Cross-language identification of non-native lexical tone

Abstract

We extend to lexical-tone systems a model of second-language perception, the Perceptual Assimilation Model (PAM) (Best & Tyler, 2007), to examine whether native-language lexical-tone experience influences identification of novel tone. Native listeners of Cantonese, Thai, Mandarin, and Yoruba hear six CV syllables, each produced with the three phonemic Yoruba tones (High-level/H, Mid-level/M, Low-level/L), presented randomly three times. In a 3-AFC task, participants indicate a syllable’s tone by selecting from a set of arrows the one that illustrates its pitch trajectory. Accuracy scores (proportion correct) were submitted to a two-way rANOVA with L1-Group (x4) as the between-subjects factor and Tone (x3) as the within-subjects factor. There was no main effect of Tone or Group. The Tone-by-Group interaction was significant (p = 0.031) but driven by one group: Thai listeners identified H and M more accurately than L (both p < 0.05), though L accuracy was above chance (59%; chance = 33.33%). Tone-error patterns indicate that Thai listeners primarily confused L with M (two-way L1-Group x Response-pattern rANOVA p < 0.05). Overall, despite their different tonal-L1 backgrounds, listeners performed comparably. As predicted by the PAM, listeners attended to gradient phonetic detail and acoustic cues relevant to L1 phoneme distinctions (F0 height/direction) in order to classify non-native contrasts. [NSF grant #0965227.]