A task dynamic approach to the coda-voicing effect on vowel duration

The coda-voicing effect on vowel duration (CVE), in which vowels are longer with voiced codas and shorter with voiceless, is well attested. However, the precise mechanism of this relationship is not well studied. Task-dynamic models such as articulatory phonology (e.g., Browman & Goldstein, 1986) offer two ways to affect the duration of a vowel in a CVC syllable: the relative phasing of the onset, vowel, and coda gestures (when they start relative to one another), and the relative stiffness of an individual gesture (roughly how quickly it is executed). Onosson (2010) argued that for /ai/, the principal mechanism of Canadian Raising (CR) was vowel shortening via increased overlap of the onset gesture and the vowel, rather than vowel-coda phasing or stiffening the vowel gesture. This study investigates whether this explanation holds generally for 15 vowels in Canadian English, using controlled wordlist-style data (originally from Hagiwara, 2006). Preliminary investigation suggests that onset-phasing may hold for the diphthongs, but does not adequately characterize CVE across the vowels. The observable mechanism(s) of CVE will be discussed, as well as implications for CVE and related effects (e.g., CR) across dialects/languages generally.