Sources of Variability Affecting Speech Perception and Spoken Word Recognition

Author: David B. Pisoni

Abstract:
This paper reviews recent studies on the perception, encoding and retention of stimulus variability in speech perception. Experiments on talker variability, speaking rate and perceptual learning provide evidence for the encoding of very fine perceptual details of the speech signal. Listeners apparently encode specific attributes of the talker's voice and speaking rate into long-term memory. The process of perceptual normalization in speech perception therefore appears to involve the encoding of specific instances or "episodes" of the stimulus input and the processing operations used in perceptual analysis. The present set of findings is consistent with non-analytic accounts of perception, memory and cognition which emphasize the contribution of episodic or exemplar-based encoding in long-term memory. The results also raise questions about the long-standing dissociation in phonetics between the linguistic and indexical properties of speech. Listeners apparently do encode and retain non-linguistic information in long-term memory about the speaker's gender, dialect, speaking rate and emotional state, attributes of speech signals that are not traditionally considered part of phonetic or lexical properties of words. The findings reported here have important implications for current theoretical accounts of how the nervous encodes speech signals and what kinds of information are stored in the mental lexicon.