A top-down analysis for reconstruction

In this paper, I argue for a top-down derivational approach to reconstruction facts. I defend two related proposals to account for some puzzling and unresolved data on reconstruction regarding resumption, relativization and cyclicity. First, I argue for a syntactic top-down derivation in the sense of Phillips (1996) and Richards (1999). Second, I argue that the assignment of semantic values to any referential expression throughout the interpretation procedure also operates derivationally and in parallel with the syntactic derivation. This analysis will be supported mainly by data from Breton, but also from Welsh (Rouveret, 2002), Italian (Bianchi, 1995), Lebanese Arabic (Aoun et al., 2001), and English (Fox, 2000).