Lietuvių kalbos sakinio sintaksinės ir semantinės struktūros vienetai

LIETUVIŲ KALBOS SINTAKSĖS TYRINĖJIMAI LIETUVIŲ KALBOTYROS KLAUSIMAI XXV (1986) VYTAUTAS AMBRAZAS SYNTACTIC AND SEMANTIC UNITS OF THE STRUCTURE OF THE LITHUANIAN SENTENCE 0.1. Studies of Lithuanian sentence structure have become more intensive in recent years, and the methods applied in this field are changing and becoming more diverse. They are directly or indirectly conditioned by the general theoretical principles, approaches, and models that prevail throughout contemporary linguistics. When intensive attempts to describe the structure of language while bypassing the meaning of its units failed to produce the expected results, the problem of the relationship between syntactic structure and semantics came to the fore. In one way or another, it is reflected in the work of researchers of Lithuanian syntax as well. Some rely first and foremost on the grammatical form of the sentence and its components, define the units of inquiry on that basis, and then seek to interpret them semantically, that is, to connect them with concepts at the semantic level. This approach was taken, for example, in *Lietuvių kalbos gramatika* (1965, 1971, 1976), and this article, too, intends to follow it. Others seek syntactic counterparts for semantic categories, or for categories of the sentence’s internal structure directly related to semantics (Валейка 1980, partly Paulauskienė 1979). A third path has also been proposed: when studying the sentence from a single perspective (syntactic form or meaning), to keep the other perspective in view as well (Sirtautas 1978; 1980; Labutis 1981). Each of these paths, explored to varying degrees, has its peaks and pitfalls; ultimately, the choice depends on the aim, purpose, and objectives of the work. However, in order for results reached by one route or another to be compared and used in further research, efforts should be made to define as clearly as possible and distinguish on a common basis at least the most important categories at the levels of form and semantics. This is an elementary requirement for any scientific theory, but it is also important for school grammar, which is increasingly lagging behind the level of contemporary linguistics. Unfortunately, in the work of researchers of syntax, especially what is called “semantic syntax,” we continually observe the conflation or ambiguous use of logical, pragmatic, semantic, and syntactic concepts, sometimes disguised by hyphenated terms: categories are described as logical-grammatical, syntactic-semantic, logical-syntactic, and so on. This gives rise to misunderstandings and fruitless discussions. The indeterminacy of syntactic and semantic categories cannot be excused by the specific nature of linguistic phenomena, their connection with thought, or their social character. The syntactic structure of language is far too complex to be studied successfully using vague or ambiguous concepts. In an effort to contribute to the refinement of the methodology of research into Lithuanian syntax, we shall discuss separately several of the most important units of the syntactic and semantic structure of the sentence. 4 0.2. Although syntacticians are concerned with the structure of both larger and smaller textual units than the sentence, the sentence is almost universally regarded as the primary object of syntax. No one has yet managed to formulate a comprehensive definition of the sentence that would meet the requirements of the various approaches to syntax. This is understandable: the primary concepts of all fields of science are difficult to define. Sometimes an attempt is made to define the sentence by listing the features considered most important (cf. Русская грамматика 1980, 85–90), but the number and nature of those features are conditioned by the system chosen. For specific syntactic research, what matters most is not so much a comprehensive definition of the sentence as its differentiation from other objects. The most important distinguishing features of the sentence are related to both the grammatical and communicative levels: the sentence is the smallest grammatically formed unit of communication. In this respect, the sentence differs from the morpheme or word, which also have grammatical form and are used in communication but cannot be regarded as its units; a word becomes such a unit only when it functions as a sentence in its own right. Nevertheless, it is not enough to state that a sentence has grammatical form. Linguistic forms (in the broadest sense) are of two kinds: dependent and independent. Dependent forms are those included in the composition of other forms by grammatical means, while independent forms are not so included. On this basis, the definition given above can be refined as follows: a sentence is the smallest unit of communication distinguished by intonation and having a grammatically independent form. Tokia sakinio samprata yra artima tai, kurią randame suformuluotą A. Meillet (1934, 355) ir L. Bloomfieldo (1935, 170) darbuose. Ji leidžia atskirti sakinį nuo kitų, platesnių komunikacijos vienetų (sakinių grupių, pastraipų ir pan.), kuriuos tyrinėja teksto lingvistika. Norint aiškiau diferencijuoti kalbos (langue) ir šnekos (parole) lygmenis, šnekos komunikaciniu vienetu derėtų laikyti pasakymą (énoncé, utterance), o kalbos vienetu — sakiniu vadinti pačią gramatiškai nepriklausomą pasakymo formą. Many features included in various definitions of the sentence have not been mentioned here: expressing a complete thought, having a structural schema (model), functioning as a unit of predication (or simply—predicativity), and so on. However, these features are not distinctive. Thoughts are expressed not only by sentences but also by word combinations, sentence constituents, and groups of sentences (the concept of a “complete thought,” often invoked, is itself vague). Not only sentences but also word combinations are formed according to structural schemas or models. Of all the features used in attempts to define the sentence, predication (or “predicativity”) is the least clear. Modern logic uses predication to mean the assignment of a property (predicate) to an object; this is how the term should be used. Such predication is also found in many word combinations (e.g., attributive ones). Some attempt to give the term predicativity a special, supposedly linguistic meaning of “relation to reality” (this is often done in Russian linguistics, cf. Виноградов 1975, 267; Русская грамматика 1980, 86). But not only sentences but also their parts express a relation to reality (cf. Ломтев 1979, 65). It is even harder to understand predicativity as some special force or indeterminate relation that turns a word combination or construction into a sentence (Мухин 1968, 77). It is no coincidence that M. Steblin-Kamensky compared this “predicativity” to phlogiston, once invented by chemists (Стеблин-Каменский 5 1974, 136—140). Such features, of course, do not help distinguish the sentence from other syntactic units. 0.3. Both simple and complex sentences have a grammatically independent form. Therefore, the term sentence is also applied to the latter, which some authors regard as combinations of sentences. The clauses of a complex sentence may correspond in their structure to sentences, but through intonation and, most often, conjunctions (that is, certain grammatical means), they are incorporated into a larger unit and are therefore not grammatically independent. The most important structural feature distinguishing a complex sentence from a simple one is its bicentric or polycentric nature. A simple sentence is one whose constituent parts are grouped around a single syntactic centre, while a complex sentence has two or more such centres. Since what follows will deal primarily with the simple sentence as the basic and simplest type of sentence, for convenience it will be referred to simply as the sentence. 0.4. In contemporary linguistics, the sentence is studied from various perspectives: as a form for realizing a thought expressed by a proposition (the logical perspective), as a means of expressing certain meanings and the relations between them (the semantic and semiological perspectives), and as the totality of grammatical relations between word forms (the syntactic perspective). From the standpoint of communication, it is important to establish which sentence components carry the greatest information load and how the relations among participants in the speech situation, and their relations to the content of the utterance, are reflected in the sentence (the pragmatic perspective). Each perspective yields different units, different “parts of the sentence.” Given the sentence’s complexity and multifaceted nature, it would be difficult to expect a method of division and analysis that revealed its properties from all the perspectives mentioned, or even several at once. The long experience of grammar shows that such efforts generally give rise to numerous uncertainties, contradictions, and ambiguities. Consequently, when investigating the relation between the syntactic form of the sentence and semantics, it is important to differentiate clearly between these perspectives. Only by identifying the principal units of inquiry at each level of analysis can we hope to determine their relations to one another more precisely. 1. UNITS OF THE SYNTACTIC STRUCTURE OF THE SENTENCE 1.0. The syntactic structure of a sentence consists of the totality of relations among word forms. We see relations among word forms in all sentences except one-word sentences (e.g., It is dawning; Attention!), whose structure consists of a single word form. A word form is a lexeme used in one of its characteristic morphological forms and defined in morphological terms (the genitive of a noun, the present indicative, first or third person of a verb, and so on). When a lexeme is invariable, the scope of the concepts word form and word coincides; otherwise, the word refers to the set of its paradigmatic forms. Word forms are the smallest syntactic units among which syntactic relations are established; they link morphology to syntax in much the same way as phonemes link phonology to morphology. In analyzing sentence structure, it is most important to identify relations among independent word forms, since function words (e.g., conjunctions or prepositions) are used in sentences as means of expressing syntactic relations and in this respect are similar to inflectional morphemes (Kurylowicz 1964, 21). Therefore, hereafter we will speak of syntactic relations as 6 relations among independent word forms (free forms), while treating function words (bound forms) only as indicators or modifiers of those relations. 1.1.0. Syntactic relations among word forms may be direct (i.e., immediate) or indirect (between word forms through other word forms). A word form is incorporated into a sentence through a direct syntactic relation; all its other relations are indirect. For example, in the sentence A miraculous light appeared on a high mountain (J. Biliūnas), the word form high is incorporated into the sentence through its relation with mountain, and light through its relation with appeared. L. Barčhudarov calls such relations syntactic functions (Бархударов 1966, 32). Hereafter we will call direct syntactic relations strictly syntactic relations. Indirect relations among word forms are not a special type of syntactic relation; they are more distant relations between word forms directly related at another level of division. A sentence containing direct and indirect syntactic relations forms a hierarchically organized structure of two or more levels (ranks). The following principal types of syntactic relations are distinguished: subordinate, interdependence (correlative), and coordinate relations1. 1.1.1. Subordinate relations occur between word forms such that one—the dependent form—is incorporated into the sentence through the other—the head form—but not vice versa. In typical cases, the dependent form2 has a particular formal marker of dependence. This marker may be an inflection, a function word, a particular position relative to the head word form, or intonation. In Lithuanian, the most important markers are inflections and function words (prepositions); the position of the dependent word usually plays a secondary role. Subordinate relations are unidirectional; this is represented by an arrow directed from the head (determined) form to the dependent (determining) form3. Subordinate relations may be strong (obligatory) or weak (non-obligatory). Strong subordinate relations occur when the dependent form is essential to preserving the grammatical structure of the sentence, e.g.: Mother finished eating; The soldier reported the news. In weak subordinate relations, the dependent form is optional: it may be omitted without violating the grammatical structure of the sentence, although its communicative content would thereby change, e.g.: Beware of bad people. We returned late. According to the formal expression of subordinate relations, three types of subordination can be distinguished in Lithuanian (as in many other inflected languages): agreement, government, and adjunction. 1 Plg. jau L. Hjelmslevo (1953) išskirtus determinacijos, interdependencijos bei konstelacijos santykius ir Trubetzkojaus (1939) žodžių junginių-sintagmų klasifikaciją į determinatyvines, predikatyvines ir sociatyvines (panašiai Pauliny 1981, 24–26). 2 In atypical cases, a form with a characteristic marker is replaced by another form without it, which has the same distribution. For example, in the sentence Mother brought something to drink, the infinitive form is used in a syntactic position of the accusative or genitive; although it does not have the inflection of these cases, the relation between it and the determining word brought is nevertheless considered subordinate. 3 Žodžių sintaksiniai ryšiai žymimi pagal Martinet (1973) siūlomus principus, plg. dar Мухин 1964, 42 tt.; 1968, 68 tt., tik prijungiamųjų santykių rodyklės kryptis atitinka priklausomybių gramatiką. 7 Agreement is the adjustment of a dependent word form to the head according to a morphological category. The dependent word form takes the case, number, and gender markers characteristic of the head form (cf. Sirtautas 1978, 50—54, with Lithuanian examples): a copper handle, all the friends, nine brothers, a flying bird. A special case of agreement is the coordination with other nouns of nouns that independently bear gender, number, and case categories and function as appositives: King Mindaugas, the late father, the widowed mother. Some propose treating coordination as a separate type of dependency because of the independence of the noun’s morphological categories (Скобликова 1971, 206—225; Labutis 1976, 30—32). However, this should not prevent us from treating coordination as a subtype of agreement, since in such cases we likewise see a dependent form being adapted to the head. Adjective and participle forms are similarly adapted to indefinite pronouns: someone angry, someone invisible, nobody foreign. Agreement usually expresses a weak dependency relation. Government is the selection of an inflected form of a dependent word (usually a particular one), determined by the valency and grammatical properties of the head word. The governed word’s form may be one of various cases without prepositions (prepositionless government) or with prepositions (prepositional government). Verbs and verbal derivatives are especially often the head word (writes a letter, while I am writing, the writing of a letter…), but nouns, adjectives, numerals, pronouns, some adverbs, and occasionally interjections and ideophones may also serve as heads, for example: a troop of soldiers, a field tree, a gold ring, dear to everyone, full of grain, a hundred soldiers, a trio of horses, several children, someone (from among) the neighbors, much water, far from home, shoo the money, a thump on the head. Where the dependency relation is strong, we have strong government, for example: I am finishing the work, there is a shortage of water, we represented the workers, I am interested in art, we lived in the city, I became a beggar, we jumped over the ditch. In such cases, the form of the governed word is most often determined by the valency of the head word, sometimes also depending on derivational properties (e.g., a prefix; cf. I swam across the river and I swam across the river). Less often, the form of the governed word depends on the grammatical properties of the head word, for example: bigger than a mountain, the highest of all, I felt sorry, as the sun was setting, or on the semantic relation between the governed word and the head, cf.: he brought the goods and he brought some goods. Where the dependency relation is weak, we have weak government, for example: I came for an axe, we work for ourselves, we traveled through forests, I went gray from worry, I sat with friends. Here the form of the governed word depends especially on its semantic relation to the head word. Adjunction is dependency that is not expressed by inflectional endings on words. In this respect, adjunction contrasts with agreement and government taken together. Most commonly, non-inflected word forms are adjoined: infinitives, adverbs, adjectives- 8 and participles. Their head words are verbs, their derivatives, adjectives, some nouns, pronouns, and adverbs. All the non-inflected word forms mentioned above may be adjoined to verbs. Adverbs, and to a limited extent infinitives, are adjoined to adjectives, for example: pleasant to look at, relatively small. Infinitives may be adjoined to verbal and some other nouns (hope to return, an opportunity to visit), and only in certain cases may adverbs be adjoined to them (a step back, life abroad). Only adverbs may be adjoined to adverbs, for example: quite well, very much. The absence of an inflected form is a feature of most adjoined forms, particularly prominent in inflectional languages. Together with the derivational formatives of adjoined forms and their position in the sentence, it helps identify the nature of the dependency relation. The frequently encountered view that the dependency relation of adjunction is indicated only by the meaning of words needs to be qualified. Sometimes we also find adjunction between inflected forms that are linked to other forms by double syntactic relations. In such cases, however, the inflectional endings of the adjoined words indicate not adjunction but a syntactic relation of another kind with another word. For example, in the sentences The passenger fell asleep while traveling; The boy came back happy, the inflectional endings of the word forms traveling, happy are markers of agreement with the forms passenger, boy, since they are not conditioned by an adjunction relation with the verbs fell asleep, came back. Adjunction is usually a weak relation. Strong adjunction occurs between certain semantically light (modal, phase, and similar) verbs and an infinitive, for example: stopped working, I tried to return, I can eat. Adverbs are also adjoined to certain verbs by a strong dependency relation, for example: I was here, we looked beautiful, behave like a man. 1.1.2. Reciprocal dependency (linkage) relations are bidirectional: the word form (the head) through which another form is incorporated into the sentence is formally dependent on the latter. In typical cases, a reciprocal dependency relation expresses coordination between forms. It is indicated by a two-headed arrow. The clearest example of a reciprocal dependency relation is the link between a verb and the nominative form of a noun or pronoun (the subject). The nominative form of the noun or pronoun depends on the verb, but in turn determines the number form of the verb. The finite form of the verb is also linked to the expression of the nominal constituent by first-, second-, or third-person pronouns (cf. aš grįžau, mes grįžome, jie grįžo); their coordination is also bidirectional, since there is no basis for treating the verbal category of person as dependent on pronouns. The finite form of the verb būti, when functioning as a copula, also determines the nominative form (e.g., Brolis buvo piktas); when the finite form is replaced by a participle or infinitive, the nominative must likewise be replaced by another case, cf. Brolio būta pikto, (Kaip) broliui būti piktam. Meanwhile, nominal (pronominal) forms linked by a copula agree with one another in case, number, and sometimes gender (for further discussion, see Sirtautas 1978, 57 ff.). The nominative position of a noun linked to a verb through a reciprocal relation may also be occupied by forms without inflectional endings, such as a verb infinitive or an indeclinable participle, cf.: Gyventi (gyvenimas) pakyrėjo; Buvo girdėti griudžiant 9 (griaustinis). In such cases, reciprocal dependency has no clear inflectional markers, but is indicated by substitution and the highly restricted mutual selection of the forms linked (cf. Sirtautas 1978, 56–57 with references). The question of reciprocal dependency relations between word forms is sometimes addressed differently. Some consider this relation to be subordinative, and the verb form, which agrees in number (and partly in gender in Russian) with the nominative form of a noun or pronoun, to be dependent and subordinated (as in Шахматов 1941, 31; Карцевский 1948, 33; more recently Елизаренкова, Топоров 1965; Цивьян 1979; Ломтев 1979, 52–53). The dependence of the nominative form on the verb is considered not syntactic but semantic or content-based (Ломтев 1979, 67); to explain it, predicative relations are distinguished alongside object, subject, locative, and other relations (Цивьян 1979, 239–241). Other researchers hold that the verb depends on the nominal form at the syntactic level, while the nominal nominative depends on the verb at the communicative level (Попов 1974, 42–55 with references; Щубик 1982, 132). However, numerous circumstances prevent us from denying the syntactic nature of nominative dependence. The nominative form in a sentence is conditioned by the syntactic valency of verbs: it is possible only with verbs that open a nominative position, and is impossible in sentences with verbs such as reikėti, užtekti, norėtis and the like. The syntactic dependence of the nominative form is also shown by pairs of constructions such as subėgo vanduo – pribėgo vandens, mes norime – mums norisi, liko pelenų – neliko pelenų, as well as by case change when the verb form changes, e.g.: mes einame – mums einant, tėvas atėjo – tėvo ateita and so on. Some compare this dependence to government (Мразек 1964, 60; Белошапкова 1981, 389–390). On the other hand, the verb’s dependence on the nominative form is much more limited: in Lithuanian, only first- and second-person verb forms used with personal pronouns agree in number, while the category of number itself is clearly secondary for the verb (cf. Lyons 1968, 281; Скобликова 1971, 201). Thus, the syntactic nature of the reciprocal relation between the constituents is quite clear. 1.1.3. Coordinative relations hold between two or more mutually independent word forms, each of which is incorporated into the sentence through an identical relation with another word form. This relation may be either reciprocal dependency (Pušys siūravo, braškėjo) or subordination (Atėjo ilgas ir lietingas ruduo). Because of their distinctive nature, coordinative relations are opposed to subordinative and reciprocal dependency relations taken together. Each word form linked by a coordinative relation can function without the other (or others); their relation has no direction. Coordinative relations are expressed not by word inflections but by coordinating conjunctions, word order, and intonation. They may be closed or open. A closed coordinative relation links only two word forms, usually joined by a coordinating conjunction placed before the second form, e.g.: Žiemos saulė šviečia, bet nešildo; Kieme mergaitės žaidė ir dainavo. An open coordinative relation links an indefinite number of word forms and is usually expressed either by repeated conjunctions or by word order alone (a contact 10 by the arrangement of forms and intonation, e.g.: Such roofs are pierced by neither stones, nor swords, nor axes, nor spears; She spoke timidly, peacefully, weakly. Word combinations linked by closed and open coordinating relations are often coordinated with one another, e.g.: He can be stern and angry, good and forgiving. 1.2. The forms of independent words that occur at the same level of syntactic expansion of a sentence and are linked by subordinating, mutual-dependence, or coordinating relations form word combinations. According to the types of syntactic relations, three largest groups of word combinations can be distinguished: subordinating, mutual-dependence (correlative), and coordinating combinations. Although all linguists agree that word combinations are units at the level of syntactic relations, they distinguish and interpret them differently. Often, only subordinating combinations are considered word combinations. This view was, and in part still is, quite widespread in Russian studies, especially after it was implemented in the first academic grammar of the Russian language (Грамматика русского языка 1952—1954). It is based on the hypothesis that so-called “word” relations differ from “sentence” relations (Виноградов 1947, 12, 78; 1954, 10 ff.; 1975, 231). According to this hypothesis, a word combination is a semi-lexical, nominative unit formed by “word” subordinating relations—agreement, government, and adjunction. It is regarded not as a component of sentence structure, but only as the “building material” of a sentence, similar to a word. Accordingly, word combinations themselves are classified and described according to the lexical and morphological features of the head and dependent members. “Word” relations, on the basis of which a word combination is formed, are strictly distinguished from “sentence” relations—predication (defined as the relation of an utterance to reality) and relations between secondary sentence parts. Thus syntax is divided into two parts: the syntax of word combinations, which studies the relations of words as lexical units to other words, and sentence syntax, which studies sentence structure and its constituents. This view, which has also been applied to Lithuanian (Balkevičius 1963; Labutis 1976, 42—50), is difficult to justify theoretically, while its practical implementation leads to considerable confusion, repetition, and contradiction. The distinction between “word” and “sentence” relations is precluded above all by the fact that the lexical meaning of words, especially verbs, determines not only the word combination but unquestionably also the structure and model of the sentence (cf. Шмелев, 1976, 43 ff.). Syntactic processes, above all passive transformation, clearly show that government relations between a verb and dependent word forms, which are regarded as “word” relations, play no lesser role in sentence structure than “sentence” mutual-dependence relations, and sometimes an even greater one (cf. the elimination of the nominative form in a passive or nominalized construction: the father writes a letter, but the letter is being written, the writing of a letter (see also Apytūfonova, 1976, 125—126, 134 with references). A verb’s valency with dependent words is also reflected in sentence models. It is quite impossible to prove that combinations such as I am writing a letter, I am going to the forest are compound names. At most, some lexicalized subordinating word combinations, especially two-word names of things and persons, can be regarded as nominative units. On the other hand, the analysis of verb valency shows that mutual-dependence relations, considered “sentence” relations, also depend on the lexical meaning of the verb; these are relations between certain word forms, expressed through agreement and having formal markers. It is true that a sentence can be formed on the basis of these relations. 11 but this too is not their distinguishing feature, since a sentence can also consist of subordinating combinations (I see danger; I am cold; There is a shortage of money; Catch the thief!), whereas many mutual-dependence combinations do not form a sentence on their own (*I thought; *The child jumped, etc.). Even in grammars that declare a contrast between subordinating word combinations and sentence structure, so-called secondary sentence parts are described on the basis of the same relations that link word forms in combinations. It is therefore quite understandable that the aforementioned hypothesis of two types of relations has had no significant impact on the study of the syntactic structure of other languages (except Russian). Almost all syntactic theories currently dominant in world linguistics regard word combinations as units of sentence syntactic structure. Some recent works in Lithuanian linguistics have also taken this approach (Lietuvių kalbos gramatika 1976; Sirtautas 1978). Word combinations are identified at various levels of sentence analysis. They are part of a sentence and, in certain cases, may correspond to a sentence in structure, but they are not sentences themselves. Word combinations, or hierarchically organized groups of word combinations (like individual word forms), become sentences only when they are formalized according to certain models, linked to a speech situation, and given the intonation characteristic of a sentence. Thus, a description of word combinations does not yet reveal the entire structure of a simple sentence. It is also necessary to establish the models according to which the core of a sentence is formed from word forms and their combinations, as well as the rules for expanding those models. 1.3. A sentence model includes those word forms that are necessary for the structure of the sentence as a grammatical unit. A sentence model may consist of one word form (It is snowing), two (Mother arrived), three (Uncle shot the fox), and sometimes more (Grandfather asked the orphan for bread). Two-member or multi-member models are formed on the basis of mutual-dependence and/or subordinating relations. Mutual-dependence relations (alone or together with subordinating ones) are typically found in models of personal sentences, while only subordinating relations are found in impersonal sentence models (The neighbor is doing well; People gathered, etc.). Linguists who follow the aforementioned hypothesis distinguishing “word” relations from “sentence” relations do not include many obligatory members linked by subordinating relations in the sentence model (cf. Шведова 1966; Белошапкова 1977; 1981; Русская грамматика 1980); however, such a restriction is neither necessary nor sufficiently well-founded theoretically (cf. Кржижкова 1967, 76; Helbig, Schenkel 1973, 34; Москальская 1974, 27; Распопов 1976; Золотова 1982, 83 ff.; Ломтев 1979, 178 ff.). The members of a sentence model are usually designated either according to the grammatical features of word forms (approaches based on valency theory) or by the terms for sentence parts (relational grammar, so-called universal grammar, and some approaches to functional grammar). This brings us to the much-discussed but still highly topical problem of sentence parts. 1.4.1. The traditional conception of sentence parts, inherited from K. F. Becker’s system (1827), has long been justifiably criticized (cf. as early as Steinthal 1855; more recently Glinz 1962; Бархударов 1966; Helbig 1982, 45–54; Золотова 1979, among others). Its principal shortcoming is the conflation of grammatical and semantic criteria, which gives rise to contradictions 12 in every description of sentence structure constructed on their basis4. They have not been avoided even in the most extensive descriptions of Lithuanian sentence structure, where sentence constituents are either defined according to meaning but practically distinguished by syntactic position (Balkevičius, 1963), or some are defined according to syntactic position and form (the subject), and others according to meaning and their relation to other words (Lietuvių kalbos gramatika 1976). In fact, the concepts of sentence constituents are not necessary for describing the structure of a simple sentence. Units called subject, complement, modifier, or adverbial in school grammars and many linguistic works can be described as dependent components of phrases that have a particular meaning (for the meanings of syntactic units, or semantic relations, see 2). The information found in chapters on sentence constituents can be presented consistently in discussions of phrases (in terms of syntactic links and semantic relations), sentence patterns, their modifications, and expansion. Therefore, works on syntax that seek methodological precision either dispense entirely with sentence constituents (especially so-called secondary ones) or attempt to distinguish them according to other criteria. The first approach is followed mainly by researchers representing certain strands of dependency grammar and valency theory, as well as role and referential grammar (cf. Van Valin, Foley 1980). Among those taking the second approach, one may first mention the representatives of so-called standard transformational theory. According to this theory, sentence constituents (subject, predicate, complement) are taken to be relations between elements of the basic sentence structure: the subject is defined as the relation of a noun phrase (NP) to the sentence as a whole, the complement as the relation of an NP to a higher-ranking verb phrase (VP), and so on (Chomsky 1965, 10 ff., 13). This conception of sentence constituents is based on the hypothesis of a deep syntactic sentence structure having the form NP∧Aux∩VP, which theoretical research over the past decade has considerably undermined. Applying the terms for sentence constituents to the units of immediate sentence structure is inconvenient, if only because in that case they would have to be given an entirely different meaning: for example, the whole subject group would have to be called the subject, the predicate group the predicate, and so on. At present, the terms for sentence constituents are increasingly used to denote sentence-structure units dependent on the predicate or simply on the verb. This view is followed by approaches whose principles are close to dependency grammar; researchers who treat the sentence as a “system with relations” take a similar view of sentence constituents (Ломтев 1979). The technique of distributional analysis established by Ch. C. Fries (1952) is used to identify individual sentence constituents (Erben 1964, Flämig 1970). In typologically comparing languages of different systems, researchers seek universal properties of sentence constituents (Comrie 1974; 1981; Johnson-Postal 1980; Perlmutter 1980). Particularly great efforts have been made to find a universal definition of the subject suitable for languages with different structures. For this purpose, J. Keenan (1976) proposed first distinguishing basic (primary) sentences from derived (secondary) ones (307), and seeking the subject only in basic sentences, using a scale of 30 properties (312—331). None of these properties is considered necessary, but the sentence constituent that has the greatest number of them is considered closest to the subject, “most subject-like” (333). It is understandable that such an approxi- 4 The shortcomings of the traditional classification of sentence constituents were also clearly recognized by J. Jablonskis, who emphatically pointed out its conditional nature (1957, 406, 450 ff.). 13 mate conception of the subject is difficult to apply to the description and study of the grammatical structure of particular languages and has attracted justified criticism (Johnson 1977 a; Schmalstieg 1980, 166). Word order is also invoked in the search for universal properties of the subject; cf. J. Stepanov’s proposed definition of the subject (and topic): “the first noun group in sequence or the first noun in a sentence with neutral word order” (Степанов 1981, 173). In languages where word order depends considerably on the actual division of the sentence (as in Lithuanian), a sentence with the same lexical composition and grammatical structure may have more than one neutral-order pattern; in such a case, identifying the subject/topic by the above definition becomes dependent on what is regarded as the sentence topic. It should be noted that other researchers seeking universal properties of the subject and other sentence constituents are currently trying to rely not so much on syntactic as on pragmatic and semantic criteria. The subject/topic is considered to be the intersection of “topic” (close to theme) with the semantic role of agent (Bechert 1978, 19; Comrie 1980, 101), a combination of agent and “topic” with the nominative (Abraham 1978, 716), “the initial identifiable point to which information is added” (Chafe 1976, 44), and so on. Representatives of relational grammar completely refrain from defining sentence constituents (“grammatical relations”), regarding them as primitive concepts (primitives) that emerge in specific syntactic operations (Johnson 1977, 153; Perlmutter 1980, 197; for criticism of this view, see Schachter 1977, 301 ff.; Abraham 1978; Van Valin, Foley 1980, 348; Кибрик 1982, 27–28). Is it possible at all to formulate syntactic and, at the same time, universal definitions of sentence constituents that would allow sentence-structure units to be identified in languages of different structures? It is difficult to answer this question affirmatively. Given the great diversity of the syntactic structures of natural languages, sentence constituents distinguished by syntactic criteria in one language will have clear counterparts in another only in certain cases, and the differences will be greater the more typologically distant the languages being compared are. A system of semantic coordinates, defined independently of syntactic properties, is necessary for the typological comparison of languages. The position of syntactic units with respect to that system will differ to a greater or lesser degree in each language. To avoid a vicious circle, we must also define sentence constituents as syntactic units independently of categories at the semantic and communicative levels. 1.4.2. When attempting to distinguish sentence constituents from a syntactic perspective, we first encounter the question of what should be regarded as the center of a simple sentence. The model of two main sentence constituents, still prevalent in many grammars, is difficult to justify theoretically and poses many problems, especially in describing ergative languages, where not only the subject but also the complement is often linked to the verb by a relation of mutual dependence, so that the syntactic distinction between these two sentence constituents becomes blurred (Чикобава 1967, 19 ff.; Tschekhoff 1978, 28–30; Генадзе-Гайдарова 1982, 174 ff.). However, even in accusative (or nominative) languages, there are plenty of examples showing that the distinction between subject and complement is in many cases not fundamental; compare, for example, sentences such as Skauda galvą and Skauda galva. That the subject is not essential to sentence structure is shown by impersonal sentences and, in many languages 14 (including Lithuanian) are characterized by personal sentences without an expressed agent (I live in the countryside, etc.). The predicate, by contrast, is indispensable in every sentence (cf. Danės 1978, 8). In terms of the organizing role of the sentence, the subject is essentially no different from the object: apart from the predicate, the other components of the sentence’s structural core are not directly connected to it; moreover, the subject can have only the same dependent words as the object. The subject cannot be assigned the role of representing the sentence merely because it lacks the modal categories characteristic of a sentence. The subject is drawn to the center of the sentence mainly not by syntactic arguments, but by the tradition based on Aristotelian logic (the division of the sentence into subject and predicate), by actual division, and by considerations of a semantic nature (its connection with the theme, its being the principal means of expressing the agent, etc.). Attempts by transformational grammar to revive the principle of the bicentral division of the sentence were strongly conditioned by an orientation toward the structure of English and other modern European languages (cf. Lehmann 1973, 47 ff.; 1974, 14–15). On the other hand, the organizing role of the predicate as the center of syntactic relations in the sentence of any language is beyond doubt. The predicate is autonomous in the sentence in the sense that its function is evident from itself (cf. Martinet 1966, 56); having modal categories associated with the speech act, it represents a simple sentence in a broader context (Kurylowicz 1960, 38). The decisive importance of the predicate in the sentence is also evident when variations in word order are studied from the perspective of syntactic typology (cf. Steele 1978, 604). It is therefore quite understandable that most current theories of syntax model the structure of the simple sentence according to the principle of monocentricity: the predicate is regarded as the principal part of the sentence, and the other parts of the sentence are related to it in one way or another. The theory of valency is developed in accordance with this principle, most clearly formalized in so-called dependency grammar (e.g., Hays 1964; Падучева 1964; Robinson 1970; Гладкий 1973; Hudson 1980); it also underlies the typological study of languages on the basis of diatheses (in the works of A. Cholodovičius, V. Chrakovskis, G. Silnickis, and others). Recently, the monocentric model of the sentence has also become increasingly established in Lithuanian linguistics (Balkevičius 1983; Генюшене 1983; Slizienė 1978, 1980). It is also superior in that it can be interpreted semantically in accordance with contemporary predicate logic, which distinguishes the predicate and its associated arguments in a sentence. By contrast, the subject–predicate dichotomy characteristic of Aristotelian logic corresponds more closely not to the syntactic, but to the communicative division of the sentence into theme and rheme. 1.4.3. Researchers who view sentence parts as syntactic units endeavor to define them according to syntactic position, which is established using distributional analysis and the substitution test (classic examples are Fries 1952; Hockett 1958). Questions asked on the basis of the verb may also be regarded as a simplified use of substitution; this method of identifying sentence parts is especially widespread in school practice. Pronouns and pronominal adverbs used to form questions, owing to their highly generalized meaning, are often fairly good indicators of syntactic position (cf. Ревзин 1977, 186–190). In Lithuanian linguistics, V. Sirtautas (1978, 1981) distinguishes sentence parts most consistently according to syntactic features; his ideas on this matter therefore deserve particular attention. 15 V. Sirtautas points out that the syntactic position characteristic of a sentence part, manifested in its relations with other sentence parts, depends on various factors: valency, the syntactic environment, the morphological nature of the word, the lexical meaning and semantics of the word, the nature of grammatical relations, word order, and intonation (1981, 25 — 27). Although not all these factors are considered equally important, it is nevertheless proposed that all should be taken into account. Sentence parts are characterized by bundles of syntactic features (1981, 33). It should be noted that some of these features are attributed to at least two sentence parts, while others are specific to one or another. For example, the organizing role of the sentence, independence, and the relation of one to the other (the subject being “right-hand,” the predicate “left-hand”) are considered features common to the subject and predicate; their distinguishing features are considered to be morphological basis (nominative for the subject, verb for the predicate) and the fact that the subject optionally requires an attribute, while the predicate requires an object and adverbials. Of the three features subsequently listed for the object and adverbials, two are shared (dependence on the predicate, optional requirement of an attribute), while only one is specific—the morphological basis (an oblique case for the object, an adverb for the adverbial). Thus, the criteria for distinguishing sentence parts vary considerably and are not applied equally to all of them. Nor are all equally clear and indisputable. It has already been mentioned that assigning the organizing role to the subject is difficult. The feature of independence attributed to the subject and predicate may also raise doubts, since the two sentence parts are linked by a relation of mutual dependence, which prevents either from being regarded as completely independent. On the other hand, only the predicate is autonomous in the sense that it can by itself form a sentence. The proposed criteria for distinguishing sentence parts can therefore be refined. It would be best to define sentence parts as syntactic units of sentence structure on the basis of a single principal criterion—namely, syntactic relations, which determine their position. The predicate is the syntactic center of a simple sentence, the head of all its syntactic relations. Because of this special position, it does not stand in opposition to the other sentence parts at the same level, and therefore a more detailed definition is not required here. The subject should be defined as a sentence part that, in typical cases (cf. 1.1.1.), is related to the predicate by a relation of mutual dependence (interconnection). In Lithuanian (and many other languages), the subject is also readily identified by its characteristic nominative position, although by no means every nominative form in a sentence should be considered a subject. The object is a sentence part linked to the predicate by a subordinating relation. Sentence parts connected to one another by a coordinating relation and equally linked to other sentence parts may be called homogeneous. Since syntactic relations may be one-place or two-place, in addition to the aforementioned sentence parts connected to the predicate by one-place relations, it would also be necessary to distinguish a sentence part connected both to the predicate and to the subject or object—that is, a predicative attribute (e.g., The child got up cheerful; I knew her younger, cf. Valec­kienė 1967). All other attributes have no direct relation to the syntactic center of the sentence—the predicate; they are distinguished at lower levels of sentence analysis and should be regarded not as sentence parts but as certain modifiers of sentence parts. Word forms expressing objective and adverbial relations and dependent on the subject, object, predicate, or other attributes are not sentence parts either. 16 Sentence parts defined in the manner described are conditioned by certain specific features of the syntactic structure of a language and are therefore not universal. This conception of them is convenient to apply to inflectional languages of the accusative (or nominative) structural type, to which Lithuanian also belongs. In many languages of the ergative structural type, distinguishing the subject from the object on the basis of their mutual dependency relationship with the verbal predicate is inappropriate, if only because forms of nouns denoting an object, corresponding to the object in accusative languages, are also linked by this relationship (cf. the so-called object agreement in Caucasian and other languages; see Чикобава 1967, 19–20; 1968, 243; Moravcsik 1978, 364 ff.). 1.4.4. Modifiers of sentence parts, which we call attributes5, are linked to the words they modify by subordinating relations. Attributes, in turn, may themselves have other attributes, and so on. Thus, in a typical sentence we see several hierarchical levels—or ranks—of syntactic relations. At the first level are the sentence parts directly linked to the predicate—the subject and objects; at lower levels are attributes of various ranks. Take, for example, the sentence: On Sunday evening, a cart full of old harness stopped beside a completely dry well. Its syntactic relational structure can be represented as follows: Here, attributes of three ranks are linked to the subject and attributes of two ranks to the object, forming certain groups (cf. Sirtautas 1981, 31). Attributes are linked to sentence parts and to one another by various types of subordination: agreement (full cart, old harness, dry well), government (full of harness, on Sunday evening), and adjacency (completely dry). Syntactic relations are more problematic in cases where the positions of sentence parts are occupied by constructions with non-finite verb forms, whose meanings are close to those of subordinate clauses in complex sentences and which may be regarded as transformations of embedded clauses, e.g.: As I was drawing water, a narrow wind rose; Kastis thought her to be asleep. Words dependent on non-finite verb forms are linked in the same way as those dependent on finite forms, except that subordinating relations correspond to the relations of mutual dependency; cf.: I draw — as I was drawing, she sleeps — her to be asleep (cf. aut. 1979, 168 ff.). Apparently, words directly linked to non-finite forms 5They are also called components of sentence parts (Balkevičius 1963). 17 should nevertheless be regarded as objects in such cases, and when syntactic relations are represented graphically, the construction itself should be marked with the symbol for an embedded clause, e.g.: 1.4.5. When modeling the structure of a simple sentence in terms of sentence parts, the question arises: should only the verb (expressed or functioning in a null form) be regarded as the syntactic center of the sentence—the predicate—or should other word forms closely associated with the verb be included in the predicate? This is the problem of the compound predicate, whose solution depends on the criteria selected for sentence analysis and on how consistently they are applied. Lithuanian grammars and many grammatical studies distinguish two types of compound predicate: verbal (or analytic) and nominal (or compound). A fairly diverse range of constructions is classified as compound predicates. We will briefly discuss their main groups, relying on syntactic features. a) Verb constructions with the so-called object infinitive (ordered to return, asked for mercy, brought food and so on; see Lietuvių kalbos gramatika 1976, 341—342) raise no doubts. Syntactically, they can be divided into a verb—the predicate—and an object dependent on it. The heterogeneity of these constructions is clearly demonstrated by substitution with subordinate clauses or case forms; cf. ordered that he return; asked that they show mercy; brought food to eat, asked for mercy. Constructions with the so-called subject infinitive are more problematic, especially those formed with modal and phase verbs (e.g.: we want to eat, I can work, must know, one must believe; began/finished singing). Their analysis has been seriously debated in Lithuanian linguistics (Sirtautas 1968, 346 ff.; 1978, 41—42). Although there are arguments both for and against, strict adherence to syntactic criteria makes it entirely possible to divide many of these constructions into a predicate and an object dependent on it (cf. we want bread, one needs faith; began/finished a song); only with verbs such as have, can, and the like are action nouns generally not used (but cf. I can do everything, I can do nothing). The fact that constructions with phase verbs begin, start can be replaced by a single inchoative verb (start to sing, chirp, begin to cry) is not decisive, since a verb often stands for a phrase in other cases as well; cf. go outside — exit, go around — go around (circumvent). The infinitive is sometimes included in the predicate because in the constructions mentioned it carries the main informational load, while the verb is semantically non-autonomous. However, at the level of syntactic sentence analysis, it is necessary to set aside matters pertaining to the semantics of word forms and the logical structure of the sentence. Constructions such as came to realize, decided to leave and the like are clearly divisible into a predicate and an object. 18 b) In verb constructions with participles (felt condemned, said that he had eaten, believed he would recover, thought he knew, pretended to be invisible, and so on), we likewise distinguish two components: the predicate and a predicative attribute or complement associated with it. That the participle here occupies the position of a separate sentence constituent is evident from its substitution by a subordinate clause, cf.: felt that he was condemned, said that he had eaten, believed (that) he would recover, etc. The verbs mentioned are generally not used without participles or subordinate clauses, but this indicates not the syntactic indivisibility of the construction, but simply a strong subordinating relation (cf. 1.1.1.) connecting the verb with the participle in a way similar to its connection with a case-marked complement, cf. I know that he has returned and I know the way, but not *I know. (For more on the syntactic structure of these constructions, see the author 1978, 7 ff.). c) Constructions of neuter-gender adjective forms with infinitives (it is pleasant to write, difficult to work, good to travel, boring to rest...) have been syntactically analyzed by V. Sirtautas (1978, 37—41); one can only endorse the author's firmly substantiated conclusion that they, too, are divisible into a predicate and a subject. It should only be added here that in the constructions it is a pity to part, little to believe, respectively, one should distinguish a predicate and a complement (cf. a pity about the parting, little belief). d) In constructions of the verb būti with nouns and adjectives, some researchers regard only the verb functioning as a copula as the predicate (Glinz 1965; Erben 1964; Grebe 1959; Flämig 1970; Рожновская 1979, etc.). This view can be motivated by the nature of syntactic relations: in Lithuanian, as in many other languages, the verb būti forms with nominals constructions of the same form as those formed by independent verbs, which are all considered predicates, cf. he walked cheerfully and he was cheerful. However, in a sentence the verb būti, when functioning as a copula, operates as a syntactically empty morpheme, imparting to the property expressed by the nominal only temporal and modal meanings (cf. Lyons 1968, 322—323; Babby 1975, 76 ff., 89). In many sentences it has a zero form, and in some types of sentences (e.g., the impersonal sentences mentioned in 1.4.5 c) the zero form of the present-tense copula is predominant or even the only form. It is difficult to regard such a word, lacking an independent syntactic function, as the sole main constituent of a sentence, especially given that in many languages the copula is not a verb but a word form or morpheme of pronominal origin. The structure of sentences that never had a copula and are inherently nominal also contradicts the view that the copula is the main constituent of the sentence (author 1979, 184 with references). Therefore, the verb būti, functioning as a copula in personal or impersonal sentences (e.g., it was good, it will be pleasant), is best regarded not as an independent predicate but as part of the predicate, as is done in existing Lithuanian grammars and most studies of syntax, including L. Tesnière’s model (1959, 47, 72), which is based on syntactic relations. We should view the verbs pasidaryti, tapti, virsti, similarly; they differ from the “pure copula” būti only in expressing a change of state. Other verbs, hitherto classified in Lithuanian grammars as “impure copulas,” are much more independent and, when serving as predicates, often have attached word forms and constructions, e.g.: seemed unattractive, stayed in the forest, felt bad, was called strangely, pretended not to see. Their combinations with nominal nominative or instrumental forms (e.g., seemed unattractive, remained lonely, felt— 19 continued offended, was called oak/oaks, pretended to be drunk) can likewise be syntactically divided into a predicate and a predicative attribute associated with it, just like the constructions of other verbs (cf.: lived lonely, got up cheerful, walked offended). If constructions of būti with nominals are regarded as compound predicates, it is nevertheless necessary to bear in mind that each component of this predicate has its own distinct syntactic relations with other word forms (cf. Бурлакова 1975, 62—63; Шмелев 1976, 60—63). For example, in the sentence The old man was strong, the verb buvo is connected to the noun senis by a mutual-dependence relation, while the adjective stiprus has two-way syntactic relations: with both buvo and senis. Similar relations are found in the corresponding passive construction, The old man was strong, except that here the form būta is connected with senio by a subordinating, not a mutual-dependence, relation (see also 3.3). The verb būti functions not only as a copula marking tense and modality, but is also used in the sense “exist” (on the relation between these meanings, see Benveniste 1960; Kahn 1966; Babby 1975, 79—80). In Lithuanian, the verb būti with an existential meaning can be regarded as the same kind of predicate as the verbs egzistuoti, gyventi, augti, atsirasti, dingti, and so on. Existential meaning is organically connected with coordinates of place and time (cf. Селиверстова 1977, 11 ff.), which in a sentence are often expressed by adverbs, noun cases, or prepositional constructions, e.g.: The guard is at school; The tree was by the road; The meeting will be tomorrow. In such sentences the verb būti plays a role similar to that of other verbs of state in the sentences The guard lives at school; The tree grew by the road; The meeting will take place tomorrow; thus, here too it is best regarded as a predicate. Of course, this syntactic analysis of sentences of the type under discussion does not require that they be divided correspondingly from a logical and semantic point of view (on the close relation between locative and property meanings, see Lyons 1977, 473). 1.4.6. Further subclassification is possible within sentence constituents. This is particularly relevant for the sentence constituent called the complement, distinguished on the basis of highly varied subordinating relations. The question of adverbials arises first. So far, no one has succeeded in defining adverbials as a separate sentence constituent on syntactic criteria and clearly distinguishing them from complements, although considerable effort has been made. The traditional conception of adverbials, also characteristic of school grammar, is based on meaning and thus belongs to an entirely different—semantic—level of analysis. At the level of syntactic relations, adverbials should be viewed as a particular subgroup of complements. The following ways of distinguishing adverbials seem possible: a) Treat adjunct complements as adverbials and distinguish them from other—governed—complements. However, this approach is questionable because of the very formal, “surface” distinction between government and adjunction (cf. 1.1.1.), which is determined by the morphological properties of the forms connected. Adverbials distinguished by this criterion would be very unlike those usually classified as such today; for example, infinitives in the combinations need to go, started to travel, want to live, and so on, would have to be considered adverbials, whereas accusative forms in combinations such as arrive at night, weighed three kilograms would have to be excluded from that category. Such a change in the scope of the term would offer little benefit, since forms attached by adjunction can simply be called adjunct complements. 20 b) To treat as adverbials complements that are not necessary to the sentence’s syntactic structure (i.e., that do not belong to the sentence nucleus). This approach can be justified by valency theory, which distinguishes obligatory and optional constituents (actants) from so-called free constituents (circonstants or situants). Only constituents governed by verb valency are considered complements in works representing the Leningrad School of Typology, which advocates the theory of diatheses (e.g., Храковский 1974; 1981); in Lithuanian linguistics this principle was implemented by E. Geniušienė (cf. Генюшене 1983, 16). Distinguishing adverbials from complements according to whether or not they belong to the nucleus of sentence structure is undoubtedly meaningful from a syntactic point of view. However, when applying this criterion, the term adverbial would also have to be used in a quite different sense from that in which it has been used up to now. For example, in the sentences The brother behaves well; We live in the city, the word forms well, in the city could not be considered adverbials, since they are necessary to the structure of the sentences in question and are governed by verb valency; whereas the same word forms in the sentences The brother studies well; We spent the night in the city would have to be considered adverbials. Since the term adverbial is intended to be retained primarily because it is familiar, transferring it to designate entirely different things would hardly seem acceptable, especially since the distinction between obligatory and non-obligatory constituents can be expressed by other terms used for this purpose (actants–circonstants, obligatory and optional–free constituents, etc.). c) To treat as adverbials complements occupying a position characteristic of adverbs. This approach is based on the theory of primary and secondary functions of word forms (Slotty 1932; Kurylowicz 1936), which is also applied to the analysis of Lithuanian sentences. A comparison of the sentence-part properties identified by V. Sirtautas (1981, 33) shows that the sole distinguishing feature between complements and adverbials is the difference in their morphological base. Thus, adverbials may be understood as those complements that either themselves have an adverbial form or can be replaced in a sentence by adverbs without disrupting its structure. Adverbs with generalized meanings would be good indicators of an adverbial position: then (respectively, from then, until then…), there (from there, up to there…), so, therefore, so much, and so on. If adverbial substitution is taken as the distinguishing feature of adverbials, word forms or prepositional constructions that can be replaced both by adverbs and by case forms (with or without prepositions) should be classified as adverbials, while those that cannot be replaced by adverbs with generalized meanings should be classified as direct or indirect complements. For example, the following prepositional constructions would fall among adverbials: we returned from the party; better from the jug; I pulled my hand out of my pocket (from where/from what); would go to her mother (where/to whom); we sat down around the little lake (where/around what); we went out into the fields (where/into what); he read a letter under the table (where/under what); he laughed through his tears (how/through what); he built the buildings according to the plan (how/according to what), and so on. By contrast, indirect complements should be identified in expressions such as: he laughed at his father, he made a snowman out of snow (out of what), he fought with the enemies (with whom), he hid from the dudra (from what), he is driving without a matchmaker (without whom), it smelled of tar (of what), and so on. Such a conception of adverbials would deviate least from the traditional one while also being grounded in a grammatical criterion. Moreover, in this case one can make use of the question test applied in school practice, which, as already noted, is a simplified version of the substitution test. Only those adverbials that lack adverbial correlates (e.g., invited for lunch; postponed 21 the meeting until tomorrow), would have to be distinguished by means of other generalized questions. Nevertheless, it is not always possible to draw a clear boundary between adverbials and indirect complements in the manner described, and constructions at the periphery of usage will occur that can be interpreted in two ways. Given the field structure characteristic of grammatical-unit classes (cf. Адмони 1964, 47–51), such fluctuation at the boundary may be theoretically justified. 1.4.7. By tradition, complements are also divided into direct and indirect. Research in relational grammar shows that this distinction is syntactically significant in many cases (Johnson 1979; Johnson, Postal 1980, Perlmutter, Postal 1977; Perlmutter 1980; cf. also Gary, Keenan 1977; Comrie 1981, 60–67). However, it is not easy to define the difference between direct and indirect complements in terms of grammatical properties. External morphological indicators are far from sufficient, cf. we returned at night (adverbial), but we are waiting for night (direct complement). Apparently, one should rely primarily on syntactic processes here, among which passive transformation plays the most important role. When an active construction is changed into a passive one, the direct complement in Lithuanian becomes the nominative subject, whereas the indirect complement, like an adverbial, generally retains its adpositional relation, case form, and syntactic position (cf. the father bought the child a hat and the hat was bought for the child by the father). On this criterion, direct complements should include: a) accusative forms linked to verbs by a strong adpositional relation, which change the type of relation and become nominative forms in passive constructions; b) the corresponding genitive forms in constructions with negation (he did not buy fruit, I did not see the road); c) genitive forms dependent on the verbs want, ask, desire, covet, miss, avoid, seek, look for, wait for, and so on (cf.: desired food, requested book, achieved goal, awaited news). Apparently, the dative associated with the verb represent should also be included in this group (cf. the company represented by the director). Nevertheless, there are several verbs with which associated complements are changed into nominative forms in passive constructions (especially in Samogitia) in certain cases, but should not be considered direct complements, particularly since another, clearly direct complement is used alongside them; cf.: I give the child an apple – an apple given to the child; she gave the old woman a spoon – a spoon given to the old woman; cf. also: I did not sleep all night – the whole night was not slept by me. Dative-form complements in constructions such as help the pupil (cf. the pupil helped by the teacher), order/command the soldier (cf. the soldier ordered/commanded by you) are also problematic in this respect, as are genitive-form complements alongside accusative ones, e.g.: ask the father for a knife. In addition to the indirect complement, relational grammar also distinguishes the so-called oblique object, which roughly corresponds to most syntactically non-obligatory adverbials and to complements with instrumental and benefactive meanings. Although the criterion for identifying an oblique object is not clearly stated, it is essentially semantic: the term denotes relations having “independent semantic content” (Johnson 1977, 153). Therefore, there is no need to distinguish an oblique object in addition to adverbials. Thus, the following subclassification of complements seems possible: at the first level of division, according to their differing participation in syntactic processes (first and foremost—passive- 22 in the transformation) distinguish direct and indirect objects; at the second stage of classification, distinguish adverbials from indirect objects on the basis of the syntactic position characteristic of adverbs. 1.4.8. The sentence constituents discussed are defined according to the same criterion as word combinations—that is, according to the syntactic relations between independent word forms. The units referred to here as sentence constituents can also be described in terms of word forms, with the corresponding syntactic relations specified. For example, the subject is a particular word form occupying a nominative position linked to the verb by a relation of mutual dependence; the object is a word form linked to the verb by a subordinating relation. However, the terms for sentence constituents, although considerably compromised by the mixing of syntactic and semantic criteria characteristic of traditional grammar, are not only shorter but also more convenient in many cases. Of course, they need to be made more precise and given clear syntactic content. If distinguished on the basis of defined syntactic features, the units of the sentence’s syntactic structure could continue to be called sentence constituents both in linguistic scholarship and in school practice. 2. UNITS OF THE SEMANTIC STRUCTURE OF THE SENTENCE 2.0. The semantic structure of a sentence consists of the meanings expressed by mutually related word forms and the totality of relations among them. Relations between word meanings in a sentence are realized not directly, but through a network of syntactic units and their relations specific to each language. Therefore, there is no direct correspondence between the semantic and syntactic structures of a sentence. If we tried to define semantic relations in terms of syntactic relations, we would inevitably fall into the same vicious circle as when distinguishing syntactic units on the basis of meaning. The most general units of the semantic structure of a sentence are distinguished with reference to the universal logical division of the sentence and the generalized meanings of its components. 2.1. In current logic, the semantic center of a sentence (S) is represented as a propositional function (P) relating a predicate, or property (F), to one or more arguments, or things conceived of: P ~ F(x...). The linking of a thing (object) with a property is called predication. The communicative orientation of the sentence (K) and the categories of modality associated with the speech act, which together constitute the so-called modal frame of the sentence (M), are usually placed outside the propositional function, i.e. S ~ K(M)P)). Depending on the number of arguments, propositional functions with one-place predicates—properties—of the form F(x) are distinguished from propositional functions with multi-place predicates—relations (R)—of the form F(xy...) or xRy... By defining one argument in terms of another, multi-place functions can be converted into one-place functions (Kappu 1969, 62; McCawley 1981, 395–396), but this transformation of the function is not useful for semantic analysis of language, since it oversimplifies the original schema. Arguments are variable quantities (“empty positions”), whose number depends on the nature of the predicate; when they are replaced with fixed quantities (constants), the propositional function becomes a proposition and takes the form of a sentence. In linguistics, arguments expressed by lexemes have been called actants since the time of L. Tesnière. 23 Actants can be not only names of concrete or abstract things, but also other propositional functions (so-called embedded predicates). Semantic relations between the verbalized predicate and the actants form the basis of the sentence’s semantic structure. A great deal of attention is devoted to studying these relations in contemporary linguistics. 2.2. Many linguists describe the semantic relations between predicates and actants using the traditional terms for subject, object, and adverbial relations, which are also widespread in Lithuanian linguistics. In recent years, this approach has been strongly represented by the academic grammar of Russian. It distinguishes the following principal semantic categories, termed elementary: predicative property (a category elsewhere referred to simply as the predicate), subject, and object. The subject is defined as the bearer of a predicative property (Русская грамматика 1980, 127). An object is a living or nonliving thing toward which an action or relation is directly directed; this includes objects of physical action, thought, perception, approach and departure, the addressee, and the object of an emotional relation (ib. 129). Qualifiers with attributive and circumstantial meanings, which expand the elementary semantic structure, are considered non-elementary semantic categories. The elementary semantic structure as thus understood roughly corresponds to the syntactic core of the sentence determined by predicate properties, while non-elementary categories correspond to free elements and components dependent on sentence constituents. However, the relation of the object category to the “non-elementary” categories remains unclear. For example, the feature of direct direction of an action does not apply to instruments, while components with adverbial meanings determined by the predicate’s meaning (e.g., We live in a city) do not fit into any category. In many other works, the concepts of subject and object are not defined at all and are not clearly distinguished from the syntactic concepts of subject and object. The predicate’s relations with actants are described more specifically in terms of semantic functions (or semantic roles, deep cases). Ch. Fillmore called the meanings of positions dependent on the predicate—the verb—deep cases, which were intended to designate universal types of semantic relations; more recently, they have most often been called semantic functions or roles. The number and nature of semantic functions vary; different researchers’ works (and sometimes even the same researchers’ works) identify from 5–9 (Fillmore 1968; 1977; Cook 1978; Anderson 1971; Abraham 1978; Starosta 1978; Dik 1980, 1980 a) up to 25 (Апресян 1974, 125–126). This is partly because the very broad range of actant meanings is divided according to the research approach and specific objectives. For example, when describing the valency of verbal lexemes, it is useful to subdivide semantic functions more finely than when investigating the general features of sentence structure. Much more important differences arise from divergent views about the level of analysis represented by semantic functions. Both Ch. Fillmore and many other researchers in this field sought to distinguish and characterize semantic functions on the basis of the situation denoted by the sentence, the “state of affairs” in reality beyond language (cf. Pleines 1978, 367–368). The predicate (first and foremost, the verb) was viewed as the name of an extralinguistic situation, and semantic functions (roles) as relations reflected in the speaker’s consciousness 24 among the participants in that situation. On the basis of the view of a situation as the denotatum or referent of a sentence (Гак 1969, 78–80; 1973, 358; Москальская 1981, 9–10 with references), attempts were also made to distinguish and differentiate semantic functions according to how and in what respect the property expressed by the predicate concerns the participants in that situation. If it turned out that two or more sentences with different predicates could be associated with the same situation, they were considered to have the same semantic but different syntactic structure (Апресян 1974, 258 ff.). This view required modeling in terms of semantic functions not so much the meanings of predicates as the external situations themselves, which in turn are accessible only through the meanings of sentence constituents. Applying it to the analysis of particular sentences (especially so-called conversives) proved difficult. Closer examination of predicate semantics revealed that there is no theoretical basis for regarding an external situation as the direct denotatum or referent of a sentence, since the predicate, or property, is a unit of the signification plane and therefore cannot serve as a unit of nomination—a name for a situation. “Predicates do not name” (Степанов 1980, 321). Unlike actants, which have both denotative and significative meaning, the predicate in a sentence realizes only its significative (abstract, conceptual) content (Арутюнова 1976, 11). The meaning of a sentence is not the situation itself, but what is asserted about the situation, the properties selected within it by means of the resources of one language or another. Therefore, the same situation can be described by sentences with different semantic structures (cf. Starosta 1978, 512; Lyons 1981, 491 ff.; Běličová 1982, 26; Hlavsa 1982, 30 ff.). For example, exactly the same external situation can be described by the sentences: (1) The father sold the horse to the neighbor; (2) The neighbor bought the horse from the father. The actants denoted by the nouns father and neighbor have the same significata and denotata in both cases, but their semantic function (i.e., their relation to the predicate) is by no means the same: in the first sentence, father denotes the performer of the action (the agent), while neighbor denotes the recipient of the action (the beneficiary); in the second, the agent role is borne by neighbor, while the form of the word father denotes the source of the action, which may also be unexpressed in the sentence. Thus, the semantic structures of the two sentences differ; it is also difficult to find a common denominator for them by including a causation operator, since sell ≠ CAUS + buy. The logical and semantic structure of a sentence is determined primarily by the choice of predicate, since both the number and the nature of actants depend on the meanings of words serving as predicates. Every language has its own system of verb semantic fields and specific tendencies to describe certain phenomena as active actions, as states of the things experiencing them, or as ambient processes (cf.: It snowed and Быпал снег alongside The snow caused it). Meanwhile, semantic functions are known to us only through the meaning of the predicate, and “there is no other way to reach them” (Potts 1978, 399). This is presumably why researchers have sought the units of a sentence’s semantic structure at the significative level of language, focusing on the meanings of the forms of words serving as predicates and actants in various languages. It is precisely in this way that attempts are now being made to identify the most general semantic functions (roles), which would reflect the relations between predicates and actants and be relevant to the analysis of sentence structure in many languages (Abraham, 1978; Яхонтов 1978; Храковский 1978; 1981). The following can be regarded as the most general semantic functions, shared by many of the classifications presented: agent – A (initiator and performer of an action, including natural forces), patient – P (recipient of an action or bearer of a state), together with percipient – Pcp (bearer of a mental state, or experiencer), descriptive – Descr (bearer of a property or name), beneficiary – B (recipient of an action or state), result – R, content – Con (content of an action or state), and instrument – I (means). 25 ), patient – P (recipient of an action or bearer of a state), together with percipient – Pcp (bearer of a mental state, or experiencer), descriptive – Descr (bearer of a property or name), beneficiary – B (recipient of an action or state), result – R, content – Con (content of an action or state), and instrument – I (means). The semantic relations of actants more distant from the predicate (but conditioned by it in certain cases) are denoted by the functions of causative (Caus), finitive (F), locative (L, to which references to direction and time may also be assigned), and mode (M), which may also be regarded as separate predicates. Since the semantic functions listed here denote the relations between predicates and actants at the significative level, they cannot be defined by the referential properties of actants: animate/inanimate, whole/part, and so on. Those properties belong not to the sentence’s but to the actants’ referential structure and can only indirectly affect semantic functions. Because of this interaction and other factors, borderline, transitional cases occur between some semantic functions, for example, between the causative and content functions in sentences such as The sky glitters with stars. 2.3. Thus, we have two models of semantic relations: in the first, these relations are described in terms of subject, object, and adverbial constituents; in the second, in terms of semantic functions (or roles, deep cases). What do these models have in common, and can they be reconciled? If we regard the sentence predicate as a semantically indivisible unit, the two classifications will appear incompatible: what is called the subject will in some cases be an agent (The child writes), in others a patient (The child lies), and in yet other cases the patient will correspond to the object (The child carries an apple). Predicates, however, are also divided into classes. In addition to the familiar logical division of predicates into one-place and many-place predicates, methods and criteria for classifying semantic predicates are currently being actively discussed. Authors studying the componential structure of predicates seek to identify more elementary units within it—components of meaning or semantic features (cf. Bendix 1966; Jackendoff 1972; 1975; Апресян 1974; Wierzbicka 1972; 1976)—that could serve as a basis for dividing predicates into separate classes. These classes are distinguished in different ways and have varying degrees of generality. For example, L. Ščerba distinguished predicates of action, state, and quality (Щерба 1974, 90). W. Chafe (1971), J. Pleines (1976), and R. Zimek (1980) distinguish predicates of action, state, and process (similarly Grenda 1982, 33 ff.); J. Stepanov considers the three main types of predicates to correspond to Aristotle’s categories of substance, state, and relation (Степанов 1981, 131–132), while V. Bogdanov presents classes of action, state, property, and relation predicates as generally accepted (Богданов 1977, 51; for an overview of various classifications, see Селиверстова 1982, 88 ff.). A consistent classification of predicates based on several differential features was recently developed by T. Bulygina. In this classification, properties are first distinguished from phenomena according to the feature of temporality (similarly Rosengren 1978, 379–380); the latter are then divided according to the feature of dynamism into actions (dynamic phenomena) and states and situations (static phenomena), while actions are divided according to duration into processes and events (Булыгина 1982). Setting aside the feature of temporality, which may be associated with the verb and noun as grammatical classes of words, di- 26 chotomija (plg. Benveniste 1950, 21 tt.) ir predikatinėje sakinio struktūroje apskritai užima ypatingą vietą, ypatybes ir būsenas bei situacijas būtų galima jungti į vieną sinkretinį tipą (plg. Vendler 1967, 103) ir supriešinti su veiksmais. Jeigu būsenos kaitą reiškiančius predikatus, L. Chafe’o vadinamus procesais, priskirtume irgi prie būsenų, turėtume dvi didžiausias predikatų grupes – veiksmo ir būsenos (plačiausia prasme) predikatus. Net ir tokio bendriausio predikatų suskirstymo pakanka tam, kad subjekto ir objekto sąvokas būtų galima susieti su semantinėmis funkcijomis (aut. 1982). Atskirai paimtų veiksmo ir būsenos predikatų semantinės funkcijos nesiskyrinėdamos įeina į bendresnes subjektinių, objektinių bei adverbialinių santykių kategorijas: Semantic relations and functions (roles) Semantic relations | Semantic functions associated with action predicates | Semantic functions associated with state predicates Subject relations | A (Guests arrived; The rain beat down the rye) | P (The glass broke); Pep (Everyone heard the noise); B (The parents owned land); Descr (The daughter was beautiful; The father was a teacher) Object relations | P (The girl broke the glass); R (The neighbor dug a ditch); B (I sent a letter to my parents); I (Play the violin) | Con (I do not know the way; There is a lack of money); R (The money turned into coal) Adverbial relations | Caus (People died of hunger); F (I came for an axe); M (We will live differently; The cheese weighed a kilogram); L (I lived in the city; We went into the city; It warmed up in the evening) In view of this correlation, subject, object, and adverbial relations can be expressed in terms of semantic functions (roles). The subject can be defined as the agent of action predicates and the patient, percipient, beneficiary, and descriptor of state predicates. This conception of the subject essentially corresponds to the definition already mentioned: “the performer of an action or the holder of a state.” Actants directly determined by the semantics of the predicate and performing other semantic functions are considered objects. The term adverbial relations denotes the functions (roles) of cause, purpose, location, direction, time, and manner, characteristic of actants more remote from the predicate, which in many cases can be regarded as separate embedded predicates (cf. Helbig 1982, 28). The function of instrument is rather problematic, since expressions of the means of an action are determined by predicate semantics in some sentences (e.g., with the verbs to play, to ring, to rattle, to wave), while in others they are optional (e.g., with verbs of motion). Many of them, too, can be understood as embedded predicates. Some researchers attempt to distinguish them from objects and assign them to the same group as expressions of time, place, or cause (cf. Cook 1978, 299; Dik 1978, 39 ff., 1980, 9; in Lithuanian linguistics—Drukteinis 1982, 16–17 with references). Nevertheless, given the semantic relation of the instrument to the agent, its frequently observed dependence on predicate meaning, its nominal expression, and its meaning of concreteness (Арбат 1976, 6), for now it seems more appropriate to retain the instrument function within the group of object relations. 27 They can be understood as embedded predicates. Some researchers attempt to distinguish them from objects and assign them to the same group as expressions of time, place, or cause (cf. Cook 1978, 299; Dik 1978, 39 ff., 1980, 9; in Lithuanian linguistics—Drukteinis 1982, 16–17 with references). Nevertheless, given the semantic relation of the instrument to the agent, its frequently observed dependence on predicate meaning, its nominal expression, and its meaning of concreteness (Арбат 1976, 6), for now it seems more appropriate to retain the instrument function within the group of object relations. The concepts of subject, object, and adverbial constituents, correlated with semantic functions (roles) in the manner described, acquire a more clearly defined content that essentially does not conflict with ordinary usage; they correspond to the groups of semantic functions associated with action and state predicates as units of a higher degree of abstraction. Harmonizing the models of sentence semantic structure currently in widest use is important also because their areas of application are not the same. In cases where syntactic processes need to be described, layers of semantic relations generalized by the concepts of subject or object are more useful; on the other hand, to characterize those relations more specifically (e.g., in valency dictionaries), the model of semantic functions is more convenient. Other researchers, too, have drawn attention to the relevance of grouping semantic functions for describing sentence structure and for typology. A. Kibrik uses the term hyperroles for this purpose, denoting by it layers of semantic functions (roles) characteristic of certain language-structure types and realized differently in external structure: the Factitive in ergative languages and the Actor in accusative languages; he groups the functions of experiencer (resp. percipient), recipient, and goal under the hyperrole of Addressee (Кибрик 1979, 314–315; 1980, 328 ff.). E. Geniušienė applies the terms subject and object hyperroles (resp. hyperfunctions) to the classification of reflexive verbs in the Baltic languages (Генюшене 1983, 15). However, without distinguishing action and state predicates, the content of those hyperfunctions could not be clearly defined from a semantic point of view; they were considered semi-syntactic and characterized with reference also to communicative features (cf. definitions such as “the principal participant in a situation”; “the immediate, closest, most affected participant in a situation” Кибрик 1979, 314). Groups of subject, object, and adverbial relations are needed because in Lithuanian, as in other languages of the accusative structural type, semantic functions are not directly encoded. In constructions with two-place predicates, the patient is not only expressed differently than in constructions with one-place predicates (cf. The girl broke the glass and The glass broke), but also participates differently in syntactic processes that would be difficult to describe semantically in terms of functions (roles). For example, in passive transformation, the most important role is played not by the change in the agent, patient, or other functions, but by the change in the subject’s relation to the verb. For languages whose grammatical structure directly encodes semantic functions (Van Valin, Foley 1980, 348–350); Кибрик 1980, 328 ff. Нунэн 1982, 363), the concepts of subject and object (as well as the corresponding syntactic concepts of subject and complement) may also be unnecessary. Besides the semantic relations mentioned above, sentence analysis also requires attention to the componential structure of the predicate’s meaning and the referential properties of actants; here it is also necessary to take into account— 28 to consider also categories of the communicative and pragmatic levels. Examining these matters is a task for a separate study. Here, the aim was only to discuss in the broadest terms those units of the sentence’s semantic structure that are necessary for modeling predicate–actant relations. 3. RELATIONS BETWEEN UNITS OF SYNTACTIC AND SEMANTIC STRUCTURE 3.0. The syntactic and semantic structures of a sentence are not autonomous. In language, the meaning of a sentence is realized only through the syntactic relations between word forms. Everything that can be said about the semantic structure of a sentence derives from observing the composition of concrete sentences; even logic, in which we seek principles for modeling sentence meaning, ultimately rests on generalized and formalized natural language. Distinguishing the syntactic and semantic levels at the first stage of analysis is necessary not so that we can examine the form and content of a sentence separately, but so that we can define their relations more precisely—that is, the network of mutual correlations among syntactic and semantic units, which differs from one language to another and determines the distinctive structural features of individual languages. Situations perceived by humans are encoded differently not only across languages, but also within the same language. These differences are easiest to observe in the grammatical structure of sentences, but they extend to much deeper layers as well. It has already been mentioned that the same external situation can be described by selecting predicates with differing numbers of arguments that have different semantic functions (see 2.2.). Sentence structure depends on what one wishes to say about the situation, from the perspective of which participant it is evaluated, which information about it is considered given in advance, and which information one wishes to present as new; finally, to some extent, it also depends on the relations between the participants in the speech act that become salient in that act. Differences in the external forms of sentences almost always reflect certain differences in their content. However, in the grammatical structure of natural languages, individual units at the semantic, communicative, and pragmatic levels do not each have separate means of expression: they are encoded physically. This is the main reason why attempts to directly connect syntactic structures with categories of thought and cognition can have only limited success. Following this path, one either describes classes of syntactic constructions in semantic terms, or projects onto the realm of cognition features of sentence semantics and syntax that are characteristic of a particular language. It seems that anything reliable about universal tendencies in natural languages to link categories of thought with certain syntactic structures can be said only after the mutual relations among syntactic, semantic, communicative, and pragmatic units—defined separately at each level of analysis—have been studied across as many languages as possible. Setting aside the communicative and pragmatic levels for now, we will try to illustrate with a few examples how units of the syntactic and semantic structures of a sentence interact in Lithuanian sentences. 3.1. The view, customary in contemporary logic, of the internal structure of a proposition as a propositional function linking a predicate (property) with one or more arguments 29 through certain semantic relations essentially corresponds to the monocentric model of the syntactic structure of a simple sentence discussed in 1.4.2. In many sentences, the predicate of a proposition is expressed by the verbal predicate, and its arguments (actants) by the sentence constituents connected with that predicate. This makes it possible to interpret the syntactic relations between the subject and complements, on the one hand, and the verbal predicate, on the other, in terms of semantic functions and the bundles of semantic relations—subject (Sb), object (Ob), and adverbial (Ad)—distinguished according to the predicate’s most general meanings. For example, the logical and semantic structure of the sentence Mother handed the child a toy can be represented as follows: In this case, the sentence predicate (T) corresponds to the predicate; the subject (V) to the first actant, which has the semantic function of agent; the direct complement (Ppₜ) to the second actant, which has the function of patient; and the indirect complement (Ppₙ) to the third actant, which has the function of beneficiary. The mutual-dependency relation between T and V reflects the subject relation (Sb), while the adjunctive relation between T and Ppₜ and Ppₙ reflects the object relation (Ob). Thus, although the syntactic and semantic units of sentence structure are defined at different levels of analysis and independently of one another, we observe an unmistakable correlation between them. On the basis of basic sentences (setting aside syntactic processes), one could even claim that the subject is typically associated with a subject relation and the complement with an object relation. If sentence constituents were merely labels for formal relations having nothing to do with the content plane of language, it would be difficult to understand their major role in many languages. However, the same syntactic relations and the sentence constituents distinguished on their basis express not only categories at the semantic level, but also categories at other levels. For example, passive constructions have the same structure of semantic functions as the corresponding active constructions, but their meaning is by no means identical: the reorganization of syntactic structure (passivization) is associated with a different perspective on the situation (cf. the elimination of the agentive actant), a different information structure, and is conditioned by factors such as the relation to context. We do not always find a correspondence between the syntactic and semantic structures of a sentence. It is precisely in cases of mismatch that various doubts and difficulties most often arise. For example, in the sentence The child began to play, the predicate began expresses not the main predicate, but only its phase operator (for the term, cf. Черч 1960, 42 ff.), denoting a certain aspect of a property (such operators are also called quasi-predicates; cf. Кацевич, Храковский 1981, 14—16; Бирюлин 1981, 74). The main predicate is expressed here by the infinitive, which occupies the position of a direct complement, as in sentences with modal operators, e.g.: The child can play. It is precisely because of this relation between the predicate and the operator that infinitival constructions with phase and modal 30 Verbal forms of this kind are generally considered compound predicates, although in the syntactic structure of sentences it is possible to distinguish a simple predicate and a complement associated with it (cf. The child started a game; The child can do everything, see 1.4.4); in impersonal sentences with phase operators, the infinitive denoting the predicate can also function as the subject, e.g. I am allowed to go (cf. I am allowed everything). We see another kind of relationship between semantic and syntactic structure in sentences with nominalized word forms. For example, in the sentence The basketball players’ competition ended, the predicate corresponds to the predicate’s phasal operator (Op), the subject to the predicate, and the subject attribute (Pž) to an actant that has the semantic function of agent and is related to the predicate by a subject relation: Thus, predication is a characteristic, but not a distinguishing, feature of the sentence; it is also expressed by sentence constituents. Not every predicate corresponds to a grammatical predicate, and not every grammatical predicate to a predicate. 3.2. The relationships between syntactic and semantic units defined at separate levels of analysis also make it easy to explain some Lithuanian constructions whose nature remains a matter of debate. For example, in sentences such as Students are already working; A person went and stopped, some authors discern an active meaning and regard the genitive form as the subject (Balkevičius 1963, 78; Paulauskienė 1979, 111), while others deny the active meaning and regard the genitive as a complement (Sirtautas 1970; 1978, 35; Geniušienė 1972). The arguments advanced in support of the first view (the genitive denotes an agent, precedes the predicate, and cannot be omitted) reflect an undifferentiated approach to syntactic and semantic categories. Here the genitive case occupies a syntactic position in which the word form cannot be related to the predicate through a relation of mutual dependence and therefore cannot be regarded as the subject. It depends on the predicate in the same way as genitives functioning as complements in other passive constructions (cf. The work is already being done by the students). The obligatory nature of the genitive marks a strong syntactic relation, not a function; nor is its usual position relative to the predicate a distinguishing feature of the subject or complement in Lithuanian. It is equally clear that the genitive form in the sentences cited has the semantic function of agent and denotes the participant in the action (or, in other cases, the state), while the sentences themselves are semantically close (though not identical) to corresponding sentences with active-form predicates (The students are already working; The person went and stopped). Apparently, this subject relation and the meaning close to that of active constructions prompt researchers to transfer semantics to the syntactic level and to seek, in the cases mentioned, a subject position that is altogether unavailable to the genitive form. Only the nominative form of nouns or pronouns can function here as a subject denoting the object of an action, e.g.: The old people loved forests. 31 Differences between syntactic and semantic structure are prominent in many impersonal sentences whose core consists of subordinate word combinations. In them, the subject of a state is often expressed by oblique cases (especially the dative), e.g.: The family ran out of bread; The child feels sleepy. In meaning, these sentences are close to personal ones, cf.: The family ran out of bread; The child wants to sleep. Their syntactic composition, however, is quite different: in the first case, subject relations are expressed by subordination, and the subject corresponds to a complement; in the second, they are expressed by mutual-dependence relations, and the subject corresponds to the grammatical subject. Yet in personal sentences too, the semantic subject may correspond not to the grammatical subject but to a complement (cf. Katkuvienė 1980). For example, in the personal sentence The girl broke out in pimples, as in the impersonal sentence with the same semantic structure The girl broke out in pimples, the subject of the state is expressed by a direct object, but the syntactic expression of the content of the state differs; cf.: Impersonal sentences and personal sentences with corresponding lexical composition may differ not only in how semantic relations are expressed at the syntactic level, but also in their semantic structure. Differences in semantic structure can be seen by comparing even the following sentences: (1) Water ran into the room and (2) The room filled with water. The external situation referred to in both cases is apparently the same, but it is interpreted differently. In the first sentence, the predicate describes a process occurring in a particular place (the room), and the genitive complement denotes its patient. The verb pribėgo is synonymous here with prisirinko; it may also be replaced by the near-synonymous verb atsirado. Cf.: 6 To make the relationship between semantic and syntactic structure clearer, the subject position is not shifted to the right in the diagram. 7 The actant with a locative function can also be considered an incorporated predicate here: There was water running in (the room). 32 The second sentence describes the state of the room itself (the patient) and its content. The verb pribėgo here is synonymous with prisipildė (cf. Vanduo pripildė kambarį ‘Water filled the room’); using the verb atsirado ‘appeared’ or the like is impossible here. Cf.: It is difficult to explain the difference in meaning between these sentences by means of communicative sentence articulation, since in both cases, depending on word order and intonation, the components that function as theme and rheme may differ; cf.: Vandens pribėgo į kambarį; Vandens pribėgo (pilnas) kambarys. Differences in meaning between personal and corresponding impersonal sentences may be somewhat more closely related to the tendency to express certain meanings through the syntactic structures predominant in a language. Lithuanian, like other languages of the accusative structural type, has a tendency to express the subject’s “hyperfunction” through relations of mutual dependence and, at the same time, to associate it with the sentence topic; this distinguishes the subject position from the others and makes it “privileged” in this sense. The subject’s predominant position before the predicate also reflects its connection with the topic. Therefore, a sentence constituent expressed in the nominative, preceding the predicate and connected to it by a relation of mutual dependence, is naturally understood as the topic and often also as the subject of an action or state. Because of the aforementioned physical coding of communicative, pragmatic, and semantic functions, this also affects the meaning of the sentence as a whole, lending the things described one perspective or another. 3.3. The same semantic relations that we observe between the predicate and the sentence constituents linked to it also become apparent at lower levels of sentence articulation—between attributes and the words they modify. The semantic structure of modified (attributive) phrases is highly diverse. Some attributes combine with the words they modify to form lexicalized phrases functioning as compound names (Didžioji Lietuvos Kunigaikštystė, Grįžulo Ratai). Many pronouns functioning as attributes denote quantifiers associated with predicate arguments (visi draugai, kiekvienas keleivis, kai kurie mokiniai) or serve a deictic and emphatic function (tas vyras, toks medis). In phrases containing word forms derived from verbs, attributes may denote the arguments of embedded predicates (laiško rašymas, žemės trauka). Most attributes, however, are themselves embedded, “degraded” predicates (graži diena, aukšta pušis). Modified (attributive) relations are relations between a property and its bearer that are not associated with the modal categories of the speech act and do not form the basis of a sentence. According to their semantic functions and bundles of functions—semantic relations and the meanings of individual components—attributive phrases (like the components of the sentence nucleus) are discussed in greater detail in the new “Grammar of the Lithuanian Language” (Gramatika 1985). Here we would only like to note that a differentiated approach to the criteria for analy- 2. 120 33 sis helps explain some word combinations that are variously interpreted and give rise to doubt. For example, some treat noun phrases with a genitive such as aukso žiedas, tėvo gyvenimas, paukščio giesmė, brolio kepurė, drobių audėja as formed by government (Balkevičius 1963, 33–34; Lietuvių kalbos gramatika 1976, 16; Белошапкова (ред.) 1981, 393), others as formed by juxtaposition (Русская грамматика 1980, 42), while still others regard them as problematic (Labutis 1976, 36). Their dependent constituents are also classified differently in terms of sentence constituents: some classify those with an object meaning (drobių audėja, eilėraščių rašymas) as complements and those without one as attributes (Balkevičius 1963, 186, 170–171); others classify them all as attributes (Lietuvių kalbos gramatika 1976, 417–419; Sirtautas 1981, 58); still others identify here a certain syncretic group—the attributive complement (Адмони 1955, 75–76). The reason for these variations is not so much syncretism in the linguistic phenomenon as syncretism in the criteria used to analyze it. Applying the concepts of units at the semantic and syntactic levels consistently makes it clear that all the phrases mentioned above are formed by government, since the dependent word has a form that differs from the head and is inflected (i.e. an oblique case form), a form determined by the head noun’s capacity to combine with the genitive (cf. 1.1.1). Since the dependent word form is not directly linked to the predicate, it functions as an attribute. Only at the level of transformational (or derivational) sentence analysis can genitives dependent on deverbal nouns (saulės laida, rato sukimas) be regarded as subjects or complements of secondary predicates embedded in the sentence. According to their semantic relations, attributive phrases with the genitive can be subjective (rugių žydėjimas), objective (tėvynės ilgesys), or adverbial (pavasario šventė); closer analysis also reveals purpose and intended-use relations (darbo laikas, batų tepalas), temporal relations (ryto saulė), locative relations (miško garsai), and other semantic functions. Objective and adverbial relations may also be expressed by phrases with other cases, e.g.: dovana svečiui, kelionė dviračiu (objective relations); namai kaime, kelionė jūra (adverbial relations). Of course, phrases with different cases and different prepositional constructions each have their own scale of semantic functions. 3.4. The approach to relating syntactic sentence structure to semantic structure discussed here is not the only one. For a long time, generative grammar used the concept of deep syntactic structure for this purpose; it was assumed that semantic interpretation could be applied only to deep syntactic structure, which was considered universal in nature. Research over the past decade has shown that in some respects this deep sentence structure corresponds to its logical and semantic structure (and is therefore unnecessary), while in other respects (binarity, linearity) it is insufficiently motivated typologically (and thus questionable). Consequently, more recent sentence models seek to dispense with it (cf. Stockwell 1980, 380), and transformations are used mainly as a means of syntactic analysis for determining how more complex sentences relate to basic sentences of simpler structure and for linking them to semantic structure (cf. McCawley 1980, 177 ff.). Relations between syntactic and semantic units distinguished at separate levels have recently been used as a basis for many grammatical studies (Abraham 1978, 709 ff. with references; Běličová 1982, 45 ff.; Lamprecht 1983). Researchers are seeking correlations between these units. 34 regularities—a certain hierarchy according to which the syntactic functions of subject and object can be assigned to actants having one or another semantic function (Dik 1980, 14—15; 1980 a, 63—64). Consistently guided by the relations between syntactic and semantic units, A. Cholodovičius, V. Chrakovskis, S. Jachontovas, G. Silnickis, and other authors developed a diathesis model used in the typological study of many languages. One obstacle in research in this area is the uncertainty and undifferentiated use of the terms subject and object, especially characteristic of linguistic works on languages that lack commonly used equivalents for the syntactic concepts of subject and object. The existence of two sets of terms in Lithuanian (as in many Slavic languages) creates particularly favorable conditions for describing units at different levels of analysis and the relations between them. Conclusions 1. Units of the syntactic and semantic structure of sentences are defined separately at each level of analysis: syntactic units according to the syntactic relations between word forms in a sentence, and semantic units according to semantic functions and more general semantic relations. 2. Three main types of syntactic relations are distinguished: dependency, mutual dependency (interdependence), and coordination. Through a dependency relation, a word form is incorporated into a sentence by means of another word form that is independent of it; through a mutual dependency relation, it is incorporated by means of a word form mutually connected with it. Through coordination, word forms are incorporated into a sentence by means of an identical relation to another word form. Syntactic relations are formally realized through agreement and concord, government, and adjunction. 3. Word forms connected by direct relations constitute phrases. Phrases (except lexicalized ones) are not nominative units but components of the syntactic structure of a sentence, identified at its various levels of analysis. According to the nature of the syntactic relations, dependency, mutual-dependency, and coordinate phrases are distinguished. 4. The terms for sentence constituents may be used to describe units of a sentence’s syntactic structure. In that case, it is also proposed that a sentence constituent be defined according to the syntactic relations of word forms with the structural center of the sentence—the predicate. A sentence constituent typically connected to the predicate by a mutual-dependency relation is considered the subject; one connected by a dependency relation is considered an object. Objects, in turn, can be divided into direct objects, indirect objects, and adverbials. Sentence constituents connected to one another by coordination are coordinate constituents. A sentence constituent connected both to the predicate and to the subject or object by two-way dependency relations is considered a predicative attribute. Other attributes, which have no direct relation to the predicate, are not sentence constituents but merely their modifiers. Following syntactic criteria, the so-called compound predicate can in many cases be divided into a predicate and a subject or object dependent on it; however, copular constructions with nouns should be regarded as compound predicates. 2* 35 5. The semantic units of a sentence’s semantic structure are defined according to the predicate’s meaning relations with arguments (actants) expressed by lexemes. The semantic functions of actants (agent, patient, beneficiary, etc.) are determined at the significative level, taking into account the predicate’s meaning, which reflects the speaker’s attitude toward the situation. 6. Two most general semantic groups of predicates are distinguished: action and state predicates (state predicates also include predicates denoting classes of things, properties, and processes). On the basis of this subcategorization of predicates, semantic functions (roles) can be grouped into more general bundles (hyperfunctions) corresponding to subject, object, and adverbial relations. Subject relations correspond to the semantic functions of agent for action predicates and patient, percipient, beneficiary, and descriptive for state predicates; object relations correspond to the functions of patient, result, beneficiary, and instrument for action predicates, and content and result for state predicates. Adverbial relations correspond to the functions of cause, purpose, place, direction, time, and manner of action, characteristic of actants more remote from the predicate, which may themselves be considered embedded predicates. Thus, subject, object, and adverbial relations are defined in terms of the most general groups of predicates and semantic functions (roles). 7. The monocentric model of syntactic relations is correlative to the logical schema of the sentence core. Although there is no direct correspondence between the syntactic and semantic structures of many sentences, each sentence constituent or phrase component can be described in terms of the relations between differentiated syntactic and semantic units. LITERATŪRA Abraham W. (Ed.) Valence, Semantic Case and Grammatical Relations. – Amsterdam, 1978. Abraham W. Valence and Case: Remarks on their Contribution to the Identification of Grammatical Relations. – In: Abraham (ed.) 1978, p. 695–723. Ambrazas V. Lietuvių kalbos vardininko su dalyviu (nominativus cum participio) struktūra ir raida. – Kn.: Lietuvių kalbotyros klausimai. V., 1978, t. 18, p. 7–68. Ambrazas V. Lietuvių kalbos dalyvių istorinė sintaksė. – V., 1979. Ambrazas V. Dėl sintaksinių ir semantinių terminų santykio. – Kn.: Sintaksės ir semantikos klausimai. Šiauliai, 1982, p. 6–7. Anderson J. M. The Grammar of Case: Towards a Localistic Theory. – Cambridge, 1971. Babby L. H. A Transformational Grammar of Russian Adjectives. – The Hague, Paris, 1975. Balkevičius J. Dabartinės lietuvių kalbos sintaksė. – V., 1963. Balkevičius J. Sakinio pamatas veiksmažodis. – Mūsų kalba, 1983, Nr. 3, p. 3–6. Bechert J. Ergativity and the Constitution of Grammatical Relations. – In: Plank (ed.) 1980. Becker K. F. Organism der Sprache als Einleitung zur deutschen Grammatik. – Frankfurt am Main, 1827. Beličová H. Sémantická struktura věty a kategorie pádu. – Praha, 1982. Bendix E. H. Componential Analysis of General Vocabulary. – Bloomington, 1966. Benveniste E. La phrase nominale. – Bulletin de la Société de linguistique, 1950, t. 46, p. 19–36. 36 Benveniste E. „Être“ et „avoir“ dans leurs fonctions linguistiques. — Bulletin de la Société de linguistique, 1960, vol. 55. Bloomfield L. Language. — New York, 1933. Chafe W. L. Meaning and the Structure of Language. — Chicago, London, 1971. Chafe W. L. Giveness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View. — In: Li (ed.) 1976, p. 27—55. Chomsky N. Aspects of the Theory of Syntax. — Cambridge (Mass.), 1965. Chomsky N. Topics in the Theory of Generative Grammar. — In: Sebeok (ed.) 1966. Cole P., Sadock J. M. (Ed.) Syntax and Semantics. Vol. 8. Grammatical Relations. — New York, San Francisco, London, 1977. Comrie B. Causatives and Universal Grammar. — In: Transactions of the Philological Society, 1974, p. 1—32. Comrie B. Language Universals and Linguistic Typology. — Oxford, 1981. Cook W. A. A Case Grammar Matrix Model. — In: Abraham (ed.) 1978, p. 295—309. Daneš F. Satzglieder und Satzmuster. — In: Beiträge zu Problemen der Satzglieder. Leipzig, 1978. Dik S. C. Functional Grammar. — Amsterdam, 1978. Dik S. C. Studies in Functional Grammar. — London, 1980. Dik S. C. Seventeen Sentences: Basic Principles and Application of Functional Grammar. — In: Moravcsik, Wirth (ed.) 1980 (a), p. 45—75. Drukteinis A. Objekto sakovo ir kai kurie požymiai. — Kalbotyra, 1982, t. 23(1), p. 15—20. Erben J. Abrib der deutschen Grammatik. — Berlin, 1964. Fillmore Ch. J. The Case for Case. — In: Universals in Linguistic Theory. New York, 1968, p. 1—88. Fillmore Ch. J. The Case for Case Reopened. — In: Cole, Sadock (ed.) 1977, p. 59—82. Flämig W. Der Satzbau (die Syntax). — In: Die deutsche Sprache (Kleine Enzyklopädie), Leipzig, 1970, Bd. 2, S. 908—978. Fries C. The Structure of English. — New York, 1952. Gary J. O., Keenan E. L. On Collapsing Grammatical Relations in Universal Grammar. — In: Cole, Sadock (ed.) 1977, p. 83—120. Geniušienė E. Dėl pasyvinės intranzityvinių veiksmažodžių formos. — Kalbotyra, 1972, t. 24(1), p. 27—31. Glinz H. Grammatik und Sprache. — In: Moser (Hrsg.) 1962, p. 42—60. Glinz H. Der deutsche Satz. — Düsseldorf, 1965. Gołąb Z. Próba klasyfikacji syntaktycznej czasowników polskich (na zasadzie konotacji). — Biuletyn Polskiego Towarzystwa Językoznawczego, 1967, t. 25, p. 3—43. Grebe P. Der Große Duden. Grammatik der deutschen Gegenwartssprache. — Mannheim, 1959. Greenberg J. H. (Ed.) Universals of Human Language. Vol. 4. Syntax. — Stanford, 1978. Grenda Č. Veiksmažodžio semantika ir žodžių junginio struktūra. — Kn.: Vientisinio sakinio semantika. V., 1982, p. 31—56. Hays D. G. Dependency Theory. A Formalism and some Observations. — Language, 1964, N 4, p. 511—525. Helbig G. Valenz—Satzglieder—Semantische Kasus—Satzmodelle. — Leipzig, 1982. Helbig G., Schenkel W. Wörterbuch zur Valenz und Distribution deutscher Verben. — Leipzig, 1973. Heringer H. J. Theorie der deutschen Syntax. — München, 1970. Hjelmslev L. Prolegomena to a Theory of Language. — Indiana, 1953. Hockett C. F. A Course in Modern Linguistics. — New York, 1958. Hudson R. A. Constituency and Dependency. — Linguistics, 1980, vol. 18, p. 179—198. Jablonskis J. Rinktiniai raštai. T. 1 / red. Palionis J. / — V., 1957. Jackendoff R. S. Semantic Interpretation in Generative Grammar. — Cambridge (Mass.), 1972. 37 Jackendoff R. S. Morphological and Semantic Regularities in the Lexicon. — Language, 1975, vol. 51, N 3, p. 639—671. Johnson D. E. On Relational Constraints on Grammars. — In: Cole, Sadock. (ed.) 1977, p. 151—178. Johnson D. E. On Keenan’s Definition of „Subject of“. — Linguistic Inquiry, 1977 (a), N 8, p. 673—692. Johnson D. E. Toward a Theory of Relationally Based Grammar. — New York, London, 1979. Johnson D. E., Postal P. M. Arc Pair Grammar. — Princeton, 1980. Kahn Ch. H. The Greek Verb „to be“ and the Concept of Being. — Foundations of Language, 1966, vol. 2, N 3. Karcevskij S. Sur la parataxe et la syntaxe en russe. — Cahiers F. de Saussure, 1948, N 7. Katkuvienė L. Dabartinės lietuvių kalbos beasmeniai veiksmažodiniai sakiniai su įnagininku. — LTSR MA darbai. Serija A, 1980, t. 1 (70), p. 107—116. Keenan E. L. Towards a Universal Definition of „Subject“. — In: Li (ed.) 1976, p. 303—333. Kurylowicz J. Esquisses linguistiques. — Wrocław-Kraków, 1960. Kurylowicz J. The Inflectional Categories of Indo-European. — Heidelberg, 1964. Labutis V. Žodžių junginių problemos. — V., 1976. Labutis V. Sintaksės aprašymo galimybės ir uždaviniai. — Kalbotyra, 1981, t. 32 (1), p. 54—59. Lamprecht A. Relationale Satzanalyse. — München, 1983. Lehmann W. P. A Structural Principle of Language and Its Implications. — Language, 1973, vol. 49, p. 47—66. Lehmann W. P. Proto-Indo-European Syntax. — Austin, London, 1974. Li Ch. (Ed.) Subject and Topic. — New York, 1976. Lietuvių kalbos gramatika. — V., 1965, t. 1; 1971, t. 2; 1976, t. 3. Lyons J. Introduction to Theoretical Linguistics. — Cambridge, 1968. Lyons J. Semantics. — Cambridge, 1977, vol. 1—2. Martinet A. L’Autonomie syntaxique. — In: Méthodes de la grammaire. Paris, 1966. Martinet A. Conventions pour une visualisation des rapports syntaxiques. — Linguistique, 1973, vol. 9, f. 1, p. 5—16. McCawley J. D. An Un-Syntax. — In: Moravcsik, Wirth (ed.) 1980, p. 167—194. McCawley J. D. Everything that Linguists Have Always Wanted to Know about Logic (but were ashamed to ask). — Chicago, 1981. Meillet A. Introduction à l’étude comparative des langues indo-européennes. — Paris, 1934. Moravcsik E. A. Agreement. — In: Greenberg (ed.), 1978, p. 331—374. Moravcsik E. A., Wirth J. R. (Ed.) Syntax and Semantics. Vol. 13. Current Approaches to Syntax. — New York, London, etc., 1980. Moser H. (Hrsg.) Das Ringen um eine neue deutsche Grammatik. — Darmstadt, 1962. Paulauskienė A. Gramatinės lietuvių kalbos veiksmažodžio kategorijos. — V., 1979. Pauliny E. Slovenská gramatika (Opis jazykového systému). — Bratislava, 1981. Perlmutter D. M. Relational Grammar. — In: Moravcsik, Wirth (ed.) 1980, p. 195—229. Perlmutter D. M., Postal P. M. Toward a Universal Characterization of Passivization. — In: Proceedings of the Third Annual Meeting of the Berkeley Linguistic Society. Berkeley, 1977. Plank F. (Ed.) Ergativity. Towards a Theory of Grammatical Relations. — New York, 1980. Pleines J. Handlung, Kausalität, Intention: Probleme der Beschreibung semantischer Relationen. — Tübingen, 1976. Potts T. C. Case-Grammar as Componential Analysis. — In: Abraham (ed.) 1978, p. 399—458. Robinson J. J. Dependency Structures and Transformational Rules. — Language, 1970, vol. 46, N 2, p. 259—285. Rosengren I. Die Beziehung zwischen semantischen Kasusrelationen und syntaktischen Satzgliedfunktionen: der freie Dativ. — In: Abraham (Hrsg.). 1978, p. 377—398. 38 Schachter P. Reference-Related and Role-Related Properties of Subjects. – In: Cole, Sadock (ed.) 1977, p. 279–306. Schmalstieg W. R. Indo-European Linguistics. A New Synthesis. – University Park and London, 1980. Schmitt Jensen J. Les syntagmemes. – In: Méthodes de la grammaire. Paris, 1966, p. 161–164. Sebeok T. A. (Ed.) Current Trends in Linguistics. Vol. 3. Linguistic Theory. – The Hague, 1966. Sirtautas V. Veiksmažodinis tarinys dabartinėje lietuvių literatūrinėje kalboje. Filol. kand. dis. (rankr.) – V., 1968. Sirtautas V. Dėl pasyvinių formų aktyvinės reikšmės. – Kalbotyra, 1970, t. 21(1), p. 71–74. Sirtautas V. Pagrindinių sakinio dalių derinimas. – V., 1978. Sirtautas V. Sakinio dalys. – K., 1981. Sližienė N. Veiksmažodžių valentingumo klausimu. – Lietuvių kalbotyros klausimai, 1978, t. 18, p. 107–124. Sližienė N. Subjektinimo ir objektinimo vaidmenų nustatant lietuvių kalbos veiksmažodžių sintaksinį valentingumą. – LTSR MA darbai. Serija A, 1980, t. 1 (70), p. 93–105. Starosta S. The One Per Cent Solution. – In: Abraham (ed.) 1978, p. 459–575. Steele S. Word Order Variation. A Typological Study. – In: Greenberg (ed.) 1978, p. 585–624. Steinthal H. Grammatik, Logik und Psychologie. – Berlin, 1855. Stockwell R. P. Summation and Assessment of Theories. – In: Moravcsik, Wirth (ed.) 1980, p. 353–381. Tchekhoff C. Aux fondaments de la syntaxe: l’ergatif. – Paris, 1978. Tesnière L. Éléments de syntaxe structurale. – Paris, 1959. Trubetzkoy N. Le rapport entre le déterminé, le déterminant et le défini. – In: Mélanges de linguistique offerts à Ch. Bally. Genève, 1939, p. 75–82. Valeckienė A. Predikatyvinis pažyminys kaip atskira sakinio dalis. – Kn.: Lietuvių kalbotyros klausimai, 1967, t. 9, p. 97–114. Van Valin R. D., Foley W. A. Role and Reference Grammar. – In: Moravcsik, Wirth (ed.) 1980, p. 329–352. Van Valin R. Grammatical Relations in Ergative Languages. – Studies in Language. Amsterdam, 1981, vol. 5, p. 361–394. Vendler Z. Verbs and Times. – In: Z. Vendler, Linguistics in Philosophy. Ithaca, 1967. Wierzbicka A. Semantic Primitives. – Frankfurt am Main, 1972. Wierzbicka A. Dociekania semantyczne. – Wrocław-Warszawa-Kraków, 1976. Zimek R. Sémantická výstavba věty: Obecnělingvistické problémy metodologie syntaktické sémantiky ruštinu. – Praha, 1980. Адмони В. Г. Введение в синтаксис современного немецкого языка. – М., 1955. Адмони В. Г. Основы теории грамматики. – М. – Л., 1964. Александров Н. М. Проблема второстепенных членов предложения в русском языке. – Уч. зап. ЛГПИ им. А. И. Герцена. Л., 1963, т. 236. Апресян Ю. Д. Лексическая семантика. – М., 1974. Арват Н. Н. Компонентный анализ семантической структуры простого предложения. – Черновцы, 1976. Арутюнова Н. Д. Предложение и его смысл. – М., 1976. Бархударов Л. С. Структура простого предложения в английском языке. – М., 1966. Белошапкова В. А. Современный русский язык. Синтаксис. – М., 1977. Белошапкова В. А. Синтаксис. – В кн.: Современный русский язык / Под ред. Белошапковой В. А. М., 1981, с. 363–552. Бирюлин Л. А. Конструкции с фазовыми, модальными и каузативными глаголами. – В кн.: Семантика и синтаксис конструкций с предикативными актантами. Л., 1981, с. 71–88. Булыгина Т. В. К построению типологии предикатов в русском языке. – В кн.: Семантические типы предикатов. М., 1982, с. 7–85. 39 Богданов В. В. Семантико-синтаксическая организация предложения. — Л., 1977. Валейка Л. Транспозиция в литовском и английском языках. — Вильнюс, 1980. Вардуль И. Ф. Основы описательной лингвистики. — М., 1977. Виноградов В. В. Русский язык. — М.—Л., 1947. Виноградов В. В. Введение. — В кн.: Грамматика русского языка. М., 1954, т. 2, ч. 1, с. 5—111. Виноградов В. В. Избр. тр. Исследования по русской грамматике. — М., 1975. Гак В. Г. К проблеме синтаксической семантики (семантическая интерпретация „глубинных“ и „поверхностных“ структур). — В кн.: Инвариантные синтаксические значения и структура предложения. М., 1969, с. 77—85. Гак В. Г. Высказывание и ситуация. — В кн.: Проблемы структурной лингвистики — 1972. М., 1973, с. 349—372. Генадзе И. О., Гайдарова Ф. А. О выражении субъектно-объектных отношений в иберийско-кавказских языках. — В кн.: Категория субъекта и объекта в языках различных типов. Л., 1982, с. 154—188. Генюшене Э. Рефлексивные глаголы в балтийских языках и типология рефлексивов. — Вильнюс, 1983. Гладкий А. В. Формальные грамматики и языки. — М., 1973. Грамматика литовского языка. — Вильнюс, 1985. Грамматика русского языка. — М., 1952, т. 1; 1954, т. 2. Гухман М. И. Позиция подлежащего в языках различных типов. — В кн.: Члены предложения в языках различных типов. Л., 1972, с. 5—27. Елизаренкова Т. Я., Топоров В. Н. Язык пали. — М., 1965. Золотова Г. А. К вопросу об объекте синтаксических исследований. — Изв. АН СССР. Сер. лит-ры и яз., 1979, № 1. Золотова Г. А. Коммуникативные аспекты русского синтаксиса. — М., 1982. Карри X. Основания математической логики (Пер. с англ.). — М., 1969. Касевич В. Б., Храковский В. С. Общие вопросы семантики конструкций с предикатными актантами. — В кн.: Семантика и синтаксис конструкций с предикатными актантами. Л., 1981, с. 7—23. Кибрик А. Е. Подлежащее и проблема универсальной модели языка. — Изв. АН СССР. Сер. лит-ры и яз., 1979, т. 38, № 4, с. 309—317. Кибрик А. Е. Предикатно-аргументные отношения в семантически эргативных языках. — Изв. АН СССР. Сер. лит-ры и яз., 1980, т. 39, № 4, с. 324—335. Кибрик А. Е. Проблема синтаксических отношений в универсальной грамматике. — В кн.: Новое в зарубежной лингвистике. М., 1982, вып. 11, с. 5—36. Кржижкова Е. Проблема простого предложения. — Československá rusistika, 1967, N 2. Ломтев Т. П. Структура предложения в современном русском языке. — М., 1979. Лопашов Ю. А. О грамматических средствах выражения субъекта и объектов в албанском языке. — В кн.: Категория субъекта и объекта в языках различных типов. Л., 1982, с. 101—124. Москальская О. И. Проблемы системного описания синтаксиса. — М., 1974. Мразек Р. Синтаксическая дистрибуция глаголов и их классы. — ВЯ, 1964, № 3. Мухин А. М. Функциональный анализ синтаксических элементов. — М.—Л., 1964. Мухин А. М. Структура предложений и их модели. — Л., 1968. Нунь М. О подлежащих и топиках. — В кн.: Новое в зарубежной лингвистике. М., 1982, вып. 11, с. 356—375. Падучева Е. В. О способах представления синтаксической структуры предложения. — ВЯ, 1964, № 2. Пешковский А. М. Русский синтаксис в научном освещении. — М., 1956. Рожновская М. Г. Очерки по синтаксису болгарского литературного языка. Предложения с предикативным определением. — М., 1979. 40 Русская грамматика. Т. 2: Синтаксис. – М., 1980. Попов А. С. Подлежащее и сказуемое в структуре простого предложения современного русского литературного языка. – Пермь, 1974. Распопов И. П. Что такое структурная схема предложения. – ВЯ, 1976, № 2. Распопов И. П. Сказуемое как конструктивный центр предложения. – В кн.: Теоретические проблемы синтаксиса современных индоевропейских языков. Л., 1975, с. 162–168. Ревзин И. И. Современная структурная лингвистика. – М., 1977. Селиверстова О. Н. Семантический анализ экзистенциальных и посессивных конструкций в английском языке. – В кн.: Категории бытия и обладания в языке. М., 1977, с. 5–67. Селиверстова О. Н. Второй вариант классификационной сетки и описание некоторых предикатных типов русского языка. – В кн.: Семантические типы предикатов. М., 1982, с. 86–157. Скобликова Е. С. Согласование и управление в русском языке. – М., 1971. Скобликова Е. С. Современный русский язык. Синтаксис простого предложения. – М., 1973. Смирницкий А. И. Синтаксис английского языка. – М., 1957. Стеблин-Каменский М. М. Спорное в языкознании. – Л., 1974. Степанов Ю. С. К универсальной классификации предикатов. – Изв. АН СССР. Сер. лит-ры и яз., 1980, т. 39, № 4, с. 311–323. Степанов Ю. С. Имена. Предикаты. Предложения. – М., 1981. Сусов И. П. Семантическая структура предложения. – Тула, 1973. Фитялов С. Я. О моделировании синтаксиса в структурной лингвистике. – В кн.: Проблемы структурной лингвистики. М., 1962, с. 100–113. Холодович А. А. Проблемы грамматической теории. – Л., 1979. Храковский В. С. Пассивные конструкции. – В кн.: Типология пассивных конструкций. Диатемы и залоги. Л., 1974, с. 5–45. Храковский В. С. Проблемы классификации синтаксических актантов. – В кн.: Диате-зы и залоги. Л., 1975, с. 274–280. Храковский В. С. Залог и рефлексив. – В кн.: Проблемы теории грамматического залога. Л., 1978. Храковский В. С. Диатеза и референтность. – В кн.: Залоговые конструкции в разноструктурных языках. Л., 1981, с. 5–38. Черч А. Введение в математическую логику. – М., 1960, т. 1. Чикобава А. С. Проблема эргативной конструкции в иберийско-кавказских языках. – В кн.: Эргативная конструкция предложения в языках различных типов. Л., 1967, с. 10–32. Шахматов А. А. Синтаксис русского языка. – М.—Л., 1941. Шведова Н. Ю. Построение раздела „Синтаксис словосочетания и простого предложения“. – В кн.: Основы построения описательной грамматики современного русского литературного языка. Л., 1966, с. 167–205. Шмелев Д. Н. Синтаксическая членимость высказывания в современном русском языке. – М., 1976. Шубик С. А. Иерархия членов предложения в индоевропейских языках. – В кн.: Категория субъекта и объекта в языках различных типов. Л., 1982, с. 125–134. Щерба Л. В. Языковая система и речевая деятельность. – М.—Л., 1974. Яхонтов С. Е. Классы глаголов и падежное оформление актантов. – В кн.: Проблемы теории грамматического залога. Л., 1978. В. АМБРАЗАС ЕДИНИЦЫ СИНТАКСИЧЕСКОЙ И СЕМАНТИЧЕСКОЙ СТРУКТУРЫ ПРЕДЛОЖЕНИЯ В ЛИТОВСКОМ ЯЗЫКЕ Резюме Единицы синтаксической и семантической структуры предложения определяются отдельно на каждом уровне анализа: единицы синтаксической структуры — на основании синтаксических связей словоформ в предложении, единицы семантической структуры — на основании семантических функций (ролей) и их более общих групп — семантических отношений. В синтаксической структуре предложения различаются три основных типа синтаксических связей: подчинение, взаимозависимость и сочинение. Словоформа с подчинительной связью включается в предложение при посредстве независимой от нее другой словоформы, словоформа со связью взаимозависимости — при посредстве взаимосвязанной с ней словоформы. Словоформы с сочинительной связью включаются в предложение при посредстве одинаковой связи с другой словоформой. Согласование (включающее координацию), управление и примыкание представляют собой формальные способы выражения синтаксических связей. Словоформы, связанные в предложении непосредственной синтаксической связью, образуют словосочетания. Словосочетания (за исключением лексикализированных оборотов) являются не номинативными единицами, а компонентами синтаксической структуры предложения, выделяемыми на разных ступенях его членения. По характеру синтаксической связи различаются подчинительные, взаимозависимые и сочинительные словосочетания. Структура подчинительных и взаимозависимых словосочетаний обусловлена грамматическими и семантическими особенностями их компонентов. Подчинительные словосочетания в литовском языке играют первостепенную роль при построении многих типов предложений, входят в их структурное ядро и чередуются с взаимозависимыми словосочетаниями в синтаксических процессах. Это не оправдывает деления синтаксических связей на „присловные“ и „предложенческие“. Предложения отличаются от словосочетаний не характером синтаксической связи, а оформлением по определенным моделям, соотнесением с речевой ситуацией, модальностью и интонационным контуром. Для характеристики синтаксических единиц при описании структуры предложения могут быть использованы термины членов предложения, определенные на основании синтаксических связей словоформ с синтаксическим центром предложения — сказуемым. Член предложения, в типовых случаях связанный со сказуемым связью взаимозависимости, определяется как подлежащее, связанный подчинительной связью — как дополнение. (Типовыми считаются случаи, в которых словоформа выступает в характерной для данной позиции грамматической форме.) По разному участию в синтаксических процессах дополнения в свою очередь подразделяются на прямые (преобразующие синтаксическую связь в пассивной трансформации) и непрямые; разновидностью непрямых дополнений можно считать обстоятельства, выделяемые на основании характерной для наречий синтаксической позиции (на нее указывает возможность субституции наречием с обобщенным значением). Член предложения, обладающий двойной синтаксической связью (со сказуемым и подлежащим или дополнением), характеризуется как сказуемостное определение. Другие определения, не имеющие прямой связи со сказуемым, являются не членами предложения, а их модификаторами. Единицы семантической структуры предложения определяются на основании отношений предиката к выраженным лексемами аргументам (актантам). Семантические функции актантов (агенс, пациенс, бенефициент, результат и др.) устанавливаются на сигнификативном уровне с учетом значения предиката, отражающего оценку ситуации с точки зрения говорящего. 42 Различаются следующие наиболее крупные семантические типы предикатов: предикаты действия и предикаты состояния (к последним относятся также предикаты, обозначающие включение в определенный класс, признаки и процессы). Такая субкатегоризация предикатов позволяет группировать семантические функции (роли) актантов в более общие пучки (гиперроли), соответствующие субъектным, объектным и адвербиальным отношениям. Субъектные отношения объединяют функцию агенса при предикатах действия и функции пациенса, перципиента, бенефициента и дескриптива при предикатах состояния; объектные отношения — функции пациенса, результата, бенефициента, инструмента при предикатах действия и функции контентива, результата при предикатах состояния. Адвербиальные отношения объединяют функции причины, цели, места, направления, времени и способа действия, характерные для более отдаленных от предиката актантов, которые сами могут рассматриваться в качестве включенных предикатов. Таким образом, субъектные, объектные и адвербиальные отношения определяются терминами типов предиката и семантических функций (ролей). Моноцентрическая модель синтаксических связей в принципе соответствует логической и семантической структуре предложения. Однако единицы синтаксической структуры не могут быть прямо соотнесены с категориями мышления. Предикат может выражаться не только сказуемым, но и другими членами предложения, субъект — дополнением, объект — подлежащим. Различия между семантической и синтаксической структурой предложений обусловлены разными в каждом языке особенностями грамматического и лексического строя, а также фузионным характером кодирования семантических и коммуникативно-прагматических категорий. Каждый компонент предложения характеризуется определенным набором синтаксических, семантических и коммуникативно-прагматических признаков, возможные комбинации которых определяют специфику языка. Последовательный учет этих сложных отношений между единицами разных уровней позволяет уточнить интерпретацию многих дискуссионных конструкций. V. AMBRAZAS THE UNITS OF SYNTACTIC AND SEMANTIC SENTENCE STRUCTURE IN LITHUANIAN Summary The article deals with the definition and relations of the main units of sentence structure distinguished on each level of analysis. A differentiation of the syntactic concepts of surface subject, object and adverbial from the corresponding semantic concepts is proposed. The units of the syntactic structure are defined on the basis of the syntactic relations of subordination, interdependency and coordination. The syntactic centre of a sentence (the apex of all syntactic relations) is the grammatical predicate. The word form in the relation of interdependency with the predicate is defined as the grammatical (surface) subject, the word form subordinated to the predicate is termed a grammatical (surface) object. The direct object is distinguished from indirect by participation in syntactic processes such as passivization. Adverbial modifiers are regarded as a subcategory of indirect objects characterized by distribution typical of adverbs. The units of the semantic structure are defined on the basis of semantic relations between the predicate and verbalized arguments (actants). These relations are distinguished on the level of signification and conditioned by the meaning of predicate reflecting the evaluation of a cognitive situation by the speaker. Predicates are subcategorized into two main groups: predicates of action and predicates of state (including the indications of qualities, classes and processes). This subcategorization of predicates permits the roles of actants (“deep cases”) to be related to the concepts of semantic subject, object and adverbial. The subject relation covers the agent of the predicate of action and the patient, recipient, benefactive and descriptive of the predicates of state. The object 43 relation covers the roles of patient, result, benefactive, instrument of the predicate of action and content, result of the predicates of state. The adverbial relation covers the indications of cause, purpose, direction, time, and manner characteristic of participants which may be interpreted as incorporated (secondary) predicates. The notions of semantic subject, object and adverbial defined in terms of role groups and type of predicate are not universal but they are relevant to the description of sentence components and syntactic processes in Lithuanian and other languages of the accusative sentence type. Each component of a sentence can be characterized as the point of intersection of syntactic, semantic and pragmatic relations forming a specific network of possible combinations in the sentence structure of each language. The approach suggested helps to solve a number of questions concerning the interpretation of ambiguous constructions.