Article révisé par les pairs
Résumé : Previous studies on the perception of French stop consonants in isolated utterances have demonstrated that the timing relationship between the onset/offet of voice and the release of the closure provides a very reliable acoustic criterion to separate voiced from voiceless stops, and a major cue for the perception of the voicing feature. The aim of this work is to examine stops pronounced spontaneously during a conversation. Experiment 1, an acoustic analysis of spontaneous productions, largely confirms the high reliability of voice timing cues but also shows that they can be occasionally misleading. Experiment 2, where subjects were asked to identify stops contained in excerpts from the conversation, confirms the major part played by voice timing in the perception of voicing. It also suggests that the reliability of the voice timing cues is not optimally exploited by the perceptual system. Experiment 3 and a control condition, where subjects were asked to write down the contents of excerpts varying in duration, demonstrate that the supplementary information that allows correct identification in the frame of the conversation is provided by top-down processes. These experiments also suggest that secondary acoustic cues play a decisive part in the case of conflict between voice timing cues and top-down information. © 1989, SAGE Publications. All rights reserved.