sC Clusters are (almost always) coda-initial

This article defends the position that syllables have internal structure, through an examination of sC clusters. Although perceptual factors will be shown to account for why it is sibilants that pattern in unexpected ways in clusters, it will be argued that the behavior of sC clusters cannot be explained solely by functional considerations. Among structural approaches to the syllable, it is argued that sC clusters are best analyzed as coda+onset, not as appendix+onset. The typological patterns of sC cluster well-formedness on the sonority dimension and sC cluster repair are shown to follow only from a coda analysis of s: the patterns follow from constraints on syllable contact. In view of this, it will be shown that the two most commonly defended options for the organization of s as an appendix, the syllable and the prosodic word, can be straightforwardly captured under a coda approach, through a comparative examination of English and Italian. It will further be shown that the distribution of aspiration in English is amenable to a coda analysis of s. Finally, it is argued that some languages require an analysis of sC other than coda+onset. This situation holds in Acoma: an empty nucleus interrupts putative sC clusters in this language.