Applying educational data mining to explore individual experiences in digital games

Research involving digital games and language learning is rapidly growing. One advantage of using digital games to support language learning is the ability to collect data on students learning in real time. In this study, we use educational data mining methods to explore the relationship between in-game data and elementary students’ Chinese language learning. Thirty-six students in the sixth grade played a digital game for eight 25-minute sessions as part of their Chinese Dual Language Immersion classroom instruction. We used classification and regression tree analyses and cluster analyses to explore how in- game indicators, such as battles, time spent reading a text, and the use of an in-game glossing tool are associated with language learning and change in affect. The results indicate that time on task and use of the glossing tool were the most important variables in determining language learning gains. We also identified four subgroups of gameplay styles. While there were no significant differences in learning or affective factors based on the subgroups, these gameplay styles allow for a more individualized approach to analyzing learning within digital environments