Duration, voice, and dispersion in stop contrasts from Latin to Spanish

This paper takes a new look at the evolution of stops from Latin to Spanish. In a series of sound changes that have often been described as a chain shift, geminates reduced to singletons, voiceless stops voiced, and voiced stops approximantized. These changes have generally been attributed to lenition, which I assume here. I analyze these changes in Optimality Theory as variant constraint ranking over time, couching the analysis in the terms of Flemming’s (1995) Dispersion Theory, a functionalist approach that takes whole systems of contrast into account rather than focusing on piecemeal change. I thus view the changes in question not only as the predictable effects of lenition in word-medial, intervocalic position (where they took place) but as part of a suite of changes that weakened stops in the same way while maintaining extant contrasts in the process. I focus on the phonetic cue of closure duration in addition to voicing distinctions, based on the findings of numerous researchers that duration functions crosslinguistically as a major means of marking voiceless/voiced stop contrasts. I moreover show that modifications to Dispersion Theory recognizing a more direct role for faithfulness (as in Padgett 2003) allow us to motivate constraint reranking over time as an effect of language acquisition and reanalysis of underlying forms (see Holt 2003a). Reference here to traditionally unspecified phonetic detail (e.g. stop closure duration) is argued to provide both a richer model and a fuller understanding of the nature and mechanisms of contrast underlying the changes in question. Moreover, since lenition — an often vaguely defined concept — is here shown to operate in highly similar ways across these changes, the current approach provides for a more coherent, unified analysis than those heretofore presented.