The challenge of learning new highly variable prosodic categories: Training English listeners on pitch accent in Tokyo Japanese.

Abstract

The high‐variability multi‐talker paradigm has been successful for perceptual training of segments and tones. Here, it is extended to training native English listeners on Tokyo Japanese pitch accent using multiple lexical items, talkers, and sentential contexts. Native English listeners were randomly assigned to control or training groups. Their test task was to categorize pitch patterns of two syllable words as first syllable accented, second syllable accented, or unaccented. The same 3AFC identification task, but with feedback, was used in three training sessions (one hour total duration) over a course of one week. The identification accuracy scores of Japanese native speakers and the control group were used to assess the trainees’ improvement. Preliminary results showed a small effect of training that was also generalized to new stimuli in new sentential contexts spoken by a familiar talker. The non‐native listeners tended to misidentify the unaccented pattern as second syllable accented, and the nativ...