The role of lexical tone in L2 Mandarin spoken word recognition
MetadataShow full item record
The thesis consists of two experimental studies, each of which investigates how Mandarin lexical tone is processed by both native speakers and English learners of Mandarin. The first experiment was an auditory lexical decision task. Tone neighborhood density was controlled. Words in dense tone neighborhoods were defined as syllables that can be associated with three of the four Mandarin tones (i.e., T55, T35, T214 and T51) while words located in sparse tone neighborhoods were syllables that can be associated with only one tone. Both native and L2 listeners were asked to make a word/nonword decision on the syllable they heard. Tone neighborhood density was found to influence spoken word recognition processes in a way similar to the effects observed with segmental neighborhood (Luce & Pisoni, 1998). Both listener groups recognized fewer words from dense tone neighborhoods than from sparse tone neighborhoods. However, L2 listeners' performance was inferior (having lower accuracy and longer RTs) to native listeners. This is likely due to their relative inexperience with Mandarin tones and their consequently having difficulties perceiving tone accurately and activating and selecting the correct lexical item promptly. In the second experiment, a form priming task similar to Lee's (2007) study was conducted. Both groups of listeners were asked to make a lexical decision on a monosyllabic target preceded by one of the four types of primes in which prime and target were identical (i.e., ma 1 'mother' - ma 1 'mother'), shared only segmental structure (i.e., ma 3 'horse'- ma 1 'mother'), shared only tone (i.e., tu 1 'bald'- ma 1 'mother') or were unrelated (i.e., fo 2 'Buddha' - ma 1 'mother'). Results showed that although L2 listeners were slower and less accurate to recognize words, their processing patterns were essentially the same as that produced by native listeners (c.f. Lee, 2007). Reliable facilitation of lexical decisions to the targets was observed only when the prime and the target were identical, while monosyllabic Mandarin words differing only in tone failed to speed the response to the target. That is, listeners speeded their responses to the target ma 1 'mother' when the target was preceded by the prime ma 1 'mother', but not when preceded by the prime ma 3 'horse', even though the words were segmentally identical. In addition, there was an increase in RTs to respond to the target when it was preceded by tone overlap primes, although the effect was not significant in comparison with the condition of unrelated primes. The results of these two experiments demonstrate that tone neighborhood is an important factor in the recognition of Mandarin spoken words observed in both native and L2 listeners. The overall similarity in L1 and L2 data suggests that recognition of spoken word may universally involve a process of simultaneous activation of phonologically similar lexical items (i.e. segmental and tonal neighbors), and subsequent competition among the activated items. Tonal information is obviously used to inhibit tonally incompatible candidates, leading to the selection of the best matching lexical candidate. On the other hand, the L1/L2 differences in word recognition, as evidenced by lower accuracy and longer RTs observed in L2 listeners, clearly point to the impact of the listeners' respective native phonology: the L2 learners, although quite proficient in Mandarin, have yet to achieve native-like competence in regard to lexical tones.