GRAMATINĖS KATEGORIJOS IR JŲ RAIDA

LIETUVIŲ KALBOTYROS KLAUSIMAI XVIII (1978)

N. SLIŽIENĖ

ON THE QUESTION OF VERB VALENCY

Of all parts of speech, the verb is particularly distinguished by its structural properties. For this reason, in recent years many linguists have regarded it as the center of the sentence, organizing the other words in it. B. Abramov, having divided all the potentials for word combination into centripetal ones, which indicate the dependence of one word on another, and centrifugal ones, which indicate the dominance of one word over another, noted that the verb is the only element in a sentence with exclusively centrifugal potentials (Абрамов 1966). This property of the verb—its ability to attach other words to itself—has increasingly been called verb valency, although linguists have not yet reached a consensus on the nature and scope of this phenomenon. In this article, we will attempt to outline the main conceptions of valency researchers and indicate our own view on the subject. The focus will be primarily on studies of verb valency.

1. FROM THE HISTORY OF VALENCY RESEARCH

The term valency is borrowed from chemistry and applied to designate certain potentials for word combination. In linguistics, it was first used by L. Tesnière, one of the best-known founders of valency theory (1934; 1953; 1959). L. Tesnière discussed only the valency of the verb as the center of the sentence. He compared the verb group to a small drama in which one can distinguish a process, actors, and certain circumstances. In terms of structural syntax, these are the verb, the participants in the action (actants), and circumstances (circonstants) (Tesnière 1959: 102—29). Actants participate in the action regardless of their degree of activity; for example, in the sentence Alfredas duoda knygą Šarliui, besides Alfredas, the participants in the action are Šarlis and the book. The number of actants is limited. L. Tesnière compared the verb’s capacity to take a certain number of actants with the valency of an atom, and likewise called it valency. According to the number of actants, he divided all verbs into avalent (without actants), monovalent (with one actant), bivalent (with two actants), and trivalent (with three actants) (Tesnière 1959: 238—82). L. Tesnière considered only nouns or their equivalents to be actants, and only three of their cases: the nominative for the subject and the accusative and dative for the object. Everything else, for him, consisted of circumstances (circonstants), whose number was supposedly unlimited and which were expressed by adverbs or their equivalents. L. Tesnière did not take into account that even the elements he called circumstances are sometimes necessary to the verb, and he disregarded them when classifying verbs.

Valency is now understood much more broadly. It is discussed not only in relation to verbs, but also to nouns, adjectives, and adverbs (Sommerfeldt 1971; Sommerfeldt — Schreiber 1974; 1975; Bondzio 1971: 90—1). Soviet linguistics distinguishes external and internal valency; the latter is determined at the level of a word’s components (Степанова 1973: 31—61). However, both Soviet,

107

and foreign linguists devote most of their attention to verb valency, since it is of decisive importance for sentence structure. German linguists have made particularly substantial contributions in this field¹. In recent years, interest has also turned to the valency of Lithuanian verbs (Енюшене 1970; Geniušienė 1971).

Although much has been written about verb valency, there is still no consensus as to what should be considered valency, what its limits are, or what level it belongs to; under such circumstances, no single definition acceptable to all can be proposed.

1.1. On understanding the scope of valency

1.1.1. Some linguists tend to regard valency as the grammatical relations of a verb with all the words dependent on it, that is, what is also called the verb’s capacity for combining (“сочетательная способность”, “Fügungspotenz”). For example, S. Katsnelson, the first to use the term valency in Soviet linguistics, defined this phenomenon generally as “the property of words to be realized in a certain way in a sentence and to enter into certain combinations with other words” (Кацнельсон 1948: 132)². Valency understood in this way encompasses not only the subject and object relations, but also all adverbial relations between a verb and other words. Such a multitude of connections cannot be relevant to the syntactic classification of verbs. For example, many adverbial specifications can occur with almost any verb (yesterday went, ate, thought, rained, planted, worked, etc.)³.

With valency understood so broadly, verbs are generally classified only according to what is called obligatory valency. Only those relations between a verb and other words whose realization is sufficient to make a sentence grammatical are considered obligatory valency. All other potential relations of the verb are considered facultative valency.

Nevertheless, the boundaries between obligatory and facultative valency are usually unclear, because distinguishing them conflates two criteria that do not fully coincide. It is said that the verb itself determines its obligatory environment, which is required by the verb’s semantics, whereas the facultative environment is merely potential (Почепцов 1969: 66–9; Pocheptsov 1970: 867; Geniušienė 1971: 8). On the other hand, it is claimed that obligatory valency is always realized in a sentence, that the obligatory environment of a verb constitutes the obligatory minimum of a verbal sentence, and that if any required form is omitted, the entire syntactic construction is incomplete; facultative elements, by contrast, are irregular, can combine with almost any verb, and their absence does not deprive the syntactic construction of either semantic or structural completeness (Адмони 1958: 113; Почепцов 1969: 69; Geniušienė 1971: 9; Лейкина 1961: 11; Болдырева 1970: 20; Иванчикова 1965: 85–6). In fact, the syntactic environment determined by a verb’s meaning is not fully realized in every case; some of its elements may remain unrealized without impairing either the meaning or the structure of the construction. For example, the instrumental case with an instrumental meaning is very often absent from a syntactic construction, and is therefore even considered facultative (Geniušienė 1971: 12). But it is, after all, determined by that

¹ East German linguists were the first to compile and publish a valency dictionary of German verbs (Helbig–Schenkel 1969). In 1973, the second, expanded and improved edition of this dictionary appeared; it was reprinted in 1975 (Helbig–Schenkel 1973; 1975). They also prepared a publication devoted specifically to questions of valency; see: Beiträge zur Valenztheorie. Halle (Saale), 1971.

² E. Geniušienė defined valency in a similar way: “The valency of a verb is its capacity to combine with nouns, adjectives, and other parts of speech” (Geniušienė 1971: 8).

³ “The diagnostic power of adverbial positions is small and, in many cases, practically nil. These positions can occur in the distributional formula of verbs belonging to the most diverse classes,” writes J. Fomenko (Фоменко 1975: 6). On the other hand: “Of the entire environment of a verb, only those constituents are relevant that form the verb’s obligatory distribution, especially those whose presence in the sentence is required by the verb’s semantics” (Pocheptsov 1970: 867).

108

the meaning of certain verbs (plowed with a tractor, cut with a knife, struck with a stick, and so on). Moreover, the instrumental case sometimes has significance for the syntactic classification of verbs. For example, in sentence (1), although it can be omitted, it indicates a different meaning and a different syntactic use of the verb blaškyti than in sentence (2), where it is unnecessary.

(1) The tail flicks flies away. (2) The girl shakes (dusts) flour-covered sacks.

Likewise, the prepositional constructions in sentences (3) and (4) could be omitted without making the sentences ungrammatical, yet they are determined by the verb’s meaning; they even signal different meanings of the verb džiauti:

(3) Mother dried clothes on a line. (4) She fleeced people of money, clothes, and other things.

Veiksmažodžio susiginčyti būtinaisiais determinantais E. Geniušenė (1971: 13) laiko tik vardininką ir konstrukciją dėl + kilmininkas, bet juk šio veiksmažodžio reikšmė suponuoja ir kontrაგენტo reikšmės realizaciją, pavyzdžiui, su + įnagininkas (susiginčijo — kas? su kuo? dėl ko?). Taip pat ir veiksmažodis pasakoti determinuoja ne tik agento vardininką ir turinio reikšmės prielinksninę konstrukciją (apie + galininkas), bet ir adresato naudininką (pasakoja — kas? kam? apie ką?).

Thus, we see that the obligatory environment of a verb, determined in light of its determining power, and the obligatory environment identified on the basis of its required syntactic realization do not coincide. How, then, can there be clear boundaries between obligatory and facultative valency?

1.1.2.

Other linguists regard only those combinatory potentialities that are presupposed by the meaning of the verb and are relevant to the classification of verb syntax as the verb’s valency. For example, G. Helbig, who has perhaps written more than anyone else about verb valency, described this phenomenon as the verb’s ability (Fähigkeit) to open certain “empty slots” for itself, which must be occupied by obligatory or facultative partners (Mitspieler) (Helbig – Schenkel 1975: 49). G. Helbig calls those sentence constituents that are attached to the verb but do not occupy the “empty slots” associated with it free indications (freie Angaben). For example, the verb wohnen “to live” requires an indication of place, whereas sterben “to die” does not, cf.: Er wohnte in Dresden; Er starb in Dresden (Helbig—Schenkel 1975: 38).

G. Helbig determines the verb’s valency partners on the basis of the deep structure of the sentence: it contains obligatory and facultative partners, but no free indications. In analyzing a sentence, he identifies constituents that correspond to the arguments of the logical predicate — these are the verb’s partners. He considers those constituents that can be treated as potential independent predicates to be free indications. For example, G. Helbig divides the sentence Die Kinder badeten an der Spree into Die Kinder badeten and Das Baden war (geschah) an der Spree; consequently, the indication of place is not included in the predicative structure of the verb baden as an argument. The sentence Berlin liegt an der Spree, however, cannot be divided in this way, because it corresponds to a single two-place predicate R (a, b) (Helbig 1971 b: 274).

To distinguish obligatory from facultative partners, G. Helbig uses the transformation of eliminating constituents of the sentence’s surface structure: facultative partners can be omitted without making the sentence ungrammatical, whereas obligatory partners cannot be omitted.

These criteria proposed by G. Helbig are interesting and worthy of attention, but the problem is that the first of them, which requires a clear understanding of deep structure, is difficult to apply. The concept of deep structure is generally still not very clear in linguistics, and G. Helbig understands it in his own way, too, treating it as some kind of “auxiliary syntactic level” close to logical structure (1971b: 274) and differing from the deep structure of generative grammar in being even more abstract and universal (1969: 162).

109

1.2. Three levels of valency and three conceptions of valency researchers

Currently, the question of which level valency relations belong to is topical and much debated. The lack of clarity arose through the fault of L. Tesnière, the pioneer of valency theory, who left the matter unclear and also, as W. Bondzio (1971: 88) has pointed out, conflated different levels by identifying actants belonging to the semantic-logical level with cases-constituents belonging to the grammatical level.

The question is now usually posed as follows: is valency a formal or an abstract category; does it belong to the level of expression or to that of content (Helbig 1971b: 279; Helbig – Schenkel 1975: 63)? If it is a formal category, it can be studied in each individual language as a specific phenomenon; if it is abstract, then it is universal and can be described using the tools of the logic of relations.

1.2.1. G. Helbig speaks separately of logical, semantic, and syntactic valency, which are determined at the corresponding levels and reflect the relation between reality, consciousness, and language (Helbig 1971b: 280; Helbig – Schenkel 1975: 65–6). Logical valency is the circumstances of reality formulated as propositional structures, i.e. as predicates with several empty slots; verbs are usually regarded as propositional operators, and nouns as arguments of the logical predicate, but what matters most is not this linguistic realization, but the fact that we are dealing with a logical predicate with a certain number of arguments. Logical valency is non-linguistic and therefore universal.

Semantic valency is the verb’s requirement for certain context partners with specific distinctive semantic features and its inability to have partners with different semantic features. Thus, G. Helbig regards semantic valency as selectional restrictions governed by the semantic compatibility between the verb and its partners.

G. Helbig understands semantic valency more broadly than logical and, as we shall see below, syntactic valency. Logical and syntactic valency encompass only the relations between the verb and its partners, whereas semantic valency also includes the verb’s relations with free indications; the latter, G. Helbig maintains, “can be freely added and removed syntactically, and are restricted only by semantic valency” (Helbig 1971b: 280).

Other authors define semantic valency more narrowly. For example, W. Flämig (1971: 109) describes it as follows: “The semantic valency of the verb regulates the filling of empty slots by classes of semantically motivated partners, i.e. verbs require partners — words with specific distinctive semantic features corresponding to their role in the circumstances of reality: agent, patient, product, source or endpoint, location... The appropriate partner — a word — is selected on the basis of the semantic compatibility between the verb and its complements, which is in turn motivated by the structure of the circumstances of reality.” Understood in this way, semantic valency does not differ in scope from logical and syntactic valency.

G. Helbig characterizes syntactic valency as the filling of obligatory or facultative empty slots; the number and nature of the slots are determined by what the verb requires in each individual language.

G. Helbig emphasizes that these levels, although closely interconnected, are not identical and cannot be represented isomorphically as following one after another, since not all abstract relations are realized in language in the same way, let alone realized in different languages. For example, the verbs helfen “to help” and unterstützen “ditto” presuppose the same abstract logical relation, which can be written R (a, b), but argument b is realized syntactically in different ways (helfen requires the dative, whereas unterstützen requires the accusative). Special rules for transferring from one level to another are needed, but they would have to be very complex, and for this reason no one has yet devised them (Helbig 1969: 167; 1971a: 43; Helbig – Schenkel 1975: 68–9). Therefore, G. Helbig studies valency as a formal phenomenon at the level of expression, specific to each individual language. His point of departure is the verb’s relations with

110

in other words in the external structure of the sentence, but he nevertheless seeks the criteria for determining valency in deep structure (Helbig 1971b: 273—4).

1.2.2. W. Bondzio and W. Flämig approach this matter from an entirely different angle. They take the actual meaning of the word as their point of departure and locate the basis of valency at the logical level (Bondzio 1969; 1971: 88—95; Flämig 1971: 107—11). The verb’s constructive role in the sentence corresponds to the role of a logical predicate in a propositional function, which consists of a logical predicate and its open slots, e.g.: x sleeps, x defeats y. Propositional functions record circumstances in reality that have been reflected in consciousness as structures of thought. In logical terms, circumstances in reality are individuals and their properties (attributes or relations). A propositional function shows whether an individual’s property is reflected, e.g., P(x), that is, x has the property P (Gold shines), or a relation between individuals, e.g., R (x, y), that is, x bears the relation R to y (Peter is looking for Paul). In the first case the predicate is one-place, and in the second, two-place. The logical relation in the structure of a proposition corresponds to the semantic relation in the structure of a sentence. Thus, propositional structures are certain intermediate auxiliary constructions that help link the sentence with the circumstances in reality reflected in consciousness, although neither system is entirely unambiguous or derivable from the other.

W. Flämigas (1971: 110) savo valentingumo koncepciją atvaizdavo trijų lygmenų schema. Sąmonėje atsispindėjusių tikrovės aplinkybių pagrindu susidaro loginės (minties) struktūros, kurias reprezentuoja predikatas ir jo argumentai. Toliau eina semantinės struktūros. Jas reprezentuoja žodžių žymima tarpusavio derinimo santykiai, atsižvelgiant į tam tikrus vaidmenis tikrovės aplinkybėse. Trečiasis lygmuo — sintaksinės struktūros, kurias reprezentuoja hierarchiniai santykiai tarp sakinio konstituentų. Loginės struktūros priklauso abstrakčiam nekalbiniam lygmeniui, o semantinės ir sintaksinės struktūros — konkretiems kalbos lygmenims.

Abstract relational structures at the logical level form the basis of word valency. The number of open slots depends on the abstract content of the predicate; when discussing verb valency, it depends on the content of the verb. In W. Bondzio’s words, “has valency” primarily means that a linguistic expression, on the basis of its abstract content, has open slots in the sense of relational logic (1969: 234—5). Thus, the actual meaning of the word must be the point of departure, and valency is a property of meaning and, as such, characteristic not only of verbs but also of other parts of speech (cf. also Bondzio 1971: 89). Similarly, K. E. Sommerfeldt and H. Schreiber (1974: 15) define valency as a word’s capacity, arising from its meaning, to combine with other words.

If logical structures are taken as the basis of valency, the question inevitably arises of how to move from the abstract logical level to the concrete linguistic level. W. Flämig himself (1971: 111) noted that hypothetical valency models do not always coincide with concrete sentence structures. Some models, it turns out, can be reduced. W. Flämig, drawing on Ch. F. Hockett, explains this phenomenon as follows. A morpheme’s valency may remain unsaturated, in which case it also encompasses the situational context (“… daß dann das Morphem seine Valenz in den Situationskontext erstrecke”). Unoccupied open slots in concrete sentences are understood as contextual economy, while the constituents are regarded as known and presupposed throughout the context. The number of open slots, as a phenomenon at the logical level, does not depend on whether they are filled by concrete words at the external level or not, since abstract valency models are in each case represented by complete structures. When these do not correspond to linguistic realization, Flämig’s view is that one may speak of their optional reduction, but not of optional fillers of open slots or optional valency. This view follows from a purely logical understanding of valency.

Kalbėdamas apie abstrakčių valentingumo modelių realizavimą, W. Bondzio (1969: 236—9; 1971: 95—100) siūlo lyg ir tam tikras perėjimo iš vieno lygmens į kitą taisykles, tačiau reikia sutikti su G. Helbigu, kad jos šiam reikalui toli gražu nėra pakankamos. Užpildant laisvas vietas, neužtenka žinoti žodžio — valentingumo

111

the possessor’s nature and grammatical form. W. Bondzio also stresses the need to take into account the possibility of alternative realizations, when the same abstract relational structure corresponds not to one but, let us say, to two linguistic realizations; however, it is unclear how this can be anticipated on the basis of the logical level. It is true that W. Bondzio remarked in one place that the results of semantic analysis can be compared with language practice, or that analysis of open slots can begin with linguistic usage so as to determine which constituents occupy them (Bondzio 1969: 238), but he did not develop or specify these ideas further.

1.2.3. J. Apresian’s conception of valency is also close to the views of W. Bondzio and W. Flämig, though his approach is more semantic⁴. His focus is the meaning of a word and the semantic valencies that derive from it, which enable the word to take syntactically dependent words (Апресян 1974: 119 ff.). The number of a word’s semantic valencies coincides with the number of participants in the situation it denotes. In analyzing a situation, J. Apresian uses the tools of logic and represents it as a complex predicate. He divides complex predicates into simpler ones and then determines the total number of valencies of those predicates. For example, the author describes the situation denoted by the verb сообщать as follows: person A causes person B to know fact C. The predicate сообщать is divided into two simpler two-place predicates, cause and know, which share the valency B. Thus, there are three valencies, and сообщать is a three-place verb (1974: 122)⁵. This method of determining the number of valencies is well motivated semantically, but it must be said that it is by no means easy to describe semantically the situation denoted by each word (for example, each verb) and to divide the corresponding complex predicate. Perhaps this will become possible once semantic research has reached an appropriate level.

In analyzing the situation denoted by a word, J. Apresian determines not only the number of its semantic valencies but also their nature (the meanings of subject, object, addressee, counteragent, instrument, etc.). On the basis of this information, he constructs a government model that reflects the possibilities for the morphological realization of valencies. The government model must record all possible means of morphological expression for each semantic valency; for example, a valency of the content (i.e. explicative) meaning of the verb надеяться may be realized by a subordinate clause (надеюсь, что успешно окончу университет) or by a prepositional construction (надеюсь на успешное окончание университета) (1974: 143). The same semantic valencies of different words are realized in still more varied ways; cf. instrument realizations: резать ножом, прыгать с парашютом, стрелять из ружья, курить трубку, проехать через марлю, тереть на терке (1974: 146). Finally, one or another semantic valency of some words is not realized at all. For example, the verb разбинтовать has the same number of valencies as забинтовать, but one of them remains unrealized; cf. забинтовать руку марлей, but разбинтовать руку (1974: 147). It is hardly possible to anticipate all such variation in morphological realizations in a government model that takes word meaning and the semantic valencies deriving from it as its point of departure and relies solely on knowledge of the language.

Just as G. Helbig speaks of obligatory and optional syntactic valency, J. Apresian (1974: 149 ff.) speaks of the syntactic obligatoriness and optionality of valencies, determined in the sentence by means of elimination. He links these semantic properties with the content and morphological expression of valencies and notes, for example, that the subjec-

⁴ J. Apresian studies valency only to the extent required by his semantic studies. Valency interests him as a certain auxiliary means that helps to describe word meanings semantically.

⁵ Panašiai skaidyti žodžių reikšmes siūlo ir W. Bondzio (1971: 92–4). Pavyzdžiui, veiksmažodžio duoti reikšmę jis suskaido šitaip: A daro, kad B turėtų C. Šią išraišką jis užrašo kaip funkcinės struktūrą A daro: B turi C. Tais atvejais, kai objektas atsiranda kaip veiksmo rezultatas funkcinė struktūra yra A daro: B yra (egzistuoja), kai objektas dėl veiksmo poveikio kinta —, A daro: B turi ypatybę e arba A daro: B yra e.

112

The valencies of the goal and object are syntactically obligatory more often than those of instrument and place, and the latter two more often than those of cause and time; moreover, bare cases are more often obligatory than prepositional constructions. Most importantly, however, any valency may be syntactically obligatory for some words and optional for others.

Unlike other researchers, J. Apresian also discusses the semantic obligatoriness and optionality of valencies (1974: 124–5). Some valencies seem highly probable, although not all are semantically obligatory; for example, the goal valency with verbs of directed motion (to go to eat, to run to buy something). When determining a word’s valency structure, it is suggested that such valencies be taken into account, while their semantic optionality is discussed separately.

1.2.4. Three approaches to verb valency emerge from the concepts discussed: syntactic, logical, and semantic. For G. Helbig, what matters most is the verb’s relations with other words at the syntactic, external level. He approaches valency as a practitioner who needs to define the usage environment of each verb, that is, to specify with words belonging to which classes and in which forms the verb is used in a sentence, as well as to indicate which of its partners are obligatory and which optional. All these specifications are very useful for constructing well-formed sentence patterns and are especially helpful when learning foreign languages. G. Helbig does not attempt to answer the question of why a verb requires certain word forms and not others, or what determines the number of its free positions—in short, what constitutes the basis of valency. He leaves this question aside, although he acknowledges that in many cases there is nevertheless a direct relationship between valency and meaning, and that valency and distributional relations are, in general, a reflection of semantic phenomena. Thus, it is no coincidence, for example, that most verbs meaning ‘give’ and ‘inform’ in the Indo-European language family are trivalent, since their content implies a giver (informant), something given (communicated), and a recipient, and semantically requires an agent, an object, and a goal. Finally, semantic relations can be structurally grasped and formally described precisely through valency and distributional relations (Helbig – Schenkel 1975: 60).

On the other hand, G. Helbig’s deep structure nevertheless seems very close to the logical level (if not entirely identical with it). Thus, with certain reservations, it can be said that he, too, sees the basis of valency at an abstract logical level, while associating its realization with concrete levels of language.

J. Apresian approaches valency as a practitioner as well, but his aim is different—not syntax, but semantics. Since a word’s valency properties follow from its lexical meaning, knowing the word’s valency model makes it possible to describe its meaning semantically.

W. Bondzio and W. Flämig are more concerned with what constitutes the basis of valency; they seek to answer why a verb (or any other word, since valency is not solely a property of verbs) has a particular number of free positions that must be filled by appropriate word forms. However, because they focus primarily on constructing hypothetical valency models on the basis of word meaning, questions of syntactic realization seem to be left somewhat aside by W. Bondzio and W. Flämig.

Thus, according to all three of the aforementioned conceptions (logical, semantic, and syntactic), valency is understood in roughly the same way: it has the same scope and the same conditions of existence, but the approaches to it differ and, consequently, so do the definitions.

2. WHAT IS VERB VALENCY, AND WHAT ARE ITS CHARACTERISTIC FEATURES

2.1. In studying verb valency, the first question that arises is what its scope should be: does it encompass all possible relations between a verb and other words, or only those entailed by the verb’s meaning? It has already been

8. Gramatinės kategorijos ir jų raida

113

It has been noted that a broad understanding of valency offers little practical benefit (see 1.1.1.); moreover, it lacks a common theoretical foundation. Both the combinatory potentials conditioned by a verb’s meaning, which have a subclassifying function, and the purely grammatical capacity of a verb to combine with other words and their forms, which is characteristic of verbs in general, are lumped together (cf. Apresian 1974: 120). The realization of all a verb’s potentials creates the environment of its syntactic use, or its distribution, usually defined as the set of contexts in which a particular linguistic element can occur, as opposed to those contexts in which it cannot (Axmanova 1966: 137)⁶.

Thus, it seems more appropriate, when discussing verb valency, to restrict ourselves to those combinatory potentials that follow from the verb’s meaning.

2.2. Each of the three valency conceptions mentioned above emphasizes one aspect of valency—logical, semantic, or syntactic. However, by absolutizing any one of them, insufficient attention is paid to the others. It therefore seems natural to consider valency a multilayered phenomenon, grounded at an abstract level and realized at concrete linguistic levels⁷. Logical, semantic, and syntactic valency cannot be discussed in isolation, since they are different aspects of one and the same phenomenon, forming a dialectical unity. If we study valency only as a syntactic phenomenon, we cannot determine the verb’s open slots and thus the number of its partners; if we study it only as a logical or semantic phenomenon, we will lack a clear picture of its linguistic realization. Thus, valency must be studied at all three levels⁸.

Since what matters for valency is not only the number of a predicate’s arguments but also their semantic roles (agent, patient, addressee, instrument, and so on), it seems appropriate to speak of an abstract logical-semantic level, at which the number of open slots in a verb as predicate and their semantic content are established. At the concrete semantic level of language, semantically motivated word classes, or even individual words, are identified as potential partners capable of filling the verb’s open slots. At the syntactic level of language, the morphological forms of the verb’s partners are established; moreover, it is only at this level that obligatory and facultative partners are distinguished.

3. DETERMINING VERB VALENCY

3.1. Since verb valency includes only combinatory potentials that follow from meaning, the researcher’s point of departure must therefore be the meaning of verbs. Each verb denotes a particular situation involving a certain number of participants with defined meanings—participants in the situation, or actants. For example, the verb pjauti ‘to separate a part by cutting with a sharp object’ denotes a situation in which three actants must participate: someone who cuts (agent), what is cut (patient), and what is used to cut (instrument). Thus, this situation can be represented as a predicate with three open slots: R (x, y, z). These slots must be filled by arguments, one with the meaning of agent, another of patient, and the third of instrument.

⁶ G. Helbigas ir W. Schenkels (1975: 50) distribuciją laiko tik veiksmažodžio partnerių morfologine ir semantine charakteristika, t. y. kokybine valentingumo puse, nors čia pat, remdamiesi Z. S. Harrisu, kalbinio elemento distribuciją jie apibūdina kaip sumą visų apsupčių, kuriose jis pasitaiko. Šį prieštaravimą jau yra pastebėjusi M. Stepanova (1973: 19).

⁷ Plg. Sommerfeldt—Schreiber 1971: 227, kur teigiama, kad loginiu valentingumu grindžiamas sintaksinis ir semantinis valentingumas. Vėliau (1974: 21) autoriai šią mintį kiek modifikavo, sakydami, kad sintaksinis valentingumas yra sukurtas loginio-semantinio valentingumo pagrindu.

⁸ Panašiai valentingumą interpretuoja W. A. Buchbinderis (1971: 284), kuris sako, kad valentingumas savo esme apima tiek loginį, tiek ir visus tris kalbinius lygmenis. Trys jo skiriami kalbiniai lygmenys – tai gramatinis, leksinis-semantinis ir uzualinis, kuriuo autorius laiko žodžių specifines jungimo taisykles, naudojamas „kalbos jausmo“ pagrindu.

114

At the concrete level of language, the predicate is a verb, while its arguments are words or constructions in particular forms whose meanings correspond to the aforementioned agent, patient, and instrument roles and are semantically compatible with the verb’s meaning. Since the situation involves an instrument, the agent role is filled by a person; the patient is an object that can be cut, while the instrument role is usually filled by a knife, saw, scythe, certain machines, and so on. Thus, the verb pjauti has three partners—nouns (or their equivalents) which, when used in the appropriate forms, respectively denote the person performing the action and bear the agent role, the object affected and bear the patient role, and the object used to perform the action and bear the instrument role. On the basis of our knowledge of the language, we can construct the corresponding syntactic realizations:

(5) Father cuts the hay with a scythe. (6) The collective farmers harvest the grain with combine harvesters.

We test whether all the verb’s partners are obligatory: (5a) Father cuts the hay. (5b) *Father cuts with a scythe. (5c) *Father cuts. (5d) *Cuts the hay with a scythe. (6a) The collective farmers harvest the grain. (6b) *The collective farmers harvest with combine harvesters. (6c) *The collective farmers harvest. (6d) *Harvests the grain with combine harvesters.

We see that the instrumental case of the noun denoting the instrument, either on its own or with the preposition su, can be omitted without making the sentences ungrammatical; therefore, this partner of the verb pjauti is facultative. If the nominative denoting the agent or the accusative denoting the affected object is omitted, the sentences become incomplete (such sentences are possible only in a broader context, when the omitted elements have been mentioned earlier or are well known to the participants in the conversation). In certain cases, when a sentence has no nominative denoting the agent, the verb’s endings indicate the agent (cf. K. Kolпачский 1965: 173 ff.), e.g.: When will you harvest the rye?; Get on with the harvesting; Yesterday we harvested the hay until dusk; In the old days they used to harvest rye with sickles (the agent is generic).

Thus, in this case the syntactic valency model of the verb pjauti can be represented by the formula SₙVSₐ (Sᵢ / suSᵢ)⁹, with the facultative partner enclosed in parentheses. The facultative element may be either a noun in the instrumental case or an instrumental form with the preposition su. The slash / indicates alternation between these forms.

However, other realizations of the verb pjauti are also possible, e.g.: (7) The combine harvester not only cuts the grain but also threshes it.

Here, the noun in the nominative has the meaning of instrument, not agent (i.e. an actor who performs an action on their own initiative), and in this case the agent can no longer be realized in the sentence (cf. Fillmore 1969a); the valency of the verb pjauti is syntactically realized according to the model SₙVSₐ. Here the interrelation and conditioning between semantic and syntactic valency are especially clear. The syntactic realization of the verb depends on the semantic role of the nominative it requires.

When the verb pjauti means that an object is capable of cutting, that it is sharp, it is most often used with only one partner—the nominative; however, the accusative is also entirely possible, since a cutting situation cannot be imagined without something being cut (cf. Апресян 1974: 149).

(8) Does your scythe still cut? (9) This saw cuts even oak.

⁹ The following symbols are used in the formulas: S — noun (or its equivalent), n — nominative, g — genitive, d — dative, a — accusative, i — instrumental, loc — locative, il — illative, pSᵣₑc — prepositional construction expressing direction, pSₗₒc — prepositional construction expressing location, and so on.

8*

115

The accusative here is an optional verb partner, so the syntactic valency model of the verb pjauti is, in this case, SₙV (Sₐ).

Similarly, the valency of the verb papjauti, “to be able, to manage to cut,” is syntactically realized according to the model SₙV (Sₐ) (Sᵢ / with Sᵢ), where not only the accusative but also the instrumental or the alternative prepositional construction is optional.

(10) Our children can already cut hay.

(11) Grandfather can no longer cut with a scythe.

When discussing the realization of verb valency, it is not enough to establish only its syntactic model. The same model may fit, for example, two meanings of a verb. Compare (5) and (6) with (12) and (13), where the verb pjauti means “to kill with a sharp tool”:

(12) Tomorrow mother will slaughter a goose.

(13) Raugys slaughters pigs only with his own knife.

In both cases, the valency of the verb pjauti is realized according to the model SₙVSₐ (Sᵢ / with Sᵢ), but the lexical filling of the Sₐ position differs: in (5) and (6), this position is occupied by words denoting inanimate things, whereas in (12) and (13) it is occupied by words denoting living beings. (14), in turn, differs from (12) and (13) not only in that the Sₙ position contains a word denoting not a person but merely a living creature, but also in the resulting absence of a partner with an instrumental meaning: SₙVSₐ.

(14) Wolves slaughter sheep.

Some verbs denote the same situation but differ in their significative meanings. Their syntactic usage also differs. Such verbs form certain pairs because the actions they denote condition one another, e.g.: give – take, give – receive, sell – buy, frighten – become frightened, anger – become angry, and so on.

(15) The father gives an apple to his son.

(16) The son takes an apple from his father.

(17) The hare frightened the sheep.

(18) The sheep were frightened by the hare.

Both sentence (15) and sentence (16) denote the same situation, involving a father, a son, and an apple, and in both cases the father gives and the son takes the apple. However, when expressed by the verbs duoti and imti, it is interpreted differently from a significative point of view (cf. Аппресян 1967: 5 – 6, where such sentences as Petras — Ivano tėvas and Ivanas — Petro sūnus are compared). When the verb duoti is used, the nominative foregrounds the actant who gives, whereas with imti it foregrounds the actant who takes. This is precisely what constitutes their significative difference, which determines the different syntactic valency models of these verbs: SₙVSₐSᵢ and SₙVSₐ (from Sₙ)¹⁰. The optionality of the third position in the second model is illustrated, for example, by sentences such as:

(16a) The son takes an apple that is bigger.

(16b) Here, son, take an apple.

Between the verbs in sentences (17) and (18), there is likewise only a significational difference, which determines the different syntactic valency patterns of these verbs: SₙVSₐ and SₙV(Sg). In sentence (17), the hare’s role is expressed in the nominative, whereas in (18) it is the state of the sheep. Realization of the Sg position in the pattern of the verb be frightened is optional, since sentence (18a) is also grammatical:

(18a) The sheep were very frightened.

¹⁰ The verb take is also used according to a different syntactic pattern, namely SₙVSₐ (pSₗₒₒ), e.g.: The son takes an apple from the basket / off the shelf / from under the table / from behind the board. In this case, the third actant in the situation it denotes plays the role of the point of origin; used in this way, the verb take does not form a pair with the verb give. In addition, it should be noted that, for convenience, incomplete valency patterns are presented here. Both the verb give and take require either an accusative or a genitive with the meaning of Patient, so alternation between them is possible; compare sentences (32), (33), (34), and (35) with the verb put and its valency syntactic pattern given below.

116

Sometimes one and the same verb can perform different significational functions.

(19) Mother covered the table with a tablecloth.

(20) Mother covered the tablecloth over the table.

Both sentences denote the same situation, involving the mother, the table, and the tablecloth; in both cases, the mother places the tablecloth on the table. However, this situation is interpreted significationally in one way in one case and in another way in the other, and the different syntactic uses of the verb cover depend on this. In one case, the table is expressed in the accusative, indicating what is done to it; in the other, the tablecloth is expressed, indicating what is done with it. The different significational functions correspond to different syntactic valency patterns of this verb: SₙVSₐ (Sᵢ / with Sᵢ) and SₙVSₐ over Sg. In sentence (19), the instrumental, either alone or with the preposition “with,” may be omitted without affecting the grammaticality of the sentence; therefore, this position is considered optional.

3.2. This method of determining valency, which proceeds from the abstract logical-semantic level to the concrete linguistic levels—the semantic and syntactic—as G. Helbig has noted more than once (see 1.2.1), is not convenient in practice and is unlikely to provide a complete picture of semantic and especially syntactic realization. It can be used in cases where we have a verb of known meaning and want to construct one or another well-formed sentence with it.

Looking at just one or another relatively simple example, we have already seen that there may be no complete correspondence (1:1) between predicate arguments at the abstract level and a verb’s partners at the concrete level of language; there may be more predicate arguments than verb partners (see, e.g., 7). Moreover, even with the best command of a language, it would be difficult to anticipate all possible forms of syntactic realization.

(21) Father beat the horse with a whip.

(22) Ona beat the rug against the wall.

(23) The pupils beat their feet against the floor.

Comparing these sentences, we see that in all three cases the situation denoted by the verb mušti involves three participants: an agent, a patient, and an instrument, but some of them are realized grammatically in different ways. In all three sentences, the agent is expressed by a noun in the nominative, but the patient is realized in (21) and (22) by a noun in the accusative, and in (23) by a prepositional construction; the instrument is realized in (21) and (23) by a noun in the instrumental, and in (22) by a prepositional construction. It is true that the different forms denote somewhat different instruments: in (21) and (23), the instrument is unstable and movable, whereas in (22) it is stable. Compare also (24), (25), (26), and (27), where several participants in the situation of the verb draugauti are realized grammatically in different ways, and (28), (29), (30), and (31), with various realizations of the content:

(24) Sauliukas is friends with the children from the yard.

(25) The children from the yard are friends with Sauliukas.

(26) Sauliukas and the children from the yard are friends.

(27) The children are friends.

(28) Mother learned some good news.

(29) Mother learned about her son’s arrival.

(30) Mother learned that her son was arriving.

(31) Mother learned of her son’s arrival.

It is much more reliable to determine verb valency on the basis of a sufficiently large body of collected material illustrating the syntactic use of individual verbs. Following this approach, one can compile a verb valency dictionary that records all possible valency patterns for each verb and also reflects their different meanings.

The verb is examined in various syntactic environments. It is immediately apparent that in many cases different meanings are realized in different environments

117

the meaning of the verb or nuances of meaning. Cf. the meanings of the verb dėti in the following sentences:

(32) I put the brushes on the shelf. (33) They put plantain by the wound. (34) Did you put salt in the borscht? (35) Mother put eggs into the pot. (36) Spiders lay round, whitish or yellowish eggs. (37) Petras puts teeth on the rakes.

We can see three clearly distinct meanings and, correspondingly, different valency patterns of the verb. In the first four sentences, the verb dėti denotes a situation in which objects are moved to a certain place through the efforts of an agent. This situation corresponds to a logical-semantic structure consisting of a predicate and three arguments, one of which has the meaning of agent, another that of patient, and the third that of endpoint. By comparing the logical-semantic structure with the syntactic one, we can see which sentence constituents correspond to the arguments with the aforementioned meanings¹¹. In (32), (33), and (34), the agent is expressed by the ending of the verb dėti, whereas in (35) it is expressed by the nominative of the noun. In (32) and (33), the patient meaning is borne by the accusative of the noun, and in (34) and (35) by the genitive. The meaning of the genitive is somewhat modified—it denotes an indefinite object or quantity of objects and thus differs from the accusative. However, the syntactic relation of the verb to the accusative and to the genitive is the same, and the meaning of the verb is identical in both cases; therefore, from a syntactic point of view, the accusative and genitive are alternative forms. This means that the verb requires either one form or the other, while the choice between them depends on communicative aims (e.g., whether the reference is to a definite or indefinite quantity of objects, to the whole object or to an indefinite part of it). It is by no means necessary that in every case the accusative could be replaced by the genitive, or vice versa. There are nouns whose genitive is not even possible in the patient position; for example, in the sentence Vaikas deda galvą ant pagalvės, only the accusative is possible.

Similarly, in (35) the endpoint meaning is expressed by the illative, whereas in the other three sentences it is expressed by prepositional constructions with prepositions indicating various directions. These too are alternative forms.

Thus, when used with the meaning mentioned above, the verb dėti requires three partners: 1) a noun in the nominative, 2) the accusative or genitive, and 3) the illative or a prepositional construction denoting direction. All the partners are obligatory. Consequently, the verb’s syntactic valency pattern is SₙVSₐ / SₙSₐSᵢ / SₙSᵣₑc. All other sentence constituents dependent on the verb can only be free modifiers, since they do not correspond to any argument in the verb’s predicative structure. Cf.:

(32a) I always put the brushes neatly on the shelf. (33a) They immediately put plantain by the wound so that it would draw out the heat.

In sentence (36), dėti means “to produce from oneself,” and the situation it denotes involves an agent and a result. In the sentence, these are realized by the nominative and accusative of the noun. Consequently, in this case the verb dėti requires two partners: a noun in the nominative and one in the accusative, and both partners are obligatory. Its syntactic valency pattern is SₙVSₐ. Since the semantic realization of position Sₐ is very restricted (only one lexeme is possible), when speaking of domestic birds, for which laying eggs is an almost constant characteristic, the accusative is often not expressed at all, since the result of the action is very clear and self-evident, e.g.:

(36a) This hen lays every day. (36b) The hen is laying—don’t poke it.

¹¹ The meanings of agent, patient, addressee, and so on are considered semantic meanings of cases (and of cases with prepositions), belonging to deep structure; case forms, by contrast, belong to surface structure (cf. Fillmore 1968; 1969a). It is precisely on this basis that the logical-semantic structure can be compared with the syntactic one.

118

The meaning of these sentences is clear without the support of context or situation, because when the verb deda is uttered, only one possible realization of the Sₐ position is intended. Since the sentences are complete in meaning, their structural incompleteness is not felt either. However, we cannot regard the Sₐ position as optional, because when speaking about other animals that lay eggs, the accusative cannot be omitted without a broader context; for example, it would be impossible to say: In autumn, spiders lay; Sparrows lay; Flies lay.

In sentence (37), the verb dėti means “to fit a missing part onto an object,” and the situation it denotes involves an agent, a result, and a recipient. In the syntactic structure, these are realized respectively by the nominative, accusative, and dative cases of nouns, which are the obligatory partners of the verb dėti when used in this sense. The syntactic valency model is SₙVₛAₛ. The dative can sometimes be left unexpressed even without a broader context, but only when the recipient is a well-known and self-evident object, e.g.:

(37a) He puts good half-soles in.

(37b) This dentist puts teeth in well.

In many other cases, the dative cannot be omitted. For example, if we omitted it in (37), the sentence would be understood quite differently, while a sentence such as Petras puts on a new leg would be entirely unclear without knowing for whom the leg is being put on. Therefore, the dative cannot be regarded as optional in this valency model.

If one or another partner of a verb occurs in some sentences but not in others, then it may be optional. However, from (36a), (36b), and (37a), (37b), we saw that in certain cases an obligatory partner of a verb may also be left unexpressed. Moreover, in sentences in spontaneous dialogue, when the speakers are talking about familiar things, almost any obligatory partner can be omitted, even the verb itself. Obligatory partners may be absent from a sentence taken from a broader context if the relevant things were mentioned earlier and are now known (cf. Адмони 1966: 167). Such sentences would be unclear without context. For example, the sentence He is already finishing is perfectly understandable if the preceding discussion concerned someone’s work, but taken in isolation it is clearly structurally incomplete and incomplete in meaning. Therefore, for the purposes of valency, simple, complete sentences are used, spoken in a calm voice, without any particular emotion, to quote V. Admonis, when there is no additional influence from the situation or context (Адмони 1966: 167).

The absence of optional partners does not make a sentence incomplete either structurally or semantically (see, e.g., 5a, 6a, 8, 10, 11, 12, 16a, 16b, 18a). Moreover, expressed optional verb partners can be omitted not merely in some other sentence, but in every sentence in which the verb is used with the same meaning, except when they are necessary from a communicative point of view, since structural necessity and optionality do not by any means always coincide with communicative necessity and optionality (for a fuller discussion, see Учехонова 1973). For example, in sentence (38), the instrumental case expressing an instrument is structurally optional and could be omitted (cf. e.g. 39), unless the rest of the sentence indicates that it is communicatively necessary (omitting it would make the sentence meaningless).

(38) Eat away that stain with a damp brush, otherwise you won’t get it clean.

(39) Eat away that stain.

Sometimes one of a predicate’s arguments is not realized, and cannot be realized, at the external level of language because it is incorporated into the predicate’s meaning as part of it. For example, the situations denoted by the verbs uogauti, grybauti, riešutauti, obuoliauti, vartliauti, kidusiniaduti, pietauti, pusryčiauti, vakarieniauti involve an agent and a patient, but at the external level these verbs require only one partner—the nominative expressing the agent—because their meaning specifies the patient (cf. Fillmore 1969b: 118–20). The syntactic valency model of these verbs is SₙV.

3.3. It is important to establish not only the syntactic but also the semantic realization of valency at the external level of language, and this too is easier when specific material is at hand. The semantic realization of valency is determined primarily by the arguments

119

semantic roles (Agent, Patient, etc.). For example, only living beings can perform the Agent role, since only they act on their own initiative, whereas the Content role is filled only by abstract entities (what is thought, said, learned, and so on). Moreover, valency is realized through semantic compatibility between the verb and its partners. For example, the Agent partner of some verbs can be the name of any living being (e.g., people, livestock, wild animals, birds, fish, and parasites feed, etc.); for others, only animal names, and often not just any animals but only certain ones (e.g., a horse neighs; a wolf howls; cows, goats, and sheep ruminate; bees, hornets, and mosquitoes buzz; birds peck); and for still others, only human names (e.g., only people cry, eat, dream, and speak). Of course, metaphorically one can also say the lad neighs, the cows buzz, the pines laugh, the owls cry; however, when determining the valency of verbs, their metaphorical use should be left aside, since it gives a false picture of the verb’s semantic environment.

The realization of the Patient also depends on the meaning of the verb: for example, to hit, to touch can apply to all concrete objects, to eat, to feed on, to drink only to some, while to peel can apply only to bark or skin. Different realizations of the Patient may indicate different meanings of the same verb; cf. (5) and (12).

Constraints on semantic compatibility are particularly relevant when verbs with very similar meanings differ only in the semantic realization of one of their partners. For example, to wash can be used with hands, dishes, clothes, floors, and many other things; to wash with face and hands; to shampoo only with the head; and to launder only with clothes (cf. Lyberis 1975: 156).

In describing the semantic realization of verb valency, it is customary to indicate which semantic classes of words can occupy the verb’s partner positions. For example, the semantic realization of the syntactic valency model SₙVSₐ (S₁ / with S₁) for the verb to cut (see Figs. 5 and 6) can be described as follows: Sₙ — Hum, Sₐ — Anim, Sᵢ — Anim¹². In certain cases, however—for example, when the realization of a partner is very restricted (only one or a few lexemes are possible), or when synonymous verbs differ only in the semantic realization of one of their partners—more specific word meanings could also be given.

4. CONCLUDING REMARKS

A few of the most general issues related to the problem of verb valency have been touched on here. The aim was to indicate, in a few respects, how complex a phenomenon it is and how differently it is interpreted in linguistics. Without attempting to address all the problematic issues, we have focused on several fundamental questions: what is considered verb valency, what constitutes its essence, what assumptions underlie its existence and what conditions govern its realization, and what methods can be used to determine verb valency. The realization of verb valency at particular linguistic levels, especially the syntactic level, is highly complex, and therefore has only been touched upon here. After reviewing various researchers’ views on the matters mentioned, the author of this article sought to highlight those points that appear most important for the study of Lithuanian verb valency. The proposed approach to verb valency can be briefly summarized in the following statements:

1. Verb valency is understood as a property conditioned by its meaning, whereby it opens a certain number of vacant slots, which at a particular linguistic level are filled by obligatory or optional partners. Vacant slots are the combinatorial possibilities arising from the verb’s meaning, while verb partners are—

¹² The symbols used for this purpose are abbreviations of the corresponding Latin terms: Hum — human, Anim — living being, Abstr — abstract entity, and so on. These symbols are used by many researchers; cf. Helbig—Schenkel 1975, where, at the third stage of the valency model, the authors use them to characterize the semantic aspect of valency.

120

that is, certain forms of words with a defined meaning or constructions with corresponding meanings.

2. The basis of verb valency lies at the abstract logical-semantic level, while its realization occurs at specific language levels, semantic and syntactic. At the abstract level, the number of verb’s free slots and the semantic meanings of their potential fillers are determined; these meanings correspond to certain roles in the situation denoted by the verb (agent, patient, etc.). At the specific language levels, the appropriate fillers of the verb’s free slots—partners—are selected. At the syntactic level, the grammatical form of the partners and whether they are obligatory or optional are determined, while at the semantic level, their semantics are determined. Thus, valency is studied at all three levels, each of which brings one of its aspects into focus.

3. In practice, the starting point for determining verb valency is the meaning of the verb. It reflects a certain action situation with a defined number of participants performing particular roles. By analyzing the situation denoted by the verb, its logical-semantic valency structure is established and compared with specific syntactic constructions, thereby revealing how many and what kinds of partners the verb requires. Applying the elimination transformation to syntactic realizations of valency establishes which of the verb’s partners are obligatory and which are optional; the verb’s syntactic valency model is then constructed. It shows the number of the verb’s partners, their grammatical form, and whether they are obligatory or optional. Finally, a semantic characterization of valency is provided, that is, what semantic types of words can fill the positions specified in the valency model.

4. Verb valency is highly relevant to linguistic science, since the structure of verbal sentences depends on it. Grammatically correct sentences are constructed on the basis of verb valency models.

LITERATŪRA

Bondzio W. Das Wesen der Valenz und ihre Stellung im Rahmen der Satzstruktur. – „Wissenschaftliche Zeitschrift der Humboldt-Universität zu Berlin. Gesellschafts- und sprachwissenschaftliche Reihe“, 1969, Hf. 2, S. 233–240.

Bondzio W. Valenz, Bedeutung und Satzmodelle. – In: Beiträge zur Valenztheorie. Halle (Saale), 1971, S. 85–103.

Buchbinder W. A. Die Valenz und ihre Berücksichtigung bei der Auswahl lexikalischen Materials. – „Deutsch als Fremdsprache“, 1971, Hf. 5, S. 282–286.

Fillmore Ch. J. The Case for Case. – In: Universals in Linguistic Theory. New York, 1968, p. 1–88.

Fillmore Ch. J. Toward a Modern Theory of Case. – In: Modern Studies in English. Readings in Transformational Grammar. New Jersey, 1969a, p. 361–376.

Fillmore Ch. J. Types of Lexical Information. – In: Studies in Syntax and Semantics. Dordrecht-Holland, 1969b, p. 109–37.

Flämig W. Valenztheorie und Schulgrammatik. – In: Beiträge zur Valenztheorie. Halle (Saale), 1971, p. 105–21.

Geniušienė E. Lietuvių kalbos veiksmažodžių sintaksinė klasifikacija. – „Kalbotyra“, 1971, t. XXIII (1), p. 7–16.

Helbig G. Valenz und Tiefenstruktur. – „Deutsch als Fremdsprache“, 1969, Hf. 3, S. 159–169.

Helbig G. Theoretische und praktische Aspekte eines Valenzmodells. – In: Beiträge zur Valenztheorie. Halle (Saale), 1971a, S. 31–49.

Helbig G. Zu einigen Specialproblemen der Valenztheorie. – „Deutsch als Fremdsprache“, 1971b, Hf. 5, S. 269–82.

Helbig G., Schenkel W. Wörterbuch zur Valenz und Distribution deutscher Verben. Leipzig, 1969.

Helbig G., Schenkel W. Wörterbuch zur Valenz und Distribution deutscher Verben. Leipzig, 1973.

Helbig G., Schenkel W. Wörterbuch zur Valenz und Distribution deutscher Verben. Leipzig, 1975.

Lyberis A. Lietuvių kalbos sinonimų žodyno sudarymo klausimai. – „LTSR Mokslų akademijos darbai. A serija“, 1975, t. 4(53), p. 153–165.

121

Pocheptsov G. G. Constructional Analysis of Sentence Structure. — In: Actes Du Xe Congrès International des Linguistes. T. II. Bucarest, 1970, p. 865—9.

Sommerfeldt K. E. Zur Valenz des Adjektivs. — „Deutsch als Fremdsprache“, 1971, Hf. 2, S. 113—17.

Sommerfeldt K. E., Schreiber H. Untersuchungen zur syntaktischen und semantischen Valenz deutscher Adjektive (1). — „Deutsch als Fremdsprache“, 1971, Hf. 4, S. 227—231.

Sommerfeldt K. E., Schreiber H. Wörterbuch zur Valenz und Distribution deutscher Adjektive. Leipzig, 1974.

Sommerfeldt K. E., Schreiber H. Zu einem Wörterbuch der Valenz und Distribution der Substantive. — „Deutsch als Fremdsprache“, 1975, Hf. 2, S. 112—119.

Tesnière L. Comment construire une syntaxe. — In: Bulletin de la Faculté des lettres à Strasbourg, 1934.

Tesnière L. Esquisse d’une syntaxe structurale. Paris, 1953.

Tesnière L. Éléments de syntaxe structurale. Paris, 1959.

Абрамов Б. А. Синтаксические потенции глагола (в сопоставлении с потенциями других частей речи). — „Научные доклады высшей школы. Филологические науки“, 1966, т. 3, с. 35—44.

Адмони В. Г. Завершенность конструкции как явление синтаксической формы. — „Вопросы языкознания“, 1958, № 1, с. 111—117.

Адмони В. Г. Структурный каркас элементарного предложения в современных германских языках. — В кн.: Структурно-типологическое описание современных германских языков. М., 1966, с. 167—197.

Апресян Ю. Д. Экспериментальное исследование семантики русского глагола. М., 1967.

Апресян Ю. Д. Лексическая семантика. М., 1974.

Ахманова О. С. Словарь лингвистических терминов. М., 1966.

Бондарко М. М. О сочетаемости, дистрибуции, валентности в синтаксисе. — „Ученые записки МГПИЯ. Вопросы романо-германской филологии“, 1970, т. 55, с. 12—23.

Генюшене Е. Ш. Валентностные классы литовского глагола и система синтаксических структур. — В кн.: Синтагматика, парадигматика и их взаимодействие на уровне синтаксиса. Материалы конференции. Рига, 1970, с. 58—60.

Иванчикова Е. А. О структурной факультативности и структурной обязательности в синтаксисе. — „Вопросы языкознания“, 1965, № 5, с. 84—94.

Кацнельсон С. Д. О грамматической категории. — „Вестник ЛГУ“, 1948, № 2, с. 114—134.

Колшанский Г. В. Логика и структура языка. М., 1965.

Лейкина Б. М. Некоторые аспекты характеристики валентностей. — „Доклады на конференции по обработке информации, машинному переводу и автоматическому чтению текста“, 1961, вып. 5.

Почепцов Г. Г. О принципах синтагматической классификации глаголов. — „Научные доклады высшей школы. Филологические науки“, 1969, № 3, с. 65—76.

Степанова М. Д. Теория валентности и валентный анализ. М., 1973.

Фоменко Ю. В. Глаголы психического воздействия в русском языке. — В кн.: Вопросы теории русского языка. Новосибирск, 1975, с. 3—22.

Чеснокова Л. Д. О факультативности и обязательности компонентов синтаксической структуры. — В кн.: Вопросы синтаксиса русского языка. Ростов-на-Дону, 1973, с. 3—10.

Н. СЛИЖЕНЕ

К ВОПРОСУ О ГЛАГОЛЬНОЙ ВАЛЕНТНОСТИ

Резюме

На основе анализа трех существующих концепций валентности, а именно: логической, семантической и синтаксической, представителями которых являются соответственно В. Бонди, С. В. Фэлмэн, Ю. Д. Апресян и Г. Хельбиг, в статье делается вывод, что различия между ними состоят только в интерпретации валентности, а сущность этого явления ими понимается одинаково. Каждая концепция выдвигает вперед одну сторону валентности — логическую, семантическую или синтаксическую — и недооценивает другие. Поэтому автору показалось возможным и целесообразным эти концепции объединить, и глагольную валентность рассматривать на всех трех уровнях — абстрактном логико-семантическом и конкретных языковых семантическом и синтаксическом, так как все эти стороны валентности образуют диалектическое единство. Основа глагольной валентности находится на абстрактном, а ее реализация — на конкретных уровнях. На абстрактном логико-семантическом уровне с помощью предикатных структур устанавливается число свободных мест при глаголе и определяются семантические значения их потенциальных заполнителей соответственно определенным ролям (агента, пациента и т.д.) участников обозначаемой глаголом ситуации. На конкретном синтаксическом уровне определяется грамматическая

122

форма заполнителей свободных мест – партнеров глагола и устанавливается их обязательность или факультативность, а на семантическом – дается их семантическая характеристика.

Таким образом, под валентностью глагола автором понимается его способность на основании своего значения открыть определенное число свободных мест, которые на конкретном языковом уровне заполняются облигаторными или факультативными партнерами. Свободные места – это сочетательные потенции глагола, обусловленные его семантикой, а партнеры – словоформы или конструкции с определенным семантическим значением.

При практическом определении валентности глагола исходной точкой является глагольное значение. Глаголом обозначается ситуация действия, в которой участвует определенное число актантов, выполняющих определенные роли. Анализом ситуации устанавливается логико-семантическая структура валентности глагола, которая показывает число свободных мест при глаголе и их семантическое содержание. На основании того, что у грамматических падежей и предложных конструкций тоже имеется определенное семантическое содержание (см. Fillmore 1968), абстрактная логико-семантическая структура сравнивается с конкретными синтаксическими конструкциями, и таким образом определяется число и грамматическая форма требуемых глаголом партнеров. Применением к синтаксической конструкции трансформации элиминирования устанавливается их обязательность или факультативность. Тогда записывается синтаксическая модель валентности глагола. Так как у большинства глаголов имеется несколько значений, притом некоторые глаголы способны выполнять различные синдикационные функции, а синдикативные различия связаны с различиями синтаксического употребления, у глаголов бывает по нескольку моделей валентности. Выявить все модели валентности глагола возможно только при наличии достаточного материала, иллюстрирующего синтаксическое употребление этого глагола. Наконец, дается семантическая характеристика валентности, т.е. определяется, какие семантические классы слов (или отдельные слова) могут заполнить позиции в синтаксической модели.

N. SLIŽIENĖ

ZUR FRAGE DER VALENZ DES VERBS

Zusammenfassung

Auf Grund der Analyse der drei bestehenden Auffassungen der Valenz, nämlich der logischen, der semantischen und der syntaktischen, ist die Schlußfolgerung gezogen, daß sie sich nur in der Interpretierung der Valenz unterscheiden, während sie in ihrem Wesen gleich sind. Jede Auffassung hebt nur eine Seite der Valenz hervor, während die anderen nicht vollständig berücksichtigt werden. Die Autorin erachtet es für möglich und zweckmäßig alle drei Auffassungen zu vereinigen und die verbale Valenz auf drei Ebenen zu betrachten, nämlich auf der abstrakten logisch-semantischen Ebene und auf den konkreten sprachlichen Ebenen, der semantischen und der syntaktischen, weil alle erwähnten Aspekte der Valenz eine dialektische Einheit bilden. Die Grundlage der verbalen Valenz besteht auf abstrakten, ihre Realisation aber auf den konkreten Ebenen. Auf der abstrakten logisch-semantischen Ebene wird mit Hilfe der prädikativen Strukturen die Zahl der Leerstellen beim Verb festgestellt und die semantische Bedeutung der potentiellen Ergänzungen bestimmt in Abhängigkeit von der Rolle der Komponenten der verbalen Situation. Auf der konkreten syntaktischen Ebene wird die grammatische Form der Mitspieler des Verbs und ihr obligatorischer bzw. fakultativer Charakter bestimmt, auf der semantischen Ebene wird ihre Bedeutung charakterisiert.

Auf diese Weise versteht man unter der Valenz des Verbs seine Fähigkeit auf Grund seiner Bedeutung eine bestimmte Zahl von Leerstellen um sich herum zu eröffnen, die auf einer konkreten Sprachebene durch obligatorische oder fakultative Mitspieler zu besetzen sind.