Production of Parisian stops word‐initial stops in the vicinity of a voiceless stop

This study examines how French speakers produced stop voicing contrasts in the sentence répète CVC tout‐de‐suite. Durations (sentence, vowel before and after word‐initial stops, closure voicing, VOT) and number of produced stop closures (/t/, /C/) were collected. There was a strong positive correlation between speaker’s sentence durations and number of closure produced: Rapid speakers produced mostly one closure, whereas slow speakers generally produced two closures. One‐closure and two‐closure productions were separately analyzed. In the two‐closure productions, significant (longer voicing in the /t/ closure in anticipation of /bdg/ than /ptk/, longer closure voicing for /bdg/ than for /ptk/, longer vowel duration following /bdg/ than /ptk/) and nonsignificant voicing‐related duration differences (e.g., closure) were found. Unlike in the two‐closure productions, average closure duration was significantly longer before /ptk/ than before /bdg/, while there was no significant voicing‐conditioned effect of the word‐initial stop on the duration of the following vowel in the one‐closure productions. As in the two‐closure productions, average closure voicing duration remained significantly longer before /bdg/ than before /ptk/. Findings suggest that closure voicing remains the most salient cue to the voicing distinction irrespective of contextual and speaking rate variations.