Research On The Physiological Mandarin Speech Corpus' Application In Teaching Chinese As A Foreign Language

Language is a symbol of human civilization, it is also an indispensable tool for the people to exchange ideas and expressing their feelings. Today, the whole world blew up "Chinese fever" frenzy. Many pronunciation and inaccurate intonation problems have been found in Chinese learning of foreign students, as well as bias errors. It's top-priority for foreign students learning Chinese to have pure pronunciation, if they want to express and communicate fluently in Chinese.Take pinyin-teaching as an example: TCSL(Teaching Chinese as a Second Language) once taught pinyin via imitating teacher's intonation with help of pronunciation sketch map. As the pronunciation of pinyin is often accompanied with complex motion of vocal organs and manner of articulation, there is a significant limitation to present the state of a special point of articulation for pronunciation sketch map. Teachers also tell their students how to move the vocal organs such as tongue and lips, but without explicit and direct-viewing watching on situation of vocal organ's motion, the students can only rely on their feeling to adjust their pronunciation state. This kind of method has it's limitation in TCSL.In order to visualize pronunciation teaching and explore the law of articulation evolution in TCSL(Teaching Chinese as a Second Language), we proposed an approach to design a physiology Mandarin speech corpus for the international preparatory students in China to research and develop new methods in teaching Chinese as a Foreign language. Preparatory students are usually beginners to learn Mandarin, so they are selected as testers to gain the most real law of Mandarin articulation evolution. The corpus was different from an ordinary speech corpus. It consisted of not only text information(such as age, gender, nationality, mother language, Mandarin learning experience, etc.) and audio information, but also the physiological data(such as speech organs moving morphology).There are many methods to observe and record the motion of interior speech organs such as ultrasound equipment, EMA or MRI(Magnetic Resonance Imaging).In this paper, we choose a ultrasound system to gather data. Hence, compared with EMA and MRI, the ultrasound equipment is cheaper, more portable, more comfortable, and high frame frequency. The ultrasound system is designed and developed by School of Computer Science and Technology in Tianjin University based on the portableTERASON T3000.When the corpus was constructed, data were gathered by a supersonic collecting system because of its safety and low cost. In the initial experiment, 21 initial consonants, 36 simple or compound vowels and tones were tested. The results showed the corpus has some positive significance in Mandarin teaching.