The perception of sequences of identical and different stops in natural and synthetic speech

Abstract

Previous studies using synthetic nonsense syllables have shown that listeners require a longer closure interval to report hearing a sequence of two identical stops (e.g., /ɛddɛ/) than a sequence of two different stops (e.g., /ɛbdɛ/). Acoustic measurements of real words in natural speech do not reveal an analogous difference in production. That is, speakers produce closure intervals in utterances such as paid Daley and paid Bailey that are essentially of the same duration. A series of experiments using natural speech is reported in which comparisons are made between the perception of (a) identical stop sequences in nonsense syllables versus real words, and (b) identical versus different stop sequences in real words. The results indicate that listeners do not require a greater closure duration to report hearing sequences of identical stops in either real words or nonsense syllables. This suggests that natural speech contains cues, not included in the synthetic syllables previously used, which inform the listener that two identical stops are intended. [Work supported by NICHD.]