Abstract:
In the literature, the general assumption is that passives of unaccusative predicates, and therefore passives of passives, or double passives are cross-linguistically unavailable. Current theories of syntax and semantics rule out such constructions in various ways. The most recent advancement in this endeavor is to suggest that passivization is necessarily restricted to the Voice domain, which is only available to unergative and (di)transitive structures. However, Turkish systematically allows both passives of unaccusative predicates and double passives, which we argue to pose a serious problem to the syntactic and semantic theory because current theories are founded on the premise that such constructions are prohibited. In this thesis, we will show that passive clauses are not derived from their active counterparts. More specifically, we will suggest that passive clauses are formed with items merged from the passive domain. This domain may consist of more than one passive head, subject to different licensing conditions in a language or may not be available for independent reasons. Hence, we argue that some languages may allow passives of unaccusatives and double passives if they fulfill these conditions. Particularly, we will argue that the head not merged in the active structure may be compensated for in the passive domain. However, if a head must somehow be projected before the passive domain, its corresponding passive form cannot be merged in the passive domain because it would cause two predicates of the same semantic contribution to be present in the same structure.
