Orofacial somatosensory inputs enhance the change of speech perception due to speech motor learning
Résumé
While somatosensory inputs are usually absent during speech perception, recent studies found somatosensory influence on phonetic perception when the somatosensory inputs applied during the process of speech sounds (Ito et al., 2009; Gick & Derrick 2009). One possible reason for these alternations is that somatosensory inputs during speech motor learning serve to structure the perceptual processing of speech sounds and these effects carry over into perceptual testing when somatosensory inputs are absent. To test this hypothesis, we manipulated somatosensory inputs associated with facial skin deformation during speech motor learning and examined if this additional somatosensory input during the learning produces changes of speech perception between before and after the learning. We tested native speakers of American English. We here focused on the production and perception of /s/ and /sh/ consonants. During the speech training, the subjects were asked to repeat a task utterance ‘a shed’ while the sound /sh/ was played back with an alteration toward /s/, as in ‘a said’, by increasing the spectral centroid in real time (Shiller et al., 2009). A facial skin stretch was applied during the training. The skin was pulled in backward direction, which is an opposite to a primary direction of skin deformation for the production of training task, ‘a shed’. Prior to and immediately after the speech motor learning, perceptual tests using /s/-/sh/ continuum were carried out to associate changes in perceptual performance with the speech motor learning. The skin stretch was not delivered during the perceptual tests. We also tested the control group whose are involved in the exactly the same test besides the absent of facial skin stretch during the training. The amount of the speech motor learning changes was quantified as changes in spectral centroid of the fricative consonants. The perceptual performance were evaluated the threshold level between /s/-/sh/ sounds. The subjects in both groups showed significant compensations for the altered auditory feedback perturbation and the change of perceptual boundaries between the fricative consonants after the speech learning. We found that the changes of perceptual shift in the skin-stretched group were greater than that in the control group. This indicates that the additional somatosensory inputs during the training enhance the perceptual change induced by the speech motor learning. The result suggests that the perceptual classification of speech sounds would be linked speech production through the influence of the somatosensory system during learning.