Recognition of spoken words with mispronounced lexical prosody in Japanese.

Abstract

Lexical prosody plays a crucial role in Japanese spoken word recognition. However, listeners of Japanese can still recognize spoken words easily even when they are pronounced with mispronounced lexical prosody, i.e., prosody that differs from their lexical knowledge. The present study investigated how listeners recognize spoken words pronounced with mispronounced lexical pitch accent in Japanese. Two cross-modal semantic priming experiments addressed the process of lexical access during listening to prosodically mispronounced words under a sentential context. When words were presented with mispronounced prosody, semantic priming effects that reflected the access to contextually congruent words were obtained not under interstimulus interval (ISI) = 0 ms (experiment 1) but under ISI = 750 ms (experiment 2). These results suggested that a prosodically mispronounced input temporarily disturbed access to the appropriate meaning of the word, whereas appropriate access was achieved later within 750 ms with reference to the contextual information rather than the mispronounced lexical prosodic information. Thus, contextual information overrides lexical prosodic information to recognize spoken words appropriately even when lexical prosody is mispronounced.