Phonetic Adaptation to Foreign-Accented Speech

Over the past few decades, there has been considerable effort to find the mechanisms through which adult listeners can accommodate the rampant phonetic variation in natural speech. My dissertation concerns one source of variability: phonetic variation in speech produced by individuals with foreign accents. Mounting evidence shows that listeners not only adapt to specific speakers by adjusting acoustic-phonetic mappings, they also sometimes generalize the remapping to novel talkers. In this dissertation, I present a series of experiments examining the mechanism of rapid phonetic adaptation and its generalization across talkers. I tested native-English listeners’ adaptation to Mandarin-accented English words, focusing on /d/ in word-final position. The first set of experiments (Experiments 1-3) investigated talker-specific adaptation. I found that perceptual learning for speech was not just a matter of adjusting phonetic boundaries in face of noncanonical tokens; it also promoted a reorganization of the internal category structure. The learning resulted in changes in cue-weighting functions that may prepare listeners for adapting to similar variation in other acoustic environments. The second set of experiments (Experiments 4, 5A and 5B) examined generalization of learning across talkers following single-talker exposure or multiple-talker exposure. Single-talker exposure failed to produce generalization to a novel talker. Following multiple-talker exposure, cross-talker generalization was evident only when the test talker (a novel talker) was acoustically similar to (one or more of) the exposure talkers. Lastly, Experiments 6 and 7 present case studies of talker-specific adaptation to foreign-accented speakers, showing a role of speaker intelligibility and within-talker variability in phonetic adaptation. In summary, the results of these experiments demonstrate that the lexically-guided phonetic reorganization mechanism that substantiates the adaptation to idiolect differences of native speakers also supports adaptation to natural foreign accents. In addition, bottom-up similarity at the acoustic-phonetic level explains a range of situations in which adaptation effects may or may not generalize to novel talkers. Taken together, the findings advance our understanding of the reorganization of the perceptual architecture that listeners experience when they adjust to unfamiliar speech