San Diego Linguistic Papers

This work argues against the claim that a structural distinction is the necessary source of the divergent patterns of behavior attested in /s/- and non-/s/ consonant sequences. Previous treatments of linguistic phenomena as wide-ranging and manifold as Sanskrit reduplication (Steriade 1988), Italian allomorphy (Davis 1990), and the acquisition of English word-initial clusters (Barlow 2001) have all converged on the assumption that /s/-sequences require structural representations that are different from those assigned to other consonant clusters. In each of these cases however, I show that analyses of the data that do not assume a structural distinction are simpler, and either have more explanatory power, or generate more accurate predictions.