There has been a long-standing debate on the acoustical difference between second tone and sandhi third tone in Mandarin Chinese. Except [1], most of the studies focus on the tonal realization in carefully controlled speech, and the relationship between demographic features and tonal realization has not been addressed by existing studies using large-scale speech data. In this paper, we investigate the tonal realization in three large speech corpora of Mandarin Chinese, which contain around 2,300 speakers from different regions of China. The relation between demographic factors (e.g., gender, age, and place of origin) and tonal realization is addressed. The results show that even though the sandhi third tone is close to the second tone in terms of acoustic features, it still displays significant differences in large data. We also report some demographic influences on tonal realization, including age, gender, and places of origin.
Cite as: Tian, Z., Dong, X., Gao, F., Wang, H., Lin, C. (2022) Mandarin Tone Sandhi Realization: Evidence from Large Speech Corpora. Proc. Interspeech 2022, 5273-5277, doi: 10.21437/Interspeech.2022-10897
@inproceedings{tian22e_interspeech,
author={Zuoyu Tian and Xiao Dong and Feier Gao and Haining Wang and Charles Lin},
title={{Mandarin Tone Sandhi Realization: Evidence from Large Speech Corpora}},
year=2022,
booktitle={Proc. Interspeech 2022},
pages={5273--5277},
doi={10.21437/Interspeech.2022-10897}
}