ABSTRACT
Introduction
Models of spoken word recognition can be divided into two classes: models such as the Cohort model (Marslen-Wilson, 1987) in which competition among lexical alternatives occurs in a strictly "left-to-right fashion (e.g., "casket" competes with its cohort, "castle," but not with its rhyme, "basket") and activation models - e.g., TRACE (McClelland & Elman, 1986) - which specify that competition can occur throughout words (e.g., "casket" can compete with both "castle" and "basket"). While there is evidence that cohorts compete during on-line processing, evidence for competition between rhymes is less clear.
