Quadros, Ronice Müller de. Phrase Structure of Brazilian Sign Language.

In Cross-
linguistic perspectives in sign language research. Selected papers from TISLR 2000.
Signum Press: Hamburg. 2003. p.141-162

Phrase Structure of Brazilian Sign Language

Ronice M. de Quadrosi

This work presents an overview of the syntactic structure of Brazilian

Sign Language (LSB), with the main goal of outlining the architecture of the phrase
structure of this language. This is accomplished by the investigation of basic word
order and various syntactic operations1. It is shown that SVO is the underlying word
order and that word order is influenced by the type of the verb that is used: plain or
inflecting. The phrase structure proposed for Brazilian Sign Language combines two
different approaches: (a) the functional projections of agreement and tense (AgrP-
TP) for inflecting verbs and (b) the functional projection of only one inflectional
category (IP) for plain verbs. I assume Lasnik’s (1995) proposal on the asymmetry
between have and be and main verbs in English, explained by featural and affixal
processes. Also, Bobaljik’s (1995) parameter of agreement functional projections is
also incorporated into my analysis. He claims that if affixation and merger under
adjacency take place in the derivation, syntax is blind to them; therefore, there will
not be an agreement functional projection in the phrase structure. On the other hand,
if there are feature affixes associated with verbs, they must be checked against
projected agreement features at some point of the derivation. A combination of these
two views delineates my proposal of verbal asymmetry observed in sign languages,
such as LSB. In accordance with generative investigation, all grammaticality
judgements used in our analysis were provided by native signers from the south of
Brazil2.

1 Word order

There are two works that mention the flexibility of word order in
LSB: Felipe (1989) and Ferreira-Brito (1995). They point out that this language has
different possibilities for ordering the words in the sentence, but even with this
flexibility, there seems to be a basic order of Subject – Verb – Object (SVO). I
provide empirical support for this intuition and propose a representation of phrase
structure for this language. The evidence comes from simple sentences, sentences
with embedded clauses, sentences with adverbs, with modals and auxiliaries. The
other word orders allowed in LSB result from the interaction of grammatical
mechanisms such as the ones mentioned for ASL (Chen, 1998; Matsuoka, 1997;

1
Non-manual markers are considered important to determine the structures under consideration. They
follow, with the respective conventions for transcription: head nod for emphasis/focus (____hn); wh
questions head movement and facial expression (____wh); negative head movement (____neg);
topicalization facial expression (____top) and eyegaze associated with the direction of the hands
and/or the direction established in agreement (____eg). Also used in the transcriptions is IX to
indicate an index with a, b, c letters to indicate third persons incorporated in the verb and 1 and 2
indicating the first and second persons respectively.
2
A certain degree of regional variation in LSB exists, however, this variation was not a problem for
my investigation, since I was concerned only with abstract representations of LSB phrase structure.
Braze, 1997; Fischer, 1975; Padden, 1990; Liddell 1980; Aarons, 1994; Lillo-Martin,
1986). The conclusions of these studies of ASL include the following topics:
• SVO is the underlying word order in ASL;
• SOV order seems to follow from object shift because of the presence of handling
verbs (Chen, 1998), aspectual verbs (Matsuoka, 1997; Braze, 1997), and
agreement (Fischer, 1975); there are also special spatial-plain verbs with
‘clitics’ and the possibility of analyzing a single case of SOV as consisting of
three sentences instead of one (Padden, 1990);
• OSV order follows from topicalization (Fischer 1975; Liddell 1980, Aarons,
1994); object shift because of the presence of handling verbs (Chen, 1998) and
aspectual verbs (Matsuoka, 1997; Braze, 1997);
• (S)V(O) order follows from possible null arguments, because ASL is a pro-drop
language (Lillo-Martin, 1986)3.

I will consider some of these phenomena, since LSB shows OSV and
SOV word orders besides SVO word order. The following facts give support to
assume that SVO word order is the underlying order in LSB:

(i) All SVO sentences are grammatical;

(1) a) ______________eg
JOHNa aWATCHb bTV (SVO)
John watches TV.
b) JOHN LIKE SOCCER (SVO)
John likes soccer.

Sentences like (1) are very natural in LSB and examples using this
order are always considered grammatical.
The (a) example from (1) has agreement signed on the verb. Also, this
example has in common the special non-manual marker that is associated with the
subject and object of the sentence. We see body shift and eyegaze toward the object
from the subject. Eyegaze is marked in the glosses by the line over the sentence. It is
important to clarify that it is possible to have other non-manual markers associated
with each sentence. For example, head nod can be simultaneously associated with
this example sentence. However, these other non-manuals are not obligatory. For our
purpose, we are considering only eyegaze and body shift with non-plain verbs.
This special non-manual marker that can combine eyegaze and body
shift was first described by Bahan (1996) for ASL agreement non-manual markers.
He observed that this non-manual marker cooccurs with the whole sentence when
agreement is signed, while it is optional with sentences where no agreement is
signed. It is interesting to analyze this special non-manual marker because it seems
to behave similarly in LSB being associated with the asymmetry between plain and
non-plain verbs discussed latter. Agreement together with the non-manual marker
seems to be important for the occurrence of changes in the basic structure. It seems
to have something to do with the information in the verb that licenses different

3
Lillo-Martin (1986, 1991) shows that ASL is a pro-drop language and allows null subjects as well as
null objects. In LSB, we observe the same distribution (Quadros, 1995).

1
derivations. From a minimalist point of view, it might be interpreted as indicating
specific features of the verbs that license object movement to a higher position.
Agreement seems to be one of these, as well as aspect (Fischer, 1975; Liddell, 1980;
Matsuoka, 1997).
This is illustrated in LSB through the next examples. The only
difference that appears is the fact that in (1a), we may have the sentence signed with
no special non-manual marker (eye gaze and body shift), even though this results in a
slight degradation of the sentence’s acceptability. However, with orders other than
SVO, sentences with agreement and no non-manual markers become much worse. In
LSB, it seems that the non-manual marker brings an additional feature that licenses
changes in the word order. This also gives support to the hypothesis that the basic
word order in LSB is SVO.

(ii) OSV and SOV word orders are allowed only when there is some special
characteristic (or feature), such as agreement and the non-manual markers,
including eye gaze or head nod:

(2) a)_eg _____eg ________eg

TVb JOHNa aWATCHb (OSV)
b) __eg ________eg
JOHNa TVb aWATCHb (SOV)

a’) * TVb JOHNa aWATCHb (OSV)

b’) * JOHNa TVb aWATCHb (SOV)

John watches TV.

(3) a) _________hn
SOCCER JOHN LIKE (OSV)
b) __________hn
JOHN SOCCER LIKE (SOV)

a’) *SOCCER JOHN LIKE (OSV)

b') *JOHN SOCCER LIKE (SOV)

John likes soccer.

Comparing the sentences with the non-manual marker with sentences without the
non-manual marker, we conclude that it is the special marker that allows constituents
to be moved in LSB. In (1) we have different behavior that can be used as an
argument in favor of the SVO basic order in LSB. If we assume this conclusion, then
we can explain why (2a’, b') and (3a’, b') are ungrammatical sentences: examples
(2a’, b') and (3a’, b') show us that movement of the object seems to be blocked,
because there is no ‘force to push’ the movement; therefore any movement violates
principles of economy, and the derivation crashes. This is not the case in (2a, b) and
(3a, b), because these sentences have the non-manual marker that seems to have the
feature that allows movement.

2
(iii) There is no doubt about the interpretation of SVO sentences with ‘reversible’
arguments, as there is with OSV and SOV word orders; however, the
sentences containing such reversible arguments are considered unacceptable
with transitive verbs because both arguments are interpreted as subject, and
the object is eliminated (sentences in (4))4. This is not the case with optional
transitive verbs, since the sentences with ‘reversible’ arguments are
ambiguous (sentences in (5));

(4) a) _______ hn
JOHN LIKE MARY (SVO)
John likes Mary.
b) _____________hn
*JOHN MARY LIKE (OSV)

(5) a) ____________hn
LION RABBIT EAT
Possible interpretations:
(i) ∃x & ∃y (x is LION & y is RABBIT), x & y eat something – SV(O)
(ii) ∃x (x is LION & y is RABBIT), x eats y - SOV
b) ________hn
RABBIT LION EAT
Possible interpretations:
(i) ∃x & ∃y (x is RABBIT & y is LION), x & y eat something – SV(O)
(ii) ∃y (y is LION & x is RABBIT), y eats x - OSV

Because these interpretations are possible, we exclude the possibility of a syntactic

restriction on OSV and SOV word orders concerning reversible and non-reversible
arguments. OSV and SOV word orders seem to be generated by movement from a
syntactic point of view. The unacceptability of (4b) seems to be a result of the fact
that LIKE is a transitive verb and, for semantic reasons, the only interpretation
available is JOHN and MARY as subjects of the sentence; these sentences are
excluded because of their lack of object. This is not the case with ‘non-reversible’
arguments in which the semantic interpretation of grammatical relations is plausible
such as in (5).

(iv) The word orders OSV and SOV with reversible arguments in sentences with
plain verbs need an auxiliary:

(6) a) ___________eg _________________eg ___hn

IX<a> JOHNa IX<b> MARYb aAUXb LIKE
John likes Mary (aAUXb)

b) __________eg ___________eg _____eg ___hn

4
Fischer (1975) analyzed analogous examples in ASL to (4b) and (5) in LSB, as ungrammatical,
because the object and the subject are ‘reversible’. JOHN and MARY are reversible because each one
can be the subject or the object of the sentence, while JOHN and SOCCER are not in a sentence like
JOHN LIKE SOCCER. She observed that if the object and the subject are non-reversible, changes in
order can occur.

3
 IX<b> MARYb IX<a> JOHNa aAUXb LIKE
John likes Mary (bAUXa)

The behavior of AUX in LSB is quite similar to the auxiliaries in

Taiwanese Sign Language - TSL - (Smith, 1990). Smith observed that there are three
different auxiliaries in TSL. These appear in fixed positions before the main verbs
with the primary function of conveying subject-object relationships. Moreover, these
auxiliaries in TSL seem to have features-ϕ (like gender, number). The AUX in LSB
is a pure expression of agreement by movement established from one point to
another point (these points are those of the subject and the object of the sentence). It
is not an independent lexical item, but an item that must be signed together with a
plain verb, a verb that lacks agreement information. Through its direction of
movement, the AUX expresses the relation established between the arguments of the
sentence. It seems to compensate for the lack of agreement in these sentences. AUX
is required only when the irregular word order has no way of identifying the subject
and the object in a sentence with a plain verb. As with agreement verbs, the
sentences in (6) need to be pronounced with non-manual marker. Following Lasnik’s
proposal (1995a,b) for auxiliaries in English, in LSB it seems that the auxiliary AUX
is in the head of IP position (or the head of TP, considering the split of IP into
AgrSP, TP and AgrOP).

(v) There is no extraction of the clausal object to a higher position in complex

structures as it is possible in simple ones (examples (7) and (8)).

(7) a) I THINK [IP MARYa aLEAVEloc].

I think that Mary left.

b) *I [IP MARYa aLEAVE] THINK.

(8) a) I WANT [IP MARY WORK BETTER].

I want Mary to work better.

b) *I [IP MARY WORK BETTER] WANT.

The sentences (7) and (8) show us that it is not possible to have SOV word order
when the object is a subordinate clause, unlike what we have seen in simple
sentences such as JOHN [SOCCER] <LIKE>hn and MARY [BOOK]
<aGIVEb>eg/bs. (7) and (8) illustrate that the order SOV has an additional
restriction, providing a strong argument again in favor of SVO as the basic word
order in LSB.

(vi) Temporal and frequency adverbs bring us the fact that there is a VP
constituent in LSB that includes the verb and the object: [VP[V NP]]. This
relationship cannot be interrupted by an adverb, an additional argument to
conceive that SVO is the basic word order in LSB.

(9) a) *JOHN [VP [V BUY [AdvP YESTERDAY [NP CAR]]].

John bought a car yesterday.
b) *I [VP [V DRINK [AdvP SOMETIMES [NP MILK]]]].

4
 Sometimes I drink milk.

I assume that temporal adverbs are right- or left-adjoined to IP (AgrP) and frequency
adverbs are right- or left-adjoined to VP (as shown in (10) and (11) respectively).

(10) a) JOHN BUY CAR YESTERDAY.

b) YESTERDAY JOHN BUY CAR.
c) *JOHN YESTERDAY BUY CAR.
John bought a car yesterday.

(11) a) I DRINK MILK SOMETIMES.

b) I SOMETIMES DRINK MILK.
c)?? SOMETIMES I DRINK MILK.
Sometimes I drink milk.

(vii) Topicalization allows changes in the word order5

A possible explanation to the ‘apparent free word order’ in LSB is related to a
topicalization phenomenon where there is a specific non-manual marker associated
with the topic. Topicalization in LSB is marked with raised brows and the head tilted
slightly back followed by a head nod (if the sentence is affirmative), or by a negative
nod (if the sentence is negative), or still by a question non-manual marker (if the
sentence is an interrogative). The following examples illustrate each of these cases.

(12) ________topic _____________hn

SOCCER, IX<the>JOHN LIKE
As for soccer, John likes it.

(13) _______topic ___________________neg

SOCCER, IX<the>JOHN DESIRE NO
As for soccer, John does not like it.

(14) ________topic _____________y/n

SOCCER, IX<the>JOHN LIKE
As for soccer, does John like it?

(15) ___________topic ___________________________wh

SOCCER-BALL, WHERE IX<the>JOHNa <a>GET
As for soccer ball, where will John get it?

(viii) Focus allows changes in the word orders with plain verbs

In LSB, there is a kind of construction with double elements. In this sentences, there
is an emphasis associated with these double constituents: 'JOHN NEVER GO
NEVER' (John never went to that place). In ASL, Petronio and Lillo-Martin (1997)
restrict the ‘double construction’ “to sentences in which a significant pause does not

5
I do not discuss here in detail the topic, focus and object shift constructions, since this is not the
point for the present proposal. These discussions are detailed in my Ph.D. Dissertation (Quadros,
1999).

5
precede the final ‘double’. When there is a significant pause, the construction has
different syntactic properties” (footnote 12). This note makes a distinction between
focus constructions on the one hand, and constructions such as tag questions and use
of the discourse strategies to confirm some part of the sentence on the other hand.
These researchers pointed out double constructions in ASL to explain why subject
wh-question can appear in final position (‘WHO BUY C-A-R WHO’, from which we
can derive ‘BUY C-A-R WHO’ in ASL). Like ASL, LSB shows double
constructions with modals, quantifiers and verbs. In addition, it is very common to
have a double construction with wh-questions, negation and adverbs. For my
proposal in this paper, I will show only double constructions with verbs:
________hn
(16) IX<1> LOSE BOOK LOSE
I LOST the book
There is a Focus Phrase (FP) functional category projected to allow this kind of
construction in LSB. That is, there is a full projection of focus between CP and IP as
in Portuguese (Kato and Raposo, 1994) and Korean (Kim, 1997). This position is
projected when accented information with interpretation in PF and LF is inserted in
the structure in the head position associated with a strong [+focus] feature. This
strong feature must check and be checked for the feature associated with the stressed
element in the phrase, i.e., IP (or AgrSP) which is also associated with a [+focus]
feature. Therefore, IP will raise (extraposition of IP) to Spec of FP to check its
feature with the head F associated with [+focus]. This is an instance of obligatory
movement in LSB. The [+focus] feature also licenses null elements that meet the
requirement of identity with the head of FP as shown through the following example:
________hn
(17) IX<1> LOSE BOOK LOSE
I LOST the book

Focus constructions in LSB explains the difference between SOV derivations with
plain and non-plain verbs. When we have SOV with a plain verb, in fact, we do not
actually have SOV word order in LSB, but S(V)OV. The final verb is just the
doubled verb that permits the omission of the first verb in the sentence, a null
element licensed by [+focus]. This final verb is marked with a final head nod as
observed in almost every instance of focus. On the other hand, with non-plain-verbs
we explained this word order through object shift. This different analysis for these
two kind of verbs in LSB is explained by the asymmetry discussed in the next
section. There is no instances of SVOV with non-plain-verbs.

(ix) Object shift allows changes in the word orders with non-plain verbs
The word order SOV with a non-plain verb is obligatorily associated with an
agreement non-manual marker. This word order results from short verb movement
and short object movement to AgrO and to Spec of AgrOP, respectively, an option
that is not available for plain verbs (since it projects an IP construction as discussed
later). Considering this analysis, we explain the different behavior of the double
construction between plain and non-plain verbs. The following representation
captures the fact stated above and confirms that there is at least short movement in

6
this language to agreement projections following Holmberg’s (1986)
generalization67.

(18) [AgrSP [AgrS’ [TP MARYa [T’ [AgrIOP JOHNbi [AgrIO’ aGIVEbj [AgrDOP [AgrDO’ BOOKl
[VP ti [V’ tj [DP tl ]]]]]]]]
Mary gave the book to John

(x) Modals also provide evidence to the basic word order in LSB. They can
occupy a position to the right and/or to the left of IP: (i) modals may not
break up the VP constituent; (ii) modals can be doubled as a result of a
double construction ; and (iii) as a consequence of (ii), a modal in the original
position can be null (for OSV and SOV word order: OS(M)VM, S(M)OVM;
for SVO word order: S(M)VOM).

(xi) (S)V(O) order follows from possible null arguments, because LSB is a pro-
drop language (Quadros, 1995)

(19) aGIVEb
(she/he) gives to (her/him)

(xii) Other combinations such as VSO, OVS and VOS are not possible in LSB,
even in the presence of a special marker.

Until now, it seems that SVO is the underlying word order in LSB,
and that OSV and SOV word orders are derived from SVO. In particular, I have seen
that these orders result from syntactic operations motivated by some additional
feature, like agreement or non-manual markers. I have not discussed the specific
operations because this is not the proposal for the present paper.

The following table summarizes the findings:

Table 1 Distribution of word order in LSB

WORD ORDER YES NO WITH
RESTRICTION
SVO X
OSV X
SOV X
VOS X
OVS X
VSO X

6
This conclusion will give support for the asymmetry between plain and non-plain verbs in my
analysis. When we consider the characteristics from Non-Free-Agr languages and plain verbs in the
next step, this point will become clearer.
7
Note that in French, NegP must be higher than AgrO. This difference follows because French verbs
raise overtly to T.

7
 These facts give support for a representation of the phrase structure as
a head initial language. However, there is a clear asymmetry between sentences with
plain verbs and non-plain verbs. I have not made any distinctions so far, but this is a
fact that requires special attention. This is the next step of my work.

2 Asymmetry of two verbal classes

I analyzed the facts related to the two kinds of verbs in LSB, plain
verbs and inflecting verbs (non-plain verbs), which seem to have different behavior
in their phrase structure. This difference is clearly recognized at the morphological
level. However, I analyze possible consequences from this asymmetry for the
syntactic structure. I assume Lasnik’s (1995) proposal on the asymmetry between
have and be and main verbs in English, explained by featural and affixal processes.
Lasnik claims that verbs can be taken from the lexicon either inflected or bare. If
they are introduced fully inflected, Infl is featural, following Chomsky’s proposal in
the minimalist approach. If verbs are introduced bare, their inflectional affixes are
introduced separately under Infl, and the association of affixes with the verb goes
through a PF operation called Merge, restricted by the adjacency requirement. This
approach follows the spirit of Chomsky’s (1957) proposal. Bobaljik’s (1995)
parameter of agreement functional projection is also incorporated into my analysis.
He claims that if affixation and merger under adjacency take place in the derivation,
syntax is blind to them; therefore, there will not be an agreement functional
projection in the phrase structure. On the other hand, if there are feature affixes
associated with the verbs, they must be checked against projected agreement features
at some point of the derivation. A combination of these two views delineates my
proposal of verbal asymmetry observed in sign languages, such as LSB. As a
consequence, a revision of phrase structure in LSB is needed.
Let us look at the facts that show the asymmetry between the verbs
that present inflection and the ones that do not in LSB.

(xiii) Sentences with agreement verbs seem to have more freedom in the structure
than those with plain verbs:

(20) a) ____________eyegaze
MARYb JOHNa aLOOKb (OSV)
John looks Mary.
b) _____________eyegaze
*JOHN MARY LIKE (OSV)
Mary likes John.

(xiv) Non-manual markers are obligatory with agreement verbs and optional with
plain verbs:

(21) a) ____________eyegaze
JOHN LIKE MARY.
John likes Mary.

8
 b) JOHN LIKE MARY.

c) _______eyegaze
*JOHN MARY LIKE

(22) a) ____________eyegaze
JOHN aHELPb MARY.
John helps Mary.

b) ??JOHN aHELPb MARY.

c) _______eyegaze
JOHN MARY aHELPb

(xv) Null arguments with agreement verbs take place in different syntactic
contexts from the ones that are allowed in sentences with plain verbs:

(23) _________eyegaze
TOMORROW 2GIVEa BOOK.
(You) give (her) the book tomorrow.

(24) a) *TOMORROW TALK

a’) TOMORROW IX(you) TALK IX(her)
b) ______eyegaze
*TOMORROW TALK
b’) ______eyegaze
TOMORROW IX(you) TALK IX(her)
(You) talk with (her) tomorrow.

It is important to mention that the examples (24 a, b) can be considered grammatical

in discourse contexts that make the arguments clear. This is expected, since there
seems to have ‘discourse-licensed’ null arguments in LSB (see Lillo-Martin (1986)
for details about this analysis for ASL).

(xvi) There is a different negation distribution between sentences with plain verbs
and agreement verbs:

(25) ____________neg
JOHNa NO aGIVEb BOOK
John does not give the book to (her).

(26) ___________neg
*JOHN NO DESIRE CAR
John does not like the car.

9
 Considering Lasnik’s (1995) proposal for verbal morphology, I
checked the analogy between main verbs in English and plain verbs in LSB. Main
verbs, as affixal verbs in English, present the following characteristics8:

(xvii) Main verbs can not precede negation (*John likes not Mary). The same is
observed in LSB:

(27) ____________neg
*JOHN DESIRE NO CAR9

(xviii) Main verbs can not follow the negation without do-support (*John not likes
Mary and John does not like Mary). This is also observed in LSB. With
negation preceding an agreement verb the sentence is fine (example (28)), but
with a plain verb the sentence will be fine only with an AUX (examples (29)
and (30)):

(28) ____________neg
JOHNa NO aGIVEb BOOK
(29) ___________neg
*JOHN NO DESIRE CAR
(30) _________neg
JOHNa aAUXb pro NO DESIRE.
John does not like (her).

(xix) Main verbs can not precede adverbs that are adjoined to the left of VP (*John
plays always soccer). In LSB, this is also observed:

(31) *JOHN BUY ALWAYS CANDIES10.

John always buys candies.

(xx) Main verbs can be elided from complex sentences through identity with their
bare form (John slept, and Mary will too). This happens in the same way in
LSB:

(32) a) _______eg ____hn _____eg ___neg

aMARY bJOHN aAUXb [LIKE]i, bAUXa NOT [e]i
Mary likes John and (he) does not.
_____eg ___hn ____eg ____hn
b) aMARY bJOHN aAUXb [LIKE]i, bAUXa TOO [e]i

8
English examples from Lasnik (1995).
9
However, this distribution is always ungrammatical, even with agreement verbs, since the VP-unit is
broken. Recall that verbs in LSB do not move overtly. The best examples to show the contrast
between plain and non-plain verbs as well as to show the analogy between plain verbs and main verbs
in English are sentences that present a virtual affix, in which the negation will block adjacency
between T and V, and therefore the sentence will crash (see below this analysis).
10
Again, this is not good evidence to show differences between plain and non-plain verbs. I have seen
that adverb distribution is the same for both plain and agreement verbs. Temporal adverbs are right-
or left-adjoined to IP, and frequency adverbs are right- or left-adjoined to VP. Therefore, the position
between the verb and the object will always be ungrammatical.

10
 Mary likes John and he does also.

_____eg ___hn ____eg ___hn

c) *aMARY bJOHN aAUXb [LIKE]i, JOHN TOO [e]I
Mary likes John and he also.

In LSB, ellipsis without an auxiliary is not allowed:

__hn
(33) a) *JOHN LIKE MARY, MARY TOO.
John likes Mary and Mary also
____eg ___hn
b) bJOHN LIKE aMARY, aAUXb TOO
John likes Mary and (she) does also.

I have seen so far that plain verbs behave in the same way as main
verbs in English. That is, following Lasnik’s proposal, these verbs are affixal and
must merge with their respective affixes as a consequence of a phonological
requirement under adjacency. I also observed in (18) that the affixal approach can
not explain the lexical negation that takes place in sentences with inflecting verbs,
which suggests the existence of an asymmetry. If I am on the right track, I must now
explain how the structure works with inflecting verbs. Let us see if there is an
analogy between inflecting verbs and the featural approach for verbal morphology
presented by Lasnik (1995).
The featural approach observed with auxiliaries in English and main
verbs in French, as well as in Swedish, presents the following characteristics:

(xxi) These verbs precede negation when they have “strong” features and follow
negation when they have “weak” ones (the first case is observed in English
and French: John has no idea about the game and Jean (n’)aime pas Marie;
the second case is observed in Swedish, …, om hon inte ofta har sett
honnom)11. This also is observed in LSB with agreement verbs (examples
(34) and (35)):

(34) ______________________neg/eg
JOHNa NO aGIVEb BOOK MARYb.
John did not give the book to Mary.
(35) _______________neg/eg
JOHNa NO bCARRY-BOXc.
John did not carry the box (from here to there).

(xxii) Verbs as have and be in English do not need do-support (John has no idea
about the game results). Considering AUX in LSB as something similar to
do-support the same distribution is observed (example (36)):

11
Glosses to French and Swedish examples respectively:
John likes not Mary
…, whether she not often has seen him

11
(36) a) ________________eyegaze
*JOHNa MARYb aAUXb aMEETb

b) ___eyegaze __________eyegaze
* MARYb JOHNa aAUXb aCALLb

(xxiii) Verbs as in French precede adverbs adjoined to the left of VP when they have
strong features to be checked (Jean embrasse souvent Marie)12.

(37) *IX<the> JOHN BUY YESTERDAY CAR

(38) * IX<the> JOHN <a>MEET<b> ALREADY MARYb

(xxiv) Verbs as have and be in English can not be elided from complex sentences
because there is no identity between the verb in the main clause and the
elided verb (*John is here, and Mary will too). On the other hand, these verbs
can be elided when there is something present in the sentence that guarantees
the identity between them and their antecedent (John will be here, and Mary
will be too). These facts can also be observed in LSB as shown in the
examples (39) and (40):

(39) a) __________________hn ________hn

_________eyegaze
MARY CAN [aGIVEb BOOK]i, I TOO CAN [e]i
Mary can give the book (to someone), I can also

b) __________________hn _________hn
_________eyegaze
MARY WANT [aGIVEb BOOK]i, I TOO WANT [e]i
Mary wants to give (to someone) the book, I want too

(40) a) __________hn _______hn

_________eyegaze
*MARY aGIVEb BOOK, I TOO WILL

__________hn _______hn
_________eyegaze
b) *MARY aGIVEb BOOK, I TOO WANT.

In LSB, a modal guarantees the identity between the elided verb and
its antecedent.
I have enough evidence to assume that plain verbs have a virtual affix
that requires adjacency and that inflecting verbs (agreement and spatial verbs) are
inserted fully-inflected from the lexicon.

12
Glosses to French example:
John kisses often Mary

12
 The proposal is based on the featural and affixal approaches
developed by Lasnik (1995), and also on the morphological parameter proposed by
Bobaljik (1995). I relate these analyses and present the phrase structures in LSB that
account for the asymmetric behavior of the verbs. This approach is compatible with
the minimalist program, since phrase structure is projected as a consequence of the
interface motivations for meaning and sound/sign. Moreover, I state two general
assumptions. First, all main sentences in LSB are tensed, as assumed by Aarons,
Bahan, Kegl and Neidle – ABKN – (1995) for ASL, and therefore there is a
functional projection for tense. There are some lexical items that are generated in this
position, such as tense markers13. This assumption allows the subject to be base-
generated in Spec of T, then it will observe the EPP, since the subject position will
be filled out, the second assumption made here (following Bobaljik, 1995). Bobaljik
shows that there are enough reasons to assume Koizumi’s (1993) and Chomsky’s
(1995) proposals that the subject must be generated outside of VP. He also observed,
however, that there is no reason to stipulate the split of VP, as done by the original
approaches.
The basic idea put forward by Bobaljik (1995) is that some languages
project agreement and tense, because there are agreement and tense features that
must be checked during the derivation (Free-Agreement languages). Other languages
solely project inflection for checking related to agreement or tense (Non-Free-
Agreement languages). I suggest that there are also languages which can set both
possibilities. This is caused by the existence of features related to both kinds of
information in different verbal classes. For instance, main verbs in English are
affixal, and there is no Agr projection, whereas French verbs, have and be in English,
and inflected verbs in LSB do present AgrP. The grammar has the two options
available and one and/or the other will be set considering the features inserted with
the lexical items during Numeration.
One of the characteristics observed by Bobaljik for languages that do
project agreement is that in these languages there is a target in the phrase structure
for possible object-shift when the verb moves to a higher position overtly (observing
Holmberg’s generalization, 1986).
LSB, as well as ASL, allows object shift. Independent motivations for
this movement have not been analyzed for LSB yet. As for ASL, Matsuoka (1998),
Braze (ongoing) and Chen (ongoing) have provided some explanations. I give some
examples to show that object-shift is a common syntactic operation in these
languages.

(41) ASL
a) [AgrSP [AgrS’ [TP WOMAN [T’ [AgrOP PIE i [AgrO’ PUT-IN-OVENj [VP ti [V’ tj ]]]]]]]]
The woman put the pie in the oven

(42) LSB14
a) _________________eyegaze ______hn

13
This assumption must be investigated more deeply, since there is no morphological marker for
tense attached to the verb.
14
In these LSB examples, I assume the V raises to AgrO. When the V moves, the object may move to
Spec of AgrOP. This explains the word order SVO with agreement verbs including handling verbs.

13
[AgrSP [AgrS’ [TP MARYa [T’ [AgrOP JOHNb i [AgrO’ aGIVEb j [VP ti [V’ tj [DP BOOK]]]]]]]]]
Mary given to John the book

b) ________________eyegaze
[AgrSP [AgrS’ [TP MARY [T’ [AgrOP PIE i [AgrO’ PUT-IN-OVENj [VP ti [V’ tj ]]]]]]]]
Mary put the pie in the oven

Free-Agr languages must present a target position for object

movement related to the agreement projection. This is what I observe, since, on the
other hand, with plain verbs (that I assume generate a non-Free Agreement
representation), object shift is not allowed in the same contexts, even in the presence
of a non-manual marker. Object-shift with plain verbs is allowed when there is an
Aspect Projection or other projections, but not in cases analogous to object-shift with
inflecting verbs (agreement and spatial verbs). As I have said before, these sentences
must have non-manual markers related to agreement, as claimed by Bahan (1996) for
ASL. This is said to provide evidence for agreement projection, because there are V-
features related to this projection that need to be checked.

Another characteristic pointed out by Bobaljik for Free-Agr languages

is that agreement and tense can co-occur in these languages, unlike Non-Free
Agreement languages in which these elements are in complementary distribution.
The data that follow show LSB co-occurrence of Agr and T with non-plain verbs in
(43b) and the ungrammaticality of this co-occurrence with plain verbs. Also, I add
some examples from Brazilian Portuguese, Icelandic and English.

(43) LSB
a) _______eyeg/body shift
* JOHN MARY [I [FUTURE-TNS/aAUXb] ] CALL
John will call Mary.15
b) _____eyeg/body shift
JOHN [T FUTURE-TNS [Agr [aGIVEb]]] BOOK
John will give the book (to her).

(44) Brazilian Portuguese

a) Ela/ele comeu [3rd singular person, past tense]
She/he ate
b) Nós comemos [1st plural person, present tense]
We ate

(45) Icelandic (Bobaljik, 1995:264)

kasta-oi-r (past tense, second singular agreement)
throw-past-2s

5 The sentence with only FUTURE present would be grammatical:

_______eg
* JOHN MARY [I [FUTURE-TNS] ] CALL
It seems that the position can be occupied by only one element realized or by the virtual affix.

14
(46) English
a) She/he likes (agreement)
b) She/he liked (past tense)
c) *She/he likesed (agreement, past tense)

The final characteristic that I want to discuss is presented by Bobaljik

for Non-Free Agr languages. He shows that in these languages, verbs surface in V,
since they are adjacent to the inflection projection. Although this is hard to see in
LSB, I assume the existence of a virtual affix that must be adjacent to V. The
examples that follow present the structure assumed for plain verbs.

(47) English (Bobaljik, 1995:272, example number (7))

[IP [D(P) Sam [I’ [I [past] [VP [V eat [D(P) lunch]]]]]]]
Sam ate lunch
(48) LSB
[IP [D(P) JOHN [I’ [I [∅] [VP [V EAT [D(P) RICE]]]]]]]
John eats rice.

3 Phrase Structures of LSB

Based on the evidence discussed up to this point, I assume that there

are two possible representations that languages can access: the functional projection
of agreement and tense, as proposed in the spirit of Pollock (1989), and the
functional projection of only one inflection category IP, in the spirit of Chomsky
(1995). Instead of putting these two analyses in opposition, I put them together to
capture different behavior of languages. Actually, this is a natural result from the
minimalist point of view, since projections result from features that have some
independent motivation from the interfaces. When there is no motivation, there is no
reason to have a functional projection. Also, I assume both possibilities for LSB,
since there are plain verbs that activate an affixal approach, and inflecting verbs that
activate a featural approach (as is the case with English (Lasnik, 1995)). The relation
that I establish between these two analyses is in accord with the minimalist
investigation that is based on economy principles.

Figure 1 – Phrase Structure Projected with Plain Verbs

IP

Spec I’
Position in which
The subject is generated

I VP

15
 Position occupied by
affixes, modals and tense markers

V DP
Position occupied by
Position occupied by the verb the object

Figure 2 – Phrase Structure Projected wit Inflected Verbs

AGRsP

Spec AGRs’
Position
for the subject

AGRs TP
Position in which
agreement with
the subject will
be checked
Spec T’

T AGRoP
Position that can be
occupied by modals,
auxiliary and tense
Spec AGRo’
Position occupied
by the object when it raises

AGRo VP
Position in which agreement with
the object will be checked

V DP
Position occupied by the verb
Position occupied by the
object when it doesn't move

4 References

AARONS, D. (1994) Aspects of the syntax of American Sign Language. Ph.D. Dissertation,
Boston University, Boston, MA.
AARONS, D.; BAHAN, B.; KEGL, J.; NEIDLE, C. (1995) Lexical Tense Markers in
American Sign Language. In: Language, Gesture and Space. Emmorey,K. and Reilly, J.
S. (eds.). Hillside, NJ: Lawrence Erlbaum Assoc.
BAHAN, B. (1996) Non-manual realization of agreement in American Sign Language.
Ph.D. Dissertation, Boston University, Boston, MA.
BELLETTI, (1990) Generalized Verb Movement: Aspects of Verb Syntax. Turin: Rosenberg
and Sellier.
BOBALJIK, J. D. (1995) The syntax of verbal inflection. Ph.D. Dissertation. MIT. MIT
Working Papers in Linguistics.

16
BRAZE, D. (1997) Objects, Adverbs and Aspect in ASL. In Is the Logic Clear? Papers in
Honor of Howard Lasnik. Kim, J-S. and Stjepanovic (eds.) University of Connecticut.
Working Papers in Linguistics 8. 21-54.
CHEN, D. (1998) Investigation of word order acquisition in early ASL. Unpublished
manuscript. University of Connecticut.
CHOMSKY, N. (1957) Syntactic Structures. The Hague: Mounton.
CHOMSKY, N. (1993) A Minimalist Program for Linguistic Theory. In K. Hale & S, J.
Keyser (eds.) The View from building 20: essays in linguistics in honor of Sylvain
Bromberger. MIT. Cambridge, Massachusetts. 1-52.
CHOMSKY, N. (1995) The Minimalist Program. MIT Press.
CHOMSKY, N. & LASNIK, H. (1993) Principles and Parameters Theory. Syntax: An
International Handbook of Contemporary Research. Walter de Gruyter. Berlin.
CHOMSKY, N. (1986) Knowledge of Language. Praeger. New York.
FISHER, S. (1973) Verb Inflections in American Sign Language and Their Acquisition by
the Deaf Child. Paper presented at the Winter Meeting of the Linguistic Society of
America. [s.l.,s.n.].
FISHER, S. & GOUGH, B. (1978) Verbs in American Sign Language. In: SLS 18. [s.l.,s.n.]
p. 17-48.
FELIPE, T. (1989) A estrutura frasal na LSCB. In Anais do IV Encontro Nacional da
ANPOLL. Recife.
FERREIRA-BRITO, L. (1995) Por uma gramática das línguas de sinais. Tempo Brasileiro.
UFRJ. Rio de Janeiro.
HOLMBERG, A. (1986) Word order and Syntactic Features in the Scandinavian Languages
and English. Ph.D. Dissertation. University of Stockholm.
KOIZUMI, M. (1993) Phrase Structure in Minimalist Syntax. Ph.D. Dissertation. MIT.
[Distributed by MIT Working Papers in Linguistics].
LASNIK, H. (1995) Verbal Morphology: Syntactic Structures Meets the Minimalist
Program. In Evolution and Revolution in Linguistic Theory: Essays in Honor of Carlos
Otero. Georgetown University Press.
LIDDELL, S. (1980) American Sign Language Syntax. Mouton Publisher. The Hague.
LILLO-MARTIN, D. C. (1986) Parameter setting: evidence from use, acquisition, and
breakdown in American Sign Language. Doctoral Dissertation. University of
California, San Diego. University Microfilms International. Ann Arbor. Michigan.
LILLO-MARTIN, D. C. (1991) Universal Grammar and American Sign Language. Kluwer
Academic Publishers. Dordrecht. Boston. London.
MATSUOKA, K. (1997) Verb Raising in American Sign Language. In Lingua. 103:127-
149.
NUNES, J. (1999) Linearization of Chains and Phonetic realization of Chain Links. In
Epstein and Horstein (eds.): Working Minimalism. MIT Press. Crambridge. Mass.
PADDEN, C. (1983) Interaction of Morphology and Syntax in ASL. Doctoral Dissertation.
University of California, San Diego.
PADDEN, C. (1990) The Relation Between Space and Grammar in ASL Verb Morphology.
In Sign Language Research - Theorical Issues. Gallaudet University Press. Washington.
118-132.
PETRONIO, K. (1993) Clause Structure in ASL. Ph.D. Dissertation. University of
Washington.

17
PETRONIO, K. & LILLO-MARTIN, D. (1996) WH Movement and the Spec of CP:
Evidence from American Sign Language. In Languages.
POLLOCK, J.Y. Verb Movement, Universal Grammar, and the Structure of IP. In Linguistic
Inquiry. Volume 20. Number 3. Summer. 1989. 365-424.
QUADROS, R. M. de. (1995) As categorias vazias pronominais:uma análise alternativa
com base na língua de sinais brasileira e reflexos no processo de aquisição. Dissertação
de Mestrado. PUCRS. Porto Alegre.
QUADROS, R. M. de Phrase Structure of Brazilian Sign Language. Tese de Doutorado.
PUC/RS. Porto Alegre. 1999.
SMITH, W. H. Evidence for auxiliaries in Taiwan Sign Language. In Theoretical Issues in
Sign Language Research. Suzan D. Fischer and Patricia Siple (ed.) Volume 1:
Linguistics. The University of Chicago Press. Chicago. 1990.

i
The present work is part of the requirements for my PhD degree at Pontíficia Universidade Católica do RS/Brasil (Quadros,
1999). It had been elaborated at the University of Connecticut under Dr. Diane Lillo-Martin’s supervision to whom I am very
thankful. Also, I am grateful to my supervisor in Brazil, Dr. Jorge Campos. I am very thankful to Deborah Chen for always
being ready to help me with my English and above all for being my friend. I cannot miss to mention my colleagues and friends
Gaurav Mathur, Ingrid Finger and Stephanie Berk with whom I enjoy to exchange ideas about my work, and professors Howard
Lasnik, Zeijko Boskovic, and Diane Lillo-Martin, with whom I had the opportunity to learn a lot from during my stay at
University of Connecticut. Needless to say, I am fully responsible for all errors. Suggestions and comments are welcome. Please
e-mail me at [email protected]

18