Abstract:

During speaking practice, it is very important for learners to receive instant feedback from their teachers. However, it requires teachers to help them practicing at the same time. This constraint still exists in case of online learning. If we can eliminate the constraint, we will be able to practice speaking with some immediate feedback whenever we want.

Although existing systems adapt different measures, such as comprehension level, degree of concentration or emotion, their fundamental functions are basically the same. That is, they provide contents based on estimation of their comprehension levels. However, many of them can not directly apply to speaking practice because they just recommend contents according to the estimated measures, and they can not estimate their speaking levels. Also, some systems use sophisticated devices to obtain information about learners' movement or physiological response, which are used to estimate their comprehension level. This limits time and environment to learn.

Our distant goal of this study is to develop a system which can give learners flexible feedback in speaking practice without limiting time and environment to practice. Realizing our system will be able to reduce teachers' costs and for learners to practice speaking as many times as we like.

However, there are two challenges to this goal. (1) Developing such a system that checks atypical answers, like free speaking, is more difficult than one that checks typical answers. (2) Few systems can estimate learners' status using various information such as their face, voice, etc., which are usually used by teachers.

The contribution of this paper is threefold. (1) We sorted out existing learning support systems based on two aspects, synchronous vs. asynchronous, and interactive vs. non-interactive, and found a missing part on the segmentation. (2) We developed a prototype that fits this missing part and meets other requirements, including no limiting time and environment for practicing. (3) We conducted a preliminary experiment using our system for 14 weeks on 14 international students participating in a Japanese language class, and confirmed some positive feedback and improvements in our system from a questionnaire.

Keywords:

Learning support system, web application, preliminary experiment, Japanese language class, questionnaire.