Geometric Approaches in Time-Frequency to Grade Tones in Chinese As a Foreign Language Students

Siyu Chen,Guoqin Tang,Ziyi Chen,John J. Healy
DOI: https://doi.org/10.1109/spin48934.2020.9070940
2020-01-01
Abstract:Mandarin is a tonal language, meaning pitch is used to distinguish lexical meaning. Tone recognition occupies a significant role in learning Chinese. Automated grading metrics can provide language students with feedback in a non-threatening environment and without the high time and/or financial cost of individual feedback from a teacher. In this paper, we present some preliminary work on geometric measures in time-frequency to grade the quality of tone production in Chinese as a foreign language students. Our preliminary results show that the metrics can distinguish native from non-native speakers for all tones. The low computational cost of these metrics makes them especially suitable for mobile platforms.
What problem does this paper attempt to address?