Production and perception of laryngeal contrasts in Mandarin and English by Mandarin speakers

Abstract

Word initial stops contrast on many acoustic dimensions, and the acoustic realization of the laryngeal contrast is different across languages. The language-specific hybrid of acoustic cues for the laryngeal contrast shapes how speakers and listeners represent and identify the contrast. This dissertation explores how Mandarin speakers produce and perceive the laryngeal contrasts in their native language (L1) and the second language (L2), focusing on F0 (fundamental frequency) perturbation patterns. By doing so, this study aims to contribute to the theoretical discussion of the production-perception link, the L1-L2 interface, and the cross-linguistic comparison between tonal and non-tonal languages. This dissertation features two sets of experiments: Mandarin speakers’ L1 production and perception experiments (§3), and Mandarin speakers’ L2 production and perception experiments (§4). Mandarin speakers’ L1 experiments (§3) examine the acoustic cues that Mandarin speakers use to produce and perceive the Mandarin aspiration contrast. The study observes a high initial tone and low initial tone effect in producing the aspiration contrast. The F0 following aspirated stops (F0-aspirated stops) is found to be significantly higher than the F0 following the unaspirated stops (F0-unaspirated stops) in the high-level tone and the falling tone but is significantly lower than the F0-unaspirated stops in the rising and the dipping tone. The duration of significant F0 differences between the F0-aspirated stops and the F0-unaspirated stops is limited to the onset of the vowel, ranging from 11 ms to 75 ms. In perception, the voice onset time (VOT) is the primary cue for the aspiration judgement. Moreover, L1 Mandarin listeners are observed to extract both tonal and consonantal information from the post-onset F0. The listeners tend to associate high pitch with the aspirated stops and low pitch with the unaspirated stops across the four tonal contexts. The high initial tone and low initial tone effect is observed in the perception task as well. The low initial tones elicit significant more unaspirated responses than the high initial tones. Mandarin speakers’ L2 experiments (§4) investigate L1 Mandarin speakers’ perception and production of the English voicing contrast with parallel tasks from the L1 experiments. Overall, the F0 following voiceless stops (F0-voiceless stops) is found to be produced significantly higher than the F0 following voiced stops (F0-voiced stops). The duration of significant F0 differences between the F0-voiceless stops and the F0-voiced stops is shorter than that produced by L1 English speakers. In perception, the L1 Mandarin listeners use VOT as a primary cue and pitch as a secondary cue for the English voicing contrast. They tend to associate high F0 with the voiceless stops and low F0 with the voiced stops. The intrinsic F0 of the vowels also play a role and a perceptual compensation effect is observed in the English stop identification task. It seems that the listeners tend to attribute the high F0 they hear to the intrinsic F0 of the vowel rather than the voicelessness feature of the preceding stop when VOT is ambiguous. The findings in this dissertation indicate that F0 perturbation effect is primarily an automatic effect. The acoustic features such as tone and vowel height that realized with F0 can influence F0 perturbation patterns as different tones and vowels require different coordination of articulators. In addition, this study sheds light on why the F0 perturbation duration in tonal languages is shorter than that in non-tonal languages. A comparison between Mandarin speakers’ L1 and L2 stop production suggests that speakers from a tonal language inhibit the F0 perturbation effect to keep the tonal information intact. In sum, the parallel studies of the laryngeal contrasts across languages and modalities in this dissertation offer insight into between-language (tonal vs. non-tonal) and within-language (production vs. perception) variations of how Mandarin speakers-listeners use different acoustic properties to contrast laryngeal features in their L1 and how they adapt the information of individual acoustic cues when they learn an L2. Along with such findings, this dissertation also provides a balanced corpus for testing models of the perception-production link as well as the L1-L2 interface. This work also exhibits the acoustic representation of physiological and aerodynamic factors that contributing to the speakers-listeners’ ways of categorizing sound categories in their L1 and L2.