Chinese Pronunciation Training System "Seicho Nigo"

-Educational TV: "Chinese Conversation"-

Toru TAKAGI,
Associate Director, Human Science

The conventional language learning process has mainly relied on a student listening to a teacher's voice, or to a recorded voice, to pronounce the sound in imitation of what he or she has heard. However, scientific analysis involving comparison of the model's pronunciation with the student's pronunciation is expected to enhance the effectiveness of current learning methods. For this purpose, STRL has developed "Seicho Nigo," a software program for learning the Chinese language. Accent and intonation in Chinese is expressed by the speaker raising and lowering the pitch of his or her utterance. Each Chinese character represents a syllable having a certain intonation. Chinese has four basic forms of accent, with each having a different meaning. Conventional programs describe the accent and intonation of the model voice only in the form of an arrow, as shown in Figure 1. "Seicho Nigo" uses overlapping curves representing the model's utterances and the student's, thereby visually assisting the learning process. It was first used on the NHK educational TV program "Chinese Conversation," and has since gained favorable comments.

Figure 1: Four tones of Chinese

Characteristics
Presentation of an easy-to-understand pronunciation curve
To make the pronunciation difference per syllable easier to grasp, this system can automatically syllabify a pronunciation curve for display, even when the actual pronunciation is of continuous words.
Intuitive model and student pronunciation comparison
Even when a student pronounces a word at a different tempo from the model sound, the portion for the same syllable can be overlapped during display to help students to discern intuitively where they have made pronunciation errors.
Pink: Native Chinese speaker
Yellow: Japanese student
Figure 2: Seicho Nigo display example
Figure 3: Seicho Nigo being used in "Chinese Conversation"
During the broadcast program, an instructor gives a supplemental explanation on a pronunciation point while looking at a displayed result. This interaction, we felt, could be incorporated into the system, and we are currently working with program producers and the instructor towards this goal. By programming the software with the instructor's expertise so it gives a "tip" during each lesson, we hope to create a self-learning tool.