Effects of automatic hidden caption classification on a content-based computer-assisted language learning system for foreign language listening

This study develops a content-based computer-assisted language learning tool for training English listening through multimedia. The system automatically provides instant vocabulary translation and classification to users whenever they pause while watching videos. Experiments were conducted to compare the novel system with the traditional solution using a translation machine to support subtitle comprehension. The first goal of the novel system is to support listening comprehension, and the second goal is to confirm its usability. These two goals were successfully achieved in the study. Eye tracking was used to observe variations in gazing and reliance on subtitles. The results of vision analysis show that the degrees of reliance on subtitles remain unchanged.