Speaker
Description
Introduction: Understanding speech in everyday life often requires adapting to unfamiliar speakers, a process known as perceptual adaptation. This adaptation is not effortless and likely draws on cognitive resources that are also needed for other aspects of language comprehension. Prominent models of language comprehension assume that listeners predict linguistic information at various levels (e.g., phonological, lexical, semantic). However, it remains unclear how predictive processes operate under heightened cognitive demands. This study explores how talker variability affects the brain’s ability to generate predictions during natural speech comprehension.
Methods: Thirty native Italian speakers listened to two stories while EEG data were recorded. In the single-talker condition, one speaker narrated the entire story; in the multi-talker condition, different speakers narrated different sections of the story. Temporal Response Function analysis was used to map the relationship between stimulus features and neural responses over time. We analyzed the encoding of speech envelope, phonemic entropy (i.e., the uncertainty of the next phoneme), and semantic surprisal (i.e., the unpredictability of lexical elements) over time.
Results: Neural response to speech envelope exhibited a greater N1-P2 TRF peak-to-peak amplitude in the multi-talker condition compared to the single-talker condition. Phonemic entropy elicited earlier and stronger responses in the multi-talker condition, while no differences between conditions emerged for semantic surprisal.
Conclusions: These findings suggest that speech adaptation involves increased allocation of cognitive resources and greater reliance on acoustic cues, as reflected in the N1-P2 results. Listeners are more sensitive to phonological uncertainty under increased talker variability, while lexical predictions appear unaffected.
