lynn. 2014. Grammaire de la langue Innue. Quebec: Presses de l’Université du Québec.
Eastern James Bay Dictionary on the web. 2018. https://dictionary.eastcree.org/. (Accessed 24 October
2021.)
Ellis, C. Douglas. 2000. Spoken Cree level I: West Coast of James Bay. Edmonton: University of Alberta Press.
Ellis, C. Douglas. 2004. Spoken Cree level II: West Coast of James Bay. Edmonton: University of Alberta Press.
Frantz, Donald G. 1991. Blackfoot grammar. Toronto: University of Toronto Press.
Goddard, Ives. 1974. Remarks on the Algonquian independent indicative. International Journal of American
Linguistics 40(4). 317–327.
Goddard, Ives. 1979. Delaware verbal morphology: A descriptive and comparative study. New York: Garland.
Goddard, Ives. 1980. Eastern Algonquian as a genetic subgroup. In William Cowan (ed.), Papers of the 11th
Algonquian Conference, 143–158. Ottawa: Carleton University.
Goddard, Ives. 1995. Notes on Fox (Mesquakie) inflection: Minor modes and incompletely described
morphemes. In David H. Pentland (ed.), Papers of the 26th Algonquian Conference, 124–150. Winnipeg:
University of Manitoba.
Goddard, Ives. 2002. Grammatical gender in Algonquian. In H. C. Wolfart (ed.), Papers of the 33rd Algonquian
Conference, 195–231. Winnipeg: University of Manitoba.
Goddard, Ives. 2021. A grammar of Southern Unami Delaware (Lenape). Petosky, Michigan: Mundart Press.
Hamilton, Michael David. 2017. Ditransitives and possessor raising in Mi’gmaq. In Monica Macaulay &
Margaret Noodin (eds.), Papers of the 46th Algonquian Conference, 81–100. East Lansing: Michigan State
University Press.
Hewson, John. 1991. Verbal derivation in Micmac. Journal of the Atlantic Provinces Linguistic Association 13.
21–33.
Hewson, John. n.d. Proto Algonkian word formatives. http://www.mun.ca/linguistics/people/faculty/protoal-
gonkian.php. (Accessed 4 July, 2017.)
Hoffman, Sarah & Will Oxford. 2021. Derivational functions of theme signs in Oji-Cree. In Monica Macaulay
& Margaret Noodin (eds.), Papers of the 50th Algonquian Conference, 135–153. East Lansing: Michigan
University Press.
Inglis, Stephanie Heather. 1986. The fundamentals of Micmac word formation. St. John’s, Newfoundland: MUN
dissertation.
Itwêwina Plains Cree Dictionary. 2019–2021. https://itwewina.altlab.app/. (Accessed 13 September 2021.)
Johnson, Meredith, Monica Macaulay, Bryan Rosen & Rachel Wang. 2015. A survey of word order in
Menominee. In Monica Macaulay & J. Randolph Valentine (eds.), Papers of the 43rd Algonquian
Conference, 154–178. Albany: Albany: State University of New York Press.
 18 Applicatives across Algonquian 643

Junker, Marie-Odile. 2002. La grammaire des relations: constructions possessives et obviation in Cri de l’Est.
In Sophie Burelle & Stanca Somesfalean (eds.), Canadian Linguistic Association Conference Proceedings /
Actes du Congrès de l’Association canadienne de linguistique, 145–155. https://cla-acl.ca/actes/
actes-2002-proceedings.html.
Junker, Marie-Odile. 2003a. East Cree relational verbs. International Journal of American Linguistics 69(3).
307–329.
Junker, Marie-Odile. 2003b. East Cree dependent nouns and disjoint reference. Algonquian and Iroquian
Linguistics 28(1). 11–13.
Junker, Marie-Odile & Ida Toivonen. 2015. East Cree ghost participants. In Miriam Butt & Tracy H. King (eds.),
Proceedings of the LFG15 Conference, 125–144. Stanford: CSLI Publications.
Kim, Kyumin. 2020. Oblique nominals, a verbal affix and late merge. Linguistics 58(6). 1737–1773. https://doi.
org/10.1515/ling-2020-0179.
Leman, Wayne. 2014. The Cheyenne language. Unpublished manuscript.
LeSourd, Philip S. 1995. Diminutive verb forms in Passamaquoddy. International Journal of American
Linguistics 61(1). 103–134.
LeSourd, Philip S. 2006. Problems for the Pronominal Argument Hypothesis in Maliseet-Passamaquoddy.
Language 82(3). 486–514.
Lochbihler, Bethany Christina. 2012. Aspects of argument licensing. Montreal: McGill University
dissertation.
Macaulay, Monica. 2009. On prominence hierarchies: Evidence from Algonquian. Linguistic Typology 13(3).
357–389.
Macaulay, Monica & Joseph Salmons. 2017. Synchrony and diachrony in Menominee derivational
morphology. Morphology 27(2). 179–215.
Malchukov, Andrej, Martin Haspelmath & Bernard Comrie. 2010. Ditransitive constructions: a typological
overview. In Andrej Malchukov, Martin Haspelmath & Bernard Comrie (eds.), Studies in ditransitive
constructions: A comparative handbook, 1–65. Berlin: Mouton de Gruyter.
Meadows, Kimberley. On the role of sentience in Blackfoot: Evidence from the accompaniment and associative
constructions. Calgary: University of Calgary MA thesis.
Mühlbauer, Jeffrey Thomas. 2008. kâ-yôskâtahk ôma nêhiyawêwin: The representation of intentionality in Plains
Cree. Vancouver: University of British Columbia dissertation.
Myaamia-Peewaalia Dictionary. 2021. https://mc.miamioh.edu/ilda-myaamia/dictionary. (Accessed 25
October 2021.)
Nishnaabemwin dictionary (Odawa and Eastern Ojibwe online dictionary). 2015–2021. https://dictionary.
nishnaabemwin.atlas-ling.ca. (Accessed 5 October 2021.)
Ojibwe People’s Dictionary. 2012–2021. https://ojibwe.lib.umn.edu. (Accessed 4 October 2021.)
O’Meara, John. 1992. Intransitive verbs with secondary objects in Munsee Delaware. In William Cowan (ed.),
Papers of the 23rd Algonquian Conference, 322–333. Ottawa: Carleton University.
Oxford, Will. 2019. Inverse marking and multiple agree in Algonquin: Complementarity and variability.
Natural Language and Linguistic Theory 37. 955–996.
Pentland, David H. 1999. The morphology of the Algonquian independent order. In David H. Pentland (ed.),
Papers of the 30th Algonquian Conference, 222–266. Winnipeg: University of Manitoba.
Plains Cree dictionary. 2015–2021. https://dictionary.plainscree.atlas-ling.ca. (Accessed 13 December 2021.)
Reinholtz, Charlotte. 1999. On the characterization of discontinuous constituents: evidence from Swampy
Cree. International Journal of American Linguistics 65(2). 201–227.
Rhodes, Richard A. 1990. Ojibwe secondary objects. In Katarzyna Dziwirek, Patrick Farrell & Errapel
Mejías-Bikandi (eds.), Grammatical relations: A cross-theoretical perspective, 401–414. Stanford: CSLI
Publications.
Rhodes, Richard A. 2006. Clause structure, core arguments and the Algonquian relative root construction.
Belcourt Lecture. Winnipeg: Voices of Rupert’s Land.
644 Hunter Thompson Lockwood and Monica Macaulay

Rhodes, Richard A. 2010. Missing obliques: Some anomalies in Ojibwe syntax. In Donna B. Gerdts, John C.
Moore & Maria Polinsky (eds.), Hypothesis A / Hypothesis B: Linguistic explorations in honor of David M.
Perlmutter, 427–456. (Current Studies in Linguistics 49). Cambridge, Massachusetts: MIT Press.
Rhodes, Richard A. & Rand Valentine. 2015. Transitivity in Ojibwe. In Andrej Malchukov & Bernard Comrie
(eds.), Valency classes in the world’s languages, Vol. 2: Case studies from Austronesia, the Pacific, the
Americas, and theoretical outlook, 1205–1264. Berlin: De Gruyter Mouton.
Slavin, Tanya. 2012. The syntax and semantics of stem composition in Ojicree. Toronto: University of Toronto
dissertation.
Sullivan, Michael. 2012. Making statements in Ojibwe: A survey of word order in spontaneous sentences.
In Monica Macaulay, Margaret Noodin & J. Randolph Valentine (eds.), Papers of the 44th Algonquian
Conference, 329–347. Albany: State University of New York Press.
Taylor, Allan Ross. 1969. A grammar of Blackfoot. Berkeley: University of California dissertation.
Valentine, J. Randolph. 2001. Nishnaabemwin reference grammar. Toronto: University of Toronto Press.
Wolfart, H. Christoph. 1973. Plains Cree: A grammatical study. Transactions of the American Philosophical
Society, n.s. 63(5). 1–90. Philadelphia: American Philosophical Society.
Wolvengrey, Arok Ellessar. 2011. Semantic and pragmatic functions in Plains Cree syntax. Utrecht: LOT
Publications.
Zúñiga, Fernando. 2008. How many hierarchies, really? Evidence from several Algonquian languages.
Linguistische Arbeits Berichte 86. 277–294.
Brad Montgomery-Anderson
19 Applicative constructions in Mayan
languages: An overview with special focus
on Chontal
Abstract: This chapter describes applicative constructions in Mayan languages, one of
the largest and most well-known Native American language families. Mayan languages
abound in valency-altering verb affixes that produce passive, antipassive, causative,
and applicative constructions. The first part of this study presents the relevant features
of Mayan morphosyntax using Kaqchikel, a Guatemalan language from the Eastern
branch, and Chontal, a Mexican language from the Western branch. The second part of
the study reviews the published data to examine examples of applicative and applica-
tive-like constructions in the four branches of the family. This overview is followed by
an in-depth analysis of applicative constructions in Chontal Mayan; these examples are
from the author’s field work with the Project for the Documentation of the Languages
of Mesoamerica. The final part of the paper briefly summarizes the morphosyntactic
and semantic properties of applicative construction across the Mayan language family.

1 Introduction
This chapter describes applicative constructions in Mayan languages, one of the largest
and most well-known Native American language families. The commonly accepted
grouping of the thirty languages in this family posits four branches (Campbell 2017: 44).
The Eastern Branch languages are found in Guatemala; this grouping includes several
large languages (e.g. Q'eqchi', K'iche', and Kaqchikel) with speakers numbering in the
hundreds of thousands. Languages in the Western Branch are spoken in southern
Mexico. The northernmost Mayan language, Huastec, comprises its own group and
is thought to have split off first from the other languages. The fourth branch, Yucate-
can, comprises several languages, among which is Yucatec, the most-spoken Mayan
language of Mexico. Languages in the Mayan family have a rich inventory of valen-
cy-altering verb affixes that produce passive, antipassive, causative, and applicative
constructions. Applicatives and applicative lookalikes are found throughout the Mayan
language family.

https://doi.org/10.1515/9783110730951-019
646 Brad Montgomery-Anderson

2 The basics of Mayan verbal morphosyntax

In this section I will describe relevant Mayan morphosyntax using examples from an
Eastern Mayan language, Kaqchikel, and a Western Mayan language, Chontal.1 Unless
stated otherwise, all Chontal data comes from my own fieldwork with the Project for
the Documentation of the Languages of Mesoamerica (PDLMA), and all Kaqchikel data
comes from my teaching material at the Center for Latin American Studies at the Uni-
versity of Kansas.

2.1 Basic verb morphology

Mayan languages are agglutinating, and verbs consist of roots with affixes that express
person as well as tense, aspect, mood, and voice. Verb roots are either transitive or
intransitive. Person marking has an ergative rather than an accusative alignment. The
labels Set A and Set B are used with Mayan languages for the two sets of affixes: tran-
sitive verb subjects take the ergative Set A affixes, and objects and intransitive verb
subjects take the absolutive Set B affixes. Table 1 shows the Set A forms from the two
languages alongside the reconstructed Proto-Mayan forms (Mora-Marín 2009: 103).

Table 1: Set A person affixes.

Chontal Proto-Mayan Kaqchikel

1sg kä- (k-) ✶
nu nu- (i-, in-, w-)
2sg a- (aw-) ✶
aa a- (aw-)
3sg u- (uy-, y-) ✶
ru ru- (r-, u-)
1pl kä-. . .laj/ t'oko' ✶
qa qa- (q-)
2pl a-. . .laj ✶
ee i- (iw-)
3pl u-. . .laj ✶
ki ki- (k-)

Kaqchikel has only undergone slight change from the reconstructed proto-Mayan forms.
Chontal, on the other hand, has undergone serious innovation of its person-marking
paradigm. For example, it has used the former first-person plural marker as its first-per-
son singular marker. To fill in the gap this creates in the plural, it uses the new singular
form plus the morpheme laj, the grammaticalized word for ‘all.’ Moreover, Chontal has

1 Languages from Guatemala, including Kaqchikel, follow orthographic conventions laid out by the
Guatemalan Academy of Mayan Languages, known as the Academia de Lenguas Mayas de Guatemala
(ALMG). Languages from Mexico, including Chontal, follow orthographic conventions laid out by the
National Indigenous Languages Institute, known as the Instituto Nacional de Lenguas Indígenas (INALI).
For ALMG the letter <ä> represents the phoneme [ɐ]; INALI uses the letter <ä> to represent [ɘ]. Stress
in Chontal is unpredictable, and I represent it with an accent; stress is typically not represented in the
various orthographies.
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 647

an inclusive/exclusive distinction: the exclusive is formed with t'oko'. As described by

Law (2014: 84–92), this distinction is an areally spread innovation that is also found in
Yucatecan, Chol, Tojolabal, Tseltal, and Tsotsil. In Chontal, the second- and third-person
singular affixes look similar to the proto forms, but their plural counterparts have been
simplified; they are the singular form with the laj morpheme (“all”) added. The Set B
affixes have singular markers similar to the proto forms; once again, however, we see
that Chontal bases the plural form on the singular and adds a new inclusive/exclusive
distinction. Another important difference is that in Chontal the Set B affixes come after
the verb, whereas in Kaqchikel they precede it. Table 2 shows the Set B forms from the
two languages alongside Proto-Mayan reconstructed forms (Mora-Marín 2009: 103).

Table 2: Set B person affixes.

Chontal Proto-Mayan Kaqchikel

1sg -on ✶
iin in-
2sg -et ✶
at at-
3sg ∅ ✶
∅ ∅
1pl -on-. . .laj/ t'oko' ✶
o'n oj-
2pl -et-. . .laj ✶
ex ix-
3pl ∅. . .laj ✶
ob' e-

In the verbal systems of the two languages there is a sharp contrast in their expression
of tense and aspect. Kaqchikel uses a relatively straightforward system where aspect is
marked by a prefix on the verb. There are three aspectual prefixes: x- for completive,
y- for incompletive (the allomorph n- is used before the Set B third person singular
marker), and xk- for future. These aspect prefixes are the same whether the verb is
transitive (1a) or intransitive (1b–c).

(1) Kaqchikel
a. x-in-a-to'
compl-B1-A2-help
‘You helped me.’
b. x-∅-kos
compl-B3-get.tired
‘She got tired.’
c. xk-e-wär
fut-B3.pl-sleep
‘They will sleep.’

In Chontal, markers placed before the verb help to indicate tense and aspect. There is no
verbal tense in Chontal; the particles, together with aspect markers, serve to specify the
time frame. For example, the aspectual marker mu' is used with the incompletive suffix
648 Brad Montgomery-Anderson

to indicate a present progressive time frame, while ja is used with the completive status
marker to indicate the immediate future. As many other Mayan languages, Chontal has
two distinct classes of verbs: root verbs and derived verbs. The aspect suffixes differ
depending on the verb’s classification as transitive or intransitive, root or derived. A
root transitive verb has an incompletive -e' suffix (2a), whereas a root intransitive has
an incompletive -e suffix (2b).

(2) Chontal
a. kä-män-é'-∅
A1-buy-inc-B3
‘I buy it.’
b. kä-bix-é
A1-go-inc
‘I go.’

Derived verbs use the suffix -n to indicate incompletive aspect on both transitive (3a)
and intransitive verbs (3b).

(3) Chontal
a. u-pok'má-n
A3-become.fat-inc
‘He becomes fat.’
b. u-láchä-n-∅
A3-scratch-inc-B3
‘She scratches it.’

The completive suffix is always the same regardless of the verb being root or derived,
transitive or intransitive. It can, however, change according to person. The third person
Set B suffix is not phonetically expressed and the completive is -i (4a). When a Set B
suffix is phonetically expressed—first or second person—then it is the completive
aspect that is zero-marked (4b and 4c).

(4) Chontal
a. a=t'äb-í-∅
asp=go.up-compl-B3
‘He went up.’
b. a=t'äb-∅-ón
asp=go.up-compl-B1
‘I went up.’
c. a=t'äb-∅-ét
asp=go.up-compl-B2
‘You went up.’
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 649

Kaqchikel, like other Mayan languages, has ergative alignment for person-marking on
verbs. Set A prefixes index transitive subjects (5a), and Set B prefixes index objects (5a)
and intransitive subjects (5b).

(5) Kaqchikel
a. x-in-ru-chäp
compl-B1-A3-grab
‘She grabbed me.’
b. x-in-jalon
compl-B1-get.dressed
‘I got dressed.’

Chontal has developed a pattern of split ergativity. Intransitive verbs that are positive
incompletive follow an accusative pattern and use Set A prefixes to index the subject
(6a). Intransitive verbs that are completive (6b) or negative incompletive (6c) follow an
ergative pattern and use Set B prefixes to index the subject.

(6) Chontal
a. kä-wäy-é
A1-sleep-inc
‘I’m sleeping.’
b. a=wäy-∅-ón
asp=sleep-compl-B1
‘I slept.’
c. mach jul-∅-ón tä otót
neg come-compl-B1 to house
‘I didn’t come to the house.’

A common feature of Mayan languages is a robust system of valency-changing construc-

tions. In Kaqchikel a root verb forms a passive by changing the vowel (7). The demoted
subject in the passive is indicated by a Set A prefix on the relational noun -uma.

(7) Kaqchikel
a. x-e-ki-chäp
compl-B3.pl-A3.pl-grab
‘They grabbed them.’
b. x-e-chap k-uma ri ixoq-i'
compl-B3.pl-grab.pass A3.pl-by det woman-pl
‘They were grabbed by the women.’

In Chontal the passive is formed by adding a suffix to the verb. A typical active to passive
operation is in (8). In (8b) the demoted subject is part of a prepositional phrase.
650 Brad Montgomery-Anderson

(8) Chontal
a. kä-k'ux-é'-∅
A1-eat-inc-B3
‘I eat it.’
b. u-k'ux-ká-n k'a no'on
A3-eat-pass-inc by 1pro
‘It is eaten by me.’

Passivization in Chontal also distinguishes between root and derived verbs: a root verb
takes the suffix -ka (9a) and a derived verb takes the suffix -int (9b). Occasionally a verb
takes both (9c).

(9) Chontal
a. u-k'ech-ká-n
A3-grab-pass-inc
‘It is grabbed.’
b. u-tsäms-ínt-e
A3-kill-pass-inc
‘It is killed.’
c. uy-ä'-k-ínt-e tan lówen
A3-put-pass-inc in hole
‘It is put in the hole.’

2.2 Basic noun morphosyntax

In Mayan languages, some nouns are always possessed (e.g. body parts), while other
nouns never are. Moreover, some nouns alter their form and/or meaning to some extent
when possessed. Mayan languages are head-marking, and the Set A prefix is attached
to the possessed noun rather that the possessor noun. In the Kaqchikel example in (10),
‘hand’ is possessed with a first-person singular prefix, and ‘corner’ is prefixed by the
third-person prefix ru-; this prefix indicates the following noun is its possessor.

(10) Kaqchikel
x-∅-i-jös ri nu-q'a' ch-wäch ru-tzaltemal ri tz'alän
compl-B3-A1-scratch det A1-hand prep-front A3-corner det table
‘I scratched my hand on the corner of the table.’ (Macario et al. 1998: 301)

In the Chontal example (11), ‘house’ is possessed by third person singular prefix; the
possessor ‘Daria’ follows ‘house.’
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 651

(11) Chontal
t-uy-otót ix-dach' u-x-é tä ajtä lotoj-es-yá
prep-A3-house fem-Daria A3-go-inc prep be marry-caus-nmlz
‘In Daria’s house there’s going to be a wedding.’

Nouns in Mayan languages are not marked for case; core arguments are marked on the
verb, and the relationship of a non-core noun to the verb is expressed through preposi-
tions and relational nouns. A relational noun is a grammaticalized noun bearing a Set
A prefix that cross-references a non-core argument. For example, the Set A prefix on
the Kaqchikel relational noun -ichin indexes ‘my mother’ (12a); in (12b) the Set A prefix
appears on the relational noun -uma to index the demoted agent of a passive construc-
tion. In (12c) the recipient of the verb ‘give’ is not expressed on the verb, but rather with
a following relational noun. Relational nouns perform the role of prepositions but are
possessed nouns that have acquired a preposition-like meaning. Some relational nouns
are based on body parts, while for others the original meaning has been lost.

(12) Kaqchikel
a. x-i-loq'on r-ichin nu-te'
compl-A1-shop A3-with A1-mother
‘I shopped with my mother.’ (Brown, Maxwell, and Little 2006: 175)
b. x-in-qetëx k-uma
compl-B1-hug.pass A3.pl-by
‘I was hugged by them.’
c. n-∅-in-ya' chawe chwa'q
inc-B3-A1-give prep.A2 something
I give you something.’

The two true prepositions in Kaqchikel are distinguished from relational nouns by their
inability to take a Set A prefix. These two prepositions are grammaticalized from words
indicating parts of the body. Chi comes from chi’aj ‘mouth’ and pa from pamaj ‘stomach.’
An example with pa is in (13).

(13) Kaqchikel
jantape’ y-oj-ch'on pa Kaqchikel
always inc-B1.pl-speak in Kaqchikel
‘We always speak in Kaqchikel.’

In Chontal we see a similar use of relational nouns and prepositions. As in Kaqchikel,

some possessed body parts have become grammaticalized as relational nouns; for
example, ‘back’ (14a) and ‘mouth’ (14b). In both these examples the possessed body part
is used with one of the handful of true prepositions, in this case a shortened form of tä
‘in, on, at.’ Prepositions can also appear by themselves (14c).
652 Brad Montgomery-Anderson

(14) Chontal
a. ni yum+ká-jo' u-che-n-∅-jo' ch'úyu t-u-pat té'
det lord+earth-pl A3-do-inc-B3-pl whistle prep-A3-back tree
‘The duendes whistle behind the tree.’
b. kol-∅-ón kä-pítä-n-et t-u-ti' ch'uj
remain-compl-B1 A1-wait-inc-B2 prep-A3-mouth church
‘I stayed waiting for you in front of the church.’
c. ixkúne uy-äl-é'-∅ tä kraxtán t'an
what A3-say-inc-B3 in Spanish language
‘What does it mean in Spanish?’

3 The Mayan applicatives

3.1 Mayan comparative data on the applicative

The definition of applicative construction (AC) and base construction (BC) used in this
study is from Zúñiga and Creissels (this volume). An applicative verb construction
indexes a participant that, in the BC, either has different verb indexing or is not indexed
at all. The applicative verb itself is derived from a base verb. The applied phrase is not
a subject or agent in either construction. A canonical applicative construction treats the
phrase as a core argument. Some Mayanists use the term “registration applicative” to
indicate an applicative lookalike construction that focuses the phrase but does not have
verb indexing different from the BC (Smith-Stark 1994); in this paper I will refer to such
constructions as oblique registration constructions.
Dixon (2012: 336–337) lists eight areas that need to be addressed when describing
the applicative construction (AC): 1) the formal marking for the application construction,
2) whether a corresponding base construction (BC) for the applicative construction exists
(if not, it is not a canonical applicative construction), 3) if there are factors that determine
when the applicative construction is preferred or required (e.g., a human applicative
argument), 4) whether the applicative argument can only be expressed in an applicative
construction (if so, the construction is a quasi-applicative), 5) the semantic roles coded
by the applicative construction, 6) whether the construction applies to intransitive verbs,
transitive verbs, or both, 7) the status of the original object from the BC after the AC has
licensed the applied argument (i.e. is it equal in status to the new object, or has it lost
some of its primary object properties), and 8) the origin of the applicative marker.
Bearing in mind these criteria, I have found that the current published research
does not allow us to describe satisfactorily applicative constructions and applicative
lookalikes across the Mayan language family. In the published descriptions that do
exist, the applicative is illustrated with but a few examples; often these cursory descrip-
tions fail to make it clear if the AC has a corresponding BC. For any given language,
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 653

we would need to analyze many examples from varied discourse contexts in order to
attain a true understanding of this construction. The most detailed descriptions of the
applicative are my own inadequate description of its manifestation in a single Mayan
language (Montgomery-Anderson 2010) and Mora-Marín’s 2003 diachronic study of the
Mayan language family. Given these limitations, I will endeavor in this chapter to give
an account of what is currently known about this construction by summarizing the
available data for each of the four branches: Eastern (§ 3.2), Huastecan (§ 3.3), Yucate-
can (§ 3.4), and Western (§ 3.5).2

3.2 Eastern Mayan

3.2.1 Mamean

The Mamean subbranch of Eastern Mayan consists of four languages: Mam, Tekitek,
Awakatek, and Ixil (Campbell 2017: 44). Mam, the largest language, has no applicative
construction (England 2017). Ixil is the only language in this subgroup of Eastern Mayan
where we have evidence of an applicative-like construction. Mora-Marín reports that the
AC advances an instrument (15a) that in the BC is in an oblique phrase with a relational
noun (15b). The applicative morpheme is b'e, by far the most common applicative mor-
pheme across the Mayan language family and the topic of Mora-Marín’s diachronic study
(2003). The applicative morpheme has focused the instrument, but it has not marked it as
a core argument; in neither the AC nor the BC does the verb index it with person affixes.3

(15) Ixil (Mora-Marín 2003: 207)

a. uula a-k'oni-b'e=in
sling A2-shoot-appl=B1
‘With a sling you shot me.’ AC
b. a-k'oni=in t'-an uula
A2-shoot=B1 A3-rn sling
‘You shot me with a sling.’ BC

2 For many examples I have slightly altered the parsing abbreviations and symbols to make them more
consistent with those used in the paper. I have also standardized some of the letters used for the languag-
es themselves; for example, I have replaced any instances of <š> with <x>, a letter used in many Mayan
languages to represent a voiceless post-alveolar fricative. Any mistakes are my own.
3 In (15a) the instrument phrase is not introduced by a relational noun, so this construction is not a
typical oblique registration construction.
654 Brad Montgomery-Anderson

3.2.2 K'iche'an

The K'iche'an subbranch consists of nine languages: Q'eqchi', Uspantek, Poqomam,

Poqomchi', K'iche', Kaqchikel, Tz'utujil, Sakapultek, and Sipakapense (Campbell 2017:
44). For some of these languages we have evidence that the applicative morpheme
b'e appears on a transitive verb when an instrument is focused. In the BC in (16b) the
instrument follows the verb and is introduced by a relational noun. In (16a) the instru-
ment is in focus position, is not introduced by a relative noun, and is indexed on the
verb by the Set B prefix; the argument that was indexed by the Set B prefix in the BC is
now introduced by a relational noun.

(16) K'iche' (Campbell 2000: 278)

a. ch'iich' x-∅-u-rami-b'e-j lee achih r-eeh lee chee'
metal compl-B3-A3-cut-appl-tr det man A3-gen det tree
‘The man used a machete to cut the tree.’ / ‘A machete is what a man used to
cut a tree.’ (AC)
b. x-∅-u-rami-j lee chee' lee achih ch-eeh jun ch'iich'
compl-B3-A3-cut-tr det tree det man to.3poss-to det metal
‘The man cut the tree with a machete.’ (BC)

In (17a) the demoted object is a local person; this argument is expressed as a Set A prefix
on a relational noun.

(17) K'iche' (Campbell 2000: 278)

a. chee' x-∅-in-ch'aya-b'e-j aaw-eej
wood compl-B3-A1-hit-appl-tr 2poss-gen
‘I used a stick to hit you.’ (AC)
b. x-at-in-ch'ay chi chee'
compl-B2-A1-hit with wood
‘I hit you with a stick.’ (BC)

Henderson reports instances in Kaqchikel where the b'e applicative morpheme allows
the verb to index comitatives (18a), locatives (18b), datives (18c), and themes / topics of
speech (18d). Unfortunately, I do not have examples of corresponding BCs, so it is not
known if these instances of b’e are lexicalized or represent true applicative construc-
tions.

(18) Kaqchikel (Henderson 2007: 10)

a. x-∅-u-samaji-b’e-j w-achb'il
compl-B3-A3-work-appl-tr A1-friend
‘He worked with my friend.’
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 655

b. x-∅-u-pal-b'e-j jun pop ri ixtän

compl-B3-A3-stand-appl-tr det mat det girl
‘The girl stood on a mat.’
c. x-∅-u-ch'o-b'e-j ri ixöq
compl-B3-A3-speak-appl-tr det woman
‘He spoke to the woman.’
d. x-at-ru-tzijo-b'e-j ri ixöq
compl-B2-A3-say-appl-tr det woman
‘The woman talked about you.’

Kaqchikel uses the b'e to focus an instrument in an oblique registration construction

(England 2001: 139). In (19) ikäj ‘axe’ is fronted and its former position is marked by the
particle wi; however, it is still part of an oblique phrase and it is not indexed on the verb.

(19) Kaqchikel (England 2001: 139)

chi ikäj x-∅-u-choyo-b'e-j wi ri che' ri achi
prep axe compl-B3-A3-cut-appl-tr foc det tree det man
‘With an axe the man cut the tree.’

Kaqchikel uses prepositions and relational nouns to introduce an instrument (20a), a

beneficiary (20b), a recipient (20c), and a target (20d).

(20) Kaqchikel
a. aPala's x-∅-u-paxij ri b'ojo'y chi ab'äj
Francisco compl-B3-A3-break det pot prep stone
‘Francisco broke the pot with stones’ (García Matzar et al. 1992: 103)
b. x-∅-ch'ajon k-ichin ru-ch'utit-e
compl-B3-wash.clothes A3.pl-for A3-aunt-pl
‘She washed clothes for her aunts.’ (Brown, Maxwell, and Little 2006: 175)
c. yin x-∅-in-tz'ib'aj el jun wuj chre jun w-ach'al
1pro compl-B3-A1-write dir det paper prep.A3 det A1-friend
‘I wrote a letter to a friend.’ (Macario et al. 1998: 383)
d. yin x-∅-in-q'eb'a' jun che' pa ru-wi' raqänya'
1pro compl-B3-A1-lay.across det wood on A3-top river
‘I laid a plank across the river.’ (Macario et al. 1998: 263)

Stout argues that Kaqchikel has an adnominal strategy where the third argument of the
verb appears as a possessor of one of the verb’s core arguments. In both (21a) and (21b)
656 Brad Montgomery-Anderson

the Set A prefix on the noun can be interpreted as both a possessor and a recipient.
In (21c) the Set A possessor is a maleficiary, an interpretation reflected in the Spanish
translation (the English translation reflects more literally the Kaqchikel).

(21) Kaqchikel
a. Juan x-∅-u-ya' jun nu-wuj
Juan compl-B3-A3-give det A1-book
‘Juan gave me a book.’ (Lit. ‘Juan gave my book.’) (Stout 2015: 71)
b. x-∅-u-taq pa jun nu-tzibanik
compl-B3-A3-send prep det A1-letter
‘She sent me a letter.’ (Macario et al. 1998: 436)
c. jun achin x-∅-r-eleq'aj jun w-ikäj
det man compl-B3-A3-rob det A1-axe
‘A man robbed my axe.’ / ‘Un señor me robó un hacha.’ (Macario et al. 1998: 83)

Tz'utujil and several varieties of K'iche' use the -b'e suffix. In (22b) the addressee ‘his
wife’ is marked with the relational noun -uk'iin ‘with’ and is not marked on the verb; in
(22a) the applicative allows ‘his wife’ to be indexed by the Set B prefix.

(22) Tz'utujil (England 2001: 140)

a. x-u-tzijo-b’eej ja r-ixoqil jar aachi chi r-iij jar uleew
compl-A3-speak-appl det A3-wife det man prep A3-back det land
‘The man spoke with his wife about the land.’ (AC)
b. x-∅-tzij-oon-i jar aachi r-uk'iin ja r-ixoqiil chi
compl-B3-speak-antip-pfs det man A3-with det A3-wife prep
r-iij jar uleew
A3-back det land
‘The man spoke with his wife about the land.’ (BC)

In (23a) the applicative form of the K'iche' verb allows the Set B prefix to index ‘me’; in
the non-applicative construction in (23b) the Set B prefix indexes ‘the money.’

(23) K'iche' (England 2001: 140)

a. x-in-u-tzaqi-b'eej ri nu-chaaq'
compl-B1-A3-appl det A1-brother
‘My brother lost it for me.’
b. x-∅-u-tzaq ri nu-pwaq ri nu-chaaq'
compl-B3-A3-lose det A1-money det A1-brother
‘My brother lost my money.’
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 657

Both Poqomam and Poqomchi' have an applicative construction that indexes an instru-
ment. The applicative suffix 'i allows the instrument to be fronted for questioning; in
(24a) the patient ‘me’ is still expressed on the verb with a Set B prefix, while in (24b) the
same patient is demoted to a relational noun phrase and the instrument is indexed on
the verb with the Set B prefix.

(24) Poqomam (Dayley 1981: 33)

a. hila' x-in-a-q'at-'i-eh
what compl-Bl-A2-cut-ins-tr
‘What did you cut me with?’
b. hila' x-∅-a-q'at-'i-eh w-ihchin
what compl-B3-A2-cut-ins-tr Al-to
‘What did you cut me with?’

3.3 Huastecan

The Huastecan branch split off first from the other Mayan languages and is the most
divergent; this branch is comprised of one living language (Campbell 2017: 44). There
are three varieties of Huastec; the description of the variety from Xiloxúchil (in the state
of Veracruz) does not contain much information on ditransitive constructions, but we
can find an example of an AC (25a) and a corresponding BC (25b). In this language transi-
tive verbs are marked with the morpheme ch(i) (reduplicated as chinch) to form the AC.

(25) Xiloxúchil (Ochoa Peralta 1984: 68)

a. ∅ in t'aj-ch-al
B3 A3 do-appl-inc
‘He does it to him.’ (AC)
b. hahaa' ∅ in t'aj-al
3pro B3 A3 do-inc
‘He does it.’ (BC)

This applicative morpheme can license a beneficiary (26a), recipient (26b), or malefi-
ciary (26c).

(26) Xiloxúchil (Dayley 1981: 55)

a. ∅ u nuj-chinch-al
B3 Al B3 sell-appl-inc
‘I sell it for him.’
b. ∅ u nuj-ch-al
B3 Al sell-appl-inc
‘I sell it to him.’
658 Brad Montgomery-Anderson

c. ∅ in kwe'-ch-al
B3 A3 rob-appl-inc
‘He robs him of it.’

This applied argument can be passivized (27).

(27) Xiloxúchil (Dayley 1981: 55)

a. in kwe'-chi-n-al
Bl rob-appl-pass-inc
‘I am (being) robbed of it.’
b. in nuj-chinch-aab'
Bl sell-appl-pass
‘I am being sold it.’

Kondić (2012) provides much more detail about applicative constructions for Southeast-
ern Huastec. The applicative in this variety has several distinctive features that make
it unlike the applicative in the other Mayan languages. Southeastern Huastec uses two
applicatives, both of which are etymologically unrelated to the morpheme b'e found
in the other branches of the Mayan languages. Kondić refers to the morpheme -tx as
a dative applicative; this morpheme appears on transitive verbs and references recip-
ients and beneficiaries. As demonstrated in (28), this construction is not obligatory. In
the base construction the beneficiary/recipient is indexed with a preposition (28b); in
the dative applicative construction the direct object of the base construction is demoted
while the applied object is treated as a core argument (28a).

(28) Southeastern Huastec (Kondić 2012: 250)

a. u nuj-tx-iy an ti kwita' na josee
A1 sell-dat.appl-trc det prep chicken hum José
‘I sold José the chicken.’ (AC)
b. u nuj-uw an kwita' baal na josee
A1 sell-trc det chicken to/for hum José
‘I sold the chicken to/for José.’ (BC)

A unique feature of the dative applicative in this variety of Huastecan is the role of defi-
niteness. If the direct object in the base construction is indefinite (29b), it is not demoted
with a preposition in the corresponding applicative constriction (29a).

(29) Southeastern Huastec (Kondić 2012: 250)

a. tu ch'a'-tx-iy juun i maango
1>2 buy-dat.appl-trc det nm mango
‘I bought you a mango.’ AC
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 659

b. u ch'a'-iy juun i maango baal tataa'

A1 buy-trc det nm mango for you
‘I bought a mango for you.’ BC

Kondić demonstrates that Southeastern Huastec has a distinct morpheme -n for instru-
mental applicative constructions. An intriguing feature of this morpheme is that it is
only used on intransitive and detransitivized verbs. As shown in (30), this instrumental
applicative construction is not obligatory. In the base construction the instrument is
introduced with a preposition (30b); in the applicative construction the direct object
of the base construction is demoted while the applied object is treated as a core argu-
ment (28a).

(30) Southeastern Huastec (Kondić 2012: 256)

a. in t'ak-ax-n-a' axee' i kwaa'txim
A3 wash-antip-ins-trc dem nm cloth
‘He washed (things/it/him) with this cloth.’
b. na martiin in t'ak-a-al n-in chikaam-il
hum Martin A3 wash-tr-inc dem-A3 child-poss
k'aal juun i kwaa'txim
with det nm cloth
‘Martin washes his child with a cloth.’

The instrumental applicative construction seems to be the most productive with the
antipassive. Kondić does report uses of the instrumental on intransitive verbs, but she
is unsure how productive such constructions are. She does give an example where the
instrumental applicative and dative applicative appear together (31); she even reports
instances where the dative morpheme appears twice on the same verb (32).

(31) Southeastern Huastec (Kondić 2012: 259)

a. ne'ech k-u t'aj-a' juniil axee' i te'neel
fut irr-A1 do-tr again dem nm food
‘I am going to make that dish again,
b. jaach n-u ne'ech ay-tx-ix-n-a-al taam ch'ejel
that rel-A1 hab wait-dat.appl-antip-ins-tr-inc when half
k'iitxaaj
day
‘It is what I usually wait (for my family) with at noon.
(=offer to my family for lunch)’
660 Brad Montgomery-Anderson

(32) Southeastern Huastec (Kondić 2012: 260)

axee' i kwitx in
dem nm wrap A1
nuj-tx-in-tx-in-al k'aal n-u atk'imaath
sell-dat.appl-ep-dat.appl-mid-inc by dem-Al neighbor
‘My neighbor sells (to people) these wraps for me.’
(lit: I am sold (to people) these wraps by my neighbor.)

Kondić notes that the instrumental construction is much less common than the dative
construction; she also suggests that, in many instances, the instrumental may represent
lexicalization.

3.4 Yucatecan

The Yucatecan branch consists of four languages: Yucatec Maya, Lacandón, Itzaj, and
Mopan (Campbell 2017: 44). So far, no evidence has been found of an applicative con-
struction in any of the Yucatecan languages (Mora-Marín 2003: 199). To express a third
argument, Itzaj can use an adnominal strategy. In (33a) the recipient is part of a prep-
ositional phrase, while in (33b) the possessive construction expresses the same event
without the preposition.

(33) Itzaj
a. k-in-ts'(ah)-ik-∅ in-ts'on ti'ih in-suku'un
inc-A1-give-ppm-B3 A1-gun prep A1-brother
‘I give my gun to my older brother.’ (Hofling 1990: 550)
b. k-in-ts'(ah)-ik-∅ u-ts'on in-suku'un
inc-A1-give-ppm-B3 A3-gun A1-brother
‘I give my brother’s gun (to him).’ (Hofling 1990: 551)

We see in Itzaj that the two semantic roles most commonly indexed by the applicative in
Mayan languages—recipients (34a) and instruments (34b)—are introduced with rela-
tional nouns.

(34) Itzaj
a. b'a'ax ki-b'el ki-k'ub'-u' t-a' winik ka' tak-ej
what A1.pl-go A1.pl-deliver-dts to-det man when come/dis-top
‘What are we going to deliver to the man when he comes?’(Hofling 2000: 316)
b. u-chun a' che'-ej b'äk'-a'an et-e sum
A3-trunk det tree-top wind-ptcp with-poss rope
‘The trunk of the tree is wound with rope.’(Hofling 2000: 317)
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 661

3.5 Western Mayan

3.5.1 Q'anjob'alan

The Greater Q'anjob'alan subbranch consists of six languages: Q'anjobal, Akatek,

Jakaltek, Mocho', Chuj, and Tojolabal (Campbell 2017: 44). There is not clear evidence of
an applicative construction in these languages. Tojolabal does have a serial verb con-
struction; the verb i' ‘to carry’ has been lexicalized to introduce a third argument as part
of a dative construction (35). The verb ‘to carry’ can appear in the same clause with its
lexicalized counterpart (35b).

(35) Tojolabal (Ramírez del Prado 2017: 586)

a. oj wa-le' jan y-i' jun s-k'u' ja=k-ijts'n=i
irr A2-look.for dir A3-dat det A3-blouse det=A1-younger.sibling=top
‘Will you look for a blouse for my little sister when you come back?’
b. oj=b'i y-i' jan y-i' gancho ja=karla=i
irr=rep A3-carry dir A3-dat hair.clip det=Karla=top
‘They say, she’s going to bring hair-clips to Karla.’

3.5.2 Cholan-Tseltalan

The Cholan-Tseltalan subbranch consists of six languages: Ch'ol, Chontal, Cholti, Ch'orti',
Tseltal, and Tsotsil (Campbell 2017: 44). I will discuss Chontal separately in Section 3.5.3,
as I have much more data on the applicative for this language.
Tseltal—as well as Tsotsil—uses a be morpheme to derive an AC from a BC; i.e.
a canonical applicative construction (36a). This morpheme is used only on transitive
stems and licenses the expression of a recipient, beneficiary, or external possessor. The
applicative object is available for passivization (36b). This applicativization is obliga-
tory; i.e. there is not an alternate non-applicative means to express this argument.

(36) Tseltal (Polian 2017: 627)

a. la j-pas-b-at 'ul
pfv A1-make-appl-B2 atole
‘I made atole for you.’
b. ∅-pas-b-ot-at 'ul
pfv-make-appl-pass-B2 atole
‘You were made atole.’
662 Brad Montgomery-Anderson

Tsotsil uses the suffix -be on transitive verbs to license the semantic role of recipient
(37); this suffix also licenses an addressee, beneficiary, maleficiary, and instrument
(Mora-Marín 2003: 211).

(37) Tsotsil (Mora-Marín 2003: 211)

'a li Xun=e ba y-ak'-be chitom li 'ants=e
top det Xun=encl go A3-give-appl pig det woman=encl
‘Xun went to give the pig to the woman.’

The other extant member of the Cholan family is Ch'orti', a language spoken in Eastern
Guatemala. This language has lost the applicative morpheme and replaced it with a
prepositional phrase strategy (38).

(38) Ch'orti'
a. in-man-i-∅ e b'ujk twa' ni-maxtak
A1-buy-sf-B3 det clothes for A1-children
‘I bought clothes for my children.’ (Pérez Martínez 1994: 48)
b. inw-ajk'-u-∅ ingojr muy a-tu'
A1-give-sf-B3 medlar.fruit to A2-mother
‘I gave medlar fruit to your mother.’(Pérez Martínez 1994: 142)4

In Ch'ol the applicative licenses the expression of the same semantic roles as those
found for Tsotsil; i.e. beneficiary, maleficiary, addressee, recipient, and goal (Vázquez
Álvarez 2002: 287). We can find pairs of sentences exemplifying a canonical applicative
construction in this language (39).

(39) Ch'ol (Vázquez Álvarez 2002: 53)

a. mi k-p'is-b-eñ-∅ waj x-ixik
ipfv A1-weigh-appl-inc-B3 tortilla cl-woman
‘I weigh tortillas for the woman.’ AC
b. mi k-p'is-∅ waj cha'añ x-ixik
ipfv A1-weigh-B3 tortilla prep cl-woman
‘I weigh tortillas for the woman.’ BC

4 To segment and gloss the Ch'orti' examples I use the conventions found in Dugan (2013).
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 663

3.5.3 Chontal Mayan

Because my own field work has focused on Chontal—a language of the Cholan-Tseltalan
subbranch of Western Mayan with a productive applicative construction—I will devote
more space to an exposition of the applicative in this language.
Chontal forms the applicative by using the be morpheme that we have already seen
with several other Mayan languages. It appears only on transitive verbs. As described at
the beginning of this chapter, Chontal root and derived verbs take different aspect and
voice marking. For the applicative suffix there is no distinction, and the resulting con-
struction takes the aspect suffixes used for derived verbs; for example, the transitive
root verb ts'ä' ‘to light’ (40a) and the transitive derived verb chän ‘to look at’ (40b) both
form the applicative by adding be.

(40) Chontal
a. kä-ts'ä'-bé-n-∅ ni beladóra ni kä-páp-la
A1-light-appl-inc-B3 det candle det A1-father-pl.incl
‘I light the candle to Our Lord.’
b. n-ix-ts'ak u-chäm-bé-n-∅ u-ni'ok che'án xelkóm
det-fem-medicine A3-look.at-appl-inc-B3 A3-toe if.be uneven
‘The healer looks to see if the toes are uneven.’

The be morpheme attaches to the CVC transitive verb root; it is preceded by causa-
tive, versive, assumptive and/or depositive suffixes if those suffixes are present.5 A
verb with both a causative and an applicative is in (41a). (41b) shows the applica-
tive preceded by a versive—a suffix that transforms an adjective into an intransitive
verb—and a causative. In (41c) the applicative is preceded by an assumptive—a suffix
that transforms a positional root into an intransitive verb—and a causative. In (41d)
the applicative follows a depositive, a suffix that transitivizes an intransitive posi-
tional verb.

5 The terms ‘assumptive’ and ‘depositive’ are terms that Terrence Kaufman suggested to me specifically
for Chontal; I do not know of their use for any other language. The assumptive suffix transforms a po-
sitional root into an intransitive verb (e.g. ‘to assume X position’), and a depositive suffix transforms an
intransitive positional verb into a transitive verb (e.g. ‘to put something in X position’).
664 Brad Montgomery-Anderson

(41) Chontal
a. ni kóya' u-chäk-es-bé-n-∅ u-k'a'
det tomato A3-red-caus-appl-inc-B3 A3-soup
‘The tomato makes the soup red.’
b. dáli a-mux-m-es-bé-n-on tan ni semét
asp A2-toasted-vrs-caus-appl-inc-B1 in det comal
‘You are going to toast it for me in the comal.’
c. jink'ín u-tsäms-í-∅-jo' jol-wän-es-b-ínt-i-∅ y-ak'
when A3-kill-compl-B3-pl pull-asm-caus-appl-pass-compl-B3 A3-tongue
‘When they kill it, its tongue is pulled out.’
d. p'ul-jats'-bé-n-∅ u-näk' t'ok k'élän k'ux
puffed.up-dps-appl-inc-B3 A3-belly with much food
‘Puff up his belly with a lot of food.’

The applicative morpheme has the allomorph b before a vowel; for example, before a
Set B suffix (42a) and before a passive suffix (42b).

(42) Chontal
a. k-äl-b-∅-ét ke' máchin ni yíchu'
A1-say-appl-compl-B2 that neg det dog
‘I told you that it’s not the dog.’
b. chin-wän-∅-ón tä úk'-e ká xuch'-b-ínt-∅-on
sit-asm-compl-B1 prep cry-inc because steal-appl-pass-compl-B1
kä-tak'ín
A1-money
‘I sat down and cried because my money was stolen.’

Chontal has a canonical applicative; i.e. the applicative construction (43a) has a corre-
sponding non-applicative base construction (43b) that codes the applied phrase differ-
ently from that of the AC. Because the possessor of the direct object in the BC (43b) is
interpreted as a third argument beneficiary, we could translate it as “My sister went to
Nacajuca to buy a shirt for me.” This is the adnominal strategy described previously for
Kaqchikel and Itzaj.

(43) Chontal
a. ni kä-chich x-i-∅ tä yäxtúp u-män-bé-n-on kä-búk
det A1-sister go-compl-B3 to Nacajuca A3-buy-appl-inc-B1 A1-shirt
‘My sister went to Nacajuca to buy a shirt for me.’ AC
b. ni kä-chich x-i-∅ tä yäxtúp u-män-é'-∅ kä-búk
det A1-sister go-compl-B3 to Nacajuca A3-buy-inc-B3 A1-shirt
‘My sister went to Nacajuca to buy my shirt.’ BC
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 665

The verb ‘to follow’ is used with the possessed body part ‘back’ and can be used with the
applicative (44a) or without it (44b).

(44) Chontal
a. ni äjín u-tsäk'-bé-n-∅ u-pát ni yiník-o'
det alligator A3-follow-appl-inc-B3 A3-back det person-pl
‘The alligator follows the people.
b. u-ts'äk-é'-∅ kä-pát
A3-follow-inc-B3 A1-back
‘It follows me.’

The applicative can license the expression a third argument that is a beneficiary (45a),
maleficiary (45b), recipient (45c), or addressee (45d).

(45) Chontal
a. u-jäk-sä-b-∅-ón u-cho'án
A3-lower-caus-appl-compl-B1 A3-price
‘They lowered the price for me.’
b. k-uch'-b-í-∅ u-buk'á kä-lot
A1-drink-appl-compl-B3 A3-pozol A1-friend
‘I drank my friend’s pozol.’
c. kä-x-é k-ä'-bé-n-et a-tak'ín
A1-go-inc A1-give-appl-inc-B2 A2-money
‘I’m going to give you money.’
d. aw-äl-bé-n-∅ adyós
A2-say-appl-inc-B3 goodbye
‘You say goodbye to him.’

The semantic roles of target (46a) and instrument (46b) are not expressed by means of
an applicative strategy; instead, these semantic roles are placed in an oblique phrase.
(NB: the applicativized positional root in (46a) is indexing a beneficiary or maleficiary.)
The semantic roles of accompaniment (46c) and location (46d) likewise are not indexed
on the verb. Note that in both these examples the verb is the transitive verb chen ‘to do,
make’; this verb is used as a light verb, a common construction discussed later in this
section. In Chontal, the applicative is never formed on an intransitive base.

(46) Chontal
a. pok'-jats'-bé-n-∅ u-nok' tä ka'
thrown[wet]-dps-appl-inc-B3 A3-clothing on ground
‘Throw his wet clothes on the ground.’
666 Brad Montgomery-Anderson

b. ajchojó' u-low-é'-∅ te' t'ok y-ej

woodpecker A3-make.hole-inc-B3 branch with A3-beak
‘The woodpecker makes a hole in the branch with its beak.’
c. kä-che-n-∅ chukbá t'ok ixík
A1-do-inc-B3 lovemaking with woman
‘I make love with the woman.’
d. n-aj-tiburón u-che-n-∅ bidajlé tan nap'
det-masc-shark A3-do-inc-B3 living in sea
‘A shark lives in the sea.’

When the be morpheme is present, the Set B suffix indexes the applied argument. The
direct object of the BC is no longer indexed with an affix on the verb. The applied argu-
ment is available for passivization; the applicative verb uses the same passive suffix
used for derived verbs (47). Since Chontal has split-ergative morphology, the passive
applied argument can be indexed by a Set A prefix (47a) or a Set B prefix (47b).

(47) Chontal
a. a-top'-jäts'-b-ínt-∅-on kä-choj
asp-sound.with.mouth-dps-appl-pass-compl-B1 A1-cheek
‘I was slapped in the cheek.’
b. ané aw-äl-b-ínt-e aj-täl-täl-nál-et
2pro A2-say-appl-pass-inc masc-touch-touch-and-B2
‘You are called “Mr. Touchy”.’

The applied object can also be questioned (48a) and relativized (48b).

(48) Chontal
a. ixkuné aw-ä'-b-í-∅ tak'ín
who A2-give-appl-compl-B3 money
‘Who did you give the money to?’
b. ni chéwa jin ni
det dough dem det
buk'á ke' mach uy-ä'-b-ínt-i-∅ ni käkáw
pozol that neg A3-give-appl-pass-compl-B3 det cacao
‘It’s the white pozol that you don’t put the cacao in.’

In Chontal a transitive verb has two strategies to express a third argument. The first is
the AC with the -be suffix (49a); the second is a non-applicative adnominal construction.
This second strategy is used when a possessed noun is present. The possessed noun
itself is indexed on the verb with a third person Set B suffix; the possessor of the noun
denotes the third argument and is not cross-referenced on the verb. In (49) both objects
are third person singular.
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 667

(49) Chontal
a. yech-bé-n-∅ n-u-be'tá ni kóko
take-appl-inc-B3 det-A3-meat det coconut
‘Take the meat out of the coconut.’ (AC)
b. yech-é-∅ n-u-be'tá ni kóko
take-inc-B3 det-A3-meat det coconut
‘Take the meat out of the coconut.’ (BC)

In (50) one of the arguments is a local person. In the applicative construction (50a) it is
indexed by the Set B suffix, while in the non-applicative (50b) it is not expressed on the
verb, but as the possessor of the noun.

(50) Chontal
a. kä-tu'-bé-n-et aw-ok
A1-spit-appl-inc-B2 A2-foot
‘I spit on your foot.’ (AC)
b. kä-túbä-n-∅ aw-ok
A1-spit-inc-B3 A2-foot
‘I spit on your foot.’ (BC)

The applicative is not used when the possessor and the subject are coreferential. Both
non-applicative verbs in the sentence (51) illustrate this.

(51) Chontal
jindá untu ixík ya'án u-jok'-é'-∅ u-lot
dem one woman be A1-call-inc-B3 A3-friend
ke' ya'án nat uné uy-äk'-é'-∅ u-k'ä' u-ti'
that be far 3pro A1-put-inc-B3 A3-hand A3-mouth
‘This is a woman who is calling her companion who is far away, she puts her hand
to her mouth.’

With the applicative construction the demoted object is no longer indexed on the verb
with a Set B prefix; it is, however, available for relativization (52a), questioning (52b)
and focus (52c). The lengthier fourth example (52d) shows that the demoted object ‘bel-
ladonna’ is available for passivization; it is the topic of a description for curing angina.

(52) Chontal
a. kä-k'ajti'í-n-∅ ni primer ts'uts'óm ke' kä-b-∅-et
A1-remember-inc-B3 det first kiss that A1-give.appl-compl-B2
‘I remember the first kiss that I gave you.’
668 Brad Montgomery-Anderson

b. kuné a-kä-cher-b-∅-ét
what asp-A1-do-appl-compl-B2
‘What did I do to/for you?’
c. a-t'ox-bé-n-on ni tak'ín=da
A2-break.down-appl-inc-B1 det money=foc
‘You make change for me from this money.’
d. u-ts'äkäl-ká-n t'ok ix-beyadóna
A3-cure-pass-inc with fem-belladonna
u-num-s-ínt-e pan k'ak' k'a k'äs-chäj-má-k-∅
A3-pass-caus-pass-inc in fire so.that half-cooked-asm-subj-B3
yá'i u-täk'-b-ínt-e t-u-choj n-aj-yaj
then A3-put.in-appl-pass-inc prep-A3-cheek det-masc-sick
‘It is cured with belladonna. It is passed through the fire so that it cooks
halfway, then it is put in the cheek of the patient.’

The most common verb base for the applicative is äk' ‘to give.’ The be morpheme is
obligatory on this verb to express a recipient. The addition of be causes a predicta-
ble phonological change to the verb root, and the derived applicative form is ä' (53a).
Although this verb with the meaning ‘to give’ typically has a recipient and therefore an
applicative suffix, we do see non-applicative uses of it (53b).

(53) Chontal
a. n-aj-chon+ye'é namás un-tsul ye'é uy-ä'-b-∅-ón
det-masc-sell+meat only one-nc meat A3-give-appl-compl-B1
‘The meat seller gave me only a little piece of meat.’
b. ajbits' jin um-p'e te' ke' uy-äk'-é'-∅ u-jut ka' bujté'
cochite dem one-nc tree that A3-give-inc-B3 A3-fruit like quinquil
‘The cochite is a tree that gives fruit like the quinquil.’

This verb also commonly has the meaning ‘to put, place,’ and the applicative appears
in this usage to express a target semantic role.6 Since the Chontal applicative normally
does not index a target semantic role, I assume that the verb has extended the meaning
‘to give to someone’ to signify something like ‘to give to a location’. In (54) the applied
argument is clearly not the first-person beneficiary (who is coreferential with the
subject and therefore not available for applicative promotion), but rather the third-per-
son target; i.e. the place where the lotion is applied.

6 Polian (2013: 274–275) describes similar meanings for the verb ak’ in Tseltal; i.e. the verb can indicate
a change of owner (‘to give’) as well as change of location (‘to put, place’).
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 669

(54) Chontal
kä-x-é k-ä'-bé-n-∅ bapurú bajká u-k'ux-∅-ón apixk'ok'
A1-go-inc A1-put-appl-inc-B3 VapoRub where A3-sting-compl-B1 wasp
‘I’m going to put on VapoRub where the wasp stung me.’

The meaning of ‘to put, place’ also appears in non-applicative constructions with the
target semantic role in an oblique phrase (55). This alternation is unexpected and seems
to be particular to this verb.

(55) Chontal
a. uy-äk'-í-∅ n-aj-báyu tä ja' tuba u-k'ech-é'-∅ sits'ák
A3-put-compl-B3 det-masc-trap in water to A3-catch-inc-B3 turtle
‘He put the trap in the water to catch the turtle.’
b. ya'í=ba ka'án äk'-bitá a-x-é aw-äk'-é'-∅ a-xänäk'
there=foc where put-nmlz A2-go-inc A2-put-inc-B3 A2-shoe
‘There is the place where you leave your shoes.’

Other transitive verbs with a target sematic role do not use the applicative to index this
role (56).

(56) Chontal
kä-x-e kä-t'um-jul-é'-∅ ni tak'ín tä ja'
A1-go-inc A1-aff-throw-inc-B3 det money in water
‘I’m going to throw the money in the water.’

In (57) the pair of sentences express the same event with applicative morpheme (57a)
and without the applicative morpheme (57b); neither sentence, however, is an applica-
tive construction.

(57) Chontal
a. ni ixík uy-ä'-bé-n-∅ xápum tan ni nók'
det woman A3-put-appl-inc-B3 soap on det clothing
‘The woman puts soap on the clothing.’
b. ni ixík 'uy-äk'-é'-∅ xápum tan ni nók'
det woman A3-put-inc-B3 soap on det clothing
‘The woman puts soap on the clothing.’

The applicative form of the verb äk' also appears in more complex sentences with the
meaning ‘to cause someone to do something’ (58a) or ‘to allow someone to do some-
thing’ (58b). In the negative, this verb can mean ‘to prevent’ (58c).
670 Brad Montgomery-Anderson

(58) Chontal
a. uy-ä'-bé-n-∅ ak'ojná-k-∅
A3-give-appl-inc-B3 dance-subj-B3
‘He makes her dance.’
b. k-ä'-bé-n-∅ uy-uch'é-n-∅
A1-give-appl-inc-B3 A3-drink-inc-B3
‘I make him drink.’
c. n-ajlo' mach uy-ä'-b-i-∅ u-che-n-∅
det-boy neg A3-give-appl-compl-B3 A3-do-inc-B3
‘The boy prevented it (lit. that he/she/it do it).’

The verb can be used with this meaning without the applicative (59). It is unclear why
an AC is not used; the inanimate nature of the agent may be a factor.

(59) Chontal
uy-úts'u ni k'ux mu'-uy-äk'-é'-∅ kä-xéjä-n-∅
A3-smell det food asp-A3-cause-inc-B3 A1-vomit-inc-B3
‘The smell of the food makes me vomit.’

For some speakers the verb äk’ has a special shortened applicative base for the comple-
tive (60a) and imperative (60b); the base appears to consist entirely of the applicative
morpheme.

(60) Chontal
a. kä-b-∅-et ni yentax
A1-give.appl-compl-B2 det necklace
‘I gave you the necklace.’
b. bé-n-∅ um-p'é jek'óm de ixím n-ajló'
give.appl-inc-B3 one-nc pile of corn det-boy
‘Give a little bit of corn to this boy.’

The applicative form of äk' has become lexicalized for some meanings. For example, in
both sentences in (61) it is unclear what the applied object is since the target semantic
role is part of a relational phrase.

(61) Chontal
a. u-k'än-ká-n-∅ tuba b-int-ik-∅ tamá ni tsajel ja'
A3-have-pass-inc-B3 to put.appl-pass-subj-B3 in det sweet water
‘It is used to put in sweet water.’
b. k-äk'-bé-n-∅ tan kä-pa' bú'u
A1-put-appl-inc-B3 in A1-cl bean
‘I put it in my beans.’
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 671

A lexicalized applicative on the verb ‘to give’ is the usual way to talk about the weather
events of rain (62a) and wind (62b). With such examples I was unsure to gloss it with
the Set B suffix as there seems to be no third (or even second) argument for this verb.

(62) Chontal
a. ixkak'ín u-x-e tä xup-ó uy-ä'-bé-n-∅(?) ja'
when A3-go-inc prep stop-inc A3-give-appl-inc-B3(?) water
‘When is it going to stop raining?’
b. mu'-uy-ä'-bé-n-∅(?) noj re'í ik'
asp-A3-give-appl-inc-B3(?) big very wind
‘It’s very windy.’

Two common verbs form the applicative with an irregular base. The verb ch' ‘to take’
uses its historic CVC root ch'äm as the applicative base (63a). The verb chen ‘to do’ uses
the stem cher as a base. In (63c) verb base cher with the applicative is nominalized.

(63) Chontal
a. ni sántu-jo' u-ch'äm-bé-n-∅ u-ch'új-le ni ye'é
det saint-pl A3-take-appl-inc-B3 A3-holy-poss det meat
‘The images [of saints] absorb the spirit of the meat.’
b. aj-toch u-cher-b-í-∅ ts'ak ix-pet
masc-Antonio A3-do-appl-compl-B3 medicine fem-Petrona
‘Antonio made Petrona medicine.’
c. u-xoy-í-∅-jo' n-u-cher-b-int-e u-jobäle
A3-surround-compl-B3-pl det-A3-make-appl-pass-inc A3-party
‘They surrounded “she-for-whom-the party-is-made”.’

In Chontal chen is frequently used in light verb constructions (64a). A semantically

“light” verb (i.e., a verb not very specific in meaning) relies on an accompanying noun
to form a more specific meaning (compare the semantically specific English verb ‘to
shower’ with its light verb counterpart in the construction ‘to take a shower’). The object
of chen is a noun or a nominalized verb; a shortened form of the Spanish infinitive can
also act as the object (64b). Use of the light verb construction has probably increased
through contact with Spanish as it is a productive means of incorporating Spanish verbs
into the language (Montgomery-Anderson 2006).

(64) Chontal
a. tä ch'uj ya'án u-ché-n-∅-jo' ch'uj+t'án t'ok yéntax
in church be A3-do-inc-B3-pl holy+word with rosary
‘The rosary is being prayed in the church.’
672 Brad Montgomery-Anderson

b. mu-'u-ché-n-∅ u-ba prekupá n-ajlo' jíni

asp-A3-do-inc-B3 A3-refl worry det-boy dem
‘The boy is worrying.’ (cf. Sp. preocuparse ‘to worry’)

The light verb construction frequently uses an applicative; in these instances, the applica-
tive morpheme attaches to the base cher. In this construction the demoted object is a shorte-
ned Spanish infinitive (65a) or a non-specific noun (65b), i.e., a noun without a determiner.

(65) Chontal
a. si a-be-n-ón ts'itá ixím kä-cher-bé-n-et agradesé
if A2-give.appl-inc-B1 a.little corn A1-do-appl-inc-B2 thank
‘If you give me a little corn, I will thank you.’ (cf. Sp. agradecer ‘to thank’)
b. n-ajló' y-o kä-cher-bé-n-∅ t'an
det-boy A3-want A1-do-appl-inc-B3 word
‘The boy wants me to write to him.’

More work needs to be done on the Chontal applicative. The Chontal I encountered
in my own fieldwork was full of Spanish borrowings, particularly of the kind using
the light verb construction. Spanish is lacking in the Mayan type of complex verb mor-
phology that marks objects; instead, it employs prepositions. Modern Chontal may be
borrowing from a Spanish syntactic template by expanding its use of prepositions. The
example in (66a) has such a preposition marking the beneficiary ‘San Lázaro’ instead
of the expected applicative construction. In the longer example in (66b), the first light
verb construction with the Spanish infinitive does not have the applicative, while the
second such construction does. The use of the preposition in (66c) seems unnecessary.

(66) Chontal
a. kä-ché-n-∅ cherajbuk'á chäm-p'é k'in t-aj-láchu
A1-make-inc-B3 offering four-nc day prep-masc-Lázaro
‘I made an offering of four days to San Lázaro.’
b. jink'ín kä-ché-n-∅ t'an t'ok n-aj-yokt'án-o' tä makuspána
when A1-do-inc-B3 word with det-masc-Chontal-pl of Macuspana
nadamás kä-cher-bé-n-∅ entendé la mitá u-t'án-o'
only A1-do-appl-inc-B3 understand the half A3-word-pl
‘When I speak with the people of Macuspana, I only understand half their
words.’ (cf. Sp. entender ‘to understand’)
c. ni yíchu' mu-'u-käm-é'-on täk wok
det dog asp-A3-bite-inc-B1 prep.A1 leg
‘The dog is biting my leg.’

My own field work on Chontal was on one specific dialect around Nacajuca and Maz-
ateupa. There are examples in the literature from other dialects where the applicative
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 673

is used (or not) in ways that are different from what I have encountered. For example,
Osorio May (2005) describes a variety in Tecoluta in which a local-person argument that
I would expect to be applicativized is introduced with a preposition (67b). Note that
ta’a ‘your, yours’ is a contraction of the preposition with the Set A affix. The Chontal
speaker I worked with from Mazateupa thought this sentence sounded odd. The sen-
tence with the non-local person (67a) is the same applicative construction encountered
in the variety I studied.

(67) Chontal (Osorio May 2005: 26–27)

a. kä-män-bé-n-∅ nok' ix-ch'upim
A1-buy-appl-inc-B3 clothes fem-old.woman
‘I buy clothes for the old woman.’
b. kä-män-í-∅ buk t'a'a
A1-buy-compl-B3 clothes prep.A2
I bought clothes for you.’

It should be emphasized just how common the applicative is in Chontal. The following
relatively short description of a cure for an ear infection contains six such constructions
(68).

(68) Chontal
n-ajts'ak u-kol-bé-n-∅ un-ts'itá' ch'ú'ul ja' tan u-chikín
det-healer A3-put-appl-inc-B3 one-little holy water in A3-ear
‘The healer puts a little holy water in the ear,
ya ke' tikw-∅-í=ba
now that heat-B3-compl=foc
after it has heated up,
u-chel-bé-n-∅ u-pam k'a pas-ík-∅ n-u-pujú
A3-tilt-appl-inc-B3 A3-head so.that leave-subj-B3 det-A3-pus
he tilts the head so that the pus comes out,
yá'i t'ok pits' u-suk-bé-n-∅
then with cotton A3-dry-appl-inc-B3
then he dries it with cotton
yá'i u-ju'-bé-n-∅ yok k'uts
then A3-put.in-appl-inc-B3 dim tobacco
then he puts in it tobacco
k'o-pa'-sä-bé-n-∅ ni ik'
so.that.A3-leave-caus-appl-inc-B3 det air
so that it takes out the air,
u-pa'-sä-bé-n-∅ u-péte puju
A3-leave-caus-appl-inc-B3 A3-all pus
in order to take out all the pus.’
674 Brad Montgomery-Anderson

4 Conclusion
In this overview, we have seen that applicative constructions and applicative lookalikes
are found across the Mayan language family, with three of its four branches showing
evidence of these constructions. It must be emphasized that in most cases the amount
of applicative data we have on any given Mayan language is quite small; this construc-
tion has not been methodically and comprehensively investigated in most of the Mayan
languages. For future investigations, it will be especially important to test the known
applicative structures to see if they are obligatory or optional. This overview should
thus be taken as a starting point for research rather than as a conclusive and compre-
hensive study. According to the data that is available, Mayan applicative constructions
can be characterized as follows:

Morphology
– There are four branches in the Mayan family. The morpheme be appears in the
Eastern branch and the Western branch; these two groupings contain the largest
number of languages as well as speakers.
– The Yucatecan branch is the only branch that has no evidence of an applicative
morpheme.
– The fourth and most divergent branch of the Mayan languages—Huastecan—has
developed two distinct applicative morphemes etymologically unrelated to the
applicative morpheme in the other branches.
– This paper focuses on data from Chontal Mayan, a language in the Western branch.
In Chontal the be morpheme has the allomorph b before vowels. This morpheme
can be used with other valency-changing suffixes (passive, causative, and deposi-
tive).

Syntax
– In the Eastern Mayan branch be appears in both true applicative constructions as
well as in lookalike oblique registration constructions. Kaqchikel, for example, uses
the b'e to focus an instrument in an oblique registration construction.
– In the Mamean subbranch of Eastern Mayan, Ixil is the only language with an
applicative-like construction. In this language the AC advances an instrument that
in the BC is in an oblique phrase with a relational noun. The applicative morpheme
focuses the instrument, but it is not marked as a core argument.
– In the Western branch we have the most evidence of be used in true applicative
constructions, i.e., the AC has a corresponding BC and the AP is a core argument.
Cholan-Tseltalan languages use the be morpheme extensively.
– Although the Yucatecan branch is the only branch that has no evidence of an
applicative morpheme, there is evidence that Itzaj uses an adnominal strategy to
index a third argument.
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 675

– In at least one variety of Huastecan, the applicative construction is not obligatory.

Demotion of the BC object is affected by the definiteness of the object.
– Except for Southeastern Huastec, the applicative construction seems to be only
formed on transitive verbs, not intransitive verbs. In Southeastern Huastec, the
instrumental applicative is found only on intransitive and detransitivized verbs.

Semantics
– In all the Mayan languages except Huastec, the applicative morpheme is semanti-
cally underspecified; that is, the morpheme does not vary based on the semantic
role it indexes.
– The Huastecan branch has applicative morphemes distinct from the other Mayan
languages; these morphemes license recipients, beneficiaries/maleficiaries, and
instruments.
– In the Western branch, Cholan-Tseltalan languages use be to index a recipient,
addressee, beneficiary, maleficiary, and instrument.
– We have data showing that Kaqchikel uses the b'e applicative morpheme to index
comitatives, locatives, datives, and themes / topics of speech. These examples need
to be tested for corresponding BCs to show that these instances of b'e represent true
applicative constructions.
– In Chontal, lexicalized applicatives are used for some high frequency utterances
such as weather events.
– In one variety of Huastecan there are two distinct applicative morphemes: one for
beneficiaries and recipients and the other for instruments.
– For many Mayan languages further research needs to be done to determine if
applicative constructions are obligatory; if they are not obligatory, the pragmatic
and discursive implications of the applicative construction need further explanation.

Abbreviations
A set A affix
AC applicative construction
aff affect root
and andative
appl applicative
asm assumptive
asp aspect
antip antipassive
B set
B affix
BC base (i.e. non-applicative) construction
cl classifier
676 Brad Montgomery-Anderson

caus causative
compl completive
dat dative
dat.appl dative applicative
dem demonstrative
det determiner
dim diminutive
dir directional
dis dependent intransitive status
dps depositive
dts dependent transitive status
ep epenthesis
encl enclitic
fem feminine
foc focus
fut future
gen genitive
hab habitual
hum human
ipfv imperfective
inc incompletive
incl inclusive
ins instrumental
irr irrealis
masc masculine
mid middle voice
nc numeral classifier
neg negative
nm nominal modifier
nmlz Nominalization
pass passive
pfs phrase-final suffix
pfv perfective
pl plural
ppm proximal patient marker
pro pronoun
poss possessive
prep preposition
ptcp participle
refl reflexive
rel relative
rep reportative
rn relational noun
sf stem formative
sg singular
subj subjunctive
top topic
tr transitive
trc transitive completive
 19 Applicative constructions in Mayan languages: An overview with special focus on Chontal 677

- default affix boundary

= clitic boundary
+ compounding boundary

References
Brown, R. McKenna, Judith M. Maxwell & Walter E. Little. 2006. La Ütz Awäch?: Introduction to Kaqchikel Maya
language. Austin: University of Texas Press.
Campbell, Lyle. 2000.Valency-changing derivations in K'iche'. In R.M.W. Dixon & Alexandra Y. Aikhenvald
(eds.), Changing valency: Case studies in transitivity, 237–281. Cambridge: Cambridge University Press.
Campbell, Lyle. 2017. History and reconstruction of the Mayan languages. In Judith Aissen, Nora C. England
& Roberto Zavala Maldonado (eds.), The Mayan languages, 43–61. London: Routledge.
Dayley, Jon. 1981. Voice and ergativity in Mayan languages. Journal of Mayan Linguistics 2. 6–82.
Dixon, R.M.W. 2012. Basic Linguistic Theory, Volume 3: Further grammatical topics. Oxford: Oxford University
Press.
Dryer, Matthew S. 1986. Primary objects, secondary objects, and antidative. Language 62(4). 808–845.
Dugan, James Timothy. 2013. The grammar of Ch'orti' Maya folktales. New Orleans: Tulane University
dissertation.
England, Nora C. 1983. A grammar of Mam, a Mayan language. Austin: University of Texas Press.
England, Nora C. 2001. Introducción a la gramática de los idiomas mayas. Guatemala: Cholsamaj.
England, Nora C. 2017. Mam. In Judith Aissen, Nora C. England & Roberto Zavala Maldonado (eds.), The
Mayan languages, 500–532. London: Routledge.
García Matzar, Pedro & José Obispo Rodríguez Guaján. 1997. Rukemik ri Kaqchikel chi’: Gramática Kaqchikel.
Guatemala: Cholsamaj.
García Matzar, Pedro Oscar, Valerio Toj Cotzaja & Domingo Coc Tuiz. 1992. Gramática del idioma Kaqchikel.
Guatemala: Proyecto Lingüístico Francisco Marroquín.
Henderson, Robert. 2007. Observations on the syntax of adjunct extraction in Kaqchikel. Proceedings of the
Conference on Indigenous Languages of Latin America 3. 1–21.
Hofling, Charles Andrew. 1990. Possession and ergativity in Itzá Maya. International Journal of American
Linguistics 56(4). 542–560.
Hofling, Charles Andrew. 2000. Itzaj Maya grammar. Salt Lake City: University of Utah Press.
Keller, Kathryn & Plácido Gerónimo Luciano.1997. Diccionario Chontal de Tabasco. Tucson: Summer Institute
of Linguistics.
Knowles, Susan. 1984. A descriptive grammar of Chontal Maya (San Carlos dialect). New Orleans: Tulane
University dissertation.
Kondić, Snježana. 2012. A grammar of South Eastern Huastec, a Mayan language from Mexico. Sydney:
University of Sydney dissertation.
Law, Danny. 2014. Language contact, inherited similarity and social difference: The story of linguistic interaction
in the Maya Lowlands. Amsterdam: John Benjamins.
Macario, Narciso Cojti, Martin Chacach Cutzal & Marcos Armando Cali. 1998. Diccionario Kaqchikel.
Guatemala: Proyecto Lingüístico Francisco Marroquín.
Montgomery-Anderson, Brad. 2006. Light verbs and split ergativity in the Cholan languages. Kansas
Working Papers in Linguistics 28. 11–32.
Montgomery-Anderson, Brad. 2010. Alternatives to the applicative in Chontal Mayan. Southwest Journal of
Linguistics 29(1). 75–110.
678 Brad Montgomery-Anderson

Montgomery-Anderson, Brad. 2015. Diccionario de la lengua Yokot’an (Maya Chontal) de Tabasco: Yokot’an-Cas-
tellano. Unpublished manuscript.
Mora-Marín, David F. 2003. Historical reconstruction of Mayan applicative and antidative constructions.
International Journal of American Linguistics 69(2). 186–228.
Mora-Marín, David F. 2009. Reconstruction of the Proto-Ch’olan demonstrative pronouns, deictic enclitics,
and definite articles. Transactions of the Philological Society 107(1). 98–129.
Ochoa Peralta, María Angela. 1984. El idioma Huasteco de Xiloxúchil, Veracruz. Mexico City: Instituto Nacional
de Antropología e Historia.
Osorio May, José del Carmen. 2005. Análisis de la morfología del Yokot’an, “Chontal” del poblado de Tecoluta,
Nacajuca, Tabasco. Mexico City: CIESAS MA thesis.
Pérez Martínez, Vitalino. 1994. Gramática del idioma Ch'orti'. Guatemala: Proyecto Lingüístico Francisco
Marroquín.
Peterson, David. 2007. Applicative constructions. Oxford: Oxford University Press.
Polian, Gilles. 2012. Grámatica del tseltal de Oxchuc. Mexico City: Universidad Nacional Autónoma de México.
Polian, Gilles. 2017. Tseltal and Tstotsil. In Judith Aissen, Nora C. England & Roberto Zavala Maldonado
(eds.), The Mayan languages, 610–647. London: Routledge.
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Ramírez del Prado, Alejandro Curiel. 2017. Tojoloabal. In Judith Aissen, Nora C. England & Roberto Zavala
Maldonado (eds.), The Mayan languages, 570–609. London: Routledge.
Smith-Stark, Thomas. 1994. Instrumental voice in Jilotepequeño Pocomam. Función 15–16. 231–260.
Stout, Tammi. 2015. The proleptic possessive construction in Kaqchikel Maya. In Natalie Weber & Sihwei
Chen (eds.), Proceedings of the Workshop on Structure and Constituency in the Languages of the Americas
18 & 19 (University of British Columbia Working Papers in Linguistics 39), 67–75.
Vázquez Álvarez, Juan J. 2002. La morfología del verbo de la lengua Chol de Tila, Chiapas. Mexico City: CIESAS
MA thesis.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Néstor Hernández-Green and Oscar López Nicolás
20 Applicative constructions in two
Otomanguean families: Otomi and Zapotec
Abstract: This chapter describes several applicative constructions (ACs) in two Oto-
manguean language families, Otomi (Otopamean, Western Otomanguean) and Zapotec
(Zapotecan, Eastern Otomanguean). As both families are highly diverse internally, a
wide range of morphological and syntactic phenomena can be observed in their ACs.
With respect to morphology, Zapotec languages tend to mark ACs via concatenative mor-
phology, whereas Otomi makes use of cumulative exponence, stem alternations and,
arguably, lexical alternations. With respect to syntax, we observe a close correlation
between (non-)promotional constructions and extraction (~ wh-movement). Further-
more, applicative morphology is also found in constructions that display a promotion/
non-promotion continuum, both within a single language (e.g., Northern Zapotec) and
among different language varieties (e.g., Otomi languages).

1 Introduction
Otomanguean (Mexico) is the most diverse language stock in the Meso-American lin-
guistic area (Campbell, Kaufman, and Smith-Stark 1986), containing 178 languages
according to Ethnologue,1 181 according to Glottolog,2 and 220 according to Mexico’s
National Institute of Indigenous Languages (INALI 2009). These language families
(listed in Figure 1) are nowadays spoken in the states of San Luis Potosí, Guanajuato,
Michoacán, Querétaro, Hidalgo, Veracruz, Puebla, Tlaxcala, Oaxaca, Guerrero, and
Mexico, by approximately 2,148,000 people;3 some extinct Otomanguean languages
(marked with “†” in Figure 1) were once spoken in Southern Mexico and in Costa Rica
(Kaufman 1994).
Otomanguean languages are tonal (2 to 11 tonemes), and tend to have CV-type syl-
lable structures, nasal vowels, and complex phonation types. They are fusional (some
more synthetic than others), their derivational morphology is not very productive,
and most (if not all) of them have inflectional classes. Otomanguean languages lack
non-finite verb forms, their word order is verb-initial, and they are all head-marking
(Baerman, Palancar, and Feist 2019: 3–4). The head-marking feature is illustrated in (1)
below. As can be seen in (1a), the possession relation in Otomi languages is indicated in

1 https://www.ethnologue.com/subgroups/otomanguean
2 https://glottolog.org/resource/languoid/id/otom1299
3 https://site.inali.gob.mx/Micrositios/estadistica_basica/estadisticas2015/estadisticas2015.html

https://doi.org/10.1515/9783110730951-020
680 Néstor Hernández-Green and Oscar López Nicolás

Mè’pàà (Tlapanec)
Mè’pàà-Subtiaba Subtiaba†
Tlapanec-Chorotegan Mangue (Chorotega)†
Chorotegan Chiapanec†
Otomi
Western Otomanguean Mazahua
Oto-Pamean Matlatzinca
Tlahuica
Oto-Pamean-Chinantec Chichimeco Jonaz
Pame
Chinantec Chinantec
Otomanguean
Trique
Mixtecan Cuicatec
Amuzgo-Mixtecan Mixtec
Amuzgo Amuzgo
Eastern Otomanguean Chatino
Zapotecan Zapotec
Popolocan-Zapotecan Ngigua (Popoloca)
Popolocan Ngigua (Chocho)
Ixcatec
Mazatec

Figure 1: The Otomanguean stock (Campbell 2017).4

the possessum noun phrase (kár thühü ‘his name’) rather than in the possessor noun
phrase (kar ánima ‘the dead person’). As for the verb, person of the subject is indi-
cated in the preverbal inflectional formative,5 as in (1b), while object is encoded via an
enclitic or a suffix, as in (1c) and (1d), respectively.

(1) San Felipe Otomi

a. da de̠ n ka=ŕ thühü ka=r ánima
irr ss\follow det.sg.nv=sg.3psr name det.sg.nv=sg dead.person
‘So as the dead father’s bloodline continues.’
(LBI 2010: Matthew 22:24)
b. xkí ma libre, ꞌbu̠
2.imm ss\go free then
‘You’d be free to go then.’
(LBI 2010: John 19:10)

4 Please note that the branches in the diagram are Campbell’s (2017) interpretation of one of Kaufman’s
non-peer reviewed manuscripts. We thank an anonymous reviewer for pointing this out to us.
5 Inflectional formatives are clitics, but they are written as separate words in the original source from
which these examples are taken.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 681

c. gege=ká i ꞌo̠ꞌti=bi ka i ꞌa̠di

3=3sg.nv ipfv make.appl:ben=3io 3sg.nv ipfv ask.for
‘To him it will be done as he asks for.’
(LBI 2010: James 5:16)
d. bi zah-pi kí ngo̠ ku̠ ánima
pfv ss\bite.appl:ben-3io pl.nv3psr meat det.pl.nv dead.person
‘They (i.e. the birds) ate the corpses’ flesh.’
(LBI 2010: Revelation 19:21)

Obliques and other adjuncts are expressed either in PPs, as illustrated in (2a), in PPs
headed by so-called relational nouns, as in (2b), with adverbs and bare nouns serving as
locative adverbials, as shown in (2c), or via serial-verb constructions, as in (2d).6

(2) a. Acazulco Otomi

ra=di=pe̠ ꞌ=[ꞌa Márquesa]PP kꞌa=ḿ mini=ga
icp=cl=work=loc La.Marquesa det.sg.nv=1psr comadre=1
‘My comadre works in La Marquesa.’
(Hernández-Green 2015: 48)
b. Quiegolani Zapotec
w-nii men disa [lo noo]PP
compl-speak 3 language rn:face 1excl
‘She spoke Zapotec to me.’
(Black 2000: 46)
c. Chalcatongo Mixtec
ni-xã́ʔã́=Ø nundùa íkú
compl-go2=3 Oaxaca yesterday
‘She went to Oaxaca yesterday (and has returned).’
(Macaulay 1996: 170)
d. Teotitlán Zapotec
ru-ta’w=an gæt ri-kā’ Ánn
hab-sell=3sg.infor tortilla hab-get Ana
‘S/he sells tortillas to Ana.’
(Gutiérrez 2021: 147)

This chapter centers on applicative constructions in two language families of the Oto-
manguean stock (in bold in Figure 1 above): Otomi (Otopamean, Western Otomanguean)
and Zapotec (Zapotecan, Eastern Otomanguean). Applicative constructions and their
syntax have been described in these two families, and their morphological and seman-
tic features are representative of the internal diversity of Otomanguean.

6 Glosses from some secondary sources have been modified.

682 Néstor Hernández-Green and Oscar López Nicolás

As is stated in the position paper (Zúñiga and Creissels, this volume), descriptions of
applicative constructions relate a base construction (BC) to an applicative construction
(AC). These constructions have the following characteristics:

(3) a. The predicates in both the BC and the AC are built upon the same root, but
the one in the AC bears additional overt marking with respect to the one in
the BC.
b. The S/A participant in the BC remains S/A in the applicative construction.
c. The AC includes an applied phrase (AppP), which refers to a participant that
either requires a non-core coding different from its coding in the AC (e.g. in a
PP) or cannot be expressed at all in the BC.

These relationships between the BC and the AC are illustrated in (4). The verb root is
bɛ́rɛ́k ‘work’ in both (4a) and (4b), but in (4b) (i.e., the AC) it receives the applicative
suffix -ɛ́l, as per the characteristics in (3a) above. According to characteristic (3b), the 1st
person singular remains the subject in both constructions (prefix kɪ ̀-). Finally, as per (3c),
the applicative suffix -ɛ́l in the AC in (4b) allows the expression of the (beneficiary) AppP
kítsɔ́ ‘Kitso’, which, crucially, cannot be expressed in the BC in (4a).

(4) Tswana
a. kɪ ̀-tɬàà-bɛ́rɛ́k-á màítsíbʊ̀ːá.
sI:1sg-fut-work-fv evening(6)
‘I’ll work this evening.’
b. kɪ ̀-tɬàà-bɛ́rɛ́k-ɛ́l-à [kítsɔ́]AppP màítsíbʊ̀ːá.
sI:1sg-fut-work-appl-fv Kitso(1) evening(6)
‘I’ll work for Kitso this evening.’
(adapted from Creissels, this volume)

In this chapter, we attempt to describe the wide range of phenomenon types that can
be observed in the grammar and functions of ACs in Otomi and Zapotec languages. We
think that the internal diversity of these two families is very interesting from a cross-
linguistic perspective. On the one hand, with respect to morphology, Zapotec languages
tend to mark ACs via concatenative morphology, whereas Otomi makes use of cumula-
tive exponence, stems alternations, and possibly lexical alternations. On the other hand,
with respect to syntax, we observe a promotion/non-promotion continuum both within
a single language (Northern Zapotec) and among different language varieties (Otomi
languages).
In the following sections, we describe the morphological (Section 2), syntactic
(Section 3), and semantic (Section 4) features of ACs in Otomi and Zapotec languages. In
Section 5 we enlist a series of constructions that are like ACs either in their form or their
functions. Some final remarks are given in Section 6.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 683

2 Morphology
ACs in Otomi and Zapotec involve a wide range of morphological phenomena. These
are illustrated in the following sections separately, Otomi on one side, and Zapotec on
the other.

2.1 Otomi applicative morphology

Otomi ACs are typologically unusual in their morphology, as they make extensive use of
cumulative exponence and stem alternations. These two phenomena are presented in
the following two sections. In addition, contrasts in verb paradigms between BCs and
ACs are presented in Section 2.1.3.

2.1.1 Cumulative exponence

In Otomi-Mazahua, TAM and person morphosyntactic features are cumulatively

encoded in the inflectional formative (cf. Palancar 2011a), a clitic that precedes the verb
stem. Inflectional formatives, which are represented as prefixes by some authors (e.g.
Andrews 1993; Palancar 2011b), can also fuse an applicative-like category called regis-
tration in Hernández-Green (2016) and adverbial inflection in Palancar (2017). In this
chapter, we adopt the latter term to refer to this morphosyntactic category. Although
adverbial inflection is not an AC in modern Otomi languages, it seems to have been
one in Old Otomi (16th century).7 The goal applicative in Old Otomi was encoded in pre-
fixes that also encoded grammatical person in a cumulative way: ä- ‘1.appl:goal’ for 1st
person, and e- ‘2/3.appl:goal’ for other persons, as shown in (5).

(5) Old Otomi

a. ko=tä-n-ä-ë-kꞌï
cop=1-prs-1.appl:goal-come-2po
‘I come to you.’
b. ko=kä-n-e-yë-gï
cop=2-prs-2/3.appl:goal-ss\come-1po
‘you come to me’
c. ko=n-e-yëhë
cop=prs-2/3.appl:goal-ss\come
‘S/he comes to him/her.’
(Cárceres 1580/1907: 98)

7 In this chapter, we do not attempt to reconstruct or theorize about the historical development of ACs
in the Otomi family, but we merely present data from different variants and different diachronic stages.
684 Néstor Hernández-Green and Oscar López Nicolás

2.1.2 Stem/lexical alternations

Otomi languages (along with Mazahua, their sister language) have benefactive appli-
cative verb stems that can be traced back to a valency-increasing suffix ✶-H in Proto-
Otomi-Mazahua (cf. Bartholomew 1965: 100).8 Some Otomi transitive verbs alternate
with applicative stems characterized by a stem medial (or final, in monosyllabic stems)
glottal segment, as shown in (6) with data from Acazulco Otomi.

(6) Acazulco Otomi

a. óra=da ꞌyo̠ꞌtꞌi=bi=nu̠=ŕ cómida
now=irr ss\make.appl:ben=3io=det.sg.dist=sg.3psr food
‘now she’ll make food for them (i.e., the farmhands)’ {txt}
b. po̠=xo=bi t-he̠ hki=bi=kꞌa nzǎfi=a=ꞌna
so=so=pfv imprs-release.appl:ben=3io=det.sg.nv firearm=encl=quot
‘so they say that someone discharged the firearm on him’ {txt}
c. ¿tébe̠ =di jaꞌ-kꞌi=a=nu̠?
what=irr do.appl:ben-2po=encl=3sg.dist
‘What is she going to do for you?’ {txt}

Other transitive verbs in Acazulco Otomi alternate with applicative stems that, although
traceable back to ✶-H, cannot be described simply as a glottal segment. To illustrate this,
consider the verb pairs in (7).

(7) Acazulco Otomi

base applicative
a. höí höꞌ́ mbi ‘take out’
b. hòni hòndi ‘look for’
pèni pèndi ‘wash (clothes)’
c. kú̠hu̠ kú̠nti ‘bring’

Stem alternations in Otomi also involve tone alternations. The verb stem pǎ̠ ‘sell’ bears
rising tone in the BC in (8a), but low tone pa̠h̀ ‘sell.appl:ben’ in the AC in (8b).9

8 This suffix’s reflexes also yield causative verbs in the modern Otomi and Mazahua languages, but
these are dealt with in Section 5.1.
9 Another case of non-productive, applicative morphology involving stem alternation is found in So-
chiápam Chinantec. The monotransitive stem cué³² ‘give’ alternates with the ditransitive stem cuéh³²
‘give’, as can be seen by comparing (ia) to (ib). The recipient is expressed as a PP in the former, but as a
NP in the latter.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 685

(8) Acazulco Otomi

a. ra=pǎ̠
ipfv=sell
‘S/he sells it.’
b. ra=pa̠h̀ -pi
ipfv=sell.appl:ben-3io
‘S/he sells it to him/her.’

Unlike the Zapotec applicatives presented in Section 2.2.1, below, the Otomi benefactive
applicative is no longer productive nowadays. The form contrasts between BCs and ACs
illustrated in this section are considered as stem alternations in this chapter, as they all
can be traced back to the suffix ✶-H, and all base/applicative pairs share the same root
(i.e., the first syllable of the stem). However, such alternations have features that resem-
ble lexical alternations comparable to intransitive/causative pairs like die/kill (provided
that the fact is ignored that this pair does not share root): a) not all ✶-H reflexes result
in applicative stems, as in the data in (9a); b) verbs with ✶-H reflex may not have a
counterpart without ✶-H, as in (9b); and c) not all applicative alternations involve ✶-H
reflexes, as shown in (9c).

(9) a. ꞌuayi ‘break (intr)’ ꞌuahki ‘break (tr)’ (causative)

ju̠tsꞌi ‘pull up’ ju̠ꞌtsꞌi ‘pull up (forcibly) (intensive)
b. – pě̠ ꞌtsꞌi ‘put away’
– ndo̠ꞌtsꞌe ‘look at (uphill)’
c. paꞌtꞌi(=bi) ‘heat up (for sb.)’
xálte(=bi) ‘fry (for sb.)’

Given the facts shown in (9) above, the question remains whether the alternations
observed are to be described at the inflectional level (i.e., stem alternations) or at the
lexical level (i.e., lexeme pairs).

(i) a. cué³² tsú² quɨe³ [ñí¹con² jon²]R.PP tsa³háu²

give.tr.inan.fut.3 3 money to child.3 tomorrow
‘S/he will give money to her/his child tomorrow.’
b. cuéh³² tsú² [jon²]R.NP quɨe³ tsa³háu²
give.appl.inan.fut.3 3 child.3 money tomorrow
‘S/he will give her/his child money tomorrow.’
(Foris 1993: 370)
686 Néstor Hernández-Green and Oscar López Nicolás

2.1.3 Paradigm structure of applicative verbs

In Old Otomi, the goal applicative seems to have involved less TAM distinctions than
their BC counterparts. Consider the data in Table 1. On the one hand, transitive verbs
(verbo actiuo) in the tänä conjugation10 are inflected in four different tenses in the
indicative mood (indicatiuo modo), as well as in imperative (imperatiuo modo), optative
(optatiuo modo), and subjunctive moods (subjuntiuo modo; Cárceres 1580/1907: 73–78).
Constructions with the goal applicative, on the other hand, are attested in only three
tenses in the indicative (Cárceres 1580/1907: 98).

Table 1: Defectiveness of the goal applicative subparadigm

(compared to tänä transitives of Old Otomi).

Basic appl:goal
Indicative Present
Imperfect preterite
Perfect preterite
Future
Imperative
Optative
Subjunctive

2.2 Zapotec applicative morphology

Applicative morphology in Zapotec languages is rather agglutinative, in contrast with

the Otomi data presented in Section 2.1. These morphological characteristics are
described in Section 2.2.1. Section 2.2.2 shows that Zapotec ACs have verb paradigms
identical to those of BCs, unlike Old Otomi (see § 2.1.3).

2.2.1 Bound morphology

All ACs reported in Zapotec languages involve distinct bound morphology in the verb,
either clitics or affixes. The exponent of the recipient/benefactive applicative in North-
ern Zapotec is the enclitic =d ‘appl:ben’ illustrated in (10).11

10 One of the two verb conjugations Cárceres (1580/1907) describes for Old Otomi; the other conjuga-
tion is called tati. Both conjugations are named after their inflectional formative for 1st person present.
11 It is worth to notice that the applicative =d in Northern Zapotec may be found lexicalized as a stem
formative in some verbs. For example, the stem àkd ‘reckon’ is formed by the root àk ‘happen’ plus the
applicative =d, as illustrated below.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 687

(10) Northern Zapotec

sh-ghe-shàb=d=tòˀ=nhéˀ góˀn=nhàˀ
icp-go-offer=appl:ben=1pl:excl=3form.objt bull=def
‘We were going to offer them the bulls.’ {txt}

Applicative affixes are illustrated with data from Northern Zapotec in (11) and (12). The
comitative suffix -lhénh ‘appl:com’ in (11a) is cognate with the comitative relational
noun lhénh ‘prep:with’ in (11b).12 The instrumental suffix -é ‘appl:instr’ in Northern
Zapotec must co-occur with the applicative =d in transitive stems, as is shown in (12).

(11) Northern Zapotec

a. sh-yêgh-lhénh=áˀ=nhéˀ. . .
irr-leave.for.origo-appl:com=1sg.nom=3form.objt
‘I will go back with her/him. . .’
b. sh-yêgh=áˀ lhénh bíˀ=nhàˀ
irr-leave.for.origo=1sg.nom prep:with clf.pro:infor=def
‘I will go back with them (her/him).’

(12) Northern Zapotec

w-dxíxé-é+d=éˀ=nh yìnhˀ yàˀà=nhàˀ
irr-measure-appl:instr[tr]=3form.nom=3inan chili green=def
‘She would weigh the green chili peppers with that (i.e., the scale).’

Some applicative affixes are traceable back to adpositions, and ultimately to content
words such as nouns or verbs. The instrumental suffix -lhénh of Northern Zapotec is an
incorporated form of the preposition lhénh ‘with’, which in turn derived from the verb
lhénh ‘join’ (López Nicolás 2016: 162, 349).

(ii) dx-àk+d=áˀ bèòˀ jûnìoh=nhàˀ

icp-reckon=1sg.nom month june=def
‘I reckon it was in the month of June.’ {txt}
12 Some authors analyze such cases as preposition incorporation. In Zenzontepec Chatino, for instance,
“[t]he applicative is coded on the verb by the incorporation of the relational noun lóʔō ‘with’, which
otherwise typically flags comitative and instrument NPs” (Campbell 2015: 1417).

(iii) a. tz-aa=na nakwę=ūʔ

pot-go=1pl.incl say=3pl
‘“Let’s go”, they said.’
b. tz-a+lóʔō nāáʔ kichi
pot-go+rn:with 1sg quern
‘I will take (lit. go with) a quern.’ (Campbell 2015: 1417)

Nevertheless, in this chapter we regard the Chatino incorporation as a case of affixation.

688 Néstor Hernández-Green and Oscar López Nicolás

2.2.2 Paradigm structure of applicative verbs

The paradigm structure of verbs in Zapotec ACs is the same as that of their correspond-
ing BCs. This is illustrated with the basic verb paradigm of Northern Zapotec in (13) and
(14), which consists of a completive form in the a-examples, an incompletive form in
the b-examples, and an irrealis form in the c-examples. This structure remains the same
with the applicative stems, as can be seen in Examples a’, b’ and c’.

(13) Northern Zapotec: shàb ‘offer’

a. b-shàb ‘compl-offer’ a’. b-shàb=d ‘compl-offer=appl:ben’
b. t-shàb ‘icp-offer’ b’. t-shàb=d ‘icp-offer=appl:ben’
c. w-shàb ‘irr-offer’ c’. w-shàb=d ‘irr-offer=appl:ben’

(14) Northern Zapotec: chòg ‘cut’

a. b-chòg ‘compl-cut’ a’. b-chòg-é+d ‘compl-cut-appl:instr[tr]’
b. sh-chòg ‘icp-cut’ b’. sh-chòg-é+d ‘icp-cut-appl:instr[tr]’
c. chòg ‘irr.cut’ c’. chòg-é+d ‘irr.cut-appl:instr[tr]’

2.3 Summary

The data presented in this section shows how morphologically diverse ACs can be. On
the one hand, Otomi applicative morphology is mostly non-concatenative (i.e., cumu-
lative exponence, stem alternations) or lexical (i.e., lexeme alternations). Old Otomi
ACs also seem to have presented a reduced inflectional paradigm compared to that of
BCs. Zapotec ACs, on the other hand, are canonically agglutinative, and their inflec-
tional paradigms are identical to those of BCs. Thus, it can be argued that Zapotec
and Otomi ACs cover many morphological types among those that can be expected
crosslinguistically.

3 Syntax
Some syntactic properties of AppPs and ACs in Otomi and Northern Zapotec are pre-
sented separately in Section 3.1 and Section 3.2, respectively. The syntactic status of
AppPs with respect of their corresponding expression in BCs is presented first, and then
the interactions of ACs with valency of the base verb and with other valency-chang-
ing constructions are presented second. Some Otomi-particular idiosyncrasies of the
expression of POs in ACs are shown in Section 3.1.3. Finally, in Section 3.2.3, dedicated to
Zapotec syntax, we discuss some constructions where the access of adjuncts to extrac-
tion operations is only possible if they are applied.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 689

3.1 Otomi applicative syntax

3.1.1 Syntactic status of the AppP

The Otomi benefactive is an obligatory AC, i.e., it is the only non-periphrastic strategy
the language has to express a recipient/beneficiary participant, and therefore such
participant cannot be expressed in the BC. The beneficiary AppP has the status of an
object, whose key morphosyntactic feature is the encoding via person suffixes in tran-
sitive verbs. Consider the monotransitive examples from Acazulco Otomi in (15a) and
(15b), where the 1st and 2nd person patient is encoded in the person suffixes; 3rd person
patients are seldom overtly encoded, as can be seen in (15c). The same verb root with
the applicative stem alternation in (15a’) and (15b’) receives the same person suffixes,13
which this time refer to the recipient; 3rd person recipient/beneficiaries are cross-refer-
enced by the suffix bi ‘3io’ in most cases.14

(15) Acazulco Otomi

monotransitive applied R/ben
a. bi=mbǎ̠ -gi a’. bi=mba̠h-ki
pfv=ss\sell-1po pfv=ss\sell.appl:ben-1po
‘s/he sold me’ ‘s/he sold it to me’
b. bi=mbǎ̠ -kꞌi b’. bi=mba̠ꞌ-kꞌi
pfv=ss\sell-2po pfv=ss\sell.appl:ben-2po
‘s/he sold you’ ‘s/he sold it to you’

c. bi=mbǎ̠ c’. bi=mba̠h-pi

pfv=ss\sell pfv=ss\sell.appl:ben-3io
‘s/he sold him/her/it’ ‘s/he sold it to him/her’

Other Otomian languages have similar constructions to those in (15a’-c’), although the
exact status of the recipient/beneficiary object is analyzed in different ways. For example,
it has been analyzed as primary object (PO) in Acazulco Otomi (Hernández-Green
2016) and in Tlahuica (Martínez Ortega 2016), while it is analyzed as indirect object in
Querétaro Otomi (Palancar 2009) and in Mazahua (Knapp 2008; Mora-Bustos 2019: 520,
among others).
AppPs licensed via the goal applicative seem to have been objects in 16th century
Old Otomi (OO) as well, as they could also be cross-referenced via person suffixes. An

13 The person suffixes -gi ‘1po’ and -bi ‘3io’ have the allomorphs -ki and -pi after voiceless segments,
respectively; reflexes of ✶-H are /ꞌ/ (= glottal stop) before glottalized stops/affricates.
14 As Hernández-Green (2022) shows for San Felipe Otomi, the object-marking pattern for 1st and 2nd
person is secundative, whereas the pattern for 3rd person is indirective. The cases where a 3rd person
object is overtly encoded by the suffix -bi ‘3io’ are discussed in Section 3.3.
690 Néstor Hernández-Green and Oscar López Nicolás

example of this is shown in (16). However, the frequency in day-to-day language use and
obligatoriness of such constructions remains unknown.

(16) Old Otomi

ko=kä-n-e-yë-gi
cop=2-prs-2/3.appl:goal-ss\come-1po
‘you come to me’
(Cárceres 1580/1907: 98)

In modern Otomi languages, adverbial inflection cognate with the goal applicative illus-
trated above in (16) does not promote adjunct phrases the way OO used to, as we will
show in Section 5.1.

3.1.2 ACs and valency in Otomi

In this section we discuss some interactions between ACs and valency in Otomi. On the
one hand, ACs may be restricted to verbs with certain valency, while others may not
have such restrictions. On the other hand, some valency-changing processes (causative,
reflexive, and the like) may co-occur with some ACs, but not with others. This subsection
does not contain examples of the Old Otomi goal applicative, as it is only attested with
the monovalent verb ëhë ‘come’ in the available data in Cárceres (1580/1907).
The Otomi benefactive applicative is restricted to bivalent and trivalent verb stems,
as is shown in (17), (18), and (19) below. The bivalent verb ꞌo̠tꞌi ‘make’ in (17a) can occur
in benefactive ACs as in (17b). Similarly, the trivalent verb xitꞌi ‘pour (liquid in recipi-
ent)’ illustrated in (18a) receives a fourth beneficiary participant in the AC in (18b). In
contrast, the monovalent verb ꞌuayi ‘break’ shown in (19a) does not acquire a second,
beneficiary-type participant with the stem-medial segment /h/, but a causer agent, as
can be seen in (19b).

(17) Acazulco Otomi

a. ja=da ꞌyo̠t=kꞌa=ch cómida
and=irr ss\make=det.sg.nv=dim food
‘And she used to cook food.’ {txt}
b. óra=da ꞌyo̠ꞌtꞌi=bi=nu̠=ŕ cómida
now=irr ss\make.appl:ben=3io=det.sg.dist=sg.3psr food
‘Now she’s going to cook them their food.’ {txt}

(18) Acazulco Otomi

a. da=xit=kꞌa téhe=kꞌa gásila
irr=pour=det.sg.nv water=det.sg.nv pan
‘She would pour the water in the pan.’ {txt}
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 691

b. da=xiꞌtꞌi=bi=ŕ gásila
irr=pour.appl:ben=3io=sg.3psr pan
‘S/he would pour it in their pan for them.’

(19) Acazulco Otomi

a. bi=ꞌuayi
pfv=break
‘It broke.’
b. bi=ꞌuahki
pfv=break
‘S/he broke it.’
(Not: ‘It broke for/despite someone.’)

Intransitive/causative verb pairs involving the same stem alternant process as the one
found in benefactive ACs of Otomi (both reflexes of the valency-increasing suffix ✶-H of
Proto-Otomi-Mazahua; Bartholomew 1965: 100) are discussed in Section 5.1.
Otomi ACs are observed to interact with reflexive-reciprocal constructions.15 The
benefactive applicative of Otomi does not seem to have restrictions to combine with
reflexive-reciprocal constructions; we do not have enough data from Old Otomi to assess
interactions between the goal applicative and such constructions. The following exam-
ples show the Otomi middle marker N- (see Palancar 2006), which has middle, antipas-
sive, reflexive, and reciprocal functions, co-occurring with the benefactive applicative.
The monotransitive stem ꞌǘni ‘give away’ in Acazulco Otomi in (20a) alternates with
the applicative (ditransitive) stem ꞌǘndi ‘give’ in (20b); the latter can receive the middle
prefix N- (allomorph nch- before glottals) in the reciprocal construction in (20c).

(20) Acazulco Otomi

a. bi=ꞌǘni
pfv=give.away
‘S/he gave it (away).’
b. bi=ꞌǘndi(=bi)
pfv=give.appl:ben=3io
‘S/he gave it to him/her.’
c. bi=nch-ꞌǘndi
pfv=mid-give.appl:ben
‘They gave it to each other.’

Lastly, Otomi languages have an impersonal voice contrast that suppresses the agent
in the verb that is used for topical patients without promoting them to subject (see

15 No such interactions are possible in Northern Zapotec.

692 Néstor Hernández-Green and Oscar López Nicolás

Hernández-Green 2018).16 This agent-suppressing construction can co-occur with the

benefactive applicative construction, as is illustrated in (21). In this example, the appli-
cative exponent is the alternate ditransitive stem ta̠m of the monotransitive stem ta̠i
‘buy’; the impersonal passive is marked by the aspiration of the initial consonant /t/.

(21) Eastern Highlands Otomi

a. xó̠n=do̠n-xä’t’ä=ꞌa=rá pahni=ꞌa=zí
sta=flower-cactus=det.sg.nv=sg.3psr shirt=det.sg.nv=dim
hmute=bi tha̠m-bi
girl=pfv imprs\buy.appl:ben-3io
‘The blouse that was bought for the girl is cactus-flower pink.’
(Voigtlander, Echegoyen, and Bartholomew, ms.)

3.1.3 The morphological expression of beneficiary AppPs in Otomi

The overt encoding of 3rd person objects in Otomi—which we mentioned in passing in §

3.1.1—is at least partially determined by the basic-applied distinction. This distinction is
not entirely straightforward, and it is described in the following paragraphs.
In general, basic 3rd person objects tend to be unmarked, while applied 3rd person
objects tend to be encoded in the verb via the suffix -bi ‘3io’ (or its clitic alternate form
=bi). These tendencies are based on the number of lexemes that follow one pattern
or the other, from which very few verbs deviate. For example, only one monotransi-
tive verb has been identified in Acazulco Otomi that admits, and even requires, the
3rd person marker, as illustrated in (22). In this example, the object NP in brackets is
cross-referenced in the verb by the enclitic =bi ‘3io’.

(22) Acazulco Otomi

dí=ꞌmbá̠ꞌtꞌi=bi [gáꞌtho=na hnǐhni]O
1.pfv=go.around=3io all=det.sg.prox village
‘We went around this village’ {txt}

Labile verbs that are either monotransitive or ditransitive, such as ꞌëń ä ‘say’ and ꞌa̠di
‘ask (for)’ are rare in Acazulco Otomi, and they take the 3rd person marker only in dit-

16 Northern Zapotec lacks such morphological strategy. Agent suppressing in this language is achieved
by using the 3rd person formal pronoun in the subject position, plus a suffix that indicates 3rd person
plural subject, as illustrated in the example below. The out-of-context interpretation of this example is
ambiguous between an impersonal and a definite 3rd person plural agent.

(v) b-s+ˀ-òt=lhát=éˀ béné=dáóˀ=nhàˀ

compl-pl.3s-beat=adv=3form.nom person=aff=def
‘They/someone beat that person.’ {txt}
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 693

ransitive uses. The monotransitive construction with the verb ꞌa̠di ‘ask (for)’ in (23a)
does not admit the suffix -bi ‘3io’, but the ditransitive construction in (23b) requires it.

(23) Acazulco Otomi

a. da=ꞌya̠ꞌ=kꞌa=ŕ ólla
irr=ss\ask.for=det.sg.nv=sg.3psr saucepan
‘They ask for their (corresponding) saucepan (full of mole).’ {txt}
b. nu̠xu̠ ntꞌa, bi=t-ꞌa̠h-p=ru̠ cúbeta
contr one pfv=imprs-ask.for-3io=det.sg.dist bucket
‘As for the other one, they asked him for the bucket.’ {txt}

Acazulco Otomi has only a few basic ditransitive verbs (xifi ‘tell’, ꞌa̠ndi ‘ask’, héhte
‘put [clothes] on [sb.]’, tühti ‘load [burden] on [sb.]’, xitꞌi ‘pour [liquid in recipient]’,
kǒtsꞌi ‘daub [substance] on [surface]’), none of which has been attested to receive the
3rd person marker in a 12-hour corpus. Consider the ditransitive constructions in (24)
without suffix bi ‘3io’.

(24) Acazulco Otomi

a. da=xit=[kꞌa téhe]T=[kꞌa gásila]Goal
irr=pour=det.sg.nv water=det.sg.nv pan
‘She would pour the water in the pan.’ {txt}
b. ja=ꞌya̠nd-ui=[na mbehtsi=a]R
and=ss\ask-2pl=det.sg.prox boy=encl
[jáma méro mjön ́ i ko=ra nde=na xühtsi=a]T
whether precisely really foc=ipfv want=det.sg.prox girl=encl
‘Ask this young man whether he really wants this young woman.’ {txt}

In contrast with basic monotransitives and ditransitives, where the overt expression of
the 3rd person object in the verb is very infrequent, the vast majority of applied ditran-
sitives require the 3rd person marker to cross-reference the recipient/beneficiary. The
3rd person marker is optional for only three derived ditransitive verbs, listed in (25)
below. Examples (26a) and (26b) illustrate the applied verb ꞌǘndi ‘give.appl:ben’ with
and without the 3rd person object marker, respectively.

(25) Acazulco Otomi

monotransitive applicative
ꞌǘni ‘give away’ ꞌǘndi(=bi) ‘give (to sb.)’
ꞌǘdi ‘show’ ꞌǘhti(=bi) ‘show (to sb.)’
jǜtꞌi ‘pay’ jǜꞌtꞌi(=bi) ‘pay (to sb.)’
694 Néstor Hernández-Green and Oscar López Nicolás

(26) Acazulco Otomi

a. ¿tébe̠ ꞌ=kꞌa gidi=ꞌǘndi=bi m-tꞌǎ?
what=3sg.nv 2.irr=give.appl:ben=3io other-one
‘What else are you going to give them?’ {txt}
b. xo=gidi=ꞌǘnd-ga
so=1.irr=give.appl:ben-1
‘So, I used to give that (i.e., nourishment) to them (i.e., my children).’ {txt}

Table 2 below summarizes the conditions in which 3rd person objects are overtly
encoded in the verb in Acazulco Otomi. In general, the 3rd person marker is ungrammat-
ical in monotransitives and basic ditransitives, except for three lexemes (numbers “1”
and “2” in parentheses). In contrast, applied ditransitives tend to encode the 3rd person
recipient/beneficiary overtly, except for three verbs (number “3” in parentheses) where
its expression is optional.

Table 2: Overt expression of 3rd person object per valency types in Acazulco Otomi.

Obligatory Optional Ungrammatical

Monotransitive (1)
Basic ditransitive (2)
Applied ditransitive (3)

Tendencies like those shown in Table 2 are also observed in other Otomi languages,
such as Eastern Highlands Otomi. In this language, (di)transitive verbs are classified
into two groups: “direct/indirect complement” (complemento directo/indirecto), where
3rd person objects remain unmarked, and “benefactive complement” (complemento
benefactivo), where it is marked in most (if not all) cases (Voigtlander and Echegoyen
1985: 170–198). All verbs in the latter group present reflexes of Proto-Otomi-Mazahua
✶
-H, which is associated to applicative morphology in Acazulco Otomi (see § 2.1.2).

3.2 Zapotec applicative syntax

3.2.1 Syntactic status of the AppP

The Zapotec comitative AC optionally alternates with a BC where the comitative partic-
ipant is introduced by the preposition lhénh ‘with’, illustrated in (27). Northern Zapotec
POs have four morphosyntactic properties that comitative AppPs acquire once they
are promoted via the suffixation of -lhénh ‘appl:com’, as shown in (28). The AppP dàˀ
Ísídôrònhàˀ ‘the late Isidoro’ in (28a) is a non-oblique phrase that immediately follows
the subject (in this case, the enclitic =áˀ ‘1sg’); these are two of the key properties of POs.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 695

The possibility of the AppP being pronominalized by an object pronoun is illustrated in

(28b). Finally, the construction in (28c) shows the possibility of the AppP being fronted
to the preverbal position without the need of an in situ resumptive pronoun—another
property of POs. Example (29a) shows the occurrence of a resumptive pronoun when
S/A is fronted; the absence of this pronoun in this syntactic context results in ungram-
maticality, as in (29b).

(27) Northern Zapotec

y-ídé=òˀ [lhénh nhàdàˀ]PP
irr-come=2sg prep:with 1sg
‘You’re coming with me.’ {txt}

(28) Northern Zapotec

a. b-ey+zàˀ-lhénh=áˀ [dàˀ Ísídôrò=nhàˀ]AppP
compl-leave.for.origo-appl:com=1sg.nom dead Isidoro=def
‘I went back with the late Isidoro.’ {txt}
b. sh-yêgh-lhénh=áˀ=[nhéˀ]AppP we-òtèˀ zhîlh=nhàˀ
icp-go-appl:com=1sg.nom=3form.objt nmlz-sell comal=def
‘I was going with her to sell comales.’ {txt}
c. [bíˀ=nhàˀ]AppP go-àt-lhénh=éˀ=x
clf.pro:infor=foc compl-die-appl:com=3form.nom=evid
‘With it (i.e., the baby) did she die, indeed.’ {txt}

(29) Northern Zapotec

a. nhètòˀ=nhàˀ ba=dxìˀ=tòˀ
1pl.excl=foc term=be.seated=1pl.excl
‘we were already seated’ {txt}
b. ✶ nhètòˀ=nhàˀ ba=dxìˀ
1pl.excl=foc term=be.seated
Intended meaning: ‘we were already seated’

Lastly, Northern Zapotec has an optional recipient/benefactive applicative marked with

the clitic =d ‘appl:ben’. AppPs in this AC acquire all the object morphosyntactic proper-
ties illustrated in (28) above. Only the pronominalization of the recipient AppP via an
object pronoun (=ndàˀ ‘1sg.objt’) is illustrated in (30a). The corresponding BC, where
the source participant is introduced by a relational noun, is given in (30b).

(30) Northern Zapotec

a. t-shàb=d=báˀ=ndàˀ shkʷé góˀn
icp-offer=appl:ben=3infor=1sg.objt pair bull
‘He’s offering me a couple of bulls.’ {txt}
696 Néstor Hernández-Green and Oscar López Nicolás

b. shêgh-nàb=sházé=òˀ=b lhàò bínhàˀ

irr.go-ask.for=adv=2sg=3an rn:to clf.pro:infor
‘You’ll have to go ask him (for the bulls).’

3.2.2 ACs and valency in Zapotec

In Northern Zapotec, applicative morphology occurs more frequently with mono- and
bivalent predicates. However, trivalent predicates can occasionally be seen taking
applicatives, as illustrated in (31). The trivalent predicate in (31a) can take the comita-
tive applicative -lhénh ‘appl:com’, as can be seen in (31b).

(31) Northern Zapotec

a. sh-yèb [ _ ]CS chìxh=òˀ=nhàˀ xhògh
icp-spread pf.tortilla=psr2sg=def hot.sauce
‘You spread hot sauce on your tortilla.’ {txt}
b. sh-yèb-lhénh=òˀ=báˀ chìxh=òˀ=nhàˀ xhògh
icp-spread-appl:com=2sg=3infor pf.tortilla=psr2sg=def hot.sauce
‘You and he/she spread hot sauce on your tortilla.’

The low frequency of trivalent predicates with applicatives is probably because most
underived predicates in Zapotec languages are monovalent, only a few are bivalent,
and far fewer are trivalent.
In addition, Northern Zapotec has a causative construction that can co-occur with
both the comitative and the recipient/benefactive applicatives. Only the former is illus-
trated here. The Northern Zapotec prefix gʷ- ‘caus’ in (32a) introduces a causer to the
bivalent stem èˀègh ‘drink’. In (32b), a causer and a concomitant agent are introduced by
the causative prefix gʷ- and the comitative applicative -lhénh, respectively.

(32) Northern Zapotec

a. bìtò b-lhéˀy+d=àˀ shí z-ídé-gʷ-èˀègh=báˀ=b nhìs=nhàˀ
neg compl-see=1sg.nom sub prf-come-caus-drink=3infor=3an water=def
‘I didn’t see that he came to make them (i.e., the bulls) drink water.’ {txt}
b. z-ídé-gʷ-èˀègh-lhénh=báˀ=nhéˀ=b nhìs=nhàˀ
prf-come-caus-drink-appl:com=3infor=3form.objt=3an water=def
‘He came with him to make them drink water.’

3.2.3 Applicative-conditioned constructions

Both Northern Zapotec ACs from in Section 3.2.2 are described as optional. As it turns
out, those optional ACs are only so in simple clauses, but they are obligatory in most
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 697

adjunct extraction constructions (i.e., interrogation, relativization, focalization, where a

phrase is said to be fronted, or dislocated to the left, to a preverbal position)17 and with
high-topicality adjuncts (see Peterson 2007: 83, 159). In other words, most adjuncts have
access to such constructions only if they are applied. In the following paragraphs, we
provide examples of ACs allowing for the extraction of adjuncts in Northern Zapotec.
One example of interrogated adjunct is shown in (33) below. The interrogated com-
itative in the Northern Zapotec example in (33a) requires the applicative suffix in the
verb, otherwise the construction is ungrammatical, as shown in (33b), and (33c) with
the comitative adjunct introduced by the comitative preposition.

(33) Northern Zapotec

a. nhő=nhàˀ shêgh-lhénh=òˀ
who=foc irr.go-appl:com=2sg
‘Who are you going with?’ {txt}
b. ✶nhő=nhàˀ shêgh=òˀ
who=foc irr.go=2sg
Intended meaning: ‘Who are you going with?’
c. ✶lhénh nhő=nhàˀ shêgh=òˀ
prep:with who=foc irr.go=2sg
Intended meaning: ‘Who are you going with?’

Relativization of adjuncts is illustrated in (34). Maleficiary relativization requires the

use of the applicative =d in Northern Zapotec in (34a); the same construction without
the applicative, as in (34b), is ungrammatical.

(34) Northern Zapotec

a. béné=nhàˀ [nhó=nhàˀ b-kʷàshèˀ=d=áˀ
person=def pro.rel=foc compl-hide=appl:ben=1sg.nom
mêdxoh=nhàˀ]RC
money=def
‘The person from whom I hid the money.’
b. ✶béné=nhàˀ nhó=nhàˀ b-kʷàshéˀ=áˀ mêdxoh=nhàˀ
person=def pro.rel=foc compl-hide=1sg.nom money=def
Intended meaning: ‘The person from whom I hid the money.’

As for focalization, consider the Northern Zapotec example in (35a), where the suffix
-lhénh ‘appl:com’ is used to put the comitative phrase dàˀ táwánhàˀ ‘my late grand-
mother’ in focus. The absence of the applicative enclitic results in an ungrammatical

17 Extraction seems to involve what generative approaches describe as wh-movement (see Carnie
2013: 362).
698 Néstor Hernández-Green and Oscar López Nicolás

construction, as in (35b), even with the preposition lhénh ‘with’ heading the comitative
phrase, as in (35c).18

(35) Northern Zapotec

a. dàˀ táó=á=nháˀ dx-òtèˀ-lhénh=áˀ zhîlh=nhàˀ
dead pf.grandmother=psr.1sg=foc icp-sell-appl:com=1sg.nom comal=def
‘with my late grandmother I used to sell comales.’ {txt}
b. ✶dàˀ táó=á=nháˀ dx-òtèˀ=áˀ zhîlh=nhàˀ
dead pf.grandmother=psr.1sg=foc icp-sell=1sg.nom comal=def
Intended meaning: ‘with my late grandmother I used to sell comales.’
c. ✶lhénh dàˀ táó=á=nháˀ dx-òtèˀ=áˀ zhîlh=nhàˀ
prep:with dead pf.grandmother=psr.1sg=foc icp-sell=1sg.nom comal=def
Intended meaning: ‘with my late grandmother I used to sell comales.’

3.3 Summary

In this section we have illustrated the (primary) object status AppPs acquire in ACs in
Otomi and Northern Zapotec. In the case of Zapotec, object morphosyntactic properties
include both coding and behavior-and-control properties (see Givón 2001: 175–178),
while Otomi primary objects can be identified by only one coding property (i.e., person
suffixes). As for obligatoriness, while the Otomi benefactive AC is obligatory, both the
comitative and the benefactive ACs in Northern Zapotec are optional (in simple clauses;
see below for details).
We also have discussed the interactions of ACs with the valency of the base verb, as
well as with valency-changing constructions. The possible co-occurrences of the North-
ern Zapotec and Otomi ACs with predicates with a certain valency and valency-changing
constructions are summarized in Table 3 below. The symbol “ ” indicates the possibility
of the combination of the constructions and categories that intersect in each cell. The

18 Many examples of AC in works about other Otomanguean languages contain an extracted adjunct.
Consider the examples from Peñoles Mixtec (Otomanguean > Mixtecan) and Temalacayuca Popoloca
(Otomanguean > Popolocan) in (via) and (vib), respectively, where the applicative markers and fronted
adjuncts have been put in bold.

(vi) a. ni-ki-xi ña¹dɨ’ɨ²³ [ni³-x-e²nde²-ndɨ’ɨ²³=n tɨ³kwe²³]RC

compl-r-come woman compl-r-pick-appl:com=2sg orange
‘the woman you picked the orange with came.’
(Ramírez Pérez 2014: 67–68)
b. méꞌè rí-t-jì-xį nújáꞌì
that prog-hab-go.a1-appl:instr thither
‘That’s why I’m going there.’
(adapted from Nakamoto 2017: 129)
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 699

shaded cells indicate that the valency-changing construction from the first column is not
found in the language.

Table 3: Co-occurring constructions in Northern Zapotec and Otomi.

Northern Otomi
Zapotec
rec/ben com ben goal

monovalent
bivalent ?
trivalent ?
reflexive-reciprocal ?
impersonal passive ?
causative

Apart from certain valency-changing constructions not being found in both languages,
there are some interesting differences between the two. First, the Otomi benefactive AC
co-occurs with all the valency-changing constructions that were investigated, while the
Northern Zapotec ACs are more restricted. Second, some semantic motivations may lie
beneath these restrictions in Northern Zapotec: while events that combine comitative
and reflexive do not seem to make much sense, comitative and reciprocal semantics may
be conceived so close to each other that their co-occurrence may be deemed redundant.
We also have shown the special morphological expression that 3rd person applied
POs have in Otomi ACs (mostly encoded by the object marker -bi ‘3obj’) with respect to
BCs (left unmarked in most cases). Finally, we have presented examples of the obligato-
riness of applicative morphology in the verb in Northern Zapotec when optionally appli-
cable adjuncts are extracted (i.e., interrogated, relativized, or focalized pre-verbally).

4 Semantics
This section deals with the semantic properties of AppPs in Northern Zapotec and in
Otomi. Section 4.1 and Section 4.2.1 review the semantic types AppPs cover in Otomi
and in Northern Zapotec, respectively. The interplay of ACs with propositional seman-
tics (in Sothern Zapotec) is briefly presented in Section 4.2.2.

4.1 Semantic roles in Otomi ACs

All Otomi ACs are semantically dedicated, i.e., they apply semantically homogeneous
paradigms of adjuncts. For instance, all the attested examples of Old Otomi ACs seem to
700 Néstor Hernández-Green and Oscar López Nicolás

be dedicated for coding goal-like participants as primary objects. One of such examples
in shown in (36).

(36) Old Otomi

ko=kä-n-e-yë-gi
cop=2-prs-2/3.appl:goal-ss\come-1po
‘You come to me.’
(Cárceres 1580/1907: 98)

Otomi (recipient/)benefactive ACs are dedicated to dative-like AppPs, among which

several closely-related semantic subtypes can be identified. In Acazulco Otomi, the ben-
efactive AC applies recipient/beneficiary participants, including semantically (and/or
typologically) related roles such as maleficiaries, sources, and causees/manipulees. One
example of each is provided in (37a–e), respectively. The applied participant is often
coded as possessor of the T(heme) phrase in an external possessor construction (Payne
and Barshi 1999: 3), as in (37c) with the clitic =í ‘pl.3psr’.

(37) Acazulco Otomi

a. dádi=güꞌtꞌi=bi=ga=ꞌmbe=nu̠
1.hab=ss\pay.appl:ben=3io=1=pl.excl=3sg.dist
‘We used to pay her.’ {txt}
b. tébe̠ ꞌ=kꞌa=xo=ndádi ta̠ꞌtꞌi=bi=ga
what=3sg.nv=ever=1.hab find.appl:ben=3io=1
‘Whatever I found for him.’ {txt}
c. porque=ngi ꞌmbëh-p=kꞌu̠=í ́ i=ꞌna
ndön
because=pst.irr imprs\steal.appl:ben-3io=det.pl.nv=pl.3psr cow=quot
‘Because his cattle were going to be stolen (from him).’ {txt}
́
d. höndi=kꞌu ̠ ze̠ t́ u=ꞌna kꞌu̠=álaja
take.appl:ben=det.pl.nv cloth=quot det.pl.nv=jewel
‘They took the clothes and jewelry away from them.’ {txt}
e. jón=gu̠ he̠ hki=bi=ga=gwa=du kú̠hu̠
nobody=1.irr release.appl:ben=3io=1=loc.prox=irr.ven enter.hither
‘I will not let anybody enter here.’ {txt}

4.2 Zapotec ACs

4.2.1 Semantic roles

Northern Zapotec’s comitative AC, as its name suggests, is dedicated for AppPs in the
role of companion. Example (38) illustrates this dedicated comitative AC, encoded by
the suffix -lhénh ‘appl:com’ (cf. lhénh ‘prep:with; join’).
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 701

(38) Northern Zapotec

kátèˀ b-s-ey+lháˀ-lhénh=éˀ=[nhéˀ]Com
adv compl-pl.3s-return.to.origo-appl:com=3form.nom=3form.objt
‘When they came back (to their village) with him.’ {txt}

As for the recipient/benefactive AC of Northern Zapotec, a dedicated AC involving the

clitic =d ‘appl:ben’, it applies participants with roles of recipient (39a), source (39b), and
maleficiary (39c); as for the maleficiary, this reading is given by pragmatic inference
from the participant’s affected interests in the event.

(39) Northern Zapotec

a. t-shàb=d=báˀ=[ndàˀ]R shkʷé góˀn
icp-offer=appl:ben=3infor=1sg.objt pair bull
‘He’s offering me a couple of bulls.’ {txt}
b. [léˀ=nhàˀ]S sh-yêgh-nàb=d=tòˀ yà=káˀ
3form=foc icp-go-borrow=appl:ben=1pl.excl firearm=pl:dist
‘It was him we were going to borrow the guns from.’ {txt}
c. b-kʷàshèˀ=d=òˀ[=báˀ]M mêdxoh=nhàˀ
compl-hide=appl:ben=2sg=3infor money=def
‘You hid the money from him/her.’

4.2.2 Interplay of ACs with meaning

No AC in Northern Zapotec (nor in Otomi) have been reported to modify the propositional
interpretation with respect to BCs beyond the addition of one participant. However, one
case of interplay between an AC and semantics is observed in Southern Zapotec. This
language has a comitative applicative construction that adds a concomitant subject to
monovalent predicates (Vásquez 2016: 153). The exponent of this AC is the suffix -nie̋
‘appl:com’. Consider the contrast between the one-argument event in (40a) with the
two-argument event in (40b) with the comitative suffix (in bold). The latter construction
allows both a comitative interpretation (number 1) and an affected-concomitant inter-
pretation (number 2).19

(40) Southern Zapotec

a. lè xèy g-uêy
sf man compl-go
‘The man left.’

19 This ambiguity (or perhaps rather, vagueness) seems to be pervasive in comitative constructions,
regardless of the possible involvement of applicative marking (Denis Creissels, p.c.).
702 Néstor Hernández-Green and Oscar López Nicolás

b. lè xěy g-uèy-nie̋ Mǐngw ló dzí’n

sf man compl-go-appl:com Domingo rn:face work
1. ‘The man went to work with Domingo.’
2. ‘The man took Domingo to work.’
(Vásquez 2016: 154)

4.3 Summary

Table 4 below summarizes the semantic roles associated (indicated with ) with each
AC in Northern Zapotec and in Acazulco Otomi. The first column contains general role-
types, and more specific role-types are listed in the second column.

Table 4: Semantic roles applied by each AC in Acazulco Otomi and Northern Zapotec.

Otomi Northern Zapotec

ben goal rec/ben com

a. Dative-like Recipient
Maleficiary
Source
Beneficiary
Causee/manipulee
b. Comitative
c. Goal

Table 4 shows that applicatives in both languages follow the typological trend of having
at least a benefactive and/or a comitative AC (along with instrumental ACs as well;
Zúñiga and Creissels, this volume). Applicativization of locative relations (row “Goal”)
is only found in Old Otomi.
In this section we saw that ACs in Otomi (beneficiary and goal) and Northern Zapotec
(beneficiary and comitative) are all semantically dedicated, that is, they are specialized
to apply semantically homogeneous paradigms of adjuncts. Additionally, we saw that AC
can sometimes switch the (expected) propositional semantics of a sentence with respect
to the BC, as is the case of the comitative AC of Southern Zapotec. This has not been
reported in either Otomi languages or in Northern Zapotec.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 703

5 Lookalikes
In the following sections, we present constructions in Otomi and Zapotec that are mor-
phologically similar to ACs, but that do not have their syntactic properties (Zúñiga and
Creissels, this volume). No syntactic AC lookalikes (i.e., syntactically, but not morpholog-
ically, similar to ACs) were identified in the languages described.
Phrases that are either promoted or registered (i.e., highlighted but not promoted;
cf. Norman 1978; Aissen 1990; Zavala 2000: 859–860; López Nicolás 2009: 104–105;
Hernández-Green 2016) via morphological devices similar to those of ACs have dif-
ferent syntactic statuses according to the construction involved and the grammar of
the individual languages. These syntactic statuses range from core grammatical roles
such as subject (promotion) to mere non-core grammatical roles (registration), with at
least one intermediate category in between. This continuum can be represented as is
shown in Figure 2.

promotion registration

subject non-core oblique

Figure 2: Promotion-registration continuum.

5.1 Otomi morphological lookalikes

The following paragraphs describe constructions where the relevant participant is coded
as subject. The valency-increasing suffix ✶-H of Proto-Otomi-Mazahua (Bartholomew
1965: 100) yielded applicative stems on the one hand (§§ 2.1.2, 3.1.1, 3.1.3), and causative
stems on the other hand. Similar to the applicative stems with ✶-H reflexes, the causative
pairs with this morphology seem to be lexicalized, but they present more idiosyncrasies.
First, while practically any bivalent verb can accept an extra beneficiary-like participant
(with or without overt applicative exponence), not every monovalent verb has a causa-
tive counterpart with ✶-H reflexes. For example, Acazulco Otomi has intransitive/causa-
tive pairs involving ✶-H, illustrated in (41a), but also analogous pairs without ✶-H, as in
(41b). The roman numerals in parentheses indicate the conjugation each verb belongs to.

(41) Acazulco Otomi

intransitive causative
a. ꞌuayi (I) ‘break’ ꞌuahki (I) ‘break’
zátꞌi (IV) ‘stick’ záꞌtꞌi (III) ‘stick’
pǒ̠ tsꞌe (I) ‘go up’ pó̠ꞌtsꞌe (III) ‘carry up’
b. tá̠yi (I) ‘fall’ tá̠yi (III) ‘drop’ (cf. ✶tá̠hki)
ꞌotꞌi (I) ‘dry’ ꞌotꞌi (III) ‘dry’ (cf. ✶ꞌoꞌtꞌi)
kú̠tꞌi (I) ‘enter’ kú̠tꞌi (III) ‘enter’ (cf. ✶kú̠ꞌtꞌi)
704 Néstor Hernández-Green and Oscar López Nicolás

Second, intransitive/causative pairs—with or without ✶-H reflexes—often involve a

change of conjugation, which is a lexical property of verbs. This is shown in 5 out of
the 6 pairs in (41) above. In contrast, the benefactive applicative construction never
involves conjugation alternations. Third, many bivalent verbs that seem to have
reflexes of ✶-H do not have intransitive counterparts (42a) or may even be intransi-
tives themselves (42b).

(42) Acazulco Otomi

a. ju̠ꞌtsꞌi ‘pull up (forcibly)’ (cf. ju̠tsꞌi ‘pull up; carry hanging from hand’)
ndo̠ꞌtsꞌe ‘look at (uphill)’
pě̠ ꞌtsꞌi ‘keep; put away’
hëꞌtꞌí ‘look at (downhill)’
he̠ h́ ki ‘cut off’
b. gá̠hki ‘lose weight’
në̌hki ‘be visible’
záhki ‘get stronger’

The registration of instrument phrases via adverbial inflection (called circumstantial

mood in Voigtlander and Echegoyen 1985: 222) in Eastern Highlands Otomi [otm] is
characterized by unrestricted non-oblique expression, but no other observable object
properties. This is an optional construction where the oblique expression of the instru-
ment in the BC alternates with its non-oblique expression in the construction with
registration, as illustrated between (43a) (basic inflection bi ‘pfv’) and (43b) (adverbial
inflection í ‘pfv.adv’), respectively. An instrument registered phrase (henceforth RegP)
can also be expressed by non-oblique pronouns, as is the case with ꞌu̠ ‘3pl’ in (44). Note
that this pronoun is not a morpheme on the verb, but a clitic hosted by the subject NP
gogu ‘deafness’.

(43) Eastern Highlands Otomi

a. bi=ga̠hk=ra dähpo [nange ra juai]InstrPP
pfv =clear=sg forest with sg machete
‘He cleared the forest with a machete.’
b. í=ga̠hk=ra dähpo [ra juai]Instr-NP
pfv.adv=clear=sg forest sg machete
‘He cleared the forest with a machete.’
(Palancar 2012)

(44) Eastern Highlands Otomi

ní=yän=ra gogu=[ꞌu̠]Instr-PRO
ipfv.adv=be.cured=sg deafness=3pl
‘Those (leaves) cure deafness.’ (Lit. ‘Deafness is cured with those.’)
(Voigtlander, Echegoyen, and Bartholomew, ms.)
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 705

No morphosyntactic properties like those of patients (or other core grammatical roles)
can be observed in the construction illustrated in (43b) and (44) above, other than the
expression of the instrument as a non-oblique phrase (see Palancar 2012). As was men-
tioned in Section 3.1.1, one key morphosyntactic property of core grammatical roles
in Otomi is the cross-referencing in the verb morphology via person markers. This is
a somewhat elusive feature of Otomi grammar, as instruments (and other adverb-like
semantic roles) tend to be 3rd person, and 3rd person participants are overtly encoded
in the verb only rarely, outside of recipient/beneficiary participants. In the available
corpus of Eastern Highlands Otomi, no examples of instruments (or other peripheral
participants) have been identified that bear person markers (1st, 2nd, or 3rd) in co-refer-
ence to the RegP. This fact is in line with Palancar’s (2012) assessment of the syntactic
status of such phrases.
In the last construction, Acazulco Otomi adverbial inflection, the RegP neither
acquires object (or subject) properties nor can it drop the preposition if the adjunct
noun is present, either in situ or in a fronted position. Consider the data in (45). The
instrument PP in the BC (basic inflection dí ‘1.pfv’) in (45a) must keep the preposition ko
‘with’ (< Spanish con ‘with’) in the corresponding construction with registration (adver-
bial inflection dá ‘1.pfv.gen’), shown in (45b). The latter is used to put the instrument in
focus, while the former is unspecified in this respect.

(45) Acazulco Otomi

a. dí=kꞌú̠hki=ga [ko=ya=ḿ tsꞌí]instr-PP
1.pfv=snap=1 with=det.pl.prox=1psr tooth
‘I snapped it with my teeth’
b. dá=kꞌú̠hki=ga [ko=ya=ḿ tsꞌí]instr-PP
1.pfv.gen=snap=1 with=det.pl.prox=1psr tooth
‘I snapped it with my teeth.’ {txt}

The only contexts where the preposition is dropped are relativization and interroga-
tion (i.e., extraction constructions), which involve the pronominal forms in boldface in
(46) below. The preposition is not allowed in relative constructions as the one in (46a),
and it is optional (and only attested in elicitation of grammaticality judgements) in the
interrogative construction in (46b). Adverbial inflection is obligatory in these examples,
which is reminiscent of the obligatoriness of the Northern Zapotec ACs in similar con-
texts (see § 3.2.3).

(46) Acazulco Otomi

a. gen=kꞌa=dá kꞌú̠hki
cop=3sg.nv=1.pfv.gen snap
‘That’s what I snapped it with.’
706 Néstor Hernández-Green and Oscar López Nicolás

b. ¿(ko=)tébe̠=dá kꞌú̠hki?
with=what=1.pfv.gen snap
‘What did I snap it with?’

The situation illustrated in (46) above is different to what is observed with instrument
RegPs in constructions with the adverbial inflection in Eastern Highlands Otomi, where
the preposition is dropped (see examples [43b] and [44] above). However, both lan-
guages coincide in the lack of cross-reference to the instrument RegP in the verb. For
example, a 2nd person reason participant in Acazulco Otomi can only be expressed via
a PP as in (47a), and never cross-referenced by the person marker -kꞌi ‘2po’ in the verb,
even if it is inflected with the general (i.e., instrumental) feature of adverbial inflection
(in bold), as shown in the ungrammatical construction in (47b).

(47) Acazulco Otomi

a. [por=gen=kꞌe]Reason, jan=ní=zongi=a
for=cop=2 because.of.that=ipfv.gen=weep=encl
‘S/he’s crying for you.’ (Hernández-Green 2016: 369)
(Lit. ‘For (the one that is) you, because of that s/he’s crying for.’)
b. ✶[por=gen=kꞌe]Reason, jan=ní=zon-kꞌi=a
for=cop=2 because.of.that=ipfv.gen=weep-2po=encl
Intended meaning: ‘S/he’s crying for you.’

5.2 Zapotec morphological lookalikes

The Northern Zapotec applied experiencer construction (AEC) is an obligatory, non-al-

ternating construction where an experiencer is encoded as subject. This construction
involves the clitic =d ‘appl:exp’, and it is thus more comparable to constructions with
subject undergoer nucleatives in Mapudungun and Yupik (Zúñiga and Kittilä 2019:
80) than to applicative constructions as defined in Zúñiga and Creissels (this volume).
Therefore, the phrase “applied experiencer construction” is clearly a misnomer, as it
would be more appropriately named “A-nucleativized experiencer construction”. The
AEC involves intransitive stems as yàxhgh ‘be necessary’, illustrated in (48a). An expe-
riencer participant to whom something is necessary cannot be expressed in a PP, as is
shown in the ungrammatical examples in (48b) and (48c).

(48) Northern Zapotec

a. dx-yàxhgh bárrét=nhàˀ
icp-be.necessary wrench=def
‘A wrench is necessary.’ {txt}
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 707

b. kátèˀ dx-yàxhgh
✶
lhéˀè [nìchè=áˀ]PP
adv icp-be.necessary 2pl prep:rea=1sg.nom
Intended: ‘When you are necessary because of me.’
c. kátèˀ dx-yàxhgh
✶
lhéˀè [pàr nhàdàˀ]PP
adv icp-be.necessary 2pl for 1sg
Intended: ‘When you are necessary to me.’

Like any other subject in Northen Zapotec, experiencers in the AEC can be crossrefer-
enced by nominative clitics, as in (49a) below. Such nominative clitics are required as
resumptive pronouns in the focus construction for subjects, as shown in (49b).

(49) Northern Zapotec

a. kátèˀ dx-yàxhgh=d=áˀ lhéˀè
adv icp-be.necessary=appl:exp=1sg.nom 2pl
‘When I need you.’ {txt}
b. nhètòˀ=nhàˀ dx-yàxhgh=d=tòˀ mêdxoh=nhàˀ
1pl.excl=foc icp-be.necessary=appl:exp=1pl.excl money=def
‘We need the money.’

The last subject property observed in these experiencer participants is the covert
subject (CS) construction, where the possessor of the (object) genitive NP is co-referent
with the logic subject, and the latter is not overtly expressed in the canonical position
for the subject. This is illustrated in the examples in (50). The experiencer is expressed
in the subject position in (50a). In (50b) there is no NP in that position, as the logic
subject is expressed within the object NP as possessor.

(50) Northern Zapotec

a. dx-s+ˀ-yàxhgh=d [bénéˀ=káˀ]S [mêdxoh=nhàˀ]O
icp-pl.3s-be.necessary=appl:exp person=pl:dist money=def
‘People need the money.’
b. dx-s+ˀ-yàxhgh=d [ _i ]CS [x-mêdxoh bénéˀi=káˀ]O
icp-pl.3s-be.necessary=appl:exp psr-money person=pl:dist
‘Peoplei need theiri money.’ (Lit. ‘Theyi need the peoplei’s money.’)

Besides the promotion of an experiencer to subject, the AEC often implies a rearrange-
ment in the mapping between semantic and grammatical roles (i.e., diathesis), repre-
sented in Figure 3. This rearrangement cannot happen, of course, with zero-valency
predicates, such as meteorological verbs, illustrated in (51).
708 Néstor Hernández-Green and Oscar López Nicolás

base S

AEC SEXP O Figure 3: Diathesis alternation in Northern Zapotec’s AEC.

(51) Northern Zapotec

wxê sh-yèˀ+nhìˀ=d=òˀ nhàˀ
tomorrow icp-dawn=appl:exp=2sg loc
‘You wake up there tomorrow.’ (Lit. ‘Tomorrow it dawns on you there.’)

Southern Zapotec presents a case of applicative-reciprocal polysemy, that is cross-lin-

guistically a rare type of co-expression pattern in the valency domain. The verb stem
kà’y ‘stain’ is bivalent, as shown in (52a) with both the agent and the patient expressed
with pronouns. The verb accepts the applicative suffix -nie̋ ‘appl:com’, but with syntac-
tic effects different than those expected, as can be seen in (52b). The first interpretation
is a comitative event with an intransitive—not transitive—predicate, while the second
interpretation is also intransitive but with a reciprocal interpretation.

(52) Southern Zapotec

a. p-kà’y nà lù’
compl-stain 1sg 2sg
‘I stained you.’
b. p-ká’y-nie̋ ná lù’
compl-stain-appl:com 1sg 2sg
‘You and I got stained together.’ /
‘You and I stained each other.’
(Vásquez 2016: 166)

In the following paragraphs we present two alternating constructions of Northern

Zapotec where a RegP is expressed as a direct NP without gaining the full range of object
properties. These constructions are in the mid-section of the continuum proposed in
Figure 2.
Instrument RegPs in Northern Zapotec (marked by -é ‘appl:instr’ plus =d) with
transitive stems do not have all the object properties canonical patients display. On the
one hand, unlike canonical objects, they are optionally direct in the preverbal position
(when focused), as shown in (53a), but obligatorily oblique when they occur in situ, as
in (53b).

(53) Northern Zapotec

a. [(#lhénh) lhákó=nhàˀ] dx-o+yèˀ-é+d=éˀ=nh
prep:with bark=foc icp-cook-appl:instr[tr]=3form.nom=3inan
‘It’s the bark they cook it with.’
(adapted from López Nicolás 2017: 217)
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 709

b. dx-o+yèˀ-é+d=éˀ=nh [✶(lhénh) lhákó=nhàˀ]

icp-cook-appl:instr[tr]=3form.nom=3inan prep:with bark=def
‘They cook it with the bark.’
(adapted from López Nicolás 2017: 216)

On the other hand, the instrument RegP can be cross-referenced by clitic pronouns in
the verb, as canonical objects are, provided that the patient NP is also expressed in the
sentence, as can be seen in (54).

(54) Northern Zapotec

w-dxíxé-é+d=éˀ=[nh]Instr [yìnhˀ yàˀà=nhàˀ]Pat
irr-measure-appl:instr[tr]=3form.nom=3inan chili green=def
‘She would weigh the green chili peppers with that (i.e., the scale).’ {txt}

The morphosyntactic properties of instrument RegPs with transitive stems combine

object encoding (i.e., object clitics in certain conditions) with oblique encoding (i.e.,
obligatory in situ). Instrument RegPs with intransitive predicates in Northern Zapotec
have even less object properties. The instrumental BC with the preposition lhénh ‘with’
is illustrated in (55a). In the alternate construction, the RegP can drop the preposition
only when fronted, as is shown in (55b). The ungrammatical example in (55c) shows
that the RegP cannot appear as a NP in the object position, even if the applicative mor-
pheme =d is used.

(55) Northern Zapotec

a. sh-dâ=chèˀ=áˀ lhénh lênh
icp-walk=adv=1sg.nom prep:with 3inan
‘I walk (with more confidence) with it (i.e., the walking cane).’ {txt}
b. (lhénh) mêdxoh=nhàˀ zhình=d=tòˀ nhàˀ=tè
prep:with dinero=foc irr.arrive.thither=appl:instr=1pl.excl loc=ints
‘With the money (i.e., by using it) we would get there.’ {txt}
c. ✶zhình=d=tòˀ mêdxoh=nhàˀ nhàˀ=tè
irr.arrive.thither=appl:instr=1pl.excl money=def loc=ints
Intended meaning: ‘We would get there with (i.e., by using) the money.’

Unlike instrument RegPs with transitives, RegPs with intransitives cannot be cross-ref-
erenced via clitic pronouns, as is shown in the ungrammatical example in (56) below.

(56) Northern Zapotec

✶
zhình=d=tòˀ=nh nhàˀ=tè
irr.arrive.thither=appl:instr=1pl.excl=3inan loc=ints
Intended meaning: ‘We would get there with (i.e., by using) that.’
710 Néstor Hernández-Green and Oscar López Nicolás

To end this section, we present some cases where a morpheme cognate with an applica-
tive seems to be lexicalized in certain predicates in Northern Zapotec. Consider the con-
struction with a trivalent verb in (57a), root in boldface, where all three participants are
overtly expressed. The applicative =d with this same verb root does not add an AppP,
but it is a lexicalized stem formative to yield an idiomatic expression (together with the
noun dìzhèˀ ‘word’) that describes the transference of information (rather than physical
objects), illustrated in (57b).

(57) Northern Zapotec

a. g-òèˀ=báˀ=nhéˀ mêdxoh
irr-give.to.3=3infor=3form.objt money
‘S/he’ll give him money.’
b. b-s+ˀ-òèˀ=d béné gólhé=nhàˀ nhàdàˀ dìzhèˀ
compl-pl.3s-tell=appl clf.pro:form old=def 1sg word
‘The elders told me that. {txt}

Similarly, the enclitic =d ‘appl:exp’ is lexicalized in the causative verb shown in (58a).
This verb form is the causative counterpart of the monovalent verb in (58b) via fortition
of the initial consonant (xh [ʐ] → x [ʂ]), so the enclitic here is redundant as a valen-
cy-changing strategy. For reference, the corresponding periphrastic causative construc-
tion is shown in (58c).

(58) Northern Zapotec

a. nhá b-xîzh=d=òˀ Júàn=nhàˀ
and compl-caus.laugh=appl:exp=2sg Juan=def
‘And you made Juan laugh.’
b. w-xhízh=báˀ chè Júành
irr-laugh=3infor gen Juan
‘S/he’ll laugh at Juan.’
c. be-ônh=òˀ [gâ b-xhìzh=báˀ]CC
compl-make=2sg comp compl-laugh=3infor
‘You made him/her laugh.’

The presence of the enclitic =d in the constructions in (57b) and (58a) above is not
clearly related to valency alternations (applicative or otherwise). However, it could
have semantic (rather than syntactic) motivations: the association of the enclitic with
recipients (see Table 4 in § 4.3) and experiencers may have licensed the lexicalization
observed in those examples.
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 711

5.3 Summary

Table 5 below summarizes the syntactic status of the added phrase in the constructions
presented in the previous two sections. Otomi constructions are shaded; Zapotec con-
structions are unshaded. The ticks indicate where the relevant phrase is with respect
to the continuum proposed in Figure 2.

Table 5: Syntactic status of the added phrase in Otomi and Zapotec applicative lookalikes.

Syntactic status
S Non-core Oblique

Otomi causative
Northern Zapotec applied experiencer
Southern Zapotec reciprocal
Eastern Highlands Otomi adverbial inflection
Northern Zapotec instrumental (tr.)
Acazulco Otomi adverbial inflection

By looking at the data in Table 5, it becomes evident that the continuum from Figure 2
is observable on different dimensions in the languages analyzed. On the one hand, the
continuum can be found within a single language (i.e., Northern Zapotec), where dif-
ferent constructions give a different status to the added phrase. On the other hand,
we can observe the continuum among languages in a family (i.e., Otomi languages), as
different languages can give a different status to the phrase introduced by a cognate
construction (i.e., adverbial inflection). Additionally, according to the object status the
goal AppP seems to have had in Old Otomi, variation on the continuum may even be
observable diachronically. By looking at two language families from the Otomanguean
stock, we have presented a wide range of syntactic (and morphological) phenomena
that will certainly contribute to our knowledge about ACs and related constructions in
a cross-linguistic perspective.

6 General summary and final remarks

The present chapter has described ACs in Zapotec and Otomi, two language families
from the Otomanguean stock (Central, Southern Mexico), which is one of the more
diverse and widespread groups of languages indigenous to Mexico. The description has
included morphological, syntactic, and semantic aspects in the Otomi and Zapotec fam-
ilies, as well as morphological lookalikes.
712 Néstor Hernández-Green and Oscar López Nicolás

Morphology
– Applicative morphology in Northern Zapotec and Otomi displays a great variety
of formal strategies, ranging from cliticization to affixation and stem alternations.
Affixation is found in other Otomanguean families; stem alternations are rarer, as
they have been attested in Otomi (and Mazahua, its sister language) and Chinantec.
All these applicative exponents are found to be lexicalized in some verbs in both
Northern Zapotec and Otomi, especially in the latter. Adverbial inflection, appar-
ently an AC in Old Otomi, displays cumulative exponence of applicative marking
together with grammatical person.
– Northern Zapotec applicative morphemes do not present any allomorphy, phono-
logical or otherwise. The Otomi benefactive applicative, in contrast, is often marked
by stem alternations that may or may not involve stem-medial or stem-final glottal
segments /h/ or /ʔ/; in some cases, verb stems in ACs are not formally distinguished
from those in BCs (i.e., they are labile).
– Old Otomi goal applicative seems to have had a reduced inflectional paradigm in
comparison with non-applicative verb forms: only three tense distinctions (out of
four in the indicative for verbs without the applicative), and no mood distinctions.
Northern Zapotec applicatives, in contrast, present no differences in inflectional
paradigm structure with respect to verb forms in BCs.
– The morphological variation of ACs within the Otomanguean stock is remarka-
ble from a cross-linguistic perspective, as it covers a wide range of phenomena
related to the concatenative/non-concatenative spectrum, allomorphy, and para-
digm structure.

Syntax
– The AppPs in all the applicative constructions described in this chapter have a gram-
matical status of object, both in Zapotec and in Otomi. Otomi languages have person
suffixes as the only reliable morphosyntactic (primary) object test, while Northern
Zapotec has both coding (i.e., non-oblique phrase, object marker on the verb, posi-
tion with respect to subject) and behavior (i.e., fronting without in situ resumptive
pronoun) tests of objecthood.
– Both the Otomi and Northern Zapotec (recipient/)benefactive ACs can be observed
in both bivalent and trivalent verbs but they are incompatible with monovalent
verbs; Northern Zapotec comitative AC, in contrast, has no such restriction. As for
diathesis alternation operations, the Otomi benefactive AC shows no restrictions of
combination, while Northern Zapotec ACs combine with the causative but not with
the reciprocal-reflexive construction.
– Beneficiary AppPs in Otomi are indexed differently from recipient NPs in basic dit-
ransitives: these recipient NPs are obligatorily zero-indexed (with two exceptions
where the zero is ungrammatical), but beneficiary AppPs are obligatorily indexed
with -bi ‘3io’ (with three exceptions where the suffix is optional).
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 713

– Although Northern Zapotec ACs are optional, they are so only in some unmarked
clause types: applicativization of the peripheral participant is obligatory in con-
structions that involve extraction (focalization, interrogation, relativization).

Semantics
– The Northern Zapotec comitative marker is a dedicated morpheme for comitative
AppPs; Old Otomi goal applicative seems to be a dedicated morphological device as
well. Otomi and Zapotec (recipient/)benefactive ACs are also dedicated to “dative-
like” participants, a label that encompasses a number of semantically and/or typo-
logically related semantic roles such as recipient, maleficiary, source, beneficiary,
and causee/manipulee.
– Zapotec ACs are optional. In these constructions, the AppP in the AC is an oblique
phrase in the corresponding BC. In contrast, the Otomi benefactive applicative is
obligatory, so there is no BC to compare the status of the phrase in it to the corre-
sponding AppP in the AC.
– In Otomi, the benefactive AC is the only strategy for expressing dative-like periph-
eral participants. Corresponding BCs do not exist in the language, either by means
of an oblique phrase (there is no preposition for dative-like PPs) or by means of the
codification of the dative-like participant as a possessor (these constructions are
rejected by speakers, and an AC is always preferred).
– ACs occasionally have idiosyncratic semantic effects with respect to the BC. For
instance, the comitative may have “affected concomitant” interpretations in South-
ern Zapotec.

Lookalikes
– Non-applicative constructions marked with applicative morphology (i.e., morpho-
logical lookalikes in Section 5) in both Zapotec and Otomi are arranged along a pro-
motion-registration continuum, based on how many core morphosyntactic proper-
ties (if any) non-core arguments acquire in such constructions. On the promotion
end of the continuum are, on the one hand, Zapotec Applied Experiencer Construc-
tion, which is a subject undergoer nucleative construction (Zúñiga and Kittilä 2019:
80) where an experiencer is coded as subject, and, on the other hand, Otomi caus-
atives that have a (however lexicalized) reflex of Proto-Otomi-Mazahua ✶-H. On
the registration end we have constructions where the adjunct phrase is registered
in the verb without acquiring an object status. Constructions in the middle of the
continuum display only some object properties acquired by the registered phrase.
The promotion-registration continuum can be observed not only within a single
language (i.e., Northern Zapotec) but also among languages in the same family (i.e.,
Otomi languages), which is worth highlighting from a cross-linguistic perspective.
– All non-promotional morphological lookalike constructions referred to in the
previous bullet point have syntactic properties similar to those of ACs that have
714 Néstor Hernández-Green and Oscar López Nicolás

nothing to do with object codification. The obligatoriness of Zapotec ACs in extrac-

tion constructions (i.e., interrogation, relativization, focalization) is also observed
among the non-promotional constructions from Section 5. Consider the following
examples from both Otomi and Zapotec interrogation and focalization in (59a) and
(59b), respectively.

(59) a. ¿tébe̠ =dí ꞌuingi=a?

what=2.irr.gen feed=encl
‘What are you going to feed them with?’ {txt}
b. [lhénh lhákó=nhàˀ]Instr dx-o+yèˀ-é+d=éˀ=nh
prep:with bark=foc icp-cook-appl:instr[tr]=3form.nom=obj3inan
‘It’s the bark they cook it with.’
(adapted from López Nicolás 2017: 217)

The obligatory registration of instrument phrases in extraction constructions like

those in (59) has led some authors to group registration and applicative constructions
together, often giving registration the label “registration applicative” (Norman 1978;
Zavala 2000: 859–860, among others).

Abbreviations
adv adverbial (registration)/adverb
aff affective
an animal (pronoun)
appl applicative
ben benefactive
caus causative
cl inflectional clitic
clf.pro pronominal classifier
com comitative
comp complementizer
compl completive
contr contrastive
cop copula
def definite
det determiner
dim diminutive
dist distal
du dual
encl enclitic
evid evidential
excl exclusive
exp experiencer
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 715

foc focus
form formal
fut future
fv final vowel
gen general
hab habitual
icp incompletive
imm immediative
imprs impersonal
inan inanimate
incl inclusive
infor informal
instr instrumental
ints intensifier
io indirect object
ipfv imperfective
irr irrealis
loc locative
mid middle
neg negative
nmlz nominalizer
nom nominative
nv non-visible
objt object(ive)
pf possessed form
pfv perfective
pl plural
psr possessor
po primary object
pot potential
pp prepositional phrase
prep preposition
prf perfect
prs present
prog progressive
pro.rel relative pronoun
prox proximal
pst past
quot quotative
r realis
RC relative clause
rea reason
rn relational noun
s subject
sf sentence focus
sg singular
ss secondary stem
sta stative
tr transitive
716 Néstor Hernández-Green and Oscar López Nicolás

txt example from text

ven venitive

References
Aissen, Judith. 1990. Una teoría de voz para idiomas mayas. In Nora C. England & Stephen R. Elliott (eds.),
Lecturas sobre la lingüística maya, 399–419. Guatemala: Centro de Investigaciones Regionales de
Mesoamérica.
Andrews, Henrietta. 1993. The function of verb prefixes in Southwestern Otomi. Arlington: Summer Institute of
Linguistics / University of Texas at Arlington.
Baerman, Matthew, Enrique L. Palancar & Timothy Feist. 2019. Inflectional class complexity in the
Oto-Manguean languages. Amerindia 41. 1–18.
Bartholomew, Doris A. 1965. The reconstruction of Otopamean. Chicago: University of Chicago dissertation.
Black, Cheryl A. 2000. Quiegolani Zapotec syntax: A Principles and Parameters account. Arlington: Summer
Institute of Linguistics / University of Texas at Arlington.
Campbell, Eric W. 2017. Otomanguean historical linguistics: Exploring the subgroups. Language and
Linguistics Compass 11(7). https://onlinelibrary.wiley.com/doi/epdf/10.1111/lnc3.12244 (accessed 9 April
2022).
Campbell, Lyle, Terrence Kaufman & Thomas C. Smith-Stark. 1986. Meso-America as a linguistic area.
Language 62(3). 530–570.
Cárceres, Pedro de. 1907 [1580]. Arte de la Lengua Othomi. Boletín del Instituto Bibliográfico Mexicano 6.
38–155.
Carnie, Andrew. 2013. Syntax: A generative introduction. Chichester, UK: Wiley-Blackwell.
Creissels, Denis. This volume. Applicative constructions and non-applicative uses of applicative morphology
in Tswana (Bantu).
Duncan, Phillip T. 2017. The role of argument structure in Me’phaa verbal agreement. Lawrence: University of
Kansas dissertation.
Foris, David P. 1993. A Grammar of Sochiapam Chinantec. Auckland: University of Auckland dissertation.
Gutiérrez, Ambrocio. 2021. A description and analysis of the syntax and functions of subordinate clauses in
Teotitlán del Valle Zapotec. Austin: University of Texas dissertation.
Hernández-Green, Néstor. 2015. Morfosintaxis verbal del otomí de Acazulco. Mexico City: Centro de
Investigaciones y Estudios Superiores en Antropología Social dissertation.
Hernández-Green, Néstor. 2016. Registration versus applicative constructions in Acazulco Otomi.
International Journal of American Linguistics 82(3). 353–383.
Hernández-Green, Néstor. 2018. Tres fuentes de las construcciones impersonales en otomí. Línguas
Indígenas Americanas 18(1). 7–29.
Hernández-Green, Néstor. 2022. Exploración sobre roles gramaticales en otomí de San Felipe Santiago a
partir de datos de corpus. Cuadernos de Lingüística de El Colegio de México 9. 1–64.
INALI. 2009. Catálogo de las lenguas indígenas nacionales. Variantes lingüísticas de México con sus
autodenominaciones y referencias geoestadísticas. Mexico City: Instituto Nacional de Lenguas Indígenas.
Kaufman, Terrence. 1994. The native languages of Mesoamerica. In R. E. Asher & Christopher Moseley
(eds.), Atlas of the world’s languages, 31–34. London: Routledge.
Knapp, Michael. 2008. Fonología segmental y léxica del mazahua. Mexico City: Instituto Nacional de
Antropología e Historia.
LBI. 2010. El Nuevo Testamento en el otomí del estado de México. <https://www.scriptureearth.org/data/ots/
PDF/00-WNT-ots-web.pdf>
 20 Applicative constructions in two Otomanguean families: Otomi and Zapotec 717

López Nicolás, Oscar. 2009. Construcciones de doble objeto en zapoteco de Zoochina. Mexico City: Centro de
Investigaciones y Estudios Superiores en Antropología Social MA thesis.
López Nicolás, Oscar. 2017. Las construcciones aplicativas en el zapoteco de Zoochina. In Gilles Polian,
Lourdes de León Pasquel, Roberto Zavala Maldonado, Eladio Mateo Toledo, Claudine Chamoreau &
Regina Martínez Casas (eds.), Nuevas miradas a la lenguas mesoamericanas, 191–235. Mexico City:
Publicaciones de la Casa Chata.
Macaulay, Monica. 1996. A grammar of Chalcatongo Mixtec. Berkeley: University of California Press.
Martínez Ortega, Aileen Patricia. 2016. Clases verbales, transitividad y valencia verbal en el pjyekakjó, tlahuica
de San Juan Atzingo. Mexico City: Instituto Nacional de Antropología e Historia / El Colegio de México.
Mora-Bustos, Armando. 2019. Cleft sentences in Mazahua (Oto-Manguean). Alfa 63(3). 517–539.
Nakamoto, Shun. 2017. El hombre que resucitó a su esposa. Tlalocan XXI. 105–142.
Norman, William M. 1978. Advancement rules and syntactic change: The loss of instrumental voice in
Mayan. Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society. 458–476.
Palancar, Enrique L. 2006. Intransitivity and the origins of middle voice in Otomi. Linguistics 44(3). 613–643.
Palancar, Enrique L. 2011a. Revisiting the conjugation classes of Eastern Highlands Otomi. Sprachtypologie
und Universalienforschung 64(3). 213–236.
Palancar, Enrique L. 2011b. The conjugations of Colonial Otomi. Transactions of the Philological Society 109(3).
246–264.
Palancar, Enrique L. 2012. Verbal inflection with applicative-like function? The adverbial tenses in
Otomi-Mazahua. Paper presented at the Syntax of the World’s Languages V, Dubrovnik, 1–4 October.
Palancar, Enrique L. 2013. The evolution of number in Otomi: The many faces of the dual. Studies in
Language 37(1). 94–142.
Palancar, Enrique L. 2017. Oto-Pamean. In Enrique Palancar & Søren Wichmann (eds.), The languages and
linguistics of Middle and Central America: A comprehensive guide. <https://hal.archives-ouvertes.fr/
hal-01493977>.
Payne, Doris L. & Immanuel Barshi. 1999. External possession: What, where, how, and why. In D. L. Payne &
Immanuel Barshi (eds.), External possession, 3–29. Amsterdam: John Benjamins.
Peterson, David A. 2007. Applicative constructions. New York: Oxford University Press.
Ramírez Pérez, Elodia. 2014. La predicación no verbal y las construcciones copulares en el tnu’u²³ dau²³ de
Santa María Peñoles, Etla, Oaxaca. Mexico City: Centro de Investigaciones y Estudios Superiores en
Antropología Social MA thesis.
Voigtlander, Katherine, Artemisa Echegoyen & Doris A. Bartholomew. Diccionario yuhú — español. Ms.
Zavala, Roberto. 2000. Inversion and other topics in the grammar of Olutec (Mixean). Eugene: University of
Oregon dissertation.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Zúñiga, Fernando & Seppo Kittilä. 2019. Grammatical voice. Cambridge: Cambridge University Press.
Sara Pacchiarotti
21 The polyfunctional applicative ✶-ɪd in Bantu
languages
Abstract: This chapter is about a verbal derivational suffix reconstructed as ✶-ɪd to Pro-
to-Bantu and labeled applicative. Reflexes of this suffix are found in a great number of
present-day Bantu languages. I show that the Bantu applicative suffix is highly poly-
functional. Applicative morphology is often the only means to introduce any semantic
role other than Agent, Patient, Instrument and Possessor into a main clause. Obligatori-
ness vs. optionality of applicativization to express any given semantic role depends on
the language, the lexical meaning of the verb root and the communicative context. Syn-
tactically, the applied phrase can be, among others, a direct object or an oblique. This
means that increased syntactic valence is not a defining feature of Bantu applicative
constructions. This chapter also highlights that suffixes formally identical to ✶-ɪd and
behaving morphophonologically as reflexes of ✶-ɪd display several non-syntactic func-
tions across the Bantu domain, such as adding aspectual nuances to the meaning of the
verb root (e.g., repetitiveness, thoroughness, excess, persistence, intensity), placing an
applied phrase with a Location(-related) semantic role under narrow constituent focus,
and widening/shifting the scope of a Location(-related) applied phrase with respect to
subject and object arguments.

1 Introduction
The Bantu languages branch off from one of the lowest nodes of the Benue-Congo
branch of the Niger-Congo phylum (Greenberg 1963; Mukarovsky 1976–1977; William-
son and Blench 2000). The fact that they are a relatively recent Niger-Congo offshoot
contrasts sharply with their remarkable geographic spread throughout Sub-Saharan
Africa (Bostoen 2018). Bantu languages span from southwest Cameroon in the northwest
all the way to southern Somalia in the northeast and until the southernmost tip of the
continent (Bostoen and Van de Velde 2019: 3–4). As such, they are the largest language
family in Africa by number of speakers and geographic extension. Estimates about
their exact number vary anywhere between more than four hundred to almost seven
hundred, one reason being the arbitrary demarcation between dialects and languages.
Due to their considerable number, there are several geographically-based referential
(i.e. non-genetic) classifications of Bantu languages. The most widely used is the one by
Guthrie (1971), which divides Bantu languages into fifteen zones each labelled with a
letter from A to S, where A is used for the languages currently located around the home-
land area in the borderland between Nigeria and Cameroon and S for the southernmost
languages of southern Africa. Within each zone, sets of ten after a letter refer to a group

https://doi.org/10.1515/9783110730951-021
720 Sara Pacchiarotti

of languages, e.g., B70, B80 etc., whereas a number following a decimal point refers to
specific varieties within a group, e.g., the Bantu language Ding is assigned the alpha-
numeric code B86 within the B80 referential group, see Hammarström (2019), Maho
(2009). The Bantu languages of Guthrie’s referential classification are known as Narrow
Bantu languages. This term is used in contrast to Wide Bantu, which includes both Bantu
and Bantoid languages, that is, about 150 varieties around the Bantu homeland area
which are genetically related to Bantu but are conventionally not considered to be part
of “Narrow Bantu” (Bostoen and Van de Velde 2019; Williamson and Blench 2000: 8–9).1
To date, there is no internal subclassification of Bantu languages based on the
Comparative Method (see however the tentative historical classification of Nurse and
Philippson 2003 based on shared innovations of non-lexical features). Bantu scholars
often make a non-genetic and approximate division between the northwestern Bantu
languages spoken in Cameroon, Gabon, Congo and parts of the Democratic Republic of
the Congo, from the rest of Bantu languages spoken to the east, southeast and southwest
(see, e.g., Schadeberg 2003b). Geographically speaking, northwestern Bantu is ill-de-
fined. Most authors include languages of zones A, B and C under the geographic label
“northwestern”, while others also add small adjacent pockets of zones D and H. This
latter grouping is known as “Forest Bantu” (Grégoire 2003; Nurse and Philippson 2003:
177) and is often used interchangeably with northwestern Bantu. The geographically
less widespread northwestern languages display a much higher linguistic diversity
compared to those spoken further east and south (Bostoen 2018). Most lexicostatistical
and phylogenetic classifications (e.g., Bastin, Coupez, and Mann 1999; Coupez, Evrard,
and Vansina 1975; Grollemund et al. 2015; Heine, Hoff, and Voßen 1977; Vansina 1995)
point to an initial split between some of the northwestern languages (especially zone
A and parts of zone B) and the rest, while there is variability in the internal subgroup-
ings of the rest. Given the number of languages, the general(izing) claims made in this
chapter should not be taken at face value, but rather understood as tendencies for
which exceptions can always be found.
In the remainder of this section, I discuss morphosyntactic features which are of
immediate relevance to the discussion of applicative constructions (ACs). Bantu nouns
are not marked for case, though tonal case marking has been claimed to exist in some
languages, see, e.g., Schadeberg (1986), Blanchon (1999). As is typical of Niger-Congo
(Hyman 2014), Bantu languages feature a system of nominal classification where nouns
belong to different, partially semantically motivated classes. Nouns tend to take a dis-
tinct noun class prefix in the singular and the plural and trigger agreement on other
constituents of the NP as well as the verb, as shown in (1) and (2). In both cases, the head
noun ò-mpûy ‘pangolin’ (singular cl3 ò-) in (1) and è-mpûy ‘pangolins’ (plural cl4 è-) in
(2) triggers class agreement on all modifiers of the NP as well as on the verb through

1 The term Bantoid is also used to name the node within the Benue-Congo branch of Niger-Congo to
which both Narrow Bantu and Bantoid languages belong.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 721

a class-determined subject index. In Bantu, a third person referent can belong to any
of the noun classes existing in a particular language. Nouns denoting humans usually
belong to class pairing 1/2.

(1) Ngwi B861 (Pacchiarotti and Bostoen 2021: 28)

òmpûy òmɔ̂ʁ ònɛ̂n òídzə́ːm
ò-mpûy ò-mɔ̂ʁ ò-nɛ̂n ó-tí-dzə̂mH
cl3-pangolin ap3-one ap3-big 3.subj-prf-disappear
‘One big pangolin has disappeared.’

(2) Ngwi B861 (Pacchiarotti and Bostoen 2021: 28)

èmpûy ènɛ̂n èɲí èídzə́ːm
è-mpûy è-nɛ̂n è-ɲí é-tí-dzə̂mH
cl4-pangolin ap4-big ap4-that 4.subj-prf-disappear
‘Those big pangolins have disappeared.’

The noun class system reconstructed for Proto-Bantu (PB) featured at least three loca-
tive noun classes whose semantic content varies depending on the author of the recon-
struction. These are PB cl16 ✶pà- ‘at, on’, cl17 ✶kʊ̀- ‘outside’ and cl18 ✶mʊ̀- ‘in’ (Meinhof
and van Warmelo 1932: 40). In many modern Bantu languages, nouns already carrying
a noun class prefix may additionally be prefixed with one of the historically locative
noun class prefixes. Some Bantu languages developed a locative suffix reconstructable
to some node as ✶-inɪ (Grégoire 1975: 187). These two strategies can be combined as
shown in (3), where the cl10 noun ‘gardens’ takes a historical cl16 locative prefix va-
(< PB cl16 ✶pà-) and a locative suffix -ni (< ✶-inɪ).

(3) Lomwe P32 (Güldemann 1999b: 52)

va-i-macha-ni
cl16-cl10-garden-loc
‘in the vegetable/fruit gardens’

In this chapter, I use the term locative phrase as a general cover term for a phrase with
locative semantics which displays morphological material Bantu nouns must take when
they function as semantic adjuncts specifying the location of an event, as in (3), without
specifying the syntactic category (NP, PP, or something in between) to which the locative
phrase belongs (see discussion below in this section).
In terms of morphological verb profile, Bantu languages spoken to the (south)east
and southwest of the northwestern area uniformly have morphologically complex verb
forms hosting numerous bound morphemes with a wide range of grammatical functions
such as participant cross-reference, Tense Aspect Mood and Polarity (tamp) marking,
voice and derivation (causative, applicative, passive, reflexive, reciprocal, middle, posi-
tional, separative, stative, etc.). The languages in the northwestern area tend to have
722 Sara Pacchiarotti

a much lower ratio of morphemes per word. These different morphological profiles
sometimes coexist in one and the same phylogenetic branch. This is illustrated in (4)
and (5) with two languages belonging to the West-Coastal Bantu branch, both spoken in
the Democratic Republic of the Congo (DRC).

(4) Ntandu H16b (Daeleman 1966: 253)

kabatátúsádísáánga ko
ka-ba-tá-tú-sál-ís-áng-a ko
neg-2.subj-prs.prog-1pl.obj-work-caus-ipfv-fv neg
‘They are not helping us.ʼ

(5) Nzadi B865 (Crane et al. 2011: 170)

mǐ ke pá fùfú kó yá bɔ
1sg neg.pst give fufu to 2sg neg
‘I didn’t give fufu to you.’

The verbal morphological template reconstructed for PB by Meeussen (1967: 108–111)

undoubtedly reflects the structure found in present-day languages such as Ntandu in
(4). A version of the PB verbal template adapted by Güldemann (1999a: 546) is given in
Figure 1, where the row immediately after the numbered verb slots indicates how these
are traditionally labelled in Bantu studies.2 The reconstruction of the template in Figure
1 to PB stage has been subject to a heated debated in recent years. One side of the debate
argues that Figure 1 actually represents the proto verbal form of a much later node in
the Bantu family tree (see, e.g., Güldemann 2011, 2022), and that PB would have looked
much more like Nzadi in (5) than Ntandu in (4).

–4 –3 –2 –1 0 +1 +2 +3

preinitial initial postinitial preradical radical prefinal final postfinal

tamp subj tamp obj root derivation/tam final vowel clause type/Obj/P

Figure 1: Proto-Bantu verbal template adapted from Güldemann (1999a: 546).

While subject indexes (subj) are usually present on the verb form in all TAMP finite con-
structions except the imperative, the presence of an object index (obj) and its co-occur-
rence with the lexical object NP it coreferences is subject to considerable variation (for
a typology see Beaudoin-Lietz, Nurse, and Rose 2004; on the reconstruction of object

2 In his original PB verbal template Meeussen (1967: 108–111) distinguishes two slots between the rad-
ical and the final, i.e., a suffixal slot for verbal derivation (where the applicative belongs) and a prefinal
slot for a specific inflectional suffix. Güldemann (1999a) conflates these two suffixal slots under the label
“prefinal” as he does with several distinct prefixal slots under the label “preinitial”.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 723

indexation in PB see Wald 2022). The applicative as well as numerous other PB deriva-
tional suffixes (Schadeberg and Bostoen 2019; Schadeberg 2003a) appear in position +1.
As we will see in Section 2, northwestern languages often display verbal phonotactic
maximality constraints which affect the segmental realization of the suffixal portion of
the template in Figure 1.
In terms of main clause syntax, the vast majority of Bantu languages displays an
SVOX word order (where X stands for a syntactic oblique) when clause constituents are
lexical NPs (Creissels 2000: 250). The immediately postverbal position is often a new
information focus position for object NPs (Buell, Riedel, and Van der Wal 2011), but in
some branches it is rather immediately preverbal (Bostoen and Mundeke 2012; Koni
Muluwa and Bostoen 2014). Changes to these word orders, often in combination with
morphological and suprasegmental strategies, correlate with changes in information
structure (Bearth 2003: 130; Downing and Hyman 2016). All known Bantu languages
display a robust nominative-accusative alignment system. Three overt properties pro-
viding evidence for this claim are: (i) the existence of two (supra)segmentally distinct
sets of bound pronominal forms used for participant cross-referencing, one for the S/A
category and one for P;3 (ii) verbs usually must index the S/A category in most tamp
constructions but not P (see above); and (iii) while the S/A category appears prever-
bally, P occurs postverbally. Zero-anaphora for anaphorically retrievable object argu-
ments of syntactically transitive verb stems is very common; on the other hand, subject
arguments usually do not undergo zero anaphora. While the grammatical relation of
subject is easily identifiable as the target of specific morphosyntactic operations, the
grammatical relation of object is less so. Typical Bantu objecthood diagnostics include,
among others, adjacency to the verb, ability to be cross-referenced on the verb, and
subjectivization by means of a passive construction (Hyman and Duranti 1982). In fact,
the by now well-known distinction between symmetrical and asymmetrical object-type
languages, originating in data from the eastern Bantu languages Chaga and Chewa
within the Lexical Functional Grammar framework (e.g., Alsina and Mchombo 1990;
Bresnan and Moshi 1990), was established on the basis of the behavior that multiple
postverbal object NPs show with respect to the aforementioned diagnostics. Neverthe-
less, their reliability is often disputed due to the arbitrariness of their choice, their
language-specific partial applicability (Rugemalira 1991), the high degree of morpho-
syntactic variation they display (especially object indexation on the verb), and the
intraspeaker variation in answers obtained in elicitation contexts (Pacchiarotti 2020:
59–66 for an overview).
An idiosyncrasy of Bantu languages immediately relevant for the discussion of ACs
is that the boundaries between syntactic object arguments and adjuncts with locative
semantics are far from discrete (Creissels 1998; Kuperus and Mpunga wa Ilunga 1990

3 I use S, A, P following Dixon (1994), among others. A stands for the subject of a transitive verb, P for
the object of a transitive verb and S for the subject of an intransitive verb.
724 Sara Pacchiarotti

and references therein; Schadeberg 1995). Bantu locative phrases can be either NPs,
PPs, or entities in between these two syntactic categories. This is because present-day
Bantu languages either preserved or restructured the original system where locatives
were part of the noun class system (see discussion above). Eastern Bantu languages
such as Chewa N31 by and large retained the original system. In these languages loc-
ative phrases tend to behave morphosyntactically as nouns and are thus likely to be
syntactic arguments, e.g., they trigger agreement within the NP just like non-locative
nouns, and they can bear the grammatical relations of subject and object to their verb,
as shown for instance by subject agreement and preverbal position of the locative
phrase ku San José in (6).

(6) Chewa N31 (Bresnan 1994: 111)

ku San José kú-ma-ndi-sangalâts-a
cl17 S.J. 17.subj-prs.hab-1sg.obj-please-fv
‘It pleases me in San José / (Being in) San José pleases me.’

Some northwestern and southern Bantu languages have developed locative preposi-
tions either from PB locative noun classes or other sources such as locative demonstra-
tives (Grégoire 1975: 106–134). Unlike locative-marked NPs, prepositions trigger neither
agreement on other constituents of the noun phrase, nor can they bear the grammatical
relation of subject and/or object to their verb. This is illustrated with the Londo data
in (7) where ò derives historically from PB locative class 17 ✶kʊ̀- but is now a phono-
logically independent element which does not trigger agreement in the following con-
nective construction: the morpheme wá linking the cl14 noun ‘face’ with the cl7 noun
‘hole’ agrees with the inherent class of the first linear noun (cl14).

(7) Londo A11 (Kuperus 1985: 182, glosses uniformized)

ì-sɛ́rì í-mòkɔ ̀mɛ́ ò bò-só (w-á) è-yùkù
cl19-antelope 19.subj-stopped prep cl14-face cl14-conn cl7-hole
‘The small antelope stopped in front of the hole.’

Yet other languages show a fluctuating situation where, depending on the construction,
locative phrases display morphosyntactic behavior which makes them less like an NP
and more like a PP, see Kuperus and Mpunga wa Ilunga (1990) for a detailed discussion
of this situation in Luba-Kasai L31a (see also Section 3). Because NPs are more likely to
be syntactic arguments than PPs, the morphosyntactic nature of Bantu locative phrases
has direct implications for claims related to valence, especially because, as we will see
in Section 3, applicative morphology can/must be used to introduce object NPs but also
locative phrases on a language-specific, root-by-root basis.
Because eastern Bantu data have dominated applicative studies in different the-
oretical syntactic frameworks (cf. supra) and because this handbook features a case
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 725

study on the eastern Bantu language Tswana (see Creissels, this volume), this chapter
tries to include as many northwestern Bantu data as possible.

2 Morphology
The only verbal derivational suffix reconstructed in PB with the label applicative is ✶-ɪd
(Meeussen 1967: 110).4 Its reflexes in present-day Bantu languages are quite uniform,
minor variations being conditioned by language-specific morphophonology and pho-
notactics. For instance, in Lengola D12, ✶-ɪd has three allomorphs according to Stappers
(1971: 265): -i when the preceding morpheme ends in /l/ (8a), -li when it ends in a vowel
(8b), and -il in all other contexts (8c).

(8) Lengola D12 (Stappers 1971: 265)

a. i-yoβul-a ‘to play drum’ i-yoβul-i-a ‘to play drum-appl’
b. i-tú-a ‘to sprout’ i-tú-li-a ‘to sprout-appl’
c. i-ɸam-a ‘to shout’ i-ɸam-il-a ‘to shout-appl’

As is common with reflexes of PB derivational suffixes having a near-close vowel (✶ɪ or

✶
ʊ) and/or ending in ✶d, reflexes of PB ✶-ɪd often undergo root-conditioned vowel and/
or nasal harmony. The Sikongo data in (9) show that the language’s direct reflex of ✶-ɪd,
i.e. -il, is subject to vowel harmony whenever the root contains /e/ or /o/, compare (9c–d)
with (9a–b) and nasal harmony whenever the root contains a nasal segment, see (9f–g).
Note that vowel and nasal harmony can also co-occur, see (9e).

(9) Sikongo H16a (Ndonga Mfuwa 1995: 315, 331)

a. kús ‘smear’ kús-ìl ‘smear-appl’
b. tál ‘look’ tál-ìl ‘look-appl’
c. vóv ‘talk’ vóv-èl ‘talk-appl’
d. vét ‘throw’ vét-èl ‘throw-appl’
e. nón ‘gather’ nón-èn ‘gather-appl’
f. kún ‘plant’ kún-ìn ‘plant-appl’
g. símb ‘touch’ símb-ìn ‘touch-appl’

4 Nevertheless, other PB verbal derivational suffixes might develop applicative functions overtime. For
instance, in some zone A languages reflexes of PB ✶-an ‘associative/reciprocal’ commonly introduce In-
strument and Comitative applied phrases (Schadeberg 1980). Similarly, the Great Lakes Bantu languages
of zone J (e.g., Rwanda, Ganda, Haya, etc.) use reflexes of PB ✶-ici ‘causative’ to introduce Instrument
applied phrases into a main clause (see, e.g., Kimenyi 1988: 367–368 for Rwanda).
726 Sara Pacchiarotti

In some Bantu languages, applicative derivation with monosyllabic verb roots can
trigger suffix doubling and/or vowel lengthening, as in (10), usually to satisfy minimal
length constraints (e.g. disyllabic or bimoraic) on verb stems (see Hyman and Mtenje
1999 and references therein).

(10) Lunda L52 (Kawasha 2003: 35)

d-á ‘eat’ d-íil-a ‘eat at/for/on, enjoy’
nw-á ‘drink’ nw-íin-a ‘drink for/at’

Due to the phenomenon known as “imbrication” in Bantu studies (Bastin 1983), reflexes
of applicative ✶-ɪd can be phonologically less transparent than those presented so far.
Imbrication is a process whereby certain -(V)CV verbal suffixes conflate or a suffix gets
infixed into a verb root. This morpheme fusion is usually accompanied by the loss of
the consonant of the infixed CV suffix and the insertion of its vowel portion in front the
final root consonant. This process commonly occurs with perfect ✶-ide, applicative ✶-ɪd
and causative ✶-icɪ. Typical instances of imbrication are in (11).

(11) Luba-Kasai L31a (Lukusa 1993: 58)

Underlying form Surface realization
mòn-ɪl ‘see-appl’ mwèːn
tabal-ɪl ‘open eyes-appl’ tabeel
tabeel-ɪʃ ‘open eyes.appl-caus’ tàbeʒ ‘cause too see clearly’

More extreme, less typical examples of imbrication occur in the northwestern area
(especially zones A, B, and C), mainly due to maximality constraints on verb stems,
such as maximum number of syllables and restrictions on the positional distribution
of consonants in verb roots according to place/manner of articulation (Ellington 1977;
Hyman 2008, 2010).5 An instance of such complex (morpho)phonological reflexes of
✶
-ɪd, reminiscent of classical imbrication, is illustrated in (12) with data from Ding B86.
Depending on the quality (and length) of the root vowel, the applicative -il (as well as
any other derivational or inflectional suffix featuring /i/) can trigger umlaut – with or
without concomitant diphthongization – and/or lengthening (for a discussion of these
phenomena in the area where Ding is spoken see Bostoen and Koni Muluwa 2014; Koni
Muluwa and Bostoen 2012). The consonant portion of -il is generally realized when the
root has a CV or CVV shape, but see pá vs. kú in (12c).

5 Besides loss and imbricated reflexes, phonological mergers among historically distinct verbal deriva-
tional suffixes in the northwestern area have created a situation where there is no longer a one-to-one
relationship between form and meaning, e.g. a suffix with a given phonological shape can express sev-
eral semantically distinct, unrelated meanings, or one single meaning can be expressed by three or four
formally distinct derivational morphemes (Ellington 1975). Although particularly common in the north-
west, mergers causing homophony are also attested elsewhere, see, e.g., Guérois and Bostoen (2018).
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 727

(12) Ding B86 (Ebalantshim 1980)

Root shape Underlying form Surface realization
a. C(G)V(ː)C túβ-il ‘drill-appl’ týýβ
lɔ ̀m ‘ask-appl’ lœ̀ œ̀m
lám ‘prepare-appl’ lɛ́àm
sààl ‘weed-appl’ sɛ́ɛ ̀l ~ syɛ̂l
bííl ‘call-appl’ bííl
kyɛ ̀l ‘cut-appl’ kyèèl
b. CVː píí ‘throw-appl’ pííl
lòò ‘bewitch-appl’ lø̀øl̀
c. CV pá ‘give-appl’ pɛ́à ~ pɛ́
kú ‘die-appl’ kwél

An even more extreme case of imbricated-like reflexes of ✶-ɪd is found in Tiene B81. In
this language C1VC2VC3V is the longest possible verb form on which the following con-
straints apply: C2 must be coronal, C3 must be non-coronal, and C2 and C3 must agree
in nasality (Hyman 2010: 153 see also Ellington 1977: 111–113 for derivational rules).

(13) Tiene B81 (Ellington 1977; Hyman 2010)

Underlying form Surface realization
a. yal-el-a ‘spread-appl’ yaal-a
b. sɔ́n-el-ɔ ‘write-appl’ sɔ́ɔn-ɔ
c. yók-el-a ‘listen-appl’ yólek-ɛ
d. yɔb-el-ɔ ‘bathe-appl’ yɔlɔb-ɔ
e. lɔŋ-el-ɔ ‘load-appl’ lɔnɔŋ-ɔ
f. dum-el-a ‘run fast-appl’ dunem-ɛ
g. t-el-a ‘throw-appl’ téel-ɛ
h. dí-el-a ‘wrap-appl’ díil-ɛ

In (13a–b), the Tiene applicative suffix -el surfaces as vowel lengthening to compensate
for the fact that C3, i.e. the /l/ of -el cannot be realized because it is coronal. In (13c–f),
C2 and C3 metathesize because C3 must be a non-coronal consonant. In (13e–f), the /l/
of the applicative becomes /n/ because C2 and C3 must agree in nasality. Additionally,
in (13d–e), the vowel portion of applicative -el harmonizes to the root vowel ɔ. Finally,
verb stems without a C2 such in (13g–h) show a more phonologically transparent man-
ifestation of -el (still subject to coalescence with the root vowel).
In Bantu languages without maximality constraints on verb stem length, there are
usually no restrictions on the stacking of verbal derivational suffixes, as shown in (14).
728 Sara Pacchiarotti

(14) Yao P21 (Hyman 2004: 70)

taam-uk-ul-igw-aasy-an-il-a
sit-imps-sep.tr-pass-caus-rec-appl-fv
‘cause each other to be unseated for/at’

On the other hand, in languages with maximality constraints on verb stem length, an
applicativized causative stem (15b) might be formally identical to a causativized stem
(15a). Note that while in (15a) the verb optionally cross-references the cl1 object NP
mwàn, in (15b) the applied 1sg Beneficiary object gets priority over the base Patient
object.

(15) Ding B86 (Kamtsha variety) (Sidonie Mayuma Mangwem and Donatien Musimar
Aleben p.c.)6
a. mùyœ́ ːy mwàn! b. ŋgyœ́ ːy mwàn!
mù-yɔ́b-iy mù-àn N-yɔ́b-iy-il mù-àn
1sg.obj-wash.intr-caus cl1-child 1sg.obj-wash.intr-caus-appl cl1-child
‘Make the child wash himself!’ ‘Make the child wash himself for me!’

Hyman (2003) reconstructs a default PB and Pan-Bantu suffix templatic ordering Caus-
ative-Applicative-Reciprocal-Passive (CARP). While in some languages this templatic
order is strictly fixed, in others compositionality and/or semantic scope constraints
allow for some combinations of suffixes to override the CARP template. For instance, in
the Chewa reciprocalized causative in (16a), the causative suffix precedes the recipro-
cal, while in the causativized reciprocal in (16b), the reciprocal precedes the causative
and thus violates the CARP template in favor of semantic scope.

(16) Chewa N31 (adapted from Hyman and Mchombo 1992: 350)
a. mang-its-an b. mang-an-its
tie-caus-rec tie-rec-caus
[Xi cause [each otheri to tie Y]] [X cause to [Yi tie each otheri]

On the other hand, as shown in (17b), in the same language the applicative suffix can
never precede the causative. The sequence root-caus-appl in (17a) can be interpreted
either as an applicativized causative, i.e. [[X cause Y to tie Z] for/with/at], or as a caus-
ativized applicative, i.e. [X cause [Y tie Z for/with/at]] (see Good 2005 and Hyman 2003
for further discussion).7

6 The Ding data not taken from Ebalantshim (1980) come from own fieldwork with two native speak-
ers of the eastern variety of Ding B86 (a.k.a. Ding Kamtsha), Mrs. Sidonie Mayuma Mangwem and Mr.
Donatien Musimar Aleben.
7 In his survey of morphological exponency of applicativization and causativization in Bantu, Good
(2005: 19, 35–40) reports very few languages which apparently allow applicative-causative (AC) order
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 729

(17) Chewa N31 (adapted from Hyman and Mchombo 1992: 352)
a. mang-its-ir b. ✶mang-ir-its
tie-caus-appl tie-appl-caus
[[X cause Y to tie Z] for/with/at] ✶
(intended meaning: [X cause [Y tie Z
[X cause [Y tie Z for/with/at]] for/with/at]])

The combination of applicative and reflexive, which unlike other verbal deriva-
tion occurs in the same slot as object indexes in Bantu, is often reported to develop
non-compositional meanings, special semantic nuances or discourse functions. For
instance, the combination of the applicative and reflexive in Rwanda (18) conveys
the expectations of the speaker with respect to the event they are narrating (Trithart
1983: 184–187).

(18) Rwanda JD61 (Kimenyi 1980: 57-58 cited in Trithart 1983: 186)
u-mu-gabo a-r-íi-ryaam-i-ye
aug-cl1-man 1.subj-prs-refl-sleep-appl-prf
‘The man is asleep.’ (when expected to be doing something else)8

In general, applicativized verbs show the same inflectional paradigms as those of their
corresponding underived roots, as illustrated with North Boma B82 in (19).9 Note that
the imbricated realization of applicative -il in North Boma also encompasses the har-
monization of a final vowel /a/ to /ɛ/. Hence, whenever a particular tamp construction
features a final -a as in the present and past habitual in (19a–d), the final vowel of the
applicativized stem is -ɛ. However, final vowels such as -i in the negative remote past
“override” the expected final -ɛ of the applicativized stem.

alongside causative-applicative. However, in these languages, AC order is not always fully productive
and/or semantically compositional.
8 In Rwanda as in many other Bantu languages, the noun class prefix on the head noun may be preced-
ed by another prefix, traditionally known as augment or pre-prefix (De Blois 1970; Meeussen 1969: 96-
99), see u- in (18). In languages such as Rwanda, nouns carry both the noun class prefix and the augment
by “default” (Van de Velde 2019: 249).
9 As will be shown in Section 5, applicative morphology in Bantu languages can be used to focalize
Location phrases. In some (especially) eastern Bantu languages, the so-called conjoint verb form is used,
among others, when a given clause constituent is the target of narrow focus, while the so-called disjoint
verb form is used when no specific element is focused within the clause (van der Wal and Hyman 2017).
The dynamics (and possible co-occurence restrictions) of the conjoint/disjoint forms in combination
with the focalizing function of the applicative are largely unknown, but see Misago et al. (forthcoming)
for a first exploration in the Eastern Bantu language Rundi.
730 Sara Pacchiarotti

(19) North Boma B82 (Stappers 1986: 52–57 glosses added)

a. namulaːʁatɔ́máː
na-mu-laːʁa-tɔ́m-áː
1sg.subj-1.obj-prs.hab-send-prs.hab
‘I often send him.’
b. namulaːʁatúmíɳɛ́ː
na-mu-laːʁa-túmíɳ-áː
1sg.subj-1.obj-prs.hab-send.appl-prs.hab
‘I often send for him.’
c. namututɔ́ma
na-mu-tu~tɔ́m-a
1sg.subj-1.obj-pst.hab~send-pst.hab
‘I used to send him.’
d. namututúmiɳɛ
na-mu-tu~túmíɳ-a
1sg.subj-1.obj-pst.hab~send.appl-pst.hab
‘I used to send for him.’
e. abɔ́túmí kɔ́
a-bɔ́-tɔ́m-í kɔ́
neg-2sg.subj-send-rem.pst neg
‘You did not send.’
f. abɔ́mutúmíɳí kɔ́
a-bɔ́-mu-túmíɳ-í kɔ́
neg-2sg.subj-1.obj-send.appl-rem.pst neg
‘You did not send for him.’

Bantu languages usually do not make use of applicative serial verb or converb construc-
tions. In fact, while other PB verbal derivational suffixes have been regularly innovated
since PB (Bostoen and Guérois 2022), the applicative has not. There are even languages
such as Eton A71 where the loss of the inherited applicative suffix has not been compen-
sated by the development of any alternative strategy, so that participants such as Bene-
ficiaries, commonly encoded in Bantu as core arguments via applicative constructions,
are simply encoded as objects that do not require licensing by any special form of the
verb (Van de Velde 2008). Nevertheless, applicative serial verb constructions expressing
instrumental (serial verb ‘take’), beneficiary (serial verb ‘give’), and comparative (serial
verb ‘(sur)pass’) meanings appear to be extremely common in a Bantoid group known
as Grassfields Bantu (Kießling 2021).
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 731

3 Syntax
Applicative constructions (ACs) in Bantu languages can be optional or obligatory. Lan-
guages where the applicative is obligatory on a root-by-root basis to introduce any given
set of semantic roles except Agent (and occasionally Instrument) have a very restricted
set of prepositions or no prepositions at all, e.g., the Chaga E60 language group (Bresnan
and Moshi 1993). Languages with optional applicative constructions usually have a
fairly developed system of prepositions, but the applicative might still be obligatory
with certain verb roots to introduce certain semantic roles, e.g. Mongo-Nkundo C61
(see Section 4).10 Usually there are no restrictions related to the syntactic valence of
the root: the applicative suffix in Bantu languages can appear on syntactically intran-
sitive, monotransitive and ditransitive roots. Nevertheless, in cases where the applica-
tive introduces a syntactic object, languages might show restrictions on the number
of postverbal object NPs and/or object indexes on the verb (see De Kind and Bostoen
2012 on Luba-Kasai L31a). Additionally, there might be language-specific applicativiza-
tion restrictions linked to the lexical meaning/semantic class of certain verb roots (see,
e.g., Machobane 1989 who argues that experiencer verb roots in Sotho S33 cannot be
applicativized).
Following Zúñiga and Creissels’s introduction to this volume, Bantu languages have
P-applicatives (= direct object applicatives), X-applicatives (=oblique applicatives) and
P~X applicatives, meaning that the syntactic status of an applied phrase (AppP) with
locative semantics can be in between an object and a syntactic adjunct (see Section 1).
In languages with optional ACs, the morphosyntactic entity introduced by the applica-
tive can alternatively be expressed as a prepositional phrase in the construction of the
underived root, i.e., the base construction (BC), with a concomitant semantic and/or
pragmatic difference(s) between the AC and the BC (see Section 4).11 To my knowledge,
there are no redirecting ACs in Bantu languages. The Tswana obligatory AC in (20) is an
instance of an X-applicative: the transitive verb root ‘kill’ requires applicative deriva-
tion in order to co-occur with a phrase specifying the exact Location of the killing. Syn-
tactically, mó lɪ́tɬàpéŋ̀ ‘on the stone’ is an obligatorily present oblique: it cannot be made
the subject of a passive construction, it cannot be indexed on the verb, nor can it appear
in immediate postverbal position (see Bantu objecthood diagnostics in Section 1).

10 Additionally, there are also languages like Tswana S31 where a developing system of (quasi-)prepo-
sitions co-exists with obligatory applicative constructions for all peripheral participant roles except In-
strument (see Creissels, this volume). However, synchronically, the presence of (quasi-)prepositions has
no incidence whatsoever on the syntactic use of the applicative in Tswana. Their development can only
be analyzed in terms of reinforcement of locative marking on locative phrases (Denis Creissels, p.c.).
11 Certain semantic roles might have a bi-clausal structure as an alternative way of expression. For in-
stance, in Bantu and Niger-Congo languages more generally, the Source and the Goal of movement with
a verb such as ‘move’ cannot be expressed simultaneously in the frame of a single-verb construction;
rather, a sequence of two verbs is necessary (Creissels 2006: 146–147; Creissels et al. 2008).
732 Sara Pacchiarotti

The valence of the applicativized stem in (20) is therefore not increased with respect to
the corresponding root.

(20) Tswana S31 (Creissels 1998: 133)

ke bolaetse noga mo letlapeng
kɪ ̀-bʊ́ lá-éts-ɪ́ nɔ́χà mó lɪ́tɬàpé-ŋ̀
1sg.subj-kill-appl.prf-fv cl9.snake loc cl5.stone-loc
‘I killed the snake on the stone.’

The obligatory Luba-Kasai AC in (21) is an example of a P~X applicative. The verb root
y ‘go’ requires the applicative to co-occur with the Path locative phrase ku cisalu ‘via
the market’. Locative markers such as ku (< PB cl17 ✶kʊ̀-) in this language “straddle the
categories of noun [class] prefixes and a type of preposition” (Kuperus and Mpunga wa
Ilunga 1990: 9); although they can appear in subject and object position and be indexed
on the verb, they display NP-level morphological properties which make them more
similar to prepositional phrases (see Kuperus and Mpunga wa Ilunga 1990: 11-26 for a
detailed discussion of their formal properties).

(21) Luba-Kasai L31a (De Kind and Bostoen 2012: 110)

ng-èndààmùshingà ù-di ù-y-il-a ku ci-salu
cl1n-merchant 1.subj-be 1.subj-go-appl-fv loc.cl17 cl7-market
‘The businessman is going via the market.’

Note that the Luba-Kasai root y ‘go’ can also optionally co-occur with the locative phrase
ku cisalu in its underived form, but in this case the locative phrase can only be under-
stood as a Spatial Goal (22).

(22) Luba-Kasai L31a (De Kind and Bostoen 2012: 110)

ng-èndààmùshingà ù-di ù-y-a (ku ci-salu)
cl1n-merchant 1.subj-be 1.subj-go-appl-fv loc.cl17 cl7-market
‘The businessman is going (to the market).’

These language-specific, root-by-root idiosyncrasies in the obligatory use of the applica-

tive virtually always revolve around location/spatial semantic roles. To give another
example, the Lunda root hólok ‘fall downʼ does not require applicative derivation to
co-occur with a locative phrase expressing the Location of the event of falling, but does
require it to co-occur with a Goal/Endpoint locative phrase.12

12 Some authors argue that in these cases, the applicative “changes” the semantic role of the locative
phrase. For arguments against this analysis see Pacchiarotti (2020: 126–132).
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 733

(23) Lunda L52 (Kawasha 2003: 261)

wahóloka mukaloña
wu-a-hólok-a (mu-ka-loña)
1.subj-pst-fall.down-fv loc-cl12-river
‘He fell down ([while he was standing] in the river).’

(24) Lunda L52 (Kawasha 2003: 261)

wahóloka mukaloña
wu-a-hólok-el-a mu-ka-loña
1.subj-pst-fall.down-fv loc-cl12-river
‘He fell down into the river.’

The optional Mbuun AC in (26) is an instance of a P-applicative. The syntactically ditran-

sitive root p ‘give’ in (25) can optionally co-occur with the prepositional phrase ɔ́ŋgírá
mwan ‘for the child’. Alternatively, the Beneficiary can also be introduced by applica-
tive derivation.13 In (26), ‘child’ is “promoted” to objecthood status compared to (25):
it appears in immediate postverbal position, can be indexed on the verb and can be
omitted from the construction (i.e., zero anaphora) just like objects of syntactically tran-
sitive verb roots in Mbuun (Bostoen and Mundeke 2011: 189–191). However, the base
objects of the root loose some of their object properties. Hence, the applicativized stem
in (26) is arguably tritransitive even though not all postverbal NPs display the same
object properties.

(25) Mbuun B87 (Léon Mundeke, p.c.)

maam wápa táár mats ɔ́ŋgírá mwan
maam o-á-p-a táár ma-ts ɔ́ŋgírá mo-an
cl1.mother 1.subj-prs.prog-give-fv cl1.father cl6-water prep cl1-child
‘Mother is giving father water for the child.’

(26) Mbuun B87 (Bostoen and Mundeke 2011: 190)

maam wápyéllé mwan táár mats
maam o-á-pyéllé mo-an táár ma-ts
cl1.mother 1.subj-prs.prog-give.appl cl1-child cl1.father cl6-water
‘Mother is giving father water for the child.’

Generally P-applicatives formed on a syntactically transitive verb root deriving a dit-

ransitive verb stem are identical to ditransitive/double object constructions featuring

13 Léon Mundeke (p.c.) informs me that in fact there is a slight semantic difference in the propositional
meaning of (25) compared to (26). When the Beneficiary is introduced as a prepositional phrase, the
subject is doing the action with special affection. This semantic nuance is absent in (25).
734 Sara Pacchiarotti

syntactically ditransitive verb roots such as p ‘give’ in (25). Depending on whether the
original object of a transitive verb root and the object brought about by the applica-
tive behave the same against a set of objecthood diagnostics, the terms symmetrical vs.
asymmetrical objects/object-type language are used after the seminal work of Bresnan
and Moshi (1990). For instance, (26) would be considered an asymmetrical object con-
struction.
An AppP in Bantu can also be an NP followed by an infinitival form of the verb as
in (27), where the syntactically transitive root lâm ‘prepare’ co-occurs with a Purpose
AppP. More research is needed to determine whether these clause-like AppPs behave
syntactically as objects.

(27) Ding B86 (Kamtsha variety) (Donatien Musimar Aleben p.c.)

bìyálɛ́àm mwàn kùdí
bi-ya-lɛ́àm mù-àn kù-dí
1pl.subj-prs.prog-prepare.appl cl1-child cl15-eat
‘We prepare (food) so that the child can eat.’ (lit: we prepare for the child eating)

In languages with optional ACs, applicativization may condition the access of syntac-
tic adjuncts to topicalization and focalization.14 Consider the Nsong proverb in (28),
extracted from a corpus of over two hundred spontaneously recorded proverbs.

(28) Nsong B85d (Koni Muluwa and Bostoen 2007: 553)

mokíl a nkím, ɛ́wá bawakɛ́ndɛ́l nkim
[mɔ-kíl a N-kím]TOP [ɛ́wá]FOC ba-wa-kánd-íl N-kim
cl3-tail conn cl9-monkey cl3.dem 2.subj-tam-tie-appl cl9-monkey
‘It is with its own tail that they tie up the monkey.’ (lit: as for the monkey’s tail,
with that they tie up the monkey)15

The syntactically transitive verb root kand ‘tie up’ undergoes optional applicative
derivation to introduce the Instrument AppP mokíl a nkím ‘the tail of the monkeyʼ.
The AppP appears dislocated in sentence initial position and is then anaphorically
referred to by a demonstrative pronoun which agrees in class with the head noun

14 I am not aware of a Bantu language where the applicative conditions the access of a syntactic adjunct
to relativization. There are usually dedicated constructions to relativize core as well as non-core argu-
ments without the need of applicative derivation (Meeussen 1967; Nsuka-Nkutsi 1982).
15 The free translation of (28) renders the French translation in the original which reads C’est avec sa
propre queue qu’on lie un singe. In fact, because the cl2 subject index ba- on the verb is used in Nsong as
a 3pl functional passive, another possible translation would be ‘It is with its own tail that the monkey is
tied up’. The meaning of the proverb is as follows: if someone has never paid the bride price for his wife
to his in-laws but receives bride price for his own daughter, he should pay to his in-laws what he has
received from his son-in-law for his daughter.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 735

(cl3) and appears in immediately preverbal position. In Nsong, clause initial posi-
tion is the topic position, while narrow-focused subject and object arguments appear
immediately before the verb (Koni Muluwa and Bostoen 2014). The Instrument AppP
in (28) could alternatively be expressed as a prepositional phrase headed by the prep-
osition eyí ‘withʼ, i.e., eyí mokíl a nkím, but this adjunct cannot be topicalized in clause
initial position (Joseph Koni Muluwa p.c.). Hence, in (28) the applicative introduces
an object AppP which can then be topicalized and focalized. In fact, most AppPs in
the corpus of proverbs appear in clause-initial position. For a parallel example of
the use of optional ACs to topicalize Instruments in Chewa N31 proverbs see Trithart
(1983: 183).

4 Semantics
In (mostly eastern) Bantu languages where applicative derivation is the only structural
means with a given verb root to express a given non-Actor semantic role into a main
clause (including Themes), the applicative suffix is semantically underspecified in that
many distinct semantic roles can be mapped onto the AppP. This mapping depends
on lexical meaning of the verb root, the meaning of other constituents present in the
clause, and the communicative intention of the speaker. For example, in the eastern
Bantu language group Chaga (E60), applicative derivation is required to express all
semantic roles except Instrument, Agent, Patient and Possessor (see also Tswana in Cre-
issels, this volume). In languages where ACs are optional, applicative morphology can
become restricted in the kinds of semantic roles it optionally introduces. For instance,
in Mbuun B87 (Bostoen and Mundeke 2011: 187–188), the applicative can only intro-
duce Recipient, Beneficiary, Maleficiary and Reason AppPs. On the other hand, in Mon-
go-Nkundo C61, a much wider variety of semantic roles (including Instrument, Purpose,
and Location) can be optionally expressed by the applicative (Hulstaert 1965: 257–263).
In languages with optional ACs such as Mongo, however, the applicative can still be
obligatory with certain verb roots to express certain semantic roles. For instance, with
a root such as ‘fall’, the only way to express the Endpoint of the event of falling in (29)
is with applicative derivation.

(29) Mongo-Nkundo C61 (Mbandaka variety) (Hulstaert 1965: 262, glosses added)
mbúla ɛ̌ɔdʒwɛ̂la bofaya
mbúla ɛ̌-ɔ-dʒw-ɛ̂l-a bo-faya
cl9.rain 9.subj-prf-fall-appl-fv cl1-visitor
‘It has rained on the visitor.’ (lit: rain has fallen on the visitor)

None of the prepositions present in Mongo-Nkundo can be used to express the meaning
‘on the visitor’ in the construction of the underived root (Gertrude Ekombe, p.c.),
736 Sara Pacchiarotti

although other interpretations (such as the Location where the rain falls) are possible
as shown in (30).

(30) Mongo-Kundo C61 (Mbandaka variety) (Gertrude Ekombe, p.c.)

mbúla ɛ̌-ɔ-jw-a ěle bo-faya
cl9.rain 9.subj-prf-fall-fv loc.prep cl1-visitor
‘It rained at the visitor’s place.’
✶
(intended: It rained on the visitor)

On the other hand, the Mongo root kaf ‘share, distribute’ can participate in the optional
AC in (31) where the applicative introduces the Recipient AppP baékoli ‘to the pupils’.

(31) Mongo-Nkundo C61 (Mbandaka variety) (Hulstaert 1965: 260, glosses added)
kafela baékoli mbɔmbɔ
kaf-él-á ba-ékoli mbɔmbɔ
distribute-appl-fv cl2-student cl9.chikwangue
‘distribute chikwangue to the pupils’16

This semantic role could also be expressed in the construction of the underived root
as a prepositional phrase as in (32) without any semantic differences which could be
captured in an elicitation context.

(32) Mongo-Nkundo C61 (Mbandaka variety) (Gertrude Ekombe p.c.)

kaf-á mbɔmbɔ ěle ba-ékoli
distribute-fv cl9.chikwangue loc.prep cl2-student
‘distribute chikwangue to the pupils’

In languages with optional applicative constructions, the differences between the

BC and the AC can be semantic and/or discourse-oriented, although there is still too
little research on this topic to make reliable generalizations. In terms of semantics,
optional ACs such as (34) can convey a higher degree of involvement/volitionality
of the Agent in the situation with respect to the BC in (33). According to Donatien
Musimar Aleben (p.c.), (33) would be uttered if the person who died did not know that
there was a snake somewhere: they got bit by the snake by accident and this caused
their death. On the other hand, (34) would be uttered if the person who died already
knew there was a snake out there and purposefully went to kill it, but the snake bit
them to defend itself.

16 Chikwangue, also known as kwanga, is a bitter fermented dough of manioc which is a staple food in
the DRC and other Central and West African countries.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 737

(33) Ding B86 (Kamtsha variety) (Donatien Musimar Aleben p.c.)

ndé àkýý sámà ntyɛ ̀l
ndé à-kýý sámà ntyɛ ̀l
3sg 1.subj-die.pst prep cl9.snake
‘He died because of the snake.’

(34) Ding B86 (Kamtsha variety) (Donatien Musimar Aleben p.c.)

ndé àkwéél ntyɛ ̀l
ndé à-kwéél ntyɛ ̀l
3sg 1.subj-die.appl.pst cl9.snake
‘He died because of the snake.’

Other reported semantic differences between ACs and corresponding BCs have to do
with the idea of “achieved goal”. According to Mabugu (2001: 120), optional ACs express-
ing an Animate Goal in Shona (e.g., Mother sent child toward grandmother) differ
semantically from their corresponding BCs in that they entail that the event described
by the root culminates at the endpoint (i.e. the child reaches grandmother), while the
BCs do not necessarily imply an achieved goal.
In terms of information structure, optional ACs in Bantu are used when the AppP
is a discourse topic, see (28). For instance, Rapold (1997: 43–44) finds that in a corpus of
spontaneous discourse optional ACs expressing animate Beneficiaries in Lingala C30B
are used whenever the Beneficiary is pronominal, i.e., given in discourse. Similarly,
Kisseberth and Abasheikh (1977) and Trithart (1983: 181–183) show that optional ACs
introducing Instrument AppPs in Mwiini G412 and Chewa N31 respectively are used
only when Instruments are asserted or presupposed in previous discourse.

5 Lookalikes
Suffixes formally identical to and behaving morphophonologically the same as reflexes
of PB applicative ✶-ɪd can have several valence-neutral semantic and discourse func-
tions. Crucially, they are available on a language-specific, root-specific basis whenever
the applicative is not required to introduce a non-Actor semantic role. Compared to
the classic valence-increasing function, they are much less researched and understood.
Some of these valence neutral functions are conceptually (and possibly diachronically)
closely related to the syntactic functions of PB ✶-ɪd and point to the polyfunctionality of
the suffix. Others are rather at odds with those, which leaves one wondering whether
there were not originally two or more formally and functionally distinct suffixes in PB
(or earlier on) which merged into or became homophonous with ✶-ɪd (Hyman 2007;
Pacchiarotti 2020: 279–286; 2022).
738 Sara Pacchiarotti

Semantically, reflexes of PB ✶-ɪd are used virtually everywhere in present-day

Bantu languages to indicate that the action/state described by the root is performed to
completion, or that the action is performed continuously, with intensity, persistence,
excess, or repetition, among other aspectual nuances.17 Depending on the language-spe-
cific phonotactics and minimality/maximality constraints (see Section 2), one, two, or
even three applicative suffixes might be required to express this Aktionsart function,
which is usually not fully productive, but rather restricted to a certain number of roots
in the lexicon. In (35), the transitive Mongo root ɔtsw ‘penetrate’ undergoes applicative
derivation and the meaning of the derived stem is penetrate the forest deeply, far away.

(35) Mongo-Nkundo C61 (Hulstaert 1965: 263)

ǎɔlɔtsw(ɛl)a ngonda
ǎ-ɔ-ɔtsw-(ɛl)-a ngonda
1.subj-prf-penetrate-(appl)-fv cl9.forest
‘He has penetrated (deeply, far away) the forest.’

The two functions related to information structure always involve a Location AppP.
A first widespread function of the applicative is placing an AppP with a Location-re-
lated semantic role (usually the Location where the event takes place) under some kind
of narrow or constituent focus. This function is usually available only for those roots
which do not require applicative derivation to co-occur with the Location-related AppP
which is being narrow-focused, such as the Kongo ya Leta root lámb ‘prepare food’ in
(36). While (36) is a thetic statement which could be a felicitous answer to a question
like What’s happening?, the construction in (37), structurally identical to (36) except for
the presence of the applicative on the verb stem, places the prepositional phrase nà
nzúngù yà nɛ́nɛ ̀ ‘in the big potʼ under selective (choosing one item from among a pre-
supposed set of possible values) or replacing focus (removing an item in the pragmatic
information of the addressee and replacing it with the correct one), following the
terminology of Dik et al. (1981). Different speakers consistently report that (37) would
be used to correct the expectation of a hearer who believes that the women are cooking
in another pot or to inform the hearer that among other possible cooking vessels, the
women are preparing the food in the big pot. As such, (37) would be a felicitous answer
to In which pot are they preparing? or Are they preparing in the small pot?18

17 Some authors report meanings such as that the action described by the verb root is performed in
vain or little by little (Hulstaert 1966; Sharman 1963).
18 In the Kongo ya Leta spoken in Kikwit, applicative morphology can also be used on where questions
with a similar meaning nuance to the one described by the contrast between (36) and (37). For example,
mwánà kɛ ̀dílà wápí? | mù-ánà kɛ ̀-díl-à wápì | cl1-child prs.prog-cry-fv where| ‘Where is the child cry-
ing?’ would be used by someone who does not know where the crying child is located. However, mwánà
kɛ ̀dídilà wápí? | mù-ánà kɛ ̀-díl-il-à wápì | cl1-child prs.prog-cry-appl-fv where| ‘Where is the child cry-
ing?’ would be used when the speaker expects the child to be somewhere (e.g., in the house), but because
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 739

(36) Kongo ya Leta spoken in Kikwit (Joseph Koni Muluwa p.c.)

bànkɛ́ntɔ ̀ kɛ ̀lámbà nà nzúngù yà nɛ́nɛ ̀
bà-nkɛ́ntɔ ̀ kɛ ̀-lámb-à nà n-zúngù yà nɛ́nɛ ̀
cl2-woman prs.prog-prepare-fv prep cl9-pot conn big
‘The women are preparing (food) in the big pot.’

(37) Kongo ya Leta spoken in Kikwit (Joseph Koni Muluwa p.c.)

bànkɛ́ntɔ ̀ kɛ ̀lámbílà nà nzúngù yà nɛ́nɛ ̀
bà-nkɛ́ntɔ ̀ kɛ ̀-lámb-íl-à nà n-zúngù yà nɛ́nɛ ̀
cl2-woman prs.prog-prepare-appl-fv prep cl9-pot conn big
‘The women are preparing (food) in the big pot.’

As far as available sources allow to say, new information focus, or “completive” in the
terminology of Dik et al. (1981), can also be signaled by applicative derivation (on an
Event Location AppP), see Pacchiarotti (2020: 144–157) for an overview.
A second discourse function, attested so far only in eastern Bantu languages, is
what has been called “implicit contrast” (Trithart 1983: 170), “widening the scope of the
locative phrase” (Grégoire 1998; Hyman, Duranti, and Morolong 1980; Trithart 1977),
or “argument orientation” (Pacchiarotti 2020: 141). To see this, consider the contrast
between (38) and (39). In (38), the root uumv is followed by an object NP ‘snake’ and an
adjunct-like locative phrase ‘in the forest’ which has scope over the object NP ‘snake’
but leaves the position of the subject vague: the man could or not be in the forest. When
the same root co-occurs with the applicative in (39), the scope of the locative phrase is
now on the subject NP ‘the man’ while the position of the object is vague. In this case
too, the only structural difference between (38) and (39) is that in (39) the Location AppP
can no longer be omitted.

(38) Rundi JD62 (Misago et al., forthcoming)

Umugabo yuúmvye inzóka (mw'iishaamba)
u-mu-gabo a-á-uúmv-ye i-n-zóka (mu i-∅-shaamba)
aug-cl1-man 1.subj-rem.pst-hear-pfv aug-cl9-snake cl18 aug-cl5-forest
‘The man heard the snake in the forest.’
Implication: The snake is in the forest, the man could be or not in the forest.

of where the crying comes from he/she realizes that the child is elsewhere (e.g., outside). Elsewhere in
Bantu, applicative morphology commonly occurs on where, why and how questions (Trithart 1983: 148).
740 Sara Pacchiarotti

(39) Rundi JD62 (Misago et al., forthcoming)

Umugabo yuúmviye inzóka mw'iishaamba
u-mu-gabo a-á-uúmv-i-ye i-n-zóka mu i-∅-shaamba
aug-cl1-man 1.subj-rem.pst-hear-appl-pfv aug-cl9-snake cl18 aug-cl5-forest
‘The man heard the snake in the forest.’
Implication: The man is in the forest, the snake could be or not in the forest.

The few data available for other languages indicate that there are cases where the scope
of the locative phrase is not shifted from the object to the subject NP, but is widened to
include both subject and object NPs (see Pacchiarotti 2020: 141–144).19
While lexical aspectual meanings can be conveyed by other verbal derivational
suffixes in Bantu (see, e.g., Bostoen, Dom, and Segerer 2015) and even Bantoid (Kießling
2004), the discourse functions of narrow focus and argument orientation are exclusive
to the applicative. What is more, the applicative is the only verbal derivational suffix
which can serve opposite discourse functions, namely topicalization (see Section 3) and
focalization, depending on the construction, the verb root and the language.

6 Conclusions
According to the analytical levels set out in the questionnaire underlying the contribu-
tions to this handbook, applicative constructions in Bantu languages can be summa-
rized as follows:

Morphology
– Bantu languages usually have only one dedicated applicative verbal suffix recon-
structed as ✶-ɪd in Proto-Bantu. Nevertheless, verbal derivational suffixes with
other primary functions (e.g., PB ✶-an ‘reciprocal’, PB ✶-ici ‘causative’) can take on
applicative functions over time.
– Bantu languages usually do not make use of applicative serial verb or converb con-
structions.
– Reflexes of PB ✶-ɪd in present-day Bantu languages range from phonologically
transparent to phonologically complex, involving phenomena such as metathesis,
assimilation, and vowel lengthening, umlaut and diphthongization among other
processes. These phonologically complex reflexes are usually due to phonotactic

19 There is actually a third semantico-pragmatic function of applicative morphology in relation to Lo-

cation AppPs, namely, indicating that the action described by the verb root habitually occurs at a certain
location. This has been reported in a handful of eastern Bantu languages (see, e.g., Creissels 2004) and
might be an effect of the interaction between the applicative and tense/aspect morphemes (Mabugu
2001).
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 741

maximality constraints (number of syllables allowed per verb stem, restrictions in

the place of articulation of consonants in specific positions within the stem, etc.)
especially common in northwestern languages. As it happens with other recon-
structed PB derivational suffixes, reflexes of PB ✶-ɪd very often undergo nasal and/
or vowel height harmony. Especially in eastern Bantu languages, the applicative
in combination with monosyllabic verb roots can be doubled or undergo vowel
lengthening to satisfy minimality length constraints on the verb stem.
– In general, applicativized verb stems take the same tamp morphology as their
underived counterparts, although future research should investigate the relation-
ship between the information-structure sensitive conjoint-disjoint morphology and
the applicative in its focalizing function.

Syntax
– An AppP can be an oblique, a direct object, an embedded clause featuring a non-fi-
nite form of the verb, or a morphosyntactic entity in between an oblique and a core
object argument. The latter are usually phrases with locative semantics which were
presumably part of the noun class system in PB (and thus being NPs were more
likely to be core syntactic arguments) and have been preserved or restructured in
different ways in present-day languages. Perhaps the only generalization that can
be made about the syntactic status of AppPs in different ACs is that those which
introduce a Beneficiary, Recipient or Human Goal often have the syntactic status
of (direct) object. This implies that ACs are not always valence-increasing in Bantu
languages. Valence-rearranging ACs are not attested in Bantu languages.
– In languages with optional ACs the semantic role introduced by the applicative can be
alternatively expressed in the BC as a prepositional phrase (with concomitant seman-
tic and/or discourse differences between the BC and the optional AC). In the optional
ACs, this erstwhile oblique constituent is syntactically promoted to objecthood.
– In eastern Bantu languages where minimality constraints predominate, there are
no restrictions on the stacking of voice operations on the verb stem. The templatic
order in which voice operations occur is Causative Applicative Reciprocal Passive,
but compositionality or semantic scope constraints allow for variations of this tem-
platic ordering in some languages. In the northwestern languages, the same maxi-
mality constraints affecting the phonological realization of reflexes of PB applica-
tive ✶-ɪd also affect the realization of a sequence of verbal derivational suffixes.
– When applicative derivation occurs on a syntactically transitive verb root and
the AppP is syntactically a direct object, the resulting applicativized construction
is identical (except for the presence vs. absence of the applicative) to a construc-
tion where a syntactically ditransitive verb root occurs with two postverbal object
NPs. Traditionally, depending on whether these two objects behave the same or not
according to a given set of objecthood diagnostics, they are called symmetrical or
asymmetrical.
742 Sara Pacchiarotti

– In languages with optional ACs, applicativization can condition the access of non-
core syntactic arguments to operations such as topicalization and focalization.
These operations are typically available to subjects and objects in Bantu but not to
syntactic adjuncts. These need to become objects through applicativization in order
to be topicalized (and sometimes focalized).

Semantics
– Bantu languages typically have only one applicative verbal suffix. Usually, there
are no restrictions on applicativization depending on the syntactic valence of the
verb root, but occasional restrictions dictated by the lexical meaning of the verb
root have been reported.
– ACs in Bantu languages can be optional or obligatory depending on the language,
the verb root and the non-Actor semantic role which needs to co-occur with that
root in a given communicative context. In languages with obligatory ACs, reflexes
of PB applicative ✶-ɪd are semantically underspecified in the sense that they can
introduce any non-Actor (sometimes excluding Instruments) semantic role depend-
ing on the lexical meaning of the verb root and the communicative context. In lan-
guages with optional ACs, the applicative might become restricted in the types of
semantic roles it can introduce and/or there might be semantic roles which can
only be introduced by the applicative. In general, there is remarkable language-spe-
cific, root-specific variation and idiosyncrasy as to whether a verb root requires
the applicative to co-occur with a phrase expressing Spatial Goal and other types
of Location- related semantic roles such as Event Location, Location of a specific
participant of the event, Path, Source, etc.
– The semantic differences between a BC with a semantic role expressed as a prepo-
sitional phrase and the corresponding optional AC where that semantic role usually
becomes an object argument are still poorly understood. Nevertheless, optional ACs
involve notions such as greater involvement/agentivity of the S/A argument and
“achieved” goal/endpoint.
– Pragmatically, optional ACs are often used to topicalize an AppP.
– Although studies based on discourse corpora are limited, available evidence sug-
gests that optional ACs are typically used when the AppP is discourse-given.

Lookalikes
– Suffixal forms identical to and displaying the same morphophonogical behavior as
the applicative suffix have developed at least three valence-neutral functions.
– Semantically, morphology identical to reflexes of ✶-ɪd can add aspectual nuances
to the meaning of the verb root, including repetitiveness, completeness, thorough-
ness, excess, persistence, intensity, intentionality, among many others.
– The two valence-neutral discourse functions involve only Location(-related) AppPs
and are usually available only when in a given language and with a given root
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 743

applicative morphology is not required to introduce the Location(-related) AppPs

targeted by the discourse functions. These functions are: placing a Location(-re-
lated) AppP under narrow constituent focus, and widening/shifting the scope of a
Location(-related) AppP with respect to subject and object arguments.

Abbreviations
In the following, x stands for a number and parentheses indicate optionally present elements.

1 first person
2 second person
3 third person
apx agreement prefix of class x
appl applicative
aug augment
caus causative
clx noun class prefix of class x
conn connective
cop copula
dem demonstrative
dj disjoint verb form
fv final vowel
H
melodic H tone
hab habitual
imps impositive
intr intransitive
ipfv imperfective
loc locative
neg negation
x.obj object (of class x)
pass passive
pfv perfective
pl plural
prep preposition
prf perfect
prog progressive
prs present
pst past
rec reciprocal
refl reflexive
rem remote
sep separative
sg singular
x.subj subject (of class x)
tam(p) tense-aspect-mood(-polarity)
tr transitive
744 Sara Pacchiarotti

References
Alsina, Alex & Sam Mchombo. 1990. The syntax of applicatives in Chicheŵa: Problems for a Theta Theoretic
Asymmetry. Natural Language & Linguistic Theory 8(4). 493–506.
Bastin, Yvonne. 1983. La finale verbale -ide et l'imbrication en bantou. Tervuren: Royal Museum for Central
Africa.
Bastin, Yvonne, André Coupez & Michael Mann. 1999. Continuity and divergence in the Bantu languages:
Perspectives from a lexicostatistic study. Tervuren: Royal Museum for Central Africa.
Bearth, Thomas. 2003. Syntax. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 121–142.
(Routledge Language Family Series). London & New York: Routledge.
Beaudoin-Lietz, Christa, Derek Nurse & Sarah Rose. 2004. Pronominal object marking in Bantu. In Akinbiyi
Akinlabi & Oluseye Adesola (eds.), Proceedings of the World Congress of African Linguistics, New Brunswick
2003, 175–188. Cologne: Rüdiger Köppe.
Blanchon, Jean Alain. 1999. ‘Tone cases’ in Bantu group B.40. In Jean Alain Blanchon & Denis Creissels
(eds.), Issues in Bantu tonology, 37–82. Cologne: Rüdiger Köppe.
Bostoen, Koen. 2018. The Bantu expansion. In Thomas Spear (ed.), Oxford research encyclopedia of African
history. Oxford: Oxford University Press, Oxford Research Encyclopedias Online.
Bostoen, Koen, Sebastian Dom & Guillaume Segerer. 2015. The antipassive in Bantu. Linguistics 53(4).
731–772.
Bostoen, Koen & Rozenn Guérois. 2022. Reconstructing suffixal phrasemes in Bantu verbal derivation. In
Koen Bostoen, Gilles-Maurice de Schryver, Rozenn Guérois & Sara Pacchiarotti (eds.), Reconstructing
Proto-Bantu grammar, 343–383. Berlin: Language Science Press.
Bostoen, Koen & Joseph Koni Muluwa. 2014. Umlaut in the Bantu B70/80 languages of the Kwilu (DRC).
Transactions of the Philological Society 112(2). 209–230.
Bostoen, Koen & Léon Mundeke. 2011. The causative/applicative syncretism in Mbuun (Bantu B87, DRC):
Semantic split or phonemic merger? Journal of African Languages and Linguistics 32(2). 179–218.
Bostoen, Koen & Léon Mundeke. 2012. Subject marking and focus in Mbuun (Bantu, B87). Southern African
Linguistics and Applied Language Studies 30(2). 139–154.
Bostoen, Koen & Mark Van de Velde. 2019. Introduction. In Mark Van de Velde, Koen Bostoen, Derek
Nurse & Gérard Philippson (eds.), The Bantu languages, 1–13. 2nd edn. London: Routledge.
Bresnan, Joan. 1994. Locative inversion and the architecture of Universal Grammar. Language 70(1). 72–131.
Bresnan, Joan & Lioba Moshi. 1990. Object asymmetries in comparative Bantu syntax. Linguistic Inquiry
21(2). 147–185.
Bresnan, Joan & Lioba Moshi. 1993. Object asymmetries in comparative Bantu syntax. In Sam A. Mchombo
(ed.), Theoretical aspects of Bantu grammar, 47–92. Stanford: CSLI Publications.
Buell, Leston C., Kristina Riedel & Jenneke van der Wal. 2011. What the Bantu languages can tell us about
word order and movement. Lingua 121(5). 689–701.
Coupez, André , E. Evrard & Jan Vansina. 1975. Classification d’un échantillon de langues bantoues d’après la
lexicostatistique. Africana Linguistica 6. 131–158.
Crane, Thera Marie, Larry M. Hyman & Simon Nsielanga Tukumu. 2011. A grammar of Nzadi (B.865), a
Bantu language of the Democratic Republic of Congo with an appendix of Proto-Bantu-Nzadi sound
correspondences by Clara Cohen. Berkeley: University of California Publications.
Creissels, Denis. 1998. La problématique du circonstant dans une langue où il existe une forme applicative
du verbe (sur l’exemple du tswana). In Sylvianne Rémi-Giraud & André Roman (eds.), Autour du
circonstant, 115–137. Lyon: Presses Universitaires de Lyon.
Creissels, Denis. 2000. Typology. In Bernd Heine & Derek Nurse (eds.), African languages: An introduction,
231–258. Cambridge: Cambridge University Press.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 745

Creissels, Denis. 2004. Non-canonical applicatives and focalization in Tswana. Paper presented at the
Syntax of the World's Languages, Leipzig, Available at http://email.eva.mpg.de/~cschmidt/SWL1/
handouts/Creissels.pdf.
Creissels, Denis. 2006. Syntaxe générale, une introduction typologique, vol. 2: La phrase. Paris: Lavoisier.
Creissels, Denis , Gerrit J. Dimmendaal, Zygmunt Frajzyngier & Christa König. 2008. Africa as a
morphosyntactic area. In Bernd Heine & Derek Nurse (eds.), A linguistic geography of Africa, 86–150.
(Cambridge Approaches to Language Contact). Cambridge: Cambridge University Press.
Daeleman, Jan. 1966. Morfologie van naamwoord en werkwoord in het Kongo (Ntandu) met ontleding van het
foneemsysteem. Louvain: Katholieke Universiteit Leuven dissertation.
De Blois, K.F. 1970. The augment in the Bantu languages. Africana Linguistica 4. 85–165.
De Kind, Jasper & Koen Bostoen. 2012. The applicative in Cilubà grammar and discourse: A semantic goal
analysis. Southern African linguistics and applied language studies 30(1). 101–124.
Dik, Simon C. , Maria E. Hoffman, Jan R. de Jong, Sie Ing Djiang, Harry Stroomer & Lourens de Vries. 1981.
On the typology of focus phenomena. In Teun Hoekstra, Harry van der Hulst & Michael Moortgat
(eds.), Perspectives on functional grammar, 41–74. Dordrecht: Foris.
Downing, Laura J. & Larry Hyman. 2016. Information structure in Bantu languages. In Caroline Féry &
Shinichiro Ishihara (eds.), The Oxford handbook of information structure, 790–813. Oxford: Oxford
University Press.
Ebalantshim, Masuwan. 1980. Esquisse grammaticale de la langue idiŋ à Káantsa. Lubumbashi: Université
Nationale du Zaïre MA thesis.
Ellington, John Ernest. 1977. Aspects of the Tiene language. Madison: University of Wisconsin dissertation.
Good, Jeff Craig. 2005. Reconstructing morpheme order in Bantu: the case of causativization and
applicativization. Diachronica 22. 55–109.
Greenberg, Joseph Harold. 1963. The languages of Africa. Bloomington: Indiana University.
Grégoire, Claire. 1975. Les locatifs en bantou. Tervuren: Royal Museum for Central Africa.
Grégoire, Claire. 1998. L’expression du lieu dans les langues africaines. Faits des langues 11/12. 285–303.
Grégoire, Claire. 2003. The Bantu languages of the forest. In Derek Nurse & Gérard Philippson (eds.), The
Bantu languages, 349–370. London & New York: Routledge.
Grollemund, Rebecca, Simon Branford, Koen Bostoen, Andrew Meade, Chris Venditti & Mark Pagel. 2015.
Bantu expansion shows that habitat alters the route and pace of human dispersals. Proceedings of the
National Academy of Sciences of the United States of America 112(43). 13296–13301.
Guérois, Rozenn & Koen Bostoen. 2018. From separative to middle/passive polysemy in Cuwabo. Southern
African Linguistics and Applied Language Studies 36(3). 211–233.
Güldemann, Tom. 1999a. The genesis of verbal negation in Bantu and its dependency on functional
features of clause types. In Jean-Marie Hombert & Larry M. Hyman (eds.), Bantu historical linguistics:
Theoretical and empirical perspectives, 545–587. Stanford: CSLI Publications.
Güldemann, Tom. 1999b. Head-initial meets head-final: Nominal suffixes in Eastern and Southern Bantu
from a historical perspective. Studies in African Linguistics 28(1). 49–91.
Güldemann, Tom. 2011. Proto-Bantu and Proto-Niger-Congo: Macro-areal typology and linguistic
reconstruction. In Osamu Hieda, Christa König & Hiroshi Nakagawa (eds.), Geographical typology and
linguistic areas with special reference to Africa, 109–141. Amsterdam: John Benjamins.
Güldemann, Tom. 2022. Predicate structure and argument indexing in early Bantu. In Koen Bostoen, Gilles-
Maurice de Schryver, Rozenn Guérois & Sara Pacchiarotti (eds.), Reconstructing Proto-Bantu grammar
387–421. Berlin: Language Science Press.
Guthrie, Malcolm. 1971. Comparative Bantu: An introduction to the comparative linguistics and prehistory of the
Bantu languages, Volume 2: Bantu prehistory, inventory and indexes. London: Gregg International.
Hammarström, Harald. 2019. An inventory of Bantu languages. In Mark Van de Velde, Koen Bostoen, Derek
Nurse & Gérard Philippson (eds.), The Bantu languages, 17–78. 2nd edn. Oxford: Routledge.
746 Sara Pacchiarotti

Heine, Bernd, H. Hoff & Rainer Voßen. 1977. Neuere Ergebnisse zur Territorialgeschichte der Bantu. In
Wilhelm J.G. Möhlig, Franz Rottland & Bernd Heine (eds.), Zur Sprachgeschichte und Ethnohistorie in
Afrika, 52–72. Berlin: Dietrich Reimer.
Hulstaert, Gustaaf. 1965. Grammaire du lomongo. Deuxieme partie, Morphologie. Tervuren: Royal Museum for
Central Africa.
Hulstaert, Gustaaf. 1966. Grammaire du lomongo. Troisième partie, Syntaxe. Tervuren: Royal Museum for
Central Africa.
Hyman, Larry M. 2003. Suffix ordering in Bantu: A morphocentric approach. In Geert Booij & Jaap van Marle
(eds.), Yearbook of Morphology 2002, 245–281. Dordrecht: Kluwer.
Hyman, Larry M. 2004. How to become a Kwa verb. Journal of West African Languages 30(2). 69–88.
Hyman, Larry M. 2007. Niger-Congo verb extensions: Overview and discussion. In Doris L. Payne &
Jaime Peña (eds.), Selected Proceedings of the 37th Annual Conference on African Linguistics, 149–163.
Somerville: Cascadilla Press.
Hyman, Larry M. 2008. Directional asymmetries in the morphology and phonology of words, with special
reference to Bantu. Linguistics 46(2). 309–350.
Hyman, Larry M. 2010. Affixation by place of articulation: The case of Tiene. In Jan Wohlgemuth & Michael
Cysouw (eds.), Rara & rarissima: Documenting the fringes of linguistic diversity, 145–184. Berlin: Mouton
de Gruyter.
Hyman, Larry M. 2014. Reconstructing the Niger-Congo verb extension paradigm: What’s cognate, copied
or renewed? In Martine Robbeets & Walter Bisang (eds.), Paradigm change: In the Transeurasian
languages and beyond, 103–125. Amsterdam: John Benjamins.
Hyman, Larry M. & Alessandro Duranti. 1982. On the object relation in Bantu. In Sandra A. Thompson &
Paul J. Hopper (eds.), Studies in transitivity, 217–239. New York: Academic Press.
Hyman, Larry M., Alessandro Duranti & Malillo Morolong. 1980. Towards a typology of the direct object
in Bantu. In Luc Bouquiaux (ed.), L’expansion Bantoue: Actes de Colloque International de CNRS Viviers
(France): 4–16 avril 1977, vol. 2, theme 3 (Problèmes spécifiques de grammaire comparée du bantou :
Innovations dans le système grammatical: Dérivation, formes verbales, système des classes nominales),
563–582. Paris: SELAF.
Hyman, Larry M. & Sam Mchombo. 1992. Morphotactic constraints in the Chichewa verb stem. In Laura
A. Buszard-Welcher (ed.), Proceedings of the Eighteenth Annual Meeting of the Berkeley Linguistics
Society: General session and parasession on the place of morphology in a grammar, 350–364. Berkeley:
ELanguage.
Hyman, Larry M. & Al Mtenje. 1999. Prosodic morphology and tone: The case of Chichewa. In Harry van
der Hulst, René Kager & Wim Zonneveld (eds.), The prosody-morphology interface, 90–133. Cambridge:
Cambridge University Press.
Kawasha, Boniface Kaumba. 2003. Lunda grammar: A morphosyntactic and semantic analysis. Eugene:
University of Oregon dissertation.
Kießling, Roland. 2004. Kausation, Wille und Wiederholung in der verbalen Derivation der westlichen
Ring-Sprachen (Weh, Isu). In Raimund Kastenholz & Anne Storch (eds.), Sprache und Wissen in Afrika:
Beiträge zum 15. Afrikanistentag, Frankfurt am Main und Mainz: 30. September – 2. Oktober 2002, 159–181.
Cologne: Rüdiger Köppe.
Kießling, Roland. 2021. Looking for areal features in the Grassfields/Bantoid region. Paper presented at the
West-central African linguistic history between Macro-Sudan Belt and Niger-Congo: commemorating
the 100th anniversary of the Berlin professorship for African languages and the legacy of Diedrich
Westermann, Humboldt-Universität zu Berlin, Institut für Asien- und Afrikawissenschaften.
Kimenyi, Alexandre. 1980. A relational grammar of Kinyarwanda. Berkeley: University of California Press.
Kimenyi, Alexandre. 1988. Passives in Kinyarwanda. In Masayoshi Shibatani (ed.), Passive and voice,
355–386. Amsterdam: John Benjamins.
 21 The polyfunctional applicative ✶-ɪd in Bantu languages 747

Kisseberth, Charles W. & Mohammad Imam Abasheikh. 1977. The object relation in Chi-Mwi:ni, a Bantu
language. In Peter Cole & Jerrold M. Sadock (eds.), Syntax and semantics, Vol. 8: Grammatical relations,
179–218. New York: Academic Press.
Koni Muluwa, Joseph & Koen Bostoen. 2007. Un recueil de proverbes nsong (RD Congo, bantu B85d).
Annales Aequatoria 28. 521–578.
Koni Muluwa, Joseph & Koen Bostoen. 2012. La diphtongaison dans les langues bantu B70–80 (Bandundu,
RDC): typologie et classification historique. Africana Linguistica 18. 355–386.
Koni Muluwa, Joseph & Koen Bostoen. 2014. The immediate before the verb focus position in Nsong (Bantu
B85d, DR Congo): A corpus-based exploration. ZAS Papers in Linguistics 57. 123–135.
Kuperus, Juliana. 1985. The Londo word: Its phonological and morphological structure. Tervuren: Royal
Museum for Central Africa.
Kuperus, Julianna & Antoinette Mpunga wa Ilunga. 1990. Locative markers in Luba. Tervuren: Royal Museum
for Central Africa.
Lukusa, Stephen Tchidiessa-Mulami. 1993. The imbrication of extensions in Ciluba. Afrikanistische
Arbeitspapiere 36. 55–77.
Mabugu, Patricia Ruramisai. 2001. Polysemy and the applicative verb construction in Chishona. Edinburgh:
University of Edinburgh dissertation.
Machobane, Malillo Matšepo. 1989. Some restrictions on the Sesotho transitivizing morphemes. Montreal:
McGill University dissertation.
Maho, Jouni Filip. 2009. NUGL Online: The online version of the new updated Guthrie List, a referential
classification of the Bantu languages (4 Juni 2009). Online file: http://goto.glocalnet.net/mahopapers/
nuglonline.pdf (accessed December 13, 2010).
Meeussen, Achiel Emiel. 1967. Bantu grammatical reconstructions. Africana Linguistica 3. 79–121.
Meeussen, Achiel Emiel. 1969. Bantu lexical reconstructions. Tervuren: Royal Museum for Central Africa.
Meinhof, Carl & Nicolaas Jacobus van Warmelo (eds.). 1932. Introduction to the Phonology of the Bantu
Languages Being the English Version of “Grundriss Einer Lautlehre Der Bantusprachen” by Carl
Meinhof, Translated, Revised and Enlarged in Collaboration with the Author and Dr. Alice Werner by
N. J. Van Warmelo. Dietrich Reimer/Ernst Vohsen, published under the auspices of The International
Institute of African Languages and Cultures, The Carnegie Corporation of New York, and The
Witwatersrand Council of Education, Johannesburg.
Misago, Manoah-Joël, Ernest Nshemezimana, Koen Bostoen & Sara Pacchiarotti. forthcoming. The
applicative and the focalization of Event Location in Rundi (Bantu JD62). In Danh-Thành Do-Hurinville,
Daniel Petit & Huy-Linh Dao (eds.), L’applicatif dans les langues: regard typologique. Paris: Éditions de la
Société de Linguistique de Paris.
Mukarovsky, Hans G. 1976–1977. A study of Western Nigritic. Wien: Institut für Ägyptologie und Afrikanistik,
Universität Wien.
Ndonga Mfuwa, Manuel. 1995. Systématique grammaticale du kisikongo (Angola). Paris: Université René
Descartes Paris 5 dissertation.
Nsuka-Nkutsi, François. 1982. Les structures fondamentales du relatif dans les langues bantoues. Tervuren:
Royal Museum for Central Africa.
Nurse, Derek & Gérard Philippson (eds.). 2003. The Bantu languages. London & New York: Routledge.
Pacchiarotti, Sara. 2020. Bantu applicative constructions. Stanford: CSLI.
Pacchiarotti, Sara. 2020. On the reconstructable main clause functions of Proto-Bantu applicative ✶-ɪd.
In Gilles-Maurice De Schryver Koen Bostoen, Rozenn Guérois, Sara Pacchiarotti (ed.), Reconstructing
Proto-Bantu grammar, 309–341. Berlin: Language Science Press.
Pacchiarotti, Sara & Koen Bostoen. 2021. The evolution of the noun class system of Ngwi (West-Coastal
Bantu, B861, DRC). Language in Africa 2(3). 11–67.
Rapold, Christian. 1997. The applicative construction in Lingala. Leiden: Universiteit Leiden MA thesis.
748 Sara Pacchiarotti

Rugemalira, Josephat. 1991. What is a symmetrical language? Multiple object constructions in Bantu.
Proceedings of the annual meeting of the Berkeley Linguistics Society 17. 200–209.
Schadeberg, Thilo. 1995. Object diagnostics in Bantu. In E. N. Emenanjo & Ozo-Mekuri Ndimele (eds.), Issues
in African languages and linguistics: Essays in honour of Kay Williamson, 173–180. Aba: National Institute
for Nigerian Languages.
Schadeberg, Thilo & Koen Bostoen. 2019. Word Formation. In Mark Van de Velde, Koen Bostoen, Derek
Nurse & Gérard Philippson (eds.), The Bantu languages, 2nd edn., 172–203. New York: Routledge.
Schadeberg, Thilo C. 1980. La morphologie verbale du bantu commun et les langues bantoues du
Cameroun. In Luc Bouquiaux (ed.), L’expansion bantoue. Actes du Colloque International du CNRS,
Viviers (France), 4–16 avril 1977, Volume 2, 503–509. (Colloques internationaux du Centre national de la
recherche scientifique). Paris: Société des Etudes Linguistiques et Anthropologiques.
Schadeberg, Thilo C. 1986. Tone cases in Umbundu. Africana Linguistica 10. 423–447.
Schadeberg, Thilo C. 2003a. Derivation. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages,
71–89. (Routledge Language Family Series). London & New York: Routledge.
Schadeberg, Thilo C. 2003b. Historical linguistics. In Derek Nurse & Gérard Philippson (eds.), The Bantu
languages, 143–163. London & New York: Routledge.
Sharman, John Campton. 1963. Morphology, morphophonology and meaning in the single–word verb-forms in
Bemba. Pretoria: University of South Africa dissertation.
Stappers, Leo. 1971. Esquisse de la langue lengola. Africana Linguistica 5. 255–307.
Stappers, Leo. 1986. Boma: eine Sprachskizze. Hamburg: Helmut Buske.
Trithart, Mary Lee. 1977. Locatives. In Ernest Rugwa Byarushengo, Alessandro Duranti & Larry M. Hyman
(eds.), Haya grammatical structure, 89–98. (Southern California Occasional Papers in Linguistics 6). Los
Angeles: Linguistics Department, University of Southern California.
Trithart, Mary Lee. 1983. The applied suffix and transitivity: A historical study in Bantu. Los Angeles: University
of California dissertation.
Van de Velde, Mark. 2019. Nominal morphology and syntax. In Mark Van de Velde, Koen Bostoen, Derek
Nurse & Gerard Philippson (eds.), The Bantu languages, 237–269. (Routledge Language Family Series).
London: Routledge.
Van de Velde, Mark. 2008. A grammar of Eton. Berlin: Mouton de Gruyter.
van der Wal, Jenneke & Larry M. Hyman (eds.). 2017. The conjoint/disjoint alternation in Bantu. Berlin & New
York: De Gruyter.
Vansina, Jan. 1995. New Linguistic Evidence and the Bantu Expansion. Journal of African History 36(2).
173–195.
Wald, Benji. 2022. On reconstructing the Proto-Bantu object marking system. In Koen Bostoen, Gilles-
Maurice de Schryver, Rozenn Guérois & Sara Pacchiarotti (eds.), Reconstructing Proto-Bantu grammar
423–463. Berlin: Language Science Press.
Williamson, Kay & Roger Blench. 2000. Niger-Congo. In Bernd Heine & Derek Nurse (eds.), African
Languages: An Introduction, 11–42. Cambridge & New York: Cambridge University Press.
Sylvie Voisin and Denis Creissels
22 B-applicatives and I-applicatives in Atlantic
languages (Niger-Congo)
Abstract: This chapter provides an overview of applicative constructions in a sample
of 27 languages belonging to the Atlantic branch of the Niger-Congo family. Most of the
languages of the sample have two distinct applicative suffixes, and none of them has an
applicative suffix that would be equally available to license applied phrases express-
ing the roles of beneficiary and instrument. After a relatively detailed description of
the applicative constructions of Wolof (a language with two applicative suffixes) and
Jóola Fóoñi (a languages with just one applicative suffix, in which the applicative strat-
egy is not used to encode beneficiaries), the chapter discusses possible generalizations
concerning the Atlantic applicatives that have the ability to license applied phrases
expressing the role of beneficiary (B-applicatives) and those that have the ability to
license applied phrases expressing the role of instrument (I-applicatives).

1 Introduction
This chapter is devoted to a typological investigation of applicative constructions in
Atlantic languages.1 An interesting particularity of Atlantic languages is that they share
the following limitation to the polysemy of applicative markers: none of the applicative
markers attested in Atlantic languages is found both in benefactive and instrumental
applicative constructions. At the same time, applicative markers exclusively found in
functions other than benefactive or instrumental are rare. This suggests organizing the
search for regularities in the properties of Atlantic applicatives with a focus on possi-
ble contrasts in the behavior of the applicative markers that have the ability to license
applied phrases in benefactive role (B-applicative markers) and those having the ability
to license applied phrases in instrumental role (I-applicative markers).

1 In this paper, we adopt the delimitation of the Atlantic language family put forward by Konstantin
Pozdniakov and Guillaume Segerer (Pozdniakov and Segerer, forthcoming). Atlantic languages as delim-
ited by Konstantin Pozdniakov and Guillaume Segerer are a proper subset of the Atlantic language fam-
ily as delimited by Greenberg and Sapir. There is consensus that the languages classified as “Atlantic” by
Greenberg and Sapir belong to the Niger-Congo macro-family, but Pozdniakov and Segerer argue that
there is no convincing evidence that they constitute a genetic unit within Niger-Congo. The label Atlantic
as they use it applies to a subset of Greenberg/Sapir’s “Atlantic” that excludes Mel languages and a few
other languages that, in the present state of our knowledge, are best considered as Niger-Congo isolates.
As reflected in Table 1, two branches of (New) Atlantic are recognized: Northern and Bak.

https://doi.org/10.1515/9783110730951-022
750 Sylvie Voisin and Denis Creissels

The chapter is organized as follows. Section 2 provides some basic information

about clause structure in Atlantic languages and introduces the distinction between
B-applicatives and I-applicatives. Section 3 consists of two sub-sections in which we
describe the applicative constructions of two languages for which we have detailed
first-hand data: Wolof and Jóola Fóoñi. Section 4 and Section 5 are devoted to the dis-
cussion of possible generalizations about B-applicatives and I-applicatives, respectively.
Section 6 summarizes the main conclusions.

2 Clause structure in Atlantic languages, and first

observations on Atlantic applicatives
Atlantic languages are among the languages for which the recognition of a grammatical
relation “subject” conflating A (the term of the transitive construction representing the
agent of prototypical transitive verbs) and S (the sole core NP in clauses projected by
monovalent verbs) is unproblematic. A grammatical relation “object” corresponding to
the typological notion of P can also be recognized. The Atlantic languages do not have
a grammatical relation “indirect object” comparable to that found in most European
languages, but make extensive use of multiple-object constructions in which two or
more NPs show coding characteristics identical to those of the object in monotransitive
constructions.2
In all Atlantic languages, a distinctive property of the subject NP (as opposed to the
other nominal terms of verbal clauses) is its fixed position immediately before the verb,
contrasting with the default postverbal position of objects and obliques. In most Atlan-
tic languages if not all, subjects have indexation properties which clearly distinguish
them from objects and obliques, but there is important cross-linguistic variation in the
details of subject and object indexation. In general, subjects and objects are unflagged,
whereas obliques are standardly flagged by prepositions.
With very few exceptions, Atlantic languages have multiple-gender systems with
the kind of relationship between noun morphology and the division of nouns into
genders typically found in several branches of the Niger-Congo family, traditionally
described in terms of “noun classes”.
In the languages in which the recognition of a grammatical relation “subject” is
unproblematic, applicatives can be defined as V>V derivations in which the derived
verb occurs in constructions characterized by a change in the mapping of semantic
roles onto syntactic roles meeting the following two conditions:

2 This definition does not imply the lack of hierarchy in the behavioral properties of the objects in
multiple-object constructions. However, in this respect, we are not in a position to propose any general-
ization about the multiple-object constructions of Atlantic languages.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 751

– the subject of the derived verb expresses the same semantic role as that of the base
verb;
– the construction of the derived verb differs from that of the base verb in the expres-
sion of a semantic role that cannot be encoded as an object in the construction of
the base verb;3 the phrase expressing the semantic role in question in the construc-
tion of the derived verb can be designated as the applied phrase.

Two subtypes of constructions meeting this definition must be distinguished:

– in optional applicatives, the semantic role expressed by the applied phrase can
also be expressed in monoclausal constructions whose nucleus is the base verb,
but with a non-core coding distinct from its coding in the applicative construction;
– in obligatory applicatives, the semantic role expressed by the applied phrase cannot
be expressed at all in monoclausal constructions whose nucleus is the base verb.

An important aspect of the broad definition of applicatives adopted in this volume is

that it does not restrict the notion of applicative construction to constructions in which
the applied phrase fulfills the syntactic role of object. The only condition is that, if the
semantic role expressed by the applied phrase can be encoded in the construction of
the base verb, it requires a non-core coding distinct from its coding in the applicative
construction.
Another important aspect of the definition formulated above is that it leaves open
the possibility that the derived verbs identifiable as fulfilling an applicative function in
some of their uses also have other functions. It is indeed very common that the same
derived verb forms are found both in constructions meeting the definition of applica-
tive constructions formulated above, and in constructions in which they fulfill other
functions, either related to valency change or not.
The data we have gathered on Atlantic applicatives include no applicative marker
having the ability to license both applied phrases expressing the role of beneficiary and
applied phrases expressing the role of instrument, and very few applicative markers
used exclusively for semantic roles other than beneficiary and instrument. Conse-
quently, a general discussion of Atlantic applicatives can conveniently be based on the
distinction between B-applicatives or I-applicatives, defined as follows:
– B-applicative markers may license applied phrases expressing the role of benefi-
ciary (and possibly other semantic roles), but not the role of instrument;
– I-applicative markers may license applied phrases expressing the role of instru-
ment (and possibly other semantic roles), but not the role of beneficiary.

3 This condition is essential to distinguish applicative constructions from antipassive constructions,

since applicative and antipassive constructions share the lack of change in the semantic role assigned
to the subject.
752 Sylvie Voisin and Denis Creissels

As summarized in Table 1, B-applicative markers are found in all the languages of our
sample, with the exception of those belonging to the Jóola group. I-applicative markers
are attested in the vast majority of the languages of our sample (21 out of 27), but the
languages whose description does not mention the existence of an I-applicative marker
are scattered across various branches of the Atlantic family. Applicative constructions
involving markers exclusively used to license applied phrases with roles other than
beneficiary and instrument are attested in 5 languages.4

Table 1: B-applicatives and I-applicatives in Atlantic languages.

B-appl I-appl others

North Wolof Wolof -al -e –

Nyun-Buy Nyun Guñaamolo -Vr -um –

Gubëeher -ur -um -un
Gujaahar -d (-it, -in) – –

Buy Kobiana -ar -Vm -al

Tenda-Jaad Tenda Bedik -ɤ́n -ál –

Basari -ʌn (-ə́n) – –
Konyagi -nə́l̰ -xǽl -ryǽry, -ryéry, -ǽry 5

Jaad-Biafada Jaad -ii, -rik -een –

Biafada -iig -en –

Fula-Seereer Fula -an -ir (-r, -d, -or) –

Seereer -an -it –

Cangin Noon -iɗ -oh –

Laalaa -eɗ (-iɗ) -oh –
Saafi -iɗ -oh –
Palor -ɪd -aʔ –

4 The sources we consulted for languages other than Wolof and Jóola Fóoñi are as follows. Guñaam-
olo: Bao Diop (2017), Bodian (2017); Gubëeher: Cobbinah (2013); Gujaahar: Goudiaby (2017); Kobiana:
Doneux (1991); Bedik: Ferry (1991); Basari: Ferry (1991), Perrin (2019); Konyagi: Ferry (1991), Santos
(1996); Jaad: Ducos (1971), Meyer (2001); Biafada: Wilson (1993); Fula: Arnott (1970), McIntosh (1984);
Seereer: Faye (1979), Renaudier (2012); Noon: Soukka (2008), Lopis-Sylla Jeanne (2010), Wane (2017);
Laalaa: Dieye (2011); Saafi: Mbodj (1983), Stanton (2011), Pouye (2015); Palor: D’Alton (1987); Ndut: Mor-
gan (1996); Nalu: Seidel (Forthcoming); Jóola Esuulaalur: Sambou (1979); Jóola Banjal: Bassène (2007);
Jóola Kujireray: Watson (2015); Karon: Sambou (2007); Kuwaataay: Coly (2010); Bayot: Diagne (2009);
Manjaku: Karlik (1972); Mankanya: Trifkovic (1969), Gaved (2020); Ganja: Creissels and Biaye (2016);
Bijogo: Segerer (2002).
5 In Santos’ (1996: 36) description of Konyagi, ry is the transcription of a single phoneme she describes
as a particular trill phonemically distinct from plain r.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 753

Table 1 (continued)

B-appl I-appl others

Ndut -iɗ -aʔ 6 –

Nalu Nalu -er (-ir, -ar) – –
Bak Jóola Central Jóola Fóoñi – -úm -en
Esuulaalur – -úm –
Banjal – – -en
Kujireray – – –
Jóola outliers Karon – -anan –
Kuwaataay – – –
Bayot – – –
Manjaku-Mankanya Manjaku -ar -na –
Mankanya -ar -na –

Balant Ganja -Vd – –

Bijogo Bijogo -an -at –

In this chapter, we present and analyze data showing that the operational definition
of B-applicatives and I-applicatives formulated above helps capture interesting gener-
alizations about Atlantic applicatives, since in Atlantic languages, B-applicatives and
I-applicatives as we define them consistently differ in several important respects.

3 Two case studies

3.1 Wolof

3.1.1 Introductory remarks

Wolof has two applicative suffixes, the B-applicative marker -al and the I-applicative
marker -e. In its benefactive function, the B-applicative marker occurs in an obligatory
applicative construction, whereas in its instrumental function, the I-applicative marker
occurs in an optional applicative construction. In (1), there is no alternative way to
mention a beneficiary in a clause projected by jénd ‘buy’,7 whereas in (2), the instru-

6 In Ndut, the applicative use of -aɁ can be characterized as vestigial (see § 5.6.2).
7 Church (1981) and Njie (1982) mention the possibility of expressing beneficiaries by means of a loc-
ative preposition in a non-applicative construction, but we never came across such a construction in
texts spontaneously produced by Wolof speakers. The problem is that Wolof is so widely used as a lingua
franca that it is sometimes difficult to draw a dividing line between traditional Wolof as spoken in rural
areas and usages found in Urban Wolof but deemed incorrect by traditional speakers.
754 Sylvie Voisin and Denis Creissels

ment can be encoded either as an applied object (2b), or as an oblique introduced by

the preposition ak (2c). In both cases, the applied phrase is syntactically an object, and
applicativization does not affect the expression of the other participants, which is made
possible by the high productivity of double-object constructions in Wolof.

(1) Wolof (Diouf 2003)

a. Woto b-u xonq la jénd.
car clB-rel be.red focO/X.sI:3sg buy
‘He bought a red car.’
b. Woto b-u xonq la jénd-al jabar-am.
car clB-rel be.red focO/X.sI:3sg buy-appl wife-poss:3sg
‘He bought a red car for his wife.’
c. ✶Woto b-u xonq la jénd (prep) jabar-am.
car clB-rel be.red focO/X.sI:3sg buy prep wife-poss:3sg
Intended: ‘He bought a red car for his wife.’

(2) Wolof (Diouf 2003)

a. Dafa dem c-a kër coro l-i.
focV.sI:3sg go loc-dist house girlfriend clL-d.prox
‘He went to her girlfriend’s place.’
b. Woto la dem-e.
car focO/X.sI:3sg go-appl
‘He went by car.’
c. Boo fa-y dem ak woto,
when.sI:2sg there-icpl go with car
dafa-y mel ni l-u jége.
focV.sI:3sg-icpl resemble like clL-rel be.near
‘If you go there by car, it seems to be close.’

Both applicative markers can license applied phrases expressing other roles than those
illustrated in (1) and (2). However, depending on the semantic role it expresses, the applied
phrase does not always display the properties expected from a direct object. There is also
variation in the obligatory or optional nature of the applicative construction.

3.1.2 Other uses of the B-applicative marker -al

3.1.2.1 -al and the expression of roles semantically close to the role of beneficiary
In addition to the role of beneficiary already illustrated above, the B-applicative
marker licenses applied phrases expressing other roles that can be subsumed under
the general notion of orientation of the action, such as goal (3), purpose (4), or con-
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 755

cernee (5).8 Applied phrases in constructions involving the B-applicative marker may
also express roles that meet this general characterization but are difficult to classify in
general terms because they are tightly bound to the lexical meaning of the verb, as in
(6). In all cases, the applied phrase displays the same behavior as the object of underived
transitive verbs, and there is no other means but the applicative to introduce these roles.

(3) Wolof (Diouf 2003)

Soo def-al-ul màngo s-i cëslaay, dina daanu.
if.sI:2sg put-appl-neg mango.tree clS-d.prox stake fut.sI:3sg fall
‘If you don’t put a stake to the young mango tree, it will fall.’

(4) Wolof (Voisin-Nouguier 2002)

Waa dëkk b-i yépp dem-al-oon benn dëkkandoo
people.of village clB-d.prox all go-appl-pst one neighbor
santaane.
collective.work
‘All the villagers had gone to a neighboring village for a collective work.’

(5) Wolof (Diouf 2003)

Fomp-al ko lex y-i.
wipe-appl oI:3sg cheek clY-d.prox
‘Wipe his cheeks.’ (Lit. ‘Wipe him the cheeks.’)

(6) Wolof (Diouf 2003)

Sos-al nga ma.
say.bad.things-appl prf.sI:2sg oI:1sg
‘You bad-mouthed me.’

3.1.2.2 The comitative function of -al

The B-applicative marker -al also licenses applied phrases expressing the role of com-
panion. However, in contrast to its other uses, the comitative use of the B-applicative
marker -al is only possible if the applied phrase is focalized, questioned, or relativized,
as in (7a), otherwise the role of companion can only be expressed by an oblique phrase
introduced by the preposition ak ‘with’, without any verbal marking, as shown in (7b-c).
As seen in (7d), the preposition ak ‘with’ may also be maintained in the presence of -al,

8 Following Van de Velde (2020), we designate as concernee a person concerned by the event by virtue
of an inherent relationship (s)he has with a participant more directly involved in the event (the ‘con-
cern’). Concernees are commonly designated as ‘external possessors’. As discussed in detail by Van de
Velde (2020), concernee-concern constructions are prototypically motivated by a whole-part relation-
ship between the concernee and the concern.
756 Sylvie Voisin and Denis Creissels

in which case -al does not act as an applicative marker, but rather as a verbal marker
highlighting the pragmatic saliency of an oblique without modifying its syntactic status.
However, this construction is not accepted by all informants.

(7) Wolof (Voisin-Nouguier 2002)

a. Sama rakk laa génn-al.
poss:1sg younger.sibling focO/X.sI:1sg go.out-appl
‘It’s with my sister that I went out.’
b. ✶Génn-al naa sama rakk.
go.out-appl prf.sI:1sg poss:1sg younger.sibling
Intended: ‘I went out with my sister.’
c. Génn naa ak sama rakk.
go.out prf.sI:1sg with poss:1sg younger.sibling
‘I went out (with) my sister.’
d. ?Ak yow la soxna s-i di wax-al.
with 2sg focO/X.sI:3sg woman clS-d.prox icpl speak-appl
‘It’s with you that the woman is speaking.’

3.1.3 Other uses of the I-applicative marker -e

3.1.3.1 -e and the expression of roles semantically close to the role of instrument
In applicative constructions with the suffix -e, the applied phrases referring to roles
semantically close to the role of instrument (such as means [8–9], material [10], or price
[11]) behave syntactically as objects, and there are no other ways to express these roles.

(8) Wolof (Voisin-Nouguier 2002)

Alal-am j-i, amin w-u sell la ko
wealth-poss:3sg clJ-d.prox way clW-rel be.pure focO/X.sI:3sg oI:3sg
am-e.
get-appl
‘His fortune, he got it in an irreproachable way.’

(9) Wolof (Voisin-Nouguier 2002)

Gaal g-i, wiir la dem-e.
pirogue clG-d.prox sail focO/X.sI:3sg go-appl
‘It is under sail that the pirogue moves forward.’

(10) Wolof (Diouf 2003)

Ndaa l-i, ban lañu ko tabax-e.
pot clL-d.prox clay focO/X.sI:3pl oI:3sg build-appl
‘The pot was made from clay.’
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 757

(11) Wolof (Church 1981)

Ñaata nga leen jënd-e?
how.much sI:2sg oI:3pl buy-appl
‘How much did you pay for them?’

3.1.3.2 -e and the expression of source

The suffix -e also marks applicative constructions in which the applied phrase expresses
the role of source (12). Contrary to the instrumental use of -e, the applied phrase is a
prepositional phrase or an adverb displaying no object-like property, and there is no
alternative way of expressing the role of source.

(12) Wolof (Diouf 2003)

War nañu fi jél-e jal-u mbalit b-i.
have.to prf.sI:3pl here remove-appl heap-cstr garbage clB-d.prox
‘They should remove the garbage heap from here.’

3.1.3.3 -e and the expression of path

With verbs of motion, the suffixation of -e licenses applied phrases expressing the role
of path (perlative). In this construction, illustrated in (13b), the applied phrase is a prep-
ositional phrase introduced by a locative preposition or an adverb displaying no object-
like properties, and there is no alternative way of expressing the role of path.

(13) Wolof (Diouf 2003)

a. Dugg na c-i nég b-i.
enter prf.sI:3sg loc-prox room clB-d.prox
‘He entered the room.’
b. Dugg-e na c-i bunt b-i.
enter-appl prf.sI:3sg loc-prox door clB-d.prox
‘He entered through the door.’

3.1.3.4 -e and the expression of circumstantial roles (manner or location

of the event)
The use of the I-applicative marker -e may also be conditioned by the presence of
phrases that do not refer to participants, but to circumstances of the event: manner, as
in (14–15), or location of the event, as in (16). There is, however, an interesting contrast
with the uses of -e described in § 3.1.3.2–3.
As in the ablative and perlative uses of -e (see § 3.1.3.2–3), the phrases referring
to circumstances of the event are prepositional phrases or adverbs that do not show
object-like properties.
758 Sylvie Voisin and Denis Creissels

(14) Wolof (Voisin-Nouguier 2002)

Na def-e nan?
oblg.sI:3sg do-appl how
‘How should he do?’

(15) Wolof (Diouf 2003)

Naka lañu tëral-e eksame y-i
how focO/X.sI:3pl organize-appl exam clY-d.prox
‘How did they organize the exams?’

(16) Wolof (Voisin-Nouguier 2002)

C-i teraas b-i lañu-y lekk-e.
loc-prox terrace clB-d.prox focO/X.sI:3pl-icpl eat-appl
‘It’s on the terrace that they eat.’

Such examples could suggest that, in Wolof, manner phrases and location phrases are
treated as applied obliques in an applicative construction. However, the same phrases
in similar functions also occur in constructions in which the I-applicative marker is
not present. The problem is that the available data are not sufficient to draw a general
conclusion about the exact role of -e in such cases.
In the case of phrases expressing location, there is some evidence that the use of the
applicative marker might reflect nuances within the semantic role of location.
On the one hand, pairs of sentences such as (17a–b) suggest that, with the same verb
dëkk ‘live somewhere’, the absence of -e marks reference to a type of environment defined
in general terms (a), whereas the presence of -e marks reference to a specific place (b).

(17) Wolof (Diouf 2003, Voisin-Nouguier 2002)

a. Saafàndu c-i àll rekk la-y dëkk.
wild.dog loc-prox bush only focO/X.sI:3sg-icpl live
‘Wild dogs live only in the bush.’
b. Ñu ngi dëkk-e c-i jéeri j-i,
sI:3pl prog live-appl loc-prox inland clJ-d.prox
dex g-i sore na leen.
river clG-C.prox be.far prf.sI:3sg oI:3pl
‘They live inland, the river is far from them.

On the other hand, (18a–b) suggests that, with a verb like ‘pound’, the absence of -e may
mark participant location (a) as opposed to event location (b).9

9 See Pacchiarotti (this volume) for a discussion of possible relationships between applicativization and
the participant vs. event location contrast in Bantu.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 759

(18) Wolof (Church 1981)

a. Gis naa fu mu-y dëbb.
see prf.sI:1sg where sI:3sg-icpl pound
‘I saw the thing in which she pounds.’
b. Gis naa fu mu-y dëbb-e.
see prf.sI:1sg where sI:3sg-icpl pound-appl
‘I saw the place where she pounds.’

However, much more data than we have at our disposal would be necessary to confirm
the contrasts suggested by such examples. In the absence of more precise and systematic
information about the interaction between the lexical meanings of individual verbs, the
semantic nuances that circumstantial phrases may express, and the constraints on the
presence of the verbal suffix -e, we leave open the question of the extent to which the
use of -e related to the presence of a manner or location phrase can be analyzed as an
instance of applicativization or of optional highlighting of circumstantial phrases.

3.1.4 Lexicalized applicatives

Wolof has quite a few verbs that look like applicative verbs but are best analyzed as lex-
icalized applicatives. For example, the last vowel of jóge ‘come from’ could be the I-ap-
plicative marker -e, especially as the role of source is among those that can be expressed
by the applied phrase in the applicative constructions marked by -e. However, synchron-
ically, jóg can only mean ‘stand up’, whereas jóge has the general meaning ‘come from’,
without any reference to the change of posture expressed by jóg.

3.1.5 -al as a polysemous causative-applicative marker

The verbal suffix -al is used productively, not only as a B-applicative marker, but also as
a causative marker, as in (19).

(19) Wolof (Diouf 2003)

a. Paan b-i fees na.
basin clB-d.prox be.full prf.sI:3sg
‘The basin is full.’
b. Bul fees-al paan b-i.
proh.sI:2sg be.full-caus basin clB-d.prox
‘Don’t’fill the basin.’

With toog ‘sit’, the al-form toog-al can be interpreted as ‘seat’ (causation) or ‘represent’
(deputative benefaction, lit. ‘sit on behalf of s.o.’).
760 Sylvie Voisin and Denis Creissels

3.1.6 Co-expression patterns involving the I-applicative marker -e

A verbal suffix -e showing the same morphophonological behavior as the I-applicative

marker is also found with some verbs as a marker of reciprocalization (20), antipassiv-
ization (21), causativization (22), or state-passivization (23). It is unclear to what extent
these uses of valency-changing suffixes identical to the I-applicative marker -e can be
analyzed as reflecting grammaticalization from a common source, or are just cases of
accidental homonymy.

(20) Wolof (Diouf 2003)

a. Dafa-y nuyu boroom kër g-i.
focV.sI:3sg-icpl greet owner house clG-d.prox
‘He is greeting the householder.’
b. Ñaar-i sefdetaa yaa ngiy nuyoo.
two-cstr head.of.state sI:3pl prog greet.recp
‘The two heads of state are greeting each other.’ (nuyu-e > nuyoo)

(21) Wolof (Diouf 2003)

a. Xaj b-i dafa ko màtt.
dog clB-d.prox focV.sI:3sg oI:3sg bite
‘The dog bit him.’
b. Xaj b-i dafa-y màtt-e.
dog clB-d.prox focV.sI:3sg-icpl bite-antip
‘The dog bites.’

(22) Wolof (Diouf 2003)

a. Saxaar s-ee tax mu génn.
smoke clS-d.prox.focS cause sI:3sg go.out
‘He went out because of the smoke.’
b. Dama-y génn-e woto b-i.
focV.sI:1sg-icpl go.out-caus car clB-d.prox
‘I’m taking the car out.’

(23) Wolof (Diouf 2003)

a. Ub naa néeg b-i.
close prf.sI:1sg room clB-d.prox
‘I closed the room.’
b. Néeg b-i dafa ub-e bëccëg b-i yépp.
room clB-d.prox focV.sI:3sg close-stat day clB-d.prox all
‘The room remained closed all day long.’
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 761

3.1.7 Lexicalized uses of -e

There are also isolated verb pairs involving a suffix -e that are best analyzed in terms of
lexicalization. For example, in the case of des (intransitive) ‘remain somewhere’ / des-e
(transitive) ‘still have’, the subject of des is the figure in a figure-ground relationship,
whereas the subject of des-e is a possessor. This precludes an applicative analysis, since
by definition, applicativization does not affect the semantic role of the subject, and at
the same time, the relationship between the argument structures and coding frames
of des and des-e cannot be analyzed as an instance of any other general type of voice
alternation.

3.1.8 Summary

Leaving apart some isolated cases of verbs that are formed by means of suffixes identi-
cal to an applicative marker but occur in constructions lending themselves to no gener-
alization, it is possible to analyze as productive the following uses of the B-applicative
and I-applicative markers of Wolof:
– The semantic role of beneficiary can only be encoded by means of the B-applicative
marker in an obligatory applicative construction with the beneficiary encoded as
an applied object. The B-applicative marker is used in the same way for the expres-
sion of roles semantically close to the role of beneficiary.
– The semantic role of comitative can be encoded in the construction of the unde-
rived verb by means of the preposition ak; it can also be encoded in a construc-
tion involving the B-applicative marker, with the comitative phrase in the role of
applied object, but only if the comitative phrase is focalized, questioned, or relativ-
ized; moreover, for some speakers, it is possible to maintain the preposition in the
presence of the B-applicative marker, which brings into question the analysis of the
construction as an applicative construction.
– The semantic role of instrument can be encoded in the construction of the unde-
rived verb by means of the preposition ak, but also by means of the I-applicative
marker in an optional applicative construction with the instrument phrase in the
role of applied object. The I-applicative marker is used in the same way for the
expression of semantic roles close to the role of instrument, such as means or
material.
– The expression of the roles of ablative and perlative requires the I-applicative
marker in an obligatory applicative construction in which the applied phrase does
not display object-like properties.
– The use of the I-applicative marker may also be related to the presence of an oblique
expressing manner or location of the event, but further investigation would be nec-
essary to establish to what extent such constructions qualify as applicative con-
structions or not.
762 Sylvie Voisin and Denis Creissels

– A verbal suffix identical to the B-applicative marker is found in causative function,

whereas a verbal suffix identical to the I-applicative marker is found in reciprocal,
antipassive, causative, and state-passive function.

3.2 Applicative constructions in Jóola Fóoñi

3.2.1 Introductory remarks

Jóola Fóoñi does not have applicative constructions with the applied phrase expressing
the semantic role of beneficiary. In Jóola Fóoñi, beneficiaries are encoded as objects
indistinguishable from the objects representing the patientive argument of transitive
verbs. For example, in (24), nothing indicates whether the object index suffixed to the
verb must be interpreted as referring to the patient or to a beneficiary.10

(24) Jóola Fóoñi

Pan i-pos-ool.
fut sI:1sg-wash-I:clA
‘I’ll wash him/her.’ OR ‘I’ll do the washing for him/her.’

Jóola Fóoñi has two verbal suffixes that can mark applicative constructions: -úm and
-en.11 However, they differ considerably in terms of productivity.
The suffix -en is productive as a causative marker, but we are aware of only four
verbs, all denoting bodily excretions, with which it may have an applicative function.
With these verbs, -en yields derived transitive verbs whose object denotes the place
towards which the excretion is directed.12

10 If not otherwise indicated, the Jóola Fóoñi examples quoted in this chapter are from Denis Creissels
and Alain Christian Bassène’s work on a corpus of oral texts they transcribed and analyzed with the help
of Boubacar Sambou as part of a project of a reference grammar of Jóola Fóoñi.
11 In the Jóola orthography, the acute accent marks +ATR vowels. Jóola languages have a system of
vowel harmony that can be described by positing that every formative is either underlyingly speci-
fied as +ATR, or underlyingly unspecified for the ATR feature. The formatives underlyingly specified
as +ATR (for example, the verbal suffix -úm) are consistently realized with +ATR vowels, and tend to
spread the +ATR feature to neighboring formatives, whereas the vowels of the formatives underlyingly
unspecified for the +ATR feature are -ATR by default, but may acquire the +ATR feature in contact with
underlyingly +ATR formatives.
12 The same pattern is described by Cobbinah (2013: 256) in the Nyun language Gubëeher, where -un is
a productive causative marker also found in applicative function with the following verbs: sel ‘urinate’ >
sel-un ‘urinate on’, reej ‘defecate’ > reej-un ‘defecate on’, and loot ‘spit/vomit’ > loot-un ‘spit/vomit on’.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 763

Table 2: Applicative use of the Jóola Fóoñi causative suffix –en.

base verb en-derivation

-lac ‘vomit’ > -lac-en ‘vomit on’

-púus ‘spit’ > -púus-én ‘spit on’
-sur ‘urinate’ > -sur-en ‘urinate on’
-wooy ‘fart’ > -wooy-en ‘fart in the direction of’

The suffix -úm is a productive I-applicative marker whose use is described in detail in
the remainder of this section.

3.2.2 Optional applicative constructions with the applied phrase in object role

In Jóola Fóoñi, instruments can be encoded as obliques introduced by the multifunc-

tional preposition di without any modification of the verb form (25a), but they can alter-
natively be encoded in an applicative construction (25b) in which the applied phrase
fulfills the syntactic role of object, as evidenced by the fact that it has the form of a
canonical NP or pronoun bearing no mark of its function, and can be indexed on the
verb in the same way as the patients of prototypical transitive verbs. The presence of
the applied object has no incidence on the coding of the other participants.

(25) Jóola Fóoñi

a. E-jala-a-y u-y-e, di-yo ni-sof-e-m si-wol-a-s
sg-net-d-clE dem-clE-prox with-I:clE sI:1sg-catch-icpl-r/f pl-fish-d-clS
‘This net, it’s with it that I catch fish.’
b. E-jala-a-y u-y-e, y-oo ní-sóf-úm-é-m si-wol-a-s
sg-net-d-clE dem-clE-prox clE-PRO sI:1sg-catch-appl-icpl-r/f pl-fish-d-clS
same meaning as (a)

As illustrated by example (25), in the instrumental use of the applicative marker, the
applicative construction is particularly frequent if the instrument is focalized, although
this is by no means a strict constraint.
The semantic role of material from which something is made can also be expressed,
either as an oblique phrase introduced by the preposition di without any modification
of the verb form, or as the applied object in an applicative construction, as in (26).
764 Sylvie Voisin and Denis Creissels

(26) Jóola Fóoñi

Bu-sees-a-b, ká-búmp-á-k k-ati bu-saana-a-b
sg-potash-d-clB sg-ash-d-clK clK-gen sg-ceiba.tree-d-clB
kú-rók-úm-é-bo-m.
sI:clBK-produce-appl-icpl-I:clB-r/f
‘Potash is produced from the ashes of the ceiba tree.’

The semantic role of cause can also be expressed, either as a prepositional oblique, or as the
applied object in an applicative construction, but the preposition used in the non-applica-
tive construction (mati) is distinct from that used for the roles of instrument or material.

(27) Jóola Fóoñi

a. Mati u-y-e ku-tey-e.
because.of dem-clE-prox sI:clBK-run-cpl
‘They ran away because of that.’
b. U-y-e kú-téy-úm-é.
dem-clE-prox sI:clBK-run-appl-cpl
same meaning as (a)

3.2.3 Obligatory applicatives with the applied phrase in object role

This configuration is only found with two verbs, and no generalization seems to be
possible.
In the obligatory applicative construction of -roŋ ‘live’, the applied object refers to
the way the referent of the subject earns his/her livelihood. This can just be viewed as
an exception to the rule according to which the role of means is expressed by means of
applicative constructions with the applied phrase in oblique role (see § 3.2.4).

(28) Jóola Fóoñi13

Pan ú-róŋ-úm e-jeena-a-y y-ool-i.
fut sI:2sg-live-appl sg-sweat-d-clE clE-poss-I:2sg
‘You will leave by your sweat.’

The verb -lako ‘settle, sit, remain’ has an obligatory applicative construction with the
applied phrase in object role expressing the meaning ‘maintain (something which is already
in place), live with something’. The semantic role expressed by the applied phrase in this
construction does not seem to be analyzable as a particular case of a more general type of
semantic role regularly expressed by means of applicative constructions in Jóola Fóoñi.

13 Example taken from the Jóola Fóoñi version of Genesis.

 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 765

(29) Jóola Fóoñi

E-cil-e-y y-ati fucen, let u-ŋoolen-aa
inf-possess-d-clE CLe-gen yesterday fut.neg sI:1pl-be.able-incl
ú-lákó-úm-aa-yo jaat.
sI:1pl-stay-appl-incl-I:clE today
‘Yesterday’s ownership rules, we cannot maintain them today.’

3.2.4 Obligatory applicatives with the applied phrase in oblique role

In this configuration, the applied phrase is a prepositional oblique expressing a role

that cannot be expressed within the limits of a clause projected by the base form of the
same verb. Two productive uses can be distinguished, perlative and mediative.
As illustrated in (30), prepositional obliques referring to a place through/along
which something or someone moves must be licensed by the applicative form of the
verb denoting motion.

(30) Jóola Fóoñi

N-an i-nag-u-m e-jangoon-e-y, n-é-báal-úm di
clN-rel sI:1sg-hit-ep-r/f sg-cat-d-clE seq-sI:clE-jump-appl prep
e-palanteer-e-y e-jaw e-maal.
sg-window-d-clE Is:clE-go Is:clE-go.away
‘When I hit the cat, it escaped by jumping through the window.’

As illustrated in (31), the same construction is also fully productive in mediative func-
tion, ‘mediative’ referring to anything that may contribute to the realization of an event
involving the will of the participant encoded as the subject.

(31) Jóola Fóoñi

Di ma-lilla-a-m m-ool-a ná-pák-úm-é.
prep nn-intelligence-d-clM clM-poss-I:clA sI:clA-escape-appl-cpl
‘It is through his intelligence that he escaped.’

Semantically, the mediative use of -úm has obvious affinities with both the perlative use
and the instrumental use of the same marker. However, syntactically, the properties of
the mediative applicative are identical to those of the perlative applicative, and very
different from those of the instrumental applicative: the instrumental applicative is
optional, whereas the mediative applicative is obligatory, and the applied phrase is syn-
tactically an object in the instrumental applicative, whereas it is a prepositional oblique
in the mediative applicative.
766 Sylvie Voisin and Denis Creissels

In addition to the two productive cases of obligatory applicative constructions with

the applied phrase in oblique role, an isolated case of a similar construction is found
with the verb ‘come’, in which the applied phrase, introduced by the allative preposi-
tion bee, expresses purpose, as in (32).

(32) Jóola Fóoñi

Sanka-a-y é-jáw-úm-úló-ót bee fu-jamara-a-f ceb.
mosquito.net-d-clE sI:clE-go-appl-ven-neg all sg-rainy.season-d-clF only
‘The mosquito net is not only for the rainy season.’
(Lit. ‘The mosquito net did not come for the rainy season only.’)

3.2.5 Lexicalized applicatives

As already signaled in the previous sections, the applicative form of some Jóola Fóoñi
verbs can be analyzed as lexicalized in the sense that, in a construction whose analysis
as an applicative construction is uncontroversial, the semantic role expressed by the
applied phrase does not lend itself to any generalization. The verb pair kaan ‘do, make’ /
káan-úm ‘be careful with’ illustrates a more opaque case of lexicalization, in which
it is even difficult to imagine semantic shifts that might account for the relationship
between the meanings expressed synchronically by kaan and káan-úm.

3.2.6 Non-applicative use of the applicative marker -úm

In the presence of manner adverbs such as buu ‘how?’ or moomu ‘thus’, -úm can option-
ally be added to the verb without triggering any formal change in the construction, and
without any change in the denotative meaning. In this particular case, -úm does not
meet the definition of an applicative marker, since its presence does not correlate with
a change in the expression of semantic roles, and seems to just add some emphasis.
Moreover, in this use (and only in this use), -úm has an optional variant -óorúm.
Formally, it is tempting to decompose this variant as -oor + -úm, but semantically, this
decomposition is not very plausible, since otherwise, a formative -oor is only attested,
either as a middle marker mainly used in reciprocal function, or as the first formative
of the complex negative marker -oor-ut ‘not yet’, i.e., with meanings which can hardly
be related with the expression of emphasis on manner.

(33) Jóola Fóoñi

Ni-mam-maŋ bee e-manj buu ∅-káan-(óor)úm-é.
sI:1sg-want-asrt all inf-know how sI:cld-happen-emph-cpl
‘I want to know how this happened.’
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 767

3.2.7 Summary

Jóola Fóoñi has a productive I-applicative marker, and a very marginal applicative use of
the causative marker -en, but no B-applicative marker. Leaving apart some isolated cases
of verbs whose applicative form occurs in a construction lending itself to no generaliza-
tion, four productive uses of the I-applicative marker -úm can be recognized in Jóola Fóoñi:
– the semantic role of instrument can be encoded, without any verbal marking, by
means of the multifunctional preposition di, but also in an applicative construction
with the instrument phrase in the role of applied object;
– the semantic role of cause can be encoded, without any verbal marking, as a causal
adjunct introduced by the preposition mati ‘because of’, but also in an applicative
construction with the cause phrase in the role of applied object ;
– the expression of the semantic role of perlative requires an applicative construc-
tion in which the applied phrase is an oblique introduced by the multifunctional
preposition di;
– the expression of the semantic role of mediative also requires an applicative con-
struction in which the applied phrase is an oblique introduced by the multifunc-
tional preposition di.

4 Commonalities and differences among Atlantic

B-applicatives
4.1 The range of semantic roles expressed by the applied phrase
in applicative constructions involving a B-applicative marker
Unsurprisingly, given the cross-linguistic tendencies in the coding of benefactive and
closely related semantic roles, the Atlantic applicative markers licensing applied
phrases expressing the role of beneficiary stricto sensu are also commonly found
in applicative constructions in which the applied phrase refers to a variety of roles
that can be subsumed under the general notion of orientation of the action, such as
addressee, purpose, destination.
In Fula (Arnott 1970: 353–4), the B-applicative marker may license applied phrases
expressing not only the roles commonly encoded by means of a B-applicative marker,
but also the role of cause, which is more unexpected. Example (34) illustrates the causal
use of the Fula B-applicative marker -an.

(34) Gombe Fula (Arnott 1970: 354)

’O-maay-an-ii weelo.
sI:clO-die-appl-cpl hunger
‘He died from hunger.’
768 Sylvie Voisin and Denis Creissels

We came across no other unambiguous case of a causal use of a B-applicative marker,

but example (35) illustrates a use of the B-applicative marker of Seereer which is in fact
ambiguous between purpose and cause (in the event to which this example refers, the
referent of the subject was in search for money, and this caused his death). One may
hypothesize that this is the kind of context in which an applicative marker licensing
applied phrases referring to the orientation of the action may acquire a causal use.

(35) Seereer (Renaudier 2012: 177)

A-xon-an-a xaalis.
sI:3sg-die-appl-cpl money
‘He died for money.’

Wolof attests the possibility of applicative constructions involving a B-applicative

marker in which the applied phrase expresses the role of companion (comitative).
However, this seems to be rather exceptional among Atlantic languages. Apart from
Wolof, the only mention we have found of this possibility is in Cobbinah’s description of
Gubëeher, and according to Cobbinah (2013: 258), the B-applicative suffix of Gubëeher
-ur is used in comitative function with just one verb: dëëk ‘go’ > dëëk-ur ‘accompany’.

4.2 The syntactic role of the applied phrase in applicative

constructions involving B-applicative markers

The data we have gathered include no case of a B-applicative marker involved in an

applicative construction with the applied phrase showing evidence of a syntactic role
other than object.

4.3 B-applicative markers and the distinction between obligatory

and optional applicative constructions

In general, applicative constructions involving B-applicatives markers are obligatory

applicatives.14 Not all descriptions mention this property of B-applicatives explicitly, but
the absence of any mention of an adposition having a benefactive use in most descrip-
tions can be viewed as evidence that applicative constructions with an applied phrase
in the role of beneficiary are obligatory applicatives. Note however that two sources
mention the possible use of a locative preposition to encode beneficiaries in Wolof (see
footnote 5), and a benefactive preposition ure is found in Jaad.

14 As already mentioned in § 3.1.2.2, the B-applicative marker of Wolof behaves differently in comi-
tative function, but we came across no other clear case of a comitative use of a B-applicative marker.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 769

Interestingly, the Jóola languages do not have B-applicatives and have an allative
preposition, but they do not use this preposition productively in benefactive func-
tion, and they do not have a dedicated benefactive preposition either. As mentioned
in Section 2.4 for Jóola Fóoñi, in Jóola languages, beneficiaries are standardly encoded
as objects that nothing distinguishes from the objects representing the patientive argu-
ment of transitive verbs.

4.4 B-applicatives and the notion of valency increase

In Atlantic languages, apart from isolated cases that are best analyzed in terms of lexi-
calization, B-applicative markers are unambiguously valency-increasing. Multiple-object
constructions are very productive in Atlantic languages, and in their productive uses, B-ap-
plicative markers license applied objects whose introduction does not affect the expres-
sion of participants that may be encoded as objects in the coding frame of the base verb.

4.5 Other valency-related functions of B-applicatives

4.5.1 The applicative-causative co-expression pattern

The suffixes that meet the definition of B-applicative markers are also used as causative
markers in languages belonging to the following three groups of Atlantic languages:
Wolof (see § 3.1.6), Tenda (Bedik, Bassari), and Cangin (Laalaa, Saafi, Paloor, Ndut).
Example (36) illustrates this co-expression pattern in Ndut.

(36) Ndut (Morgan 1996: 111, 28)

a. Lah won-e Biram saam-iɗ ɗo gúm.
have say-imp Biram find-appl 2sg calabash
‘You must tell Biram to find you a calabash.’
b. Fu tëëk-iɗ ɗi laɁ-a, . . .
sI:2sg sit-caus 3sg rock-d
‘You sit him on the rock, . . .’

In Gubëeher (Nyun), according to Cobbinah (2013: 258), the productive B-applicative

suffix -ur is found in causative function with just one verb: jir ‘run’ > jidd-ur ‘conduct a
vehicle’.15

15 In West-African languages, the verbs glossed ‘run’ commonly have the wider meaning ‘go fast’
(which means that, for example, a boat may ‘run’), and the causative form of these verbs is commonly
used as the equivalent of English ‘ride’ (a horse or a vehicle).
770 Sylvie Voisin and Denis Creissels

There are also languages with a B-applicative marker formally similar to a caus-
ative marker, but not completely identical. Unfortunately, the available data are not
sufficient to discuss the historical significance of such similarities.
Jóola languages do not have productive B-applicatives, but their causative markers
also have a very marginal applicative use restricted to verbs referring to bodily excre-
tions (see § 3.2.1 on Jóola Fóoñi). One may wonder whether this might be the vestige
of a formerly productive applicative use of the causative markers in question, since
it is difficult to imagine a semantic shift from causative to this very particular type of
applicative meaning.

4.5.2 The applicative-reciprocal co-expression pattern

This co-expression pattern is attested in Mankanya and Bijogo, whose B-applicative

and reciprocal markers are formally identical. Example (37) illustrates this situation
in Mankanya.

(37) Mankanya (Gaved 2020: 72)

a. Mankañ a-fiŋ-ar u-pi Dama.
Mankanya sI:clA-kill-appl sg-goat Dama
‘Mankanya killed a goat for Dama.’
b. Ba-ntohi bik-i Bula ba-fiŋ-ar
pl-elder clBA-gen Bula sI:clB-kill-recp
‘The elders of Bula were killing each other.’

In Balant Ganja, the B-applicative marker and the reciprocal marker are formally
similar, but nevertheless distinct, at least from a synchronic point of view, and the his-
torical significance of this similarity is unclear.

4.6 Uses of B-applicative markers not related to valency

operations

In Wolof, the B-applicative marker in comitative function (and only in comitative func-
tion) may depart from the behavior expected from an applicative marker, and behave
as a verbal marker highlighting the saliency of an oblique phrase without changing
anything in the syntax, at least for some speakers (see § 3.1.2.2). However, the data we
have been able to gather include no other case of a construction in which a marker
that otherwise meets the definition of a B-applicative marker would not act as a valen-
cy-changing operator.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 771

5 Commonalities and differences among Atlantic

I-applicatives
5.1 The range of semantic roles expressed by the applied phrase
in applicative constructions involving an I-applicative marker

As already illustrated by Wolof and Jóola Fóoñi, in Atlantic languages, the applicative
markers licensing applied phrases in instrumental role are commonly also involved in
the expression of the following roles: material, means (mediative), source of motion
(ablative), or path (perlative). I-applicative markers licensing applied phrases express-
ing accompaniment are also attested.
Example (38) illustrates the involvement of the I-applicative marker of Fula in the
expression of means, source of motion, and path.

(38) Gombe Fula (Arnott 1970: 349)

a. Mi ’anndaa no ’o-wurt-or-ii.
1sg know.neg how sI:clO-go.out-appl-cpl
‘I don’t know how he got out.’
b. Ndiyam ’yiw-r-ii fuuna.
rain come-appl-cpl east
‘The rain came from the east.’
c. Naaŋge fuɗ-ir-ay fuuna, mut-ir-ay hiirna.
sun rise-appl-cpl east set-appl-cpl west
‘The sun rises in the east, sets in the west.’

According to Renaudier (2012: 190), the I-applicative marker of Seereer is productive in

ablative and comitative roles.
In Gubëeher (Nyun), according to Cobbinah (2013: 257), the applied phrase in the
construction of yaax-um ‘eat with’ < yaax ‘eat’ may refer not only to the instrument
used by the eater (spoon, etc.), but also to the side dish. With lód-um < lód ‘build’, the
applied phrase refers to the material used for the construction, whereas with fur-um <
fur ‘leave’, the applied phrase refers to the source of motion.
We saw in § 3.2.2 that the I-applicative marker of Jóola Fóoñi is productively used
in causal function. However, in the documentation we have gathered on Atlantic applic-
atives, the only other unambiguous example of an applicative marker fulfilling a causal
function concerns the B-applicative marker of Fula (see example (34) above).
Dieye (2010: 245) mentions a particular use of the I-applicative marker of Laalaa
(Cangin) in a special type of causative construction implying that the person acting as
the causer is at the same time the beneficiary of the action performed by an unmen-
tioned causee (autobenefaction).
772 Sylvie Voisin and Denis Creissels

The construction in question is marked by a verbal suffix -elok whose addition to

transitive verbs does not change anything in the syntax, but carries the following impli-
cations: in the presence of -elok, the subject is not interpreted as the immediate agent,
but rather as a causer, and at the same time as the beneficiary of the event denoted
by the verb, whereas the causee must remain implicit (39b). However, as illustrated in
(39c), the addition of the I-applicative marker -oh (surfacing as -o in (39c) for phono-
logical reasons) licenses an additional object phrase interpreted as expressing the role
of causee (or immediate agent). This particular use of an I-applicative marker can be
analyzed as motivated by the fact that, in the causative construction of transitive verbs,
the causee can be viewed as a kind of animate instrument.

(39) Laalaa (Dieye 2010: 231, 245)

a. Oomah-c-aa soob-en too-t-aa.
child-clC-d pound-prf millet-clT-d
‘The children pounded the millet.’
b. Clotilde soob-elok-en too-t-aa.
Clotilde pound-caus.autob-prf millet-clT-d
‘Clotildei had the millet pounded for heri.’
c. Clotilde soob-elok-o-en oomah-c-aa too-t-aa.
Clotilde pound-caus.autob-appl-prf child-clC-d millet-clT-d
‘Clotildei made the children pound the millet for heri.’

Several descriptions of Atlantic languages also mention manner and location of the event
among the semantic roles that can be expressed by the applied phrase in applicative
constructions involving an I-applicative marker. However, a closer look at the examples
they provide casts some doubt on this analysis. The problem is that most descriptions do
not distinguish manner from means, and the expression of location from other spatial
notions such as source or path. As already discussed for Wolof and Jóola Fooñi, once
these distinctions are taken into account, it turns out that means, source and path are
unquestionably possible semantic roles for the applied phrase in applicative construc-
tions involving an I-applicative marker, whereas one may have doubts about the exact
nature of the constructions in which the presence of a verbal marker otherwise analyz-
able as an I-applicative marker is conditioned by the presence of a phrase expressing
manner or location of the event.

5.2 The syntactic role of the applied phrase in applicative

constructions involving I-applicative markers

Bijogo illustrates the case of an I-applicative marker whose syntactic behavior, at least
with applied phrases in instrumental role, corresponds to what is commonly consid-
ered as the prototypical behavior of applicative markers: in Bijogo, instruments may be
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 773

encoded as obliques introduced by the instrumental preposition ta, or as objects of an

applicative verb form, but it is not possible to use the instrumental preposition and the
I-applicative marker in the same clause (Segerer 2002: 219). Note incidentally that the
formal resemblance between the I-applicative suffix and the instrumental preposition
suggests that, historically, the I-applicative suffix may have resulted from encliticization
of the instrumental preposition.

(40) Bijogo (Segerer 2002: 219)

a. ni-mεs ni-dεndɔk ta nɔ-ɔgɔ.
sg-knife sI:clNV.icpl-sharpen.mid with sg-stone
‘A knife can be sharpened with a stone.’
b. ni-mεs ni-dεndɔk-at nɔ-ɔgɔ.
sg-knife sI:clNV.icpl-sharpen.mid-appl sg-stone
same meaning as (a)

The same synonymy between a non-applicative construction in which the instru-

ment is encoded as a prepositional oblique and an applicative construction in which
it is encoded as an object is also observed in Wolof, Jóola Fóoñi (see § 3) and Seereer
(Renaudier 2012: 186–188).
However, the syntactic behavior of Atlantic I-applicatives does not always follow
this pattern. As already illustrated by Wolof and Jóola Fóoñi, depending on the semantic
role expressed by the applied phrase, the same I-applicative marker may occur both in
applicative constructions in which the applied phrase can be analyzed as an object, and
in applicative constructions in which the applied phrase has an oblique status.
The details of the relationship between the semantic role of the applied phrase and
its status as an object or an oblique vary from one language to another, at least to some
extent. Not all descriptions provide detailed data on this question, but the general ten-
dency in Atlantic languages seems to be that applied phrases expressing the semantic
role of instrument have a strong tendency to take the syntactic status of object, whereas
in some other uses of the I-applicative markers, applied phrases encoded as preposi-
tional obliques are more common. However, these are only tendencies, and variation
can be observed even with the role of instrument.
For example, in Fula, the expression of the role of instrument requires an applica-
tive construction, in which, however, the instrument may optionally be encoded as
noun phrase in object role or as a prepositional oblique.

(41) Gombe Fula (Arnott 1970: 348)

’O-haɓɓ-ir-ii. gujjo (’e) ɓoggol.
sI:clO-tie-appl-cpl thief with rope
‘He tied the thief with a rope.’
774 Sylvie Voisin and Denis Creissels

Interestingly, in closely related Laalaa and Noon, the applicative constructions with
an applied phrase in the role of instrument behave differently. In Laalaa (Dieye 2010:
245), in the same way as in Wolof, Jóola Fóoñi, Seereer, or Bijogo, the applicative con-
struction with the instrument phrase encoded as a syntactic object is in competition
with the coding of the instrument as a prepositional oblique without any specific verbal
marking, whereas in Noon, according to Wane (2017: 133), the coding of instruments
requires simultaneously the applicative marker -oh and the preposition në ‘with’.

(42) Noon (Wane 2017: 133)

Zan ñam-oh haawë në kutu.
Jean eat-appl couscous with spoon
‘Jean is eating couscous with a spoon.’

Constructions in which the coding of instruments involves both applicative marking

and the use of a preposition are found in two other languages of the Cangin branch:
Saafi (Stanton 2011: 49) and Palor (see below).
Example (43b) illustrates the use of prepositional phrases as applied phrases in the
ablative use of the Seereer I-applicative marker -it. The analysis of this construction as
an applicative construction follows from the fact that the source of motion cannot be
expressed in the construction of the base verb. However, the applied phrase is obliga-
torily flagged by a preposition, and its oblique nature is confirmed by the fact that it
cannot be indexed on the verb (Renaudier 2012: 183). This is all the more remarkable
given that, in the construction of the base verb, the same preposition is optional with
phrases denoting destination of motion, as in (43a).

(43) Seereer (Renaudier 2012: 183)

a. A-ret-a (na) marse.
sI:3sg-go-cpl prep market
‘S/he went to the market.’
b. A-ret-it-a na marse.
sI:3sg-go-appl-cpl prep market
‘S/he left the market.’

5.3 I-applicatives and the distinction between obligatory

and optional applicatives

As already illustrated by Wolof and Jóola Fóoñi, in Atlantic languages, there is important
variation in the optional or obligatory nature of applicative constructions involving
I-applicative markers, and this variation correlates with the semantic role expressed by
the applied phrase.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 775

For example, in Seereer, the same I-applicative marker -it is obligatory to express
the source of motion (43b) but optional to express instrument. By contrast, as already
mentioned, there are also Atlantic languages in which the applicative constructions
with an applied phrase expressing the role of instrument are obligatory applicatives.
As regards the instrumental use of I-applicative markers, several descriptions
mention that applicative constructions involving an I-applicative marker are in princi-
ple optional but tend to be preferred, and may even be required, when the instrument
is focalized or relativized.
As regards the ablative use of I-applicative markers, it is noteworthy that Atlantic
languages do not have ablative adpositions. This implies that phrases expressing the
role of source of motion can only be licensed, either by motion verbs that have the
ability to assign the role of source, or by applicative derivation, in constructions that,
consequently, meet the definition of obligatory applicatives.

5.4 I-applicatives and the notion of valency increase

It follows from the variation observed in the syntactic status of the applied phrase that
applicative constructions involving an I-applicative marker do not necessarily imply
an increase in valency, in the sense that the number of core syntactic terms in the con-
struction of the applicative verb is not necessarily greater than in the construction of
the base verb.

5.5 Other valency-related functions of I-applicatives

In Atlantic languages, markers identical to I-applicative markers may have valency-re-

lated functions other than applicative, most commonly (although not exclusively) recip-
rocal and/or antipassive.
Markers fulfilling a reciprocal function identical to I-applicative markers are
attested in Wolof, Noon, Laalaa, and Palor.
Markers fulfilling an antipassive function identical to I-applicative markers are
attested in Wolof, Seereer, Noon, Laalaa, and Palor.
A marker identical to the Wolof I-applicative marker -e is found not only with recip-
rocal and antipassive functions, but also with a causative function. However, its pro-
ductivity as a causative marker is relatively limited in comparison with the causative
marker identical to the Wolof B-applicative marker -al (see § 3).
776 Sylvie Voisin and Denis Creissels

5.6 Uses of I-applicative markers not related to valency

operations

With I-applicatives (in contrast to B-applicatives), it is relatively common that the same
verbal suffix has a valency-changing function in some of its uses, but also has uses
involving no apparent change in the expression of semantic roles. Unfortunately, as
already mentioned, the data provided by most descriptions are not sufficient to eval-
uate the exact extent of this phenomenon, even for a relatively well-documented lan-
guage such as Wolof.
In such cases, it is not always clear whether this is an instance of more or less free
variation, or the I-applicative marker fulfills a function not related to valency.

5.6.1 I-applicative markers and focalized location

In Noon, as in Wolof (see § 3.1.3.4), a marker identical to the I-applicative marker -oh
may co-occur with phrases expressing location, and according to Wane (2017: 132), it
then marks focalization of the phrase expressing location, without any change in its
coding properties.

(44) Noon (Wane 2017: 132)

a. Më en ngë kaan Lamin.
1sg be prep house Lamine
‘I am at Lamine’s place.’
b. Më en-oh ngë kaan Lamin.
1sg be-foc prep house Lamine
‘I am AT LAMINE’S PLACE.’

5.6.2 I-applicative markers and habitual aspect

Closely related Palor and Ndut (which together constitute a sub-branch of the Cangin
branch of Atlantic) share a verbal suffix -aɁ but show interesting contrasts in its possi-
ble functions.
In Palor, -aɁ is an I-applicative marker (possibly cognate with the preposition Ɂa
‘with’) productively used in an obligatory applicative construction in which the applied
phrase is a prepositional phrase expressing the role of instrument, as in (45).

(45) Palor (D’Alton 1987: 144)

Ɗi Ɂool-aɁ-te xar Ɂa paaka
3sg cut.the.throat-appl-cpl sheep with knife
‘He cut the sheep’s throat with a knife.’
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 777

In Ndut, according to Morgan (1996: 92–95), -aʔ shows only vestiges of a formerly pro-
ductive use as an I-applicative marker. Synchronically, the only uses of -aɁ that show
some productivity are its use to mark the saliency of a place or time adjunct, and its use
as a marker of habitual aspect, as in (46).

(46) Ndut (Morgan 1987: 144)

Ɗi lom-aɁ too.
3sg buy-hab millet
‘He buys (habitually) millet.’

5.6.3 I-applicative markers and the expression of manner

In the descriptions of Atlantic languages, the semantic role of manner is often men-
tioned among the semantic roles expressed by the applied phrase in constructions
involving the same verbal suffix as instrumental applicative constructions. However,
when detailed data on this use of I-applicative markers are available, they cast doubts
about the validity of an applicative analysis of this particular use of verbal suffixes oth-
erwise acting as applicative markers.
For example, in Seereer, according to Renaudier (2012: 183–184), the I-applicative
marker -it must be present if a manner adjunct is relativized or focalized, but the mere
presence of a manner adjunct does not require the use of the applicative form of the
verb, which confirms our own observations on Jóola Fóoñi and Wolof.
In fact, the exact nature of the relationship between manner adjuncts and the I-ap-
plicative markers in Atlantic languages could only be clarified on the basis of corpus
studies, and the only thing we can do here is to leave this question open.

5.7 I-applicative markers and nominalization

The use of the same suffixes, or of very similar suffixes, as applicative markers licensing
applied phrases in the role of instrument and as instrument nominalization markers is
found in Jóola languages, for example Jóola Fóoñi -sonten ‘heal’ > bú-sóntén-úm ‘medical
treatment’.
The same phenomenon is observed in Fula (Arnott 1970: 251). Moreover, in Fula,
the same suffix is also used to derive nouns referring to places dedicated to a particular
activity from verbs (as in loot-ir-de ‘place for washing (clothes, etc.)’ < loot-a ‘wash).
778 Sylvie Voisin and Denis Creissels

5.8 A particularity in the expression of the semantic role

of instrument in Laalaa

As already mentioned above, Laalaa (Cangin) has an I-applicative marker -oh licensing
the expression of the semantic role of instrument as an applied object. Interestingly,
Laalaa has another verbal suffix, -ah, licensing the expression of the semantic role of
instrument, but in the syntactic role of subject, as in (47).

(47) Laalaa (Dieye 2010: 206)

Fetal-aa ap-ah-an paloom.
gun-d kill-be.used.to-fut antelope
‘The gun will be used to kill antelopes.’

In the languages that have both instrumental-applicative and passive constructions

(which is the case of Laalaa), such a construction can be expected to involve the com-
bination of an applicative marker licensing an applied object with the semantic role of
instrument and a passive marker converting the applied object into the subject of an
applicative-passive construction. However, in Laalaa, -ah can hardly be decomposed
as -oh (I-applicative marker) + -uu (passive marker). Unfortunately, nothing similar is
evoked in the descriptions of the other Atlantic languages (even in the other languages
of the Cangin group), and consequently, we have nothing to propose about a possible
relationship between the two verbal suffixes of Laalaa involved in the expression of the
semantic role of instrument.

6 Conclusion
An interesting particularity of Atlantic languages is that most of them have applicative
constructions with applied phrases expressing the roles of beneficiary and instrument,
but at the same time, none of the applicative markers attested in Atlantic languages
has the ability to license applied phrases expressing the role of beneficiary and applied
phrases expressing the role of instrument. In this chapter, we have shown that recur-
rent contrasts can be observed between the applicative markers of Atlantic languages
that have the ability to license applied phrases in benefactive role and those having the
ability to license applied phrases in instrumental role:
– Apart from the role of beneficiary, B-applicative markers license almost exclusively
applied phrases expressing roles that can be subsumed under the general notion of
orientation of the action, whereas I-applicative markers commonly license applied
phrases expressing roles such as material, means, source, or path.
– Neither B-applicatives nor I-applicatives show a particular propensity to license
applied phrases expressing the role of companion (comitative).
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 779

– Applied phrases expressing the role of cause are rarely mentioned in descriptions
of Atlantic languages, but the examples we came across involve B-applicative as
well as I-applicative markers, depending on the individual languages.
– B-applicative markers license almost exclusively applied phrases in the syntactic
role of object, whereas oblique applied phrases are common with I-applicative
markers.
– The co-expression pattern in which B-applicative markers are most commonly
involved is the applicative-causative co-expression pattern, whereas I-applicative
markers are mainly involved in the applicative-reciprocal and applicative-antipas-
sive co-expression patterns.
– It is relatively common for I-applicative markers to be also found in constructions
in which they cannot be analyzed as marking a valency-changing operation; this is
much less common for B-applicative markers.

Abbreviations
all allative
antip antipassive
appl applicative
asrt assertion
autob autobenefactive
caus causative
clX class X,16
cpl completive
cstr construct form
d definite
dem demonstrative
dist distal
emph emphatic
ep epenthetic
foc focalization
focO/X object or oblique focalization
focS subject focalization
focV verb focalization
fut future
gen genitive
hab habitual
I index
icpl incompletive
imp imperative

16 “Class” refers to gender-number agreement patterns conventionally designated by capital letters

that evoke the phonological forms of agreement markers in the language in question.
780 Sylvie Voisin and Denis Creissels

incl inclusive
inf infinitive
loc locative
mid middle voice
neg negation
nn number-neutral noun prefix
oblg obligative
oI object index
pl plural
poss possessive
prep multifunctional preposition
prf perfect
prog progressive
proh prohibitive
prox proximal
pst past
recp reciprocal
rel relativizer
r/f marker of verb forms used exclusively in relative clauses or focus constructions
seq sequential
sg singular
sI subject index
stat stative
ven venitive

References
Arnott, David. W. 1970. The nominal and verbal systems of Fula. London: Oxford University Press.
Bao Diop, Sokhna. 2017. Le baynunk guñaamolo, une langue du sud du Sénégal. Paris: L’Harmattan.
Bassène, Alain-Christian. 2007. Morphosyntaxe du jóola banjal, langue atlantique du Sénégal. Cologne:
Rüdiger Köppe.
Bodian, Lamine. 2017. Morphosyntaxe du guñaamolo, parler baïnounk de Niamone. Munich: LINCOM.
Church, Eric. 1981. Le système verbal du wolof. Dakar: University of Dakar.
Cobbinah, Alexander. 2013. Nominal classification and verbal nouns in Baïnounk Gubëeher. London: SOAS
University of London dissertation.
Coly, Jules. 2010. Morphosyntaxe du kuwaataay (Sénégal). Cologne: University of Cologne dissertation.
Creissels, Denis & Séckou Biaye. 2016. Le balant ganja: phonologie, morphosyntaxe, liste lexicale, textes. Dakar:
IFAN.
D’Alton, Paula. 1987. Le palor, esquisse phonologique et grammaticale d’une langue cangin du Sénégal. Paris:
Editions du CNRS.
Dieye, El Hadji. 2011. Description d’une langue cangin du Sénégal : le laalaa (léhar). Dakar and Paris: UCAD and
INALCO dissertation.
Diouf, Jean-Léopold. 2003. Dictionnaire wolof-français et français-wolof. Paris: Karthala.
Doneux, Jean Léonce.1984. La place de la langue buy dans le groupe atlantique de la famille kongo-kordofan.
Brussels: Université Libre de Bruxelles dissertation.
Ducos, Gisèle. 1971. La structure du badiaranké de Guinée et du Sénégal: phonologie et syntaxe. Paris:
Klincksieck.
 22 B-applicatives and I-applicatives in Atlantic languages (Niger-Congo) 781

Faye, Waly C. 1979. Etude morphosyntaxique du sereer singandum (région de Jaxaaw-Ñaaxar). Grenoble:
Stendhal University dissertation.
Ferry, Marie-Paule. 1991. Thesaurus Tenda: dictionnaire ethnolinguistique de langues sénégalo-guinéennes
(basari, bedik, konyagi). Paris: SELAF.
Gaved, Timothy. 2020. A grammar of Mankanya. Leiden: University of Leiden dissertation.
Goudiaby, Arame. 2017. Eléments de grammaire du gújáahár, parler Baïnounck de Niaguiss (Casamance). Dakar:
UCAD dissertation.
Karlik, Jan. 1972. A Manjako grammar with special reference to the nominal group. London: SOAS University of
London dissertation.
Lopis-Sylla, Jeanne. 2010. La langue noon. Dakar: UCAD-IFAN.
Mbodj, Chérif. 1983. Recherches sur la phonologie et la morphologie de la langue saafi (le parler de Boukhou).
Nice: University of Nice dissertation.
McIntosh, Mary. 1984. Fulfude syntax and verbal morphology. London: St. Edmundsbury Press.
Meyer, Gérard. 2001. Eléments de grammaire du badiaranké: parler de la région de Koundara, Guinée. Dakar,
self-published.
Morgan, Daniel Ray. 1996. Overview of grammatical structure of Ndut: A Cangin language of Senegal. Arlington:
The University of Texas dissertation.
Njie, Codu Mbassy. 1982. Description syntaxique du wolof de Gambie. Dakar: Les Nouvelles Éditions Africaines.
Perrin, Loïc-Michel. 2019. Description grammaticale du basari (oniyan). Langue atlantique du Sénégal Oriental.
Paris: Presses de l’INALCO.
Pouye, Abdoulaye. 2015. Pour une sauvegarde des langues en danger au Sénégal – Description synchronique du
saafi-saafi. Paris: Edilivre.
Pozdniakov, Konstantin & Guillaume Segerer. Forthcoming. A genealogical classification of Atlantic
languages. In Friedrike Lüpke (ed.), The Oxford guide to the Atlantic languages of West Africa. Oxford:
Oxford University Press.
Renaudier, Marie. 2012. Dérivation et valence en sereer, variété de Mar Lodj (Atlantique, Sénégal). Lyon:
University of Lyon dissertation.
Santos, Rosine. 1996. Le mey: langue ouest-atlantique de Guinée. Paris: Université Paris 3 habilitation thesis.
Sambou, Pierre. 2007. Morphosyntaxe du joola karon. Dakar: UCAD dissertation.
Sambou, Pierre-Marie. 1979. Diola kaasa esuulaalur: phonologie, morphophonologie et morphologie. Dakar:
Université de Dakar dissertation.
Segerer, Guillaume. 2002. La langue bijogo de Bubaque (Guinée Bissau). Louvain/Paris: Peeters.
Seidel, Frank. To appear. Nalu. In Friederike Lüpke (ed.), Atlantic languages. Oxford: Oxford University Press.
Soukka, Maria. 2008. A descriptive grammar of Noon, a Cangin language of Senegal. Munich: LINCOM.
Stanton, Juliet. 2011. A grammatical sketch of Saafi. Unpublished manuscript, Indiana University.
Trifkovic, Mirjana. 1969. Le mancagne: étude phonologique et morphologique. Dakar: IFAN.
Van de Velde, Mark. 2020. Concernee-Concern constructions: A comparative study of external possession in
the Bantu languages. Studies in Language 44(1). 70–94.
Voisin-Nouguier, Sylvie 2002. Relations entre fonctions syntaxiques et fonctions sémantiques en wolof. Lyon:
University of Lyon dissertation.
Wane, Mohamadou Hamine. 2017. La grammaire du noon. Utrecht: LOT Publications.
Watson, Rachel. 2015. Kujireray: morphosyntax, noun classification and verbal nouns. London: SOAS University
of London dissertation.
Wilson, William André. 1993. An outline description of Biafada. Journal of West African Languages XXIII(2).
59–90.
Doris L. Payne
23 Nilotic applicatives
Abstract: The Nilotic family divides into three branches. Eastern, Southern, and Western
branches have dative applicatives covering roughly beneficiary, recipient, addressee
and goal. The dative applicative is usually required for benefactive phrases, but in a few
languages it can syntactically alternate with prepositional expression. Eastern, South-
ern and at least one Western Nilotic languages have syntactially optional instrumen-
tal/locative applicatives, covering instrument, time, location, result, etc. All branches
have directionals with a redirective applicative function with verbs that have an <agent,
goal/source> base argument structure. In Southern Nilotic, the ventive versus the dative
applicative have specialized for person of the beneficary; and in some, the itive marks
a nonspecific 3rd person beneficiary, and in at least one other an applied comitative.

1 Introduction to Nilotic languages

Nilotic languages extend from the southernmost part of Sudan to southern Tanzania.
The family divides in three branches (Köhler 1955), sketched in Figures 1 through 3.
Western Nilotic (WN) languages (about 22) extend from southernmost Sudan, through
South Sudan and Uganda, into the border regions of western Ethiopia, and to north-
ern Tanzania. Eastern Nilotic (EN) languages (about 17) are in South Sudan, its border
regions with Ethiopia, and in Uganda, Kenya, and Tanzania; with the notable exception
of Atlasi ya lugha za Tanzania (Languages of Tanzania Project 2009), most maps fail to
show the modern extension of the EN Maa language complex into southern Tanzania.
Southern Nilotic (SN) languages (about 16) are found in Uganda, Kenya, and Tanzania,
near Mbeya. The approximate number of languages in each branch is from Hammar-
ström et al. Number of speakers per language varies widely, from highly endangered
ones like Akie with perhaps under 200 speakers (König, Heine, and Legère 2015: 10), to
around a million if, for example, the Maa complex were considered a single language.
This chapter draws on selected languages from each major sub-branch of the family,
seen in Figures 1–3. Main sub-branches of the three main branches are in bold, further
sub-branches are in plain type, and individual languages are in italics.

Barian: Bari, Kukú, Mundari

Teso-Lotuxo-Maa
Teso-Turkana: Ateso, Turkana
Lotuxo-Maa: Lopit, Maa (including Maasai)

Figure 1: Eastern Nilotic (EN): main branches and selected languages.

https://doi.org/10.1515/9783110730951-023
784 Doris L. Payne

Kalenjin: Akie, Nandi, Cherang’any

Datooga-Omotik: Barbayiiga, Gisamjanga, Asimjeeg

Figure 2: Southern Nilotic (SN): main branches and selected languages.

Research on Nilotic varieties varies widely. For some, just word lists exist, while the mor-
phosyntax of others was documented starting in the 19th Century. Work on applicatives
is uneven, especially regarding syntax and semantic extensions. For some languages, a
few examples demonstrate that applicatives exist though they may be mentioned under
various labels, but little more is known. A few WN languages may lack applicatives,
insofar as current literature suggests.

Dinka-Nuer
Dinka: Agar Dinka
Nuer-Reel: Nuer, Reel
Luo-Burun
Southern Lwoo: Lango, Kumam, Dholuo, Acholi
Northern Lwoo: Anuak (or Anywa), Päri, Shilluk
Mabaan-Jumjum: Mabaan, Jumjum
Burun: Kurmuk

Figure 3: Western Nilotic (WN): main branches and selected languages.

Section 2 introduces the morphosyntactic typology of the three main branches relevant
to applicative constructions. As there is considerable diversity across the family, the
rest of the chapter is organized largely by semantics and probable cognate applicative
morphemes: the so-called dative and other forms for beneficiary (with various seman-
tic extensions; Section 3), instrument or location applicatives (with various semantic
extensions; Section 4), and the use of directionals in applicative function (Section 5).
Section 6 notes a few “applicative look-alike” constructions, and Section 7 addresses
non-applicative and lexicalized functions of the applicative morphology. Section 8 sum-
marizes the findings.1

2 Basic morphosyntactic overview

The morphosyntax of EN and SN languages is typologically similar, while that of WN is
quite distinct. Whether this is due to genetic closeness or more intense contact between
EN and SN, or to coincidental development is not established.

1 I mostly regularize glosses of applicative forms and adapt authors’ original glosses to Leipzig glossing
conventions. I generally retain authors’ representation of data.
 23 Nilotic applicatives 785

In all branches, applicatives do not add new clause types to the syntactic inven-
tory, except that four-argument clauses can be created in at least some languages (basic
four-argument clauses do not occur). Choice of applicative form does not generally
depend on base predicate valence, though some WN languages have distinct directional
forms for intransitive versus transitive bases. Each applicative has some degree of pol-
ysemy, but none are all-purpose applicatives. That is, none is semantically as broad as
the oft-cited Bantu applicative ✶-ɪd or Indonesian -kan.

2.1 Morphology

SN and EN verbs are broadly agglutinative, but with some fusional features, co-oc-
currence restrictions, and significant grammatical tone. (For SN, cf., among others,
Rottland 1982; Creider and Creider 1989; Mietzner 2016; Griscom 2019; Bruckhaus
2021. For EN, cf., among others, Tucker and Mpaayie 1955; Dimmendaal 1983a; Payne
2015; Barasa 2017; Moodie and Billington 2020.) Verbs have a conjugation class dis-
tinction between what are called Class I versus Class II (Rottland 1982; Tucker and
Mpaayei 1955; Dimmendaal 1983b). Allomorphy for applicatives and directionals is
largely phonological, determined by ATR harmony, tone sandhi, and consonant and
vowel deletion due to interaction with other morphemes. Subjects and speech-act
objects are indexed on the verb; pronominal 3rd person objects are zero-marked on
the verb.
WN verbs carry fewer parsable affixes than EN and SN languages. There is con-
siderable stem variation for derivation and some inflection. Most derivational catego-
ries, including applicatives and directionals, involve stem alternations in vowel length,
vowel and voice quality; consonant alternation, gemination, and deletion; and/or tone
variations. Specific stem forms can interact with verb class, which is usually discussed
in terms of the (sometimes abstract) phonological nature of the base stem (Andersen
1988a, 1992–1994; Reh 1996; Noonan 1992: 87–102; Remijsen, Miller-Naudé, and Gilley
2016; among others). More segmentable affixes index core participants, but especially
inanimate 3rd person objects may have zero verb indexation.

2.2 Basic syntax

SN and EN languages have mostly predicate-initial clause syntax, though order can
depend on discourse and constructional issues. However, Bari (EN; Spagnolo 1933)
and arguably Asimjeeg Datooga (SN; Griscom 2019: 269) have basic SVO order. The
verb initial languages display marked-nominative tonal case marking on post-verbal
NPs (König 2008). In such Nilotic systems, there is a morphologically and/or distribu-
tionally marked form for post-verbal transitive and intransitive subjects, but a mor-
phologically and/or distributionally unmarked form for citation, objects, and (most)
786 Doris L. Payne

pre-verbal subjects.2 In contrast, SVO languages do not appear to have marked nomi-
native case patterns.
Some WN languages are described as having SVO basic order, as might be suggested
by (1).3

(1) Dinka (Remijsen, Miller-Naudé, and Gilley 2016: 239)

bòol ǎ-tï ŋ̀ mìiir
Bol agr-see giraffe
‘Bol sees the giraffe.’

However, there is significant diversity across WN. Jumjum has intransitive SV but tran-
sitive AOV and OVA orders (Andersen 2019). Remijsen, Miller-Naudé, and Gilley (2016:
239) comment that in Dinka, Päri and Shilluk, subject/agent can occur post-verbally
and is then case-marked, but these languages may differ as to the syntactico-pragmatic
status of this arrangement. In Dinka, the post-verbal agent case is analyzed as ‘oblique’,
as in (2); but in Päri and Shilluk it has been analyzed as a core ergative, as in (3) (Ander-
sen 1988b; Miller and Gilley 2001; Remijsen, Miller-Naudé, and Gilley 2016).4

(2) Dinka (Remijsen, Miller-Naudé, and Gilley 2016: 239)

acóol ǎ-tï̌ ïŋ mîiir
Achol agr-see.nts giraffe.obl
‘The giraffe sees Achol.’

(3) Shilluk (Miller and Gilley 2001: 36)

byɛ́l á-ˈrākk ̀ yī ɲān d̪ájɔ ̀
grain.pl pst.evid-grind.tr.rep erg person female
‘The woman ground the durra.’

The core clause structure of some WN languages, including Lango (Noonan 1992: 119),
Kurmuk (Andersen 2015) and Agar Dinka (Andersen 2012a), is described as having

2 Most Niloticists use nominative for the marked case form, but various terms occur for the unmarked
case form including absolute (Tucker and Bryan 1962, 1964/1965; Toweett 1975; Dimmendaal 1983b;
Mietzner 2016; Barasa 2017), absolutive (Rottland 1982; Dimmendaal 2009; Bruckhaus 2021), accusative
(Tucker and Mpaayie 1955, Griscom 2019), and oblique (Creider and Creider 1989). Departing from all
the preceding, Mel’čuk (2006: 266–269) uses nominative for the unmarked case form, and oblique for the
marked one (which other Niloticists call the nominative). Here I use absolute for the unmarked form
(leaving this form unglossed for case), and nominative in accord with dominant Nilotic practice.
3 Remijsen, Miller-Naudé, and Gilley (2016) do not indicate the variety of Dinka for this example.
4 Remijsen and Ayoker (2018) argue for an alternative “object voice” analysis for the Shilluk construc-
tion. While agreeing with Miller and Gilley (2001) that it does not have the informational structure
properties of a passive, they argue that the so-called ergative phrase lacks properties of a core argument.
 23 Nilotic applicatives 787

an initial structural “Topic” position. In particular, the Agar Dinka basic declarative
clause scheme is (roughly) topic–verb–subject–object; while the Lango basic clause
scheme is topic–subject–verb–object(s). Alternative orders of arguments/semantic
roles typically require alternative verb forms (cf. the basic versus nts verb forms in
[1] versus [2]).
Nilotic languages have basic intransitive, transitive and ditransitive clause con-
structions. Simple verb roots correspond to each valence construction type, and der-
ivational processes change valence. Typical valence-reducing morphemes include
impersonal and/or passive, middle and/or reflexive, and some have an antipassive.
Regarding transitivity of Lango verb stems, Noonan (1992: 87, 123–132) distinguishes
transitive, intransitive (“activity naming” or roughly antipassive in sense), and “second-
ary argument” (roughly anticausative in sense) forms. Valence-increasing constructions
include causative and applicative(s). The various applicative constructions appear to
behave like basic syntactic constructions (transitive or ditransitive), with the aplicative
phrase (AppP) usually serving as a direct or primary object.5 In some languages, loca-
tive applied phrases may still carry an adposition, but this is not universal.

3 Dative applicative concepts: beneficiary, recipient,

addressee, goal, etc.
Nearly all Nilotic languages have what I will call a “dative” applicative. This allows a
verb to have a beneficiary/maleficiary, recipient, addressee, goal/goal-reached/spe-
cific-goal participant as a core argument, which the verb would otherwise not have.
The general forms (with additional allomorphs) are -(V)kɪ(n) in EN, -chi(n)/-sii(n) in SN,
and stem modification plus sometimes suffix elements in WN. Andersen (1988a) argues
that components of the WN applicative stem modifications are likely cognate with the
EN/SN forms (see also Dimmendaal 2009). In SN, the beneficiary function of the dative
applicative is partially taken over by other morphemes (Sections 3.2 and 5.2). Some
WN languages may be on the verge of developing a new applicative in this functional
domain (Section 3.3).6

5 I use “primary object” in the sense of Dryer (1986). Dimmendaal (2009: 11) calls the Turkana dative
AppP a “secondary object”, but does not appear to follow Dryer’s usage.
6 Creissels and Voisin (this volume) use “B-applicative” for applicatives that include benefactive but
not instrumental function, and “I-applicative” for those that include instrumental but not benefactive
function. While this terminology generally fits EN applicative patterns, the term “dative (applicative)”
is quite widely used in Nilotic studies for cognates of -(V)kɪ(n) and for some WN verb forms which may
be cognate. I retain the term “dative” here for continuity with the existing Nilotic literature and because
benefactive functions are not exclusive to cognates of -(V)kɪ(n) in SN.
788 Doris L. Payne

The dative applicative is usually valence-increasing, creating transitive stems from

intransitive roots, and ditransitives from transitives. With derived ditransitives, the
added participant does not typically appear to displace a base argument, though this is
not well researched and does happen in some languages. For a very few documented
transitive roots, the dative derivation is not valence-increasing but affects the semantic
reading of the object; that is, the dative can have an argument “re-arranging” function
(cf. Lamoureaux 2004 on Maasai). In the clear majority of languages, the dative is oblig-
atory for expressing a beneficiary. In contrast, it is generally syntactically optional for
expressing a general goal with movement and caused-motion verbs.

3.1 Eastern Nilotic

In EN, the dative has the general shape -(V)kɪ(n) and is obligatory for a beneficiary
and possibly recipient and addressee, but syntactically optional for goal. The syntactic
status of beneficiary versus base objects is not so well researched across the sub-family,
but an applied beneficiary is generally the privileged object (likely due to privileging
animate participants).
Spagnolo (1933), Nyombe (1987), and Yokwe (1987: 24–28) describe a Bari suffix -akin
(with various allomorphs).7 Spagnolo describes it as meaning ‘on behalf of’, ‘limited to
s.o.’, ‘general motion toward a central place’, etc. We may infer it has a transitivizing
effect from sets like those in (4).

(4) Bari (Spagnolo 1933: 165)

“Long” stem Dative
rʊma ‘walk fast’ rʊmakin ‘hasten for s.o. or to some place’
jara ‘be absent’ jarakin ‘delay in some place’
miö ‘be painful’ mikin ‘hurt’

In (5), the theme ná precedes the verb and the AppP follows the verb (in the presumably
basic object position). In (6), the theme follows the verb. Either the base theme or the
applied beneficiary can be pronominal. If 3rd person and known from context, these
can be expressed as a null like the beneficiary in (6).

(5) Bari (Yokwe 1987: 416; my glosses)

Pòní ná mòk-ákìn Jàdà
Poni dem.fsg hold-dat Jada
‘Poni is holding it for Jada.’

7 Spagnolo parses the first vowel as part of the “long form” of the preceding stem.
 23 Nilotic applicatives 789

(6) Bari (Yokwe 1987: 478; my glosses)

Pòní à dér-á-kín súkùrì
Poni pst cook-link-dat chicken
‘Poni cooked chicken for him.’

Example (7) suggests that some adpositions might still co-occur with some added par-
ticipants since both ‘1sg’ and ‘leg’ appear to be in adpositional phrases with the dative
stem; but this needs investigation (Dimmendaal 2009: 13).

(7) Bari (EN; Spagnolo 1933: 165; my glosses based on Spagnolo)

mɛjɛkɛ mi~mikin köyö i mokot
guinea.worm asp~hurt.dat to.me in leg
‘The guinea-worm hurts me in the leg.’

Kukú is mutually intelligible with Bari. Cohen (2000: 9, 54–56) documents an applica-
tive -akɪn/-ɨkɨn (the latter is a +ATR allomorph). This is said to indicate “ditransitivity”
and beneficiary, and it is lexicalized in some stems (e.g., ‘arrive’, ‘support’). There is
little information on valence or syntactic effects of the Kukú dative, but Cohen (2000:
124–125) gives examples with base and dative forms of ‘give’. The applicative stem tíkín
‘give to’ can occur with a pronoun recipient immediately after the verb, which is the
normal position for base pronoun objects. (Applied examples with a lexical recipient
are missing.) Contrariwise, an example with the non-applicative stem tín ‘give’ has the
order Subject–Verb–theme–recipient when all arguments are lexical. (Examples are
missing with tin and pronoun recipients.)
Dimmendaal (2009: 1–10) provides a detailed account of the Turkana dative applica-
tive. It creates transitives from intransitive bases (like ‘be angry with’ from ‘be angry’),
and ditransitives from transitive bases. In (8) with a transitive root, for instance, the
applied beneficiary is indexed on the verb by kà-, showing that 1sg is an object (also
expressed here by a free unmarked-case pronoun).8

(8) Turkana (Dimmendaal 2009: 2)

kà-ɪ ̀nɔ ̀k-ak(ɪ ̀) Ŋesì ayɔŋ̀ akɪ ̀m(ɪ)
3>1-light-dat 3sg.nom 1sg fire
‘S/he has lighted a fire for me.’

In a second construction, the Turkana dative can add a locative AppP. In (9), the dative
is absent and the locative is preceded by the preposition à. In (10), the dative occurs and
the preposition disappears. In both, the location is marked for locative case by na-.

8 Dimmendaal (2009) calls the unmarked Turkana case “absolutive”, which has wider distribution in
the grammar. This does not pattern as absolutive in the sense of an ergative/absolutive system.
790 Doris L. Payne

According to Dimmendaal (2009: 6–7), the prepositional construction introduces new

or contrastive information, while the dative construction tracks already-established
information.

(9) Turkana (Dimmendaal 2009: 4)

k-ìbòy-è-te à na-wuy(è) kɛc(ɪ ̀)
subs-stay-asp-pl prep loc.f-home their
‘and they stayed at their homestead’

(10) Turkana (Dimmendaal 2009: 4)

k-ìbò-ìkìn-o-s(ì) nà-wuy(è) kɛc(ɪ ̀)
subs-stay-dat-mid-pl loc.f-homestead their
‘and they stayed in their homestead’

Example (11) shows similar facts with a transitive root. As with ɪnɔk ‘light’ in (8), the
dative yields a three-argument clause from the transitive root ɪrɛp ‘insert, put’. But
unlike (8) which has a benefactive interpretation, (11) has one of its non-nominative
arguments in a locative case.

(11) Turkana (Dimmendaal 2009: 3)

k-ɪ ̀rɛ ̀p-ak(ɪ ̀) ŋimòyo ̀ kɛŋ ̀ na-ki‵
subs-put-dat fingers 3sg.poss loc.pl-ears
‘S/he has put his/her fingers in his/her ears.’

In a third construction type and contrary to (10), a locative phrase sometimes still
carries a preposition with the dative verb derivation, as in (12). Note, however, that the
locative in (12) is a source, rather than a ‘place-at-which’ or goal; also, adpositional
source phrases do not always require co-occurrence of the dative verb suffix. Whether
(12) should be considered an applicative construction merits further study.

(12) Turkana (Dimmendaal 2009: 5)

ɛ ̀làk-àkɪ ̀n-ɪt ŋikurùdoi à na-kɪtʊ̀k
3.release-dat-asp foam prep loc.f-mouth
‘S/he released foam from the mouth.’ (i.e., s/he was in a convulsion)

Fourthly in Turkana, the dative can yield comitative meaning in combination with the
preposition kà; compare (13) and (14).

(13) Turkana (Dimmendaal 2009: 6)

è-bùn-it-o ̀ ka ̀ nakwèe
3-come-asp-pl prep Nakwee
‘They are coming together with Nakwee.’
 23 Nilotic applicatives 791

(14) Turkana (Dimmendaal 2009: 6)

ɲ-ɪmɔrɪn-àkin-o-ì ŋitùrkwanà ka ̀ ŋɪtɔpɔ ̀sa
neg-3.mix-dat-mid-pl Turkana.nom prep Toposa
‘The Turkana do not associate/mix with Toposa people.’

In Lopit, dative applicative and prepositional forms can alternate for expressing bene-
ficiary and goal (Moodie and Billington 2020: 190–191). Further, the Lopit dative may
co-occur with a prepositional goal in something like ‘I hid the cow in the mountains’.
In such instances, the authors speculate that the dative may be lexicalized as part of
the verb stem and is not valence-increasing, though interestingly they also state that a
location must be expressed when the dative relates to a goal (p. 192).
In Ateso (Barasa 2017: 186–188), the dative occurs for beneficiary as in (15), recip-
ient (even with the verb ‘give’) as in (16), or addressee as in (17). It also occurs with
the verb ‘meet’ to express the party met, which we may consider a goal-reached, as in
(18). But there is a difference in how the AppP is treated depending on semantic role.
An applied beneficiary, recipient or addressee occurs in the unmarked absolute case
immediately after a lexical subject (if any) and before the theme. The goal of ‘meet’
occurs in a prepositional phrase despite the applicative on the verb.9

(15) Ateso (Barasa 2017: 187)

é-tjàk-àkìn è-múrwòk àkàtòlà
3sg-cut.pst-dat m-doctors sugarcane
‘S/he cut (divided) sugarcane for the (traditional) doctors.’

(16) Ateso (Barasa 2017: 186)

é-ìn-àkìn-ì èmús ̀gòt Pétérò àpàkí
3sg-give.pst-dat-ipfv butcher.nom Petero permission
‘The butcher gave Petero permission.’

(17) Ateso (Barasa 2017: 186)

é-sòm-ak-ìt ɟáɟá ébàrwàìt
3sg-read-dat-pfv aunt letter
‘S/he read the letters to (the) aunt.’

9 Barasa (2017: 170) describes a tonal locative case which can occur after the preposition k=. He does
not indicate what case occurs on ‘chief’ in (18) but it is tonally different from the apparent absolute case
form èɟákàít.
792 Doris L. Payne

(18) Ateso (Barasa 2017: 188)

ŋàíbó kɪ́=rjàm-àkìn-ò-s(i) k=éɟàkàìt
who prep=2sg-meet.pst-dat-rec-pl prep=chief
‘Who met the chief?’

For Maasai, Lamoureaux (2004) quite comprehensively treats the morphosyntax and
semantics of applicative constructions (see Andrason and Karani 2019 on the argument
status of dative AppPs in Arusa Maa). If the dative is added to a transitive root, the
AppP patterns as the primary object and tends to immediately follow the verb (but this
may be due to animacy features of a beneficiary more than the effect of the applica-
tive; cf. Payne 2022a). 1st and 2nd person objects (base or applied) are indexed by verb
prefixes. Example (19) shows this and also demonstrates the possible goal(-reached)
semantics of the Maa dative. This is a complex example. The dative AppP is “raised” to
be the object of an otherwise intransitive verb ‘go.pl’ (marked by the “inverse” prefix
kí- which expresses 3/1pl>2sg or 2>1sg; Payne, Hamaya and Jacobs 1994), even though
the applicative is on the infinitive verb ‘take’.

(19) Maasai
K-á-yīēū ní-kí-púó-puo áa-y-akɪ ɪl=pɪ́dɪlá tɛ síaŋau
cn2-1sg-want cn-inv-go.pl-go.pl inf.pl-take-dat mpl=fleas obl calabash.nom
‘I want you (pl) to go and bring me fleas in a calabash.’ (arinkoi.006)

The Maasai dative can create four-argument clauses, as in (20). Pɪk ‘put’ is a three-argu-
ment root; the dative adds a beneficiary, indexed by the verb prefix.

(20) Maasai (fieldnotes)

áa-pɪk-akɪ̂ ɔl=mʊrraní ɛnk=árɛ́ e=motí
3>1sg-put-dat msg=warrior.nom fsg=water fsg=pot
‘The warrior will put water into the pot on my behalf.’

The Maasai dative can also have a valence-neutral redirective applicative function with
<agent path> translational motion verbs. Compare (21)–(22), which together show that
the dative derives an <agent goal-reached> stem.

(21) Maasai (adapted from Lamoureaux 2004: 58)

ɛ-ɪ́d ɔl=páyian o=sóít
3-jump.over msg=man.nom msg=rock
‘The man will jump (over) the rock.’
 23 Nilotic applicatives 793

(22) Maasai (adapted from Lamoureaux 2004: 58)

ɛ-ɪ́d-ákɪ ɔl=páyian o=sóít
3-jump.over-dat msg=man.nom msg=rock
‘The man will jump on (top of) the rock.’

For Arusa Maa, Andrason and Karani (2019: 183) show a goal ApplP preceded by a
relational noun ‘inside’.10 Additional non-applicative functions of the Maa dative are
mentioned in Section 7.

3.2 Southern Nilotic

Both SN sub-branches have an applicative which appears to be cognate with the EN

dative.11 As we will see, SN languages use the dative or a ventive directional for benefi-
ciary depending on person of the beneficiary.

3.2.1 Kalenjin varieties

The Akie dative is -ci(n)/-in(i) (among other allomorphs; c represents a palatal conso-
nant; König, Heine, and Legère 2015: 21). Examples show this can express beneficiary
or goal.

(23) Akie (König, Heine, and Legère 2015: 56)

ki-rá ál-ci láákwɛɛ amtɪ
1pl-pfv buy-dat.pfv child.acc food.acc
‘We have bought food for the child.’

(24) Akie (König, Heine, and Legère 2015: 56)

a ŋútt-ini ŋʊ́ llɛɛ
1sg spit-dat saliva.acc
‘I spit (at him).’ [for blessing]

10 Andrason and Karani claim that ‘inside’ is a preposition, but it varies for case like a relational noun.
Another dative example which they claim shows a prepositional goal could instead be analyzed as
‘Father will send mother the boy at home’, where ‘mother’ (which doesn’t carry a preposition) is the
goal AppP.
11 It is called “dative”, “terminative”, “allative” or just “applicative” by SN scholars. Bruckhaus (2021:
122) hypothesizes that it derives from sii ‘somebody’ in G/B Datooga, though he also says it “corresponds
to” the “goal marking suffix ✶-k(In)” (p. 50).
794 Doris L. Payne

In Nandi, the dative expresses applied recipient, addressee, goal, beneficiary, and
maleficiary as in (25), and reason (Creider and Creider 1989; Creider 2002).12

(25) Nandi (Creider 2002: 174)

ka-rér-cí ce:ka cesí:lû:t
pst1-be.scarce-dat milk Jesilut (woman’s name)
‘The milk has run out on Jesilut.’

Creider and Creider (1989: 90) state that the Nandi dative is restricted to 3rd person
objects. However, (26) with a 2nd person addressee shows this apparently does not
always hold; (27) shows the same root ‘say’ without the dative.

(26) Nandi (Creider 2002: 174)

ká:-lé:n-cí:n ib-ú: kikô:mpé:t
pst1.1sg-say-dat bring-ven cup
‘I said to you to bring the cup.’

(27) Nandi (Creider 2002: 174)

ké:-le nè:
pst1.2sg-say what
‘What did you say?’

Interpretation of the semantic role of the AppP depends on verb semantics (e.g.,
addressee with ‘say’ as just seen) and on features of the AppP such as animacy. Compare
constructionally identical (28)–(29), the first with beneficiary and the second with
goal(-reached) readings.

(28) Nandi (Creider 2002: 175)

itú:-cí kwe:nìk ce:pe:t
put.down-dat firewood Jebet
‘S/he is putting the firewood down for Jebet.’

(29) Nandi (Creider 2002: 175)

ki:-tu:-cì kwe:nìk ngecerê:t
inf-put.down-dat firewood chair
‘to set the firewood on the chair’

12 The Nandi data do not represent ATR contrasts. The colon represents a long vowel. Creider (2002:
172) states there are three degrees of past tense. I assume his “past 1” (pst1) is closest to the temporal
point of reference.
 23 Nilotic applicatives 795

The Nandi dative is lexicalized with some vebs including ‘give’ (Section 7). However, it
increases valence when structurally optional.
In Cherang’any, the dative is used for 3rd person goal and beneficiary AppP objects
(which Mietzner 2016: 134 calls “indirect” objects), as in (30). In contrast, 1st and 2nd
person applied beneficiaries involve directionals (Section 5.2).

(30) Cherang’any (Mietzner 2016: 134)

ʌ́-tèèk-cíní sɪ́kɪ́k-cùù kɔ ̀
1sg-build-dat parents-psr.sg+psm3pl house
‘I am building a house for my parents.’

3.2.2 Datooga varieties

Within the Datooga SN sub-branch, Barbayiiga and Gisamjanga (B/G) are very closely
related, over against Asimjeeg Datooga (Griscom 2019).13 As in Cherang’any, there is
generally a person-based split for benefactive applicatives: dative and sometimes itive
cognates are used for 3rd person beneficiary AppPs, while the ventive directional is
used for 1st and 2nd person (Section 5.2). Syntactically, the dative increases valence in all
Datooga varieties.
For B/G Datooga, dative -s/-sii adds a beneficiary, recipient or concrete or fictive
specific goal that is not the current deictic center (Bruckhaus 2021: 117, 122); a moving
entity may or may not actually arrive at the goal. This is seen in (31)–(33) for 3rd and 2nd
person goals of movement verbs.

(31) B/G Datooga (Bruckhaus 2021: 122)

béegà gá-gúl-s-á bàsôoda
water 3sbj-flow-dat-is lake
‘The water flows to the lake.’

(32) B/G Datooga (Bruckhaus 2021: 122)

gá-wèe-s-à
3sbj-go-dat-is
‘S/he goes to her/him.’

13 For B/G Datooga, the dative and -an applicatives (Section 4.3.2) may co-occur (Bruckhaus (2021). For
Asimjeeg Datooga, there may be limits on their co-occurrence, and -an can occur with directionals (Gri-
scom 2019: 240).
796 Doris L. Payne

(33) B/G Datooga (Bruckhaus 2021: 122)

gá-bíik-s-èey
3sbj-return-dat-2sg.obj
‘He returns to you (sg).’

For most verbs, however, the dative is not used for 1st or 2nd person, as the contrasts in
(34)–(35) demonstrate for goal, and in (36)–(39) for beneficiary.14

(34) B/G Datooga (Bruckhaus 2021: 68)

ní-bàlàg-ù míiyɛ ̀ɛndɛ ̀-ɛɲù
3.prf-shift-ven sickness-sg.poss.1sg
‘My sickness has returned hither/to me.’

(35) B/G Datooga (Bruckhaus 2021: 68)

nì-bálák-s-á gɛ́ɛskà-scì
3.prf-shift-dat-is coughing-pl.poss.3sg
‘His coughing has returned to him.’

(36) B/G Datooga (Bruckhaus 2021: 117)

gwá-lák-n-ɛ ̀ɛn
3-cut-ven-1sg.obj
‘S/he is cutting for me.’

(37) B/G Datooga (Bruckhaus 2021: 117)

gwá-làk-n-éeyì
3-cut-ven-2sg.obj
‘S/he is cutting for you (sg).’

(38) B/G Datooga (Bruckhaus 2021: 117)

gwá-làk-s-à
3-cut-dat-is
‘S/he is cutting for her/him.’

(39) B/G Datooga (Bruckhaus 2021: 124)

síi-sí gá-gɛ ̀ɛs-s-à bétrò bêega
person-prox 3sbj-fetch-dat-is Peter water
‘This person is fetching water for Peter.’

14 It is not clear whether valence has increased in (34) given the lack of an object suffix on the verb.
However, (34) is informative in its contrast with (35).
 23 Nilotic applicatives 797

In addition to 3rd person beneficiary, the dative can add a location as in (40)–(41).15

(40) B/G Datooga (Bruckhaus 2021: 124)

gòo-réer-s-à Héeydòm
3sbj-feed-dat-is Haydom
‘S/he is feeding at Haydom.’

(41) B/G Datooga (Bruckhaus 2021: 124)

gá-kèek-íi-s-á qɔ ̀ɔh
3sbj-vomit-plrc-dat-is home
‘S/he always vomits at home.’

Some dative stems are restricted to animate goals; others must co-occur with héedà
‘place’ + animate noun for an animate goal. Kießling (2007: 132) notes sense-extension
of the dative across recipient and beneficiary in instances like ‘He gave out the meat
to the youth’. He further finds that with verbs which lexically have a recipient role in
their basic predicate frame, the dative “adds an idea of finality and introduces a cause
or a purpose” (p. 134). He illustrates with maar ‘give a present (cattle) to someone in
acknowledgement of a heroic deed’ which, in its underived form, “already assigns the
recipient to the object role”. In (42), the dative stem maars adds the meaning ‘for a
special purpose’ (‘finality’ is not particularly evident here).

(42) B/G Datooga (Kießling 2007: 134)

gɪ́ɪ-màars-ínàa
sbj2.sg-donate.dat-what
‘What do you give a hero’s present to him for?’

Griscom (2019: 125–126, 237) summarizes functions of the Asimjeeg Datooga dative
-s(V:n) ~ -s(a) ~ s(i) as adding 3rd person goal in a macro-sense. More specifically, the
added argument can be an endpoint as seen by comparing (43)–(44); beneficiary as in
(45); occasionally location as seen by comparing (46)–(47); and some addressees as in
(48)–(49).16

(43) Asimjeeg Datooga (Griscom 2019: 238)

àníːn g-ò-tʃág-n-àːn gíʃín
1sg aff-3-send-ven-1sg Gishina
‘Gishna sent me.’

15 Bruckhaus (2021: 124) also suggests the dative (his “allative”) can indicate purpose, but perhaps only
in combination with the question word ‘what’.
16 Griscom (2019: 126) reads as if it can also add an instrument, but no examples support this. In (48),
the double colon represents extra length, and (.) represents a pause.
798 Doris L. Payne

(44) Asimjeeg Datooga (Griscom 2019: 238)

∅-wún máːŋòl g-ì-dà-tʃàg-síːn-eːɲ
2sg-come Mang’ola aff-fut-1sg-send-dat-2sg
‘Come, I’m sending you to Mang’ola.’

(45) Asimjeeg Datooga (Griscom 2019: 240)

g-àː-ràm-s mùh-óːg áb qùwàrí
aff-1sg-fetch-dat calf-pl 17
prep home
‘I fetched (water) for the calves at home.’
17
(46) Asimjeeg Datooga (Griscom 2019: 239)
q-àː-dà-j síː-d sí-bàr
aff-1sg-see-fs person-sg impers-hit
‘I saw a person get beaten (killed).’

(47) Asimjeeg Datooga (Griscom 2019: 239)

m-à-núŋ dá-jíɲ àsìm-dʒànd sí-bár-s dʒérkʷ-èːd
neg-3-let 1sg-hear asimjeeg-sg impers-hit-dat field-sg
‘I’ve never heard of an Asimjeeg person being beaten in the field.’

(48) Asimjeeg Datooga (Griscom 2019: 240)

há iːdúːː (.) g-à-gùr-s bùn-éːd áb màdʒ-òːd
well later aff-3-call-dat people-sg prep magic-sg
‘Later he called people by using magic. . .’

(49) Asimjeeg Datooga (Griscom 2019: 107)

íːs g-èː-rùg-s síː-d ŋáʃ-ánd
often aff-impers-tell-dat person-sg word-sg
‘Often a person is told something. . .’

Unlike varieties which use the ventive for speech-act addresse, in Asimjeeg Datooga the
dative sometimes is used (Griscom 2019: 237, 259) as in (50). However, the ventive still
generally occurs for this (see Section 6.3).18

17 SN nouns have complex number affixation, which Griscom (2019) thoroughly parses. As the number
system does not bear on issues addressed here, I collapse multiple number affixes into one parsed ele-
ment and gloss it simply as SG or PL.
18 Griscom speculates on whether the dative versus ventive could be lexicalized with different roots, or
perhaps reflects an aspectual feature.
 23 Nilotic applicatives 799

(50) Asimjeeg Datooga (Griscom 2019: 237)

g-à-gùr-síːn-áːn sìː-d qàh-áɲ
aff-3-call-dat-1sg person-sg home-1sg.poss
‘. . .and a person from home called me. . .’

In addition to using the ventive for 1st and 2nd person beneficiary and addressee,
Datooga varieties have an applicative -an which seems to have some association with
1st and 2nd person addresses and goals (Section 4.3.2).

3.3 Western Nilotic

The morpological interactions of derivational categories (e.g., benefactive/dative, direc-

tionals, voice) are relatively well-studied for some WN languages (Andersen 1988a;
Andersen 1992–1994; Reh 1996; Remijsen, Miller-Naudé, and Gilley 2016; and others).
Languages in all WN sub-branches have applicative verb stems that include expression
of beneficiary/goal (our broad dative category). As noted in Section 2.1, WN verb dervi-
ation is largely achieved by stem-internal changes; but some languages have a ‘bene-
factive’ suffix -ɪ co-occurring with stem changes and Jumjum has a dative suffix -k/-g. I
treat the WN applicative category that includes functions of the EN/SN dative here, as
they are similar in semantics and valence effects. Further, Andersen (1988a: 106–110)
reconstructs a pre-Päri dative suffix ✶-V̤C (where V̤ is a +ATR vowel and C is a stop). He
suggests it goes back to Proto-WN, and that this and the EN/SN forms may reconstruct
to a single source.
The following pair shows the valence-increasing effect of the Agar Dinka dative.
Observe the differences in tone and voice quality of the verb stem vowel.19

(51) Agar Dinka (Andersen 1992–1994: 9)

d̻ ɔ̤ ɔk
̀ à̰-mḭ it̀ wḛŋ́
boy decl-pull cow
‘The boy is pulling the cow.’

(52) Agar Dinka (Andersen 1992–1994: 9)

d̻ ɔ̤ ɔk
̀ à̰-mî̤ it wḛŋ́ mò̰c
boy decl-pull.dat cow man
‘The boy is pulling the cow for the man.’

19 Andersen (1992–1994) calls this the “benefactive” derivation; for some other WN languages the
presumably cognate derivation is just called “applicative”, and for others “dative”. Under-dots mark a
breathy vowel and a tilde under a vowel marks creaky voice.
800 Doris L. Payne

Cien et al.’s (2016: 124, 215) pedagogically-oriented grammar of Reel states that an
“applicative” verb can add a location, direction, recipient (for ‘give’), or beneficiary
participant. Verb classes can affect form of the applicative stem, and some applica-
tive forms may be homophonous with non-derived transitive stems.20 The applicative
increases valence, as in (53)–(54).

(53) Reel (Cien et al. 2016: 124)

nuɛ̈ɛ̈r liiŋ gat
person listen.incompl child
‘Person is listening to child.’

(54) Reel (Cien et al. 2016: 124)

nuɛ̈ɛ̈r lïŋ gat ruɛ̈th
person listen.incompl.dat child chief
‘Person is listening to child for chief.’

Examples (55)–(56) demonstrate the transitivizing effect of the applicative with an

intransitive root. The citation form of ‘well’ is kath, while kaath means ‘in well’ (Cien
et al. 2016: 41). Thus, the applicative stem appears to license a locative role but in
a locative case. The locative case form of a noun is distinct from the prepositions kë
for goal/direction and ke for accompaniment, instrument, time, goal/direction (Cien
et al. 2016: 60).

(55) Reel (Cien et al. 2016: 124)

nuɛ̈ɛ̈r pen
person fall.incompl
‘Person is falling.’

(56) Reel (Cien et al. 2016: 124)

nuɛ̈ɛ̈r peen kaath
person fall.incompl.dat well.loc
‘Person is falling in well.’

The following pair demonstrates the locative case with a transitive root.

20 Cien et al. do not write tone, but it is possible it changes (cf. Reid 2010: 28–29). Diereses mark breathy
vowels.
 23 Nilotic applicatives 801

(57) Reel (Cien et al. 2016: 73)

juii ɛ Ayak
saw 3 Lion
‘He saw Lion.’

(58) Reel (Cien et al. 2016: 41)

kue Ayak juic kaath
then.3 Lion saw.dat well.loc
‘Then he saw Lion in well.’

Reid (2019: 99, 169) briefly describes the Nuer “applicative” derivation, with vowel fea-
tures varying by verb class and finiteness. This derivation adds various semantic roles
including beneficiary/maleficiary and recipient, as follows.21

(59) Nuer (Reid 2019: 119)

ɰɔ̤́aaaɰ-ʌ̤́
push(tr)-1sg
(free translation not provided; presumably ‘I push it.’)

(60) Nuer (Reid 2019: 99)

ɰʌ ̤́n ɰó̤oɰ-ʌ ̤́ rāaan Bó̤ool
1sg push(tr).dat-1sg person Bool
‘I am pushing Bool for the person.’

(61) Nuer (Reid 2019: 99)

ɟɛ̤n
́ lʌ́t̪-ɛ̤́ twé̤e et̪ wi̤ ì j-d-ʌ̤́
3sg work(tr).dat-3sg spell head loc-poss.sg-1sg
‘S/he is putting a spell on me’.

(62) Nuer (Reid 2019: 73)

mʊ̤́ɔɔj-ʌ̤́
give(tr)-1sg
(free translation not provided; presumably ‘I give sth./it.’)

21 Crazzolara (1933: 112–129) uses “applicative” for certain Nuer forms which are “transitive verbs
when actually used transitively, i.e., when followed by a direct object” (p. 112). (I am grateful to a
reviewer who notes that Tucker and Bryan 1966: 424 similarly use “applicative” simply to designate
transitive action “applied to a specific Object”.) Crazzolara’s translations do not indicate benefactive or
recipient semantics. In sum, his “applicative” verb forms are likely not applicative in the sense used in
this volume.
802 Doris L. Payne

(63) Nuer (Reid 2019: 99)

ɰʌ̤́n mṳ́ ɔɔj-ʌ̤́ rāaan dwɔ̤ ̀ɔɔr
1sg give(tr).dat-1sg person thing
‘I am giving a thing to the person.’

In Reel, Cien et al. (2016: 40) call muɔɔc ‘give’ an applicative verb and show it with three
participants in (64). Elsewhere the apparently identical form is called a “derived tran-
sitive” and sometimes has just an overt recipient (p. 85) or just an overt theme (p. 39);
it is conceivable that both recipient and theme could be zero if known from context.
The evidence is inconclusive as to whether Reel ‘give’ is a lexicalized applicative stem
or productively adds the dative.

(64) Reel (Cien et al. 2016: 197)

yɛn ɣä yïn kuï gaatku muɔɔc ɣɛ̈m më kel,
it if you.sg then.you.sg 2sg.children give.dat leg that one
‘. . .if you can give your children one leg, . . .’

In Lango (Southern Lwoo), “benefactive stems” (also malefactive) can be derived from
verbs with an agentive subject. I retain Noonan’s gloss ben(efactive) here, since he does
not comment on any broader meanings. Compare:

(65) Lango (Noonan 1992: 120)

dákô òtèdò rìŋó
woman 3sg.cook.pfv meat
‘The woman cooked the meat.’

(66) Lango (Noonan 1992: 120)

dákô òtèddɪ ̀ lócə ̀ rìŋó
woman 3sg.cook.ben.pfv man meat
‘The woman cooked the meat for the man.’

The benefactive stem geminates the second consonant of the base verb root/stem. A
suffix -ɪ ̀ occurs when there is an NP or 3rd person singular non-human independent
pronoun beneficiary; otherwise, “the appropriate [direct object] pronoun suffix is sub-
stituted” (Noonan 1992: 136). For example, (67) and (68) are formed from lego ‘to pray’
(intransitive) and lɛ ̀ggò ‘to pray to’ (transitive).

(67) Lango (Noonan 1992: 136)

òlèggə́
3sg.pray.intr.ben.pfv.1sg
‘He prayed for me.’
 23 Nilotic applicatives 803

(68) Lango (Noonan 1992: 136)

òlɛ ̀ggá òbáŋá
3sg.pray.tr.ben.pfv.1sg God
‘He prayed to God for me.’

With a benefactive applicative stem, only the applied argument can be indexed as object
on the verb, as in (69). A 3rd person human theme-object can pronominalize with the
applicative stem only if it is the sentence topic in preverbal position. Apparently, 1st and
2nd persons simply cannot be themes with benefactive stems (Noonan 1992: 122, 141).

(69) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀ɪ ̀yɛ́ bòtə́
man 3sg.give.ben.pfv.3sg to.1sg
‘The man gave it to me for him.’

In Anywa (Northern Lwoo), a beneficiary may be expressed in a dative applicative con-

struction or in a prepositional phrase (Reh 1996: 229–230). With the applicative deri-
vation, the beneficiary occurs in the privileged object position and the base patient
cannot remain in the clause.
In Shilluk from the same sub-branch, an applicative stem is apparently required
to express a beneficiary and recipient, as in (70). The applicative stem is marked by
a combination of affixal, tone and stem-internal changes, and varies by verb class and
tense/aspect (Remijsen, Miller-Naudé, and Gilley 2016: 225–227). The past tense applica-
tive template is described as á-CV́[+ATR]C-ɪ ̀ for most verbs. The -ɪ suffix also occurs for
the future in this derivation, while the perfect and imperfective have different suffix
elements.
The Shilluk dative derivation allows both a recipient and theme to remain in the
clause. Argument order is partly affected by tense/aspect. In a basic transitive clause,
the patient/theme precedes the verb, as in (70a). However, in imperfective aspect an
applied beneficiary/recipient follows the verb (p. 227); in other tense/aspects it can
follow or precede the verb, as in (70b).

(70) Shilluk (Remijsen, Miller-Naudé, and Gilley 2016: 225)

a. ɟám-ánɪ́ kàa rû́m gɛ́ jôoot-ɔ ̀ ɪ ̀ɪ jáa
things-previous.mention when finish 3pl find-nmlz erg 1sg
‘When I had found those things’
b. jáa á-t̪ ôot̪ -ɪ ̀ tɔ́ŋ
1sg pst-hand-dat spear
‘they handed me a spear.’
804 Doris L. Payne

In Jumjum (Southern Burun), the dative applicative uses the suffix -k/-g (Andersen 2018;
2019: 179). The order facts are reminiscent of those for Shilluk non-imperfective clauses
(Andersen does not comment on any tense/aspect order issues). In the basic transitive
clause in (71), the patient object is before the verb. With the dative applicative in (72),
the beneficiary precedes the verb while the patient follows it in the position of demoted
patients in antipassive clauses.

(71) Jumjum (Andersen 2019: 179)

ʔɔ̂ɔn ʔʌ̂n bán̪-n̪-ɛ́ t̻ ɪ ̀ɪnʊ̀k
man house plaster-pst-3sg>3 yesterday
‘The man plastered the house yesterday.’

(72) Jumjum (Andersen 2019: 179)

kíilgà báabà bʌ́t̻ -k-ɛ́d̻ -ɛ́ ʔʌ̂n
Kiilga Baaba plaster-dat-ipfv-3sg>3 house
‘Kiilga is plastering a house for Baaba.’

Finally, Dimmendaal (2009: 16) states that Dholuo (Luo) (Southern Lwoo branch) appears
to have lost the dative morphological alternation, perhaps due to loosing gemination
and thus neutralizing the type of difference seen in Lango. However, data from Tucker
(1994: 347–348) suggests that Dholuo may be poised to develop a new benefactive/goal
applicative from a cliticized preposition ni. This is evident when the AppP is pronomi-
nal. Example (73) has a beneficiary in a prepositional phrase; (74) has the same verb
root with pronominal beneficiary, phonologically attached to the verb. In (75)–(76), the
variable order of what I gloss ‘dat.3pl’ and just ‘3pl’ suggests the clitic nature of these
elements.22

(73) Dholuo (Tucker 1994: 348; my glossing)22

o-bíró kélo bél ní món
3prf-come bring millet prep:for women
‘He has come to bring millet to the women.’

(74) Dholuo (Tucker 1994: 348; my glossing)

o-bíró kélo-nígí bél
3prf-has.come bring-dat.3pl millet
‘He has come to bring them millet.’

22 I have omitted Tucker’s raised dot after certain vowels as it doesn’t appear to represent a phonemic
or morphemic element.
 23 Nilotic applicatives 805

(75) Dholuo (Tucker 1994: 348; my glossing)

o-bíró tédo-gí-nígí
3prf-come cook-3pl-dat.3pl
‘He has come to cook them for them.’

(76) Dholuo (Tucker 1994: 348; my glossing)

o-bíró tédo-nígí-gí
3prf-come cook-dat.3pl-3pl
‘He has come to cook them for them.’

In the next section, we turn to applicatives which express instrument, locative, and
other non-dative meanings.

4 Instrument and locative applicatives

Whether an instrumental applicative can be reconstructed for Proto-Nilotic is presently
unclear.23 EN verb suffixes or suffix-complexes -rɪ, -rɛ, -arɛ/-ore, and -ie(k) are charac-
terized as expressing instrument, with semantic extensions to location, time, reason,
comitative (and cause), depending on the language and form. SN -ɛ:(w), -è, -èèn, -èyyà,
-èèyyèn, and -an forms have been called instrument and locative, also with various
semantic extensions. Instrumental applicatives are syntactically optional in EN and
SN, alternating with oblique expressions. In WN, an instrumental/locative applicative,
separate from the stem variations described for benefactive/recipient (Section 3.3), is
documented for Shilluk; otherwise, the WN literature is silent on instrument/locative
applicatives.

4.1 Eastern Nilotic instrumentals

Information on a Barian instrument applicative is somewhat conflicting. Spagnolo’s

(1933) examples said to contain one are often also described as “passive” (I suspect these
might be middle forms fused with an instrumental or combined with an impersonal
or even a subjunctive morpheme like -ɪ).24 For Bari, Yokwe (1987: 58–60) mentions an
instrumental suffix -rɪ, but does not address its function or morphosyntax beyond the
label “instrumental”; he notes it can co-occur with the dative suffix. Spagnolo (1933)

23 Dimmendaal (1981: 69) proposed that ✶-E ‘instrumental’ was a common Nilotic heritage. Subsequent
work has not, to my knowledge, furthered this idea.
24 Spagnolo (1933: 149–150) does not write tone, which might help clarify the morphological categories
present in a verb.
806 Doris L. Payne

says the Bari instrumental can convert an intransitive into a transitive such that the
semantic instrument loses its preposition ko ‘with’. Compare the forms in (77).

(77) Bari (Spagnolo 1933: 169–170)

“Long” stem form “Transitive” form
rɔma ‘make salute’ rɔmari ‘salute with, use for saluting’
lyöŋö ‘be joyful’ lyöŋöri ‘rejoice in/for, delight in/by means of’
doto ‘sleep’ dotori ‘use for sleeping, sleep with/in’

In (78), -rɪ would appear to increase valence by adding an instrument. The instrument
is in the subject grammatical role due to what is called the “passive” in Yokwe’s treat-
ment (but I suspect -á could be a middle suffix).25

(78) Bari (Yokwe 1987: 480; my parsing and glosses based on Yokwe’s description)
mú’dâ à dér-á-rî àmbàtà
pot pst cook-pass/link-ins bread
‘The pot was used for cooking bread.’

For the closely related Kukú variety of Bari, Cohen (2000: 56–58) rejects that apparent
cognate(s) of Spagnolo’s “instrumental” have an instrument function. He suggests they
are aspectual forms or derive subordinate adverbial clauses (e.g., with temporal ‘when’
meaning).
The differences between Spagnolo/Yokwe’s and Cohen’s analyses are reminiscent of
Dimmendaal’s treatments of Turkana. Dimmendaal (1981: 64) presents examples with
-ɪa and -are/-ore glossed as “instrumental”. But the same examples have revised parsing
and glossing in Dimmendaal (1983: 189–192) where -ɪa is parsed into two elements,
-ɪ “aspect” and -a/o “voice”, and the combination is considered a type of subjunctive
marking. -Rɛ ̀/-rɪ̥ ̀ is then called an “instrumental (subjunctive)”, and is discussed under
the heading of subjunctive mood. The examples provided do not show clear evidence
that these forms function like applicative(s).
Despite Dimmendaal’s (1983) revised analysis for Turkana, Barasa (2017: 129)
asserts that for closely related Ateso, -ɪa/-io increases valence by adding an instrument
(e.g., from ‘pour’ to ‘pour with [sth.]’). The instrumental applicative can follow the dative
to doubly increase valence (pp. 142–143). The instrument AppP carries ‘instrumental’
tone, a reduced preposition k=, and linearly follows the base object, all seen in (79).26
Whether -ɪa/-io increases valence when the preposition is retained needs research.

25 Cohen (2000: 9) presents a Bari suffix -a that he simply glosses as “applicative” without further ex-
planation.
26 Ka has senses of ‘genitive’, ‘with/instrument’, ‘location’, ‘comitative’, ‘and/addition’ (Barasa 2017:
106). Tone on the following noun sometimes helps distinguish among these functions (pp. 164–173).
 23 Nilotic applicatives 807

-Ia/-io can occur without a lexical instrument phrase if the referent is understood from
context.

(79) Ateso (Barasa 2021: 173)

à-tùb-ìò émàèmbèt k=èkíléŋ
3-cut.pst-ins mango prep=knife.ins
‘S/he was cutting the mango with a knife.’

For Maa (Maasai), Tucker and Mpaayei (1955: 157) discuss a morpheme complex
-arɛ/-ore which they say describes “neuter [middle –DP] action by means of a specific
instrument, or directed to a specific person or place”. Historically, -ɛ/-e would appear
to be a valence-increasing element added to the middle -a(r)/-o(r) (which also expresses
reflexive/reciprocal). Compare the argument structures of ɛl ‘smear’ in (80) with an
oblique instrument, versus ɛl-arɛ ‘smear self with’ in (81) where the instrument is the
grammatical object.

(80) Maasai (fieldnotes)

ɛ-gɪ́rá-ɪ́ áa-ɛl ɛn=kɛ́ráí tɛ ɪlatá
3-prog-impers inf.pl-smear fsg=child obl oil.nom
‘The child is being smeared by/with oil.’

(81) Maasai (fieldnotes)

ɛ-gɪ́ra ɛn=kɛráí kɪ́tɪ a-ɛl-arɛ́ ɛn=kapianá
3-prog fsg=child.nom small.nom inf.sg-smear-mid.ins fsg=milk.fat
‘The little child is smearing herself with milk fat.’

In non-middle morphological contexts, Maa uses the instrumental applicative -ie(k), as

in (82). This is distinct from the /ɛ/~/e/ element in -arɛ/-ore in having consistent +ATR
behavior, and it has a final /k/.27 -Ie(k) can add arguments with roles of instrument,
means, comitative, reason, source, or location.

(82) Maasai
e-te-yiaŋ-ak-ɪ́ ɔl=kɪ́tɛ́ŋ o-el-íék-i
cvb3-pf-slaughter-pf-impers msg=bovine msg.rel-smear-ins-impers
ɔ-sɪ́nya
msg.rel-be.perfect
‘when a perfect ox for smearing him with [its fat] has been slaughtered’
(enkeeya.017)

27 I know of no suffixes following -arɛ/-ore, which makes it difficult to test for old consonants after
/ɛ/~/e/.
808 Doris L. Payne

There is no difference in syntactic status between base objects and applied instruments
with -ie(k). Speech-act objects (base or applied) are indexed on the verb, but if the AppP
is human, it is typically interpreted as a causee.28 In (83), the instrument ‘cups’ follows
the subject and is before the theme ‘tea’, but this order can vary (Payne 2022a).

(83) Maasai
m-e-ok-íé oshî ol=porrór l=áŋ in=kikompení sháai
neg-3-drink-ins always msg=age.set.nom mpl=our.nom fpl=cups tea
‘Our age-set never drinks tea using cups.’ (enkang-enkai 1.112)

-Ie(k) can combine with three-argument stems to create four-argument clauses. Compare
(84a–b) with the root pɪk ‘put’. In (84b) the AppP ‘that ox’ is fronted before the applica-
tive verb, while the two just-established 3rd person objects ‘bracelets’ and ‘his sons’ are
definite nulls.

(84) Maasai
a. amʊ̂ ɛ́-ɪ́dɪ́p-á apá ɔl=páyian
because 3-finish-pf before msg=elder.nom
a-tɪ-pɪ́k-a ɪl=ayîôk l=ɛnyɛ́nak ɪl=kataarrí
inf.sg-pf-put-pf mpl=boys m=3sg.psr+pl.psm mpl=metal.bracelets
‘because the man had already put bracelets on his sons’
b. amʊ̂ ɪ́lɔ̂ kɪ́tɛ́ŋ náají apá e-pik-íék-i
because that.m bovine possibly before 3-put-ins-impers
‘because that could be the ox he uses to put them
[bracelets] on them [the sons]’ (enkeeya.038a–b)

The base object of transitive la(k) ‘untie, pay’ can be a debt/item paid for, as in (85), or
the means used to pay with, as in (86). In (87), the instrumental plus dative applicatives
on la(k) create a four-argument clause.

(85) Maasai
amʊ̂ ɛ-tá-lá-á e=síle
because 3-pf-pay-pf fsg=debt
‘because they have paid the (marriage) debt’ (enkiama.043)

28 With Maa Class II verb stems, -ie(k) is the only way to create a morphological causative. Class I verb
stems take a prefixal causative (Tucker and Mpaayei 1955).
 23 Nilotic applicatives 809

(86) Maasai
tá-la-a ɛ=lʊ́ kʊ́ nyá ɛ́=n=kɪ́tɛ́ŋ
imp.sg-pay-sbjv fsg=head fsg.psr=fsg=bovine
‘You pay the head of the cow.’ [to compensate a forgiven homicide] (iloikop.052)

Given that la(k) has two base senses, ‘pay price’ and ‘pay debt’, it is not entirely clear in
(87b) whether -ie(k) licenses adding the debt ‘school (fees)’ versus ‘money’ (expressed
as a definite null). Regardless, the double applicative clause has four participants: ‘I’ as
agent=subject, ‘child’ as beneficiary=dative applied object, ‘money’ as a definite null
instrument object, and ‘school (fees)’ as the overt debt object.

(87) Maasai
a. nɛ́ākʊ̄ ɛ́n-chɔɔ́=kɪ náají ɪ=ropiyianí é=síl . . .
so pl.sbjv-give=1sg.obj possibly fpl=money of=debt
‘so possibly give me some money as a loan . . .’
b. m-a-ta-la-ákín-yíé ɛn=kɛ́ráí sukúul
prosp-1sg-sbjv-pay-dat-ins fsg=child school
‘so that I may pay school (fees) with it for the child’ (ilomon.0099)

Similar to Maa which has an instrumental form with r after a middle but -ie(k) else-
where, Lopit has two instrumental applicatives: -rɪ and -ije (j is a palatal glide; Moodie
and Billington 2020: 194–195). -Ije can also express result and habituality.

4.2 Southern Nilotic instrumental/locative applicatives

Various SN verb suffixes have been glossed as ‘locative’ and ‘instrument’. Whether these
are all etymologically related (or are allomorphs), or whether any are cognate with the
EN -(r)ɪ/(r)ɛ forms or with Maa -íé(k), has not been determined.

4.2.1 Kalenjin

Nandi has an applicative -e: which Creider (2002: 176, 186) states is etymologically
related to a semantically broad preposition e:ng (p. 179). The suffix adds an argument
with roles of instrument (including ‘means’ and ‘amount’); location (including static
locations like ‘in, on, around’, as well as motion-related ‘path’, ‘source’, ‘goal’ – but not
‘clear endpoint’); with respect to; and reason.29 Compare (88)–(89) and (90)–(91).

29 Creider includes “benefactive” as a function of -e:. The example with -e: that he identifies as benefac-
tive might be ‘The child feels sympathy for the calf’ (p. 178); but ‘calf’ could be considered stimulus. The
Nandi dative expresses prototypical beneficiary.
810 Doris L. Payne

(88) Nandi (Creider 2002: 176)

ká:-wí:r sê:sé:t
pst1+1sg-throw dog
‘I threw at / hit the dog.’

(89) Nandi (Creider 2002: 176)

ká:-wí:r-ê: ké:tít sê:sé:t
pst1+1sg-throw-ins stick dog
‘I threw a stick at the dog / I hit the dog with a stick.’

(90) Nandi (Creider 2002: 178)

ke:-et pé:k
inf-refuse water
‘to refuse water’

(91) Nandi (Creider 2002: 178)

ke:-et-e: pé:k
inf-refuse-ins water
‘to refuse water to (s.o.)’

Akie has what König, Heine, and Legère (2015: 53–54) simply call an “applicative”. It has
four forms (plus ATR variations) depending on interaction with aspect/mode, person,
and number: -è, -èèn, -èyyà, -èèyyèn. It can add a typically inanimate locative or instru-
ment. The following pair shows prepositional and applied paraphrases for instrument.

(92) Akie (König, Heine, and Legère 2015: 54)

i llúú-e nen rúnkuu láákwɛɛ
2sg beat-ipfv loc club.acc child.acc
‘You are hitting the child with the club.’

(93) Akie (König, Heine, and Legère 2015: 53)

i llúú-eyyen rúnkuu láákwɛɛ
2sg beat-appl club.acc child.acc
‘You are hitting the child with the club.’

4.2.2 Datooga

Discussions of Datooga instrument/location applicatives mention forms ending in /n/

and /w/. It is not currently clear whether there are two etymologically separate ‘instru-
ment/locative’ applicatives in this sub-branch or just one.
 23 Nilotic applicatives 811

For G/B Datooga, Bruckhaus (2021: 129) introduces -a(n)/-aaw as a single applicative
and describes the distribution as morphologically conditioned: “While -an surfaces in
combination with simplex forms and ventive stems, the variant -aaw occurs on itive
stems. The mid-low-vowel form of the allomorph -aaw is the -ATR variant, and the
fronted reflex -ɛɛw surfaces on +ATR verbs”. The -an form may co-occur with the dative
(p. 123). He relates at least the -an element to Rottland’s (1982: 125, 184) discussion of a
Kalenjin -ɛɛ(n) instrumental and Omotik -een, and notes that Rottland reconstructs ✶-a
for Proto-SN.
The semantic role added by -an/-aaw can be instrument, means, source, site, or
inanimate concomitant. The intricate relationship between the added semantic role
and other elements is discussed as follows:

The role of the peripheral argument introduced by -an depends in particular on the lexical seman-
tics of the verbal root, on the semantics of the added noun, and on the presence of a directional
suffix on the verb. The site-introducing function of -an is confined to non-motion verbs and a few
stationary movement verbs. Source-indication occurs only in combination with itive and ventive
stems. . . . (Bruckhaus 2021: 127)

Examples (94)–(95) contrast an oblique and applied instrument; (96) shows an inani-
mate concomitant. Animates must be expressed with the oblique preposition sɛ́ɛ ‘with’
(p. 128).

(94) B/G Datooga (Bruckhaus 2021: 145)

qɔ ̀ɔ-ŋwáal àbà màttíiŋgòodà
3sbj-stir with cooking.stick
‘S/he is stirring with a cooking stick.’

(95) B/G Datooga (Bruckhaus 2021: 145)

góo-ŋòol-á màttíiŋgòodà
3sbj-stir.plrc-appl cooking.stick
‘S/he always stirs with a cooking stick.’

(96) B/G Datooga (Bruckhaus 2021: 128)

qá-fwáj-á ŋùtta
3sbj-run.away-appl spear
‘S/he runs away with a spear.’

For Gisamjanga Datooga, Kießling (2007: 135–136) also describes the use of -an for instru-
ment (including what I would call means), locative, and ablative applied arguments.
Interestingly, -an seems to interact with person of the object in some complex situations.
First, roots like rukt ‘tell sth., foretell, give away information/a secret’ can combine with
the itive to suppress the base goal and allow the theme to be the object (Section 5).
The simple itive stem is used for a 3rd person patient/theme object. To have a 1st or 2nd
812 Doris L. Payne

person patient object, -an must be added to the itive stem. Kießling presents (97)–(98)
to support these claims.30

(97) G Datooga (Kießling 2007: 135; partially my glossing)

qáyí rúkt-à ʃèedḁ
long.ago tell.itv-is moon.nom
‘(Finally, a small thing killed him) as had been foretold long ago by the moon.’

(98) G Datooga (Kießling 2007: 135; partially my glossing)

àd-ù-rúktàn-àan
neg.sbjv-sbj2.sg-tell.itv.appl-1sg.obj
‘Don’t sell me out!’ / ‘Don’t tell anything about me!’

Unlike Brookhouse who treats -an and -aaw as allomorphs, Griscom (2019: 236) treats
these as distinct morphemes in Asimjeeg Datooga. He characterizes -an as a semanti-
cally somewhat opaque ‘oblique’ applicative (obl.appl) for AppP objects of any person
(p. 126, 240–241), mostly with roles of time and location. According to Griscom, in (99)
it adds a time argument; (100) has the same itive form of ‘send’ without -an and with an
adverbial ‘often’ rather than a specific time expression.31

(99) Asimjeeg Datooga (Griscom 2019: 126)

g-ʷà-jéʃ àní:n gʷátʃ g-ò-tʃàg-d-án-à:n gʷàláɲ-ànd
aff-3-say 1sg that.time aff-3-send-itv-obl.appl-1sg elder-sg
‘He said that me, at that time, the elder sent me.’

(100) Asimjeeg Datooga (Griscom 2019: 105, 128)

nìɲ ɲíːs g-ó-tʃág-d síː-d
3sg often aff-3-send-itv person-sg
‘Often he (would) send someone (away).’

Griscom states that -an may occur in negative copular clauses, as in (101); but note that
here also it correlates with roles of location or time.

30 He further states that verbs with a goal in their basic predicate frame can use -an to highlight a
specific 3rd person patient (p. 136); however, the argumentation is cursory.
31 Griscom (2019) has examples of other verbs with gʷátʃ and without an applicative. Whether -an
licenses ‘that time’ in (95), as opposed to just highlighting it, needs study.
 23 Nilotic applicatives 813

(101) Asimjeeg Datooga (Griscom 2019: 119)

nàʃ-ò:d máʃí:n hídʒ m-ànd-án dàràb-è:t
grind-nmlz machine loc.dem neg-3-cop-obl.appl wilderness-sg
‘There was no grinding with a machine in the wilderness at that time.’
(Lit. ‘No grinding machine there / that time (in) the wilderness.’)

Asimjeeg uses -an with a 1st or 2nd person goal (fictive or literal) or addressee, contrast-
ing with the dative for 3rd persons for this function. Compare (102)–(104), which carry
-an, with (48)–(49) above.32

(102) Asimjeeg Datooga (Griscom 2019: 230)

g-è:-nàl-nál-à:d-àn-é:s-àj sí-ɲàg-d-í:n
aff-impers-teach-teach-am.itv-obl.appl-2pl-plrc 2pl-meet-itv-dat
‘We were taught (again and again) when we met (them).’

(103) Asimjeeg Datooga (Griscom 2019: 279)

q-à-ŋùl-àn-a:n qàŋ-d-éɲ
aff-3-see-obl.appl-1sg eye-sg-1sg.poss
‘They will look at me in my eye.’

(104) Asimjeeg Datooga (Griscom 2019: 271)

q-á-ŋúl-àn-è:ɲ qàŋ-d háw díjá
aff-3-see-obl.appl-2sg.obj eye-sg big a.lot
‘They look at you with bad eyes.’

In contrast to -an, Asimjeeg Datooga -e:(w)/-ɛ:(w) is used for AppP objects with roles of
location as with ʃúl ‘school’ in (105); manner (106); and animate and inanimate accom-
paniment (107)–(108). Griscom states that this form also can add an instrument, but he
presents no examples of this (Griscom 201: 241–242). Here I retain Griscom’s gloss of
loc for -e:(w)/-ɛ:(w).

(105) Asimjeeg Datooga (Griscom 2019: 127)

q-à:-sɔ ̀m-ɛ́:w ʃúl èd máŋól!á
aff-3-study-loc.appl school prep Mang’ola
‘I studied at the school in Mang’ola.’

32 Regarding addressee, also recall Griscom’s comments about (50) above. The ventive is used for 1st or
2nd and the dative for 3rd person beneficiary.
814 Doris L. Payne

(106) Asimjeeg Datooga (Griscom 2019: 242)

g-é:-sín-é: fúqár-è:d
aff-impers-do-loc.appl cunning-sg
‘It was created with skill/cunning.’

(107) Asimjeeg Datooga (Griscom 2019: 242)

∅-jád-é:w múʃódà-k múʃ-k
2sg-send-loc.appl bag-pl skin-pl
‘(You) send them with animal skin bags.’

(108) Asimjeeg Datooga (Griscom 2019: 242)

g-ʷá-líl-é:w ásìmdʒ-é:g
aff-3-sleep-loc.appl asimjeeg-pl
‘They sleep with the Asimjeeg.’

There is sketchy information relevant to possible EN cognates of the SN -an applicative.

For the Barian variety Mundari, Lutwori et al. (2013: 110) briefly note a suffix -an/-än/-on
‘for the sake of someone/for some reason’. The discussion and examples are hard to
parse and free translations are sometimes approximate. Nevertheless, brief examples
like bo’de ‘stop’ versus bobo’dan ‘stop for the sake of [an elephant in the road] (p. 175)
are suggestive.33

4.3 Western Nilotic

The literature on most WN languages is silent about applicatives with instrumental

function, suggesting such may not exist. In Shilluk, however, a syntactically optional
valence-increasing applicative verb stem adds instrument, accompaniment and loca-
tion. Remijsen, Miller-Naudé, and Gilley (2016: 222) and Remijsen and Ayoker (2018)
detail stem-internal variations according to verb class, showing that applicative forms
expressing these meanings are different from those that cover dative applicative mean-
ings (Section 3.3) For example, (110) contains an instrumental applicative form of ‘eat’,
but the dative stem for ‘eat’ is short-voweled cʌm (p. 225). Syntactically, the Shilluk
instrument AppP precedes the verb, displacing the patient object that normally occurs
there. Compare (109)–(110).

33 The Turkana verb suffixes -een (for dynamic verbs) and -aan/-oon (for statives) express habitual, iter-
ative, or protracted-in-time aspect (Dimmendaal 1983: 107). Maa -an derives abstract nouns from stative
verbs. These are not applicative in nature, despite phonological similarity to SN -an.
 23 Nilotic applicatives 815

(109) Shilluk (Remijsen, Miller-Naudé, and Gilley 2016: 222)

kwʌ̄n á-cấm kɪ́ pâal
porridge pst-eat prep spoon
‘Somebody ate the porridge with a spoon.’

(110) Shilluk (Remijsen, Miller-Naudé, and Gilley 2016: 222)

pâal á-cāaam kwʌ̄n
spoon pst-eat.ins porridge
‘Somebody ate the porridge with a spoon.’

The instrumental stem can also indicate accompaniment, time and location. Though
the verb stem is the same, a locative AppP requires a focus particle; compare (111)–
(112) with (109)–(110).

(111) Shilluk (Remijsen, Miller-Naudé, and Gilley 2016: 222)

kwʌ̄n á-cấm kɪ ̀ kâal
porridge pst-eat prep cattle.camp
‘Somebody ate porridge in the cattle camp.’

(112) Shilluk (Remijsen, Miller-Naudé, and Gilley 2016: 223)

kâal-à á-cāaam kwʌ̄n
cattle.camp-foc pst-eat.ins porridge
‘The cattle camp is where somebody ate the porridge.’

In WN Luo, a verbal enclitic =gô/=gódô expresses ‘with it’. As a clitic, this arguably
should not be considered an applicative.

(113) Luo (Tucker 1994: 348; my clitic indication)

ó-dwa̍ tedo-ná=gó(dó) kuon
3ipfv-want cook-dat.1sg=ins.3 mush
‘She wants to cook mush for me with it.’

5 Directionals as applicatives
Nilotic languages have itive and ventive directionals. These have a “redirecting” (Kiyo-
sawa 2006) applicative function (usually without increasing valence), but only with
a restricted set of verbs. In particular, this happens with transitive verbs having a
base <agent source/goal> argument frame. A directional derives an <agent theme>
stem. Any goal or source then occurs in an oblique phrase. Payne (2022b) presents
a cross-Nilotic study of this phenomenon and argues it can be explained by an earlier
816 Doris L. Payne

associated motion function of the directionals which perforce profiles a moving theme.
In SN, the directionals have additional valence-related applicative functions and extend
into marking beneficiary.
The directionals almost certainly reconstruct to Proto-Nilotic (Reh 1996: 261). The
Proto-Nilotic ventive likely had a high vowel plus nasal. The Proto-EN and Proto-SN
itive likely had a coronal consonant plus low vowel. The Proto-WN itive possibly had a
breathy vowel and final-consonant allomorphs. Payne (2021: 704) summarizes various
reconstructions.

5.1 Eastern Nilotic

All EN languages have a ventive containing the sequence /ʊn/ or /un/ (plus allomorphs).
Spagnolo (1933: 146) describes the Bari itive as -rVʔ, Turkana has -arɪ̥ , Ateso has -Vr(ɪ),
Lopit has an itive containing /ɾɪ/ (segmentally like one of its instrumental applicative
forms) plus other variants, and the Maa itive is -áa but with allomorphs that include
r. Not all sources give sufficient examples to show the redirecting applicative function
(e.g., Lutwori et al. 2013: 106–110 for Mundari; Moodie and Billington 2020 for Lopit),
but I expect this function exists throughout EN. The following from Maa with an <agent
source> root are representative.

(114) Maasai (fieldnotes)

á-púrr ol=dúka
1sg-rob msg=shop
‘I (will) rob the shop.’

(115) Maasai (fieldnotes)

n-é-purr-óo ɔl=áyíóní il=mósorr
cn-3-rob-itv msg=boy.nom mpl=eggs
‘The boy stole / will steal eggs.’

(116) Maasai (fieldnotes)

á-púrr-ú ɛnk=alámu tɔ l=dúkâ
1sg-rob-ven fsg=pen obl msg=shop.nom
‘I (will) steal a pen from the market.’

Examples (117)–(118) demonstrate the redirecting function with the <agent goal> root
naŋ ‘hit by throwing at’. In (117c), the subject is 3rd person, and the definite null object
of naŋ is the anaphoric goal ‘house’, mentioned in (117a–b). What hits the house is not
expressed or anaphorically understood. The house is clearly not the item thrown.
 23 Nilotic applicatives 817

(117) Maasai
a. n-é-jo á-nyɪ́k-ákɪ ɛnk=áŋ
cn-3-try inf.sg.sbjv-approach-dat fsg=home
‘He (a warrior) tried to approach to the home.’
b. n-ɛ́-ɪ́ŋat-áa
cn-3-withdraw-itv
‘and it [a magical house] withdrew’
c. n-ɛ́-naŋ
cn-3-throw.at
‘He (the warrior) hit it (the house, by throwing).’ (enamuke2.0049)

In contrast to (117c), the itive occurs on naŋ in (118). Now the syntactic object is the
theme which undergoes movement.

(118) Maasai
ɪ́-wa taá ɛnâ kɛráí shɔ́mɔ tá-naŋ-á-í
2-take.sbjv emph this.f child go.sbjv imp.sg-throw.at-itv-sbjv
‘Take this child and go throw it [the child] away.’ (kitejine.040)

5.2 Southern Nilotic

SN languages have ventive forms with /n/ and itives with a coronal stop. The same redi-
rective applicative function occurs with the same verb root types. However, the direc-
tionals have additional applicative functions not so far attested for directionals in other
Nilotic branches.

5.2.1 Kalenjin

Directionals can derive <agent theme> verbs from <agent source/goal> verbs in both
SN sub-branches. In (119) from Nandi, sè:sé:t ‘dog’ is the base goal object of the action.
Given the lexical meaning of wi:r ‘throw at’, it is implicit that something is thrown but
this is not expressed with the simple root. In (120) with the itive, koytà ‘stone’ is the
theme object that undergoes movement; now the goal is suppressed.

(119) Nandi (Creider 2002: 184)

keː-wiːr sèːséːt
inf-throw.at dog
‘to throw at the dog’
818 Doris L. Payne

(120) Nandi (Creider 2002: 184)

keː-wiːr-tá koytà
inf-throw.at-itv stone
‘to throw the stone thither’

Beyond the redirecting applicative function, Creider (2002: 172) documents use of the
Nandi itive -ta for an applied co-theme comitative (never co-agent); compare (121)
and (122).

(121) Nandi (Creider 2002: 172)

am kímyé:t
eat polenta
‘Eat polenta!’

(122) Nandi (Creider 2002: 172)

ke:-am-ta kímyé:t ínkwê:k
inf-eat-itv polenta vegetables
‘to eat polenta with vegetables’

A major applicative use of the SN ventive is for speech-act beneficiary/goal. This is

illustrated by the Cherang’any examples in (123)–(124) (Mietzner 2016: 129–135).
Cherang’any indexes 1st and 2nd person objects by suffixes. Thus, applied beneficiary
objects are marked by a complex involving the ventive plus an object suffix (which
Mietzner calls an “indirect object” but apparently due to the semantics of beneficiary/
goal). In contrast, 3rd person beneficiary objects are expressed by the dative applica-
tive without person indexation (Mietzner 2016: 106–111, 134; cf. Section 3.2).

(123) Cherang’any (Mietzner 2016: 111)

kwɔŋ-wééc cebióse-ɲu kɔ ̀ɔ ̀sɔ ̀ɔ ̀lɪ ̀ɲ
3.cook-ven.1pl wife-psr.sg:psm1sg evening.loc
‘In the evening my wife cooks food for us.’
(Lit. ‘My wife cooks toward us in the evening.’)

(124) Cherang’any (Mietzner 2016: 111)

ʌ̀-wek-uúŋ kóngoi
1sg-return-ven.2sg thanks
‘I will thank you.’
 23 Nilotic applicatives 819

5.2.2 Datooga

Kießling (2007: 136–137) describes partially similar redirecting facts for the Gisam-
janga Datooga itive with verbs that he says allow both a patient and a goal in their
predicate frame. The itive suppresses the goal, leaving just the patient.34 Bruckhaus
(2021: 67) mentions that the itive removes the goal from the <agent goal (theme)>
argument structure of daw ‘give’ (presumably he places theme in parentheses because
it is implicit in the non-itive form).
Similarly to Cherang’any, the Gisamjanga Datooga ventive is obligatory for 1st and
2 person beneficiary/goal, while the dative occurs for 3rd person (Kießling 2007: 134,
nd

138). However, Bruckhaus (2001) notes that the itive can at least sometimes occur for 3rd
person goal/beneficiary, in paradigmatic opposition with both the ventive and dative.
For instance, with the verb qaw ‘milk’ in (125), the ventive stem gaw-un expresses a 1st
or 2nd person beneficiary, the itive stem qaw-d indicates the action is done for a nonspe-
cific 3rd person beneficiary, and the dative stem gaw-s marks a specific 3rd person ben-
eficiary. The final suffixes in (125b–e) index person and number of the applied object.

(125) B/G Datooga (Bruckhaus 2021: 165)

a. qá-qàw ‘s/he is milking’ simplex
b. qá-qàw-d-à ‘s/he is milking for another person’ itv
c. gá-gàw-n-èey ‘s/he is milking for you (sg)’ ven
d. gá-gàw-n-ɛ́ɛsà ‘s/he is milking for us’ ven
e. gá-gàw-s-à ‘s/he is milking for her/him’ dat

For Asimjeeg Datooga, directionals can modify the verb meaning and/or increase valence
by adding a new object (Griscom 2019: 225). Examples (126)–(127) show the ventive licens-
ing speech-act addressees. The addressee is perhaps just implied in (128). But unlike G/B
Datooga where apparently only the ventive occurs for speech-act adressees, Griscom
(2019: 236–237) notes some variability in use of dative and ventive for this function.35

(126) Asimjeeg Datooga (Griscom 2019: 107)36

qámá-t-àŋʷ g-íː-g-ʷá-rúŋ-n-òːɲ ŋàʃ-ǃánd
mother-sg-2pl.poss aff-fut-aff-3-say-ven-2sg word-sg
‘. . .your mother will tell you something. . .’

34 Kießling does not discuss whether the verbs are syntactically trivalent, allowing both patient and
goal as core objects. His examples all carry a directional suffix.
35 He also observes that across Datooga studies, examples with ventive versus dative for applied goal/
beneficiary/addressee are not controlled for verb root, nor for aspect which applicative/directional
forms can help express.
36 Griscom (2019: 107) glosses -n as OBL in (126); based on personal communication (1/11/2022) I gloss
it here as VEN.
820 Doris L. Payne

(127) Asimjeeg Datooga (Griscom 2019: 237)

àníːn g-àː-gàs-àj dà-rùg-n-óːg gìdàb m-ád-gʷ-á-nd
1sg aff-1sg-want-plrc 1sg-tell-ven-2pl comp neg-pers-aff-3-cop
‘I want to tell you all that there isn’t anything anymore. . .’

(128) Asimjeeg Datooga (Griscom 2019: 221)

∅-rúŋ-ní
2sg-tell-ven
‘Tell (me).’

Griscom argues that the ventive can increase valence in Asimjeeg Datooga. In (129),
the root ʃa occurs with no directional, meaning ‘buy’. In (130), ʃa occurs with the itive,
yielding ‘sell’. In (131) it occurs with the ventive but now also with a 1sg object suffix.
This is an impersonal construction which has a prefix resembling that of a 1pl subject
(Griscom 2016: 180); but the end result in (131) has two objects, ‘shawl’ and 1sg.

(129) Asimjeeg Datooga (Griscom 2019: 225)

g-ì-dà-ʃà déː-d ...
aff-fut-1sg-buy cow-sg
‘I will buy a cow. . .’

(130) Asimjeeg Datooga (Griscom 2019: 225)

q-àː-ʃà-d dájéːg íːjèɲ
aff-1sg-buy-itv goat.pl two
‘I sold two baby goats.’

(131) Asimjeeg Datooga (Griscom 2019: 225)

g-éː-ʃà-n-àːn háŋ-d
aff-impers-buy-ven-1sg shawl-sg
‘I was bought a shawl.’

Griscom (2019: 226) suggests the itive can also increase valence by adding an instru-
ment to bar ‘hit’ in (132). However, this single example is not entirely convincing; the
itive perhaps creates the meaning of ‘farm’ (lit. ‘hit away’) and hence an instrument
may be contextually evoked because, pragmatically, someone must use something to
‘hit away’ with.

(132) Asimjeeg Datooga (Griscom 2019: 226)

m-à-nd-án gísír-dʒànd m-ɛɛ:-bár-dà
neg-3-cop-obl.appl hoe-sg neg-impers-hit-itv
‘There weren’t any hoes, they weren’t used to farm.’
(Perhaps lit. ‘They weren’t hit away / [people] didn’t hit away’.)
 23 Nilotic applicatives 821

5.3 Western Nilotic

For WN, Reh (1996: 261) suggests that ventive ✶-V(̀ V)[-BRV]n or ✶-nV(̀ V)[-BRV]) and itive ✶-tV ̀
or ✶-Vt̀ forms reconstruct to Proto-Nilotic; these correspond to intransitive directional
forms in WN. Transitive verbs have distinct directional forms, ventive ✶-V́[+BRV] and
itive ✶-V́[-BRV]; Reh suggests these reconstruct to Proto-WN. Modernly for various WN
languages and verb classes, ventive and itive versus base roots may be marked just
by tone, alternation or deletion of a final stem consonant, and stem-internal vowel
changes.
Not all discussions of WN directionals indicate whether they have applicative
functions (cf. Reh 1996: 249–258 on Anywa; Reid 2019 on Nuer). However, Agar Dinka
(Andersen 1992–1994; Andersen 2012b) and Mabaan (Andersen 1999) directionals
clearly have redirecting applicative functions. In (133) there is no directional and the
place or source from which something is removed is the object. In both (134) with the
itive and (135) with the ventive, the moved theme is the grammatical object.

(133) Mabaan (Andersen 1999: 109)

ʔɛ́kkɛ ̀n ʔʌ̄n wiiɛ̂j-ɛ́
3pl house sweep-pst.3pl.3
‘They swept the house.’

(134) Mabaan (Andersen 1999: 109)

ʔɛ́kkɛ ̀n ɟîik-én wiiɛ̂c-ɛ́
3pl rubbish-pl sweep.itv-pst.3pl.3
‘They swept the rubbish (into something).’

(135) Mabaan (Andersen 1999: 109)

ʔɛ́kkɛ ̀n ɟîik-én wɛ̂ɛw-w-ɛ́
3pl rubbish-pl sweep-ven-pst.3pl.3
‘They swept the rubbish hither.’

For Lango, Noonan (1992) mentions ventive derivations with certain motion verbs
(he is silent about an itive). Discussion of a potential valence effect is minimal, but
the ventive always “refers to motion toward the speaker” (p. 135). In the following
examples there is no indication of 1sg other than what is implied by the ventive
(compare Lango examples in Section 6 which overtly mark 1sg objects). Thus, we
might say the ventive increases semantic valence in at least cases like (136) and
(137); perhaps it just clarifies what the goal is in (138) since ‘send’ already implies a
destination.
822 Doris L. Payne

(136) Lango (Noonan 1992: 135)

àtɪ̂ n òrìŋŋô
child 3sg.run.ven.pfv
‘The child ran to me.’37

(137) Lango (Noonan 1992: 135)

yàt òmòllô
wood 3sg.float.ven.pfv
‘The wood floated toward me.’38

(138) Lango (Noonan 1992: 135)

Dákó òcwàllô búk
woman 3sg.send.ven.pfv book
‘The woman sent the book to me.’

6 Look-alike syntactic constructions

WN Lango distinguishes the dative applicative construction (Section 3.3) and a non-ap-
plicative “dative shift” construction. I am not aware of other Nilotic languages which
have such an alternation. In (139) the recipient has the preposition bòt ‘to’ and follows
the theme object. In (140) it occurs without a preposition directly after the verb and
before the theme. No object is indexed on the verb when the object is an NP (p. 141) or
is inanimate.

(139) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀yò búk bòt dákô
man 3sg.give.pfv book to woman
‘The man gave the book to the woman.’

(140) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀yò dákô búk
man 3sg.give.pfv woman book
‘The man gave the woman the book.’

37 Noonan (1992: 125) subsumes the non-ventive counterpart of (136) to his “activity naming” verb
form which “refer[s] solely to a subject’s participation in an activity, but not activity directed toward any
particular object”. At least some such forms are antipassive in sense.
38 Noonan (1992: 125) relates the non-ventive counterpart of (137) to his “secondary argument” (SA)
form, in which “[t]he argument that would be the DO of a transitive is the Su[bject] of the corresponding
SA form”. These appear to be anticausatives.
 23 Nilotic applicatives 823

At most one object is indexed on the Lango verb at a time, but this can be theme, or recip-
ient in the “dative shifting” construction. In (141) the human recipient is indexed and the
non-human theme is expressed as a zero. In (142) the human theme is indexed while the
recipient is in a prepositional phrase headed by bòt. Note the contrast between òmɪyɛ́ ̀ in
(142) versus òmɪyò̀ in (140) above; and between òmɪyɛ́ ̀ and òmɪyá
̀ in (143). Preferential
indexing of human (or perhaps animate) objects on the verb is generally the norm in Nilotic.

(141) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀yá
man 3sg.give.pfv.1sg
‘The man gave it to me.’

(142) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀yɛ́ bòtə́
man 3sg.give.pfv.3sg to.1sg
‘The man gave him (e.g., a slave) to me.’

(143) Lango (Noonan 1992: 121)

lócə ̀ òmɪ ̀yá búk
man 3sg.give.pfv.1sg book
‘The man gave me the book.’

Per definitions in Zúñiga and Creissels (this volume), the EN Lopit and Maa external pos-
session constructions are morphologically unmarked but syntactic “applicative-look-
alike” constructions. Maa verbs are almost never labile but in the external possession
construction, the clause has one more object than normally allowed by the valence of
the root or stem (Payne 1997). If the base verb is intransitive, as in (144), the possessed
item in the external possession construction is the subject, and the possessor NP is the
grammatical object, indexed on the verb if it is a speech-act participant, as in (145).
Thus, valence increases with adding a possessor object. If the base is transitive, the pos-
sessor is the primary grammatical object, again increasing valence. This construction
expresses that there is some effect on the possessor, whether positive or negative.

(144) Maasai (adapted from Payne 1997: 104)

é-íshú en=kíne
3-be.alive fsg=goat.nom
‘The goat is alive.’

(145) Maasai (adapted from Payne 1997: 104)

áa-ishú en=kíne
3>1sg-be.alive fsg=goat.nom
‘My goat is / will be alive (and I am benefited thereby).’
824 Doris L. Payne

EN Lopit (Moodie and Billington 2020: 287–288) and WN Jumjum (Andersen 2019) are
similar to Maa in treating an external possessor as the object with no derivational
marking on the verb. The phenomenon has not, to my knowledge, been explored for
other Nilotic languages.

7 Non-applicative and lexicalized functions

of the Nilotic morphology
Finally, we comment on major non-applicative functions of morphology that also has
applicative functions.
Non-applicative uses of the dative are minimally discussed for most Nilotic lan-
guages. In Maa (EN), the dative can mark ‘intensity, persistence’ or intention to do some-
thing with translational-motion verbs like lo(t) ‘go’, kʊɛt ‘run’, sʊj ‘follow’. These can
take geographical goals with no special marking, as in (146).

(146) Maasai (fieldnotes)

áa-sʊ́ j ol=dóínyó
3>1sg-follow msg=mountain
‘S/he will follow me to the mountain.’

With the dative, the otherwise identical sentence might imply that the one following me
has bad intentions and I am seeking refuge in the mountain, as in (147).

(147) Maasai (Wuasinkishu dialect; fieldnotes)

áa-sʊj-akɪ́ ol=dóínyó
3>1sg-follow-dat msg=mountain
‘S/he will track me (all over, through whatever routes I might take) to the
mountain.’

Also note the multiple readings possible in (148). The second reading has the same argu-
ment structure as (146), but with a greater sense of intensity.

(148) Maasai (Wuasinkishu dialect; fieldnotes)

áa-sʊj-akɪ́ ɛn=kítɛ́ŋ
3>1sg-follow-dat fsg=bovine
(i) ‘He will follow the cow for me.’ (e.g., it is lost and I am unable to go after it.)
(ii) ‘He will pursue me all the way to the cow.’ (e.g., I know that a dog pursuing
me is afraid of cows, so I run to a cow seeking safety, but the dog pursues
me all the way there.)
 23 Nilotic applicatives 825

The fact that sʊj ‘follow’ in its simple form can already occur with an overt goal pre-
sumably makes the dative available for an extended function. Whether the ‘intensity/
high-intention’ function extends to verbs without an implicit sense of goal in their
argument structure is unknown.
As we have seen, instrumental applicative forms have a range of semantic func-
tions across Nilotic languages, including location, reason, and more. In Maa, the -ie(k)
instrumental applicative is the causative with Class II verbs. Example (149) shows the
base <theme source> argument frame of ɪŋat ‘move away from’; (150) demonstrates the
valence-increasing causative effect of what is otherwise the instrumental applicative.

(149) Maasai (fieldnotes)

ɛ́-ɪ́ŋát-á ɛn=kɛ́jʊ ɛn=kɔ́p
3-move.from-pf fsg=leg.nom fsg=land
‘The leg has lifted up from the ground.’ (e.g., when running)

(150) Maasai (Wuasinkishu dialect; fieldnotes)

áa-ɪŋat-ie ɔl=ŋátúny
3>1sg-move.from-ins/caus msg=lion
‘S/he will make me flee from the lion.’

Section 5 discussed applicative functions of the directionals; but their core functions
are to express literal and fictional direction, orientation, spatial deixis, and associated
motion. In SN languages, directionals co-occur with a distinct associated motion mor-
pheme for the associated motion function; but in EN and WN, directionals by them-
selves function as such (Payne 2021).
Directionals also extend into the domain of aspect. The aspectual functions are rel-
atively less studied, but appear to be quite diverse from one language to another. For
instance, Spagnolo (1933: 143) says the Bari ventive can communicate ‘perfective/finish-
ing off’, while in other EN languages the ventive has likely developed into the inchoa-
tive. Payne (2021) provides one overview of the functions of directionals across Nilotic.
Perhaps most remarkably, we have seen that in SN, the ventive, itive, and even
the dative are beginning to participate in person indexation. In general, the ventive
correlates with 1st and 2nd person beneficiary and related notions, while the itive and
dative may correlate with 3rd person beneficiary and related notions. Most notably, for
G/B Datooga, Bruckhaus (2021: 118) states that though 1st and 2nd person beneficiaries
can be indexed on ventive verb stems by object suffixes, they can also be infered from
“blank” ventive stems.
Finally, in probably all Nilotic languages that have them, dative, instrumental and
directional affixes are lexicalized into some verbs. That is, certain simple root forms
do not occur without one of these (historical) affixes. Remarkably, this extends even to
verbs for ‘give’ in some Nilotic languages as in Nandi and Turkana, seen in (151)–(152).
This is so even when other three-argument roots do exist, such as Turkana ‘beg’.
826 Doris L. Payne

(151) Turkana (Dimmendaal 2009: 3)

kà-ɪ ̀n-ak(ɪ ̀) ŋesì ayɔŋ̀ ŋakipì
3>1-give-dat 3sg.nom 1sg water
‘S/he has given me water.’

(152) Nandi (Creider 2002: 174)

mé:-ka:-cín cî:
neg-give-dat person
‘(S/he) doesn’t give (it) to anyone.’

8 Summary of major findings

Much research remains to be done on the Nilotic family, and details about applicative
constructions are no exception. However, major findings that we can draw from the
available literature follow.

Inventory
– Dative applicatives occur in all Nilotic branches. Dholuo (WN) appears to have lost
this but may be newly developing one by cliticizing a preposition.
– Instrumental/locative applicatives are well-attested in EN and SN, but minimally
in WN. WN Shilluk has distinct instrumental/locative versus benefactive/dative
applicative constructions.
– SN Datooga languages may have a third -an applicative.
– Ventive and itive directionals have redirective applicative functions in all Nilotic
branches and additional applicative functions in SN. Serial or converbal construc-
tions are not used for applicative jobs. However, ventive and itive directional
affixes conceivably developed from movement verbs.

Morphology
– EN and SN applicative suffixes have considerable allomorphy, motivated by vowel
harmony and consonant changes including deletion. WN languages mostly (but not
exclusively) use stem alternations involving tone, length, vowel quality, and con-
sonant alternations and deletion. WN phonological verb classes affect applicative
stem forms, and Anywa has different directional forms for intransitive versus tran-
sitive verbs.
– Applicativized verbs behave like regular verbs in their inflection for person,
number, tense, aspect, mood.
– In EN and SN, dative and instrumental applicatives can co-occur.
 23 Nilotic applicatives 827

– Dative/benefactive applicatives generally do not co-occur with directionals. There

is no restriction on combining the instrumental with directionals in at least EN
and SN.

Syntax
– Dative and instrumental applicatives normally increase valence on both intransi-
tive and transitive bases.
– Animacy is likely significant in what is treated as the more privileged object, as
much or more so than semantic role and base versus applied status.
– Dative AppPs which express beneficiary/maleficiary and addressee roles are
primary/direct objects, being indexed on the verb and/or occurring in the privi-
leged object position.
– The status of dative AppPs that express goals or other locative arguments is more
varied in at least EN. In Maa they are the privileged object. But in Turkana and
Ateso they may occur with a preposition and/or in a locative case; this is also true
of WN Reel.
– With instrumental/locative applicatives, the AppP is usually the privileged object,
but in at least Ateso, the AppP keeps a reduced preposition plus has instrumental
tonal case. Little work has been done on the required versus optional nature of
location, time, and other semantically “oblique” AppP arguments.
– Dative and instrumental applicatives generally don’t change the syntactic status of the
companion arguments between the base and applied constructions (at least in lan-
guages with nominative-accusative syntax), aside from perhaps affecting linear order
(though order may depend on relative topicality). (Some WN languages are argued to
have ergative features.) WN Anywa and Jumjum are languages for which the applied
beneficiary is argued to syntactically displace the base-construction object.
– Directionals can have a redirecting applicative function with basic <agent source/
goal> verbs, without changing valence. The applied theme is then the primary or
direct object. The goal/source is no longer a core argument, and can only occur in
an oblique phrase.
– There are no known limitations on combining applicatives with voice-type opera-
tions such as antipassive, causative, middle, and impersonal/passive.
– Applied constructions generally correspond to the regular valence patterns of
a language. However, in at least some languages they can create four-argument
clauses which otherwise do not exist.
– With reference to case and verb indexation frames, in some languages all AppPs
occur in the unmarked case for primary/direct objects. In EN Turkana and Ateso
and WN Reel, the dative applicative licenses a locative AppP in a locative case form.
– In Shilluk, a pre-verbal focused location requires an applicative verb. (There is
little information on whether applicativization is required to relativize or focalize
otherwise non-core syntactic phrases.)
828 Doris L. Payne

Semantics and pragmatics

– Different applicative derivations are specialized for different ranges of semantic
roles. Dative applicatives add beneficiary/maleficiary, recipient, addressee, and
(specific/reached-)goal and occasionally other locative notions.
– The dative is typically obligatory for expressing a beneficiary/maleficiary partici-
pant. Adpositional phrases are not usually possible for this function but at least EN
Lopit is an exception. SN languages have two or more forms for applied benefac-
tives according to person (and sometimes specificity for 3rd persons) of the benefi-
ciary.
– The dative applicative is syntactically optional for expressing a goal with move-
ment and caused-motion verbs. It is sometimes lexicalized even with ‘give’.
– With some movement verbs the Maasai dative can yield a semantic nuance relative
to the semantic role of the object (goal-reached with the dative versus path with
the non-applied construction), or intensity of the action vis-à-vis the object, but
without changing valence. Nevertheless, no morphology has turned from erstwhile
applicative into a strictly valence-neutral function.
– In EN, SN, and WN Shilluk, the instrumental applicative can add a range of roles
such as instrument, means, concomitant, location, source, time, reason, result,
depending on interaction with lexical and contextual factors.
– Datooga varieties have a possible distinction between an ‘instrumental’ and a more
opaque ‘oblique’ applicative -an; their differentiation is complicated by interaction
with other morphemes including aspect and directionals.
– Little is known about what motivates the choice when a given semantic role can be
expressed in an applicative or prepositional construction. Sometimes it may have
to do with specificity, or topical versus new information status. In WN Shilluk, the
instrumental applicative is required if a non-core argument is focused in pre-ver-
bal position.
– Directionals have valence-neutral redirective applicative functions with <agent
source/goal> verbs, suppressing base goal/source in favor of theme.
– In some SN languages, directionals have extended into adding a comitative object.
– In SN, the ventive has specialized for adding speech-act beneficiary/addressee
objects.

Look-alike constructions
– WN Lango has an English-style “dative alternation”.
– EN Maasai and Lopit and WN Jumjum have external possession constructions.
These increase valence by adding an object, with no valence-increasing morphol-
ogy on the verb.

Finally, we must underscore that few extensive studies of Nilotic applicatives currently
exist, notable exceptions being Lamoureaux (2004) for EN Maasai, Dimmendaal (2009)
 23 Nilotic applicatives 829

on EN Turkana, and Bruckhaus (2021) for SN Datooga. Research is especially needed on

syntactic and semantic extensions, the interrelationship of applicatives with particular
verb types and participant features, and on discourse-pragmatic issues.

Abbreviations
acc accusative
aff affirmative
agr agreement
am associated motion
appl applicative
AppP applied phrase
asp aspect
ATR advanced/retracted tongue root
ben benefactive/beneficiary
brv breathy voice
caus causative
cn connective
comp complementizer
cop copula
cvb converb
decl declarative
dat dative
dem demonstrative
emph emphatic
EN Eastern Nilotic
erg ergative
evid evidential
f feminine
foc focus
fpl feminine plural
fs final suffix
fsg feminine singular
fut future
imp imperative
impers impersonal
incompl incomplete
inf infinitive
ins instrument
intr intransitive
inv inverse
ipfv imperfective
is inflectional suffix
itv itive
link linker
loc locative
830 Doris L. Payne

m masculine
mid middle
mpl masculine plural
msg masculine singular
neg negative
nmlz nominalzer
nom nominative
nts non-topical subject
obj object
obl oblique
pass passive
pers persistive
pf perfect(ive)
pfv perfective
pl plural
plrc pluractional
poss possessive
prep preposition
prf perfect
prog progressive
prosp prospective
prox proximal demonstrative
psm possessum
psr possessor
pst past
rec reciprocal
rel relative
rep repeated action
sbj subject
sbjv subjunctive
sg singular
SN Southern Nilotic
subs subsequent
tr transitive
ven ventive
WN Western Nilotic
x>y x acts on y

References
Andersen, Torben. 1988a. Consonant alternation in the verbal morphology of Päri. Afrika und Ubsersee 71.
63–113.
Andersen, Torben. 1988b. Ergativity in Päri, a Nilotic OVS language. Lingua 75(4). 289–324.
Andersen, Torben. 1992–1994. Morphological stratification in Dinka: On the alternations of voice quality,
vowel length and tone in the morphology of transitive verbal roots in a monosyllabic language.
Studies in African Linguistics 23(1). 1–63.
 23 Nilotic applicatives 831

Andersen, Torben. 1999. Vowel quality alternation in Mabaan and its Western Nilotic history. Journal of
African Languages and Linguistics 20. 97–120.
Andersen, Torben. 2012a. Spatial roles and verbal directionality in Dinka. Journal of African Languages and
Linguistics 33. 143–179. DOI 10.1515/jall-2012-0007.
Andersen, Torben. 2012b. Verbal directionality and argument alternation in Dinka. In Angelika Mietzner &
Ulrike Claudi (eds.), Directionality in grammar and discourse: Case studies from Africa, 35–53. Cologne:
Rüdiger Köppe.
Andersen, Torben. 2015. Syntacticized topics in Kurmuk: A ternary voice-like system in Nilotic. Studies in
Language 39(3). 508–554.
Andersen, Torben. 2018. The encoding of subjects and objects in Jumjum, a Nilotic OV language. Lingua
204.78–116.
Andersen, Torben. 2019. External possession of body-part nouns in Jumjum. Journal of African Languages
and Linguistics 40(2). 171–203.
Andrason, Alexander & Michael Karani. 2019. Dative applicative elements in Arusa (Maa): A canonical
approach to the argument-adjunct distinction. Stellenbosch Papers in Linguistics Plus 58. 177–204.
doi: 10.5842/58-0-842.
Barasa, David. 2017. Ateso grammar: A descriptive account of an Eastern Nilotic language. Munich: LINCOM.
Bruckhaus, Stefan. 2021. Motion in Datooga. Hamburg: University of Hamburg dissertation (revised
manuscript).
Cien, Mark Akec, Andrew Agok Igai Piiny, David Mabor Makuei Angong, John Malak Balang Kerjok, Daniel
Majier Johou & David Kuacreng Nhial. 2016. Reel grammar book. (Trial edition.) Juba: SIL-South Sudan.
Cohen, Kevin Bretonnel. 2000. Aspects of the grammar of Kukú. Munich: LINCOM.
Crazzolara, J. P. 1933. Outlines of a Nuer grammar. Wien: Anthropos.
Creider, Chet. 2002. The semantics of participant types in derived verbs in Nandi. Revue québécoise de
linguistique 31. 171–190.
Creider, Chet & Jane Tapsubei Creider. 1989. A grammar of Nandi. Hamburg: Helmut Buske.
Dimmendaal, Gerrit J. 1981. On verbal derivation in Nilotic: The case of Turkana. In Thilo C. Schadeberg &
M. Lionel Bender (eds.), Nilo-Saharan. Proceedings of the First Nilo-Saharan Linguistics Colloquium, Leiden,
September 8–10, 1980, 59–73. Dordrecht: Foris.
Dimmendaal, Gerrit. 1983a. The Turkana language. Dordrecht: Foris.
Dimmendaal, Gerrit. 1983b. The two morphological verb classes in Nilotic. In Rainer Vossen & Marianne
Bechhaus-Gerst (eds.), Nilotic Studies. Proceedings of the International Symposium on Languages and
History of the Nilotic Peoples, Cologne, January 4–6, 1982, 269–309. Berlin: Dietrich Reimer.
Dimmendaal, Gerrit. 2009. Datives in Nilotic in a typological perspective. Afrikanistik Aegyptologie Online.
https://www.afrikanistik-aegyptologie-online.de/archiv/2009/2355. urn:nbn:de:0009-10-23558.
Dryer, Matthew. 1986. Primary objects, secondary objects, and antidative. Language 62(4). 808–845.
Griscom, Richard. 2019. Topics in Asimjeeg Datooga verbal morphosyntax. Eugene: University of Oregon
dissertation.
Hammarström, Harald, Robert Forkel, Martin Haspelmath, & Sebastian Bank (eds.). Family: Nilotic.
Glottolog 4.4. Leipzig: Max Planck Institute for Evolutionary Anthropology. https://doi.org/10.5281/
zenodo.4761960 (Available online at https://glottolog.org/resource/languoid/id/nilo1247, Accessed on
2021–10–15.)
Kießling, Roland. 2007. Space and reference in Datooga verbal morphosyntax. In Doris Payne & Mechthild
Reh (eds.), Advances in Nilo-Saharan linguistics: Proceedings of the 8th Nilo-Saharan Linguistics
Colloquium. Hamburg, August 22–25, 2001, 123–142. Cologne: Rüdiger Köppe.
Kiyosawa, Kaoru. 2006. Applicatives in Salish languages. Vancouver: Simon Fraser University dissertation.
Köhler, Oswin. 1955. Geschichte der Erforschung der nilotischen Sprachen. Afrika und Übersee, Beiheft 28. Berlin:
Dietrich Reimer.
König, Christa. 2008. Case in Africa. Oxford: Oxford University Press.
832 Doris L. Payne

König, Christa, Bernd Heine & Kersten Legère. 2015. The Akie language of Tanzania, a sketch of discourse
grammar (Studies in Nilotic Linguistics 9). Tokyo: Research Institute for Languages and Cultures of
Asia and Africa.
Lamoureaux, Siri Van Dorn. 2004. Applicative constructions in Maasai. Eugene: University of Oregon MA
thesis.
Languages of Tanzania Project. 2009. Atlasi ya lugha za Tanzania. Dar es Salaam: Chuo Kikuu cha Dar es
Salaam [University of Dar es Salaam].
Lutwori, Allen Pitya, Enike Amina Wani, Lodu Philip Jembeke, Robert Gajuk Paul Wani, Martin Lomu Goke &
Augustino Laku Buli. 2013. Mundari grammar book (trial edition). Juba: SIL-Sudan.
Mel’čuk, Igor. 2006. Aspects of the theory of morphology. Berlin: Mouton de Gruyter.
Mietzner, Angelika. 2016. Cherang’any, a Kalenjin language of Kenya. Cologne: Rüdiger Köppe.
Miller, Cynthia & Leoma Gilley. 2001. Evidence for ergativity in Shilluk. Journal of African Languages and
Linguistics 22. 33–68.
Moodie, Jonathan & Rosey Billington. 2020. A grammar of Lopit: An Eastern Nilotic language of South Sudan.
Leiden: Brill.
Noonan, Michael. 1992. A grammar of Lango. Berlin: Mouton de Gruyter.
Nyombe, Bureng G. V. 1987. Argument bearing affixes in Bari. New York: City University of New York
dissertation.
Payne, Doris L. 1997. Argument structure and locus of affect in the Maasai External Possession
construction. Proceedings of the Annual Meeting of the Berkeley Linguistics Society. 23S Special Volume on
Syntax in African Languages, 98–115.
Payne, Doris L. 2015. Perspectives on Nilotic verb composition: Why can’t we agree on the Maa verb? In
Angelika Mietzner & Anne Storch (eds.), Nilo-Saharan – Models and descriptions, 211–229. Cologne:
Rudiger Köppe.
Payne, Doris L. 2021. The extension of “associated motion” to aspect and argument structure in Nilotic
languages. In Antoine Guillaume & Harold Koch (eds.), Associated motion, 695–746. Berlin: Mouton de
Gruyter.
Payne, Doris L. 2022a. Proximal to distal: Information flow and order in Maa. In Hiwa Asadpour & Thomas
Jügel (eds.), Word order variation: Semitic, Turkic and Indo-European languages in contact, 15–38. (Studia
Typologica 31). Berlin: De Gruyter Mouton.
Payne, Doris L. 2022b. The applicative(-like) function of Nilotic directionals: Introducing THEMES. In Sara
Pacchiarotti & Fernando Zúñiga (eds.), Applicative morphology: Neglected syntactic and non-syntactic
functions, 227–260. (Trends in Linguistics. Studies and Monographs 373). Berlin: De Gruyter Mouton.
Payne, Doris L., Mitsuyo Hamaya & Peter Jacobs. 1994. Active, inverse, and passive in Maasai. In T. Givón
(ed.), Voice in discourse, 283–315. Amsterdam: John Benjamins.
Reh, Mechthild. 1996. Anywa language. Cologne: Rüdiger Köppe.
Reid, Tatiana. 2010. Aspects of phonetics, phonology and morhophonology of Thok Reel. Edinburgh: University
of Edinburgh MSc dissertation.
Reid, Tatiana. 2019. The phonology and morphology of the Nuer verb. Guildford, UK: University of Surrey
dissertation.
Remijsen, Bert & Ayoker, Otto Gwado. 2018. Forms and functions of the base paradigm of Shilluk transitive
verbs. In Bert Remijsen & Otto Gwado Ayoker (eds.), A descriptive grammar of Shilluk, pp. 1–80.
̜(Language Documentation Conservation Special Publication 14). <http://hdl.handle.net/10125/24777>
Remijsen, Bert, Cynthia Miller-Naudé & Leoma Gilley. 2016. The morphology of Shilluk transitive verbs.
Journal of African Languages and Linguistics 37. 201–245.
Rottland, Franz. 1982. Die südnilotischen Sprachen: Beschreibung, Vergleichung und Rekonstruktion (Kölner
Beiträge zur Afrikanistik 7). Berlin: Dietrich Reimer.
Spagnolo, L. M. 1933. Bari grammar. Verona: Missioni Africane.
Toweett, Taitta. 1975. Kalenjin nouns and their classification. Nairobi: University of Nairobi MA thesis.
 23 Nilotic applicatives 833

Tucker, Archibald N. 1994. A grammar of Kenya Luo (Dholuo), Vols. 1–2. Edited by Chet A. Creider. Cologne:
Rüdiger Köppe.
Tucker, Archibald N. & Margaret A. Brian. 1962. Noun classification in Kalenjin: Päkot. African Language
Studies 3. 137–181.
Tucker, Archibald N. & Margaret A. Brian. 1964/1965. Noun classification in Kalenjin: Nandi-Kipsigis. African
Language Studies 5. 192–297.
Tucker, Archibald N. & Margaret A. Bryan. 1966. Linguistic analyses: The non-Bantu languages of north-eastern
Africa. London: Oxford University Press.
Tucker, Archibald N. & John T. O. Mpaayei. 1955. Maasai grammar – with vocabulary. London: Longman,
Green.
Yokwe, Eluzai Moga. 1987. The tonal grammar of Bari. Urbana-Champaign: University of Illinois dissertation.
Voisin, Sylvie & Denis Creissels. This volume. B-applicatives and I-applicatives in Atlantic languages (Niger-
Congo).
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Martine Vanhove
24 Applicative constructions in Cushitic
Abstract: This chapter provides an overview of the morphology, syntax, and semantics
of applicative constructions and applicative lookalikes in the five branches of Cushitic
languages for which enough data show that such constructions exist: Central Cushitic,
Northern Cushitic, and three groups within Eastern Cushitic (the Iraqw-Alagwa-Bu-
runge cluster, Omo-Tana and Oromo). These constructions have hitherto been largely
underestimated, or analysed differently. In general, applicative lookalikes predomi-
nate, and even though there is a good number of commonalities among the languages,
each of them has developed its own system. The two main strategies used are preverbal
constructions and periphrastic constructions, while morphological derivation is rather
marginal. The most frequent semantic roles of applied phrases are beneficiary, instru-
ment, and location (ventive, ablative); the roles of maleficiary and cause are marginal.

1 Introduction
Cushitic languages constitute one of the six branches of the Afroasiatic phylum, along-
side Berber, Chadic, Egyptian, Omotic and Semitic. They are spread in Northeast and
East Africa from southern Egypt, in the north, to Tanzania, in the south; the majority of
these languages are spoken in Ethiopia. They are traditionally divided into three sub-
branches: northern, central, and eastern, with further subdivisions as shown in Figure 1.
The number of speakers varies greatly from one language to the other. According to
Ethnologue (Eberhard, Simons, and Fennig 2022), Oromo, which is also used as a lingua
franca, has over 36 million speakers, Somali over 21 million. Afar, Beja, Hadiyya, Kambaata
and Sidaama have between approx. 750,000 and 4,3 million speakers; Awngi, Xamtanga,
Konso, and Iraqw between approx. 200,000 and 500,000 speakers (Ethnologue), while
the number of speakers goes down to some 10,000 for e.g. Alagwa (Mous 2016: 1). Yaaku
and Elmolo in Kenya are highly endangered with very few old (semi-)speakers left (even
maybe none today for Elmolo), as well as Kemant and Bilin in Ethiopia, and probably also
Dahalo, with a few hundred speakers left for the latter.

Acknowledgements: My gratitude goes to my colleagues Yvonne Treis for her careful reading of the first
version of this chapter and her very constructive comments, to Denis Creissels for his helpful remarks, and to
Girma Mengistu Desta, who drew my attention to the possible applicative lookalike construction in Oromo.
I am also indebted to my Sudanese colleague Mohamed-Tahir Hamid Ahmed, and my Beja consultants, in
particular to Ahmed Abdallah Mohamed-Tahir and his family at Sinkat.

https://doi.org/10.1515/9783110730951-024
836 Martine Vanhove

Cushitic

Northern (= Beja) Central (=Agaw) Eastern

Bilin Xamtanga Kemant Awngi Highland Lowland Dahalo [Iraqw Alagwa] Burunge
(† Kw’adza † Aasáx)

[Hadiyya Kambaata] Sidaama Gedeo Burji Southern [Saho Afar]

Nuclear Transversal

Omo-Tana Oromoid Dullay Yaaku

cluster

Dhaasanac Baiso Rendille Oromo [Gidole, Konso]

Arbore Boni
Elmolo Somali

Figure 1: The classification of Cushitic (adapted from Tosco 2000a: 89, 108; Mous 2012: 342)
[ = extinct language].

Depending on the authors’ stance about the status of dialects vs. languages, one counts
between 30 and 45 Cushitic languages. For instance, Ethnologue distinguishes five Oromo
languages and seven Somali languages, while other scholars count two languages, with
dialectal variants.
A good proportion of the languages have gained national or official recognition
during the 20th century. They may even be taught at school in the countries where they
are spoken, as is the case for most of them in Ethiopia, for Somali in Somalia, Afar in
Eritrea (and Ethiopia), or Beja in Eritrea, while the same (and other) languages in other
(or the same) countries still remain unscripted and non-official, as e.g. Beja in Sudan.
Language contact, and bi- or multilingualism between Cushitic languages, with
various Ethio-Semitic languages (or with Arabic in Sudan and Egypt) are widespread,
as well as with Omotic in Southwest-Ethiopia, Bantu and Nilotic languages in the south
of the Cushitic domain, in Kenya and Tanzania. More often than not, language contact
not only impacted the phonology and the lexicon, but also led to morphological borrow-
ings, syntactic calques, and various convergence phenomena at the morphological level
(Tosco 2000b, 2009; Crass and Meyer 2008; Vanhove 2012, 2020a).
Cushitic languages are morphologically rich, with suffixes (a majority), prefixes,
infixes (rare), ablaut, stem alternations, reduplication and suprasegmental morphemes.
They are predominantly verb-final. There is a robust noun-verb distinction. Verbal
predicates can be finite, semi-finite or non-finite (converbs, relative verbs). Valency-
changing derivation (causative, middle and/or passive) exists in almost all languages,
as well as morphological devices to express pluractionality. Inchoative and verbalizing
 24 Applicative constructions in Cushitic 837

morphological devices are also widespread. See Mous (2012) for a detailed typological
profile of Cushitic.
The derived middle voice is often used with an auto-benefactive semantic value,
among other senses as listed by Kemmer (1993). In several languages, the auto-benefac-
tive value is even the most productive one. Mous (2004: 85) mentions Somali, Oromo,
Afar, to which Kambaata (Treis 2023) and a few others could be added. Below is an
example from Gedeo, where the auto-benefactive value is also productive.

(1) Gedeo (Eyob 2015: 278)

manʤ-i-ʧʧ-i abokatto k’oʧʧʼ-eɗ-∅-e
person-ep-sg-nom.m avocado collect-mid-3sg.m-pfv
‘The man collected avocado for himself.’1

The auto-benefactive value of the middle voice won’t be further dealt with in this
chapter.
The term “applicative” is rarely used in Cushitic linguistics, except, as far as I know,
in Heine (1980a), Kießling (1994), Mous (2016), and Darmon (2015), neither are “bene-
factive” or “malefactive” often mentioned in descriptions. Consequently, this chapter
presents an endeavour to identify applicative constructions (or applicative lookalikes)
in three branches where some constructions meet (part of) the criteria presented in
Zúñiga and Creissels (this volume). This concerns central (§ 2.1) and northern Cushitic
(§ 2.2) for periphrastic constructions with auxiliaries, parts of the eastern branch, the
Iraqw-Alagwa-Burunge cluster (§ 3.1), and the Omo-Tana branch (§ 3.2) for construc-
tions with preverbs, and marginally with verbal derivation (§ 4). From the available
descriptions, it seems that the other branches only use the adjunct strategy with dative
cases or adpositions. § 5 presents a possible lookalike in Oromo.
The definition provided by Zúñiga and Creissels in the introductory chapter,
repeated below, is followed to identify applicative constructions (and their lookalikes):

The base construction (BC) and the applicative construction (AC) are related as follows:
i) The predicates in both constructions are built upon the same root, but the one in the AC bears
additional overt marking that distinguishes it from the one in the BC.
ii) The participant encoded as S or A in the BC appears as S or A in the AC.
iii) The AC includes a noun phrase in a role other than S or A, the applied phrase (AppP), which
refers to a participant that either requires a non-core coding different from its coding in the
BC or cannot be expressed at all in the BC. (Zúñiga and Creissels, this volume).

1 Throughout this chapter, glosses have been harmonized whatever the system used by the authors. In
particular, preverbs used with an applicative value are always glossed appl. Whenever possible, pre-
verbs relevant for the applicative domain have been hyphenated if they were not segmented in the
source. Translations from German sources are mine. In the Iraqw-Alagwa-Burunge cluster’s transcrip-
tions, / corresponds to the voiced laryngalized pharyngeal fricative ʢ.
838 Martine Vanhove

2 Periphrastic applicatives with auxiliaries

Following David Cohen’s unpublished approach to auxiliaries (summed up in Sime-
one-Senelle and Vanhove 2003), several criteria are used to identify auxiliaries (also
called verb-operators) within a verbal periphrastic construction, defined as a bi-verbal
mono-clausal construction denoting a single event:
(i) A verbal periphrastic construction, made of an auxiliary verb and a lexical verb,
constitutes a morphosyntactic and semantic unit.
(ii) Within the utterance, the syntactic scope of the auxiliary is the lexical verb, not the
utterance.
(iii) No coordinating or subordinating element can intervene between the two verbs.
(iv) Both verbs have the same S (or A).
(v) The P argument of the periphrasis, if any, is that of the lexical verb.
(vi) The auxiliary usually undergoes desemanticization: its meaning is more general
than that of the lexical verb, since it needs to combine with all (or many) semantic
classes of verbs.
(vii) Paradigmatic substitution is only possible for the auxiliary.

All these criteria point to the mono-clausal status of the periphrasis as opposed to
bi-clausal constructions.
Note that finiteness and the autonomous properties of the two verbs are not criteria
for the identification of auxiliaries. Thus, this broad characterization also includes what
is traditional termed serial verb constructions, minimally characterized as follows
by Creissels (2010: 37): (a) no linking element is present between the verbs involved
in the construction, and (b) none of the verbs involved in the construction is in a form
implying a non-autonomous status. Such an approach of verbal periphrastic construc-
tions is more in line with Shibatani’s (2009) criticism of serial verb constructions “as a
cross-linguistically valid type of construction” (Creissels 2010: 38) because they share
many properties with other types of verbal periphrastic constructions.

2.1 Central Cushitic

For Xamtanga, Darmon (2015) reports four periphrastic applicative constructions with
the following auxiliaries: yɨw- ‘give’, näy- ‘give to the speaker / here’, bär- ‘leave, abandon’,
and y- ‘say’.

2.1.1 The auxiliary yɨw- ‘give’

In the first periphrastic construction, the lexical verb is in the converb form, followed by
the auxiliary yɨw- ‘give’, which conforms to the seven criteria mentioned in § 2 above. In
 24 Applicative constructions in Cushitic 839

addition, Darmon (2015: 193) mentions language specific properties also showing that
the applicative construction is a single event (and not several, as when converbs are
used in dependent clauses): (i) only one converb can precede the finite ‘give’ verb, unlike
in deranked subordinate clauses; and (ii) the negative marker on ‘give’ has scope over
the whole construction (not only on ‘give’ if ‘give’ had been the verb of a matrix clause).
In Xamtanga, recipient, goal, and beneficiary semantic roles are systematically
marked with a non-core coding, the dative case, but with the ‘give’ construction the
recipient and goal interpretations are excluded. However, since the applied phrase has
the same non-core marking as in the base construction (compare 2a, 2b and 2c), it does
not comply with the third criteria of Zúñiga and Creissels’s definition. The ‘give’ con-
struction could thus better qualify as an applicative lookalike of the privative-mark-
ing type, in which “the A[pplied] C[onstruction] predicate shows higher morphological
complexity than the B[ase] C[onstruction] predicate—more precisely, an element ana-
lyzable as an applicative marker” (Zúñiga and Creissels, this volume). However, there
are also cases in which the applied phrase cannot occur in the base construction (3),
thus also meeting the third criterion of a well-behaved applicative construction. The
syntactic marking of the arguments is a clear trace of the origin of the construction.
Semantically, what the auxiliary yɨw- ‘give’ adds to the base construction is that
the action is done in somebody’s interest, on somebody’s behalf, in his or her favour,
showing that the semantic role of the dative argument is indeed that of a beneficiary.
The ‘give’ auxiliary is restricted to transitive lexical verbs, and it may be contiguous to it
(4), or not (2c–3). No examples with a pronominal applied phrase were found.

(2) Xamtanga
a. s’ɨqa kɨn-d-ɨŋ ŋɨn bɨz-n-u-n
ten learn-medp-nmlz house[acc] open-1pl-pfv-1pl
‘We opened ten schools.’ (Darmon 2015: 150)
b. abɨn-i-z bir-ɨd bɨz-ɨ-č
guest-def-dat door-def[acc] open-pfv-3sg.f
‘She opened the door to the guest.’ (Darmon 2015: 194)
c. bir-ɨd bɨz-ɨr šɨmir-yän-s yɨw-ɨ-č
door-def[acc] open-3sg.f[cvb] old.woman-def-dat give-pfv-3sg.f
‘She opened the door for the old woman.’ (she did not manage to do it herself)
(Darmon 2015: 192)

(3) Xamtanga (Darmon 2015: 192)

mi-dyän-t abɨz Dawit-ɨs yɨw-u-n
injera-def-acc finish[1sg.cvb] Dawit-dat give-pfv-1sg
‘I finished the injera for the sake of Dawit.’ (otherwise he would have been
punished)
840 Martine Vanhove

(4) Xamtanga (Darmon 2015: 193)

ŋɨ=xʷɨrä-sɨ ŋɨn täss yɨw-ɨ-y-äw=ɨm
3sg.m.poss=child-dat house[acc] build[3sg.m.cvb] give-pfv-neg-3sg.m=mulcl
‘He did not build a house for his son.’

2.1.2 The auxiliary näy- ‘give to the speaker / here’

If the beneficiary is the speaker, i.e. a first person, Xamtanga uses a deitic ‘give’ auxiliary
näy- ‘give to the speaker / here’ instead of yɨw-.2 The lexical verb is in the converb form,
preceding the auxiliary, and no overt 1st person pronoun is mentioned. This is a differ-
ent case of an applicative lookalike construction, where the applied phrase is inherent
to the meaning of the auxiliary.

(5) Xamtanga (Darmon 2015: 194)

ŋɨn täss näy-u
house[acc] build[3sg.m.cvb] give.to.the.speaker-pfv[3sg.m]
‘He built a house for me.’

2.1.3 The auxiliary bär- ‘leave, abandon’

The periphrastic construction with bär- ‘leave, abandon’ is morphosyntactically similar

to the one with yɨw- ‘give’: the lexical verb in the converb form precedes the auxiliary.
Both transitive and intransitive verbs are licensed with this auxiliary.
Semantically, the applied phrase may have a malefactive or a benefactive interpre-
tation, depending on the semantics of the verb, the subject NP, or the extra-linguistic
context, judging by the examples provided in Darmon (2015). Based on the semantics
of the auxiliary when used as a lexical verb, Darmon (2015: 204) hypothesizes that the
malefactive reading is primary. When the participant is positively affected, it is always
understood as a plain beneficiary.
Unlike with yɨw-, the introduction of bär- after transitive verbs does not necessar-
ily require the presence of a third non-core argument. When the direct object of the
base construction becomes the applied phrase of the verbal periphrasis, it keeps its
accusative case (6). This construction could thus also be interpreted as an applicative
lookalike. On the other hand, the locative case of a non-core argument with intransitive
verbs in the base construction is turned into the accusative case with the applicative
periphrasis (7). It thus promotes the adjunct participant to the status of a core syntactic
term. With transitive verbs, the non-core argument of the base construction can be
either promoted to a core argument with accusative case (8–9)—in line with the narrow

2 See Creissels (2010: 50) for crosslinguistic attestations of ‘give’ verbs with a deictic component.
 24 Applicative constructions in Cushitic 841

definitions of applicatives (e.g. Dixon and Aikhenvald 2000), but not of Zúñiga and Cre-
issels (this volume)—or remain in the non-core dative case in the applied construction,
while only the non-core dative case is possible in the base construction (9).

(6) Xamtanga
a. däŋä-d ɨǧɨr-ɨdyän-t däŋt-u
judge-def man-def-acc judge-pfv[3sg.m]
‘The judge judged the man.’ (Darmon 2015: 198)
b. däŋä-d ɨǧɨr-ɨdyän-t däŋt bär-u
judge-def man-def-acc judge[3sg.m.cvb] leave-pfv[3sg.m]
‘The judge judged the man to his detriment.’ (he was probably innocent)
(Darmon 2015: 197)

(7) Xamtanga (Darmon 2015: 197)

a. Bɨrtukʷan-tɨl qas’-u
Bɨrtukʷan-loc shout-pfv[3sg.m]
‘He yelled at Bɨrtukʷan.’
b. ŋɨta=nyä-yän ɨfär-dyän-t qas’-ɨr bär-ɨ-č
3pl.poss=mother-def child.pl-def-acc shout-3sg.f[cvb] leave-pfv-3sg.f
‘The mother scolded her children.’ (she shouted at their detriment)

(8) Xamtanga (Darmon 2015: 199)

a. ɨfärä-y-z-u s’ɨbk’ä-d k’äb-u-n
child-def-gen-head.sg hair-def[acc] cut-pfv-1sg
‘I cut the child’s hair.’
b. ɨfärä-y-z-u s’ɨbk’ä-d k’äb bät’-u-n
child-def-gen-head.sg hair-def[acc] cut[1sg.cvb] leave-pfv-1sg
‘I cut the child’s hair.’ (to make him happy / against his will)

As with yɨw-, when a third argument is added to a transitive verb, the applied phrase
can be placed between the core verb and the auxiliary or precede the whole periphrasis.

(9) Xamtanga
a. Abbäbä Bɨrtukʷan-sɨ dɨbdabe s’af-u
Abbäbä Bɨrtukʷan-dat letter[acc] write-pfv[3sg.m]
‘Abbäbä wrote a letter to Bɨrtukʷan.’ (Darmon 2015: 194)
b. Guläšu Fre-t / Fre-s dɨbdabe-d s’af
Guläšu Fre-acc Fre-dat letter-def[acc] write[3sg.m.cvb]
bär-u
leave-pfv[3sg.m]
‘Guläšu wrote the letter to the detriment of Fre.’
(Lit. ‘Guläšu wrote the letter and left Fre / (it) to Fre’)
842 Martine Vanhove

c. Guläšu dɨbdabe-d s’af Fre-s bär-u

Guläšu letter-def write[3sg.m.cvb] Fre-dat leave-pfv[3sg.m]
‘Guläšu wrote the letter to the detriment of Fre.’ (Darmon 2015: 198)

The positively or negatively affected element can also be the speaker, which is left unex-
pressed. In such a case, the construction is again an applicative lookalike. In (10) the
auxiliary bär ‘leave’ induces the applicative reading of the sentences, but it is the adjec-
tive or the verb modifying the subject that entails the benefactive (10a) or malefactive
(10b) reading.

(10) Xamtanga (Darmon 2015: 199)

a. ɨd č’ɨqa gɨzɨŋ-ɨd kɨr bär-u
dem2.sg.m bad dog-def die[3sg.m.cvb] leave-pfv[3sg.m]
‘That bad dog died.’ (speaker positively affected)
b. ɨd k’äs-äw gɨzɨŋ-ɨd (k’al) kɨr bär-u
dem2.sg.m be.good-3sg.m dog-def intrj die[3sg.m.cvb] leave-pfv[3sg.m]
‘That good dog died!’ (speaker negatively affected)

2.1.4 The ‘say’ construction

The case of the ‘say’ construction is problematic from an applicative verbal periphrasis
perspective. Morphosyntactically, conversely to the previous constructions, ‘say’ is a
converb in V1 position, following a dative-marked noun or pronoun, and the lexical
verb is finite, in V2 position. This suggests a different grammaticalization path, from a
quotative construction with an overt addressee in the dative case,3 and a different syn-
chronic morphosyntactic analysis. The grammaticalized quotative converb functions
more like a postposition following the dative argument than as an applicative verbal
periphrasis.

(11) Xamtanga (Darmon 2015: 195)

Almaz Dawit-ɨs y-ɨr qas’ɨŋä-dyän-t k’ɨw-ɨ-č
Almaz Dawit-dat say-3sg.m[cvb] thief-def-acc kill-pfv-3sg.f
‘Almaz killed the thief for Dawit.’ (Lit. ‘Almaz killed the thief saying to Dawit.’)

The postpositional analysis is also favoured by the fact that the ‘say’ construction can
combine with the ‘leave’ construction (12). In this case, the 1st person applied phrase is
not left unexpressed as in (10), but is marked with a pronoun in the dative case.

3 The addressee of ‘say’ used as a quotative verb is also marked with the dative case (Darmon 2015: 324).
 24 Applicative constructions in Cushitic 843

(12) Xamtanga (Darmon 2015: 199)

ɨd k’äsäw ɨǧɨr yɨggɨs y-äw kɨr
dem.sg.m good.3sg.m man 1sg.dat say-3sg.m[cvb] die[3sg.m.cvb]
bär-u
leave-pfv[3sg.m]
‘That good man sacrificed himself for me.’

This is in line with Creissels’s (2010: 61–62) analysis of similar constructions in Ethio-Se-
mitic languages as being not inherently benefactive applicative periphrases.

2.2 North Cushitic

In Beja, nouns with the semantic role of beneficiary are only constructed with the di-
rectional postposition dhaːj ‘towards’, which governs the genitive case, without any
applicative marking on the verb. But when the benefactive argument is a pronoun, a
periphrastic applicative construction with the (highly irregular) auxiliary verb hi ‘give’,
the sole applicative auxiliary of the language, is used instead of the adjunct construc-
tion. Compared with the base construction, hi obligatorily licenses an additional argu-
ment (Vanhove 2017: 143), and can be used with both transitive (13–14) and intransitive
verbs (15).
As a lexical verb, hi is a ditransitive verb, and if the recipient is a pronoun, the
latter belongs to the bound set of object or possessive pronouns (depending on the verb
paradigm) and is enclitic to ‘give’.4 This morphosyntactic feature is kept for the applied
phrase when ‘give’ is used as an applicative auxiliary. The Beja construction would thus
be best described as an applicative lookalike.
The lexical verb, which precedes the auxiliary, is either the general (same subject)
converb (13b), or a finite verb form (14b) (optionally followed by the—non-coordinat-
ing or subordinating—linker -i for the Perfective [15b]), which shares its subject with
the lexical verb (Vanhove 2017: 143–144).
The beneficiary of the ‘give’ construction is mostly of the deputative type, express-
ing that an action is done on someone else’s behalf.

4 Among hundreds of examples, there are only two where the pronominal recipient is followed by the
directional postposition, or is an independent object pronoun.
844 Martine Vanhove

(13) Beja
a. ʤabanaː=t=oː i-dʔi
coffee=indf.f=poss.3sg.acc 3sg.m-do\pfv
‘He made his coffee.’ (BEJ_MV_NARR_15_leopard_066)5
b. mhaj=t ʃarti-ja dʔi-ti a-niːw=hoːk
three=indf.f line-pl do-cvb.gnrl 1sg-give\ipfv=2sg.obj
‘I’ll draw three lines for you.’ (BEJ_MV_NARR_06_foreigner_29)

(14) Beja
a. eː=mak a-gʷʔad
def.pl.m.acc=donkey\pl 1sg-watch\pfv
‘I watched after the donkeys.’ (BEJ_MV_NARR_190_poorDonkeys_14)
b. eː=mak gʷʔad-a heː=hoːn
def.pl.m.acc=donkey\pl watch-imp.2sg.m give\imp.2sg.m=1pl.obj
‘Watch after the donkeys for us!’ (BEJ_MV_NARR_190_poorDonkeys_11)

(15) Beja
a. geːb=oːn tikʷ-iːni
next=poss.1pl.acc go.down-ipfv.3sg
‘They go down next to us.’ (BEJ_MV_NARR_57_Ababda_246)
b. i=ʔaraːw=uːn dhaːj tikʷ-ja-i
def.m=friend=poss.1pl.nom dir go.down-pfv.3sg.m-lnk
i-heː=heːb
3sg.m-give\pfv=1sg.obj
‘Our friend went down to him for me.’
(BEJ_MV_NARR_43_hyena_man_042-044)

In a few instances, the ‘give’ construction has a plain beneficiary semantic role:

(16) Beja
a. ti=ʤhalaj fif-tiːt
def.f=coal\pl pour-cvb.seq
‘They poured the coal and. . .’ (BEJ_MV_NARR_42_hunter_bread_115)

5 The Beja examples are mostly extracted from Vanhove’s online corpus (2020b); the indications at the
end of the free translation refer to the texts and prosodic units where they are found. In the glosses, \
signals stem alternation.
 24 Applicative constructions in Cushitic 845

b. ʤaːhil-a i-ʔaʃiʃ-n=eːk halaːwaː=t

small.child-pl 3sg.m-face.each.other\pfv-pl=if sweet\pl=indf.f
i-fif i-hi
3sg.m-pour\pfv 3sg.m-give\pfv
‘When the small children met him, he poured sweets for them.’
(BEJ_MV_NARR_12_witch_122-124)

A malefactive interpretation is possible only in negative utterances:

(17) Beja
a. ki=t-kan=heːb han
neg.ipfv=2sg.m-know\mid.pfv=1sg.obj q.plr
‘Don’t you know me?’ (BEJ_MV_NARR_31_king_40)
b. tak kan-ti hi-it=heːb i-rib
man know\mid-cvb.gnrl give-vn=1sg.obj 3sg.m-refuse\pfv
‘No one got information for me.’ (Lit. ‘a man refused giving me knowing’)
(BEJ_MV_NARR_04_djinn_075)

The periphrastic construction can also be used, albeit rarely, to express a causal relation-
ship between two events, another type of grammaticalization of benefactives (see Creissels
2010: 35–36 for examples in São-Tomense Creole and Yoruba). In (18), there is no bound
object pronoun on the auxiliary because the 3rd persons of this set are zero morphemes:

(18) Beja
ti=bʔaʃi ti-ʃbib oː=jaːs ti-fdin
def.f=fox 3sg.f-look\pfv def.sg.m.acc=dog 3sg.f-move.away\pfv
tal-ta-i t-hi
trot-pfv.3sg.f-lnk 3sg.f-give\pfv
‘The vixen looked at the dog and trotted away because of it (the dog).’ (BEJ_
MV_NARR_50_fox_hunt_050-051)

3 Applicatives with preverbs

The verb forms of a number of Cushitic languages are made of two independent or,
more often, clitic elements,6 consisting of a (partly) inflected lexical core, preceded by
what some Cushiticists call a “selector” (e.g. Mous 2006: 306), defined as “an inflectional
complex” in Mous (2016: 173). In this chapter, I will use “preverb” (also used by some
authors) as a neutral cover term, since the smallest common denominator between all

6 Most often, cliticization is not represented in the transcriptions of the authors.

846 Martine Vanhove

the languages concerned is that this inflectional element is obligatorily placed before the
verb (although in some languages it may be separated from the verb by other elements,
or cliticize to any preceding element).7 Preverbs may consist of only one morpheme,
e.g. a special set of object pronoun, or be, more often, multimorphemic. However, the
type and number of grammatical categories that can be marked and often stacked on
preverbs vary across languages: subject, impersonal subject, object, sentence type (e.g.
dependent or independent), focus, direction, tam, and obliques (sometimes termed
“case clitics”). In none of these languages do the preverbs bear all these grammatical
categories. Among the thirteen languages with a preverb reported in Mous (2006: 306),
I could identify nine with morphemes that are good candidates to be applicative (looka-
like) markers: Alagwa, Arbore, Boni, Burunge, Dhaasanac, Elmolo, Iraqw, Rendille, and
Somali.8
Diachronically, some of the morphemes composing the preverbs have stemmed
from phonetically reduced subject pronouns (Banti 1997: 103–104), or, as in the case of
applicative morphemes, they are cognate with instrumental, directional or dative post-
positions (see e.g. Kießling 2001). It seems no preverbal morpheme has a verbal origin,
unlike the auxiliary constructions in § 2.

3.1 East Cushitic: The Iraqw-Alagwa-Burunge cluster

Four preverbs contain morphemes used in constructions that meet the criteria of the
definition proposed in the introductory chapter for applicative constructions, or are
more of the lookalike type.

3.1.1 The instrumental applicative -rar

The applicative -rar (and its cognate forms), is the sole marker that has one and the
same instrumental semantic role across the three languages.9
In the Iraqw example below, the peripheral argument of the base construction
bears the instrumental postposition (19a), cognate with the applicative marker, while in

7 In addition to the idiosyncratic “selector”, preverbs have been termed in many various ways in Cush-
itic studies, such as “preverb”, “derivative prefix”, “indicator particle”, “focus particle”, “inflectional
particle”, “prepositional particle”, “preverbal particle”, “Incorporated Object Prefix”, “adposition”, “aux-
iliary”, “be verb”, “case marker”, “case clitic”. Obviously, such a wealth of terminology does not ease
cross-linguistic comparison, even within the Cushitic family.
8 This list is shorter than the ones provided in Mous (2006: 322) since what he labels as “dative” does not
have an applicative function in all the languages he mentions.
9 A cognate preverb ar- is also found in Elmolo (Omo-Tana), but as a ventive and focus marker (Heine
1980a: 197). A possible applicative function is unclear.
 24 Applicative constructions in Cushitic 847

(19b) the preverb stacks an object pronoun, a tam marker, and the instrumental marker,
while the noun does not bear the instrumental postposition. This construction is again
an applicative lookalike.

(19) Iraqw
a. qaymo ga kurmo-’éen-ar doohl
field 3.obj:f.obj hoe-l.sg.poss-ins dig:3.sg.m:prs
‘He cultivates the field with my hoe.’ (Mous 1993: 245)
b. aníng kurmo u-na-rar dóohl
lsg hoe m.obj-pst-appl:ins dig:lsg
‘I was digging with a hoe.’ (Mous 1993: 153)

Burunge also makes use of an instrumental-comitative preverbal marker -ri, and its
grammatical allomorph (20b) -r-, cognate with the preposition hari ‘with, to, at’ (20a)
(Kießling 1994: 171, 192–193). The base construction (20a) uses a different preverb.

(20) Burunge
a. ’ana haati ’alung ki/a
1.sg 1/2.sbj:fut:sep back return:1sg.ipfv
‘I will go all the way back to the starting point.’ (Kießling 1994: 237)
b. ’ana ha-gu-r-aa ki/a ’ayaa-ge
1sg 1/2.sbj-2sg.m.obj-appl:com-fut return:1sg.ipfv house-all
‘I’ll go back home with you.’ (Kießling 1994: 172)

Alagwa illustrates a similar situation where the instrumental-comitative marker -ra on

preverbs is cognate with the preposition hara ‘in, at, to, with’ (Mous 2016: 186, 211–215).
(21) is the sole example in Mous’s grammar with this applicative marker stacked on the
preverb, but there is no corresponding base construction. However, clauses with the
cognate preposition before a noun do not use the preverb with the applicative marker
(22). Since the preposition used in (21) is different from the cognate preposition, it might
be the case that Alagwa has here a well-behaved applicative, according to Zúñiga and
Creissels’s definition.

(21) Alagwa (Mous 2016: 186)

nongo-ra ibíiti na hati hare tée-lo’
ips.cons.m.obj-appl stay.3.m with relative wife F-heavy
‘And he was left with a pregnant woman.’

(22) Alagwa (Mous 2016: 252, 257)

nanka máax hara du/i
cons:ips:f.obj smear with oil
‘And she was anointed with oil.’
848 Martine Vanhove

3.1.2 The benefactive applicative s(V)-

Alagwa has an applicative marker s- which is prefixed to the preverb.10 The marker can
be used for the addressee of a ‘say’ verb, the recipient argument of ‘give’ verbs, and as
a benefactive device. As a benefactive, it is restricted to two contexts: when a nominal
beneficiary precedes the verb and when it is a pronoun. When the lexical NP follows
the verb, Alagwa uses instead the cognate preposition as in (25), i.e. an adjunct strategy
(Kießling 1994: 171). Compare the preverbal applicative constructions where the pro-
nominal applied phrase is the last suffix of the preverb in (23b), and a noun in (24b),
here as the head of a relative clause, with the base constructions with less complex
preverbs in (23a) and (24a). Only examples with transitive verbs were found. Could it
be that the word order change operates as a different non-core coding of the applied
phrase, or could the possible absence of the applied phrase in the base construction
indicate that the corresponding applicative construction meets the criteria of Zúñiga
and Creissels’s definition? Since no strict corresponding constructions were found in
the grammar, the final answer is left open.

(23) Alagwa
a. qoo yawa k-i /agagin?
emph cattle ips-n.obj eat:hab:dur:q
‘Are cattle not eaten?’ (Mous 2001: 240, 245)
b. daankíi si-k-i koonka /ag
then appl:ben-ips-3pl chicken eat
‘And then a chicken was slaughtered for them.’ (Lit. ‘One ate for them a chicken.’)
(Mous 2001: 131)

(24) Alagwa
a. ra'amu k-a ra'am-an na heku
song sub-1/2 sing-1pl cop dem1.pro
‘The song that we sing is this one.’ (Mous 2001: 189)
b. hiru sa-k-a-n raa’ na heeku
man.m appl:ben-sub-1/2.sbj-pfv sing.1sg cop dem1.m
‘The man that I sang to is this one.’ (Mous 2016: 191)

(25) Alagwa (Kießling 1996: 37)

kuu marée hhaab-it sa taatáà
2sg proh inform-2sg appl:ben father
‘You don’t tell father!’

10 It is called “dative case marker” in Mous (2006: 322), and “beneficient selector prefix” in Mous (2016:
176).
 24 Applicative constructions in Cushitic 849

A cognate applicative sV- (the vowel is subject to regressive assimilation) is used in

Burunge (Kießling 1994: 169–171) (26b–27b). Unlike in Alagwa, the applied phrase is
expressed twice, as an independent NP (pronouns in 26b and 27b) and prefixed on the
preverb. Since the applied phrase cannot be expressed in the base construction, sV- can
be viewed as a true applicative marker.

(26) Burunge (Kießling 1994: 170)

a. ’ana ha-ga laahhama hhada
1sg 1/2.sbj-3sg.f.obj put.up:1sg:ipfv stick
‘I put up a stick.’
b. ’ana ugu su-gu laahhama hhada
1sg 2sg.m appl:ben-2sg.m.obj put.up:1sg:ipfv stick
‘I put up a stick (as a target) for you.’

(27) Burunge (Kießling 1994: 170)

a. 'ina ha-gw-áa fiisi kooloo
3sg 1/2.sbj-3sg.m.obj-pret steal:3sg.m:pfv hoe
‘He stole a hoe.’
b. 'ina 'ana si-n-áa-ni fiisi kooloo
3sg lsg appl:ben-1sg.obj-pret-vent steal:3sg.m:pfv hoe
‘He stole a hoe for me.’

In Burunge, sV- can also be used with a causal value:11

(28) Burunge (Kießling 1994: 171)

taataa-gway 'idoo si-ni11 qaase hing
father.m-poss.1sg.m thing appl:reas-1sg.obj burry:1sg.sbjv 3sg:comp
gwaa’i
die:3sg.m:pfv
‘The reason why I bury my father is that he is dead.’

A cognate preverb -s (in final position on a preverb) / sa- when cliticizing to the verb,
exists also in Iraqw as a suffix to the other preverbal elements. It has a causal value (29).
Combined with the locative applicative ni- (§ 3.1.4.) on complex preverbs, s(a) marks the
pronominal beneficiary of an intransitive verb (30a). Alternatively, it can be prefixed
to the verb, without ni- (30b). s(a) is cognate with the reason case marker and the prep-
osition as ‘because’ (Mous 1993: 102, 261–262). The exact applicative status is unclear.

11 Kießling’s example has sugu, obviously a typo, since this preverb corresponds to the applicative with
a 2sg.obj.
850 Martine Vanhove

(29) Iraqw (Mous 1993: 153)

bar-ti-na-sa caacaam-ín
cond-ips.lsg.obj-pst-reas cry-dur.3sg.m
‘If they were crying because of me.’

(30) Iraqw (Mous 1993: 154)

a. ngu-s gadyúus
3.obj:appl:loc:m.obj-appl:ben work.1sg
‘I work for him.’ (Lit. ‘The reason I work is for him.’)
b. u sa-gadiyúus
3m.obj appl:ben-work.1sg
‘I work for him.’ (Lit. ‘The reason I work is him.’)

3.1.3 The locative applicative ee

For Alagwa, Mous (2016: 185–186) reports on a morpheme ee, that he labels “applica-
tive” with a locative meaning. It can encliticize to the other preverbal elements, but can
also be prefixed on verbs. The latter strategy is treated under the section about com-
pound verbs (not derived verbs) in which Mous states that “[t]he applicative clitic can
also fuse with the verb or adjective and lexicalise to a new verb often with a different
and partly unpredictable meaning” (Mous 2016: 150–151), suggesting that the device as
a verbal prefix is not very productive.
Only applicative constructions with intransitive verbs are exemplified: (31b) shows
the use of ee prefixed to a motion verb, and (31c) as a clitic to the preverb, with a loca-
tive phrase which cannot be expressed in the base construction (31a), i.e. a true applica-
tive; (32a) illustrates the applicative prefixed to a middle verb, and (32b) cliticized on
the preverb. In (32a–b) no overt location is expressed, but the applicative implies that
space is somehow involved; in this case the constructions may better qualify as applica-
tive lookalikes.

(31) Alagwa
a. ningi daaf
cons:3 enter.3m
‘And he entered.’ (Mous 2016: 186)
b. ningi qoroo ee-daaf do’o-li
cons:3 emph appl:loc-enter.3m house-in
‘And he entered the house / went inside.’
c. ning=ee qoroo daaf do’o-li
cons:3=appl:loc emph enter.3m house-in
‘And he entered the house / went inside.’ (Mous 2016: 185)
 24 Applicative constructions in Cushitic 851

(32) Alagwa (Mous 2016: 185)

a. hadithi-r-i i-ni asa ee-fak-it
story-f-poss.1sg 3-prf already appl:loc-finish-mid
‘My story finishes here.’
b. hadithi-r-i i-n=ee fak-iti
story-f-poss.1sg 3-prf=appl:loc finish-mid.prs.3sg.m12
‘My story is finished.’

Mous mentions that ee is compatible with the s- applicative preverb (§ 3.13) only if
space is involved. In the sole example provided, ee is prefixed to the verb:

(33) Alagwa (Mous 2016: 186)

s-oo fa/a ee-qaas
appl:ben-m.obj mash appl:loc-put
‘I will provide mash for you.’

Although not mentioned in the list of prepositions (Mous 2016: 211), it seems, judging
by the sole example I found in the texts, that ee may also function as a preposition, dou-
bling the allative case of the noun:

(34) Alagwa (Mous 2016: 253, 257)

kimaa ín nanga xab ee mfalmee-li, xwaylá-s
then 3sg cons:f.obj marry appl king-all child-dem2
‘Then she was married to the king, that child.’

No cognate form of this applicative is reported for the other two languages.

3.1.4 The locative and benefactive applicative ni

The three languages of this group have a marker ni (and its cognate forms)—labelled
“hither” in Iraqw, “ventive” in Alagwa and Burunge—for which an applicative analysis
can be considered.
In Burunge, this morpheme, which is the last element of the preverb, is described
in terms of pragmatics:

Das Ventivmorphem {ni} im Flexionskomplex zeigt an, daß die Handlung auf ein zuvor etabliertes
Aufmerksamkeitszentrum gerichtet ist. Dieses Zentrum fällt im alltäglichen Diskurs in der Regel
mit dem Standpunkt des Sprechers zusammen. In einer fortlaufenden Erzählung wird der Ventiv

12 The two examples with fakit and fakiti are simply glossed ‘finish’ by Mous; I have restored the full
glossing.
852 Martine Vanhove

meist auf die Person des Hauptakteurs bezogen. [The ventive morpheme {ni} in the inflectional
complex indicates that the action is directed toward a previously established center of attention. In
everyday discourse, this center usually coincides with the speaker’s point of view. In a continuous
narrative, the ventive usually refers to the person of the main actor.] (Kießling 1994: 166)

The construction is used with intransitive (35b) and transitive (36b) verbs to mark a
proximal locative meaning. (37b) is a case where the construction refers “to the person
of the main actor”. Since there is no applied phrase, ni- is better analyzed as an applica-
tive lookalike.

(35) Burunge (Kießling 1994: 166)

a. ’ugu haa ki/ida
2sg.m 1/2.sbj:fut1 return.2sg.ipfv
‘You will return.’
b. ’ugu haa-ni ki/ida
2sg.m 1/2.sbj:fut1-appl:loc return.2sg.ipfv
‘You will return here.’

(36) Burunge (Kießling 1994: 166)

a. ’ana haa satisi fa/a
1sg 1/2.sbj:pret move:caus:1sg.pfv porridge
‘I moved the porridge (there).’
b. ’ana háa-ni satisi fa/a
1sg 1/2.sbj:pret-appl:loc move:caus:1sg.pfv porridge
‘I moved the porridge here.’

(37) Burunge (Kießling 1994: 224)

a. /anta dagáa tipi hari qoro’u
termite.mound ips:3sg.f.obj:pret cover.3sg.m.pfv with dry.leaves
sa hiica/a da Laa’ay
for escape sub.f Laa’ay
‘The termite mound was prepared with dry leaves for Laa’ay’s escape.’
b. /anta higáa-ni tipi hari
termite.mound 3.sbj:3sg.f.obj:pret-appl:ben cover.3sg.m.pfv with
qoro’u sa hiica/adosi
dry.leaves for escape:f:poss.3sg
‘He covered the termite mound with dry leaves for his escape.’

Alagwa has a marker n-, which, similarly to its Burunge cognate, is used for ventive
actions relative to the hinc et nunc (Mous 2016: 179), as well as with ‘give’ verbs. I found
one example in Mous’s texts which could be interpreted as a benefactive:
 24 Applicative constructions in Cushitic 853

(38) Alagwa (Mous 2016: 253, 257)

haa’uw n-í hhútlit
come.imp opt:appl:ben-sg.f.obj.2 plait:1sg
‘Come let me braid your hair!’ (Lit. ‘Let me braid to you.’)

In Iraqw, n(i-), which is the first morpheme of the preverb, is used, among others, for
recipients of ‘give’ verbs and addressees of ‘say’ verbs, and as a directional with motion
verbs (Mous 1993: 134–136). It also has the same pragmatic functions as in Burunge and
Alagwa, and can mark the pronominal beneficiary of transitive verbs (39b–40b), thus
behaving as a benefactive applicative. First person beneficiaries are unexpressed (41b).

(39) Iraqw (Mous 1993: 136)

a. aníng kurmo u tlaaxw
1sg hoe(m) m.obj buy:1sg
‘I’ll buy a hoe.’
b. aníng kurmo ngu tlaaxw
1sg hoe(m) 3.obj:appl:m.obj13 buy:lsg
‘I’ll buy him a hoe.’

(40) Iraqw (Mous 1993: 136)

a. da’angw gu-na óo’
song(m) 3.obj:m.obj-pst sing:3sg.m.pst
‘He sang a song.’
b. da’angw ngu-na di-r-ós-i ó’
song(m) 3.obj:appl:m.obj-pst place-f-3sg.poss-dir14 sing:2sg.m
‘You sang a song for him.’

(41) Iraqw
a. garma gu-na barwadu leehha-r-wa yacaaw
boy(m) 3.obj:m.obj-pst letters catching-f-abl send:3.sg.m
‘He sent a boy to get the letters.’ (Mous 1993: 271)
b. garma u-na ya’aaw imboru barwadu ngi-wa
boy(m) m.obj-pst send:1sg Mbulu letters 3.obj:appl:n.obj-bckg
oh-i
catch-3sg.m:sbjv
‘I sent a boy to Mbulu to collect letters for me.’ (Mous 1993: 132)

13 There is a typo in Mous’s glossing which reads “DEP.S.1.SG:O.M”.

14 Motion verbs and ‘say’ verbs often require the use of the feminine place noun dí-r with the direction-
al suffix -i (Mous 1993: 99).
854 Martine Vanhove

3.2 Omo-Tana

3.2.1 The applicative ka/ká

In Elmolo, there is an applicative preverb ka (also a dative with ‘give’ verbs [43]), which
either cliticizes to the verb (42b, 43a), or is separated from it by a direct object (43b). The
ka- preverb, which follows the subject index and tam affixes, marks a benefactive with
animates (42b), an instrumental with inanimates (42c), and a locative with ablative
phrases (42d) (Heine 1980a: 197). Only with the latter meaning does the applied phrase
occur with a non-core marking, but no corresponding base construction was found.

(42) Elmolo (Heine 1980a: 197)

a. yesé mín ’án-dís-a
1sg house 1sg-build-ipfv
‘I build a house.’
b. yesé mín hꞌéle aŋ-ká-dis-a
1sg house children 1sg-appl:ben-build-ipfv
‘I build a house for the children.’
c. yesé mín elle áŋ-k’a-dís-a
1sg house stones 1sg-appl:ins-build-ipfv
‘I build a house with stones.’
d. yesé mín-lu áŋ-ká-pé-e
1sg house-abl 1sg-appl:loc-get.out-ipfv
‘I (shall) get out of the house.’

(43) Elmolo (Heine 1980a: 196–197)

a. yesé núúm hꞌéle an-ꞌáŋ-ka-koon-e
1sg food children 1sg-pfv-appl:ben-bring-pfv
b. yesé núúm an-ꞌáŋ-ka hꞌéle koon-e
1sg food 1sg-pfv-appl:ben children bring-pfv
‘I brought the children food.’

When the applied phrase is a pronoun, it belongs to a special set of bound object pro-
nouns (Heine 1980a: 197), different from independent object pronouns (kesé or kéló for
2sg.obj vs. éké- with the applicative in [44]):

(44) Elmolo (Heine 1980a: 197)

isé dáfárné éké-ka-waak-as
she dress 2sg.obj-appl:ben-sew-ipfv
‘She sews a dress for you.’

For Arbore, Mous (2006: 322) mentions an applicative preverb ka but does not provide
examples. It covers ablative (source), dative (beneficiary), locative and directional
 24 Applicative constructions in Cushitic 855

meanings. There is unfortunately only one example in Hayward’s (1984) grammar, with
a beneficiary semantic role, in the following relative clause:

(45) Arbore (Hayward 1984: 312)

náag gidiká mod áɦa-ttóy naag ka ḍá-wʔw-elo
boy then those people-3sg.m girl appl beg-pfv-def
ʔay ka hóǧǧeta
15
3sg.m.ind.indf appl:ben work:ipfv:3sg.m
‘Then the boy will work for those people who asked for the girl for (him).’

The second occurrence of ka is analyzable as a suffix of the complex preverb ʔay which
merges a person index and a tam marker, while the first occurrence could be inter-
preted as a postposition, since cognates of ka are numerous in Cushitic as case markers
or postpositions, with either ablative, locative, comparative, instrumental or dative
functions (Appleyard 1990: 28). It could be that both constructions as a dative postposi-
tion and an applicative preverbal element coexist in Arbore.
In any case, the related ka morpheme of Rendille is analyzed by Pillinger and Gal-
boran (1999) as both a postposition with ablative, instrumental, and directional mean-
ings, when following directly a noun (46a), and as an applicative instrumental preverb
when preceding a verb (46b). Judging by the translations, pragmatics may be involved
in the choice of one or the other construction.

(46) Rendille (Pillinger and Galboran 1999: 28)

a. wúl ká ínam jahe
stick with def.boy 3sg.m.hit
‘It was with a stick that he hit the boy.’
b. wúl ínam á-ká-jahe.
stick def.boy foc-appl:ins-3sg.m.hit
‘He hit the boy with a stick.’

Heine (1980b) provides an example with an ablative reading:

(47) Rendille (Heine 1980b: 238)

a. usu ayimi
he came
‘He has come.’

15 Glossing is mine. ʔay is simply glossed as 3sg.m by Hayward. ʔay is not in his list of pronouns, but it is
duly reported as a “selector” of “3sg indicative indefinite” (p. 109).
856 Martine Vanhove

b. usu min-é ka-yimi

he house-foc appl:abl-came
‘He has come from the house.’

The syntactic status of the Dhaasanac cognate morpheme ká is debated, and it is unclear
whether it marks an adjunct strategy or a preverbal strategy: Sasse (1976: 209) and
Mous (2006) analyze ká as a preverbal element, while Tosco (2001: 232–233, 259–261)
is more in favour of considering it an allative postposition.16 According to Tosco, ká
occurs in a fixed position, immediately before the verb, and follows directly a noun, to
which it is suffixed (even though he always transcribes ká as an independent word),
except when the noun is topicalized. Two arguments, duly mentioned by Tosco, could
support the preverbal analysis: (i) when the noun is topicalized, ká stays in its preverbal
position; (ii) object pronouns following the focus marker license a special set of “verbal
pronouns” directly preceding ká.
However, it might be the case that both analyses hold. There is one indisputable
case of a postpositional behaviour in one example (48) in Tosco’s texts in which ká is
separated from the verb by a direct object noun, thus breaking the rule of verbal adja-
cency.

(48) Dhaasanac (Tosco 2001: 317–318)

mé ká fás ɠaa ruut
head to blood have.prf goat
‘The he-goat had blood on its head.’

Apart from its use as an allative and dative marker with motion and ‘give’ verbs, there
is one example of ká with an intransitive middle verb which might be interpreted as an
applicative construction, but there are no other examples of this verb or similar con-
structions to confirm this interpretation.

(49) Dhaasanac (Tosco 2001: 260)

diidic h
a hí ká raaða
dung.beetle foc vpro.3.obj appl? take.by.surprise.mid.ipfv
‘The dung-beetle takes him by surprise.’

16 This disagreement is also the case for the other preverbal elements of Dhaasanac in the following
sections.
 24 Applicative constructions in Cushitic 857

3.2.2 The applicative í-/ú/u

Tunni (a dialect of Somali), has an applicative preverb í, with several grammatical

allomorphs depending on the pronoun it combines with. It has directional and dative
meanings with motion, ‘give’, and ‘say’ verbs, and has a benefactive interpretation with
transitive (50b–51) and intransitive verbs (52). It always precedes the verb, whatever
the position of the applied phrase. It can combine with the other applicative preverbs
(51). No corresponding base constructions were found.

(50) Tunni
a. áy sagáar[a] árrabkíi kú qáatə a soo nágt
Mrs. dikdik:art tongue:m.anp from take:3f juss here give.back:3f
‘Let Mrs. Dikdik, who took his tongue from him, give it back.’ (Tosco 1997: 158)
b. isín beesódán n-oo-kú taabtèen
2pl money:f:prox 1pl-appl:ben-abl take.2pl
‘You took this money from him for us.’ (Tosco 1997: 110)

(51) Tunni (Tosco 1997: 108)

rúbbunkán sóorta i-i-kí laabàayt
spoon.m:art food.f:art 1sg.obj-appl:ben-appl:ins stir:prog:2sg
‘You are stirring the food for me with this spoon.’

(52) Tunni (Tosco 1997: 160–162)

áw Mayow hidáa má kíi róon[o]
Aw Mayow so neg 2sg:appl:ben be.good
‘Aw Mayow, this is not good for you.’

In Standard Somali, the cognate applicative preverb is -u, and it is structurally similar
(with minor differences) to Tunni. No corresponding base constructions were found.

(53) Standard Somali (Biber 1984: 49)

Cabdi nin-ka b-uu suuq-a u-tagay
Abdi man-art foc-3sg market-art appl:ben.3sg-go.to.pfv
‘Abdi went for the man to the market.’

Boni as described by Sasse (1981: 257) has a cognate applicative preverb ú-, which is a
well-behaved applicative since no counterpart of the applied phrase can occur in the
base construction.
858 Martine Vanhove

(54) Boni (Sasse 1981: 277)

a. filí kadi
comb bought
‘He bought a comb.’
b. hác-hablo fíli ú-kadi
girl-def comb appl:ben.3sg-buy.pfv
‘He bought a comb for the girl.’

In Dhaasanac, notwithstanding the issue of the exact morphosyntactic status of pre-

verbal elements in this language (see § 3.2.1), (k)í, one of Tosco’s “core adpositions”,
has three readings: dative, benefactive (55b), and instrumental (56). The preverb can
encliticize to any preceding word; in such a case, the initial k is dropped.

(55) Dhaasanac
a. yáa bie ʔoŋolic fii
1sg.sbj water calabash pour.pfv
‘I poured water in the calabash.’ (Tosco 2001: 259)
b. bie ɗáat h
a yú ko (k)í fafaa
water calabash foc 1sg 2sg.obj appl:ben pour.rdp.ipfv
‘I’ll pour for you water in the calabash.’ (Tosco 2001: 237)

(56) Dhaasanac (Tosco 2001: 233)

gáal ʔéðe húol=í tuni
people guns refl=appl:ins hit.pfv
‘People fought (each other) with guns.’

Rendille uses an applicative preverb í- (for 1st and 3rd oblique pronouns) and kí- (for
2nd oblique pronouns).17 Those preverbs differ from the bound direct object pronouns,
which have low tones (57a). Heine (1980b: 238) only mentions the í- form.18 This
applicative construction is used with both transitive (57b) and intransitive verbs (58b),
with a benefactive interpretation.

(57) Rendille (Pillinger and Galboran 1999: 29)

a. usú i’-helé / ki’-helé
3sg.m.foc 1sg.obj-find.pfv 2sg.obj-find.pfv
‘He’s the one who found me/you.’

17 They are labelled “incorporated object prefixes” by Pillinger and Galboran (1999: 29).
18 Heine labels it a “verbal extension”.
 24 Applicative constructions in Cushitic 859

b. usú í’-helé / kí’-helé

3sg.m.foc appl:ben.1sg-1sg.m:find:pfv appl:ben.3sg.m-1sg.m:find:pfv
‘He’s the one who found [it] for me/him.’

(58) Rendille (Heine 1980b: 238)

a. usú írt-a
3sg.m go-3sg.m.ipfv
‘He will go.’
b. usú í-írt-a
3sg.m appl:ben-go-3sg.m.ipfv
‘He will go for him.’

3.2.3 The applicative kí

Tunni has an applicative preverb kí (which is not an allomorph of í, as in Rendille and

Dhaasanac) that indicates locative (59) and instrumental semantic roles (60) (Tosco
1997: 101) with inanimate applied phrases, as well as the addressee of ‘say’ verbs.

(59) Tunni (Tosco 1997: 102)

lámma is jéešə magàar sagáar sə kí ʄiifə́t[ə]
two refl love.3f skin dikdik npst appl:loc lie.3f
‘Two who love each other lie on a dikdik hide.’ (proverb)

(60) Tunni (Tosco 1997: 177–178)

rubbùun sóort la-kí lába
paddle porridge.f.art ips-appl:ins stir
‘The paddle with which they stir their porridge.’

3.2.4 The applicative íl/la-

Another applicative preverb of Tunni is the comitative íl (61). Its cognate form in stand-
ard Somali is la (62). No corresponding base constructions were found.

(61) Tunni (Tosco 1997: 102)

dád má s[i] íl də́gnə
people neg npst appl:com dwell.1pl
‘We don’t live with other people.’
860 Martine Vanhove

(62) Standard Somali (Biber 1984: 50)

Cabdi b-uu Cali geela ka-la-dhacay
Abdi foc-3sg.m Ali camels abl-appl:com-rob.pfv.3sg.m
‘He robbed some camels from Ali together with Abdi.’

Rendille leé- may also be a cognate preverb, but the sole example—leejira ‘be with
(someone)’, cf. jira ‘be, live’—and the comments in Pillinger and Galboran (1999: 29) are
too sketchy to be sure.

4 Applicative derivation
Kießling (1984: 94) mentions a derived applicative verb form in Burunge marked by the
prefix hii-/hay-, which “zeigt an, daß die Handlung in Bezug auf ein Objekt, im Hinblick
auf ein bestimmtes Ergebnis oder in Richtung auf ein bestimmtes Ziel hin unternom-
men wird” [indicates that the action is taken with respect to an object, with respect to a
particular result, or toward a particular goal]. He provides a short first list of base forms
with their corresponding applicative forms, and longer lists of applicatives with “spe-
cialized semantics” and fossilized forms. Utterances with only two intransitive verbs
of the first list could be found in his grammar. In (63b) the applied phrase keeps its
non-core marking, and there is no applied phrase in (64); the construction is thus better
viewed as an applicative lookalike with ventive semantics.

(63) Burunge
a. kalaamay ma gesaa day ’ayaage
better cons-l/2.sbj first come.1sg.pfv house:all
‘It is better I go home first.’ (Kießling 1984: 227)
b. higi hii-day maraage
3.sbj:seq appl-come:3sg.m:pfv:decl house:all
‘And he entered into the house.’ (Kießling 1984: 266)

(64) Burunge (Kießling 1984: 218)

yee dahagway hoo malaalee hii-tlayda
voc:m guest:m:poss.1sg 1/2.sbj:pros when appl-stand:2sg:ipfv
‘Oh my dearest, when will you rise?’

Kießling (1994: 94) relates the applicative prefix to a few fossilized verb forms with a
prefix ii-/in- in Iraqw, which Mous (1994: 199) tentatively relates to the Iraqw directive
case marker -i. Kießling (1994: 109) also reports on a malefactive derived verb with a
prefix ’afa-, which only occurs with some lexicalized items, not as a productive device.
 24 Applicative constructions in Cushitic 861

5 Dative doubling in Oromo: A possible lookalike?

Oromo uses an adjunct strategy with a dative case that marks the beneficiary. In addi-
tion, in both the Wollegga and Harar dialects, the dative case may occur twice, obliga-
torily on the beneficiary and optionally on the verb, after the inflectional morpheme:

(65) Oromo (Wollega dialect; Girma Mengistu Desta, p.c.)

ani ilma-koo-f konkolaataa=n bit-e(-ef)
1sg son-poss.1sg-dat car=1sg.sbj buy-pfv-dat
‘I bought a car for my son.’

(66) Oromo (Harar dialect; Owens 1985: 119)

sheex-níi isaaf-f hiddá heezabée-f
sheikh-nom 3sg.obj-dat medicine prescribe-dat
‘The Sheikh prescribed some (native) medicine for him.’

The construction with repetition of dative marking on the verb can be viewed as an
applicative lookalike that might constitute the initial stage of a grammaticalization
process towards a suffixal applicative strategy.

6 Conclusion
In what follows, I recapitulate the above analyses, trying to answer as best as I possibly
can the questions asked in the questionnaire about the morphology, syntax, and seman-
tics of applicative constructions, as well as the issue of lookalikes.

Morphology
– The survey of applicative constructions in eleven Cushitic languages showed that
the preverb strategy is more widespread than the auxiliary strategy (9/11 vs. 2/11
languages), and that verbal derivation is highly marginal in the sole language that
uses it. Although there is a good number of commonalities, each language has
developed its own system.
– Two main types of preverbs are attested: (i) preverbs, often clitics, which can bear
tam markers (the Iraqw-Alagwa-Burunge cluster, and two Omo-Tana languages,
Elmolo, and probably also Arbore); and (ii) preverbs, also often clitics, that cannot:
all Omo-Tana languages, except Tunni. Therefore, this bi-partition corresponds only
partly to the genealogical subgroupings. Most applicative preverbs have stemmed
from adpositions. However, degrees of grammaticalization seem to vary across lan-
guages, although it is not always easy or feasible to assess them from the available
grammars. For instance, no examples of applied lexical phrases were found for
some markers in Iraqw, Burunge, Alagwa, Arbore, Tunni and Dhaasanac, but the
862 Martine Vanhove

descriptions rarely specify whether they are licensed or not, except for Alagwa,
where the applicative s- is said to be restricted to pronominal applied phrases and
to nouns in preverbal position. Descriptions do not mention either possible polarity
restrictions, but one negative example (52) was found in Tunni.
– As for the two languages with auxiliaries, Xamtanga and Beja, both share the ‘give’
strategy, but only the former has in addition a ‘leave’ strategy. In Xamtanga they
always combine with a preceding lexical verb in the converb form, while in Beja,
finite lexical verbs are also licensed. There are limitations to their grammatical-
ized status. In both Xamtanga and Beja, the auxiliaries cannot co-occur with them-
selves as lexical verbs. On the other hand, examples show that applicativization
is compatible with the negative polarity in Beja, while it not the case in Xamtanga
(Darmon 2015: 196). In Beja, the periphrastic applicative construction is limited to
pronominal applicative phrases. Conversely, Xamtanga provides no examples with
pronouns, but this issue is not discussed.
– Verbal derivation, with a prefix, is limited to one of the applicative constructions of
Burunge and does not seem very productive.
– Allomorphy is attested in six languages with the preverb strategy for some applica-
tive morphemes. In most cases allomorphy is grammatically conditioned (by the
presence of bound pronouns and/or other preverbal elements), in a few others it is
due to phonological processes of vowel harmony or cliticization.
– In all languages, applicativized verbs show comparable inflectional paradigms to
those of their base counterparts.

Syntax
– In most cases, the applied phrase is a P. With the preverbal strategy, P may belong
to a special set of bound object pronouns. In some languages, it is an X, usually a
dative, as in Xamtanga, rarely an adpositional phrase.
– Depending on the language, and the marker, the syntactic status of the applied
phrase’s companion arguments or adjuncts may change or not between the base
constructions and the applicative construction. In the languages with the preverb
strategy, the applied phrase is mostly not marked with an adposition, contrary to
the base construction, or it combines with different preverbal elements than in
the base construction. However, with some preverbs, the adposition or case is kept
in the applicative construction. In Xamtanga, the dative case applies to both con-
structions, but there is an optional alternation with the accusative case with the
auxiliary bär- ‘leave, abandon’. Beja retains the object marking of the beneficiary
in the applicative construction.
– In the descriptions, little is mentioned about voice operations which can combine
with applicativization, but sporadic examples are found. Such is the case with the
causative and the middle forms in Beja. Conversely, the Xamtanga applicative ‘give’
construction is incompatible with the causative and the mediopassive (Darmon
 24 Applicative constructions in Cushitic 863

2015: 196). In Alagwa and Tunni, the applicative preverbs are used with an imper-
sonal subject (a passive lookalike). In some languages, there are very few examples
of applied phrases with reflexive and reciprocals.
– In Beja, the applied construction is a special subset of a ditransitive.
– From the available descriptions, it is difficult to know how rigid or flexible the
assignment of case or agreement frames to applied constructions is when com-
pared to base constructions, but my general impression is that it is rigid.
– It seems that applicativization does not condition the access of non-core syntactic
arguments to operations such as relativization or focalization (several examples in
relative clauses, clefts, and with focus markers are mentioned for Alagwa, Arbore,
Dhaasanac, Rendille, and Somali, and this is also the case for relativization in Beja).

Semantics
– Although Cushitic languages do not have all-purpose applicative constructions com-
parable to the Bantu applicative using ✶-ɪd (Pacchiarotti, this volume), it is possible
to distinguish two types of applicative (or applicative lookalike) markers, some
being more semantically specialized than others. The highly specialized ones are
the following: -rar is instrumental in Iraqw (and comitative with animate applied
phrases in Burunge and Alagwa); s(V)- is benefactive in Alagwa; ni is ventive in
Burunge; i/ú/u is benefactive in Tunni, Boni, and Rendille; íl/la is comitative in
Tunni and Standard Somali; and the ‘give’ auxiliary is benefactive in Xamtanga. On
the other hand, descriptions rarely specify whether the applicative construction is
limited or not to transitive or intransitive verbs. A restriction to transitive verbs is
specifically mentioned for the yiw ‘give’ auxiliary construction in Xamtanga, and
the absence of valency restrictions is the case in Beja, as well as for part of the
markers of the other languages. However, the absence of one verb type or the other
for some markers in Iraqw, Alagwa, and Burunge may be just due to a gap in the
documentation.
– The most frequent semantic interpretation of applied phrases is benefactive, then
instrumental(-comitative), and finally locative (ventive, ablative). Malefactive
(Xamtanga, Beja) and causal (Burunge, Beja) are marginal.
– From the available data, it is difficult to assess whether there are semantic roles
that can only be expressed by means of an applicative construction for each lan-
guage. At most, some applied phrase cannot occur in the corresponding base con-
structions. This is the case for the locative in Alagwa with the preverb ee-, for the
benefactive with ú- in Boni, and for the benefactive, instrumental and locative with
ka- in Elmolo.
– With the exception of Xamtanga, where affectedness is specific to the benefactive
construction, as opposed to the base construction, nothing in the descriptions shows
that there is a semantic difference between the applied phrase and its counterpart
in the base construction. Once again, this may be due to gaps in the documentation.
864 Martine Vanhove

– In four languages—Burunge, Alagwa, Iraqw for the ni-applicative, and Rendille for
the ka-applicative—the comments of the authors show that there is a pragmatic
difference between the applied phrase and its counterpart in the base construction,
but the exact link with focus or topic is often unclear. Dhaasanac may also be a case
where pragmatics plays a role, for all applicative preverbs, if Tosco’s analysis in
terms of topicalization of the applied phrase is correct.
– Xamtanga is the sole language where discourse-sensitive specificity of the applica-
tive construction is mentioned. The yiw ‘give’ auxiliary construction adds a specific
engager / discourse prominent recipient-like semantic role to the dative argument.

Lookalikes
– In the Cushitic languages studied, there is always an applicative marking on the
verb. However, if one sticks to the third criteria of the definition of applicative con-
structions proposed in Zúñiga and Creissels (this volume), which requires a non-
core marking on the applied phrase different from its non-core marking in the
base construction, then most of the constructions are lookalikes. Additionally, the
morphemes for which an applicative analysis can be considered, at least in some
constructions, are often also found in constructions in which no applied phrase can
be identified. They are best analyzed as directionals.
– The case of Oromo is different, and, in some respects, reminiscent of the situation
found in Amharic (Ethio-Semitic), although not entirely identical (Amberber, this
volume). In Oromo, an adjunct strategy with a dative case, marking the applied phrase,
is the rule. Optionally the dative case can also occur suffixed on the verb, which may
be an incipient grammaticalization towards a suffixal applicative construction.
– Historical or comparative data are lacking, which makes it difficult to assess whether
some markers may have turned from erstwhile applicatives into strictly valen-
cy-neutral markers, but synchronic data does not seem to point in this direction.

Abbreviations
abl ablative cond conditional
acc accusative cons consecutive
all allative cop copula
anp anaphoric cvb converb
appl applicative dat dative
art article decl declarative
ben benefactive def definite
bckg background dem demonstrative
caus causative dir directional
com comitative dur durative
comp completive emph emphatic
 24 Applicative constructions in Cushitic 865

ep epenthetic vowel pfv perfective

f feminine pl plural
foc focus marker plr polar
fut future poss possessive
gen genitive pret preterit
gnrl general prf perfect
hab habitual pro pronoun
head head agreement prog progressive
imp imperative proh prohibitive
ind indicative pros prospective
indf indefinite prox proximal
ins instrumental prs present
intrj interjection pst past
ips impersonal subject q question
ipfv imperfective rdp reduplication
juss jussive reas reason
lnk linker refl reflexive
loc locative sbj subject
medp medio-passive sbjv subjunctive
mid middle sep separative-intrusive
mulcl multifunctional clitic seq sequential
n neuter sg singular
neg negative sub subordinator
nmlz nominalizer sbj subject
nom nominative vent ventive
npst non-past vn verbo-nominal
obj object voc vocative
opt optative vpro verbal pronoun

References
Appleyard, David. 1990. Prepositional particles in Somali and their cognates in other Cushitic languages.
African Languages and Cultures 3(1). 15–31.
Banti, Giorgio. 1997. Review of Maarten Mous: “A Grammar of Iraqw”. Journal of African Languages and
Linguistics 18. 95–105.
Biber, Douglas. 1984. The diachronic development of preverbal case markers in Somali. Journal of African
Languages and Linguistics 6. 47–61.
Crass, Joachim & Ronny Meyer. 2008. Ethiopia. In Bernd Heine & Derek Nurse (eds.), A linguistic geography of
Africa, 228–250. Cambridge: Cambridge University Press.
Creissels, Denis. 2010. Benefactive applicative periphrases: A typological approach. In Fernando Zúñiga
& Seppo Kittilä (eds.), Benefactives and malefactives: Typological perspectives and case studies, 29–70.
Amsterdam: John Benjamins.
Darmon, Chloé. 2015. A morphosyntactic description of Xamtanga: an Agaw (Central Cushitic) language of the
northern Ethiopian highlands. Lyon: Université Lumière Lyon 2 dissertation.
Dixon, R. M. W. & Alexandra Aikhenvald. 2000. Introduction. In R. M. W. Dixon & Alexandra Aikhenvald
(eds.), Changing Valency: Case studies in transitivity, 1–29. Cambridge: Cambridge University Press.
866 Martine Vanhove

Eberhard, David M., Gary F. Simons & Charles D. Fennig (eds.). 2022. Ethnologue: Languages of the world.
25th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com. [accessed
10/11/2022]
Eyob Kelemwork Wolde. 2015. A grammar of Gedeo: A Cushitic language of Ethiopia. Hyderabad: Osmania
University dissertation.
Hayward, Dick. 1984. The Arbore language: A first investigation including a vocabulary. Cologne: Rüdiger Köppe.
Heine, Bernd. 1980a. Elmolo. In Bernd Heine (ed.), The non-Bantu languages of Kenya, 173–218. Berlin:
Dietrich Reimer.
Heine, Bernd. 1980b. Rendille. In Bernd Heine (ed.), The non-Bantu languages of Kenya, 219–243. Berlin:
Dietrich Reimer.
Kemmer, Suzanne. 1993. The middle voice. Amsterdam: John Benjamins.
Kießling, Roland. 1994. Eine Grammatik des Burunge. Hamburg: Research and Progress.
Kießling, Roland. 1996. Verbal inflectional suffixes in the West Rift group of southern Cushitic. In Catherine
Griefenow-Mewis & Rainer Voigt (eds.), Cushitic and Omotic languages: Proceedings of the 3rd
International Symposium, Berlin, March 17–19, 1994, 59–70. Cologne: Rüdiger Köppe.
Kießling, Roland. 2001. South Cushitic links to East Cushitic. In Andrzej Zaborski (ed.), New data and new
methods in Afroasiatic linguistics: Robert Hetzron in memoriam, 95–102. Wiesbaden: Otto Harrassowitz.
Mengistu Amberber. This volume. Applicativization in Amharic.
Mous, Maarten. 1993. A grammar of Iraqw. Hamburg: Helmut Buske.
Mous, Maarten. 2001. Basic Alagwa syntax. In Andrzej Zaborski (ed.), New data and new methods in
Afroasiatic linguistics: Robert Hetzron in memoriam, 125–135. Wiesbaden: Otto Harrassowitz.
Mous Maarten. 2004. The middle in Cushitic languages. In Andrew Simpson (ed.), Proceedings of the twenty-
seventh annual meeting of the Berkeley Linguistics Society, March 22–25, 2001: Special session on Afroasiatic
languages, 75–86. Berkeley: Berkeley Linguistic Society.
Mous, Maarten. 2006. Selectors in Cushitic. In F. K. Erhard Voeltz (ed.), Studies in African lLinguistic typology,
303–325. Amsterdam: John Benjamins.
Mous, Maarten. 2012. Cushitic. In Zygmunt Frajzyngier & Erin Shay (eds.), The Afroasiatic languages,
342–422. Cambridge: Cambridge University Press.
Mous, Maarten. 2016. Alagwa: A South Cushitic language of Tanzania. Cologne: Rüdiger Köppe.
Owens, Jonathan. 1985. A grammar of Harar Oromo (Northeastern Ethiopia). Hamburg: Helmut Buske.
Pacchiarotti, Sara. This volume. The polyfunctional applicative ✶-ɪd in Bantu languages.
Pillinger, Steve & Letiwa Galboran. 1999. A Rendille dictionary. Cologne: Rüdiger Köppe.
Sasse, Hans-Jürgen. 1976. Dasenech. In Lionel M. Bender (ed.), The Non-Semitic languages of Ethiopia,
196–221. East Lansing, MI: African Studies Center, Michigan State University.
Sasse, Hans-Jürgen. 1981. Basic word order and functional sentence perspective in Boni. Folia Linguistica
15(3–4). 253–290.
Simeone-Senelle, Marie-Claude & Martine Vanhove. 2003. Transcatégorialité, auxiliarité et auxiliarisation. In
Stéphane Robert (ed.), Perspectives synchroniques sur la grammaticalisation. Polysémie, transcatégorialité
et échelles syntaxiques, 123–126. Leuven/Paris: Peeters.
Shibatani, Masayoshi. 2009. On the form of complex predicates: Toward demystifying serial verbs. In
Johannes Helmbrecht, Yoko Nishina, Yong-Min Shin, Stavros Skopeteas & Elisabeth Verhoeven (eds.),
Form and function in language: Research papers in honour of Christian Lehmann, 255–282. Berlin/Boston:
De Gruyter Mouton.
Tosco, Mauro. 1997. Af Tunni: Grammar, texts and glossary of a Southern Somali dialect. Cologne: Rüdiger
Köppe.
Tosco, Mauro. 2000a. Cushitic overview. Journal of Ethiopian Studies 33(2). 87–121.
Tosco, Mauro. 2000b. Is there an “Ethiopian language area”? Anthropological Linguistics 42(3). 329–365.
Tosco, Mauro. 2001. The Dhaasanac language: Grammar, texts, vocabulary of a Cushitic language of Ethiopia.
Cologne: Rüdiger Köppe.
 24 Applicative constructions in Cushitic 867

Treis, Yvonne. 2023. Reflexive constructions in Kambaata. In Katarzyna Janic, Nicoletta Puddu & Martin
Haspelmath (eds.), Reflexive constructions in the world’s languages, 155–184. Berlin: Language Science
Press. DOI: 10.5281/zenodo.7874940
Vanhove, Martine. 2012. Roots and patterns in Beja (Cushitic): The issue of language contact with Arabic. In
Martine Vanhove, Thomas Stolz, Hitomi Otsuka & Aina Urdze (eds.), Morphologies in contact, 311–326.
Berlin: Akademie Verlag.
Vanhove, Martine. 2017. Le Bedja. Leuven: Peeters.
Vanhove, Martine. 2020a. Beja. In Christopher Lucas & Stefano Manfredi (eds.), Arabic and contact-induced
change, 419–439. Berlin: Language Science Press.
Vanhove, Martine. 2020b. Corpus de bedja. https://corporan.huma-num.fr/Archives/corpus.php.
Zúñiga, Fernando & Denis Creissels. This volume. Applicative constructions: An introductory overview.
Peter Arkadiev, Yury Lander, and Irina Bagirokova
25 Applicative constructions in the Northwest
Caucasian languages
Abstract: This chapter describes applicative constructions in the polysynthetic North-
west Caucasian languages, which are typologically unusual in several respects. First,
these languages possess an extraordinarily rich system of applicatives whose seman-
tic functions range from benefactive, comitative and malefactive to fairly specialized
spatial meanings. Second, the Northwest Caucasian applicatives invariably introduce
indirect objects, thus almost never affecting the ergative-absolutive alignment of core
arguments and serving as important and often only means of integrating peripheral
participants into clausal structure. We describe the morphology, syntax and semantics
of applicatives, as well as a range of non-trivial phenomena such as the semantically
empoverished and morphosyntactically special “dative” applicative and the uses of
applicatives in agent demotion and clause combining.

1 Introduction
This chapter describes applicative constructions in the Northwest Caucasian (NWC) lan-
guages. NWC is interesting and instructive for the typology of applicative constructions
for at least two reasons:
– these languages possess extraordinarily rich systems of applicative markers whose
semantics ranges from the cross-linguistically common benefactive and comitative
applicatives to applicatives with fairly specialized spatial meanings, and
– the NWC applicatives differ from canonical applicatives as discussed, for example,
by Peterson (2007) in many respects, most notably in that the syntactic status of the
AppP in NWC is indirect rather than direct object and that applicatives serve as
important and often only means of integrating peripheral participants into clausal
structure.

Applicatives in NWC languages are relatively well described. This survey is based
mostly on our own fieldwork, but we also use data from various other sources, in particu-
lar, Smeets (1992), Paris (1987), Lomtatidze (1976), O’Herin (2001), Letuchiy (2009), Fell

Acknowledgments: The first and the second authors contributed to writing up the main text, while all authors
are responsible for collecting and analyzing the data presented here. We thank all our consultants for their
generous help and the editors of the volume for their invitation, patience, useful comments and meticulous
editing. Yury Lander gratefully acknowledges the support from the Basic Research Program of the National
Research University Higher School of Economics. All errors are ours.

https://doi.org/10.1515/9783110730951-025
870 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(2012), Ponomareva (2013) and Chirikba (2020). For the sake of exposition, we illustrate
the system primarily with examples from West Circassian, a language for which we have
more detailed data and can use large corpora, but add examples from some other NWC
languages to illustrate the parallel or distinct behavior. Whenever unmarked, examples
come from corpora (Arkhangelskiy et al. 2018–2022; Bagirokova et al. 2020; Arkadiev
et al. 2020; Panova et al. 2019), examples elicited or taken from other sources are marked
as such.
The structure of this chapter is as follows. Section 2 provides the necessary back-
ground on NWC languages. In Section 3 we discuss morphological and syntactic aspects
of applicative constructions in these languages. Section 4 is devoted to the semantic
diversity of grammatical and locative applicatives, while Section 5 deals with dative
applicatives, which show specific behavior. Section 6 focuses on the morpheme order
in forms containing multiple applicatives. In Section 7 we touch upon non-applicative
functions of applicatives, while in Section 8 we look at a phenomenon that is function-
ally similar to applicatives but is not related to the grammatical subsystem discussed
here. The last section summarizes the main typological characteristics of NWC applic-
atives.

2 Background on Northwest Caucasian languages

2.1 General information

The NWC (or Abkhaz-Adyghean) family is one of the three autochthonous language
families of the Caucasus (for a general background on NWC, see Hewitt 2005 and Ark-
adiev and Lander 2020). It comprises at least four living languages, namely Abkhaz
(ISO 639–3: abk) and Abaza (abq), which constitute the Abkhaz-Abaza branch, and West
Circassian (also known as Adyghe, ady) and Kabardian (sometimes called East Circas-
sian, kbd), which constitute the Circassian branch. In addition, there is one extinct NWC
language which has been thoroughly documented (though from few speakers) in the
20th century, namely Ubykh (uby), which is usually thought to be closer to Circassian
languages than to Abkhaz-Abaza.
Originally, NWC languages were spoken in the West Caucasus: to the South of
the Greater Caucasus Mountains in the territory of the Republic of Abkhazia and to
the North of the mountains in the now territories of several regions belonging to the
Russian Federation, namely Krasnodarskij Kraj, Adygea, Karachaevo-Cherkessia and
Kabardino-Balkaria. In the 19th century, however, when these lands were occupied by
the Russian Empire as a result of the Caucasian War, most speakers of NWC languages
migrated into the Ottoman Empire, and the remaining speakers were resettled into a
number of disconnected areas. As a result, currently there is also a large diaspora which
uses NWC languages (but to different extents) in Turkey, Syria, Jordan, and Israel.
 25 Applicative constructions in the Northwest Caucasian languages 871

The sociolinguistic situation of the NWC languages is ambivalent. On the one hand,
Abkhaz is the state language of Abkhazia, and West Circassian, Kabardian and Abaza
are recognized as official languages with written standards and some presence in the
media and education in the Russian republics of Adygea, Karachaevo-Cherkessia and
Kabardino-Balkaria. On the other hand, both in Russia and in Abkhazia all NWC lan-
guages experience pressure from Russian, and their use is largely limited to rural areas
and informal settings. Most if not all speakers of NWC languages in Russia and Abkhazia
are bilingual in Russian, and for the younger generations Russian is becoming the dom-
inant language. Bilingualism and/or the shift to a dominant language is also the norm
in the diaspora, where the sociolinguistic situation varies from good and stable (e.g.,
among Circassians in Israel) to the complete language shift (e.g., in many communities
in Turkey).

2.2 Basics of morphosyntax

NWC languages are generally characterized as polysynthetic, both prefixing and suf-
fixing, ergative (morphologically but possibly also syntactically), predominantly head-
(Abkhaz-Abaza) or double-marking (Circassian and Ubykh). Further, the languages are
consistently left-branching, i.e. have possessors preceding the possessa, postpositions
rather than prepositions and the basic Actor-Undergoer-Predicate word order. Exаmple
(1) illustrates some of these features: here we find two noun phrases marked with an
“oblique” case expressing an ergative and an indirect object arguments, an absolutive
noun phrase and a polysynthetic predicate whose morphology indexes (at least) two
arguments.

(1) West Circassian

djeɮe-m djeɮe-m məŝe-r ∅-r-j-e-ʁe-wəč̣ʼə
fool-obl fool-obl bear-abs 3sg.io-dat-3sg.erg-dyn-caus-kill
‘A fool makes a fool kill a bear.’

The core of the clause is constituted by the predicate, which includes indexing of
up to four (in Abkhaz, Abaza and Ubykh) or probably even more (in Circassian lan-
guages) participants primarily expressed by prefixes; some 3rd person prefixes in some
languages are null. All content words may serve as predicates and take predicate mor-
phology, hence the distinction between nouns and verbs in this position is very weak,
if exists at all; therefore, we prefer to use the term ‘predicate’ rather than ‘verb’. Much
more important for the morphology is the distinction between stative and dynamic
predicates, the former class including both noun-like words and some words expressing
non-nominal concepts (e.g., certain posture predicates).
The make-up of the predicate is quite complex and includes several morphological
zones, see Figure 1.
872 Peter Arkadiev, Yury Lander, and Irina Bagirokova

prefixes root suffixes

pre-stem
argument structure zone stem endings
elements

subordinators
illocutionary
operators or
and indirect
applicatives

modal and
absolutive

evaluative
preradical

aspectual,
temporal,

operators
causative
negation

negation
dinators

ergative

suffixal
objects
subor-

root
1 1 >1 1 1 1 or 2 >1 1 >1

Figure 1: The simplified structure of the NWC predicate (cf. Arkadiev and Lander 2020: 404).

Here we are concerned mainly with the argument structure zone, which occupies most
of the prefixal part of the predicate, contains person/number indexing of all basic par-
ticipants of the state-of-affairs and includes the applicative morphology (if any). The
argument structure zone opens with the absolutive indexing prefix (which can be
absent under conditions different for different languages) and ends with the ergative
indexing (with transitive predicates). In-between we find directional prefixes (‘hither’
and ‘thither’), whose position varies across languages, indirect objects and applicative
complexes discussed in detail in subsequent sections, some subordination morphology
(probably related to applicatives) and in Abaza and Abkhaz also some negation and
aspectual prefixes. Cf. comparable forms (2) and (3), which demonstrate most kinds of
prefixal morphology; square brackets indicate the boundaries of the argument struc-
ture zone.

(2) West Circassian

[z-ja-ŝʷə]-mə-t
rfl.abs-3pl.io+dat-2pl.erg-neg-give[imp]
‘Do not give yourselves to them.’

(3) Abaza
[j-g’-ʕa-sə-rə]-m-t-χ-ṭ
3sg.n.abs-neg.emp-csl-1sg.io-3pl.erg-neg-give-re[aor]-dcl
‘They didn’t give it back to me.’

Cross-reference morphology indexes person-number and in Abkhaz-Abaza also gender.

The same paradigms also include prefixes with the reflexive, reciprocal, and relative
functions. It is commonly assumed that at least three series of indexing prefixes can
be distinguished, namely absolutive, indirect object and ergative, although the latter
two are hardly distinguishable in most contexts and/or person-number combinations.
In addition, Ubykh distinguishes a separate series of possessive prefixes (also distinct
mainly in the 3rd person), while West Circassian has a series of prefixes indexing post-
positional objects.
Flagging morphology is limited. Abkhaz-Abaza only display adverbial and instru-
mental suffixes, while core noun phrases as well as adnominal possessors and objects
 25 Applicative constructions in the Northwest Caucasian languages 873

of postpositions are unmarked. By contrast, Circassian and Ubykh also have markers for
core cases, i.e. absolutive (S/P) and oblique, the latter marking basically all non-absolutive
arguments, including the ergative A, indirect objects, adnominal possessors and objects
of postpositions, and in some marginal constructions even the absolutive argument (see
Lander, Belyaev, and Bagirokova 2021). For example, in (1) above we observe a phrase
referring to P and marked with the absolutive as well as two oblique-marked phrases,
one referring to the ergative A and the other to the indirect object. Note, however, that in
Ubykh the absolutive is unmarked in both singular and plural, while in Circassian core
case markers are normally absent on personal pronouns, possessed nominals and are
usually absent on non-specific nominals (cf. Arkadiev and Testelets 2019).

2.3 A note on the transcription

Throughout this paper, examples from all NWC languages are given in the unified Cau-
casological transcription commonly employed in works on NWC and Kartvelian lan-
guages rather than in IPA (see Arkadiev and Lander 2020: 373–376). The most important
divergences from IPA are as follows: ejective consonants are marked by a dot below
or above the symbol; palatalization is marked by an apostrophe; c, č, š, ʒ, ǯ, ž denote
dento-alveolar affricates and fricatives; ŝ, ẑ, ĉ denote the so-called hissing-hushing con-
sonants usually identified as alveolo-palatal but whose exact phonetic interpretation
is subject to variation and controversy. Glossing of examples is also unified or added if
absent in the source.

3 Formal aspects of Northwest Caucasian

applicatives
3.1 Morphology: Applicative complexes

NWC languages possess rich systems of applicative prefixes which typically specify the
semantic roles of applied objects (see Section 4). Applicative markers usually appear
together with indirect object prefixes indexing the participant they introduce within
so-called applicative complexes. The canonical applicative complex, then, consists of
an indexing prefix and an applicative prefix, which immediately follows it. (Note that
in Circassian languages and in Ubykh 3sg indirect prefixes are null, so it is quite typical
to have an applicative without an overt indexing prefix.) Both the applicative prefixes
and the cross-referencing prefixes can show (morpho)phonologically conditioned allo-
morphy, such as consonant assimilation, vowel alternation or elision.
The following examples display simple applicative complexes taken in brackets:
874 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(4) West Circassian

qə-[d-de]-psewə-ʁ
csl-1pl.io-com-live-pst
‘S/he lived with us.’

(5) Kabardian, Besleney dialect

q̇ ə-[s-xʷe]-f-xə-ž’
csl-1sg.io-ben-2pl.erg-carry-re[imp]
‘Bring it back to me.’

(6) Ubykh (Fenwick 2011: 111)

š’əʁʷe š’ə-məz a-[š’-ĉǝ]-dw-ewtw-q̇ e-jṭ
1pl.poss 1pl.pr-child 3sg.abs-1pl.io-mal-die-fut-pst-rs.sg
‘Our child would have died [and been taken] from us.’

(7) Abkhaz (Hewitt 1979: 114)

a-žahwa s-[a-la]-jǝ-sǝ-jṭ
def-hammer 1sg.abs-3sg.io.n-instr-3sg.m.io-hit[aor]-dcl
‘I hit him with the/a hammer.’

There are several deviations from this canon. First, an applicative complex may
contain more than one prefix specifying the role of the applied object. In such cases, the
general rule is that a prefix has wider semantic scope with respect to the preceding part
of the applicative complex. In the following example an applicative preverb introduc-
ing “a flat, broad or open surface or area” (Fenwick 2011: 112) is followed by a transla-
tive preverb, while their combination is introduced by a general locative applicative:

(8) Ubykh (Fenwick 2011: 115)

ze-q̇ ʕaŝə-n [∅-g’ə-ʁe-le]-χʷe-g’ə
one-village-obl 3sg.io-loc:area-trans-loc-pass-cvb
‘As he was passing through a village. . .’

Where a single participant is introduced by several applicatives proper, as in (9a), we

may still assume a layered structure (albeit such examples also allow other interpreta-
tions such as postulating a kind of coordination of applicatives). Reversing the order
of applicatives in such examples does not change the meaning, cf. a similar Kabardian
example with the opposite order of applicatives, (9b).

(9) West Circassian (elicited) and Kabardian (Applebaum 2013: 52)

a. WC [[p-fə]-de]-s-ṣ̂ə-ʁ
2sg.io-ben-com-1sg.erg-do-pst
‘I did it with and for you.’
 25 Applicative constructions in the Northwest Caucasian languages 875

b. K sə-ŝə-[[b-də]-xʷe]-ɮeẑ-a-m
1sg.abs-temp-2sg.io-com-ben-work-pst-obl
‘When I had worked with you, for you.’

The second deviation relates to the fact that in Circassian some cross-reference pre-
fixes may “move to the left” and become separated from their applicatives. A widely
attested situation is the separation of the 3rd person plural affix from its applicative
by a directional preverb, as in (10b). However, under circumstances which still wait
for research such indexing prefix can be separated from its applicative by another
applicative (11). In some Circassian subdialects, the speakers also allow the separation
of the reflexive/relative prefix (12b). Moreover, some varieties of West Circassian even
display the rearrangement of indexing prefixes before the applicative (Kumakhova
1972: 72–76), but this phenomenon is still worthy of investigation.

(10) West Circassian (Kumakhov and Vamling 2009: 110)

a. we ŝʷəzə-xe-m wə-q-[a-de]-ča-ʁ
2sg wife-pl-obl 2sg.abs-csl-3pl.io-com-run-pst
b. we ŝʷəzə-xe-m w-[a]-qə-[de]-ča-ʁ
2sg wife-pl-obl 2sg.abs-3pl.io-csl-com-run-pst
Both: ‘You ran here together with women.’

(11) West Circassian

a-ʁe-wəcʷə-ʁe qʷaǯ’e-m meš’ət
3pl.erg-caus-stand-pst village-obl mosque
q-[a]-š’ə-[f]-a-ṣ̂ə-ʁa-ʁ
csl-3pl.io-loc-ben-3pl.erg-make-pst-pst
‘They made a mosque for them in the village founded by them.’

(12) Kabardian, Besleney dialect (elicited)

a. q̇ ə-[z-ḟ ]-je-s-a
csl-rfl.io-mal-dat-swim-pst
b. [zə]-q̇ ə-[ḟ ]-je-s-a
rfl.io-csl-mal-dat-swim-pst
Both: ‘He swam here against his own will.’

The third deviation is found in Abaza and Abkhaz and concerns a number of
applicative prefixes with spatial meanings that normally do not take person-number
prefixes cross-referencing the AppP (see e.g. Avidzba 2017: 115–122), at least when the
latter is singular non-human, cf. (13) with a “regular” and “bare” occurrence of the
same applicative.
876 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(13) Abaza (Klychev 1995: 111–112)

a. a-ʒə a-s j-[a-kʷ]-ḳʷḳʷ-əj-ṭ
def-water def-shirt 3sg.n.abs-3sg.n.io-loc:top-drip-prs-dcl
‘Water is dripping on the shirt.’
b. a-divan də-[kʷ]-naʕa-ṭ
def-sofa 3sg.h.abs-loc:top-recline[aor]-dcl
‘S/he reclined on the sofa.’

In fact, as argued by Lomtatidze (1983) and Avidzba (2017: 115), even those applicatives
that never co-occur with the third person singular non-human indirect object prefix
a- (14a) can be preceded by other cross-referencing prefixes, such as the third person
plural prefix r- (14b) or the relativization prefix z- (14c), on which see Section 3.2.

(14) Abaza (Klychev 1995: 57)

a. a-wasa-kʷa a-ḳʷar j-[bž’a]-r-ḳʷab-əw-n
def-sheep-pl def-canyon 3pl.abs-loc:middle-3pl.erg-bathe-ipf-pst
‘They were bathing sheep in the canyon.’
b. awat c̣la–dəw-ḳ1 [r-bž’a]-jə-ṭ
dem.pl tree–big-indf 3pl.io-loc:middle-come[aor]-dcl
‘A large tree grew between them.’1
c. a-c j-ʕa-[z-bž’a]-kša-z c̣la-ta
def-thunder 3sg.n.abs-csl-rel.io-loc:middle-hit-pst.nfin tree-adv
‘like a tree struck by a thunder’

Usually, applicative complexes modify their bases in a manner akin to adjuncts

in syntax (cf. Colarusso 2006: 29–30 for Kabardian and Lander 2015 for West Circas-
sian). There are two pieces of evidence for this. First, applicative complexes are nor-
mally non-obligatory, although there are some applicativa tantum predicates involving
“bound” roots (e.g., in Circassian posture roots ‘sit’, ‘stand’, ‘lie’, as well as ‘be’, roots
expressing directed motion such as ‘go in’ and ‘go out’ and some others) which cannot
be used without locative preverbs introducing the landmark argument, as in examples
(15a) and (15b). Example (15c), by contrast, shows a clearly optional and even occa-
sional use of an applicative.

(15) West Circassian

a. ʔaše [z-gʷe]-λ-ep / ✶λ-ep
weapon 1sg.io-loc:near-lie-neg lie-neg
‘I have no weapon with me (lit. weapon does not lie next to me).’

1 We use the en-dash to separate stems within productive nominal compounds, the so-called nominal
complexes (see Lander 2017).
 25 Applicative constructions in the Northwest Caucasian languages 877

b. c̣əf–weterə-m [pxə-rə]-č̣ʼə-ʁ / ✶č̣ʼə-ʁ

person–crowd-obl [3sg.io]loc:through-trans-go_out-pst go_out-pst
‘(He) went through the human crowd.’
c. wešʼx q-[a-f]-je-šʼxə-r-ep
rain csl-3pl.io-ben-dat-rain-dyn-neg
‘It does not rain for them.’

Second, there may be several applicative complexes within the same predicate, i.e.
they allow recursion (see e.g. Lander and Letuchiy 2010); see Section 6 for the order of
complexes in such forms. In fact, even the same applicative can occur twice, albeit in
different meanings, cf. (16), with two instances of the benefactive, one introducing the
addressee and the other the beneficiary:

(16) West Circassian (Lander and Letuchiy 2010: 269)

s-[a-fə]-[∅-f]-e-txe
1sg.abs-3pl.io-ben-3sg.io-ben-dyn-write
‘I write to him for their benefit.’
‘I write to them for his benefit.’

However, the stacking of applicative complexes has different productivity in different

languages of the family. Thus, while Circassian texts present numerous forms contain-
ing several complexes, for Abaza, O’Herin (2001: 483–485) presents only elicited exam-
ples of multiple applicatives and reports that “many speakers consider some forms
either degraded or completely unacceptable” (O’Herin 2001: 483, fn. 8). Our own field-
work experience on Abaza confirms this impression.
In all examples provided above, applicative complexes occur in the argument struc-
ture zone. It is worth noting, however, that there are some exceptions to this. In particu-
lar, “neutralized applicatives” whose combinations with the root are lexicalized appear
in the stem (see Section 7.3). Furthermore, Hewitt (2008a) provides Abkhaz examples like
(17), where applicative complexes exceptionally occur to the right of the negative prefix:

(17) Abkhaz (Hewitt 2008a: 310)

dǝ-m-[rǝ-ĉwχa]-sə-jṭ
3sg.h.abs-neg-3pl.io-appl-hit[aor]-dcl
‘S/he did not shove them.’

3.2 Syntax

All applicatives in NWC can be considered D-applicatives, since the argument they intro-
duce is invariably an indirect object sharing its morphosyntactic properties with the
recipient/goal argument of ditransitive predicates. The latter statement, however, can be
878 Peter Arkadiev, Yury Lander, and Irina Bagirokova

considered to some extent circular, since, as will be shown in Section 5, at least in Circas-
sian languages there are simply no indirect objects not introduced by some applicative,
including the recipient of ‘give’. Taking this into account, we could perhaps say that Cir-
cassian applicatives introduce a special type of argument having no parallels with argu-
ments of simplex verbs. In any case, the syntactic status and morphosyntactic encoding
of the arguments of the BC normally remain intact in the AC; in particular, applicativiza-
tion never affects the absolutive argument (cf. Letuchiy 2012 on West Circassian).
As shown in the following examples, NWC applicatives in principle combine with
predicates of any valency, so any restrictions on such combinations are motivated by
the degree of complexity allowed by the language:

(18) West Circassian: applicative + monovalent intransitive

fe-ǯʼegʷə-šʼtə-ʁe-x
[3sg.io]ben-play-aux-pst-pl
‘They were playing for him.’

(19) West Circassian: applicative + bivalent intransitive

w-jə-gʷəxeλə–ṣ̂ ʷə qə-b-de-χʷə-n-ew
2sg.io-poss-intention–good csl-2sg.io-com-happen-mod-adv
the-m sə-p-f-j-e-λeʔʷə
god-obl 1sg.abs-2sg.io-ben-[3sg.io]dat-dyn-ask
‘I ask God for you to make your good intentions come true.’

(20) Besleney Kabardian: applicative + bivalent intransitive

sadikə-m mjeste-xe-r ja-xʷə-r-jəq̇ ʷ-q̇ əm
kindergarten-obl place-pl-abs 3pl.io-ben-[3sg.io]loc:inside-be_enough-neg
‘There are not enough vacant places in the kindergarten for them.’

(21) West Circassian: applicative + monotransitive

χʷeχʷə–daxe-xe-r qə-f-a-ʔʷe-šʼtə-ʁe
toast–beautiful-pl-abs csl-[3sg.io]ben-3pl.erg-say-aux-pst
‘They said beautiful toasts for him.’

(22) West Circassian: applicative + ditransitive

se p-f-je-s-ʔʷe-n
1sg 2sg.io-ben-[3sg.io]dat-1sg.erg-say-mod
‘I will tell him (that) instead of (lit. for) you.’

(23) Abaza: applicative + ditransitive

j-ŝə-z-j-á-s-hʷ-ṗ
3sg.n.abs-2pl.io-ben-3sg.m.io-dat-1sg.erg-say-npst.dcl
‘I will tell this to him about you.’
 25 Applicative constructions in the Northwest Caucasian languages 879

In BCs many potential applied objects can be marked either with peripheral cases
(24) and postpositions (25) or with constructions involving subordinate clauses (26).

(24) Abaza (elicited)

a. a-čə́ arq̇ án-la jə-z-ʁə́č’-ṭ
def-horse rope-ins 3sg.n.abs-1sg.erg-steal[aor]-dcl
b. a-čə́ arq̇ án j-á-la-z-ʁəč’-ṭ
def-horse rope 3sg.n.abs-3sg.n.io-instr-1sg.erg-steal[aor]-dcl
Both: ‘I stole the horse by means of a rope.’

(25) West Circassian (elicited)

a. a-xe-m a-dežʼ ḳʷa-ʁe
that-pl-obl 3pl.pp-at go-pst
b. a-xe-m a-fe-ḳʷa-ʁ
that-pl-obl 3pl.io-ben-go-pst
Both: ‘S/he went to their place.’

(26) West Circassian (elicited)

a. tə-w-jə-ʁʷəs-ew tə-šxe-šʼtə-ʁ
1pl.abs-2sg.io-poss-attendant-adv 1pl.abs-eat-aux-pst
əč̣ʼjə t-je-ŝʷe-šʼtə-ʁ
and 1pl.abs-dat-drink-aux-pst
b. tə-qə-b-da-šxe-šʼtə-ʁ əč̣ʼjə
1pl.abs-csl-2sg.io-com-eat-aux-pst and
tə-qə-b-d-je-ŝʷe-šʼtə-ʁ
1pl.abs-csl-2sg.io-com-dat-drink-aux-pst
Both: ‘We were eating and drinking together with you.’

In non-standard speech in Circassian languages, locations sometimes can appear without

any locative markers (27), or with a spatial noun but without any locative preverb on the
predicate (28a), but this is not a norm at all.

(27) West Circassian (elicited)

ǯʼegʷə-xe-m weredə–be q-(a-šʼ)-a-ʔʷe
wedding-pl-obl song–many csl-3pl.io-loc-3pl.erg-say
‘They sing a lot of songs at weddings.’

(28) Kabardian (Bagov (ed.) 1999: 42)

a. zə-gʷerə-m jə-bʁʷə-č̣ʼe wəvə-n
one-some-obl poss-side-ins stand_up-msd
‘to stand up beside someone/something’
880 Peter Arkadiev, Yury Lander, and Irina Bagirokova

b. ṣ̂aɮe–c̣əḳʷə-r jə-ade-m bʁʷə-rə-wəv-a-ŝ

boy–little-abs poss-father-obl [3sg.io]loc:side-trans-stand_up-pst-dcl
‘The little boy stood up beside his father.’

In ACs, applied objects normally behave as core non-absolutive arguments. In Circas-

sian and Ubykh AppPs appear in the oblique case and in Abkhaz-Abaza they remain
unmarked, just as all core arguments.
Curiously, however, some participant expressions may (yet do not need to) retain
their “adjunct” marking even when they are cross-referenced as applied objects. For
instance, in (29) the beneficiary is introduced both by a postposition and an applicative,
while in (30) the accompanier is present both in the adverbial clause and in the matrix
predicate. Such doubling is also reported in locative constructions such as (31), where
‘the shadow of their grandfather’ is virtually introduced by an incorporating spatial
noun and a cognate locative applicative in the predicate. In examples like the latter, it
may also be the case that the whole phrase ‘the bottom of the shadow of their grand-
father’ serves as an AppP of the applicative complex. The range of constructions that
allow such doubling varies across NWC languages.

(29) West Circassian (Lander 2015: 21)

dax-jə ʔeṣ̂ʷə-ʁ-jə zə-p-λeʁʷə-č̣ʼe,
beautiful-add sweet-nmz-add rel.temp-2sg.erg-see-ins
se-šʼ paje s-fe-w-e-ʁe-ṣ̂ʷe-žʼə
1sg-obl for 1sg.io-ben-2sg.erg-dyn-caus-good-re
‘When you see beautiful and sweet things, you keep (them) for me.’

(30) West Circassian

wə-s-jə-ʁʷəs-ew wəne-m
2sg.abs-1sg.pr-poss-attendant-adv house-obl
qə-z-d-jə-h-jə ʁʷeməle-m xa-ʔ
csl-1sg.io-com-loc:inside-go_in[imp]-add meal-obl loc:mass-taste[imp]
‘Come home with me and have a snack.’

(31) Ubykh (Fell 2012: 81)

eʁe-tw(ə)gʼəʒe ʁe-žʼawe-beĉ̣e-n beĉ̣e-χe-ne-n
3pl.pr-grandfather 3sg.pr-shadow-under-obl loc:under-be-prs-pl
‘(sitting) under the shadow of their grandfather’

Little is known about the syntactic differences between ACs and BCs in those cases
when the same content can be expressed by both. For instance, O’Herin (2001: 487)
reports that in Abaza the indirect object introduced by the instrumental applicative
must be definite while no such restriction exists for the independent nominal in the
instrumental case in the BC. However, he himself adds that such a contrast does not
 25 Applicative constructions in the Northwest Caucasian languages 881

exist for the other applicatives; moreover, our own elicited data from Abaza suggest
that there is no systematic difference in definiteness between BC and AC even for the
instrumental applicative.
The only clear difference concerns constructions involving coreference (in a broad
sense, also including any kinds of coindexation), such as reflexive, reciprocal and rel-
ative clause constructions. AppP can undergo reflexivization and reciprocalization by
regular or specialized means available in the individual languages. This is especially
evident in Circassian (and to a certain extent also in Ubykh, see Fenwick 2011: 107),
where the reflexive and reciprocal prefixes simply replace the person-number prefixes
in their canonical position (see e.g. Letuchiy 2007 and Ershova 2019 on West Circassian
and Kazenin 2007 on Kabardian), cf. examples in (32).

(32) West Circassian

a. wə-zə-fe-gʷəbžə-žʼ-a?
2sg.abs-rfl.io-ben-be_angry-re-q
‘Are you angry at yourself?’ (Ershova 2019: 55)
b. te wəne-xe-r ze-fe-t-ṣ̂ə-ʁ
1pl house-pl-abs rec.io-ben-1pl.erg-do-pst
‘We built houses for each other.’ (Letuchiy 2007: 790)

Reflexivization of AppPs in Abaza and Abkhaz is usually achieved by employing the

same regular person-number-gender prefixes (O’Herin 2001: 490–491; Arkadiev and
Durneva 2023), see (33a), or by means of a free reflexive element (the noun ‘head’ with
the possessive prefix coindexed with the agent) corresponding to the AppP (33b). The
latter strategy is also used in Ubykh for transitive verbs, see Fenwick (2011: 82).

(33) Abaza
a. jə-w-zə-w-χʷʕ-əj-ṭ
3sg.n.abs-2sg.m.io-ben-2sg.m.erg-buy-prs-dcl
‘You (m) buy it for yourself.’ (elicited)
b. s-qa a-ĉə-s-χč’a-ṭ
1sg.io-head 3sg.n.io-mal-1sg.erg-protect[aor]-dcl
‘I protected myself / my head from it.’ (Arkadiev and Durneva 2023: 249)

Reciprocalization of AppPs in Abaza is achieved by a non-trivial morphological strategy,

i.e. reduplication of the combination of the applicative complex with the frozen prefix
a- (34).

(34) Abaza (elicited)

a. sawʕá a.z~a.zə́-h-χʷʕa-d
present ben~rec-1pl.erg-buy[aor]-dcl
‘We bought each other presents.’
882 Peter Arkadiev, Yury Lander, and Irina Bagirokova

b. h-a.ĉ~a.ĉ-bah-əj-ṭ
1pl.abs-mal~rec-be_angry-prs-dcl
‘We are angry at each other.’

Finally, relativization of AppPs is generally fulfilled simply by replacing the correspond-

ing indexing prefix with the relative prefix (which in Circassian coincides with the
reflexive prefix; see Lander and Daniel 2019 for discussion), see (35).

(35) Abaza (elicited)

a. a-phʷəspa sə-[l-c]-qraʕa-ṭ
def-girl 1sg.abs-3sg.f.io-com-help[aor]-dcl
‘I helped the girl.’
b. sə-[z-c]-qraʕa-z a-phʷəspa
1sg.abs-rel.io-com-help-pst.nfin def-girl
‘the girl whom I helped’

Now, at least in Circassian, reflexivization, reciprocalization and relativization of

an AppP marked in the predicate are preferred over similar operations with a parallel
element of a BC. For example, relativization of an applied goal as in (36a) is considered
much better than relativization of a goal introduced by a postposition like (36b) (the
latter construction is found in corpora but is much rarer and is considered infelicitous
by many speakers):

(36) West Circassian (elicited)

a. zə-fe-ḳʷa-ʁe-xe-r
rel.io-ben-go-pst-pl-abs
b. zə-dežʼ ḳʷa-ʁe-xe-r
rel.pp-at go-pst-pl-abs
Both: ‘the place where they went’

In fact, there are even cases of relativization of AppPs, where the corresponding BCs do
not exist; see Section 7.2.

4 Semantics
4.1 General information and etymology

NWC languages possess rich systems of applicative prefixes, which can be roughly
divided into grammatical applicatives (benefactive, malefactive, comitative and pos-
sibly some others), locative applicatives and the dative applicative. The latter, which is
 25 Applicative constructions in the Northwest Caucasian languages 883

in a sense even more grammaticalized than so-called grammatical applicatives, differs

from other applicatives in many important respects, so we discuss it in a dedicated
section (Section 5).
The boundary between grammatical (sometimes called “relational”; cf. also the
term “version” used in the works inspired by the tradition of the description of Geor-
gian, e.g. most of the descriptions published during the Soviet period, cf. Lomtatidze
1976) and locative applicatives is, however, fuzzy, since most grammatical applicatives
clearly originate from locative ones. For example, the West Circassian malefactive ṣ̂we-
goes back to the locative preverb with the very specific meaning ‘on the tip of smth.’ (see
Mazurova 2009), the benefactive fe- is sometimes used with reference to the direction of
motion, the comitative de- coincides with one of the basic locative applicatives, and the
instrumental rə- can be treated as the same morpheme as the translative prefix intro-
ducing the path of motion. Hence in many cases the glosses contrasting the grammat-
ical and locative applicatives may reflect a functional contrast rather than distinguish
between different morphemes.
Locative applicatives include a bunch of so-called locative preverbs and their com-
binations, too numerous to be listed here. The systems of locative preverbs (including
polymorphemic ones) range from about two or three dozen in Circassian (see e.g. Smeets
1984: 259–261) to more than a hundred in Abkhaz-Abaza (see e.g. Spruit 1986: 22–31);
however, in the latter only a subset of locative preverbs behave as genuine applicatives
able to introduce arguments and host cross-referencing prefixes (Lomtatidze 1983;
Avidzba 2017).
While the etymology of some of the most grammaticalized applicatives is obscure,
a large part of them shows clear resemblance to postpositions, locational nouns and
body-part terms (see e.g. Kumakhov 1964: 164–182, 1989: 200–228; Lomtatidze 1983;
Avidzba 2017; Arkadiev and Maisak 2018: 121–125). While examples like (37)–(38) from
Abkhaz and Abaza may tempt one to assume that NWC applicatives have arisen via
incorporation of postpositions (a putatively universal scenario proposed by Peterson
2007: 140–141, cf. Fell 2012 on Ubykh, Chirikba 2020: 575 fn. 3 on Abkhaz, or O’Herin
2001 on Abaza), there is evidence that at least some applicatives have rather arisen
from incorporation of nouns while the cognate postpositions may have developed inde-
pendently (see e.g. Arkadiev and Maisak 2018: 125). For instance, many postpositions
feature extra morphology (sometimes fossilized) as compared to the applicatives, and,
still more importantly, a number of applicatives do not have any cognate postpositions
at all.

(37) Abkhaz (Hewitt 1979: 113, transcription adapted)

a. aχra jə-zə jə-q̇ a-s-c̣e-jṭ
A. 3sg.m.io-for 3sg.n.abs-pvb-1sg.erg-do[aor]-dcl
b. aχra jə-zə-q̇ a-s-c̣e-jṭ
A. 3sg.n.abs+3sg.m.io-ben-pvb-1sg.erg-do[aor]-dcl
Both: ‘I did it for Axra.’
884 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(38) Abaza
a. a-warba j-š’aṗ-kʷa rə-ʒqa
def-eagle 3sg.m.pr-foot-pl 3pl.io-near
j-ʕa-ḳa-ŝa-ṭ
3sg.n.abs-csl-loc:down-fall[aor]-dcl
‘The eagle fell at his feet.’ (Tabulova 1976: 278)
b. d-ʕa-hə-ʒqa-jə-r-gəl-ṭ
3sg.h.abs-csl-1pl.io-loc:beside-3sg.erg-caus-stand[aor]-dcl
‘He caused him/her to stand near to us.’ (O’Herin 2001: 481)

The etymological sources of selected applicatives (mainly of locative preverbs) are

given in Table 1, see Kumakhov (1964: 165–182) on Circassian, Dumézil and Esenç (1975:
103–130, 139–144) on Ubykh and Klychev (1994) on Abaza. Chirikba (1996: 380–381)
also points to some putative relations between a number of Circassian, Ubykh and Abk-
haz-Abaza preverbs.

Table 1: Lexical sources of selected NWC applicatives.

language applicative lexical source

West Circassian ʔeč̣ʼe- inadvertitive ʔe ‘hand’ + č̣ʼe ‘bottom’

pe- ‘on the frontal part of the landmark’ pe ‘nose’
λe- ‘following the landmark’ λe ‘foot’
ḳʷec̣ə- ‘inside/through the landmark’ ḳʷec̣ə ‘intestines’
ʔʷə- ‘beside or near the front of the landmark’ ʔʷə ‘mouth, lips’
Ubykh g’ǝ- ‘on the surface of the landmark’ g’ǝ ‘heart’
λeje- ‘under the landmark’ λe ‘foot’
pš’e ‘behind’ pš’e ‘buttocks’
Abaza cǝ- comitative cǝ ‘be together with’
ĉə- malefactive ?ĉa- ‘skin’
čpə-nə- ‘edge of the landmark’ čpə ‘bank, side’
č̣a- ‘slope, frontal surface’ č̣a ‘mouth, face’
ḳʷa- ‘beside the landmark’ ḳʷa- ‘bosom’
qa- ‘above the landmark’ qa ‘head’
š’ta- ‘after the landmark’ š’ta ‘trace’

4.2 Grammatical applicatives

Grammatical applicatives, i.e. applicatives that have regular non-locative functions,

include at least benefactive, malefactive, comitative, and instrumental.
The semantics of the benefactive and malefactive applicatives in West Circassian
has been described by Letuchiy (2009), who identifies the following functions for the
benefactive: benefactive proper (39a), deputative ‘instead of’ (39b), inanimate goal
 25 Applicative constructions in the Northwest Caucasian languages 885

(39c) or animate recipient (39d), purpose (39e), external possessor (39f), stimulus of
feeling or emotion (39g), judicans (person judging) (39h), and content of speech or
thought ‘about’ (39i).

(39) West Circassian

a. a zə-r qə-s-fa-ŝʷ-ṣ̂
dem one-abs csl-1sg.io-ben-2pl.erg-do[imp]
‘Do this one thing for me!’
b. aχšʼe-r q-a-f-jə-gʷešʼə-n-ew je-λeʔʷə-ʁe-x
money-abs csl-3pl.io-ben-3sg.erg-divide-mod-adv dat-ask-pst-pl
‘They asked him to divide the money for them [because they could not agree
how they should do it themselves].’
c. bze–ṣ̂enəʁe-m fe-ʁe-ze-ʁe nəbž’ə-č̣’e-xe-r
language–knowledge-obl ben-caus-turn-res age-new-pl-abs
‘the young people devoted to linguistics’
d. qebar–gʷəxeč̣’ qə-p-fe-t-hə-ʁe-r
news–sorrow csl-2sg.io-ben-1pl.erg-bring-pst-abs
‘We have brought you bad news.’
e. ə-ṣ̂e-re-ba sə-qə-z-fe-ḳʷa-ʁe-r
3sg.erg-know-dyn-emp 1sg.abs-csl-rel.io-ben-go-pst-abs
‘Doesn’t he know why (lit. what for) I have come?’
f. pəj-xe-m s-qʷe s-f-a-wəč̣ʼə-ʁ
enemy-pl-obl 1sg.pr-son 1sg.io-ben-3pl.erg-kill-pst
‘The enemies killed my son.’ (Letuchiy 2009: 344)
g. bw-ew dwənajə-m sə-fe-reze-n-jə!
much-adv world-obl 1sg.abs-ben-be_content-mod-add
‘I would be very content with the world!’
h. ṣ̂enəʁe–deʁʷə əč̣ʼjə ʔepeʔesenəʁe z-jə-ʔe
knowledge–good and talent rel.io-poss-be
ekskursovod-xe-r t-fe-mač̣ʼe-x
guide-pl-abs 1pl.io-ben-little-pl
‘We are short of knowledgeable and talented guides’
i. wečʼepšʼəje, pčʼəhaλ̣əqʷaje, nečʼerezəje, neməč̣ʼ–qʷaǯʼe-xe-m-jə
W. P. N. other–village-pl-obl-add
č̣ʼərʁ-jure wered-xe-r a-f-jə-wəsə-ʁe-x
Ch.-Y. song-pl-abs 3pl.io-ben-3sg.erg-compose-pst-pl
‘Yura Chirg composed songs about Wechepshiy, Pchyhatlukay, Nechereziy
and other villages.’

The range of functions of the benefactive attested in the other NWC languages is similar
to those observed in West Circassian, although details may vary (cf. e.g. Chirikba 2020
on Abkhaz).
886 Peter Arkadiev, Yury Lander, and Irina Bagirokova

The malefactive, according to Letuchiy (2009), has the following functions in West
Circassian: malefactive proper (40a), involuntary agent (40b), and judicans (40c) (the last
is seemingly possible only with nominal predicates; cf. Lander and Bagirokova 2021).

(40) West Circassian

a. nepsə-r qə-s-ṣ̂ʷa-ḳʷe
tear-abs csl-1sg.io-mal+dyn-go
‘Tears appear against my will.’
b. a-šʼ čʼaške-r ṣ̂we-qʷəta-ʁ
dem-obl cup-abs mal-break-pst
‘He accidentally broke a cup.’2 (Letuchiy 2009: 363)
c. a-šʼ fedjəz-ew p-pse p-ṣ̂we-ʔeṣ̂w-me
dem-obl such-adv 2sg.pr-soul 2sg.io-mal-sweet-cond
se wə-qe-z-ʁe-ne-žʼə-n
1sg 2sg.abs-csl-1sg.erg-caus-remain-re-mod
‘If your life is so dear to you, I’ll save you.’

Other uses of the malefactive are also attested in NWC. For example, in Abaza it can
introduce the stimulus of negative emotions:

(41) Abaza
sə-z-ĉ-pχaš’a-wa s-satər-kʷa
1sg.abs-rel.io-mal-be.ashamed-ipf 1sg.io-line-pl
‘my verses, which I am ashamed of’

The comitative is basically monosemous and expresses the co-participant, which can
be either an agent, intransitive (42a) or transitive (42b), or a patient (42c) of the event:

(42) West Circassian

a. apere-me a-də-de-č̣’ə-ʁa-ʁe-x
first-obl.pl 3pl.io-com-loc:enclosure-go_out-pst-pst-pl
‘They left together with those who went first.’
b. a-ʔʷe-re-r a-d-jə-ʔʷa-ʁ,
3pl.erg-say-dyn-abs 3pl.io-com-3sg.erg-say-pst
a-ṣ̂ e-re-r a-d-jə-ṣ̂ a-ʁ
3pl.erg-do-dyn-abs 3pl.io-com-3sg.erg-do-pst
‘He said what they were saying together with them, he did what they were
doing together with them.’

2 Note that in this example the verbal root is labile; compare the dedicated inadvertitive constructions
with agent demotion below.
 25 Applicative constructions in the Northwest Caucasian languages 887

c. fəgʷə-r-jə de-p-hə-n faje

millet-abs-add com-2sg.erg-take-mod must
‘You must also take millet [together with meat]!’

The instrumental applicative introduces the instrument (43a) or means, e.g. lan-
guage (43b).

(43) West Circassian

a. njepe kʷembajn-jə-tfə-m ʔʷef rə-t-e-ṣ̂ e
today harvester-lnk-five-obl work instr-1pl.erg-dyn-do
‘Today we are doing our work with five harvesters.’
b. adəga–bze-m qə-rə-mə-ʔʷe-ŝʷə-n
Circassian–language-obl csl-instr-2sg.erg+neg-say-pot-mod
gʷəpšəse šʼə-ʔ-ep
thought loc-be-neg
‘There is no thought that one cannot express in Circassian.’

In Abkhaz, according to Chirikba (2020: 578), the means of transport can be introduced
alternatively either by the comitative (44a) or by the instrumental (44b) applicatives.

(44) Abkhaz (Chirikba 2020: 578)

a. sará a-mašína s-á-c-aa-jṭ
1sg def-car 1sg.abs-3sg.n.io-com-come[aor]-dcl
b. sará a-mašína s-á-l-aa-jṭ
1sg def-car 1sg.abs-3sg.n.io-instr-come[aor]-dcl
Both: ‘I arrived by car.’

Besides that, Circassian languages have a possessive applicative (homonymous with

one of the frequent locative applicatives, yet different from it in morphonological behav-
ior) which is used with the predicate ‘be’ to introduce the possessor (45). In Circassian the
same applicative appears on nouns in adnominal possessive constructions (in West Circas-
sian and formerly in Besleney Kabardian only in alienable possessive constructions) (46).

(45) West Circassian

w-jane–w-jate-xe-r w-jə-ʔe-x-a?
2sg.pr-mother–2sg.pr-father-pl-abs 2sg.io-poss-be-pl-q
‘Do you have parents?’

(46) West Circassian

w-jə-č’ezəw qe-sə-ʁ, s-jə-ŝaw
2sg.pr-poss-time csl-reach-pst 1sg.pr-poss-son
‘Your time has come, my son.’
888 Peter Arkadiev, Yury Lander, and Irina Bagirokova

Abaza has a dedicated estimative (Jacques 2023) applicative ma- (presumably going
back to the root ‘hold, have’) occurring with nominal stems and introducing a judicans
participant (47) just like the malefactive in Circassian.

(47) Abaza (elicited)

a. arə́j d-gʷəbzəʁa-b
dem 3sg.h.abs-clever-npst.dcl
‘S/he is clever.’
b. arə́j d-sə́-ma-gʷəbzəʁa-b
dem 3sg.h.abs-1sg.io-est-clever-npst.dcl
‘I consider her/him clever.’

Another curious phenomenon observed in Abaza and Abkhaz concerns the recip-
rocal prefix aba- whose basic use is illustrated in (48a). As shown in (48b), the same
prefix can function as a sociative applicative and introduce an indirect object; note that
the base predicate ‘know’ in this example is transitive, while its applicativized version
is bivalent intransitive, obviously in line with the general rule that reciprocals detran-
sitivize predicates (see also Section 7.1), cf. (48a). Such a use of the reciprocal does not
seem to be productive, although is attested with a number of predicates denoting inter-
personal interaction (e.g. ‘be acquainted’, ‘fight’).

(48) Abaza
a. a-sabəj-kʷa j-aba-ʒ̂ʒ̂a-ṭ
def-child-pl 3pl.abs-rec-wash[aor]-dcl
‘The children washed each other.’ (elicited)
b. rqʷəχəʕʷ-ĉa-g’əj h-r-aba-dər-nəs h-ajgʷʁ-əj-ṭ
researcher-hpl-add 1pl.abs-3pl.io-rec-know-purp 1pl.abs-hope-prs-dcl
‘We hope to get acquainted with the researchers.’

In all NWC languages we find irregular and non-predictable lexicalized combina-

tions of grammatical applicatives and roots. Some examples include the use of the mal-
efactive for an experiencer (49a) (which, however, may be related to its judicans use as
in [40c] above) and the benefactive in combination with ‘happen’ for ‘enough’ in West
Circassian (49b), the appearance of a benefactive in combination with the root ‘look’ for
‘wait’ in Abkhaz and Abaza (50), etc.

(49) West Circassian

a. sawəserəqʷe səd ə-ṣ̂e-ʁa-ʁ-ew qə-p-ṣ̂we-ṣ̂ə-re we?
S. what 3sg.erg-do-pst-pst-adv csl-2sg.io-mal-do-dyn 2sg
‘Do you guess what Sosruko did?’
 25 Applicative constructions in the Northwest Caucasian languages 889

b. p-fe-χʷə-me t-je-pλə-n!
2sg.io-ben-happen-cond 1pl.abs-dat-look-mod
‘We will see if that is enough for you.’

(50) Abkhaz (Hewitt 1979: 151)

χ-ɥə-ḳ ḳʷətolaa a-mašina j-a-zə-pš’ə-w-ṗ
three-clh-num Kwitolian def-bus 3pl.abs-3sg.n.io-ben-look-st-npst.dcl
‘Three Kwitolians are waiting for the bus.’

Moreover, one can find bound roots whose semantics cannot be determined without
applicatives: in West Circassian and Kabardian, for example, there are predicates like
fe-je- / xʷ-jej ‘must, want’ (with the benefactive prefix) and jə-je- / jej- (< jə-jej-) ‘belong
to’ (with the possessive prefix), where the meaning of the root cannot be precisely for-
mulated.

4.3 Locative applicatives

Basically, locative applicatives specify the spatial configuration of the event with respect
to the landmark which they introduce as the indirect object, consider examples in (51).

(51) West Circassian

a. maṣ̂we-m pe-t
fire-obl loc:front-stand
‘He is standing in front of the fire.’ (Rogava and Kerasheva 1966: 121)
b. he-r pče-m ʔʷə-λ
dog-abs door-obl loc:near-lie
‘The dog is lying near the door.’ (Rogava and Kerasheva 1966: 126)
c. čər–ɮenəste-r λ-jə-ʒə-ʁ
steel–scissors-abs loc:after-3sg.erg-throw-pst
‘He threw steel scissors after him.’ (Rogava and Kerasheva 1966: 128)

However, in many cases the choice of a particular preverb can also be described as
dependent on the semantic features of the landmark itself (Kerasheva 1957, 1992; Paris
1995). Consider the examples in (52), where the spatial configuration is apparently con-
stant while the preverbs differ with respect to the topological properties of the locative
argument they introduce.

(52) Standard Kabardian (Kumakhov 1964: 165)

a. tjepŝečʼə-m jə-λə-n
plate-obl loc:container-lie-msd
‘to be on a plate’
890 Peter Arkadiev, Yury Lander, and Irina Bagirokova

b. škamṗə-m de-λə-n
cupboard-obl loc:enclosure-lie-msd
‘to be in a cupboard’
c. daʁe-m xe-λə-n
oil-obl loc:mass-lie-msd
‘to be in oil’
d. šxəʔenə-m ḳʷec̣ə-λə-n
blanket-obl loc:inside-lie-msd
‘to be in a blanket’

Locative applicatives are too numerous to be adequately treated here; moreover,

many aspects of their semantics and usage are not yet fully described, despite the con-
tinuing interest of linguists. On the Circassian preverbs, see primarily Kumakhov (1964:
164–182), Kerasheva (1957, 1992) and Adyshesova (1999); on Abaza and Abkhaz, see
Klychev (1994, 1995) and Avidzba (2017); on Ubykh, see Dumézil and Esenç (1975: 103–
130). Some of the locative applicatives have very specific meanings, e.g., West Circassian
qwe- ‘corner’ or Abaza χ’ə- ‘down a vertical surface’. A number of locative applicatives,
especially in Abaza and Abkhaz, can be considered incorporated body-part nouns, the
argument they introduce corresponding to the possessor of the body-part, e.g. (53). In
Circassian this is only possible when a noun is part of a complex preverb (54), even
though many simplex preverbs, as already been said above, are etymologically related
to body-part nouns.

(53) Abkhaz (Spruit 1986: 28)

a-maĉwaz lə-mʁa-s-χə-jṭ
def-ring 3sg.f.io-loc:finger-1sg.erg-take[aor]-dcl
‘I took the ring from her finger.’

(54) West Circassian

wex-wex gʷəšʼəʔe-r pstewə-m-jə q-a-že-de-zə-ʁ
oh-oh word-abs all-obl-add csl-3pl.io-mouth-loc:enclosure-fall-pst
‘Everyone blurted out the word “Oh-oh!”’
(Lit. For all, the word “Oh-oh!” fell from their mouth(s).’)

All languages have a translative preverb which introduces the path of motion (55)–
(56). In Circassian and Abkhaz-Abaza these preverbs also have instrumental functions
described above, while the Ubykh ʁe- lacks it and, moreover, only appears in the trans-
lative meaning following other preverbs, see (8) above (Fenwick 2011: 115). Complex
applicatives containing the translative are also common in Circassian (57).
 25 Applicative constructions in the Northwest Caucasian languages 891

(55) West Circassian

gʷə-r ma-pλe, ne-r λaʁʷe-m r-e-ḳʷe
heart-abs dyn-look eye-abs path-obl trans-dyn-go
‘The heart watches, the eye goes along the path.’ (a proverb)

(56) Abaza (elicited)

aslán dačá–mʕʷa-ḳ d-á-la-ʕa-j-χ-ṭ
A. other–road-indf 3sg.h.abs-3sg.n.io-trans-csl-go-re[aor]-dcl
‘Aslan returned by another road.’

(57) West Circassian (Rogava and Kerasheva 1966: 133)

mezə-m ḳʷec̣ə-rə-č̣ʼə-ʁe-x
forest-obl loc:inside-trans-go_out-pst-pl
‘They went through the forest.’

Circassian and Ubykh also have general locative preverbs which introduce location
without specifying its details (58). Such preverbs can further co-occur with other loca-
tive preverbs following them in an applicative complex (59).

(58) West Circassian

nahə-č̣ʼ-ew anzawər bjerlin šʼ-e-psewə, šʼ-e-ɮažʼe
comp-new-adv A. B. loc-dyn-live loc-dyn-work
‘The younger, Anzaur, lives and works in Berlin.’

(59) Ubykh (Fenwick 2011: 116)

a-č’ə-n a-wawe–mʁ’aṭwe λeje-ʁe-le-sə-χʷe-q̇ e
def-horse-obl def-saddle–strap loc:under-trans-loc-1sg.erg-[caus]pass-pst
‘I passed the saddle strap under the horse.’

Locative (and dative) applicatives may interact with partly grammaticalized roots
conveying the semantics of directional motion and occurring with roots of different
semantic types (see Arkadiev and Maisak 2018: 125–127 on Circassian). For instance,
in Circassian, a fixed combination of the locative preverb de- ‘enclosure’ and the direc-
tional suffix -je creates predicates expressing upwards motion; in (60a) the preverb
introduces the landmark argument, while in (60b) no landmark is apparently implied.

(60) West Circassian

a. ʔʷeŝhe–λage–gʷere-m de-ḳʷe-ja-ʁe-x
hill–high–some-obl loc:enclosure-go-up-pst-pl
‘They climbed some high hill.’
892 Peter Arkadiev, Yury Lander, and Irina Bagirokova

b. č̣əʁeṣ̂wə-xe-m a-wase λeš-ew de-ḳʷe-ja-ʁ

fertilizer-pl-obl 3pl.pr-price strong-adv loc:enclosure-go-up-pst
‘The price of fertilizers has risen considerably.’

Locative applicatives can participate in lexicalized non-compositional root-applica-

tive combinations just like grammatical ones (see e.g. Spruit 1986: 33–34 for Abkhaz),
and sometimes such combinations contain roots that do not occur elsewhere; cf. the
predicate in (61) which is based on a root that is not found outside of this combination:

(61) West Circassian

čʼetəw–ʁase-r-jə dəŝe-m xe-nə-ʁ
cat–trainer-abs-add gold-obl loc:mass-be.deprived-pst
‘The cat trainer was left without gold, too.’

Finally, for Ubykh Fenwick (2011: 113–114) notes a number of locative preverbs which
only combine with a single root, and such preverbs are also attested in Abaza and
Abkhaz as well (for Abaza, see e.g. Klychev 1995).

5 Dative applicatives
All NWC languages can introduce an indirect object by means of a so-called dative
applicative. Dative applicatives follow all other applicative complexes in the prefixal
string and show considerable differences from them in their behavior.
Semantically, dative applicatives are unspecified, i.e. the thematic relation of a
“dative” indirect object is determined by the semantics of the stem. Typical arguments
introduced by means of the dative applicative include the recipient of ‘give’ (62a), the
addressee of ‘say’ (62b), and the causees in causatives derived from transitive stems
(62c).

(62) West Circassian

a. qə-s-e-ŝʷ-t
csl-1sg.io-dat-2pl.erg-give[imp]
‘Give it to me!’
b. zə-gʷere qə-w-e-s-ʔʷe-šʼt
one-some csl-2sg.io-dat-1sg.erg-say-fut
‘I will tell you something.’
c. the-m ʁʷegʷə–nefə-r qə-r-jə-ʁe-λeʁʷə-ʁ
god-obl road–light-abs csl-[3sg.io]dat-3sg.erg-caus-see-pst
‘God has shown him the radiant road.’
 25 Applicative constructions in the Northwest Caucasian languages 893

At least in Circassian languages, dative applicatives are the main means of express-
ing the goal-like participant of ditransitive predicates, as well as of the second argument
of many bivalent intransitive predicates denoting events with low semantic transitivity,
in particular, not implying any salient change of state. These include both physical and
mental activities, speech, and perception, see a representative list of West Circassian
predicates in (63) based on the dictionary Tharkaho (1991: 74–84).

(63) West Circassian

a. physical activity: je-we ‘hit, strike, shoot’, je-ceqe ‘bite’, je-ṭeχʷə ‘scratch’,
je-ʔʷənč̣ʼə ‘push’, je-bewǝ ‘kiss’, je-benǝ ‘wrestle’, je-ṗesḳʷə ‘pinch’, je-ʔe
‘touch’, je-ŝʷe ‘drink’, je-pšʼe ‘blow’;
b. speech: je-λeʔʷə ‘ask (a favor)’, je-wəpč̣ə ‘ask (a question)’, je-χʷenə ‘curse’,
je-gəjə ‘scold’, je-ǯʼe ‘call’, je-pseλe ‘talk to’, je-ʔʷəŝeŝe ‘whisper to’;
c. perception: je-pλə ‘look, watch’, je-deʔʷə ‘listen to’, je-pemə ‘smell’;
d. mental activity: je-gʷəpšəse ‘think about’, je-ǯʼe ‘read, learn’, je-se ‘get used to’.

Formally, dative applicatives manifest manifold peculiarities. First, in Circassian,

they display complex morphophonologically conditioned allomorphy (see e.g. Smeets
1984: 217–226, 264–267), partly shown in the examples above. The basic allomorph of
the dative applicative is je-; however, depending on the morphological context in can
show up as e-, jǝ-, r(ǝ)- and even null, as in (64). In Abkhaz-Abaza the occurrence of
the dative prefix appears to be lexically determined, cf. ‘say’ vs. ‘give’ in (65). Probably
because of all this, the dative complex is sometimes described just as a peculiar expres-
sion of an indirect object coming without an applicative (Smeets 1984: 264; Kumakhov
1971: 266–269, 297–308; Kumakhov and Vamling 2009: 37–40).

(64) West Circassian

xetə qə-s-∅-jə-ʔʷe-n
who csl-1sg.io-dat-3sg.erg-say-pot
‘Who will tell me [how you were killed]?’
(65) Abkhaz
a. aḳ-g’ə s-a-lə-m-hʷa-jṭ
one-add 1sg.io-dat-3sg.f.erg-neg-say[aor]-dcl
‘She said nothing to me.’ (Hewitt 2008a: 311)
b. jǝ-wǝ-s-ta-wa-jṭ
3sg.n.abs-2sg.m.io-1sg.erg-give-ipf-dcl
‘I am giving it to you.’ (Chirikba 2003: 50)

Second, like other applicative complexes, dative complexes can normally be omitted
(66)–(67), but there are exceptions and complications. In particular, dative complexes
introducing causees in causative predicates, like (62c) above, are usually required (68).
894 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(66) West Circassian

qe-s-tə-šʼt-ep zə aχšʼ-jə
csl-1sg.erg-give-fut-neg one money-add
‘I will give no money.’

(67) Kabardian, Besleney dialect (Arkadiev and Letuchiy 2021: 494)

a. ha-r q̇ ə-šʼə-w-e-ʒaq̇ e-č̣ʼe vračə-m=dej ḳʷe
dog-abs csl-temp-2sg.io-dat-bite-ins doctor-obl=to go[imp]
‘If a dog bites you, go to the doctor.’
b. ha-r me-ʒaq̇ e
dog-abs dyn-bite
‘The dog bites.’

(68) West Circassian

✶
q-jə-ʁe-λeʁʷə-ʁ
csl-3sg.erg-caus-see-pst
Intended: ‘S/he has shown.’

Third, in Circassian with many predicates, omission of the dative applicative is

accompanied by the change of the stem-final vowel /ǝ/ into /e/ (69), which is also found
with antipassive derivatives from transitive predicates (70), see Arkadiev and Letuchiy
(2021). This shows that the dative applicative and the argument it introduces in some
way belong to the lexical representation of the predicate.

(69) West Circassian

a. adre-xe-r qe-zə-bgəne-xe-re šʼəʔenəʁe-m je-bgə-x
other-pl-abs csl-rel.erg-leave-pl-dyn life-obl dat-curse-pl
‘Others are cursing the life that leaves them.’
b. ad-č̣ʼ-jə feqʷeλ̣-gʷere ma-bge
there-ins-add peasant-some dyn-curse
‘Some peasant is cursing there as well.’

(70) Kabardian, Kuban dialect (Arkadiev and Letuchiy 2021: 491)

a. se ǯʼane-r z-də-ne
1sg dress-abs 1sg.erg-sew-fut
‘I will sew a dress.’
b. zə-z-ʁe-psexʷ-me jəṭane sə-de-ne
rfl.io-1sg.erg-caus-relax-cond then 1sg.abs-sew.antip-fut
‘I will take a rest and then will do my sewing.’
 25 Applicative constructions in the Northwest Caucasian languages 895

Finally, in Ubykh (Fenwick 2011: 115–116) the dative is the only applicative preverb
that takes cross-reference prefixes from the possessive series rather than from the
regular indirect object one, compare (71a) and (71b).

(71) Ubykh (Fenwick 2011: 115, 110)

a. sə-ʁ-a-ʒʁe-n
1sg.abs-3sg.pr-dat-ask-prs
‘I am asking him.’
b. ze-nejnš wə-n ∅-ǯ’ə-na-ḳ’e-q̇ e
one-young_man-obl 3sg.io-com-3pl.erg[caus]-go-pst
‘They married her (lit. made her go with) a young man.’

Dative complexes also show some peculiarities related to their ordering, see the next
section.

6 Order of applicative complexes

When a predicate contains several applicative complexes, their order, at least partially,
can be described via a default template. In particular, the following template seems to
hold for all NWC languages: grammatical applicative(s) < special locative applica-
tive(s) < dative applicative(s). (72) shows a form containing all these kinds of applica-
tives:

(72) West Circassian

t-jə-wəram asfal’t
1pl.pr-poss-street asphalt
qə-t-fə-tər-a-r-jə-ʁe-λ-ha-ʁ
csl-1pl.io-ben-[3sg.io]loc:on-3pl.io-dat-3sg.erg-caus-lie-lat-pst
‘He made them cover our street with asphalt (lit. put asphalt on our street) for us.’

In Abkhaz and Abaza, there is additional evidence that grammatical and locative applic-
atives occupy distinct slots in the wordform (see O’Herin 2001: 481–482), with the former
farther from the root than the latter and separated from them by directional prefixes:

(73) Abaza
j-[s-zə]-ʕa-[n]-χa-ṭ
3sg.n.abs-1sg.io-ben-csl-loc:inside-remain[aor]-dcl
‘It has remained for me.’
896 Peter Arkadiev, Yury Lander, and Irina Bagirokova

The general locative applicatives found in Circassian and Ubykh behave differently
from special locative applicatives. In West Circassian, according to Lander and Arkhan-
gelskiy (2015), who provided the results of an experimental study of the ordering of the
general locative, benefactive and comitative complexes, these preverbs tend (albeit are
not obliged) to occur in that order (74). As a mirror image of that, for Ubykh Fenwick
(2011: 98) argues that the general locative follows all other preverbs (although he notes
that Charachidzé 1989: 384 proposed a different template, where the general locative
applicative preceded all other applicatives), cf. (8) above and (75).

(74) West Circassian (elicited)

ǯʼane šʼə-fə-d-jə-də-ʁ
dress loc-ben-com-3sg.erg-sew-pst
‘S/he sewed a dress there for him/her together with him/her’

(75) Ubykh (Fenwick 2011: 98)

ŝweλe-g’ə ʁe-beẑe ŝw-χ’e-le-g’ətw-q̇ e-n
2pl-add 3sg.pr-penis 2pl.io-ben-loc-remain-pst-pl
‘His penis remained for you all.’

Indeed, the order of applicative complexes cannot be reduced to any template, as

first noticed by Jakovlev and Ashkhamaf (1941: 103–106). Importantly, at least in Circas-
sian, we also observe scope-based effects in the order of applicatives. For example, the
default association of a situation with some participant may require positioning of its
applicative closer to the stem, probably that is why in (76a) the benefactive applicative
expressing the addressee occurs to the right of the comitative applicative. Similarly, in
causatives, a complex occurring closer to the stem is more likely to be interpreted with
respect to the caused rather than the causing situation (76b). At the same time, if an
applied object is relativized, i.e. presumably has wide scope over other arguments, it is
normal to “move” its complex to the left of other complexes (76c).

(76) West Circassian (elicited)

a. mwe s-jə-č̣ʼaɮe pis’me-r a-xe-me
dem 1sg.io-poss-boy letter-abs dem-pl-obl.pl
[a-də]-[fe]-s-txə-š’tə-ʁ / ✶[f]-[a-de]-s-txə-š’tə-ʁ
3pl.io-com-[3sg.io]ben-1sg.erg-write-aux-pst [3sg.io]ben-3pl.io-com. . .
‘I was writing a letter to my son together with them.’
b. sportsm’en-xe-r spartakiade-m
athlete-pl-abs competition-obl
[š’ə]-[ze-de]-d-ʁe-bena-ʁ
[3sg.io]loc-rec.io-com-1pl.erg-caus-compete-pst
‘We made the athletes participate in the competition together.’
✶
‘We together made the athletes participate in the competition.’
 25 Applicative constructions in the Northwest Caucasian languages 897

c. mwe pis’me-r [zə-f]-[a-de]-p-txə-š’tə-ʁe-r

dem letter-abs rel.io-ben-3pl.io-com-2sg.erg-write-aux-pst-abs
‘The one whom you were writing this letter together with them’

Caponigro and Polinsky (2011: 80–81) also reported that the order of the complexes
in West Circassian may correspond to the scope of the quantifiers, so that the appear-
ance of an applicative complex farther from the root implies its broader scope. In (77)
the beneficiary phrase has scope over the comitative phrase (i.e. only the interpretation
‘There is a girl for whom he made it with all the boys’ is preferred with respect to the
interpretation ‘For all the boys with whom he made it there is some girl for whom this
was done’). However, our consultants only partly confirm this.

(77) West Circassian (Caponigro and Polinsky 2011: 81)

zeč̣ʼe-m-jə č̣ʼaɮe-xe-m zə pŝaŝe-m [fə]-[ra-d]-jə-ṣ̂ə-ʁ
all-obl-add boy-pl-obl one girl-obl [3sg.io]ben-3pl.io-com-3sg.erg-do-pst
‘He made it for one girl with all the boys.’ (one > all, ✶all > one)

Finally, it is worth noting that at least in West Circassian the order of multiple
dative complexes (possible when one of them introduces the causee in a causative
construction involving a ditransitive stem) may depend on various factors including
a person/number hierarchy (cf. Bagirokova, Lander, and Moroz 2017). Normally, the
dative complex expressing the causee follows the dative indirect object belonging to
the caused situation (78a), but if the latter is higher than the causee in the hierarchy 1sg
> 2sg > 1pl > 2pl (presumably combined from the hierarchies 1 > 2 and sg > pl) both
orders are possible, as shown by the ambiguity in (78b):

(78) West Circassian (elicited)

a. qə-w-a-r-jə-ʁe-ʔʷa-ʁ
csl-2sg.io[dat]-3pl.io-dat-3sg.erg-caus-say-pst
‘s/he made them tell you that’ / ✶‘s/he made you tell them that’
b. qə-w-ə-s-jə-ʁe-ʔʷa-ʁ
csl-2sg.io-dat-1sg.io[dat]-3sg.erg-caus-say-pst
‘s/he made me tell you that’ / ‘s/he made you tell me that’

7 Special uses of applicatives

7.1 A-demotion

Probably the least expected function of applicatives, given their basic function of pro-
motion of arguments, is the demotion of ergative arguments to indirect objects. This is
898 Peter Arkadiev, Yury Lander, and Irina Bagirokova

found mainly in potential constructions expressing ability, where the potential erga-
tive argument is introduced via the benefactive applicative complex (79), and in inad-
vertitive constructions expressing that the expected ergative argument behaves as an
accidental causer of the event and hence appears as an indirect object (80), but also
probably in some reciprocal constructions (see below). Note that in most typical cases
such indirect objects retain some properties of the transitive agent – e.g., the use of such
constructions is almost restricted to transitive stems (i.e. stems that otherwise require
an ergative agent) and, unlike prototypical applicative complexes, such expressions of
the agent cannot be omitted. This goes against the idea that such constructions involve
agentless (i.e. intransitive) stems which combine with applicatives introducing a com-
pletely distinct semantic role (see Lander 2022 for discussion).

(79) West Circassian (Letuchiy 2009: 355)

a. se harəfə-xe-r s-e-txə
1sg character-pl-abs 1sg.erg-dyn-write
‘I am writing characters.’
b. se harəfə-xe-r s-fe-txə-r-ep
1sg character-pl-abs 1sg.io-ben-write-dyn-neg
‘I cannot write characters.’

(80) West Circassian (Arkadiev and Letuchiy 2011: 503–504)

a. se s-jə-š’eweʁʷə-r sə-wəʔa-ʁ
1sg 1sg.pr-poss-friend-abs 1sg.erg-wound-pst
‘I wounded my friend’
b. se səməṣ̂axew s-jə-š’eweʁʷə-r s-ʔeč̣’e-wəʔa-ʁ
1sg unintentionally 1sg.pr-poss-friend-abs 1sg.io-inadv-wound-pst
‘I unintentionally wounded my friend’

In Circassian, the potential use of the benefactive is only available for transitive
predicates, cf. an ungrammatical example based on an intransitive predicate in (81), see
also Letuchiy (2012: 336–339). In general, it is also impossible to attach the inadvertitive
to an intransitive predicate adding to it an involuntary agent or cause; however, the
borderline between such putative constructions and the use of the same preverb in its
etymological meaning ‘from under hand’ attested in Circassian and Ubykh is sometimes
fuzzy, as shown in (82).

(81) West Circassian (Letuchiy 2009: 358)

✶
č̣ʼaɮe-r / ✶č̣ʼaɮe-m fa-ḳʷe-r-ep
boy-abs boy-obl ben-go-dyn-neg
Ungrammatical in the intended meaning: ‘The boy cannot go.’
 25 Applicative constructions in the Northwest Caucasian languages 899

(82) West Circassian

a. senefə–bẑe-r qə-ʔe-č̣ʼ-e-zə
wine–horn-abs csl-hand-loc:under-dyn-fall
‘The horn with wine falls from his hand.’ (Rogava and Kerasheva 1966: 130)
b. pšʼə-m sə-ʔe-č̣ʼe-ḳʷede-n
prince-obl 1sg.abs-hand-loc:under-vanish.lat-mod
‘I’ll perish from the hand of the prince.’ (Rogava and Kerasheva 1966: 282)

The situation in Abkhaz-Abaza and Ubykh is different. Here the potential and inad-
vertitive markers appear both with transitive (83a), (84b) (usually behaving similarly
to their Circassian counterparts, but see below) and intransitive (83b), (84a) predicates
(see Hewitt 2008b inter alia). In the latter case, the most agentive absolutive argument
retains its syntactic status while the potential and inadvertitive markers do not function
as applicatives introducing any indirect object anymore.

(83) Abaza
a. knigá g’-s-zə́-m-χʷʕ-əw-z-ṭ
book neg-1sg.io-pot-neg-buy-ipf-pst.nfin-dcl
‘I could not buy books.’
b. jará d-g’ə-z-ʕá-mə-j-ṭ
3sg.m 3sg.h.abs-neg.emp-pot-csl-neg-come[aor]-dcl
‘He could not come himself.’

(84) Ubykh (Fenwick 2011: 114)

a. q̇eʁe-š weč’e-q̇ e
inadv-laugh-pst
‘He could not help but burst out laughing.’
b. jə-χ’ə́-n ze-tə́t-gʷere q̇eʁe-ḳʷ-q̇ e
dem-prince-obl one-man-certain inadv-kill-pst
‘This prince accidentally killed a man.’

Curiously, Hewitt (1999) reports that in Abkhaz, in potential and inadvertitive

forms derived from transitive stems, the ergative indexing can even be retained on a
par (85a) or instead of (85b) the indirect object indexing.

(85) Abkhaz (Hewitt 1999: 201)

a. j-a-z-a-m-ga-jṭ
3sg.n.abs-3sg.n.io-pot-3sg.n.erg-neg-take[aor]-dcl
‘It could not take it.’
b. a-ʁba f-ɥə-ḳ z-a-m-ga-jṭ
def-boat five-clh-num pot-3sg.n.erg-neg-take[aor]-dcl
‘The boat could not carry five persons.’
900 Peter Arkadiev, Yury Lander, and Irina Bagirokova

Finally, according to one of the interpretations (see e.g. Lander and Letuchiy 2010:
270), reciprocal constructions coindexing the absolutive and ergative arguments are
basically formed by demoting the ergative argument by means of some applicative
prefix (in Circassian probably related to the instrumental applicative) and replacing
the corresponding indexing prefix with a reciprocal morpheme. This interpretation,
which is illustrated by glosses in (86a), explains the typologically unusual binding of the
agent by the patient (under such an account treated as binding of an indirect object by
the absolutive argument) as well as some morphophonological facts not to be discussed
here, but perhaps implies a violation of the rule stating that the dative applicative
follows all others (see Section 5). Moreover, some speakers of West Circassian margin-
ally allow even a combination of this “reciprocal” applicative affix with the inadverti-
tive applicative (86b). Yet the standard description presented, for example, in Letuchiy
(2007) treats the sequences such as ze-re- in (86a) as single reciprocal prefixes, and it
cannot be excluded that examples like (86b) result from morphological reanalysis of
reciprocal markers as applicative complexes.

(86) West Circassian

a. ade təde tə-šʼə-ze-re-ʁʷetə-žʼə-šʼt?
but where 1pl.abs-loc-rec.io-rec-find-re-fut
‘But where will we find each other again?’
b. tə-ze-ʔeč̣ʼe-re-wəʔa-ʁe-x
1pl.abs-rec.io-inadv-rec-wound-pst-pl
‘We wounded each other accidentally.’ (elicited; Lander and Letuchiy 2010:
270)

7.2 Last resort applicative relativization in Circassian

As mentioned in Section 3.2, applicatives may facilitate relativization. Besides that, in

Circassian there are cases where the appearance of an applicative is possible only if the
corresponding argument is relativized, e.g. in embedded clauses referring to place, time
or reason. For example, the destination with the predicate ‘go’ is not normally introduced
via an applicative (87a), but relativization of this participant requires an applicative (87b):

(87) West Circassian (elicited)

a. a-šʼ sə(✶-de)-ḳʷa-ʁ
that-obl 1pl.abs-loc:enclosure-go-pst
‘I went there.’
b. sə-zə-de-ḳʷa-ʁe-r
1sg.abs-rel.io-loc:enclosure-go-pst-abs
‘(the place) where I went’
 25 Applicative constructions in the Northwest Caucasian languages 901

In other cases, an argument cannot be expressed in independent clauses at all but

can be relativized after applicativization. For example, reason cannot appear as an
argument in independent clauses (88a) but appears as a relativized applied indirect
object in relative clauses (88b). The applicative morpheme introducing reason coincides
with the locative preverb ‘under’.

(88) West Circassian (elicited, Lander 2012: 290)

a. we–dejə-m wəɮape sə-ḳʷa-ʁ-ep / ✶sə-č̣’e-ḳʷa-ʁ-ep
weather–bad-obl U. 1sg.abs-go-pst 1sg.abs-rsn-go-pst-neg
‘I did not go to Ulyap due to bad weather.’
b. [wəɮape sə-(z)-č̣’e-mə-ḳʷa-ʁe we-r]
U. 1sg.abs-rel.io-rsn-neg-go-pst weather-abs
qə-ze-č̣’e-č̣’ə-ž’ə-ʁ
csl-rec.io-loc:under-go_out-re-pst
‘The weather, due to which I did not go to Ulyap, improved.’

Interestingly, since an applicative occurs only if the applied object is relativized, the
appearance of the relative prefix turns out to be optional. The subsequent development
where a (former?) applicative becomes the sole marker of relativization is observed
in Kabardian constructions with relativization of time. In both Circassian languages
relativization of time can exploit the general locative applicative, but in Kabardian it is
regularly used as the only marker of subordinate temporal clauses (89):

(89) Kabardian, Besleney dialect

bzəλxʷəʁe-r š’ə-ʔʷə-č̣’ə-ž’ə-m psəne-m jə-h-a
woman-abs temp-loc:near-go_out-re-obl well-obl loc:inside-go_in-pst
pŝeχʷə-m-č̣’e
chain-obl-ins
‘When the woman went away, he used the chain to get into the well.’

A different path of development which retains the relative prefix is found in Cir-
cassian subordinate clauses describing the manner (90) and the fact of the event (91).
Such clauses display properties of relativization but contain a dedicated marker (zere-
in West Circassian, zerə- in Kabardian). At least diachronically but probably synchron-
ically as well this marker can be analyzed as a sequence of the relative prefix and an
applicative introducing it and presumably related to the instrumental applicative (for
a discussion, see Bizhoev 1991: 89–91; Gerasimov and Lander 2008; Arkadiev and Ger-
asimov 2019):
902 Peter Arkadiev, Yury Lander, and Irina Bagirokova

(90) West Circassian

muzəke-r qə-b-gʷə-rə-ʔʷe-nə-m feṣ̂
music-abs csl-2sg.io-heart-instr-say-mod-obl for
maqe-xe-r zere-txə-ʁe-xe-m wə-q-je-ǯʼe-n (. . .) faje
sound-pl-abs rel.mnr-write-res-pl-obl 2sg.abs-csl-dat-read-mod must
‘To understand music, you should (be able to) read how sounds are written down.’

(91) West Circassian

s-jə-ŝeweʁʷ ɮažʼe ze.r-jə-mə-ʔe-r d-ʁe-wənefə-ʁe
1sg.pr-poss-friend fault rel.fact-poss-neg-be-abs 1pl.erg-caus-turn_out-pst
‘We found out that it was not my friend’s fault (lit. that my friend has no fault).’

In Abkhaz-Abaza, subordinate clauses parallel to the ones described in this section also
have syntactic properties of relative clauses, but there is no morphological evidence
that their markers can be treated as applicatives synchronically or diachronically.

7.3 Neutralized applicatives

In some examples, we observe frozen applicatives whose combinations with the root are
lexicalized to the extent that they appear together with the root in a stem rather than in
the argument structure zone. Such applicatives do not introduce any indirect objects.
An example is presented by the Circassian monovalent intransitive predicate ‘search’,
whose root combines with the “former” locative applicative λə- ‘after’ (as shown in
[92a], where it follows the negative prefix and hence belongs to the stem). Interestingly,
Circassian languages also have an intransitive predicate ‘search’, where the argument
being searched is introduced by the same applicative as a genuine indirect object (92b).

(92) West Circassian

a. bere mə-λə.χʷa-ʁ-ew ʔeχʷeʁʷə-r q-ə-ʁʷetə-žʼə-ʁ
long neg-search-pst-adv herd-abs csl-3sg.erg-find-re-pst
‘He found the herd without a long search (lit. not having searched for a long
time).’
b. təʁʷaḳʷe-m bere λə-mə-χʷə-x-ew q-a-ʁʷetə-ʁ
thief-obl long loc:after-neg-search-pl-adv csl-3pl.erg-find-pst
‘They found the thief without searching him for a long time.’

In fact, Gishev (1983: 109) also provides some other examples where special locative
preverbs no longer appear in the argument zone but come together with the root.
However, Circassian languages even show an example where a whole complex involv-
ing a dative applicative does not change the argument structure but rather fulfills a
derivational function. In (93) the combination of the reciprocal suffix introduced by
 25 Applicative constructions in the Northwest Caucasian languages 903

the dative applicative and the root ‘hit’ is interpreted as ‘fight’, but the corresponding
patient-like argument may be introduced by a further dative applicative:

(93) West Circassian

je-[z-e-wa]-ʁ-a, ə-wəč̣ʼə-ʁ-a?
dat-rec.io-dat-hit-pst-q 3sg.erg-kill-pst-q
‘Did he fight with him? Did he kill him?’

A similar case is found in Abaza. Here the prefix a(j)- (regularly used as reciprocal)
combining with the root ‘hit’ occupies the slot immediately preceding the root rather
than the regular slot in the middle of the prefixal chain; this is shown by the fact that it
can be separated from the cross-referencing prefix it introduces by other material such
as negation, see (94).

(94) Abaza (elicited)

w-ʕa-s-m-áj-sə-n
2sg.m.abs-csl-1sg.io-neg-appl-hit-neg.imp
‘Don’t beat me!’

While such frozen applicatives usually do not introduce any arguments anymore, there
is no evidence that they have been reanalyzed as parts of the root, either.

8 Lookalike: transitivizing ablaut

In Circassian languages, there are applicative-like constructions not fitting into the
canonical picture described above. They mainly concern intransitive predicates of
motion like ḳʷe ‘go’, which can be transitivized by changing the final vowel /e/ into /ǝ/,
whereby the absolutive argument of the BC denoting the moving entity becomes the
ergative A, while the new absolutive object denotes the path or distance covered by
motion, compare (95a) and (95b).

(95) West Circassian

a. a-r ḳʷa-ʁe
dem-abs go-pst
‘S/he went.’ (elicited)
b. a-šʼ ʁʷegʷə–be ə-ḳʷə-ʁ
dem-obl way–many[abs] 3sg.erg-go.tr-pst
‘He has traveled many roads.’
904 Peter Arkadiev, Yury Lander, and Irina Bagirokova

The transitive versions of such predicates are systematically used in constructions

denoting circular or perambulatory motion, which employ directional suffixes. Such
predicates can be used both with an absolutive argument denoting the spatial region
encircled or covered by motion (96a), as well as without any referential second argu-
ment (96b) or with such an argument introduced as a locative AppP rather than an
absolutive (96c).

(96) West Circassian

a. t-jə-gʷap-ew jeǯʼaṗe-r qe-t-ḳʷə-ha-ʁ
1pl.pr-poss-pleasure-adv school-abs csl-1pl.erg-go.tr-circum-pst
‘We visited (lit. went over) the school with pleasure.’
b. mefe–rjenə-m qe-s-ḳʷə-ha-ʁe-m-jə
day–whole-obl csl-1sg.erg-go.tr-circum-pst-cond-add
šə-xe-m-re zeweλ̣-xe-m-re a-neməč̣ʼ
horse-pl-obl-coord soldier-pl-obl-coord 3pl.io-except
s-λeʁʷə-r-ep
1sg.erg-see-dyn-neg
‘Even though I have walked around / traveled for the whole day, I don’t
see anybody but horses and soldiers.’
c. məjeqʷape jə-wəram-xe-m q-a-šʼ-a-ḳʷə-ha-ʁ
M. poss-street-pl-obl csl-3pl.io-loc-3pl.erg-go.tr-circum-pst
‘They walked around the streets of Maykop.’

This use of ablaut for transitivization of basic intransitive predicates (for more on
the /e/ ~ /ǝ/ ablaut in Circassian, see Kumakhov 1974; Kumakhov and Vamling 2009:
34–35) is a mirror-image of the more productive antipassive formation mentioned
above in Section 5, which changes the root-final /ǝ/ into /e/. Cf. (97) and Arkadiev and
Letuchiy (2021) for more details and a discussion of the problematic directionality of
these derivations.

(97) West Circassian

a. haləʁʷ-jə-blə-r se s-šxə-ʁe
bread-lnk-seven-abs 1sg 1sg.erg-eat-pst
‘I ate seven pieces of bread.’
b. a-xe-r ma-šxe-xe-me, te-rjə t-e-šxe
dem-pl-abs dyn-eat.antip-pl-cond 1pl-add 1pl.abs-dyn-eat.antip
‘If they eat, we eat, too.’

The same pattern of ablaut has another productive use only indirectly related
to valency and more clearly equipollent than the ones manipulating transitivity, i.e.
the so-called introvert (lative) and extravert (elative) forms of verbs denoting real or
 25 Applicative constructions in the Northwest Caucasian languages 905

metaphorical motion and always requiring a locative applicative (see e.g. Smeets 1984:
442–445; Arkadiev and Letuchiy 2011: 500), shown in examples (98).

(98) West Circassian (Smeets 1984: 442)

a. məẑʷe-r tje-sə-ʒe-šʼt
stone-abs loc:top-1sg.erg-throw.lat-fut
‘I will throw the stone on it.’
b. məẑʷe-r tje-sə-ʒǝ-šʼt
stone-abs loc:top-1sg.erg-throw.elat-fut
‘I will throw the stone from it.’

9 Conclusions
In this chapter we have offered a necessarily incomplete survey of the rich system of
applicatives in the Northwest Caucasian languages, focusing primarily on West Circas-
sian. According to the questionnaire proposed as a guideline for this volume’s contribu-
tions, the constructions presented can be characterized as follows:

Morphology
– All NWC applicatives are prefixes occurring in the argument structure zone of the
prefixal chain.
– Canonically, applicative prefixes are immediately preceded by person-number(-gen-
der) prefixes indexing the applicativized participant. Deviations from this pattern
include prefix displacement in Circassian, absence of third person singular non-hu-
man prefixes with some applicatives in Abkhaz and Abaza, and instances of complex
applicative prefixes introducing the same argument.
– There are special “dative” applicatives that show idiosyncratic allomorphy and
occupy a dedicated slot in the verbal template closest to the root.
– Applicativized predicates do not show any morphological idiosyncrasies in their
inflection.

Syntax
– Abaza, Abkhaz and Ubykh applicative constructions are D-applicatives introducing
indirect objects encoded like the recipient of ‘give’; applicative constructions in Cir-
cassian, however, introduce arguments which have no parallels with non-derived
verbs, for the simple reason that even the recipient of ‘give’ in these languages is
introduced by an applicative.
– Applicatives combine with both intransitive and transitive base predicates.
906 Peter Arkadiev, Yury Lander, and Irina Bagirokova

– In general, applicativization is optional in that in many cases the participant

expressed by an AppP can be encoded by alternative means.
– Apart from the potential, inadvertitive and (arguably) detransitivizing reciprocal
applicatives that reassign the original transitive agent from the ergative slot to that
of the indirect object, applicativization does not in any way affect the encoding and
syntactic status of the core arguments (S, A and P).
– With the same exception, applicativization is valency-increasing.
– Applicativization does not show restrictions in combination with such valency-
changing operations as causativization, reflexivization and reciprocalization; in
Circassian, bivalent intransitive predicates whose patient-like argument is intro-
duced by the dative applicative can undergo antipassivization eliminating that par-
ticipant.

Semantics
– The applied phrase bears such semantic relations as beneficiary, external pos-
sessor, maleficiary, co-participant, instrument, means, path, location, and a large
number of more concrete spatial meanings. The semantic roles of the AppPs intro-
duced by the dative applicatives include ditransitive recipients, causees of causa-
tive constructions based on transitive predicates, and low-affected non-agentive
participants of verbs of impact, speech, perception, and cognition.
– The special uses of applicatives in relative clause constructions include the expres-
sion of such meanings as location, time, reason, manner and, by extension, fact.

Lookalikes
A potential lookalike involves an unproductive transitivizing derivation attested with
a few verbs of motion and introducing the P participant expressing path or distance.

Abbreviations
abs absolutive
add additive
adv adverbial marker
antip antipassive
aor aorist
appl applicative
aux auxiliary marker
ben benefactive
caus causative
circum circumferential ‘around’
clh human classifier
com comitative
 25 Applicative constructions in the Northwest Caucasian languages 907

comp comparative
cond conditional
coord coordination marker
csl cislocative
cvb converb
dat dative applicative
dcl declarative
def definite
dem demonstrative
dyn dynamic
elat elative
emp emphatic
erg ergative
est estimative
f feminine
fact factual
fut future
h human
hpl human plural
imp imperative
inadv inadvertitive
indf indefinite
ins instrumental case
instr instrumental applicative
io indirect object
ipf imperfective
lat lative
lnk linking morpheme
loc locative preverb
m masculine
mal malefactive
mnr manner
mod modal marker
msd masdar
n nonhuman
neg negation
nfin nonfinite
nmz nominalization
npst nonpast
num numeral marker
obl oblique case
pl plural
poss possessive
pot potential
pp postpositional series of personal prefixes
pr possessor
prs present
pst past
purp purposive
908 Peter Arkadiev, Yury Lander, and Irina Bagirokova

pvb preverb
q interrogative
re refactive
rec reciprocal
rel relativization
res resultative
rfl reflexive
rs retrospective shift
rsn reason
sg singular
st stative
temp temporal
tr transitive
trans translative
up motion upwards
- default morpheme boundary
– productive nominal compound boundary
= clitic boundary

Sources
Arkadiev, Peter, Irina Bagirokova, Anna Sorokina & Elena Sokur. 2020. Corpus of oral texts in Besleney
Kabardian. Moscow: Linguistic Convergence Laboratory, HSE University. https://lingconlab.ru/spoken_
besleney (accessed 12 April 2022).
Arkhangelskiy, Timofey, Irina Bagirokova, Anna Lander & Yury Lander. 2018–2023. West Circassian (Adyghe)
Corpus. http://adyghe.web-corpora.net/ (accessed 7 September 2023).
Bagirokova, Irina, Yury Lander, Anna Sorokina & Elena Sokur. 2020. Corpus of oral texts in West Circassian
(Temirgoi dialect). Moscow: Linguistic Convergence Laboratory, HSE University. https://lingconlab.ru/
spoken_adyghe/ (accessed 12 April 2022).
Panova, Anastasia, Anna Sorokina, Peter Arkadiev & Elena Sokur. 2019. Spoken corpus of Abaza. Moscow:
School of Linguistics, HSE University; Linguistic Convergence Laboratory, HSE University. http://
lingconlab.ru/spoken_abaza/ (accessed 12 April 2022).

References
Adyshesova, Marina R. 1999. Lokativnye preverby v kabardino-čerkesskom jazyke [Locative preverbs in
Kabardian]. Maykop: Adyghe State University dissertation.
Applebaum, Ayla Ayda Bozkurt. 2013. Prosody and grammar in Kabardian. Santa Barbara: University of
California dissertation.
Arkadiev, Peter & Sonia Durneva. 2023. Reflexive constructions in Abaza. In Martin Haspelmath, Katarzyna
Janic & Nicoletta Puddu (eds.), Reflexive constructions in the world’s languages, 233–257. Berlin:
Language Science Press.
 25 Applicative constructions in the Northwest Caucasian languages 909

Arkadiev, Peter & Dmitry Gerasimov. 2019. Ot reljativizacii k aspektu: konstrukcii s prefiksom zere- v
adygskix jazykax [From relativisation to aspect: Constructions with the prefix zere- in Circassian
languages]. In Dmitry Gerasimov, Sergej Dmitrenko & Natalia Zaika (eds.), Sbornik statej k 85-letiju V. S.
Xrakovskogo [A festschrift for Viktor Xrakovskij on the occasion of his 85th birthday], 11–42. Moscow:
LRC Publishing.
Arkadiev, Peter & Yury Lander. 2020. The Northwest Caucasian languages. In Maria Polinsky (ed.), The
Oxford handbook of the languages of the Caucasus, 369–446. Oxford: Oxford University Press.
Arkadiev, Peter & Alexander Letuchiy. 2011. Prefixes and suffixes in the Adyghe polysynthetic wordform:
Types of interaction. In Vittorio Springfield Tomelleri, Manana Topadze & Anna Lukianowicz (eds.),
Languages and cultures in the Caucasus, 495–514. Munich & Berlin: Otto Sagner.
Arkadiev, Peter & Alexander Letuchiy. 2021. Indirect antipassive in Circassian. In Katarzyna Janic & Alena
Witzlack-Makarevich (eds.), Antipassive: Typology, diachrony, and related constructions, 483–514.
Amsterdam & Philadelphia: John Benjamins.
Arkadiev, Peter & Timur Maisak. 2018. Grammaticalization in the North Caucasian languages. In Heiko
Narrog & Bernd Heine (eds.), Grammaticalization from a typological perspective, 116–145. Oxford: Oxford
University Press.
Arkadiev, Peter & Yakov Testelets. 2019. Differential nominal marking in Circassian. Studies in Language
43(3). 715–751.
Avidzba, Asmat V. 2017. Lokalʹnye preverby v abxazskom i abazinskom jazykax [Locative preverbs in
Abkhaz and Abaza]. Sukhum: The D. I. Gulia Abkhazian Institute for Research in the Humanities
dissertation.
Bagirokova, Irina G., Yury A. Lander & Georgi A. Moroz. 2017. O vyraženii množestvennyx neprjamyx
objektov v adygejskom glagole [On the expression of multiple indirect objects in the West Circassian
verb]. In Stanovlenie i razvitie mladopis’mennyx jazykov. K 120-letiju so dnja roždenija vydajuščegosja
jazykoveda, osnovopoložnika adygejskogo jazykoznanija D.A. Ašxamafa: materialy meždunarodnoj naučnoj
konferencii (Majkop, 21–23 ijunja 2017 g.) [Formation and development of languages with recent writing
traditions. On the occasion of the 120th anniversary of the great linguist and the founder of the
Circassian linguistics D. A. Ashkhamaf: proceedings of the international conference (Maykop, 21–23
June 2017)], 23–27. Maykop.
Bagov, P. M. (ed.). 1999. Adəgebze psaλaλe. Slovar’ kabardino-čerkesskogo jazyka [A dictionary of Kabardian].
Moscow: Digora.
Bizhoev, Boris Ch. 1991. Pričastie v adygskix jazykax v sravnitel’nom osveščenii [Participle in Circassian
languages in a comparative perspective]. Nalchik: Nart.
Caponigro, Ivano & Maria Polinsky. 2011. Relative embeddings: A Circassian puzzle for the syntax/semantics
interface. Natural Language & Linguistic Theory 29(1). 71–122.
Charachidzé, Georges. 1989. Ubykh. In B. George Hewitt (ed.), The indigenous languages of the Caucasus.
Vol. 2: The Northwest Caucasian languages, 357–459. Delmar, New York: Caravan Books.
Chirikba, Viacheslav A. 1996. Common West Caucasian: The reconstruction of its phonological system and parts
of its lexicon and morphology. Leiden: CNWS.
Chirikba, Viacheslav A. 2020. Applikativy v abxazskom jazyke [Applicatives in Abkhaz]. In Andrej Kibrik,
Ksenia Semenova, Dmitry Sitchinava, Sergei Tatevosov & Anna Urmanchieva (eds.), VAProsy
jazykoznanija: Megasbornik nanostatej k jubileju V. A. Plungjana [VAProsy jazykoznanija: A megacollection
of nanoarticles dedicated to Vladimir Plungian’s birthday], 574–580. Moscow: Buki-Vedi.
Colarusso, John. 2006. Kabardian (East Circassian). Munich: LINCOM.
Dumézil, Georges & Tevfik Esenç. 1975. Le verbe oubykh. Études descriptives et comparatives. Paris:
Klincksieck.
Ershova, Ksenia. 2019. Syntactic ergativity in West Circassian. Chicago: University of Chicago dissertation.
Fell, Brian. 2012. Applicatives and incorporation in Ubykh. SKY Journal of Linguistics 25. 61–92
Fenwick, Rohan S. H. 2011. A Grammar of Ubykh. Munich: LINCOM.
910 Peter Arkadiev, Yury Lander, and Irina Bagirokova

Gerasimov, Dmitry V. & Yury A. Lander. 2008. Reljativizacija pod maskoj nominalizacii i faktivnyj argument
v adygejskom jazyke [Relativization in the guise of nominalization and the fact argument in West
Circassian]. In Vladimir A. Plungian & Sergei G. Tatevosov (eds.), Issledovanija po glagol’noj [Studies in
verbal derivation], 290–313. Moscow: Jazyki slavjanskoj kul’tury.
Gishev, Nukh T. 1983. Nekotorye voprosy glagol’noj affiksacii v adygejskom jazyke [Some issues of verbal
affixation in West Circassian]. In Nurja T. Tabulova & Raisa X. Temirova (eds.), Sistema preverbov i
poslelogov v iberijsko-kavkazskix jazykax [The system of preverbs and postpositions in Ibero-Caucasian
languages], 55–111. Cherkessk: Karačaevo-čerkesskij NII istorii, filologii i èkonomiki.
Hewitt, B. George. 1979. Abkhaz. Amsterdam: North-Holland.
Hewitt, B. George 1999. Morphology revisited: Some peculiarities of the Abkhaz verb. In Helma van den
Berg (ed.), Studies in Caucasian linguistics, 197–208. Leiden: CNWS.
Hewitt, B. George 2005. North West Caucasian. Lingua 115. 91–145.
Hewitt, B. George 2008a. Are verbs always what they seem to be? Iran and the Caucasus 12(2). 307–323.
Hewitt, B. George 2008b. Cases, arguments and verbs in Abkhaz, Georgian and Mingrelian. In Greville G.
Corbett & Michael Noonan (eds.), Case and grammatical relations: Studies in honor of Bernard Comrie,
75–104. Amsterdam & Philadelphia: John Benjamins.
Jacques, Guillaume. 2022 Estimative constructions in crosslinguistic perspective. Linguistic Typology 27(1).
157–194.
Jakovlev, Nikolay F. & Daud Ashkhamaf. 1941. Grammatika adygejskogo literaturnogo jazyka [A grammar of
Standard West Circassian]. Moscow & Leningrad: Izdatel’stvo AN SSSR.
Kazenin, Konstantin I. 2007. Reciprocal, comitative, sociative and reflexive in Kabardian. In Vladimir
Nedjalkov (with the assistance of Emma Geniušienė & Zlatka Guentchéva) (eds.), Typology of reciprocal
constructions, 739–772. Amsterdam & Philadelphia: John Benjamins.
Kerasheva, Zainab I. 1957. Problema preverbov v adygskix jazykax [The problem of preverbs in Circassian
languages]. Učenye zapiski AGPU 1. 233–249.
Kerasheva, Zainab I. 1992. Lokativnye i napravitel’nye preverby v adygskix jazykax [Locative and directional
preverbs in Circassian languages]. In Catherine Paris (ed.), Caucasologie et mythologie comparée. Actes
du Colloque international du C.N.R.S. – IVe Colloque de Caucasologie (Sèvres, 27–29 juin 1988), 249–263.
Paris: Peeters.
Klychev, Rauf N. 1994. Lokalʹno-preverbnoe obrazovanie glagolov abazinskogo jazyka [The locative preverbial
derivation of verbs in Abaza]. Cherkessk: Adžʹpa.
Klychev, Rauf N. 1995. Slovarʹ sočetaemosti lokal’nyx preverbov s suffiksoidami i glagol’nymi kornjami v
abazinskom jazyke [The collocational dictionary of locative preverbs with suffixoids and verbal roots in
Abaza]. Cherkessk: Karačaevo-čerkesskoe knižnoe izdatelʹstvo.
Kumakhov, Mukhadin A. 1964. Morfologija adygskix jazykov. Sinxronno-diaxronnaja xarakteristika [Morphology
of Circassian languages: Synchronic-diachronic characteristics]. I. Nalchik: Kabardino-Balkarskoe
knižnoe izdatel’stvo.
Kumakhov, Mukhadin A. 1974. K probleme ablauta v abxazo-adygskix jazykax [On the problem of ablaut in
West Caucasian]. Annual of Ibero-Caucasian Linguistics 1, 80–90.
Kumakhov, Mukhadin & Karina Vamling. 2009. Circassian clause structure. Malmö: Malmö University.
Kumakhova, Zara Y. 1972. Abadzexskij dialekt i ego mesto sredi drugix adygskix dialektov. [The Abzakh dialect
and its place among other Circassian dialects.] Nalchik: Elbrus.
Lander, Yury A. 2012. Reljativizacija v polisintetičeskom jazyke: adygejskie otnositelʹnye konstrukcii v tipologičeskoj
perspektive [Relativization in a polysynthetic language: West Circassian relative clauses in typological
perspective]. Moscow: Institute of Oriental Studies dissertation.
Lander, Yury. 2015. Aktanty i sirkonstanty v morfologii i v sintaksise adygejskogo jazyka [Arguments
and adjuncts in morphology and syntax of West Circassian]. Vestnik RGGU: ser. Istorija. Filologija.
Kul’turologija. Vostokovedenie 1. 7–31.
 25 Applicative constructions in the Northwest Caucasian languages 911

Lander, Yury. 2017. Nominal complex in West Circassian: between morphology and syntax. Studies in
Language 41(1). 76–98.
Lander, Yury. 2022. O poniženii agensa (v adygejskom jazyke) [On agent demotion (in West Circassian)]. In
Dmitry V. Gerasimov, Sergey Y. Dmitrenko, Natalia M. Zaika & Sergey S. Say (eds.), Opuscula linguistica
Magistro sapientissimo dedicata. Sbornik statej k 90-letiju Viktora Samuiloviča [Opuscula linguistica
Magistro sapientissimo dedicata. A festschrift for Viktor Samuilovich Xrakovskij on the occasion of his
90th birthday], 114–124. St. Petersburg: ILI RAN.
Lander, Yury & Timofey Arkhangelskiy. 2015. Producing polysynthetic verb forms in West Circassian
(Adyghe): an experimental study. Working papers by NRU HSE. Series WP BRP “Linguistics” 23/LNG/2015.
Lander, Yury A. & Irina Bagirokova. 2021. Cilitivy (‘legko’ i ‘trudno’) v adygejskom jazyke: semantika,
argumentnaja struktura i časterečnye xarakteristiki [Cilitives (‘easy’ and ‘difficult’) in West Circassian:
Semantics, argument structure and part-of-speech characteristics]. Rhema 1. 56–75.
Lander, Yury & Michael Daniel. 2019. West Circassian relative prefixes as resumptives. Linguistics 57(6).
1239–1270.
Lander, Yury & Alexander Letuchiy. 2010. Kinds of recursion in Adyghe morphology. In Harry van der Hulst
(ed.), Recursion and human language, 263–284. Berlin: Mouton de Gruyter.
Lander, Yury, Oleg Belyaev & Irina Bagirokova. 2021. (Almost) everything is oblique in West Circassian.
In Miriam Butt, Jamie Y. Findlay & Ida Toivonen (eds.), Proceedings of the LFG’21 Conference, On-Line,
223–242. Stanford, CA: CSLI Publications.
Letuchiy, Alexander. 2007. Reciprocals, reflexives, comitatives, and sociatives in Adyghe. In Vladimir
Nedjalkov (with the assistance of Emma Geniušienė & Zlatka Guentchéva) (eds.), Typology of reciprocal
constructions, 773–811. Amsterdam: John Benjamins.
Letuchiy, Alexander B. 2009. Affiksy benefaktiva i malefaktiva: sintaksičeskije osobennosti i krug
upotreblenij [Affixes of benefactive and malefactive: Syntactic properties and use]. In: Yakov
G. Testelets (ed.), Aspekty polisintetizma: očerki po grammatike adygejskogo jazyka [Aspects of
polysynthesis: Essays in the grammar of West Circassian], 329–372. Moscow: RGGU.
Letuchiy, Alexander. 2012. Ergativity in the Adyghe system of valency-changing derivations. In Gilles Authier
& Katharina Haude (eds.), Ergativity, valency and voice, 323–354. Berlin: Mouton de Gruyter.
Lomtatidze, Ketevan V. 1976. Kategorija versii v kartvel’skix i abxazo-adygskix jazykax [The category of
version in Kartvelian and Abkhaz-Adyghe languages]. Annual of Ibero-Caucasian Linguistics, III. 87–100.
Lomtatidze, Ketevan V. 1983. Osnovnye tipy lokal’nyx preverbov v abxazskom i abazinskom jazykax [The
basic types of locative preverbs in Abkhaz and Abaza]. In Nurja T. Tabulova & Raisa X. Temirova (eds.),
Sistema preverbov i poslelogov v iberijsko-kavkazskix jazykax [The system of preverbs and postpositions
in Ibero-Caucasian languages], 10–13. Cherkessk: Karachaevo-Cherkesskij NII istorii, filologii i
èkonomiki.
Mazurova, Julia V. (2009). Semantika lokativnyx preverbov pə- i ṣ̂ ʷe- [Semantics of the locative preverbs pə- i
ṣ̂ ʷe-]. In Jakov G. Testelec (ed.), Aspekty polisintetizma: Očerki po grammatike adygejskogo jazyka [Aspects
of polysynthesis: Essays in the grammar of West Circassian], 429–453. Moscow: RGGU.
O’Herin, Brian. 2001. Abaza applicatives. Language 77 (3). 477–493.
Paris, Catherine. 1987. Comment sont remplies en tcherkesse les fonctions dévolues dans d'autres langues
aux variations de diathèse. Actances 3. 14–72.
Paris, Catherine. 1995. Localisation en tcherkesse: forme et substance du référent. In André Rousseau (ed.),
Les préverbes dans les langues d’Europe. Introduction à l’étude de la préverbation, 345–379. Lille: Presses
Universitaires de Septentrion.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Ponomareva, Maria A. 2013. Voprosy funkcionirovanija objektnoj derivacii v abxazo-adygskix jazykax [Issues in
functioning of object derivation in Abkhaz-Adyghean languages]. Moscow: Russian State University
for Humanities diploma thesis.
912 Peter Arkadiev, Yury Lander, and Irina Bagirokova

Rogava, Georgi V. & Zainab I. Kerasheva. 1966. Grammatika adygejskogo jazyka [Grammar of West
Circassian]. Krasnodar/Maykop: Krasnodarskoe knižnoe izdatelʼstvo.
Smeets, Rieks. 1984. Studies in West Circassian phonology and morphology. Leiden: The Hakuchi Press.
Smeets, Rieks. 1992. On valencies, actants and actant coding in Circassian. In B. George Hewitt (ed.),
Caucasian perspectives, 98–144. Unterschleissheim: LINCOM.
Spruit, Arie. 1986. Abkhaz studies. Leiden: University of Leiden dissertation.
Tabulova, Nurja T. 1976. Grammatika abazinskogo jazyka. Fonetika i morfologija [A grammar of Abaza:
Phonetics and morphology]. Cherkessk: Karačaevo-Čerkesskoe otdelenie Stavropol’skogo knižnogo
izdatel’stva.
Tharkaho, Yury A. 1991. Adygejsko-russkij slovar’ [West Circassian-Russian dictionary]. Maykop: Adygejskoe
knižnoe izdatel'stvo.
Kevin Tuite
26 Applicative constructions in Kartvelian
Abstract: Two types of applicative constructions—benefactive and superessive (or loc-
ative)—occur in all four languages of the Kartvelian family (Georgian, Svan, Mingre-
lian and Laz). The principal marker of Kartvelian applicatives is a single-vowel mor-
pheme preceding the verb root (“preradical vowel”). In this chapter, the two types of
applicatives are illustrated with examples from Georgian and its sister languages. The
applicatives are compared to “version”, a grammatical category commonly employed in
Kartvelian linguistics, which indicates the orientation of the action denoted by the verb
either toward the referent of the subject, or that of the indirect object. Also presented
are applicativa tantum with lexically-specified benefactive or superessive applicative
markers; double applicatives; and morphological and syntactic lookalikes. The chapter
includes a discussion of the origin of the preradical vowels which mark applicatives in
Kartvelian. Whereas the applicative markers of other language families tend to come
from adpositions or serial verbs, no such source can be identified for Kartvelian prerad-
ical vowels, which are clearly very ancient in this language family.

1 The Kartvelian language family

Kartvelian or South Caucasian is one of the three language families endemic to the
Caucasus region, along with the Abkhaz-Adyghean or West Caucasian family, and the
Nakh-Daghestanian or East Caucasian family. Despite numerous attempts, no conclu-
sive demonstration has yet been made that these families are genetically related to each
other, nor to any known languages spoken elsewhere (Tuite 2008; Comrie 2008; Daniel
and Lander 2011). The Nostratic hypothesis, according to which Kartvelian is distantly
related to Indo-European, Uralic, Altaic and several other Eurasian families, remains
controversial (Klimov 1991, Manaster Ramer 1995).
The Kartvelian family comprises four languages, all spoken in or adjacent to the
Transcaucasian republic of Georgia. With close to 4 million speakers (Ethnologue:
3,898,550), Georgian is by far the largest language in the family, and the only one with
a long tradition of use in writing, since at least the 5th century AD. Numbers for the
other Kartvelian languages cannot be ascertained with precision. In Georgian censuses
since Soviet times, speakers of any Kartvelian language are counted as Georgian speak-
ers. Estimates of the size of the Mingrelian (a.k.a. Megrelian) speech community, based

Acknowledgments: I wish to thank my colleagues Winfried Boeder, Balkiz Öztürk, and Alexander Rostovtsev-
Popiel for sharing their works in progress with me, and for their helpful and insightful comments on earlier
drafts of this paper.

https://doi.org/10.1515/9783110730951-026
914 Kevin Tuite

in western Georgia, range from 300,000 to 500,000 (Ethnologue: 345,530). Almost all
speakers of Laz are in northeastern Turkey, and estimates of their numbers vary from
22,000 (Ethnologue) to over 20 times that many (Holisky 1991; Lacroix 2009; Kavakli
2015). Svan is the outlier in the Kartvelian family, having separated from the pro-
to-language as early as the Bronze Age. The speech community is estimated at between
14,000 (Ethnologue) and 50,000 (Gippert 2005). Both Laz and Svan are considered to be
“threatened” languages by Ethnologue, that is, the number of speakers is believed to
be declining.
Comparative work on Kartvelian goes back to the 19th century, and several ety-
mological dictionaries have been compiled (Klimov 1964; Fähnrich/Sarjveladze 2007).
In terms of morphology and syntax, the four languages share a significant number of
traits, especially with respect to the structure of the verb, but also some striking differ-
ences in case marking and person/number agreement (Harris 1991; Boeder 2005).
The Kartvelian verb is primarily agglutinative (with some morphophonemic com-
plexity, especially in Svan). Its basic architecture can be described in terms of morphe-
mic zones centered around the root (Zone I, slot 0, in Table 1). Zone II, the verb stem
includes the root and a string of suffixes (mostly of VC shape) encoding valence, Aktion-
sart, and verbal plurality. For the most part, Zone II elements occur in participles and
verbal nouns as well as finite verbs. Bracketing the stem are the zone III morphemes,
which include tense, aspect and mood suffixes limited to finite verbs (slots 8 and 9),
and the preradical vowels (slot –1) which will discussed in detail in the remainder of
this chapter. Zone III in turn is flanked by person and number markers (slots –2 and
10), which can reference one or two clausal arguments. The outermost layer includes
preverbs denoting direction, orientation and/or aspect, and clitics (especially abundant
in Old Georgian and Svan).

Table 1: The internal structure of the Kartvelian verb.

Slots –4, –3 –2 –1 0 1 to 7 8, 9 10 11
I. root Root
II. stem causative/
formants passive, verbal
plurality, series
marker
III. verb class, “version” imperfect,
tense/mood tense/
mood
IV. person and person person/
number number
V. clitics and preverbs, clitics
preverbs clitics
 26 Applicative constructions in Kartvelian 915

Here is a Georgian verb composed of morphemes from all five zones:

V
[še–4-IV[v–2-III[e–1-II[I[b0]-m2-ev3-in5-eb6-o7]-d8-e9]-t10]]
pvb-S1-prv-bind-sm-caus-caus-sm-ext-impf-opt-pl
‘were we to let ourselves be bound to it’

Of particular relevance for the analysis of Kartvelian applicatives are the morphemes
in slots –2 and –1. The personal prefixes in slot –2 are commonly separated into sets of
“subject” and “object” markers, as shown in Table 2:

Table 2: Person prefixes in the Old

Georgian and Svan verb (slot –2).

Old Georgian Svan

“Subject”
1st v- 1sg xw-
1excl xw-
1incl l-
2nd x/h/∅- 2nd x-
3rd — 3rd —
“Object”
1sg/excl m- 1sg m-
1excl n-
1incl gw- 1incl gw-
2nd g- 2nd ǰ-
3rd x/h/∅- 3rd x/∅-

However, only one prefix at a time can occupy slot –2, with the sole exception of S1 v- +
O3 x/h- (in Georgian only). Which prefix appears is conditioned by hierarchies of syntac-
tic role (O > S) and person (1,2 > 3). For example, in the Svan verb ǰ-i-t’q’b-e [O2-prv-roast-
prs] ‘I/he/she roasts it for you’, the only person marker is O2 ǰ-, which could just as well
be analyzed as a marker of both 2nd-person object and 1st- or 3rd-person subject. In
this respect, the Kartvelian person-prefix system resembles the inverse/direct or hier-
archical person-marking systems of Algonquian and some other New World languages
(Zúñiga 2006; Tuite 2021), but in the glosses, the traditional designations of these pre-
fixes as “subject” (S) and “object” (O) markers will be used.
The principal markers of applicatives are the preradical vowels (henceforth, prv) in
slot –1 (see Table 3). The Mingrelian and Laz reflexes of ✶a- and ✶e- reflect regular sound
correspondences. The irregular correspondence between the Svan prefix o-, and u- in
the other languages, remains unexplained.
916 Kevin Tuite

Table 3: prvs in the Kartvelian languages and the reconstructed

ancestral language.

Proto-Kartvelian Georgian Svan Mingrelian Laz

✶
a- a- a- o- o-
✶
i- i- i- i- i-
✶
u- u- o- u- u-
✶
e- e- e- a- a-

Georgian is well known for its intricate system of case-marking and agreement. Transi-
tive verbs and a large class of “active” intransitives assign different cases to their sub-
jects and direct objects according to tense and aspect, whereas the remaining “inactive”
intransitives assign nominative case in all tenses. In the present-series tenses (Series I:
present, future, imperfect, conditional, present and future conjunctive), case is assigned
according to a nominative-accusative pattern, with dative case marking both direct and
indirect objects. In the aorist-series paradigms (Series II: aorist and optative), however,
transitive and active intransitive verbs assign ergative case to the subject and nomina-
tive to the direct object. Since inactive intransitives do not undergo this shift of case-as-
signment properties, the resulting alignment can be characterized as split-S or active
(Harris 1990; Tuite 2017). Finally, in the perfect-series tenses (Series III: present perfect,
pluperfect, perfect conjunctive), the clause undergoes “inversion” (Shanidze 1953/1980
§ 241; Harris 1981: 117–127): the subject takes many of the attributes of an indirect
object (dative case, object agreement in the verb), whereas the direct object is assigned
nominative case and is linked to subject agreement.

(1) Georgian
a. Series I: present
bič’-i leks-s ∅-u-c’er-s deda-s
boy-nom poem-dat O3-prv-write-prs.S3sg mother-dat
‘The boy writes a poem for his mother.’
b. Series II: aorist
bič’-ma leks-i da-∅-u-c’er-a deda-s
boy-erg poem-nom pvb-O3-prv-write-aor.S3sg mother-dat
‘The boy wrote a poem for his mother.’
c. Series III: present perfect
bič’-s leks-i da-∅-u-c’er-i-a ded-is-tvis
boy-dat poem-nom pvb-O3-prv-write-perf-S3sg mother-gen-for
‘The boy apparently wrote a poem for his mother.’

Svan morphosyntax is very similar to that of Georgian, as far as split-S patterning

and inversion are concerned. Mingrelian and Laz, although closely-related languages,
diverge in interesting ways from Georgian and from each other. In Mingrelian, the
 26 Applicative constructions in Kartvelian 917

so-called ergative ending has become for all intents and purposes the aorist-series allo-
morph of the nominative, since it occurs with all types of intransitives, regardless of
their semantic traits (e.g. dzapi-k dačxir-s kimtič’u [thread-erg fire-dat was.burnt] ‘the
thread was burnt up in the fire’; Q’ipshidze 1914: 11). As for Laz, split-S alignment has
been extended to almost all tenses (Lacroix 2009), including those of the present series,
in most dialects; whereas case-marking of core arguments has disappeared completely
in the Ardeşen varieties (Kutscher, Mattissen and Wodarg 1995).

2 Applicatives in Kartvelian
In this section I will present the two types of applicatives which I ascribe to the Kart-
velian languages. Both are what Zúñiga and Creissels label as “D-applicatives”, in that
they accord indirect-object status to less prominent arguments.

2.1 Applicatives and the category of “version”

In most previous work on applicatives in the Kartvelian languages, the object of inves-
tigation is defined according to criteria specific to syntactic frameworks in the GB/
Minimalist tradition (McGinnis 2004; Lomashvili 2010; Öztürk 2013; Bondarenko 2015;
Nash 2017). While these criteria are not necessarily relevant to the approach I take
here, two of the constructions the above-mentioned authors classify as applicative are
identified as such in this chapter as well. One characteristic common to all inventories
of applicatives in Kartvelian is their grounding in the grammatical category known as
“version” (G kceva). This term was coined by Shanidze (1920, 1925) a century ago, and
has been employed in almost all descriptive and pedagogical grammars of Georgian
published since then. In his initial definition of version, Shanidze (1925), drawing upon
earlier classifications of Kartvelian valence-marking phenomena, such as those of Uslar
(1861/1887) and Marr (1925: 136–141), distinguished five types of version, signaled by
prvs: (i) “objective” (sasxviso ‘for another’) in /i/ and /u/; (ii) “subjective” (sataviso ‘for
oneself’) in /i/; (iii) “superessive” (sazedao ‘for upon’) in /a/; (iv) satanao ‘for taking
along’, to designate indirect object markers unaccompanied by a prv (which I will label
as “unmarked version”); and (v) “neutral” (saarviso ‘for no one’) for the basic construc-
tion. Some years later, Shanidze (1953/1980 § 393) revised his definition of version, shift-
ing the focus from valence to the signalling of a relation of “possession” (k’utvnileba) or
“designation” (danišnuleba) between the theme (the direct object of a transitive verb or
the subject of an intransitive verb), on the one hand, and the indirect object or agent,
on the other (see also Boeder 1969, 2021). Version in this newer sense was reduced to
a three-way contrast among objective, subjective and neutral versions. The superes-
sive was assigned to a new category, called “situation” (Shanidze 1953 § 434), and the
918 Kevin Tuite

unmarked satanao was reanalyzed as a sub-type of the neutral version. Here are exam-
ples of each type of version, as well as neutral version, in Georgian and Svan (there is
no distinct unmarked version in Svan):

(2) Georgian and Svan

a. Basic construction (neutral version)
G kali c’eril-s ∅-c’er-s
S zuräl läir-s ä-yr-i
woman:nom letter-dat prv-write-S3sg.prs/prs1
‘The woman writes a letter.’
b. Objective version, 1st p. indirect object
G kali c’eril-s m-i-c’er-s
S zuräl läir-s m-i-yr-i
woman:nom letter-dat O1sg-prv-write-S3sg.prs/prs
‘The woman writes a letter for me.’
c. Objective version, 3rd p. indirect object
G kali c’eril-s bavšv-s ∅-u-c’er-s
S zuräl läir-s bepšw-s x-o-yr-i
woman:nom letter-dat child-dat O3-prv-write-S3sg.prs/prs
‘The woman writes a letter for the child.’
d. Subjective version
G kali p’irǰvar-s i-c’er-s
S zuräl st’ārun-s i-yr-i
woman:nom cross-dat prv-write-S3sg.prs/prs
‘The woman makes the sign of the cross (lit. writes the cross on herself).’
e. Superessive version
G kali saxel-s kva-s ∅-a-c’er-s
S zuräl žaxa-s bač-s x-ä-yr-i
woman:nom name-dat stone-dat O3-prv-write-S3sg.prs/prs
‘The woman writes her name on the stone.’
f. Unmarked version
G kali c’eril-s bavšv-s s-∅-c’er-s
woman:nom letter-dat child-dat O3-prv-write-S3sg.prs/prs
‘The woman writes a letter to the child.’

Of the above-mentioned types of version, only two correspond to applicatives in the

strict sense advocated by Zúñiga and Creissels: objective, which will be labelled as “ben-
efactive applicative” in this chapter, and superessive. Subjective and unmarked ver-

1 In all these examples, the Georgian verbal suffix -s denotes person, number, role, and present tense;
the Svan verbal suffix -i only denotes present tense.
 26 Applicative constructions in Kartvelian 919

sions represent what Zúñiga and Creissels designate as morphological and syntactic
lookalikes, respectively, and will be discussed in Section 4 (Table 4).

Table 4: Shanidze’s versions and the associated applicatives.

prv version (Shanidze 1925) this chapter

i-/ u-
✶ ✶
objective (sasxviso) benefactive applicative, § 2.2
✶
a- superessive (sazedao) superessive applicative, § 2.3
✶
i- subjective (sataviso) (morphological lookalike, § 4.1)
∅- unmarked (satanao) (syntactic lookalike, § 4.2)

2.2 Objective version

2.2.1 The benefactive applicative

Shanidze’s objective version (sasxviso kceva) is identified as a type of applicative in

all studies of Kartvelian morphosyntax that employ the term “applicative”, whatever
the author’s theoretical orientation might be. One of its more noteworthy features is
the allomorphy of the prv, which is ✶i- with 1st and 2nd person indirect objects, and
✶
u- with 3rd person objects (except in Svan, which has 3rd-person o-). Compared to the
basic construction, the benefactive applicative takes an additional argument, which has
the characteristics of an indirect object: it is assigned dative case and is linked to object
agreement in the verb.
With respect to semantics, Boeder (2021 § 3.6.17) identifies two types of indirect
objects added in the benefactive applicative construction, which he labels “allative/
adessive” and “beneficiary/experiencer”. The first type of applied object denotes a ref-
erent toward, near or at which the action occurs. Here is a passage from an Old Geor-
gian translation of the Book of Genesis which contains two such constructions:

(3) Old Georgian

da-a-g-eb-s ‘spreads/lays sthg. out’ > da-m-i-g-eb-s ‘spreads/lays sthg. out for/
before me’
dg-a-s ‘stands’ > m-i-dg-a-s ‘stands near/by me’
da da-∅-u-g-o mat da čam-es. xolo twit c’ina
and pvb-O3-prv-set-pst.S3sg 3pl:dat and eat-pst.S3pl but self before
∅-u-dg-a mat xe-sa kweše
pvb-O3-prv-stand-pst.S3sg 3pl:dat tree-dat beneath
‘(He brought butter and milk and the calf that he prepared), and he set these before
them, and they ate. But he stood nearby before them under a tree.’ (Gen 18:8)
920 Kevin Tuite

The beneficiary/experiencer type of indirect object denotes the one for whose benefit,
detriment or interest the action occurs. Here are two more Old Georgian examples:

(4) Old Georgian

aɣ-a-šen-eb-s ‘builds sthg.’ > aɣ-m-i-šen-eb-s ‘builds sthg. for me’
aɣ-∅-u-šen-a mun abraam sak’urtxevel-i upal-sa
pvb-O3-prv-build-pst.S3sg there A.[erg] altar-nom lord-dat
‘Abraham built an altar there to the Lord.’ (Gen 12:7)

(5) Old Georgian

mo-drek’-s ‘bends sthg.’ > mo-m-i-drek’-s ‘bends sthg. to/for me’
mo-m-i-drik’-e me sarc’q’wal-i šeni
pvb-O1-prv-bend-pst.S1/2 1sg water.jug-nom 2sg.poss[nom]
‘Lean down your water-jug for me (so that I may drink).’ (Gen 24:14)

(6) Svan
twep-n-i ‘is lost’ > m-i-twp-en-i ‘is lost to me’
isk’wi iɣbäl ešiy dem ǰ-i-twep-n-i
your fate[nom] nonetheless not O2-prv-lose-intr-sm
‘Your fate will nonetheless not be lost to you.’ (TK 644)

(7) Svan
sgur ‘sits’ > m-i-sgur ‘sits by/next to me’
därǰəl nensga x-o-sgur-x
D.[nom] between O3-prv-sit-pl
‘Darjil sits between them.’ (TK 654)

The benefactive applicative in Laz and Mingrelian covers essentially the same semantic
range as that of Georgian and Svan. In his grammar of Laz, Lacroix (2009: 492–495)
identifies recipient and deputative types of beneficiary, and also maleficiaries (see also
Gérardin and Rostovtsev-Popiel 2016):

(8) Laz
a. k’od-um-s ‘builds sthg.’ > m-i-k’od-um-s ‘builds sthg. for me’
sumi-s-ti ayi-ayi oxoyi d-∅-u-k’od-u-doren.
three-dat-add one-one house:nom pvb-O3-prv-build-aor.S3sg-indirev
(Someone told me) ‘He built a house for each of the three of them.’
b. nax-um-s ‘washes sthg.’ > m-i-nax-um-s ‘washes sthg. for me’
mo-m-č-i do ma do-g-i-naxv-a-ya
pvb-O1-give-S1/2 and 1sg pvb-O2-prv-wash-opt-quot
(She said) ‘Give me (the clothes) and I will wash them for you.’
 26 Applicative constructions in Kartvelian 921

c. ntxo-um-s ‘digs sthg.’ > m-i-ntxo-um-s ‘digs sthg. for me’

k’ui g-i-ntxo-es nek’na-s tudele
hole:nom O2-prv-dig-aor.S3pl door-dat under
‘They dug a hole for you under the door.’ (in order to throw you in it)

The benefactive applicative construction is not only of relatively high frequency in the
Kartvelian languages, it is productive, and can readily be formed from new verbs, if
the context is appropriate. Here are some recent examples from the Georgian-language
social media. The verbs are all of recent coinage, and—to the dismay of grammarians
and purists—commonly used by young people in oral and electronic communication.

(9) Georgian
a-mesiǰ-eb-s ‘sends a text message’ > m-i-mesiǰ-eb-s ‘sends a text message to me’
Gabriel-ma da-m-i-mesiǰ-a sad xar amden xan-s?
G.-erg pvb-O1-prv-message-aor.S3sg where you.are so-long time-dat
‘Gabriel texted me: “where have you been all this time?”’

(10) Georgian
a-laik-eb-s ‘likes a post/photo (on social media)’ > m-i-laik-eb-s ‘likes my post/photo’
roca vinme mo-m-c’on-s da is sxva-s
when someone:nom pvb-O1-like-S3sg and that:nom other-dat
∅-u-laik-eb-s pot’o-s
O3-prv-like-sm-S3sg photo-dat
‘when I am attracted to someone, and that person likes someone else’s photo’

(11) Georgian
link’-av-s ‘supplies sthg. with a hyperlink’ > m-i-link’-av-s ‘sends me the link to sthg.’
dɣe-s da-g-i-link’-e es unda nax-o
day-dat pvb-O2-prv-link-aor.S1/2 this:nom must see-opt.S1/2
‘Today I sent you the link to it, you have to see this.’

Within the GB camp, there has been a side debate over whether Georgian objective
version constitutes a “low” or “high” applicative. Leaving aside the theory-specific
details, the distinction comes down to whether the applied object is primarily linked to
the theme argument, or to the verb phrase as a whole. Shanidze’s later, narrower defi-
nition of version, as a relation of possession or designation between the theme and the
applied object, would seem to support the low-applicative interpretation. Lomashvili
(2010: 150) argued for this position, characterizing the core semantics of the benefactive
applicative as one of “transfer of possession” between the two arguments. Bondarenko
(2015), on the other hand, analyses the Georgian benefactive as a high applicative,
although on the basis of syntactic, not semantic, criteria. Finally, Öztürk (2013, 2016)
922 Kevin Tuite

segments the benefactive applicative of the Pazar dialect of Laz into both types, depend-
ing on the role of the applied object: Beneficiary indirect objects are generated by a
high applicative configuration, whereas recipient and goal indirect objects (Boeder’s
allative/adessive) are the output of low applicatives.

2.2.2 Applicativa tantum with lexically-specified benefactive markers

In addition to verbs with contrasting basic and benefactive-applicative forms, Georgian

and its sister languages have a certain number of verbs for which the basic form con-
tains objective version markers.

(a) Primary statives in objective version. Most such benefactive applicativa tantum are
intransitive verbs with dat-case experiencer subjects (Table 5). These cluster in the
semantic fields of cognition and positive emotions, whereas the primary statives with
no prv or the prv ✶a-, on the whole, denote psycho-physiological states, negative emo-
tions and possession (cf. M. Mach’avariani 1987: 33–34).

Table 5: Cognate primary statives with benefactive-

applicative markers.

Meaning Georgian Zan Svan

‘I love sb./sthg.’ m-i-q’var-s m-i-ʔor-s ––
‘I prefer sthg.’ –– m-i-sx-un-u m-i-cx-a

(b) Transitive verbs with lexically-specified objective version, but no indirect object.
Shanidze (1953/1980 § 400) identified a small number of Georgian bivalent transitive
verbs with benefactive-applicative markers in their basic forms, such da-∅-u-k’rav-s
‘plays sthg. (instrument or piece of music)’, which in most varieties of Modern Georgian
take no indirect object, despite the apparent O3 marking.

2.3 Superessive version

2.3.1 The superessive applicative

The superessive applicative, expressed by the prv ✶a- (= o- in Mingrelian and Laz),
is less common and less productive than the benefactive applicative, but it is by no
means rare. This construction typically indicates that the described action took place
on a surface, denoted by the applicative indirect object. According to Boeder (2021 §
3.6.17.vi), the superessive applicative “is connected with a specific relationship between
the subject/direct object and the indirect object: a part-whole relationship . . ., a close
 26 Applicative constructions in Kartvelian 923

attachment . . ., a continuous physical or emotional pressure on sth/sb . . ., or a nui-

sance”. With respect to (13) below, Rostovtsev-Popiel (2015) specified that the choice of
the superessive construction, rather than the basic construction with a postpositional
object, implies a degree of prominence or foregrounding of the site of contact: “It is
more or less obvious that the horns are to be added in this case on top of the heads of
the people painted on the pictures, not on their sides, shoulders etc.”.

(12) Old Georgian

ps-am-s ‘urinates’ > ∅-a-ps-am-s ‘urinates on sthg.’
mo-v-sp’-o ierobuam-is-i romel-i ∅-a-ps-m-i-d-e-s
pvb-S1-destroy-opt J.-gen-nom which-nom O3-prv-piss-sm-ext-impf-opt-S3sg
k’edel-sa
wall-dat
‘I will destroy him of Jeroboam, that pisses against the wall.’ (III Kings 14:10)

(13) Georgian
xat’av-s ‘paints/draws sthg.’ > ∅-a-xat’av-s ‘paints/draws sthg. on sthg.’
bavšv-eb-ma p’ort’ret’-eb-s rk-eb-i mi-∅-a-xat’-es
child-pl-erg portrait-pl-dat horn-pl-nom pvb-O3-prv-paint-aor.S3pl
‘The children drew horns on the portraits.’ (Rostovtsev-Popiel 2015)

(14) Mingrelian
č’k’ad-ən-s ‘hammers sthg.’ > ∅-o-č’k’ad-an-s ‘hammers sthg. onto sthg.’
k’učxi-s ečdoxut-xuti putiani nal-ep-i
foot-dat 25-each pood horseshoe-pl-nom
ku-m-m-o-č’k’ad-i-a
pvb-pvb-O1-prv-hammer-S1/2-quot
(The magic horse said): ‘Nail 25 pood (= 400 kg!) horseshoes onto each of my feet.’
(Xubua 1976: 167)

(15) Svan
a. sgur ‘sits’ > x-a-sgur ‘sits on sthg.’
katal ži x-a-sgur ləgr-ol-s
chicken:nom up O3-prv-sit egg-pl-dat
‘The hen is sitting on the eggs.’ (TK 575)
b. bə̄d-n-i ‘is poured, [liquid] falls’ > x-a-bə̄d-n-i ‘is poured, falls on sthg.’
mananay bal-ar-s x-a-bə̄d-n-i
dew:nom leaf-pl-dat O3-prv-pour-intr-sm
‘Dew falls on the leaves.’ (TK 139)

As regards present-day usage with new verbs, the superessive is less frequent, but can
occur when the context calls for it:
924 Kevin Tuite

(16) Georgian
a-st’ep’ler-eb-s ‘staples sthg.’ > ∅-a-st’ep’ler-eb-s ‘staples sthg. to sthg.’
dana-m q’ur-is bibilo-ti mi-∅-a-st’ep’ler-a k’edel-s
knife-erg ear-gen lobe-inst pvb-O3-prv-staple-aor.S3sg wall-dat
čven-i temo
our-nom T.[nom]
‘The knife stapled our Temo to the wall by his ear-lobe.’

In Svan, the superessive applicatives formed from ablauting intransitive verbs have the
prv e- rather than a-, as in the other Kartvelian languages (Table 6). This is most likely an
innovation in Svan, although its cause remains unclear (Topuria 1967: 49–50; Tuite 2021):

Table 6: Svan superessive applicatives: paired root intransitives in e- and transitives in a-.

Ablauting intransitive with prv e- Transitive with prv a- (superessive applicative)

x-e-šgb-en-i ‘slips off from sthg.’ x-a-šgb-e ‘tears sthg. from sb./sthg.’
x-e-t’q’wp-en-i ‘(skin) comes off from sb./sthg.’ x-a-t’q’wp-e ‘tears sthg. (skin, body part) off from sb./sthg.’
x-e-q’wl-en-i ‘departs from sb./sthg.’ x-a-q’wl-e ‘separates sthg. from sb./sthg.’

Unlike the benefactive applicative, the superessive can be marked by morphological

changes other than the addition of a prv. In Georgian, Mingrelian and Laz, the forma-
tion of the superessive applicative of many transitive verbs is accompanied by a change
of the series marker (sm, slot 6), a morpheme which occurs in the present-series stem of
most verbs. The Mingrelian verb in (14) above illustrates this phenomenon, as does its
Georgian cognate (Table 7). The preverb is also different in many superessive verbs. The
preverb mi- is especially common in Georgian superessives, as in (13) and (16).

Table 7: Series marker change in the superessive applicative.

basic transitive (sm ✶-aw-) superessive (prv ✶a- + sm ✶-ew-)

G. č’ed-(av)-s ‘forges, hammers sthg.’ G. a-č’ed-eb-s ‘forges, nails sthg. onto sthg.’
M. č’k’ad-ən-s M. o-č’k’ad-an-s
G. par-av-s ‘covers’ G. a-par-eb-s ‘covers sb./sthg. with sthg.’
M. por-un-s M. o-por-an-s

2.3.2 Applicativa tantum with lexically-specified superessive markers

As was the case with the benefactive applicative, there are verbs of different types
with the prv ✶a- which, at least from a synchronic standpoint, cannot be considered the
output of applicativization.
 26 Applicative constructions in Kartvelian 925

(a) Primary statives with lexically-specified prv ✶a-. Alongside the primary statives
which have the form of benefactive applicatives (§ 2.2.1 above), there are a small
number of primary statives with the PR ✶a-. Here are some examples with cognate roots
in two or more Kartvelian languages (Table 8):

Table 8: Primary stative verbs in ✶a-.

Georgian Mingrelian Svan

‘I lack sthg.’ m-a-k’l-i-a m-o-rk’-u-n m-a-k’l-i

‘I want sthg.’ —— m-o-k’-on m-a-k’u

(b) Trivalent transitive verbs with lexically-specified prv ✶a-. Lacroix (2009: 463, 525)
identified several trivalent transitive verbs in Laz with basic forms in the superessive,
such as dolo-∅-o-kun-am-s ‘puts sthg. (clothes) on sb.’; and mo-∅-o-k’id-am-s ‘hangs sthg.
on sthg.’. Georgian has many superessive applicativa tantum, most of which have mean-
ings implying transfer, e.g. a-dzl-ev-s ‘gives sthg. to sb.’, a-c’vd-i-s ‘hands/passes sthg. to
sb.’, a-bar-eb-s ‘entrusts sb./sthg. to sb.’, a-dar-eb-s ‘compares sb./sthg. to sb./sthg.’.

(c) Bivalent transitive verbs with lexically-specified prv ✶a-, but no indirect object.
An archaic class of bivalent transitive verbs, many of them with nonsyllabic roots,
take the prv ✶a- in their basic “neutral version” form (Shanidze 1953/1980 § 458).
Primary ✶a-transitives cluster around the meanings of (i) building, setting up; (ii)
touching; (iii) bringing into contact (e.g., flame to a candle, a brush dipped in paint),
which makes it likely that the ✶a- prefix in these verbs is cognate with the superessive
applicative marker. Here are some verbs of this type with cognate roots in Georgian
and Mingrelian (Table 9):

Table 9: Cognate primary ✶a-transitives in Georgian and Mingrelian.

Georgian Mingrelian
✶
a-g-ew- ‘builds’ a-g-eb-s o-g-an-s
✶
a-gz-ew- ‘lights, incites’ a-gz(n)-eb-s o-rz-an-s
✶
a-c’(w)-ew- ‘dips’ a-c’-eb-s u-c’u-an-s
✶
a-x-ew- ‘touches’ a-x-eb-s o-x-u(n) ‘concerns’

2.3.3 The prv ✶a- and transitivity

The most productive use of the prefix ✶a- in the Kartvelian languages would appear at
first to have nothing to do with superessive meaning. Along with certain suffixes, the
prv ✶a- is a component of derived transitives and causatives. The ✶a-prefixed deriv-
atives of monovalent verbs, nouns and other parts of speech are bivalent transitives
926 Kevin Tuite

without indirect objects, such as these Georgian examples: a-c’ux-eb-s ‘bothers, causes
to worry’ < c’ux-s ‘is worried’; a-lamaz-eb-s ‘beautifies’ < lamaz- ‘beautiful’; a-ortkl-eb-s
‘makes evaporate’ < ortkl- ‘steam’. Those derived from transitives are causatives with
an indirect object denoting a second agent or instigator: a-c’er-in-eb-s ‘causes to write’ <
c’er-s ‘writes’; a-č’m-ev-s ‘feeds’ < č’am-s ‘eats’ (G. Mach’avariani 1988; M. Mach’avariani
1987: 87–115). The possibility of a deeper diachronic link between these two functions
of the prv ✶a-—superessive and transitivity—will be discussed below (§ 5).

2.4 Applicatives in the Series III tenses

The morphological and syntactic distinction between the applicative and basic construc-
tions for both types of applicatives presented above is maintained in all tenses except
those of Series III (Shanidze 1953/1980 §§ 403, 410, 435). In all Kartvelian languages,
transitive verbs, as was mentioned in § 1, undergo “inversion” in the present-perfect
and other Series III tenses, that is, the subject is marked like the indirect object of a
benefactive applicative, as far as case and agreement are concerned. One consequence
of inversion is the neutralization of the morphological and syntactic signs of applicativ-
ization, and in fact of all four types of version identified in § 2.1 above. The applicative
indirect objects and their associated prvs are replaced by postpositional phrases which
do not agree with the verb (Table 10):

Table 10: Basic and applicative transitive verbs in Series I and Series III.

Series I (present) Series III (present-perfect)

Basic v-t’ex ‘I break sthg.’ mo-m-i-t’ex-av-s ‘I have broken sthg.’
Benefactive appl v-∅-u-t’ex ‘I break sthg. for sb.’ mo-m-i-t’ex-av-s mis-tvis [3sg:gen-for]
Superessive appl v-∅-a-t’ex ‘I break sthg. on sthg.’ mo-m-i-t’ex-av-s mas-ze [3sg:dat-on]
Subjective version v-i-t’ex ‘I break sthg. for myself’ mo-m-i-t’ex-av-s čem-tvis [1sg:gen-for]
Unmarked version v-s-t’ex ‘I break sb.’s sthg.’ mo-m-i-t’ex-av-s mis-tvis [3sg:gen-for]

In morphosyntactic terms, the present perfect of a transitive verb such as G c’er-,

M č’ar-, S ir- ‘write’ is a bivalent resultative-passive, with an indirect object referring to
the agent. The verb forms G m-i-c’er-i-a, M m-i-č’ar-u-n, S mīra < m-i-ir-a could, accord-
ing to the context, be interpreted as either benefactive applicatives of stative verbs (‘it
is written for me, in my (e.g. book), etc.’), or as present perfects of a transitive verb (‘I
have written it’). As for applicative intransitive verbs, they too undergo neutralization
in Series III (Shanidze 1953/1980 § 410). For example, both the benefactive ∅-u-t’q’d-eb-a
‘sthg. breaks for/on sb.’, and superessive ∅-a-t’q’d-eb-a ‘sthg. breaks on/off from sthg.’,
have the same present-perfect mo-s-t’q’d-om-i-a.
 26 Applicative constructions in Kartvelian 927

3 Stacking/combination of voice operations

In the Kartvelian languages, verbs which have undergone other types of valence-alter-
ing operations, such as intransitivization and causative formation, can subsequently be
applicativized. It is also the case that certain types of applicative verbs can undergo a
second applicativization.

3.1 The prv ✶e-

So far, nothing has been said concerning the fourth of the prvs in Table 3, ✶e-. The
primary function of this vowel is to form the bivalent counterparts of intransitives
which have the prv ✶i- in their basic forms, in order “to relate the action/event/state to a
new participant in a way that the latter becomes either directly or indirectly involved”
(Gérardin and Rostovtsev-Popiel 2016). The resulting verbs typically govern a theme in
the nom case and an indirect object, although the latter often has many of the syntac-
tic privileges of a grammatical subject. In terms of their relation to basic forms, three
subtypes can be distinguished: intransitivized applicatives, applicativized intransitives,
and primary ✶e- verbs.

3.1.1 Intransitives formed from applicativized transitives

These are most commonly superessive applicatives and quite frequent in the Old Geor-
gian corpus, e.g.

(17) Old Georgian

mi-∅-a-axl-eb-s ‘brings sb./sthg. near to sb./sthg.’ > mi-∅-e-axl-eb-i-s ‘approaches,
nears sb./sthg.’
da mi-∅-e-axl-a iak’ob isak’-s
and pvb-O3-prv-near-aor.S3sg J.[nom] I.-dat
‘And Jacob went near to Isaac.’ (Gen 27: 22)

(18) Svan
x-o-cwm-i ‘smears sthg. on/for sb./sthg.’ > x-e-cwm-i ‘sthg. is smeared on sb./sthg.’
äpicer-s ulmaš-är-s ži-ad-x-e-com-ēn-a2 nacmun
officer-dat moustache-pl-dat pvb-pvb-O3-prv-smear-pass-aor grease:nom
‘Grease was smeared on the officer’s moustache.’ (TK 234)

2 The surface form is žäxcomǟn.

928 Kevin Tuite

3.1.2 Applicatives of ✶i-intransitives

Another common use of the prv ✶e- is to form applicatives from intransitives with the
prv ✶i-. The distinction between benefactive and superessive applicatives is neutralized
in this case. Here are two examples formed from passive verbs:

(19) Old Georgian

še-i-cvl-eb-i-s ‘is changed’ > še-∅-e-cvl-eb-i-s ‘sb.’s sthg. is changed; sthg. is changed
for sb.’
q’ovel-i ese . . . codvil-ta še-∅-e-cval-o-s borot’-sa šina
all-nom this sinner-obl.pl pvb-O3-prv-change-opt-S3sg evil-dat in
‘All these things . . . are turned into evil for sinners.’ (Ecclesiasticus 39:27)

(20) Svan
i-dgär-i ‘dies’ > x-e-dgär-i ‘sb.’s (relative) dies; sb. dies (accidentally) by sb.’s action’
dīna-s diutwra ad-x-e-dag-an3
girl-dat stepmother:nom pvb-O3-prv-die-aor
‘The girl’s stepmother died.’ (TK 193)

(21) Laz
i-čod-e-n ‘ends, is finished’ > ∅-a-čod-e-n ‘sb.’s sthg. is finished, sthg. ends for sb.’
bič’i-s xorci d-∅-a-čod-u
boy-dat meat:nom pvb-O3-prv-end-aor.S3sg
‘The meat ended for the boy (i.e. the boy had no more meat).’ (Lacroix 2007)

Primary medial verbs in ✶i- can also form applicatives in ✶e-. This formation is espe-
cially common in Svan, less so in Old Georgian, Mingrelian and Laz (Lacroix 2007).

(22) Old Georgian

i-glov-s ‘mourns’ > ∅-e-glov-s ‘mourns sb.’
da ∅-e-glov-d-a mas egwip’t’e sameoc-da-at dɣe
and O3-prv-mourn-impf-S3sg 3sg:dat Egypt:nom 60-and-10 day
‘And Egypt mourned him seventy days.’ (Genesis 50:3)

(23) Svan
i-mzir ‘prays’ > x-e-mzir ‘prays for sb.’
megza x-e-mzir-x naɣwžurgezl-äš lipširi-s
family:dat O3-prv-pray-pl male-child-gen multiply-dat
‘They prayed for an abundance of sons for the family.’ (TK 451)

3 The surface form is atdagan.

 26 Applicative constructions in Kartvelian 929

3.1.3 Primary verbs in ✶e-

Each of the Kartvelian languages has a sizeable, and productive, set of intransitive
verbs in ✶e- derived from noun, adjective and verb stems. Shanidze (1953: 299–301)
groups these into verbs of possibility (šesadzlebloba), assessment (mičneva) and mood
(guneba). Here are examples of each kind:

(24) Laz: possibility

gam-i-l-e-n ‘goes out’ > gama-∅-a-l-e-n ‘sb./sthg. can go out’
gama-g-a-l-e-n
pvb-O2-prv-go-sm-S3sg
‘You can go out, are allowed to go out.’ (Lacroix 2007)

(25) Georgian: assessment

pot’ošop’-i ‘Photoshop’ > ∅-e-pot’ošop’-eb-a ‘(photo) seems altered by software to sb.’
cot’a m-e-pot’ošop’-eb-a es surat-i
little O1-prv-photoshop-sm-S3sg this picture-nom
‘I have a slight impression that this picture was photoshopped.’

(26) Georgian: mood

mɣer-i-s ‘sings’ > ∅-e-mɣer-eb-a ‘sb. feels like singing’
m-e-mɣer-eb-a da v-i-mɣer-i
O1-prv-sing-sm-S3sg and S1-prv-sing-sm
‘I feel like singing, and I will sing.’ (title of poem by Vazha-Pshavela, 1903)

3.2 Applicatives of causatives

In principle, Kartvelian causatives should have the same range of applicatives as ordi-
nary transitive verbs. For bivalent causatives formed from intransitive verbs, this is
more or less the case. As for trivalent causatives of transitive verbs, Makharoblidze
(2012: 155–156) provides conjugation tables for quadrivalent benefactive applicatives
such as v-u-šen-eb-in-eb ‘I am making him/her/it build it (for him/her/it)’. In practice,
such verbs are uncommon. Shanidze (1980 § 428–429) provides some examples from
Georgian literary sources:
930 Kevin Tuite

(27) Georgian
c’er-s ‘writes sthg.’ > a-c’er-in-eb-s ‘makes sb. write sthg.’ > m-i-c’er-in-eb-s ‘makes
sb. write sthg. for me’
kavtarišvil-ma sigel-i mikel teodat’e-s švil-s
K.-erg charter-nom M. T.-gen son-dat
da-m-i-c’er-in-a
pvb-O1-prv-write-caus-aor.S3sg
‘Kavtarishvili had Mikel, son of Theodate, write a charter for me.’ (Iese Osesshvili
c. 1770)

3.3 Double applicatives

In addition to being applied to verbs that have undergone valence change through
passivization or causativization, applicatives can also be layered on verbs that are
themselves the product of applicativization. Double applicatives are not common, espe-
cially those that result in quadrivalent verbs. In general, they consist of a benefactive
applicative superimposed on a superessive applicative (Shanidze 1953/1980 §§ 402, 411,
416, 442; Harris 1981: 99–100, 286; Singer 2003; Lomashvili 2005: 205–207), as in this
example:

(28) Georgian
c’eb-av-s ‘glues sthg.’ > ∅-a-c’eb-eb-s ‘glues sthg. to sthg.’ > m-i-c’eb-eb-s ‘glues sthg.
to my sthg.’
viɣaca-m ertmanet-s c’ebo-ti mi-m-i-c’eb-a
someone-erg each.other-dat glue-inst pvb-O1-prv-glue-aor.S3sg
tit-eb-i
finger-pl-nom
‘Someone stuck my fingers together with glue.’ (T. Jangulashvili Mnatobi #5, 1986)

In present-day usage, double applicatives with two indirect objects tend to be avoided,
with the object of the inner applicative marked by a postposition rather than the bare
dative case (Lomashvili 2005: 205–207), e.g.

(29) Georgian
k’er-av-s ‘sews sthg.’ > ∅-a-k’er-eb-s ‘sews sthg. onto sthg.’ > m-i-k’er-eb-s ‘sews
sthg. onto my sthg.’
ɣil-i p’iǰak’-ze // (p’iǰak’-s) mi-m-i-k’er-a
button-nom jacket-on jacket-dat pvb-O1-prv-sew-aor.S3sg
‘She sewed a button onto my jacket.’
 26 Applicative constructions in Kartvelian 931

In principle, verbs that are the product of benefactive applicativization cannot undergo
the same operation a second time, but some apparent exceptions have been attested
(Singer 2003). The few examples attested in Georgian are applicatives of benefactive
applicativa tantum, that is basic verbs which contain the functionless or invariant
prv ✶u- (see § 2.2.2 above). The verb da-∅-u-k’rav-s ‘plays sthg. (instrument or piece of
music)’, which contains a functionless O3 marker, can undergo the addition of a benefi-
ciary argument (Boeder 1968: 120–121):

(30) Georgian
∅-u-k’r-av-s ‘plays sthg.’ > m-i-k’r-av-s ‘plays sthg. for me’
git’ara-ze da-gv-i-k’r-a ramdenime simɣera
guitar-on pvb-O1pl-prv-play-aor.S3sg several song:nom
‘He played us several songs on the guitar.’

According to Shanidze (1953/1980 § 414), in earlier times, and in some conservative

Georgian dialects, the indirect-object marker of da-∅-u-k’rav-s referred to the instru-
ment played. In this example from the Khevsurian dialect, spoken in the northeastern
highlands, the benefactive applicativized form is quadrivalent, with a direct object des-
ignating the piece that is performed, an indirect object denoting the instrument, and
another indirect object denoting the beneficiary. In this context, the O3 marker in the
basic form is functional but invariant, since the referent is necessarily inanimate, thus
3rd person:

(31) Georgian
∅-u-k’r-av-s ‘plays sthg. on sthg.’ > m-i-k’r-av-s ‘plays sthg. on sthg. for me’
pandur-s da-m-i-k’ar-∅
pandur-dat pvb-O1-prv-play-aor.S1/2
‘Play the pandur [name of instrument] for me!’ (Shanidze 1953/1980 § 414)

Öztürk (2013, 2016) elicited sentences in the Pazar variety of Laz, which appear to result
from two operations of benefactive applicativization, should the basic form of this verb
be bivalent (as is Arhavi Laz o-şku-me ‘sends/releases sb./sthg.’; Lacroix 2009: 437, 445):

(32) Laz
o-šk-u ‘sent sb./sthg.’ > ∅-u-šk-u ‘sent sb./sthg. to sb.’ > ∅-u-šk-u ‘sent sb./sthg. to
sb. for sb.’
Xordza-k Ali-s k’oçi-s bere ∅-u-şk-u.
woman-erg A.-dat man-dat child:nom O3-prv-send-aor.S3sg
‘The woman sent the child to the man for Ali.’

There are nonetheless semantic constraints: the first applied object must denote a recip-
ient, and the second a beneficiary.
932 Kevin Tuite

4 Morphological and syntactic lookalikes

As mentioned above (§ 2.1), the Kartvelian subjective version and unmarked (satanao)
version do not correspond to applicatives as defined in this volume. In the case of
subjective version, the morphology undergoes change, by addition of the prv i-, but
the surface syntax remains the same, in that no overt argument is added. Unmarked
version, by contrast, is characterized by the addition of an indirect object, but without
any change to the morphology of the verb (other than the addition of an object marker).

4.1 Subjective version

Alongside its function as the marker of objective version with 1st- and 2nd-person indi-
rect objects, the prv ✶i- also marks subjective version (sataviso kceva). The contrast
between the basic and subjective version constructions is limited to transitive verbs.
The Kartvelian subjective version indicates that the action is performed by the referent
of the subject (1) on his/her own body, or clothing, or an object he/she is carrying; or
(2) for the subject’s own benefit, in some sense (Shanidze 1953/1980 § 396). Although
it is sometimes described as the reflexive counterpart of the benefactive applicative
(e.g. Bondarenko 2015), the semantic range of the subjective version is wider. Boeder
(2021 § 3.6.17 vi) notes that it “occurs with any reflexive indirect object. It neutralizes
the opposition between objective version, superessive version and unspecified indirect
objecthood [= unmarked version—KT]”. The reflexive applicative construction has the
same valence as the basic construction, at least as far as surface structure is concerned
(more on this below).

(33) Old Georgian

tes-av-s ‘sows sthg.’ > i-tes-av-s ‘sows sthg. for oneself’
mk’-i-s ‘mows, reaps sthg.’ > i-mk’-i-s ‘mows, reaps sthg. for oneself’
∅-i-tes-e-t ipkl-i da ek’al-sa mo-∅-i-mk’-i-t
S2-prv-sow-aor-S1/2pl wheat-nom and thorn-dat pvb-S2-prv-reap-sm-S1/2pl
‘You sowed (for yourselves) wheat but will reap (for yourselves) thorns.’
(Jeremiah 12:13)

(34) Svan
a-tī ‘mows sthg.’ > i-tī ‘mows sthg. for oneself, in one’s own fields’
k’wecen-s našt’ak-wš xw-i-tī-d
wheat-dat sickle-inst S1excl-prv-mow-S1/2pl
‘We mow (for ourselves) wheat with sickles.’ (TK 275)
 26 Applicative constructions in Kartvelian 933

(35) Mingrelian
k’vat-un-s ‘cuts sthg.’ > i-k’vat-un-s ‘cuts one’s own sthg. (e.g. body part)’
k’it-i gi-i-k’vat-u xam-it
finger-nom pvb-prv-cut-aor.S3sg knife-inst
‘He cut his (own) finger with a knife.’

In his grammar of Laz, Chikobava (1936/2008: 119–121) stated that the use of the subjec-
tive version in that language was limited to situations where the referent of the gram-
matical subject acted on his/her own body; in other words, the Laz subjective version
could be described more accurately as the reflexive counterpart of superessive or
unmarked version. Whereas Georgian i-c’er-s usually means ‘writes (down) for oneself’
(e.g. takes notes, records something), its Laz cognate i-č’ar-up-s, according to Chikoba-
va’s informants, “would be used if one started to write on one’s own body, and who
would ever do that?” (Chikobava 1936/2008: 120). This restriction on the semantic range
of the subjective version may not (or no longer) apply to the present-day dialects of
Laz. While most of Lacroix’s examples of verbs in the subjective version, collected from
contemporary Laz speakers in Turkey, are consistent with Chikobava’s observation, he
also recorded instances with benefactive or possessive meaning, e.g.

(36) Laz
ma oxoi b-i-k’od-um
1sg house:nom S1-prv-build-sm
‘I build a house for myself.’ (Lacroix 2012)

As was noted previously for the benefactive applicative, the subjective version is com-
monly used in all Kartvelian languages, and readily appears with newly-minted verbs:

(37) Georgian
a-pot’ošop’-eb-s ‘uses software to modify (image)’ > i-pot’ošop’-eb-s ‘uses software
to modify one’s own (image)’
sax-is k’an-s i-pot’ošop’-eb-s
face-gen skin-dat prv-photoshop-sm-prs.S3sg
‘She photoshops her (own) facial skin.’

Unlike the other constructions presented here, the subjective version in ✶i- does not
add an indirect object to the surface structure of the clause. This has led to an inter-
esting divergence of opinion concerning how to classify this construction. Beginning
with Shanidze, most linguists have grouped it in the same category as the benefactive
applicative, either under the heading of “version”, or as a type of applicative (“reflexive
applicative”, according to Lomashvili 2010: 191ff and Bondarenko 2015). With respect
934 Kevin Tuite

to valence change, Boeder (1968) noted that the subjective version construction can be
paraphrased with a benefactive or another type of applicative and an explicit reflexive
indirect object, e.g.

(38) Georgian: reflexive applicative (bivalent) = benefactive applicative (trivalent)

v-i-k’rep vašl-s = v-u-k’rep vašl-s čem-s tav-s
S1-prv-pick apple-dat S1-prv-pick apple-dat my-dat head-dat
‘I pick myself an apple.’ = ‘I pick an apple for my self (lit. my head).’

On this basis, the Kartvelian subjective version / reflexive applicative has been analyzed
as a construction with an implicit indirect object (“impliziten i-Dativ”) that is coreferen-
tial with the grammatical subject (see also Harris’s “Coreferential Version Object Dele-
tion”, 1981: 95–99; Harris 1991: 46).
Shanidze also noted significant semantic overlap between the subjective version
and certain uses of the middle voice in Indo-European languages such as Greek and
Sanskrit (1953/1980 § 417; see also Schmidt 1965). Lacroix (2009: 456–483; 2012) takes
this observation a step further, and classifies the prv ✶i- as the morphological marker of
middle voice in Kartvelian. He therefore separates subjective-version transitives from
the applicatives, and groups them with the large class of Kartvelian intransitive verbs
also marked by the prv ✶i-, which cover the semantic domains of: passive (G i-c’er-eb-a
‘is written’), potential (i-č’m-eba ‘can be eaten’), anticausative (i-c’v-eb-a ‘burns (intr)’),
autocausative (i-ndzr-ev-a ‘moves (intr)’), and antipassive (i-ɣeč’-eb-a ‘(person or
animal) chews in a leisurely or annoying manner’; Tuite 2002). Taken together, subjec-
tive-version transitives and ✶i-prefixed intransitives cover most of the meanings associ-
ated by Kemmer (1993) with the middle voice on a cross-linguistic basis.

4.2 Unmarked version

It was mentioned at the outset of this chapter that Shanidze, in his initial definition of
the category of version, included a type he labelled satanao ‘for taking along’, which
specified an indirect object but without the addition of a prv to the verb morphology.
Such verbs occur in Laz and Mingrelian as well as Georgian, but not in Svan. In the
Svan cognates of verbs with unmarked version, the prv a- or o-/i- appears after the indi-
rect-object marker, e.g. G m-q’id-i-s, S mäq’di < m-a-q’id-i ‘sells sthg. to me’; G m-c’er-s,
M m-č’ar-un-s, S mīyri < m-i-ir-i ‘writes (and sends) sthg. to me’. In my view, there are
grounds for hypothesizing that Svan lost a distinction between indirect objects with and
without prvs, which has been retained in its sister languages (Tuite 2021).
Turning to Georgian, Laz and Mingrelian, the verbs in unmarked version can be
divided into two groups. On the one side are those for which unmarked version repre-
sents their basic form, that is, fundamentally trivalent transitives (e.g. G mo-m-c-em-s
 26 Applicative constructions in Kartvelian 935

‘will give sthg. to me’) and bivalent intransitives (e.g. G m-dzul-s ‘I hate sb./sthg.’). Such
verbs do not have a more basic form which lacks an indirect object, i.e. there are no
such verbs as †mo-c-em-s ‘will give sthg.’ or †dzul-s ‘sb./sthg. is hateful’. The second,
larger, group comprises verbs in unmarked version for which there exist basic forms
lacking an indirect object. The principal semantic fields associated with verbs taking
prv-less indirect objects are: (i) transfer, transmission or taking, with the indirect object
denoting addressees or recipients (Jorbenadze 1983: 219–226); and (ii) action implying
body contact, often violent, with the indirect object denoting the participant intimately
effected by the action. Here are some examples from Old Georgian and Mingrelian:

(39) Old Georgian

k’wet-s ‘cuts sthg.’ > m-k’wet-s ‘cuts my sthg. (esp. body part)’
mo-g-k’wet-o-s parao tav-i šeni šen-gan
pvb-O2-cut-opt-S3sg pharaoh[erg] head-nom 2sg.poss[nom] 2sg-from
‘Pharaoh will cut off your head from you.’ (Gen 40: 19)

(40) Mingrelian
č’ar-un-s ‘writes sthg.’ > m-č’ar-un-s ‘writes (and sends) sthg. to me’
minǰe-s me-∅-č’ar-ə dzɣabi-k
owner-dat pvb-O3-write-aor.S3sg girl-erg
‘The girl wrote to the owner.’ (Xubua 1976: 74)

According to Shanidze (1953/1980 § 440) and Deeters (1930: 79–80), the semantic range
of prv-less indirect objects overlaps that of superessive objects marked by the prv ✶a-,
as attested by parallel Old Georgian translations of the same Biblical passage (e.g., da=h-
k’wet-a /da=∅-a-k’wet-a ‘threw him down [to the ground]’, Mark 9: 20). Nonetheless,
comparison of a corpus of verbs which allow both superessive and unmarked indirect
objects reveals consistent semantic differences between the two, especially as regards
the animacy of the applied object. The unmarked version has a strong association with
animate arguments (possessors of body parts, experiencers), whereas the superessive
applicative covers a broader semantic range, including verbs denoting physical move-
ment or removal from an inanimate surface (Table 11).

Table 11: Contrasting superessive and unmarked version in Georgian.

Root Superessive applicative Unmarked version

gleǰ/gliǰ- ‘tear’ mo=∅-a-gliǰ-a ‘tore sthg. (e.g. mo=h-gliǰ-a ‘tore sthg. off sb.’s body (e.g. hat
knob) off sthg.’ off head, mask off face)’

t’q’d- ‘break (intr)’ mo=∅-a-t’q’d-a ‘sthg. (e.g. handle) mo=s-t’q’d-a ‘sb.’s sthg. (e.g. fingernail, arm)
broke off sthg.’ broke’
936 Kevin Tuite

When used with some verbs of transfer or communication, the unmarked version “can
be more or less synonymous” with the benefactive; e.g. G m-q’ep-s / m-i-q’ep-s ‘it barks
at me’ (Boeder 2021 § 3.6.17.ix).
Although somewhat less common than the other version types, an Internet search
has yielded at least one newly-created Georgian verb which allows the unmarked
version:

(41) Georgian
p’ost’-av-s ‘posts sthg. (on social media)’ > m-p’ost’-av-s ‘posts sthg. to me’
me ro uk’ve mo-g-p’ost’-e, k’ide mo-g-p’ost’-o?
1sg that already pvb-O2-post-aor.1/2 again pvb-O2-post-opt
‘Since I already posted it to you, do I have to post it to you again?’

5 The origin of the Kartvelian applicative markers

In his crosslinguistic survey of applicatives, Peterson (2007: 123) identified two primary
sources of applicative morphology: adpositions and serial verbs. Other sources, such as
body-part nouns, have been proposed for the applicative morphemes of certain languages,
but Peterson (2007: 140–141) considers the arguments unconvincing. The prvs which mark
Kartvelian applicatives, however, have no evident link to any such lexical category. The
most promising path toward elucidating the origin of the preradical vowels was suggested
initially by Topuria (1947) and Vogt (1974), who pointed to evidence that the prvs were not
limited to finite verbs in Proto-Kartvelian. The prefixes ✶a- and ✶i- occur in participles and
a small number of nouns, as in the following examples from Georgian (Table 12):

Table 12: prvs in verbs, participles and nouns (Georgian)

✶
a- (root -xl- ‘touch’) ✶
i- (root -s(v)r- ‘shoot’)
Finite verbs (“version”) a-xl-eb-s ‘touches’ i-svr-i-s ‘shoots’
Participle in s- s-a-xl- ‘house’ (‘site of closeness’) OG s-i-sr-a ‘shooting’
Nouns (frozen prefix) a-xl-o- ‘close, near’ i-sar- ‘arrow’ (‘it is shot’)

Of special interest for reconstructing the original functions of the prvs is a small,
archaic class of vowel-initial nouns based on verbal roots (Fähnrich and Sarjveladze
2007: 27–28, 210). The initial vowels in these nouns appear to be frozen prvs, a hypoth-
esis which draws support from their meanings. In the ✶a-prefixed nouns, one detects a
semantics of space and attachment (a-xl-o ‘near’ < ‘touching sb./sthg.’), corresponding
to the core uses of the superessive applicative marker; whereas the ✶i-prefixed nouns,
when compared to their verbal roots, have middle-voice or passive meaning (G i-sar-, L
i-siǰ- ‘arrow’ < ‘ce qui est lancé’; see Vogt 1974 and Klimov 1964: 102) (Table 13).
 26 Applicative constructions in Kartvelian 937

Table 13: Georgian nouns with frozen prvs

Form Noun Verb

✶
a-
a-√-∅ a-lag- ‘place, position’ (‘is arranged on it’) a-lag-eb-s ‘arranges’
a-√-il- a-dg-il- ‘place” (‘is put on it’) a-dg-am-s ‘puts on’
a-√-o a-s-o ‘(body) limb, member; letter’ (‘is affixed to it’) a-sv-am-s ‘sets on, affixes’
✶
i-
i-√-∅ i-gav- ‘parable, fable, riddle’ (‘it resembles’) h-gav-s ‘resembles’
i-k’ank’el- ‘zigzag line” (‘it shivers’) k’ank’al-eb-s ‘shivers’
i-√-al- i-dum-al- ‘secret, unspoken’ (‘is kept silent’) dum-s ‘is silent’
i-pk-l- ‘autumn-sown wheat’ (‘is ground [into flour]’) pkv-av-s ‘grinds’

In a recent essay on the origins of the Kartvelian category of version, I argue that it
emerged from an older distinction between the primary prvs ✶a- and ✶i-, which was
correlated with the trajectory or orientation of the action denoted by the verb (Tuite
2021). The prv ✶a- was primarily associated with locative or superessive meaning—
situating an event on a surface or target—from which emerged its link to transitiv-
ity, in the sense of action directed toward a goal (§ 2.3.3). Jorbenadze (1983: 115–122)
characterized the core meaning of the prv ✶i- as “reflexivity” or “turning back” (uk’uk-
cevitoba), as reflected in the clusters of meanings linked to this prefix: (i) attributes
associated with the middle voice, such as intransitivity, reflexivity, passive/antipassive;
(ii) “introversion” as understood by M. Mach’araviani (1987), that is, orientation toward
either the grammatical subject (subjective version) or a speech-act participant (1st or
2nd person objects of benefactive applicatives). The prvs thus indicate the trajectory
of the denoted activity vis-à-vis the referent of the subject, and secondarily, the speech-
act participants. The primary contrast of (intro-/extra-vert) trajectory also has implica-
tions for the animacy of the participant toward which the trajectory is oriented, and the
valence of the associated verb, as summarized below (Table 14):

Table 14: Contrasting characteristics of prvs ✶a- and ✶i-.

✶
a- ✶
i-
Trajectory subject → affected surface subject (orientation toward grammatical
(superessive, transitive; extraversion) subject, speech context; introversion)
Animacy inanimate human
Valence adds argument (superessive, causative) replaces overt actant with implicit reflexive

The secondary prvs ✶u- and ✶e- can be considered specialized alternants of ✶i- in particu-
lar contexts. The prv ✶e- marks the addition of a dative-case argument to an ✶i-medial
or ✶i-intransitive verb: i-cin-i-s ‘laughs’ > e-cin-i-s ‘laughs at her/him/them’. The prv ✶u-,
which signals a 3rd-person non-reflexive argument for benefactive applicative verbs,
938 Kevin Tuite

could be characterized as “indexically creative” in Silverstein’s (1976) sense. Whereas

✶
i- is linked to the highly-presupposable referents associated with the speech context
(speaker and addressee), and the content of the utterance (the grammatical subject), the
prefix ✶u- entails the addition of a new argument outside of this circle.

6 Conclusion
Two types of applicatives can be ascribed to the Kartvelian languages: benefactive and
superessive. Here is a summary of their principal characteristics:

Morphology
The morphological marker associated with both types of applicatives is a preradical
vowel (prv), which intervenes between the person prefixes and the verb root. In the case
of benefactive applicatives, the prefix is ✶i- with a 1st- or 2nd-person indirect object, and
✶
u- in the 3rd person. This alternation cannot be explained on phonological grounds,
and might reflect the semantic feature of introversion (§ 5). Applicativized verbs have
the same inflectional paradigms as their base counterparts, but the distinction between
the basic construction and the two types of applicatives is neutralized in the Series III
tenses (present-perfect, pluperfect, etc.; § 2.4).

Syntax
– Both types of applicatives are “D-applicatives”, which add an indirect object to
the construction. Kartvelian applicatives can combine with other valence-altering
transformations, such as causative and intransitivization. Double applicatives are
possible under certain restrictions.
– The superessive, unlike the benefactive applicative, can bring about morphological
changes other than the addition of a prv (§ 2.3.1). It is also less strict with respect
to the marking of applied objects, which not infrequently appear as objects of post-
positions rather than indirect objects in the dative case (Boeder 1968: 112; Aronson
1982: 75; Kojima 2012: 230). This is especially common in verbs that would other-
wise have two indirect objects (§ 3.3).

Semantics
The Kartvelian applicative types have semantic associations, which are reflected in
the names assigned to them: benefactive and superessive. Although applicative con-
structions can be paraphrased by basic constructions with postpositional phrases—
and indeed, this becomes a necessity in the Series III tenses, for which distinctions of
applicativity are neutralized (§ 2.4)—subtle contrasts between applicative and basic
constructions have been noted (§ 2.4).
 26 Applicative constructions in Kartvelian 939

Lookalikes
Kartvelian subjective version has the morphological attributes of an applicative, in that it
is signalled by the addition of a prv, but it has the same surface valence as the basic con-
struction. This has led some linguists to analyze it as a type of applicative with an implicit
reflexive indirect object, whereas others compare it to the middle voice (§ 4.1). Unmarked
or satanao version, by contrast, has the syntactic attributes of an applicative, but not the
morphology, since it is not associated with a prv or any other marker (§ 4.2). Valency-neu-
tral lexically-specified prvs occur in particular classes of verbs, which were presented
above. Of particular relevance for understanding the evolution of applicative morphology
in Kartvelian are transitive verbs with basic forms containing the prvs ✶a- and ✶i-, as well
as a handful of vowel-initial nouns with what appear to be frozen prvs, discussed in § 5.

Abbreviations
add additive
aor aorist
caus causative
dat dative
erg ergative
excl exclusive
ext extension
gen genitive
impf imperfect
incl inclusive
indevid indirect evidential
inst instrumental
intr intransitive
nom nominative
O object
obl oblique
opt optative
pass passive
perf perfect
pl plural
poss possessive
prs present
prv preradical vowel
pst past
pvb preverb
quot quotative
S subject
sg singular
sm series marker
TK Topuria & Kaldani (2000)
940 Kevin Tuite

References
Aronson, Howard I. 1982. On the status of version as a grammatical category in Georgian. Folia Slavica
5(1–3). 66–80.
Boeder, Winfried. 1968. Über die Versionen des georgischen Verbs. Folia Linguistica 2. 82–152.
Boeder, Winfried. 2003. The South Caucasian languages. Lingua 115(1-2). 5–89.
Boeder, Winfried. 2021. Grammatical sketch of Georgian. Unpublished manuscript, version 5.8.21a.
Bondarenko, T. I. 2015. Bitranzitivnye glagoly i applikativnye konstrukcii ot perexodnyx glagolov v
gruzinskom jazyke [Ditransitive verbs and applicative constructions from transitive verbs in Georgian].
In E. A. Ljutikova, A. V. Cimmerling & M. B. Konoshenko (eds.), Tipologija morfosintaksičeskix parametrov
[Typology of morphological parameters: Proceedings from the international conference], Vol. 2,
25–46. Moscow: MPGU Press.
Chikobava, Arnold. 1936. č’anuris gramat’ik’uli analizi t’ekst’ebiturt [A grammatical analysis of Laz, with texts].
Tbilisi: Mecniereba. (Reprinted in Šromebi III.)
Comrie, Bernard. 2008. Linguistic diversity in the Caucasus. Annual Reviews in Anthropology 37. 131–143.
Daniel, Michael & Yuri Lander. 2011. The Caucasian languages. In Johan van der Auwera & Bernd Kortmann
(eds.), The languages and linguistics of Europe: A comprehensive guide, 125–157. Berlin: Mouton de
Gruyter.
Deeters, Gerhard. 1930. Das kharthwelische Verbum: vergleichende Darstellung des Verbalbaus der südkauka-
sischen Sprachen. Leipzig: Markert und Petters.
Fähnrich, Heinz & Zurab Sarjveladze. 2007. Kartwelisches etymologisches Wörterbuch. Leiden: Brill.
Gérardin, Hélène & Alexander Rostovtsev-Popiel. 2016. Valency and valency change in the Caucasus. Paper
presented at the Higher School of Economics, Moscow, 28–30 November.
Gippert, Jost. 2008. Endangered Caucasian languages in Georgia. In David K. Harrison, David S. Rood &
Arienne Dwyer (eds.), Lessons from documented endangered languages, 159–194. Amsterdam: John
Benjamins.
Harris, Alice C. 1981. Georgian syntax: A study in relational grammar. Cambridge: Cambridge University Press
Harris, Alice C. 1990. Georgian: A language with active case marking. Lingua 80. 347–365.
Harris, Alice C. 1991. On the history of the Kartvelian languages. In Alice C. Harris (ed.), The indigenous
languages of the Caucasus: Kartvelian, 7–83. Delmar, N.Y.: Caravan Press.
Holisky, Dee Ann. 1991. Laz. In Alice C. Harris (ed.), The indigenous languages of the Caucasus: Kartvelian,
395–472. Delmar, N.Y.: Caravan Press.
Jorbenadze, Bessarion. 1983. zmnis xmovanp’repiksuli c’armoeba kartulši [Vowel-prefix formation of the verb
in Georgian]. Tbilisi: Tbilisi State University Press.
Kajaia, Otar. 2004. Megrul-kartuli leksik’oni [Mingrelian-Georgian Dictionary]. Tbilisi: Nek’eri.
Kavakli, Nurdan. 2015. Novus Ortus: The awakening of Laz language in Turkey. Idil Journal of Art and
Language, 4(16). 133–146.
Kemmer, Suzanne. 1993. The middle voice. Amsterdam: John Benjamins.
Klimov, G. A. 1964. Etimologičeskij slovar’ kartvel’skix jazykov [Etymological dictionary of the Kartvelian
languages]. Moscow: Nauka.
Klimov, G. A. 1991. Some thoughts on Indo-European-Kartvelian relations. Journal of Indo-European Studies
19. 325–348.
Klimov, Georgij A. 1998. Etymological dictionary of the Kartvelian languages. Berlin: Mouton de Gruyter.
Kojima, Yasuhiro. 2012. Version and object marking in Georgian verbs. In Wataru Nakamura & Ritsuko
Kikusawa (eds.), Objectivization and subjectivization: A typology of voice systems, 221–235. Suita, Osaka:
National Museum of Ethnology.
Kutscher, Silvia, Johanna Mattissen & Anke Wodarg. 2012. Das Muťafi-Lazische (Arbeitspapier Nr. 24, Neue
Folge). Cologne: Institut für Sprachwissenschaft der Universität zu Köln.
 26 Applicative constructions in Kartvelian 941

Lacroix, René. 2007 The preroot vowel a- in the Laz verb. Unpublished manuscript, Université Lumière
Lyon 2.
Lacroix, René. 2009. Description du dialecte laze d’Arhavi: Grammaire et textes. Lyon: Université Lumière Lyon
2 dissertation.
Lacroix, René. 2011. Ditransitive constructions in Laz. Linguistic Discovery 9(2). 78–103.
Lomashvili, Leila. 2010. Morphosyntax of complex predicates in South Caucasian languages. Tucson: University
of Arizona dissertation.
Mach’avariani, Maia. 1980. kcevis k’at’egoriis sak’itxisatvis [On the category of version]. Iberiul-k’avk’asiuri
enatmecniereba 22. 39–66.
Mach’avariani, Maia. 1987. Kcevis gramat’ik’uli k’at’egoriis semant’ik’a [The semantics of the grammatical
category of version]. Tbilisi: Mecniereba.
Mach’avariani, Givi. 1988. k’ausat’ivis k’at’egoria kartvelur enebši [The category of the causative in the
Kartvelian languages]. Iberiul-k’avk’asiuri enatmecniereba 27. 62–106
Makharoblidze, Tamar. 2012. The Georgian verb (Studies in Caucasian Linguistics 20). Munich: LINCOM.
Manaster Ramer, Alexis. 1995. On ‘Some thoughts on Indo-European-Kartvelian relations’. Journal of
Indo-European Studies 23. 195–208.
Marr, N. Ja. 1925. Grammatika drevneliteraturnogo gruzinskogo jazyka [Grammar of the Old Georgian literary
language]. (Materialy po jafetičeskomy jazykoznaniju XII [Materials on Japhethic linguistics XII]).
St. Petersburg: Akademija.
Marr, N., & M. Brière. 1931. La langue géorgienne. Paris: Firmin-Didot.
McGinnis, Martha. 2004. Lethal ambiguity. Linguistic Inquiry 35(1). 47–95.
Nash, Léa. 2017. The structural source of split ergativity and ergative case in Georgian. In Jessica Coon,
Diane Massam & Lisa Travis (eds.). The Oxford handbook of ergativity, 175–200. Oxford: Oxford
University Press.
Öztürk, Balkiz. 2013. Low, high and higher applicatives: Evidence from Pazar Laz. In Victoria Camacho-
Taboada, Ángel L. Jiménez-Fernández, Javier Martín-González & Mariano Reyes-Tejedor (eds.),
Information structure and agreement, 275–295. Amsterdam: John Benjamins.
Öztürk, Balkiz. 2016. Applicatives in Pazar Laz. Paper presented at the South Caucasian Chalk Circle,
University of Chicago Research Center, Paris. 22 September 2016.
Öztürk, Balkız. 2021. Transitive unergatives in Pazar Laz. Glossa 6(1). 1–24.
Peterson, David. 2007. Applicative constructions. Oxford: Oxford University Press.
Q’ipšidze, I. Grammatika mingrel’skago (iverskago) jazyka s xrestomatieju i slovarem [Grammar of the
Mingrelian (Iberian) language with chrestomathy and a dictionary]. St. Petersburg: Akademia Nauk.
Rostovtsev-Popiel, Alexander. 2015. Subjects of decreased control in Kartvelian anticausatives. Paper
presented at Diversity Linguistics: Retrospect and Prospect, University of Mainz / MPI EVA, May 1–3.
Schmidt, Karl Horst. 1965. Indogermanisches Medium und Sataviso im Georgischen. Bedi Kartlisa 19-20.
129–135.
Shanidze, Ak’ak’i. 1925/1981. kartuli zmnis sakcevi [The category of version of the Georgian verb], reprinted
in Txzulebani, II. kartuli enis st’rukt’urisa da ist’oriis sak’itxebi [Collected works, II. Issues in the structure
and history of the Georgian language], 375–401. Tbilisi: Mecniereba.
Shanidze, Ak’ak’i. 1953/1980. kartuli gramat’ik’is sapudzvlebi, I: morpologia [The fundamentals of Georgian
grammar I: morphology]. Tbilisi: Tbilisi State University Press.
Silverstein, Michael. 1976. Shifters, linguistic categories, and cultural description. In Keith Basso & Henry A.
Selby (eds.), Meaning in anthropology, 11–55. Albuquerque: University of New Mexico Press.
Singer, Kora. 2003. On double dative constructions in Georgian. In Dee Ann Holisky & Kevin Tuite (eds.),
Current trends in Caucasian, East European and Inner Asian linguistics. Papers in honor of Howard I.
Aronson, 349–362. Amsterdam: John Benjamins.
Topuria, Varlam 1947/2002. xmovantavsartovani saxelebi [Vowel-prefixed nouns]. In Varlam Topuria,
Šromebi III [Works III], 92–101. Tbilisi: Kartuli Ena.
942 Kevin Tuite

Topuria, Varlam. 1967. svanuri ena, 1: zmna [The Svan language, 1: Verb]. Tbilisi: Mecniereba.
Topuria, Varlam & Maksime Kaldani. 2000. svanuri leksik’oni. [Svan dictionary]. Tbilisi: Kartuli Ena.
Tuite, Kevin. 2002. Deponent verbs in Georgian. In Wolfram Bublitz, Manfred von Roncador & Heinz
Vater (eds.), Philologie, Typologie und Sprachstruktur. Festschrift für Winfried Boeder zum 65. Geburtstag,
375–389. Frankfurt am Main: Peter Lang.
Tuite, Kevin. 2004. Early Georgian. In Roger D. Woodard (ed.), Cambridge Encyclopedia of the World’s Ancient
Languages, 967–987. Cambridge: Cambridge University Press.
Tuite, Kevin. 2008. The rise and fall and revival of the Ibero-Caucasian hypothesis. Historiographia Linguistica
35(1). 23–82.
Tuite, Kevin. 2017. Alignment and orientation in Kartvelian (South Caucasian). In Jessica Coon, Diane
Massam & Lisa Travis (eds.), The Oxford handbook of ergativity, 1114–1138. Oxford: Oxford University
Press.
Tuite, Kevin. 2021. On the origin of Kartvelian version. (Studies in Caucasian Linguistics 22). Munich: LINCOM.
Uslar, P. K. 1861/1887. Grammatičeskij očerk svanetskago jazyka [Grammatical sketch of the Svan language].
Ètnografija Kavkaza. Jazykoznanie, I. Abxazskij jazyk. Otdel II [Ethnography of the Caucasus. Linguistics, I.
The Abkhaz Language], 103–120. Tbilisi: Upravlenie Kavkazskago Učebnago Okruga.
Vogt, Hans. 1974. Notes d’étymologie géorgienne. Norsk tidsskrift for sprogvidenskap 28. 103–111. (Reprinted
in Studia Caucasologica II. 513–521.)
Xubua, Mak’ar. 1976. Megruli t’ekst’ebi [Mingrelian texts]. Tbilisi: Tbilisi State University Press.
Zúñiga, Fernando. 2006. Deixis and alignment: Inverse systems in indigenous languages of the Americas.
Amsterdam: John Benjamins.
Guillaume Jacques, Aimée Lahaussois
27 Applicative derivations in Kiranti
Abstract: Among the languages of the Trans-Himalayan family, the Kiranti subgroup
stands out in terms of the richness of its applicative morphology. Most Kiranti languages
have both an inherited valency-increasing ✶-t suffix with applicative and causative func-
tions (with cognates in the rest of the family) and innovated bipartite verb constructions
with one or several applicative auxiliaries. This chapter illustrates the general properties
of applicative constructions in Kiranti. It shows that both ✶-t applicatives and bipartite
verbs involve a specific set of redundant stem alternations in some languages, in par-
ticular Khaling, and that one language, Hayu, has developed a specific applicative conju-
gation class. The applied object is generally indexed as the direct object of a monotran-
sitive verb, but it can receive the dative marker in some languages, like Thulung. All
Kiranti languages have at least one benefactive applicative construction, but we also
find applied objects with the semantic roles of patient, source, stimulus and instrument.

1 Introduction
AC’s have been reported in at least eleven of the thirty-odd subgroups of the Trans-Him-
alayan (or Sino-Tibetan) family.1 Given the size and typological diversity of the
Trans-Himalayan family (Arcodia & Basciano 2020), it is counterproductive to attempt
a comprehensive overview of all these subgroups in this chapter, and we restrict our
discussion to Kiranti languages, with a focus on Khaling (khal1275), a language spoken
by more than 10000 speakers in Solukhumbu, Eastern Nepal.2
The chapter first presents background information on Kiranti languages, and on
morphosyntactic constructions relevant to the description of the applicatives, includ-
ing case marking, person indexation, stem alternation and complex predicates. It then
describes the two applicative constructions found in Kiranti languages, the -t suffix and
the applicative auxiliaries, which are fused with the verb stem as a bipartite verb in
most languages of the subgroup.
Then follows a description of the syntactic constructions where applicative con-
structions are used, including the effects of this valency-increasing derivation with

1 Branches of Trans-Himalayan where AC’s have been described include Gyalrongic (Sun 2006, Jacques
2013, Zhang 2020), Old Chinese (Downer 1959, Jacques 2016), Dulong-Rawang (LaPolla 2000: 304–308),
Jinghpo (Peng & Chappell 2011), Kuki-Chin (Peterson 2007), Karbi (Konnerth 2014: 260–261), Tani (Post
2007: 521), Dhimal (King 2009: 198–200), Kham (Watters 2002: 249), in addition to Kiranti.
2 Our Khaling teacher is Dhan Bahadur Rai, coauthor of several publications on the Khaling language,
including Jacques et al. (2012, 2016, 2015).

https://doi.org/10.1515/9783110730951-027
944 Guillaume Jacques, Aimée Lahaussois

verbs roots of various valencies, the use of applicative with complement clauses, and its
combination with other voices and derivations.
The next section comprises an inventory of the semantic roles that the applied
object (ApplO) can receive in the various applicative constructions in Khaling and other
Kiranti languages.
Finally, the chapter concludes with a brief account of verbal lability, which has
applicative-like semantics in some languages.

2 Background information
2.1 Kiranti languages

Kiranti is a subgroup of the Trans-Himalayan family, comprising about 30 languages,

spoken in Eastern Nepal and neighbouring Sikkim.
Although the monophyly of the subgroup has been questioned (Gerber & Groll-
mann 2018), there are potential lexical, phonological and morphological innovations
defining Kiranti (Michailovsky 2017, Jacques 2017) and this clade is also supported by
Bayesian phylogenies (Sagart et al. 2019).
This section focuses on a number of morphosyntactic features that are needed to
discuss applicative constructions in Kiranti languages: case marking, person indexa-
tion, stem alternation, and complex predicates (in particular bipartite verbs).

2.2 Case marking

All Kiranti languages mark transitive subjects with an ergative case and use an unmarked
absolutive form for intransitive subjects. The ergative also serves to mark instruments.
Direct objects are generally in absolutive form as in (1a), but some languages, including
Bantawa (bant1281, Doornenbal 2009: 211), Puma (puma1239, Sharma 2014: 293–308)
or Thulung (thul1246, Lahaussois 2002: 66–68), mark animate objects with the dative
-lai borrowed from Nepali (1b).

(1) Bantawa (Doornenbal 2009: 229)

a. seluwa-ʔa kʰaŋ kʰɨk-mett-u.
bitter.herb-erg vegetables be.bitter-caus-3O
‘The herbs made the vegetables bitter.’
b. seluwa-ʔa o ŋa-lai kʰɨkt-u.
bitter.herb- erg this fish-dat numb-3O
‘The herbs numbed the fish.’
 27 Applicative derivations in Kiranti 945

Beneficiaries can be encoded by a benefactive applicative (§5), but also simply by means
of a postposition. In Khaling (khal1275) for instance, either lagi ‘for’ (borrowed from
Nepali) or dɵ̂l ‘for’ (in higher register) combines with the genitive to mark beneficiaries,
both animate and inanimate (2).

(2) Khaling
kʰɛ̂l brâː-po dɵ̂l dʉk mɛ
Khaling language-gen for suffering do:2sg → 3:n.pst
‘He is working hard for the Khaling language.’

The ablative -lʌkʌ is also attested to mark the beneficiary in applicative constructions
involving transitive or ditransitive bases (§4.1.3).

2.3 Person indexation

Kiranti languages distinguish singular, plural and dual numbers, and have a robust
clusivity contrast. They encode transitivity in their verbal morphology, and transitive
verbs can index the person and number of two arguments, with the possible exception
of a few varieties which have undergone rapid morphological simplification in recent
decades (Borchers 2008).
Ditransitive verbs generally index the recipient as direct object rather than the
theme, as illustrated by (3) with the verb |butt| ‘give to drink’.

(3) Khaling (Jacques et al. 2015)

ʔʌ̄m-ʔɛ ʔûŋ ku ʔi-bʉs-tʌ.
3sgerg 1sg water 2/inv-give-1sg:pst
‘He gave me water to drink.’

In order to describe specific slots in the polypersonal paradigm in a condensed fashion,

transitive configurations are presented in the form X → Y, where X corresponds to the
transitive subject and Y to the object. For instance, the person configuration in (3) is
3SG → 1SG; the verb form ʔi-bʉs-tʌ ‘He gave it to me to drink’ indexes a third singular
subject with first singular object.
Kiranti morphology is mainly suffixal, but Puma for instance allows up to three
prefixes (Sharma 2014: 135), as shown in (4).

(4) Puma (Sharma 2014: 149)

khokkuci-a ke-lai kha-ni-pʌ-sont-en
3pl-erg 1du-dat 1n.sg.O-3S/A-neg-persuade-neg;pst
‘They did not persuade us.’
946 Guillaume Jacques, Aimée Lahaussois

Some languages such as Khaling only have a single prefixal slot, which can be filled by
a negative prefix, or by the prefix ʔi-, which occurs in all forms involving second person
as subject or object (except the 1SG → 2) and in the inverse 3 → 1 forms, hence the gloss
2/INV in (3).
There is some interaction between person indexation and tense-aspect-modality
(TAM), and in most languages, distinct paradigms have to be posited at least for past,
non-past and imperative tenses. The segmentation of forms into affixes and bases is not
always trivial, and requires decisions that are sometimes arbitrary (Ackerman et al.
2009). For instance, in the Khaling form lʉ̂ːtɛ ‘he said it’ (3SG→3SG.PST of |lut| ‘say’), it
is unclear whether the -t- belongs to the verb stem (lʉ̂ːt-) or the past tense suffix (-tɛ).3
Voice also affects person indexation in Kiranti. Reflexive-middle and applica-
tive-causative forms have special conjugations involving non-predictable morpholog-
ical alternations, and complete paradigms have to be listed in their entirety.

2.4 Stem alternations

The morphology of Kiranti verbs is not exclusively concatenative, and argument index-
ation, TAM and voice are not only encoded by affixes, but also by stem alternations,
which present considerable variety across the subgroup (Herce 2021).
The partial paradigms from Khaling in Table 3 are representative of the type of stem
alternations and affixation used to index person, number, tense and voice in Kiranti.
In the following discussion, verb forms are referred to using abstract root forms
from which the paradigms can be derived by a set of regular expressions (see Jacques et
al. 2012 and Jacques 2017 on the procedure to identify root forms in Khaling and other
languages). These roots, which almost never occur as independent forms, and are not
pronounceable as such, are written between vertical bars |. . .|.

2.5 Complex predicates

All Kiranti languages have complex predicates involving lexical verbs followed by
closed class of auxiliaries, with various degrees of morphophonological and morpho-
syntactic integration. In the following, lexical/main verbs are referred to as V1, and
auxiliaries (also called ‘vector verbs’ or ‘aspectualizers’) as V2. The V2’s have various
functions, including aspect, associated motion (Jacques et al. 2021), but also causative

3 An anonymous reviewer objected that the past prefix was -ɛ, cognate to the -a past suffix found in other
Kiranti languages. However, from the point of view of the synchronic analysis, this would amount to
positing a suffix -ɛ, in complementary distribution with -tɛ, only appearing after verb roots with a -t coda.
 27 Applicative derivations in Kiranti 947

and applicative. A preliminary discussion of the general morphological properties of

complex predicates is necessary before presenting the applicative V2’s.
In some languages such as Limbu, the complex predicates are serial verb construc-
tions, both verb forms remaining phonologically distinct, and identical to independent
forms (5).

(5) Limbu (van Driem 1987: 129)

a-ndzum-ille yəllik samaːn-haʔ khuːtt-aŋ
1sg.poss-friend-erg much belongings-pl rob-1SG:S/O:PST
piːr-aŋ
give-1sg:S/O:pst
‘My friend robbed me of many belongings.’

Bantawa V1’s and V2’s are represented as distinct words in Doornenbal’s grammar, but
they do not entirely behave like independent words, since prefixes (for instance second
person tɨ- in 6) only appear on the V1, and some of the indexation suffixes (as first
person exclusive -ka) only occur on the V2.

(6) Bantawa (Doornenbal 2009:168)

tɨ-man-nin kʰan-nin
2-lose-1ns↔2 send.away-1ns↔2
‘Yousg have forgotten uspl.excl’

(7) Bantawa (Doornenbal 2009: 254)

kʰar-in lont-in-ka
go-1/2pl come.out-1/2pl-excl
‘We shall rise again.’

Rather than being serial verb constructions as in Limbu, these complex predicates are a
subtype of bipartite verbs, as each of the verbal roots only has partial phonological and
morphological autonomy (Jacques 2018, Lahaussois 2020).
In Western and Northern Kiranti languages, such as Khaling, the auxiliaries have
become considerably more integrated and are phonologically fused with the lexical
verb, as illustrated in the partial paradigms in Table 1.
In Khaling bipartite verbs, both V1 and V2 conjugate for person, number and tense,
and indexation suffixes can surface as codas when the root of the lexical verb is an open
syllable as in the case of |dza| ‘eat’. For instance, the 1PL.INCL suffix -ki in its reduced
form -k- is inserted between the V1 dzɵ- and the V2 -kʰʌ-, and person/number is thus
redundantly marked two times, an example of multiple exponence. Similar phenomena
are found in other Kiranti languages (see in particular Lahaussois 2020 on Thulung).
In the case of closed syllable verb stems, the indexation suffixes cannot surface as
independent segments, but the stem alternations they cause are still visible in some
948 Guillaume Jacques, Aimée Lahaussois

Table 1: Comparison of the paradigm of |dza|

‘eat’ with that of the compound verb |dza-kʰʌ|
‘eat completely; eat and go’ in Khaling.

Form |dza| ‘eat’ |dza-kʰʌ|

1SG→3SG dzʌ-ŋʌ dzʌ̂-ŋ-kʰʌt-u
1DU.INCL→3 dzɵ-ji dzɵ̂-j-kʰɵts-i
1PL.INCL→3 dzɵ-ki dzɵ-k-kʰʌ-ki
2SG→3 ʔi-dzɛ ʔi-dzɛ-kʰʌt-ʉ

cases. For instance, in (8), the verb stem -lʌ̂m- from |lipt| ‘light a fire’ shows nasaliza-
tion of the coda -p to -m (with a falling tone), as a consequence of the merger of -p with
the 1SG -ŋ(ʌ), which does surface here as a distinct coda. This form also illustrates the
fact that in Khaling, as in Bantawa, there is a single prefixal slot for the bipartite verb.4

(8) Khaling (Jacques et al. 2015)

ʔʌ̄m -ʔɛ mi-ʔɛ ʦuroʈ ʔi-lʌ̂m -sʌ-ŋʌ.
3sg-erg fire-erg cigarette 2/inv-light.up:1sg-appl:n.pst-1sg
‘He lights a cigarette for me.’

3 Morphology
There are two distinct applicative morphemes in Kiranti languages: an ancient ✶-t tran-
sitivizing suffix, which is fused with the verb roots and displays extensive morphopho-
nological alternation, and the more recent applicative auxiliaries, which in most Kiranti
languages form bipartite verbs with the root of the verb to which they are attached.

3.1 -t causative-applicative

The suffix ✶-t serves to build applicative verbs in most, if not all, Kiranti languages.
This section first presents the polyfunctionality of this suffix, then focuses on the stem
alternations that occur in the paradigms of suffixed verb roots, and describes the dis-
tinct conjugation classes that came into being in some languages through a reflex of this
suffix.

4 Note the instrumental use of the ergative form mi-ʔɛ ‘by/using fire’ in (8).
 27 Applicative derivations in Kiranti 949

3.1.1 Applicative and causative suffixes in Kiranti

A pair of valency-increasing dental suffixes ✶-t and ✶-s are reconstructible to proto-
Kiranti (Michailovsky 1985, Bickel et al. 2010, Michailovsky 2017, Jacques 2017), with cog-
nates in other branches of the Trans-Himalayan family (Wolfenden 1929), in particular
Dulong-Rawang (LaPolla 2000: 308), Old Chinese (Jacques 2016), Gyalrongic (Zhangshuya
2020: 161) and Jinghpo (Dai & Xu 1992:78).
The suffix ✶-t has reflexes in all Kiranti languages, but it is unclear whether it is
productive in any of them. In Khaling (khal1275), Bantawa (bant1281) and Yakkha
(yakk1236), although it is no longer productive with independent verbs, it appears in
the applicative bipartite verb construction, which is fully productive (§3.2.2).
In most languages, it is only found in a limited number of verbs, with either caus-
ative or applicative function. A second suffix ✶-s, which only has a causative function,
has been lost in most Kiranti languages, but preserved in Southern (Bantawa bant1281,
Doornenbal 2009: 230) and Eastern Kiranti (Limbu limb1266, van Driem 1987: 245-265,
Chintang chhi1245, Bickel et al. 2010).
These two suffixes can be illustrated with the following examples from Limbu
(Michailovsky 1985).

(9) Limbu
a. |haːp|‘weep’ (intransitive, base form)
haːb-ɛ
weep-pst
‘S/he was weeping/wept.’
b. |haːpt|‘mourn, weep for’ (transitive, applicative)
haːp-t-u
weep-appl-3:O
‘S/he is/was mourning him/her.’
c. |haːps| ‘cause to weep’ (transitive, causative)
haːp-s-u
weep-caus-3:O
‘S/he makes/made him/her cry.’

In addition to the applicative function illustrated in (9b), the -t suffix also has a causa-
tive function with some verb bases, in particular motion verbs (Table 2). These forms
are found in all Kiranti languages.
It is possible that motion verbs are the pivot through which the originally applica-
tive prefix was reinterpreted as a causative, as meanings such as ‘bring’ or ‘take’ can be
derived from ‘come’ and ‘go’ through both a comitative applicative (10) or a causative (11).
950 Guillaume Jacques, Aimée Lahaussois

Table 2: The causative function of the -t suffix in motion verbs in Limbu.

|thaŋ| ‘come up’ |thak-t| ‘bring up’

|yu| ‘come down’ |yuː-t| ‘bring down’
|phɛn| ‘come (same level)’ |phɛt-t| ‘bring (same level)’
|ta| ‘come’ |taː-t| ‘bring’

(10) ‘come with X’ → ‘bring’

(11) ‘cause X to come’ → ‘bring’

There is also an unrelated formative -t occurring on intransitive verbs, which

Michailovsky (2017) refers to as ‘deponent’ verbs. Its only identifiable derivational func-
tion is denominal verbalizing (Jacques 2017).

3.1.2 Applicative and stem alternations

Due to language-specific morphophonological alternations, the applicative ✶-t suffix is

not realized as a distinct segment in all the slots of the paradigms, and its presence
is sometimes only indirectly detectable from stem alternations. In Khaling (khal1275)
for instance, the Kiranti language with the highest number of verb stem alternations
(Jacques et al. 2012), the contrast between applicative conjugations and the correspond-
ing monotransitive conjugations is only visible in forms combining a first or second
person singular subject with a third person object (1sg→3, 1sg→3) and in non-local con-
figurations (when both arguments are third person). In all other configurations, includ-
ing inverse (3→1, 3→2), local (1→2, 2→1), and direct forms with a dual or first/second
plural transitive subject, applicative forms cannot be distinguished from monotransi-
tive conjugations, as illustrated in Table 3, where ambiguous forms are shaded in grey.
The applicative suffix surfaces as -d- or -t- in non-past forms with a singular subject
and third person object. In the past tense, which is marked by a suffix -t- , the applicative -t
cannot be realized as a segment, but triggers the ‘strong’ stem in singular subject forms with
a falling tone (in the case of |kur| ‘carry’, the strong stem is kʌr- , and the weak stem kʉr- ).

Table 3: Comparison of monotransitive and applicative paradigms

in Khaling (selected forms).

Form |kur| ‘carry’ |kurt| ‘bring for’

1sg3→sg.n.pst kur-u kʌ̄rd-u
1du.incl→3sg.n.pst kʉr-i kʉr-i
1pl.incl→3sg.n.pst kʌ̄r-ki kʌ̄r-ki
2sg→3sg.n.pst ʔi-kʉ̄ːr-ʉ ʔi-kʌ̄rd-ʉ
 27 Applicative derivations in Kiranti 951

Table 3 (continued)

Form |kur| ‘carry’ |kurt| ‘bring for’

2pl→3sg.n.pst ʔi-kʌ̄r-ni ʔi-kʌ̂r-ni
3sg→3sg.n.pst kʉːr-ʉ̄ kʌ̄rd-ʉ̄
1sg→3sg.pst kur-u-tʌ kʌ̂r-tʌ
1du.incl→3sg.n.pst kʉr-iti kʉr-iti
1pl.incl→3sg.n.pst kʌ̄r-tiki kʌ̄r-tiki
2sg→3sg.pst ʔi-kʉ̂r-tɛ ʔi-kʌ̂r-tɛ
2pl→3sg.pst ʔi-kʉr-tɛ-nu ʔi-kʉr-tɛ-nu
3sg→3sg.pst kʉ̂r-tɛ kʌ̂r-tɛ

3.1.3 Coda alternations

In addition to the vocalic and tonal alternations illustrated above, the codas of some
verb roots undergo alternation when combined with the applicative suffix. In Khaling,
the only such alternation involves the assimilation of -ŋ to -n- (Jacques 2015).5
Bantawa has the greatest number of alternations of this type (Doornenbal 2009:
233), illustrated in Table 4: -s assimilates to t , -r and -l merge with n before t or assimi-
late to t , and some nasal codas are denasalized. These alternations are not regular, and
have exceptions.

Table 4: Examples of alternations in the coda of verb

roots triggered by the applicative -t in Bantawa.

Base verb Applicative verb

|t om| ‘dance’ (tr)
h
|thop-t| ‘dance for someone’
|kus| ‘heat’ (tr) |kut-t| ‘heat for someone’
|khur| ‘carry’ |khut-t| ‘carry for someone’

3.1.4 Applicative conjugation in Hayu

Hayu (wayu1241) is the only Kiranti language (and possibly the only Trans-Himala-
yan language) in which the -t applicative is compatible with the majority of verb roots
(Michailovsky 1988: 89), in the form of a special conjugation in non-dual direct forms (5),
which neutralizes the Past/Non Past contrast except in first person plural forms. As in
Khaling, base forms and applicative ones are identical in local and inverse configurations.

5 For instance, the motion verb |khoŋ| ‘come (upwards)’ has the causative |khoŋt| ‘bring (upwards)’,
whose 1sg→3sg.n.pst is khond-u ‘I bring it upwards’.
952 Guillaume Jacques, Aimée Lahaussois

Table 5: Selected forms of the applicative

person-indexing paradigm in Hayu.

Form base applicative

n.pst pst n.pst pst
1sg→3sg -ŋ/-N/-sʊŋ -kʊŋ -tʊŋ -tʊŋ
1pl.incl→3 -ke -kikeŋ -tike -tikeŋ
1pl.excl→3 -kok -kikoŋ -tikok -tikoŋ
2/3sg→3sg ∅ -ko -to -to
2/3sg→3pl -me -kome -tome -tome

Michailovsky (1988) analyzes this paradigm as a distinct conjugation class, and refrains
from segmenting the applicative -t- from the rest of the ending.

3.2 Applicative complex predicates

3.2.1 Applicative auxiliaries

All Kiranti languages for which a detailed description is available (except Hayu) have at
least one complex predicate construction (§2.5) with a applicative/benefactive V2 (Table 6).
Some Eastern Kiranti languages, Yakkha, Belhare and Yamphu, have two distinct
V2’s used in applicative constructions. In Yamphu for instance, the V2 -pett is the default
applicative, while -khitt only occurs with manipulation verbs, and specifically means
‘bring for’.

Table 6: Kiranti complex predicates with applicative function.

Language Applicative Source Reference

Wambule -gwakt |gwakt| ‘give’ Opgenort (2004: 424–426)
Jero -gɔkt |gɔkt| ‘give’ Opgenort (2005: 213)
Khaling -sʌ(t)/-sɵt |sott| ‘greet, pass to’
Dumi -khotnd- khotnnɨ ‘proffer’ van Driem (1993: 205)
Koyi -khond ? Lahaussois (2009: 19)
Thulung -sa(ʈ) Lahaussois (2002: 212–214)
Kulung -pi |pi| ‘give’ Tolsma (2006: 95–96)
Bantawa -pɨ | pɨ| ‘give’ Doornenbal (2009: 284–285)
Puma -idt |it| ‘give’ Sharma (2014: 299)
Camling -pid |it| ‘give’ Ebert (2017: 731)
Chintang -bid |pit| Paudyal (2015: 122)
-dhett
Yakkha -piʔ |piʔ| ‘give’ Schackow (2015: 297–302)
-ni Schackow (2015: 299)
Belhare -pir |pir| Bickel (2017: 710)
 27 Applicative derivations in Kiranti 953

Table 6 (continued)

Language Applicative Source Reference

-pett
Yamphu -pett Rutgers (1998: 178)
-khitt Rutgers (1998: 179)
Limbu -pi |pi| ‘give’ van Driem (1987: 128–129)

Most of these V2’s are transparently grammaticalized from verbs meaning ‘give’, and
are not cognate across Kiranti. In Thulung, Allen (1975) reports the existence of a verb
samu ‘give’ as the lexical source for the applicative V2, but it appears to be restricted
to complex predicates. In Dumi, both the applicative V2 and its source verb khotnnɨ
‘proffer’ have highly irregular alternations, and no root form can be reconstructed
(Michailovsky 2012). In Khaling, the V2 -sʌ(t)/-sɵt also presents complex alternations
between the stems -sʌ- , -sa- , -sʌt- , -sɵts- and -sɵs- , which are not found in any inde-
pendent verb form, and look like a patchwork from |-a|, |-ut| and |-ot| conjugation
classes. There is no historical explanation for this pattern, but it is also shared by other
V2’s such as the associated motion khʌ(t) ‘do X and go’ (Jacques et al. 2021), which origi-
nates from the verb |khot| ‘go’.
The fact that the verbs that were grammaticalized as applicative V2’s are not cognate
across all of Kiranti languages suggests that this construction is not reconstructible to
the proto-language, and developed independently in each of the subbranches of this
subgroup. It is possible that they arose as a calque of the benefactive construction in
Nepali, which involves the verb dinu ‘give’ (Pokharel 2005).
Most of the V2’s in Table 6 are dedicated applicative derivations, but some can also
be used as causatives. In Thulung for instance, the V2-sa(ʈ) has a permissive function
in (12).

(12) Thulung (Allen 1975)

luŋ-ka ne pi khlos-ta ma ʈhãu ɖi-saʈ-ɖʉtsi
stone-erg top ideo return-3sg.pst conj place leave-appl-3sg→3du.pst
‘A stone suddenly turned round and allowed them to leave.’

3.2.2 Double marking of applicative derivation

In Khaling, the applicative -sʌ(t) differs from other V2’s in that the V1 with which it
combines follows the conjugation of -t suffixed verb roots, even when the applicative -t
suffixed form is not attested as an independent verb. The applicative derivation is thus
doubly marked by the -t suffix and the V2.
The -t suffix is directly visible in the paradigms of open syllable verbs, where it
surfaces as -s-, -j- or -ç- , as illustrated in Table 7, due to a series of regular morphoph-
954 Guillaume Jacques, Aimée Lahaussois

onological rules (described in Jacques et al. 2012). The verb |phlo| ‘help’ lacks a corre-
sponding -t applicative verb form †|phlo-t|.
The applicative paradigms are complicated by the fact that some V1’s have pre-
served all expected stem alternations (for instance |mu| ‘do’ in Table 7), while other
V1’s have generalized the same stem to most of the paradigm. For instance, |phlo-(t)-
sʌ(t)| ‘help on X’s behalf’ has the stem form phlɵ- in nearly all forms. As it is unpredict-
able, the proportion of analogical forms has to be specified in the lexical entry of each
verb.
In closed syllable verb roots, the -t suffix never surfaces directly, but the presence
of the strong stems (Jacques et al. 2012) in singular subject forms (kʌr- in Table 8) indi-
cates that the paradigm of the V1 is based on that of the -t applicative (compare with the
conjugation of |kurt| ‘bring for’ in Table 3).

Table 7: Selected forms of the paradigm of the sʌ- applicative from |mu| ‘do’
and |phlo| ‘help’ in Khaling.

Form |mu-(t)-sʌ(t)| |phlo-(t)- sʌ(t)|

‘do on X’s behalf’ ‘help on X’s behalf’
1sg→3sg.n.pst mʌ-s-sʌt-u phlɵ-s-sʌt-u
1du.incl→3sg.n.pst mʉs-sɵts-i phlɵ-s-sɵts-u
1pl.incl→3sg.n.pst mʌ-ç-sʌ-ki phlɵ-s-sʌ-ki (phlɵɔ-ç-sʌ-ki)
2sg→3sg.n.pst ʔi-mʌ-s-sʌt-ʉ ʔi-phlɵ-s-sʌt-ʉ
2pl→3sg.n.pst ʔi-mʌ̂n-sʌ-ni ʔi-phlɵ̂n-sʌ-ni
3sg→3sg.n.pst mʌ-s-sʌt-ʉ phlɵ-s-sʌt-ʉ
2/3sg→1sg.n.pst ʔi-mʌ̂-j-sʌ-ŋʌ ʔi-phlɵɔ̂ -j-sʌ-ŋʌ
3sg→2sg.n.pst ʔi-mʌ-ç-sa ʔi-phlɵ-sa
1sg→3sg.pst mʌ-s-sʌt-ʌ phlɵ-s-sʌt-ʌ
1du.incl→3sg.n.pst mʉ-s-sɵs-ti phlɵ-sɵs-ti
1pl.incl→3sg.n.pst mʌ-ç-sʌ-ktiki phlɵ-s-sʌ-ktiki
2sg→3sg.pst ʔi-mʌ-s-sʌt-ɛ ʔi-phlɵ-s-sʌt-ɛ
2pl→3sg.pst ʔi-mʉ-s-sɵs-tɛ-nu ʔi-phlɵ-s-sɵs-tɛ-nu
3sg→3sg.pst mʌ-s-sʌt-ɛ phlɵ-s-sʌt-ɛ

Table 8: Selected forms of the paradigm of the

sʌ- applicative from |kur| ‘carry’ in Khaling.

Form |kur-(t)-sʌ(t)| ‘carry for X ’

1sg→3sg.n.pst kʌ̂r-sʌt-u
1du.incl→3sg.n.pst kʉr-sɵts-i
1pl.incl3→sg.n.pst kʌ̄r-sʌ-ki
2sg→3sg.n.pst ʔi-kʌ̂r-sʌt-ʉ
2pl→3sg.n.pst ʔi-kʌ̄r-sʌ-ni
3sg→3sg.n.pst kʌ̂r-sʌt-ʉ
 27 Applicative derivations in Kiranti 955

Table 8 (continued)

Form |kur-(t)-sʌ(t)| ‘carry for X ’

2/3sg→1sg.n.pst ʔi-kʌ̄r-sʌ-ŋʌ
3sg→2sg.n.pst ʔi-kʌ̄r-sa
1sg→3sg.pst kʌ̂r-sʌt-ʌ
1du.incl→3sg.n.pst kʉr-sɵs-ti
1pl.incl→3sg.n.pst kʌ̄r-sʌ-ktiki
2sg→3sg.pst ʔi-kʌ̂r-sʌt-ɛ
2pl→3sg.pst ʔi-kʉr-sɵs-tɛ-nu
3sg→3sg.pst kʌ̂r-sʌt-ɛ

Khaling is not the only language with double marking of applicative derivations. In
Yakkha, the V2 -piʔ requires the applicative suffix -t on the V1 when used in applicative/
benefactive function (Schackow 2015: 371). The V2 -piʔ lacks valency-increasing func-
tion when used without the -t suffix, and expresses either telicity or that ‘some partici-
pant is affected by the event in undesirable ways’, without introducing a new argument
(Schackow 2015:299). In such cases, an applicative construction with the transitivizer
-ni can be used as the applicative function of the intransitive -piʔ construction; compare
(13a) with (13b). This type of equipollent derivation is restricted to a handful of verbs.

(13) Yakkha (Schackow 2015: 299)

a. ka mund-a-by-a-ŋ=na.
1sg forget-pst-V2.give-pst-1sg=nmlz.sg
‘I was forgetful.’
b. muʔ-ni-nen=na.
forget-transitivizer-1→2=nmlz.sg
‘I forgot you.’

3.2.3 Applicative derivations from intransitive bases

When the -sʌ(t)/-sɵt applicative is applied to intransitive verbs in Khaling, their stem
also receives the additional -t. When these intransitive verbs already have a -t applica-
tive or causative counterpart, the -sʌ(t)/-sɵt applicative of the intransitive verb is iden-
tical to that of its transitive form. For instance, the motion verb |khot| ‘go’ and its caus-
ative |khott| ‘take’ have the same applicative form (Table 9).6

6 In addition, Table 7 illustrates the fact that V1 stems ending in dental obstruents originating from ✶t by
morphological alternations (-Vt/d/ts- , but not -ç-) assimilate to -Vs- when followed by the V2 -sʌ(t)/-sɵts .
956 Guillaume Jacques, Aimée Lahaussois

Table 9: Partial paradigm of |khott| ‘take’ and the applicative of

|khott| ‘take’ and |khot| ‘go’.

Form |khot| ‘go’ |khott| ‘take’ Applicative

1sg(→3sg.n.pst) k oɔ̂ j-ŋʌ
h
k oɔtt-u
h
khoɔs-sʌt-u
1du.incl(→3sg.n.pst) khɵts-i khɵts-i khɵs-sɵts-i
1pl.incl(→3sg.n.pst) khoɔç-ki khoɔç-ki khoɔç-sʌ-ki
2sg(→3sg.n.pst) ʔi-khoɔ̂ j ʔi-khoɔtt-ʉ ʔi-khoɔs-sʌt-ʉ
3sg(→3sg.n.pst) khoɔ̂ j khoɔtt-ʉ khoɔs-sʌt-ʉ

For instance, the form khoɔs-sʌt-u can either be interpreted as the applicative form of
khoɔtt-u ‘I will take it’ (14a) or that of the intransitive khoɔ̂j-ŋʌ ‘I will go’ (14b).7

(14) Khaling
a. kitap khoɔs-sʌt-u
book take-appl-1sg→3
‘I will take the book for him/her.’
b. kitap ʔʌ̄ŋ-bi khoɔs-sʌt-u
book buy:inf-loc go:appl-appl-1sg→3
‘I will go to buy the book for him/her.’

4 Syntax
While grammars of Kiranti provide extensive data on the morphology of applicative
constructions, their syntax remains imperfectly described, and this section mainly
focuses on Khaling.
Since beneficiaries can be marked by means of a postposition (§2.2), applicative
derivation selecting a beneficiary as ApplO (on which see §5) cannot be considered to be
obligatory. This is not the case with other types of semantic roles (including goal or stim-
ulus), for which no alternative construction is possible to express the same meaning
(see Creissels on Tswana, §3.1, this volume).

4.1 Transitivity of the base verb

4.1.1 Intransitive verbs

In all Kiranti languages, the -t suffix (§3.1) can derive applicative verbs from intran-
sitive verb roots. The intransitive subject of the BC (in the absolutive, see 15a) corre-

7 For an account of the intransitive conjugation, see Jacques et al. (2012).

 27 Applicative derivations in Kiranti 957

sponds to the transitive subject in the AC, marked in the ergative (15b),8 and the added
argument is the direct object of the AC.

(15) Khaling
a. ʌ̄m ŋʌ̄i
3sg be.afraid:n.pst:3sg
‘He is afraid.’
b. ʌ̄m-ʔɛ nɵ̂r ŋʌ̄n-d-ʉ
3sg-erg tiger be.afraid-appl-3sg→3:n.pst
‘He is afraid of the tiger.’

V2 applicatives (§3.2, §3.2.3) occur with intransitive bases in most Kiranti languages,
with the same reorganization of argument structure, as illustrated in (16).

(16) Khaling
a. tsɵttsɵ tsêr-tɛ
child piss-2/3sg:pst
‘The child pissed.’
b. tsɵttsɵ-ʔɛ ʔi-tsêr-sɵs-tʌ
child-erg 2/inv-piss-appl-1sg:pst
‘The child pissed on me.

In Limbu (van Driem 1987: 128) and Bantawa (Doornenbal 2009: 284), however, V2
applicatives are only found with transitive bases.

4.1.2 Transitive verbs

While in Eastern and Southern Kiranti languages, the -t applicative often occurs with
transitive bases, in Khaling, only one such example is attested: |kurt| ‘bring for’ from
|kur| ‘carry’ (Table 3).
The V2 -sʌ(t)- is used instead to build the applicative of all other transitive verbs.
For instance, the transitive verb |jok| ‘distribute, share’ selects as its object the entity
being distributed, and the only possibility to promote the people who receive the shares
from the distribution to argument status is by using the applicative V2 -sʌ(t), as in (17b).
Both the ApplO and the object of the BC (henceforth BO) are in absolutive form, but the
verb indexes the number of the ApplO.

8 Examples (15a) and (15b) illustrate the verb root s |ŋin| ‘be afraid’ and |ŋint| ‘be afraid of’, respec-
tively, with regular morphophonological alternations.
958 Guillaume Jacques, Aimée Lahaussois

(17) Khaling (Jacques et al. 2015)

a. ʔuŋʌ lēmpɛ jog-u.
1sg:erg sweet distribute-1sg→3:n.pst
‘I distribute sweets.’
b. ʔuŋʌ ʦɵʦʦɵ-ɦɛm lēmpɛ jok-sʌt-u-nu.
1sg:erg child-pl sweet distribute-appl-1sg→3:n.pst-pl
‘I distribute sweets to the children.’

In languages such as Bantawa and Thulung, which mark direct objects with the dative
suffix -lai (§2.2), both the ApplO (18) and the BO(19) can receive dative case.

(18) Thulung
go oram nem a-lwak-lai
1sg prox.dem house 1sg.poss-younger.sibling-dat
ɖi-saʈ-pu
leave-appl-1sg→3sg
‘I leave this house to my brother.’

(19) Thulung
go i:nima tsɵttsɵ-lai ɖʉlʉmtsa-ka jal-sa-nini
1sg 2pl.poss child-dat stick-instr strike-appl-1sg→2pl
‘I will strike your child for you with a stick.’

In Khaling, applicative verbs derived from a transitive base generally become ditransi-
tive, with both the BO and the ApplO in absolutive form, though the latter can optionally
be marked with the benefactive case when the applicative has a benefactive interpre-
tation (§2.2). The applied phrase is thus not completely assimilated to the grammatical
function of object.
For instance, in (20), the ApplO ʔʌ-tsɵ-su ‘my two sons’ can either be in absolutive
form or receive the complex marker -po lagi ‘for’.9

(20) Khaling
ʔʌ-tsɵ-su(-po lagi) kitap
1sg.poss-son-du(-gen for) book
ʔʌ̂n-sʌt-ʌ-su
buy:appl-appl-1sg→3sg.pst-du
‘I bought book(s) for my two sons.’

9 The ablative -lʌkʌ, which occurs in ditransitive constructions to mark the beneficiary (§4.1.3) is not
possible with this verb.
 27 Applicative derivations in Kiranti 959

Although the ApplO is most often indexed as direct object as in (17b),(20) and (21a),
when the BO is first or second person and the ApplO is third person, the former can be
indexed instead (as in 21b) due to its being higher in the person hierarchy.

(21) Khaling
a. ʔuŋʌ ʔi-tsɵ tʰʌ̄i-sʌ-nɛ
1sg:erg 2sg.poss-son wake-appl-1sg→2sg:n.pst
‘I will wake your son for you.’
b. ʔuŋʌ ʔi-mɛ̂m-po laɡi ʔīn
1sg:erg 2sg.poss-mother-gen for 2sg
tʰʌ̄i-sʌ-nɛ
wake-appl-1sg→2sg:n.pst
‘I wake you on your mother’s behalf.’

In Hayu (§3.1.4), the -t applicative can be applied to most transitive verbs, in the form
of a specific applicative conjugation (see Table 5). Since both the applied object and
the original object are marked in the absolutive, examples as (22) are ambiguous as to
whether the overt noun is the applied phrase or not (Michailovsky 1988: 142).

(22) Hayu (Michailovsky 1988: 142)

gɑ ɑŋ uxpʊ pʊk-t-ʊŋ-mi
1sg:erg 1sg.poss father lift-appl-1sg → 3-assert
(a) ‘I wake/woke my father.’
(b) ‘I wake/woke my father for him/her (someone else).’
(c) ‘I wake/woke him/her/it for my father.’

In addition, interpretation (a) in (22) is very close to the meaning of the BC (23), but has
the additional nuance of forced causation (‘I forced my father to get up’). Intensive or
coercive meaning without increase in the number of participants is attested in causa-
tive derivations (especially double causatives, see Kulikov 1993), but appears to be rare
in applicative derivations.

(23) Hayu
gɑ ɑŋ uxpʊ pʊk-k-ʊŋ-mi
1sg:erg 1sg.poss father lift-pst-1sg3-assert
‘I woke my father.’

Person hierarchy effects are also observed: either the beneficiary (24a) or the object of
the BC (24b) can be indexed on the verb. When the original object is a first or second
person and the beneficiary a third person, the beneficiary is marked with the suffix
-leːsi (24b). In such forms, the morphological contrast between applicative and non-ap-
960 Guillaume Jacques, Aimée Lahaussois

plicative conjugations is neutralized (§3.1.4), so that (24b) does not count as a genuine
case of applicative construction.

(24) Hayu (Michailovsky 1988: 142)

a. ga gon co puk-no-m
1sg 2sg child raise- 1→2-assert
‘I will wake the child up for you.’
b. minoŋ-leːsi ga gon puk-no-m
3sg-for 1sg 2sg raise-1→2-assert
‘I will wake you up for him.’

Michailovsky (1988: 140) also reports a conflicting example, where the verb is marked
in the 3sg→3 applicative suffix -to (Table 5), but the beneficiary (1sg) is not indexed,
and receives instead the suffix -leːsi. This isolated example, which runs counter to the
person hierarchy observed in the rest of Kiranti, is difficult to interpret in the absence
of additional data from this language.

(25) Hayu
komi-ha aŋ-leːsi kolu xoːco six-to-m
3sg-erg 1sg-for one chicken kill-appl:2/3→3-assert
‘He killed a chicken for me.’

4.1.3 Ditransitive verbs

Applicative derivations with ditransitive verbs are poorly documented in existing

grammars of Kiranti languages. In Khaling, the applicative of |bi| ‘give’ expresses a
double transfer of property, with a direct recipient in the absolutive, and an indirect
recipient which may be optionally marked with the benefactive postpositions. In (26),
the 2sg indirect recipient is indexed as object in the verb morphology, though it receives
optional marking of the ablative suffix -lʌkʌ, marking here the beneficiary.

(26) Khaling
ʔuŋʌ ʔīn-lʌkʌ Boyd kitap bîn-san-tɛni
1sg:erg 2sg-abl Boyd book give:appl-appl-1sg→2sg:pst
‘I gave Boyd a book for you.’

When the direct recipient is higher in the person hierarchy than the indirect one,
speakers have hesitations about whether the direct recipient is to be encoded as object
in verbal indexation rather than the indirect one. Example (27), where the 2sg direct
recipient is encoded as an object, is the consensus obtained after a thorough discussion
between speakers.
 27 Applicative derivations in Kiranti 961

(27) Khaling
ʔuŋʌ Boyd-lʌkʌ kitap ʔīn bîn-san-tɛni
1sg:erg Boyd-abl book 2sg give-1sg→2sg:pst
‘I gave you a book for Boyd.’

4.2 Combination with other voices and complex constructions

The Khaling V2 applicative -sʌ(t) is not compatible with reflexive derivations. For
instance, the applicative |mo-(t)-sʌ(t)| ‘vomit on’ (see 38 below), from the intransi-
tive verb |mo| ‘vomit’, cannot be combined with reflexive-middle -(N)si derivation
(Jacques et al. 2016) to express the meaning ‘vomit on oneself’.
The form expressing this meaning (28) appears at first glance to only contain a verb
stem from the root |mo| ‘vomit’ followed by the reflexive -nsi suffix.

(28) Khaling
moɔ̂-nsi-ŋʌ-tʌ
vomit:appl-refl-1sg-pst:1sg
‘I vomited on myself.’

However, there is a morphological complication here: the stem mɵ- rather than moɔ̂-
would have been expected from the root |mo| in the 1sg. The stem form moɔ̂- corre-
sponds to the conjugation class |-ot| (Jacques et al. 2016). This piece of indirect evi-
dence indicates that moɔ̂nsiŋʌtʌ ‘I vomited on myself’ is not directly derived from
the intransitive |mo| ‘vomit’, but rather from the non-attested -t applicative †|mo-t|,
which occurs as first element of the applicative |mo-(t)-sʌ(t)| ‘vomit on’. We thus have
a partial preservation of applicative derivation in this reflexive form.
The V2 applicative -sʌ(t) can occur in the reciprocal periphrastic construction,
which combines the auxiliary |lu| ‘feel’ with the bare infinitive of the verb root of the
BC. This infinitive selects the reduced strong stem (Jacques et al. 2012: 1119) of the V1, for
instance ʔʌŋ- from the root |ʔiŋ| ‘buy’ in (29).

(29) Khaling
ʔʌ̄m-su kitap ʔʌ̄ŋ-sʌ lʉ̂-iti
3-du book buy-appl feel-du:pst
‘They two bought books for each other.’
962 Guillaume Jacques, Aimée Lahaussois

Yakkha presents a similar situation: applicative verbs can undergo reciprocal deri-
vation, as can be seen in example 30 (Schackow 2015: p 274), but it is not possible to
combine applicative V2’s and reflexive derivations.10

(30) Yakkha
Kanciŋ moja pham-bi-khusa ca-me-ci=ha
1du sock knit-appl-recip eat.aux-n.pst-[1]du=nmlz.n.sg
‘We knit socks for each other.’

In Khaling, the applicative can also be employed in the periphrastic desiderative con-
struction with the impersonal verb |dhak| ‘want’, involving optional reduplication of
the last syllable of the bare infinitive (in this case, the V2 -sʌ)

(31) Khaling
ʔʌ̄m kitap ʔʌ-ʔʌ̄ŋ-sʌ~sʌ dâː
3sg book 1sg-buy-appl~desid want:3sg:n.pst
‘I want to buy a book for him.’

5 Semantics
The -t applicative conveys various semantic roles to the ApplO, including goal, stimulus,
instrument or beneficiary/maleficiary. These general properties are illustrated below
with data from Khaling and Hayu.
In Khaling, the applied object of applicative verbs derived by the -t suffix can be a
goal (or an addressee), a stimulus or a beneficiary depending on the base verb, but each
applicative verb only has one fixed interpretation (Table 10) and these verbs are highly
lexicalized (Jacques 2015).
For instance, the applicative |ŋint| of the intransitive verb |ŋin| ‘be afraid’ can
only be interpreted as ‘be afraid of’ (see 15b above, §4.1.1): its ApplO is necessarily a
stimulus, and cannot be beneficiary (entailing an interpretation ‘be afraid for X’).
The only -t applicative with a beneficiary interpretation in Khaling, |kurt| ‘bring
for’ (Table 3), has a counterpart with a V2 (Table 8) with the more compositional meaning
of ‘carry for’.
Instrumental applicatives are not found in Khaling, but in Hayu (§3.1.4), the applica-
tive conjugation (32b) can either convey beneficiary (i) or instrumental (ii) roles to the
ApplO for most verbs (32a).

10 Belhare has a specific V2 marking autobenefactive (Bickel 2017: 710), and this meaning is not ex-
pressed by combining the applicative V2 with the regular reflexive.
 27 Applicative derivations in Kiranti 963

Table 10: -t applicatives in Khaling.

base meaning applicative meaning semantic

role

|ŋur| roar |ŋurt| roar at goal

|bhur| be angry |bhurt| scold goal
|bhrot| shout |bhrott| call goal
|ret| laugh |rett| laugh at goal
|lem| be sweet |lemt| coax goal
|ŋin| be afraid |ŋint| be afraid of stimulus
|tshil| be frustrated |tshilt| be dissatisfied with stimulus
|kur| carry |kurt| bring for beneficiary

(32) Hayu (Michailovsky 1988: 141)

a. gɑ rʊk-k-ʊŋ-mi
1sg:erg plough-pst-1sg→3-assert
‘I ploughed it (of a field).’
b. gɑ rʊk-t-ʊŋ-mi
1sg:erg plough-pst:appl-1sg→3-assert
(i) ‘I ploughed for him/her.’
(ii)‘I ploughed using it (of an ox).’

In addition, the applicative can also have a forced causation interpretation in Hayu, as
illustrated by (22) above.
As for the V2 applicatives (Table 6), their default interpretation in all Kiranti lan-
guages is that of benefactive (‘for X’, ‘on X’s behalf’) illustrated below with data from
Khaling. Example (33) shows a typical example of benefactive V2 construction, with a
1sg beneficiary indexed on the verb.

(33) Khaling
tikîm poɔpoɔp ni sʉ̂ː-ʔɛ go
dem:auditory owl top who-erg foc
ʔi-sêj-sʌ-ŋʌ-nu
inv/2-kill-appl:n.pst-1sg-pl
‘Who is going to kill that (noisy) owl for me? ’ (https://doi.org/ 10.24397/pangloss-
0000608#S54)

When the applicative V2 has a benefactive interpretation, the ApplO can be optionally
marked with benefactive case marking, even while being indexed on the verb (20).
This construction is used to specify the recipient of an action (for instance, with the
verb |jok| ‘distribute’, see 17b). The benefactive function fo the applicative V2 has in some
cases meanings that are not entirely compositional. In particular, it can turn verbs of
964 Guillaume Jacques, Aimée Lahaussois

manipulation into verbs of transfer of property. For instance, the V2 applicative of the tran-
sitive verb |pum| ‘hold in one’s fist’ specifically means ‘give something by putting it into
someone’s fist’ (34), an action in which the giver, holding the object to be given in his fist,
places it in the palm of the recipient (so that the latter does not see what is being given).

(34) Khaling (Jacques et al. 2015)

ʔʌ̄m-ʔɛ khɛpʦi pʌ̂m-sʌ-tɛ.
3sg-erg money hold.in.fist:appl-appl-3sg:pst
‘He put money into his hand.’

Similarly, |lott| ‘reach into’, a verb which selects bags or containers as object, has an
applicative form meaning ‘go somewhere to fetch something for someone’ (35).

(35) Khaling (Jacques et al. 2015)

ʔuŋʌ pʌsʌl-bî-m saman ʔīn
1sg:erg shop-loc-nmlz thing 2sg
loɔ̂n-sʌ-nɛ.
reach.into:appl-appl-1sg→2:n.pst
‘I will fetch your things in the shop.’

Apart from the benefactive/recipient function, the V2 applicative in Khaling has three
other possible interpretations.
First, when the applied object is the possessor of the original object, as in (36a), the
V2 often has a malefactive meaning. When the BO is not possessed, a beneficiary inter-
pretation is favoured (36b).

(36) Khaling
a. ʔʌ̄m-ʔɛ ʔʌ-kitap ʔi-kʰe-s-sɵs-tʌ
3sg-erg 1sg.poss-book 2/inv-steal-appl-appl-1sg:pst
‘He stole my book’.
b. ʔʌ̄m-ʔɛ kitap ʔi-kʰe-s-sɵs-tʌ
3sg-erg book 2/inv-steal-appl-appl-1sg:pst
‘He stole a book for me.’ (possible interpretation)

Second, with a limited number of verbs (such as |iŋ| ‘buy’), the applied object can also
refer to the source from which the BO is obtained, as in (37).

(37) Khaling
kitap ʔʌ̂n-sʌ-nɛ
book buy:appl-appl-1sg→2sg
‘I will buy the book for you.’ (as a present, benefactive)
‘I will buy the book from you.’ (you are a bookseller, source applicative)
 27 Applicative derivations in Kiranti 965

Third, it can express the goal of the action with intransitive verbs of physical excretion
such as |mo| ‘vomit’ (38), |ʔe| ‘shit’ or |tser| ‘piss’ (16b).

(38) Khaling
tsɵttsɵ-ʔɛ ʔûŋ ʔi-mɵ-s-sɵs-tʌ
child-erg 1sg 2/inv-vomit-appl-appl-1sg:pst
‘The child vomited on me.’

The semantic role of stimulus is almost never attributed to a ApplO by V2 applicative

derivations in Kiranti. The only exception is the transitivizer -ni in Yakkha, which does
have this function with a handful of examples (see 13b, §3.2).

6 Lookalikes
All Kiranti languages have examples of object-preserving lability, illustrated by the root
|khutt| ‘steal’ in Limbu, which means ‘steal, rob of’ when conjugated transitively (39c,
39d), and can be interpreted as ‘be stolen’ when occurring with intransitive morphology
(39a).
A few Kiranti languages, including Puma and Limbu (but not Khaling), also have
subject-preserving lability (Bickel et al. 2007), as in (39b), where |khutt| also shows the
meaning ‘commit a theft’, with an intransitive subject corresponding to the transitive
subject of (39c) and (39d).

(39) Limbu (van Driem 1991: 527)

a. Sapla khutt-ɛ
book steal-pst:intr
‘The book was stolen.’
b. A-ndzum-in khutt-ɛ
1sg.poss-friend-def steal-pst:intr
‘My friend committed a theft.’
c. mɛ-n-ni-baŋ-ba məna-lle
neg-neg-see-1sg→3:pst-nmlz man-erg
a-yaŋ-in khutt-u
1sg.poss-money-def steal-3:O
‘A man I didn’t see stole my money.’
d. A-ndzum-ille sapla khutt-aŋ
1sg.poss-friend-erg book steal-1sg.O.pst
‘My friend robbed me of my book.’
966 Guillaume Jacques, Aimée Lahaussois

The transitive construction in (39c) selects as transitive subject the same semantic role
as the intransitive subject of (39b). The additional argument in (39c) has the semantic
role of patient, and the relationship between (39b) and (39c) resembles that of an intran-
sitive BC with an AC selecting a patient as ApplO, though without overt derivation.
The verb in example (39d) takes two arguments in addition to its subject: an absolu-
tive argument (patient) not encoded in the verb morphology, and a second one indexed
as direct object with a semantic role of maleficiary, possessor of the patient. Its seman-
tics resembles that of a malefactive applicative construction (compare in particular 5
and 36a above), though again without any overt derivation.

7 Conclusion
The applicative constructions in Khaling and other Kiranti languages present features
that are fairly widespread crosslinguistically, but a few observations are of wider
interest.
First, a striking feature of one of the applicative markers (the suffix -t ) is the fact
that it appears in a specific set of applicative conjugation classes. Applicative forms
are distinct from non-applicative ones only in a subsection of the direct configurations,
involving a third person object and a non-dual subject, as observed in Khaling (§3.1.2)
or Hayu (§3.1.4). This results in ambiguous forms in inverse configurations (3→1/2) and
in local scenarios (1→2 and 2→1), as illustrated by (24) in §4.1.2.
Second, the parallel grammaticalization of V2 applicatives across Kiranti (§3.2.1),
which are not reconstructible to the proto-language renewed the applicative construc-
tions throughout the whole subgroup (except Hayu). The morphological paradigms of
these auxiliaries are highly irregular, but the irregularities do not appear to be shared
across languages.
Third, the bipartite applicative constructions in Khaling, Bantawa and Yakkha
(§3.2.2) exhibit double marking, expressing applicative derivation by a combination of
the -t suffix (or its secondary effects on stem alternations) and of the applicative V2.
Fourth, the V2 applicatives are not only compatible with intransitive and monotran-
sitive verbs, but also occur on ditransitives, resulting in four-argument predicates
(§4.1.3).
Fifth, the applicatives in Kiranti can convey a broad range of semantic roles to the
ApplO on a verb-to-verb basis (§5).
 27 Applicative derivations in Kiranti 967

Abbreviations
ApplO applied object
BO object of the base construction
abl ablative
appl applicative
assert assertive
caus causative
conj conjunction
dat dative
def definite
dem demonstrative
desid desiderative
du dual
erg ergative
excl exclusive
foc focus
incl inclusive
inf infinitive
inv inverse
instr instrumental
loc locative
n. non-
neg negation
nmlz nominalization
ns non-singular
pl plural
pst past
recip reciprocal
sg singular
top topic

References
Ackerman, Farrell, Jim Blevins & Rob Malouf. 2009. Parts and wholes: implicative patterns in inflectional
paradigms. In Jim Blevins & Juliette Blevins (eds.), Analogy in grammar, 54–82. Oxford: Oxford
University Press.
Allen, Nicholas J. 1975. Sketch of Thulung grammar, with three texts and a glossary. Ithaca: China-Japan
Program.
Arcodia, Giorgio Francesco & Bianca Basciano. 2020. Morphology in Sino-Tibetan Languages. In Mark
Aronoff (ed.), Oxford Research Encyclopedia of Linguistics, Oxford: Oxford University Press. doi:10.1093/
acrefore/9780199384655.013.530.
Bickel, Balthasar. 2017. Belhare. In Graham Thurgood & Randy LaPolla (eds.), The Sino-Tibetan languages,
696–721. London: Routledge.
968 Guillaume Jacques, Aimée Lahaussois

Bickel, Balthasar, Martin Gaenszle, Arjun Rai, Prem Shoj Rai, Shree Kumar Rai, Vishnu Rai & Narayan P.
Sharma. 2007. Two ways of suspending object agreement in Puma: Between incorporation, antipas-
sivization, and optional agreement. Himalayan Linguistics 7. 1–19.
Bickel, Balthasar, Manoj Rai, Netra Paudyal, Goma Banjade, Toya Nath Bhatta, Martin Gaenszle, Elena
Lieven, Iccha Purna Rai, Novel K Rai & Sabine Stoll. 2010. The syntax of three-argument verbs in
Chintang and Belhare (Southeastern Kiranti). In Andrej Malchukov, Martin Haspelmath & Bernard
Comrie (eds.), Studies in ditransitive constructions: A comparative handbook, 382–408. Berlin: Mouton de
Gruyter.
Borchers, Dörte. 2008. A grammar of Sunwar: Descriptive grammar, paradigms, texts and glossary. Leiden: Brill.
Dai, Qingxia & Xijian Xu. 1992. Jǐngpōyǔ yǔfǎ 景頗語語法 [A grammar of Jingpo]. Beijing: Zhōngyāng mínzú
xuéyuàn chūbǎnshè.
Doornenbal, Marius. 2009. A grammar of Bantawa. Leiden: Leiden University PhD dissertation.
Downer, Gordon B. 1959. Derivation by tone-change in Classical Chinese. Bulletin of the School of Oriental and
African Studies 22(1/3). 258–290.
van Driem, George. 1987. A grammar of Limbu. Berlin and New York: Mouton De Gruyter.
van Driem, George. 1991. Tangut verbal agreement and the patient category in Tibeto-Burman. Bulletin of
the School of Oriental and African Studies 54.3. 520–534.
van Driem, George. 1993. A Grammar of Dumi. Berlin and New York: Mouton De Gruyter.
Ebert, Karen. 2017. Camling. In Graham Thurgood & Randy LaPolla (eds.), The Sino-Tibetan languages,
722–735. London: Routledge.
Gerber, Pascal & Selin Grollmann. 2018. What is Kiranti? Bulletin of Chinese Linguistics 11(2). 99–152.
Herce, Borja. 2021. Stem alternations in Kiranti and their implications for the morphology-phonology
interface. Journal of Linguistics 57(2). 321– 363.
Jacques, Guillaume. 2013. Applicative and tropative derivations in Japhug Rgyalrong. Linguistics of the
Tibeto-Burman Area 36(2). 1–13.
Jacques, Guillaume. 2015. Derivational morphology in Khaling. Bulletin of Chinese Linguistics 8(1). 78–85.
Jacques, Guillaume. 2016. How many ✶-s suffixes in Old Chinese? Bulletin of Chinese Linguistics 9(2). 205–217.
Jacques, Guillaume. 2017. A reconstruction of Proto-Kiranti verb roots. Folia Linguistica Historica 38. 177–215.
Jacques, Guillaume. 2018. Bipartite verbs in Japhug and other Trans- Himalayan languages. Linguistics of the
Tibeto Burman Area 41(2). 175–191.
Jacques, Guillaume, Aimée Lahaussois, Boyd Michailovsky & Dhan Bahadur Rai. 2012. An overview of
Khaling verbal morphology. Language and Linguistics 13(6). 1095–1170.
Jacques, Guillaume, Aimée Lahaussois & Dhan Bahadur Rai. 2016. Reflexive paradigms in Khaling. Linguistics
of the Tibeto-Burman Area 39(1). 33–48.
Jacques, Guillaume, Aimée Lahaussois, Dhan Bahadur Rai & Yadav Kumar. 2015. Khaling-Nepali-English
dictionary, version 1.0. Paris: Projet HimalCo. http://himalco.huma-num.fr/.
Jacques, Guillaume, Aimée Lahaussois & Shuya Zhang. 2021. Associated motion in Sino-Tibetan/
Trans-Himalayan. In Antoine Guillaume & Harold Koch (eds.), Associated motion, 819–853. Berlin:
Mouton de Gruyter.
King, John T. 2009. Grammar of Dhimal. Leiden: Brill.
Konnerth, Linda. 2014. A grammar of Karbi. Eugene: University of Oregon PhD dissertation.
Kulikov, Leonid. 1993. The ‘second causative’: A typological sketch. In Causatives and transitivity, 121–154.
Amsterdam: John Benjamins.
Lahaussois, Aimée. 2002. Aspects of the grammar of Thulung Rai: An endangered Himalayan language.
Berkeley: University of California PhD dissertation.
Lahaussois, Aimée. 2009. Koyi Rai: An initial grammatical sketch. Himalayan Linguistics Archive 4. 1–33.
Lahaussois, Aimée. 2020. Predicate derivations in Thulung. Journal of South Asian Languages and Linguistics
7(1). 57–84.
 27 Applicative derivations in Kiranti 969

LaPolla, Randy. 2000. Valency-changing derivations in Dulong/Rawang. In R.M.W. Dixon & A.Y. Aikhenvald
(eds.), Changing valency: Case studies in transitivity, 282–311. Cambridge: Cambridge University Press.
Michailovsky, Boyd. 1985. Tibeto-Burman dental suffixes: Evidence from Limbu (Nepal). In Graham
Thurgood, James A. Matisoff & David Bradley (eds.), Linguistics of the Sino-Tibetan area: The state of the
art, 363–375. Canberra: Pacific Linguistics.
Michailovsky, Boyd. 1988. La langue hayu. Paris: Editions du CNRS.
Michailovsky, Boyd. 2012. Internal reconstruction of the Dumi verb : Lexical bases and stem formation.
Linguistics of the Tibeto-Burman Area 35(2). 49–87.
Michailovsky, Boyd. 2017. Kiranti languages. In Graham Thurgood & Randy LaPolla (eds.), The Sino-Tibetan
languages, 646–679. London: Routledge.
Opgenort, Jean-Robert. 2004. A grammar of Wambule. Leiden: Brill.
Opgenort, Jean-Robert. 2005. A grammar of Jero, with a historical comparative study of the Kiranti languages.
Leiden: Brill.
Paudyal, Netra. 2015. Aspects of Chintang syntax. Zürich: University of Zürich PhD dissertation.
Peng, Guozhen & Hilary Chappell. 2011. Ya33 ‘ give’ as a valency increaser in jinghpo nuclear serialization.
Studies in Language 35(1). 128–167.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Pokharel, Madhav P. 2005. Benefactive constructions in Nepali. Kobe papers in linguistics 2(149–179).
Post, Mark W. 2007. A grammar of Galo. Melbourne: LaTrobe University PhD dissertation.
Rutgers, Roland. 1998. Yamphu, grammar, texts and lexicon. Leiden: Research School CNWS.
Sagart, Laurent, Guillaume Jacques, Yunfan Lai, Robin J. Ryder, Valentin Thouzeau, Simon J. Greenhill &
Johann-Mattis List. 2019. Dated language phylogenies shed light on the ancestry of Sino-Tibetan.
Proceedings of the National Academy of Sciences 116(21). 10317–10322. doi:10.1073/ pnas.1817972116.
https://www.pnas.org/content/116/21/10317.
Schackow, Diana. 2015. A grammar of Yakkha. Berlin: Language Science Press. http://langsci-press.org/
catalog/book/66.
Sharma, Narayan Prasad. 2014. Morphosyntax of Puma, a Tibeto-Burman language of Nepal. London: School
of Oriental and African Studies, University of London PhD dissertation.
Sun, Jackson T.-S. 2006. Jiāróngyǔ dòngcí de pàishēng xíngtài 嘉戎语动 词的派生形态 [Derivational
morphology in the Rgyalrong verb]. Mínzú yǔwén 民族语文 4(3). 3–14.
Tolsma, Gerard Jacobus. 2006. A grammar of Kulung. Leiden: Brill.
Watters, David E. 2002. A grammar of Kham. Cambridge: Cambridge University Press.
Wolfenden, Stuart. 1929. Outlines of Tibeto-Burman linguistic morphology, with special reference to the prefixes,
infixes and suffixes of classical Tibetan and the languages of the Kachin, Bodo, Nǎgǎ, Kuki-Chin and Burma
groups. London: Royal Asiatic Society.
Zhang, Shuya. 2020. Le dialecte de Brag-bar du rgyalrong situ et sa contribution à la typologie de l’expression
des relations spatiales: le movement associé et l’orientation. Paris: Inalco PhD dissertation.
Bradley McDonnell and Christina L. Truong
28 Applicative constructions in languages
of western Indonesia
Abstract: This chapter provides an overview of applicative constructions in a sample
of eight Austronesian languages of western Indonesia. Following an orientation to the
languages (§ 2) and the forms of their applicatives affixes (§ 3), we describe the seman-
tic and syntactic properties of applicative constructions according to possible roles
for the applied phrase. These include beneficiaries and recipients (§ 4), instruments
and themes (§ 5), goals, locations, and addressees (§ 6), and other roles found in tran-
sitivizing constructions, e.g. content phrases and stimuli (§ 7). For each type, we note
the syntactic status of the AppP and any companion phrase (the participant expressed
as P in a corresponding base construction), and semantic characteristics of the AC and
compatible base verbs. We find that all languages of the sample allow a beneficiary
AppP and a theme companion phrase to both be expressed as core arguments in dit-
ransitive clauses. However, when the AppP is an instrument or goal, some languages
require that the companion phrase be realized as an oblique or unexpressed. Remain-
ing sections discuss lookalike constructions where an applicative suffix shows only
an aspectual or semantic effect (§ 8), and describe interactions between applicatives
and causative morphology (§ 9) as well as applicatives and voice (§ 10).

1 Introduction
In this chapter we examine a sample of Austronesian languages of a region that we refer
to as western Indonesia, which includes Malayo-Polynesian languages spoken in Malay-
sia, Brunei, and many parts of western and central Indonesia (Sumatra, Java, Kaliman-
tan, Sulawesi, Bali, and Lombok).1 In these languages, applicative constructions (ACs)
are typically marked by one of a small number of verbal affixes that while serving many

1 We would like to recognize four individuals who shared their expertise on some of the languages in
this paper: Khairunnisa for Sasak, Dewi Setiani for Sundanese, and Wawan Sahrozi and Johan Safri
for Nasal. Additionally, we are grateful to the editors of this volume as well as an anonymous reviewer
for comments on an earlier version of this paper. The first author is also grateful to his research coun-
terpart in Indonesia, Yanti (Atma Jaya Catholic University of Indonesia) and the Ministry of Research
and Technology in Indonesia for allowing him to conduct research on Nasal. Discussion of the Nasal
data is based upon work supported by the National Science Foundation under Grant BCS-1911641. Any
opinions, findings, and conclusions or recommendations expressed in this material are those of the
author(s) and do not necessarily reflect the views of the National Science Foundation. The second au-
thor would like to acknowledge the Bilinski Educational Foundation, which supported the collection of
Sundanese data.

https://doi.org/10.1515/9783110730951-028
972 Bradley McDonnell and Christina L. Truong

Table 1: Language Sample.

Language Subgroup Location Sources

Toba Batak Batak (Sumatran) North Sumatra van der Tuuk (1971 [1864–1867])
Nasal Sumatran Bengkulu McDonnell fieldnotes
Sundanese Sundic West Java Truong fieldnotes
Salako (Kendayan) Malayic Borneo Adelaar (2005)
West Coast Bajau Greater Barito Sabah Miller (2007)
Sasak Bali-Sasak-Sumbawa Lombok Khairunnisa & McDonnell (in prep)
Pendau Celebic Central Sulawesi Quick (2007)
Makasar South Sulawesi South Sulawesi Jukes (2020)

non-applicative functions are clearly separate from voice morphology. To explore the
range of diversity and commonality found within applicative systems in this region, we
describe the syntax and semantics of ACs and applicative lookalikes in a sample of eight
western Indonesian languages, listed here with their ISO 639–3 codes and glottocodes:
Toba Batak [bbc/bata1289], Nasal [nsy/nasa1239], Sundanese [sun/sund1252], Salako
[knx/kend1254], West Coast Bajau [bdr/west2560], (Ampenan) Sasak [sas/sasa1249],
Pendau [ums/pend1242], and Makasar [mak/maka1311]. These languages all belong to the
Malayo-Polynesian branch of the Austronesian family. While this selection of languages
represents a convenience sample, limited in part by the availability of the relevant infor-
mation, these languages are not closely related to one another and are geographically
dispersed across major islands and island groupings of western Indonesia and eastern
Malaysia.
This paper is organized as follows. Section 2 presents general information about
the languages examined and their basic morphosyntax. This includes a critical descrip-
tion of transitive alternations often coded on the verb, which we refer to as voice.
Section 3 gives a general introduction to the applicative suffixes in the languages of
the sample. The following four sections describe ACs according to the semantic roles
which map to the phrase licensed by the applicative marker. These include beneficiar-
ies and recipients (Section 4), instruments and themes (Section 5), goals, locations, and
addressees (Section 6), and other types of participants (Section 7). Section 8 describes
morphological lookalikes in which an applicative suffix appears on the verb without
resulting in a change in argument structure. Section 9 describes how applicative suf-
fixes combine with causative morphology and other affixes. Section 10 discusses how
voice and applicative morphology are used together in these languages. A conclusion
follows in Section 11.
 28 Applicative constructions in languages of western Indonesia 973

2 The languages and their basic morphosyntactic

properties
The eight languages included in the sample for this chapter are listed in Table 1, along
with basic information including genetic subgroup, geographic location and the sources
we have primarily drawn on in the descriptions that follow. The authors have firsthand
experience working with Sundanese, Nasal, and the Ampenan variety of Sasak, which
differs to some extent from the varieties described in Austin (2001).
The western Indonesian languages in this sample range from moderately agglu-
tinative (e.g. Pendau with the potential to have several prefixes and suffixes) to rela-
tively isolating (e.g. Ampenan Sasak which at most can take a single prefix and single
suffix). There is little to no casemarking and arguments of the verb are often unrealized
when their reference is recoverable in the discourse. In general, core arguments are
unmarked, while obliques are marked with a preposition. Intransitive predicates may
be unmarked or marked with one of several affixes that express different semantic
properties, such as stative, dynamic, reciprocal, non-volitional among others. Transitive
predicates are typically marked for voice.
The voice systems of western Indonesian languages, however, differ in important
ways from well-known voice systems with active-passive or ergative-antipassive alter-
nations (see Chen & McDonnell 2019 for a recent summary). With some exceptions,
these systems have been labeled symmetrical voice, which involve a two-way distinc-
tion between A-oriented and P-oriented transitive constructions. These constructions
have been described as ‘symmetrical’ because they show little difference in terms of
morphosyntactic complexity, syntactic transitivity, and frequency in discourse (Arka
2003, Himmelmann 2005, Riesberg 2014). While there are a number of salient similari-
ties among the voice systems in the eight languages of the sample, there is considerable
variation even in the principal components of symmetrical voice. Consider the exam-
ples from Toba Batak in (1).2

(1) Toba Batak, Voice alternations (Schachter 1984: 127–128)

a. Man-jaha buku guru i.
av-read book teacher def
‘The teacher read a book.’
b. Di-jaha guru buku i.
pv-read teacher book def
‘A teacher read the book.’

2 In examples, we have generally kept the orthographic conventions of original sources, but adopted
unified glossing conventions throughout.
974 Bradley McDonnell and Christina L. Truong

In A-oriented constructions, the predicate is typically marked by a prefix that contains

a homorganic nasal that assimilates to—and in some cases replaces—the first conso-
nant in the root to which it attaches.3 The abstract nasal segment is represented as N,
and the AV prefix in Toba Batak is maN-. In this chapter, A-oriented constructions are
referred to as A-Voice (AV) because they privilege A (i.e., the most agent-like argument
in a transitive construction), which is guru ‘teacher’ in (1a). The P argument buku ‘book’
is unmarked and is considered a core argument.
Likewise, P-oriented constructions may be marked with various prefixes (e.g. di- in
Toba Batak) and are referred to as P-Voice (PV) as they privilege P (i.e., the most patient-
like argument in a transitive construction), which is buku ‘book’ in (1b). The A argument
guru ‘teacher’ is unmarked and is considered core. The privileged argument in each
construction has been referred to by many names (e.g. pivot, trigger, primary argu-
ment), but here we will follow the Indonesianist tradition (e.g. Musgrave 2001, Riesberg
2014) and refer to it as the subject. Since the core argument that is not privileged can
be P as in (1a) or A as in (1b), the label ‘object’ is not appropriate. Thus, we again follow
the Indonesianist tradition and refer to this argument as the non-subject core argument
or simply non-subject.
The examples from Toba Batak demonstrates symmetrical voice clearly: the two
transitive constructions are equally marked by prefixes maN- av and di- pv, respectively,
and there is no apparent difference in transitivity between the two. Furthermore, Wouk
(1984) finds that in small corpus of narratives, AV and PV have similar frequencies: PV
makes up 52% and AV 48% of transitive constructions.4 However, the voice systems in
the languages in the sample—and in western Indonesia more generally—vary along
several dimensions. In this section, we focus on two: (i) the marking of non-subject A
arguments and (ii) (overt) voice morphology.
In regards to (i), Sundanese marks both AV and PV constructions with prefixes N-
and di-, respectively, as in (2). However, the A argument in PV is most commonly marked
by a preposition ku ‘by’, as in (2b).

(2) Sundanese, Voice alternations (Truong fieldnotes)

a. Asep m-(b)euli baju.
A. av-buy clothes
‘Asep bought clothes.’
b. Baju di-beuli ku Asep.
clothes pv-buy by A.
‘Asep bought clothes.’

3 In examples where the first consonant of the root has been replaced by the homorganic nasal prefix,
we include the underlying segment in parentheses for clarity, e.g. m-(b)euli for the A-oriented form of
beuli ‘to buy’ in example (2a) below.
4 These percentages include only cases where the predicate was marked by maN- or di-.
 28 Applicative constructions in languages of western Indonesia 975

On the surface these examples look to be an active-passive alternation, but there are
syntactic properties (e.g. the ability for A to bind P arguments in PV; see Kurniawan
2013 and Kroeger & Riesberg 2023) as well as discourse properties (e.g. AV and PV have
similar frequencies in discourse) that call such an analysis into question.
In regards to (ii), the PV in West Coast Bajau, for example, is not overtly marked by
morphology, while AV is prefixed with N-, as in (3). Miller (2007), however, analyzes PV
as being marked by a null prefix ∅-.

(3) West Coast Bajau, Voice alternations (Miller 2007: 135–136)

a. Azizy boi n-(s)embali sapi’ e dilaw.
A. cmpl av-slaughter cow dem yesterday
‘Azizy slaughtered the cow yesterday.’
b. Boi ∅-sembali Azizy sapi’ e dilaw.
cmpl pv-slaughter A. cow dem yesterday
‘Azizy slaughtered the cow yesterday.’

In Salako, PV is also unmarked when the non-subject A argument is realized, but it is

marked with a proclitic di= when it is not realized.5 In the Nasal examples in (4), PV is
unmarked when A is first or second person, as in (4a), and optionally marked with a
dedicated prefix when A is third person, as in (4b).

(4) Nasal, PV marking (McDonnell fieldnotes)

a. lahan ni kak khadu kam=suah.
field that pfv finish 1pl.excl.nsbj=[pv]burn
‘We already burned the field.’
b. lahan ni kak khadu (di-)suah=nyo.
field that pfv finish pv-burn=3sg
‘He already burned the field.’

Pendau shares similarities to Nasal with several additional layers of complexity. Pendau
distinguishes two sets of voice prefixes, a realis set and an irrealis set. Additionally,
there are irrealis PV constructions for first and second person non-subject A arguments,
which appear in the preverbal position without a dedicated voice prefix (see Quick
2007: 374–375).
In Ampenan Sasak and Makasar, the situation is more extreme. In Ampenan Sasak,
there still appears to be an opposition between A-oriented and P-oriented construc-
tions, but the predicate is typically unmarked; it is only optionally marked by the AV
prefix N- (see Khairunnisa 2022). Rather, transitive alternations are signaled through

5 The construction marked with di= is likely a separate passive construction, although Adelaar (2005)
does not label it as such.
976 Bradley McDonnell and Christina L. Truong

a combination of word order and the marking of A arguments. Consider the examples
in (5).

(5) Ampenan Sasak, Transitive alternations (Khairunnisa & McDonnell in prep)

a. Andii paléq(=nei) aku.
A. chase(=3) 1sg
‘Andi chased me.’
b. Aku siq=nei paléq siq Andii
1sg agt=3 chase agt A.
‘Andi chased me.’

In the A-oriented construction in (5a), A occurs in the preverbal position with cross-ref-
erencing of an optional second position pronominal clitic and is considered the subject.
P occurs post-verbally and is considered the non-subject argument. In the P-oriented con-
struction in (5b), P is in the preverbal position and is considered the subject, while A occurs
post-verbally and is marked with the agent marker siq. A is also cross-referenced on a
second position pronominal clitic. In both constructions, the A argument is co-referential
with a second position clitic that invariably expresses A. Since this alternation is not marked
on the verb, voice may not be the most appropriate term, but functionally these construc-
tions are analogous to AV and PV constructions in the other languages of the sample.6
Finally, Makasar has been described differently than the rest of the languages of
the sample. According to Jukes (2013, 2020), Makasar has an unmarked, basic transi-
tive clause alongside several marked constructions. In the basic transitive clause, the
verb receives no verbal marking, but A and P arguments are indexed with pronominal
clitics, as in (6).

(6) Makasar, Transitive construction

ku=kanre=i taipa=nu
1=eat=3 mango=2f.poss
‘I eat your mangoes.’

Makasar additionally has a number of verbal prefixes that are analogous to voice pre-
fixes in the other languages of the sample, although Jukes (2013) describes these pre-
fixes as valency- signaling prefixes. These include two prefixes that are analogous to AV:
aN(N)- and aN-. The formal differences between these prefixes is that aN(N)- triggers
nasal substitution of the first consonant of the root, whereas aN- does not. aN(N)- marks
a so-called semi-transitive clause where P is not indexed on the verb and may not be
definite, as in (7).

6 Khairunnisa (2022) provides a more detailed description of the different properties of the second
position clitic and the preverbal A or P argument.
 28 Applicative constructions in languages of western Indonesia 977

(7) Makasar, Semi-transitive construction (Jukes 2020: 257)

angng-(k)anre=a’ taipa
str-eat=1 mango
‘I eat mangoes’

The prefix aN- marks Actor Focus, where A must occur in the preverbal position but
is not indexed on the verb. Rather, P is indexed, as in (8). See Jukes (2020: 240–242) for
further discussion of argument indexing in Makasar.

(8) Makasar, Actor-focus construction (Jukes 2020: 269)

kongkong am-buno=ii miong=kui
dog af-kill=3 cat=1.poss
‘A dog killed my cat.’

Laskowske (2016) analyzes a closely related language, Bugis, as having a symmetrical

voice system. Under this analysis, the basic transitive clause is a P-oriented construc-
tion, while the the semi-transitive and Actor Focus constructions would be A-oriented
constructions. It is not necessary in this chapter to decide whether a particular voice
system is considered symmetrical, but it is helpful to see that while the languages in
our sample have diverse systems for marking transitive constructions, there are clear
similarities that allow for a more or less straightforward comparison.
In addition to transitive alternations in at least seven languages in the sample,
many of the languages have a ‘true passive’ construction with the exception of Pendau
and Toba Batak. The passive may have a dedicated affix, as in West Coast Bajau (-in-),
Amepenan Sasak (te-), Makasar (ni-), and Salako (di=), or it may be marked with the PV
prefix, as in Nasal and Sundanese. (See Chen & McDonnell 2019 for a discussion of the
difficulties in teasing apart PV and passive constructions in languages that mark both
constructions with the same affix).
Languages in the sample vary in terms of word order. With the exception of Ampenan
Sasak and Makasar, the non-subject argument is typically adjacent to the verb, forming a
constituent that we refer to as the predicate complex. The subject typically has a freer dis-
tribution and in these six languages it may occur before or after the predicate complex.
However, each language has a preferred word order. Toba Batak, exemplified in (1), is
typically subject-final, as is West Coast Bajau. However, in West Coast Bajau, PV clauses
more typically have subject-final order, while AV clauses typically have subject-initial
order (Miller 2007: 146–157). The other languages are predominately subject-initial,
including Nasal, Pendau, Sundanese, and Salako. In Ampenan Sasak and Makasar there
does not appear to be strong evidence for a predicate complex. Ampenan Sasak is pre-
dominately subject-initial. Makasar is canonically predicate-initial, although the order
of NP arguments after the predicate in transitive clauses is apparently completely free.
978 Bradley McDonnell and Christina L. Truong

Such arguments are indexed on the verb, and in cases where there are two third
person arguments indexed with pronominal clitics, the order of NPs does not resolve
any ambiguity (Jukes 2020: 231).
Finally, Pendau is unique in the sample as it has a number of stem-former pre-
fixes. These prefixes, which have the shapes pong-, po-, pe-, and popo- (and allomorphs
showing vowel harmony), have no semantic content of their own, but form an aug-
mented stem which can then take voice, applicative, and other derivational prefixes
(Quick 2007: 99–108). The use of stem-former prefixes is found in many Austronesian
languages of the Philippines and Sulawesi (see e.g. the discussion of stem-formers in
Himmelmann & Wolff 1999).

3 Applicative morphology
In this chapter, we generally follow Zuñiga & Creissels’ (this volume) definition of applica-
tive construction (AC) with the following key properties: (i) an AC is contrasted with a
base construction (BC) in that an AC shows additional overt morphological marking, (ii)
the applicative marker licenses an applied phrase (AppP) that expresses a non-agentive
semantic role that either cannot be expressed in the BC or is expressed as an oblique
phrase, and (iii) the participant encoded as A or S in the BC is encoded as A in the AC (if
it is expressed).
Each language of the sample has from one to three verbal affixes that mark ACs,
which are listed in Table 2. In two languages—West Coast Bajau and Ampenan Sasak—
there is only one applicative suffix. The other six languages have at least two applica-
tives, each of which licenses AppPs with different semantic roles. One suffix licenses
beneficiary and instrument AppPs and the other licenses goal and locative AppPs (with
the exception of Salako, which does not appear to mark instrumental AppPs at all).
However, for AppPs with other semantic roles such as stimulus or content, there is no
such specialization in the morphology (see Section 7). In addition to two applicative
suffixes, Sundanese makes use of a circumfix, pang- -keun that is used exclusively for
substitutive benefactive constructions. We treat this form as a circumfix (as opposed
to a separate prefix and a suffix) because it has a distinct semantic meaning that is not
compositional, and it is extremely productive across verbal forms in Sundanese. Note
that certain less productive ACs found in languages of the sample also require the use
of a fossilized prefix or stem-forming prefix in addition to an applicative suffix (see
Section 9 below).
Phonologically conditioned allomorphy is found for a number of these affixes. The
Salako suffix -AN may show rounding of the vowel to [ɔ] (written as à) and/or preplo-
sion of the final nasal (written as tn) depending on the shape of the root (Adelaar 2005:
30–32). For this reason, Adelaar represents the applicative suffix as an abstract form
written with capital letters. The vowels in Makasar -ang and -i coalesce with an iden-
 28 Applicative constructions in languages of western Indonesia 979

tical stem-final vowel, which also triggers a shift from penultimate to final stress, e.g.
jappa+ang → jappáng ‘walk with’ (Jukes 2020: 102). The vowel in West Coast Bajau -an
harmonizes with preceding /o/ or /e/ in the stem. It also weakens after a stem-final
vowel, coalescing with stem-final /o/, and being centralized to [ə] following stem-fi-
nal /e/ (Miller 2007: 54). Both stem and affixal consonants may undergo assimilatory
processes at a morpheme boundary. In Toba Batak, a sequence consisting of a mor-
pheme-final stop, nasal or fricative and morpheme-initial /h/ is realized as a geminate
stop, e.g. di+dalan+hon → didalatton ‘he carries out’ (Nababan 1981: 61). The Sundanese
circumfix pang- -keun causes a voiceless obstruent (and sometimes /b/) in stem-initial
position to become a nasal, as in teundeun ‘to place’ → pang-neunduen-keun ‘to place
for s.o.’.

Table 2: Applicative suffixes by semantic role of AppP.

Language Suffix Semantic role

ben inst goal loc
Toba Batak -hon
-i
Nasal -kun
-i
Sundanese -keun
-an
pang- -keun
Salako -AN
-iʔ
West Coast Bajau -an
Ampenan Sasak -an
Pendau -a’
-i
Makasar -ang
-i

Applicatives in these languages are predominantly optional because the phrase cor-
responding to the AppP in a BC may be expressed as an oblique, most commonly with
a prepositional phrase. In a few cases, however, the applicative is obligatory as there
is no (monoclausal) equivalent BC. These exceptions are noted in the sections below.
Throughout the paper, we are also concerned with the semantic participant expressed
by the P argument in a BC, and its syntactic realization in the corresponding AC. We
refer to this participant as the companion phrase to the AppP in describing the syntax
of ACs. While this chapter is focused on ACs marked with the affixes described here,
it is important to note that the these morphemes are highly polyfunctional. Besides
forming ACs and applicative lookalikes, they also form causative constructions, and
function to derive transitive verbs from non-verbal bases. In a few languages of the
980 Bradley McDonnell and Christina L. Truong

sample, they also form comparative constructions. For more on the non-applicative
functions of applicative morphology in western Indonesian languages, see Truong &
McDonnell (2022).

4 Beneficiaries and recipients

As mentioned in the previous section, all eight languages in our sample have an AC
that licenses a beneficiary and/or recipient AppP. Descriptions of the languages in the
sample do not distinguish between beneficiary and recipient, but the vast majority of
examples from grammatical descriptions refer to recipients. Despite this, we follow the
terminology of the descriptions of western Indonesian languages and refer to them as
benefactive applicative/AC. In all but Sundanese, benefactive applicatives may only
attach to a subset of transitive bases, including roots with the meanings ‘bring’, ‘take’,
‘make’, ‘buy’, ‘search’, ‘cook’. However, languages in the sample differ in their produc-
tivity across different roots. Consider the examples in (9) from Nasal with the suffix
-kun. The BC is in (a), and the AC is in (b).

(9) Nasal, Recipient AppP (McDonnell fieldnotes)

a. Azma ny-(s)anik buwak (gin anak=nyo).
A. av-make snack for child=3sg
‘Azma made snacks for her children.’ (BC)
b. Azma ny-(s)anik-kun anak=nyo buwak.
A. av-make-appl child=3sg snack
‘Azma made her children snacks.’ (AC)

In the BC, the recipient is expressed in an adjunct PP, which also occurs in Sundanese,
Toba Batak, and West Coast Bajau. In Makasar and Salako, marking the recipient with
a PP is only possible with an apparently borrowed preposition untu’/untuk ‘for’ from
Indonesian. For Makasar, it is apparently only used by younger speakers of the language
(Jukes 2020: 315). In Ampenan Sasak the oblique PP is possible with a preposition umaq,
but it is apparently rare and has only come about in elicitation (Khairunnisa & McDon-
nell in prep). Finally, Quick (2007) does not report an oblique PP construction, and we
did not find any instances of a preposition that would mark a recipient or beneficiary.
Sundanese differs from the other languages in the sample with two different ben-
efactive applicatives: -keun and pang- -keun. The prior is much less productive as an
applicative. It is restricted to a small number of the transitive bases, and licenses a
recipient AppP, as in (10). It may also act as a causative, as in (11).
 28 Applicative constructions in languages of western Indonesia 981

(10) Sundanese, Benefactive -keun (Hanafi 1997: 23)

a. Udi m-(b)uka panto keur kuring.
U. av-open door for 1sg
‘Udi opens the door for me.’
b. Udi m-(b)uka-keun kuring panto
U. av-(b)uka-appl 1sg door
‘Udi opens the door for me.’

(11) Sundanese, Causative -keun (Truong fieldnotes)

a. Jandela peupeus.
window break
‘The window breaks / is broken.’ (BC)
b. Abi m-(p)eupeus-keun jandela.
1sg av-break-caus window
‘I broke the window.’ (Causative)

The benefactive applicative circumfix pang- -keun is much more productive, occurs
with a wide-range of intransitive and transitive bases, and licenses a substitutive ben-
eficiary AppP. However, when the pang- -keun benefactive applicative attaches to an
intransitive base, such as peupeus ‘break’ in (11a), the AC is not monotransitive as we
might expect. Instead, the construction is ditransitive, as in the example in (12).

(12) Sundanese, Beneficiary AppP (Truong fieldnotes)

Euis di-pang-meupeus-keun kalapa ku abi.
E. pv-appl-break-appl coconut by 1sg
‘I broke open a coconut for Euis.’ (AC)

While the BC is intransitive, the AC is ditransitive and appears to have undergone both
causativization and applicativization. One may think that the prefixal element of the
circumfix pang- is a benefactive applicative, but this is not the case. The prefix pang-
never occurs on its own as a benefactive applicative, and with transitive bases pang-
-keun does not have a causative meaning. See examples in (13) below.
In all languages, the benefactive AC may be ditransitive. While few descriptions (i.e.
Quick 2007, Miller 2007, Jukes 2013) discuss the core/oblique status of the beneficiary/
recipient AppP and patient/theme companion phrase, both may be considered core
since they are unmarked (i.e. not in a PP) in all eight languages.
Languages in the sample differ somewhat in how the benefactive AC interacts with
voice. Seven of the eight languages allow benefactive applicatives in AV. Sundanese has
a general constraint on ditransitive constructions, which must occur in PV for a major-
ity of verbal bases. In the example in (13) below, the recipient is the subject and the
theme is balanjaan ‘shopping purchases’.
982 Bradley McDonnell and Christina L. Truong

(13) Sundanese, Beneficiary AppP expressed as possessed NP (Truong fieldnotes)

a. Icih ny-(c)okot duit.
I. av-take money
‘Icih took the money.’ (BC)
b. Indung di-pang-nyokot-keun balanja-an ku Udi.
mother pv-appl-take-appl shopping-nmlz by U.
‘Udi took the shopping purchases (in the house) for mother.’ (AC)

When this construction occurs in AV, it is most often monotransitive with a theme or
patient being selected as the P argument. If the beneficiary is overtly mentioned, it is
most commonly expressed as the possessor of the theme or patient NP, which is shown
in (14) below. Note that the beneficiary may also be unrealized when understood from
context or overtly expressed as a PP (see Nasal examples in (20)–(21) below).

(14) Sundanese, Beneficiary AppP (Truong fieldnotes)

Udi m-(p)ang-nyokot-keun balanja-an indung.
U. av-appl-take-appl shop-nmlz mother
‘Udi took mother’s shopping purchases (in the house for her).’ (AC)

In examples like (14), the beneficiary is understood to be the same entity as that expressed
by the possessor, likely resolved through pragmatic inference. While these possessive
beneficative AC constructions are common cross-linguistically (Kittilä & Zúñiga 2010:
19–20), they have not been reported in other languages in our sample. They have been
reported in at least one other western Indonesian language, Balantak (see van den Berg
& Busenitz 2012: 100–101).
In AV constructions in Pendau, the recipient AppP and its companion phrase may
occur in either post-verbal position, as in (15).

(15) Pendau, Recipient AppP (Quick 2007: 305)

a. Nong-oli-a’ io vea a’u.
av.rls.sf.tr-buy-appl 3sg rice 1sg
‘I bought him rice.’ (AC)
b. Nong-oli-a’ vea io a’u.
av.rls.sf.tr-buy-appl rice 3sg 1sg
‘I bought him rice.’ (AC)

In the remaining six languages in the sample, the recipient AppP occurs before the com-
panion phrase in AV, as we saw in (9b).
The status of the beneficiary/recipient AppP in PV or passive constructions is not
reported for Makasar, Toba Batak, or Salako. However, in the languages for which we
can describe this property, the beneficiary/recipient AppP must be the subject, as in the
examples from Nasal in (16) and West Coast Bajau in (17).
 28 Applicative constructions in languages of western Indonesia 983

(16) Nasal, Recipient AppP in PV

anak=nyo di-sanik-kun Azma buwak.
child=3sg pv-make-appl A. snack
‘Azma made her children snacks.’ (AC)

(17) West Coast Bajau, Recipient AppP in PV and passive (Miller 2007: 278–280)
a. Boi ∅-sembali emma’=ku kambing e ta’ Saiman.
cmpl pv-slaughter father=1sg.nsbj goat dem loc S.
‘My father slaughtered the goat for Saiman.’ (BC)
b. Boi ∅-sembali-an emma’=ku Saiman kambing tu.
cmpl pv-slaughter-appl father=1sg.nsbj S. goat dem
‘My father slaughtered (for) Saiman the goat.’ (AC)
c. Saiman boi s<in>embali-an kambing le’ emma’=ku.
S. cmpl <pass>slaughter-appl goat loc father=1sg.poss
‘(For) Saiman was slaughtered a goat by my father.’ (AC)

In both languages it does not appear to be possible for the patient companion phrase to
be subject in either a PV or a passive construction (see Miller 2007: 280 for discussion
of West Coast Bajau). In Pendau, either the AppP or the patient companion phrase may
occur in the preverbal position, as in (18). In his description of these examples, Quick
(2007: 291–292) states that the recipient AppP io 3sg is the subject whether it is in the
preverbal subject position as in (18a) or the patient companion phrase is in the prever-
bal position as in (18b). It is unclear to us why this is the case.

(18) Pendau, Recipient AppP in PV (Quick 2007: 292)

a. io ni-po-gabu-a’=’u vea
3sg pv.rls-st.fa-cook-appl=1sg.nsbj raw.rice
‘I cooked rice for him/her.’ (AC)
b. vea ni-po-gabu-a’=’u io
raw.rice pv.rls-st.fa-cook-appl=1sg.nsbj 3sg
‘I cooked rice for him/her.’ (Passive AC)

In Makasar, both arguments in transitive constructions are indexed on the verb, as

in the BC in (19a). However, the maximum number of arguments indexed on the verb
is two, and in ditransitive constructions, it is the A argument and the recipient that
are indexed (Jukes 2020: 314–315). In the AC, the verb is suffixed with -ang, and the
recipient/beneficiary AppP is indexed on the verb (but the patient/theme argument
is not).
984 Bradley McDonnell and Christina L. Truong

(19) Makasar, Recipient AppP (Jukes 2020: 314–315)

a. ku=balli=i baju=a
1=buy=3 shirt=def
‘I bought the shirt.’ (BC)
b. ku=balli-ang=ko baju.
1=buy-appl=2fam shirt
‘I bought you a shirt.’ (AC)

As mentioned in Section 2, arguments of the verb are commonly unrealized in the lan-
guages of the sample. In benefactive ACs, it is possible—and even common in the case
of Ampenan Sasak—for the beneficiary/recipient AppP to be unrealized (Khairunnisa &
McDonnell in prep). In such cases the benefactive meaning is still present in the clause.
Consider the Nasal example in (20). Sundanese is the only other language that we know
of where this is possible, the remaining five languages do not report this possibility.

(20) Nasal, Unrealized recipient AppP (McDonnell fieldnotes)

Azma ny-(s)anik-kun buwak.
A. av-make-appl snack
‘Azma made snacks (for them).’ (AC)

In Nasal and Sundanese benefactive ACs, it is possible for the beneficiary/recipient

AppP to occur as an oblique PP even though the verb is suffixed with the benefactive
applicative, as in Nasal in (21).

(21) Nasal, PP Recipient AppP (McDonnell fieldnotes)

Azma ny-(s)anik-kun buwak gin anak=nyo.
A. av-make-appl snack for child=3sg
‘Azma made snacks for her child.’ (AC)

Thus, it is possible for the beneficiary to occur in a PP with an applicative suffix on the
verb, as in (21), or without an applicative suffix, as in (9a). In some sense, the double
marking of the applicative and the preposition may call into question the applicative
function of the suffix in such constructions. However, if the recipient argument is
unmarked, the applicative suffix must be present, as in (9b), and if the applicative suffix
is present but the recipient is unrealized, the AC must be interpreted as benefactive,
as in (20).7 Taken together these properties provide evidence that these affixes indeed
mark ACs.

7 Analogous constructions have been described at length in Standard Indonesian (see e.g. Cole & Son
2004).
 28 Applicative constructions in languages of western Indonesia 985

5 Instruments and themes

Six of eight languages in our sample have an AC that licenses an instrument AppP: Toba
Batak, Nasal, West Coast Bajau, Sundanese, Pendau, and Makasar. In all six languages in
which instrumental ACs are attested, the applicative suffix used is the same as the one
that licenses a beneficiary/recipient AppP.8 Instrumental ACs are not reported for Salako
or Ampenan Sasak. The base verb in an instrumental AC is commonly transitive. Exam-
ples with base verbs meaning ‘to hit’, ‘to strike’, ‘to chop’, and ‘to buy’ are frequently
found, but also attested are ‘to spear’, ‘to shoot’ , ‘to scoop up’, ‘to pay for’, ‘to write’, ‘to
make’, and ‘to kill’. Two examples are given below.

(22) Sundanese, Instrument AppP (Truong fieldnotes)

a. Udi ny-(c)oel sambel maké témpé.
U. av-scoop chili.sauce av.use soybean.cake
‘Udi scooped up chili sauce using (a piece of) soybean cake.’ (BC)
b. Udi ny-(c)oel-keun témpé kana sambel.
U. av-scoop-appl soybean.cake onto chili.sauce
‘Udi used (a piece of) soybean cake to scoop up chili sauce.’ (AC)

(23) Toba Batak, Instrument AppP (van der Tuuk 1971 [1864–1867]: 103)
a. mangombak tanggurung ni horbo dohot indalu
av.strike back of buffalo with rice.pestle
‘to strike the back of a buffalo with a rice pounder’ (BC)
b. mangombak-kon indalu tu tanggurung ni horbo
av.strike-appl rice.pestle onto back of buffalo
‘to strike a rice pounder on the back of a buffalo’ (AC)

As seen in these examples, the companion phrase for an instrument AppP is typically a
patient or goal (i.e., an endpoint of directed motion). In BCs, this patient/goal participant
is the P argument. The instrument is realized as an oblique PP or adverbial phrase, if it
is expressed.9 In Nasal, West Coast Bajau, and Toba Batak, it appears that instrumental
ACs are exclusively monotransitive. Sundanese follows the same pattern with a single
exception; when -keun is suffixed to beuli ‘buy’ the AC is ditransitive. As seen in the AC
examples above, the instrument AppP is an unmarked core argument and the compan-
ion phrase is expressed as an oblique PP. Following Zúñiga & Kittilä (2019), we refer
to these as remapping ACs because the companion phrase shows a change in coding
compared to the BC.

8 In Sundanese, the -keun suffix marks instrumental ACs as well as benefactive ACs with certain verbs.
9 Instruments may also be expressed as the complement of the verb meaning ‘to use, to wear’, e.g. Sun-
danese maké. In some languages, this verb appears to be grammaticalized as a preposition.
986 Bradley McDonnell and Christina L. Truong

In Pendau, all instrumental ACs are ditransitive and are only found in PV. In such
constructions, the instrument AppP and the companion phrase are both considered
core arguments, as in (24).

(24) Pendau, Instrument AppP (Quick 2007: 297–298)

a. Paee ro-sunung nijimo nu=uram.
rice pv.irr-burn 3pl.nsubj instr=medicine
‘They burned (or smoked) the rice with medicine (for medicinal purposes).’
(BC)
b. Uram ro-po-sunun-a’ nijimo paee
medicine pv.irr-sf.instr-burn-appl 3pl.nsubj rice
‘They burned (or smoked) the rice with medicine (for medicinal purposes).’
(AC)

In the BC in (24a), the patient argument paee ‘rice’ is the subject. The instrument uram
‘medicine’ is expressed with the genitive case marker nu=, which also marks instru-
ments. In the AC in (24b), the instrument AppP is the subject and the patient compan-
ion phrase is a postverbal core argument. The companion phrase does not receive any
marking, which suggests that it is not oblique. In Makasar, it also appears that instru-
mental ACs are ditransitive (see Jukes 2020: 315–316), as in (25).

(25) Makasar, Instrument AppP (Jukes 2020: 316)

anjo selek=a na=buno-ang=i bali=a
that kris=def 3=kill-appl=3 enemy=def
‘He killed the enemy with that kris. (AC)’

In languages with instrumental ACs, it is not reported whether the instrument AppP has
a special pragmatic status. However, in Sundanese, we find that the instrumental AC is
preferred over the BC if the instrument represents salient or unexpected information;
this is especially the case in PV, where the instrument AppP is the subject (see further
discussion of the realization of AppPs as subject in Section 10 below).
In five of the six languages that have instrumental ACs—Toba Batak, Nasal, West
Coast Bajau, Sundanese, and Makasar—the same applicative marker can also license a
theme AppP, i.e. an entity which changes location in a directed motion event. No clear
examples of theme AppPs are found in Salako, Pendau, and Ampenan Sasak (however,
see the discussion below in Section 7, where the product of a bodily function verb may
be considered a theme).
Instrument AppPs in these languages share semantic characteristics with themes,
because instruments used for chopping, hitting, and similar actions are directed along
a path of motion (see also Kroeger 2007). However, we note that this is not true for all
bases; in ACs formed with verbs like ‘buy’ the instrument AppP (typically some type of
currency) is not necessarily in motion. In these languages, we find multiple types of
 28 Applicative constructions in languages of western Indonesia 987

constructions in which applicative marking coincides with a theme AppP. These are
discussed below according to properties of the base verb. First, theme AppPs are found
in ACs with transitive bases that describe an action of directed motion, e.g. ‘to pelt/
throw (at)’, ‘to spray (at)’, ‘to shoot (at)’. The companion phrase in these cases is a goal
which is expressed as the P argument of the BC. The theme in a BC is an oblique if it is
expressed. ACs in which a theme is selected as an AppP are monotransitive and consid-
ered remapping, with the companion phrase expressed as an oblique PP. An example
from West Coast Bajau is given in (26) with the verb seput ‘spray’. In the BC in (26a) the
goal using e ‘the cat’ is the P argument, and no theme argument is expressed. In the AC
in (26b), the theme dalit ‘venom’ is the AppP and is realized as core argument, while the
goal companion phrase is expressed as an oblique PP with the locative preposition ta’.

(26) West Coast Bajau, Theme AppP (Miller 2007: 290)

a. Using e ai ∅-seput soo dilaw.
cat dem pfv pv-spray snake yesterday
‘A snake sprayed the cat (with venom) yesterday.’ (BC)
b. Ai ∅-seput-an soo dalit ta’ using e
pfv pv-spray-appl snake venom loc cat dem
‘A snake sprayed venom at the cat.’ (AC)

Van der Tuuk (1971 [1864–1867]: 104) describes a similar alternation in Toba Batak for
verbs mamodil ‘to shoot with a gun (av)’ and mangultop ‘to shoot with a blowpipe (av)’.
In both cases, the goal (i.e. the target of the shot) is the P argument. When suffixed with
the applicative -hon, the AppP of “mamodilhon is the bullet or that which acts as such,
as, for example, inal [‘wooden rod for shooting at birds’]”, and the AppP of “mangultop-
pon is the arrow (nakkat)” (van der Tuuk 1971 [1864–1867]: 104). Van der Tuuk goes on
to state that the AppP with these same verbs may be the instrument, i.e. a blowpipe and
gun, respectively. The companion phrase in these ACs is the goal.
Second, theme AppPs are found in ACs with base verbs that describe an act of loco-
motion. Such ACs are found in Toba Batak and West Coast Bajau. In this type of AC, the
theme is semantically similar to a comitative or causand. The base verb in such ACs may
be intransitive or transitive, as illustrated below.
If the base verb is an intransitive locomotion verb, the AC is monotransitive, and
the theme AppP is licensed as a core argument. For example, in Toba Batak, the verb
suffixed with the applicative makkabakkon means ‘to fly away with (s.t.)’ and takes as
a core argument the entity that is flown with (van der Tuuk 1971 [1864–1867]: 1977).
This verb is formed by the addition of the AV prefix maN- and the applicative -hon to the
intransitive verb root habang ‘to fly’.
If the base verb is transitive, the AC is monotransitive and considered remapping,
with the theme AppP realized as a core argument and the companion phrase (a goal
or path) expressed as an oblique PP. Examples are given from West Coast Bajau in (27)
and (28).
988 Bradley McDonnell and Christina L. Truong

(27) West Coast Bajau, Theme AppP (Miller 2007: 238)

a. Aku boi n-(s)embet Azizy engko’ surat.
1sg cmpl av-chase A. with letter
‘I chased Azizy with the letter.’
b. Aku boi n-(s)embet-en surat e ta’ Azizy.
1sg cmpl av-chase-appl letter dem loc A.
‘I rushed the letter to Azizy.’

(28) a. West Coast Bajau, Theme App (Miller 2007: 237)

Ai ∅-keta Pirik suang e.
perf pv-cross P. river dem
‘Pirik crossed the river.’
b. Ai ∅-keta-an Pirik using e pe dembila’ suang.
perf pv-cross-appl P. cat dem to.there across river
‘Pirik carried the cat across the river.’

Note that in (27b) the AppP, surat e ‘the letter’, is semantically similar to a comitative or
causand. The meaning of (27b) is similar to ‘I chased Azizy with the letter’ or ‘I caused
the letter to chase Azizy.’ Likewise in (28b), the AppP, using e ‘the cat’, is semantically
similar to a comitative or causand. In the situation described in (28b), the agent crosses
the river together with the cat and in doing so causes the cat to cross the river.
Third, the applicative suffix is also found on bases that normally select a theme as
the P argument when the verb is unsuffixed. Thus, there is no change in the participant
selected as P in the marked construction, however there is usually a semantic emphasis
on the event as an act of directed motion. This type of construction is found with certain
verbs of transfer in Sundanese, West Coast Bajau, Toba Batak, and Makasar e.g. ‘send/
send to’, ‘to give’, ‘to push’. For instance, with the Sundanese verb surung ‘push’, both
the unaffixed form and the form suffixed with -keun take a theme as P. However, the
applicative verb is only used if the event described is an act of directed motion. Con-
sider the example in (29).

(29) Sundanese, Directed motion construction with -keun (Truong fieldnotes)

a. Keur abi ulin ka lapang aya Pak Haji keur ny-(s)urung roda.
while 1sg play to field exist mister Haji while av-push cart
‘When I was playing at the field, Pak Haji was there pushing his cart.’
(unmarked)
b. Udi tos ny-(s)urung-keun mobil ka imah.
U. already av-push-appl car to house
‘Udi is done pushing the car to the house (i.e. it is now at the house).’ (marked)
 28 Applicative constructions in languages of western Indonesia 989

In (29a), the unsuffixed verb nyurung is used because the act described does not direct
the theme towards some endpoint.10 In (29b), the verb suffixed with -keun is used
because the event represents an act of directed motion. The completive marker tos can
only be used with this verb if the theme has already undergone the intended change
in location. A similar example is given by van der Tuuk for the Toba Batak verb tongos
‘to send’. This verb may take a theme (e.g. sorat ‘letter’) as a core argument whether or
not it bears the applicative suffix -hon, but the suffixed verb is preferred when directed
motion is emphasized and a recipient or goal “is either stated or in the mind of the
speaker” (van der Tuuk 1971 [1864–1867]: 104).

6 Goals, locations, and addressees

The applicative affix that licenses goals, locations, and addressees in the majority of lan-
guages differs from those that license the semantic roles discussed thus far (i.e. instru-
ments, themes, beneficiaries, recipients). We refer to these as goal ACs. The only excep-
tions are West Coast Bajau and Ampenan Sasak, both of which have a single applicative
suffix -an. Goal ACs occur on intransitive and transitive bases in seven of the eight lan-
guages. In Makasar, the goal applicative -i appears to only attach to intransitive bases,
but Jukes (2020) does not explicitly state this restriction. Common intransitive bases
to which the goal applicative attaches include some activity verbs (e.g. ‘swim’, ‘jump’),
posture verbs (e.g. ‘sit’,‘stand’), bodily function verbs (e.g. ‘cough’, ‘vomit’), and verbs of
speaking among others. For all languages except for Makasar, transitive verbs include
‘put’, ‘pour’, ‘plant’, ‘teach’, and ‘pay’. Consider the examples from West Coast Bajau in
(30)–(32).11

(30) West Coast Bajau, Locative AppP (Miller 2007: 283)

a. Sesok e ai pe-rekot ta’ jing.
house.lizard dem pfv intr-stick loc zinc
‘The house lizard has stuck to the zinc.’ (BC)
b. Ai ∅-rekot-on sesok jing e.
pfv pv-stick-appl house.lizard zinc dem
‘The house lizard has stuck to the zinc.’ (AC)

10 The term Pak Haji refers to a distinguished man who has completed a religious pilgrimage.
11 Miller (2007: 283) states that the suffix -an applies “vacuously” and is optional in the BC without any
change in the argument or oblique (see Section 8 for discussion). Thus, in (31a), it is possible to use the
-an suffix without any change in argument structure. However, in (31b), the ditransitive construction
requires the applicative suffix -an.
990 Bradley McDonnell and Christina L. Truong

(31) West Coast Bajau, Goal AppP (Miller 2007: 285, slightly modified)
a. Ai ∅-enna’(-an)=ni gula’ diam kupi’.
pfv pv-place-appl=3sg.nsbj sugar inside coffee
‘(S)he put sugar in the coffee.’ (BC)
b. Ai ∅-enna’-an=ni kupi’ e gula’.
pfv pv-place-appl=3sg.nsbj coffee dem sugar
‘(S)he put sugar in the coffee.’ (AC)

(32) West Coast Bajau, Addressee AppP (Miller 2007: 286)

a. “Buat-in do’ aku bue’ susu, too’ bana kelong=ku
make-pv.imp emph 1sg water milk dry very throat=1sg.nsubj
tu,” ∅-bara’=ni m-aku.
dem pv-tell=3sg.nsbj loc-1sg
“‘Make me some milk, I am very thirsty,” she said to me.’ (BC)
b. Bila teko me-ruma’ bara-an=ni emma’=ni
when arrive loc-house PV-tell-appl=3sg.nsbj father=3sg.poss
uun jomo mu’ lawa’ bana.
exist person there beautiful very
‘When she arrived home, she told her father that there was a very handsome
man there.’ (AC)

In these examples, the locative in (30), goal in (31), and addressee in (32) are expressed
in oblique PPs in the BC, but as the AppP, they are unmarked core arguments in the AC.
In West Coast Bajau, the goal AC shows an increase in transitivity in each case compared
to the BC. In Nasal, transitivity increases when the applicative attaches to intransitive
bases and a limited number of transitive bases. Compare the examples of the transitive
base ajakh ‘teach’ in (33) to takhuk ‘plant’ (34).

(33) Nasal, Goal AppP, valence-increasing AC (McDonnell fieldnotes)

a. yo agi ng-ajakh baso Nasal khan anak=ku.
3sg prog av-teach language Nasal with child=1sg.poss
‘I am teaching Nasal to my child.’ (BC)
b. yo agi ng-ajakh-i anak=ku baso Nasal.
3sg prog av-teach-appl child=1sg.poss language Nasal
‘I am teaching my child Nasal.’ (AC)

(34) Nasal, Locative AppP, remapping AC (McDonnell fieldnotes)

a. be-bibai-an n-(t)akhuk jagung di sawah.
distr-woman-distr av-plant corn loc rice.paddy
‘The women planted corn in the rice paddy.’ (BC)
 28 Applicative constructions in languages of western Indonesia 991

b. be-bibai-an n-(t)akhuk-i sawah khan jagung.

distr-woman-distr av-plant-appl rice.paddy with corn
‘The women planted corn in the rice paddy.’ (AC)

In the BC in (33a), P expresses what is being taught, while the goal companion phrase
(i.e. the person being taught) is an oblique PP. In the AC, both the goal and the compan-
ion phrase are unmarked and the construction is ditransitive.
In the BC in (34a), P is expressed as a core argument and the location compan-
ion phrase (if it is expressed) occurs in an oblique PP. In the goal AC in (34b), the loc-
ative is expressed as an unmarked core argument while the companion phrase (if it
is expressed) occurs in an oblique PP. Sundanese goal ACs are similar to Nasal. Some
transitive bases result in a ditransitive AC, while others are considered remapping.
However, it is unclear how transitive bases pattern when suffixed with a goal applica-
tive in Salako and Toba Batak.
Pendau behaves similarly to West Coast Bajau in increasing transitivity in all goal
ACs with the additional complication that the goal applicative suffix -i only increases
the transitivity of transitive bases when it co-occurs with a stem-former prefix, which
was mentioned in Section 2. The AC in (35) requires both the stem-former and the goal
applicative -i.

(35) Pendau, Locative AppP with stem-former (Quick 2007: 301)

a. bau ’uo ni-alap ni=kai ri=payangan.
fish yonder pv.rls-take pn=grandfather loc=boat
‘The grandfather took the fish in the boat.’ (BC)
b. payangan ni-pong-alap-i ni=kai bau ’uo
boat pv.rls-sf.tr-take-appl pn=grandfather fish yonder
‘The grandfather took the fish in the boat.’ (AC)

Another exceptional example is found in Pendau where the AppP in goal ACs may also
occur in a PP. According to Quick (2007: 300), this PP can even be the subject, evidenced
by its preverbal position in (36).

(36) Pendau, Locative AppP in PP (Quick 2007: 300)

ri=bongkarong=’u ni-pong-soput-i=’u
loc=hut=1sg.poss pv.rls-sf.tr-shoot-appl=1sg.nsbj
‘I shot (it) at/beside my hut.’ (AC)
992 Bradley McDonnell and Christina L. Truong

7 Other applied arguments

When applicative suffixes attach to intransitive bases, they often licence a P argument
that takes on various semantic roles. Such transitivizing ACs are common across all
eight languages. In many cases, there is no clear monoclausal BC equivalent. To express
the same semantic role in a BC, a subordinate clause or parallel clause often must be
used.
With bases that describe acts of speaking and cognition, an applicative suffix com-
monly licenses a content AppP that is realized as the P argument. The referent of the AppP
may be a topic, proposition, or reported speech. An example is given from Sundanese with
the verb carios ‘talk’ in (37) below. Other examples include Toba Batak mang-hata-hon ‘to
talk about’ from hata ‘word, talk’ (see Schachter 1984: 103) and Sasak pikir-an ‘to think
about’ from pikir ‘to think’ (Khairunnisa & McDonnell in prep).

(37) Sundanese, Content AppP (Truong fieldnotes)

a. Abi ny-(c)arios ka mama, “Ma, abi hoyong miliarian damel.”
1sg av-talk to mother mom 1sg want av.look.iter work
‘I said to my mother, “Ma, I want to look for work.”’ (BC)
b. Hayang urang kempel ny-(c)arios-keun pa-damel-an.
let 1pl gather av.talk-appl nmlz-work- nmlz
‘Let’s meet up and talk about the job.’ (AC)

A number of languages also show an applicative alternation with the verb meaning ‘to
tell (a story)’. This is found in Salako, where ba-curità means ‘to tell (intr.)’ but ny-(c)
urità-ʔàtn means ‘to tell (s.t.)’, and also in Sundanese, i.e. ny-(c)arita ‘to tell a story’
cf. ny-(c)arita-keun ‘to tell (s.t.), to tell about (s.t.)’. A similar derived verb is found in
Makasar, i.e. pau-ang ‘to tell (s.t.)’ from pau ‘story’.
With verbs describing emotional states or responses, applicative suffixes commonly
license a stimulus AppP that is realized as the P argument. This is found in Toba Batak,
Nasal, Sundanese, Sasak, Pendau and Salako. In corresponding BCs in these languages,
the stimulus may be realized as an oblique PP, as in the Salako example in (38) but in
some cases there is no monoclausal equivalent to the AC as in the Sasak example in (39).

(38) Salako, Stimulus AppP (Adelaar 2005: 92)

a. Berà sidi ià ka Neʔ Kulup.
angry very 3 loc pn K.
‘He was extremely angry at Kulup.’ (BC)
b. Tarutama bapaʔ=e karas sidi m-(b)era-iʔ ià. . .
especially father-3.poss hard very av-angry-appl 3
‘Especially his father got very angry at him. . .’ (AC) (Adelaar 2005: 86)
 28 Applicative constructions in languages of western Indonesia 993

(39) Sasak, Stimulus AppP (Khairunnisa & McDonnell in prep)

a. Ie takut.
3.sg afraid
‘(S)he is afraid.’ (BC)
b. Ie takut-an berarak.
3.sg afraid-appl spider
‘(S)he is afraid of spiders.’ (AC)

In Toba Batak, emotion verbs in this type of AC take the fossilized prefix ha- in addition
to the applicative suffix -i on the verb as in ma-tahut ‘to be afraid’ > mak-ka-tahut-i ‘to
be afraid of (s.t.)’ (van der Tuuk 1971 [1864–1867]: 134). In Sundanese, similar construc-
tions require the use of the prefix CVng- (partial reduplication) which also indicates
greater intensity, as in (40) below.

(40) Sundanese Stimulus AppP (Truong fieldnotes, based on Hanafi 1997: 22)
a. Mariam ceurik lantaran indung=na maot.
M. cry because mother=3sg.poss die
‘Mariam cried because her mother died.’ (BC)
b. Mariam ny-(c)eung-ceurik-an indung=na.
M. av-rdp-cry-appl mother=3s.poss
‘Mariam cried intensely about her mother.’ (AC)

In Sundanese, Makasar and Sasak, an applicative suffix licenses a P argument with

intransitive base verbs describing bodily processes. Examples include Sasak batok-an
‘to cough up (s.t.)’ from batok ‘to cough’, Sundanese utah-keun ‘to throw up (s.t.)’ from
utah ‘to vomit’, and Makassar nata’-me-áng=i cera’ ‘he is pissing blood’ cf. at-ta’-mea=i
‘he is urinating’. In such cases the AppP may be considered a type of theme. An example
of an intransitive BC and a transitive AC with a bodily function verb is given below.

(41) Sundanese, AC with bodily function verb (Truong fieldnotes)

a. Icih utah.
I. vomit
‘Icih vomited.’ (BC)
b. Icih ng-utah-keun udang=na
I. av-vomit-appl shrimp=def
‘Icih vomited up the shrimp.’ (AC)

Applicative suffixes are also found on intransitive verbs of perception. For example,
in Toba Batak, the intransitive verb marnangi means ‘to have ears, to be able to hear’
and the transitive applicative verb manangikan means ‘to hear (s.t.), to listen to (s.t.)’
(Van der Tuuk 1977:101). Van der Tuuk writes that the AppP of manangihon “is some-
thing to/for which one listens in order to catch it, either a distant sound, or a word
994 Bradley McDonnell and Christina L. Truong

towards which one directs one’s hearing” (1977:101). See Section 8 below for discussion
of applicative suffixes used with an intensifying effect on transitive verbs of perception.
Van der Tuuk (1971 [1864–1867]: 103) also reports that Toba Batak -hon may license
a P argument expressing a reason but does not give clear examples showing the clausal
syntax of such constructions. However, ACs where the AppP expresses a reason are
attested elsewhere in the region, as is the case in Tukang Besi (Donohue 1999) and Bal-
atank (van den Berg & Busenitz 2012).

8 Morphological lookalikes
In western Indonesian languages we also find various cases in which the verb is marked
with an applicative suffix, but no syntactic alternation or valency modulation can be
identified in comparison to unmarked forms of the verb.
First, a number of verb roots only occur with an applicative suffix in predicative
use. For instance, in West Coast Bajau the verb sepak-an means ‘to kick (s.t.) backwards’
but there is no non-suffixed predicate ✶sepak, and leba-an means ‘to set (s.t.) down’ but
there is no predicate ✶leba (Miller 2007: 293). The Sundanese verb alung-keun means ‘to
throw (s.t.)’, but there is no non-suffixed predicate ✶alung (at least in some varieties). In
these cases, which appear to result from lexicalization of root + affix, a theme partici-
pant is selected as P, and directed motion is part of the semantic meaning of the suffixed
verb. Quick (2007: 288–289) writes that the Pendau verb bagi ‘to give’ does not occur
without an applicative suffix. Quick shows that bagi-i with the goal applicative –i selects
the theme as the subject in PV and the recipient as an additional core argument, while
bagi-a’ with the benefactive applicative -a’ selects the recipient as the subject in PV and
theme as an additional core argument.
Second, we find sets of related clauses in which the marked and unmarked clauses
show no difference in argument structure or modulation in valency but instead some
purely semantic difference in aspect, manner, or characteristics of the P argument.
The suffix that otherwise marks goal and locative ACs can indicate repeated, itera-
tive, habitual or pluractional aspect without any change in argument structure. Aspec-
tual effects of this type are found with Toba Batak -i, Salako -iʔ, Nasal -i and Sundanese
-an (van der Tuuk 1971 [1864–1867]: 99– 100, Adelaar 2005: 49–50). Aspectual effects
are also attested in Makasar with -i, though this use is “most likely no longer produc-
tive” (Jukes 2020: 306).
Applicative suffixes can also mark intensive or careful action. This is found with
transitive verbs of perception as in Salako nanang-an ‘to watch, to look at (av)’ cf.
nanang ‘to see (av)’ (Adelaar 2005). In Toba Batak, -hon may be used without a change
of argument structure but only intensive/emphatic meaning. Thus pasak means ‘beat’,
and pasak-kon may mean ‘beat with s.t.’ (instrumental applicative function) or ‘do beat
s.t.’ (intensive/emphatic function) (Nababan 1981: 70).
 28 Applicative constructions in languages of western Indonesia 995

With certain verbs in Sundanese, the suffix -keun is associated with individua-
tion or specificity of P. In the example given below, both melak and melakkeun mean
‘to plant’ and select the thing planted as the P argument. In (42a) the unsuffixed verb
is used when the planting of rice is described in general. On the other hand, in (42b),
the suffixed verb is used because the clause refers to planting of a more individuated
referent, in this case, rice seeds which have been prepared by the farmer ahead of
time.

(42) Sundanese, Higher individuation of P (Truong fieldnotes, based on Kustian n.d.)

a. M-(p)elak paré ayeuna mah di sawah, di-sebut=na ny-(s)awah
av-plant rice now prt in rice.field pv-call=3sg av-rice.field
‘Now planting rice in a paddy is called nyawah (making paddies).’
b. Saméméh m-(p)acul ilaharna patani sok sa-sadia-an binih
before av-hoe usually farmer go.ahead rdp-ready-caus seed
paré heula. M-(p)elak-keun binih paré (gabah) di-sebut=na
rice first av-plant-appl seed rice rice.grain PV-call=3sg
tebar.
make.rice.seedling
‘Before tilling the ground usually the farmer prepares rice seeds first. Planting
the rice seeds (grains of rice) is called tebar (making rice seedlings)’.

In Makasar, the verb sare ‘give’ is a ditransitive verb that takes three arguments. When
the unsuffixed form of this verb is used, the theme is always indefinite, is not indexed
on the verb, and may not be unexpressed; an example is shown in (43a) below. When
the applicative suffix -ang is added to sare, the mapping of participant roles does not
change; however there is a semantic difference in that the theme is definite; this is
shown in (43b) below. There may also be a difference in syntactic properties of the
theme argument, as it may now be unexpressed.

(43) Makasar, Change in definiteness of the theme (Jukes 2020: 254)

a. La=ku=sare=ko doe’
fut=1sg=give=2sg money
‘I’ll give you some money.’
b. La=ku=saré-ang=ko doek=ku
fut=1sg=give-appl=2sg money=1sg.poss
‘I’ll give you my money.’

A similar semantic effect is found in West Coast Bajau. The suffix -an is found on many
bivalent and trivalent verbs in AV without a change in argument structure. Miller (2007:
293) analyses this by noting that “when the -an1 suffix does occur, a specific/referential
argument and/or a particular event is involved.”
996 Bradley McDonnell and Christina L. Truong

Finally, lexicalized changes in verbal meaning are also frequently attested. For
example, the Sundanese verb béré means ‘give’, while béré-keun means ‘to hand over’.
For further examples and discussion of the semantic functions of applicative suffixes in
western Indonesian languages see Truong & McDonnell (2022).

9 Applicatives combining with causatives

and other affixes
In western Indonesian languages, causative constructions are typically formed by use
of verbal prefixes and/or by use of the same verbal suffixes that mark applicatives. Pro-
ductive causative prefixes are found in Toba Batak, Makasar and Pendau. In the first
two of these three languages, causative prefixes may combine with applicative suffixes.
However, in Pendau, it does not appear that causative prefixes freely combine with
applicative suffixes.12
In Toba Batak, Van der Tuuk (1977:130) gives examples of verbs formed on transi-
tive bases with pa- and -hon where it appears that -hon licenses a theme AppP, while pa-
introduces a causer argument. Examples include pa-pahat-ton ‘to give (s.t.) to animals
to eat’ from pahan ‘to eat’ and pa-djudjuk-kon ‘to give (s.t.) to someone to carry on the
head’ from djudjung ‘to carry on one’s head’. However, note that -hon is also required
to appear on certain other causative verbs formed with pa- in AV. These appear to be
formed on intransitive or stative bases, and in such cases, the use of -hon does not have
any licensing effect, e.g. jóngjong ‘to stand’ → pa-jóngjong ‘to make to stand (up), pv’ /
pa-jóngok-kon ‘to make to stand (up), av’ (Nababan 1981: 103).
In Makasar, the causative prefix pa- combines with both applicative suffixes -ang
and -i. In some cases the applicative suffix licenses an additional definite argument, as
shown in (44) below.

(44) Makasar, Causative + applicative (Jukes 2020: 290, 296)

a. ku=pa-kanre=i bembe
1=caus-eat=3 goat
‘I made/let him eat goat (meat).’
b. ku=pa–kanre–ang=i bembe=a
1=caus-eat-appl=3 goat=def
‘I made/let him eat the goat.’

12 However, see Quick (2007: 284) for one example in which pa- is analyzed as both a causative marker
and a stem forming prefix when used with the applicative suffix -a’.
 28 Applicative constructions in languages of western Indonesia 997

In other cases, it is difficult to tease apart the functions of the causative prefix and
applicative suffix (see Jukes 2020: 295–297). Similar constructions are also found with
the prefix pi- and the applicative suffixes in Makasar. The prefix pi- has a variety of func-
tions; it forms causative constructions and “derives forms with meanings like ‘(examine/
inspect/listen) carefully or intently’” (Jukes 2020: 299).
In all eight languages one or both applicative suffixes also forms causative con-
structions. The only applicative suffixes for which a causative function is not attested
in the languages of the sample are Makasar -ang and Salako -AN. When the same suffix
has both causative and applicative functions, the distribution of functions is largely
determined by syntactic and semantic properties of the base. The use of such suffixes
with stative base verbs and most intransitive dynamic base verbs commonly results in
only the licensing of a causer argument. With transitive bases, and intransitive bases of
certain semantic subclasses (e.g. speech, perception, emotion as described in the previ-
ous section), the use of such suffixes commonly results in the licensing of a non-A AppP.
Additionally, with certain bases, a single suffix can apparently license both a causer and
an AppP that is not a causand. An example is given from Sundanese in (45).

(45) Sundanese, Portmanteau use of -an (Truong fieldnotes)

a. Cai hujan ng-(k)ucur=na ka solokan.
water rain av-flow=3sg to gutter
‘Rain water flows to the gutter.’
b. Icih ng-(k)ucur-keun cai kana gelas.
I. av-flow-caus water into glass.
‘Icih poured water into a glass.’
c. Gelas di-kucur-an cai ku Icih.
glass pv-flow-caus/appl water by I.
‘The glass had water poured into it by Icih.’

Example (45a) shows that the base verb kucur ‘flow’ is intransitive without any suffix
and takes a single S argument, cai hujan ‘rain water’. In example (45b), -keun has a caus-
ative function with kucur; the verb ng-(k)ucur-keun takes an A argument that expresses
the causer Icih, and a P argument that expresses the causand cai ‘water’. On the other
hand, in example (45c), the suffix -an with kucur has both a causative and applicative
function. It licenses a goal AppP gelas ‘drinking cup’ realized as the subject, while the
causand cai ‘water’ is realized as a core argument. In some languages, the licensing of
a stimulus AppP with a base verb of perception or emotional states requires both an
applicative suffix and another prefix. This was noted earlier for Toba Batak ha-, a fos-
silized prefix of unclear function, and Sundanese CVng-, an intensifier or simulfactive
marker (see also the discussion of Makasar pi- in experiencer-oriented constructions
above).
998 Bradley McDonnell and Christina L. Truong

10 Applicatives and voice

ACs in the languages in the sample show few syntactic restrictions when combining
with AV, PV, and passive voice morphology. In fact, throughout this chapter we have pro-
vided examples where the benefactive, instrumental, theme, and goal ACs occur in both
AV and PV constructions in various languages. Even in Makasar, which has been been
described to have an asymmetrical voice system, applicative suffixes combine with the
actor focus aN- and semi-transitive prefix aN(N)- (see Jukes 2020: 306). Furthermore,
ACs freely combine with passive morphology in the languages that have a ‘true passive’
construction, as in Ampenan Sasak in (46).

(46) Ampenan Sasak, Passive with Goal AppP (Khairunnisa & McDonnell in prep)
a. Dengan tólóq babak bajur leq ramuan.
people put bark bajur loc potion
‘People put bajur tree bark in the potion.’ (BC)
b. Ramuan te-tólóq-an babak bajur siq dengan.
potion pass-put-appl bark bajur by people
‘The tree bark was added to the potion (lit. made thing) by people.’ (AC)

In (46a), the BC is an A-oriented construction: A is the subject and the goal is expressed
in an oblique PP. In the AC in (46b), the verb is prefixed with the passive marker te- and
suffixed with the applicative -an. In this construction, the goal AppP is promoted to the
subject position.
The only syntactic restriction we have noted thus far is in Sundanese where ditran-
sitive constructions may occur in PV but not AV (see Section 4). In Pendau, Quick (2007:
304–305) notes that there are some roots that require the goal applicative -i in PV, as in
(47a), but the same applicative cannot occur in AV, as in (47b). Quick even shows that
it is ungrammatical without the applicative suffix -i in PV and ungrammatical with the
applicative suffix in AV.

(47) Pendau, voice restrictions on AC (Quick 2007: 304)

a. palan ro-guntung-i=nyo
lamp pv.ir-light-appl=3sg.nsbj
‘He/she will light the lamp.’
b. a’u mong-guntung palan
1sg av.irr.sf.tr-light lamp.
‘I will light the lamp.’

A similar restriction is reported by Miller (2007: 192–193) for West Coast Bajau where
the applicative suffix -an is required for some roots to occur in PV but are optional in AV.
Such restrictions likely arise from the grammaticalization of a more general tendency
for PV constructions to be higher in semantic transitivity where P (or the AppP in ACs)
 28 Applicative constructions in languages of western Indonesia 999

is likely to be more highly individuated and affected in discourse. This is not to imply
that applicatives do not occur in AV, but that there appears to be a correlation between
PV and applicative affixes in at least some western Indonesian languages (see McDon-
nell 2016: 214–215). See also Donohue (2001) who shows that there is an overwhelming
preference in Tukang Besi for AppPs to be subjects in discourse.
Despite a general lack of syntactic restrictions on voice in ACs across the sample, the
combination of voice and applicative morphology allows a participant with a periph-
eral semantic role, such as a beneficiary, instrument, goal, or location among others to
be the subject (see Davies 2005 for in-depth discussion of these points in Madurese).
Further, it is well-known that subjects in western Indonesian languages are the syn-
tactically privileged argument and thus play an important role in syntactic operations,
such as relativization and focus (see e.g. Arka 2003, Riesberg 2014). In many of the lan-
guages of western Indonesia, such operations are restricted to the subject or at least
core arguments.
In all eight languages, relativization is marked by a ‘gap’ in the relative clause,
which is coreferential with the head noun, and in all but Makasar and West Coast Bajau
the relative clause is introduced by some sort of linker or relativizer. With the exception
of Ampenan Sasak, the ‘gap’ is restricted to the subject or in the case of Makasar a single
argument in the clause (see below). In Ampenan Sasak, relativization is restricted only
to core arguments (see Khairunnisa 2022: 84–87). Nasal provides a clear example of
how voice and applicatives interact to allow peripheral semantic roles to head a relative
clause. The examples in (48a) and (49a) demonstrate that when P is the head noun, the
predicate in relative clause must be in PV. When the PV construction is marked with
an applicative suffix, it is the AppP that is the head of the relative clause. In (48b), the
predicate in the relative clause is suffixed with the goal applicative -i, and the goal AppP
is the head noun. In (48b), the predicate in the relative clause is suffixed with the instru-
mental applicative -kun, and the instrument AppP is the head noun.

(48) Nasal, Instrument AppP in relative clause (McDonnell fieldnotes)

a. tulis-an [sai di-tulis anak=ku jenu] kak ku=hapus.
write-nmlz rel pv-write child=1sg.poss earlier pfv 1sg.nsbj=[pv]erase
‘I erased the writing that my child wrote (on the wall).’ (BC)
b. sisai [sai di-tulis-i anak=ku jenu] kak
wall rel pv-write-loc.appl child=1sg.poss earlier pfv
ku=bekhesih-kun.
1sg.nsbj=[pv]clean-caus
‘I cleaned the wall that my child wrote on.’ (AC)

(49) Nasal, Goal AppP in relative clause (McDonnell fieldnotes)

a. manuk [sai ku=panggul jenu] lijung.
chicken rel 1sg.nsbj=[pv]hit earlier flee
‘the chicken that I hit earlier ran away.’ (BC)
1000 Bradley McDonnell and Christina L. Truong

b. tungkuk [sai ku=panggul-kun khan manuk jenu] patuh.

staff rel 1sg.nsbj=[pv]hit-appl with chicken earlier break
‘the staff that I used to hit the chicken broke.’ (AC)

In Makasar, relative clauses simply follow the head noun without a relativizer, but the
end of the relative clause is typically marked with a clitic =a that marks definiteness.
It appears that only the P argument in unmarked transitive clauses may be the head
of the relative clause; when A is relativized, the verb is prefixed with the actor focus
aN- or semi-transitive aN(N)- prefix (see Jukes 2020: 228). However, just as in Nasal, a
peripheral semantic role may be the head of the relative clause when it is suffixed with
an applicative, as in (50) where the locative AppP is the head noun and in (51) where the
instrumental AppP is the head noun. These examples are analogous to Nasal examples
in (48b) and (49b), respectively.

(50) Makasar, Goal AppP in relative clause (Jukes 2020: 229)

sikola [na=mange-i=a agang=ku] baji’=i.
school 3=go-appl=def friend=1.poss good=3
‘the school my friend goes to is good.’ (AC)

(51) Makasar, Instrument AppP in Relative Clause (Jukes 2020: 229)

sele’ [ni-buno-ang=a=i] tarang=i.
kris pass-kill-appl=def=3 sharp=3
‘the kris he was killed with was sharp.’ (AC)

Aside from relativization, Jukes (2020: 311) points out that applicatives also allow
peripheral semantic roles to be focused in Makasar, as in (52). In Makasar, arguments
in canonical clauses occur after the verb, but arguments that occur in the preverbal
position receive focus.

(52) Makasar, Focused location AppP (Jukes 2020: 311)

tapper=e’ ku=empo-i.
mat=ec 1=sit-appl
‘I sit on a mat.’ (AC)

West Coast Bajau has a similar focus construction that is restricted to subjects (Miller
2007: 206–207). In this construction, the subject occurs in the preverbal position and is
focused, as in (53). In (53a), the P argument is the subject and thus can be focused in
the preverbal position, but when the predicate is suffixed with the applicative -an the
recipient AppP is subject and can now be focused, as in (53b). Miller (2007: 207) also
demonstrates how non-subject arguments cannot be fronted in the same way. Again,
with the combination of applicative and voice, peripheral semantic roles can be focused
in this way because they are the subject.
 28 Applicative constructions in languages of western Indonesia 1001

(53) West Coast Bajau, Focused beneficiary AppP (Miller 2007: 206)
a. Telumpa’ e boi ∅-beli=ni ta’ Kuzik.
shoes dem cmpl pv-buy=3sg.nsbj loc K.
‘She bought the shoes for Kuzik.’ (BC)
b. Kuzik boi ∅-beli-an=ni telumpa’ e dilaw.
K. cmpl pv-buy-appl=3sg.nsbj shoes dem yesterday
‘She bought Kuzik the shoes yesterday.’ (AC)

11 Conclusion
In this chapter, we have surveyed ACs in a sample of eight Austronesian languages of
western Indonesia. While the affixes that mark ACs are polyfunctional with numerous
non-applicative functions, the primary focus of this chapter is on the applicative uses
of this morphology. Truong & McDonnell (2022) focus on many of the non-applicative
functions.
– Each of the eight languages in the sample has between one and three verbal affixes
that marks ACs.
– The majority of languages have two applicative suffixes: one that marks beneficiary,
recipient, theme, and/or instrument AppPs and another that marks locative, goal
and/or addressee AppPs. Outliers include West Coast Bajau and Ampenan Sasak
which have a single suffix that either marks all four of these AppP (West Coast Bajau)
or all but instrumental AppPs (Ampenan Sasak).
– ACs with benefactive/recipient AppPs generally only occur with transitive bases
and result in an increase in the number of core arguments, such that these ACs are
ditransitive.
– Instrumental ACs license an instrument AppP. In Pendau instrumental ACs are dit-
ransitive, but in all other languages of the sample, instrumental ACs are monotran-
sitive and the (patient or goal) companion phrase is remapped to an oblique phrase.
– Most languages with instrumental ACs also have theme ACs marked with the same
affix. The two constructions share semantic similarities in that like themes, most
instrument AppPs in these languages express an entity that is directed into motion.
Theme ACs are generally monotransitive in the languages of the sample.
– Goals, locatives, and addressees are typically marked with the same applicative
affix. The syntax of these goal ACs is more diverse than both benefactive and instru-
mental ACs. With some exceptions, goal ACs may occur on intransitive and transi-
tive bases, and with transitive bases, may either increase the transitivity or remap
the AppP and the companion phrase.
– When the applicative affixes that mark instruments, recipients, goals, and locations
attach to intransitive bases, they also license AppPs expressing various other seman-
tic roles, including content, stimulus, and product of bodily process. These construc-
1002 Bradley McDonnell and Christina L. Truong

tions result in monotransitive ACs. For many ACs of this type there is no monocalusal
BC equivalent.
– The affixes that function as applicatives may also be morphological lookalikes. Con-
structions marked with these affixes may not license an AppP but instead have a
purely semantic effect such as indicating repeated or pluractional aspect, greater
intensity or properties such as individuation, definiteness, or specificity.
– In the languages of the sample, three of the eight languages have a separate produc-
tive causative prefix. In two of these languages the applicative markers and causa-
tive prefix can freely combine. In all languages of the sample, one or both suffixes
that mark ACs can also mark causative constructions. In a number of ACs, both the
applicative suffix and another prefix must appear on the verb. In some cases, these
appear to no longer have semantic content.
– In western Indonesian languages, the subject plays an important role in syntactic
operations, and the combination of voice and applicative morphology allows periph-
eral semantic roles, such as beneficiaries, instruments, goals, and locations to be the
subject.

Abbreviations
1 first person
2 second person
3 third person
A most agent-like argument of transitive clause
agt agent marker
AC applicative construction
appl applicative
AppP applied phrase
av voice in which A is the privileged syntactic argument
BC base construction
ben benefactive
caus causative
cmpl completive
def definite
dem demonstrative
distr distributive
ec echo VC (epenthetic syllable)
fa factitive (stem-former)
fam familiar
imp imperative
intr intransitive
irr irrealis
iter iterative
loc locative
neg negative
 28 Applicative constructions in languages of western Indonesia 1003

nmlz nominalizer
nsbj non-subject core argument
P patient-like argument of transitive clause
pfv perfective
pl plural
pn personal name marker
poss possessive
prog progressive aspect
prt particle
pv voice in which P is the privileged syntactic argument
rdp reduplication
rel relativizer
S Single argument of intransitive clause
sf stem-former
sg singular
str semi-transitive

References
Adelaar, K. Alexander. 2005. Salako or Badameá: sketch grammar, text and lexicon of a Kanayatn dialect in West
Borneo (Frankfurter Forschungen zu Südostasien Bd. 2). Wiesbaden: Harrassowitz.
Arka, I Wayan. 2003. Balinese morphosyntax: a lexical-functional approach (Pacific Linguistics 547). Canberra:
Pacific Linguistics.
Austin, Peter. 2001. Verbs, valence and voice in Balinese, Sasak and Sumbawan. La Trobe Papers in Linguistics
11(3). 47–71.
Chen, Victoria & Bradley McDonnell. 2019. Western Austronesian Voice. Annual Review of Linguistics 5(1).
173–195. https://doi.org/10.1146/annurev-linguistics-011718-011731.
Cole, Peter & Min-Jeong Son. 2004. The argument structure of verbs with the suffix -kan in Indonesian.
Oceanic Linguistics 43(2). 339–364. https://doi.org/10.1353/ol.2005.0003.
Davies, William D. 2005. The richness of Madurese voice. In I Wayan Arka & Malcolm Ross (eds.), The many
faces of Austronesian voice systems: some new empirical studies. Canberra: Pacific Linguistics.
Donohue, Mark. 1999. A grammar of Tukang Besi (Mouton Grammar Library 20). Berlin; New York: Mouton
de Gruyter.
Donohue, Mark. 2001. Coding choices in argument structure: Austronesian applicatives in texts. Studies in
Language 25(2). 217–254. https://doi.org/10.1075/sl.25.2.03don.
Hanafi, Nurachman. 1997. A typological study of Sundanese. Bundoora: La Trobe University PhD Dissertation.
Himmelmann, Nikolaus P. 2005. The Austronesian languages of Asia and Madagascar: Typological
characteristics. In K. Alexander Adelaar & Nikolaus P. Himmelmann (eds.), The Austronesian languages
of Asia and Madagascar, 110–181. (Routledge Family Language Series). New York: Routledge.
Himmelmann, Nikolaus P. & John U. Wolff. 1999. Toratán (Ratahan) (Languages of the world. Materials 130).
München: LINCOM Europa.
Jukes, Anthony. 2013. Voice, valence, and focus in Makassarese. In Alexander Adelaar (ed.), Voice variation in
Austronesian languages of Indonesia. (NUSA 54), 67–84.
Jukes, Anthony. 2020. A grammar of Makasar: a language of South Sulawesi, Indonesia (Grammars and
Sketches of the World’s Languages. Mainland and Insular South East Asia). Leiden; Boston: Brill.
1004 Bradley McDonnell and Christina L. Truong

Khairunnisa. 2022. Diathesis, grammatical relations, and clitics in ampenan sasak. Honolulu: University of
Hawai‘i at Mānoa PhD dissertation.
Khairunnisa & Bradley McDonnell. in prep. Polyfunctionality and productivity of the verbal suffix -an in
Ampenan Sasak.
Kittilä, Seppo & Fernando Zúñiga. 2010. Benefactives and Malefactives: Typological perspectives and case
studies. Vol. 92 (Typological Studies in Language). Amsterdam: John Benjamins Publishing Company.
Kroeger, Paul. 2007. Morphosyntactic vs. morphosemantic functions of Indonesian –kan. In Annie Zaenen,
Jane Simpson, Tracy Holloway King, Jane Grimshaw, Joan Maling & Christopher D. Manning (eds.),
Architectures, rules, and preferences: Variations on themes of Joan Bresnan, 229–251. Stanford, CA: CSLI
Publications.
Kroeger, Paul & Sonja Riesberg. 2023. Voice and transitivity. In Alexander Adelaar & Antoinette Schapper
(eds.), The Oxford Guide to the Malayo-Polynesian Languages of South East Asia. Oxford; New York: Oxford
University Press.
Kurniawan, Eri. 2013. Sundanese Complementation. Iowa City: University of Iowa PhD dissertation.
Kustian. n.d. Nyawa. https://basasunda.com/nyawa/.
Laskowske, Douglas. 2016. Voice in Bugis: An RRG Perspective. Grand Forks: University of North Dakota MA
thesis.
McDonnell, Bradley. 2016. Symmetrical voice constructions in Besemah: A usage-based approach.
Santa Barbara: University of California, Santa Barbara PhD dissertation.
Miller, Mark. 2007. A Grammar of West Coast Bajau. Arlington: The University of Texas at Arling- ton PhD
dissertation.
Musgrave, Simon. 2001. Non-subject arguments in Indonesian. Melbourne: University of Melbourne PhD
dissertation.
Nababan, P. W. J. 1981. A grammar of Toba-Batak (Pacific Linguistics 37). Canberra, A.C.T: Research School of
Pacific Studies, Australian National University.
Quick, Phil. 2007. A grammar of the Pendau language (Pacific Linguistics 590). Canberra: Pacific Linguistics.
Riesberg, Sonja. 2014. Symmetrical voice and linking in western Austronesian languages (Pacific Linguistics
646). Berlin: De Gruyter Mouton.
Schachter, Paul. 1984. Semantic-role-based syntax in Toba Batak. In Paul Schachter (ed.), Studies in the
Structure of Toba Batak, (UCLA Occasional Papers in Linguistics 5), 122–149.
Truong, Christina & Bradley McDonnell. 2022. Neglected functions of western Indonesian applicatives. In
Sara Pacchiarotti & Fernando Zuñiga (eds.), Neglected syntactic functions and non-syntactic functions of
applicative morphology. Berlin; Boston: De Gruyter Mouton.
van den Berg, René & Robert L. Busenitz. 2012. A grammar of Balantak, a language of Eastern Sulawesi
(SIL E-Books 40). SIL. https://www.sil.org/resources/publications/entry/49492 (29 October, 2021).
van der Tuuk, H.N. 1971 [1864–1867]. A grammar of Toba Batak [Tobasche spraakunst]. Trans. by Jeune
Scott-Kemball (Translation Series, Koninklijk Instituut Voor Taal-, Land- En Volkenkunde 13). The
Hague: Nijhoff.
Wouk, Fay. 1984. Scalar transitivity and trigger choice in Toba Batak. In Paul Schachter (ed.), Studies in the
Structure of Toba Batak (UCLA Occasional Papers in Linguistics 5), 195–219.
Zúñiga, Fernando & Seppo Kittilä. 2019. Grammatical voice (Cambridge Textbooks in Linguistics). Cambridge;
New York: Cambridge University Press.
Part III: Theoretical/Comparative outlook
Maria Polinsky
29 Understanding applicatives
Abstract: This chapter presents and analyzes the main criterial properties of applicative
constructions as proposed by Zúñiga and Creissels in their introductory chapter. These
properties are as follows: (i) applicativization, as a subtype of nucleativization, alters
the status of non-agentive non-subject arguments; (ii) the predicate in the applicative
construction has overt morphological marking that sets it apart from the base-construc-
tion predicate; (iii) the applied phrase has a role other than S or A and refers to a par-
ticipant that either requires a non-core coding different from its coding in the base con-
struction or cannot be expressed at all in the latter. While generally in agreement with
these criteria, the chapter proposes some adjustments to and expansions of all three of
these properties. Concerning property (i), the chapter discusses the connection between
causativization and applicativization, which is manifested through common causa-
tive-applicative syncretism. With regard to (ii), it argues for loosening the restriction on
overt morphological marking of the applied verb, which would allow us to capture lan-
guage-internal inconsistency in applicative marking and include agreement exponents
as signals of the applicative as well. With respect to property (iii), the chapter advo-
cates the inclusion of constructions with an external object possessor (object possessor-
raising constructions) in the range of applicatives.

1 Introduction
In their introductory chapter, which informs this volume, Fernando Zúñiga and Denis
Creissels (henceforth Z and C) define the relation between the applicative construction
and the base construction using three criteria. The order in which I present them is
different from the order used in their chapter.
First, the participant that is agentive or subject-like (referred to as A or S,1 depend-
ing on transitivity) in the base construction retains these properties in the applicative
construction. This requirement is needed to capture the generalization that applicativi-
zation alters the status of non-agentive/non-subject arguments, in contrast, for instance,

1 A stands for Agent or most agent-like argument of a transitive/ditransitive clause, and S stands for the
sole argument of an intransitive clause.

Acknowledgments: I acknowledge the support by NSF (grants BCS-1563129 and BCS-1941733) and CNRS/
Paris-8, where I was an invited professor in October and November 2022. I am grateful to Leston Buell, Denis
Creissels, and especially Fernando Zúñiga for helpful discussions. All errors are my responsibility.

https://doi.org/10.1515/9783110730951-029
1008 Maria Polinsky

to passivization. This criterion is also needed to separate applicativization from causativ-

ization, which, as Z and C indicate, “target[s] an instigating agent that appears in A role”.
The second operational criterion has to do with overt marking of the predicate;
the predicates in the base construction (BC) and in the applicative construction (AC)
need to be marked differently. Finally, the applicative construction naturally includes
an applied phrase: “a noun phrase in a role other than S or A, . . . which refers to a
participant that either requires a non-core coding different from its coding in the AC or
cannot be expressed at all in the BC”.
It is commendable that the applicative construction is well defined operationally,
as this allows us to follow strict criteria for establishing and comparing ACs across lan-
guages and also allows us to separate true ACs from what Z and C aptly call “lookalikes”.
At the same time, there is a worry that the proposed operational criteria are too restric-
tive and may force us to miss some generalizations. In what follows, I will discuss each
criterion in turn (§§ 2–4), and it is my hope that this discussion will stimulate further
debate about the nature of applicative constructions both empirically and theoretically.

2 Generalized applicative: Applicatives

and causatives as two subtypes of nucleatives
All definitions of applicatives, no matter how narrow or wide, are constructed in such
a way that the subject, the agent, or the most agent-like argument is not affected by
applicativization. Informally, that means that applicativization involves operations on
structurally lower arguments: non-subjects. In addition to this constraint, one that refers
to grammatical relations, the applied argument does not associate with agent-like inter-
pretation. In keeping with this reasoning, Z and C further distinguish causativization
and applicativization as two distinct ways of introducing a core argument:

[A]pplicativization is a subtype of verb-coded valency alternation: it is a special case of nucleativi-

zation, an operation that allows participants not encoded as core terms in the base construction to
be encoded as such in the derived construction. More precisely, applicativization is an instance of
non-S/A nucleativization: applicatives target participants that have a non-S/A role in the derived
construction, unlike operations where nucleativized participants have an S/A role, like causativiza-
tion (which target an instigating Agent that appears in A role) and several other cases of non-caus-
ativizing alternations (which target other participants, like obliques or possessors).

I agree with Z and C in assuming that nucleativization (their term) amounts to adding
a new argument, one that is not going to be external in the resulting applicative con-
struction. Thus, there is an operation that increases valency by adding a non-subject
argument to a particular verb. However, in contrast to Z and C’s approach, I would like
to treat this valency-increasing operation as a more general mechanism, one that does
 29 Understanding applicatives 1009

not distinguish between causativization and applicativization. I see several arguments

in favor of this more general approach.
The first two arguments are empirical; they have to do with the cross-linguistically
common syncretism of applicative and causative and with the interpretative vague-
ness of the newly added argument in a number of cases. I discuss these two arguments
in Sections 2.1 and 2.3. The other two arguments, discussed in Sections 2.2 and 2.4, have
to do with the conceptual premises of grammatical theory and with the interpretation
of the causee role.

2.1 Generalized applicative: Applicativization as addition

of a syntactic object

Quite a few languages use the same affix to form causatives and applicatives (Dixon
2012: 332, 336). This phenomenon has been described as “causative/applicative syncre-
tism” (Shibatani and Pardeshi 2002: 116): a single morpheme has two different func-
tions, typically distinguished based on the semantics of the verb it combines with; see
also Zúñiga and Kittilä (2019: Ch. 2). Alternatively, this phenomenon can be referred to as
generalized applicative or, to follow Z and C’s terminology, as unspecified nucleative.
It appears that such unspecified nucleativization is as common as the more specified
applicativization, and the two can actually co-occur within a single language, as I show
below. Early studies of applicatives were very much inspired by Bantu applicatives, and
those are typically distinct from causatives, so there may also be an element of tradition
in separating the two processes.
Just a cursory look across the globe shows that causative-applicative syncretism is
not limited to a genetic or areal grouping. Austin (2005) and Austin (this volume) discuss
a range of Australian Aboriginal languages where the same affix can have different syn-
tactic effects depending on the verb root that it attaches to. In Pitta-Pitta, for example,
the affix -la- forms causatives when added to non-volitional intransitives (such as ‘fall’)
but forms applicatives when added to volitional intransitives (such as ‘play’) (Austin
2005: 12). In Arawakan languages, the applicative reading is only one of the functions
of a particular exponent, “and probably a secondary one, as its main function is causa-
tive” (Van Gijn, this volume). The causative-applicative syncretism is observed in several
Austronesian languages: Javanese (Hemmings 2013), Old Malay, colloquial Indonesian
(Arka et al. 2009), several languages of Western Indonesia (McDonnell and Truong, this
volume), Mori Belait (Himmelmann 2005: 170), Kambera (Klamer 1998), Kapampangan
(Zúñiga and Kittilä 2019: Ch. 2), Boumaa Fijian (Dixon 1988; Creissels, to appear: Ch.
14), Niuean (Massam 1998, 2020), and Tongan (Ball 2008). It is also reported in Hakha
Lai, spoken in Myanmar (Peterson 2003; 2007). In the Papua New Guinean language
Tauya, the same marker introduces causees and applied objects (Creissels, to appear: Ch.
14). Foley (this volume) also notes the syncretism of applicative and causative marking
in Papuan languages. Mithun (this volume) mentions that the addition of the general
1010 Maria Polinsky

applicative marker -ute- to the verb in Yupik can signal a variety of roles, including
causee.2 In Chukchi, the same prefix r-/-n-, which combines with intransitives verbs, can
have either the causative or the comitative/instrumental applicative meaning (Dunn
1999: 198–201, 210–212); a similar syncretism is observed in the closely related language
Alutor (Koptjevskaja-Tamm and Muravyova 1993). Kartvelian languages use a number
of prefixes (also known as character vowels, pre-radical vowels, thematic prefixes, or
versionizers, the term I use below; see Tuite, this volume, for an extensive discussion
and examples) that typically indicate an increase in valency (Boeder 1969; Lomash-
vili 2011; Nash 2020; Tuite, this volume). Among them, the versionizer a- is ambiguous
between causative and locative applicative; see also Tuite (this volume); Creissels (to
appear: Ch. 14), and consider example (2) below. The causative-applicative syncretism
is found in some Uto-Aztecan languages (see Álvarez González and Estrada Fernández,
this volume; Thornes on Numic, this volume; Harley 2017; Langacker 1977; Tuggy 1983)
and in the isolate spoken in the Pacific Northwest, Ktunaxa (also known as Kutenai), as
discussed by Gatchalian (submitted) and illustrated in examples (5) and (6) below. In the
Yuman language Hualapai, the syncretic verbal marker encodes applicativization on
transitive verbs and causativization on intransitive verbs (Peterson 2007: 64–66). This
list is far from exhaustive, but it is genetically and areally quite diverse.
On the lumping approach, one that treats causativization and applicativization as
different manifestations of the operation that adds a (non-subject) argument, causa-
tive-applicative syncretism is expected, as the two phenomena are different facets of
adding an object and thus increasing valency; the interpretation of the applied object
(its semantic role) is a separate process. If, however, the two processes are considered
different, the causative-applicative syncretism is accidental and needs to be explained
using some other tools.
An argument against the lumping approach comes from the observation that both
syncretic and non-syncretic applicatives are robustly present worldwide and can even
co-occur in the same language. Given this co-occurrence of patterns, one can propose
the splitting approach to causatives and applicatives. Under this approach, we should
distinguish between two kinds of phenomena: (i) those in which causatives and applica-
tives constitute two competing readings of one and the same structure (syncretism), and
(ii) those in which the causatives and applicatives are alternative structures with their
own conventionalized readings.3 The splitting approach is based on the tacit assump-
tion that the addition of an object argument and its semantic-role assignment are parts
of the same grammatical operation. I turn to this issue in the next section, 2.2, and also
offer further considerations in Section 2.3.

2 However, the markers for causative and applicative are different in Inuktitut and Kalaallisut (Mi-
chelle Yuan, p.c.).
3 I am grateful to Fernando Zúñiga for suggesting this alternative to me.
 29 Understanding applicatives 1011

As we try to distinguish between these analytical alternatives, it is worth bearing

in mind that the splitting approach is less economical and specifically requires that we
make predictions as to where phenomenon (i) or phenomenon (ii) is expected. Overall,
the choice between lumping and splitting leads to further questions, including the inter-
pretation of the applied argument, which I will address in Sections 2.3 and 2.4.

2.2 Syntactic objects versus bearers of semantic roles

In addition to the empirical arguments, which have to do with causative-applicative

syncretism and the vagueness of the interpretation when it comes to the applied object,
we should also consider a conceptual argument, one that relies on the separation of
form and meaning. Adding an argument to a clause/verb is a purely morphosyntactic
operation, one that increases the transitivity of a predicate or leads to the rearrange-
ment of its arguments. What all the instances of this operation share is that the added
argument in the applicative construction is not the subject (highest external argument).
If we separate this operation from the operation of interpreting the applied object as
the bearer of a particular semantic role, the syncretism between causativization and
applicativization again becomes quite expected.
Next, we can account for cross-linguistic differences between languages whose
applicative or causative markers are distinct and languages with syncretic markers in
a straightforward manner. A morphological exponent can be specified only for its func-
tion as adding an argument without any reference to the semantic role of that argu-
ment; in other words, its only function is to signal that a new non-subject argument is
added. This is the characterization of a generalized applicative. In Section 2.3, I will
introduce data from Wixárika whose verbal exponent -ri(e), which can introduce either
causees or beneficiaries/maleficiaries, occurs only with two-place verbs. This is an indi-
cation that its distribution is sensitive to the number of arguments in the base construc-
tion and not to their semantic interpretation. Numerous examples of such sensitivity to
the number of arguments are found throughout this book, lending further support to
the purely structural nature of generalized applicatives.
In contrast to a generalized applicative, a morphological exponent can be specified
both morphosyntactically and semantically; in such a case, not only does it mark the
addition of a new argument, it also signals the role of that argument: causee, benefi-
ciary, source, instrument, location, or the like. Such a situation is also common across
languages. For instance, in Hiaki (Uto-Aztecan), there are separate exponents that intro-
duce the causee and the benefactive applicative, and these can co-occur:

(1) Hiaki (Harley 2017: 12)

Maria uusi-ta aa ham-ta-ria-k.
Maria child-acc 3sg.acc break-tr.caus-appl-prf
‘Maria broke it (made it break) for the child.’
1012 Maria Polinsky

Finally, we can also expect to find an applicative marker that specifies a semantic role
of a clausal constituent without adding such a constituent as an object argument. In
their chapter on applicatives in European languages, Zúñiga, Arkadiev, and Hegedűs
(this volume) specifically note that in “some instances in English, German, and Hun-
garian, [applicatives] can be valency-neutral (or even valency-reducing, like with some
be-verbs in German)”. Similarly, Toba (Qom) adds the applicative affix to the transitive
base without increasing verbal valency (Censabella, this volume). More generally, the
separation of semantic-role indexing from the valency-increasing function may be char-
acteristic of X- or D-applicatives (per Z and C’s classification); an adpositional phrase (PP)
becomes obligatory without turning into a syntactic object; hence, it is something like an
obligatory adjunct. Distinguishing arguments from adjuncts is a difficult problem, and
to determine whether a particular expression that does not look like a direct object is an
argument or an adjunct, one needs to examine language-specific characteristics such as
binding, scope, subextraction, and control.4
In sum, separating two functions of applicative exponents, namely, marking a par-
ticular semantic role and marking the general addition of an argument, allows for a
more adequate typology of applicative encoding.5

2.3 The interpretation of the argument role under

causative-applicative syncretism

Related to the syncretism of verbal marking that adds an object to the structure is the
observation that there is often vagueness in the interpretation of the applied argu-
ment’s role. It can be interpreted as a causee, location, possession, comitative, experi-
encer, recipient, or source. Consider some examples. In Georgian, the applied object in
the a-version can be interpreted either as a causee/location (Nash 2022) or as a causee/
comitative-sociative (Nash 2020):6

4 We can also expect to see lexicalized applicatives where the use of an applicative marker on the verb
is no longer associated with argument addition.
5 In formal accounts of applicatives, the co-occurrence of applicative markers is explained in terms
of the structural dominance of individual functional heads and arguments introduced by these heads
(among others, Pylkkänen 2000, 2008; Jung 2014; Harley 2017). This ordering is only partially related to
the issue discussed in this section; it also bears on another issue: the hierarchical relation between the
applied object and the base object.
6 See also Tuite (this volume), who characterizes locative applicative as superessive, and Creissels (to
appear).
 29 Understanding applicatives 1013

(2) Georgian (Nash 2022)

mzia-m txilamureb-i ertmanets mi-a-magr-a.
Mzia-erg skis-nom each.other.dat prv-vers-strong-aor.3sg
‘Mzia made the skis stick together.’
‘Mzia fixed the skis to each other.’

In Korean, the dative object added by the generalized applicative -I-, which surfaces
variously as -i, -hi, -li, or -ki, can be interpreted as adding either a causee or a location:7

(3) Korean (Kim 2009: 25)

emma-ka ai-eykey os-ul ppaliip-hi-ess-ta.
mother-nom child-dat clothes-acc wear-appl-pst-dec
‘Mother dressed the child (made the child wear clothes).’
‘Mother had the clothes put on the child.’

This generalized applicative cannot co-occur with the specialized benefactive applicative
-ecwu-, which indicates that they compete for the same position in the verb structure:

(4) Korean (Jung 2014: 50)

✶
Yenghi-ka Chelswu-eykey ai-eykey chak-ul ilk-hi-ecwu-ess-ta.
Yenghi-nom Chelswu-dat child-dat book-acc read-appl-appl-pst-dec
(Intended: ‘Yenghi made Chelswu read the book for the child.’)

In Ktunaxa, an applied animate argument can be interpreted as a causee or a benefi-

ciary/maleficiary, as shown in (5).8 Note that morphologically, there are two exponents:
the suffix glossed as causative and the suffix -t-, which specifically marks the addition
of an argument.9

(5) Ktunaxa (Gatchalian, submitted: Ex. [51a])

ʔik-ȼi-t-ap-ni kanuhusnana.
eat-caus-val-1sg.obj-ind apple
‘Someone made me eat the apple.’
‘Someone ate the apple on me.’

7 Glosses modified from the original.

8 Bella Coola (Salishan, western Canada) has similar generalized structures (Davis and Sanders 1997).
The two languages are not related but areal similarities may also be behind the common use of gener-
alized structures.
9 Ktunaxa also has an unambiguously applicative marker that adds a beneficiary (Gatchalian, submit-
ted). Thus, there is a rich inventory of argument-increasing exponents in this language, which makes the
observed causative-applicative syncretism even more remarkable.
1014 Maria Polinsky

(6) Ktunaxa (Gatchalian, submitted: Ex. [51b])

mu=hun=hukup-ȼi-t-is-ni k̓ikiⱡ.
pst=1sg.sbj=be.cooked-caus-val-2sg.obj-ind food
‘I made you cook food.’
‘I cooked food for you.’

If nucleativization involves different operations, the ambiguity or vagueness of the

interpretation associated with the applied object seems unexpected; in contrast, on the
assumption that the applicative exponent simply adds an object, rather than an argu-
ment with a particular semantic role, such ambiguity is unsurprising.
Potential support for the interpretive vagueness of the applied object may also
come from diachrony. For example, as observed by Petersen (2007: 135), parallel gram-
maticalization of the same verb ‘give’ in a causative periphrasis and in an applicative
periphrasis may be a possible source of the applicative-causative syncretism.
If the causative and applicative marking are linked, as I propose here, we should
expect the following diachronic developments:
i. narrowing: generalized applicative marker turning into a more specialized marker
of causative or of applicative
ii. expansion: dedicated causative marker assuming the functions of a generalized
applicative marker, or dedicated applicative marker assuming the functions of a
generalized applicative marker

I am not aware of either diachronic development, which could be equally a sign of my

ignorance or, in a more promising way, of an area in comparative linguistics that war-
rants further investigation.
Assuming the syncretism of causative and applicative marking, one wonders if it is
possible to make predictions concerning the semantic role of the argument introduced
by a generalized-applicative marker. As an illustration, let us consider the suffix -ri(e)
in the Uto-Aztecan language Wixárika, also known as Huichol (Bierge 2017: 257 ff.).
The main function of this suffix is to introduce a beneficiary, recipient, or maleficiary;
based on the discussion and examples in Bierge (2017), these participants are always
added to a two-place verb. In the following example, the applied beneficiary object is
cross-referenced in verbal agreement, and this cross-referencing involves the non-sub-
ject exponent mats:

(7) Wixárika (Bierge 2017: 258)

pini ne-mats-u-ti-wewi-rie.
dress 1sg.sbj-2sg.nsbj-vis-pl-make-appl.incompl
‘I made you a dress.’

On the other hand, the same affix -ri(e) is also used to introduce a causee; consider
(8), where the causee is cross-referenced in the verb by the same type of non-subject
 29 Understanding applicatives 1015

marker as in (7) (the persons are different) and the verb is again transitive, with the
inanimate base object:

(8) Wixárika (Bierge 2017: 261)

haa me-nets-u-xɨ-ri-ri-xɨ.
water 3pl.sbj-1sg.nsbj-vis-heat-res.compl-appl.compl-pfv
‘They made me boil water.’

It is worth noting that the two variants of the applicative affix differ in their aspectual
value, with the true applicative being associated with incompletive aspect or possibly
lack of telicity, and the causative -ri(e) being associated with completive aspect/telicity.
At this juncture, it is unclear how strong these correlations are for Wixárika, but they
do not seem to be limited to that single language.
In the analysis couched in terms of Cognitive Grammar, Tuggy (1983) makes a con-
nection between the choice of the causative or applicative meaning and aspect.10 To
paraphrase his analysis, if the semantics of the affixal predicate represents and empha-
sizes the end result of an event, the polysemous suffix is interpreted as causative, and
the argument introduced by the generalized applicative is interpreted as causee. This
causee is viewed as the Ground (topic), and the eventuality is its Figure, and this empha-
sis on the outcome of the event is tied to the telic interpretation.11 Conversely, if the
representation of the affixal predicate emphasizes the process, the polysemous suffix is
likely to be interpreted as applicative; the argument added by the generalized applica-
tive is interpreted as Figure, and the eventuality corresponds to Ground. The outcome
is less likely, which explains the atelic nature of the event or at least the lack of specifi-
cation in terms of telicity.
Based on these preliminary observations, causativization is partially correlated
with telicity, and applicativization is either free of such correlation or is partially cor-
related with a lack of telicity. If this is on the right track, the two interpretations are in
a privative opposition, where the values for the causatives are more specified. On this
interpretation of the data (which need to be probed further), causative-applicative syn-
cretism is more predictable; the choice of the causative interpretation is specified, and
the applicative interpretation is a sort of elsewhere condition.

10 Z and C also note possible correlations between applicativization and aspect or manner (see § 1.2.2 of
their chapter) but tend to consider them only under the rubric of lexicalized applicatives.
11 See Talmy (1978; 2000: Ch. 5), Langacker (1993) on the notions of Figure and Ground.
1016 Maria Polinsky

2.4 How agentive is the causee?

Z and C’s rationale for separating causativization from applicativization may stem from
the need to maintain the non-agentive interpretation of the applied object; a causee on
the other hand is agentive, interpreted as S or A. But is this distinction so strong that
is forces us to consider the process of causativization to be completely separate from
applicativization? Causatives constitute a broad set of constructions, of course — con-
sider the distinction between faire-infinitif and faire-par causatives in Romance (Folli
and Harley 2007; Ippolito 2000; Kayne 1975, among others);12 the discussion below only
relates to the former type.
First, there is a subset of causees that have fewer agentive properties than the pro-
totypical external argument. This lack or deficiency of agentive properties has been
noted, in particular, with respect to causatives of verbs of cognition and perception
(whose subject is not agentive to begin with and is likely to be interpreted as experi-
encer). If a generalized applicative adds such an argument as causee, it is hard to tell
it apart from other types of recipients, and assigning it the agentive interpretation is
problematic.
Going beyond the experiencer subtype, many discussions of causative construc-
tions suggest that a causee is inherently less agentive than a prototypical agent/causer;
after all, the event in which the causee appears as S or A is instigated by another partic-
ipant, and causee may share properties with a comitative or patient (e.g., Bisang 2006;
Dowty 1991; Gerdts 2003; Nash 2020; Shibatani and Pardeshi 2002; Wali 1981, to name
a few).
In particular, causees in causatives of transitives are often interpreted as being
similar to the instrument, comitative, or associate argument; consider the discussion
of “sociative causatives” by Shibatani and Pardeshi (2002). This is not surprising, and
there are at least two considerations related to this ambiguity of the causee role. First,
causer-related causation already includes the external argument, whose semantic role
is identified as the external initiator of the caused event. The causee is not acting fully
on its own volition; rather, the initial motivation for the event comes from the causer,
and the causee is more of an “associate”. This is related to the second consideration, one
that has to do with the decomposition of semantic roles into proto-properties (Dowty
1991). Among other properties, a prototypical agent causes an event or change of state
in another participant, whereas a prototypical patient is causally affected by another
participant (Dowty 1991:572). In causative constructions, the causee is construed as
affected by the causer, which can lead to de-agentivization of the former.

12 The two types of causatives differ in a number of ways, including the encoding of causee (the causee
of a transitive embedded verb is marked with dative case in faire-infinitif and by a preposition in faire-
par causative) and the omission of causee (it can only be omitted from faire-par causatives); see Folli and
Harley (2007), Kayne (1975) for a detailed list of differences.
 29 Understanding applicatives 1017

As noted by Kittilä (2013: 126; see also Kittilä 2009),

the result of causation is a division of agentive features, but the degree of ‘overall agency’ is main-
tained. Because the original agent —the causee in the caused event— is made to act by an external
causer, causation deprives it of certain agentive features and makes it more patient-like. . .. The
participant in question is somehow affected by the event it partakes in, which is also manifested
in case marking.

If we accept that causees are less agentive than prototypical agents, that removes the
original consideration for treating causativization as a separate process of nucleativi-
zation by Z and C. And in fact, considering causatives in the context of applicatives may
allow us to better understand the properties of a subset of causees.

3 Overt marking of the predicate in the applicative

construction
3.1 Types of applicative markers

Another criterial feature of applicativization has to do with overt marking of the

applicative on the predicate of the applicative construction. In their chapter, Z and C
distinguish four ways in which the predicate of the base construction can differ from
the predicate of the applicative construction (see Table 1 in Z and C’s chapter); I have
slightly modified their wording:

Table 1: Morphological valency alternation types (after Z and C).

Overt voice marking on the No overt voice marking

predicate on the predicate

Asymmetrical Applicative overtly marked —

compared to the base predicate
Symmetrical Both base and applicative Labile verbs
predicate overtly marked

For Z and C, it is crucial that the applicative predicate show greater morphological com-
plexity than the base predicate (asymmetrical overt marking). This is an important obser-
vation, one that implicitly reflects the diachronic sources of applicativization, which
often arises either (i) when an adposition or case marker is incorporated into a verb13

13 The synchronic presence of adpositions in a given language is irrelevant for this process, as an adpo-
sition may have been reanalyzed into an applicative marker on the verb at prior stages of the history of
1018 Maria Polinsky

or (ii) when a verb in a serial verb construction gets reanalyzed as an applicative mor-
pheme (see Censabella, this volume, for the latter grammaticalization route). As evidence
of these routes to applicativization, the same or similar exponents can coexist in a single
language as applicative morphemes and adpositions. As an example, consider Amberer
(this volume) on what he analyzes as the prepositions lə- and bə- and applicative markers
ll and bb observed in Amharic. This transparent correspondence between prepositions
and applicative markers may be an instantiation of diachronic path (i) in that language.
Likewise, applicatives in Northwest Caucasian languages, which make wide use of the
D-type (meaning dative-like), are typically analyzed as predicates that incorporate an
adposition and an agreement marker attached to it (see Letuchiy 2007; O’Herin 2001).14
(Note, however, Arkadiev, Lander, and Bagirokova, this volume, who discuss non-adposi-
tional sources of applicatives.)
In principle, the distinction between applicative markers that trace their parent-
age from adpositions and those that derive from grammaticalized verbs in serial verb
constructions should be relatively straightforward — as long as we are not dealing with
morphologically impoverished languages where distinguishing between lexical cate-
gories takes more work. Polynesian languages are a good case in point. In a number
of these languages, the instrumental/applicative marker Caki coexists with the prepo-
sition (‘)aki. Researchers are divided on the analysis of this marker, some analyzing it
as a preposition incorporated into the predicate (e.g., Massam 1998, 2009, 2020), others
considering it a verb (Ball 2008). The case of Tongan applicatives considered in detail
by Ball (2008) is particularly instructive. An ergative language, Tongan has three types
of constructions with the instrumental expression: the base construction with a prep-
ositional instrumental phrase, as in (9a); the applicative construction whose predicate
is overtly marked by ‘aki, exemplified by (9b); and the so-called double construction,
where ‘aki appears twice, shown in (9c):15

(9) Tongan (Sisilia Lutui, p.c.)

a. Na‘e tofi ‘e he fafine ‘a e mā ‘aki
pst cut erg det woman abs det bread ins
‘a e hele.
abs det knife
‘The woman cut (the) bread with the knife.’

that language. The development of applicative markers from adpositions has been extensively discussed
in analyses of Bantu (see, among others, Creissels, this volume; Pacchiarotti, this volume; Baker 1988;
Nakamura 1997; Buell 2003, 2004; den Dikken 2023).
14 For other sources of applicative marking, see Peterson (2007), Creissels (to appear).
15 The examples here are similar to those used in Ball (2008: 315–319) but instead of proper names, I
use two common nouns.
 29 Understanding applicatives 1019

b. Na‘e tofi ‘aki ‘e he fafine ‘a e hele

pst cut ins.appl erg det woman abs det knife
‘a e mā.
abs det bread
‘The woman cut (the) bread with the knife.’
c. Na‘e tofi ‘aki ‘e he fafine ‘a e mā ‘aki
pst cut ins.appl erg det woman abs det bread ins
‘a e hele.
abs det knife
‘The woman cut (the) bread with the knife.’

With respect to examples such as (9b), Ball (2008: 319) writes:

The term applicative ‘aki construction for [(9b)] is due to the fact that it resembles applicatives in
other languages. This construction has an applicative marker close to the verb. . ., and there is an
added core argument, the instrument.

As Ball shows, ‘aki does not pattern with other Tongan prepositions, though. First of
all, its complement appears in the absolutive (see [9a] and [9c]), which is unusual of
prepositions in general and of Tongan prepositions in particular. In addition to its argu-
ment realization, ‘aki patterns with verbs based on its co-occurrence with conjunctions,
various kinds of morphological alternations, and its co-occurrence with negation (Ball
2008: 321–327). On the other hand, unlike regular Tongan verbs, ‘aki cannot be transi-
tivized or causativized (Ball 2008: 327–330). It appears closer to a grammaticized serial
verb than a preposition, thus instantiating diachronic route (ii) toward applicativiza-
tion mentioned above. At the same time, this discussion is intended to demonstrate
that careful distributional analysis is needed to reach one conclusion or the other in
each particular case. Assuming that ‘aki is synchronically an applicative marker, one
can hypothesize that the applicative argument in (9b) corresponds to P-applicatives in
Z and C’s classification, while the double construction (9c) creates an interesting chal-
lenge for their classification: the verb is marked as applicative but the instrumental
marker does not change. Based on the discussion in Ball’s work (2008), the prepositional
instrument in the double ‘aki construction in (9c) has properties of an oblique object,
while the applied instrument in (9b) has direct object properties. So we are left with the
open question of what criteria of applicativization to use in accounting for the double
construction.
To wrap up the discussion of pathways to applicative marking (including the dia-
chrony of such marking), it is worth noting that the incorporation of adpositions/case
markers and reanalysis of serial/secondary verbs may both be available within a single
language, as these two routes are not mutually exclusive. If so, a question may arise,
one for future research, as to which semantic roles associated with applied arguments
are more likely to be introduced by adpositions, and which by serial/secondary verbs.
1020 Maria Polinsky

3.2 Dedicated applicative marker on the verb?

So far, I have been discussing the privative opposition that Z and C posit for true applic-
atives: the verb in the base construction is not marked, but the one in the applicative
construction is. (The equipollent marking, where the base verb and the applicative verb
are marked in different ways, is something that Z and C set aside. I will not pursue it
further, especially since there are no attested cases of the equipollent marking in tran-
sitives and applicatives.)
Two other issues that I would like to address have to do with inconsistent applica-
tive marking and the role of agreement as applicative marking. With respect to the
former, one can imagine that applicative marking may occur on a subset of verbs but
not on others, thus leading to what may be called “masked applicativization”. As Dixon
(2012: 301) states, some applicative derivations are found only with a subset of verbs,
and it may well be that the overt applicative marker simply alternates with a silent one.
I will return to this issue in Section 4, as I discuss some Tswana examples.
The other question raised by Z and C’s classification that is worth addressing has to
do with the role of agreement as applicative marking. On the current classification, Z
and C consider only those predicates that are overtly marked to be true applicatives. But
the Tongan example shows that it is not always clear what the nature of the applicative
marker is synchronically. In principle, valency alternations, of which applicativization
is one type of instantiation, can be marked in a number of ways: via a dedicated verbal
exponent, as in Z and C’s definition, via the indexing of verbal arguments (agreement,
cliticization), via the incorporation of the internal argument (I will not discuss this last
strategy here).
Finally, one may expect that the verbs in the base construction and the applicative
construction are equally unmarked, that is, lacking an overt affix that sets them apart.
The difference between the base and the applicative construction in such a case is solely
determined by three-way comparisons: (i) between the base object and the applied
object, (ii) between the object-like argument and a PP, and (iii) between the presence
and absence of object (primarily in case of applicatives of intransitives). On such an
approach, any construction that includes a dative/accusative argument could count as
applicative (see Marantz 1993 for such a view). The downside is that we then lose the con-
trast between true ditransitives and applicative ditransitives, whose respective dative/
accusative arguments do not always have the same properties (see, for example, Landau
1999). Further still, languages with the so-called symmetrical ditransitives (where the
two objects behave alike in tests like passivization, pronominalization, scope, or sub-
extraction) can vary in symmetry depending on the combination of semantic roles or
subtypes of applicatives (see van der Wal 2017 for an insightful discussion and further
references).
I will set aside the case where there is no marking whatsoever distinguishing
applicative verbs from base verbs; doing so brings the position discussed here closer to
that of Z and C and also allows me to flag the issues of “masked applicativization” and
 29 Understanding applicatives 1021

“flexible symmetry”, in van der Wal’s terms. But I would like to offer some considera-
tions on the requirement that the applicative-construction predicate have a dedicated
applicative marking, as I find this requirement unnecessarily stringent.
Imagine a situation where the verb is not marked for applicativization but the
added applied argument triggers agreement on the predicate; thus, the agreement pat-
terns could look like this (Table 2):

Table 2: Agreement in base and applicative constructions.

Base construction Applicative construction

Agreement with only one object Agreement with base object Agreement with applied object
Multiple agreement Agreement with base object(s) Agreement with base and applied object

All it takes to separate the privative marking of applicativization from marking applica-
tivization by agreement alone is the applicative marker being null. Amharic, which I
already mentioned above, may illustrate the case in point, if we assume the analysis by
Baker and Kramer (Kramer and Baker 2013; Baker and Kramer 2014), who argue that
Amharic applicatives are marked on the verb by a unitary morpheme, which consists
of a case marker plus a non-subject agreement marker of gender, number, and person.
The evidence in support of this analysis comes from morphosyntactic properties of
the applicative marker, which it shares with the regular object-agreement marker;
namely, in clauses that contain both a main verb and an auxiliary verb, such an agree-
ment marker attaches to the main verb; there can only be one non-subject agreement
marker, and the object agreement marker and the applicative marker cannot co-occur,
indicating that they compete for the same position. Applicative agreement, like regular
object agreement, is possible only with semantically specific noun phrases and induces
a semantic interpretation of emphasis. All these empirical observations suggest that
Amharic may actually instantiate applicativization with agreement and no special
marking; depending on how strictly one would like to apply Z and C’s marking criteria,
the Amharic applicative may either be banished to the land of lookalikes or be used
to expand the proposed typological classification. Verbal applicatives in Cushitic, dis-
cussed by Vanhove (this volume), also seem to be indexed by verbal agreement.
Cliticization may be another sign of an applicative construction, and it may again
occur in the absence of dedicated applicative marking on the verb. For example, in
Romanian, the beneficiary of a transitive can appear in the dative form either when it
is cross-referenced by the clitic on the verb, or not, as shown in (10a) and (10b), respec-
tively. Only when the recipient/beneficiary is cross-referenced by the clitic on the verb,
as in (10a), does it have the properties of an argument, and it is structurally higher than
the direct object, which is manifested in a number of morphosyntactic properties such
as binding or scope (Diaconsecu and Rivero 2007).
1022 Maria Polinsky

(10) Romanian (Cornilescu 2020: 123)

a. Mama le=a copt prăjituri copiilor.
mother.def 3pl.dat=has baked cakes children.def.dat
‘Mother has baked the children cakes.’
b. Mama a copt prăjituri copiilor.
mother.def has baked cakes children.def.dat
‘Mother has baked cakes for the children.’

These empirical data suggest that in addition to dedicated applicative marking, the dif-
ference between base and applicative constructions can also be diagnosed by system-
atic differences in agreement and cliticization.
I will return to the relevance of dedicated applicative marking in the next section,
where I address possible sources of the applied object.

4 Applied objects: Their source and structural size

In setting up the distinction between high and low applicatives, Pylkkänen (2000, 2008)
argues that in high applicatives, the applied object stands in a relation to the entire
event denoted by the predicate, whereas in the low type, the relationship between the
applied and base objects is that of possession or inclusion. Setting aside the implemen-
tation of such an idea (theoretical models can differ widely), we still have a valid empir-
ical observation: the referents of the two objects form a tight relation, which can be
described as possession or a part–whole relation. Assuming that the referent of the
base object denotes a possessed entity, the referent of the applied object can receive
that entity (hence it can be goal, recipient, or beneficiary), be affected by the loss of
that entity (malefactive, source), or stand in static possession of that entity (location,
possessor).16 Given the tight relation between the referent of the base object and the
referent of the applied object, it is not surprising to find an alternation between a pos-
sessor, which is a subconstituent inside a noun phrase denoting the base object, and a
free-standing applied object. The former is present in the base construction, while the
latter is found in the applicative construction.
Alternations of this kind (as well as the alternation between a possessor of S/A and
a free-standing argument) are known in the linguistic literature under different terms.
External possession is a common, and also theory-neutral term: a nominal is syntacti-
cally expressed as a dependent of a verb but is semantically understood as the posses-

16 The instrument and comitative can also be interpreted as being part of the possession/inclusion
relation as long as this relation is construed as spatial, involving proximity between the two entities (see
Landau 1999; 2007).
 29 Understanding applicatives 1023

sor of one of its co-arguments in a clause (Deal 2017).17 Possessor raising (or possessor
ascension, a term that goes back to Relational Grammar) is one of the most common
analyses of external possession. The idea is that the applied argument originates as a
genuine possessor in one of the verbal arguments and then gets “promoted” (raised,
advanced) to an independent argument position. The argument position may be that of
subject (subject possessor raising) or that of object (object possessor raising). Such pro-
motion may or may not be accompanied by special marking on the verb, and the case of
the raised possessor can vary accordingly.
In their definition of applicativization, Z and C include the criterion according to
which the applicative expression “either requires a non-core coding different from its
coding in the [applicative construction] or cannot be expressed at all in the [base con-
struction]”. The way I understand this definition, it allows for object possessor raising
as a type of applicativization, because the possessor in the base construction appears in
“non-core coding”. If one were to exclude external possession from the range of applica-
tive constructions, the definition proposed by Z and C would have to be modified to
exclude expressions that do not originate as verbal co-arguments.
The connection between external possession and applicativization is reinforced
by the observation that quite a few languages use the same verbal exponent (if any)
to mark genuine applicatives and external possessors. For instance, in Chickasaw, the
addition of a dative argument is indexed by what seems to be the generalized applica-
tive prefix on the verb (glossed as III in the literature on Muskogean), and this marker
accompanies regular possessor raising and beneficiary applied objects. Thus, example
(11) is ambiguous:

(11) Chickasaw (Munro 1984: 640)

hattak-at iho-a chipota i-sho’ka-tok.
man-sbj woman-nsbj child appl-kiss-pst
‘The man kissed the woman’s child.’
‘The man kissed the child for the woman’s benefit.’

Thus, it may be desirable to include subcases of external possession under the rubric
of applicativization. Adopting the classification of applied objects proposed by Z and
C, and following their desideratum that the verb be marked, we can expect exter-

17 Creissels (to appear: Ch. 13), following Van de Velde (2020), proposes replacing the notion of an exter-
nal possession construction with that of a concernee–concern construction, one that is broader in scope:

Two NPs in the construction of the same verb form a concernee–concern construction if the involve-
ment of the referent of one of them (the concern) in the event denoted by the verb is determined by its
syntactic role in the construction of the verb, whereas the involvement of the referent of the other one
(the concernee) is simply a consequence of a relationship it has with the concern independently of the
particular event referred to.
1024 Maria Polinsky

nal possessors qua applied objects to be encoded as D-applicatives (i.e., dative-like) or

P-applicatives (i.e., preposition-like), and this applicativization is likely to be optional.
The Chickasaw example has an overt applicative marker, but what happens when
such a marker is present only on some occasions? This issue goes back to the discussion
that I started in Section 3.2. How does the on-off applicative marking interact with the
requirement proposed by Z and C that having an applicative marker on the verb is a
necessary condition on applicativization? Consider the Bantu language Tswana, which
is featured prominently in this collection (see the chapter by Creissels, and also the
chapters by Pacchiarotti and by Z and C). In Tswana, the construction with an external
object possessor appears without verbal marking if the relationship between the base
and applied object is part–whole or inalienable possession, but includes applicative
marking in all other cases of external possession (Creissels, this volume; Creissels to
appear: Ch. 14). Compare the following pair:

(12) Tswana (Creissels to appear: Ch. 14)

a. kì-rém-i-lé sí-tłhàrì dì-kà:là.
1sg.s/a-chop-prf-fv sg-tree.cl17 pl-branch.cl10
‘I chopped (the) branches off the tree.’ (part-whole possession, no applicative
marking on the verb)
b. M̀pʰɔ́ !
ʊ́ -dʒ-éts-í !
kítsɔ́ dí-nàːwá.
Mpho(cl1) 1s/a.cl1-eat-appl.prf-fv Kitso(cl1) pl-beans(cl10)
‘Mpho ate Kiso’s beans.’ (alienable possession, applicative marking on the
verb necessary)

It appears that when such verb-marking alternations are present in a language, it

makes sense to group the entire construction under a single category — but that in turn
makes one revisit the requirement of a dedicated applicative marker on the verb as a
necessary condition on applicativization (see also the discussion in § 3.2).
So far, the examples I have discussed in this section showcased what applicatives
are supposed to do: add an applied argument. However, the connection with possessors
opens up another dimension of typological variation: a decrease in valency where the
argument that could appear as an applied object is coded as the possessor of an internal
argument. This anti-applicativization is found in Mayan languages where the benefi-
ciary, recipient, and sometimes locative and source are expressed as possessors.
Details vary, but in broad strokes, the generalization for several Mayan languages is
that prepositions introducing indirect arguments cannot combine with pronouns (nor
can they combine with some nominals with determiners), which suggests that they take
nominals of reduced structure (Pleshak 2022). As a consequence of the restriction against
preposition-pronoun combinations, there are two ways of introducing pronouns into
clausal structure: (i) as direct arguments (P-applicatives in Z and C’s terminology) or (ii)
as possessors, which means that they are treated as a subconstituent of another argu-
ment. The former strategy is found, for example, in Tsotsil, where first- or second-per-
 29 Understanding applicatives 1025

son pronouns in the indirect object position must be expressed as the primary object
(see Aissen 1983, 1987; Pye 2007; Montgomery-Anderson, this volume). For instance, in
(13), ‘for me’ cannot be expressed as a PP and instead becomes an applied argument in
the absolutive, such that the verb agrees with it (see 1sg.abs exponent before the root);
the base object ‘my children’ is not referenced on the verb at all.

(13) Tsotsil (after Aissen 1987: 181, 205)

L-i-s-k’el-be-ik j-ch’amaltak li Xune.
asp-1sg.abs-3sg.erg-watch-appl-pl 1sg.poss-children det Juan
‘Juan watched my children for me.’

Strategy (ii), which can be characterized as anti-applicativization, is illustrated by Q’an-

job’al where the first/second person pronouns cannot occur with prepositions intro-
ducing indirect objects or as complements of relational nouns. Consider the contrast in
the examples in (14). In (14a), the indirect object is a third-person referent, expressed
by male or female classifiers naq and ix respectively, and these classifiers can be com-
plements of a preposition; as (14b) shows, the first-person pronoun cannot occur in this
complement position and is instead expressed as the possessor of the noun ‘food’, hence
‘my food’ (14c).

(14) Q’anjob’al (Pedro Mateo Pedro, p.c.)

a. Hoq-Ø y-aq xim loy b’ay naq/ix.
pot-3sg.abs 3sg.erg-give clf food to clf[male]/clf[female]
‘S/he will give food to him/her.’
b. ✶Hoq-Ø y-aq xim loy b’ay ahim.
pot-3sg.abs 3sg.erg-give clf food to 1sg
(Intended: ‘S/he will give me food.’)
c. Hoq-Ø y-aq hin-lob’ej.
pot-3sg.abs 3sg.erg-give 1sg.poss-food
‘S/he will give me food.’ (Lit. ‘S/he will give my food.’)

I would like to underscore that anti-applicativization involves only a subset of argu-

ments (personal pronouns in case of Mayan languages), yet its presence is another
indication that possessors (or more broadly speaking, concernees) should be included
in considerations of applicatives. Another take-home message from the Mayan data is
that in considering the possible roles of applied arguments (comitatives, instruments,
sources, beneficiaries, possessors) we should also be cognizant of structural constraints
on direct and indirect complements occurring in a clause, in particular: can these
complements be expressed by pronouns, determiners, proper nouns, or just common
nouns.
1026 Maria Polinsky

5 Conclusions
This chapter builds on the overview of applicative constructions presented by Z and
C and proposes some refinements to their definition of the construction. Their oper-
ational definition includes three criterial properties: (i) applicativization alters the
status of non-agentive/non-subject arguments; (ii) the predicate in the applicative con-
struction has overt morphological marking that sets it apart from the base-construction
predicate; and (iii) the applied phrase is “a noun phrase in a role other than S or A, . . .
which refers to a participant that either requires a non-core coding different from its
coding in the AC or cannot be expressed at all in the BC”.
While generally in agreement with these criteria, I have proposed some adjust-
ments to these criterial properties. In particular, with respect to the non-A/S status of the
applied argument, I have suggested combining applicativization and causativization, the
two processes of nucleativization that Z and C treat as separate. While distinct applica-
tivization and causativization are found in a number of languages, quite a few unrelated
languages have generalized applicatives, ones that allow for causative-applicative syn-
cretism.
The typological options we can anticipate depend on the way lines are drawn
between two functions: the introduction of an object argument and the identification
of the semantic role associated with that argument. The former function is carried by
generalized applicatives (which can also be called unspecified nucleatives, following Z
and C’s terminology). The identification of a particular semantic role associated with
the applied object leads to the separation of applicativization and causativization.
Generalized and specialized applicatives can co-exist within a single language.
Thus, the morphology of applicativization can range from syncretism (generalized
applicatives) to what can be called extreme anti-syncretism whereby semantically dif-
ferent types of applicatives (benefactive, instrumental, locative, etc.) are marked dis-
tinctly. The question remains whether or not we should follow morphological marking,
in which case the syncretic and non-syncretic options are to be held separate? On this
approach, which can be thought of as “splitting”, the crucial assumption is that the
two functions, that of adding an argument and that of interpreting it, are inexorably
linked.
An alternative approach, which can be called “lumping”, relies on the assumption
that grammar allows us to add an argument; the interpretation of that argument can
vary widely, and we should not expect a one-to-one match between form and mean-
ing.18 This is the approach I have advocated in this chapter. On this approach, the start-
ing point is whether the operation in question adds an object to the syntactic structure.
If the answer to this question is affirmative, the next question is whether or not the

18 See also Arkadiev and Letuchiy (2021: 507 ff.) for this option and the discussion of its cross-linguistic
distribution.
 29 Understanding applicatives 1027

semantic role of the added object is (fully) specified. If yes, we get a distinction between
causatives and applicatives, with possible further distinctions within applicatives.
If the role of the applied object is not specified, then we can speak of a generalized
applicative. If no syntactic applied object is added, the construction is not applicative
whatsoever but may allow for what Z and C refer to as X-applicatives. The chart in
Table 3 illustrates the criteria for the lumping approach and the results this approach
predicts.

Table 3: Separating the functions of argument addition and semantic-role marking.

Object argument added No new argument added

Role of applied argument specified Applicative, causative X-applicatives, D-applicatives

Role of applied argument not specified Generalized applicative

The recognition of the causative-applicative syncretism raises at least two questions

that I would like to flag here. The first has to do with agentivity of the causee role. Z
and C’s rationale for separating causativization from applicativization may stem from
the need to maintain the non-agentive interpretation of the applied object; causee on
the other hand is agentive, interpreted as S or A. At the same time, many discussions of
causative constructions suggest that the causee role is inherently less agentive than that
of a prototypical agent/causer; after all, the event in which causee appears as S or A is
instigated by another participant, and causee may share properties with a comitative
or patient. Considering causatives in the context of applicatives may allow us to better
understand the properties of causees.
The second question has to do with the predictability of a particular interpretation
when a generalized applicative is used. In some languages, the choice of the causative
vs. applicative interpretation is lexically determined, being associated with a particu-
lar verb or subclass of verbs. Cataloguing lexical restrictions is an important empiri-
cal task. In some other languages, one and the same verb can receive the causative or
applicative interpretation, and factors that determine the choice of a particular inter-
pretation are yet to be understood.
Turning to the overt morphological marking of applicatives, we observe that the
applicative exponent on predicates may not always show up even within a single
language; it can be present only on some predicates but not on others. Furthermore,
we find that it is sometimes difficult to separate verbal exponents with a non-applica-
tive function from those that are specifically intended to mark applicatives. Finally,
giving equal weight to property (ii) on par with the other properties runs the risk of
missing some data from morphologically impoverished languages. Thus, the ques-
tion of the overt marking that signals applicativazation on predicates may need more
scrutiny.
1028 Maria Polinsky

With respect to the status of the applied argument in the base construction —crite-
rial property (iii) mentioned above—, my proposal is to include constructions with an
external object possessor (object possessor raising constructions) in the range of applic-
atives. Such constructions often have the same verb marking as applicative construc-
tions proper. Including external possessors among applied objects also captures the
observation that the referents of the original (base) and applied object are understood
to be in a semantic relationship that resembles possession or inclusion.

Abbreviations
abs absolutive
aor aorist
appl applicative
caus causative
cl noun class
clf classifier
compl completive
dat dative
dec declarative
det determiner
erg ergative
fv final vowel
incompl incompletive
ind indicative
ins instrument(al)
nom nominative
nsbj non-subject
obj object
pfv perfective
pl plural
poss possessive
pot potential
prv preverb
pst past
res resultative
sbj subject
sg singular
val valency morpheme
vers versionizer
vis visible / speaker’s area
 29 Understanding applicatives 1029

References
Aissen, Judith. 1983. Indirect object advancement in Tzotzil. In David Perlmutter (ed.), Studies in Relational
Grammar 1, 272–302. Chicago: University of Chicago Press.
Aissen, Judith. 1987. Tzotzil clause structure. Dordrecht: Reidel.
Arka, I Wayan, Mary Dalrymple, Meladel Mistica, Suriel Mofu, Avery Andrews & Jane Simpson. 2009. A
linguistic and computational morphosyntactic analysis for the applicative -i in Indonesian. In Miriam
Butt & Tracy H. King (eds.), Proceedings of the LFG09 Conference, 85–105. Stanford: CSLI Publications.
Arkadiev, Peter & Alexander Letuchiy. 2021. Indirect antipassive in Circassian. In Katarzyna Janic & Alena
Witzlack-Makarevich (eds.), Antipassive. Typology, diachrony, and related constructions, 483–514.
Amsterdam: John Benjamins.
Austin, Peter. 2005. Causatives and applicatives in Australian Aboriginal languages. In Kazuto Matsumura &
Tooru Hayasi (eds.), The dative and related phenomena, 165– 225. Tokyo: Hituzi Syobo.
Baker, Mark C. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago
Press.
Baker, Mark C. & Ruth Kramer. 2014. Rethinking Amharic prepositions as case markers inserted at PF. Lingua
145. 141–172.
Ball, Douglas. 2008. Clause structure and argument realization in Tongan. Stanford: Stanford University
dissertation.
Bierge, Stephanie Ramos. 2017. Clause types and transitivity in Wixárika (Huichol): A Uto-Aztecan language.
Boulder: University of Colorado dissertation.
Bisang, Walter. 2006. From meaning to syntax: Semantic roles and beyond. In Ina Bornkessel, Michael
Schlesewsky, Bernard Comrie & Angela D. Friederici (eds.), Semantic role universals and argument
linking, 237–262. Berlin: Mouton de Gruyter.
Boeder, Winfried. 1969. Über die Versionen des georgischen Verbs. Folia Linguistica 2. 82–152.
Buell, Leston. 2003. Merging arguments above the agent: The case of Zulu locative applicatives. WCCFL 22,
Proceedings, 109–122. Somerville, MA: Cascadilla Press.
Buell, Leston. 2004. Bantu locative applicatives, high and low. Paper presented at the LSA Annual Meeting,
Boston, January 9, 2004.
Cornilescu, Alexandra. 2020. Ditransitive constructions with differentially marked direct objects in
Romanian. In Anna Pineda & Jaume Mateu (eds.), Dative constructions in Romance and beyond, 117–142.
Berlin: Language Science Press.
Creissels, Denis. To appear. Transitivity, valency and voice.
Davis, Philip W. & Ross Saunders. 1997. A grammar of Bella Coola. (University of Montana Occasional Papers
in Linguistics 13). Missoula: University of Montana Press.
Deal, Amy Rose. 2017. External Possession and Possessor Raising. In Martin Everaert & Henk
C. van Riemsdijk (eds.), The Wiley Blackwell Companion to Syntax. 2nd edn. https://doi.
org/10.1002/9781118358733.wbsyncom047
Diaconescu, Constanţa Rodica & Rivero, María Luisa. 2007. An applicative analysis of double object
constructions in Romanian. Probus 19: 200–233.
Dikken, den, Marcel. 2023. High and low applicatives of unaccusatives: Dependent case and the phase.
Linguistic Inquiry 54 (3). 479–503. https://doi.org/10.1162/ling_a_00450
Dixon, R. M. W. 1988. The grammar of Boumaa Fijian. Chicago: University of Chicago Press.
Dixon, R. M. W. 2012. Basic Linguistic Theory, Vol. 3: Further grammatical topics. Oxford: Oxford University
Press.
Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3). 547–619.
Dunn, Michael J. 1999. A grammar of Chukchi. Canberra: Australian National University dissertation.
Folli, Raffaella & Heidi Harley. 2007. Causation, obligation, and argument structure: On the nature of little v.
Linguistic Inquiry 38. 197–238.
1030 Maria Polinsky

Gatchalian, Terrance. Submitted. Building Ktunaxa causatives.

Gerdts, Donna B. 2003. The morphosyntax of Halkomelem lexical suffixes. International Journal of American
Linguistics 69(4). 345–356.
Harley, Heidi. 2017. The “bundling” hypothesis and the disparate functions of little v. In Roberta D’Alessandro,
Irene Franco & Ángel J. Gallego (eds.), The verbal domain, 3–28. Oxford: Oxford University Press.
Hemmings, Charlotte. 2013. Causatives and applicatives: The case for polysemy in Javanese. SOAS Working
Papers for Linguistics 16. 167–194.
Himmelmann, Nikolaus P. 2005. The Austronesian languages of Asia and Madagascar: Typological
characteristics. In Alexander Adelaar & Nikolaus P. Himmelmann (eds.), The Austronesian languages of
Asia and Madagascar, 110–181. London: Routledge.
Ippolito, Michela. 2000. Remarks on the argument structure of Romance causatives. Ms., MIT, Cambridge,
Mass.
Jung, Hyun Kyoung. 2014. On the syntax of applicative and causative constructions. Tucson: University of
Arizona dissertation.
Kayne, Richard. 1975. French syntax. Cambridge, MA: MIT Press.
Kim, Kyumin. 2009. Applicatives in Korean causatives and passives. Toronto Working Papers in Linguistics.
Retrieved from https://twpl.library.utoronto.ca/index.php/twpl/article/view/656
Kittilä, Seppo. 2009. Causative morphemes as non-valency increasing devices. Folia Linguistica 43(1). 67–94.
Kittilä, Seppo. 2013. Causative morphemes as a de-transitivizing device: What do non-canonical instances
reveal about causation and causativization? Folia Linguistica 47. 113–138.
Klamer, Marian. 1998. Kambera. Berlin: Mouton de Gruyter.
Koptjevskaja-Tamm, Maria & Irina Muravyova. 1993. Alutor causatives, noun Incorporation, and the Mirror
Principle. In Bernard Comrie & Maria Polinsky (eds.), Causatives and transitivity, 287–314. Amsterdam:
John Benjamins
Kramer, Ruth & Mark C. Baker. 2013. Applicative verbs in Amharic. Paper presented at NACAL 41, February
13, 2013.
Landau, Idan. 1999. Possessor raising and the structure of VP. Lingua 107. 1–37.
Landau, Idan. 2007. The locative syntax of experiencers. Cambridge, MA: MIT Press.
Langacker, Ronald W. 1977. Studies in Uto-Aztecan grammar, Volume 1: An overview of Uto-Aztecan grammar.
Dallas: Summer Institute of Linguistics & University of Texas at Arlington.
Langacker, Ronald W. 1993. Reference point constructions. Cognitive Linguistics 4. 1–38.
Letuchiy, Alexander B. 2007. Reciprocals, reflexives, comitatives, and sociatives in Adyghe. In Vladimir P.
Nedjalkov (ed.), Reciprocal constructions, 773–811. Amsterdam: John Benjamins.
Lomashvili, Leila. 2011. Complex predicates: The syntax-morphology interface. Amsterdam: John Benjamins.
Marantz, Alec. 1993. Implications of asymmetries in double object constructions. In Sam Mchombo (ed.),
Theoretical aspects of Bantu grammar I, 113–151. Stanford: CSLI Publications.
Massam, Diane. 1998. Instrumental aki and the nature of Niuean transitivity. Oceanic Linguistics 37(1). 12–28.
Massam, Diane. 2009. The structure of unergatives. The Proceedings of AFLA 16. 125–135.
Massam, Diane. 2020. The lingering DP in Niuean. In Lauren Clemens & Diane Massam (eds.), Polynesian
syntax and its interfaces, 282–302. Oxford: Oxford University Press.
Munro, Pamela. 1984. The syntactic status of Object Possessor Raising in Western Muskogean. Proceedings
of the Tenth Annual Meeting of the Berkeley Linguistics Society, 634–649. Berkeley: Berkeley Linguistic
Society.
Nakamura, Masanori. 1997. Object extraction in Bantu applicatives: Some implications for Minimalism.
Linguistic Inquiry 28(2). 252–280.
Nash, Léa. 2020. Causees are not agents. In Elitzur A. Bar-Asher Siegal & Nora Boneh (eds.), Perspectives on
causation, 349–395. Dordrecht: Springer.
 29 Understanding applicatives 1031

Nash, Léa. 2022. The architecture of argument introducing categories in Georgian: The case of voice
marker a. Paper presented at the conference on Theoretical Linguistics and Languages of the
Caucasus (TLLC), Istanbul, June 18–19, 2022.
O’Herin, Brian. 2001 Abaza applicatives. Language 77(3). 477–493.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Pleshak, Polina. 2022. The status of Dative case in the Moksha case paradigm. Voprosy jazykoznanija 5.
108–130.
Pye, Clifton. 2007. The genetic matrix of Mayan applicative acquisition. Linguistics 45(3). 653–681.
Pylkkänen, Liina. 2000. What applicative heads apply to. University of Pennsylvania Working Papers in
Linguistics 7(1). Article 18.
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
Shibatani, Masayoshi & Prashant Pardeshi. 2002. The causative continuum. In Masayoshi Shibatani (ed.),
The grammar of causation and interpersonal manipulation, 85–126. Amsterdam: John Benjamins.
Talmy, Leonard. 1978. Figure and Ground in complex sentences. In Joseph Greenberg (ed.), Universals of
human language, Vol. 4: Syntax, 625–649. Stanford: Stanford University Press.
Talmy, Leonard. 2000. Toward a cognitive semantics: Concept structuring systems, Vol. 1. Cambridge, MA: MIT
Press.
Tuggy, David. 1983. Aztec causative/applicatives in Space Grammar. Working Papers of the Summer Institute
of Linguistics, University of North Dakota Session 27. Article 9. DOI: 10.31356/silwp.vol27.09
Van de Velde, Mark L. 2020. Concernee-Concern constructions: A comparative study of external possession
in the Bantu languages. Studies in Language 44(1). 70–94.
van der Wal, Jenneke. 2017. Flexibility in symmetry: An implicational relation in Bantu double object
constructions. In Michelle Sheehan & Laura R. Bailey (eds.), Order and structure in syntax II: Subjecthood
and argument structure, 115–152. Berlin: Language Science Press.
Wali, Kashi. 1981. Cause, causer and causee: A semantic perspective. Journal of Linguistics 17(2). 289–308.
Zúñiga, Fernando & Seppo Kittilä. 2019. Grammatical voice. Cambridge: Cambridge University Press.
George Moroz and Maria Polinsky
30 Applicatives cross-linguistically: Features
and distribution
Abstract: This chapter constitutes an update on the overview chapter on applicative
constructions that appeared in the World Atlas of Language Structures (Polinsky 2003,
2013). The update has been possible thanks to the novel empirical material represented
in the current handbook and to new descriptive and theoretical work on applicatives
that has appeared since the compilation of the original atlas. The chapter includes a
static map showing the distribution of main values in applicatives and a link to the cor-
responding dynamic map. The values reflected in the map include the main semantic
roles of the applied object (we distinguish between the beneficiary, as the most common
role, and everything else) and transitivity of the base (transitive, intransitive, or both).
The most common type of attested applicative constructions is the one where applica-
tives of different roles are formed from both verbal bases, transitive and intransitive.
We also show several values that are unattested. The chapter concludes with a brief
discussion of current approaches to applicative constructions.

1 Introduction
This chapter constitutes an update on the chapter originally submitted to the World
Atlas of Language Structures, WALS (Polinsky 2003, 2013), building both on the rich
material represented in this handbook and on new descriptive and theoretical work on
applicatives that has appeared since the compilation of the original atlas. In light of the
growing work on applicatives, some of the data points presented in the WALS chapter
could be revised, to reflect the improved and expanded empirical base of applicatives.
For instance, Mapudungun (Mapuche) was represented in WALS as lacking applicatives;
however, as Zúñiga (this volume) convincingly shows, it does have an applicative con-
struction. In general, the data presented in this chapter supersede the WALS maps.
On the research side, several overviews and analytical treatments appeared after the
publication of the WALS chapter, in particular, Georgala (2012), McGinnis (2008, 2017),
Peterson (2007, 2019), Pylkkänen (2008). Some of the current approaches to applicatives
are sketched in Section 4 of this chapter.

Acknowledgments: This work was funded in part by NSF grants BCS-1563129 and BCS-1941733 to the second
author. The first author has received funding from the Basic Research Program at the National Research
University Higher School of Economics. We thank Denis Creissels, Monica Macaulay, and Fernando Zúñiga
for helpful comments on this chapter. All errors are our responsibility.

https://doi.org/10.1515/9783110730951-030
1034 George Moroz and Maria Polinsky

Following the definition by Zúñiga and Creissels (this volume), applicativization

relates two constructions, the base construction (BC) and the applicative construction
(AC) in the following way: (i) the predicates in both constructions are built upon the
same root, but the one in the AC bears additional overt marking that distinguishes it
from the one in the BC; (ii) the participant encoded as S or A in the BC appears as S or A
in the AC; and (iii) the AC includes an applied phrase (AppP), which refers to a partici-
pant that either requires a non-core coding in the BC different from its coding in the AC
or cannot be expressed at all in the BC.
For example, in Tukang Besi (Austronesian; Sulawesi, Indonesia), the verb ‘fetch’
takes one theme object in the basic construction (as shown in [1a]), but with the applica-
tive marker it takes two objects, theme and beneficiary (as shown in [1b]).

(1) Tukang Besi (Donohue 1999: 256)

a. Basic construction, two-place predicate
no-ala te kau
3.realis-fetch the wood
‘She fetched the wood.’
b. Applicative construction, three-place predicate
no-ala-ako te ina-su te kau
3.realis-fetch-appl the mother-my the wood
‘She fetched the wood (as a favor) for my mother.’

The term applicative is used to denote either the applicative construction or the verb
in that construction. The term was first used by early Spanish missionary grammar-
ians in the description of Native American languages, in particular Uto-Aztecan; it
was later adopted by Bantuists and is now used for similar constructions all over the
world, although a popular perception of applicatives is primarily tied to these lan-
guage families.
It is customary to restrict the designation applicative to those cases where the
addition of an object is overtly marked on the predicate (see Zúñiga and Creissels,
this volume), which why English pairs such as she baked a cake vs. she baked Oscar a
cake typically do not count as basic-applicative alternation (but see Zúñiga, Arkadiev
and Hegedűs, this volume, and Polinsky, this volume, for further discussion). And
in fact, there are clear instances where the derived applied object is different in its
grammatical properties from the underived one, as in dative-object or double object
constructions.
In contrast to other valency-changing alternations, applicativization is typically
marked by an affix on the predicate, and does not involve reduplication, deletion of
verbal morphology, ablaut, or stress shifts (see Dixon 2012: 301). However, the stem alter-
nations observed in Otomanguean languages (Hernández-Green and López Nicolás, this
volume) may present a counterexample to this observation.
 30 Applicatives cross-linguistically: Features and distribution 1035

Difficulties in identifying applicatives on the basis of verbal marking may arise as

in a number of languages, some verbs can be overtly marked as applicative and others,
are not, thus creating what can be called “masked applicatives”, lexicalized applica-
tives, or deponents (see Zúñiga and Creissels, this volume). All told, we face the situation
with applicative which may be described as “now you see it, now you don’t”. Some chap-
ters in this volume address this issue by introducing the distinction between obligatory
and optional applicatives. In mapping out the applicatives, we have adopted a more
conservative approach, relying on reports that identify at least some verbal marking;
we have also chosen to exclude clearly lexicalized occurrences and some constructions
that are still under debate (for example, locative-object verbs or relational verb con-
structions of Algonquian — see Lockwood and Macaulay, this volume).
In terms of typical semantic roles associated with more specialized applicatives, the
most common role is that of beneficiary. This role of the applied argument is often close
to or get collapsed with goal, recipient, maleficiary (some researchers group all four
under the rubric of goal or recipient, consider Dixon 2012: Ch. 25). The other common
roles include instrument; comitative/sociative; location, broadly understood, and sub-
stitutive (a participant on whose behalf the action is performed). A number of terms
have been proposed for the latter, including “deputative beneficiary” (Van Valin and
LaPolla 1997) and “surrogate” (Zúñiga 2014).
Comitative/sociative applicatives formed from intransitives seem to be particularly
common in languages of Australia (see Austin, this volume). Comitatives and substitu-
tives are quite common in applicatives of intransitives, consider a comitative applica-
tive in Mapudungun and a substitutive/surrogate applicative in Kinyarwanda (3):

(2) Mapudungun (Zúñiga, this volume)

Amu-ye-fi-n ñi wenüy.
go-appl-3.obj-1sg.sbj.ind 1sg.psr friend
‘I went with my friend.’ (comitative)

(3) Kinyarwanda (Maria Polinsky, field notes)

Umugabo a-ra-geend-er-a umugóre.
man 3sg-pres-travel-appl-asp woman
‘The man is travelling instead/on behalf of the woman.’ (substitute)

The applicative derivation can also add causee and possessor as separate arguments,
but there is less consensus among researchers as to whether or not these arguments
should count as applied or not (see Polinsky, this volume, for more discussion).
Some languages have dedicated markers for different roles (consider the discus-
sion of “applicative arrays” in Dixon 2012: 312–315, and see Beck, this volume, on
1036 George Moroz and Maria Polinsky

Totonac, which has a rich array of dedicated applicative markers),1 whereas others
group many roles under the generalized applicative, which is a syncretic exponent
used to add an argument with a variety of interpretations. If different exponents are
used to index different semantic roles, we commonly find a contrast between benefi-
ciary (often with the associated meanings of goal and recipient), location, and instru-
ment/comitative.
Generalized applicative and dedicated applicative markers can co-occur in a single
language, as for example, in Ktunaxa (Gatchalian, submitted), or Uto-Aztecan languages
(Álvarez González and Estrada Fernández, this volume; Thornes, this volume). Varia-
tion is observed even in such “classic” applicative languages as Bantu languages; for
instance, Pacchiarotti (this volume) writes:

Languages where the applicative is obligatory on a root-by-root basis to introduce any given set
of semantic roles except Agent (and occasionally Instrument) have a very restricted set of prepo-
sitions or no prepositions at all, e.g., the Chaga E60 language group. . . . Languages with optional
applicative constructions usually have a fairly developed system of prepositions, but the applica-
tive might still be obligatory with certain verb roots to introduce certain semantic roles, e.g. Mon-
go-Nkundo C61.

Languages also vary depending on whether or not more than one applicative marker
can be used within a single verb. Applicative stacking is not uncommon (see § 4.3 below);
it is found in Bantu (Creissels, this volume; Pacchiarotti, this volume), Kartvelian (Tuite,
this volume),2 Northwest Caucasian (Arkadiev, Lander, and Bagirokova, this volume),
Mapudungun (this volume), Totonac (Beck, this volume), just to give a few examples.

2 Definition of map values

At the time the original WALS maps were created, the editors of the Atlas asked the authors
of individual chapters to survey a list of about two hundred languages, mapping the rele-
vant feature in that sample. As a result, the WALS applicative atlas divided languages into
those that have applicative constructions and those that do not. In this chapter, we adopt
a different approach, mapping only those languages which have been claimed to have
applicative constructions. Languages that are not on the map may therefore be excluded
for two different reasons: either they genuinely lack applicatives or no definitive data to
this point are available. We believe that this approach is more appropriate at the present
stage of our knowledge; as new languages with applicatives are found, they could be
added to the current map.

1 In mapping languages with multiple dedicated applicatives, we used the option “Benefactive and
other roles”. The semantics of individual roles indexed by applicative markers tends to be quite fluid, so
grouping them all together appears to be a descriptively safe option.
2 Tuite refers to applicative stacking as “double applicatives”.
 30 Applicatives cross-linguistically: Features and distribution 1037

Two main parameters in which applicatives vary cross-linguistically include the

transitivity of the base and the semantic role of the applied object. The map presented
in this chapter reflects the values of these parameters.
With respect to the transitivity of the base, the main distinctions are between
applicatives formed (i) from a transitive base only, (ii) from an intransitive base
only, and (iii) from both bases (these distinctions are shown by different shapes of the
symbols on the map). There are also constraints on the number of arguments of the
base verb; they are not shown in the map because they are harder to determine on the
basis of grammars.
With respect to the semantic roles of the applied object, the most common role of
the applied object is that of beneficiary (sometimes also associated with recipient/goal).
Accordingly, the map differentiates applicatives whose applied object (i) is limited to
beneficiary; (ii) corresponds to the beneficiary and some other roles; (iii) corresponds
to other roles to the exclusion of the beneficiary. These distinctions are shown by
the different colors of the symbols on the map.
Of nine logically possible values, six are actually attested. We summarize all the
options in Table 1 (we also include the unattested values), with each cell showing the
number of languages instantiating that particular option:

Table 1: Main values in the distribution of applicatives.

Verb base

Intransitive Transitive Both intransitive and transitive

Benefactive object only Unattested Attested (24) Attested (32)

Benefactive and other objects Unattested Attested (5) Attested (68)
Non-benefactive object only Attested (2) Unattested Attested (13)

With respect to the unattested values, it remains to be seen if they are not found for
principled reasons or for lack of data. We return to some of the reasons behind lack of
attestation in Section 4.2.
The static map below (Figure 1) presents all the attested values in their geographi-
cal distribution; the attested numbers shown in Table 1 are also included. The map was
created with R (R Core Team 2022) package lingtypology (Moroz 2017). The dynamic
version and raw data are available at: https://lingconlab.github.io/supplementary_
applicative_constructions/; the language data on the dynamic map also include their
genetic affiliations. While the static map has to remain as is, the dynamic map can be
updated with new or corrected data (directly on github https://github.com/LingConLab/
supplementary_applicative_constructions/issues/new or by emailing the authors of this
chapter).
1038 George Moroz and Maria Polinsky

Figure 1: The crosslinguistic distribution of applicative constructions.

3 Areal and genetic distribution of applicatives

Applicatives are common in three geographical areas: Africa (mostly in Bantu), the
western Pacific region (Austronesian), and North and Meso-America (Salish, Mayan,
Uto-Aztecan). The main generalization seems to be that applicatives are commonly
found in those languages that have little or no case-marking of noun phrases in a
clause and that have sufficiently rich verbal morphology to mark applicative for-
mation on the predicate. There may be a weak correlation between the use of the
applicatives and impoverished nominal marking; this would account for what can
be called typical applicative languages, such as Bantu, but it is still worth noting
that not all languages with applicatives lack nominal morphology — consider Kart-
velian, where rich verb morphology coexists with rich nominal marking (see Tuite,
this volume).

4 Linguistic puzzles related to applicatives

Applicative verbs and constructions have generated a significant amount of research
dealing with the morphosyntax of applicatives, transitivity, VP structure, distinct object
relations, argument structure, and mapping from argument to syntactic structure. For
an overview of theoretical issues, see Alsina (1996), Alsina and Mchombo (1993), Peter-
son (1999, 2007, 2019), and McGinnis (2008, 2017).
 30 Applicatives cross-linguistically: Features and distribution 1039

4.1 Transitivity of the base

The intransitive base of applicatives is less common than the transitive base. This is
quite clear from the map; few languages in the sample that form applicatives exclu-
sively from the intransitive base (see Table 1 and Figure 1). The overall tendency is that
if a language has applicatives formed from the intransitive base, it also has applicatives
formed from the transitive base.
The following explanation may be offered as to why the applicative is uncommon
with intransitives. Adding an object to an intransitive base amounts to creating a tran-
sitive verb. In order to transitivize an intransitive, languages typically use causativi-
zation, thus increasing the complexity of the event structure (V > CAUSE V); with the
applicative, the event structure is not modified, just another participant is added. Under
causativization, the argument added to the argument structure of the verb is agent;
under simple transitivization, the added argument is theme (or patient). Note that
applicative formation results in adding arguments other than agent and theme, and
may thus be constrained by the general hierarchy of semantic roles (see, among others,
Dowty 1991; Baker 1998; Primus 1999; Levin and Rappaport Hovav 2055):

(4) agent > theme (patient) > goal (recipient, beneficiary) > location > other roles

Although the constraint on applicative formation from intransitives seems not to be

absolute, a particular subset of intransitives, namely unaccusative predicates (those
whose subject originates as an object in the underlying structure), resist applicativi-
zation (see Baker 1998). However, even this generalization does not hold in some lan-
guages, for instance, in Halkomelem (Gerdts 1988 and this volume), Lai (Peterson 2007)
and Sesotho (Machobane 1989), which suggests that it is just a strong tendency.

4.2 Morphosyntax

In applicatives derived from the transitive base, the question arises which object, basic
or applied, has more object-like properties, which can point to its being structurally
superior. Three possibilities can be anticipated:

(5) a. the base object and the applied object have same/similar grammatical properties
b. the applied object has more object properties than the base object
c. the base object has more object properties than the applied object

In some languages both objects, basic and applied, are accessible for passivization and
relativization, can bind a reflexive, can trigger agreement on the verb, or can license
coreferential deletion across clauses. Such languages are called “symmetrical”, instan-
tiating scenario (5a). In other languages, only one object, either applied or basic, can
1040 George Moroz and Maria Polinsky

show the relevant grammatical behaviors. Such languages are called “asymmetrical”,
instantiating options (5b) and (5c).
Thanks to the influential work by Bresnan and co-authors (Bresnan and Mchombo
1987; Bresnan and Moshi 1990), the symmetrical-asymmetrical distinction has been
tested primarily in Bantu languages. A number of chapters in this handbook explore
it with respect to other languages, demonstrating a common tendency for object asym-
metry. Furthermore, more recent work on Bantu suggests that treating individual lan-
guages as symmetrical or asymmetrical leads to missed generalizations because sym-
metry and asymmetry may co-occur within the same language, with respect to different
verb classes or object types (van der Waal 2017).
The morphosyntax of applicatives and object asymmetries can be captured under
Baker’s (1988) approach to incorporation. According to Baker, applicative formation is
the result of the incorporation of a preposition into a verb — thus, what appears to be an
applicative affix starts out as an adposition introducing an oblique object. The oblique
becomes an object following adposition incorporation. This account fits well with the
observation that many (but not all) applicative markers originate as adpositions, and
may also explain the empirical generalization that applicatives are not marked via
reduplication, stress shift or stem reduction (but see § 1 for possible exceptions).
In exploring the syntax of applicatives, Pylkkänen (2008) proposes a dedicated func-
tional head, ‘Appl’, which can be merged in different places in the structure, namely, high
and low, hence the distinction between high and low applicatives. When the applicative
head is added to the structure low, inside the verb phrase, it serves a particular func-
tion of relating the two objects, the base object (patient) and the applied one; very often
this is the relationship which can be captured in terms of possession/inclusion, transfer,
or affectedness of the applied object. This explains the common association between
applicativization and transitive base, and the common albeit not obligatory benefactive
interpretation of the applied object (the subscripts on the object denotations in (6) indi-
cate the semantic roles):

(6) [VP V [ApplP APPLIED OBJECTBEN [Appl Appl [BASE OBJECTPATIENT/THEME]]]]

In high applicatives, the Appl functional head is added outside (above) the verb phrase,
and the applied object is causally connected to the entire eventuality denoted by that
verb phrase, not to the object inside that verb phrase. Accordingly, high applicatives
can occur with intransitive predicates which do not introduce an object or with transi-
tive verbs where no clear semantic relation between the two objects is presumed. The
structure is as follows (by putting the base object in parentheses we indicate that it is
not always present in this structure).

(7) [ApplP APPLIED OBJECT [Appl Appl [VP V (BASE OBJECT)]]]

 30 Applicatives cross-linguistically: Features and distribution 1041

The structure in (7) allows for a variety of semantic roles associated with the applied
object, as long as this object can be construed as associated with the entire event (as a
location, associate, etc.).
Pylkkänen’s analysis has been criticized on semantic grounds, primarily because
it does not establish a stronger connection between the event expressed by the low
applicative and the possession (Larson 2010). On the structural side, however, this
account has been influential and predictive with respect to argument properties of the
applied object. It is worth noting that Pylkkänen’s account predicts that low applicatives
would occur with transitive bases, and that benefactive would be particularly common
among such applicatives. This is confirmed by the attested cross-linguistic distribution
(see the map in Figure 1 and also Table 1).

4.3 Multiple applicatives: Applicative-marker stacking

The view of applicative formation as preposition incorporation has played an important

role in approaches to applicative morphology. The basic idea is that morphological der-
ivations must directly reflect syntactic derivations (and vice versa) (Baker 1985; Alsina
1999), and this isomorphism should be reflected in affix ordering (so-called Mirror Prin-
ciple). Crucial evidence for the Mirror Principle has been drawn from the interaction
between applicative morphology and morphology of other morphosyntactic operations
(passive, reciprocal) in Bantu.
Instances of multiple applicative marking and multiple applied objects are found in
Bantu languages, Tukang Besi (and possibly other Austronesian languages), and North-
west Caucasian languages (Arkadiev, Lander and Bagirokova, this volume; O’Herin
2001).3 In Cahuilla, applicative formation is achieved either via prefixation (for adding
a locative object) or via suffixation (for adding a beneficiary/recipient), which suggests
that the diachronic origin of these markers may be different. In those cases where iden-
tical applicative morphemes allow iteration, as in Bantu, it is unclear whether their
number is constrained by grammar or by processing limitations.

4.4 Semantics of the applied object

It is sometimes hard to tell from grammatical descriptions whether the beneficiary is dif-
ferentiated from the recipient/goal, or other common roles associated with the applied
object. It is also unclear how many languages merge the beneficiary and the malefi-

3 Koyraboro Senni is another language with multiple-applicative marking mentioned in Polinsky (2013),
whose data were based on Heath (1999: 168–169). However, a closer examination of this language sug-
gests that it may actually lack applicatives (Denis Creissels, p.c.), so we do not include it in the list pre-
sented here.
1042 George Moroz and Maria Polinsky

ciary (i.e., the adversely affected object), and whether their separation or their merger
is more typical. In addition, the applied object can be mapped onto a possessor, whose
place in the hierarchy of semantic roles is not quite clear (see Polinsky, this volume, for
more discussion of the possessor role in applied objects). The beneficiary, the goal, and
the possessor are typically animate participants, which may create an impression that
the applied object has to be an animate participant. Indeed, in some languages (e.g.,
Halkomelem) the referent of the applied object must be animate regardless of its seman-
tic role (Gerdts 1988, 1993, and this volume). In some languages, e.g. Kinyarwanda, the
linear order of the base and applied object varies depending on animacy (Kimenyi 1988;
Polinsky 1995); animacy plays a big role in the structure and interpretation of Algon-
quian applicatives (see Lockwood and Macaulay, this volume; Rhodes 2010). For a more
detailed discussion of semantic roles of the applied object, see Zúñiga and Creissels (this
volume) and Dixon (2012: Ch. 25).

Abbreviations
appl applicative
asp aspect
ind indicative
obj object
pres present
psr possessor
sbj subject
sg singular

References
Alsina, Alex. 1996. Passive types and the theory of object asymmetries. Natural Language and Linguistic
Theory 14. 673–723.
Alsina, Alex. 1999. Where’s the Mirror Principle? The Linguistic Review 16. 1–42.
Alsina, Alex & Sam Mchombo. 1993. Object asymmetries and the Chichewa applicative construction. In Sam
Mchombo (ed.), Theoretical aspects of Bantu grammar, 17–45. Stanford: CSLI Publications.
Baker, Mark C. 1985. The Mirror Principle and morphosyntactic explanation. Linguistic Inquiry 16. 373–416.
Baker, Mark C. 1988. Incorporation: A theory of grammatical function changing. Chicago: University of Chicago
Press.
Baker, Mark C. 1998. On the structural positions of themes and goals. In Johan Rooryck & Laurie Zaring
(eds.), Phrase structure and the lexicon, 7–34. Dordrecht: Kluwer.
Bresnan, Joan & Sam Mchombo. 1987. Topic, pronoun, and agreement in Chicheŵa. Language 63(4).
741–782.
Bresnan, Joan & Lioba Moshi. 1990. Object asymmetries in comparative Bantu syntax. Linguistic Inquiry 21.
147–185.
 30 Applicatives cross-linguistically: Features and distribution 1043

Dixon, R. M. W. 2012. Basic Linguistic Theory, Vol. 3: Further grammatical topics. Oxford: Oxford University
Press.
Donohue, Mark. 1999. A grammar of Tukang Besi. Berlin: Mouton de Gruyter.
Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67(3). 547–619.
Gatchalian, Terrance. Submitted. Building Ktunaxa causatives.
Georgala, Effi. 2012. Applicatives in their structural and thematic function: A minimalist account of
multitransitivity. Ithaca, NY: Cornell University dissertation.
Gerdts, Donna B. 1988. Object and absolutive in Halkomelem Salish. New York: Garland.
Gerdts, Donna B. 1993. Mapping Halkomelem grammatical relations. Linguistics 31. 591–621.
Heath, Jeffrey. 1999. Grammar of Koyraboro (Koroboro) Senni, the Songhay of Gao. Cologne: Rüdiger Köppe
Verlag.
Kimenyi, Alexandre. 1988. Passives in Kinyarwanda. In Masayoshi Shibatani (ed.), Passive and voice,
355–386. Amsterdam: John Benjamins.
Larson, Richard K. 2010. On Pylkkänen’s semantics for low applicatives. Linguistic Inquiry 41. 701–704.
Levin, Beth & Malka Rappaport Hovav. 2005. Argument realization. Cambridge: Cambridge University Press.
Machobane, Malillo. 1989. Some restrictions on the Sesotho transitivizing morphemes. Montreal: McGill
University dissertation.
McGinnis, Martha. 2008, Applicatives. Language and Linguistics Compass 2. 1225–1245.
McGinnis, Martha. 2017. Applicatives. In Martin Everaert & Henk C. van Riemsdijk (eds.), The Wiley Blackwell
Companion to Syntax, Second Edition https://doi.org/10.1002/9781118358733.wbsyncom124
Moroz, George. 2017. lingtypology: Easy mapping for Linguistic Typology. https://CRAN.R-project.org/
package=lingtypology
O’Herin, Brian. 2002. Abaza applicatives. Language 77(3). 477–493.
Peterson, David A. 1999. Applicative constructions. Berkeley: University of California dissertation.
Peterson, David A. 2007. Applicative constructions. Oxford: Oxford University Press.
Peterson, David A. 2019. Applicatives. Oxford Bibliographies Online. DOI: 10.1093/OBO/9780199772810-0227
Polinsky, Maria. 1995. Double objects in causatives. Studies in Language 19(1). 121–221.
Polinsky, Maria. 2003. Applicative constructions. In Martin Haspelmath, Matthew Dryer, David Gil & Bernard
Comrie (eds.), World Atlas of Language Structures, 442–445. Oxford: Oxford University Press.
Polinsky, Maria. 2013. Applicative constructions. In Matthew S. Dryer & Martin Haspelmath (eds.), The World
Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology.
Primus, Beatrice. 1999. Case and thematic roles. Berlin: Max Niemeyer.
Pylkkänen, Liina. 2008. Introducing arguments. Cambridge, MA: MIT Press.
R Core Team. 2022. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
Rhodes, Richard. 2010. Relative root complement: a unique grammatical relation in Algonquian syntax. In
Jan Wohlgemuth & Michael Cysouw (eds.), Rara & rarissima, 305–324. Berlin: De Gruyter Mouton.
van der Wal, Jenneke. 2017. Flexibility in symmetry: An implicational relation in Bantu double object
constructions. In Michelle Sheehan & Laura R. Bailey (eds.), Order and structure in syntax II: Subjecthood
and argument structure, 115–152. Berlin: Language Science Press.
Van Valin, Robert D. & Randy LaPolla. 1997. Syntax: Structure, meaning, and function. Cambridge: Cambridge
University Press.
Zúñiga, Fernando. 2014. Benefaction proper and surrogation. Studies in Language 38(3). 543–565.
Denis Creissels and Fernando Zúñiga
31 Applicative and related constructions:
Results and perspectives
What distinguishes this chapter from the other survey chapters (the introductory
chapter, Chapter 29 by Polinsky and Chapter 30 by Moroz & Polinsky) is that it summa-
rizes the important points of the case studies that constitute this book in the perspective
of the typology of applicative and related constructions. Its structure reflects that of the
questionnaire, and it is basically conceived as a guide for readers looking for precise
information on particular aspects of applicative constructions, with the help of which
they will be able to select the case studies particularly relevant for the questions in
which they are interested. However, we cannot possibly summarize here in an appro-
priate way all noteworthy features unearthed, systematized, or called into question by
the authors of the individual chapters. Instead, we concentrate on some of the most
interesting findings related to the non-canonical features of applicatives, as well as to
the lookalikes.

1 Morphology
As regards possible interactions between applicative marking and the structure of verb
inflection, the general rule is that applicative marking does not affect the inflectional
possibilities of verbs. However, Hernández-Green and López Nicolás mention that, in
Old Otomi, the goal applicative seems to have involved a reduced tam paradigm, and
an extreme case of reduction of the inflectional possibilities of applicativized verbs is
signaled by Van Gijn for Kakua. In this Amazonian language, verbs have a benefactive
imperative whose syntactic properties meet the definition of applicativization, but the
benefactive counterpart of the imperative has no equivalent of the other forms that
constitute the tam paradigm of Kakua verbs.
Two of the chapters that constitute this book analyze languages characterized by a
complex morphophonological interaction between applicative marking and verb inflec-
tion: Gerdts on Hul’q’umi’num’ and Jacques and Lahaussois on Kiranti languages. In
Kiranti languages, it is common that morphophonological processes result in neutrali-
zation of the distinction between applicative and non-applicative verb forms.

Acknowledgments: We are grateful to the series editors for their comments on a previous version of this
chapter.

https://doi.org/10.1515/9783110730951-031
1046 Denis Creissels and Fernando Zúñiga

Systems of applicative marking involving complex stem alternations are found in

Otomi (Hernández-Green and López Nicolás), in some Bantu languages (Pacchiaro-
tti), and in Western Nilotic languages (Payne).
Regarding morphological issues, readers are invited to bear in mind that the present
book is clearly biased, in the sense that non-affixal applicative markers are under-repre-
sented here. Only a few chapters address applicativization via verb compounding (Foley
for Papuan), verb compouding and auxiliaries (Jacques and Lahaussois for Kiranti),
applicativizing particles (Zúñiga, Arkadiev, and Hegedűs for Germanic, Baltic, Slavic,
and Hungarian), or applicativizing particles and auxiliaries (Vanhove for Cushitic).
Serial verb constructions functionally equivalent to applicative constructions are
mentioned by Foley for some Papuan languages, by Montgomery-Anderson for the
Mayan language Tojolabal, and by Pacchiarotti for Grassfields Bantu languages. Pos-
sible vestiges of an applicative periphrasis are also signaled by Álvarez González and
Estrada Fernández in the Uto-Aztecan language Yaqui.
Among the cases of non-affixal applicative marking discussed in this book, the
grammaticalization of a verb ‘say’ in periphrastic applicatives, discussed by Vanhove
for Cushitic languages and also noted in Amharic by Amberber, is worthy of mention,
since this use of ‘say’ is uncommon (if attested at all) outside of the Ethiopian area.
We address the cross-linguistically common fact that applicative markers may
encode other types of morphologically-oriented valency alternations in Section 5 below.

2 Syntax
2.1 Constraints on the transitivity properties of the base
construction
The languages whose applicative constructions are investigated in this book illustrate
all possible configurations as regards the conditioning of applicativization by the tran-
sitivity properties of the BC: some applicative markers only operate on intransitive BCs,
some others only on transitive BCs, and still others operate indiscriminately on intran-
sitive and transitive BCs. For example, Hul’q’umi’num’ has two applicative markers
attaching exclusively to intransitive verbs and two other attaching exclusively to transi-
tive verbs (Gerdts). By contrast, the transitivity properties of the BC play no role in the
conditioning of applicativization in the Bantu language Tswana (Creissels).

2.2 Optional vs. obligatory applicatives

Optional applicatives are generally acknowledged to be widespread, and we do not

dispute that; in some language groups they are either the only or the predominant type,
 31 Applicative and related constructions: Results and perspectives 1047

as mentioned by Austin for Australian and by McDonnell and Truong for Western
Indonesian.
Several chapters show, however, that obligatory applicatives are by no means mar-
ginal from a cross-linguistic perspective. Pacchiarotti and Creissels confirm this for
Bantu, Voisin and Creissels for Atlantic languages, Payne for Nilotic, Foley for Papuan,
Gerdts for Salishan, Thornes and Álvarez González and Estrada Fernández for Uto-
Aztecan, Hernández-Green and López Nicolás for Otomi and Zapotec, Van Gijn for
Tukanoan languages and Zúñiga for Mapudungun, as well as Montgomery-Anderson—
albeit with some reserves due to the scarcity of data in several cases—for Mayan.
The data analyzed in some chapters show that applicatives apparently analyzable
as optional are in fact obligatory applicatives. For example, the mechanism found in
Tswana (Creissels) according to which some motion verbs assign the role of Source of
motion to locative expressions in their base form, and the role of Destination in their
applicative form, could be analyzed superficially as optional applicative marking in the
presence of a locative expression. However, since the verbs in question in their base
form simply cannot combine with a locative expression expressing the role of Desti-
nation, and cannot combine with a locative expression expressing the role of Source
in their applicative form, the only possible analysis is that this is a case of obligatory
applicativization of the redirecting type.
Several other chapters discuss situations in which applicative constructions carry
semantic nuances that may make it difficult to classify them as optional or obligatory,
since it is not always obvious whether the nuances in question should be analyzed as
involving a difference in the semantic roles expressed by the BC and the AC. However, at
least in some cases, a thorough analysis of the semantic roles assigned by the base form
and the applicative form of a verb leads to the conclusion that applicatives that at first
sight look like optional applicatives are in fact obligatory applicatives.
A consequence of this situation for a general typology of valency operations is that it
may make sense to treat optional applicatives and obligatory applicatives as two distinct
types of valency operations, but the still widespread view according to which optional
applicatives are the canonical variety of a type of valency operation that includes oblig-
atory applicatives as a non-canonical variety must definitely be abandoned, since the
cross-linguistic distribution of these two varieties of applicatives provides no justifi-
cation for regarding optional applicatives as canonical, and obligatory applicatives as
non-canonical, rather than the other way round.
Tswana (Creissels) illustrates the extreme case of a language having only obliga-
tory applicatives, i.e. a language in which all applicative constructions are straightfor-
wardly conditioned by the impossibility of expressing the semantic role expressed by
the applied phrase in an alternative construction with the same verb in its non-applica-
tive form. An even more extreme case is that of the languages whose situation can be
characterized in terms of across-the-board applicativization (see § 2.3 below).
In the languages analyzed in this book, it is not uncommon that the same applica-
tive marker appears in contructions that differ in their status as obligatory or optional
1048 Denis Creissels and Fernando Zúñiga

applicatives. For example, in Jóola Fóoñi (Voisin and Creissels), applicative construc-
tions involving the same applicative marker -úm are obligatory if the applied phrase
expresses the roles of Path (perlative) or Means (mediative), but optional with applied
phrases expressing the roles of Instrument or Cause. Similar facts are mentioned by
Payne in Nilotic languages, and by Foley in the Papuan language Yimas, where the ACs
involving the applicative marker taŋ- are obligatory if the applied phrase expresses the
role of Beneficiary, but optional if the applied phrase expresses the role of Companion.
The data analyzed in this book also show that applicative constructions may have
an ambiguous status with respect to the distinction between obligatory and optional
applicatives, in the sense that, for example, an applicative construction that is optional in
pragmatically neutral, affirmative, clauses may become obligatory in some other types
of clauses.
Such a situation is described by Hernández-Green and López Nicolás in Northern
Zapotec, where ACs that are optional in pragmatically neuter clauses are obligatory
in most adjunct extraction constructions (interrogation, relativization, focalization) in
which the phrase optionally coded as an applied phrase in the corresponding pragmat-
ically neutral clause moves to a preverbal position.
Similarly, Payne signals that, in the Western Nilotic language Shilluk, the instru-
mental applicative is required if a non-core argument is focused in preverbal position.
In Wolof (Voisin and Creissels), the applicative construction in which the applied
phrase expresses the role of Companion (comitative applicative) differs from the other
applicative constructions found in the language in that it is impossible in pragmatically
neuter clauses, but obligatory if the participant fulfilling the role of Companion is focal-
ized, questioned, or relativized.

2.3 Across-the-board applicativization

In languages that have obligatory applicatives, the participants that cannot feature
as core syntactic terms in clauses projected by non-applicative verb forms divide into
those that can be coded as obliques without any special verbal marking and those that
can only be coded as applied phrases in an applicative construction. For example, in
Tswana (Creissels), Beneficiaries are coded as applied objects, whereas Instruments
are coded as prepositional phrases whose presence has no effect on verb morphology.
The question that arises is whether there are languages making a particularly sys-
tematic use of the obligatory applicativization strategy, in which the coding of non-core
participants as obliques able to combine with the base form of the verb would be mar-
ginal, or even completely inexistent. Interestingly, this book includes a language (Toba/
Qom) providing a perfect illustration of this possibility, and another (Upper Necaxa
Totonac), whose situation is less extreme, but nevertheless quite close to across-the-
board applicativization.
 31 Applicative and related constructions: Results and perspectives 1049

As described in detail by Censabella, Toba/Qom has a wide array of semantically

specific verbal affixes occurring in obligatory applicative constructions, and applicativi-
zation is the only available strategy to encode semantic roles other than those licensed
as subjects or objects of non-applicative verb forms.
Contrary to Toba/Qom, Upper Necaxa Totonac (Beck) does not have locative applic-
atives and expresses static location by means of a locative proclitic, but the other seman-
tic types of participants that cannot be coded as subjects or objects of non-applicative
verb forms can only be introduced via applicativization
Among the other languages surveyed in this book, the Papuan language Barupu
shares with Toba/Qom the total lack of NP flagging by means of either case affixes or
adpositions, and makes extensive use of applicativization (Foley). However, in Barupu,
the serialization strategy is also productive, and the distinction between serialization
and applicativization is not clear-cut, since some of the applicative markers have an
obvious verbal origin and still “very much look like verbs due to their inflections”.

2.4 Maintenance vs. demotion of base Ps in P-applicative

constructions (transitivizing vs. redirecting applicatives)

Another syntactic issue of interest is the variation with respect to the status of the
non-S/A arguments in the applicative construction. The earlier literature has already
noted the existence of different kinds of P-applicatives in that some are transitivizing
(or “valency-increasing”) while others are redirecting (sometimes termed “remapping”,
“redirective”, or “valency-neutral”); while the former are simply promotional, the latter
necessarily combine the introduction of an applied P with the demotion of any base P
(see Zúñiga and Creissels, this volume). Both possibilities are well attested in this book.

2.4.1 P-applicatives from transitives in languages that have double-P constructions

In the languages in which double-P constructions are attested with underived verbs,
the general rule is that P-applicatives are transitivizing, and it may even happen that
applicatives from ditransitives are constructions of a type not found with underived
verbs, with three terms coded like monotransitive Ps, as reported by Álvarez González
and Estrada Fernández for the Uto-Aztecan languages from Northwestern Mexico, by
Payne for the Nilotic language Maasai and by Creissels for Tswana.1

1 What is at issue in this and the following sections is in fact not a relation to (mono)transitive typology,
but to ditransitive typology, namely dependency of applicative constructions on the indirective vs. neu-
tral vs. secundative alignment, a topic addressed by Malchukov (2017).
1050 Denis Creissels and Fernando Zúñiga

In the languages in which base Ps can be maintained in P role in P-applicative con-

structions, it must be taken into consideration that double-P constructions are rarely
if ever perfectly symmetrical. Several chapters dealing with languages in which base
Ps are maintained in P role in applicative constructions discuss the extent to which, in
applicatives from transitives, base Ps and applied Ps manifest the properties typical for
monotransitive Ps (such as accessibility to the role of A/S in passive constructions); see,
among others, Álvarez González and Estrada Fernández, Van Gijn, and Creissels.
The situation is not uniform across the languages for which the question of
asymmetries in double-P constructions resulting from applicativization is discussed.
However, situations in which base Ps are treated as less-privileged Ps in the applicative
construction seem to be particularly frequent, the applied P taking the role of syntac-
tically privileged P (see in particular the discussion in Thornes’s chapter on Northern
Uto-Aztecan languages). This is probably related to the general preference for animate
applied Ps, since, more generally, animacy seems to play a crucial role in the distinction
between syntactically privileged and less-privileged Ps in the languages that have dou-
ble-P constructions. See, among others, Woolford’s (1993) observation that, in asymmet-
ric double-P constructions of ‘give’ verbs, the syntactically privileged P in passivization
is always the Recipient, and Malchukov, Haspelmath, and Comrie’s (2010) observation
that ‘give’ verbs “normally have an animate R and inanimate T”.
Among the languages in which base Ps in applicatives from transitives are main-
tained in P role, the possibility of triple-P constructions resulting from applicativization
of double-P BCs is mentioned among others for the Uto-Aztecan languages from North-
western Mexico (Álvarez González and Estrada Fernández), for Tswana (Creissels)
and other Bantu languages (Pacchiarotti), and for Nilotic languages (Payne). In Upper
Necaxa Totonac, the stacking of applicative and causative markers may yield construc-
tions with up to five phrases coded like monotransitive Ps (Beck).

2.4.2 P-applicatives from transitives in languages that do not have double-P

constructions stricto sensu but have a grammatical relation “secondary
object”

Algonquian languages (Lockwood and Macaulay) have a class of bivalent verbs with
which the non-subject argument fulfills a syntactic role O2 distinct from both the syntac-
tic role O1 that characterizes the non-subject argument of typical transitive verbs and
from oblique roles. Interestingly, in the coding frame of trivalent verbs such as ‘give’,
the Recipient is coded as O1, and the Theme as O2, and in P-applicatives from transi-
tives, the O1 role is taken over by the applied phrase, whereas the base O1 is demoted to
O2, which constitutes a particular type of redirecting applicative.
A situation similar to that of Algonquian languages is described by Zúñiga for
Mapudungun.
 31 Applicative and related constructions: Results and perspectives 1051

2.4.3 P-applicatives from transitives in languages that have neither double-P

constructions nor a grammatical relation “secondary object”

In the languages which have neither double-P construction stricto sensu (i.e., construc-
tions in which two terms show the coding properties that characterize monotransitive
Ps) nor a grammatical relation “secondary object”, it can be expected that P-applicatives
from transitives are exclusively of the redirecting type with demotion of the P term of
the BC to oblique, and this is confirmed by several of the chapters that constitute this
book; see, among others, Mithun on Inuit-Yupik-Unangan languages and Gerdts on
Hul’q’umi’num’.
A remarkable type of redirecting P-applicative is mentioned by Austin in the Aus-
tralian language Yidiny and in some dialects of Dyirbal, where transitive verbs must be
overtly intransitivized via antipassivization before being applicativized.

2.4.4 P-Applicatives in languages that do not have oblique NPs and do not have
double-P constructions or a grammatical relation “secondary object” either

Toba/Qom (Censabella) illustrates the case of languages that don’t have oblique NPs at
all (see § 3.3), and in which double-P constructions are extremely marginal, since apart
from ‘give’, no Toba/Qom verb (either derived or underived) can be found in a double-P
construction. Contrary to Algonquian languages or Mapudungun (see § 3.4.3). Toba/Qom
doesn’t have a grammatical relation “secondary object” either. Consequently, in applic-
atives from transitives, the participant coded as the P of the BC cannot be expressed
at all in the AC, and can only be expressed alongside with the applied P by means of a
clause chain in which the verb is repeated in its base form an in its applicative form. For
example, the allative applicative form of ‘throw’ cannot combine directly with a noun
phrase referring to the thing being thrown, and ‘A throws B to C’ is expressed literally
as A throws B throws.appl C.

2.5 D-applicatives

Several chapters of this book identify a number of D-applicatives (i.e., applicative con-
structions in which the applied phrase is a dative/indirect object), either prominent
ones (like those discussed by Arkadiev, Lander, and Bagirokova for Northwest Cauca-
sian and Tuite for Kartvelian) or more sporadic ones (like those mentioned by Austin
for Australian, Foley for Papuan, Vanhove for Cushitic, and Zúñiga, Arkadiev, and
Hegedűs for European languages).
The data analyzed by Van Gijn also suggest that the notion of D-applicative might
be relevant for Hup (Nadahup) and for Tukanoan languages, since in these languages,
1052 Denis Creissels and Fernando Zúñiga

“applied objects are generally not subject to differential case marking, unlike direct
objects (but like indirect objects)”.

2.6 X-applicatives

The instances of X-applicatives (i.e., applicative constructions in which the applied phrase
is an oblique) identified in this book confirm the interest of this notion in a general
typology of voice, for example the X-applicatives found in Algonquian (Lockwood and
Macaulay), Totonac (Beck), Bantu (Pacchiarotti and Creissels), Nilotic (Payne) and
Atlantic (Voisin and Creissels). Such constructions are also spotted in Cushitic languages
(Vanhove), Papuan languages (Foley) and Australian languages (Austin).
In Algonquian, the markers occurring in X-applicative constructions and those
involved in P-applicative constructions are distinct and occupy a distinct slot in the
structure of verb forms: the former (traditionally known as “relative roots”) are pre-
fixed to verb stems, whereas the latter are suffixed.
A different situation is found for example in Bantu and Atlantic, where the same
markers can be found in P-applicative and X-applicative constructions, depending on
the lexical meaning of verb and the semantic role of the applied phrase.
Interestingly, Creissels shows that the X-applicatives of Tswana are involved in a
mechanism showing some similarity with redirecting P-applicatives. This mechanism
concerns motion verbs whose non-applicative form assigns the role of Source of motion
to locatives. As already mentioned above, the applicative form of the same motion
verbs assigns the role of Destination to locatives, and at the same time cannot combine
with a locative expressing the Source, so that expressing the Source and Destination of
the same motion event requires a clause chain in which the same verb shows up succes-
sively in its non-applicative and applicative forms.

2.7 Multiple applicatives

The data analyzed in this book show that there is cross-linguistic variation in the pos-
sibility of stacking applicative markers in the same verb form, each of them licensing a
corresponding applied phrase. Multiples applicatives are found among others in Tswana
(Creissels), where for example ‘write’ in ‘write a letter’ can take two occurrences of the
applicative marker -ɛl, one licensing an applied object in Recipient role, the other licens-
ing an applied object in Beneficiary role.
However, on the whole, the prevailing tendency is that multiple applicatives are
either impossible or at least dispreferred. A possible explanation is that constructions
with several non-essential participants encoded as unflagged NPs whose semantic role
must be retrieved from verbal marking may be more difficult to process than sequences
of adpositional phrases (or case-marked NPs) in which a marker adjacent to each
 31 Applicative and related constructions: Results and perspectives 1053

nominal term (or the form of the nominal term itself) provides indications about the
semantic role of nominal expressions.
Interestingly, the languages among those represented in this volume whose situ-
ation can be characterized in terms of across-the-board applicativization (Toba/Qom
and Upper Necaxa Totonac) behave quite differently in this respect. Toba/Qom has a
strict ban on multiple applicatives (Censabella), whereas Upper Necaxa Totonac has no
restriction on the stacking of applicatives (Beck).

2.8 Combinability of applicative markers with markers of other

valency operations

As regards the combinability of applicative markers with markers of other valency

operations within the same verb forms, the default situation is the absence of arbitrary
ban on semantically plausible combinations, apart from the avoidance of accumula-
tions of voice markers that could make the construction difficult to process by speakers
or hearers. Cross-linguistically uncommon combinations of applicative markers and
markers of other valency operations are even mentioned in some chapters.
This is in particular the case in Upper Necaxa Totonac (Beck), where the combina-
tion of reciprocalization and comitative applicativization has the effect of transitivizing
reciprocal constructions, converting for example the intransitive construction A and B
love each other into a transitive construction that can be glossed as A is.in.mutual.love.
with B.
The combination of applicativization and causativization in Northwestern Cauca-
sian languages is another example. Cross-linguistically, applicativization of causative
constructions is common with the function of introducing a Beneficiary, but an addi-
tional function of the applicativization of causative constructions is found in North-
western Caucasian languages, where Causees can only be coded as the applied phrase
in a D-applicative construction corresponding to a causative BC (Arkadiev, Lander, and
Bagirokova).
However, some chapters mention the existence of arbitrary limitations on the com-
bination of applicative markers with markers of other valency operations.
The possibility of combining applicativization and antipassivization is illustrated,
among others, by Classical Nahuatl in the introductory chapter. By contrast, Gerdts
notes that, in Hul’q’umi’num’, applicative verbs do not form antipassives.
Several of the chapters that constitute this book explicitly mention the possibil-
ity of combining applicativization with reflexivization and/or reciprocalization, but
Hernández-Green and López Nicolás indicate that Northern Zapotec ACs can combine
with the causative construction but not with the reciprocal-reflexive construction. In
Hu’lq’umi’num’, transitivizing applicatives lend themselves to reflexivization and recip-
rocalization, whereas redirecting applicatives can be reciprocalized but do not lend
themselves to reflexivization (Gerdts).
1054 Denis Creissels and Fernando Zúñiga

In the Kiranti language Yakkha, applicative verbs can undergo reciprocal deriva-
tion, but are incompatible with reflexive derivation (Jacques and Lahaussois). Interest-
ingly, in another Kiranti language (Khaling), derived verbs expressing reflexivization
of an applicative construction seem at first sight to consist only of a root followed by
the reflexive suffix, but a closer look at their morphological particularities leads to the
conclusion that, historically, an applicative marker was present.
In this connection, it is also interesting to note that the combination of benefactive
applicativization and reflexivization, in addition to its compositional meaning of auto-
benefaction, may develop non-compositional meanings, as observed by Pacchiarotti
for Bantu languages.
Three of the chapters that constitute this book provide data about applicativization
in languages with symmetrical voice systems: Zúñiga on Mapudungun, McDonnell
and Truong on languages of Western Indonesia, and Musgrave, Arka, and Rajeg on
Standard Indonesian.

3 Semantics
3.1 Specialized applicatives vs. catch-all applicatives

Regarding semantic issues, and rather unsurprisingly, the languages covered here
confirm the findings of previous studies with respect to the existence of specialized
applicatives vis-à-vis broad markers/constructions. Both are widely attested across the
languages analyzed in this book, as well as situations involving moderately polysemous
applicative markers.

3.2 The semantic roles expressed as applied phrases

3.2.1 Semantic roles commonly expressed as applied phrases

The languages covered here also confirm the findings of previous studies with respect
to the cross-linguistic recurrence of applied phrases expressing the semantic roles of
Beneficiary/Maleficiary, Instrument, Companion (alias Concomitant), and the semantic
roles relate to space (Location, Source, Path, and Goal).
Applied phrases expressing the semantic roles of Cause and Stimulus are also rela-
tively common among the languages analyzed in this book.
Benefactive applicatives are particularly common, but not universal. Jóola Fóoñi
(Voisin and Creissels) has applicative constructions for Instruments and other seman-
tic roles, but not for Beneficiaries. Interestingly, in Jóola Fóoñi, Beneficiaries are not
encoded as adpositional phrases either, but as objects whose coding properties are
 31 Applicative and related constructions: Results and perspectives 1055

identical to those of the P argument of typical transitive verbs, resulting in ambiguity

between, for example, ‘I’ll do the washing for you’ and ‘I’ll wash you’. A similar situation
is mentioned by Foley for the Papuan language Coastal Marind.
Some of the Australian languages dealt with in Austin’s chapter are also counterex-
amples to the generalization according to which, if a language has applicative construc-
tions at all, Beneficiary is always among the possible roles for applied phrases.
In this respect, the European languages analyzed by Zúñiga, Arkadiev, and Hegedűs
are particularly exotic, since they virtually lack three of the four cross-linguistically
common functions of applicatives: benefactive, comitative and instrumental, and at the
same time attest quite a few uncommon functions (see § 4.2.2).
As a rule, in the languages surveyed in this book, benefactive applicatives lend
themselves to a malefactive interpretation, depending on the lexical semantics of the
verb. However, benefactive applicatives that cannot have a malefactive interpretation
are signaled in Tukanoan languages (Van Gijn), Conversely, the Arawakan language
Yukuna has an applicative that can only be interpreted as malefactive or “relinquitive”
(‘leaving behind’) (Van Gijn), and the Australian language Murrinhpatha has an applica-
tive whose only possible interpretations are malefactive and ablative (Austin).

3.2.2 Semantic roles less commonly expressed as applied phrases

Applied phrases expressing some remarkable, albeit possibly uncommon, semantic

roles are mentioned in several chapters.
The Kawapanan language Shiwilu has a relinquitive applicative (‘leaving behind’)
and an applicative expressing ‘in the vicinity of someone’ (Van Gijn). As mentioned
above, the relinquitive meaning is also expressed via applicativization in the Arawakan
language Yukuna, connected to malefactive.
In the Papuan language Barupu, the applicative marker licensing applied phrases
referring to negatively affected participants (malefactive) can also license applied
phrases referring to participants excluded from action (‘without’).
Algonquian languages have X-applicatives expressing ‘in the appearance of an X’
(Lockwood and Macaulay).
Algonquian languages also have X-applicatives with time adjuncts as applied
phrases (Lockwood and Macaulay).
The Inuit language Kalaallisut has a similative applicative (Mithun).
ACs with the applied phrase expressing the role of Object of Empathy or Object
of Emotion are mentioned in the Uto-Aztecan language Yaqui (Álvarez González and
Estrada Fernández), in the Papuan language Yimas (Foley), and in some Australian
languages (Austin).
The Bantu language Tswana has an obligatory applicative construction in which
the applied phrase is an adverb that can be glossed as ‘for a long time’, ‘for ever’, ‘for
real’, ‘irrevocably’ (Creissels).
1056 Denis Creissels and Fernando Zúñiga

The role of Containing Instrument (as in ‘grind in a mortar’ or ‘drink from a cup’)
is relevant for the description of the ACs of Upper Necaxa Totonac (Beck) and Tswana
(Creissels). Upper Necaxa Totonac has a dedicated applicative marker for Containing
Instruments, distinct from that used for ordinary Instruments. Tswana codes Contain-
ing Instruments via applicativization, whereas ordinary Instruments are coded as prep-
ositional phrases, without any verbal marking.
Toba/Qom has four distinct locative applicatives and five distinct directional applic-
atives, some of them expressing very specific nuances rarely expressed by dedicated
markers cross-linguistically (Censabella). A similar situation is found in the Papuan
language Barupu (Foley).
With motion verbs, German and other European languages have ACs with applied
phrases expressing the space or distance covered by motion (Zúñiga, Arkadiev, and
Hegedűs).
Applied phrases expressing the role of Viewpoint Holder (alias Judicans) are sig-
naled in some Northwestern Caucasian and European ACs (Arkadiev, Lander, and
Bagirokova; Zúñiga, Arkadiev, and Hegedűs).
Applied phrases expressing the role of Causee are found in the causative construc-
tions of Northwestern Caucasian languages (Arkadiev, Lander, and Bagirokova), in an
analytical causative construction of Otomi (Hernández-Green and López Nicolás), and
in the causative-autobenefactive construction of the Atlantic language Laalaa (Voisin
and Creissels).
English out-applicatives express the role of exceeded threshold / surpassed compet-
itor (Zúñiga, Arkadiev, and Hegedűs).
Baltic and Slavic languages have ACs with applied phrases expressing the roles
of Created Object, Eliminated Object or Exhausted Object (Zúñiga, Arkadiev, and
Hegedűs).
With bodily function verbs, the languages of Western Indonesia have ACs in which
the applied phrase specifies the thing or substance expelled (as in ‘urinate blood’).
ACs with applied phrases expressing the role of Speech Topic are signaled in the
Uto-Aztecan language Yaqui (Álvarez González and Estrada Fernández), in the Mayan
language Kaqchikel (Montgomery-Anderson), in Yup’ik (Mithun), and in the languages
of Western Indonesia (McDonnell and Truong).
The Papuan language Yimas has an AC whose meaning can be glossed as ‘chasing
someone into a place’ (Foley).
In Circassian, applicativization of a ‘be’ verb yields a predicative possession con-
struction with the Possessor encoded as an applied indirect object (Arkadiev, Lander,
and Bagirokova).
The expression of Manner may also involve verbal markers otherwise found in
bona fide applicative constructions, but in general, as discussed by Voisin and Creissels
for Atlantic languages, the use of the markers in question in relationship with phrases
expressing Manner seems to be an instance of oblique registration (see § 6.1.3) rather
than applicativization proper.
 31 Applicative and related constructions: Results and perspectives 1057

3.2.3 Possible correlations between the semantic roles expressed by applied

phrases and other properties of applicative constructions

The possibility of correlations between the semantic roles expressed by applied phrases
and other properties of applicative markers (syntactic properties of the applicative con-
struction, polysemy patterns involving other types of valency operations, possibility of
uses not related to valency operations) is systematically explored by Voisin and Creis-
sels for Atlantic languages. They show that, in this language family, the markers found
in benefactive applicative constructions and those found in instrumental applicative
constructions not only are consistently distinct, but also show recurrent contrasts in
several respects.
Payne also organizes her discussion of Nilotic applicatives according to a similar
distinction between the roles expressed by applied phrases.
However, the languages of Western Indonesia analyzed by McDonnell and Truong
do not confirm the tendencies observed in Atlantic and Nilotic, since those that have
two distinct applicative markers use the same marker for benefactive and instrumental
applicative constructions, whereas they consistently have two distinct markers for the
applicative constructions in which the applied phrase expresses the roles of Beneficiary
and Goal, respectively.
In this connection, it is also interesting to observe that a marker involved in ben-
efactive applicatives but not in instrumental applicatives in some Mayan languages is
found in instrumental applicatives but not in benefactive applicatives in some others
(Montgomery-Anderson).
Consequently, a larger-scale investigation of the regularities suggested by the chap-
ters on Atlantic and Nilotic would be necessary to test their cross-linguistic validity.

3.3 Essential participants coded as applied phrases in applicative

constructions

Interestingly, several of the languages analyzed in this book attest the possibility that
applicativization is involved in the coding of semantic roles characterizing essential
participants, usually considered less canonical for applied phrases.
Applied phrases expressing the roles of Patient and Theme are not uncommon with
Eskimo-Aleutian applicatives (Mithun).
The role of Recipient, which also occurs time and again with applicatives (for example
in the Nilotic language Ateso and in Northwestern Caucasian languages), can hardly be
considered peripheral. In the Mayan language Chontal, the Recipient of the verb ‘give’ can
only be expressed as an applied phrase (Montgomery-Anderson), and the non-applicative
form of ‘give’ is interpreted as ‘produce’ (as in English cows give milk). In Toba/Qom, the
Recipient of ‘give’ can only be expressed as the applied P of an applicativized verb whose
underived form can be glossed as ‘give away’ (Censabella).
1058 Denis Creissels and Fernando Zúñiga

In Tswana (Creissels), the expression of the Recipient does not require applicative
marking with verbs whose inherent argument structure includes a Recipient, but the
expression of the Recipient of ‘write a letter’(whose presence in argument structure
is due to the lexical meaning of the noun in P role) necessitates applicative marking
on ‘write’. In Hul’q’umi’num’ (Gerdts), ‘give’, ‘show’ and ‘tell’ are frozen applicativized
verbs whose roots are not attested as monotransitive verb stems and only exist is com-
bination with a suffix still acting as an applicative suffix in sem̓ət ‘sell it’ > sam̓əst ‘sell
him/her it’.
Applied phrases expressing various semantic types of essential participants are also
indicated in the Uto-Aztecan languages from Nothwestern Mexico (Álvarez González
and Estrada Fernández) and in Creissels’ chapter on Tswana.
However, none of the languages or language groups analyzed in this book have
across-the-board transitivization as attested in some Oceanic languages where all tran-
sitive verbs are formed by means of a transitivizing suffix acting as a causative operator
with unaccusative intransitives, and as an applicative marker with unergative intran-
sitives. This suggests that systems of this type, although well-attested among Oceanic
languages, are rare (if attested at all) in the remainder of the world.

3.4 Applicativization and external possession

Several of the languages analyzed in this book confirm the possible involvement of
applicative marking in so-called external possession constructions. For example, in
Tswana (Creissels) applicative marking is not found in external possession construc-
tions referring to a whole-part relationship (such as ‘he broke my leg’, expressed liter-
ally as he broke me the leg), but is obligatory in external possession constructions refer-
ring to other semantic types of relationships between the “external possessor” and the
“possessee” (such as ‘he ate my food’, expressed literaly as “he ate.appl me the food”).
In fact, as developed by Creissels (forthcoming: Chapters 2, 13, and 14), a simple way
of dealing with the relationship between external possession and applicativization is to
follow Van de Velde’s (2020) proposal to replace the notion of external possession con-
struction (which misleadingly suggests syntactic derivation from some kind of “under-
lying structure” involving adnominal possession) by the notion of Concernee-Concern
construction, whose roots can in fact be found in the old Indo-Europeanist notion of
“dativus sympatheticus” (see among others Behaghel 1923: 633–638, and more recently
the “sympatheticus” role as defined by Lehmann 2006). The crux of this alternative
approach to external possession is that so-called external possessors are characterized
in terms of a participant role labeled “Concernee” that can be viewed as a subtype of the
semantic role of Beneficiary conceived as a macro-role. What distinguishes Concernees
from other subtypes of Beneficiaries is that their possible advantage or disadvantage
in the event denoted by the verb does not follow from the will of the Agent and/or
the nature of the particular event referred to, but from some relationship they have
 31 Applicative and related constructions: Results and perspectives 1059

inherently with another participant (the Concern), regardless of the particular events
in which they may be involved.
Since Beneficiary is unquestionably the semantic role whose coding most com-
monly involves applicativization, in this alternative approach to external possession,
the frequent involvement of applicative marking in so-called external possession con-
structions does not necessitate any particular explanation.
Moreover, the notion of Concernee explains very simply why in some languages
(e.g., Tswana), Concernee-Concern constructions without verbal marking are exclu-
sively found with reference to whole-part relationships. Crucially, in contrast to pos-
session (which admittedly conflates three semantic prototypes: whole-part, kinship
and exclusive use of an object), whole-part relationships seem to constitute the
semantic core of Concernee-Concern constructions. This explains in particular why
the notion of inalienable possession one might be tempted to invoke is clearly not
relevant in the analysis of Concernee-Concern constructions. The point is that the
notion of inalienable possession encompasses whole-part relationships and kin rela-
tionships, which cross-linguistically tend to behave differently in Concernee-Concern
constructions.

3.5 Applicativization and animacy

Several chapters in this book mention a tendency for applied phrases to refer to animate
entities, even if they express semantic roles that do not imply animacy.
For example, Álvarez González and Estrada Fernández observe that, in the
Uto-Aztecan languages they analyze, applied phrases usually refer to animate partici-
pants. It is particularly suggestive that Yaqui has applicative marking in the equivalent
of ‘I spread mud on your face’, but not in ‘Goyo spread butter on the bread’.
Beck mentions a contrast in Totonac between animate Goals, which require the
applicative form of verbs of motion, and inanimate Goals, which do not require verbal
marking.
Foley describes an applicative construction of Yimas in which “the actor performs
an action while carefully visually monitoring the applied, necessarily animate, partic-
ipant”.
Austin mentions an applicative marker of Murrinhpatha that licenses applied
phrases referring specifically to animate Sources.
Gerdts also notes in Hul’q’umi’num’ “a strong tendency for noun phrases high
on the person/animacy hierarchy to occur as applied objects rather than as obliques
[whereas] noun phrases low on the person/animacy hierarchy dis-prefer applicative
constructions”, as illustrated by the pair of examples ‘the child was frightened of the
car’ (where ‘car’ is encoded as a prepositional oblique) vs. ‘the child was frightened of
the dog’ (where ‘dog’ is encoded as an applied object). However, after a closer look at
textual data, she concludes that “it is not the person or animacy of the noun phrase that
1060 Denis Creissels and Fernando Zúñiga

determines whether it appears as an applied object or an oblique”, and that “the person/
animacy effects are a by-product of the salience of the noun phrases to the discourse”.
Interestingly, Payne describes the opposite situation with the role of Concomitant in
Datooga, where inanimate concomitants may be expressed as applied objects, whereas
animate concomitants must be expressed as prepositional obliques.

3.6 The holism effect

A particularly significant finding concerns the so-called “holism effect”, namely the
fact that a particular verbal marking often encodes some sort of heightened semantic
transitivity (whether correlated with valency increase or not), typically related to plu-
ractionality and/or increased or “extended” affectedness of the Patient. This effect is
known from studies on German be-applicatives and comparatively recent studies on
Bantu lookalike constructions, but several chapters of this book mention them appear-
ing in languages related neither areally nor genealogically to either Germanic or Bantu.
Further comparative research on this topic is likely to yield very interesting results.

3.7 Applicativization and topicality

With optional applicatives, the question arises of a possible discursive conditioning of

the use of the applicative construction.
Mithun argues that, in Eskimo-Aleut languages, the referents of applied phrases
are in general topical within the discourse, and other authors in this book suggest a
similar relationship between applicativization and topicality.
Thornes explores the hypothesis that, in Northern Uto-Aztecan languages, “Transi-
tivization is a side-effect of the basic function of ACs, namely, to assign discourse prom-
inence to otherwise peripheral (or, at least, affected) arguments”.
Álvarez González and Estrada Fernández state that, in the Uto-Aztecan languages
they analyze in their chapter, optional applicativization “can be used as a topicalization
device for referential discourse continuity”.
Topicality of the referent encoded as the applied phrase is also analyzed as a crucial
factor for the use of optional applicative constructions by Gerdts.
Pacchiarotti also mentions information structure as relevant for the use of
optional ACs in Bantu languages.
Vanhove mentions four Cushitic languages for which “the comments of the authors
show that there is a pragmatic difference between the applied phrase and its couterpart
in the base construction, but the exact link with focus or topic is often unclear”.
 31 Applicative and related constructions: Results and perspectives 1061

4 Lookalikes
4.1 Syntactic lookalikes

4.1.1 Uncoded valency alternations functionally comparable to applicativization

Two phenomena familiar from the descriptive and theoretical literature make an appear-
ance in this book (see. among others. Payne for Nilotic, Amberber for Amharic, and Zúñiga,
Arkadiev, and Hegedűs for European languages), namely (i) the uncoded dative alterna-
tion and (ii) external possession constructions without verbal marking (or uncoded con-
cernativization, if one accepts the alternative approach to external possession construc-
tions noted in § 4.4 above).
It is also worth mentioning here the treatment of Beneficiaries in Joola languages
(Voisin and Creissels) and in the Bantu language Eton (Pacchiarotti). In Joola languages
and in Eton, NPs representing Beneficiaries, in spite of their semantic status of adjuncts,
have exactly the same coding properties as NPs representing the Patient of typical tran-
sitive verbs. Syntactically, this mechanism evokes typical benefactive applicative con-
structions; however, in Joola languages and in Eton, as in English I build him a house, no
particular verbal marking is required in constructions including a benefactive object.
Interestingly, Joola languages and Eton belong to families in which it is common that
Beneficiaries can only be expressed via applicativization. This suggests that, historically,
in Joola languages and in Eton, the loss of applicative marking in the presence of a ben-
efactive object is probably responsible for this situation.
A similar mechanism is evoked by Foley for some Papuan languages under the
name of “promiscuous promotion to core from oblique without applicative marking”.

4.1.2 Equipollent marking of valency alternations functionally comparable

to applicativization

Apart from systems of inflectional voices of the type attested in Ancient Greek and Latin,
not very common cross-linguistically, equipollent verbal marking of valency alternations
has been mainly signaled in the literature and investigated with respect to the noncaus-
al-causal alternation.
“Noncausal-causal alternation” is the term that came into common use in the recent
literature for verb pairs in which one of the two verbs (the causal member of the pair)
projects transitive clauses whose P term corresponds semantically to the A or S term in
the construction of the other verb (the noncausal verb), whereas A in the construction of
the causal verb represents the instigator of the event described by the noncausal verb,
as in Northern Akhvakh istaka biq’ʷari ‘the glass broke’ (non-causal) vs. mik’ide istaka
biq’ʷāri ‘the child broke the glass’ (causal). From the relatively numerous studies that
have been devoted to this topic (see in particular Haspelmath 2016), it can be concluded
1062 Denis Creissels and Fernando Zúñiga

that, among the possible semantic subtypes of noncausal-causal verb pairs, unaccusa-
tive-transitive pairs are special in that, cross-linguistically, they show maximum varia-
tion among the five possible strategies (suppletivism, ambitransitivity, causativization,
decausativization and equipollence). Several languages with a relatively strong prefer-
ence for the equipollence strategy in unaccusative-transitive pairs have been signalled
in the literature (cf. Creissels, forthcoming: Chapter 16), although they are less common
than languages with a strong preference for causativization, decausativization or ambi-
transitivity.
Things are different for the functional type of valency alternation for which Creis-
sels (forthcoming) proposes the term “undirected-directed alternation”, i.e. verb pairs
in which one of the two verbs (viz. the directed member of the pair) projects transitive
clauses whose A term corresponds semantically to the A or S term in the construction of
the other verb (viz. the undirected verb), whereas P in the construction of the directed
verb represents an additional participant towards which the activity of the referent of
A is directed.
As far as we are aware, no large-scale typological study of the undirected-directed
alternation has been published so far. A priori, symmetrically with the noncausal-causal
alternation, five strategies can be expected to be available for the undirected-directed
alternation, at least in the particular case of unergative-transitive pairs:
– applicativization: the undirected verb is morphologically less complex than its
directed counterpart, as in Boumaa Fijian -la’o ‘go’ / -la’o-va ‘go to get (something)’;
– antipassivization: the undirected verb is morphologically more complex than its
directed counterpart, as in Mandinka dómó-rì ‘eat (intr)’ / dómò ‘eat (tr)’;
– suppletivism: the undirected verb and its directed counterpart are completely dif-
ferent, or differ in such a way that their formal relationship cannot be analyzed as
a particular instance of some more or less regular pattern, as in Akhvakh ũk- ‘eat
(intr)’ / q̄’am- ‘eat (tr)’;
– equipollence: the two members of the undirected-directed pair are formally related,
but the relationship is not morphologically oriented from undirected to directed or
from directed to undirected;
– flexivalency (including ambitransitivity): the undirected verb stem and its directed
counterpart are identical, as in Mandinka jélè ‘laugh (intr)’ / ‘make fun of (someone)
(tr)’.

An obvious asymmetry between the noncausal-causal alternation and the undirect-

ed-directed alternation is that the equipollence strategy, relatively common in the coding
of the noncausal-causal alternation, is cross-linguistically rare for the undirected-directed
alternation. However, the data examined in this book include languages (Tarahumara
and Guarijío, two of the Uto-Aztecan languages from Northwestern Mexico analyzed in
Álvarez González and Estrada Fernández’s chapter) in which equipollent marking is
common for pairs of verbs involved in a valency alternation functionally comparable
to optional P-applicativization, as for example in Guarijío naósa- ‘talk’ / naóse- ‘talk to’.
 31 Applicative and related constructions: Results and perspectives 1063

What is particularly interesting in the situation described by Álvarez González

and Estrada Fernández is the fact that, in Tarahumara and Guarijío, the same alterna-
tion a/i ~ e or a ~ e encodes the undirected-directed alternation if the member of the pair
ending with a(/i) is a transitive verb or an unergative intransitive, and the non-causal
alternation if the member of the pair ending with a(/i) is an unaccusative intransitive.
In other words, this instance of equipollent marking of valency alternations follows
the same distributional pattern as that observed in many languages in which the same
morphologically oriented derivation expresses applicativization with some verbs and
causativization with others.
Other instances of equipollent marking of alternations functionally comparable to
applicativization are also signaled in the following languages:
– In German (Zúñiga, Arkadiev, and Hegedűs), the alternation XACC von YDAT weg-zie-
hen ~ YDAT XACC ent-ziehen ‘withdraw X from Y’ illustrates this possibility.
– In the Algonquian language Blackfoot (Lockwood and Macaulay), a former applica-
tive marker has been reanalyzed as a ta final (i.e., as an obligatory component of
the verb forms that include it), so that the presence vs. absence of a benefactive O1
is not conditioned by the addition/deletion of morphological material, but merely
by the alternation between two possible finals, either ta ~ ti or ta ~ ai.2
– In Circassian languages (Arkadiev, Lander, and Bagirokova), a vowel alternation
designated as “transitivizing ablaut” distinguishes the intransitive use of motion
verbs like ‘go’ from a transitive use of the same verbs with an object denoting the
path or distance covered by motion.
– Jacques and Lahaussois mention the Kiranti language Yakkha as having a handful
of verbs with equipollent marking of an alternation functionally similar to applica-
tivization (e.g., between ‘be forgetful’ and ‘forget someone’).

One may also wonder whether the complex system of stem alternations that char-
acterizes the formation of applicative verb stems in Otomi and constitutes a reflex of
a valency-increasing suffix ✶-H (Hernández-Green and López Nicolás) should not be
analyzed synchronically as an instance of equipollent marking of an alternation func-
tionally similar to applicativization, rather than applicativization proper. The same
question arises for the Western Nilotic languages in which the distinction between
applicative verbs and their non-applicative counterpart relies entirely on stem alter-
nations (Payne).

2 In Algonquian languages, verbs are morphologically marked as transitive with animate object (ta),
transitive with inanimate object (ti), intransitive with animate subject (ai), or intransitive with inani-
mate subject (ii).
1064 Denis Creissels and Fernando Zúñiga

4.2 Morphological lookalikes

4.2.1 Lexicalized applicatives

First, familiar morphological lookalikes like the lexicalized applicatives of Germanic are
also found in languages and groups as disparate as Bantu (Pacchiarotti and Creissels),
Atlantic (Voisin and Creissels), Kartvelian (Tuite), Western Indonesian (McDonnell
and Truong), Uto-Aztecan (Álvarez González and Estrada Fernández), Algonquian
(Lockwood and Macaulay), Hul’q’umi’num’ (Gerdts), Mapudungun (Zúñiga), etc.
An interesting case of lexicalized applicative is the conventionalization of the
applicative form of ‘sit’ (literally ‘sit with’) as a transitive verb of possession in the Aus-
tralian language Diyari (Austin).
Lexicalized applicatives do not appear to exist in Totonac, however (Beck).

4.2.2 Applicativization marking and conjugation class marking

Foley discusses evidence that, in some groups of Papuan languages, applicative markers
tend to lose their applicative function and to transgrammaticalize into markers whose
sole function is to distinguish conjugation classes of verbs.

4.2.3 Applicative markers also used to express syntax-neutral intensification

Possibly as something related to the holism effect mentioned above, morphological

lookalikes that do not add an applied phrase but express some kind of heightened
semantics (“syntax-neutral intensification”) are clearly not an exclusive hallmark of
Bantu; they are also found in Western Austronesian (McDonnell and Truong), in the
Nilotic language Maa (Payne), in Papuan languages (Foley), in Standard Indonesian
(Musgrave, Arka, and Rajeg), Toba/Qom (Censabella), Mapudungun (Zúñiga), etc.

4.2.4 Applicative markers and valency-neutral V>V derivation

The syntax-neutral intensifying use of verbal formatives that also have the ability to act
as applicative markers mentioned in § 6.1.3 can be viewed as a particular case of what
Foley describes for Papuan languages as the “adverbial” function of verbal formatives
also used as applicative markers. In their “adverbial” function, the verbal formatives
in question express meanings similar to those they express as applicative markers, but
do not add arguments.
The possibility of being used to express V > V derivations that imply no modifica-
tion of the syntactic properties of verbs is not uncommon for the European preverbs
 31 Applicative and related constructions: Results and perspectives 1065

that have applicative marking as one of their possible functions (Zúñiga, Arkadiev, and
Hegedűs).
Similarly, Upper Necaxa Totonac has an applicative marker licensing applied objects
expressing the role of Containing Instrument which is homonymous (and cognate) with
a “limitative” suffix that does not modify the valency properties of verbs but implies
that the action affects specifically a subpart of one of the participants (Beck).
One may also mention here a phenomenon observed in the Algonquian Relational
Construction analyzed by Lockwood and Macaulay, where a verbal suffix registers the
presence of an additional “ghost participant” concerned by the event, without, however,
licensing a phrase representing the additional participant, which distinguishes this con-
struction from bona fide applicative constructions.

4.2.5 Applicative marking and oblique registration

Morphological lookalikes that mark an oblique term as pragmatically salient without

modifying its coding characteristics (“oblique registration”) are common in Meso-Amer-
ica (see Montgomery-Anderson for Mayan, Hernández-Green and López Nicolás for
Otomanguean), but are also found in Bantu (Pacchiarotti and Creissels), Atlantic
(Voisin and Creissels), Amharic (Amberber), in Papuan languages (Foley), and in Euro-
pean languages (Zúñiga, Arkadiev, and Hegedűs).
Moreover, several chapters mention constructions that cannot be unambiguously
analyzed as oblique registration constructions or applicative constructions, since the
oblique whose pragmatic saliency is marked by the addition of a verbal formative can
optionally maintain its oblique coding or acquire P-like coding.
According to Hernández-Green and López Nicolás, both in Zapotec and Otomi,
non-applicative constructions marked with applicative morphology can be arranged
along a promotion-registration continuum, based on how many core morphosyntactic
properties (if any) non-core arguments acquire in such constructions. The Mayan lan-
guage Ixil illustrates a construction in which an instrumental phrase whose focalization
is marked by a verbal suffix acting as a bona fide applicative marker in other Mayan
languages loses the prepositional marking it has in the BC but does not acquire the
possibility of being indexed in the verb form that characterizes core arguments in Ixil
(Montgomery-Anderson).
There is cross-linguistic variation in the semantic roles lending themselves to oblique
registration constructions or to constructions halfway between bona fide applicative and
oblique registration constructions. Oblique registration targets instrumental adjuncts in
Mayan languages, but locative adjuncts in Bantu and in the European languages sur-
veyed by Zúñiga, Arkadiev, and Hegedűs, and the only example of oblique registration
mentioned in the Uto-Aztecan languages analyzed by Álvarez González and Estrada
Fernández concerns a comitative adjunct.
1066 Denis Creissels and Fernando Zúñiga

A similar use of applicative morphology is mentioned by Thornes in Northern

Paiute, with the difference that the highlighted participant is not an adjunct, but a
Recipient.

4.2.6 Applicative marking and the scope of locative phrases

Some Bantu (Pacchiarotti) and Atlantic (Voisin and Creissels) data suggest that markers
otherwise involved in applicative marking may have a syntax-neutral use with the func-
tion of widening or shifting the scope of locative phrases, as for example marking the
distinction between two possible readings of the man heard the snake in the bush (‘the
snake is in the bush, the man could be or not in the bush’ vs. ‘the man is in the bush, the
snake could be or not in the bush’).

4.2.7 Applicative markers also used to derive joint activity verbs

The derivation of joint activity verbs (‘do something together’) implies no valency oper-
ation, but commonly involves the same markers as reciprocal derivation. In the Inuit
language Kalaallisut, joint activity verbs can be derived by means of a suffix also acting
as an applicative marker (in particular in comitative applicative constructions) and as
a reciprocal marker (Mithun).
The cross-linguistic variation in the Algonquian comitative constructions also pro-
vides interesting data about the possible relationships between comitative applicativi-
zation and the derivation of joint action verbs.

4.2.8 Syntax-neutral use of applicative markers related to definiteness

McDonnell and Truong mention that, in the languages of Western Indonesia, a verbal
suffix otherwise used as an applicative marker may have with some verbs a syntax-neu-
tral use in which its presence highlights that the P argument is definite, or that the
clause refers to a specific event.

4.2.9 Directionals, applicative marking, and aspect

Applicative markers also found in constructions in which they have a valency-neu-

tral use as aspectual markers are indicated in Bantu (Pacchiarotti and Creissels), in
Atlantic (Voisin and Creissels), in the languages of Western Indonesia (McDonnell
and Truong), in Standard Indonesian (Musgrave, Arka, and Rajeg), and in European
 31 Applicative and related constructions: Results and perspectives 1067

languages (Zúñiga, Arkadiev, and Hegedűs). This possibility is also evoked in Payne’s
chapter on Nilotic and in Zúñiga’s chapter on Mapudungun.
Since the (trans)grammaticalization of directionals into aspectual markers is
cross-linguistically common, it is tempting to speculate that this type of polysemy char-
acterizes applicative markers resulting from the (trans)grammaticalization of direc-
tionals.
In this respect, as discussed by Payne, Nilotic languages show particularly interest-
ing data about the possible extension of directionals into the domains of aspect, applica-
tive marking and person indexation.
Data on the possibility that directionals develop an applicative function can also be
found in Vanhove’s chapter on Cushitic.

4.2.10 Applicative markers also used to express modal notions

An applicative marker also appearing in a modal construction that does not meet the
definition of an applicative construction is mentioned by Arkadiev, Lander, and Bagi-
rokova. The Circassian languages have a potential construction involving a verbal affix
identical to the benefactive applicative marker. However, the potential construction
does not qualify as an applicative construction, since it is an intransitive construction
whose subject corresponds to the object of the base construction, whereas the transi-
tive subject of the base construction corresponds to a dative oblique in the potential
construction.
In languages having both a middle voice lending itself to a facilitative use and a
benefactive applicative, a compositional expression of potentiality is possible by taking
a facilitative middle as the input for benefactive applicativization (something like liter-
ally ‘the letter is easy to write for me’ > ‘I can write the letter’). However, such an anal-
ysis cannot be considered for the Circassian potential construction, directly derived
from the transitive base construction via the addition of a marker otherwise used as an
applicative marker.

4.2.11 Applicative markers also used for N>V or V>N derivation

In Standard Indonesian (Musgrave, Arka, and Rajeg), the suffixes acting as applicative
and causative markers are also used as derivational suffixes whose addition to nominal
or adjectival stems yields transitive verbs. A verbalizing use of applicative morphology
is also mentioned by Álvarez González and Estrada Fernández for some Uto-Aztecan
languages, by Zúñiga, Arkadiev, and Hegedűs for Germanic languages, and by Zúñiga
for Mapudungun.
1068 Denis Creissels and Fernando Zúñiga

In some Atlantic languages, the applicative constructions with the applied phrase
in the role of Instrument involve markers also used for instrument nominalization
(Voisin and Creissels).

5 Applicativization and other morphologically-

oriented valency operations
It has already been observed in the literature that the use of the same verbal marking
to encode applicativization and other types of morphologically-oriented valency alter-
nations is cross-linguistically common.

5.1 The applicative-causative polysemy

Unsurprisingly, the data analyzed in this book confirm that the applicative-causative
polysemy is particularly common cross-linguistically.
As rightly observed by Mithun and Jacques and Lahaussois, some situations are
inherently ambiguous between the conceptualization underlying an applicative formu-
lation and a causative formulation. For example, in the kind of situation described in
English as A went with B or A took B along, there are equally good reasons to encode B
as the applied phrase in an applicative construction or as the causee in a causative con-
struction. Similar examples are also discussed by Creissels for Tswana and by Van Gijn
for the languages of northwestern Amazonia. Hernández-Green and López Nicolás
quote a Zapotec example of a comitative applicative lending itself to a causative-like
interpretation (A went to work with B > A took B to work).
That said, true instances of applicative-causative polysemy (i.e., the use of identical
markers in applicative and causative constructions referring to situations that do not
show the conceptual ambiguity mentioned above) are mentioned in the chapters by
Thornes, Álvarez González and Estrada Fernández, Hernández-Green and López
Nicolás, Pacchiarotti, Payne, McDonnell and Truong, Musgrave, Arka, and Rajeg,
Austin, and Zúñiga, Arkadiev, and Hegedűs.
Zúñiga describes for Mapudungun a situation he analyzes as “an erstwhile instance
of the causative-applicative isomorphism that has given rise to two different templatic
slots”. A situation that might also be analyzed as historically linked to applicative-caus-
ative polysemy is described by Tuite for Kartvelian languages.
The data analyzed in this book also confirm the existence of a widespread tendency
to use polysemous applicative-causative markers in causative function with monova-
lent verbs assigning a patientive role to their single argument (unaccusative intransi-
tives), and in applicative function with monovalent verbs assigning an agentive role
to their single argument (unergative intransitives) and with bivalent/transitive verbs.
 31 Applicative and related constructions: Results and perspectives 1069

There are, however, counterexamples to this generalization. For example, in Bantu

languages, the possibility of applicative constructions in which a marker predomi-
nantly used in causative function licenses applied phrases expressing the role of Instru-
ment does not affect the possibility of using the same marker in causative function with
the same verbs. In Wolof (Voisin and Creissels) the same derived form of ‘sit’ can be
used with a causative meaning (‘make sit’) or with a surrogative meaning (‘sit in behalf
of s.o.’ > ‘represent’), in Standard Indonesian (Musgrave, Arka, and Rajeg) the same
derived form of ‘sew’ can be interpreted as ‘make s.o. sew’ (causative) or ‘sew for s.o.’
(applicative), etc.
An intriguing pattern signaled by Voisin and Creissels in two groups of Atlantic
languages (Joola and Nyun) concerns markers productively used with a causative func-
tion that also fulfill an applicative function, but exclusively with a small set of verbs
denoting bodily secretions.
Another interesting phenomenon found in the Atlantic language Laalaa concerns a
verbal suffix mainly used in instrumental applicative constructions. Laalaa has a verbal
suffix that can be characterized as causative-autobenefactive, found in construction in
which the referent of the subject is both the Instigator and the Beneficiary of an action
carried out by an unspecified Causee. In this causative-autobenefactive construction,
the Causee can only be specified if the instrumental applicative marker is added to the
causative-autobenefactive marker.

5.2 The applicative-reciprocal polysemy

Among the languages represented in this book, the applicative-reciprocal polysemy is

attested in some of the Atlantic languages analyzed in Voisin and Creissels’s chapter,
in some Bantu languages (Pacchiarotti), in Yup’ik (Mithun) and in Southern Zapotec
(Hernández-Green and López Nicolás). In Southern Zapotec, the addition of a suffix
otherwise marking comitative applicative constructions to the transitive verb ‘stain’
does not seem to modify the transitivity of the construction but yields meanings that
can be glossed as A got.stained.together.with B (passive-sociative) and A is.involved.
in.mutual.staining.with B (reciprocal).

5.3 The applicative-antipassive polysemy

The applicative-antipassive polysemy is attested in some of the Atlantic languages ana-

lyzed in Voisin and Creissels’s chapter and in Yup’ik (Mithun).
In Mapudungun (Zúñiga), a verbal suffix mainly used as an applicative marker also
has an antipassive function with a limited number of verbs.
1070 Denis Creissels and Fernando Zúñiga

Van Gijn also mentions that the Kawapanan languages Shiwilu and Shawi have
cognate verbal suffixes acting as applicative markers with some verbs, and as antipas-
sive markers with some others.
In Standard Indonesian (Musgrave, Arka, and Rajeg), a suffix typically used as an
applicative marker has an antipassive function with the ditransitive verb ‘give’, con-
verting it into a monotransitive verb projecting clauses in which the Recipient can only
feature as a prepositional phrase.

5.4 Applicative markers also used to mark non-causative

A/S-nucleativization

The data analyzed in this book attest polysemy patterns involving applicativization and
less common types of valency alternations that can be subsumed under the general
notion of non-causative A/S-nucleativization. As commented in more detail in the Intro-
ductory Overview § 6.2.1, “non-causative A/S nucleativization” is the term we use for
derived constructions in which the A/S role is taken over by a non-agentive participant
that cannot be encoded as a core term (A, S or P) in the base construction.
The use of the same marker in applicative constructions and in constructions in
which an oblique referring to a non-agentive participant is promoted to A/S role is
described in Creissels’s chapter on Tswana. It concerns a suffix most commonly used
as an applicative marker but also found in a construction in which an instrumental
oblique is promoted to A role, whereas the A of the base construction (the Agent) is
obligatorily left unexpressed.
Hernández-Green and López Nicolás describe a Northern Zapotec verbal der-
ivation yielding a construction misleadingly referred to in the literature as “applied
experiencer construction”. As they themselves acknowledge, this construction is not
an instance of applicativization, but of non-causative subject-nucleativization, since for
example an intransitive verb such as ‘be necessary’ is converted into a derived transi-
tive verb glossable as ‘need’, whose subject is an experiencer that cannot be expressed
in the BC, whereas the subject of the BC becomes the object of the derived construction.
Voisin and Creissels mention a similar example from Wolof, where the combination of
a verb ‘remain’ with a suffix otherwise acting as an applicative marker yields a derived
verb meaning ‘still have’. Voisin and Creissels analyze it as a lexicalized applicative,
because no other Wolof verb lends itself to a similar alternation, but as a valency oper-
ation, ‘remain’ > ‘still have’ is comparable to ‘be necessary’ > ‘need’, since ‘A still has B’
can be paraphrased as ‘A is a person for whom B remains’.
The use of the same marker in applicative constructions and in constructions in
which the A/S role is taken over by a Concernee (or “external possessor”) is mentioned
by Mithun in Yup’ik. A similar use of applicative morphology is also mentioned by
Thornes for Northern Paiute: in his examples (61–63), applicative morphology licenses
 31 Applicative and related constructions: Results and perspectives 1071

a Concernee in subject role, whereas the subject of the BC is converted into an object
(i.e., something like literally I died.appl her for ‘I was affected by her death’).
The use of the same marker in applicative constructions but also in constructions
in which it licenses an additional non-agentive participant in subject role, functionally
similar to Japanese or Tungusic “adversative passives” (Malchukov 1993), is also found
in Mapudungun (Zúñiga), where this mechanism concerns avalent meteorological
verbs and a subclass of non-agentive monovalent verbs. Similarly, Van Gijn mentions
that in the Kawapanan language Shiwilu, a suffix that also has causative, antipassive
and applicative uses converts the avalent/impersonal meteorological verb ‘get cold’ (as
in it gets cold) into an intransitive verb whose subject expresses the role of Experiencer
(as in he/she gets cold).
In the languages that have both P-applicativization and passivization, it is common
that non-causative A/S-nucleativization is realized compositionally, by combining P-ap-
plicativization and passivization: a phrase representing a non-agentive participant is
introduced in P role via P-applicativization, and the applicative construction serves as
the input for passivization, the non-agentive participant expressed as the applied P in
the applicative construction taking thus the role of A/S. Consequently, depending on the
theoretical framework, the use of an applicative marker to also mark non-causative
A/S nucleativization can be analyzed as a case of covert passivization of an applicative
construction.
Interestingly, among the languages surveyed in this book, the Atlantic language
Laalaa has an instrumental applicative construction, but also a dedicated marker found
exclusively in a construction in which it licences the expression of Instruments in A role
(Voisin and Creissels).
Concerning the possible relationships between non-causative A/S-nucleativization
and applicativization, the English example Atlanta outrained Seattle quoted by Zúñiga,
Arkadiev, and Hegedűs is remarkable in that it attests the possibility that a single
marker licenses these two valency operations at the same time—something even more
potent, as it were, than the applicative-cum-passive known from Philippine languages
(which, roughly, introduces a new core argument as a subject). The base construction
can only be it rained more in Atlanta than in Seattle, which means that, in this particular
case, out- (otherwise widely attested in applicative function) actually promotes both
participants to the clausal core and grants one of them subject status; outrain licenses
both the promotion of Atlanta (Location) to subject (non-causative A/S-nucleativization)
and the promotion of Seattle (Standard of Comparison) to object (applicativization).

5.5 Other possible polysemy patterns

Bahrt (2021: 110–120) also mentions the possibility of applicative-passive, applica-

tive-reflexive and applicative-anticausative polysemy, but the examples he quotes are
1072 Denis Creissels and Fernando Zúñiga

not really convincing.3 Interesting data in this perspective might be provided by a group
of Papuan languages (the languages of the Tonda sub-family of the Yam family) in which
Foley signals the existence of a verbal affix acting not only as an applicative marker, but
also as a decausative, reflexive or reciprocal marker.

5.6 Applicative and autobenefactive

Autobenefactive constructions are the result of a valency operation that does not affect
the formal valency of the verb, but modifies the assignment of semantic roles by imply-
ing that the referent of A/S, in addition to its semantic role in the base construction
(typically Agent), is the Beneficiary of his/her own action.
Unsurprisingly, in the languages that have applicative constructions in which the
applied phrase expresses the role of Beneficiary, autobenefaction is commonly expressed
compositionally, i.e. by taking the benefactive applicative construction as the input of
reflexivization.
However, autobenefaction may also be one of the possible meanings of middle voices,
as mentioned by Vanhove for Cushitic languages.
Another possibility, illustrated by the Jivaroan/Chicham language Wampis (Van Gijn),
is the “semi-reflexive” interpretation of a benefactive applicative, analyzable in terms of
covert reflexivization of an applicative construction.
There are also languages, such as Kartvelian languages, which have a dedicated
autobenefactive marker that cannot be decomposed into an applicative marker and a
reflexive marker (Tuite). The Kiranti language Belhare is also mentioned by Jacques
and Lahaussois as having a dedicated autobenefactive marker that cannot be decom-
posed as “applicative + reflexive”.

6 Conclusion: Areas for future research

To conclude this chapter, primarily conceived as a summary of the aspects of applica-
tive constructions to the typology of which the case studies collected in this book contrib-
ute significantly, we would like first to briefly highlight some of the findings that can be
viewed as particularly relevant for the typology of applicatives and related constructions:

3 According to the definitions adopted in this book, the illustrations of applicative-passive polysemy
proposed by Bahrt (2021: 110–112) do not involve applicativization, but rather non-causative A/S-nucle-
ativization (see § 5.4). In the few languages quoted by Bahrt (2021: 114–115) as having applicative-re-
flexive polysemy, applicative markers and reflexive markers show only partial resemblance. Finally,
concerning the applicative-anticausative polysemy, Bahrt himself (2021: 119–120) acknowledges that
the two potential illustrations he came across are dubious.
 31 Applicative and related constructions: Results and perspectives 1073

– Obligatory applicatives are cross-linguistically common, and the analysis of the

semantic roles expressed by applicative constructions that at first sight seem to be
analyzable as optional applicatives leads sometimes to the conclusion that they are
in fact best analyzed as obligatory applicatives.
– As illustrated in this book by Toba/Qom, and to a lesser extent by Upper Necaxa
Totonac, in the languages that make a particularly systematic use of the obligatory
applicative strategy, the coding of non-core participants as obliques compatible
with the base form of the verb may be marginal, or even completely inexistent.
– In addition to the cross-linguistically common roles of Beneficiary/Maleficiary,
Instrument, Companion (alias Concomitant), and the semantic roles related to
space (Location, Source, Path, and Goal), several chapters mention remarkable,
albeit uncommon, semantic roles among those that may be expressed by applied
phrases in applicative constructions.
– European applicatives are particular in that they virtually lack three of the four
cross-linguistically common functions of applicatives: benefactive, comitative, and
instrumental, and at the same time attest quite a few uncommon functions.
– In Concernee-Concern constructions, non-prototypical Concernees are more prone
to be encoded as applied phrases than prototypical ones.
– Several chapters confirm the generalization that polysemouns applicative-causa-
tive markers tend to be used in causative function with monovalent verbs assign-
ing a patientive role to their single argument (unaccusative intransitives), and in
applicative function with monovalent verbs assigning an agentive role to their
single argument (unergative intransitives) and with bivalent/transitive verbs.
However, exceptions to this generalization also appear in some other chapters.
– Several chapters illustrate the possibility of equipollent marking of valency alterna-
tions functionally similar to the alternation between optional applicative construc-
tions and the corresponding base constructions.

That said, we would also like to point out some issues that are not tackled systemat-
ically in this book, and should constitute areas for future research.
First, as already mentioned in § 2, the present book is biased in the sense that non-af-
fixal applicative markers are under-represented. Hopefully, a later volume devoted pre-
dominantly to non-affixal applicatives will systematize the diversity found among those
constructions, including the famous cases found in West Africa, East Asia, and Southeast
Asia. Further comparative research should also bring affixal and non-affixal applicatives
together.
By a related token, even though individual chapters say something about the known
or plausible etymologies of the applicative markers,4 this book does not investigate

4 For example, the Amharic data analyzed by Amberber point to a scenario according to which an ap-
plicative construction may result from the evolution of a construction in which, initially, an NP flagged
1074 Denis Creissels and Fernando Zúñiga

possible correlations between particular sources and specific synchronic features. For
instance, whether applicatives originating in verbs systematically differ syntactically
and/or semantically from those originating in adpositions is a stimulating question
that future research will have to address. The possibility that purely aspectual uses of
markers also acting as applicative markers might be indicative of a directional origin
has been mentioned above, but would necessitate further investigation.
A systematic investigation of the diachrony of applicative marking would also be
crucial to shed some light on what constitutes cross-linguistically the most striking aspect
of applicative constructions, namely the remarkable recurrence of several non-applica-
tive uses of applicative morphology. A recently published volume (Pacchiarotti and
Zúñiga 2022) has already been devoted to this topic, but this is domain in which much
remains to be done, in particular in the perspective of diachronic typology. There is
no doubt that a systematic investigation of the non-applicative functions of applicative
morphology and their diachrony could greatly contribute to a better understanding of
the relationships, not only between applicativization and other types of valency opera-
tion, but also between valency operations and other aspects of grammatical structure.5
The issue of areality, directly adressed in only one of the case studies collected in
this book (Van Gijn), and somewhat indirectly in several others, would also certainly
deserve being investigated more systematically.

Abbreviations
AC applicative construction
ai animate intransitive
BC base construction
ii inanimate intransitive
intr intransitive
O1 primary object
O2 secondary object
ta transitive animate
ti transitive inanimate
tr transitive

by an adposition is cross-referenced by an index attached to the same adposition in a position immedi-

ately adjacent to the verb. In such a construction, the deletion of the adposition flagging the NP automat-
ically leads to the reanalysis of this NP as an applied phrase, and of what was initially the repetition of
an adposition in combination with a resumptive pronoun as an applicative marker.
5 Note that, among the chapters that constitue this book, Musgrave, Arka, and Rajeg’s chapter on
Standard Indonesian explores the possibility of a prototype theory analysis of the polyfunctionality of
the verbal markers that have applicative marking as one of their possible functions.
 31 Applicative and related constructions: Results and perspectives 1075

References
Bahrt, Nicklas N. 2021. Voice syncretism. Berlin: Language Science Press.
Behaghel, Otto. 1923. Deutsche Syntax: Eine geschichtliche Darstellung, Bd. I: Die Wortklassen und Wortformen
(A. Nomen, Pronomen). Heidelberg: C. Winter.
Creissels, Denis. Forthcoming. Transitivity, valency and voice. Oxford University Press.
Haspelmath, Martin. 2016. Universals of causative and anticausative verb formation and the spontaneity
scale. Lingua Posnaniensis 58(2). 33–63.
Lehmann, Christian. 2006. Participant roles, thematic roles and syntactic functions. In Tasaku Tsunoda &
Taro Kageyama (eds.), Voice and grammatical relations: Festschrift for Masayoshi Shibatani, 167–190.
Amsterdam: John Benjamins.
Malchukov, Andrej. 1993. Adversative constructions in Even in relation to passive and permissive.
In Bernard Comrie & Maria Polinsky (eds.), Causatives and transitivity, 369–384. Amsterdam: John
Benjamins.
Malchukov, Andrej. 2017. Markedness effects in applicative formation. In Albert Álvarez González &
Ía Navarro (eds.), Verb valency changes: Theoretical and typological perspectives, 3–29. Amsterdam:
John Benjamins.
Malchukov, Andrej, Martin Haspelmath & Bernard Comrie. 2010. Ditransitive constructions: a typological
overview. In Andrej Malchukov, Martin Haspelmath & Bernard Comrie (eds.), Studies in ditransitive
constructions: A comparative handbook, 1–64. Berlin: De Gruyter Mouton.
Pacchiarotti, Sara & Fernando Zúñiga (eds.). 2022. Applicative morphology: Neglected syntactic and
non-syntactic functions. Berlin: De Gruyter Mouton.
Van de Velde, Mark. 2020. Concernee-concern constructions: A comparative study of external possession in
the Bantu languages. Studies in Language 44(1). 70–94.
Woolford, Ellen. 1993. Symmetric and asymmetric passives. Natural Language and Linguistic Theory 11.
679–728.
Language index
Abaza 870–871, 872, 875, 876, 877, 878–879, 880, Blackfoot 7, 16, 20, 58, 608, 611, 612, 614, 616–617,
881–882, 883, 884, 886, 888, 890, 891, 892–893, 619, 621, 625, 628, 1063
895, 899, 902, 903, 905 Boma, North 729, 730
Abkhaz 870–872, 874, 875, 877, 880–881, 883, Boni 836, 846, 857, 858, 863
884–885, 887, 888, 889–890, 892, 893, 895, Bulgarian 441, 445, 446, 451
899, 902, 905 Burunge 836, 846, 847, 849, 851, 852, 853, 860,
Abui 379, 380–381, 382 861–864
Achagua 311, 338
Acjachemem 475, 487, 493 Caddo 28, 42
Adyghe see Circassian, West Cahuilla 475, 490, 1041
Aguaruna 23, 310–311, 315, 317, 323, 329, 330, Camsá 311–312, 314
337–338 Carapana 311, 315–316, 317, 322, 329
Akhvakh 6, 1061–1062 Cherang’any 784, 795, 818, 819
Akie 783–784, 793, 810 Chewa 723, 724, 728–729, 735, 737
Alagwa 835–837, 846, 847–848, 849, 850–851, 852, Cheyenne 606
853, 861–864 Chickasaw 1023, 1024
Alamblak 353, 354, 356, 366, 368 Chingoni 23, 24
Aleut 557–559, 562–563, 571–572, 576, 585, 588, Chontal 646–647, 648–652, 661, 663–673, 674–675,
589, 593–594, 595, 597 1057
Amele 353, 356, 387 Ch’ol 661, 662
Amharic 28, 243–278, 864, 1018, 1021, 1046, Ch’orti’ 661, 662
1061, 1065 Circassian, West 870, 871–872, 874–882, 883–884,
Arabana-Wangkanguru 403–404, 413 885–905
Arabela 307, 311–312, 315, 318–319, 326, 329, 332, Cofán 311–312, 314
333, 337–340 Cree
Arapaho 621, 628, 629–630 – East, Northern 614, 618, 620
Arapesh, Mountain 350–351, 364, 372, 386–387 – East, Southern 633–634
Arbore 836, 846, 854, 855, 861, 863 – Moose 636
Aruamu 383, 385 – Plains 608, 612–613, 614, 615, 618, 624, 628,
Asmat, Central 348, 352 630, 636
Ateso 783, 791–792, 806, 807, 816, 827, 1057 – Swampy 610, 628, 631, 633–634
Cupeño 475, 477–478, 480, 481, 485, 487, 488,
Bahasa Indonesia see Indonesian, Standard 490, 498
Bajau, West Coast 972, 975, 977–980, 982, 983, Czech 420, 442, 444, 451–452, 456
985–987, 988–990, 991, 994–995, 998–1000,
1001 Dalabon 395–397, 398
Bantawa 944, 947, 948–949, 951–952, 957–958, 966 Dani, Lower Grand Valley 353, 355, 375–376,
Barasano 311, 315, 316, 329 378, 381
Bardi 395, 398–399, 413 Datooga
Bari 783, 785, 788–789, 805, 806, 816, 825 – Asimjeeg 784–785, 795, 797–799, 812–814,
Barupu 347, 349, 350, 363–366, 371, 386–387, 1049, 819–820
1055–1056 – Barbayiiga 784, 793, 795–797, 811, 819, 825
Batak, Toba 972, 973, 974, 977, 979–980, 983, 985, – Gisamjanga 784, 793, 795–797, 811, 812, 819, 825
987–989, 991–997 Dâw 311, 319
Baule 15, 62 Desano 310, 315, 329
Beja 835–836, 843, 844–845, 862–863 Dhaasanac 836, 846, 856, 858, 859–860, 863–864
Bijogo 753, 770, 772, 773, 774 Dholuo 784, 804–805, 826

https://doi.org/10.1515/9783110730951-032
1078 Language index

Ding 720, 726, 727–728, 734, 737 Indonesian, Standard 6, 25–27, 40, 49, 279–303,
Dinka, Agar 784, 786, 787, 799, 821 368, 785, 980, 984, 1047, 1054, 1064,
Diyari 402–403, 413, 1064 1066–1067, 1069–1070, 1074
Dutch 119, 293, 419, 420 Innu 612, 614, 617, 618, 620, 628, 631, 634–635,
Dyirbal 394, 411–412, 414, 1051 637–638
Iquito 311, 315, 318–319, 326, 329, 331, 332, 333
Elmolo 835–836, 846, 854, 861, 863 Iraqw 835–837, 846, 847, 849, 850, 851, 853,
English 8, 15, 45, 59, 97, 119, 128, 224, 238, 248, 860–861, 863–864
270, 274, 279, 291, 293, 372–373, 420–426, Itzaj 660, 664, 674
436–437, 451, 462, 465–467, 494, 562, 566, Ixil 653, 674, 1065
591, 608–610, 656, 671, 769, 828, 1012, 1034,
1056–1057, 1061, 1068, 1071 Jóola Fóoñi 216, 749–750, 753, 762–766, 767,
769–774, 777, 1048, 1054
Fijian, Boumaa 6, 1009, 1062 Jumjum 784, 786, 799, 804, 824, 827–828
Fore 347, 348, 352, 353, 356–358
Fula, Gombe 752, 767, 771, 773, 777 Kabardian 870–871, 874–875, 876, 878–879, 881,
887, 889–890, 894, 901
Gedeo 836, 837 Kakua 310–312, 315, 320, 327–328, 329, 331,
Georgian 339, 1045
– Modern 19, 40, 883, 913–915, 916, 917, 918, 920, Kalaallisut see Greenlandic, West
921, 922, 923–924, 925–926, 929–931, 933–936, Kalkatungu 394, 405–406
937, 1012, 1013 Kaqchikel 12, 645–646, 647, 649–651, 654–656, 664,
– Old 914, 915, 919–920, 923, 927–928, 932, 935 674–675, 1056
German 8–9, 22, 24, 26, 30, 419–421, 424–426, Khaling 943–944, 945, 946–947, 948, 950–955,
427–438, 439, 452–453, 459, 461–462, 956–965, 966
464–467, 1012, 1056, 1060, 1063 Kinyarwanda see Rwanda
Goreng-Goreng 408, 409 Kikuyu 23
Greenlandic, West 557–558, 561, 563, 569–570, 571, Kongo ya Leta 738, 739
573–575, 576–577, 578–579, 583–584, 588, 590, Kopar 368–369, 371
594–595, 1010, 1055, 1066 Korean 1013
Guarijío 45, 509, 527, 529–530, 532, 534–535, 536, Koreguaje 311, 315, 329, 330, 338
538–539, 540–543, 545, 546, 549, 550–551, Kotiria 310, 315, 316, 321, 329
552–553, 1062–1063 Ktunaxa 1010, 1013–1014, 1036
Gungabula 407 Kubeo 311, 315, 317, 329
K’iche’ 15, 645, 654, 656
Hayu 943, 951–952, 959–960, 962, 963, 966
Hopi 475–477, 481–483, 486–489, 491, 498, Laalaa 752, 769, 771, 772, 774–775, 778, 1056, 1069,
501, 502 1071
Hua 352, 357–358 Lai, Hakha 27–28, 1009
Huastec, Southeastern 658–660, 675 Lango 784, 786–787, 802–803, 804, 821, 822–823,
Huichol see Wixárika 828
Hul’q’umi’num 79–113, 1045–1046, 1051, 1053, Latvian 443, 445, 449, 466
1058–1059, 1064 Laz 913–917, 920, 922, 924–925, 928–929, 931, 933,
Hungarian 419–420, 428–429, 436, 439, 453, 934
457–465, 466, 1012, 1046 Lengola 725
Hup 311, 315, 319–320, 328, 329, 331, 339–340, 1051 Limbu 947, 949, 950, 953, 957, 965
Lithuanian 439, 440, 442–444, 446–448, 449,
Ilocano 43 450–451, 452, 453, 455, 462, 464
Imonda 352 Lomwe 721
 Language index 1079

Londo 724 Nen 370

Luba-Kasai 724, 726, 731, 732 Ngamini 402–403
Lunda 726, 732, 733 Ngan’gityemerri 401
Luo, Northern 815 Ngiyambaa 409
Ngkolmpu 369–370
Maasai 783, 788, 792–793, 807–809, 816–817, Ngwi 721
823–825, 828, 1049 Noon 752, 774, 775, 776
Mabaan 784, 821 Nsong 734, 735
Macedonian 444, 446, 451–452 Ntandu 722
Makasar 972, 975, 976–977, 978–980, 982–983, Nuer 784, 801–802, 821
984, 985, 986, 988–989, 992–994, 995–996, Nyole 10, 11
997–999, 1000 Nzadi 722
Mandinka 6, 8, 1062
Mankanya 68, 752–753, 770 Ocaina 311–312, 314
Mankon 15 Ojibwe
Mapudungun 10, 15, 24, 25, 40, 43, 58, 179–209, – Nishnaabemwin 614, 624, 626, 628
706, 1033, 1035, 1036, 1050–1051, 1054, 1064, – Ojicree 625, 626
1067–1069, 1071 – Southwestern 610
Marathi 15 Okanagan 25
Margany 407 Oromo 835–837, 861, 864
Marind, Coastal 347–348, 358, 359, 362, 372, 386, Otomi
1055 – Eastern Highlands 692, 694, 704, 705–706, 711
Marrithiyel 401 – Old 683, 686, 688–689, 690, 691, 699, 700, 702,
Mbuun 733, 735 711–713, 1045
Menominee 603, 607, 611, 613, 614, 619–622, – San Felipe 680, 689
623–624, 626–627, 628, 629
Meskwaki 605–606, 607–608, 609, 610, 612–613, Pahka’anil 475–476, 478, 481, 486–487, 488, 490,
614, 615, 619, 621–622, 624, 628, 630, 638 491, 498, 501, 504
Mian 355–356, 386 Paiute, Northern 475–478, 479–480, 481, 483–486,
Mikarew see Aruamu 487, 488–489, 492, 493, 494–495, 496–497,
Mingrelian 913, 915–916, 920, 922, 923, 924–925, 498–502, 503–504, 1066, 1070
928, 933, 934, 935 Palor 752, 774–775, 776
Miraña 311–312, 314 Pendau 972–973, 975, 977–979, 982–983, 985, 986,
Mixe, Tamazulápam 12 991, 992, 994, 996, 998, 1001
Mixtec, San Miguel Chalcatongo 104, 681 Piapoco 311, 338
Mongo-Nkundo 731, 735–736, 738, 1036 Pima Bajo 509–511, 512–514, 515, 516–517,
Muinane 311–312, 314 518, 521, 534, 536, 541, 545, 547–548,
Muniche 308, 311–312, 314–315 551, 553
Murrinhpatha 400, 414, 1055, 1059 Pitta-Pitta 404, 1009
Murui 311–312, 314 Polish 440, 441, 446, 450, 453
Myaamia 612–613, 614, 621 Poqomam 654, 657
Puinave 311–312, 314, 320
Nahuatl, Classical 3, 31–33, 34–39, 47, 1053 Puma 944, 945, 952, 965
Nandi 784, 794, 795, 809, 810, 817–818,
825, 826 Quechua
Nasal 972–973, 975, 977, 979, 980, 982, 983–984, – Ayacucho 21, 313
985–986, 990, 991–992, 994, 999–1000 – Napo Lowland 311
Naskapi, Western 614, 618, 620 – Pastaza 311
Ndut 753, 769, 776, 777 – San Martín 313
1080 Language index

Quichua Sundanese 972–973, 974, 977–980, 981–982, 984,

– Calderón Highland 311 985, 986, 988, 991, 992–993, 994, 995, 996,
– Cañar Highland 311 997, 998
– Chimborazo Highland 311, 313 Svan 913–916, 918, 919, 920, 922, 923, 924–925,
– Imbabura Highland 311, 313 927–928, 932, 934
– Loja Highland 311
– Salasaca Highland 311 Tairora 353
– Tena Lowland 311 Tanimuka 311, 317
Qom see Toba Tarahumara 509, 512, 515, 527, 528–529, 530–538,
Q’anjob’al 661, 1025 540–541, 543, 545–546, 548, 549, 551–553,
1062–1063
Reel 784, 800–802, 827 Tariana 311, 318, 335, 337–339
Rendille 836, 846, 855, 858–859, 860, 863–864 Tauya 377, 1009
Resígaro 311, 338 Teiwa 349
Romanian 1021, 1022 Telefol 352, 353, 356
Rundi 729, 739–740 Tepehuan, Northern 509, 511, 512, 514–515,
Russian 438–439, 440–441, 443–447, 449–452, 453, 518–521, 545, 547, 550–553
454–457, 871 Thulung 943–944, 947, 952, 953, 958
Rwanda 17, 39, 725, 729, 1035, 1042 Tiene 727
Tikuna 311, 314
Salako 972, 975, 977–980, 982, 985–986, 991, 992, Toba 143–177, 1048–1049, 1051, 1053, 1056–1057,
994, 997 1064, 1073
Sasak, Ampenan 972–973, 975, 976, 977–980, Tojolabal 647, 661, 1046
984–986, 989, 992, 993, 998, 999, 1001 Tolai 29
Secoya 311, 315, 329 Tongan 1009, 1018, 1019–1020
Seereer 19, 752, 768, 771, 773, 774, 775, 777 Totonac, Upper Necaxa 44, 115–142, 1036,
Selepet 376–377, 378–379 1048–1050, 1052–1053, 1056, 1059, 1064–1065,
Serbian 450 1073
Serrano 475, 487–489, 490, 502 Tseltal 647, 661, 668
Shawi 311–312, 315, 318, 324, 325, 326, 329, 332, Tsotsil 40, 647, 661, 662, 1024, 1025
333, 336, 338, 340, 1070 Tswana 9, 11, 21, 22, 23, 211–241, 682, 725, 731,
Shilluk 784, 786, 803, 804–805, 814, 815, 826–828, 732, 735, 956, 1020, 1024, 1046–1050, 1052,
1048 1055–1056, 1058–1059, 1068, 1070
Shiwiar 310–311, 315, 317, 322, 323, 324, 329, 330, Tukang Besi 994, 999, 1034, 1041
334, 337–338 Tukano 310–311, 315, 316, 317, 322, 329
Shiwilu 307, 311–312, 315, 318, 324–325, 329–330, Tunni 857, 859, 861–863
331, 332, 333–334, 335, 336, 337–340, 1055, Turkana 783, 787, 789–791, 806, 814, 816, 825, 826,
1070–1071 827, 829
Shoshoni Tzeltal see Tseltal
– Gosiute 479 Tzotzil see Tsotsil
– Tümpisa 487, 491–492, 493, 494–495, 497 Tz’utujil 12, 656
– Western 484, 485, 487
Shuswap 25 Ubykh 870–873, 874, 880, 881, 883–884, 890, 891,
Sikongo 725 892, 895–896, 898, 899, 905
Siona 311, 315, 329 Unami 611, 613, 614, 618–621, 624, 628
Sliammon 17 Unangan see Aleut
Slovene 446, 448, 452 Urarina 311–312, 314
Somali 61, 836–837, 846, 857, 859, 860, 873; Urim 357, 372
see also Tunni Ute 475, 487, 489, 492–493
 Language index 1081

Waka-Waka 408 Yaqui 509–511, 512–513, 514, 521–527, 1046,

Walapai 18 1055–1056, 1059
Wambaya 395, 401 Yeri 372–375
Wampis 315, 317, 322, 323, 324, 329, 337–338, 1072 Yidiny 394, 409–410, 411–412, 414, 1051
Warekena 311, 338 Yimas 347, 348, 353, 354–355, 356, 358, 360–362,
Warrgamay 410, 412, 414 366, 367–368, 371–372, 384, 386–387, 1048,
Watam 347, 382–385 1055–1056, 1059
Wixárika 1011, 1014–1015 Yuhup 311, 319–320
Wolof 62, 749–750, 752–753, 754–760, 761, 768–777, Yukuna 311, 313, 315, 317–318, 322, 328–329, 331,
1048, 1069–1070 337–338, 1055
Wubuy 395–397, 398 Yup’ik, Central Alaskan 557–558, 559–562, 563,
564–570, 572–573, 576–583, 585–594, 595, 1056,
Xamtanga 836, 838, 839–843, 862–864 1069–1070

Yagua 308–309, 311–312, 315, 319, 326, 327, 328, Zapotec

332–334, 337–340 – Northern 679, 682, 686, 687–688, 691–692, 694,
Yakkha 949, 952, 955, 962, 965, 968, 1054, 1063 695–698, 699–700, 701, 702, 705, 706–710,
Yalarnnga 407 711–713, 1048, 1053, 1070
Yale 377, 378–379 – San Lucas Quiaviní 4
Yaluyandi 402–403 – Santa María Quiegolani 681
Yao 728 – Teotitlán 681
Subject index
ablative 356, 397, 560–562, 580–581, 586, 591, 757, anticausative, decausative 41, 58, 120, 216, 239,
761, 771, 774–775, 811, 835, 854–855, 863, 945, 249–250, 437, 787, 822, 934, 1072
958, 960, 1055 antipassive 32, 34, 58, 79, 81, 91, 94–95, 102,
accompaniment (semantic role) see comitative 110–111, 117– 122, 124, 147–148, 160, 162, 188,
across-the-board applicativization 1047–1048, 1053, 299, 314, 391, 411–412, 414, 483–484, 503,
1058 567–568, 583, 585, 589–590, 595–596, 645,
actor voice, agent voice 43, 183–186, 190, 204, 659, 691, 751, 762, 775, 787, 804, 822, 827, 894,
280–281, 297, 300 904, 934, 937, 973, 1053
adposition 58, 143, 145, 190, 248, 272, 274, 326, applicativa tantum see applicative deponent
420, 434, 436, 464, 687, 768, 775, 787, 789–790, applicative-anticausative co-expression
828, 837, 846, 858, 861–862, 914, 936, 1012, pattern 1071–1072
1017–1019, 1040, 1049, 1052, 1054, 1074 applicative-antipassive co-expression pattern 775,
addressee, hearer (semantic role) 122, 335, 396, 779, 1069–1072
591, 656, 662, 665, 675, 767, 783, 787–788, 791, applicative-causative co-expression pattern 17–18,
794, 797, 799, 813, 819, 827–828, 842, 848, 853, 187, 282, 284, 287, 294, 368, 391, 411, 438, 475,
859, 877, 892, 896, 935, 938, 962, 971–972, 485–486, 494–498, 502, 504, 509, 511, 534,
989–990, 1001 547–549, 551, 553, 759, 769, 775, 779, 825, 943,
adverbial 117, 133, 166, 179, 206, 248–249, 367, 384, 946, 948, 953, 980, 997, 1007, 1009, 1010–1015,
386, 438, 477–478, 549, 605, 609, 638, 681, 1026–1027, 1058, 1063, 1068, 1071, 1073
806, 812, 872, 880, 985, 1064 applicative deponent, applicativa tantum 9, 14, 160,
adverbial inflection see oblique registration 203, 436, 455, 467, 509, 511, 550, 553, 876, 913,
adversative 103, 111, 225, 265–266, 363, 557, 922, 924–925, 931, 1035
579– 581, 590, 595, 1071 applicative lookalike 3, 5–6, 9, 13–14, 34, 38, 45–50,
agent 5, 27– 29, 31, 42–43, 92–93, 119, 129, 181, 59, 140, 184, 200, 205–206, 240, 265, 275–276,
183– 185, 191, 200–201, 213–214, 233, 238, 286, 289–290, 318, 334–335, 337, 372, 387, 391,
250, 254–255, 260, 267, 270, 275, 287, 290, 402, 426, 429, 436–437, 455–456, 464–467,
293, 314, 394, 404, 407, 411, 425, 434, 494, 509, 511, 550, 553, 601, 616, 623–624, 626,
503, 510, 561, 563, 567–568, 581, 587–591, 630, 638–639, 645, 652, 674, 703, 706, 711, 713,
593, 595, 631, 651–652, 670, 690– 692, 696, 737, 742, 823, 835, 837, 839–840, 842–843,
708, 719, 731, 735–736, 750, 772, 783, 786, 792, 846–847, 850, 852, 860–861, 863–864, 903,
802, 809, 815– 817, 819, 827–828, 869, 881, 906, 913, 919, 932, 939, 965, 971–972, 979, 994,
886, 898– 900, 906, 917, 926, 974, 976, 988, 1002, 1008, 1021, 1045, 1060–1061, 1064–1065
1007–1008, 1016, 1017, 1027, 1036, 1039, 1058, applicative marker stacking 101, 140, 188–189, 536,
1068, 1070, 1072 727, 741, 877, 927, 1036, 1041, 1050, 1052–1053;
agent demotion 869, 886, 897 see also double–applicative construction
agentive intransitive verb see unergative verb applicative particle 14–15, 57, 419–421, 423–424,
agent voice see actor voice 426–429, 434, 436–437, 439, 457–466, 1046
aktionsart see aspect(uality) applicative-passive co-expression pattern
allative 21, 27, 116, 119, 123, 132, 134, 157, 161, 164, 1071–1072
166, 171, 352, 356, 359–360, 367, 397, 429, 464, applicative periphrasis see periphrastic applicative
494, 560– 562, 591, 766, 769, 797, 851, 856, applicative-reciprocal co-expression pattern 775,
919, 922, 1051 1069
ambitransitivity 7, 160, 172, 423, 428, 495, 562, applicative-reflexive co-expression
565– 568, 570, 575, 579–580, 582, 590–591, pattern 1071–1072
593, 595–596, 1062 applied-oblique construction see X-applicative
analytic(al) applicative see periphrastic applicative argument frame see argument structure

https://doi.org/10.1515/9783110730951-033
1084 Subject index

argument structure, argument frame 41, 255, 293, catch-all applicative see semantically
308, 364, 419, 424, 438, 443, 455–457,488, 516, underspecified applicative
562, 580, 585, 591, 607, 761, 783, 807, 815, 819, causal applicative 222, 767–768, 771, 805, 849, 863
824–825, 872, 877, 902, 905, 957, 972, 989, causal vs. noncausal 510, 533–534, 549, 1061–1063
994–995, 1038–1039, 1058 causative 5, 6, 28–29, 32–33, 35, 38–39, 41– 43,
aspect(uality), aktionsart 8, 10, 46, 58, 166–167, 58, 79, 81, 84–86, 91, 94–95, 98–99, 102,
187–188, 235, 285–286, 355–356, 419–421, 110, 115, 118–120, 129, 145–148, 160, 179, 181,
423–424, 426, 428–429, 437–438, 440, 185–188, 192, 199–200, 202, 205–206, 211–212,
456–457, 465–466, 606, 719, 738, 740, 215–216, 218, 221, 236–239, 250–251, 259–260,
742, 776–777, 798, 803–804, 806, 810, 819, 279–280, 282–284, 286, 289–293, 298,
825–826, 828, 914, 946, 971, 994, 1002, 1015, 307–308, 314–315, 318–319, 325–326, 332, 335,
1066–1067, 1074 338, 340, 368–369, 386, 391, 393, 404, 409,
autobenefactive 187, 771, 837, 932–934, 962, 1054, 411, 413, 425, 455, 466, 475–478, 482, 487–488,
1056, 1069, 1072 490–491, 494–495, 498–499, 502–504, 511,
517–518, 528–529, 536–537, 547–549, 565,
B-applicative 749–755, 759, 761–762, 767–771, 569, 604, 645, 663, 674, 684–685, 690–691,
775–776, 778–779, 787 696, 699, 703–704, 710–713, 721, 725–726,
benefactive, beneficiary 7, 16, 19, 20–21, 23, 25–29, 728–729, 740–741, 759–760, 762–763, 767,
36, 39, 43–45, 79–80, 84–85, 87, 89–90, 769, 770–772, 787, 808, 825, 827, 836, 862,
93–94, 98, 100–103, 105, 110–111, 116, 119–124, 872, 892–893, 896–897, 906, 914, 925, 927,
143, 152–153, 158, 162, 166, 169, 171–172, 192, 929–930, 937–938, 949–951, 955, 959,
198–199, 211, 214, 216, 218, 222, 225, 227, 231, 971–972, 979, 981, 996–997, 1002, 1008–1011,
239–240, 243, 247–248, 251, 253, 256–257, 1013–1017, 1019, 1026–1027, 1039, 1050, 1053,
258–260, 264, 266–267, 269–271, 274–275, 1056, 1062, 1067–1069
280, 283, 286–288, 292–293, 298–301, 307, cause (semantic role) 79, 158, 218, 222–224, 240,
316, 320321, 323, 328, 330–331, 335–337, 349, 357, 361, 365, 374–375, 406, 503, 557,
340, 347, 351–361, 365, 366, 368, 370–375, 573–574, 624, 764, 767–768, 779, 797, 805, 835,
377, 386, 391, 396–397, 400, 403–405, 898, 1048, 1054
413–414, 434, 436, 454, 456, 461, 464, 466, causee 33, 98, 118, 187, 214, 238, 251, 260, 495, 700,
475, 478, 480–481, 484, 486, 488, 490–495, 702, 713, 771–772, 808, 892–893, 897, 906,
497, 501–504, 509–511, 516, 519–525, 528, 1009–1017, 1027, 1035, 1053, 1056, 1068–1069
530–531, 533, 535–541, 543, 546, 552–553, co-actor, co-agent, co-experiencer see comitative
557, 562, 565–566, 582, 584–585, 593–594, co-expression pattern, polysemy, syncretism see
596, 601, 613, 617–618, 620–621, 638, 655, applicative-anticausative co-expression
657–658, 661–662, 664–665, 668, 672, 675, pattern, applicative-antipassive co-expression
682, 684–686, 689–696, 698–705, 712–713, pattern, applicative-passive co-expression
728, 730, 733, 735, 737, 741, 749, 751–754, pattern, applicative-reciprocal co-expression
76– 762, 767–769, 771–772, 778, 783–784, pattern, applicative-reflexive co-expression
787–797, 799–805, 809, 813, 816, 818–819, pattern
825–828, 835, 839–840, 842–844, 848–849, comitative, co-actor, co-agent, co-experiencer 4,
851–855, 857–858, 861–863, 869, 877, 880, 20–21, 27–29, 36, 116, 119, 129–132, 143,
882–885, 888–889, 896–898, 906, 913, 152–153, 159, 168, 170, 179, 196, 199, 206, 267,
918–922, 924–926, 928–934, 937–938, 943, 274, 292, 307, 308, 319, 327, 332–333, 336–337,
945, 952–953, 955–956, 958–960, 962–964, 347, 349, 354, 359–360, 362, 365, 369, 371,
971–972, 978, 980–985, 989, 994, 998–999, 374, 391, 396, 398–402, 406, 407–410, 412–414,
1001–1002, 1011, 1013–1014, 1021, 1023–1026, 434, 445, 454, 522, 539–540, 542, 545, 557,
1033–1037, 1039–1042, 1048, 1052–, 565–567, 569, 571, 583–584, 594–595, 613,
1057–1059, 1061, 1067, 1069, 1072–1073; 619, 628–630, 637–638, 654, 665, 675, 687,
see also B-applicative 694, 696–702, 712–713, 725, 755, 761, 768,
 Subject index 1085

770–771, 778, 783, 790–791, 800, 805–807, decausative see anticausative

811, 813–815, 818, 828, 847, 859, 863, 869, definiteness 119, 183, 245–247, 249, 251, 255, 263,
882–884, 886–887, 896–897, 906, 949, 267–269, 275, 321, 455, 457, 462–463, 479, 557,
987–988, 1010, 1012, 1016, 1022, 1025, 1027, 561, 567, 568, 589, 591, 596, 626, 629, 631, 658,
1035–1036, 1048, 1053–1055, 1060, 1065–1066, 675, 692, 808–809, 816, 855, 880–881, 976,
1068–1069, 1073 995–996, 1000, 1002, 1066
companion (semantic role) see comitative denominalization 46, 91, 187–188, 196–197, 199,
companion argument, companion phrase 58, 140, 203–204, 307, 423, 429, 436, 465, 476, 487,
200, 239, 552, 827, 862, 971, 979, 981–983, 500, 502, 548, 950
985–987, 991, 1001 dependent marking 348, 394
comparative applicative, prioritive applicative, deputative, substitutive, surrogate 162, 166, 169,
surpassing applicative 27, 336, 421, 424, 439, 270, 538, 553, 582, 595, 759, 843, 884, 920,
451, 466, 730, 980, 1056 978, 981, 1035, 1069
complex predicate 91, 484, 943–944, 946–947, destination of motion see allative, goal
952–953 detransitivization, intransitivization 6, 26, 41, 97,
concernee, external possessor 45, 157, 179, 190, 101, 125, 130, 162, 314, 369–370, 405, 410–412,
196, 199, 206, 224–225, 354, 371–372, 374, 596, 659, 675, 888, 906, 937–938, 1051
376, 397, 464, 516, 552, 634, 661, 700, 755, differential object marking 194, 245, 252, 321–322,
823–824, 828, 885, 906, 1007, 1022–1025, 327–328, 339–340, 1052
1028, 1058–1059, 1061, 1070–1071, 1073 directional applicative 29, 80, 83, 85–86, 90, 97,
concomitant (semantic role) see comitative 99, 102, 104, 108, 110, 143, 151–153, 155–156,
content of speech, content of thought see topic of 159–160, 166–167, 170, 172, 1056
speech directional marker 32, 59, 147, 149, 320, 457–459,
converb 15, 45, 57, 147, 162, 186, 244–245, 248, 461, 463–464, 477, 484, 513, 522–523, 544,
253–254, 274, 525, 730, 740, 826, 836, 783–785, 793, 795, 799, 811, 815– 817, 819–821,
838–840, 842–843, 862 825–828, 843, 846, 853–855, 857, 864, 872,
co-participant see comitative 875, 891, 895, 904, 1066–1067, 1074
discourse 58, 81, 105, 109, 119, 140, 144, 179, 183,
D-applicative (formal subtype of applicative 185, 195, 244, 247, 251, 257, 266, 280, 301, 321,
construction) 19, 42–45, 257, 404–405, 413, 385, 435, 476, 486, 492, 498, 500, 504, 511,
419, 433, 440, 445, 466, 510, 862, 877, 905, 917, 514, 542–543, 545–547, 552–553, 557, 559,
938, 1012, 1018, 1024, 1027, 1051, 1053 584–585, 590, 609–610, 622, 630, 632, 638,
dative 5, 10, 22, 120, 197, 358, 370–372, 382, 386, 653, 729, 736–737, 739–743, 785, 829, 852, 864,
397, 403–404, 410–412, 436, 445, 456, 461, 973, 975, 999, 1060
463–464, 466, 493, 499, 512, 545, 572, 654, ditransitive construction, ditransitive verb 27,
675, 837, 839, 841–842, 846, 848, 855–856, 31–36, 44, 58, 79–80, 87, 91, 96–98, 105, 110,
858, 861–862, 864, 916, 919, 930, 937–938, 159, 161, 166, 172, 239, 256, 270–271, 275, 280,
943–944, 958, 1013, 1016, 1020–1021, 1023, 283–288, 292, 298–299, 320–324, 327–328,
1034, 1051, 1067 352, 358, 360, 371, 373, 377, 386, 391, 397,
dative alternation, dative shift 8, 45, 426, 466, 608, 405–407, 413, 421–423, 432, 461, 463–464,
822–823, 828, 1061 482–484, 490, 492–493, 513, 525, 533, 547,
dative applicative (semantic subtype of applicative 551, 602, 607–610, 621–622, 632–633, 638, 657,
construction) 80, 84–85, 87, 89–90, 104, 110, 684, 691–694, 712, 731, 733–734, 741, 787–789,
293, 331, 333, 335, 658–661, 700, 702, 713, 843, 863, 877–878, 893, 897, 906, 945, 958,
783–784, 787–799, 802–806, 808, 809–810, 960, 966, 971, 981, 983, 985–986, 989, 991,
813–814, 818–819, 822, 824–828, 854, 857, 995, 998, 1001, 1007, 1020, 1049, 1070
869–870, 882, 891–895, 897, 900, 902–903, double-applicative construction 161, 172, 218,
905–906 325, 413, 536–538, 548, 552, 809, 913, 930,
dative shift see dative alternation 938, 1036
1086 Subject index

double marking of grammatical relations 348, grammaticalization 14–15, 29, 33, 57, 104, 139,
394, 871 143, 145, 150–151, 155, 159–162, 166–167, 170,
double-object construction, double-P 172, 181, 187–188, 239, 248, 254, 275, 316–317,
construction 22, 45, 58, 190, 220– 222, 314, 319, 349–350, 356, 367, 377, 386, 458, 476,
327, 422, 481, 513, 525, 534, 541, 563, 595, 608, 485, 487, 489, 525, 646, 651, 760, 842, 845,
613, 733, 754, 1034, 1049, 1050–1051 861–862, 864, 883, 891, 953, 966, 985, 998,
1014, 1018, 1046, 1064, 1067
equipollent marking 6, 7, 20, 183, 437, 510–511,
532, 534, 542–543, 549–551, 616, 904, 955, habituality 211, 235, 240, 425, 440, 729, 776, 777,
1020, 1061–1063, 1073 809, 814, 994
ergative 40, 79, 81, 87, 92, 370, 394, 557, 560–563, head marking 145, 347–348, 358, 394, 650, 679
580, 594–596, 646, 649, 666, 786, 789, 827, hearer (semantic role) see addressee
869, 871–873, 897–900, 903, 906, 916–917, holism effect 30, 431, 436, 1060, 1064
944, 948, 957, 973, 1018
experiencer 24, 29, 130–131, 196, 205, 261, 267, I-applicative 749–753, 756–763, 767, 771–779, 787
269–271, 294, 319, 422, 466, 580–581, 706–707, incorporation 32, 37, 45, 58, 184–185, 189, 193–194,
710–711, 713, 731, 888, 919–920, 922, 935, 997, 253, 347–348, 350–351, 381, 386, 394, 414,
1012, 1016, 1070–1071 484, 585, 595, 687, 880, 883, 890, 1017–1020,
external possessor see concernee 1040–1041
indirect object 5, 8, 19, 22, 24, 87, 120, 187, 214,
flexivalency, lability, polyvalency 6–8, 160, 172, 184, 245, 256, 270, 321–323, 328, 335, 340, 419,
192–193, 198–200, 202, 318, 426, 436, 466, 510, 430–433, 480, 499, 510, 513, 536, 553, 566,
639, 692, 712, 823, 886, 944, 965, 1017, 1062; 689, 750, 795, 818, 869, 871–873, 876–878,
see also ambitransitivity 880, 888–889, 892–893, 895, 897–902,
focalization 31, 46, 58, 195, 202, 211, 214, 216, 218, 905–906, 913, 916–920, 922, 925–927,
234–235, 240, 387, 543, 553, 697, 699, 713–714, 930–935, 938–939, 1025, 1051–1052, 1056
729, 734–735, 740–742, 755, 761, 763, 775–777, information structure 46, 140, 216, 258, 269, 273,
827, 863, 1048, 1065 382, 384–385, 723, 737–738, 741, 1060
force (semantic role) 29, 107, 267 instrument, instrumental 12, 18, 26, 27, 29, 39, 45,
83, 99–100, 116, 119, 125–129, 135–139, 157,
goal (semantic role) 19, 25, 29, 33, 36, 39, 79, 211, 220, 231, 233, 237–238, 240, 247, 253,
83, 87, 103, 108, 111, 116, 119, 132–135, 143, 257–258, 261–263, 267–268, 272–275, 280,
151–157, 159, 162–163, 166, 170, 172, 179, 196, 283, 285–286, 288, 291, 293, 301, 308, 313–314,
198–199, 205, 228–230, 235, 240, 247–248, 319, 327, 334–335, 337, 349, 359, 399–400,
263, 267, 273–274, 283–285, 291–293, 300, 406, 410–414, 429, 434, 444, 448, 453–454,
331, 334, 347, 367, 374–375, 379–381, 396, 462–463, 478, 539, 545, 557, 560–562,
399–400, 405, 407, 409, 422, 434–435, 569, 571, 573, 576, 584, 594–595, 653–655,
449, 455, 464, 509, 511, 544, 557, 566, 571, 657, 659–660, 662, 665, 674–675, 687, 702,
584, 594, 601, 608, 613, 618–619, 624, 627, 704–706, 708–709, 711, 714, 719, 725, 730–731,
638, 662, 683, 700, 702, 711, 731–732, 737, 734–735, 737, 742, 749, 751–753, 756–757,
741–742, 754, 767, 774, 783, 787–788, 790–797, 761, 763–765, 767, 771–778, 783–784, 787,
799–800, 809, 811, 812–813, 815–819, 821, 797, 800, 805–811, 813–816, 820, 825–828,
824–825, 827–829, 860, 877, 882, 884, 893, 835, 846–847, 854–855, 858–859, 863, 872,
900, 922, 937, 956, 962–963, 965, 971–972, 880–881, 883–884, 887, 890, 900–901, 906,
978–979, 985, 987, 989–991, 997–1002, 1022, 943–944, 948, 962, 971–972, 978, 985–987,
1035–1037, 1039, 1041–1042, 1047, 1052, 1054, 989, 994, 998–1002, 1010–1011, 1016,
1057, 1059, 1073 1018–1019, 1022, 1025–1026, 1035–1036, 1048,
goal of motion see allative, goal 1054–1057, 1065, 1068–1071, 1073; see also
goal-reached 737, 742, 792, 794 I-applicative
 Subject index 1087

intensity, intensification 9–10, 14, 46, 166, 196, 206, 330–331, 336–338, 340, 354, 356, 358–359,
211, 235–236, 240, 280, 285, 297, 319, 434, 454, 372, 391, 396–397, 400, 404–406, 413, 416, 436,
719, 738, 742, 824–825, 828, 993–994, 997, 456, 464, 466, 486, 516, 535, 544, 552, 564,
1002, 1064 656–657, 662, 665, 675, 697, 700–702, 713, 735,
intransitivization see detransitivization 787, 794, 801–802, 827–828, 835, 837, 840,
involuntary agent 886, 898 842, 845, 860, 863, 869, 882–884, 886, 888,
906, 920, 962, 964, 966, 1011, 1013–1014, 1022,
judicans see viewpoint holder 1035, 1041, 1054–1055, 1073
manner 10, 58, 100, 179, 187, 206, 463, 576, 624,
lability see flexivalency 638, 757–759, 761, 766, 772, 777, 813, 901, 906,
language contact 23, 338, 836 934, 994, 1015, 1056
lexical suffix 79, 81, 91, 99–104, 110 means (semantic role), mediative 125–126, 135,
lexicalization 9, 13, 147, 150, 158, 160–161, 172, 206, 448, 557, 572–574, 576, 756, 761, 764–765, 767,
211–212, 236–237, 240, 386, 439, 459, 464, 467, 771–772, 778, 807–809, 811, 828, 887, 906, 1048
477, 490, 498, 502, 509, 511, 541, 550, 553, 575, middle voice 32, 92, 94, 145–147, 153, 160, 172, 215,
590, 654, 660–661, 670–671, 675, 686, 703, 710, 370, 425, 437, 441, 691, 721, 766, 787, 805–807,
712–713, 759, 761, 766, 769, 784, 789, 791, 795, 809, 827, 836–837, 850, 856, 862, 934,
798, 802, 824–825, 828, 860, 877, 888, 892, 936–937, 939, 946, 961, 1067, 1072
902, 962, 994, 996, 1012, 1015, 1035, 1064, 1070 multiple-object construction 214–215, 221, 224, 227,
lexicalized applicative 9, 13, 38, 161, 172, 206, 524, 750, 769
211–212, 236–237, 240, 467, 509, 511, 550, 553,
590, 654, 670–671, 675, 686, 710, 712, 759, 761, nominalization 46, 90–91, 179, 429, 477–478, 480,
766, 784, 789, 791, 795, 798, 802, 824–825, 828, 522, 557, 576, 578–579, 584–588, 595, 671,
860, 877, 888, 892, 902, 962, 996, 1012, 1015, 695, 777, 1068
1035, 1064, 1070 non-valency-related functions of applicative
location, locational, locative 16, 24, 27, 29–30, morphology see aspect, definiteness,
45–46, 58, 100, 104, 108, 117, 133, 135, 138, 143, discourse, focalization, habituality,
149, 151–155, 160, 163, 165–168, 171–172, 189, information structure, intensity, oblique
198, 211, 213, 218, 228–235, 240, 247, 253, 263, registration, perfectivization, saliency,
272–275, 280, 283–285, 291, 293, 301, 318, 325, topicalization, topic continuity
328, 334–338, 340, 347, 349, 363–364, 374, nucleative, nucleativization 5, 41–43, 151, 185–186,
391, 396–398, 400–401, 406–414, 434, 436, 200–201, 204, 706, 713, 1007–1009, 1014, 1017,
438, 456, 462, 478, 539, 545, 561, 577, 579, 1026, 1070–1072
584, 595, 613, 617, 624–625, 654, 665, 668,
675, 719, 721, 723–724, 729, 731–732, 735–736, object of emotion, object of empathy 396, 522,
738–743, 757–759, 761, 772, 776, 783–784, 1055
787, 789, 791, 797, 800, 805–807, 809–815, obligatory applicative; 13, 22–23, 35, 47, 190, 201,
825, 826–828, 835, 849–851, 854, 859, 863, 216, 219–220, 240, 503, 510, 552, 731, 751, 753,
870, 874, 879–880, 882–883, 887, 889–892, 761, 764–766, 768, 775–776, 1046–1049, 1055,
895–896, 901–902, 905–906, 913, 971–972, 1073
978, 986, 989–991, 994, 999–1002, 1010–1013, oblique argument 14, 41, 160–161, 261, 301, 328,
1022, 1026, 1035–1036, 1039, 1041, 1049, 1054, 348, 382–384, 419, 455, 460, 467, 601, 609,
1056, 1065–1066, 1071, 1073 623, 625–626, 638
locative alternation 425, 430–432, 434, 462, 466 oblique-less language see across-the-board
applicativization
malefactive, maleficiary 16, 27–29, 36, 39, 75, oblique marking 79, 87, 542
152, 158, 162, 171–172, 179, 192, 197–199, 201, oblique registration, adverbial inflection 9, 11–12,
206, 243, 247, 252–253, 259–260, 264–266, 14, 50, 443, 467, 652–653, 655, 674, 683, 690,
269–271, 273–276, 287, 307, 318, 327–328, 703–706, 711–714, 1056, 1065
1088 Subject index

optional applicative 13, 190, 216, 220, 510, 539, polysynthesis 79, 81, 116, 143, 145, 347–348, 358,
542, 544, 552–553, 731, 734–, 751, 753, 761, 394, 559, 602, 610, 869, 871
763, 768, 774, 1035–1036, 1046–1048, polyvalency see flexivalency
1060, 1073 portative 28–29, 42
preverb 12, 49–50, 419–421, 426–428, 433, 436,
P-applicative 20, 23, 35, 41–44, 189, 204, 211, 257, 438–445, 448–450, 452, 454–455, 457, 459,
403, 413–414, 419, 421, 433, 440–441, 446, 450, 464–467, 601–602, 606–607, 619, 624–626,
459, 466, 510, 611, 628, 731, 733, 1019, 1024, 628, 629–630, 637–638, 835, 837, 845–851,
1049–1052 853–864, 874–876, 879, 883–884, 889–892,
passive 32, 34, 40–41, 58, 79, 81, 91–94, 102, 103, 895–896, 898, 901–902, 914, 924, 1064
109–111, 146–147, 160, 164, 170, 181, 184, 188, prioritive applicative see comparative applicative
191, 200–201, 205–206, 213–214, 216, 220, promotion(al) 11, 44, 46–47, 147, 160, 164, 170, 186,
226, 228, 240, 249–250, 259–260, 267, 279, 211, 220, 233, 313, 372–374, 383–384, 553, 668,
281, 295–297, 299–300, 314, 386, 422, 425, 679, 682, 703, 707, 713–714, 897, 1023, 1049,
437, 446, 483, 491, 503, 581, 585, 588–589, 1061, 1065, 1071
593–595, 631, 645, 649, 651, 664, 666, 674, purpose (semantic role) 218, 222–224, 240
692, 699, 721, 723, 728, 731, 734, 741, 762, 778,
786–787, 805–806, 827, 836, 863, 914, 926, reason (semantic role) 125, 127–128, 247–248, 314,
928, 934, 936–937, 973, 975, 977, 982–983, 557, 565, 571–576, 579, 584, 594–595, 624,
998, 1041, 1050, 1069, 1071–1072 626, 706, 735, 794, 805, 807, 809, 825, 828,
path (semantic role), perlative 136–137, 436, 849–850, 900–901, 906, 994
560–561, 732, 742, 757, 761, 765, 767, 771–772, rearranging applicative, redirecting applicative,
776, 792, 809, 828, 883, 890, 903, 906, redirective applicative, remapping
986–987, 1048, 1054, 1063, 1073 applicative 24, 26, 79–80, 84, 87, 89–92, 96,
patient 28–30, 46, 87, 91, 94, 119–120, 124, 126, 135, 97–98, 101–102, 105, 110–111, 161, 189–191, 193,
154, 179, 183–184, 196, 199, 205–206, 214, 238, 196, 205, 230, 239, 296, 422, 731, 741, 783, 792,
267, 283, 286, 290–291, 297, 299–301, 318, 327, 815–819, 821, 826–828, 985, 987, 990–991,
332, 380, 425, 464, 466, 484, 493, 509, 511, 532, 1047, 1049–1053; see also syntax–neutral,
536–537, 542, 546–548, 552, 561, 581–582, valency–neutral
584–585, 595–596, 608, 657, 689, 691, 705, recipient (semantic role) 19, 25, 27, 29, 45, 87, 105,
708–709, 719, 728, 735, 762–763, 769, 803–804, 111, 117–118, 122, 152, 159, 161–162, 166, 172,
811–812, 814, 819, 886, 900, 903, 906, 943, 966, 183–184, 214, 226–227, 256, 267, 270–271, 275,
974, 981–983, 985–986, 1001, 1016–1017, 1027, 283–285, 291–293, 299–300, 314, 321, 328,
1039–1040, 1057, 1060–1061 330–331, 356–359, 367, 373, 396, 434, 463,
patientive intransitive verb see unaccusative verb 484, 489, 509, 511, 516–517, 519, 521, 523, 525,
patient voice see undergoer voice 528–531, 533–536, 538, 542, 544, 552, 557,
perdurative applicative 446, 450 562, 564, 566, 582, 584–585, 591–592, 594,
perfectivization 429, 438–439, 443, 445, 454, 456, 596, 601, 608, 613, 651, 655–658, 660–662,
462, 464–465 665, 668, 675, 684, 686, 689–690, 693–696,
periphrastic applicative, analytic(al) applicative, 700–702, 705, 710, 712–713, 735–736, 741,
applicative periphrasis 4–5, 15, 45, 96, 204, 783, 787–789, 791, 794–795, 797, 800–803,
239, 413, 465, 527, 549, 551, 689, 835, 837–838, 805, 822–823, 828, 839, 843, 848, 853, 864,
840–843, 845, 862, 1014, 1046 877–878, 885, 892, 905–906, 920, 922, 931,
perlative see path 935, 945, 960, 963–964, 971–972, 980–985,
permissive causative 192, 238, 319, 499, 953 989, 994, 1000–1001, 1012, 1014, 1016,
person indexation 818, 825, 943–946, 1067 1021–1022, 1024, 1035–1037, 1039, 1041, 1050,
polyfunctional(ity) 280, 549, 719, 737, 948, 979, 1052, 1057–1058, 1066, 1070
1001, 1074 reciprocal 58, 79, 81, 91, 95, 97–98, 102, 110–111,
polysemy see co-expression pattern 130, 132, 146, 199, 205, 216, 239, 249, 259,
 Subject index 1089

319–320, 370, 400, 405, 413, 437, 466, 568, 266–267, 273–274, 283, 292, 318, 349, 365, 396,
570–571, 596, 627, 691, 699, 708, 711–712, 721, 400–401, 413–414, 448–449, 455, 509, 511, 522,
725, 728, 740–741, 760, 762, 766, 770, 775, 524, 526, 531, 533, 535, 544, 552, 609, 613, 619,
779, 807, 863, 872, 881–882, 888, 898, 900, 624, 626, 638, 695, 700–702, 713, 724, 731, 742,
902–903, 906, 961–962, 973, 1041, 1053–1054, 757, 759, 771–772, 774–775, 778, 783, 790, 807,
1066, 1069, 1072 809, 811, 815–817, 821, 825, 827–828, 854, 943,
redirecting applicative, redirective applicative 964, 1011–1012, 1022, 1024–1025, 1047, 1052,
see rearranging applicative 1054, 1059, 1073; see also ablative
reference tracking, referential continuity see topic stem alternation 515, 532, 679, 683–685, 688–689,
continuity 712, 785, 826, 836, 844, 943–944, 946–948,
reflexive 41, 58, 79, 81, 91, 95–98, 102, 110–111, 119, 950, 954, 966, 1034, 1046, 1063
146, 181, 199, 205, 214–216, 249–250, 259, 314, stimulus (semantic role) 24–25, 46, 79, 87, 103,
319–320, 323–324, 370, 400, 405, 413, 437, 107–108, 111, 118, 123, 199, 205, 294, 509, 511,
439, 441, 447, 453, 461, 465, 543, 690–691, 699, 519, 526, 528, 533, 552, 581, 809, 885–886, 943,
712, 721, 729, 787, 807, 863, 872, 875, 881–882, 956, 962–963, 965, 978, 992–993, 997, 1001, 1054
932–934, 937, 939, 946, 961–962, 1039, subject matter (semantic role) see topic of speech
1053–1054, 1071–1072 substitutive see deputative
registration see oblique registration superessive applicative 913, 917–919, 922–928, 930,
relational applicative 80, 86, 90, 94–95, 97–98, 932–933, 935–938, 1012
100–103, 105, 108, 110–111 surpassing applicative see comparative applicative
relational verb 601–602, 630, 631–637, 1035 surrogate see deputative
relative preverb 50, 601, 624–626, 637–638 symmetrical voice 179–180, 183, 200, 279, 281,
relative root 20, 50, 601, 609, 623–629, 638, 1052 973–974, 977, 998, 1054
relativization 46, 58, 89, 140, 195, 216, 586, 667, syncretism see co-expression pattern
697, 705, 713–714, 734,863, 876, 882, 900–901, syntax-neutral 9–10, 14, 189–191, 196, 202–203,
999–1000, 1039, 1048 206, 437, 1064, 1066; see also valency-neutral
remapping applicative see rearranging applicative
theme (semantic role) 27–29, 33, 46, 86–89, 91,
saliency 166, 168–169, 172, 756, 770, 777, 1065 100, 105, 110, 117–119, 124, 182–184, 191–192,
secondary verb 476, 484–486, 489, 503–504, 1019 255, 261–262, 267–271, 274, 283–285, 293,
semantically underspecified applicative, catch–all 299–300, 349, 400, 422, 435, 462–463, 477,
applicative 179, 205, 675, 735, 742, 1050 484, 489, 533, 542, 566, 584–585, 596, 608,
semantic effect 9, 120, 131, 202–204, 425, 713, 971, 735, 788–789, 791, 802–803, 808, 811, 815–819,
995, 1002 821–823, 825, 827–828, 917, 921, 927, 945,
semantic roles see addressee, agent, benefactive, 971–972, 981–983, 985–989, 993–996, 998,
cause, causee, comitative, concernee, 1001, 1034, 1039–1040, 1050, 1057
deputative, destination of motion, experiencer, theme sign (in the sense given to this term in
force, goal, instrument, involuntary agent, Algonquian linguistics) 605–606, 608, 611,
location, malefactive, manner, means, object 615–616, 618, 632–635, 637
of emotion, path, patient, purpose, reason, three-object construction see multiple- object
recipient, source, stimulus, theme, time, topic construction
of speech, viewpoint holder time applicative 625, 783, 800, 805, 812, 815,
serial verb construction, serialization 15, 57, 88, 827t828, 901, 906, 1055
106, 115, 143, 155, 166, 170, 172, 186, 253, topical, topicalization 31, 43, 58, 79, 109, 111, 119,
315–317, 319, 347, 349, 350–351, 354, 363, 365, 143, 163t164, 172, 183, 191, 193, 196, 205, 213,
367–368, 381, 484, 489, 661, 681, 730, 740, 826, 247, 275, 280t281, 459, 498, 504, 542, 543–544,
838, 913, 936, 947, 1018–1019, 1046, 1049 553, 557, 588, 590–593, 596, 609–610,
source (semantic role) 19–20, 25, 29, 103, 110–111, 622–623, 691, 697, 734–735, 737, 740, 742, 787,
152, 155–156, 198–199, 213, 228–230, 235, 263, 803, 827–828, 856, 864, 1015, 1060
1090 Subject index

topic continuity, referential continuity, reference valence see valency

tracking 46, 108, 140, 161, 269, 479, 545, 553 valency decrease 41, 146, 190, 202, 204, 249, 309,
topic of speech, topic of thought, content of speech, 314, 318, 320, 325, 335–336, 338, 424, 433,
content of thought, subject matter (semantic 1024
role) 25, 179, 199, 206, 522, 591, 654, 675, 885, valency increase 26, 35, 41, 44–45, 110, 124, 137,
971, 978, 992, 1001, 1056 169, 143, 146, 148–150, 160–161, 163, 172,
transitivization 6, 25–26, 41, 43, 47, 84, 100, 125, 180, 193, 196, 200, 203–206, 214, 230, 239,
130, 180, 190–191, 193–194, 196, 200, 206, 284, 249–250, 289–290, 300, 308–309, 314–315,
295, 308, 319, 350, 377, 403, 409–412, 422, 430, 318, 320–321, 324–326, 335–336, 338, 370,
432, 459, 464, 487–488, 490, 500, 502, 511, 386, 396, 413, 422–424, 433, 466, 477–478,
518, 520, 528, 534–535, 540, 663, 788, 800, 486, 498, 502, 527, 535, 540, 542, 545, 552,
903–904, 906, 948, 955, 965, 971, 992, 1019, 630, 684, 691, 703, 719, 732, 737, 741, 769, 775,
1039, 1049, 1053, 1058, 1060, 1063 787–788, 791, 795–796, 799–800, 806–807,
transitivity 7, 40–41, 47, 79, 81, 86, 101, 110, 124–125, 814–815, 819–821, 823, 825, 827–828, 906,
146–149, 160, 172, 181, 183, 201, 239, 243, 256, 943, 949, 955, 959, 990–991, 1001, 1008,
275, 284, 300, 329, 370, 375, 377, 379, 382, 387, 1010–1013, 1049, 1060, 1063
394, 421, 425, 487–488, 495, 502, 504, 559, valency-neutral 9, 11, 24, 26, 59, 188–190, 192–193,
562–563, 575, 580, 632–633, 787, 893, 904, 204, 206, 422–424, 433, 437, 466, 601, 630,
925–926, 937, 945, 956, 973–974, 990–991, 639, 737, 742, 792, 828, 864, 939, 1012,
998, 1001, 1007, 1011, 1033, 1037–1039, 1046, 1049, 1064, 1066 see also syntax-neutral,
1060, 1069 rearranging applicative
ventive 783, 793, 795, 798–799, 811, 813, 815–822,
unaccusative verb, patientive intransitive verb 18, 825–826, 828, 835, 846, 851–852, 860, 863
251, 293, 371, 380–381, 425, 475, 494–495, 497, version (in the sense given to this term in Kartvelian
503, 562, 565–566, 568, 570, 575, 579–580, linguistics) 19, 40, 50, 883, 913–914, 917–922,
582, 590, 595–596, 1039, 1058, 1062–1063, 925–926, 932–937, 939, 1010, 1012
1068, 1071, 1073 viewpoint holder, judicans 436, 466, 885–886,
undergoer voice, patient voice 43, 182–186, 190, 888, 1056
204, 279, 281, 285–286
unergative verb, agentive intransitive verb 18, X-applicative, applied-oblique construction 19–20,
251, 293, 379, 475, 497, 562, 565–567, 570, 22, 42–45, 211, 219, 227, 400, 413, 419,
575, 579–580, 582, 590–591, 593, 595, 1058, 440–441, 448, 466, 510, 552, 624–625, 731,
1062–1063, 1068, 1073 862, 1027, 1052, 1055