According to the PF merge hypothesis on the formation of inflected verbs in Korean (J. H.-S. Yoon 1993, 1994, 1997, Park 1994, J.-M. Yoon 1996, among others), so-called pre-final and final verbal endings independently project at syntax, and merge with the head of the preceding phrase at PF. One consequence of this hypothesis is that a predicate, i.e., a verb stem aug- mented with inflectional endings, is not a constituent at syntax. Chung (2009a, 2011) attributes some syntactic behaviors (immobility and undelet- ability) of embedded predicates to the very non-constituent status of predicates. This paper discusses two types of apparent challenges for the PF merge hypothesis: (I) Predicates in certain constructions appear to be syntactically active; and (II) a string of elements that is defined as a constituent ? la the PF merge hypothesis appears to be syntactically inert. It will be demonstrated, however, that neither type of challenges necessarily disproves the PF-merge hypothesis: As for the type (I) challenges, there are alternative derivations available; and as for the type (II) challenges, the syntactic inertness comes from independently motivated morphological requirements.

Young-Seok Choi. 1996. On the Complement of Some Unaccusative Predicates. Studies in Modern Grammatical Theories 8: 209-225. The main concern of this paper is a constraint on the complements of certain unaccusative predicates in Korean, with the primary focus on toy `become`. It will be argued that such constructions involve raising from an initially biclausal structure, consistent with Perlmutter and Postal`s (1984) hypothesis that auxiliary verbs universally occur in an initially unaccusative structure. Some syntactic evidence will be provided in support of the proposed analysis based on facts about word order, relativization, and passivization.