Consonant cluster phonotactics: a perceptual approach

Abstract

This dissertation deals with deletion and epenthesis processes conditioned or constrained by the consonantal environment, essentially consonant deletion, vowel epenthesis, and vowel deletion. It is argued that the standard generative approach to these processes, which relies on the syllable and the principle of prosodic licensing, is empirically inadequate, and an alternative sequential approach based on perceptual factors is developed. It is proposed that the likelihood that a consonant deletes, triggers epenthesis, or blocks vowel deletion correlates with the quality and quantity of the auditory cues associated to it in a given context. The approach is implemented in Optimality Theory and adopts more specifically the ‘Licensing by cue’ framework developed by Steriade (1999a,c). New empirical gneralizations concerning deletion and epenthesis processes are uncovered, in particular 1) the fact that stops are more likely than other consonants to delete, trigger epenthesis, or block deletion; 2) the role of syntagmatic contrast in deletion and epenthesis processes; 3) the role of the audibility of stop release bursts; 4) the existence of cumulative edge effects, whereby more and more phonotactic combinations are licensed at the edges of prosodic domains as we go up the prosodic hierarchy. These generalizations are elucidated in terms of internal and contextual cues, modulation in the acoustic signal, and cue enhancement processes at edges of prosodic domains. The proposed perceptual approach achieves a substantial simplification and unification of the conceptual apparatus necessary to analyze deletion and epenthesis processes. It subsumes under the more general notion of perceptual salience principles of syllable well-formedness and the Obligatory Contour Principle. Furthermore, it eliminates the need for exceptional mechanisms such as extrasyllabicity at domain edges. The analysis is based on the study of deletion and epenthesis processes in a variety of languages. Detailed investigations of schwa in Parisian French, cluster simplification in Quebec French, and stop deletion and vowel epenthesis in Ondarroa Basque are provided.