In lieu of an abstract, here is a brief excerpt of the content:

GEORGIAN AND THE UNACCUSATIVE HYPOTHESIS Alice C. Harris Vanderbilt University It has been observed that many languages exhibit a semantic, syntactic, or morphological correlation between the direct objects of transitive verbs and the surface subjects of certain inactive intransitives. The Unaccusative Hypothesis (UH) proposes, within the framework of Relational Grammar, that final subjects of this type are initial direct objects. This paper shows that several morphological and syntactic processes in Georgian refer to just these nomináis, thus supporting the UH. It further shows that, although the semantic relations of nomináis to verbs is not irrelevant, those morphological and syntactic processes which refer to initial direct objects cannot be stated simply on the basis of semantics.* A wide variety of morphological and syntactic data in diverse languages indicate a correlation between, on the one hand, direct objects of transitive verbs and subjects of certain kinds of intransitive verbs, and, on the other hand, between subjects of transitive verbs and subjects of certain intransitives. These semantic correlations and oppositions have figured in work by Fillmore (e.g. 1968, esp. p. 54), Chafe (1970a, esp. Chap. 9, and 1970b) and others.1 For example, it has been observed that the cake in Mother baked the cake and in The cake baked has the same semantic relation to bake. In contrast, mother has the same semantic relation in the first sentence as in Mother baked. Perlmutter and Postal have suggested, in recent work, that nomináis like the cake not only stand in the same semantic relation to the verbs here, but also bear the same initial grammatical relation, namely direct object. This idea has become known as the Unaccusative Hypothesis (UH; Perlmutter 1978, Perlmutter & Postal, ms). The purpose of this paper is to examine evidence from Georgian that is relevant to the UH. It is shown here that direct objects of transitives, together with subjects of certain intransitives, constitute a natural category to which the rules of Georgian grammar must refer. In addition, it is established that the relevant rules of Georgian cannot be stated on the basis of simple semantic relations, but must refer to the syntactic notions 'subject' and 'direct object'. 1. The Unaccusative Hypothesis. Sapir 1917 proposed a typology of casemarking types, partially represented as Table 1 (with the addition of more recent nomenclature). Although Sapir (p. 85) made a point of distinguishing the active from the ergative type, these types continue to be conflated in many works. For example, * I am grateful to my Georgian friends, colleagues, and consultants for the examples and insights they have provided. I wish also to thank David Perlmutter for discussing the UH with me at length, and Dee Ann Holisky and Pamela Munro for valuable criticism of an earlier version of this paper. This material is based upon work supported by The International Research and Exchanges Board, by the Sinclair Kennedy Traveling Fellowship, and by the National Science Foundation under Grant No. BNS-7923452 to Vanderbilt University. 1 Klimov has also written on these notions (e.g. 1976, 1977); however, his analysis is entirely different from that presented here. 290 GEORGIAN AND THE UNACCUSATIVE HYPOTHESIS291 object:subject: intransitivesubject: transitive inactiveactive transitive ErgativeA AB ActiveAABB AccusativeA BB Table 1. Dixon (1979:82) characterizes the active type as an irregular variant of the ergative, or as a 'grammatically untidy' ergative. Though Sapir did not go into detail about the active type, we may infer something about its nature from Dakota, the language he cited as its archetype. It has often been said that Dakota and other Siouan languages distinguish between stative and dynamic verbs. In Lakhota, a Dakota dialect (Van Valin 1977), sentences like 1-3 suggest that wa marks 1st person singular in dynamic verbs, while ma marks the same person and number in statives: (1)WAkte'? kill it.' (2)thawA'sose ? spit.' (3)MAha'ske'Iam tall.' However, compare this example: (4)MAhl'xpaye ? fall down.' This shows that ma is used not only by statives, but also by intransitive dynamic verbs that are non-agentive, involuntary, or non-controllable. Recent treatments of the syntax of Dakota and other Siouan languages have observed that the walma dichotomy is not dynamic/stative...

pdf

Share