How language familiarity facilitates speaker identification: behavioral and multivariate EEG evidence

Listeners identify voices more accurately in their native language than in an unfamiliar foreign language, a well-established phenomenon known as the language-familiarity effect (LFE). While the LFE highlights the integration of linguistic and paralinguistic information during speech perception, the underlying cognitive and neural mechanisms remain poorly understood. We conducted two experiments combining behavioral measures (Experiment 1) and electroencephalography (EEG, Experiment 2) to investigate what specific information was extracted more effectively from native-language speech to support the LFE. After voice learning with Mandarin and German speech, native Mandarin speakers with no prior exposure to German were tested on voice identification performance with the trained words, untrained words in the trained language, words in the untrained language and reversed speech. Behaviorally, a robust LFE was observed, with higher identification accuracy after Mandarin voice learning than after German voice learning across all linguistically meaningful conditions (trained words, untrained words, and untrained language) but not for the reversed speech. Critically, a time-resolved multivariate decoding analysis of the EEG data showed that the LFE occurred earlier when participants were tested in the trained language. These findings indicated that familiar language enhances voice identification by facilitating the extraction of both language-specific and language-general acoustic-phonetic features during voice learning. Our findings advanced our understanding of speech perception by elucidating how language familiarity influences speaker identification.