Identification of multi-speaker Mandarin tones in noise by native and non-native listeners

The similarities and contrasts between native and non-native identification of multi-speaker Mandarin tones in quiet and in noise were explored in a perception experiment. Mandarin tone materials produced by three male and three female speakers were presented with five levels of signal-to-noise ratios (quiet, 0, −5, −10, and −15 dB) in two presentation formats (blocked by speaker and mixed across speakers) to listeners with various Mandarin experience (native, first-year, second-year, third-year, and fourth-year students). Stimuli blocked by speaker yielded higher accuracy and shorter reaction time. The additional demand of processing mixed-speaker stimuli, however, did not compromise non-native performance more than native performance. Noise expectedly compromised identification performance, although it did not compromise non-native identification more than native identification. Native listeners expectedly outperformed non-native listeners, although identification performance did not vary systematically as a function of duration of Mandarin experience. It is speculated that sources of variability in speech would affect non-native more than native tone identification only if syllable-internal, canonical F0 information is removed or altered.