Speech Motor Learning Modifies Speech Perception

Sazzad M Nasir, David J Ostry
Time: 2009-07-02  05:10 PM – 05:30 PM
Last modified: 2009-06-04


As babies learn to talk, they form a link between what they hear and what they produce. In the process of speech acquisition, there is an apparent auditory perceptual bias in that babies, very early in life, are exposed to a plethora of speech sounds that play a crucial role in guiding the speech movements they are going to make. In this way, one sensory modality (audition) that will be involved in speech production presumably dominates speech motor learning and shapes the required patterns of the other sensory input (somatosensory) that is involved in the speech learning process. However, this relationship may not be unidirectional. In fact, it may be the case that auditory perception is modified over the course of speech motor learning. In this presentation, we report a test of this hypothesis. We tested 23 subjects who performed a speech perceptual test both before and after a related speech motor learning task. As a control, another 21 subjects performed exactly the same auditory task without the intervening speech learning procedure. In the perceptual identification test, each subject listened to an auditory stimulus that was chosen at random from a 10 step continuum between head and had and was required to indicate whether the presented item sounded more like head or like had. In this way, we were able to estimate the perceptual cross-over point between head and had and any potential perceptual changes associated with motor learning. In the accompanying speech learning procedure, a robotic device was used to alter speech movements, and in particular somatosensory feedback, by delivering velocity dependent loads to the jaw during speech production. These same mechanical loads had no effect on speech acoustical patterns during learning. With training subjects corrected for load, such that the motion path and presumably the associated somatosensory input approached that normally experienced under no-load conditions. A consistent pattern of perceptual change was observed for 17 of the 23 subjects that adapted to the mechanical load. The change in perceptual threshold following the learning task was significantly different than that observed among control subjects who performed two consecutive identification tasks but in the absence of motor learning. The control/non-adapted subjects did not show any net perceptual change. Further, the amount of perceptual changes among the adapted subjects was correlated with the amount of adaptation. These results are consistent with the view that speech motor learning modifies auditory perception.

Conference System by Open Conference Systems & MohSho Interactive Multimedia