Temporally organized lexical representations as phonological units

In traditional generative phonology, there is no role for the mechanism of lexical access as a part of linguistic competence. Lexical access is assumed to be relevant only to aspects of linguistic performance such as speech production and perception. In fact, much of what is known about lexical access comes from the study of errors in production and perception. Errors are traditionally considered part of linguistic performance, as competence is assumed to be error free. This paper has two distinct, but related goals. The first goal is to demonstrate that lexical processing does influence linguistic competence, by showing that the beginning-to-end temporal order in which the segments in a word are processed is a functional influence on the phonology. More generally, I claim that cognitive factors influence phonology, and may be the source of language particular constraints or universal linguistic tendencies. In general, cognitive influences should not be ignored or factored out of linguistic theories, rather they should be recognized as sources of explanation for linguistic patterns. The second goal is to investigate a model of gradient constraint combination in the phonotactics of the verbal roots of Arabic, based on the stochastic constraint model (Frisch, Broe, & Pierrehumbert, 1997). It will be shown that the Arabic verbal roots obey a gradient consonant cooccurrence constraint, OCP-Place, which is a dissimilarity constraint that applies to homorganic consonants (McCarthy, 1994; Pierrehumbert, 1993). This