Frequency effects on French liaison

A mainstay of the debates concerning the phonology-syntax interface are phenomena of external sandhi, that is, phonological alternations whose conditioning environment is across a word boundary. 1A recurrent problem in this area is the fact that it is usually impossible to motivate a purely syntactic account of such alternations. This has led to the widespread consensus that the relation between syntax and phonology is indirect and often seemingly arbitrary (Inkelas and Zec 1995, Nespor and Vogel 1986, Vogel and Kenesei 1990). In this paper, I propose a solution to the problem of predicting the sites for external sandhi, and I pose certain questions that are seldom raised in such discussions, in particular, how is it that syntactic structures can condition phonological alternations, and how is it that such alternations can develop and be maintained?