Is Compensation for Coarticulation Mediated by the Lexicon?

Abstract

Ambiguous stops between /t/ and /k/ tend to be heard as /k/ after /s/-final words and as /t/ after /∫/-final words. Elman and McClelland (1988,Journal of Memory and Language,27,143–165) obtained this compensation for coarticulation effect when the word-final fricatives were replaced with an ambiguous phoneme and argued that this indicated top-down lexical involvement in a prelexical process, as predicted by interactive models of speech perception. But autonomous models, which have no top-down processing but which are sensitive to the transitional probabilities between speech sounds, can also account for the effect. This study tested these two accounts. In Experiments 1, 2, and 3, listeners categorized fricatives at the ends of nonwords and words and the immediately following word-initial stops. When nonwords were the context stimuli, categorization of both the fricatives and the stops was influenced by the transitional probabilities of the vowels into the fricatives. When words were used, the transitional probabilities into the fricatives were matched. No compensation on following stops was found, even though the fricatives tended to be labeled in a lexically consistent manner. In Experiment 4, where listeners simply categorized stops at the ends of nonwords, further evidence of sensitivity to the transitional probabilities of these consonants was obtained. These results challenge interactive models, but are accounted for most parsimoniously by autonomous models in which transitional probabilities are represented independently of lexical information.