Item

Abstract

Rhythm is a natural part of speech. We normally under - stand spoken sentences in our native language’s rhythm without effort. Speech perception studies have typically taken rhythm for granted, and we therefore know little about how rhythmic information affects activity in the speech perception network. We developed a novel method for decompos - ing speech signals in order to separate phonetic information from rhyth - mic structure. Audio recordings of English sentences spoken by a Japanese native speaker were manipulated such that their rhythm was stress timed (like English), mora timed (like Japanese) or had phonemes with equal durations. Noise-vocoding was used to minimize contributions of F0 and to control intelligibility across conditions. Spectral rotation was also used to create unintelligible control conditions. Twenty-one healthy right-handed participants underwent behavioural testing and fMRI scans. In behavioural testing, native English participants judged the naturalness of speaking rhythm of the sentences. Results confirmed subjects judged English sen - tences as being most natural. FMRI was used to image the brains of par - ticipants while they listened to the sentences. Result showed that supple - mentary motor area (SMA), a region involved in speech production, was sensitive to rhythm naturalness. This suggests that integrating non-native speech rhythm with native language speech may rely on increased audito-ry-motor processing.