論文

査読有り
2019年12月

Predicting the level of linguistic knowledge from appropriately chosen learning data: A pilot study of English prepositional acquisition for Japanese EFL learners

Proceedings of the 27th International Conference on Computers in Education.
  • Ono,Yuichi

記述言語
英語
掲載種別
研究論文(国際会議プロシーディングス)

In institutions like foreign language education center, it is highly possible, given a proper online learning environment, or Computer-Assisted Language Learning (CALL) environment, that daily learning log data will be stored automatically at an institutional level. However, predicting learners’ level of overall linguistic knowledge, or performance proficiency level, is a real challenge, due to the difficulty to set up an appropriate predicting model under multiple complex factors affecting linguistic knowledge or learner's performance. Especially, it is very difficult to predict proficiency level from regular tasks in the classroom or online learning at home under the context of flipped-classroom model. In this paper, I attempt to demonstrate that the prepositional knowledge can lead to the prediction of overall linguistic knowledge including actors affecting linguistic knowledge or learner's performance. Especially, it is very diff grammatical knowledge, reading comprehension, structures, and so on. This study conducted a survey of testing Japanese EFL learner’s knowledge of English prepositions by asking 80 questions on in, on, over, above, under, and below. The analysis involves correlation analysis and implementation of Random Forest Algorithm to detect the key conceptual constructs to divide proficient-less proficient learners. The result is that a strong correlation between overall linguistic knowledge and prepositional knowledge that we collect during the classroom, and the acquisition of “special” or “metaphorical” concepts accompanied with English prepositions seems to be the key to predict overall knowledge of prepositions. Lastly, this paper concludes that this finding is interesting because it provides promising implications for collaborative data-driven or knowledge-driven research between learning analytics and theoretical linguistics, especially, the field of second language acquisition.

エクスポート
BibTeX RIS