Identification of acoustically modified Mandarin tones by native listeners

This study investigated Mandarin tone identification by 40 native listeners when only partial acoustic information was available. Twelve minimal tone pairs including all six Mandarin tonal contrasts were digitally processed to generate four syllable modifications: intact, silent-center, center-only, and onset-only. The syllables were recorded in two carrier phrases such that the offset of the carrier tone and the onset of the test tone were either a match or mismatch in fundamental frequency (f0). In three experiments, the test syllables were presented in the original carrier phrases, excised from the carrier phrases, or excised and cross-spliced with another carrier phrase. Response accuracy and reaction time (RT) were measured. Listeners identified the tones at better than 86% correct with or without the carriers except when they heard the onset-only syllables in isolation, when their identification accuracy fell to 73% but still beyond chance. The modifications impacted the four tones differently, with Tones 1 and 2 being compromised more than Tones 3 and 4. Confusion matrix analyses showed that Tone 2 was predominantly confused with Tone 3, and Tone 1 was primarily confused with Tone 4. There were no main effects of splicing or match/mismatch with the carrier tone; however, in the cross-spliced context, syllables originally produced with a matching carrier tone were identified faster and/or more accurately. The implications of these findings for lexical tone processing are discussed.