Not all feedback is created equally: The role of multisensory feedback in the perceptual learning of speech under a novel acoustic transformation

Abstract

Previous research has investigated the perceptual learning of speech modified by the frequency-inversion transformation, in which the spectral content of an acoustic pattern is rotated about an arbitrary frequency. The resulting transformed speech is almost completely unintelligible. However, given appropriate training, individuals can learn to perceive the linguistic content of the transformed spoken message. Recent work by Webster and Lachs has indicated that providing multisensory feedback during training appears to enhance the rate at which individuals adapt to the transformation. Furthermore, different types of multisensory feedback give rise to different patterns of learning. For example, learners presented with concurrent visual displays of a talkers lip movements (i.e., articulatory information) demonstrate better learning of training materials than learners presented with concurrent visual displays of text. The goal of the present experiment was to refine the technique for training perceivers in order to exaggerate differences between feedback groups in overall learning. A new control condition was also included to test the hypothesis that nonarticulatory but dynamic visual displays yield qualitatively different patterns of learning that those exhibited by perceivers trained with articulatory displays. The results provided interesting insights into the methodology necessary to test perceptual learning of frequency-inverted speech under multisensory conditions.