Xin Xie

Lexically-guided phonetic retuning helps listeners adapt to non-canonical productions produced by a given speaker (e.g., a foreign-accented speaker). Our previous work shows that when tested immediately after training on one accented talker, listeners generalized phonetic retuning only to accented talkers that were acoustically similar to the trained talker. In the present study, we tested whether sleep-mediated consolidation promotes talker generalization to a novel talker who is not acoustically similar to a trained talker in their phonetic realizations. Focusing on word-final alveolar stops, we examined native-English listeners’ 1) adaptation to a trained Mandarin-accented talker and 2) generalization to another untrained Mandarin talker. Participants were tested in a categorization task immediately after training in either the morning or evening and again after twelve hours. Morning and evening groups had comparable performance for the trained talker during the second test, suggesting maintenance of talker-specific learning. Importantly, only evening-trained participants, who had slept before the second test, gained an advantage for the novel talker, whereas morning-trained participants did not improve over the 12 hours in wake state. It is argued that sleep helped listeners to abstract away from specific acoustic properties of the trained talker and thereby facilitated generalization to the novel accented talker.