Slide Slam J8 Sandbox Series
Differential effects of temporal and spectral modulation on onset and lexical tone recognition: Evidence from Mandarin Chinese
Ruofan Wu1, Jueyao Lin2, Caicai Zhang2,3; 1Department of Linguistics and Translation, City University of Hong Kong, 2Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University, 3Research Centre for Language, Cognition and Neuroscience, The Hong Kong Polytechnic University
Various hypotheses have been proposed to explain the neural mechanism subserving hemispheric asymmetry in speech processing, including functional and acoustic accounts (Zatorre & Gandour, 2008). Among the acoustic accounts, the spectrotemporal modulation model (Flinker et al., 2019) computationally unifies concepts proposed by existing acoustic models and is found to be more compatible with the decomposition pattern of sounds by auditory neurons. The model postulates that temporal and spectral features differentially affect speech perception and explains the asymmetry in speech processing by proposing that auditory neurons in the left and right hemispheres are sensitive to different ranges of temporal and spectral information respectively. While several studies have confirmed the fragility of speech comprehension under temporal degradation and its relative robustness under spectral degradation (Flinker et al., 2019; Albouy et al., 2020) in non-tonal languages, it has not been examined before whether temporal and spectral degradation would have differential effects on the perception of different types of phonemes. For example, it is likely that lexical tones, which differentiate word meanings with systematic pitch differences, are more susceptible to spectral degradation. Mandarin Chinese, a tonal language, is thus particularly suitable to test this hypothesis and further advance our understanding of the impact of spectrotemporal modulation on speech intelligibility. We created a total of 120 meaningless Mandarin sentences, which were recorded from a native male Mandarin speaker, and filtered in the temporal and spectral domain respectively with six modulation cutoff values. For temporal modulations, the sentences were also filtered with the highest cutoff value in the spectral domain and kept constant across the six temporal modulation rates; the same was true for spectral modulations. Twelve right-handed native Mandarin speakers participated in the experiment. The filtered sentences were presented diotically to the participants in a random order, and the task was to write down the Chinese characters or homophones upon hearing each sentence. The accuracy of onset, rime and tone recognition was calculated separately for each sentence. Preliminary data showed that temporal degradation impaired onset accuracy more than rime accuracy and tone accuracy, whereas spectral degradation affected tone accuracy more than the other two, as predicted. The trend of rime accuracy fell in between onset and tone accuracy and did not exhibit a clear pattern. The preliminary results confirmed the hypothesis that temporal and spectral modulation differentially affects the perception of onsets and lexical tones. These findings have implications for understanding the relative role of temporal and spectral modulation in speech intelligibility in diverse language contexts with distinct prosodic characteristics. We plan to examine the hemispheric laterality for temporal and spectral processing in onset, rime and lexical tone perception in a dichotic listening paradigm in a second study. References A. Flinker, W. K. Doyle, A. D. Mehta, O. Devinsky, D. Poeppel, Nat Hum Behav. 3, 393-405 (2019). P. Albouy, L. Benjamin, B. Morillon, R. J. Zatorre, Sci New York N Y. 367, 1043-1047 (2020). R. J. Zatorre, J. T. Gandour, Philosophical Transactions Royal Soc B Biological Sci. 363, 1087-1104 (2007).