LITUANUS LITHUANIAN QUARTERLY JOURNAL OF ARTS AND SCIENCES   Volume 25, No.4 - Winter 1979 Editor of this issue: Antanas Klimas ISSN 0024-5089 Copyright © 1979 LITUANUS Foundation, Inc.

A CASE GRAMMAR APPROACH TO VERB CLASSIFICATION IN MODERN LITHUANIAN

CHIEN-CHING MO
National Chengchi University

This paper deals with the classification of verbs in terms of verb type and configurations of case relations. I am to present a method of classifying modern Lithuanian verbs in accordance with two criteria: (1) the verb type, and (2) the semantic roles implied by the verb. First of all, I would like to outline the salient points of Walter Cook's case grammar matrix, which serves as a useful tool in verb classification.

I. Cook's Case Grammar Matrix

To make Fillmore's and Chafe's case theories reconcilable, Cook initially arranged the possible case configuration in a 16-cell matrix in 1972. One dimension of his matrix, based on Chafe's Chapter 9, divides verbs into (a) state, (b) process, (c) action, (d) action-process. The other dimension of the matrix, based on Chafe's Chapter 12, divides verbs into (1) basic, (2) experiencer, (3) benefactive, and (4) locative verbs. This 4 by 4 array forms a 16-cell matrix of verb types. Table 1 shows the sixteen-frame matrix as it appears in Cook (1972:26):

Table 1: The Case Frame Matrix

Cook (1973b:181) points out that "further work with the language suggested that there are case roles which are partially or totally covert in the surface structure, while present in the deep structure. The discovery of covert roles caused changes in the structure of matrix as a representation of deep semantic structure."

Partially covert roles are listed as obligatory to the deep structure, but the case frame is supplemented by a deletion rule which indicates that the partially covert role is deletable from the surface structure.

Table 2: Partially Covert Roles, Optional in Surface Structure

Totally covert roles are case roles which are present in the deep structure, but generally absent from the surface structure. Totally covert roles may be simply vacant, e. g. the Experiencer role in the verb, resemble; or coreferential, e. g. the Agent and Object case in the verb, walk; or lexicalized into the verb, e. g. the Object case in the verb, question. Roles which are totally absent from surface structure are marked, e. g. O*, and the reason for its absence added as a feature of the case frame, e. g. O vacant, or O lexicalized, or, if coreferential, O = A.

Table 3: Totally Covert Roles, Absent in Surface Structure

With the advent of partially covert and totally covert case roles, it is possible to defend the Anderson (1971:37) hypothesis that Object is an obligatory case, present in every case frame. The 16-frame matrix can be reduced into a 12-frame matrix by accepting the Object as an universal case role. Case frames such as [__A], [__A E], etc. are changed into [__A, O*], [__A E O*], etc. That is, the third row in Table 1, where the Object is absent, is eliminated from the matrix in the new formation. Chafe's deprocessive and complementizer derivational units are no longer necessary with the disappearance of the action verbs from the deep structure. Table 4 shows the twelve-frame matrix as it appears in Cook (1974:8):

Table 4: Revised Case Grammar Matrix

The derivational processes and the relationships between the derivational units and the verbs which they convert according to the revised case grammar matrix are illustrated as follows:

II. The Case Frame Matrix Used for Lithuanian Verb Classification

The problem considered here is the following: Is Cook's matrix of case frames a suitable tool for the classification of Lithuanian verbs? Does it account for all the verbs that are found in the widely used textbook: Introduction to Modern Lithuanian written by L. Dambriūnas, A. Klimas and W. Schmalstieg?

The following sentences taken from L. Dambriūnas et al. (1972) are related to the analysis of predicate nominals:

(1) Mano tėvas yra mokytojas, (p. 21) 
    'My father is a teacher.'

(2) Vilnius tapo didžiausiu ir gražiausiu Lietuvos miestu, (p. 215)
    'Vilnius became the largest and the most beautiful city of Lithuania.'

(3) Šis miestas vadinasi Vilnius. (p. 43)
    'This city is called Vilnius.'

Fillmore (1968:84) suggests an Essive case whose use is restricted to copulas. Langendoen (1970:102) assigns the Essive label to the subject nominal and calls the predicate nominal Predicate. Stockwell, et al. (1973:29) define the Finno-Ugric term Essive as follows, "Essive is the case employed for predicate nominals. It is the case dominating a good teacher in 'That man is a good teacher'." From the viewpoint of simplicity, it is more economical to attempt to analyze nominal predicates within a framework already available for the analysis of verbal and adjectival predicates than to introduce an ad hoc case relation such as the Essive.

Chafe (1970:202) treats predicate nominals as state verbs whose basic structure is BE+N+Predicativizer. Thus, his analysis of sentence (1) can be shown below:

The verb in (1) is identified as mokytojas. Clearly, the copula is not considered as a part of the verb. Be is seen as merely a carrier of tense and other inflectional units of verbs whose surface structure form is incapable of being inflected as other types of verbs in the English language (Chafe 1970:160). This interpretation seems to neglect the fact that a predicate nominal construction is involved in (1). Apart from this, Chafe does not postulate a scheme for handling predicate nominal constructions which are not stative such as sentences (2) and (3). Nilsen (1973:25) analyzes

(4) John is a lackey.
(5) A hammer is an important tool.

as containing two Objects (John, lackey), and two Instruments (hammer, tool) respectively. Evidently, the assignment of the same case label to both subject nominal and predicate nominal deviates from Fillmore's one-instance-per-clause principle (i. e. a simple sentence can contain at most only one occurrence of any case). However, Nilsen (1973:25-26) comments

There is little logical support for the position that there can be only one instance of a particular case per simple predication. One example of a violation of this principle is the equational or linking sentence . . . The concept of Predicate Nominative is a concept of surface structure. It is therefore difficult to see how a deep case label (Essive) can be assigned to a surface phenomenon. On the other hand, if it is argued that Essive is not a deep case, then we must assign the correct deep case label, and we are backing to having more than one instance of a particular deep case in a simple sentence.

There is an alternative explanation. Anderson (1971:206-207) points out that in some languages (like Welsh) the presence of a locative marker with the predicate nominals is quite normal. Therefore, he assumes that the predicate nominal in the sentence like (6)

(6) He's a policeman.

is a subtype of locative.

In the following sentence,

(7) Knyga kainuoja penki doleriai.
    'The book costs 5 dollars.'

Chafe (1970:157) states that a completable verb either may or must be accompanied by a complement noun (to be abbreviated comp). According to this interpretation, the essential structure of (7) can be diagrammed as follows:

Chafe's Patient and Complement in Cook's case frame matrix are put together under the more general case label Object; therefore the action-completable verbs such as dance, play, sing are reclassified as action-process verbs with deletable objects. But Cook neglects such state-com-pletable verbs as cost, weigh, and measure. When analyzing these verbs, Nilsen (1973:148) proposes the 'Measure' case for the predicate nominal in (8):

(8) [The painting]O cost [$5,000]Measure

Due to the fact that the verb kainuoti 'cost' in (7) has an attribute specified in terms of money, and can be replaced by būti 'be' as in (9).

(9) Knyga yra penki doleriai.
     'The book is five dollars.'

We can substitute the (+stative) Objective for the Measure, and assume that the verb cost can occur in [__Os, Os].

Nilsen (1973:149) states that symmetric verbs such as look like, resemble, marry and seem must have the same case as their subject and their complement. Cook (1973a:72) offers a tentative analysis of (10).

(10) [Fred]O resembles [John]O [(to me)]E.

That is, the verb resemble occurs in +[_Es, O, O]/Es vacant. Cook calls verbs like resemble complex because they violate the one-case-per-clause principle and because they probably require an analysis of the type: (1) E experiences an event, and (2) the event is X=Y. Perhaps Cook is one of the case grammarians who believe each deep case can be represented only once; therefore, the analysis of predicate nominals is excluded from his case frame matrix.

Baer (1976), following Nilsen, analyzes the English predicate nominal in her dissertation. In order to fulfill the condition of adequacy as far as the Lithuanian verb is concerned, I put two of Cook's propositional cases, Experiencer and Benefactive, under the case label Dative and follow Baer in discussing the Lithuanian predicate nominal. The following is the case frame matrix used for Lithuanian verb classification.

This 3x4 matrix defines 12 possible Lithuanian verb types in terms of the nouns (propositional cases) that are necessarily found with these verbs. These nouns constitute the case frame of the verb.

III. Illustration of Verb Classification

As regards Fillmore's case frame classification of verbs, Cook (1974:14) points out the following weaknesses:

(1) The state form is not included in the entry.

(2) Since related but separated forms are grouped together under one entry, no attempt is made to relate these forms by derivations nor to choose the base form.

(3) The conflated frame, improperly understood, implies that all the cases listed in the frame are implied in all uses of the verb; for example, that Agent is implied in process uses of the verb.

To avoid such defects, I try to group morphologically related verbs under one entry in the lexicon. Derived forms are enumerated as subentries. Derivational relations between the forms are shown by the inclusion of the derivational semantic units. Our discussion can begin in a simple way with the following sentences as illustrations:

The verbs in (11)-(13) are classified in the lexicon as follows:

būti tirštai copula + aj, STATE +[__Os]
+ IN (=Inchoative)=tirštėti iv, PROCESS +[__O]
+ IN + CA (=Causative)=tirštinti tv, ACTION PROCESS +[__A, O]

Morphological and/or syntactic correlates are used to determine which, among the related verb forms, is basic. In Cook (1974:16), it is pointed out that the state form like būti tirštai is chosen as base form because of simplicity of form, or because of the contrast existing between the simple adjective and the past participle of the verb. Derivational relations among the related verb forms are considered languages specific.

IV. The Case Frame Matrix Applied

Most of the verbs occurring in the text: Introduction to Modern Lithuanian written by L. Dambriūnas, A. Klimas and W. Schmalstieg are listed in this section. As we know, there are three basic verb types: (1) state verbs which specify that an object is in a certain state or condition and which are accompanied by an Objective noun, (2) process verbs which specify that an object undergoes a change of state or condition and which are accompanied by an Objective noun specifying what it is that changes its state or condition, and (3) action-process verbs which express an action and which are accompanied by an Agentive noun and an Objective noun. Verbs are further classified within each case frame classification according to the presence or absence of totally covert roles in their structures. The number in the parentheses gives the lesson in which the verb appears for the first time in the text.

1. State Verbs +[_Os]

2. Process Verbs

a. No Totally Covert Roles: +[__O]

 b. O Vacant +[_O*]

lyti.........................(16)

3. Action-Process Verbs

a. No Totally Covert Roles: +[_A, O]

 b. A and O Coreferential: [_A, O*]/A=O

c. O Lexicalized: [_A, O*]/O-lex

klysti........................(10)    pietauti.....................(40)

4. State Dative Verbs

No Totally Covert Roles: +[_D, Os]

5. Process Dative Verbs

a. No Totally Covert Roles: +[_D, O]

 b. D and O Coreferential +[_A, D. O] 

išsigąsti....................(15)

6. Action-Process Dative Verbs

a. No Totally Covert Roles: +[_A, D, O]

b. A and D Coreferential: +[_A, D*, O]/A=D

c. O Lexicalized: +[_A, D, O*]/O-lex

 7. State Locative Verbs: + [_Os, L]

 8. Process Locative Verbs: +[_O, L|

 9. Action-Process Locative Verbs

a. No Totally Covert Roles: +[_A, O, L]

b. A and O Coreferential: +[_A, O*, L]/A=O

c. O Lexicalized: +[_A, O*, L] IO-lex

laistyti.......................(29)

10. +[_Os, Os]

11. +[_O, O]

12. +[_A, O, O]

NOTES

1 I am greatly indebted to Professor Antanas Klimas for his constructive criticisms and constant encouragement. I also wish to thank Professor Ronald Harrington for carefully reading an earlier version of this paper and suggesting a number of improvements. Responsibility for inadequacies and errors is, of course, entirely my own.

2 Some case grammarians argue as to whether Dative (called Experiencer by Fillmore in 1971) and Benefactive constitute distinct cases or can be subsumed under Dative in the deep structure.

Lambert (1969:124) assumes that Benefactive is a variant form of Dative case, even though both Dative and Benefactive can occur together in the surface structure of (l):

(1) He gave it to her for me.
      A         O     D        Ben
(1) is a deleted form of a complex sentence, either (2) or (3):
(2) He gave it to her** for her to give to me.
      A         O     D                               D
(3) He gave it to her// for me to have.
     A          O     D           D

Marino (1970) and Cormican (1976) also reject the Benefactive as an independent case in the Proposition.

Cook (1973b:180) says:

The meanings of the Object case have been listed as affected object (patient), effected object (factitive) and complement. But these variants do not occur within the same proposition, and may be considered as allocases of the same case unit.

That is why Cook puts two of Chafe's cases, Patient and Complement, under the more general case label Object. From (1) given by Lambert, we can also assume that Benefactive is an underlying Dative and can be considered as an allocase of Dative when it occurs in the Proposition.

For a thorough discussion of the notion of Dative, see Mo (1977:76-85)

BIBLIOGRAPHY

Anderson, John M. 1971. The Grammar of Case: Toward a Localistic Theory. Cambridge Studies in Linguistics, number 4, Cambridge: Cambridge University Press.
Baer, Louis A. 1976. A Case Grammar Analysis of the English Predicate Nominal. Ph. D. dissertation, Georgetown University.
Chafe, Wallace L. 1970. Meaning and the Structure of Language. Chicago: The University of Chicago Press.
Cook, Walter A., S.J. 1972. "A Case Grammar Matrix," Languages and Linguistics: Working Papers, No. 6:15-47.
Cook, Walter, A., S.J. 1973a. "Covert Case Roles," Languages and Linguistics: Working Papers, No. 7:52-81.
Cook, Walter, A., S.J. 1973b. "Role Structures in Content Analysis," in Meaning, a common ground of Linguistics and literature. 179-187. Edited by Don L. F. Nilsen. Cedar Falls: University of Northern Iowa.
Cook, Walter A., S.J. 1974. "Case Grammar and Generative Semantics," Languages and Linguistics: Working Papers, No. 8. 1-28.
Cormican, John D. 1976. "In Defense of a Four-Case Deep Case Hypothesis," Papers in Linguistics, 9(1-2): 127-147.
Leonardas Dambriūnas, Antanas Klimas and William H. Schmalstieg. 1972. Introduction to Modern Lithuanian. New York: Brooklyn, Franciscan Fathers Press.
Fillmore, Charles J. 1968. "The Case for Case," in Universals in Linguistic Theory, Ed. by Emmon Bach and Robert Harms. New York: Holt, Rinehart and Winston.
Fillmore, Charles J. 1971. "Some Problems for Case Grammar," in Monograph Series on Languages and Linguistics, No. 24:35-56. Ed. by Richard J. O'Brien, S.J. Washington, C.D. Georgetown University Press.
Lambert, Dorothy M. 1969. The Semantic Syntax of Metaphor: A Case Grammar Analysis. Ph. D. Dissertation. University of Michigan, Ann Arbor.
Langendoen, D. Terence. 1970. Essentials of English Grammar. New York: Holt, Rinehart and Winston.
Marino, M. 1970. An Investigation into Verb Categorization in a Generative Lexicon of Early Modern English. Ph. D. Dissertation. The University of Michigan.
Mo, Chien-ching. 1977. A Case Grammar of Spoken Lithuanian. Ph. D. dissertation. The University of Rochester.
Nilsen, Don L. F. 1973. The Instrumental Case in English, The Hague: Mouton.
Stockwell, Robert, Schachter, Pau, & Partee, Barbara. 1973. Major Suntactic Structures of English. New York: Holt, Rinehart & Winston.