The Grammar of Coartiulation
Monday, 26 October 2009, Colloquium
Abstract
It is well established that phonetic details such as patterns of coarticulation are language-specific and therefore must be specified in the grammars of languages, but there has been comparatively little work on the form of phonetic grammars. I will use case studies of coarticulatory phenomena to argue for a model of phonetic grammar based on weighted constraints. More specifically, I will show that many coarticulatory patterns can be analyzed as arising from a trade-off between effort minimization constraints and constraints requiring the faithful realization of perceptual targets for speech sounds. Effort minimization constraints generally favor longer, slower transitions between segments, whereas faithfulness constraints disfavor the deviations from targets that occur during these coarticulatory transitions. The trade-off between these conflicting constraints is determined by their relative weights.
Within a language, much contextual variation in the magnitude and temporal extent of coarticulation derives from differences in the perceptual consequences of coarticulation - e.g. in English there is substantial anticipatory velum lowering preceding a nasal when the preceding sound is a vowel but much less when the preceding sound is a stop because the acoustic consequences of significant velum lowering during a stop would be much greater.
Across languages, variation in patterns of coarticulation derives from different weightings of the same basic constraints, resulting in common or universal cross-linguistic patterns of coarticulation with language-specific variation in the details. We will explore the extent to which differences in constraint weighting correlate with other phonological differences - e.g. there is some evidence that faithfulness to vowel targets has greater weight where the vowel space is more crowded.