Dr. Michael Tanenhaus: "Fine-grained phonetic detail in spoken word recognition."

Dr. Michael Tanenhaus is Beverly Petterson Bishop and Charles W. Bishop Professor of Brain & Cognitive Sciences and Linguistics at the University of Rochester.  He is also Director of the Center for Language Sciences

Abstract
Despite considerable evidence to the contrary, it is widely assumed that some classes of speech sounds are perceived categorically in a way that exemplars from other types of non-speech categories are not.  Yet, the articulation of many sounds, including consonants, varies systematically with position in a prosodic domain.  A system that discarded sub-phonetic detail would thus be ignoring potentially useful information. I'll review recent data from eye-tracking studies demonstrating that spoken word recognition does, in fact, exploit fine-grained sub-phonetic detail to make probabilistic hypothesis about lexical candidates, including within-category variation for stop consonants--once the poster child for categorical perception.  I'll conclude by presenting evidence from perceptual learning studies suggesting that listeners might make optimal use of the distributional information provided by within-category variation.