ABSTRACT
INTRODUCTION Speech offers a viable domain for developing and testing models of perception, pattern recognition, and categorisation. There has been a tradition of fairly elaborate theories of speech perception (for recent reviews, see Jusczyk, 1986; Klatt, 1989; Massaro, 1989). Some of these theories make little direct contact with experimental results, however, and fall outside the mainstream of psychological inquiry. In addition, some theories treat speech as a unique phenomenon and their properties have very little generality beyond speech itself. Different classes of computational models, on the other hand, turn out to have general value and can be formulated to address experimental results directly.
