Page 1Acta Linguistica Lithuanica LIX (2008), 1 – 27
The Semantics of Lithuanian Primary Verb Classes in the Context of Typological Data1
Piotras Arkadjevas
Institute of Slavic Studies, Russian Academy of Sciences
It is argued in this article that at least two inflectional classes of Lithuanian primary verbs have a strong semantic motivation, and that the semantic regularities correlating with Lithuanian verbal inflection classes are typologically significant. The largest inflectional classes, j-verbs and n/st-verbs, have clear semantic dominants. The morphosyntactic classification of verbs through grammaticalisation of semantic parameters relating to event structure, such as agentivity vs. patientivity, telicity vs. atelicity, action vs. change of state, is argued to be a fact of Universal Grammar.
0. INTRODUCTORY REMARKS
Inflection classes of words, that is, declension and conjugation types, have received considerable attention in discussions of the problem of the “autonomy” of morphology as an independent component of the language system, conveying information that cannot be derived from other levels of language—phonology, semantics, or syntax (see, e.g., Carstairs 1987; Carstairs-McCarthy 1994; Plank, ed., 1991; Aronoff 1994; Müller et al., eds., 2004). In this article
1This research was carried out with the support of the Russian National Science Support Foundation and using a grant from the Presidium of the Russian Academy of Sciences. I am sincerely grateful to Aleksejus Andronovas, Anna Dybo, Axel Holvoet, Artūras Judžentis, Leonid Kulikov, Tatjana Nikolajeva, Barbara Partee, Vladimir Plungian, Jekaterina Rachilina, Jakov Testelec, Vladimir Toporov, Andrej Zalizniak, and Marija Zavjalova for the various assistance they provided at different stages of the preparation of this article. A preliminary version of this text was presented as a talk at the International Symposium on the Identification and Representation of Verb Classes and Features, held in Saarbrücken in February–March 2005, and at the Grammar Conference of the second Salos Summer School of Linguistics, held in August 2005.
1
Page 2Piotras Arkadjevas
The aim here is different: namely, to show that even in a language with an exceptionally rich and complex system of inflection classes, such as Lithuanian, the morphological properties of lexemes can to some extent be predicted from their semantics.
The article examines the so-called primary Lithuanian verbs, that is, those whose infinitive stem has no affixes, so that nearly all the morphological oppositions characteristic of personal forms are neutralized in it. I will show that a verb’s membership in one of the two main inflectional subclasses of primary verbs is largely determined by certain semantic properties, namely, the type and structure of the situation denoted by the verb. This property determines many features of verbal lexemes in the languages of the world (see Levin & Rappaport Hovav 2005), so it is no coincidence that this very parameter has proved relevant for Lithuanian as well, even in such an “unexpected” area of its grammar as the distribution of verbs among conjugation types.
The first part of the article briefly describes the system of inflection classes of primary Lithuanian verbs. The second part examines in detail the semantic characteristics of the two largest subclasses of primary verbs. The third part presents typological data, specifically, analyzing similar phenomena in Georgian.
1. MORPHOLOGICAL CLASSES OF PRIMARY LITHUANIAN VERBS
The system of paradigmatic classes of primary Lithuanian verbs stands out for its complexity, which is characterized not only by an abundance of inflectional devices but also by the seemingly entirely unmotivated choice of these devices permitted for each individual lexeme. The structure of the infinitive of these verbs (traditionally regarded as the basic form) is “root + -ti,” so the oppositions characteristic of personal forms are neutralized here. On this basis, Tatjana Bulygina (Bulygina 1970, 49–50; 1977, 244) concluded that the infinitive contains the least information about a verb’s conjugation.
Most existing works on Lithuanian conjugation provide neither a comprehensive classification of the inflectional types of primary verbs nor an adequate description of them. The English-language “Grammar of the Lithuanian Language” (Ambrazas, ed., 1997, 287–289) contains only a table presenting verb-stem
2
Page 3Pirminių veiksmažodžių klasių semantika
inflection types, but provides neither a hierarchy of the various inflectional devices or possible constraints on their interaction, nor any serious attempt to determine the distribution of verb classes. Alfred Senn (Senn 1966, 258–279) likewise provides only lists of verbs classified by stem type. The detailed classification in the academic grammar (Ulvy das, ed., 1971, 219–235), which takes into account only the most general semantic properties of verbs (“action,” “state,” etc.), cannot be considered sufficiently precise from either the morphological or semantic point of view, although it undoubtedly offers a sufficiently comprehensive picture of the variety of forms and meanings of primary Lithuanian verbs (albeit without any generalizations). T. Bulygina’s “dynamic description of Lithuanian conjugation” (Bulygina 1977, 238–269) also does not resolve the problem of paradigmatic classes; this task is not even undertaken in that work. The same can be said of Claus Friedrich Eugen Heeschen’s comprehensive outline of generative Lithuanian phonology (Heeschen 1968, 190–282). An interesting classification of Lithuanian paradigmatic verb classes—unfortunately, without any substantive commentary and containing errors—is presented in an appendix to the typological article by Wolfgang U. Dressler and co-authors (Dressler et al. 2006). An analysis of the conjugation systems of Lithuanian and Latvian is presented in Aleksej Andronov’s article (Andronov 2000), which criticizes descriptive principles conventional in various grammatical traditions and proposes criteria for classifying verbs that are suitable for both languages, but does not itself provide a classification.
This article focuses not on morphology itself, but on the search for correlations between the morphological and semantic properties of primary Lithuanian verbs; therefore, I will discuss the “surface” structure of the inflectional subclasses of the lexemes under consideration only in the most general terms.
The morphological classes of primary Lithuanian verbs are distinguished according to the following main parameters (it should be added that I examine only verbs whose roots end in a consonant, leaving aside the small and highly heterogeneous class of verbs with a vowel or diphthong at the end of the stem):
1) presence/absence of palatalization of the root-final consonant (i.e., the suffix -j-) in the past tense (PRAET) stem;
2) presence/absence of palatalization of the root-final consonant (i.e., the suffix -j-) in the present tense (PRAES) stem;
3
Page 4PIOTRAS ARKADJEVAS
3) presence/absence of a nasal infix/lengthening of a short vowel in the PRAES stem;
4) presence/absence of the suffix -st- in the PRAES stem;
5) presence/absence of the alternation i ~ e in the PRAES stem;
6) presence/absence of the short grade of ablaut in the PRAES stem, with the lengthened grade in the PRAET and infinitive (INF) stems;
7) presence/absence of the lengthened grade of ablaut in the PRAET stem, with the short grade in the PRAES and INF stems.
Some of the specified parameters are independent of one another, while others are subject to mutual constraints. For example, all four theoretically possible combinations of the features (1) and (2) (palatalization of the root-final consonant) occur, albeit with varying frequency; however, the presence of an infix or the suffix -st-, or the alternation i ~ e in the PRAES stem, rules out the possibility of palatalization in this stem; by contrast, the lengthened grade of ablaut in the PRAET stem requires palatalization. In total, thirteen conjugation types can be distinguished (excluding individual lexemes with conjugational features peculiar to them alone); these are presented schematically in Table 1 (the INF and third-person PRAES and PRAET forms of the verbs are given).
Table 1. Morphological classes of primary Lithuanian verbs (i/e-PRAES = vowel alternation i ~ e in the PRAES stem; tr-PRAES = short grade of ablaut in the PRAES stem; il-PRAET = lengthened grade of ablaut in the PRAET stem)
Feature | (1) | (2) | (3) | (4) | (5) | (6) | (7) Class | j-PRAES | j-PRAET | n-PRAES | st-PRAES | i/e-PRAES | tr-PRAES | il-PRAET (1) augti: | − | − | − | − | − | − | − grows, grew (2) piršti: | − | − | − | − | + | − | − proposes, proposed (3) leisti: | + | − | − | − | − | − | − allows, allowed (4) ak-ti: | − | − | + | − | − | − | − swells, swelled
4
Page 51 lentelės tęsinys
Feature (1) (2) (3) (4) (5) (6) (7) Class j-PRAES j-PRAET n-PRAES st-PRAES i/e-PRAES tr-PRAES il-PRAET (5) skristi: skrenda, skrido − − + − + − − (6) blankti: blanksta, blanko − − − + − − − (7) tūžti: tūžta, tūžo − − + + − − − (8) gimti: gimsta, gimė − + − + − − − (9) minti: mena, minė − + − − + − − (10) pinti: pina, pynė − + − − − − + (11) arti: aria, arė + + − − − − − (12) pūsti: pučia, pūtė + + − − − + − (13) vemti: vemia, vėmė
Some of the distinguished classes form sharper oppositions associated with the hierarchy of the aforementioned features (this hierarchy is reflected in the order in which the features are presented in the list and the table). On the basis of features (1), (2), (3), and (4), which encompass the largest number of verbs, the following “macroclasses” of primary verbs can be distinguished (I use this term to denote a group of several smaller inflectional classes formed on the basis of shared formal features, without implying that a lexeme’s membership in a smaller class is determined by any of its features unrelated to inflection (cf. Carstairs 1987, Ch. 2)):
Pirminių veiksmažodžių klasių semantika
5
Page 6PIOTRAS ARKADJEVAS
Macroclass I (classes 1–3): verbs lacking palatalization in the PRAET stem and the infix/suffix -st- in the PRAES stem; this macroclass comprises about 30 lexemes2;
Macroclass II (classes 4–7): verbs with the infix/suffix -st- and without palatalization in the PRAET stem; this macroclass comprises about 270 lexemes;
Macroclass III (classes 8–10): verbs with palatalization only in the PRAET stem; this macroclass comprises just over 30 lexemes;
Macroclass IV (classes 11–13): verbs with palatalization in both the PRAES and PRAET stems; this macroclass comprises about 380 lexemes.
We see that the distinguished macroclasses differ greatly in the number of lexemes they contain: the vast majority of primary Lithuanian verbs belong to one of two macroclasses—II (infixal/suffixal PRAES stems) or IV (palatalized PRAES and PRAET stems). I will henceforth refer to lexemes in class II as n/st-verbs3, and those in class IV as j-verbs. The next part of the article will examine in depth the question of the semantic motivation of precisely these two largest classes, setting aside their subdivision into smaller subclasses.
2. ON THE SEMANTIC MOTIVATION OF THE MAIN PARADIGMATIC CLASSES OF PRIMARY LITHUANIAN VERBS
2.1. Introductory remarks
When discussing the semantic motivation of a grammatical phenomenon—in this case, the general semantic properties of morphological classes of lexemes—it is necessary to take the following aspects into account:
1. Oppositions distinguished on the basis of formal grammatical features are very rarely one hundred percent predictable from semantics; what matters most here is a statistically significant correlation between formal and semantic parameters. This is especially
2 The statistical data presented here and below were obtained by analyzing a sample of approximately 700 verbs compiled using a Lithuanian–Russian dictionary (Lyberis 1962).
3 It should be noted that the specific conjugation type of n/st verbs is derived from the structure of the verb stem; see, e.g., Stang (1942, 132–133).
6
Page 7Pirminių veiksmažodžių klasių semantika
relevant to a phenomenon such as the morphological classification of lexemes, which is established at the lexical level and may be affected by analogy and other similar processes that obscure semantic motivation. Here, the focus should be less on semantic properties shared without exception by all members of a given formal class and more on the class’s semantic dominant.
2. Semantic features that correlate with the grammatical properties of lexemes are generally not externally visible and cannot be directly “read off” from their “naive” dictionary definitions. For example, the study by Beth Levin and Malka Rappaport Hovav (Levin & Rappaport Hovav 1990) shows that a seemingly “natural” class of verbal lexicon such as verbs of motion is in fact heterogeneous; cf. also Jelena Paducheva’s monograph (Paducheva 2004), which identifies the main parameters for the semantic classification of the verbal lexicon that are relevant to grammar.
Taking the listed aspects into account, I will say that the nontrivial and nonrandom relationships between the semantic and morphological classes of Lithuanian primary verbs identified in this article are not absolute and, like all generalizations of this kind, have exceptions. Nevertheless, the statistical significance of these correlations is beyond doubt. It should also be borne in mind that the semantic parameters which, as I will demonstrate, are responsible for Lithuanian primary verbs belonging to one or the other of the two main macroclasses are quite abstract and belong to the most general and fundamental “structural” components of verb semantics—the thematic-role and aspectual organization of predicate meaning—that is, what recent literature usually calls “event structure” (see: Butt & Geuder, eds., 1998; Rothstein, ed., 1998; Tenny & Pustejovsky, eds., 2000; Ramchand 2008).
2.2. Semantic dominance of n/st verbs
The n/st-verb class is semantically homogeneous, a fact that researchers have also noted; compare Christian Stang’s observation (Stang 1966: 339) that most of the verbs in this group are intransitive and inceptive/inchoative verbs4. The vast majority of n/st verbs are syntactically intransitive and can informally be described as—
4 Žr. taip pat Temčin (1986).
7
Page 8PIOTRAS ARKADJEVAS
as “semantic passives,” i.e., verbs denoting situations that do not depend on the subject’s will (for example, because the subject is an inanimate object). The following semantic subclasses can be distinguished within this class:
a) states: to be hungry, to doze, to become impoverished, to be angry, etc.;
b) patientive processes entailing a change in the subject’s state: to perch, to turn white, to grow mouldy, to swell, to be shod, to become cheaper, and many others; two more verbs can be included here, which form the present-tense form with the suffix -st and with palatalization in the past tense: to die and to be born5;
c) natural phenomena: to dawn, to get dark, to snow;
d) uncontrolled changes in a person’s emotional state: to rage, to become gloomy, to go mad, to be stunned, etc.;
e) movement: (1) uncontrolled: to flow, to slip; (2) unspecified with respect to control: to rise; (3) controlled: to wade, to fly, to travel;
f) other verbs that do not belong to the aforementioned classes: to sprout, to shed leaves, to become, and some others.
Finally, a few isolated transitive verbs occur among the n/st verbs: to find, to feel, to like.
The principal semantic properties characteristic of most n/st verbs are patientivity of the subject and (inherent) telicity, that is, the component of meaning “transition into a new state” is characteristic of the structure of the situations denoted by these verbs. It is necessary to distinguish the “inherent,” rather than “actual,” telicity of these verbs because of certain features of the Lithuanian aspect system (Holvoet, Čižik 2004): the meaning “transition into a new state” itself generally cannot be expressed by the same lexeme as the meaning “process leading to that state.” Prefixed derivatives are used to express the telicity of an action, while retaining the inflectional properties of the base verb; compare to perch ~ to become perched, to be shod ~ to become shod, and many others.
It is important to note that the two aforementioned properties of n/st verbs are related; moreover, their subject’s patientivity derives from the meaning “(potential) change of state.” This becomes apparent when we consider isolated lexemes in this verb class that can be understood as agentive or even transitive. For instance, the verb to rise usually denotes a change of state uncontrolled by the subject and is most often used with
5 The present-tense form of these verbs appears to be a semantically motivated innovation; see Schmalstieg (2000: 156–158).
8
Page 9Pirminių veiksmažodžių klasių semantika
with inanimate subjects. However, when combined with a living participant in the situation, this verb acquires an additional component of meaning—“the situation is controlled by the subject”; the component of meaning “change of state” also remains. Transitive verbs justi and rasti can also be characterized with respect to this feature: despite their syntactic transitivity, these verbs emphasize a change in the subject’s state, only in one case the subject changes its internal state, while in the other it becomes the possessor of the object.
One further important confirmation that the “change of state” feature is dominant for n/st verbs is provided by a fairly large subgroup of this macroclass, comprising verbs used only with prefixes. Where they exist, non-prefixed lexemes with the same root often belong to the mixed or suffixal verb type. The general meaning of this group of n/st verbs is inception, i.e. transition into a state or entry into a process; depending on the nature of that process, these verbs divide into several semantic groups, among which an important place is occupied by those that would be difficult to call patientive, for example: entry into a process of internal causation6 (sužvilgti, sukrusti), the onset of speech or the production of other sounds (prakalbti, sužvangti), the onset of a state of perception (išgirsti), the onset of an emotional or mental state (panorti, pradžiugti). Of course, this group also contains a considerable number of purely patientive verbs that do not imply any active subject: aplešti, paliegti, pražiurti, etc.
Thus, the semantic dominant of n/st verbs is a (potential) change in the subject’s state. The fact that some verbs in the group clearly do not conform to this semantic prototype can, in principle, be explained (from a synchronic perspective) by an extension of the basic meaning. For instance, verbs denoting natural phenomena and implying no subject, such as brėkšti, nevertheless denote processes that bring about a change of state. The fact that this class includes verbs with a stative meaning (interestingly, all those in my
6 The distinction between internally and externally caused eventualities (internally-caused vs. externally-caused eventualities) was introduced and argued for in detail in B. Levin and M. Rappaport Hovav’s book (1995, chs. 3–4). The distinction is made according to whether a situation is conceptualized as one whose realization depends solely on the properties of the subject (for example, on its will, if the subject is a human or living being, as well as on the subject’s physical or other properties; for instance, one might suppose that žvaigždės žiba or žolė auga because of these subjects’ inherent propensity for the corresponding processes, rather than because they are acted upon by some external force), or, conversely, as one caused in the subject by an external force (for example, a lėkštė can dužti only if acted upon by someone or something, and not because of its own internal properties).
9
Page 10PIOTRAS ARKADJEVAS
corpus, verbs of this type denote negatively evaluated states), may perhaps be related to a metonymic shift in meaning: “bounded process/transition into a state → state.”
2.3. j-verbs’ semantic dominant
“Primary” j verbs, which have a palatalized final root consonant in PRAES and PRAET forms, divide into two large groups according to the syntactic feature of transitivity:
1. Transitive verbs, the great majority of which belong to the “canonical” type (see: Hopper, Thompson 1980; Testelets 1998; Næss 2007), i.e. denote the intentional action of a living agent upon a patient undergoing an (irreversible) change of state: skersti, daužti, trypti, drėbti. Many transitive j verbs are related to intransitive n/st verbs with the same root (see Section 2.4);
2. Intransitive verbs, which in turn divide into several semantic classes:
(1) “expressive” verbs;
(2) verbs of motion;
(3) verbs denoting natural phenomena.
Let us consider these classes in turn.
“Expressive” verbs form a very large class that is homogeneous in terms of both semantics and form. They are predominantly sound verbs, often denoting processes accompanied by sounds or other perceptible effects. Several subgroups of “expressive” j verbs can be distinguished:
(a) verbs denoting the production of sound: baubti, burgzti, čirkšti, inkšti, klykti, mūkti, pypti, šnarпšti, žliumbti and many others;
(b) verbs denoting movement: čiužti, plumpti, šliaužti, etc.;
(c) other “expressive” verbs: plieksti, brūzti, etc.
“Expressive” j-verbs have the following characteristics: similarity in formal structure (a heavy stem containing a long vowel or diphthong, with a falling accent); the combination of sound and motion meanings in many lexemes: čiurkšti, žliaugti; commonalities in “expressive” semantics: these verbs denote a salient characteristic of an object or situation as perceived by the speaker—
10
Page 11Pirminių veiksmažodžių klasių semantika
its characteristic, the speaker’s psychological reaction to it, as well as an emotional or aesthetic evaluation7; sound symbolism, characterized by an abundance of complex consonant clusters, sibilants, and affricates.
Among non-expressive j-verbs, a small subclass of motion verbs is worth mentioning: dengti, kopti, kuisti, lėkti, plaukti, žengti, etc. They are all verbs of manner of motion, rather than directed motion (the direction of motion can be expressed by prefixes), and in this respect differ, for example, from n/st-verbs denoting motion or a change in spatial position, such as kilti or vykti, which presuppose the existence of a certain endpoint or result state (on the importance of this opposition for motion verbs, see Talmy 2007).
Another small subclass of intransitive j-verbs consists of verbs denoting natural phenomena: drengti, druokti, dumti, dvelkti. These lexemes denote processes with a perceptible effect that do not presuppose any result state, unlike, for example, brėkšti-type n/st-verbs.
In addition, the class of j-verbs also includes the following lexemes, which do not fall into larger subgroups: burti, lošti, žaisti, bergžti, delsti, dvokti, snausti, želti, rėkti, šaukti, verkti, etc. Most of these verbs denote processes that can be characterized as internally caused. Internal causation is probably the most important feature of j-verbs in general, but it should be noted that there are also many n/st-verbs that can be considered internally caused verbs, for example: blankti, karšti, etc.—although most of them denote externally caused situations.
The semantic dominant of intransitive j-verbs can be described as follows: these verbs denote processes brought about by the subject’s will or determined by its internal properties, which do not presuppose a change in the subject’s state and are usually accompanied by a perceptible effect. Many j-verbs denote actions that can be performed only by an animate subject (a human or an animal) or by a natural force (wind, water, etc.). It is no coincidence that among j-verbs denoting sounds there are almost none that denote sound production caused by an external force (e.g., “rustle” in reference to leaves, etc.). Non-agentive j-verbs such as dvokti denote a process or state that is, first, internally caused and, second, accompanied by a perceptible effect (that is, these verbs, together with verbs of sound and light emission, belong to the broader category of so-called
7 This formulation is borrowed from Dee Ann Holisky (Holisky 1981, 105–107).
11
Page 12Piotras Arkadjevas
of emission verbs, a class). There are very few genuinely patientive j-verbs (cf. dvesti).
Interestingly, many intransitive j-verbs have transitive meanings, cf.: kliaukti, bergžti. Verbs denoting sound production seem generally to be characterized by taking a sound-denoting complement, cf.: blerbti niekus. This shows that there is, in fact, no very clear boundary between transitive and intransitive j-verbs.
The atelic character typical of intransitive j-verbs is confirmed by the fact that, except for verbs of motion, they rarely form prefixed derivatives, and where such derivatives are possible, they are characterized not by resultativity, as with n/st-verbs (cf. vesti ~ nuvesti), but only by ingressive meaning (baubti ~ užbaubti) or punctuality (subaubti).
The entire class of j-verbs—both transitive and intransitive—is characterized by a component of the structure of the situation they denote that signifies “a process caused by the subject or its internal properties and affecting either the object (in the case of transitive verbs) or the perception of the ‘observer’ (in the case of intransitive verbs with a perceptible effect).”
2.4. Regular relationships between verbs of the n/st- and j-classes
A considerable number of “primary” verbs form same-root verb pairs whose members regularly differ in the following three parameters: transitivity, morphological class, and degree of root-vowel ablaut. An especially extensive, though probably incomplete, list of such pairs is provided by Adelė Valeckienė (Valeckienė 1998, 27–28):
(a) birti – berti;
kilti – kelti;
mirkti – merkti;
skilti – skelti;
virsti – versti;
žirti – žerti;
tikšti – tekšti;
tįsti – tęsti;
linkti – lenkti;
(b) to scratch – to scrape;
dribti – drėbti;
12
Page 13Pirminių veiksmažodžių klasių semantika
(c) to fall – to thrash;
plisti – plėsti;
(c) kisti – keisti;
visti – veisti;
jukti – jaukti8;
(d) dygti – diegti;
drykti – driekti;
skysti – skiesti;
(e) lūžti – laužti;
rūgti – raugti
The intransitive verbs in the pairs presented belong to the n/st class and generally have a short grade of root-vowel ablaut (types a–c) or a simple long vowel (types d–e), while the transitive verbs belong to the j class and have the full grade (type a and types c–e, in which we see diphthongs) or the lengthened grade of root-vowel ablaut (type b, whose verbs show the simple full grade in the present-tense stem; cf. drėbti — drebia). These pairs illustrate the oldest currently known, and, naturally, no longer productive, way of opposing verbal lexemes according to the feature of transitivity/intransitivity.
Pairs of transitive and intransitive verbs of this type are interesting from a semantic point of view. Many studies (see Nedjalkov & Sil’nickij 1969; Haspelmath 1993; Levin & Rappaport Hovav 1995) show that when the relationship between transitive and causative verbs is most morphologized (as in the cases under discussion, where it is expressed by vowel ablaut and paradigmatic conversion), the intransitive member of the verb pair denotes a situation of external causation and its patient participant, while the transitive verb denotes an action whose direct consequence is a change in the patient’s state. Looking at the list of transitive and intransitive verb pairs presented, it is easy to see that most of them conform to the stated prototype. This fact may be regarded as further confirmation that the morphological primary verb classes of Lithuanian are semantically motivated.
Attention should be drawn to the fact that the same formal relations are found in several pairs of verbs that differ not in transitivity but in inchoativity—
8 This pair is not given in Adelė Valeckienė’s book.
13
Page 14Piotras Arkadjevas
as a feature (see the discussion at the end of Section 2.2); the n/st-verbs belonging to this group are not used without prefixes:
pravirkti – to cry; prajukti – to laugh; sustugti – to howl; suklikti – to shriek; pravimti – to vomit.
Thus, the same formal relations are used to express two types of semantic opposition: (1) causation of a patient process vs. a patient process; (2) a process of internally caused change vs. becoming involved in a process of internally caused change. This fact is not accidental and, as I will demonstrate in Section 3, has typological parallels.
2.5. The Relationship between the Semantic and Morphological Classes of Lithuanian Primary Verbs
Thus, in the preceding sections, we identified the semantic dominants of Lithuanian primary verbs belonging to the largest inflectional macroclasses — n/st-verbs and j-verbs. This is represented schematically in Table 2.
Table 2. Relationship between verb semantics and morphological class
subject-initiated process j-verbs (potential) change of state of the subject n/st-verbs
Table 2 shows that the basis for assigning a verb to one of the morphological classes is a component of that verb’s semantics related to the syntactic subject. Thus, other possible arguments of the verb are essentially irrelevant to morphological classification. Indeed, as has already been shown, both macroclasses include intransitive and transitive verbs, but the latter differ substantially in their properties: the j-class consists predominantly of “canonical” transitive verbs denoting an action directed toward an object, whereas the few transitive n/st-verbs do not conform to the transitivity prototype and denote a change in the subject’s own state that is related in some way to the object.
14
Page 15Pirminių veiksmažodžių klasių semantika
The revealed correlations between the syntactic-semantic and morphological properties of Lithuanian primary verbs are characterized by a high degree of statistical significance, as can be seen from Table 3, which presents statistics for the verb sample available to me9.
Table 3. Statistics on the relationship between the semantic and morphological classes of Lithuanian primary verbs
j-verbs | n/st-verbs | others | Total transitive | 247 | 8 | 51 | 306 intransitive agentive | 121 | 7 | 7 | 135 intransitive patientive | 7 | 237 | 4 | 248 Total | 375 | 252 | 62 | 698
Statistical significance for intransitive verbs and the two classes under consideration: χ = 17.7; p <.0001
It can be seen that, for intransitive verbs, the rule presented in Table 2 has almost no exceptions.
I will briefly consider verbs that do not belong to the main macroclasses discussed here. Compared with j- and n/st-verbs, such verbs are very few, and none of these minor morphological types can be characterized in general semantic terms. Table 3 shows that the great majority of verbs outside the main macroclasses are transitive; these include both prototypical transitive verbs, for example: sukti (suka, suko), kirpti (kerpa, kirpo), mesti (meta, metė), and “noncanonical” verbs whose meaning does not imply an active subject affecting an object: minti (mena, minė), sekti (seka, sekė). Despite the fact that neither macroclass I (verbs without palatalization or the infix/suffix -st-) nor macroclass III (verbs with palatalization only in the PRAET stem-
9In the calculations, prefixed verbs were not taken into account, including those without an unprefixed base lexeme.
15
Page 16Piotras Arkadjevas
lization) has a clear semantic dominant, the membership of some verbs in these rather than the main classes can be explained semantically. For instance, a small group of lexemes denoting a living subject’s directed change of position in space can be identified in macroclass I: sėsti (sėda, sėdo), lipti (lipa, lipo), lįsti (lenda, lindo). It may be assumed that the combination of such components of meaning as “change of position in space” and pronounced agentivity characteristic of these lexemes prevents their assignment to any of the main macroclasses. Nevertheless, it must be acknowledged that it is impossible to characterize Lithuanian primary verbs outside the j- and n/st-classes semantically with any degree of reliability, especially given the evident semantic motivation of the main macroclasses.
In conclusion, attention should be drawn to one interesting and important fact: although even the largest macroclasses of Lithuanian primary verbs can hardly be considered productive synchronically, there is no doubt that they displayed a certain degree of productivity at a not very distant stage in the history of the language. First, this is confirmed by the consistency with which verbs with expressive semantics and sound-symbolic stems entered the j class and, second, by the fact that among n/st-verbs there are lexemes whose morphophonological properties clearly place them in a later lexical stratum. Examples include the verbs tolti (< tolus), senti (< senas), pagailti (< gaila) and some others. Tolti-type verbs are anomalous because they have a superlong syllable nucleus (for example, /ol/), which is otherwise not tolerated (cf. korė vs. karti, where the long vowel is shortened in the corresponding position). Senti-type verbs, by contrast, are unique in that, contrary to the morphophonological rule of Lithuanian conjugation, which essentially has no exceptions, they retain /n/ in the present-tense stem before the suffix -st: sensta, rather than *sęsta, as might be expected; cf. the evidently older and lexicalized lexeme pažinti, whose present-tense stem is regular from a morphophonological perspective—pažįsta. These facts show that the n/st-class was still productive when the aforementioned morphophonological phenomena had already lost their historical phonetic motivation. The verb pratursti is also interesting: formally, it is primary, but it is clearly derived from the morphologically non-simple noun turtas < turėti.
16
Page 17Pirminių veiksmažodžių klasių semantika
3. TYPOLOGICAL PERSPECTIVE: “SPLIT INTRANSITIVITY” IN THE LANGUAGES OF THE WORLD
3.1. General remarks
The semantic parameters identified in the previous sections as relevant to the morphological classification of Lithuanian primary verbs—the telicity/atelicity of a situation or, more precisely, the presence in a verb’s semantic structure of a component denoting “transition into a state,” as well as agentivity and the opposition between internal and external causation of a situation—are not specific to Lithuanian. Many studies, both those based on various formal theories (see Perlmutter 1978; Levin & Rappaport Hovav 1990; 1995; 2000; Dowty 1991; Zaenen 1993; Alexiadou et al., eds., 2004, among others) and those belonging to the functional-typological tradition (see Lazard 1985; Merlan 1985; Van Valin 1990; Verhaar 1990; Mithun 1991; Sorace 2000; Donohue, Wichman, eds., 2008), convincingly show that these parameters are universally significant for the morphosyntactic classification of verbs across the languages of the world.
For example, Annie Zaenen (1993) has shown that in Dutch agentivity plays a decisive role in the formation of impersonal passive constructions, while telicity determines the choice of auxiliary verb in perfect tenses; data from Germanic and Romance languages are analyzed in this respect in an article by Antonella Sorace (Sorace 2000), which shows that agentivity and telicity interact in complex ways in the choice of auxiliary verb.
Mariannos Mithun straipsnyje (Mithun 1991) parodoma, kad minėtieji parametrai didele dalimi lemia laisvųjų įvardžių ar asmens derinimo rodiklių formų pasirinkimą Amerikos čiabuvių kalbose: intranzityvinių predikatų agentinis argumentas paprastai žymimas taip pat, kaip analogiškas tranzityvinio predikato argumentas, tas pats pasakytina ir apie pacientinį argumentą; kai kuriose kalbose prie to dar prisideda „subjektas patiriamo būsenos pasikeitimo“ požymis.
To show that Lithuanian is not “alone” with respect to the phenomenon examined in this article, we will consider data from Georgian. In this language, semantic structure is clearly related to morphosyntactic verb classes; moreover, we will show that the lexical composition of these classes in Georgian and Lithuanian has nontrivial similarities.
17
Page 18PIOTRAS ARKADJEVAS
3.2. Morphosyntactic verb classes in Georgian
The morphosyntax and semantics of Georgian verb classes have been studied fairly thoroughly
The morphosyntax and semantics of Georgian verb classes have been studied fairly thoroughly
The morphosyntax and semantics of Georgian verb classes have been studied fairly thoroughly (Holisky 1981; Harris 1981; 1982; 1985; Hewitt 1987; Van Valin 1990), so in what follows I rely on the publications cited above, primarily Dee Ann Holisky’s monograph, as well as Hans Vogt’s grammar (Vogt 1973).
Three major productive verb classes can be distinguished in Georgian, which I will designate by numerals, following tradition. The differences between these classes are manifested both in morphology (different sets of inflectional markers—first and foremost, person-agreement markers) and in syntax—in the case marking of verb arguments in forms of the so-called “second tense series”: the aorist (past perfective) and the optative (irrealis mood). The differences between the classes are schematically presented in Table 4.
Table 4.Morphosyntactic verb classes in Georgian
Class Argument form in the 2nd tense series 3SgSbPraes, 3PlSbPraes, 3PlSbAor markers I <Sb: Erg; DO: Nom> -s – -en – -es II <Sb: Nom> -a – -an – -nen III <Sb: Erg> -s – -en – -es
Let us examine the examples10. Class I verbs are transitive; in the “second tense series,” their subject is marked with the ergative case and their object with the nominative.
(1) glex-ma da-tes-a simind-i. farmer-ERG PRV-sow-AOR.3SG.SB grain-NOM “The farmer sowed the grain.” (2) bavšv-eb-ma gada-q’ar-es kv-eb-i. child-PL-ERG PRV-throw-AOR.3PL.SB stone-PL-NOM “The children threw the stones.”
Class II and III verbs are intransitive. Class II verbs
10 I thank Nino Amiridze for her help in analyzing the material.
18
Page 19Pirminių veiksmažodžių klasių semantika
the subject is marked with the nominative case (examples (3), (4)), while in Class III it is marked with the ergative case (examples (5), (6)):
(3) c’q’al-i ga-tb-a. water-NOM PRV-warm-AOR.3SG.SB “The water warmed up.” (4) bavšv-eb-i da-sxd-nen. child-PL-NOM PRV-sit.down-AOR.3PL.SB “The children sat down.” (5) c’q’al-ma i-duy-a. water-ERG PRF-boil-AOR.3SG.SB “The water boiled.” (6) gogo-eb-ma i-tamaš-es. girl-PL-ERG PRF-play-AOR.3PL.SB “The girls played.”
Let us now review the semantics of Georgian verb classes. The vast majority of Class I verbs are “canonical” transitive predicates denoting the effect of an active agent on a patient. The distribution of intransitive verbs between Classes II and III is more interesting.
Class II includes verbs of the following types11:
(a) “Passive” and anticausative counterparts of Class I transitive verbs: dac’ers ‘will write’ — daic’ereba ‘will be written’; damalavs ‘will hide’ — daimaleba ‘will hide oneself’; gaamravalebs ‘will increase’ — gamravald eba ‘will increase’ (intransitive), etc.;
(b) Ingressive derivatives of Class III verbs: laklakebs ‘chatters’ — alalakldeba ‘will start chatting’; itamašobs ‘will play’ — atamašdeba ‘will start playing’; goravs ‘rolls’ — migordeba ‘will start rolling’, etc.;
(c) (non-agentive) verbs denoting a bounded action: mok’vdeba ‘will die’, darčeba ‘will remain’, dadgeba ‘will stand up’, datbeba ‘will warm up’, and others.
Most Class II verbs denote bounded situations in which a participant undergoes a change of state, usually an uncontrolled one. Interestingly, this class also includes verbs denoting involvement in a process, as well as verbs whose situation includes an agent participant. Thus, the semantic dominant of Class II verbs is the component of “a change in the subject’s state.”
11 The 3rd person singular future tense form of the verb is given.
19
Page 20PIOTRAS ARKADJEVAS
Class III divides into the following main subclasses (in this article I consider only those verbs of this class that are not derived; derived lexical items, such as denominal derivatives meaning “behave like X,” for example, art’ist’obs “behave like an artist,” fully conform to the semantic prototype presented below):
(1) “expressive” verbs;
(2) motion verbs;
(3) verbs denoting natural phenomena.
Thus, even at the level of the most general characteristics of Georgian Class III verbs, similarities with Lithuanian intransitive j-verbs are apparent.
“Expressive” Class III verbs, in turn, divide into the following subclasses:
(a) verbs denoting the production of sound: bzuk’unebs “buzz,” bubunebs “sob,” zr ialebs “crackle,” lak lakebs “chatter,” sisinebs “rustle,” grgvinavs “thunder,” zmuis “moo,” c’k’muis “whimper,” rak’unobs “knock,” xorxocebs “snore,” etc.;
(b) verbs denoting shining: bdγvrialebs “shine,” brč’q’vialebs “glimmer,” varvarebs “flame,” k’ašk’ašebs “glow,” rialebs “glimmer,” bzinavs “sparkle,” etc.;
(c) verbs denoting movement in place: babanebs “tremble,” bibinebs “wave (of grass),” taxtaxebs “rattle,” t’ivt’ivebs “float,” panckalebs “throb (of the heart),” prtxialebs “flutter,” borgavs “tumble,” ckmut’avs “flail,” trtis “tremble,” etc.;
(d) manner-of-motion verbs: bobγavs “crawl,” barbacebs “stagger,” laslasebs “drag oneself,” parpatebs “flutter,” q’ialebs “sway,” etc.;
(e) verbs not included in classes (a)–(d): tkrialebs “flow” (of blood from a wound), partipurtebs “rummage,” šxeps “splash,” c’vetavs “drip,” etc.
Thus, comparing the lexical composition of “expressive” verbs in Georgian and Lithuanian reveals both significant similarities and interesting differences. In both languages, a large proportion of “expressive” verbs consists of lexical items denoting processes that produce perceptible effects—verbs of sounding, shining, “quivering,” and so on. In Lithuanian, however, most of these verbs are agentive and can be associated only with an animate or personified subject, whereas Georgian has no such restriction, and a great many Class III verbs denote sounds produced by leaves, stones, and other objects under the action of an external force.
20
Page 21Pirminių veiksmažodžių klasių semantika
“Non-expressive” Class III verbs are also semantically heterogeneous. They include: (a) verbs of manner of motion: goravs ‘roll’, srialebs ‘crawl (of a snake)’, xt’is ‘jump’, curavs ‘swim, float, skate’, etc.; (b) verbs denoting dynamic natural phenomena: grgvinavs ‘(of thunder)’, elavs ‘(of lightning)’, tovs ‘snow’, kris ‘the wind blows’, c’vims ‘rain’, etc.; (c) several verbs that do not belong to any group: asp’arezobs ‘compete’, brazobs ‘be angry’, duys ‘boil (of water)’, tamašobs ‘play’, muša-obs ‘work’, pikrobs ‘think’, čkarobs ‘hurry’, cek’vavs ‘dance’, and some others.
Thus, we could identify as the semantic dominant of Class II verbs in Georgian an atelic process, usually agentive (the Subject is a living being or a natural force), and, if non-agentive, one that has a perceptible effect.
All this leads us to conclude that, alongside undeniable differences, there is also an evident and considerable similarity between Georgian and Lithuanian verb classes—both with respect to their most general semantic characteristics and their lexical content (for example, “expressive” verbs are highly developed in both languages; in both languages, verbs denoting processes of external causation and those denoting involvement in a process of internal causation fall into one class). These similarities are highly non-trivial and non-accidental and cannot be explained by either genetic or areal-geographical factors. Thus, the established similarities between Georgian and Lithuanian reflect universal tendencies in the organization of the lexicon, characteristic of human language in general and manifested at different levels in different languages.
4. CONCLUDING REMARKS
I hope that in this article I have succeeded in showing, first, that at least one important fragment of the system of morphological classes of Lithuanian primary verbs is strongly semantically motivated and, second, that the semantic features correlating with a verb’s morphological class in Lithuanian are typologically relevant and relate to what is often, somewhat loosely, called “universal grammar.” I will briefly reiterate the main findings of this study.
21
Page 22PIOTRAS ARKADJEVAS
The two largest classes of Lithuanian primary verbs—j-verbs (with palatalization of the root-final consonant in the present and past simple tense stems) and n/st-verbs (with a nasal infix or the suffix -st- in the present tense stem)—have pronounced semantic dominants.
The core of n/st-verbs consists of predicates denoting a potential or actual change in the subject’s state, or an atelic process leading to such a change. The vast majority of n/st-verbs, as well as the few bivalent lexemes in this class, conform to this prototype.
At first glance, the less homogeneous class of j-verbs includes both transitive verbs, the vast majority of which conform to the transitivity prototype (an active, controlling agent’s impact on a patient who is strongly affected by the situation and undergoes an uncontrolled change of state), and intransitive verbs, whose key property is that they denote an atelic process caused by the subject’s will or inherent properties.
The aforementioned correlations between the semantics of primary verbs and morphological macroclasses have a high degree of statistical significance (see Table 3) and most likely arose through the grammaticalization of corresponding semantic oppositions. This assumption is also supported by the fact that there are quite a few j-verbs and n/st-verbs that do not directly belong to the oldest lexical strata.
From a typological perspective, the phenomenon under consideration is not unique. More precisely, what is universal are the semantic parameters themselves—agentivity ~ patienthood, telicity ~ atelicity, action ~ change of state—in other words, features that classify the possible types of situations lexicalized by verbs in the world’s languages. The “surface” realization of these features—that is, the area of grammar in which they manifest themselves—depends on the particular language. This article demonstrated this by comparing the lexical composition and semantic dominants of Lithuanian verb classes with the morphosyntactic verb classes of Georgian, where semantic factors, however similar they may be to those in Lithuanian, are manifested primarily at the syntactic level.
In conclusion, it must be emphasized once again that both the very fact of the semantic motivation of Lithuanian primary verb morphological classes and, even more so, the typological non-uniqueness of this phenomenon are far from accidental and testify to the fact that languages of different origins, geographical locations, and linguistic orga-
22
Page 23Pirminių veiksmažodžių klasių semantika
The natural-language lexicon and grammar are determined by universal semantic parameters.
ABBREVIATIONS
AOR – aorist, DO – direct object, ERG – ergative, INF – infinitive, NOM – nominative, PL – plural, PRAES – present tense, PRAET – simple past tense, PRF – perfective, PRV – preverb, SB – subject, SG – singular.
LITERATŪRA
Alexiadou, A., E. Anagnostopoulou & M. Everaert, eds., 2004: The Unaccusativity Puzzle. Oxford: Oxford University Press.
Ambrazas, V., ed., 1997: Lithuanian Grammar. Vilnius: Baltos lankos.
Andronov, A. 2000: Some remarks on the system of Lithuanian and Latvian conjugation. Linguistica Baltica 8, 35–47.
Aronoff, M. 1994: Morphology by Itself. Stems and Inflectional Classes. Cambridge, MA: MIT Press.
Bulygina, T. V. [Булыгина, Т. В.] 1970: Морфологическая структура слова в современном литовском языке (в его письменной форме). In: B. M. Жирмунский, Н. Д. Арутюнова, ред., Морфологическая структура слова в индо-европейских языках. Москва: Наука, 7–70.
Bulygina, T. V. [Булыгина, Т. В.] 1977: Проблемы теории морфологических моделей. Москва: Наука.
Butt, M. & W. Geuder, eds., 1998: The Projection of Arguments. Lexical and Compositional Factors. Stanford, CA: CSLI Publications.
Carstairs, A. 1987: Allomorphy in Inflection. London: Croom Helm.
Carstairs-McCarthy, A. 1994: Inflection classes, gender, and Principle of contrast. Language 70, 737–788.
Donohue, M. & S. Wichman, eds., 2008: The Typology of Semantic Alignment Systems. Oxford: Oxford University Press.
23
Page 24Piotras Arkadjevas
Dowty, D. R. 1991: Thematic proto-roles and argument selection. Language 67, 547–619.
Dressler, W. U., M. Kilani-Schoch, L. Pestal, N. Gagarina & M. Pöchtrager 2006: On the typology of inflection class systems. Folia Linguistica 40, 51–74.
Harris, A. C. 1981: Georgian Syntax. A Study in Relational Grammar. Cambridge: Cambridge University Press.
Harris, A. C. 1982: Georgian and the unaccusative hypothesis. Language 58, 290–306.
Harris, A. C. 1985: Diachronic Syntax: The Kartvelian Case. (Syntax and Semantics 18) New York etc: Academic Press.
Haspelmath, M. 1993: More on the typology of inchoative/causative verb alternations. In: B. Comrie & M. Polinsky, eds., Causatives and Transitivity. Amsterdam, Philadelphia: John Benjamins, 87–120.
Heeschen, C. F. E. 1968: Einführung in die Grundprobleme der generativen Phonologie mit besonderer Berücksichtigung der litauischen Phonologie. Bonn: Friedrich-Wilhelms-Universität.
Holisky, D. A. 1981: Aspect and Georgian Medial Verbs. New York: Garland.
Holvoet, A. & V. Čižik 2004: Veikslo priešpriešos tipai. In: A. Holvoet & L. Semėnienė, red., Gramatinių kategorijų tyrimai (Lietuvių kalbos gramatikos darbai, 2). Vilnius: Lietuvių kalbos institutas, 141–162.
Hopper, P. J. & S. A. Thompson 1980: Transitivity in grammar and discourse. Language 56, 251–299.
Lazard, G. 1985: Anti-impersonal verbs, transitivity continuum and the notion of transitivity. In: H. Seiler & G. Brettschneider, eds., Language Invariants and Mental Operations. Tübingen: Narr, 115–123.
Levin, B. & M. Rappaport Hovav 1990: The lexical semantics of verbs of motion: The perspective from unaccusativity. In: I. M. Roca, ed., Thematic Structure: Its Role in Grammar. Berlin, New York: Mouton de Gruyter, 247–269.
24
Page 25Pirminių veiksmažodžių klasių semantika
Levin, B. & M. Rappaport Hovav 1995: Unaccusativity. At the Syntax-Lexical Semantics Interface. Cambridge, MA: MIT Press.
Levin, B. & M. Rappaport Hovav 2000: Classifying single argument verbs. In: P. Coopmans, M. Everaert & J. Grimshaw, eds., Lexical Specification and Insertion. Amsterdam, Philadelphia: John Benjamins, 269–304.
Levin, B. & M. Rappaport Hovav 2005: Argument Realization. Cambridge: Cambridge University Press.
Lyberis, A. 1962: Lietuvių-rusų kalbų žodynas. Vilnius: Valstybinė politinės ir mokslinės literatūros leidykla.
Merlan, F. 1985: Split intransitivity: Functional oppositions in intransitive inflection. In: J. Nichols & A. Woodbury, eds., Grammar Inside and Outside the Clause. Approaches to Theory from the Field. Cambridge: Cambridge University Press, 324–362.
Mithun, M. 1991: Active/agentive case marking and its motivations. Language 67, 510–546.
Muller, G., L. Gunkel & G. Zifonun, eds., 2004: Explorations in Nominal Inflection. Berlin, New York: Mouton de Gruyter.
Næss, Å. 2007: Prototypical Transitivity. Amsterdam, Philadelphia: Benjamins.
Nedjalkov, V. P. & G. G. Sil’nickij [В. П. Недялков, Г. Г. Сильницкий] 1969: Типология морфологического и лексического каузативов. In: А. А. Холодович, ред., Типология каузативных конструкций. Ленинград: Наука, 20–50.
Padučeva, E. V. [Е. В. Падучева] 2004: Динамические модели в семантике лексики. Москва: Языки славянских культур.
Perlmutter, D. M. 1978: Impersonal passives and the Unaccusative hypothesis. Proceedings of the Berkeley Linguistics Society 4, 157–189.
Plank, F., ed., 1991: Paradigms. The Economy of Inflection. Berlin, New York: Mouton de Gruyter.
25
Page 26Piotras Arkadjevas
Ramchand, G. 2008: Verb Meaning and the Lexicon: A First Phase Syntax. Cambridge: Cambridge University Press.
Rothstein, S., ed., 1998: Events and Grammar. Dordrecht: Kluwer.
Schmalstieg, W. 2000: The Historical Morphology of the Baltic Verb. Washington, DC: Institute for the Study of Man. (Journal of Indo-European Studies Monograph No. 37)
Senn, A. 1966: Handbuch der Lithauischen Sprache. Bd. I. Grammatik. Heidelberg: Carl Winter.
Sorace, A. 2000: Gradients in auxiliary selection with intransitive verbs. Language 76, 859–890.
Stang, Chr. S. 1942: Das slavische und baltische Verbum. Skrifter utgitt av Det Norske Videnskaps-Akademi i Oslo. II. Hist.-Filos. Klasse I, 1–280.
Stang, Chr. S. 1966: Vergleichende Grammatik der baltischen Sprachen. Oslo, Bergen, Tromsö: Universitetsforlaget.
Talmy, L. 2007: Lexical typologies. In: T. Shopen, ed., Language Typology and Syntactic Description 3. Grammatical Categories and the Lexicon. 2nd ed. Cambridge: Cambridge University Press, 66–168.
Temčin, S. [С. Темчин] 1986: Семантика -n- и -sta- глагольных основ в литовском языке. Kalbotyra 37, 87–98.
Tenny, C. & J. Pustejovsky, eds., 2000: Events as Grammatical Objects. Stanford, CA: CSLI Publications.
Testelets, Y. G. 1998: On two parameters of transitivity. In: L. Kulikov & H. Vater, eds., Typology of Verbal Categories. Papers presented to Vladimir Nedjalkov on the occasion of his 70th birthday. Tübingen: Niemeyer, 29–46.
Ulvydas, K. red., 1971: Lietuvių kalbos gramatika 2. Morfologija. Vilnius: Mokslas.
Valeckiene, A. 1998: Funkcinė lietuvių kalbos gramatika. Vilnius: Mokslo ir enciklopedijos leidybos institutas.
26
Page 27Pirminių veiksmažodžių klasių semantika
Van Valin, R. D., Jr. 1990: Semantic parameters of split intransitivity. Language 66, 221–260.
Vogt, H. 1971: Grammaire de la langue géorgienne. Oslo: Universitetsforlaget.
Zaenen, A. 1993: Unaccusativity in Dutch: Integrating syntax and lexical semantics. In: J. Pustejovsky, ed., Semantics and the Lexicon. Dordrecht: Kluwer, 129–161.
Vertė Andrius Raskazovas
Petr Arkadiev Institut slavjanovedenija Rossijskoj akademii nauk RU-117334 Moskva, Leninskij prospekt 32-a [email protected]
27