A Pilot Study to Infer CEFR Can-Do Statements Based on a Japanese Document Classification Method Including the Pre-A1 Level

The Common European Framework of Reference for Languages (CEFR), which is designed to assess foreign language proficiency, has attracted lot of attention with can-do statements (CDSs). However, only a limited number of CEFR studies for learners of the Japanese language have been conducted. Some of them focused on the classification of Japanese example sentences into the corresponding CEFR reading comprehension indices using three inference features: length, document type, and technicality. In this study, we add seven new CDSs to the Pre-A1 level published in the CEFR Companion Volume with New Descriptors in 2017 and perform experiments to classify the CDSs. In line with the incorporation of the new CDSs, we used a new feature called Kanji rate along with the three inference features as vital indexes for assessing Japanese reading comprehension. Experimental results are presented in the study. The purpose of this study is to help teachers on site use CEFR to evaluate the Japanese language proficiency of learners. Our ultimate goal is to create a Japanese CEFR-compliant text corpus.