You are on page 1of 37

4nn.

Rev AnihmpoL 1981 ia27-62 Copynght'S- 1981 by Annual Reviews Inc Ail rights reserved

MEANING-TEXT MODELS: A R e c e n t TTend in Soviet Linguistics
Igor A. MeVcuk
Departemetit de lmguistique, Umversite de Montreal, Montreal H3C 3J7, Quebec, Canada

•9670

SOME HISTORY AND GENERAL REMARKS
The Meaning-Text Theory (henceforth MTT) was put forward by Alexander K Zholkovsky' and the present wnter in 1965 (86), 2 years later, a major presentation of the theory appeared (87) and was soon translated into Enghsh and then into French Shortly afterwards, Junj D Apresjan joined us, and thus was formed the nucleus of what was called the Moscow Semantic Circle Over a 10-year penod, some 20 people contributed to the work on a Meaning-Text Model of Russian Basic g^ieral readings, in addition to the two titles just mentioned, include (45,47, 50-52, 56,67, 88) A number of papers and books dealing with more specific topics will be indicated later MTT did not flounsh officially m the USSR Its adherents were only rarely and reluctantly admitted to conferences and colloquia, it was never taught at any of the major umversities; and the leading professional journals did not accept papers on MTT for publication The main reasons were that it was so un-Marxist, with its stress on establishing formal correspondences
'There are problems m transcnbmg Russian names into English For example, Zholkovsky's name will be wntten ^olfcovsky whtn it is transhterat«d du'ectly from Cynlhc type, moreover, vanous erraac spdlings are found in American publications, e g Zholkovsku, or ^olkovsky, which had to be adopted in the literature cited The same applies to Apresjui (sometimes spelled Apresyan), atid my own name {MeVduk-Melchuk-MeVchuk, etc) 27

28

MEL'CUK

between meanmg and sound (thus separatmg both), its extensive use of mathematical and quasi-mathematical apparatus, its heavy rehance on Western hngiustics, and so on Later, the situation detenorated sharply when the founders of the trend participated in a letter campaign protestmg the Sinyavsky-Daniel tnal and then failed to take the "nght" stand toward the Soviet invasion of Czekoslovakia Eventually I was forced to leave the USSR forever (the mterested reader can find more details in the Bntish review Survey, 1977-78,23 2,126-40,1979,24 2,213-14) Two years later, Zholkovsky followed me At present, there remain a few supporters of the Meanmg-Text theory grouped around Apresjan m Moscow My personal hope IS that the theory will recover from its transplantation to N o n h Amenca and perhaps be ennched through dn-ect contact with Amencan linguistics Conceived and developed as a general theory of human language, Meanmg-Text Theory is based on the following two postulates Postulate 1 Every speech event presupposes three mam components {a) content or pieces of information to be commumcated, which are called meaning(s), (b) certam forms or physical phenomena to be perceived, which are called text(s), (c) a many-to-many correspondence between an infinite set of meanings and an lnfimte set of texts, which constitutes language proper (or "language in the narrow sense of the term") This postulate can be diagrammed as follows 1 - l a n g u a g e proper , ^ MEANING j ^ j - < =^-{_TEXT V LANGUAGE A natural language is thus viewed as a logical device which estabhshes the correspondence between the infinite set of all possible meanmgs and the infinite set of all possible texts and vice versa This device ensures the construction of lmguistic utterances which express a given meanmg, l e speakmg, and the comprehension of possible meanmgs express^ by a given utterance, l e the understandmg of speech Postulate 2 Hypotheses about devices of the type illustrated in Postulate 1 can be formulated as functional^ or cybernetic models, with the actual
^The lenn functional, which is now a buzz word in Imguistics, as used here has nothmg to do with functional sentence perspective (topic-comment problems) nor with grammatical functions (such as "subject of," "object of," etc) We say that X is a functional or cybernetic model of an object Y if and only ^ X is a system of rules simulatmg the functioning, the "work," or the behavior of Y, with no claims as to its observable structure

MEANING-TEXT MODELS

29

language considered a "black box" where only the inputs and outputs can be observed but not the intemal structure Such modek (called MTM) are systems of rules approximating the Meaning 4=4 Text correspondence This chapter will discuss only one specific class of MTMs Proceedmg from Postulate 1 and 2, the MTM can be characterized by the foUowmg important properties First, the MTM is not a generative but rather a translative or purely transformational system It does not seek to generate (enumerate, specify) all and only grammatically correct or meanin^ul texts, but merely to match any given meamng with all texts having this meaning (synonymy) and, conversely, any given text with all meanings the text can have (homonymy) Second, the MTM is no more than a fragment of the full-fledged model of human lmguistic behavior

Reality r ^ ^ l I

M e a n i n g s > ^ = ^ ^ Texts V <=^ < L i n g u i s t i c Sounds II III

Only Fragment II, i c Language, is deemed to be the subject of linguistics proper and should be represented as an MTM Fragment I is the subject of vanous fields such as philosophy and psychology, including what is called artificial lnteUigence Fragment III is the subject of acoustic and articulatory phonetics However, even Fragment II is not represented by MTM in full To simplify our task, we made abstraction from a number of relevant aspects, properties, and phenomena of natural language—m the hope that we will obtain a clearer and more insightful picture of what remams under our lenses This is the only justification for many dehberate omissions In their actual form, MTMs observe the foUowmg six limitations (l) Functions of natural language other than the communicative one are not considered at all Language is treated exclusively as a commumcation system (ll) The extremely important problem of language acquisition and development IS deliberately ignored (ill) No attonpts have been made so far to relate the MTM expenmentally with psychological or neurological reality An MTM is no more than a model, or a handy logical means for descnbmg observable correspondences (lv) The correspondaice between meanings and texts is presented statically, 1 e M a correspondence between some elementary fragments

discrete manner. ctmventional spellmg systems—have existed for a long time The approximation they ensure for the continuous acoustic flow of speech IS generally d^med satisfactory Though no semantic transcnption exists for meanmgs. anomahes of a text but it does not deal with the semantic ones (This is done on purpose in order to reflect the essential asymmetry of meanmgs and texts) Only a very sketchy descnption of the MTM can be given I do not mention any other feasible solutions and draw no parallels with the closely related models proposed by transformational grammar. and other more recent developments. this is more or less obviom phonetic tratiscnptions—or. etc ) Thus. or grammatical. case grammar. or Russian ) sentences So it should be able to detect the formal. a qualified reader will easily recognize some common pomts and perceive the differences UTTERANCE REPRESENTATION LEVELS IN THE MEANING-TEXT MODEL I assume that both meamngs and texts can be represented m an exphcit. absurdities. lndudmg relational grammar Suffice it to state that all these have significantly influenced the author. generative semantics. using formal languages devised specially for this purpose As for texts.30 MEL'CUK of meaning and equally elementary fragments of text The procedures for actually movmg from complex meamngs to texts and vice versa. nothmg should prevent us from believmg that such a tran%ut a MT model does of course enumerate the possible kmds of changes between adjacent levels . are left out of consideration because they are believed to he outside the domam of lmgmstics ^ For this reason. within the MTM the problem of rule ordenng does not anse (v) The possible feedback between meamngs and texts in the actual process of speakmg or hstemng (changes m the ongmal message under the lnflu^ice of the text already constructed and put out. etc) IS not taken mto account (vi) The anal^is of the meaning itself go« beyond the scope of an MTM. for that matter. tnvialities. or the operations that have to put together those elementary fragments to produce actual hfe-size representations. a different type of device is needed for this purpose (By this I mean the detection of semantic anomalies such as contradictions. stratificational hnguistics." the MTM should provide for them perfect Enghsh (or French. if confronted with such deviant meamngs as "Colorless green ideas sleep furiously" or "John ate up the sincenty of his car.

if only as a workmg hypothesis A version of semantic transcnption. some of the paraphrases will be sifted out by selectional restnctions and other constraints. however. Formula I can be rewntten as 3 \ MTV # V\OT. The FDA has addressed a stem caution to ladies to the effect that while expecting a baby they should abstain from coffee. ^'epresertatio") . } • < = > ^ or "^CeBresertatior) V From now on. I speak only of representations One of the most basic facts about natural language is the following 4 A given meaning of sufficient complexity can normally be expressed by an astronomically large number of texts Consider sentence 5. or. as we call it. each of which will be represented by one column in the chart." "real" meanings and texts Therefore. the number of English sentences synonymous with 5 and 6 runs into thousands * ^ 0 see how thousands of [»iraplirases for 5 and 6 can be produced. a native speaker can easily think of new vanants.:antic. and pleasures of life. all the other words have been changed. consider the followmg Sentence 5 can be broken into roughly eight semantic fragments.. each column contains some (nearly) synonymous variants that express the fragment in question (the figure in boldface under a column represents the number of variants in it) Any variant from one column combines with nearly any other variant from another column (i. so it becomes evident that synonymy is really nch . tbat belongs to life's small joys (ii) The FDA made public a strong warning addressed to mothers-to-be they should not indulge m consuming coffee. one of the small pleasures of life Except for the function words (articles. coffee. is propo^d within the framework of MTT and is used herein Thus MTM deals with formal representations of meanings and texts rather than with "genume.MEANING-TEXT MODELS 31 scnption IS possible and from trying to construct one. 1980) 5 The Food and Drug Administration has seriously cautioned expectant mothers to avoid one of life's stmple pleasures a cup of coffee The meaning of that sentence can be expressed by 6 as well 6 Pregnant women have been earnestly warned by the FDA against drinktng coffee. (• e t l c '• 'I 3err. quoted (with slight modifications) from Ne-wsweek magazine (Sept 15. semantic language. etc rhis allows us simply to multiply the number of vanants m each column I X 4 X 9 X 7 X 6 X6> < 3 X 8 = 217. page 32. auxilianes ) and such technical terms as FDA.728 True. as has the overall structure Yet the meaning is preserved intact With additional vanations. one of the simple pleasures of life.

Surface-Morphological Representation (SMorphR). or the meamng. _. and a great many different texts have to be reduced to the same meamng representation This makes it almost impossible to establish a direct correspondence between semantic and phonological representation. Deep-Morphological Representation (DMorphR). a deep one. Deep-Syntactic Repres^tation (DSyntR). . ^Continued from page 31 1 FDA 2 seriously earnestly sternly strongly 3 4 5 to avoid to abstam fiom sh uld f^^°^^ V abstain caution warn forewarn counsel put on guard issue a warning address a caution make public /•a caution Va wammg pregnant women expectant mothers mothers-to-be during pregnancy women/ while ladies expecting a baby not to should not 6 7 8 smail 1 rpleasuresl simpl^ \joys / of life/ hfe*s coffee . geared to meamng. and so two intermediary levels of utterance representation have to be introduced— syntactic and morphological—^the former aimed at the sentence as a structural object and the latter deahng with the word All levels—with the exception of the semantic one—are spht into two sublevels. determined by physical form This gives a total of seven representation levels 1 2.rdrink(u^) 1 „ indulge in) { T ^ f coffee Vconsum<ing)J one of that belongs to that constitutes njp of coffee . the mastery of a language boils down to the control of its synonymic means The MTM has to match a given meamng with many different texts. Surface-Syntactic Representation (SSyntR).32 MEVCUK Natural language is a system capfU^le of producing a great many synonymous texts for a given meamng. 3 4 5 Semantic Representation (SemR). and a surface one. .

C2-3]. or the text A repres^itation is a set of formal objects called STRUCTURES. B.TEXT MODELS 33 6 Deep-Phonetic Representation (DPhonR. one of which IS distinguished as the mam one. etc (See section on Design of the MTM) FoUowmg are examples of successive representations for sentence 5 and some synonymous sentences. 7 Surface-Phtmetic Representation (SPhonR. which is called in the literature simply "phonetic represoitation"). m this way the Enghsh present perfect is represented] Let It be emphasized that Figure 1 does not lay claim to absolute semantic pr^ision Many flaws exist in the semantic language used. the FDA beheves that drmkmg coffee is dangerous for pregnant women [B2] This attempt took place bdbre the moment of speech yet its results persist until this moment [A2-3. 2. starting from the meaning and working upward toward the text Semantic Representation A sample Sem(antic) R(epresentation) simphfied for illustrative purposes IS given m Figure 1 Notice that the letters A. C and the numbers 1. with all the others spectfymg some of Its charactenstics Each structure depicts a certam aspect of the item considered at a given level Utterance representations are wntten m formalized languages d^ned by the researcher We can now present Formula 3 in full The top line in 7 is a sequence of the utterance representations of all seven levels. with the correspondences between any two adjacent levels shown by two-headed arrows The bottom line shows the components of the MTM and their functions Thus semantics provides for the correspondence between the semantic representation of an utterance and all the sequences of deep-syntactic representations carrying the same meaning. 3 that appear at the left side and top of the diagram are not part of the representation they are used merely for refemng to locations m the diagram The shaggy hnes mark the boundanes of theme and rheme (cf below) This SemR can be worded approximately as follows 8 The FDA has mtensively attempted to commumcate [A 1-2] to all pregnant women [B2] that they should not [A2] dnnk coffee. one of the simple pleasures of life for them [A2-3. but hopefully it serves the purposes of illustration .MEANING. or what is commonly called "phonemic representation").

a SemS is a connected graph or a network The vertices or nodes of a SemS are labeled with semantic units. the SemS tnes to depict the meaning objectively—leaving out the speaker and his intentions. complex meamngs consist of semes or less complex semantemes A complex semanteme can be represented by a semantic network.34 MEL'CUK Figure J 8 IS the SemR of 5 and 6 as wdt as of all the other sentences {or sequences of sentences) synonymous with them The SemR of an utterance consists of two structures the semantic structure and the semantico-communicative structure The Sem(antic) Structure} speafies the meaning of the utterance independent of Its linguistic form The distnbution of meaning among the words. clauses. which specifies its semantic decomposition For example. the semanteme "earher" found in 8. or sentences is ignored. so are such linguistic features as the selection of specific syntactic constructions and so on At the same time. which are taken into account in the second structure of the SemR Formally. can be decomposed as follows . or SEMANTEMES meanmgs which can be either elementary (SEMES) or complex. A3.

" "not. but only a functor can head an arrow The arrows on the arcs point from functors to their arguments The arcs or arrows of a SemS are labeled with numbers which have no meaning of their own but only serve to differentiate the vanous arguments of the same functor For instance. complex semantemes are used for the most part The names of the semantemes are word senses picked from a good dictionary For the illustration here I have taken English word senses from The American Heritage Illustrated Dictionary of the English Language. "space. until semes are reached A d^nitive list of semes. ' corniin ca te ' means that A is the first argument of "communicate" (who communicates). or semantic primitives." "time." "more. LOGICAL CONNECTIVES ("if." "say. etc). events. properties. "there exist". states. further subdivided into PREDICATES (relations." "this speech event. ." "set" (in the mathematical sense)." '*and. 79a) where 13 semantic primitives are proposed and argued for ] Since a complete analysis into semes would make the semantic network unreadable. but some likely candidates are "(some)thing. plus all numbers). lncludmg proper names Both types of semantemes can receive arcs or arrows. as deep as we need or want it. "not"). 2 NAMES (OF CLASSES) OF OBJECTS." "or " [For a different view of semantic pnmitives see Wierzbicka (79. and QUANTIFIERS ("all". actions. "and". is not available. thus "persist 3" m A3 IS "continue in existence until " rather than "to be obstinately insistent in " or "hold steadfastly to despite obstacles " Two major classes of semantemes are distinguished 1 FUNCTORS. "or".MEANING-TEXT MODELS 'Tionent 1 ' 35 Semantic decomposition can go on.

l e communicated by the speaker (c) Foregrounded (expressed as a main predication) vs backgroimded (relegated to an attnbute) (d) Emphatically stressed vs neutral Semantico-communicative information stands in approximately the same relationship to the semantic network (SemS) as do suprasegmental prosodic phenomena to the segmental phonemic stnng that makes up a sentence In . or given (known to both interlocutors). 1 and U of second order can be indicated 1 2 = "the FDA." d 2 ~ "senously cautioned pregnant women " (b) Old. its source. (a) Theme (topic." and U is that "pregnant women should not dnnk coffee smce that is dangerous for them " If we mterchange the symbols 1 and H . as opposed to what is commumcated about the topic In 8. who IS aware of B."A commumcates B to C" = "A. vs new. which is up to the speaker to choose Note that there can be different layers of topic vs comment division For example. and C the third argument (to whom the information is passed) The exact role of each argument is specified by further decomposition of the functor.1 is "the caution given by the FDA and aimed at pregnant women. the message becomes different 9 Mothers-to-be had better keep clear of coffee. l e the starting pomt of the utterance. e g. within 1 m 8. designated 1 ) vs rheme (comment.36 MEL'CUK B Its second argument (what is commumcated). designated d ). which can be harmful to them—that ts the stern warning issued by the FDA 1 and d indicate the ltmerary through the situation. on "aware" and "cause") would reveal more subtle links between the functor "communicate" and its arguments The Sem(antico)-Comm(unicative)S(tructure) specifies the intentions of the speaker with respect to the orgamzation of the message The same meanmg reflecting a given situation can be encoded m different messages according to what the speaker wants So the Sem CommS must show at least the following contrasts. exphcitly causes C to become aware of B " A deeper decomposition (bearmg.

and that on the first stratum only For more details on the SemR see (52. the deep-syntactico-communica- II Magn Figure 2 11 is the DSyntR of sentence 5 . which is one of life''s small pleasures. they should not drink coffee. Both sentences have of course the same SemR. I will give two different DSyntR^ for two different but fully synonymous sentences. the one shown in Figure 1 Their DSyntRs appear as Figures 2 and 3. pp 53-77.MEANING-TEXT MODELS 37 the simplified examples given here only the topic-comment contrast is shown. respectively A D(eep)-Synt(actic)R(epr»entation) [of a sentence] consists of four structures the deep-syntactic structure. 56) Deep-Syntactic Representation To make the contrast between a SemR and a D(eep)~Synt(actic) R(epresentation) more vivid. namely. 5 and 10 10 The FDA has issued a stern warning to pregnant women.

are not represented thus in 11 has [cautioned]. pp 3-21) It represents the syntactic organization of the sentence m terms of its constituent words and relationships between them A node of a DSyntS is labeled with a generalized lexeme of the langu^e A generalized lexeme is one of the following four items 1) A full lexeme of the language Semantically empty words. the deep-syntactico-anaphonc structure. but this kmd of umt has been widely discussed m Amencan Imguistics) . and the deep-syntactico-prosodic structure The D(eep)-Synt(actic)S(tructure) is a dependency tree (see 58. yet nonexistent (There are no fictive lexemes in my examples.38 MEL'CUK 12 Magn ^-R NAP Figure S 12 is the DSyntR of sentence 10 tive structure. l e a lexeme presupposed by the symmetry of the denvational system. to [avoid] and [cup] of [cojffee] are absent 2) A fictive lexeme. bke (strongly) governed prepositKms and conjunctions or auxiliary verbs.

Mngj^behef) = staunch ["very". = defeat [antonym]. which means that f( W) is phraseologically bound by W In the MTM. bemg respectively the subject (for Reali) or the object (Real2)" of X. are not shown in a DSyntS ) A (standard elementary) LEXICAL FUNCTION (LF) f is. bad. to shoot) = to shell. Opetiifavor) = to do ["be the subject o f ] . include) = to belong to [conversive. Magn (to illustrate) = vividly.MEANING-TEXT MODELS 39 3) A multilexemic idiom. tense. Oiperiianalysis) = to perform. the other example is expectant mother 4) A lexical function. to make good. Ao(xtt«) = solar [denved substantival/adjectival]. urgent. O^Tx{step) = to take. roughly speaking. both f{W^) and f( W^) bear an identical relationship with respect to the meaning and syntactic role to W^ and W^. Opvxjianalysis) = to undergo. respectively ib) in most cases. e g hit ifoff "h&ve good rappon" or pull a fast one on someone "gam an advantage over sa unsuspectmg person by subterfuge " I n 11 and 12 we find two examples of idioms one is The Food and Drug Administration. to machine-gun [narrower synonym]. . which is the argument. Oper2(con- trol) ~ to be under ["be the object of]. and aspect in verbs (Syntactically conditioned morphological values. O^r^attention) = to receive. M»gfk(need) = great.see below The symbol of a generahzed lexeme must be subscnbed for all the meaning-beanng morpholo^cal valu^. an lntensifier]. about 50 standard elementary LFs are used Some examples of LFs Syn(ro shoot) — to fire [synonym]. ^oriened to FDA and represented by a single node. as in This paradigm includes the locative case = The locative case belongs to this paradigm]. such as person and number m verbs. if f(ff") and t{W^) exist. M&gp{settled [area]) = thickly. l^eaiiipromise) — to keep. f( W^) ^ f( W^). So(to despise) = contempt. Op&[i(attention = to pay. such as number m nouns or mood. a relation which connects a word or phrase W—the argument of f—with a set f( (K) of other words or phrases—the value off—in such a way that fa) for any W^ and ff^. R&d^attack) = to fall to ["fulfill the requirements of X.

cf 38 also). Operi{waming) — to issue. 2nd. etc. In 11 we see a Magn(ro caution). linking any one <^ these elements to the top node (mam verb) of the correspondmg clause Let It be emphasized that there is no linear order of nodes withm the DSyntS Word order is taken to be a means for encoding syntactic structure into speech stnngs and therefore it is banned from the syntactic structure The D(eep)-Synt(actico-)Comm(unicative)S(tructure) specifies the divi- . A branch of a DSyntS is labeled with the name of a DEEP-SYNTACTIC RELATION (DSyntRel) The DSyntRels are deemed to be umversal A DSyntRel is one of nine binary relations > 1. 6th arguments. which covers all kmds of modifiers and attributes (m the broadest sense of the term). e g JOHN PETER AND 2 PAUL COORD COORD • APPEND IS an "appendancy" relation that subsumes all parentheticals. I will not touch on so-called NON-STANDARD LFs LFs play a crucial role m covenng restncted lexical cooccurrence. to moo.40 MEL'CUK Son{cow) ~ to low. there are COMPLEX LFs. lnteriections. 67. Son{windowpanes) — to jingle. and Mafftiwaming) — stem For more details about LFs see (8. AnHReak^attack) — to beat back. to rattle [typical sound] Furthermore. which is the governor. 42. YinO^eri^control) ~ to get out of. e g AntiRa^^romise) — to renege on. respectively. m 12 there are three LFs S^to warn) = warning. for instance13 SELL I who''] seller [for how much''] merchandise • ATTR IS the attributive relation. • COORD IS a relation that accounts for all coordinate or conpmed constructions hnkmg the nght conjunct to th^ left one. 6 are predicative relations connectmg a semantically predicative lexeme with its 1st. 2. etc To avoid overburdenmg the exposition. JncefOferiiflre) = U) open. adresses. l e seriously.

each descnbing a particular syntactic construction The inventory of SSyntRels for a language is estabhshed empincally(cf58. which serves to express this hierarchy . etc In our sunpbfied examples the DSynt-CommS shows the first stratum of topic-comment division only m much the same manner as withm the SemR I will not discuss the differences between the DSynt-CommS and SemCommS any further The D(eep)-Synt(actico-)Anaph(onc)S(tructure) cames the information about coreferentiahty cf the broken-line arrows in 11 and 12 that is. and for some other languages m references given in (58. all the idioms are expanded mto actual surface trees Third. pp 221-35). see below) and spelled out m place of the LFs Fourth. In 11 and 12 it is the acronym NAP. the understood subject of avoid is the expectant mothers who have been cautioned. standing for N(eutral) A(sertive) P(rosody). the values of all the lexical functions are computed (on the basis of a lexicon called an Explanatory Combmatonal Dictionary. old and new. pauses. that exemplifies the DSynt-ProsS Surface-Syntactic Representation Only sentence 5 will be represented at the surface-syntactic level (see Figure 4) The S(urface)-Synt(actic)R(epresentation) of a sentence consists of four structures correspondmg to those of the DSyntR but replacmg D(eep) by S(urface) The SSyntS is also a dependency tree. all the lexemes of the sentence are represented.pp 91-150) In 14 the reader can see examples of Engbsh SSyntRels Tentative lists of SSyntRels are found for Russian m (5) and (52. prosodies that appear at this level—intonation contours. etc TheD(e€p)-Synt(actico-)Pros((xiic)S(tructure) represents all of the meaningful. including the semantically empty ones Second. for EngUsh in (63). p 97) As IS the case with the DSyntS. the nodes of the SSyntS are not ordered linearly This enables us to keq) stnctly apart two basically different "orders" syntactic hierarchy and hnear ordenng.MEANING-TEXT MODELS 41 sion of the sentence represented mto topic and comment. all pronominal r^lacements and deletions under lexical or referential identity are earned out A branch of a SSyntS is labeled with the name of a SURFACE-SYNTACTIC RELATION (SSyntRel) A SSyntRel belongs to a set of language-specific bmary relations that obtain between the words of a sentence. and the like—^where these are not syntactically conditioned. emphatic stresses. but its composition and labelmg differ sharply from those of the DSyntS A node of a ^ y n t S is labeled with an actual lexeme of the language First.

see 15 15 THE FOODg AND g ADMINISTRATION^ CAUTION ^^ COFFEE HAVE . indic. the SSynt-AnaphS.pres. 1 e to the SSyntR of 5 . li A 15 IS the DMorphR (rf the sentence which corresponds to the SSyntR 14. and the SSynt-ProsS are analogous to their deep counterparts A detailed discussion of important divergences cannot be undertaken here because of lack of space Deep-Morphological Representation Only one DMorphR will be offered for sentence 5._ OF ppas EXPECTANT PLEASURE. fes -"once- NAP COFtEE^^ SIMPLE Figure 4 of 5 14 is the SSyntR of the Enghsh sentence that corresponds to the DSyntR 11.3sg TO SERIOUSLY ONE CUP. .3c" "Ci. l e The SSynt-CommS.

it explicitly keeps apart phenomena that appear to be different THE DESIGN OF THE MEANING-TEXT MODEL As stated above. is to descnbe all the different aspects of an actual utterance as separately and autonomously as possible That is. on the contrary. a MTM has the task of estabUshing correspondences between the semantic and the (surface-)phonetic representations of any given utterance through the five intermediate levels listed in the preceding section Accordingly. and the like In 15 the vertical bars stand for pauses of varying length. and the arrows represent the pitch contouR Representations at Other Levels Levels 5 through 7 the Surface-Morphological [Morphemic] Representation. the MTM consists of the following six basic components 1 2 3 4 5 6 The Semantic Component. intonation contours. or deep syntax The Surface-Syntactic Component. or phonemics . or surface syntax The Deep-Morphological Component.MEANING TEXT MODELS 43 The D(eep)-Morph(ological)R(epresentation) of a sentence consists of two structures the deep-morphological structure. with different structures within each. and the deep-morphologico-prosodic structure The D(eep)'Morph(ological) S(tructure) is a stnng of D(eep)-Morph(ological) R(epresentations) of all the wordforms that compose the sentence The DMorphR of a wordform W is the name of the lexeme to which W belongs sutecnbed for all its morphological values Thus the DMorphR of W unambiguously specifies this particular W (up to wordform homonymy). or surface morphology The Deep-Pbonetic Component. and the Surface-Phonetic [Phonetic proper] Representation. but see (41. 47) The aim of all these representations. a MTM does not aim for a homogeneous representation. or semantics for short The Deep-Syntactic Component. while the DMorphS of the sentence unambiguously specifies its word order (up to free vanatirai) The D(eep)-Morph(ologico-)Pros(odic) Structure J indicates pauses. 43. the Deep-Phonetic [Phonemic] Representation. will be omitted here because they are of minor lmponance. or deep morphology The Surface-Morphologica! Component.

the ongjnal SemR is not cot. Each component of the MTM is a set of rules havmg the tnvial form X <=^ Y\C. the semantic component constructs alongside it another SeniiR. but rather as permissions and prohibitions.) 2) It selects the correspondmg lexemes by means of s^nuitico-lexical rules of the type illustrated below . it performs the followmg eight operations 1) It cuts the SemR into subnetworks such that each corresponds m its semantic "size" to a sentence (In fact. *X' itself is not changed: nothmg happens to *X' while Y is bemg constructed by semantic rules under the control of 'X' The relation between a representation n and an "adjacent" representation n-hl is the same as that between the blueprint of a house and the house itself. by symbols within the rules rather than by the order of the latter The philosophy behmd this decision is that findmg the best order of rule apphcation in a specific situation goes far beyond the task of lmguistics proper The rules themselves are conceived of not as prescnptions. Iordanskaja (24. but consistmg of a sequence of smaller SemRs. which provides for the correspondence between a surface-phonetic representation and actual acoustic phenomena." not "is transformed into.25) and is nowfirmlyembedded in the research on MTMs The Semantic Component of the MTM The semantic component establishes the correspondence between the SemR of an utterance and all the synonymous sequences of DSyntR's of the sentences that make up that utterance (cf 7). To do that. when the transition from a meanmg 'X' to a DSyntR Y is performed. N. simply. where X stands for a fragment of utterance rq>resentatton at level n. but dunng construction. Y for a fragment of utterance representation at level n -h 1. or statements m a calculus Basically each rule is a filter siftmg out wrong correspondences The idea of definmg the structures m the MTM m terms of sets of rules representing well-formedness conditions was mtroduced into the meamng-text theory by L. The bluepnnt is by no means transformed into the house." Thus. equivalent to the ongmal one. falls outside the scope of the MTM model m the stnct sense. 1 e. and Cis a set of conditions (expressed by Boolean formulas) unds* which the correspondence X *=* Y holds The two-headed arrow must be mterpreted as "corresponds. it is the bluepnnt that guides the workers Another important peculianty of the rules in the MTM is that they are logically unordered All relevant information about the language is expressed explicitly.44 MEL'CUK The Surface-Phonetic Component. or instructions of an algonthm.

16 represents two dictionary entnes for two (roughly) synonymous verbs to warn J and to caution.MEANING-TEXT MODELS 45 [who] U-homj [that CAUTION ] 'dangerous 2 ' •reason 1' who] [whoir] 16 IS a semantico-lexical rule in the stnct sense (Ci and C2 stand for different sets of conditions that might determine the choice of the lexeme) Essentially. which IS ensured by two separate straightforward semantico-lexical rules 18 1intensive * This IS a semantico-functional rule responsible for the lexical function M a ^ m the DSyntS (cf m 11 and 12 seriously cautioned. stern warning) 3) It supphes meamng-beanng morphological values of lexemes by means of semantico-morphologtcal rules like the following one . etc. Y. expectant mother and mother-to-be This meamng can of coui^ be expressed also by the free phrase pregnant woman. . etc in the left-hand part of the rule represent semantic subnetworks that "hang" from the nodes labeled X. etc in the nght-hand part of the rule stand for Engbsh lexical items that express X. the vanables X'. Y. Y'. Y. The variables X. respectively." namely. respectively In rules 18-20 the vanables are used in the same way 17 !woman' EXPECTANT-MOTHER MOTHER-TO-BE •pregnant' This IS a semantico-idiomatic rule specifymg two idioms for the meaning "pregnant woman.

includes rules of two classes . all the meanhig-beanng prosodic phenomena are represented in the SemR in exactly the same way as any other meanings. etc) from the data contamed in the Sem-CommS 8) For each DSyntR produced the semantic component constructs all the synonymous DSyntRs that can be exhaustively descnbed m terms of lexical functions (Other types of hnguistic synonymy are accounted for by semantico-lexical rules that can group bundles of semantemes mto lexemes m different ways ) This is thieved by means of a paraphrasing system that defines an algebra of transformations on such DSyntR's where the DSyntS contains LF symbols The paraphrasing system. see also 67). that is.MEL'CUK 'earlier' • t h i s speech — ^ ^sres.perf JT is a vert 'results Here the meanmg of the English present perfect is roughly rendered as "the event X took place earlier than the moment of this speech event but its results persist up to this moment" 4) It forms a DSyntS (a tree) out of the lexemes it has chosen 5) It introduces the DSynt-AnaphS. but in the DSyntR their representation is different since prosodies and words are so different 7) It provides the DSynt-CommS (topic-comment. e g 'truth value' P IS a statement '—•V'commu'speaker' • •<-^ • J ('The speaker wants the hstener to communicate to him the truth value of the statement P' is rendered—along with other means of course—by the specific mterrogative pr(»odics) As we see. it indicates coreference etc for the lexical nodes that have atq>eared as a result of the duphcation of some semantic nodes 6) It computes the prosody of the sentence on the basis of semanticoprosodic rules. first descnbed m (87.

the following syntactic rules are needed 23. the number of elementary deep-tree transformations is finite and not very large Any particular syntactic rule defined on DSyntSs can be represented in terms of those specific transformations To operate lexical paraphrasing rules (21).MEANING-TEXT MODELS 47 Lexical Paraphrasing rules (about 60 of them. transfer of a node to another governor. a lexical rule 21a) \ i . i. vahd for any language) represent either semantic equivalences or semantic lmphcations Examples Equivalences 21 (a) The set contains [W\ the point M ** The point M belongs [Convji ro the set He warned [W] them ^He issued [Oper. and renaming of a branch) and only nine deep-syntactic relations. (No a warning [NQ(KO] to them (b) K' He followed [Real2(H0] her advice [W] to enroll '=> He enrolled on [Adv|g her advice [W] Implication 22 PerfCaus (X) => Pernncep {X) John started [PerfCaus (run II)] the motor => The motor started [Perflncep (ran 11)] Syntactic Paraphrasing Rules indicate what restructunng of a DSyntS is needed when a particular lexical rule is applied Smce there are only four basic syntactic opei^tions at the deep level (merger of two nodes. (for lexical rule 21c . splitting of a node.

conjunctional Rules such as 25 belong to the dictionary. to the dictionary 2) It expands the nodes of idioms mto corresponding surfiace trees 25 FDA ADMINISTRATION^ determinatT ve-/ ^s^omposi t i ve THE T coordinatlve A N D coordinate.. due to its abstract character. 52. it performs the following five operations 1) It computes the values of all lexical functions by means of rules hke 24 ^^ ^CAUTION . a syntactic rule.48 MEL'CUK Generally speaking. pp 163-66) The Deep-Syntactic Component of the MTM The deep-syntactic component establishes the correspondence between the DSyntR of a sentence and all the alternative SSyntR's which correspond to It To do that. may serve several different lexical rules A formalism has been devised for descnbing unordered dependency tree transformations—so-called A-grammar (see 17-19. too 3) It eliminates some DSyntS nodes that occur m anaphonc relations and should not appear in the actual text.yj IATTR I *Mltgn ^CAUTIONf^j *? > • •SERIOUSLY c lrcutns t a n t i a t ive ^ 2 •lA'ARNING ^ ^ I Jlst completive •< Such rules belong. of course. thus the occurrence of MOTHi under AVOID m 11 is deleted by the following rule (a kind of Equi) .

. • > Iquantitatxve t ^(Nun) (Num) Replacement of a DSynt-node by a SSyntRel • (N) \ATTR I2 ^(N) lappositive (N) Replacement of a DSynt^el by a SSynt-node j ^ [2nct completive •TO • Jprepositional . X • X • A T T n i.MEANING-TEXT MODELS 26 * X 49 X A 1st comp-/ .^=> letive>^ A \ 2 n d compN^letive T s V 4) It constructs the SSyntS by means of three types of transformations 27 Replacement of a D3yntRel by a SSyntRel X _ Y -L If Y • t V JI i n Ipredicative • • |ATTR^=> t • Imodifleative • c.

which can be illustrated with the following three rules for Enghsh 31 •^'' '^"TOl O t j + +X not r(Y. l e it determmes.50 MEL'CUK The notation X(V. the correct prosody. and Dick might too 4) Punctuation. such as the number and person of the verb 2) Lmeanzation of the SSyntS. l e it determines the actual word order of the sentence 3) SSynt-ellipsis. which. and Dick might take the course. pp 237-59) The Surface-Syntactic Component of the MTM The surface-syntactic component establishes the correspondence between the SSyntR of a sentence and all the alternative DMorphR's that are realizations of it It performs the foUowmg four main operations 1) Morphologization of the SSyntS.X) (predicative X+ -fY. is rendered by punctuation The basic tool of morphologization and lmeanzation of the SSyntS is the syntagm. on the basis of the DSynt-ProsS as well as on the basis of the resulting SSyntS. in the case of pnnted texts. too ^ * George will take the course. one can put the subject either before the verb if the standard function^ "obligatory inversions of the subject and the verb" does 'A standard function (in programming. coordination. l e it carnes out all kinds of conjunction reductions and deletions that are prescribed by the language in question. also a standard subroutine) is a closed set of rules used in solving a specific class of problems or in carrying out a specific class of transformations In a lmguistic model. e g 30 George will take the course. the corresponding information is stored m the dictionary entry of X (cf CAUTION in 11 and 14) 5) It processes the other three structures of the SSyntR A descnption of a small fragment of the DSynt-component for Russian IS found m (52. inversion. not obn v(Y. eUipsis. or SSynt-rule.3|TO]) indicates a lexeme whose 3rd DSynt-valence slot must be filled m the SSyntS by the mfinitive marker TO. standard iiinctions provide for different types of agreement. l e it determines all the syntactically conditioned morphological values of all the words.X) The rule says that to build a predicative construction with a NP as the grammatical subject. and the like .

23. apposTtive t •< > X + + Y deterp I X+ Z r(N} detern'i natT ve • ^ i . your.tepp) This rule means that a noun Y with the syntactic feature *temp(oral)' which IS an adverbial modifier of a verb can be placed before or after the verb if Y Itself has a dependent Z other than an article Last week [Y] we finished our job. ). ) and being an appositive to a human nonpronominal noun X can be placed after X in such a manner that its article or the personal adjective follows X immediately Jack [X] the [Z] editor [Y] A nearly complete list of &ighsh syntagms is found in (64). the subject must not be in objective form (relevant only for pronouns me. hurran . He wtll go there next year [Y] Y (N . and the verb must agree with it in accordance with yet another standard function "agreement of the verb with the subject noun " The NP and the standard functions are defined elsewhere since they appear m many different syntagms circuTPStantiative <^7-^ X Y v I 1 T r(N. oers) The rule states that a noun Y govermng THE or a personal adjective {my. not prop . cf also a list of basic Alutor syntagms (65)] . or after the verb if a similar function ("possible or obligatory inversions of the subject and the verb") applies. 76). us. an enlarged and revised English edition is now being prepared for publication by Benjamins Publishers Cenam types of Ru^ian syntagms have been descnbed m detail in [(22.MEANING TEXT MODELS 51 not apply.

surface syntax uses at least four additional types of rules • Word order patterns for elementary phrases. started research m this hitherto neglected field. which compute the best word order possible for the given SSyntS on the basis of vanous data* the syntactic properties of some words marked m the lexicon. and in 1964 published an issue of die Laboratory's tnilktm MaSinny) perevod i prikladnaja hngvistika devoted entirely to semantics* the now famous MPiPL 8 This modest-lookmg volume was to play a crucial role for the Soviet seman- . possible ambiguities produced by specific arrangements. I therefore mention only a few.52 MEL'CUK Besides syntagms. etc These rules try to minimize the value of a utility function that represents the "penalties" assigned (by the hnguist) to certain unfehcitous arrangements llie rules do this by reshufl9lng the constituent phrases within the limits of what the syntagms allow A fairly full description of such global word order rules for Russian. emphasis.g all those beautiful French magazines vs * French those beautiful all magazines. a group of young Soviet s^nanticists. 56) SOME LINGUISTIC RESEARCH DONE WITHIN THE FRAMEWORK OF THE MEANING-TEXT APPROACH It IS impossible to hst here all the studies conducted and pubhshed by the adherents of the Meaning-Text approach. pp 268-300) • Ellipsis rules • Prosodic or Punctuation rules The SSynt-component of the MTM for Russian is characterized at some length in (5) Other Components of the MTM These will not be dealt with m this review The interested reader is referred to (47. topicalization. the relative length of different parts of the sentence. at least m the USSR. was <^ned m the early 1960s by the work of the Machine Translation Laboratory of the M Thorez Institute of Forragn Languages m Moscow Headed by V Ju Rozencveig. a language famous for its involved word order. e. and the hke. K. with A. Zholkovsky in the forefront. was published in (40) and (52. 50. the paper offers an excellent perspective for a beginner Semantics A new trend m modem semantics. • Global word order rules. picked more or Iras at random Nichols (68) draws parallels between the MTM approach and Modem American linguistics.

of regular polysemy.MEANING TEXT MODELS 53 tic school Many papers in it (77. of semantic metalanguage. was exphcitly stated and apphed to hundreds of semantically very difficult Russian words (with meanmgs such as "goal." "help. as opposed to hsts of semantic features. 52. 80-82) are still of enormous interest for semantics. where most of the MPiPL 8 anicles are repnnted in Enghsh] Apresjan's book (1) sums up much of his work done in the late 60s and early 70s Particularly important is Apresjan's treatment of synonymy and antonymy. etc Among the minor studies are these (a) A careful analysis of two important semantic famihes (feelmgs and perceptions) is found in (28. Inc. 62) . spearheaded by A Zholkovsky. and syntactic constructions. standard lexicographic definitions of lexemes. cf also 13) (c) The highly important distinction between lingmstic anomaly and logical contradiction is convincmgly dealt with m (2) {d) Connotation m hnguistic semantics is dealt with m (30) Lexicography Lexicography is a major concern of the meamng-text approach since all the information about the relevant properties of mdtvidual words vital for the successful functKmmg of an MTM must be stored in a dictionary As soon as the work on the MTM of Russian be^in m 1965. pp 113-33. a foUer vo^on of the Russian ECD is beng prepared for pubhcatK>ii m book form by Lingu»tic Research." language-mdependent semantic representation first emerged and brilliant examples of semantic decomposition of lexical meanmgs were produced The pnnciple of using highly involved syntax in SemR's. laid the foundation for the MTM approach [cf (73)." is explored m depth m (12. Edmcaiton. but they can be summarized here only in outhne It was there that the idea of a "pure. and of modal frames as pan of lexical meanmgs A recent book by the same author (3) discusses nontnvial semantic features. wu^ sc»iie hundred dictionary entries w « e published (39) (Presoitly." "attempt. much of our effort was concentrate on the so-called Ei^danatory Comlnnatonal Dictionary (ECD) c^ Modem Rmsian Dunng a penod (^ about 12 years between 10 and ^ petite took part m our dict»mary {H'CQect." "proceed from. syntagmatic lnteractioti of meanings." etc) In a word. of conversives (such as X is behind ¥=Y is injront ofX). such as negation or "only." "manage. 29) {b) The problem of the scope of some "operator" meanings. Canada) The lexicographic th«ffy belund the ECD proposed in (87) was further elaborated and expounded m (6-8. grammemes. the semantic research of 1961-1965.

pp 9 1 150. a dictionary project for Huitchol (a Uto-Aztecan language in Mexico) IS bemg earned out by J Gnmes (Cornell University. pp 78-109). 58. NY) with extensive use of the ECD model Phraseology The notion of the lexical function and the hohstic concept of a lexical unit embodied in the BCD have contnbuted to the theory and practice of phraseology The place of LFs in linguistics is descnbed in (42) and (52. many of these have been mdicated above The following syntactic topics deserve special mention (a) A general characterization of the dependency system used m MTM theory can be found in (52. Somah (83. Ithaca. (27) is devoted to a specific class of idioms denotations of emotionally conditioned physical reactions (eg. 60) {e) See (75) for a discussion of comphcations ansmg when conjunction is described m the MTM framework. e. is given m (85) This book is. and German (71. University of Montreal (62) In addition. edited by Apresjan and orgamzed following some basic prmciples of ECDs Moreover.g Pohsh (35). while the agent phrase m the ergative case in such languages as Dyirbal or Lezghian is taken to be an agentive . 72) Recently. see (5) (</) Criteria for the types of SSynt-relations are proposed m (58. fear. pp 207-302. 84). some pioneer work has been done on £CDs for other languages. a Modem French ECD project was launched at the Department of Linguistics. His eyes started from his head— astonishment. 61) The ergative construction is interpreted 9& a predicative construction where the surface-syntactic (grammatical) subject IS marked by a case other than the nominative (frequently by the ergative). French (90). in fact. His legs gave way—fear) See also (7) Syntax A senes of concrete syntactic descnptions was completed and pubhshed in the early 1970s bearmg mostly on Russian and Enghsh. 58) {b) A full-fledged descnption of the syntax of a particular language. an excellent illustration of how our theoretical pnnciples can be apphed to an exotic hnguistic structure (c) For a systematic survey of the types of rulw and information used m the surface-syntactic component of an MTM for Russian.54 MEL'CUK Beginning m 1978. namely Somah. a hmited dictionary of Enghsh synonyms has appeared (74). Enghsh (89). an mgemous solution is suggested if) Problems of grammatical subjecthood and the ergative construction are analyzed m (57. pp 23-90.

suppletion (46). Tartar declension and ccmjugation (32-34). among others Morphology In the dommn of theoretical morphology. with special reference to Russian. Russian nominal inflection (15). see To this hst one could add many other topics of mterest such as syntactic zero (53). affix. CONCLUDING REMARKS Possible Applications of the Meamng Text Theory Some ideas of the MTM approach may find unexpected apphcations m quite different thewetical frameworks withm linguistics itself. apophony.MEANING-TEXT MODELS 55 compl^nent rather than the subject. and we consider It an asset that this theory is readily apphcable m vanous practical fielcU Let me mention only four such po^ible applications. etc). LANGUAGE TEACHING AND LEARNING LmgUlStlC descnptions based on MTT seem to be adaptable to language teaching or learmng The lexical functions and ECD-type dictionaries might be especially useful (cf 38. 36. Thus m (9) the concept of LF h^>s to solve some specific problems of an otherwise transformati(»ial descnption of Russian impersonal sentences (cf also 9a) The whole of MT theory is strongly practice-onented. a few formal morphological models have been produced for different languages Spanish conjugation (41). Alutor (Chukchee-Kamchadal stock) conjugation (49) An interesting mathematical formahzation of MT morphological models is suggested m (10. work has been m progress for a number of years on some basic morphological notions. morpheme. 11) A detailed descnption of the grammatical categones of the Enghsh verb is proposed m (69. 78) . grammatical case (55). morph. root. and a meaningbeanng sjrntactic [quotative] construction m Ri^sian (31). representabihty. etc Most of this work is s u m m ^ up (m Bngbsh or German) m (54) This author's book (59) offers a coherent syst»n of formal morphological concepts pertammg to the expression plane of morphology (such as linguistic sign. reduplication. 66). grammaUcal voice and diathesis (21. so that no ergative construction is recognized in these languages (g) For general questiims of grammatical agreement. 70). the comparative construction m Russian (76). linear segmentability. Hungarian noun inflection (43). syntactic amt»guity (26). such as conversion [like a footnote-to footnote (48)]. As for descnptive morphology. lesser or greater semantic and formal ccni^)lexity of linguistic itrans (44).

and expressing it . such an analysis is precluded by considerations of space Instead. Norman) This clearly shows that specialists in field linguistics and exotic languages do find handy and helpful tools in the realm of MT theory Perspective and Summary It IS next to impossible to set an outline of MTT in a background readily understandable to readers in related fields To do so. 37) Among the more recent developments in the USA. use of the lexical functions (16) and E Fox. an explanatory combmatonal dictionary could be used as an effective translator's tool (see 56. we find the apphcation of some MTT ideas in information retneval [in panicular. it is natural that several textprocessing computenzed systems at various hnguistic levels. one should show how MTT compares with "traditional" linguistics. dictionary. the most immediate foundations of the MTT are to be found m Chomsky's theory I believe that the MT approach can be viewed as an outgrowth and natural continuation of the generative-transformational approach On the other hand. in my view. Ithaca. its relationship to translation in general is direct from the viewpoint of MTT. MTT has developed under the strong influence of automatic text processing The problem of automatic translation puts the emphasis on the transfer of meaning and thus bnngs forth the task of representing meanmg. I will mdicate two sources of the MT approach and five basic methodological pnnciples. 78) ANTHROPOLOGICAL RESEARCH MTT was testcd on the Alutor language (49. "Lexical relations enhancing effectiveness of information retneval systems" (unpublished progress report. the structuralist school. which constitute. the most general and cogent advances SOURCES OF THE MT APPROACH On the one hand. the whole of linguistics is nothing more than the science of translation Specifically. Department of Computer Science. and transformational grammar However. extracting it from the input text.and French-to-Russian) could be mentioned (4. Cornell Umversity. NY)] TRANSLATION PRACTICE AND THEORY As MTT IS primarily concemed with the translation of meanings into texts and vice versa. and morphology. some ideas and postulates of MTT are taught in the courses offered by the Summer Institute of Linguistics (University of Oklahoma.56 MEL'CUK AUTOMATIC TEXT PROCESSING Since the MTM formahsms lend themselves easily to computer programmmg. 65) As I discovered in the summer of 1980. including syntax. 14. have been developed in the USSR within the MTT framework As an illustration. arnved at in the course of a 15-year research effort. two automatic translation systems (English.

not on grammars The only legitimate question to be asked withm this theory is "How do you express the meamng X m the language L''" or "What does the expression X of the language L mean''" It seems to us to be more rewarding to wnte a good grammar (mcluding the lexicon) at least for a language than waste one's time discussing the properties of the best theory of Language m general 2 Establishing specific facts of a specific language rather than hunting for generalizations or discovering linguistic universals To state a generalization. but no formalism is allowed to become a fetish Nobody should be afraid of replacmg 17 symbols by 439. is an entire world m itself All this means that the pnmary object of MTT is particulars. but every lexeme of a language. if need be Why should we be slaves to oversimplified and overly narrow mathematical formalisms'' Formalisms are nothing more . and if It functions correctly. therefore. let alone necessary What a MTT linguist really does is suggest practical names. i e m the dictionary. but one can never prove—m the stnct sense of the term prove—that a descnption or a solution IS sufficient. one needs something to generalize from. not universals 3 Cybernetic modeling of native linguistic intuition rather than a mathematical study of formal systems representing language The only goal of a MTT hnguist is to exteriorize his or her own intuition about language (probably checked against other speakers' intuition) Language is viewed as a "black box. useful labels.MEANING-TEXT MODELS 57 in the output text That is.rather than a grammar-onented theor>' Not only every language. there are no arguments to prove or disprove its acceptability 4 Formalisms serve the linguist rather than the other way around The formality of all representations is strongly emphasized in MTT. well-organized collections of facts must perforce precede generalizations lest the latter be scientifically void Hence the particular interest of MTT in lexical entnes." not as a formal mathematical system (semi-Thue. the meanmg-text approach is virtually required by automatic translation FIVE BASIC PRINCIPLES OF THE MEANING-TEXT APPROACH The credo of a righteous practitioner of the MT approach (l e my own credo) can be reduced to the following five operational principles 1 Description of a specific language rather than the quest for its best possible grammar MTT concentrates on languages. or whatever) Hence no rigorous proofs are possible withm the MTT Good counterexamples can show that a description is not sufficient. a grammar is considered to be nothmg more than a set of generalizations over a good dictionary MTT is thus a dictionary. etc to pass his own hnguistic intuition on to others m a most unambiguous and graphic way He builds a model.

L N lordanskaja. AlbertaLingmstic Research. and suggestions have helped me considerably to improve the presentation I wish to express my appreciation and gratitude for their efforts ABBREVIATIONS USED IN BIBLIOGRAPHY AJCL Amencan Journal of Computational Linguistics AN SSSR Akademija Nauk SSSR BSLP Bulletin de la Societe de linguistique de Paris. and at each level different formal languages are used for its different aspects and strata ACKNOWLEDGMENTS The first draft of the present review was carefully read and commented upon by L L Elmtsky. with their emphasis on relations rather than constituency. for they detach linear order from structural representation Moreover. remarks. if they so wish. Pans Khncksieck IRJa Institut Russkogo Jazyka IRSL International Review of Slavic Linguistics. in MTM a special formal language is devised to represent differently each one of the different levels of natural languages. multifanous. Moscow Mosk Gos Ped Inst Inostr Jazykov lm M Toreza NTI Naudno-Texnideskaja Informacija Moscow Vses Inst Nau6n Texn Info PGfePL Problemnaja Gnippa po eksperimental'noj l pnkladnoj hngvistike VJa Voprosy Jazykoznanija Moscow AN SSSR WSA Wiener Slawistischer Almanack Vienna . J Nichols. Izv Izvestija MPiPL MaSinnyj Perevod i Pnkladnaja Lingvistika.58 MEL'CUK than nec^sary tools. Edmonton. Inc IVSLiAP Informacionnye voprosy semiotiki. and P Roberge J D McCawley read the final version Their cnticisms. highly heterogeneous formation. to their theones ) As far as the technical nature of MTM formalisms is concerned. lingvistiki i avtomatideskogo perevoda. R I Kittredge. dependency systems. Moscow Vses Inst NauCn Texn Info. it is believed that the relation rather than the groupmg is the pivot mechanism of natural languages at all levels 5 Different formal languages for different levels of representation rather than the quest for a homogeneous representation Language is a multidimensional. are preferred. and it is up to the linguist to create tools that he feels are necessary (Let the mathematicians accommodate our formalisms. and there is no point in trying to squeeze it into the strait jacket of a single formalism Therefore.

Moscow IRJa AN SSSR (PGfePL. I A 1968 O sisteme semantideskogostntezn Ul Obra2cy siovamyx statej AT/Ser 2. I G . 2o&avskij. A V . O S. Iomdin. Mel'duk. Ju D 1974 Leksideskaja Semantika. L L . N V . No 6 42-57 U Bider. vyp 122) 45 pp 21 Iomdm. Ann Arbor. Percov. J u D 1980 Tipy Informaen Dlja Pijverxnostno-semantideskogo Komponenta Modeh "Smysl i=^ Tekst" Wien WSA (Sonderband 1) 119 pp 4 Apresjan. Mel'^uk. Ju D . No 118-21 7 Apresjan. ed M Bonllo. L L . I A 1976 Formalizaaja moribl^aeskogo komponenta mo<teli "Smysl « = ^ Tekst" I Postanovka probiemy l osnovnye ponjatija. vyp 131) 38 pp Chic^o Umv Chicago Press. I A . MerCuk. Mich KAROMA 180 pp 10 Bider. I M . II K(»npletivnye t pnsvjazoCnye konstnikcu NTI Sw 2. L H 1978 Lexical functions and syntactic constructions Russian existential sentences In CLS Parasession on the Lexicon. 364j^ 9a Babby. Bol'$akov. vyp 21) 72 pp 15 Es'kova. MeI'duk. Kulanna. IRSL 3(3)249-312 6 Apresjan. Saonikov. Ju D . A K . No 6 61-72 8 Apresjan. I A 1969 C^ odnom sposobe izuCemia soCetaemosti slov Russkij jazyk vSkole. Iomdm. L L . A V . L L. I A (971 Ob Odnoj Vozmoinoj Sisteme Majfmnogo Ferevoda. Sannikov. Mel'Cuk. Percov. Boguslavskij. Texn. L P . L N 1963 O nekotoiyx svojstvax pravil'noj sintaksid«koj struktury (na matenale russkogo jazyka) VJa No 4 102-12 25 Iordanskaja. V Z 1971 Forma/'na/a Model" Russkoj Morfologiu I Formoobrazovanie Su^estvitel'nyx i Fnlagatel'nyx. A K 1969 Semantics and lexicography towards a new type of unihnguat aicttooary In Studio in Syntax and Semantics. A K . pp 83-122 Pans CNRS 289 pp ] 23 Iomdm. Int Conf Comput Ling Stockholm (preprint No 1) 7 p p 18 Gladkij. R N 1979 A lexicon for a computer cjuesttonanswenng system AJCL (microfiche 83) 17 Gladkij. N A . N V 1978 Ob"ekty l sredstva modeli poverxnostnogo sintaksisa nisskogo jazyka. ed F Kiefer. Mel'Cuk. pp 129-51 Wroclaw Ossolmeum 3 Apresjan. A V . Mereuk. K O . Percov. No 1122-32 24 Iordanskaja.vyp 15) 71 pp 16 Evens. Moscow AN SSSR 47 pp 5 Apresjan. I A 1977 Formalizacija moTfotogi6eskO0O kommenta taoddi "Smvil ^ = ^ Tekst" II mamika morfofogi(^cskix preobrazovamj Texn. L N 1964 Svojstva pravil'noj smtaksi^eskoj struktury l algontm ee obnaru^enija (na matenale russkogo jazyka) FrobL kibern 11215^14 . I G . I A . Ser jaz. I M 1977 Osonastiaeskom c^usami russkix ckepn6astij neopredelennost' lh mnogoznadnost'^ Izv AN SSSR. Ju D . Mel'duk. Smith. kibem. Krysin. I A 1971 Grammatiki derev'ev I Opyt formallzacu pref^razovaiuj smtaksideskix struktur estestvennogo jazyka IVSLiAP 1 16-41 19 Gladkij. L H 1980 Existential Sentences and Ne^tmn in Russian. Ju D . Percov. N V 1975 Fragment modeh russkogo poverxnostnt^o sintaksisa. J Virbel. Moscow IRJa AN SSSR ( P G 6 P L . N V 1975 Fragment modeh russkogo poverxnostnogo sintaksaa I Predikativnye smtagmy NTI Sa 2. J u D 1978 Jazykovaja anomalija i logi(:^koe pronvore€ie In Tekst'Jezyk-Poetyka. Mel'Cuk. I A . V Z 1978 Lingvistideskoe Obespedenie v Sisteme Avtomatideskogo Perevoda Tret'ego Pokolenija. Kibem No I 32-50 12 Boguslavskif. I M 1978 Otncanie v predloiemjax s obstojatel'stvami v nis- B 22 Iomdm. I A 1974 Granunatiki derev'ev II K postroeniju A-grammatikidljarusskogo jazyka IVSLiAP 4 4-29 20 lomdin. M W . 1977. No 7 3 1 43 [French transl Analyse et validation dans I'etude des donnies textuelles. Moscow IRja AN SSSR (PGfePL. L L 1980 Simmetridnye Predikaty v Russkom Jaz^ i F^lema Vzaimnogo Zalt^a. Lazurskn. Moscow IRjaANSSSR(PGEPL. L L 1979 E3de Raz o Stntaksideskom Soglasovanii v Russkom Jazyke. Boi'Sakov. pp 26-33 59 skom jazyke In Studia grammatyczne 2 122-36 Wroclaw Ossolmeum 138 pp 14 Erastov. 2olkovskij. Ju D .MEANING-TEXT MODELS Literature Cited 1 Apresjan. A V . Sinonimideskie Sredstva Jazyka Moscow Nauka 367 pp 2 Apresjan. 2olkovsktj. i lit 36(3) 270-81 13 Boguslavskij. pp 1-33 Dordrecht Reidel 9 Babby. I A 1969 Tree Grammars.

No 2 65-84 Mei'Cuk. I A 1968 Model' sklonenija vengerskix suSeestvitel'nyx In Problemy Strukturnoj Lingvistiki-1967.V77 No 12 36-44 French Transl T A Informations. pp 45-90 Mel'Cuk. L N 1979 The semantics of three Russian verbs of perception vosprintmat' '(to) perceive'. I A 1974 Esquisse d'un modele linguistique du type "Sens <=^ Texte" In Problemes actuels en psychohnguistique Collogues internationaux du CNRS 206 291-317 Pans CNRS Mereuk. L N 1973 Tentative lexicographic definitions for a group of Russian words denoting emotions In Trends in Soviet Theoretical Linguistics.60 MEL'CUK 34. No 170. K E H e ^ l p h .I A 1974 Ob odnoj hngvisu£eskoj modeh tipa "Smysl ^=> Tekst" Urovni predstavlemja jazykovyx vyskazyvamj Izv AN SSSR. 1970-^1976 Vypuski NN° 2. L N 1967 Smtaksiteskaja omonimija v russkom jazyke (s to£ki zremja avtomatii^eskogo analiza l smteza teksta) NTI No 5 9-17 27 Iordanskaja. 37. 40 41 42 43 44 45 46 47 48 49 50 51 . 32(1) 15-28 ^ = > MerCuk. ed A Booth. ed F Kiefer. 23. 62. 63. 42. I A 1973 Konversija kak morfologieeskoe sredstvo Izv AN SSSR. 85.Mel'i. Ser lit ijaz. pp 344-73 Moscow Nauka 411 pp Mel'euk. pp 33-57 Dordrecht Reidel 438 pp Mel'Cuk. Nakhimovsky. ed S K Saumjan. I A 1967 O "vneimx" razheitel'nyx elementax (o semantiCeskix jjarametrax) In To Honor Roman Jakobson. E 1971 Pjaf pol'skix slov^piyx state] KLESKA. X F 1971 Model' tatarskogo spijaienija (obrazovame sinteti6eskix fcmn glagola v sovranennom tatarskom jazyke) In Sinxronno-sopostavitel'nyj analiz jazykov raznyx ststem. X F 1968 Avtomati^eskij sintez form suSJ^estvitel'nogo v tatarskom jazyke NTI Ser 2. I A 1967 Automatic translation some theoretical aspects and the desim of a translation system In Machine Translation. pp 352-63 s'Gravenhage Momon Mel'euk. 80.L N.17. Ser lit i jaz. O S. pp 389-410 Dordrecht Reidel 438 pp 29 Iordanskaja. I A 1967 Model' spqaiemja V lspanskom jazyke MPiFL vyp 10 21-53 English transl see Ref 73. ZWYCIESTWO. pp 396^38 Moscow Nauka 554 pp Enlarge. I A 1973 Model'Sprjaienija V Aljutorskom Jazyke Moscow IRJa AN SSSR (PGfePL. 35.7. Essays on the Occasion of his 70th Birthday. I A 1969 Ob opredelemi bol'sg men'Sej/smyslovoj sloinosti pn slovoobrazovaternyx otnoSemjax Izv AN SSSR. L N. 28. ed M Bierwisch.14. pp 108-26 Moscow Nauka 279 pp 35 Janus.uk. I A 1973 Towards a linguistic "Meaning ^ = ^ Text" model In Trends in Soviet Theoretical Linguistics. 33(5) 436-47 26 Iordanskaja. No 11 21-24 36 Korolev. No 3 23-26 33 Isxakova.Mel'duk. Mel'tuk. Ser lit tjaz. obozna^ajus^ fizi^eskie simptomy iuvstv MPiPL 16 3-30 28 Iordanskaja. I A 1965 Poijadok slov pn avtomatideskom sinteze russkogo teksta (predvantel'noe soobSfienie) . oSduSdat' '(to) sense' and duvstvovat' '(to) feel' Linguistics 17 825-42 30 Iordanskaja. pp 139-71 Amsterdam North Holland 529 pp 38 Lecd. ZWYCIEZAC NTI Ser 2. revised Enghsh version Linguistics 1976. 16. PASAZER. A D 1979 Lexical functions and language learning SL and East Eur J 23 104-13 39 Materialy k Tolkovo-Kombinatornomu Slovarju Russkogo Jazyka. t I 1968 O klassifikacu vozvratnyx glagolov l zapisi lx v slovare NTI Ser 2. Ref 54. 86 Moscow IRJa AN SSSR (Predvant Publikacu PGfiPL) Mel'iuk.ispugalsja bufetdik lexical polysemy or semantic syntax'') In Festschrift S0rcnscn. 28(2)126-35 Mel'tuk. WALKA. I A 1972 O su|^letivizme In Problemy strukturnoj lingvistiki1971 ed S K gaumjan. X F 1971 K voprosu o formal'nom opisami morfologu tjurkskix jazykov Sovetskaja tjurmlogija No 6 32-44 34 Isxakova. vyp 45-46) 132 PP Mel'Cuk. I A 1970 Towards a functionmg model of language In Progress in Linguistics. Copenhagen 32 Isxakova.I A 1981 On a class of Russian verbs which can introduce direct speech (constructions of the tj-pe "Ostav'te menja'" .I A 1980 Konnotacija v lmgvistifeskoj semanUke WSA 6 191-210 31 Iordanskaja. 29. No 3 19-22 37 Kulagma. 38. R L . I 43-94. pp 210-57 Mel'euk. L N 1972 Leksikografideskoe opisanie russkix vyraiemj. ed 6 A Makaev. 1967. ed F Kiefer. pp 198-207 The HaguePans Mouton 344 pp Mel'Cuk.

M 1979 Paraphrase". Percov. pp 5-39 66 Mel'cuk. T 1978 Pladoyer for das Worterbuch Linguist Ber 57 25-48 ''^ Reuther. lordanskaja. Ju D 1979 ' Anglo-Russkij SinonimiCeskij Slovar' Moscow Russkij Jazyk 543 pp . E N 1976 Fragment modeli russkoro poverxnostnogo sintaksisa III Sravnitel'nye konstrukcii (sravnitel'iiye otsojuzuye sintagmy) V77 2. 5 221^2 -. V Ju . I A 1979 Studies in Dependency Syntax Ann Arbor. Mich Karoma 163 pp 59 Mel'Cuk. N V 1977 Granunatij^eskoe predstavleme anglijskqj giagol'noj formy dlja celej avtomatieeskogo perevoda Linguistica Silesiana No 2 139-63 " ' Reuther. I A 1981 Types de dependance syntagmatique entre les motsformes d'une phrase BSLP In press 61 Mel'Cidt. I A 1974 O sintaksiCeskom nule In Tipologija passivnyx konstrukfj/ Diatezy i zalogi. vyp 90) 62 pp ^0 Percov. Spec Issue.MEANING-TEXT MODELS 52 Mel'Cuk. BLS. L N . ed A A Xolodovie. I A . vyp 43) 80 pp 64 Mel'cuk. X 1981 Un nouveau type de dictionnaire le Dictionnaire explicatif et combinatoire du frangais contemporain (six entrees de dictionnaire) Cah lexicol In press 63 Mel'duk. A K 1970 Towards a funononmg Meaning-Text model of language Linguistics '^"^ 10-47 61 68 Nichols. 5-36 56 MeRuk. Meet. ix sintaksiceskoe predstavleme I-II WS4 4 413-32. A K 1964 Predislovie MPiPL 8 3-16 si 2olkovskij. Percov. I A . I A . N V W^i Po verxnostno-syntaksideskie Otnosemja v Anghjskomjazyke Moscow IRJa AN SSSR (PGEPL. 1978. A I . A K 1970 Matenaly k russko-somahjskomu slovarju MPiPL 13 35-63 . A 1972 Semantic Primitives Frankfurt Athenaum 235 pp -•()„ Wierzbicka.. pp 384-8"^ 58 Mel'Cuk. I A . XolodoviC. N V 19^5 Model' Anglipkogo Poverxnostnogo Sintaksisa Moscow IRJa \ N SSSR (PGfePL. pp 224-60 Berkeley Umv Calif 54 Mel'^uk. I A 1978 Theone de Iangage. I A 1974 Opyt Teoni Ling-ytstideskixModelej "Smysl Tekst" Semantika. I A 1976 Das Won Munchen Fmk 461 pp 55 Mel'6uk. I A 1978 Toward a definition of the concept Ergative Construction' Proc 22nd Int Congr Ling Vienna. ed B Comne Edmonton Ling Res In press 62 Mel'Cuk. A A l*>70 2alog (Opredelenie IsCislenie) Marodv Azii I Afriki No 4 11 J-24 67 Mel'cuk. pp 147-51 Pans Institut d'etudes slaves -79 Wierzbicka. 282 pp Rozenman. I A .I A 19^7 Lecas Rev Etudes slaves 501. A K 1964 Leksika celesoobraznoj dejatel'nosti ^fPiPL 8 67-103 K3 Zoikovskij. V Z 1979-80 Sofimtel'nye 1 sravniternye konsrrukcii ix blizost'. Savvina. pp 261-76 Berkeley Umv Cahf 660 p p 69 Percov. Ju K 1964 Dve gruppy slov rubskogo jazyka MPiPL 8 50-66' 7^ Slodzian. Arbatchewsky-Jumane. T 1980 Nemeckie frazeologiceskie slovosodetanija tipa in voUiger Verzweiflung sein 1 ix russkie ekvivalenty WSA 5 207-19 Rozencvejg.7 Ser SCeglov. I A . Munchen Fink 60 Mel'Cuk. 22-24 avril 19^7). I A 1981 Grammatical subfcctand the problem of the ergative construction in Lezgian !n \on-Slayic Languages of the Soviet Union. A K 1964 O pravilax Ncmanti^eskogo analiza MPiPL 8 17-32 ^2 2oikovskij. Savvma. pp 343-61 Letungrad Nauka 383 pp English version Proc 5th Ann Meet Berkeley Ling Soc. A 1980 Lingua Mentalis The Semantics of Natural Language Sydney Academic 367 pp gQ Zoikovskij. N V 1976 O Grammatideskix Kategonjax Anghjskogo Clagola Moscow IRJa AN SSSR (PGfePL. th^nedetraducnon Meta23(4) 271-302 57 Mel'Cuk. J 1979 The meeting of East dnd West confrontatjon and convergence in contemporary linguistics Proc 5th Ann. synonymie" et traduction In lie Collogue de hnguistique russe (Paris. vyp 64-66) 232 pp 65 Mel'Cuk. No 1 3 8 ^ 3 -. Apresjan.^ Sannikov. E N 19^8 Toward a formal model of Alutor surface syntax predicative and completive constructions Linguistics. Zoikovskij. Sintaksis Moscow Nauka 314 pp 53 Mel'fuk. ed 1974 Essays on Lexical Semantics. Vols 1. 2 Stockholm Sknptor 404^ pp . I A 1981 Toward a Language of Linguistics. A K 1967 K leksikografit^eskomu opisaniju somaliiskix susdestvitel'ny\ Narody Azii i Afriki No 1 93-102 S4 2olkovbkij.

A K . VoL 19. New York 88 Zolkovskij. A K . E E 1974 O sisteme semanticeskogo smteza VI Obrazcy slovarnyx state] IVSLUP 4 36-62 85 2olkovsktj. A K . Md'Cuk. A K.62 MEL'CUK K postroeniju deistvujujee) modeli jazyka "Smysl ^=^ Tekst" MPiPL 11 89 Zolkovskij. Razlogova. I A 1965 O vozmoiann metode l uutnimentax semantiCeskogo smteza NTI No 523-28 87 Zolkovskij. 263 pp 86 Zolkovskij. I A 1969 . Md'^uk. 19177-238 English Trans Systems Theory Research. T V 1972 IitU) 1 C^fo/Ocsematitieeskomu tolkovamiu dvux anriijskix pre^ogov) IVSLiAP 3 86-96 90 2olkovskij. A K .. A K 1971 Sintaksis Somali (Ghibinnye i PO}Krxnostnye Struktury) Moscow Hauka. Pivovarova. I A 1967 Osemanbdeskomsmteze ProbL kibem. Mel'euk.