Acta Linguistica Lithuanica

LVII (2007), 17 – 44

Alternations in the Expression of Lithuanian Arguments and Applicative Grammar as a Means of Explaining Them

Dedicated to the memory of Sebastian Shaumyan (1916–2007)

KRISTINA LENARTAITĖ

Institute of the Lithuanian Language

This article deals with the notion of alternation in the domain of argument realisation or diathesis. A working definition is offered, and a survey is given of the most characteristic instances of this phenomenon in Lithuanian. A number of important approaches to this subject, including those offered in the tradition of Lithuanian grammar, are discussed. Finally, a new approach to Lithuanian alternations and their position in the Lithuanian language system is proposed within the framework of Applicative Grammar.

1. INTRODUCTION

Different diatheses realized with the same verb have been studied by Lithuanian and foreign linguists for several decades. While in the work of the latter the phenomenon under discussion has a universally accepted name and more than one interpretation has been proposed for its various manifestations, in Lithuanian research cases of argument expression alternation are discussed only fragmentarily, alongside other problematic phenomena in syntax, syntactic semantics, or lexical semantics¹. This article is therefore devoted to reviewing notable instances of

¹ In the work of N. Sližienė, who studied the valency of Lithuanian verbs and mentioned cases of argument alternation most often, the phenomenon under discussion is frequently explained and named differently, depending on the needs of the study: in some places, the variable realization of the subject or direct object permitted by certain verbs is called subject alternation and object alternation (cf. Rožių kelmaitis užaugs didžiu medaičiu

17

Kristina Lenartaitė

cases of argument expression alternation in Lithuanian and some problems arising from the interpretation of this phenomenon: first, the aim is to define the content of the phenomenon and identify and discuss the more prominent types of argument expression alternation; drawing on Lithuanian research and foreign theoretical concepts, and using the tools of applicative grammar, the article introduces another possible account and explanation of the mechanism by which alternating constructions are expressed, and seeks to locate the phenomenon under study within the language system.

and Iš rožių kelmaičio užaugs didis medaitis; Motina užtiese stalą (su) baltutėlaite staltiesė and Motina užtiesė ant stalo baltutėlaitę staltiesę (Sližienė 1980: 95–101); elsewhere, in cases of object expression alternation, emphasis is placed on the same verb’s ability to perform different significative functions, or on a different significative interpretation of the same situation, which in turn determines a different syntactic use of the verb (Sližienė 1978: 117). Sentences arising from different semantic interpretations of the same situation, in which two very similar meanings have differently realized arguments, are called paraphrases (Sližienė 1994: 50–51). Cases of subject expression alternation are called conversive paraphrases, and their significative differences are said to be connected only with differences in the verb’s syntactic valency: that is, the participants in the situation are interpreted semantically in the same way in both cases, and the differing perspective on them is determined by the fact that the nominative case corresponds to one predicate argument in one sentence and to another in the other. The relation between such conversive paraphrases is said to be similar to that between active and passive constructions (cf. krūmai apaugo krantus ir krantai apaugo krūmais) (Sližienė 1994: 52). A. Holvoet explains variation in object expression as a grammaticalized alternative option for selecting a secondary topic in a layered actual division of the sentence (cf. padovanojo mokiniams knygų: apdovanojo mokinius knygomis; pripylė arbatos į puodukus: pripylė puodukus arbatos) (Holvoet 2003: 86–87). A. Drukteinis argues that in constructions such as šiaudus ant stogo – kloti šiaudais stogą, the change in the verb’s meaning causes the locative-expressed argument to shift to a position closer to the verb and its expression to change, thereby neutralizing the meaning of the case. The verb’s meaning, meanwhile, changes through the addition of a semantic component—a holistic seme indicating that the action concerns the whole object or that the object is completely affected by the action. With the addition of the holistic seme, a verbal lexeme becomes more general than a lexeme without the corresponding seme; consequently, privative opposition relations arise between verbal lexemes (together with their dependent components) in such constructions: the more general lexeme can replace the more specific one in the latter’s construction (also acquiring the corresponding meaning) (Drukteinis 1985: 56–58, 128–129). Another issue concerns the relation between alternating structures: one is said to convey the roles in the situation directly, while the other reflects a different interpretation (perspective) of the relations among the situation’s participants, chosen by the sentence’s author. Given the various changes to the predicate and the structure as a whole identified by comparing different interpretations of the same situation, the structures may be regarded as a basic one and a derived one (the semantic functions of the arguments in those structures are classified accordingly) (Drukteinis 2007: 27, 12–14).

18

Lietuvių kalbos argumentų raiškos alternacijos

2. THE CONCEPT, SCOPE, AND NAME OF THE PHENOMENON

The fact that the position of the phenomenon under discussion in the Lithuanian language system has not been thoroughly investigated is also evidenced by the fact that there is not only no suitable term of our own for naming this phenomenon, but even the international term encompassing various types of variation in the expression of arguments—alternations—has only very recently begun to be used in Lithuanian-language studies (see Drukteinis 2007: 13). Judging by the use of the term in the works of foreign linguists, it is clear that it is understood in different ways: in broader and narrower senses. Therefore, first the content of the phenomenon must be defined: its most characteristic features must be established, and on that basis the instances of variation in expression that can be attributed to it—or distinguished from it—must be identified.

2.1. Perhaps the most salient feature of alternating constructions in Lithuanian and other languages is their paraphrastic / periphrastic nature: from the standpoint of logic² (the truth of propositions), the constructions in the diathesis alternations under study can even be regarded as equivalent (cf.: if it is true that the starlings covered the rowan tree, then it is true that the rowan tree was covered with starlings, and if it is true that the rowan tree was covered with starlings, then it is true that the starlings covered the rowan tree, that is, the equivalence of two predicative structures is established: R(x, y) ~ R(y, x)). The additional nuances of meaning in the constructions are not always sufficiently salient either (in this regard, many researchers identify so-called locative alternations, in which the meaning of one construction often acquires an additional (holistic—the complete coverage of an action / effect) seme; moreover, pairs of constructions assigned to many alternations have the same lexical composition. The view of the paraphrastic / periphrastic nature of the constructions under study, and its assessment, also depend on which theoretical model is chosen to explain the phenomenon. The grammatical conception also determines the assessment of another, perhaps most significant, aspect of the cases under discussion: the role of the verb with which the relevant constructions are formed in the process by which alternating constructions arise, and the nature of its meaning (the same or two different meanings). This assessment, in turn, also affects the question of the constructions’ periphrastic nature.

² Logical meaning must not be confused here with grammatical (structural) meaning: the structural meaning of alternating constructions is always different. Compare sentences a. the starlings covered the rowan tree and b. the rowan tree was covered with starlings: the logical meaning of the predicate is a transitive predicate—the starting point of the action is the starlings, and its endpoint is the rowan tree. In logic, the predicate of sentence b is considered the converse of the transitive predicate of sentence a; however, the structural meanings of the predicates in the two sentences differ: a has a transitive predicate, while b has an intransitive one.

19

KRISTINA LENARTAITĖ

The approaches of different theoretical models to the relevant aspects must therefore be clarified. Alongside other phenomena of syntax and syntactic semantics, theories investigating alternations in argument expression can be divided into those that prioritize the analysis of structural changes and those that focus primarily on the meanings of constructions and changes in those meanings. Among the former, the studies of representatives of transformational and relational grammars, as well as the works of M. Baker, D. R. Dowty, and R. S. Jackendoff, are particularly noteworthy³. The findings of the latter studies, summarized, attribute the following properties to alternations in argument expression and other related phenomena: alternating verbs are said to have one and the same

³ In transformational grammar, the established relationship between active and passive constructions was also applied to explaining dative and locative alternations in English: an alternating verb is said to have only one argument projection in syntax, while the second realization of the arguments is derived from the first by means of a transformation. In relational grammar, the relationships both between active and passive alternations and between dative and locative alternations, as basic and derived forms, are defined through rules for changing grammatical functions as primitives (the rules cover changes in the grammatical functions of predicate arguments, but not of the predicate itself). The point of departure for M. Baker’s non-transformational model of analysis of the phenomena under study is the assumption that dative alternations and their counterparts in other languages are “thematic paraphrases,” that is, they have the same meaning. This property of alternating constructions is grounded in the Uniformity of Theta Assignment Hypothesis (Uniformity of Theta Assignment Hypothesis, UTAH): identical thematic relations among elements are represented by identical structural relations among those elements at the level of deep structure—thus, identical deep syntactic structures are assigned to both constructions. Morphosyntax also plays an important role here: in some languages, an applicative morpheme is attached to the verb in one of the alternating constructions (e.g., the -ir affix in Chichewa), allowing an additional argument, most often a goal or benefactive, to be realized as a direct object. The shortcomings of this theoretical model are as follows: it is unclear how thematic periphrasticity is assessed—the precise means by which the semantic relation between a verb and its arguments is determined to be identical in two sentences should be defined; even if the theta hypothesis can be applied to some cases of alternation (e.g., dative alternations), it cannot explain locative alternations, for which many researchers recognize differences in meaning. According to D. R. Dowty, in English the selection of subject and object depends on the features of the agent and patient proto-roles (Proto-role) that can be assigned to arguments: the argument with the greatest number of agent proto-role features is selected as the subject, while the one to which the greatest number of patient proto-role features can be assigned is selected as the object. If neither argument has enough features to be unambiguously assigned as subject or object, variation in the expression of arguments is permitted. R. S. Jackendoff argues that a single representation of lexical semantics permits two different syntactic realizations. According to him, alternations are possible in general because a particular position in the representation of lexical semantics is optionally marked with respect to the realization of an argument (Levin & Rappaport Hovav 2005: 196–205).

20

Lietuvių kalbos argumentų raiškos alternacijos

meaning in both cases of expression; alternations are said to be based on rules defining the emergence of two morphosyntactic realizations, and, moreover, one of the variants of expression is often claimed to be basic—thus the second is considered derived (transformed) from the first. These theoretical models tend to ignore the differences in meaning characteristic of alternating constructions: a construction derived by syntactic transformation is said not to change meaning, since the differences in meaning arising from the transformation are supposedly of a different nature: they concern not the verb’s lexical content, but aspects of sentence-level meaning, such as quantifier scope or focus assignment (cf. Levin & Rappaport Hovav 2005: 196–207). Thus, according to this position, the paraphastic / periphrastic nature of alternating constructions is acknowledged but not examined in greater depth, unlike in grammatical approaches that foreground changes in meaning. The most prominent studies in this latter group are by B. Levin and M. Rappaport Hovav, S. Pinker, A. Goldberg, and W. Croft; the studies by S. Iwata and M. Fried should also be mentioned⁴. These researchers seek and explain differences in the expression of arguments—

⁴ One of the first attempts to explain locative alternations in this way is the study by B. Levin and M. Rappaport (also classified among the so-called lexical rule (Lexical rule) models, as is S. Pinker’s analysis). Here, systematic variation in the meanings of alternating constructions is claimed to reflect different lexical-semantic representations; in that case, appropriately formulated linking rules could be applied to connect lexical-semantic representations with the correct patterns of argument realization. A corresponding lexical-semantic representation is proposed for the locative type of alternation (a. load: [X CAUSE [Y TO COME TO BE AT Z] / LOAD] b. load: [[X CAUSE [Z TO COME TO BE IN STATE]] BY MEANS OF [X CAUSE [Y TO COME TO BE AT Z]] / LOAD]), where the variant with the prepositional with construction (b) is an extension of the locative variant (a); in other words, the with variant is taken as the basis for the locative variant in such a way that the main constituent of the locative variant becomes a subordinated constituent of the with variant. Derivation of the constructions is possible only in this direction (from the locative construction, but not vice versa). All this partly explains the intuitively perceived relationship between the variants, which is close to a paraphrastic one. S. Pinker’s model is very similar to the one just discussed. In addition to locative alternations, it also examines dative and causative alternations. The relationship between the lexical-semantic representations and syntactic expression of the two constructions in a pair is likewise governed by linking rules (a verb whose semantic-structure core ‘X causes Y to move into / onto Z’ is turned (converted) into a new verb whose semantic-structure core is ‘X causes Z to change state by means of moving Y into / onto it’). Here, however, unlike in B. Levin and M. Rappaport’s model, derivation of the constructions is possible in both directions, depending on which construction variant has an argument that can serve as the sole complement of the predicate (although this test is not always reliable). In A. Goldberg’s study, a verb is understood as a combination of its minimal meaning and profiled arguments (i.e., the semantic roles of only those predicate participants that must be expressed) and is called a “root”

21

KRISTINA LENARTAITĖ

by relying on differences in the meanings of alternating constructions. Thus, the paraphastic / periphrastic aspect of constructions is also significant here: it is taken into account in defining both differences in the meanings of constructions and the relationship between them. The focus here is mainly on object alternations (the most frequently mentioned cases are dative and locative alternations), and efforts are made from this perspective to maintain a unified semantic basis for objecthood. Mechanisms are developed to explain alternating constructions that allow more than one argument to satisfy the semantic definition of objecthood alternatively.

Thus, with regard to the most immediately apparent feature of alternating constructions—their paraphastic / periphrastic nature—it can be said that if differences in meaning between constructions are recognized and a theoretical model is built specifically in this direction, this feature is used in the study as a basis for explaining differences in the meanings of constructions, the conditions for their emergence, and the relations between them. If, however, the analysis proceeds from structure and the relationship between constructions is explained in terms of a basic and a derived construction, paraphastic / periphrastic nature is invoked to demonstrate that one alternating construction is derived from another by transformation and to establish the deep structure common to both constructions. Whatever theoretical approach is chosen for studying alternations, paraphastic / periphrastic nature should be considered particularly

(root); this minimal meaning is then integrated into the corresponding argument-structure construction in such a way that the process involves fusing the verb’s arguments with positions in the construction. Argument alternations arise when a verb can be combined with more than one construction. S. Iwata notes many commonalities between his research and A. Goldberg’s position, assigning alternating locative constructions a single shared highest-level lexical meaning (Lexical Head Level Meaning), which encompasses different aspects of phrase-level meaning (Phrase Level Meaning) realized in the individual constructions of a pair. However, a fundamental difference between A. Goldberg’s and S. Iwata’s theories concerns the role of the verb: in A. Goldberg’s model, the verb does not determine sentence structure, since the relevant aspects of meaning are assigned not to the verb itself. In contrast, in S. Iwata’s study the verb’s meaning occupies a central position—it is understood as a film-like scene, and the integration of the verb into a construction is based on semantic compatibility. W. Croft maintains that in many argument alternations, the constructions involve the same verb root (a combination of the verb’s minimal meaning and profiled arguments) and a single event type, but two different verb “profiles” (Levin & Rappaport Hovav 2005: 205–216; Iwata 2005: 355–403). M. Fried explains subject alternations as subject selection based on a point of view (viewpoint), dependent on the speaker’s perception and conceptualization of the event. Regular differences in morphosyntactic encoding are explained as being determined by different configurations of the corresponding semantic roles in particular scene types (Fried 2005: 475–512).

22

Lietuvių kalbos argumentų raiškos alternacijos

an important feature of this phenomenon. Moreover, the significance of this feature in investigating the pragmatic aspects of alternations (e.g., thematic-rhematic structure) is unquestionable; however, this article adopts the position that research should begin by identifying the mechanisms of expression fixed in the language system, although a comprehensive model for studying Lithuanian argument alternations will eventually need to encompass both the conditions underlying the emergence of the phenomenon under analysis and the functions it performs in the language system.

Another important question is the role of the verb in this process. Its assessment also depends on the theoretical model: some recognize the decisive role of the verb in the emergence and interpretation of alternations (almost all studies focused on differences in constructional meaning; see footnote 4); elsewhere its role is narrowed (cf. A. Goldberg’s position in footnote 4) or not discussed in greater depth (cf. footnote 3). Assessments of the meaning of the verb in alternating constructions are similarly divided: some claim it is the same in both constructions in a pair, while others posit two distinct meanings, one related to the other as a basic form to a derived one. The approach to analyzing the phenomenon discussed in this article is largely an application of S. Shaumyan’s model of Universal Applicative Grammar (Universal Applicative Grammar). Before presenting the assessment of the verb’s role in alternating constructions in applicative grammar, it should be noted that applicative grammar makes a strict distinction between two levels of linguistic phenomena especially important: genotype (genotype) and phenotype (phenotype) grammars. The genotype level comprises functional units—predicates, terms (predicate satellites—arguments and circumstantial terms), and modifiers—and the abstract operator–operand relations that connect these units. Phenotype grammar encompasses syntagmatic units—morphemes and words—and their relations with respect to linear order and morphological properties. The rules of genotype grammar should be regarded as invariant with respect to their various possible realizations in phenotype grammar. The rules of genotype grammar are intended to explain universal linguistic phenomena (Shaumyan 1987: 97). Argument expression alternation is attested in many languages; moreover, its main types (e.g., subject expression alternations or locative alternations) have analogous counterparts across different languages. Therefore, in order to properly identify the linguistic expression processes associated with this phenomenon and avoid getting lost in the abundance of phenotype-level variation realized across different languages, argument expression alternation should first be defined by rules at the genotype level

23

KRISTINA LENARTAITĖ

and only then should we investigate how they relate to the phenotype phenomena of a particular language. The following are the principles of applicative grammar relevant to the question at hand: 1) a change in the grammatical status of terms (predicate satellites) should be regarded as a consequence of a change in the predicate; changes to the predicate are determined by predicate-changing rules⁵, which must be applied to the predicate before it is applied to the terms; 2) a sentence with altered grammatical relations may be realized independently of a sentence with unchanged grammatical relations; for example, a passive sentence can be realized by directly applying a passive predicate to its terms, rather than by transforming an active sentence into a passive one (Shaumyan 1987: 272).

The second claim means that passive constructions, in relation to active ones, and other similar phenomena are not considered derived here in the sense used in transformational or relational grammars (see footnote 3). However, if predicate P₁ representing a sentence yields predicate P₂ when the relevant predicate-changing rules are applied, then a sentence with the changed predicate (P₂) is considered derived from the sentence realized with the unchanged predicate (P₁) only insofar as the structure of the non-derived sentence is represented by predicate P₁ and transferred to the structure represented by the derived predicate (P₂).

The necessity of changes to predicate structure in diathesis alternations can be illustrated by the following cases, which should be distinguished from alternations; cf.:

(1) a. Ji atidalijo mėsą nuo kaulų. b. Ji atidalijo kaulus nuo mėsos. (2) a. Jie atplėšė vaiką nuo motinos. b. Jie atplėšė motiną nuo vaiko.

Here the predicate structure does not change—the predicate opens the same positions for arguments and circumstantial terms in the sentences of each pair.

2.2. Another question important for defining the concept of argument alternations is the lexical composition of pairs of constructions. This phenomenon is usually understood to involve cases with identical lexical composition; however, B. Levin’s notion of diathesis alternations can be equated with a more broadly conceived set of possibilities for varying argument expression⁶, thereby seeking to distinguish from verb-

⁵ Predicate-changing rules are discussed in section 3.2.

⁶ Cf. B. Levin, who, classifying English verbs on the basis of the relationship between their meanings and syntactic behavior, includes among syntactic, or diathesis, alternations all

24

Lietuvių kalbos argumentų raiškos alternacijos

to identify differences in a word’s meaning from differences in its syntactic behavior. Most English examples classified as diathesis alternations would not even give rise to such assumptions in Lithuanian, since they are realized by different verbal lexemes (or different forms of the same lexeme), cf.: a. the butcher cuts the meat and b. the meat is easily sliced; a. the child broke the cup and b. the cup broke; a. that dog bites people and b. that dog bites. What is called a total transformation alternation does not change the structure of the sentence core—that is, the arguments necessary to realize the predicate’s semantics—but merely introduces an additional element into the sentence periphery⁷, cf.: a. the witch turned him into a frog and b. the witch turned him from a prince into a frog. Any element realized in the periphery should be regarded as optional, and if it is not promoted to an argument position in the predicate’s core under a change in expression, its realization or non-realization does not change the predicate’s core and therefore cannot be classified as an alternation. This article adopts the position that argument expression alternations should first and foremost be defined as regular patterns of predicate argument expression established universally across language systems (that is, as phenomena at the level of genotypic grammar), whereas any variation in the expression of predicate adjuncts (circumstantial terms) belonging to the periphery, which does not affect its arguments, is a phenomenon of phenotypic grammar (the morphosyntactic level of a particular language). On this basis, only cases of variation in the expression of predicate arguments (that is, those realized as subject, object, and indirect object) are considered alternations. However, some aspects of the issue of alternations need to be clarified: 1) it is not entirely accurate to call the phenomenon an alternation in argument expression, since changes in expression involve not only the predicate’s arguments, realized—

by verbs’ permitted cases of variation in argument expression; therefore, constructions with different lexical compositions are also included here, cf.: Middle Alternation (Middle Alternation): a. The butcher cuts the meat; b. The meat cuts easily; Causative/Inchoative Alternation (Causative/Inchoative Alternation): a. Janet broke the cup; b. The cup broke; Unspecified Object Alternation (Unspecified Object Alternation): a. Mike ate the cake; b. Mike ate; Characteristic Property of Agent Alternation (Characteristic Property of Agent Alternation): a. That dog bites people; b. That dog bites; Total Transformation Alternation (transitive) (Total Transformation Alternation (transitive)): a. The witch turned him into a frog; b. The witch turned him from a prince into a frog; and so on (Levin 1993: 25–26, 27–30, 33, 39, 57–58).

⁷ For discussion of the terms core, the arguments necessary to represent predicate semantics, and periphery, elements (non-arguments), see also Van Valin, LaPolla (1997: 25–31).

25

KRISTINA LENARTAITĖ

realized by the grammatical functions of subject, object, and indirect object, but also non-arguments—adverbial terms, which the predicate optionally permits to be realized; 2) the essential criterion for assigning a case to the diathesis alternations under investigation is that the expression alternation must affect at least one change in a predicate-argument position: that is, either the expression of the arguments themselves must change (e.g., a predicate argument functioning as a grammatical object may also be realized as a subject: fleas infested the dog and the dog was infested with fleas), or an adverbial term must be promoted from the periphery to an argument position (cf. he scatters sand on the path and he scatters the path with sand); the latter example also involves the demotion of an argument to the position of an adverbial term (cf. sand – with sand). More precisely, only sentence pairs that realize differences in the expression of the grammatical functions of subject and object belong to the diathesis alternation phenomenon under investigation—in sentences with so-called indirect objects, only their promotion to the direct-object position is relevant; this is precisely what gives rise to another type of alternation, cf. Dad puts shoes on the child and Dad shoes the child. It is precisely because alternations necessarily involve a change in the expression of the subject or object that the term “argument expression alternation” is not considered particularly apt.

B. Levin’s distinction of the unspecified object alternation encourages us to consider analogous cases in Lithuanian. Often, one construction in a pair is explained as having an unrealized argument, even though the verb’s meaning opens such a position⁸, cf. a. the child eats an apple and b. the child eats (well). However, a verb meaning that denotes an activity type or ability does not at all require complementation by an argument functioning grammatically as an object; and if an object were also realized with this meaning, it would no longer be merely an indication of ability or activity type, cf. the child already eats apples (by themself) or children (often) dream of their mother. Whatever the assessment of the second sentence in pair (b)—whether as transitive with an unrealized object function or as intransitive—this and analogous cases must be distinguished from the diathesis alternations under investigation: if the predicate is transitive in both cases, then the object is always impli-

⁸ A transitive verb, by virtue of its meaning, always opens a syntactic position for a complement, although that position need not be filled (Holvoet, Judžentis 2005: 14–15). Cf. also how direct objects of many verbs are omitted when the aim is only to describe a type of activity in general (He is writing, reading now, etc.) or an ability (The child can already read and write) (Holvoet, Čižik-Prokaševa 2005: 70).

26

Lietuvių kalbos argumentų raiškos alternacijos

implied, and if one construction is transitive and the other intransitive, then at most the question of the relationship between the constructions needs to be addressed.

Alongside the case just discussed, other similar instances of dual argument expression with different lexical compositions should also be reviewed, cf.:

(3) a. The hairdresser cut the girl’s hair.

b. The hairdresser cut the girl.

(4) a. The barber shaved the client’s beard.

b. The barber shaved the client.

(5) a. The child tied their shoelaces.

b. The child put on their shoes.

(6) a. The man buttoned his shirt buttons.

b. The man buttoned his shirt.

(7) a. Mother wiped the crumbs off the table.

b. Mother wiped the table (? of crumbs).

(8) a. We sweep the leaves off the paths.

b. We sweep the paths (? of leaves).

Clearly, in sentences 3a, 4a, 5a, and 6a, the direct object and the element realized in the dative case are related by part–whole or possessive relations; moreover, the dative-marked element can also be understood here as a modifier of the object, cf. the company sews clothes for women / women’s clothes / women’s garments. In the pairs expressing a part–whole relation, the modifiers in the first sentences (3a, 4a, 5a, 6a) metonymically represent the whole phrase in the second sentences (1b, 2b, 3b, 4b); therefore, the differences in expression here are not caused by a change in the predicate, cf.: Jablonskis’s writings – Jablonskis, and I read Jablonskis’s writings / Jablonskis. It is precisely because the predicate structure does not change that these cases are not classified as argument expression alternations.

The expression in the second (b) sentences of pairs 7 and 8 is questionable, although it is also attested in the corpus <http://donelaitis.vdu.lt>. Without deciding here whether their use is correct or incorrect, for now these cases (only those with overt oblique terms in the second sentences, 7b and 8b) are not distinguished from the other cases under investigation, although it may well be that these constructions realize different meanings of the same verb, one of which (b) no longer opens a position for an oblique term, cf.: (nu)šluostyti ‘to remove something by wiping (something) off (something)’ and ‘to clean (make clean, dry) something by wiping (it) with (something)’.

27

KRISTINA LENARTAITĖ

Thus, for sentence pairs that aspire to be called diathesis alternations, identical lexical composition is a necessary condition: if a predicate cannot open a position for some argument / oblique term, the predicate structures may be considered too distant from one another, in which case one can at most examine their relationship and the circumstances of their emergence.

2.3. Thus, the features can finally be stated that cases involving different diatheses permitted by the same verbal lexeme must all (not just one or some) exhibit in order to be called argument expression, or diathesis, alternations: 1) identical lexical composition of the sentences is obligatory; 2) changes in the predicate structure of the alternating constructions are necessary; 3) changes in expression must necessarily affect at least one of the grammatical functions realized by the predicate’s arguments—the subject and/or the direct object (the grammatical function of the indirect object plays no decisive role in this phenomenon); 4) the relationship between alternating sentences can (at a certain level of abstraction) be understood as para-/periphrastic and even as logically equivalent.

Given the listed features of alternations, it is now possible to discuss the name of the phenomenon as well. In foreign-language literature, the term alternation has become established in both a narrower and a broader sense (B. Levin, see footnote 6). Moreover, it is not used exclusively to refer to diathesis phenomena (it also denotes vowel alternation and variation in morphemes); therefore, using the term without context necessarily requires clarification by modifiers. Since the phenomenon itself is universal and should be viewed in the context of analogous phenomena in different languages, the now-established term should not be abandoned. For argument expression alternation used in the broader sense, one could also propose the Lithuanian equivalent argument expression variation, which, precisely because of its lack of specificity, is unsuitable as a name for the research object used in the narrower sense, constrained by strict requirements and denoting regular phenomena in the language system. Verbs that permit alternations are sometimes also called shift verbs (cf. English shift), e.g., dative shift verbs (cf. Van Valin 2005: 114–115). This focuses attention on a change in the position of one of the members of alternating constructions, and thus is hardly suitable as a name for the phenomenon as a whole, since its plural form would likewise be overly broad and vague. In trying to capture and describe in the most general terms what happens in alternating constructions, the first feature that stands out is the exchange, redistribution, or change in the positions / placement of the predicate’s dependents (arguments and oblique terms). Thus, for the phenomenon under investigation—

28

Lietuvių kalbos argumentų raiškos alternacijos

The Lithuanian term proposed for the phenomenon is ARGUMENT EXPRESSION ALTERNATION. Perhaps this term is not sufficiently precise either, and also requires clarification by a modifier; it could be applied not only to this phenomenon, but it seems to describe the essence of the phenomenon fairly accurately. Moreover, unlike variation, ALTERNATION allows us to infer that the process involves a limited set of cases. This term implies no evaluation of the phenomenon under investigation and should therefore be suitable for all grammatical models.

2.4. As mentioned, there are different types of argument expression alternation, which it seems most reasonable to classify according to whether they concern the expression of the grammatical functions realized by arguments—direct object and/or subject.

1. Subject and object expression alternation:

(9) a. I was covered in burdocks. b. I became covered with burdocks.

2. Subject expression alternation:

(10) a. The yard smells of lilacs. b. The yard smells of lilac blossoms.

3. Object expression alternation and its subtypes:

a) alternation in the expression of both the object and the circumstantial term denoting the location of the action⁹:

(11) a. The child spreads butter (some butter) on the bread. b. The child spreads the bread with butter.

b) alternation in the expression of both the object and the recipient of the action / effect¹⁰:

(12) a. Mother (dresses / puts on / pulls on) a shirt on the girl. b. Mother (dresses / puts on / pulls on) the little girl in a shirt.

c) alternation between the location of the action / effect and the object:

⁹ Here, the content of the various semantic functions performed by circumstantial terms, such as locative, directive, or goal (goal), is summarized.

¹⁰ Here, the so-called grammatical function of indirect object is set aside, and the content of the action / effect recipient (recipient), possessor (possessor), and addressee (target person) is generalized under the circumstantial term.

29

KRISTINA LENARTAITĖ

(13) a. The boys ran around the lake. b. The boys ran the lake.

(14) a. The child kicked the ball. b. The child kicked the ball.

The most salient types and subtypes of argument-marking alternations have been mentioned here, so it is possible that one or another case has gone unnoticed and been left out. Moreover, the types of alternations are not closed sets—the marking of arguments in constructions of different types may alternate, cf.:

(15) a. The garden smells of lilacs. b. The garden's lilacs smell fragrant. c. Lilacs smell fragrant in the garden.

The alternation between (15a) and (15b) involves changes in the marking of two predicate dependents, whereas the alternations between (15a) and (15c), and between (15b) and (15c), involve only one; accordingly, the alternation between (15a) and (15b) belongs to the subject-and-object marking type, while those between (15a) and (15c), and between (15b) and (15c), belong to the subject-marking alternation type. In sentences (15a) (=10a) and (9a), the subjecthood of lilacs and ragweeds may be questioned because they fail to satisfy one semantic property characteristic of subjects: in unmarked information-structure sentences, which is what the examples under analysis are, lilacs and ragweeds belong to the rheme of the sentence, while in the garden and in me are the topics in both sentences. Unmarked topic status in a sentence is one of the prototypical properties of the grammatical function of subject (see Holvoet, Semėnienė 2005: 49–50). Lilacs and ragweeds could become topics only if moved to the beginning of the sentence and emphasized with contrastive stress—in that case they would be marked information-structure sentences. Here, arguments expressed in the nominative are still considered subjects, even though they do not meet all the properties of a prototypical subject. In sentence (15c), the noun in the garden should be recognized as the subject, although it is realized in the locative—the locative in the garden is used here in a secondary, subject, function¹¹. The subject-and-object marking type also includes examples that leave no doubt about the subjecthood of the arguments so realized (16), whereas examples of the subject-marking alternation type (17) seem always to involve this problem, cf.:

¹¹ Moreover, applicative grammar, like many other theoretical models, recognizes that a subject is necessary in a sentence: “if a word combination is a sentence, then it has a subject” (Shaumyan 2006: 229). Impersonal sentences and other sentences with an unexpressed subject have what is called a null-subject form (dummy or zero terms).

30

Lietuvių kalbos argumentų raiškos alternacijos

(16) a. Flies swarmed around the honey.

b. Honey swarmed with flies.

(17) a. A cool breeze blew from the lake.

b. The lake blew with coolness.

The validity of distinguishing the listed types of alternation in argument marking, as well as other criteria essential to this phenomenon, should be tested by examining the mechanism by which constructions are marked.

3. ARGUMENT EXPRESSION ALTERNATION IN APPLICATIVE GRAMMAR

Argument expression alternation has been explained with reference both to differences in sentence semantics (cf. footnote 4) and to changes in their structure (cf. footnote 3). Lithuanian linguists’ fragmentary observations on the phenomenon under investigation foreground semantic differences (cf. footnote 1). All the studies provide valuable data. Studies directed toward identifying semantic differences between alternating constructions must also take extralinguistic factors into account, for example, the examination of situations that led to a different assessment of the positions of their participants. Most of the solutions proposed in this regard (profiling, differences in the coding of information structure, etc.) appear correct and applicable to the corresponding needs, which can be precisely determined only in a specific situation. However, it is clear that grammatical means for expressing corresponding needs exist in the language system (or, given the universality of alternation, in language systems); thus, the analysis of Lithuanian argument alternations is to begin precisely from this point. To investigate the phenomenon comprehensively, the next step should be to connect formal means and processes of expression with phenomena in the semantic domain. It is hoped that a properly explained process of expression will help distinguish essential from secondary factors in the semantic domain.

To achieve these aims, the method and tools of Universal Applicative Grammar, also known as the Semiotic Theory of Language, were chosen. The main architect of the model was S. Shaumyan. An important feature of this grammar is the distinction between phenomena at the genotypic and phenotypic levels (see Section 2.1). Since argument alternation operates universally, the principles governing this phenomenon are first to be defined at the level of the functional units of genotypic grammar (predicates, terms, and

31

Kristina Lenartaitė

modifiers) and the operator–operand¹² relations that connect them.

3.1. Thus, the basic principles of applicative grammar, which are applied to explain the phenomenon under investigation alongside other phenomena, must now be defined. One of the fundamental laws of the language system here is the principle of functional superposition, which creates a hierarchy of primary and secondary functions of signs and phonemes. This principle is widely applied to explain phenomena at the phenotypic level (phonology, morphology, and syntax).

“In the context of a change in meaning or function, a sign or phoneme A is applied to a sign or phoneme B by means of superposition, such that B is used as the container and A as the double-layer <B SUP A>. As a result of superposition, a functional derivation relation is created, whereby B becomes a functional derivative of A, or an alternant. The sign or phoneme B acquires a secondary meaning or function that is ‘superimposed’ on its primary, context-independent meaning or function” (Shaumyan 2006: 131). (Here and elsewhere, translated by me – K. L.)

To identify universal grammatical processes, an idealized law of grammatical autonomy with respect to the lexicon is formulated, represented by two processes of abstraction: 1) sentence structure is abstracted from the words that fill it; 2) sentence structure is abstracted from its linear representation. What remains, then, are the essential, universal elements—syntactic functions. Four basic functions are distinguished in universal syntactic structure: predicates, terms (predicate arguments and circumstantial terms), term modifiers, and predicate modifiers. At the phenotypic level, these functions are assigned as primary functions to four word classes, respectively: verbs—to the predicate function, nouns—to the term function, adjectives—to the term-modifier function, and adverbs—to the predicate-modifier function. Each word class, in addition to its primary syntactic functions,

¹² These relations are determined by the law of word-combination formation (a word combination is understood here in a manner similar to a gestalt—as a word together with its structural meaning): “If the meaning of a word combination is insufficient and must be supplemented by the meanings of other word combinations, that word combination is called an operator and combines with one or more word combinations, called operands, in order to form a new word combination, called a resultant” (Shaumyan 2006: 213).

32

Lietuvių kalbos argumentų raiškos alternacijos

may also have secondary ones. Functional superposition is a process in which a secondary function is “superimposed” on a primary function, thereby creating a new, two-layered function. The new function is a combination of the primary and secondary functions. After superposition, a word (the “superposed” one) displays the properties of both functions, primary and secondary. For example, the primary function of the word baltas is to be a term modifier, as in the sentence Baltas šunelis vizgina uodegą. But in the sentence Baltas yra geras, o juodas blogas it performs a term function. Thus, in the latter case the adjective undergoes noun superposition, and the adjective’s syntactic function merges with the noun’s syntactic function. In the sentence Vakar aš vaikščiojau sode ir staiga matau – Jonas, the structural meaning of the word matau is understood as past tense. Thus, here the present-tense function is combined with the past-tense function. In the sentence Jonas yra vairuotojas, the word vairuotojas in the phrase yra vairuotojas has a secondary, predicative function, which was “superimposed” on its primary (term) function through superposition when the word combined with the operator yra. These are examples of syntactic superposition.

In morphology, the superposition mechanism operates in the derivational process: only base words have primary syntactic functions, while derived words have only secondary functions, derived through superposition from the functions of base words; for example, the word žalias is underived and has the primary function of term modifier, whereas žalumas is derived from it by adding the corresponding outer noun marker¹³ (-um-as) and “superimposing” the term function on the base word’s primary—term-modifier—function. This yields the derivative’s secondary double function: term modifier and term.

The superposition mechanism also operates at the level of word-meaning derivation; compare the primary meaning of the word žalias, “a color of this kind”: by “superimposing” on this meaning the primary meaning of the word nepatyręs, “unskilled, unaccustomed,” the secondary meaning of žalias is derived: “inexperienced, unskilled; young.” This is a metaphorical way of transferring meaning: the secondary meaning of žalias and the primary meaning of nepatyręs link the words by synonymy, while the two meanings of žalias are linked by polysemy.

3.2. S. Shaumyan also explains the phenomenon of argument alternation under investigation

¹³ Syntactic function superposition may have no external marker, compare: nematau akmens ir akmens sieną.

33

KRISTINA LENARTAITĖ

explains it through the operation of superposition and places it alongside other syntactic phenomena based on transformation¹⁴ (e.g., passivization). Before defining the rules of superposition that govern this phenomenon, the hierarchy for applying syntactic terms to a predicate, or, in other words, the applicative hierarchy (Applicative Hierarchy), should be discussed¹⁵. Here, as in relational grammar, syntactic, or simply grammatical, functions (cf. Holvoet, Mikulskas 2005: 7) are conventionally designated as follows: 1—subject, 2—direct object, 3—indirect object, Q—circumstantial term (in applicative grammar, unlike relational grammar, the subject and direct object are not regarded as elements marking grammatical relations, but as members of the binary relation created by a two-place predicate). Syntactic terms are arranged hierarchically as follows:

[1 > 2 > 3] > Q

¹⁴ In applicative grammar, passive constructions are regarded as active conversions, but their emergence and their relation to one another are explained differently than in transformational or relational grammars: the isomorphism of the basic components of active constructions and their corresponding passive constructions (isomorphism should not be confused with equivalence) posited by S. Shaumyan means that the relation between the predicate and subject in active constructions is isomorphic to the relation between the predicate and circumstantial term in passive constructions (the latter corresponding to the subject of the active construction). The predicate is the principal element representing the sentence—the minimal sentence is a predicate. Thus, in applicative grammar, it is not the relation between constructions (active and passive: R(x, y) and R'(y, x)) that is termed a conversion; rather, applying an abstract conversion operator to the predicate of an active sentence (the base predicate) derives a conversion predicate, from which the passive predicate is then constructed (that is, the conversion predicate is derived from the base predicate, but not vice versa; the passive predicate is not a conversion predicate, but is merely derived from the conversion predicate, which, in turn, is derived from the base predicate). Therefore, if one can speak of transformation or conversion between active and passive constructions at all, it is only at the level of the structures of their predicates, since passive constructions are considered to be generated independently of the corresponding active constructions; passive constructions can only be reduced to their corresponding active constructions (Desclés, Guentchéva, Shaumyan 1985: 22–24; Shaumyan 2006: 234–235; see also the remarks in section 2.1).

¹⁵ This hierarchy follows from the laws of sentence formation. “A sentence consists of a hierarchy of binary combinations of its constituents”—that is: 1) “if X is an operator that combines with term Y to form sentence Z, then X is an intransitive predicate and Y is the subject; 2) if X is an operator that combines with term Y to form intransitive predicate Z, then X is a transitive predicate and Y is the direct object; 3) if X is an operator that combines with term Y to form transitive predicate Z, then X is a ditransitive predicate and Y is the indirect object; 4) if X is an operator that combines with phrase Y to form a phrase Z of the same type as Y, then X is an attribute of Y” (Shaumyan 2006: 228–232).

34

Lietuvių kalbos argumentų raiškos alternacijos

Brackets distinguish predicate terms, the members of the predication frame. 3 is marked with respect to 2 here, because rank 2 (the set of functional positions in the sentence) is broader: 2 is realized in both transitive and ditransitive sentences, whereas 3 is realized only in ditransitive sentences; however, 2 is marked with respect to 1, because 1 is realized in intransitive, transitive, and ditransitive sentences; and Q is marked with respect to the predication frame, whose rank is broader: the predication frame is realized in all sentences, whereas Q is not.

Thus, diathesis—the transformation of sentence A into sentence B—is determined by the following rules:

1. If 3 is identified with 2, then 2 must undergo Q superposition:

<3 SUP 2> → <2 SUP Q>

2. If 2 is identified with 1, then 1 must undergo Q superposition:

<2 SUP 1> → <1 SUP Q>

3. If 3 is identified with 1, then 1 must undergo Q superposition:

<3 SUP 1> → <1 SUP Q>

In short, if a term A in a sentence is realized as a higher-order term B, then term B will be realized as an adverbial term (Shaumyan 2006: 235–236).

The first rule covers one case of alternation in the expression of the object—alternation in the expression of both the object and the recipient of the action/effect (cf. 12ab), cf.:

(18) a. Mother wraps a scarf around the child. b. Mother wraps the child in a scarf.

In sentence (18a), the predicate arguments perform direct functions: mother – 1, scarf – 2, child – 3, whereas in sentence (18b), as a result of the superposition undergone, the arguments acquire secondary, two-layered functions: 3 undergoes 2 superposition, 2 undergoes Q, and 1 undergoes 1 superposition (i.e., zero superposition¹⁶), cf.:

(18) b. mother scarf child <1 SUP 1> <2 SUP 3> <3 SUP 2>

Other cases of alternation in the expression of the object include an adverbial term that, in the broadest sense, performs the semantic function of the location of the action/effect,

¹⁶ A transformation need not change the function of every sentence term. A situation in which a term’s function remains the same, as with the subject mother in sentences 18a and 18b, is called zero superposition.

35

KRISTINA LENARTAITĖ

a change in expression (see p. 8), therefore S. Shaumyan’s rules should be supplemented with one more:

If Q is identified with 2, then 2 must undergo Q superposition:

<Q SUP 2> → <2 SUP Q>

(19) a. The girls tuck flowers into their braids. b. The girls adorn their braids with flowers.

In sentence (19a), the primary functions of the predicate arguments and adjunct terms are as follows: the girls – 1, flowers – 2, braids – Q, whereas in sentence 19b, the following secondary functions result after superposition; cf.:

(19) b. the girls flowers braids <1 SUP 1> <2 SUP Q> <Q SUP 2>

The second rule established by S. Shaumyan is applicable not only to the subject–object expression alternation, but also to explaining the process of passivization. This attests to the functional commonality of these phenomena and indicates the place of argument alternation in the language system.

(20) a. The piglets covered the sow. b. The sow was covered with piglets. c. The sow was covered by piglets / covered in piglets.

Only sentence (20a) realizes the primary argument functions: piglets – 1, and sow – 2. In sentence (20b), the sow acquires the secondary function <2 SUP 1> and is therefore promoted to subject position, while piglets receive the secondary function <1 SUP Q> and are demoted to adjunct-term position. The passive counterpart of (20a) is (20c), which arises through the operation of the same superposition rule; cf.:

(20) c. piglets sow <1 SUP Q> <2 SUP 1>

A separate rule must also be defined for the subject-expression alternation type:

If Q is identified with 1, then 1 must undergo Q superposition:

<Q SUP 1> → <1 SUP Q>

(21) a. Mushrooms have sprung up abundantly in the forest.

36

Lietuvių kalbos argumentų raiškos alternacijos

b. The forest sprouted mushrooms.

(22) a. The kitchen reeked of garlic. b. The kitchen reeked of garlic. c. There was a smell of garlic in the kitchen.

Sentences of the type in (21b) and (22b) are precisely to be explained by the operation of this rule. It is important to note that in sentence (21a) the expression of the subject (mushrooms) is not typical, but it can likewise be explained by superposition: the verb sprouted is derived from grew; thus, relative to the base word, sprouted can have only a secondary function, which is “superimposed” by superposition on the primary function of grew. As a result, the argument required by the predicate sprouted is realized not in the nominative but in the genitive case—and this is a formally expressed marker of the argument’s secondary function: that is, the noun mushrooms, realized in the genitive, has a secondary, subject function here; cf. mushrooms grow in the forest and mushrooms sprouted in the forest. In sentence (21b), the grammatical function of the subject of sentence (21a) (mushrooms) has changed: by superposition, the grammatical function of object was “superimposed” on it, although this is not expressed by any overt markers.

Sentence (22c) is derived from sentence (22b) by superposition, with a secondary adverbial-term (Q) function “superimposed” on the terms (the function of the adverbial term garlic remains unchanged—that is, it undergoes zero superposition):

(22) c. kitchen garlic << Q SUP 1 > SUP Q > << 1 SUP Q > SUP Q >

The third rule of the operation of superposition specified by S. Shaumyan applies to the transformation of English dative alternations into passive constructions¹⁷. In Lithuanian, this rule could perhaps be applied only to cases such as: a. the hostess gave the puppy some food and b. the puppy is given some food by the hostess¹⁸. Here, in the second (b) sentence, the secondary function of the term puppy is <3 SUP 1>, while that of the term hostess is <1 SUP Q>; the function of the term food remains unchanged, that is, it

¹⁷ Cf.: a. John handed the scarf to Mary; b. John handed Mary the scarf; c. Mary was handed the scarf by John. In the first (a) sentence, the arguments perform their primary functions. John – 1, scarf – 2, Mary – 3; the second (b) sentence is transformed according to the first rule—<1 SUP 1> <2 SUP Q> <3 SUP 2>—and the third (c) is transformed from the second (b) according to the third rule established by S. Shaumyan: <<1 SUP 1> SUP Q> <<2 SUP Q> SUP Q> <<3 SUP 2> SUP 1> (Shaumyan 2006: 235–236).

¹⁸ Plg. Šunelis šiandien tebėra nieko neduotas (Jablonskis 1997: 140).

37

KRISTINA LENARTAITĖ

undergoes zero superposition—<2 SUP 2>. However, given the so-called promotion of the indirect object to subject position, it should not be assigned an exceptional function in the process of argument alternation: what matters here is the promotion to subject position of the term functioning as an indirect object, just like any other term/adverbial term (cf. 20b, 21b, 22b), rather than the underlying grammatical function of indirect object itself. Moreover, the example under discussion is a case not of alternation, but of passivization—a phenomenon closely related to argument expression alternation.

All the rules discussed involve a change in the position of two terms—both the subject/object and the adverbial term—and in every case the change begins with the latter’s promotion to subject or direct-object position, from which follows the demotion to adverbial-term position of the argument that had previously performed the grammatical function of subject or object. However, it appears that the sequence of events need not necessarily unfold in this order, since the phenomenon of argument alternation encompasses many cases in which only the position of one argument or adverbial term changes (cases of alternation between a location of action/impact and a direct object; cf. 13ab, 14ab), for example:

(23) a. The child swam across the river. b. The child swam across the river. (24) a. The boy kicks the ball. b. The boy kicks at the ball.

In sentence (23a), through derivation, the verb swam across acquires a secondary function—that of a transitive predicate—relative to the primary intransitive predicate swam. Consequently, the noun river, which is typically “adverbial” in meaning and performs the function of an adverbial term relative to the primary predicate (cf. to swim in the river, along the river), can be realized as a direct object¹⁹—thus, by superposition, the noun river is realized with the secondary, two-layer function <Q SUP 2>. In sentence (23b), after superposition, river is demoted from argument to adverbial term, and its function is now <<Q SUP 2> SUP Q>. Furthermore, sentence (23b) contains obviously redundant information.

¹⁹ According to A. Drukteinis, in constructions with motion or direction verbs, the syntactic status of the elements realized differently does not change, and constructions of this type cannot be classified as object relations (Drukteinis 1985: 59). However, the object-like status of river is also demonstrated by the passivization test; cf. the river was swum across by the child, while the two-layer content can be explained by the operation of superposition.

38

Lietuvių kalbos argumentų raiškos alternacijos

As in sentence (23a), by derivation, in sentence (24b) an intransitive predicate is realized in the secondary function of a transitive predicate without any external markers, while its argument—the former circumstantial term—is realized in the secondary function of a direct object – <Q SUP 2>. In this and other cases of alternation, evidence for the direction of transformation and for the double functions of the predicate and arguments (and/or circumstantial terms) in the transformed constructions can be regarded as the fact that the meaning of the transformed construction necessarily entails at least part of the untransformed information²⁰, but not vice versa; cf.: kaišyti kasas (19b) – “to make braids decorated by inserting things into them”; kiaulei aplipti paršeliais (20b) – “for a sow to become covered with piglets as a result of being (abundantly) covered by them”; perplaukti per upę (23b) – “to cross (swim across) a river by swimming across it”; perplaukti upę (23a) – “to cross (swim across) a river by swimming across it”; spirti kamuolį (24b) – “to strike a ball with one’s foot, thereby causing it to move.”

3.3. It is also important to return to the question of the predicate’s role in the transformations under discussion. In some cases of alternation (cf. 20ab, 23ab), a secondary intransitive-predicate function is “superimposed” on the primary transitive-predicate function. Incidentally, this also happens in all cases of passivization. Thus, in such constructions, the change in the predicate itself²¹,

²⁰ Similarly, in the lexical rules model, B. Levin, M. Rappaport Hovav, and S. Pinker ground the direction in which alternations are derived (see footnote 3), whereas S. Iwata’s denial of transformations in predicate structure and meaning in the case of locative alternations—argued on the grounds that some verbs allow only one of the variants—should be considered in the broader context of the phenomenon of argument alternation and its environment, and in deciding the assumptions underlying the possibility of the relevant realization/non-realization, rather than as an explanation of the mechanism of expression itself (Iwata 2005: 356–361).

²¹ In the formal description of the passivization process, the elimination of transitivity is explained by introducing an unspecified term (unspecified term), which, from the basic (transitive—two-place—predicate of the active construction), derives a converse predicate by applying a conversion operator (see footnote 14). This predicate is still two-place, and only when the converse predicate combines with the unspecified term as the result of the latter’s joining does a one-place intransitive passive predicate emerge. Thus, one function of the unspecified term is to turn the converse predicate into a one-place passive predicate. Other properties of the unspecified term are as follows: 1) it has the syntactic function of a secondary term (i.e., a direct object); 2) its meaning is “agent’s place” (i.e., it refers to any agent that may be specified by contextual, pragmatic-situational, or predicate lexical-meaning information); 3) it is marked by a zero sign, i.e., it is not expressed in any way (Descĺes, Guentchéva, Shaumyan 1985: 14, 55). It is not yet clear how the unfilled direct-object position should formally be marked in cases of alternation involving a change in the transitivity of the predicate.

39

Kristina Lenartaitė

which determines the realization of some of its arguments and/or circumstantial terms in different—secondary—functions is evident. Cases of alternation in which the transitivity of predicates does not change should be explained as regularities existing in the language system whereby predicate structure is realized in several ways: by a basic predicate and the positions it opens in its structure for arguments with primary functions, and by a derived predicate and the positions it opens in its structure for arguments with derived functions and/or circumstantial terms. A derived predicate always inherits the structure of the basic predicate and, by superposition, merges it with the altered predicate structure created by any of the described transformations (or various other possible combinations of them—in the end, all operations reduce to two basic ones: either an argument/circumstantial term is promoted in rank or demoted). Thus, the structure of a derived predicate is transformed from that of the basic predicate, but the constructions themselves are not derived from one another; they can only be related through their predicates. The basic structure can also be extracted (reduced) from the transformed structure, since the latter entails the former. This can be demonstrated with the following example:

(25) a. The hunters lay branches over the pit.

b. The hunters cover the pit with branches.

This pair of sentences should not be classified as an argument-expression alternation, since analogous processes here are realized not by the same, but by different verbal lexemes. The predicate apdeda in the second sentence (b) is derived from the predicate deda in the first sentence: thus, the derivative apdeda also has a secondary function in relation to the base word deda; the derivational marker at the morphological level is the prefix ap-, while the meaning of the derived verb indicates the direction of semantic derivation—apdėti ‘to place something on something, covering it on all sides’. At the abstract level of grammatical functions, the structure of the basic predicate is combined by superposition with one of the patterns for expressing the secondary functions of predicate arguments fixed in the language structure (i.e., the transformation rules discussed above and their combinations), while at the morphosyntactic level the secondary functions of the derived predicate are marked by arguments (and circumstantial terms) realized in the corresponding cases and/or prepositional constructions. In the case under discussion, it is clear that only the predicates are linked by a derivational relation; the constructions are related through the relation between their predicates, and the basic structure can be extracted (reduced) from the structure of the transformed predicate (cf. apdėti duobę šakomis ‘to place…’

40

Lietuvių kalbos argumentų raiškos alternacijos

branches over the pit so that the pit is completely covered”), but this does not mean that the realization of a derived predicate construction necessarily requires an intermediate operation realizing the base predicate construction—the second sentence can be expressed independently of the first. All the procedures discussed and their precise sequence are important only for explaining the relations between the two constructions, not for producing a particular sentence in a particular context: when a sentence with the predicate apdėti is needed, it is not necessary to recall that the latter is derived from dėti. The case discussed, along with other similar cases, is functionally equivalent to argument-expression alternation and is distinguished from it on the basis of formal morphological-level features (a morphological derivation marker). In argument-expression alternations, processes analogous to the case just discussed occur both with respect to the predicate change and with respect to its arguments (and/or adjunct terms), but are not externally expressed in the verbal lexeme in any way; therefore, it would also be appropriate to include pairs of structures with overt derivational markers (such as 25ab) in the periphery of argument-expression alternation (in some languages, e.g. Chicheŵa, processes analogous to argument alternations can be realized only through the attachment of the -ir- affix to the verbal lexeme (cf. footnote 3); and in German, changes in the expression of the argument serving as a locative are realized by attaching the affix be- to the verb, cf.: Sie sprü hte Farbe an die Wand “she sprayed paint onto the wall” and Sie besprühte die Wand mit Farbe “she sprayed the wall (with) paint” (Michaelis, Ruppenhofer 2001: 12)). Because of the analogous processes and the differences in information structure realized by the two sentences, the place of argument-expression alternation in the language system should be sought alongside the phenomenon of passivization²² (cf. 20abc). Perhaps more extensive research into the semantic functions of argument-expression alternation (especially in the field of pragmatics) would show that this phenomenon (or at least some of its cases) functionally belongs to the periphery of the category of voice.

The means of applicative grammar invoked to explain processes of argument-expression alternation only partly account for the preconditions for the emergence of pairs of constructions (these can be inferred only from the changed argument functions, without context); this article has also left undiscussed the differences in information structure realized in many cases of argument-expression alternation; furthermore, it is no less important to

²² A. Holvoet argues that voice, too, is a means of grammaticalizing information structure: the passive construction makes it possible to say nothing about the agent and to focus attention on the object, which becomes the topic of the sentence (Holvoet 2003: 87; see also Holvoet on the expression of a grammaticalized secondary topic in footnote 1).

41

KRISTINA LENARTAITE

to define the semantic properties of verbs that allow alternation—all this remains a task for further research.

4. CONCLUSIONS

Argument-expression alternation is a universal phenomenon of language structure, and therefore Lithuanian cases of argument-expression alternation should be examined in light of analogous phenomena in other languages and research on them. It has been observed that the term alternation, widely used for this phenomenon in foreign-language literature, is used in two ways: in a broader sense—as the set of possibilities for varying the expression of a predicate’s arguments (B. Levin’s position)—and in a narrower sense—as the regularities of predicate-argument expression that are grammaticalized differently in the systems of different languages (this is precisely how argument alternations are understood by most researchers, and this is the position adopted here). Thus, cases assigned to argument-expression alternation must meet all the following criteria: 1) the sentences must have the same lexical composition; 2) there must be changes in the predicate structure of the alternating constructions; 3) changes in expression must necessarily affect the grammatical functions realized by at least one of the predicate’s arguments—that is, the subject, the direct object, or both (the indirect object plays no distinctive role in the argument-expression alternation process: the grammatical function of the indirect object, like the function performed by any other term or adjunct term, matters in this phenomenon only insofar as it is promoted to the position of subject or direct object); 4) the relation between alternating sentences can be understood as paraphrastic / periphrastic and, with respect to the logical truth of propositions, even equivalent.

Considering the most general features of the phenomenon under study, the proposed Lithuanian term for it is argument-expression PERKATA.

The means of applicative grammar have been used to describe the processes by which argument-expression alternations are realized; among these, the most important role here is played by a phenomenon operating universally at all levels of language—functional superposition. It is precisely because of superposition at the level of semantic functions that possibilities arise, established in the system of each language in its own way, for predicates to be expressed in their basic structures (with primary argument functions) and in derived structures, which realize their arguments through secondary (layered) functions. The relation between the predicates of alternating constructions is established as that of a base form to a derived form, but their constr-

42

Lietuvių kalbos argumentų raiškos alternacijos

Constructions are not considered to be derived from one another, since they can be produced independently, unless the construction of the derived predicate can be extracted (reduced) from it, because the latter is always implied by the construction of the derived predicate.

For analogous processes, pairs of constructions that realize differences in argument expression through prefixed and corresponding unprefixed verb forms are also assigned to the periphery of argument-expression alternations, while the place of the phenomenon of argument alternation itself in the language system—both because of the similarity of realization processes and because of commonalities in the expression of information structure—can be seen as lying alongside the phenomenon of passivization.

LITERATŪRA

Desclés, J.-P., Guentchéva, Z., Shaumyan, S. 1985: Passivization in Applicative Grammar. Amsterdam–Philadelphia: John Benjamins Publishing Company.

Drukteinis, A. 1985: Objektiniai santykiai ir jų raiška. Disertacijos rankraštis. Klaipėda.

Drukteinis, A. 2007: Sakinio semantinių struktūrų ryšiai ir semantinių funkcijų hierarchija. Res Humanitariae 2, 7–30.

Fried, M. 2005: A frame-based approach to case alternation: the swarm-class in Czech. Cognitive Linguistics 16-3, 475–512.

Holvoet, A. 2003: Aktualiosios skaidõs problemos. In: A. Holvoet, A. Judžentis, red., Sintaksinių ryšių tyrimai. Vilnius: Lietuvių kalbos institutas, 2003 (Lietuvių kalbos gramatikos darbai, 1), 79–98.

Holvoet, Mikulskas, red., 2005: Gramatinių funkcijų tyrimai. Vilnius: Lietuvių kalbos institutas. (Lietuvių kalbos gramatikos darbai, 3.)

Holvoet, A., Judžentis, A. 2005: Sintaksinės priklausomybės tipai: papildymai ir patikslinimai. In: Holvoet, Mikulskas, red., 2005, 11–38.

Holvoet, A., Semėnienė, L. 2005: Veiksnio ir tiesioginio papildinio sąvokos. In: Holvoet, Mikulskas, red., 2005, 39–64.

Holvoet, A., Čižik-Prokaševa, V. 2005: Papildiniai ir aplinkybės. In: Holvoet, Mikulskas, red., 2005, 65–92.

43

KRISTINA LENARTAITĖ

IWATA, S. 2005: Locative alternation and two levels of verb meaning. Cognitive Linguistics 16-2, 355–407.

JABLONSKIS, J. 1997: Lietuvių kalbos gramatika. Vilnius: Mokslo ir enciklopedijų leidykla.

LEVIN, B. 1993: English Verb Classes and Alternations: a Preliminary Investigation. Chicago and London: The University of Chicago Press.

LEVIN, B., RAPPAPORT HOVAV, M. 2005: Argument Realization. Cambridge: Cambridge University Press.

MICHAELIS, L., RUPPENHOFER, J. 2001: Beyond Alternations: A Constructional Model of the German Applicative Pattern. Stanford: CSLI Publications.

SHAUMYAN, S. 1987: A Semiotic Theory of Language. Bloomington–Indianapolis: Indiana University Press.

SHAUMYAN, S. 2006: Signs, Mind and Reality: A Theory of Language as the Folk Model of the World. Amsterdam–Philadelphia: John Benjamins Publishing Company.

SLIŽIENĖ, N. 1978: Veiksmažodžių valentingumo klausimu. Lietuvių kalbotyros klausimai 18, 107–124.

SLIŽIENĖ, N. 1980: Subjektinimo ir objektinimo vaidmuo nustatant lietuvių kalbos veiksmažodžių sintaksinį valentingumą. LTSR MA darbai. Serija A, 1 (70), 93–105.

SLIŽIENĖ, N. 1994: Veiksmažodžių junglumas valentingumo teorijos požiūriu. Lietuvių kalbos veiksmažodžių junglumo žodynas 1. Vilnius: Mokslo ir enciklopedijų leidykla, 15–54.

VAN VALIN, R. D. 2005: Exploring the Syntax-Semantics Interface. Cambridge: Cambridge University Press.

VAN VALIN, R. D., LAPOLLA R. J. 1997: Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press.

Kristina Lenartaitė

Lietuvių kalbos institutas

P. Vileišio g. 5, LT-10308 Vilnius, Lietuva

[email protected]

44