Assimilation, ambiguity, and the feature parsing problem

Productive phonological processes including English coronal place assimilation appear to neutralize some lexical contrasts, and thus pose problems for spoken word recognition. The current work explores two questions: (1) can strong spontaneous assimilation create lexical ambiguity, and (2) how do listeners resolve potential lexical ambiguity. Two form priming experiments explored lexical activation created by strongly assimilated, potentially ambiguous prime items. In the first experiment listeners showed selective priming for the item corresponding to the underlying form of the probe item despite its surface similarity to another word. The second experiment replicated the first, but with post-assimilation context replaced by silence. The loss of this context lead to parallel access of items corresponding to both the underlying and apparent surface forms of the prime. It is suggested that assimilated coronals simultaneously encode coronal and non-coronal place and that listeners disambiguate conflicting place cues by associating coronality with the final segment of the assimilated item, and non-coronality with the subsequent segment.