Several studies have reported that the duration of a silent gap affects listeners' decision in identifying an auditory stimulus as rabid or rapid. It appears to be accepted that silent gap duration is a cue to stop voicing. Several implications of this asserted connection between silent gap duration and the phonetic feature of voicing deserve some discussion. First of all, since the voicing feature is commonly said to distinguish the two phoneme sets/bdg/ and /ptk/, we should like some assurance that silent gap duration operates independently of stop place of articulation. Data exists which indicate that the effectiveness of silent gap duration is far from uniform for /b/‐/p/, /d/‐/t/, and /g/‐/k/. Secondly, if a short silent gap elicits rabid responses, and /b/ is said to be voiced, i.e., characterized by glottal signal during closure, then we might expect listeners not to distinguish between presence and absence of such signal. In fact listeners can detect this difference, and some can indeed give it a phonetic interpretation. Thirdly, we may enquire whether the variation in silent gap duration needed to effect a shift in linguistic identification falls within the range observed in natural speech. A comparison of experimentally determined category boundaries with measurements of natural speech shows that the connection is not always close. [The support of the National Institute of Child Health and Human Development is gratefully acknowledged.]

This content is only available via PDF.