Influence of tempo on the closure interval cue to the voicing and place of intervocalic stops

Abstract

The silent closure interval (CI) of medial stops is known to effectively cue their phonological voicing, so that ruby converts to rupee by splicing in a longer and silent CI [L. Lisker, Language 33, 42–49 (1957)]. In addition, Dorman et al. [J. Acoust. Soc. Am. 57, Suppl. No. 1, S48(A) (1975)] demonstrated that a short enough CI for synthetic double medial stops like -bd- will produce perception of a simple flapped -d-, overriding the F2 cue for a labial implosion. We investigated the sensitivity to speaking tempo of both these effects. By varying the silent CI of the medial stop in rabid from 10 to 120 msec, a continuum of 12 words was created, from ratted to rabid to rapid. Embedded in Fast and Slow carrier sentences and identified by 16 Ss, the Slow /b-p/ boundary at 75 msec shortened about 10 msec in the Fast condition. The Slow /b/-flap boundary at 35 msec shifted only 3 msec in the Fast condition. These results (1) replicate the Dorman et al. effect with natural speech even with labial cues on both implosion and explosion, (2) show that CI as a voicing cue is sensitive to speaking tempo, and (3) suggest that the boundary at which CI overrides other place cues is less sensitive to tempo.