Converging to a common speech code: automatic imitative and perceptuo-motor recalibration processes in speech communication

International audienceUnintentional imitation is one of the major processes by which humans smooth social interactions. Such convergent behaviors also occur during speech communication, as highlighted by our tendency to automatically 'imitate' a number of phonetic characteristics in another speaker's speech. This phonetic convergence effect is thought to facilitate conversational exchange by contributing to setting a common phonetic ground between speakers, and it provides evidence for perceptuo-motor adaptive processes during speech communication. Based on f0 and F1 acoustic analyses of speech production in various laboratory tasks, the present study aimed to better characterize sensory-to-motor adaptive processes involved in unintentional as well as voluntary speech imitation, and to test possible motor plastic changes due to global auditory-motor remapping. To this aim, three groups of participants involved in speech production or imitation tasks were exposed via loudspeakers to vowel utterances spoken by different speakers. The first task was designed to induce unintentional imitation of acoustically presented vowels and to measure the magnitude of imitative changes in speech production as well as possible motor after-effects. Participants were instructed to produce vowels according to either an orthographic or an acoustic cue, without any instructions to repeat or to imitate the acoustic cues. A block design was used where participants produced the vowel target according first to an orthographic cue (baseline), then to an acoustic cue (phonetic convergence) and finally to an orthographic cue (motor after-effect). To compare phonetic convergence and voluntary imitation of the acoustic vowels, we asked the second group of participants to imitate the acoustically presented vowels. In a third task, we tested whether motor after-effects can also occur without prior unintentional or voluntary vowel imitation but only after auditory exposure of the acoustic targets. The three tasks were performed in a soundproof room using the same experimental setting and participants' productions were recorded for offline analyses. A semi-automatic analysis of the recorded vowels (around 14'000 utterances) was performed on f0 and F1 acoustic parameters. In the first task, analyses of covariance demonstrate that producing a vowel according to an acoustic cue led to a small but significant unintentional imitation of f0 and F1 (all p's < .001) and to short-term motor plastic changes in the subsequent productions based on orthographic cues (f0: p < .008; f1: p < .01). As expected, similar results were observed in the second imitation task with a much greater magnitude of imitative and after-effect changes (all p's < .001). Finally, simple auditory exposure to vowels in the third task led to similar after-effects as observed in the first task (f0: p < .001; F1: p < .02). These results demonstrate automatic imitative processes during speech communication even at a fine-grained acoustic-phonetic level and highlight the role of ambient speech in adjusting phonemic auditory-motor goals. They provide evidence for both mirror mechanisms and feedback control mechanisms as suggested by recent neurophysiological and behavioral studies on speech perception and production