A language familiarity effect for talker identification in forward but not time-reversed speech

Abstract

Listeners identify voices more accurately in their native language than an unfamiliar foreign language—a phenomenon known as the Language Familiarity Effect (LFE). The purpose of this study was to assess two hypotheses about the source of the LFE: 1) that the LFE depends on linguistic processing of speech, or 2) that it results from listeners' familiarity with language-specific but non-linguistic properties of speech acoustics. We trained native speakers of English (N = 16) and Mandarin (N = 16) to identify English and Mandarin voices. Stimuli consisted of unaltered recordings of ten sentences in those languages, as well as time-reversed versions of the same. Importantly, time-reversed speech preserves global acoustic properties of the stimuli while rendering them completely incomprehensible. Consistent with the LFE, participants identified voices in their native-language significantly more accurately than in the foreign language when listening to normal, forward speech. However, participants did not exhibit a corresponding LFE for the time-reversed speech stimuli: time-reversed native- and foreign-language voices were identified with equal accuracy. These findings support the hypothesis that processes involved in speech comprehension are the source of the LFE rather than differences in the non-linguistic analysis of speech acoustics.