On information structure and clausal architecture: Evidence from Bulgarian

Date of Completion

January 2004


Language, Linguistics




In this thesis I look at several constructions in Bulgarian, all of which involve A-bar movement, and show that they have important implications for the structure of the left periphery of the clause and the syntax/phonology interface. Empirically, the thesis focuses on multiple wh-fronting (a phenomenon, found in a number of languages whereby all wh-phrases are fronted to the beginning of the clause) and its interaction with topicalization and focalization. Another core interest is the notorious long head movement construction. ^ I argue that clausal information structure is syntactically determined; specifically, the constructions under investigation provide strong evidence for Uriagereka's (1995a,b) hypothesis that language has a special discourse-related projection dedicated to the encoding of what he calls the point-of-view of the speaker. In particular, I investigate the structural representation of information-theoretic notions such as topic and focus, and argue that they are licensed in a single functional projection. I also investigate the division of labor between syntactic structure and intonation in determining discourse information. ^ The study presents new empirical data showing that while fronted wh-phrases in Bulgarian often appear to cluster they do not necessarily form a constituent. Based on the penetrability of the wh-cluster, I motivate a dialect split regarding their standard representation. As a result, I propose a modification of Bošković's (2002) treatment of multiple wh-fronting as epiphenomenal: focus fronting and wh-movement are licensed in different projections. ^ Focus fronting is shown to be relevant to the long head movement construction, which I analyze in terms of predicate clefting. I argue that heads undergo A-bar movement with detectable semantic effects but obeying stricter locality conditions than phrases. Thus I refute the long head movement hypothesis and endorse the standard treatment of head movement. I further relate the proposed analysis to the limited availability of V(P)-fronting and VP-ellipsis in Bulgarian, a fact which remained previously unnoticed. ^ As a result, I reach a number of conclusions of broader theoretical significance concerning multiple feature checking, head movement, adjunction and excorporation. On many occasions I appeal to the activation of lower copies of movement in non-trivial chains, thereby presenting supporting evidence for Franks' (1998) pronounce-a-copy hypothesis. This, in turn, presents an argument for the copy theory of movement as revived in Chomsky 1995. The analyses advocated here fall broadly into what Bošković (2001) calls “the weak phonology approach to the syntax/phonology interface”. As a result, I provide further support for a restrictive derivational model on which syntax feeds phonology and movement is excluded from PF. I also show that interface effects do not justify lookahead from syntax to PF or LF. ^