Sensori-motor adaptation of speech production to opposing changes in auditory feedback

International audienceThe speech motor system responds to real-time changes in auditory feedback by changing the output sound pattern in a direction opposite to the perturbation [J.F. Houde and M. I. Jordan, Science 279, 1215]. The present study evaluates the hypothesis that this sensori-motor adaptation in speech is largely local by applying opposite real-time perturbations in auditory feedback during the production of the vowels ɛ in /head/ and æ in /had/. The auditory perturbation was applied in real-time during repetitions of the words /head/ and /had/ by altering the first formant (F1) frequency using a TC-Helicon Voice-One device. Subjects read the test utterances aloud as they were presented on a computer screen and received altered auditory feedback through headphones. The words were displayed in a random order and the parameters of the Voice-One were dynamically changed via a midi-connection. For half of the participants, F1 was decreased for /head/ and increased for /had/, increasing the acoustical contrast between the two vowels. For the other half, the perturbation was the opposite, reducing the contrast between the two vowels. After 60 baseline repetitions (30 /head/ and 30 /had/) with normal auditory feedback, the shift was introduced progressively over five blocks of 20 trials each (10 /head/ and 10 /had/) and then maintained for six further blocks of 60 trials each (30 /head/ and 30 /had/). The altered auditory feedback was finally removed in order to measure possible after-effects to the perturbation. The analyses compared formant frequencies during baseline trials with the formant frequencies in trials with altered feedback. When the acoustical contrast was reduced, subjects' productions showed clear changes in /head/ and /had/ in opposite directions in the F1/F2 space, such that, subjects compensate for the reduction of the phonetic contrast by changing F1 for each word in an opposite direction so as to restore the contrast. This adaptation showed a strong after-effect when the altered feedback was removed. In comparison, when the acoustical difference between the two words was increased, subjects responded by decreasing the distance between /head/ and /had/ in F1/F2 space. This initial study suggests that people are able to adapt to two opposing auditory perturbations in a fashion that is specific to the individual training utterances. The results are consistent with the hypothesis of local sensori-motor learning in speech