Work Papers of the Summer

Institute of Linguistics, University

of North Dakota Session

Volume 31 Article 4

1987

From meaning to text: Semantic representation in the Meaning-

Text linguistic theory and a new type of monolingual dictionary
Igor A. Mel'čuk
University of Montreal

Follow this and additional works at: https://commons.und.edu/sil-work-papers

Part of the Linguistics Commons

Recommended Citation
Mel'čuk, Igor A. (1987) "From meaning to text: Semantic representation in the Meaning-Text linguistic
theory and a new type of monolingual dictionary," Work Papers of the Summer Institute of Linguistics,
University of North Dakota Session: Vol. 31 , Article 4.
DOI: 10.31356/silwp.vol31.04
Available at: https://commons.und.edu/sil-work-papers/vol31/iss1/4

This Article is brought to you for free and open access by UND Scholarly Commons. It has been accepted for
inclusion in Work Papers of the Summer Institute of Linguistics, University of North Dakota Session by an
authorized editor of UND Scholarly Commons. For more information, please contact
[email protected].
 FltON MEANING TO TEXT:
Semantic Representation in the Meaning-Text Linguistic
Theory and a Rew Type of·Monolingual Dictionary

Igor A. Mel'~uk
University of Montreal

1 Introduction
2 Outline of the Meaning-Text Theory
2.1 Main postulates and general characteristics
2.2 Semantic representation in the Meaning-Text Theory
3 The Explanatory Combinatorial Dictionary as a crucial
component of a Meaning-Text model
3.1 General characteristics
3.2 Structure of an ECD entry
3.2.1 Semantic zone
3.2.2 Syntactic zone
3.2.3 Lexical cooccurrence zone
Appendix A: List of Lexical Functions
Appendix B: three sample dictionary entries

1 Introduction
We use language primarily to express meanings. And to communicate, of
course, but "sprachlich" 1 communication boils down to an exchange of
meanings expressed in and extracted from "sprachlich" messages). This
is an obvious truth or, rather, a commonplace fact that no one in the
linguistic profession has ever tried to call into question. All
linguistic schools include in their respective credos strong statements
concerning the crucial role of meaning and its expression in human
language and, consequently, in linguistics. Thus half a century ago
Leonard Bloomfield wrote: "In human speech, different sounds have
different meanings. To study this coordination of certain sounds with
certain meanings is to study language" (1933: 27). More recently, the
same tenet has been vigorously restated, by (among others) Noam Chomsky:
"A generative grammar ••• is a system of rules that relate signals to
semantic interpretations of these signals" (1966: 12). "The grammar of a
language ••• establishes a certain relation between sound and meaning"
(1968: 116); etc. For the last ten years general interest in meaning
and in the linguistic discipline dealing with meaning, i.e. semantics,
has witnessed considerable growth.

Nevertheless, semantics today still remains an underdeveloped

field. Meaning, which is de jure recognized as an important element of
language, still is de facto, if not ignored, then frequently avoided or
at least not dealt with as directly and systematically as it should be.
Let us mention, for instance, Testen et al. 1984 - a volume called

SIL-UND Workpapers 1987

 74

Lexical Semantics and representing one of the most recent publications

in the field. The volume contains many interesting papers, and is full
of astute remarks and insightful discussions of different semantic
problems, but, curiously enough, in spite of its title, it does not
offer descriptions, or even attempts at descriptions, of actual lexical
meanings. This is quite typical of the approach of modern linguistics
to semantics: First, most linguists concentrate on abstract questions
having a strong logical flavor {quantification, logical connectives,
referentiality, pragmatic functions of discourse, etc.), while
consistently shunning the description of genuine "sprachlich" meanings,
lexical or grammatical. Second, to the best of my knowledge, there is,
in the mainstream of modern linguistics, no integrated theory of
language that would consistently proceed from meaning. There is, at
least, one obvious exception: the work of A. Wierzbicka proposes,
develops and substantiates such a theory {cf. below). To discuss the
reasons for this state of affairs would be out of place; suffice it to
state that it justifies our expounding here a different approach to
natural language, namely, the Meaning-Text Theory.

The Meaning-Text Theory {MTT) is not simply another of numerous

linguistic theories, which proliferate to such an extent nowadays that
there are almost as many theories as there are practicing linguists.
The MTT is truly different in that it puts quite a new emphasis on
"sprachlich" meaning, taking it as a cornerstone of language description
as a whole; accordingly, semantics is declared and, most importantly,
actually treated as the central linguistic discipline {which, among
other things, underlies syntax and morphology).

The Meaning-Text Theory is by no means a novelty. It was launched

in 1965, in Moscow, by A. Zholkovsky {now at the University of Southern
California) and the present author: folkovskij and Mel'cuk 1965. Later,
we were joined by Ju. D. Apresjan. Since that time, the MTT has been
developing over the past 20 years {see Mel'cuk 1981 and Nakhimovsky
1983). However, the MTT still lacks an easily available and
sufficiently detailed presentation in English, a gap which this paper
tries to fill, at least in part, by outlining the two following topics:

{i) General structure of the MTT (with special attention

to its semantic chapter) {Sect. 2).

{ii) A new type of monolingual dictionary, or lexicon,

which according to the MTT constitutes the central
part of the semantic component of any full-fl~rlQ~rl
"sprachlich" model {Sect. 3).

The character of our exposition forces us to ban references,

including the most relevant ones. {Only a few are quoted, where they
were judged µecessary in order to identify the background of the MTT.)

SIL-UND Workpapers 1987

 75

2 Outline of the Meanf.ng-Text Theory

2.1 Hain Postulates and General Characteristics

The basic idea underlying the Meaning-Text Theory is as follows.

Any speech act is believed to consist of three major components:

(a) A CONTENT to be communicated by .. sprachlich"

signals; it will be called meanf.ng.

(b) A complex SIGNAL, or SIGNALS (observable

physical phenomena), used to communicate
the content in question; this signal will
be called text.

(c) A CORRESPONDENCE between the meaning and the

text; it will be called mapping.

We will shortly make the concepts of meaning and text more precise, but
before we do so we need to emphasize the following three important
points concerning our tripartite division of a speech act.

First, we presuppose the discrete character of both meanings and

texts. That is, we postulate that we can distinguish meanings and count
them; likewise for texts. Meanings and, mutatis mutandis, texts are for
us formal objects specifiable by a logical device (for instance, by a
generative grammar). This is more or less accepted with respect to
texts but seems to be rather unusual as applied to meanings.

Second, the set of possible speech acts is open: there is an

infinite number of meanings and texts in every natural language.
However, the mapping between meanings and texts is finite, i.e., it can
be fully represented ·by a finite number of correspondence rules (even
though this number is very large). This is the case since an arbitrary
meaning, as well as an arbitrary text, can be, generally speaking,
broken down into simpler meanings, viz. simpler texts, so that ultimate
correspondences can be established between elementary meanings and
elementary texts, the number of these entities being finite.

Third, there is no one-to-one correspondence between meanings and

texts: one meaning can be expressed by quite a few texts (synonymy), and
a text can express several meanings (homonymy/polysemy -- or ambiguity).

The above-stated may be symbolized as in (1), which is the main

postulate of the Meaning-Text Theory:

(1) Postu1ate 1
language
{MEANING1 } <===>

SIL-UND Workpapers 1987

 76

In the MTT, then, a natural language is viewed as a many-to-many mapping

between an infinite set of meanings and an infinite set of texts.

Let it be emphasized and re-emphasized that the words meaning and

text are to be taken, within the present framework, as technical terms,
free of the many connotations they have in every-day English or in other
terminological systems. Thus, as far as meaning is concerned, three
important properties of our approach must be constantly borne in mind:

(i) We deal only with strictly "sprachlich" meaning, i.e. with

the information which can be extracted from an utterance solely on the
basis of purely "sprachlich" knowledge, without any reference to the
context and/or extralinguistic environment (roughly, without any
encyclopedic knowledge about the universe). Our "meaning" is the
shallowest, absolutely literal meaning of utterances. Suppose a girl
on a date says to her shy partner, "Oh, I'm coldt", in order to incite
him to embrace her; for the MTT, the meaning of her utterance is still
'I'm cold' and by no means 'Go ahead and embrace met' When I tell you
that John has applied for a scholarship and you react by asking me,
"What do you mean?", you are after quite a different meaning from what
the MTT tries to describe. A meaning in the MTT is simply the invariant
of a set of all utterances that are paraphrases of each other. Or, to
put it differently, meaning is the canonical invariant of synonymous
utterances (see below). Therefore, such things as truth, incoherence or
absurdity are of no relevance to us: these properties concern the
relationship between "sprachlich" meaning and something else(= the
universe), while the task of the MTT is precisely to avoid discussing
anything that is beyond language.

(ii) Meaning is taken to be directly accessible to speakers --

much like the sounds of their language. This is not to be construed as
implying that any speaker is able to correctly analyze or explain any
given meaning; such is by no means the case. But every speaker knows
(though perhaps subconsciously) what he wants to say, even if he does
not understand the deeper meaning of his words (this last circumstance
should not bother us). Therefore meaning is open to direct
introspection; for a trained linguist describing his mother tongue,
meaning belongs to the data.

(iii) Although we talk of meaning, the MTT deals, in actual fact,

with formal representations of meanings, called Sem(antic)
R(epresentations). Meanings do have an objective existence of their own
-- as certain neurophysiological events in the speaker's brain. But
once again, this is of no relevance for us. When we write {HEANIRGi},
we mean {SeaRi}• These representations, invented and developed by the
linguist, are our only semantic reality within the limits of the MTT, of
course.

Texts possess three similar properties as well.

SIL-UND Workpapers 1987

 77

(i') The term text should by no means be understood as 'coherent

text' or 'discourse', i.e. in the sense of the so-called text grammar.
Our text is any linguistically1 valid segment of speech; it can be the
signifier (Lat. signans, Fr. signifiant) of a morph, a lexeme, a
phrase, etc., including whole volumes.

(ii') Texts are directly accessible to the speaker, in the sense

that he is able to tell a well-formed text in his language from an
ill-formed or questionable one.

(iii') We deal with representations of texts only, not with actual

texts (which consist of physical phenomena). Saying {'TKXTj}, we mean
{Phon(etic) R(epresentation)j}, and only PhonR's are the object of our
study.

Following the statements (i)-(iii) and (i')-(iii'), we can rewrite

(1) as a more precise formulation, namely (2):

(2) language
{SemRi} <===> {PhonRj} IO< i,j < oo

What has been said so far allows us to draw an important

conclusion. Given that we have to deal with representations of meanings
and texts, our first and foremost task should be to develop formal
languages to represent them. In slightly different words, the first
thing we need in order to talk about natural language in a scientific
way, so that all our statements are explicit and unambiguous, is a
semantic transcription and a phonetic/phonemic transcription. Until we
can formally present meanings and texts of a natural language we cannot,
strictly speaking, discuss it.

Now, phonetic/phonemic transcriptions are readily available; where

they are not, methods for developing them are familiar, and necessary
skills and know-how are at hand. With meanings, however, the situation
is different -- quite simply, it is disastrous. No serious efforts have
been made in mainstream linguistics to develop semantic transcriptions;
not even one artificial semantic (meta)language for one natural language
has been proposed, as far as we know, in major linguistic schools.
There are, to be sure, several developments that seem to contradict the
last statement: take, for example, the insightful and promising work of
A. Wierzbicka (1972 and 1980, to name only two of her many pioneering
studies). But in spite of its interest and availability, Wierzbicka's
research (as well as a number of similar attempts) remains little-known
and, in a sense, marginal. Modern linguistics, as a whole, has not yet
made semantic language its main concern. In sharp contrast, the MTT
lays heavy emphasis on meaning representation, i.e. on SemR and,
therefore, on a semantic language. In our opinion, such a language is a
sine qua non of today's science of language. Later on, we will discuss

SIL-UND Workpapers 1987

 78

the semantic language, i.e. the language of SemR used in the MTT, in
more detail.

Now let us make the next important point. While meanings and texts
(of a language) are immediately accessible to the linguist, rules that
correlate them (e.g. the mapping"(===>" in (1) and (2)) are not. Thus
we linguists face the classical "black box" situation: we are in front
of a running device (natural language) whose inputs and outputs are
perceptible and controllable but whose inner structure, the circuits
linking inputs with outputs, is totally hidden from us. "Sprachlich"
rules mapping meanings onto texts and vice versa are
eiectronically/chemically encoded in the speaker's brain, yet we cannot
open skulls to find out about them. The only option we have can be
expressed by the following:

(3) Postulate 2

A natural language, viewed as a many-to-many mapping

between two infinite sets (that of meanings and that of
texts), can be described only by a Cybernetical, or
Functional. Moclel.

A functional model of a device (phenomenon, event) Xis a logical

device, or a finite set of formal rules, which simulates as closely as
possible the behavior of x. In our case, Xis human language, and
therefore our model should simulate the "sprachlich" behavior of humans
in other words, it should establish correspondences between meanings
and texts. It is such models that we call Meaning-Text Models (MTM).
Strictly speaking, an MTM for language£ is a system of formal rules
which associates with a given SemR of£, all the PhonR's that, in the
judgment· of £'s speakers, can carry the corresponding meaning, and vice
versa, which associates with a given PhonR of£ all the possible SemR's
-- all meanings that this text can have. Thus our second task should be
to develop Meaning-Text models of languages. (Of course, this task is
closely related to the first one, that of developing formal languages
for linguistic representations in the first place -- for the SemR.) We
can say that the Meaning-Text Theory is a theory for building MTM's.

Note that a complete Meaning-Text model should be a "dynamic"

device in the sense that for a given SemR, it should actually produce
the set of corresponding PhonR's. However, it seems natural to
distinguish, within an MTM, two submodels: 1) the system of purely
"sprachlich" rules which specify the correspondence between meanings and
texts, and 2) the system of procedural rules which specify the process
leading from a SemR to the PhonR's (or vice versa) -- based on
"sprachlich" rules, i.e. on the first system. The second system is by
no means specific to linguistics. It includes rules that, using factual
knowledge supplied by the first system, compute the best way (or one of
the best possible ways) to construct the necessary PhonR for a given
SemR (and to construct the necessary SemR for a given PhonR); the same
type of procedural rules are needed whenever stored knowledge has to be

SIL-UND Workpapers 1987

 79

used for the solution of any problem. Therefore, in all our

deliberations, we will not concern ourselves with the second system of
rules. Thus an MTM is considered and presented only in its strictly
"sprachlich"/linguistic "static" aspects. This means, among other
things, that it appears as a Static system of correspondences between
elementary meanings and elementary texts, to the complete exclusion of
all elements of procedure. Two important remarks are needed in this
regard.

First, an MTM is by no means a generative or, for that matter,

transformational device: it is equative (or, if you like, translative).
It does not seek to generate (enumerate, specify) the set of all (and
only) grammatical or meaningful texts, rior does it transform certain
"sprachlich" entities into other entities. It simply states which
SemR's match which PhonR's in accordance with native speakers'
intuition.

Second, an MTM does not admit linear order as a formal means among
its tools. To be sure, linear order is an important expressive means of
natural languages and, as such, it has to be accounted for in the rules
and the representations used by the model. However, "sprachlich" order
should not be reflected simply by ordering the elements of the MTM.
Thus MTM rules should be unordered and all relevant information about
language should be explicitly represented by symbols and configurations
thereof. Likewise, linear order is never admitted to express any
"sprachlich" relations in the semantic or syntactic representations of
utterances. Once again, all relations should be denoted by appropriate
symbols. (Cf. Sanders 1972.)

We have mentioned above that the correspondence between meanings

and texts is many-to-many. For a fairly complex meaning, hundreds of
thousands of (nearly) synonymous texts can be constructed. Thus,
example (4)

(4) The Food and Drug Administration has seriously

cautioned expectant mothers to avoid one of
life's simple pleasures: a cup of coffee.

has more than 200,000 paraphrases (Mel'~uk 1981: 31-32). Similarly, a

text may have several meanings (i.e. have distinct readings). The
extremely involved character of the Meaning-Text correspondence makes it
practically impossible to write its rules directly from meanings to
texts. It turns out that for a better perspicuity and surveyability,
the correspondence between meanings and texts must be broken down into
simpler components. This entails the following:

(5) Postulate 3

To describe the correspondence {SemRi}<===>{PhonRj},

TWO INTERMEDIATE LEVELS of utterance representation
are introduced: Synt(actic) R(epresentation) and

SIL-UND Workpapers 1987

 80

Morph(ological) R(epresentation).

Therefore, (2) can now be rewritten, in a more developed form, as (6):

(6)
·{SemRi}<===){S~tRk}<===){MorphR1 }<===){PhonRj}
~~ ...... "V _/
Semantics Syntax Morphology
+
Phonology

The SyntR and MorphR are centered, respectively, around the sentence and
the word. Because of the obvious space constraints, we will not touch
here upon such important properties of the SyntR and the MorphR as their
division into deep and surface sublevels, the formalisms used, etc. We
will limit ourselves to insisting on the stratificational, or
multistratal, character of the proposed model (cf. Lamb 1966 and Sgall
1967). A Meaning-Text "sprachlich"/linguistic model consists, as shown
in (6), of three major components, which form, so to speak,·a production
chain:

Semantics establishes the correspondence between the

infinite set of SemR's and the infinite set of SyntR's;

Syntax establishes the correspondence between the SyntR's

and the MorphR's;

Morpho1ogy + Phonology establish the correspondence between

the MorphR's and the PhonR's.

In this presentation, we will be concerned exclusively with the semantic

component of an MTM.

Note that in principle the correspondence"<===>" is bidirectional:

an MTM should be able to go both from meanings to texts (= speech
production) and from texts to meanings (= speech understanding).
Logically, both directions are, of course, equivalent, and both
correspond to actually observable processes. Linguistically, however,
they are not equivalent; natural language gives a more prominent place
to the speaker than to the addressee:

(i) The speaker can speak (or write) even without an

explicit addressee (to himself, to God. to posterity),
while an addressee is inconceivable without an explicit
speaker.

(ii) Reference to, or value judgments by, the speaker (rather

than by the addressee) are included in the meaning of a
host of "sprachlich" units: shifters (like here= 'where
the speaker is'), speaker-oriented verbs like come vs.

SIL-UND Workpapers 1987

 81

go, performative verbs, all affective and derogatory

terms, and many others, without mentioning the cases of
empathy {the speaker identifying himself with a
character presented in his utterance), prepositional
choices {X in front of Y as opposed to Y behind X, as a
function of where the speaker mentally places himself),
etc.

{iii) Most important, the speaker knows perfectly well what he

is going to say; his only problem, when he starts from a
given meaning, is to find an appropriate "sprachlich"
form {= text) to convey this meaning. His activity qua
speaker is a purely "sprachlich" one. The addressee, on
the other hand, has no previous knowledge of the meaning
he is supposed to extract from the utterance. He has to
use his logic, his general abilities to infer and
guess, his knowledge of the world, etc.~ in addition
to his "sprachlich" skills proper. The addressee's
activity qua addressee is not purely "sprachlich": it
includes, to a considerable degree, many
non-"sprachlich" operations, with the result that
"sprachlich" operations occupy in it a rather modest
place.2

Consequently, the viewpoint of the speaker is by far the more

advantageous for linguists. Describing language as a system for
EXPRESSING meanings and not the other way around, we can concentrate on
genuine "sprachlich" phenomena, thus avoiding involvement with problems
of rule ordering and application, modalities of processing, encyclopedic
understanding, and the like, which are absolutely irrelevant to
linguistics. Let it be emphasized that certain "sprachlich" phenomena
can be properly recognized and systematically studied only if considered
in the direction from meanings to texts. Thus the following oppositions
do not represent any interesting problem from the viewpoint of the
addressee:

·1) a. strong warning vs. considerable attention

<*considerable warning, *strong attention);

b. They differ widely vs. They miss her acutely

<*differ acutely, *miss widely);

c. D1UCb/widely/greatly publicized vs. heavily

settled [area] <*heavily publicized,
*much/widely/greatly settled).

A listener or a reader easily understands the expressions in {7) and

misses the interesting fact that the boldfaced lexemes express one and
the same meaning, namely 'very', and are {complementarily) distributed
according to the lexeme modified. In fact, these lexemes are values of

SIL-UND Workpapers 1987

 82

what is called the lexical function Hagn. (The term Jfagn is from Lat.
magnus 'great, big'.)

(8) Jfagn(warning) = strong Jfagn(attention) = considerable

Magn(differ) = widely Hagn(miss) = greatly
Magn(publicized)= much, lfagn(settled) = heavily
widely,
greatly

To discover that natural languages feature a limited number of very

abstract and general meanings such that any one of these has numerous
expressions distributed as shown above, one needs to look at language
from the speaker's viewpoint. (Cf. the discussion about lexical
functions, Sect. 3.2.3 and a list of lexical functions, Appendix A.)

Therefore, in sharp contrast with most linguistic schools

(traditional gratnmar, European and American structuralism,
transformational grammar and its various outgrowths), the Meaning-Text
Theory insists on the meaning-to-text direction in linguistic research,
as well as in linguistic description.

To sum up: The Meaning-Text Theory aims at developing Meaning-Text

models of natural languages. An MTM is taken to be a system of rules
which establish correspondences between semantic and phonetic
representations of utterances, the SemR's being written in a special
semantic language, devised by the researcher. This is a multilevel, or
multicomponent, system, with a semantic component responsible for the
correspondence between semantic and syntactic representations: the
semantic component "translates" SemR's into SyntR's (and vice versa).
Last, but not least, 1 the direction of research and description 'is that
of production: from meanings to texts.

2.2 Semantic Representation in the Meaning-Text Theory

In conformity with the goal of this paper, we will now try to throw
light on the concept of semantic representation.

Let us begin with an example. In Toronto's The Globe and Mail

(Aug. 5, 1985, N6) we find the following title:

(9) The death of libido.

This is a short editorial, claiming that watching situation comedies on

American TV has a pernicious effect on the viewer's sexual drive. The
meaning of this title is clear; how can we represent it? Well, libido
means (roughly) 'urge2 to copulate' (superscripts and numbers after
words in our semantic descriptions refer to intended senses in Longman
Dictionary of Contemporary English). 'Urge2, [noun] is by necessity
someone's urge; in this context, an indefinite group of people is meant:
'people l'. Then, taking 'urge2, and 'copulate' to be two-place
predicates 3 , we can write:

SIL-UND Workpapers 1987

 83

(10) 1.~.

libido of X = L
'people 1 X'
1 ~ · 'copulate'
2
·•people 1 Y'

or. to express it in prose,

(10 1 ) libido of X = 'urge 2 of people 1 X to copulate

with people 1 Y'

The noun 'death' is used in (9) metaphorically, and means 'event 1

consisting in that X ceases to exist 1'. In our formalism, then, we
have (11):

(11)
The death of libido =
t'
1 event 1 1
'consist in'•~- 'cease'

l
1 'exist 1'

/1"A_'urge 2•

• ( 1 • 'copulate'
'people 1 X' J2
'people 1 Y'

Numbers assigned to arrows in diagrams of (10) and (11) identify

different arguments of the same predicate, underlining singles out the
main component.

We believe that (11) represents the "sprachlich" meaning of the

phrase the death of libido quite well. Two qualifications seem,
however, to be in order.

First, while the expressions in single quotes in (10) and (11) look
like English words, they are NOT English words -- rather, they are handy
designations of specific senses of the corresponding English words,
which, like almost all English words, are polysemous. To disambiguate
them, we are using sense numbers borrowed from the Longman Dictionary,
as indicated above. (For our illustrative purposes, it could be any
other dictionary.) A word taken in one well-specified sense is called a
lexeme. Then, the expressions in single quotes are semantic units, or
seaeaes, notated with English lexemes. These sememes do not feature
syntactic and morphological properties typical of English lexemes that
represent them. They are indeed units of meaning.

Second, one might ask the legitimate question about the meaning of

SIL-UND Workpapers 1987

 84

sememes themselves. For example, what does cease mean? In this case,
the answer is simple:

(12) 'begin 1' 'not' 'P'

cease to P = . •~~~~~·

1 1

[John ceased to work= 'John began not to work']

But then, what is "begin 1"? The answer is still at hand:

(13) begin 1 to P = 'at moment t 0 , P does not exist 1;

at moment t 1 , P exists 1;
t1 > to'
However, things do not always go as smoothly as this. What does exist 1
mean? And lllODlent l? And not? For the time being, I cannot answer.

I do not imply that these terms cannot ·be explained or that their
meaning cannot be described at all. It can, but maybe NOT IN
"SPRACHLiCH" TERMS. From the viewpoint of strictly "sprachlich"
meaning, these sememes could be elementary or indecomposable: they are
then semantic primitives. (Cf. 13 semantic primitives postulated by A.
Wierzbicka: 1972 and 1980.4)

Now we can proceed to specify the formal language, or semantic

metalanguage, in which semantic representations within the MTT framework
can be written. (To simplify the presentation, we will not consider the
communicative organization of meaning, i.e. contrasts of topic vs.
comment, given vs. new, etc.)

Like any other formal language, the semantic language of the MTT is
a pair of sets: a finite set of elements called alphabet (or lexicon)
and a finite set of expressions called rules; there are formation rules,
which specify well-formed configurations of elements, and transformation
rules, which specify equivalences between well-formed configurations.
Let us take these components of the semantic language in turn:

An expression in the semantic language, i.e. a Sem(antic)

R(epresentation), is, formally speaking, a completely labeled network: a
connected oriented graph whose vertices, or nodes, are labeled with
sememes and whose arcs, or branches, are labeled with arbitrary
distinctive symbols. (We will use Arabic numbers; the purpose of using
them is, as indicated above, to distinguish dif.ferent arguments of the
same predicate, cf. below}.5

The alphabet of the semantic language in question includes:

(a) A tiny list of formal elements (used to construct networks):

vertices, represented by points; arcs, represented by arrows; and a

SIL-UND Workpapers 1987

 85

small set of natural numbers as distinguishers, appearing as labels on

arcs.

(b) A huge list of sememes (several million, generally speaking):

each sememe corresponds to a specific sense of a specific word of the
language in question. A sememe can be elementary, and then it is called
a seae; otherwise, it is complex and is representable in terms of other
sememes and/or semes.

Sememes fall into two major classes:

1. functors, or sememes having "places" for other sememes, which

are then their arguments, and

2. names (of (classes of) objects), which may only serve as

arguments but may not have their own arguments.

The relationship between a functor and its i-th argument is shown,

in the semantic language, by an arrow labeled i and pointing from the
functor to the argument. For example, 'sleep' is a one-place functor,
and therefore, Mary sleeps= 'Mary' 'sleep'; 'need' is a two-place
functor, so that • 1 •

(14) 'need'

Mary is sleepy=
'Mary' -~~ 'sleep'

Functors are meanings corresponding to actions (such as 'unite',

'kiss'), states ('sleep', 'disappear', 'joy'), properties ('lovely',
'blue'), relations ('more than', 'be part of'), parameters ('the speed
of', 'the price of'), events ('rain', 'explosion', 'fall'}, quantities
('many', 'three thousand'}, etc. Names are meanings corresponding to
objects, substances, natural species, etc.

Functors are subdivided, first according to the number of the

arguments they can take (one-place, two-place, etc.; in actual fact,
natural languages do not show functors with more than 6 arguments); and
second, according to the restrictions on the semantic type of the
arguments they can take (both arguments of 'cut' must be names; the only
argument of 'begin' must be a functor; the first argument of 'see' must
be a name, but the second may be either a name or a functor). There are
three major types of functors: predicates, quantifiers and logical
connectors, but we will not delve into this matter.

Formation rules for SemR's are trivial and largely obvious. The
following are some of the more basic ones:

- a SemR must be a connected, oriented, labeled graph

SIL-UND Workpapers 1987

 86

- the number of arrows leaving a node which is labeled

with a functor must correspond to the number of places
of this functor

- no arrow should leave a node labeled with a name

- all the restrictions inherent in functors which appear

in a SemR should be observed

Transformation rules for SemR's are equations which correlate

(non-elementary) sememes with their decompositions: 'die 2' = 'cease to
exist 1' = 'begin 1 to not exist 1', etc. Note that, strfctly speaking,
we should use semantic networks and write as follows:

(15)
1 l'die
2'
<===> rf•ase'
•
'X' l' exist 1'
1
•
'X'
However, here as everywhere below, we allow ourselves to simplify the
presentation, if this does not entail ambiguities or misunderstandings.

By means of such rules we can carry out equivalent transformations

of SemR's:

'His desire 1 died 2' = 'His desire 1 ceased to exist l' = 'His
desire 1 began 1 to not-exist l'.

Let it be emphasized that a rule of the form 'X dies 2' = 'X ceases
to exist 1' is nothing more than a lexicographic definition of a
particular lexical sense of an English word. It is here that the
concept of a special dictionary enters the scene; we will return to it
shortly.

Put into a nutshell, the semantic language for, say, English, is a

hybrid using the grammar of oriented labeled networks and the English
vocabulary, but in a disambiguated form. We take as basic lexical units
of the semantic language not words, but word senses defined in terms of
more elementary senses, which in their turn are defined in terms of
still more elementary senses, and so on, until-we arrive at semantic
primitives; these latter have to be specified by a list. To achieve
this, our definitions should obey precise and rigorous rules (for
instance, vicious circles must be banned), which we cannot discuss here.
(A few words, though, will be said about them in Sect. 2.) Instead, let
us consider a sample SemR: the SemR for a sentence taken from the same
issue of The Globe and Mail as mentioned above (Aug. 5, 1985; page N6):

SIL-UND Workpapers 1987

 87

(16) Four Soviet soldiers lost on maneuvers in Czecho-

slovakia traded their tank to a tavern owner for
two cases of vodka and were found sleeping off the
liquor in a forest two days later, a West German
newspaper has reported.

We have to simplify our SemR drastically -in order to make it

surveyable. Thus, we do not even try to decompose many sememes which
are readily decomposable. For instance, we used in the SemR, without
further ado, the expressions get lost[= 'come to not know where is the
way which one should follow'], sleep off the liquor, malleUV'ers,
exchange, and others. The case of exchange would be especially
interesting. Decomposing X exchanges Y vf.th Z for V into 'X causes that
Z owns Y, which previously X owned, having the goal of thereby causing
- and X thereby causes -- that Z causes that X owns W, which previously
Y owned' would much better show the semantic link between soldiers,
vodka and their sleeping off the liquor. Furthermore, sense-
distinguishing numbers are not shown with the sememes. We cannot
explain, either, many important details of SemR (17); we will, however,
comment on four technical points.

(17)

• period

merchandise
t. ,~
'off the
1 /liquor

-------~-------
1
located
ore st
2 1• located

SIL-UND Workpapers 1987

 88

(i) All temporal relations, grammatical or otherwise, are

represented in the same homogeneous way: by means of the sememe
'before'. Thus '(to) report before now' renders (very roughly~) the
present perfect of 'has reported'; X two clays later than Y appears as 'Y
before X (and) the distance between X and Y is two days'; for X were
found sleeping ••• we have 'someone found X, the moment of finding being
included into the period of X's sleeping'. Similarly, all locative
relations are shown in a standard way as well: the sememe '(to be)
located in' •

(ii) The sememe 'use' in 'soldiers use tank' is applied very much
like the expressions 'use the bathroom' (the telephone), i.e. in the
sense 'to use X the way Xis designed to be used'; this is the meaning
of the lexical function Keal (see below).

(iii) To show the scope of the newspaper's report, a node labeled

'event consisting in ••• • is used: the arrows leaving this node identify
the two events ('{performed) a commercial transaction' and '(somebody)
found ••• ') that enter into this scope.

(iv) The sememes underlying semantic assertions (as opposed to

presuppositions and modifications) are underscored, cf. above •

.Strictly speaking, the network which appears in (17) is not a SemR;

it is only a part of a SemR, namely the so-called Sem(antic) S(tructure)
of sentence (16). A complete SemR must include two other parts: the
Comm(unicative) S(tructure), which specifies the division of the given
meaning into theme (topic) and rheme (comment), given vs. new
information, and the like. But as indicated above, for simplicity's
sake we suppress the CommS and the Rhet(orical)'S(tructure), which
specifies the "artistic intentions of the speaker, and, by obvious
metonymy, allow ourselves to call (17) a SemR.

Notice that for pedagogical reasons, it might be more advisable to

choose a simpler SemR (i.e. to present a shorter and more trivial
sentence), as is often done to illustrate a point. We have, however,
preferred not to amuse our readers with toys but rather to show them the
real thing, even at the risk of frightening some of them away. The
crucial feature of the MTT is that it does face the enormous complexity
of natural language semantics, and we feel it is important for our
readers to be fully aware of the extent of this complexity.

The SemR as proposed has a property which is central to the present

paper: the elements of a SemR are, generally speaking, "smaller" (i.e.
semantically simpler), than the lexemes which must be used in the
corresponding sentence. Thus, 'person belonging to the armed forces of
X and having the lowest rank therein' is 'soldier'; 'X {performs) a
commercial transaction with Y consisting in exchanging by X with Ya Z
as a merchandise for a Was a payment' is 'X trades Z to Y for W'; etc.
In the Synt(actic) R(epresentation), however, the elements are of course
actual lexemes. Therefore, the main bulk of semantic rules of a

SIL-UND Workpapers 1987

 89

language, that is, of the rules that correlate the SemR and the SyntR
conveying the same meaning, is composed of rules having the following
form:

(18)

in

merchandise

There are other types of semantic rules as well; e.g., rules that map
sememes onto morphological categories, such as tenses ('before now'
<===> past tense), grammatical numbers, etc., which appear in syntactic
structures. But semantic rules of type (18) are by far the most
numerous: for a given language, they number between 10 5 and 107 -- since
every such rule represents a particular lexeme, or a word sense.

Now, a semantic rule of type (18) is nothing else but (with all
additional data concerning cooccurrence) a DICTIONARY(= LEXICAL) ENTRY.
From this it follows that semantics is, roughly speaking, a dictionary,
but a very specific dictionary. We will characterize it in Sect. 3.

3 The Explanatory Combinatorial Dictionary as a Crucial Component of a

Heani.ng-Text Model

3.1 General Characteristics

The Explanatory Combinatorial Dictionary (ECD) differs from more

familiar monolingual dictionaries in that it is not a commercial
dictionary which aims at a particular public, is adapted to fulfill a
particular task, and is bound to observe various pedagogical,
typographical and financial constraints. An ECD is conceived and
developed as one part of a scientific (theoretical) description of the
language under analysis and is thus a theoretical lexicon (cf. Lakoff
1973: 162-164). An ECD is to a "normal" dictionary what a theoretical
grammar is to a language textbook. This fundamental property entails
the following five features typical of an ECD:

SIL-UND Workpapers 1987

 90

1. An ECO is a production-oriented dictionary: it is intended to

supply all the information which is conveyed by lexical units
and which is necessary to express a given thought, and is
oriented in conformity with the orientation of the meaning-text
model.

2. An ECO is a semantics-based dictionary (that is why it is called

explanatory). The definition of the entry lexeme, written in
the special semantic language discuss~d above, serves as a basis
for the description of the paradigmatic and syntagmatic
relations of this lexeme.

3. An ECO is a combinatorial dictionary: it describes the syntactic

and lexical collocations of the entry lexeme in the greatest
possible detail.

4. An ECO is a systematic dictionary: it lays heavy emphasis on the

homogeneity of the lexicographic descriptions. It is developed
by lexical fields rather than by alphabet: similar lexemes have
similar descriptions; all links existing between the definitions
and· the syntactic cooccurrence of a lexeme must be made
explicit; etc.

5. An ECO is a formal dictionary: all information is presented by

means of a rich lexicographic metalanguage, which ensures a
rigor never before attempted in lexicography.

Consistent with these five properties, an ECD features a standard,

rigid structure for its entries, as described below.

3.2 Structure of an BCD Entry

Our discussion of ECD entries will draw. on examples from the

recently published Russian ECD (Mel'~uk and Zholkovsky 1984); the
interested reader can also consult the first volume of the French ECD
(Mel'~uk et al. 1984).

The basic unit of an explanatory combinatorial dictionary is a

dictionary entry corresponding to a single lexeme_ or a single phraseae
one word or one phraseme taken in one separate sense. A family of
dictionary entries for lexemes which are sufficiently close in meaning
and which share the same signans (identical stem) is subsumed under one
vocable, which is identified in upper-case letters.

Different vocables which are the same graphically (homonyms) are

distinguished by numerical superscripts: Russ. BRAK! 'marriage' vs.
BRAK2 'defective merchandise'.
The lexemes within a single vocable are distinguished by special
indices mentioned here in order of increasing semantic proximity (cf.
the use of numerical distinguishers for sememes, illustrated above):

SIL-UND Workpapers 1987

 91

Roman numerals differentiate lexemes whose definitions have a

non-trivial common semantic component but whose semantic relationships
are not sufficiently regular:

VYSOTA I 'height or altitude [as a vertical measure]'

VYSOTA II 'level [of prices, aspirations, pressure)'
VYSOTA III 'pitch [of sound]'
VYSOTA IV 'altitude [of the sun over the horizon)'
VYSOTA V 'altitude [of a triangle]'.

Arabic numerals indicate sufficiently regular semantic relation-

ships among lexemes that share non-trivial semantic components, or
regular polysemy: e.g. SPAT'l I.l 'sleep' vs. SPAT'l I.2 'be inert, as
if sleeping'. The literal and the figurative meanings of vspyxivat'
'blaze up/flare up' are distinguished in this way, as are kipet'
'boil/seethe with', zastyt' 'congeal/become still [with fright, etc.]';
and the like.

Lower-case letters differentiate lexemes with slight semantic

differences which are maximally regular: e.g. SKOROST' la 'speed' and
SKOROST' lb 'great speed'. The same goes for vysota 'height',
teaperatura 'temperature/high temperature', k.a!estvo 'quality/excellent
quality', etc.

It should be stressed that all full idioms are entered in the ECD
separately. (A fu11 idiOlll is a phrase whose meaning cannot be computed
from the meanings of its words and such that no constituent word retains
its full meaning; cf. shoot the breeze or Hore poJrer to hf.a: in
English.) Full idioms are provided with dictionary entries as if they
were single lexemes; for example, iz ljubvi k iskusst,ru 'for the love of
the thing' or kolot' glaza 'to throw something in somebody's teeth'.
Thus the ECD does not distinguish between mono-lexemic and multi-lexemic
units as head entities in its entries.

An ECD entry is divided into three major zones (we will not even
mention other subdivisions, which are less important in the present
context, such as morphological information, usage labels, etc.):

-semantic zone
-syntactic zone
-lexical cooccurrence zone

3.2.1 Semantic Zone. The semantic zone contains the SemR of the entry
lexeme, or its defin:ltion, written in accordance with the following six
principles:

(i) Generally speaking, the definiendum is not simply an

entry lexeme but rather a propositional fora: the lexeme
with variables representing its semantic actants, or
argument roles. Thus we define not 'to help' but rather
'X helps Y in Z by/with W'; not 'aggression' but

SIL-UND Workpapers 1987

 92

'aggression of X against Y'; not 'faithful' but

faithful to Y'; etc.

(ii) The definiens is formulated in terms of lexemes which

are semantically simpler than the entry lexeme (=
definiendum). 'Xis semantically simpler than Y' means
that we need X to define Y but Y cannot (and will not)
be used to define X. Thus if 'maneuvers of X' is
defined as 'a large-scale training exercise of the armed
forces X simulating combat', then 'maneuvers' cannot
appear in the definitions of 'large-scale', 'train',
'exercise', 'armed forces', etc. This entails the
semantic decomposition of any entry wor.d, which in its
turn, precludes vicious circles -- that common plague of
practically all existing dictionaries. Consistently
applied, semantic decompositions lead to a set of
semantic primitives.

(iii) In the definiens, neither ambiguity nor synonymy of

terms are allowed. To avoid ambiguity, the definition
is written in terms of word senses, i.e. lexemes
(lexical units supplied with distinctive indices, as
specified above) rather than in terms of polysemous
words. To avoid synonymy, just one expression is chosen
and fixed explicitly for any meaning; e.g., we say only
'having the goal of' instead of '(in order) to', 'aiming
at' or 'with the aim/goal of' etc. (In this way, an ECD
ensures univocality of defining terms.)

(iv) The definiendum and the definiens are strictly

synonymous, and ABSOLUTE MUTUAL SUBSTITUTABILITY is
required of them in all possible contexts. This means
that a definition in an ECD contains only necessary
components such that, taken together, they are
sufficient to uniquely specify the definiendum. (This
is the adequacy principle.)

(v) Any lexemes that belong to the same vocable, i.e. that
are semantically related, must show this relatedness
explicitly by displaying an identical semantic
component, called semantic bridge.·

(vi) Any vocables that are semanttcally related must be

organized according to the same pattern: they show
roughly the same number and order of related lexemes,
the definitions of lexemes are written following the
same general schema, etc. (This is the uniformity
principle; it is directly linked to the systematic
character of an ECD and its field structure.)

SIL-UND Workpapers 1987

 93

As an example, let us quote the definition of Russ. SKOROST' la

'speed':

(19) SKOROST' P-a/X-a - Y 'the speed of P/X being

Y' = 'quantity II which characterizes 2 the motion
I.l P of X by the distance I.2a Y which X covers
III in unit 4 time 2, or the value 6 of this
quantity II'.

3.2.2 Syntact~c Zone. An ECD fully specifies all the expressions which
can be syntactically headed by the entry lexeme and which are "bound" by
it (which are not free but rather idiomatic,) to a lesser or greater
extent, with respect to the given entry. This is done by means of the
so-called goveruaent pattern (GP). A GP is a table in which each column
represents one semantic actant of the lexeme (marked by the
corresponding variable), and each element in the column represents one
of the possible surface realizations of the corresponding syntactic
actant. For instance, the GP for the Russian lexeme skorost' la 'speed'
has the form:

1 =P or 1 =X 2 =y
1. Ngen 1. Numnom N' (v 'per' N' 'ace>
2. Aposs 2. v' of' Numacc N'(v 'per' N' 'ace>
3. Numnom N

4. V 'of' Numacc N

s. A

6. Ngen

This GP specifies that the first deep-syntactic actant of the

lexeme skorost' la 'speed' fills the slot of the variable P or X in its
definition (1 = P or 1 = X, where X stands for the body, and P stands
for the motion of the body whose speed is described). The second
deep-syntactic actant fills the slot of the variable Y (2 = Y, Y being
the value of the speed). The first deep-syntactic actant may be
expressed on the surface either as a noun in the genitive case or as a
possessive adjective (skorost' saaolHta 'the speed of a plane', nala
skorost' 'our speed'). The second deep-syntactic actant has six surface
realizations:

1) a complex noun phrase in the nominative consisting.of two phrases: a

noun· phrase "Numeral in the nominative + Noun denoting unit distance··
and a prepositional phrase ''-v 'per' + Noun denoting unit time", for

SIL-UND Workpapers 1987

 94

instance, akorost' desjat' kiloaetrov (v las) 'a speed of ten

kilometers (per hour)'.
For the other five realizations we shall restrict ourselves to
examples:
2) skorost' v desjat' kilometrov (v ~s) 'a speed of ten kilometers (per
hour)',
3) skorost• 40 uzlov 'a speed of 40 knots',
4) skorost' v 40 uzlov 'a speed of 40 knots',
5) kosmileskaja (tysjalekilometrovaja) skorost' 'cosmic (lOOO~kilometer)
speed', and
6) skorost' sveta (zvuka) 'speed of light <of sound)'.

A GP is usually provided with a number of restrictions. These are

rules which establish the conditions under which the deep- or
surface-syntactic actants of the entry lexeme can cooccur, and give all
possible details relevant to the combinability of the lexeme in question
with its syntactic actants.

The notations used are:

M1 2 3 - 1st, 2nd, 3rd, ••• deep-syntactic actant of the lexeme

' t •••

Cl ' 2 t 3 • • • - surface realization of the 1st, 2nd, 3rd, •••

deep-syntactic actant (in most cases, this realization· is .the
grammatical subject or the 1st, 2nd, ••• complement of the lexeme but it
may also be an attribute or adverbial modifier)

Ci i - a specific means of marking the surface-syntactic

element°Ci (that is, i is the number of the column in the GP, and j is
the number of the element in that column)

For instance, the GP of the lexeme skorost' la has such

restrictions as:

2 ) C2.1,2.2: N' is unit distance [metr 'meter'

fut 'foot', ••• ],
N'' is unit time [aekunda 'second',
mesjac 'month', ••• ]

3 > c2.3,2.4: N is a unit speed [uzel 'knot',

118.X 'Mach', ••• ]
6) Impossible:

7) Undesirable: c 1 + c 2 • 1_2 • 4 , if M1 is not a movement and skorost'

does not depena on pri 'under'.

The GP and all the restrictions on it are exemplified by all

possible combinations of the entry lexeme with its actants as well as by
all the impossible combinations prohibited by those restrictions. Every

SIL-UND Workpapers 1987

 95

starred example is followed by the number of the restriction it

violates.

3.2.3 Lexical Cooccurrence Zone. This zone embodies the major novelty
proposed in an ECD: namely, exhaustive and systematic description of
restricted lexical cooccurrence of the entry lexeme. We mean the type
of lexical collocation boldfaced in (20) -- a dozen sentences collected
in ten minutes from one newspaper page:

(20) a. The President clamped (imposed) an overnight curfew

on three areas ••• to stamp out (put down) violence.

b. The panel issued a report to the Secretary of State.

c. President Reagan rejected pleas to open talks

with striking US controllers.

d. Pope released from hospital [headline]. The Pope left

the hospital yesterday, three months and one day after he
was struck by two bullets. He said a brief
prayer •••

e. The heaviest prison terms in Kentucky history (more

than 1,600 years each) have been handed down against two
men.

f. South African troops have spread a dragnet across the

country in search for three heavily armed black
guerillas.

g. The ANC has claiaed responsibility for the attack

launched last Tuesday in which four rockets were fired
at an army camp.

h. We are looking for senior consultants of proven

competence to satisfy the demands of our growing
business.

Texts, from colloquial to artistic to technical, swarm with

expressions of this type.

To describe all such collocations, the important concept of lexical

function (LF) has been introduced.

A lexical function f is, like any mathematical function, a

dependency that associates with a given "independent quantity"' (the
argument), a "dependent quantity" (the value). More precisely, an LF f
associates with a lexical unit W (a word or a phrase) a set {Wi} of
(more or less synonymous) lexical units that express, contingent on W, a
specific idea (such as 'very', 'begin', 'implement') represented by f.
For example, the LF Mago (for the present glossed roughly as 'very') in

SIL-UND Workpapers 1987

 96

conjunction with the Russian words naprja!enie 'voltage' or teaperatura

'temperatu~e' is expressed by the adjective vysoklj 'high'. In
conjunction with vysota 'height', however, the same function is
expressed by znalltel'nyj 'considerable', bol'loj 'great' or ogr011Dyj
'enormous'; and in conjunction with vibraclja 'vibration', by sll'nyj
'strong' or lntenslvnyj 'intense'. Thus we get vysokoe naprjalenie
'high voltage', vysokaja teaperatura 'high temperature', znalltel'naja
(bol'laja. ogr01BDaja) vysota 'considerable (great, enormous) height',
and sil 1 naja (lntensivnaja) vibracija 'strong (intense) vibration'; but
we do not get *sll'noe (znalitel'noe) naprjalenie. *bol'laja
(intenslvnaja. sil'naja) teaperatura. *intensivnaja (sil'naja) vysota.
*vysokaja (bol'laja) vlbraclja, etc.

The importance of LFs consists in the discovery of the following

fact: in all natural languages there is only a limited number of
meanings (about several dozen) that resemble 'ver.y' in that they also
each determine an LF.

And now, a formal definition of lexical function:

A dependency f is called lexical function if and only if it

associates with a lexical unit W (a lexeme or a phraseme, the argument
off), a set f(W) = {Wi} of lexical units (the value of f), such that
the following two conditions are simultaneously met:

For any two different wl and w2, if f(Wl) and f(W2) both exist,
then:

1. Both f(W1 ) and f(W2 ) bear an identical relationship with respect

to meaning and deep-syntactic role to wl and w2, respectively.
[This condition is language-independent.]

2. In some cases, at least, f(W1 ) I f(W2). [This condition is

completely ·language-dependent; it means that in the given
language, the value of f is phraseologically bound by its
argument.]

An important proviso: A lexical function is not a genuine semantic

unit, let alone a semantic primitive. LF's are introduced to describe
restricted lexical cooccurrence and derivation, but by no means
semantics. First, there are LF's that are semantically empty, their
values being limited to purely syntactic roles (cf. Operi, l'unci,
Laborii below). Second, the expressions making up the value of a given
LF f for a given argument need not be perfectly synonymous; it suffices
for them to share a rather general and abstract meaning 'f' while
differing in other components.

We will be interested in a particular type of LF, namely standard

lezical functions, which form a proper subset of all lexical
functions. A standard LF simultaneously satisfies two additional
conditions:

SIL-UND Workpapers 1987

 97

3. The LF f is defined for a sufficiently large number of

arguments. In other words, f has a sufficiently large semantic
cooccurrence: its meaning 'f' is sufficiently abstract to be
compatible with a large number of other meanings. [This
condition is language-independent.]

4. The LF f has a sufficiently large number of linguistic

expressions as its possible values. In other words, the set of
all f(Wi), for a vast variety of w1 , is sufficiently rich.
[This condition is completely language-dependent.]

Let us illustrate this point with two examples. The meaning

'manufactured from very dark rye flour' in Russian has three
expressions: Elrnyj 'black', r!anoj 'rye [Adj)' and iz r!anoj multi 'from
rye flour', and these expressions are phraseologically bound: Only xleb
'bread' or suxar' 'rusk' can be called Elrnyj, while very dark rye
buloi!ka 'bun', bublik 'bagel', blin 'pancake', kor!(ik) '(a kind of)
flat, dense, dry bread', le pllka 'a flat cake', etc. cannot; *Eernaja
buloEka 'black bun' is readily understandable but ungrammatical.

Only xleb 'bread' and lepllka 'flat cake' but nothing else can be
called r!anoj. (Notice that Elrnyj xleb is always r!anoj xleb, but
r!anoj xleb can be svetlyj 'light', as well as !lrnyj.)

(Speaking of buloEki. bubliki, bliny, kor!i and kor!iki, Russian

uses iz r!anoj IIIUlti (but not with reference to suxari 'rusks' and hardly
with reference to xleb): *E!rnyj/r!anoj bublik, etc.)
As we see, this meaning satisfies Conditions 1 and 2 (Elrnyj:
xleb = iz r!anoj multi: bublik = r!anaja: lepllka vs. *r!anaja buloEka,
••• ) and therefore specifies a lexical function. However, this meaning
violates Conditions 3 and 4: it is conceivable only with the names of
bread-like baked products made from dough (so that it is semantically
too specific) and it has only three expressions. Therefore, it is not a
standard LF.

The meaning 'it is necessary to P this X' has in Russian a

phraseologically bound expression nuldat'sJa 'need' that is possible
with some P's while impossible with others: Eta stat 1 ja nuldaetsja v
ispravlenii (v dorabotke> 'This paper needs to be corrected (more work)'
but not *Eta stat'ja nu!daetsja v soxraneii (v otpravke avtoru> 'This
paper needs to be saved (to be sent to the author>'. This meaning
satisfies, in addition to Conditions 1 and 2, Condition 3 (it is
extremely abstract), but it still violates Condition 4. (There are not
numerous synonymous expressions. In fact, there is only one other
expression for the meaning in question satisfying Condition 1: trebovat'
'require', which is in free variation with nu!dat'sja).

Both above-mentioned meanings, 'manufactured from very dark rye

flour' and 'it is necessary to P this X', are non-standard LF's in
Russian. In what follows, only standard LF's are considered.

SIL-UND Workpapers 1987

 98

Since both defining conditions for standard I.F's include the vague
expression 'sufficiently large number', there is no sharp borderline
between standard and non-standard I.F's. That is, there are no formal
criteria to tell them apart. This fact reflects the graduality so
typical of natural languages. The concept of standard LF· is fuzzy, as
are most linguistic concepts.

Within the class of standard lexical functions, we will distinguish

simple I.F's and coapound I.F's, the latter being built out of the former.
Notice that simp1e in this context by no means implies 'elementary' or
'further unanalyzable': some of our simple I.F's could be represen~ed in
terms of other simple LF's (and thus could be treated as compound).
Nevertheless, for purely linguistic reasons (primarily, frequency of
occurrence) we consider a particular set of LF's as simple and take this
set to be the basis of lexicographic descriptions.

The list of lexical functions which comprises the main body of our
systematic survey is found in Appendix A. It includes only simple
standard I.F's. (This allows us to omit the adjectives "simple standard"
everywhere, since the omission cannot lead to a confusion.)

Along with the LF's listed in Appendix A, two further types of LF's
are extensively used in the Explanatory Combinatorial Dictionary:
non-standard and compound I.F's.
A non-standard LF is a meaning that is idiomatically expressed
depending on a key word, but that has either a strongly limited semantic
combinability or a fairly limited range of expressions, or both. In
other words, it is too specific, too particular to be granted the status
of a standard LF. Non-standard LF's are written in standardized natural
language. Some examples:

(21) such that Y is confined to his home (arest 'arrest')=

doma§nij 'house-[arest]'

(22) such that it is the result of a loss at cards that was not
immediately paid (dolg 'debt')= karto~nyj 'card' [dolg],
obsolete: (dolg] ~esti '(debt] of honor'

(23) during a short time and/or nonintensively (with the purpose

of knowing Y somewhat better) (u~it'sja 'learn')=//
podu~it'sja 'learn a bit (of something)'

A compound LF is a combination of syntactically related simple LF's

that has a unique lexical expression covering the meaning of the
combination as a whole. I have presented numerous examples of compound
I.F's in Appendix A; let me give some more illustrations, with the key
word printed in boldface:

SIL-UND Workpapers 1987

 99

Antilfagn : !idkie aplodismenty 'thin (lit. 'liquid') applause'

slabye dovody 'weak arguments'
nizkaja temperatura 'low temperature'
nezna~itel'nye poteri 'negligible losses'

AntiVer lo!nyj styd 'false shame'

l!ivoe obel~nie 'false (lit. 'lying') promise'
o§ibo~noe predstavlenie 'a wrong conception'
bezosnovatel'nye opasenija 'unfounded misgivings/
fears'

Incep0per1 priobretat' populjarnost' 'acquire popularity'

vpadat' v ot~ajanie 'sink into despair'
vstavat' na put' predatel'stva 'take the path of
treason'
perexodit' v pike 'go into a dive [as of an
aircraft]'

sdavat 1 V ~kscluataci~U 1 put fnto operation'

vvergat' v ra stvo 1 p unge into slavery'
stavit' pod kontrol 'put under control'

provalit'sja na ~kzamene 'fail an examination'

otvergat' sovet 'reject a piece of advice'
otklonjat' xodatajstvo 'turn down an application'

The following four remarks bearing on all LF's are in order:

(i) An LF may have a fused expression, i.e. a lexical unit that

does not include the key word but covers both the meaning of the
function itself and that of its argument (the key word). The fusion is
shown by the symbol// separating all the fused values (on its right)
from all the non-fused values. For example:

Magn(do!d' 'rain')= prolivnoj 'heavy' // liven' 'shower' [i.e., liven'

= prolivnoj do!d'; cf. Engl. downpour= heavy rain].

Magn(vkusno 'delicious') = o~en' 'very' // pal'~iki obli!e§', lit.

'You'll lick your fingers' [pal 1 ~iki obli!e' = o~en' vkusno]

(ii) Several LF's which simultaneously have the same key word but
are syntactically not linked to one another may be expressed by one
lexical unit covering the meanings of all the LF's involved. This is
what we call configuration of LF's (as opposed to compound LF's, in
which all the constituent-simple LF's are syntactically linked). In a
configuration of LF's, the "+" sign is used to separate the
constituents. For example, in the entry SUD 1 BA 1 'fate, destiny' the
notation

Fact3 11 + .AntiBonz: presledovat' 'persecute'

SIL-UND Workpapers 1987

 100

means that fate really affects [= Pact 3IIJ the person in question (the
defendant, so to speak) and its verdict is bad[= AntiB~]· Two
further examples:

A1(vosxil~enie 'delight') + llagn(vosxil~enie) = preispolnennyj

[vosxil~enija] 'full [of delight]'

Oper1 (ot~ajanie 'despair')+

Nagn(ot~ajanie) = byt' vo vlasti [ot~ajanija] 'be completely in

[despair]'.

(iii) Some LF's (most often, llagn or Reali with the latter's
relatives, Pacti and Labrealij) may be subscripted with a semantic
component of the key-word's definition (in square brackets) to indicate
that the meaning of this LF interacts with exactly this component of the
key-word's meaning. Thus:

Labreal12[xranit' 'keep') (pamjat' 'computer memory') = xranit' [v

pamjati] 'store [in memory]'

Labreal12[vydavat' 'output'] (pamjat') = izvlekat' [iz pamjati] 'extract

[from the memory]'

llagn[bojat'sja 'be afraid') (strax1 'fear') = dikij 'wild'' !utkij

'terrible'

lfaan[terjat' samokontrol' ••• 'lose ••• self-control'] (strax1 ) =

pani~eskij 'panic [adj.]', !ivotnyj 'animal'

AntiVer[pora!at' 'hit'] (streljat 11 'shoot') = ploxo 'badly, skverno

'poorly'

AntiVer[cel' 'target'] (streljat 11 ) = v vo·zdux 'into the air'

(iv) Furthermore, some LF's may be superscripted with semantic

labels, like 'usual', 'loc(ation)', 'temp(oral)', 'quant(itative)', to
make their meaning more precise:

lfaantempc~ 'experience')= dlitel'nyj 'long'

lfaanquantc~ 'experience')= bol'loj 'considerable'

SIL-UND Workpapers 1987

 101

In concluding this survey of LF's we would like once again to call

attention to the fact that they are used for two main purposes:

1) for the description of idiomatic or restricted lexical cooccurrence

or derivational relations; and

2) for specifying universal synonymic transformations of utterances on

the deep-syntactic level.
It is clear, for example, that in any language (24) holds:

(24) w = 0per1-Z->s0 (w) = Oper2 -Z->s0 (w)

= Func1-l->s0 (W) = Punc2-l->s0 (w) = Labor12 -l->s0 cw); and so on.

[Numbers on the arrows stand for deep-syntactic actants; e.g., the key
word of an Oper is its second deep actant, etc.]

The transformations presented in (24) can be exemplified as

follows:

(25) vlijat' '(to) influence'= okazyvat' [= Oper1 ] vlijanie [= s0 J

'have
influence'=~' naxodit'sja [= Oper2 J pod vlijaniem 'be under
influence', etc.

Compare (26):

(26) l~an durno vlijaet na Petra

'John influences Peter badly'=

Ivan okazyvaet na Petra durnoe vlijanie

'John has a bad influence on Peter'=

P~tr naxoditsja pod durnym vlijaniem Ivana

'Peter is under the bad influence of John'.

Rules of type (24) allow one to develop a paraphrasing system for

synonymic transformations of sentences and/or discourses. Such a system
can automatically produce, for any given text, a set of its synonymous
or nearly-synonymous paraphrases. It also can automatically derive, for
a set of synonymous texts, a canonical invariant.

SIL-UND Workpapers 1987

 102

Appendix A: List of Lexical l'unctions

A lexical function, notated by a boldfaced Latin abbreviation, is

written before the parentheses enclosing its argument (a familiar
mathematical functional notation). For example:

Function Argument Function Value

llagn(shave) =
close, clean
llagn(easy) =
very, extremely, ••• as pie
Magn(scoundrel) =
unmitigated, of the first water
Magn(condemn) =
strongly, in strongest terms
llagn(cold) =
very, terribly, ••• enough to
freeze the balls off a brass monkey
(Magn is from Lat. magnus 'great, big'.)

As stated in Section 3.2.3, LF's are one of the central notions to

a new type of dictionary, the Explanatory Combinatorial Dictionary. The
simple standard lexical functions will be listed below in the order in
which they appear in a lexical entry in such a dictionary.

1. Syn - synonym; Syn:,, Syn0 and Syn 0 designate, respec-

tively, synonyms with broader, with narrower, and with intersecting
meanings. (Symbols :::,, c, andn have the same meaning when used with
Comr, Anti and other LF's.) Examples: Syn(streljat' 'shoot.') = palit'
'fire'; Syn (streljat' 'shoot') = obstrelivat' 'fire upon, shell,
machine-gun'; etc.

2. Convij - conversive, i.e., a lexical item with the same

meaning as the key word W but with deep syntactic actants i and j
permuted: Conv21(vklju~at' 'include') = prinadle!at' 'belong to [a
set]'; Conv23lc(mnenie 'opinion')= reputacija-'reputation'. ['Reputa-
tion', in contrast to 'opinion', is necessarily held by several people;
this is why it is a narrower conversive.J

3. Anti - antonym: Anti(pobeda 'victory')= pora!enie 'defeat•.

4. Gener - Generic concept such that 'Gener+ V' = 'W' (where V

is the key word): Gener(~ 'gas') = ve§~estvo 'substance' [cf.
gazoobraznoe ve§~estvo 'gaslike substance'= gaz 'gas']

5. 1igur - Standard metaphor for W:·Pigur(blokada 'blockade)=

kol'co, lit. 'ring' [kol'co blokady 'the grip of a blockade; a siege'];
Figur(tuman 'fog')= pelena 'curtain' [pelena tumana 'curtain of fog')•

6-9. s0 , An,.Adv0 , v0 - syntactic derivatives of W; that is: noun

(= substantivalJ, adjective, adverb, and verb, respectively, which have
the same meaning as w. Examples: S0 (streljat 1 'shoot')= strel'ba
'shooting'; Ao(streljat' shoot')= strelkovyj 'shooting [attrib.J'; etc.

SIL-UND Workpapers 1987

 103

10. Si - standard name of the 1-th participant in the situa-

tion described by V: S1 (u~it' 'teach')= u~itel' 'teacher'; s2 (u~it'
'teach') = u~enik 'pupil'; S3 (u~it' 'teach') = (u~ebnyj) predmet
'(subject) matter [as in high school]'.

11-15. Sinstr' 8med' Smod' Sloe' Sre~ standard name of

instrument, means, mode, location, and result of the situation described
by V: Sinst (streljat' 'shoot') = ognestrel'noe oru!ie 'firearm';
Smed(streljatf 'shoot')= boepripasy 'ammunition'; Smod(rassmatrivat'
'consider') = vzgljad [na ~to-1.] 'a view [of something]'; podxod [k
~emu-1.J 'approach [to something]'; s10 c(sra!at'sja 'fight [as of two
armies]') = pole bitvy/boja 'battlefield'; Sres<u~it'sja 'learn')=
navyki 'skills', znanija 'knowledge'.

16. Sing - 'one instance/unit of': Sing(gorox 'peas')= goro§ina

'pea'; Sing(celovat' '[to] kiss')= pocelovat' 'give a kiss'.

17. Nolt - 'aggregate of': Nalt(korabl' 'ship')= flot 'fleet';

Halt(student 'student')= studen~estvo 'student body'.6

18. Cap - 'head of': Cap(universitet 'university') = rektor

'president'; Cap{fakul'tet 'faculty, school')= dekan 'dean'.

19. Equip - 'staff/crew of'; Equip{teatr 'theatre') = truppa

'troupe'; Equip(bol'nica 'hospital')= personal 'personnel; Equip(brak.!
'marriage')= suprugi 'spouses'.

20. Centr - 'center/culmination of'; Centr(les 'forest')= ~a§~a

[lesa] 'the thick [of the forest]'; Centr(slava 'glory') = ver§ina
[slavy] 'summit [of glory]'; Centr(bor'ba 'struggle')= apogej [por'by]
'climax [of struggle]'• Centr is current in combination with Locin (see
below): LocinCentr(pustynja 'desert')= v serdce [pustyni] 'in the heart
[of the desert]'; LocinCentr(doroga 'road')= posredi [dorogi] 'in the
middle [of the road)'•

21. At - determining property of the 1-th participant of a

situation characterizing him according to his role in the situation:
A1(gnev 'anger') = v [gneve] 'in [anger]', razgnevannyj 'angry';
A1 (sl~zy 'tears')= v [slezax] 'in [tears]' A1 (skorost' 'speed') = so
iskorost'ju ... ] 'with a speed of ••• , [compare spusk s takoj skorost'ju
the descent with such a speed']; .!i(streljat' 'shoot')= pod obstrelom
'under fire'.

22. Ablei - determining property of the 1-th potential

participant of a situation ('such that it can ••• easily'/'such that it
can be ••• easily'): Able1(plakat' 'cry') = slezlivyj 'tearful';
Able2 (somnevat'sja 'doubt')= somnitel'nyj 'doubtful'.

23. Nagn - 'very', 'to a (very) high degree': Hagn(temperatura

'temperature') = vysokaja 'high'; Hagn(rassmatrivat' 'examine') =
vnimatel'no 'attentively', pristal'no 'fixedly, intently'.

SIL-UND Workpapers 1987

 104

24-25. Plus, Hf.nus - respectively, 'more' or 'less' (or 'to a

greater/lesser extent') [than something else].

26-27. Plusrefl, Hf.nusrefl - indicate that the comparison is made

with a former state of the same object: IncepPredPlusrefl(temperatura
'temperature') = povy§at'sja 'rise, increase' [For the Lf's Incep and
Pred, see below.]

28. Ver 'as it should be' [meeting intended requirements]:

Ver(udivlenie 'surprise') = iskrennee 'sincere', nepoddel'noe
'unfeigned'; Ver(sosud 'container') = celyj 'whole', germeti~nyj
'hermetical, leak-proof'; Ver(pribor 'instrument')= to~nyj 'precise'.

29. Bon - 'good' [a standard praise for V]: Bon(rez~t' 'cut')=

akkuratno 'neatly, cleanly'; Bon(sudnol 'ship) = komfortabel'noe
'comfortable' •

30. Posi - a standard praise of one of the participants of the

situation denoted by V [but not of the situation itself]: Pos?(recenzija
'review') = polo!itel'naja 'positive', while Bon(recenzija review•)=
xoro§aja 'good, zame~atel'naja 'excellent', ••• , blestja§~aja
'brilliant'; however, a blestja§~aja recenzija 'a brilliant(ly written)
review' may well be polo!itel'naja 'positive' or otricatel'naja
'negative', that is, Pos 2 or .AntiPos 2 •

[NB: The LF's Hagn, Ver, Bon and Posi are often combined with Anti.
Thus, for instance, Hagn(temperatura 'temperature') = vysokaja 'high,
and .AntiHagn(temperatura 'temperature) = nizkaja 'low'; Pos 2 (mnenie
'opinion)= polo!itel'noe 'positive', and AntiPos 2 (mnenie 'opinion') =
otricatel'noe 'negative'.]

31 • .Advi determining property of an action by the i-th

participant of a situation according to his role in the situation:
Adv1(sl~zy 'tears)= so [slezami] 'with [tears]'; .Adv1 (skorost' 'speed')
= so skorost'ju ••• 'at a speed of•••' [cf. m~at'sja so skorost'ju •••
'tear along at a ••• speed'); Adv2 (somnevat'sja 'doubt')= vrjad li
'hardly'.

32-34. Locin• Loeb• Locad - preposition governing V [= the name

of the situation] ana designating a type of localization in space with
the respective meaning position, moving away, moving toward.
Examples: Locin(vysota 'height') = na [vysote] 'at [a height]';
Locad<vrsota 'height) = na [vysotu] 'to [a height]'; Locab(vysota
'height = s [vysoty] 'from [a height]'.

35. Locfiimp - a preposition [analogous to Loe] with the meaning of

temporal localization: LocfiimP(arest 'arrest') = pri [areste] 'while
being [arrested]'; Locf~mP(analiz 'analysis')= v xode [anaUza] 'in the
course of [analysis]'.

SIL-UND Workpapers 1987

 105

36. Instr - a preposition [analogous to Loe] with the meaning of

instrumentality: Instr(pistolet 'pistol')= iz [pistoleta], lit. 'with
[a pistol]'; Instr(ma§inka 'typewriter') = na [ma§inke] 'on [a
typewriter]'•

37. Propt - a preposition with the meaning 'because of', 'as the
result of': Propt(strax 'fear') = ot [straxa], so [straxu] 'from
[fear]'; Propt(ljubov' 'love') = iz [ljubvi k ••• ] 'because of [one's
love of ••• ]'; Propt(E..EX!. I.1 'experience')= na [svo~m opyte] 'from
[one's own experience]'•

38. Copul - a copula: Copul(u~itel' 'teacher')=~. rabotat'

[u~itelem] 'be, work as [a teacher]'; Copul(primer 'example') = byt',
javljat 1 sja, slu!it' [primerom] 'be, represent, serve as [an exampleJr:-

39. Pred - a verb meaning 'be W', i.e. semantically covering

the syntactic combination of a Copul(W) with w. Thus Pred is nothing
but a 'fused' expression of Copul(W) + W (on 'fused' expressions see
below) needed for the convenience of some synonymic transformations.
For example, Pred(pjanica 'drunkard')= pjanstvovat' 'drink heavily',
Pred(rjadom 'next to')= sosedstvovat' '(to) neighbor'.

The next three LF's are verbs which are semantically empty in the
context of the entry lexeme (i.e. their key word) and which serve to
link, on the syntactic level, the name of a participant of a situation
to W - the name of the situation itself. They play important
semantico-syntactic roles and can be loosely called semi-auxiliaries.

40. Operi - the first deep actant (and the surface subject) of
this verb is the 1-th participant of the situation, and the second deep
actant (or the first surface object) of Operi is V {further actants, if
any, designate further participants of the situation); Oper1 (sl~zy
'tears')= lit', prolivat' 'shed'; Oper1 (arest 'arrest') = proizvodit'
'make [an arrest]'; Oper2(arest 'arrest')= popadat' [pod arest] 'fall
[under arrest]', podvergat 1 sja [arestu] 'undergo [arrest]';
Oper1 (soprotivlenie 'resistance') = okazyvat' 'show, put up';
Oper2 (soprotivlenie 'resistance')= vstre~at' 'meet', natalkivat 1 sja [na
soprotivlenie] 1 run [into resistance] •

41. l"unci - the first deep actant (and the surface subject) of
this verb is W - the name of the situation, and the second deep actant
(and the first surface object), the i-th participant of the situation:
Func1{udivlenie 'surprise, astonishment') = oxvatyvat', lit. 'seize'
[i.e. the person is overcome by surprise, astonishment];
Func2 (temperatura 'temperature') = ravnjat'sja 'be equal to';
Func1(predlo!enie 'proposal') = isxodit' [ot kogo-1.] 'stem from, come
from [someone]'; Func2 (predlo!enie 'proposal') = kasat 1 sja [~ego-1.]
'concern [something]'• If there is no complement at all, i.e. Fune is
an intransitive verb, the subscript O is used: FunCQ(do!d' 'rain') =
idti, lit. 'walk' [cf. Engl. fall]

SIL-UND Workpapers 1987

 106

42. Laborii - the first deep actant (and the surface subject) of
the verb is Ehe i-th participant of the situation; the second deep
actant, the j-th participant of the situation; and the third deep actant
(implemented by the second surface object) is W itself. Examples:
Labor12(dopros 'interrogation')= podvergat' [kogo-1. doprosu] 'subject
[someone to an interrogation]'; Labor32 (arenda 'lease') = sdavat'
[~to-1. v arendu] 'grant [something on lease]'•

Oper, :rune, and Labor can be paired in converse relations, that is,
Oper1 = Conv21 (Punc 1 ); Labor12 = Conv132 (0per1 ), and so on. These
relationships may be diagrammed as follows:
w

Labor 12
1st actant
Labor 21

Here, a two-participant situation designated by key word W is

presented. Arrows represent semi-auxiliary verbs; the arrow's tail
indicates the surface (grammatical) subject, the head pointing toward
the first surface object.

A different way to express the same idea is by using a matrix:

First Second
Surface surface surface
subject object object

1st/2nd deep
actant of W w
none/1st/2nd
Punco/1/2/ w deep actant
of W

1st/2nd 2nd/1st
Labor12/21 deep actant deep actant w
of W of W

SIL-UND Workpapers 1987

 107

For instance, for 'A changes B' (in Russian) we have:

proisxodit
'takes place'

A B
?

43. Involv - a verb that links a non-participant of a situation

with the name of the situation acting on him: Conv21 1uvolv(veter 'wind')
= stojat' [na vetru] 'stand [in the wind]'; Inceplnvolv(metel
'snowstorm') = zastigat' 'catch'; "2Involv(metel' 'snowstorm') = v
[meteli] 'in [a snowstorm]'.
44-46. The following three LF's represent the meanings of what are
often called phasal verbs: Incep - 'begin'; Cont - 'continue'; Fin -
'end, cease'. They are connected by obvious semantic relationships:
Fin(P) = Incep(nonP); Cont(P) = nonFin(P) = nonlncep{nonP). Incep, Cont
and Fin are used (at least in Russian) in combination with other LF's.
Examples: Oper2(vlast' 'power') = naxodit'sja (pod vlast'ju ••• ] 'be
[under the power of)', Incep0per2 (vlast') = popadat' [pod vlast' ... ]
'fall [under the power of]', Fin0per2 (vlast') = vyxodit' [iz-pod vlasti]
'get out.[from under the power of]'; Cont0per1 (vlijanie 'influence') =
soxranjat' 'maintain', Cont0per2 (vlijanie) = ostavat'sja [pod vlijaniem
.!..!.!l 'remain [under the influence of]'; ContFunc0 (zapax 'odor') =
der!at'sja 'linger'.

47. Caus - 'cause', 'do something so that a situation occurs'.

Caus is often used in combination with other verbal LF's. Examples:
Caus0per1 (mnenie 'opinion') = privodit' [kogo-1. k mneniju] 'lead
[someone to an opinion]'; CausP'unc1 (nade!da 'hope')= vseljat', vdoxnut'
[nade!du v kogo-1.J 'raise [hope in someone], inspire [someone with
hope]'; Caus0per2 {obed 'dinner')= gotovit' [~to-1. na obed] 'prepare
[something for dinner]'; CausP'unc0 {obed) = gotovit', strjapat' [obed]
'make, cook [the dinner]'.

With the LF Caus, the LF's Perm and Liqu are naturally associated:

Liqu(P) = Caus(nonP), Pera(P) = nonLiqu(P) = nonCaus(nonP). Both

Perm and Liqu are usually used in combination with other verbal LF's.

48. Perm - 'permit', 'allow': nonPeraOper2(kritika 'criticism')

= ogra!dat' [kogo-1. ot kritiki] 'protect [someone from criticism]';

SIL-UND Workpapers 1987

 108

Perm0per2 (~kzamen 'exam') = dopuskat' [kogo-1. k ~kzamenu] 'allow

[someone to (take) an exam]'.

49. Liqu - 'liquidate', 'do something so that a situation does

not occur or stops occurring': LiquFunc0 (negramotnost' 'illiteracy') =
pokon~it' [s negramotnost'ju] 'wipe out [illiteracy)'; Liqu1Func0 (kost~r
'campfire')= potu§it' [kost~r] 'extinguish [a campfire)'•

50-52. Now let us look at another triple of interrelated LF's:

Reali, Pacti and Labrealij"

The LF's Reali, Facti, and Labrealij are syntactically analogous to

the functions Operi, Punci and Laborijt respectively. This means that
the names of the situation and of its i-th participant fulfill with
respect, for example, to Reali the same syntactic roles as they do with
respect to OperH etc. However, unlike the "empty" LF's Operi, Punci,
and Laborij, the lexical functions which we are now concerned with
correspond to a specific meaning: 'fulfill a demand or requirement of
•••'• The demands can differ for different W's. For example, the
fulfillment, or realization, of a hypothesis is its confirmation;
therefore, Real2 (gipoteza 'hypothesis')= podver!dat' 'confirm' [Fakty
podtver!dajut gipotezu 'The facts confirm the hypothesis'], and
Pact2(gipoteza 'hypothesis')= sootvetstvovat' 'be in accordance with
[Gipoteza sootvetsvuet faktam 'The hypothesis is in accordance with the
facts']. Realization of an artifact is its utilization according to its
intended function; therefore, Pact 0 (no! 'knife')= rezat' 'cut' [Etot
no! re!et xoro§o 'This knife cuts well'r:- Further examples:7 ~~

Real1(obvinenie 'accusation')= dokaz;rv:at' [obvinenie] 'prove [an

accusation]'; Real1 (u~ebnoe zavedenie 'educational institution') =
prepodavat' [v u~ebnom zavedenii] 'teach [in an educational
institution] ' •

Real2(obvinenie) = sogla§at'sja [s obvineniem] 'agree [with an

accusation]'; Real2(u~ebnoe zavedenie) = izu~at'sja, prepodavat'sja [v
u~ebnom zavedenii] 'be stu4ied, be taught [in an educational institu-
tion)' (while Real3(u~ebnoe zavedenie) = u~it'sja [v u~ebnom zavedenii]
'study [in an educational institution]'); Real2 (soblazn 'temptation')=
poddavat'sja [soblaznu] 'yield [to temptation]'.

Pact0 (somnenie 'doubt') = podtver!dat'sja 'be confirmed',

opravdyvat'sja 'prove justified'; Pact 9(nade!da 'hope')= sbyvat'sja
'come true'; Fact0 (sudno 'ship')= plyt' sail'.

Pact1(o~ered' 'turn')= byt' [za kem-1.] 'be [someone's (turn)]'

[O~ered' za vami 'It's your turn']; Pact1 (~ksperiment 'experiment')=
udavat'sja [komu-1.] 'work out [for someone]'.

Pact2(sudno! 'ship') = vezti, perevozit' [gruzy, passa!irov]

'convey, transport [cargo, passengers]'; Pact 2 (sosud 'container')=
soder!at' [~to-1.) 'contain [something]'•

SIL-UND Workpapers 1987

 109

Labreal12(viselica 'gallows')= vzdernut' ko o-1. na viselicu

'string up [someone on a gallows ; La rea 12 (obed 'dinner')= est'
[~to-1. na obed] 'eat [something for dinner]'.

The LF's Reali, l'acti, and Labrealii can be superscripted with

Roman numerals to indicate the degree O·f l:he realization or fulfillment:
the superscript I means fulfillment only at the psychological level,.
with the superscript II meaning fulfillment at the physical level, cf.
ReaI21 fprigla§enie 'invitation') = prinimat' 'accept', while
Real21 (prigla§enie) = sledovat' 'follow'; or Fact11 (cuvstvo 'emotion')
= govorit',,. podskazyvat' · 'tell', while Fact1 11 (~uvstvo) = zastavljat'
'force'.

53. Manif - 'manifest itself', 'become apparent': lfanif(vina

'guilt', 'fault')= obnaru!ivat'sja 'become apparent'; lfanif{udivlenie
'amazement') = skvozit 1 'lurk'; Manif(bezgramotnost' 'ignorance')=
projavljat'sja 'manifest itself'.

S~t 'symptom', i.e. a verbal expression denoting a bodily

reaction that is the symptom of an emotional or physical function:
S:,apt(udivlenie 'amazement', glaza 'eyes') = [U nego] glaza na lob
polezli '[His] eyes started from his face'; Syapt(udivlenie 'amazement',
rot 'mouth')= razinut' rot 'open [one's] mouth wide'; Sympt{strax
'fear', volosy 'hair)= [U nego] volosy vstali dybom '[His] hair stood
on end'.

55. Prepar - 'prepare', 'get (something) ready for normal use or

functioning': Prepar 1Fact 0 (revolver 'gun') = zarja!at' 'load';
PreparII:ract0 (revolver) = vzvodit' kurok 'raise the cock'. Roman
superscripts, in much the same manner as with Reali, l'acti, and
Labrealij• express the degree of readiness. Compare also:
Prepar0per1 (obed 'dinner') = vyxodit' [k obedu] 'appear [for dinner]';
Prepar0per2 (obed) = podavat' [na obed] 'serve [something for dinner)';
Preparl'unc1(obed) = podavat' [obed komu-1.] 'serve [somebody dinner]'.

56. Prox - 'be about to/ on the verge of': Prox0per1 (ot~ajanie
'despair')= byt' na grani [ot~janija] 'be on the edge of [despair]';
Proxfunc0 (groza 'thunderstorm')= sobirat'sja 'gather, brew'.

57. Degrad - 'degradation', 'become worse or bad': Degrad(moloko

'milk')= skisnut' 'go sour'; Degrad{mjaso 'meat') = isportit 1 sja,
protuxnut' 'go bad'; Degrad{disciplina 'discipline') = ras§atat'sja
'decay'.

58. Son - 'emit characteristic sound': Son(sobaka 'dog') =

lajat' 'bark'; Sou(banknoty 'banknotes')= xrustet' 'rustle'; Son(sneg
'snow')= skripet' 'crunch'; Son(vodopad 'waterfall')= revet' 'roar'.

59. Imper 'do Wt': Imper(streljat' 'shoot')= ogon't 'fire!';

Iaper{brat' oru!ie 'seize arms) = v ru!'~t 'take up arms!';

SIL-UND Workpapers 1987

 110

Iaper(govorit' tixo 'speak lowly') = ts-s-s t, d-1-!H I Sh-h-h 1 ;

Iaper(brat' 'take')= na (te)t 'take it?', 'beret'.

60. Perf - 'perfective', i.e. 'have the process carried through

to its natural limit': Perf(vstavat' 'be standing up') = vstat• 'have
stood up'; Perf(relat' 'be solving [a problem]')= relit' 'have [it]
solved'.

61. Imperf - 'imperfective', i.e. 'be carrying out the process':

Iaperf(vstat') = vstavat'; Imperf(relit') = relat'.

62. Result - 'resultative', i.e. 'the state of affairs that

normally results from the completion of the process': Result(pokupat'
'buy')= imet' 'have'; Result(lo!it'sja 'lay down')= le!at' 'be lying';
Result(nau~it'sja 'have learnt') = umet' 'know', 'have necessary
,•ills'.

SIL-UND Workpapers 1987

 111

Appendix B: Three sample dictionary entries

In order to give the reader the full taste of an ECD, we will quote
two dictionary entries, one Russian and one French (from Mel'~uk and
Zholkovsky 1984: 651-653 and Mel'~uk et al. 1984: 148-149
respectively). These are the lexemes predlo!eniel la,b and proposition!
1, both meaning roughly 'proposal' = 'the act of proposing or what is
being proposed' and being approximate equivalents under translation.

The entries are reproduced here with slight modifications aimed at

facilitating comparison. Since we address ourselves to an
English-speaking audience, both actual metalanguages (Rusaian and
French) have been replaced by English. We are not in a position to
explain all the details of presentation; but we hope they will become
clear in due course.

For the benefit of English readers, we add also an English lexical

entry written by James Steele (Carleton University, Ottawa) and heavily
edited by the author with the help of David Weber and David Tuggy
(Summer Institute of Linguistics, Grand Forks, North Dakota). It is the
first sense of the noun hope II (hope I being the verb).

SAIIPLB 1: RUSSI.AR

Russ. PmmLOZBRI Igl, .J!.,...J.!, neuter

la,lb. Predlo!enie X-a [Z-u] delat' Y 'Proposal by X [to Z] to do
Y' = 'What is (being) proposed by X to Z' [= S2(predlagat' la,lb or
predlo!it' la,lb].

Govenment Pattern

1 =X
1. Ngen
I 2

1. Ngen
=y
I 3 = z

2. ot Ngen 2. Vinf

3. ~oss 3. ~toby +
PROPOS

4. A

SIL-UND Workpapers 1987

 112

2) c1 .4: A= Ao(N), where N is the name of a country

3) c2 •1 : only for PREDLOZENIE 1 la; N is a predicate

4) c1 •1 + c2 • 1 : impossible

c1 predlo!enie Petra (Francii) 'proposal by Peter (France>', [Ja

imeju] predlo!enie ot direktora '[I have] a proposal from the
director', na§i (francuzskie) predlo!enija 'our (French)
proposals'
c2 : predlo!enie mira <pomo§H) 'offer of peace (help)', predlo!enie
otdoxnut' <~toby my otdoxnuli) 'proposal to rest <that we have
a rest>'
Cl+ c2 : americanskoe predlo!enie pomo§~i 'American offer of help',
predlo!enie brigadira otdoxnut' <~toby my otdoxnuli) 'the
proposal by the brigade chief to rest (that we have a rest>'

Impossible: *Men a tronulo redlo!enie omo§~i ot Koli 'I was moved by

the offer to help from Kolya 1 [= Menja tronulo Kolino
predlo!enie pomo§~i; but cf. U menja est' [= Conv21Result 3 ]
predlo!enie pomo§~i ot Koli 'I have an offer of help from
Kolya']; *moskovskoe predlo!enie 'Moscow proposal' (2) [=
predlo!enie Moskuy]; *predlo!enie otdyxa <knigi) 'proposal/
offer of rest (of a book)' (3); *predlo!enie Peti pomo§~i
'offer of help by Pete' (4)

Lerlcal Functions

Vo : predlagat' la,b, predlo!it' la,b 'propose, offer'

Magn nastoj~ivoe 'insistent'
Magnquant neodnokratnye 'multiple' IP. is in pl
AntiMagn : robkoe 'timid', ostoro!noe 'cautious'
Ver 1 ser'~znoe 'serious'
AntiVer 1 legkomyslennoe 'flippant', neobdumannoe
'unthoughtful'
Bon1 ljubeznoe 'amiable'
Bon2 vygodnoe 'advantageous'
Mago+ AntiBon1 nazojlivoe 'importunate'
Oper 1 (P. 2-+ become
the wife of X) delat' 'make' [-e Ndat1
F1 = Oper 1+the same
as P.' of W : prisoedinjat'sja 'join' [k -ju
Ngen = W]I either c1 ~ A, or P. has an
iaentifying modifier
PerflncepOper3 polu~it' 'receive'[-e
(ot I from' Ngen>) - .
ContFunc 0 ostava'sja v sile 'still hold'
Liqu1Func 0 brat' obratno 'take back' [(svo~) -e]
Func 1 isxodut' 'come' [,2! 'from' Ngen1
PerfFunc1 offic posledovat 1 'follow' [so storony

SIL-UND Workpapers 1987

 113

'from the side of 'N en1 IP. cannot have a Maga

sostojat', saklju~at~sja 'consist'

... [v 'in' Nprep1

Examples

Ona otvetila otkazom na ego robkoe predlozenie pomo~' ej 'She

answered with a refusal to his shy proposal to help her'. Po
predlozeniju xozjajki my ostalis' e§~~ na dva dnja 'Following the
proposal of the lady of the house, we stayed two more days'.

SAMPLE 2: FRERCR

Fr. PROPOSITIORl, feminine

1. Proposition de X 1 Z de Yer 'Proposal by X to Z to
do Y' = 'Fact that X proposes Y to Z or the message concerning Y
by means of which X proposes Y to Z' [= s 0 /s 2 (proposer 1))

Govermaent Pattern

1 = X 2 = y 3 = Z

1. de N
- 1. de N
- 1.
-a N
2. Aposs 2. de Vinf

3. A 3. que PROPsubj

1) cl.3 A= Ao(N), where N is the name of a country, or N =

patron, gouvernement, •••
2) C2.3 M1 ·= M1 (Y)
3) C3 without C1 or c 2 : not desirable

la proposition de Pierre (de la France>

'proposal by Peter (France>, notre proposition
'our proposal', les propositions fran~aises
'French proposals'
1
la proposition de aix (d'aide> 'proposal
of peace <help) 1 , a proposition d'apporter de
l'aide 'proposal to give some help', la
proposition que son valet fasse cela ""Proposal
that his servant do that'
sa proposition (la proposition francaise> de
travail a Marie 'his (the French> proposal of
work to Mary', sa proposition (la proposition
francaise> a Marie de participer au tournoi 'his

SIL-UND Workpapers 1987

 114

(the French> proposal to Mary to participate at the

tournament', la proposition d'argent de Pierre a
Marie 'proposal of money by Peter to Mary'

Rot desirable: ·la proposition a Marie (3) [= la proposition

?

faite a Marie 'the proposal made to Mary']

Lexical Functions

Sync offre 'offer'

Vo : proposer 1 'propose'
Ver 1 s~rieuse 'serious'
AntiVer 1 a la l~g~re 'lighthearted'
Bon 1 amicale 'friendly'
Bon2 : avantageuse 'advantageous', int~ressante 'interesting'
Oper 1 : faire 'make' [ART-]
Prox0per 1 avoir 'have' [ART - (a Oper 1 )J
[J'ai une proposition-Ca faire) 'I have a proposition
(to make}')
Cont0per 1 maintenir 'maintain' [ART-]
Oper 3 recevoir 'receive' [ART-]
Result0per 3 avoir 'have' [ART-] [J'ai une proposition
avantageuse 'I have an advantageous proposal']
ContFunc 0 tenir encore <toujours> 'still hold'
Liqu1 Func 0 retirer 'take back' [ART -1~ revenir 'take back' [.!!!!,
ART-]
venir 'come' [de NJ
consister 'consist' [en NJ
s'adresser 'be adressed' [.! NJ
•••
Examples

Bien qu'officiellement accept~e, ma proposition est rest~e lettre

morte 'Although officially accepted, my proposal went unheeded'. Il
accueillit ma ro osition d'aide avec oie (froidement> 'He received my
proposal offer of help joyfully <coldly>'.

SAIIPLB 3: DGLISB

II. HOPE, noun

1. X's hope of Y = S0 (hope I.I).

SIL-UND Workpapers 1987

 115

Governaent Patte'l'll

1 = X 2 == y

1. of N 1. for N

2. N's 2. of N

3. Aposs 3. of V

4. to vrr
nf
5. that PROP

H.B.: The plural [-s] can be used in the singular sense with C2.3 4
and, if N is singular, with c2• 1 , 2 (His hopes<= hope) of winning t~e
lottery prize died when his number was not drawn; He bought ten tickets
for the lottery in hopes<= hope> of winning the prize].

1) C1.1 + C2.2 not desirable

Cl the hope of Peter, Peter's (his> hope

C2 hope for <of) victory, hope of winning (to win> a

victory, hope that victory would be won

Cl+ C2 Peter's (his) hope of being successful (to be a

success>, the hope of Peter (his hope) that success
would be achieved
Not desirable . 1the hope of Peter of success (1) = Peter's hope of
success

Lexical Functions

desire; expectation
wish, aspiration, dream; faith; prospects;
confidence; optimism
Contr fear
Anti despair
Antin doubt; fear, dread
Gener feeling, sense [of-]
Suddenly IncepFunc1
+ Magn + Figur surge [of -]IX is a person; wave
[of -]IX is many people
IncepOper1(surge of H.) 2 : have, feel [a surge of-] H. in
F1 = Involv(wave of H.) singular;
-+Z sweep [across N = Z] c1 = A
CausF1 send [a wave of - across NJ
[President's speech sent a wave of
hope across the country]

SIL-UND Workpapers 1987

 lH

Suddenly IncepFunc1
+ AntiMagn + Figur flicker, glimmer, gleam, ray,
spark [of-]; breath [of-];
grain, seed [of-] ['H. is
supposed to grow'] H. in
Incep0per 1 {flash, singular;
glimmer.gleam, Ci= A
spark of H.) have [a flicker (glimmer, gleam,
spark> of -1
Incep0per 1{ray of H.) : see [a ray ot -1
CausFunc 1{ray of H.) give (N = X a ray of -1 I Ml
is a fact
AntiMagn + Figur shadow [of {a) -]lused in negative
constructions, H. in singular [The search
party continued to look for the lost
child without a shadow of hope that the
little creature would be found alive]
full [of -]IH. in singular,
filled [with -1
void, devoid [of -1, without[-];
lit or arch out [of-] [The old H. in
king was out of hope when he learne singular;
of the disaster] C = A
i
Vo hope I.1
Culm : pinnacle [of -1
LocinCulm at the pinnacle [of -1
Magn[ 'desire'] big 1 (= 'intense'], strong (great 1
[= beyond the ordinary in strength']
[Their great hope for an early
settlement did not materialize], high 1
(= 'intense to an extraordinary
degree']; dear, earnest, passionate,
profound, sincere; eager, lively
<ardent, exultant, fervent, intense
[However passionate <exultant, fervent:
were their hopes for a coming change,
they couldn't withstand the last blow];
unconquerable, irrepressible
AntiMagn +
A1IncepPredPlus budding
CausPredPlus build up 2, bolster, encourage, expand
2, foster, increase 2, strengthen,
spur, warm 2 [~/ART-] [Do not
allow an initial success to build ·UP
your hopes]; coll get up [~oss ~]
[Do not get your hopes up too soon]
Magn{ Build up 1, 2) : [so, too ••• ] high, coll a lot
IncepPredPlus
. 1quant spread 1 [among N = X] [Hope
spread among the starving villagers
when rain clouds appeared on the
horizon]

SIL-UND Workpapers 1987

 117

CausPredPlusquant expand 2, increase 2, spread 2

[-/ART - among Ns) [The good news
spread hope among the inhabitants]
IncepPredMinus diminish 1, dwindle, lessen 1, shrink
1, sink 1, weaken 1 [Their hopes
diminished when they took·this new
consideration into account]; cool 1,
cloud 1, darken 1, dim 1, fade, fail 1
[= 'become weaker'], go down, wane,
wither 1 (Hopes faded after the defeat
of their best regiment]
CausPredMinus diminish 2, lessen 2, shrink 2, sink 2,
weaken 2, cool 2, cloud 2, dampen,
darken 2, daunt, dim 2, wither 2,
undermine, [-/ART-] (The injury
of their best player cooled their hope
of winning the game]
Oper 1 : entertain, have, hold out [-/ART-];
experience, feel [-/ART-]
AntiVer + Oper 1 indulge [ART -]IC2 ~ A
Incep0per 1 find [-] c = A
(Plus+) IncepOper1 gain [-] 2
ContOper 1 live [in-]; live, rest [in the-]
I c2 ~ A [You must live in hope even
though the odds are against you]
Caus 1 cont0per 1 cling [to the-] I c2 ~ A
[= nonPerm1 FinOper 1 )
FinOper1 lose [-/ART-] // lose heart
Caus0per 1 inspire [N with -/ART-]
Oper 1 Not much - < No - (left)
I H. in singular, Ci= A
Liqu0per 1 cheat [N of ART-]
Liqu1 oper 1 abandon, discard, give up, leave
(behind), relinquish, surrender [-/ART-]
A1nonPermOper 1 past, beyond [-] .1 H. in
singular, Ct.= A [The patient seemed
beyond hope
Able 2oper 1 of[-] I H. in singular [a child of
great hope; a country of hope]
Oper 2 be [To visit Japan was his
long-time hope]
Magn['probability') certain I c 2 ~ A , confident, firm;
lit and arch absolute I G ~ Oper1
AntiMagn('probability'] doubtful, faint, feeble,
forlorn 1 (= 'weak'], little, scant,
slim, small, unsure; all I G(H.) =
abandon, lose [Abandon all hopetl
expressed cautiously+
AntiMagn('probability'] : cautious [The cautious hope of the
Minister of Finance was that new fis-
cal measures would bring an end to the

SIL-UND Workpapers 1987

 118

recession]
Magn1 boundless, unbounded
Magn 1quant : common, general< universal
AntiMagn 1 : humble
Bon2 + Magn 2 : big 2, bright; great 2 [= 'pre-eminent'],
high 2 [= 'of exalted importance'],
monumental, utmost [To reach the moon in a
single-stage rocket was their big hope;
Everyone has such high hopes for what
can be done in Africa]
Magntemp A1contFunco/l
Bon to X dear, good, pleasant, pleasing, sweet
< fondest
Ver real, substantial
AntiVer blind, drunken, deluded, empty,
fallacious, false, fantastic, fond,
futile, idle, illusory, naive, mad,
spurious, unbridled, unfounded,
unrealistic, unsubstantiated; vain;
fallible; lit airy, lit beguiling
Magn['desire'] + AntiVer: wild
Adv 1 with [the -1 [He attended the
ball with the hope of meeting her]//
hopefully1 H. in
nonAdv1 without [-) [The little children singular
wandered in the dark forest withc,ut
hope; He continued his work, but
without hope of completing it]//hope-
lessly 1
Propt out of [- for N = Y], in [- of N/Vger1,
in [-s of V er1 I C2 ~ A
[She cherisffed his picture out of hope for his
return; He attended the ball in hope
of meeting her; She wrote to him every
week in hopes of .his returning]
lncepPredPlus build up 1, expand 1, go up, grow,
increase 1, rise, warm 1 [The rescuers'
hope grew (rose> when they heard that
the storm was subsiding]
ride [on NIN denotes the cause of H.
[Our hopes ride on his recent achievement]
in order to make X
perform P, CausFunc 1 dangle [ART - before N = X] I
H. in singular ·
be [that PROP< for V er/N)) IN
denotes an action [OuF only hope was
that if someone was there, he would be
afraid of us]
F2 = the realization
of X's H. depends on Z (X's -1 lies [on (in) NJ

SIL-UND Workpapers 1987

 119

[Hassan's hopes now lay on this little

ridge] hangs [on N = Z], hinges [on
Vger/N = Z] IN denotes an action
Caus 1 F2 . fix, lay, set, fasten, hang,
pin [Aposs - to N = Z]
attach [A oss - to N = Z]
Caus 1 (H. is based on Z) : base, buiid, rest [ART -
on N = Z)
Real 1 fulfill, realize, reach, attain
[ART -]
Real 2 satisfy, meet [ART-], meet [with
ART -1
Fact 0 . come true, materialize, be realized
Ant1Fact 0 . come to nothing, fail 3 [= 'prove
misleading'] [The farmer's hope came to
nothing when he was unable to develop a
rust-resistant species of wheat]
CausFact 0 advance, further [0/ART -]
nonPermFact 0 : disappoint, frustrate [0/ART -]
Magn + A1 caus 1Manif : bright, radiant, starry-eyed,
wide-eyed [with-] I H. in singular
Caus1Manif display, exude, show [0/ART -]
by saying, Caus 1Manif : express [0/ART -]
nonPerm1Manif : hide, conceal [ART -1
A2nonPerm1Mani 2
Adv 1caus 1Manif nonFunc 0
: secret, silent
// hopelessly 2
[He shook his head hopelessly)
Excesstrem(body)-Sympt12: tremble [with-]
Excessfulg(eyes H. in singular
face)-Sympt 12 beam, glow [with-]
CausDegrad corrupt, spoil [ART-]
situation P being
such that there is
no H. that P will
change for the
better // hopelessly 3 P [He is hopelessly
ill]
realize that Y of
X's H. may take
place later than X
expected defer [0/ART -1 [Hope deferred
makes the heart sick]
Caus1ContFuncto : cherish, harbour, nurse, sustain 1
[= 'keep hoping continuously'] [ART-]
lc 2 ., A [their generation nursed
the hope of a promised land]
Caus 1 contFunc 0 is a
spontaneous human
characteristic : - springs eternal
FinFunc 0 collapse, die, disappear, end 1, fail 2
[= 'cease'], perish, vanish [Hope for a

SIL-UND Workpapers 1987

 120

lasting peace collapsed with the

renewed fighting]
PerfFinFunc 0 be gone, be finished, be extinct,
be at an end; be dead [Our hope for
life on Mars is dead]; lit or
arch be done
s 1 causFunc 0 : source [of -1
Figurs 1 causFunc 0 : star, beacon, well-spring,
bastion [of -1 I H. in singular
being caused by W,
IncepFuncO/l arise, be born [in N < N's soul,
heart, breast)]
ContFunco/1 linger, persist, remain [in N/among
Ns]
A1ContFunco/l enduring, lasting, persistent,
steadfast, bibl unfading
CausContFuncO/l feed, keep alive, nourish, sustain
2 [= 'give strength to hopes'] [Good
news from the front fed the hopes of
the general]
LiquFunc 011 crush, demolish, destroy, wreck
[<IJ/ART -]; cut ofi [A oss -/N
from-]; blast, bligh~, break,
dash, end 2, extinguish, kill, ruin,
shatter, take away, trample down [ART -1
[The mother's hopes were shattered by
the death of her young son]; bury,
drown [Q}/ART -1 I M1 denotes a fact
there be [in N's soul (heart,
breast, ••• )]
lncepFunc1 spring [in N's soul (heart, breast,
••• >1
again IncepFunc 1 return [to NJ, revive 1 [in N]
[Hope revived in her]
nonPermContFunc 1 leave [N little (no)-] [It
leaves him no hope of escape]
FinFunc 1 abandon [NJ
CausFunc 1 arouse, awaken[-]; raise [The
serious negotiations raise hopes for an
early end to the strike]; stir up
[Q}/ART - in-NJ; engender, instill,
plant [QJ/ART - in NJ I H. in singular;
give, hold out, offer [QJ/ART - to NJ I
H. in singular
again CausFunc 1 : restore, revive 2, reawaken, rekindle
[Q}/ART -1 [Warm weather restored
the hope of the farmers]
live disregarding
adverse facts or
material necessities
as if the H. of X

SIL-UND Workpapers 1987

 121

constituted reality
or a means of X's
sustenance : live [on-] (Despite a fatal diagnosis,
the young man continued to live on hope;
After all food was consumed, the explor-
ers lived on hope)

Examples

Man is a victim of dope, in the incurable form of hope. (O.Nash) It

was evident to everyone that a strong sense of hope was the guiding
power in her life. A light of hope shone in their eyes when they heard
the good news. Rising popularity in pre-election polls gave the new
party a ray of hope. The discovery of penicillin opened a door of hope
for the wounded. The children were full of hope after visiting Santa
Claus. The young man was devoid of hope when he learned the sad fate
of his brother. She was suddenly killed while at the pinnacle of hope.
To feed the poor was the dear hope of the good monk. An ardent hope of
victory is what encouraged the competitors to do their best. Inspired
by certain hope of success, the hunter continued his search. They had
but faint hope that the buried miners would be found alive. The
children had boundless hopes that they would excel in the examination.
The common hope was for a bountiful harvest. Their best hopes were
unmercifully shattered by the death of their son. Throughout their
sinful lives, they were inspired by the sweet hope of redemption. Only
substantial hopes should be taken into consideration by the committee.
He was always preoccupied with empty hopes. Driven to extremes by her
wild hopes, she finally committed suicide. When the soldier heard of
the great victory at the front, his hopes for an end to the war went up.
The woman's budding hopes grew stronger when she received the happy
letter from her friend. The silly man indulged the hope that he could
write a novel in three days. The sick man gained hope when he learned
of the new medicine that might cure his illness. The workers found new
hope in their revolution. The young child clung to the hope that her
dead father would come back to life. Hope for a solution to the
problem must be deferred until better conditions prevail. When the
radio finally went silent, the operators in the control tower lost all
hope. A happy turn of events inspired the men with hope. Stormy
weather cheated the fishermen of their hope for a large catch. Our
hopes now lie in the skill of the surgeon. The great drought left them
little hope for a good crop. Once the oil slick was found on the
surface of the water, all hope for the submarine was gone. Hope for a
new life was born in John's soul after he read the Bible. Despite
initial losses, hope for an early victory lingered. Winning the hand
of a certain lady was the romantic gentleman's steadfast hope. The
long drought destroyed the hope of the farmer for a good harvest. Lack
of financial resources cut off his hopes. The young lover walked down
the path, hope springing in his heart. After the war, hope for
economic progress returned. The serious negotiations stirred hopes for
an early end to the strike. Let us instil in our children hope for a
better world. The physician could not hold out much hope to his

SIL-UND Workpapers 1987

 122

patient. The coach's hopes for his team rode on their recent
achievement. The Minister of Housing dangled the hope of interest-free
loans before builders. The young man pinned his hope (of quick success
(for this marriage)) on finding steady employment. The newly elected
Prime Minister was a source of hope to many citizens. Medicare was
their beacon of hope. Do not hang your hopes on impossible schemes if
you wish to avoid disappointment. Despite the dense fog,_ the explorers
never gave up hope of finding their way back to their camp. The young
bride clung to the hope that her spouse would return from the war alive.
Unconquerable hope guided the romantic poet. In composing a great
symphony, the musician realized a boyhood hope. The farmer's hopes
came true after he developed a species of wheat that was resistant to
rust. Development of the electrical automobile advanced the car
manufacturers' hopes for success. Adverse economic conditions
frustrated hopes of achieving a low rate of unemployment. She was
radiant with hope after being nominated a candidate for office. An
optimist will display hope even in adversity. The fisherman expressed
hope for a good catch. The villain concealed a hope that the plane
would crash with his mother on board. The silent hope of the innocent
victim was her ultimate vindication. When the evangelist touched the
sick woman, she trembled with hope. The children were glowing with
hope on Christmas Eve. No adversity will corrupt our hope.

Roa:es

I am happy to express my heartfelt gratitude to Lidija Iordanskaja,

who, as always, was the first to read the text of my paper and suggest
many important corrections and improvements, and to Peter Constable,
Ingrid Meyer and James Steele, who kindly accepted the task of revising
the English of the manuscript.

A part of this paper was presented as key-note address at·the 1985

annual meeting of the Linguistic Association of Canada and the United
States (LACUS), in Saskatoon, Saskachewan, Canada.

1. We make use of the distinction available in German:

sprachlich 'pertaining to or related to language',

sprachwissenschaftlich 'pertaining to or related
to linguistics'

We use the English word 'linguistics', therefore, only when the second
meaning is intended, and the German sprachlich when the first meaning is
intended.

2. The primacy of the speaker's viewpoint for linguistics is so

fundamental an issue that it seems recommendable to analyze here an
interesting argument against it, advanced by Thomas R. Hofmann. The
argument runs approximately as follows:
Within the framework of "good" communication, the speaker has
to make sure that he is properly understood; and to achieve

SIL-UND Workpapers 1987

 123

this, he must put himself in the addressee's shoes, that is,

try to decode all his own utterances and see whether they can
be interpreted the way he intends them to be. Therefore, the
speaker must use listener's device as well, with the result
that the task of speaking is harder than that of
understanding.
The reasoning is quite sound -- but it misses the point. Certainly, the
speaker has to use a listening(= understanding) device, but it by no
means follows from this fact that the speaking device is more complex
than the listening one. The task of a speaker in a communication act
may be harder than that of his listener; we do not, however, compare
communication tasks, but rather mechanisms correlating meanings and
texts (leaving aside mechanisms for constructing meanings, for
backtracking what has been said, etc. -- mechanisms that are extremely
important in .. sprachlich" communication but do not pertain to language
and are, consequently, of no interest for linguistics).

3. The logico-semantic concepts predicate, place (of a predicate) and

argument are central to the present discussion, but we are in no
position to introduce them here. We assume our readers' familiarity
with them and limit ourselves to a warning against confusing predicate
in logico-semantic sense(= 'a meaning having obligatory slots for other
meanings') with the grammatical term predicate (= 'main verb').
Likewise, place(= 'a slot of a predicate') and argument (= 'meaning
filling the slot of a predicate') should not be taken in their everyday
sense.

4. For lack of space, we cannot even touch here upon the fundamental
problem of semantic primitives.

5. Once again, we are not in a position to substantiate here the use of

purely distinctive(= meaningless) symbols instead of meaningful labels
on the arcs.

6. Interesting examples of values of the LF Nolt in English can be

found, most unexpectedly, in The Book of Lists, by D. Wallechinsky, I.
Wallace and A. Wallace, New York: w. Morrow, 1977, page 135:

A murder of crows
clowder cats
leap leopards
sloth bears
raffer turkey
smack jellyfish
skulk foxes
labor moles
crash rhinoceros
siege herons
rag colts
drift hogs
charm finches

SIL-UND Workpapers 1987

 124

trip goats
knot toads
parliament owls
troop kangaroos
gaggle geese
pride lions
muster peacocks

"Although not frequently heard in conversation, these terms are

fully correct and appropriate ways of describing the animal
listed" (ibid.).

7. A very good illustration of values of the LF Real (and .AntiR,eal) in

English is offered in Montreal's The Gazette, Sept. 28, 1985, in a
letter to Ann Landers, which we reproduce here:

Dear Ann:
I discovered this item stuck in an old book.
It was written in longhand and there is no sign of an
author. I hope you will print it.
I found it highly inspirational.
- T. D.
Dear T.D.:
So did I. Here it is:

What is life?

Life is a challenge ••• meet it.

Life is a gift ••• accept it.
Life is an adventure ••• dare it.
Life is a sorrow••• overcome it.
Life is a tragedy ••• face it.
Life is a duty••• perform it.
Life is a game••• play it.
Life is a mystery ••• unfold it.
Life is a song ••• sing it.
Life is an opportunity ••• take it.
Life is a journey ••• complete it.
Life is a promise ••• fulfill it.
Life is a beauty••• praise it.
Life is a struggle ••• fight it.
Life is a goal••• achieve it.
Life is a puzzle••• solve it.

References

Bloomfield, Leonard. 1933. Language. New York: Holt.

SIL-UND Workpapers 1987

 125

Chomsky, Noam. 1966. Topics in the theory of generative grammar. The

Hague-Paris: Mouton •

• 1968. Language and mind. New York etc.: Harcourt, Brace and
---Jovanovich.
Lakoff, ~obin. 1973. Lexicography and generative grammar. II: context
and connotation in the dictionary. In: R. I. McDavid and A. R.
Duckert (eds.), Lexicography in English [Annals of the New York
Academy of Sciences, 211], New York, 154-164.

Lamb, Sydney. 1966. Outline of Stratificational Grammar. Washington:

Georgetown University Press.

Mel'~uk, Igor A. 1981. Meaning-Text Models: A Recent Trend in Soviet

Linguistics. Annual Review of Anthropology, 10, 27~62.

---,
, with Nadia Arbatchevsky-Jumarie; L~o Elnitsky; Lidija
Iordanskaja; and Adele Lessard. 1984. Dictionnaire explicatif et
combinatoire du fran~ais contemporain. Recherches
lexico-senantiques I. Montreal: Les Presses de l'Universite de
Montreal.

and Zholkovsky, Alexander K. 1984. Tolkovo-kombinatornyj slovar'

sovremennogo russkogo jazyka. Opyty semantiko~sintaksi~eskogo
opisanija russkoj leksiki [Explanatory combinatorial dictionary of
modern russian]. Vienna: Wiener Slawistischer Almanach.

Nakhimovsky, Alexander. 1983. Meaning-text linguistics and the problem

of voice. Carbondale-Edmonton: Linguistic Research.

Sanders, Gerald A. 1972. Equational grammar. The Hague-Paris: Mouton.

Sgall, Petr. 1967. Generativnt popis jazyka a fezk& deklinace. Praha:

Academia.

Testen, David; Veena Mishra; and Joseph Drago (eds.) 1984. Papers from
the parasession on lexical semantics. Chicago: Chicago Linguistic
Society.

Wierzbicka, Anna. 1972. Semantic Primitives. Frankfurt/M.: Athen~um •

• 1980. Lingua Mentalis: The Semantics of Natural Language. Sydney

---etc.: Academic Press.

Zolkovskij, Aleksandr K., and Igor A. Mel'~uk. 1965. 0 vozmo!nom metode

i instrumentax semanti~eskogo sinteza [On a Possible Method and
Instruments for Semantic Synthesis (of Texts)]. Nau~no-texni~eskaja
informacija, No 6, 23-28.

SIL-UND Workpapers 1987