SUBJECT AND OBJECT IN TURKISH by Laura Ellen

This dissertation is a study of rules in Turkish which change grammatical relations or are sensitive to them. It addresses issues of interest to descriptive Turkish grammar and 10 general linguistic theory. Two chapters are devoted to questions about intransitive clauses. Chap' AT 2 examines the claim that impersonal passivizatlon, like personal passivization, involves the advancement of a direct object to subject. Evidence is presented that this is not the case in Turkish. Chapter 4 is an investigation of the Unaccusative Hypothesis, the proposal that some intransitive clauses have an initial direct object but no initial subject. It has been argued that there is one construction in Turkish which provides evidence for the Unaccusative Hypothesis. The control rule that operates in this construction is shown to be sensitive to thematic roles rather than to initial grammatical relations; it cannot, therefore, serve a', a diagnostic for initial unaccusativity. The topic of Chapter 3 is non-referential direct objects and subjects. Evidence is presented that a subset of such nominals, i.e., those that occur without the indefinite article, undergo incorporation with the verb, which accounts in part for the observation that sentences with non-referential subjects behave as if they were sUbjectlcss and that those with non-referential direct objects behave as 'if they were intransitive. I propose that incorporees are not final chomeurs. as has been claimed, but instead bear the final-stratum relation INC(orporated). Furthermore, I argue that sentences with incorporated subjects lack a final subject and, consequently. that the Final 1 Law is too strong. The causative construction is the subject of the final chapLer. and the cenlral question addressed is whether causative formation in Turkish is a lexical process which deri\'es one verb from another or a syntactic process which collapses clauses together (Clause Union). While the lexical account explains a class or rule interaction phenomena, I present evidence that causatives must be analyzed as underlyingly complex. A general condition is proposed which blocks syntactic rules of a particular klnd from applying on the embedded clause prior to C',~USl: Union, The discussion of causatives includes an analysis of quirky casemarking in 'furkish. Dissertation Supervisor: Dr. Kenneth Hale, Professor of Linguistics