The effect of syllabification and VOT on F2 onsets at the CV2 boundary

Abstract

The articulatory blending of vowels and consonants has long been described by a coproduction model of coarticulation [S. E. G. Ohman, J. Acoust. Soc. Am. 41, 310–319 (1967)]. Variation in intergestural timing of the C closure has been shown to produce instability in locus equation coefficients as place descriptors when simulated on the DRM vocal tract model [Chennoukh et al., J. Acoust. Soc. Am. 102, 2380–2389 (1997)]. This study attempted to achieve real-world variations in CV phasing by imposing a variety of temporal intervals separating vowel-to-vowel sequences. Variation in syllable shapes and degree of aspiration were used to alter the relative timing between V1 and V2. A coproduction account predicts little change in F2 onsets for V2 as a function of the temporal interval separating V1 and V2. In contrast, a segment-by-segment (‘‘deactivation’’) account would predict systematic changes in V2 locus frequencies, with coarticulation decreasing as the C-interval increases. Three syllable shapes (V♯CV, VC♯V, VC♯CV) were used with five vowel contexts /i,e,ae,u,ɔ/) surrounding the six stop consonants (/bdgptk/). Acoustic measures of both F2 offset frequencies for V1 as well as locus equations derived from the CV2 interface will be presented for three speakers of American English. [Work supported by NIH.]