A prosodic theory of laryngeal contrasts

Abstract

Current models of laryngeal licensing allow as many laryngeal contrasts within a syllable as there are segments, at least in principle. We show here that natural languages are much more economical in their use of laryngeal contrasts than segmental models would lead us to expect. Specifically, we show that voicing, aspiration and glottalisation occur at most once per onset, nucleus or coda in a given language, and that the order in which they are produced within onset, nucleus and coda is never contrastive. To account for these restrictions, we propose that laryngeal features are properties not of segments, but of the onsets, nuclei and codas that dominate them.