DESIGN OF AUTOMATIC PLAYBACK SPEED CONTROL SYSTEM FOR LEARNING IN SECOND LANGUAGE USING ONLINE VIDEOS

Various contributors have been creating and publishing many high-quality videos on the Internet. While learners may use them as teaching material of specific topics in higher education, it is not easy for those in non-English speaking countries to apply them to actual study because many of them are created in English. The goal of our study is to provide the video-playing scenario according to learners’ skill. The scenario is composed of vocabulary information and the APSC (automatic playback speed control).

In this paper, to derive the APSC, we discuss methods for difficulty scoring and how the changes of playback speed can affect the score. Playback speed is critical for non-native listeners because they couldn’t comprehend all of the native speakers’ words at the same speed as native speakers generally. Though subtitles, which are added in most of the online videos for learning, may help non-native listeners, many of them might not read the subtitles at the native speaker speed. For instance, Japanese college students read English documents at 120 WPM (words per minute) on average, while most people whose mother tongue is English speak at greater than 160 WPM.

We focus on the subtitles as information for controlling playback speed. In the proposed method we define the difficulty score about each unit of subtitles. The score is calculated using a duration of subtitles and word frequency in a corpus. This is based on the observation that the more frequent word is faster to comprehend. Since a unit of subtitles includes time duration and dozens of words, the comprehension can be improved by making the playback speed slower if the calculated score is too difficult for a listener.

Then another problem raises up; how slower does the system change the playback speed? We consider a theoretical evaluation of playback speed perturbation based on a queueing model and show some results of calculated sample scores. We can obtain the sequence of playback speed control from this analysis.

In addition, this paper depicts the technical issue about the whole design of our planning system. Though the proposed method is aimed at higher education in which non-English speakers study using English videos, this framework can also be useful for learning of other languages.