The role of speaker identification in auditory and visual speech recognition: Evidence from twin speech

Abstract

Understanding the content of speech does not require identifying speaker voices. Similarly, identifying speaker voices does not require understanding the content of speech. However, ample evidence has shown that speaker voices affect auditory word recognition. Familiarity with speaker voices also affects audiovisual speech recognition. This study examines whether familiarity with twin speakers, who have highly similar voices, affects the magnitude of form priming and that of the McGurk effect. In Experiment 1, participants who were familiar or unfamiliar with the twin speakers performed lexical decision and voice discrimination on pairs of auditory words that were repeated or unrelated. In Experiment 2, the same participants were asked to identify syllables in which auditory and visual information was congruent or incongruent (e.g., auditory /ba/ with visual /ga/). The auditory and visual information was also mixed between the speakers. Experiment 1 showed that familiarity resulted in reduced priming in voice discrimination but not lexical decision. Experiment 2 showed that familiarity did not result in reduced McGurk effect. It appears that the benefit of familiarity in processing auditory and visual speech is compromised when processing highly similar voices. [Work supported by Ohio University PURF and CHSP Student Research Grant.]