A Prediction Model of Foreign Language Reading Proficiency Based on Reading Time and Text Complexity.
Katsunori Kotani, Takehiko Yoshimi, Hitoshi Isahara · US-China education review · 2010
In textbooks, foreign (second) language reading proficiency is often evaluated through comprehension questions. In case, authentic texts are used as reading material, such questions should be prepared by teachers. However, preparing appropriate questions may be a very demanding task for teachers. This paper introduces a method for automatically evaluating proficiency, wherein comprehension questions are not required. This method assesses a learner's reading proficiency on the basis of the linguistic features of the text and the learner's reading time. A reading model following this method predicted reading proficiency with an ER (error rate) of 18.2%. This ER is lower than those of models proposed in previous studies. Furthermore, the ER of the authors’ reading model for various learner groups classified by their RS (reading speeds) was examined. The result of this examination showed that the error rate was the lowest for the group of learners with fast RS.