Introduction

Mandarin tones are manifested physically by different fundamental frequency (F0) values with F0 height and F0 contour as the primary acoustic parameters (Liu, 1924; Howie, 1976; Wu, 1986). In addition, amplitude and temporal properties such as overall duration and Turning Point are also effective phonetic correlates of the tones (Lin, 1965; Chuang et al., 1971; Moore & Jongman, 1997). Studies in the perceptual domain have shown that various acoustic cues are functionally integrated when Mandarin speakers identify the tones (Gandour, 1984; Massaro, Cohen & Tseng, 1985; Gårding et al., 1986; Blicher, Diehl & Cohen, 1990; Shen & Lin, 1991; Moore & Jongman, 1997). The hemispheric processing of Mandarin tones for native speakers is lateralized in the left hemisphere, suggesting that tones are processed as linguistic units, just like segmental properties (Hsieh et al., 2001; Wang, Jongman & Sereno, 2001).

These results about the processing of Mandarin tones for native speakers raise the question as to whether non-native speakers process tones auditorily (i.e. based on innate psychoacoustic mechanisms) or linguistically (i.e. resulting from language-specific experience). Indeed, for speakers whose native language is nontonal, tone has presented great difficulty, since the functional association between the F0 characteristics and the segmental structure is unfamiliar to them (e.g. Kiriloff, 1969; Bluhme & Burr, 1971; Shen, 1989). A number of crosslinguistic studies have been conducted to examine if and how nontonal speakers process the tones differently, how they improve the processing of the tones as a result of learning, and how this learning is instantiated in the brain.