COMPUTER-BASED QUESTIONS WITH CONSIDERABLE EXPLANATORY POWER TO PREDICT THE ENGLISH COMPREHENSION LEVELS OF EFL-LEARNERS

摘要

This paper discusses which type of computer-based questions could accurately predict the English reading comprehension levels of EFL (English as a Foreign Language) learners: traditional multiple-choice questions or write-in-answer questions (writing answers in the blank on the screen). Both types of questions were used in our surveys conducted on EFL learners in Japan. We conducted two types of computer-based English comprehension tests along with semi-structured interviews on how students tackled those tasks. Descriptive and qualitative analyses indicated that the write-in-answer questions had more explanatory power to identify learners' weaknesses than multiple-choice questions did. They also showed that both the learners' levels of vocabulary knowledge and awareness of their own strategies had some influence on their reading comprehension levels. The analysis suggested that educators need to take the time to judge individual learner's performances in foreign language testing based not solely on the results of common computer-based multiple-choice type questions, but also on painstaking analysis of individual write-in-answers collected by computer. This could provide an opportunity to avoid the stagnancy in advancing the learners' levels of comprehension. In accordance with our current conclusion, we will incorporate the suitable type of test in the present web-based training system, named WebIts, developed by the author's group.