Abstract

This article examines how English as a foreign language learners might be better matched to reading texts using automatic readability analysis. Specifically, I examine how the lexical decoding component of readability might be validated. In Japan, readability has been mostly determined by publishers or by professional reading organizations who only occasionally publish their lists of readability ratings for specific texts. Without transparent readability methods, candidate texts cannot be independently evaluated by practitioners. Moreover, the reliance on centralized organizations to curate from commercially available texts precludes the evaluation of the multitudes of free texts that are increasingly available on the Internet. Previous studies that have attempted to develop automatic readability formulas for Japanese learners have used surface textual features of texts, such as word and/or sentence length, and/or they have used word-frequency lists derived from large multiregister corpora. In this article, I draw upon on the findings of a study that examines how such word-lists might be validated for use in matching Japanese learners to texts (Pinchbeck, manuscript in preparation). Finally, I propose a list of general criteria that might be used to evaluate the components of readability formulas in general.

Suggested citation