This dissertation reexamines and reanalyzes the extensively studied passive voice system in Japanese within a current version of Generative Grammar—Cartographic Minimalism. Contrary to the standard assumption that Japanese passives consist of (at least) two distinct types of passives, direct and indirect (gapless), this dissertation motivates a unified movement analysis and extends Collins' (2005) smuggling analysis to Japanese passives. It is shown here that a unified movement account is not only theoretically desirable but also feasible and independently supported.

The dissertation establishes the following: (i) the dichotomy between direct and indirect passives is not only unnecessary but also empirically inadequate, (ii) the derivation of all passives involves movement, (iii) the passive morpheme -rare never assigns a &thetas;-role, and (iv) case-markers and postpositions disappear under movement, in both relativization and passives. The last property in Japanese makes it difficult to identify the source position of the derived subject in the passive. It is shown here that the derived subject always originates in the complement domain of - rare, and corresponds to an accusative, dative, genitive, or oblique source in the active counterpart.

The analysis pursued here is a modular one where interactions among the lexical properties of the morpheme -(r)are, independently-motivated principles of Universal Grammar, and the derivational path taken by the DP occupying the nominative position together give rise to different clusters of properties observed with different passive types. The dissertation also addresses the issues of interspeaker variability, the requirement of supportive context, and their implications for differences in individual grammars.