Talker and Contextual Effects On Identifying Fragmented Mandarin Tones

This study investigated identification of fragmented Mandarin tones produced by single versus multiple speakers. Six minimal pairs, including all six Mandarin tonal contrasts, were digitally processed to generate intact, silent-center, centeronly, and onset-only syllables. The syllables were produced either in isolation or with a carrier phrase qing3 shuo3 __ (“Please say __”). The stimuli were presented in four blocks: (1) single speaker, isolated syllables; (2) single speaker, syllables with the carrier; (3) multiple speakers, isolated syllables; and (4) multiple speakers, syllables with the carrier. Forty native listeners and 55 non-native listeners were put under time pressure to identify the tones of the syllables and both response accuracy and reaction time were measured. Overall, the results showed higher accuracy for the single-speaker stimuli and when the syllables were presented with the carrier. For the native listeners, context facilitated identification of multiple-speaker stimuli more than single-speaker stimuli. For the non-native listeners, in contrast, context did not interact with the speaker effect. Identification of Tone 4 was consistently most accurate and least compromised by acoustic modification among the four tones. The results indicate different processing strategies for native and non-native listeners when dealing with incomplete acoustic input.