Intonational cues to prosodic boundary influence perception of contrastive vowel length in Tokyo Japanese

We designed two experiments to test how listeners are sensitive to intonational structure in speech perception. Specifically, we tested how phrasal position, cued by pitch in a carrier phrase, mediated Tokyo Japanese listeners' perception of contrastive vowel length. We predicted that when tonal cues signal a target as phrase-final, listeners should expect it to be lengthened due to phrase-final lengthening, effectively requiring longer vowel durations for a phonemically long vowel percept in phrase-final position. We tested this in two experiments, one with an accented target word which contrasted phonemically in the length of the word-final vowel (Experiment 1, shi'sho “librarian” vs. shi'shoo “master”; Experiment 2: dookyo “housemate” vs. dookyoo “townmate”). We placed this target word in a carrier phrase, and manipulated contextual pitch to signal it either as Intonation Phrase (IP) final, or medial. As predicted, a phrase-final target required significantly longer vowel duration to be perceived as phonemically long. The results thus highlight the importance of intonational structure as a mediating factor in listeners' processing of temporal cues in speech. © 2020 International Speech Communications Association. All rights reserved.