Volume 19, No.1 - Spring 1973
Editors of this issue: Antanas Klimas, Thomas Remeikis
Copyright © 1973 LITUANUS Foundation, Inc.

(Based on Professor Pavel Trost's pioneer work)

Pennsylvania State University

Definition of the Ergative

The problem of the ergative and its construction is of universal importance. The ergative construction consists of the following components:1

1. Noun in Nominative (or generally "absolute or unmarked" case) indicates the inactive participant of the action ("-actor, non-actor") which is obligatorily present.

2. Noun in -Nominative (i.e., another case) that indicates a supplementary partner, either marginal or facultative (not necessarily present) actor (initiator of the action).

3. Verbal form which is not a passive transformation of an active form.

If Lithuanian (and similarly Latvian) were considered an ergative language, the following constructions with the verb lūžti (Latvian lūzt) "to break (intransitively) " would be interpreted according to (I):

1. medis lūžo vėjuje, nuo šalčio/šalčių, po sniegu "the tree broke in the wind, from frost, under snow"

2. keliai lūžta kareivių "the roads are overcrowded by soldiers"

I. Prime Nouns

1. "The tree" and "the roads" have a Nominative form and express the inactive partner.

2. Nouns "wind, frost, snow; soldiers" are in -Nominative form (i.e., Locative, Instrumental, prepositional complex; Genitive) and express the active partner, not necessarily present, cf., medis lūžo "the tree broke."

3. The verbal form "breaks, broke" is "active", more exactly, it is an absolute (or unmarked) form as far as the voice concerned. The Baltic verbal forms also offer other "absolute categories," here the "absolute Number" (i.e., lack of congruence in Number).

Many territorially, genetically, and even typological-ly different languages are ergative, i.e., they possess the ergative perspective of the sentence and universally can be opposed to "objective" languages,' i.e., the languages which express the same idea with the subject in Nominative (corresponding with ergative "actor" in -Nominative) and object in -Nominative (Accusative) corresponding with the ergative "non-actor" in Nominative. Thus, the "ergative" construction:

# 1. medis lūžo vėjuje, vėju (Locative-Instrumental) would correspond (or even could be transformed into) the objective construction.

# 3. vėjas laužė (Inf. laužti/laužyti) medį "the wind broke the tree," whose passive correspondence (transformation) would be

# 4. medis buvo laužtas (laužytas) vėjų / vėjuje

II. The Problem of the Indo-European Ergative

In the last ten years, attention to the problem of IE ergative was renewed and intensified,2 especially in the Soviet Union. Scholars such as M. M. Gukhman (for Germanic), I. M. Tronskii (mostly for classical languages), T. Y. Elizarenkova (for new Indian languages) L. A. Pireyko (for Iranian languages), A. N. Savchenko and Vjach. V. Ivanov (for Hittite and proto-Indo-European)3 tried to find traces of the IE ergative and reconstruct it for the primitive stage of this proto-language. However, due to the paucity of relevant data offered by the languages analyzed, the hypothesis is far from a commonly accepted one even among "ergativists."4 Thus the postulation of the IE ergative and an ad hoc application of certain facts to it (which can be explained in a different way according to other scholars) remains only a remarkable attempt rather than a fully acceptable and workable basis for further research.

III. Pavel Trost's "Lithuanian Predicative": An Original and Decisive Approach to the Ergative

Professor Trost in his article, "Das Pradikativum im Litauischem,"5 approached the problem from a completely different position. Without postulating (or even mentioning) an ergative construction, he turned his attention to a crucial language, from his point of view, namely, Lithuanian. This language has several advantages in comparison with other IE languages. It has archaic qualities which can be compared with the archaic qualities of "classical languages" of Indo-European studies, such as Greek, Sanskrit, and Hittite. On the other hand, contemporary syndromistic linguistics considers its information considerably more valuable: as a living language, it offers an inexhaustible amount of information; its structure can be analyzed in a much more detailed manner both syn-chronically and diachronically, than that of the extinct languages which offer only limited and fragmentary information about their corpus. Thus, no ad hoc ergative constructions, but fresh Lithuanian material and insur-passable synchronic and diachronic depth of this language were the main means for the success.

Analyzing several Lithuanian constructions, Pavel Trost finds that the "classic" Lithuanian construction

# 5. Senų1 miškai2 mylėta3 "The forests2 (were) loved3 by old (people)"1 (only a very rough translation)6 does not have any proper correspondence in other IE languages (p. 221). He observes that

(II) 1. The Nominative subject is the goal, not the point of departure.

2. The supplement (the noun in Genitive, i.e., -Nominative) is the actor ("agens") which, however, is not in the center of the sentence, but takes a rather marginal position ("Randstellung") and finally

3a. The predicate has a Neuter (i.e., absolute!) form.

3b. The passivity of the verb is irrelevant, (cf. eine... "Verbalform, die sowohl passiv als auch nicht passiv verwendet werden kann" p. 221).

Thus, P. Trost's deep and ingenious analysis as presented in (II), remarkably close to the ergative definition (I) .was one of the main sources for postulation of the ergative in Baltic and in Lithuanian particularly.

IV. Lithuanian Structure Confirms Baltic Ergative

Further investigation confirms that the ergative constructions that can be only hypothetically postulated in other IE languages, e.g., in Hittite, is actually very productive in modern Lithuanian. Let us consider the sentence

# 6. šeimininkės gimė sūnus "the housewife (Gen.) bore a son (Nom.)"; again in a very rough translation, with an English transitive verb replacing the Lithuanian intransitive one, a better, but not fully exact equivalent is available in Slavic languages cf. Russian u xozjajki (possessive form) rodilsja syn (other Slavic languages, such as Polish, Czech, Old Church Slavic would use Dative instead of the possessive construction).

The literary I.E., "objective" translation of the English objective sentence in 6 would be

# 7. šeimininkė (Nom.) gimdė sūnų (Ac.)

It is obvious that #7 is a transformation of #6 and not vice versa, since the verb in #7 gimdyti "to bear" is a causative (transitive) transformation of the verb in #6 gimti "to be born," zur Welt kommen (cf. exactly opposite of English, where "to be born" is an obvious transformation of "to bear"). There are, of course, other reasons for this interpretation in Lithuanian; one of the most important is the fact of the order of transformation.

Let us start from the obvious fact that both #6 and #7 are generated from a more basic structure. Since both sentences are personal, the closest basic structure must possess a subject (Nominative) and predicate so that the noun in -Nominative is the component which was newly applied. Thus for #6 we receive the following structure:

#8. gimė sūnus "a son was born", which is a grammatical sentence, while

#9. šeimininkė gimdė... iš ungrammatical or (much less usually) generated from #7. Thus, the order of the transformations must be

# 8 > # 6 > # 7 (> *# 9)                                      (1) 

and cannot be

#(#9 > #7 > #6...)                                              (1*) 

so that according to (1)

#6 > #7                                                            (1!)


Now, if we proceed from the generally accepted thesis that the transformationally earlier structure is deeper and that the synchronic depth is isomorphic with the diachronic depth, we obtain the following basic results:

(III) 1. Synchronically the (1!) proves that the deep structure of an objective construction (#7) is the ergative construction (#6).

2. Diachronically (1) proves that the development of the objective structure was the following: (a.) structure # 6 is first, (b.) structure # 7 (ergative is immediately generated from # 6), and (c.) structure # 8 was generated from # 7, so that in the proto-language from which Lithuanian developed, the objective structure was generated from an ergative. Since, however, the objective structure is obviously common IE (cf. common IE form of Accusative) the ergative structure is pre- (common IE), which stage can be indicated as IE*.

V. Indo-European and the Universal Importance of the Baltic Ergative

Consequently, the existence of ergative is not a controversial hypothesis, but a fact that shall be considered established and that, being a new firm point in the stage IE* serves as a clue to solution for several principal problems of the following stages: (1) The discovery of the IE ergative opens the door for dependable reconstructions of basic syntactic, morpho-syntactic, and paradigmatic models in IE*, (2) The original ergative structure must be postulated for other IE languages, which will cast light on their basic syntactic structures. Thus German scholars are puzzled by the development of the Genitive in Old German and its contradictory function indicating a noun that is both "object" and "motive" of the action, e.g., ziu sorget ir thanne thes anderes "why do you care for this," hungeres isterben "to die of hunger" and also thaz sie... thes wazzares giholoti "(they told her) to bring the water," cf. Lithuanian kentėti bado "to suffer from hunger" eiti vandens "to go (to get) water"7 the closest relative of Baltic.

3. Slavic, even in its present stage, preserved many ergative features. The fact that the objective (cf. Accusative) structures are established (which is true also in Baltic) does not exclude the possibility that these structures are at least partially generated from the ergative ones, and also, that the Slavic can be mixed -ergative -objective type, analogous to Georgian. The recent investigation of reflexives proved that the postulation of a purely objective type leads to unexpected paradoxes8 that can be eliminated easily by introducing the ergative category

4. It will also solve such controversial problems as the question of s.c. Balto-Slavic unity and the position of Old Prussian in the Balto-Slavo-Germanic area,9 replacing fragmentary evidence (mostly phonetic, phonemic, or lexical) with a systemic morpho-syntactic research, which is crucial for such an investigation.10

5. The results also prove that the common practice of considering the "objective" structure as a universal one (at least in European languages) leads to serious discrepancies. It is necessary to consider the ergative structure as a universal complement of the objective structure. Later investigation will prove whether these two structures can be reduced to one. This study indicates that the ergative structure has great changes to be considered deeper and consequently universal in the deep structure, while the objective construction would be only its superficial representation.

The ideas presented here would hardly be possible in this form without Professor Trost's remarkable work on this problem. Considering that this is only one of the areas in which his contribution was of crucial importance, we have a right to state that Professor Pavel Trost is one of the most original, brilliant, but nevertheless, modest scholars in his field.


1 For a good initial explanation see, e.g., J. Lyons: Introduction to Theoretical Linguistics, Cambridge 1968, esp. section 8.2, Transitivity and Ergativity pp. 350 - 371.
2 The hypothesis of the IE ergative was first brought up by H. Uhlenbeck in his article "Agens und Patiens im Kasussystem der Indogermanischen Sprachen," Indogermanische Forschungen XII, in 1901 and repeated several times, cf. A. N. Savčenko: "Ergativnaja kontrukcija predloženija v praindo-evropejskom jazyke," in Ergativnaja kontrukcija predloženija v jazykax rezlicnyx tipov, Moscow 1967, p. 81, for recent bibliography cf. Vjač. V. Ivanov: Obščeindoevropejskaja, praslavjanskaja i anatolijskaja jazykovye sistemy, Moscow, 1965.
3 In the two books indicated in the preceding footnote. 
4 The main defender of the IE ergative is A. N. Savčenko, Vjač. V. Ivanov just talks about a "hypothetical period," other scholars, especially A. A. Guxman are rather skeptical, 
5 cf. his Bibliography, 1960).
6 cf. J. Jablonskis: Rinktiniai raštai, I tomas, Vilnius, 1957, 571, cf. who calls this kind of Genitive, Genitive of origin (kilmės kilmininkas), cf. also J. Balkevičius: Dabartinės lietuvių kalbos sintaksė, Vilnius, 1963, 78, 110 passim.
7 For German, cf. e.g., V. G. Admoni: "Razvitije funkcij rodi-telnogo padeža v nemeckom jazyke," in Trudy institutą jazy-koznanija IX, Moscow, 1959, with extensive literature.
8 cf. esp. remarkable works of G. H. Schaarschmidt, e.g., in his "Passive and Pseudopassive Construction in Russian," Scan-doslavica XVII (1971) with older literature.
9 cf. J. Marvan: "Deciphering the Old Prussian Message" in The Proceedings of the Third Conference on Baltic Studies, Toronto, May 1972 (forthcoming).
10 cf. e.g., Antanas Klimas: "Baltic, Germanic and Slavic," Donum Balticum, Stockholm, 1970, 263 - 269 esp. on p. 266: "in reference to change, the most tenacious features in any language are morphological syntactical, perhaps even the syntactic-morphological structures. Only then should the phonological features be considered," The only scholar who, to my knowledge, fulfilled this requirement was Professor Trost who emphasized, in a series of articles in 1958 (cf. Bibliography), the Balto-Slavic syntactic isoglosses and their importance for the problem of the Balto-Slavic unity.