Peggy Mok

PhD thesis: Influences on vowel-to-vowel coarticulation

Department of Linguistics, University of Cambridge, 2007.

This thesis investigates two quasi-independent types of influences on V-to-V coarticulation: those stemming from the phonological structure of the language and those stemming more from phonetic or physical considerations. It has been proposed that the vowel inventory of a language determines degree of V-to-V coarticulation, but results of an experiment on Cantonese and Beijing Mandarin disconfirm this hypothesis. I instead propose that language-specific realisations of syllable structure can affect degrees of V-to-V coarticulation. Results of an experiment on Cantonese, Thai and English support the hypothesis that languages with simple syllable structure allow less V-to-V coarticulation than languages with complex syllable structure.

Four influences related to phonetic considerations are examined: 1) linguistic stress (in Cantonese and Mandarin), 2) duration (speaking rate in Cantonese and Mandarin, vowel length in Thai), 3) vowel qualities (in Thai) and 4) syllabification (in Cantonese, Thai and English). Studies based on languages with lexical stress show that stress can affect coarticulation, but its effect on tone languages without lexical stress is unclear. Results show that contrastive stress has a small effect on coarticulation in Cantonese and Mandarin. Based on the idea of target undershoot, a shorter duration caused by rate or length difference can induce more coarticulation, but little research has been done to explore its effects on V-to-V coarticulation. Results show that a shorter duration does not induce more coarticulation in Cantonese, Mandarin and Thai. Many studies show that the vowel /a/ allows more V-to-V coarticulation than /i/ and /u/ cross-linguistically, but the principles behind this pattern and the susceptibility of other vowel qualities to coarticulation are still unknown. I propose that the jaw position is responsible for vowel quality differences in V-to-V coarticulation. Finally, many studies show that onset consonants are stronger acoustically, more stable articulatorily, more prominent perceptually and more frequent typologically than coda. This suggests that syllabification can affect coarticulation: onsets should be more resistant to V-to-V coarticulation than codas. My data on Thai (singleton) and English (clusters) show that coda is indeed more transparent to V-to-V coarticulation than onset.

The thesis also discusses some important methodological issues in measuring V-to-V coarticulation.

Back to home