ABSTRACT. Mandarin is one of the most representative tonal languages in the world. It has four contrastive tone categories (Tone 1 (T1): high level (ā), Tone 2 (T2): high rising (á), Tone 3 (T3): dipping (ǎ), Tone 4 (T4): high falling (à)). Thus, incorrect use of lexical tones leads to confusion or misunderstanding (e.g. 妈mā ‘mother’ vs 马mǎ ‘horse’ or 买mǎi ‘buy’ vs 卖mài ‘sell’). It is widely ... [Show full abstract] acknowledged that learning Mandarin tones is difficult for speakers from non-tonal first language (L1) backgrounds.
The purpose of this research was to examine if listeners differ, according to their L1s, in their response to between-category and within-category pitch variations in the perception of Mandarin lexical tones. Five groups of listeners naïve to Mandarin and a control group of 10 native speakers of Mandarin participated. Non-native listeners’ L1 backgrounds included both tonal (Burmese (n = 18), Thai (n = 11), Vietnamese (n = 10)) and non-tonal (Australian English (n = 10), Japanese (n = 14)) languages.
A categorial discrimination test with a four-alternative forced-choice oddity task employed in previous second language (L2) speech research (e.g. Flege et al. 1999; Wayland & Guion, 2004) was used to assess listeners’ perception of Mandarin lexical tones. All six tone pairs (T1-T2, T1-T3, T1-T4, T2-T3, T2-T4, T3-T4) including 336 (252 change, 84 no-change) trials produced by eight (4 males, 4 females) native Mandarin speakers were presented to the listeners. Their task was to listen to the stimuli presented in triads and identify the position of an odd item (1, 2, or 3) in change trials (e.g. bā – bā – bá) or select “NO” if all items were from the same tone category in no-change trials (e.g. mā – mā – mā or dǐ – dǐ – dǐ). Percentages of incorrect discrimination for the change and no-change trials were compared across the six groups of listeners.
Overall, non-native listeners made more perception errors in change than in no-change trials (36% vs 27% on average). In other words, they failed to detect the between-category (e.g. T1 vs T2) differences more so than they failed to ignore the within-category variations. However, some non-native listeners from tonal language backgrounds (i.e. Thai and Vietnamese groups) misperceived no-change trials as frequently as (or more frequently than) change trials (Thai: 23% vs 25%, Vietnamese: 26% vs 30%) as shown in Figure 1. This suggests that tonal language listeners may be too sensitive to linguistically irrelevant pitch variations, which is suitable for processing native, but not non-native, tones.