Design of a tool to support Kanji learning based on text analysis

Abstract

The use of technology can have positive effects in second language acquisition. However, in many cases, students must learn an entirely new writing script before they can access basic learning resources in the target language. Such is the case for Mexican students learning Japanese. Unlike Spanish, which is written with a 27 character Latin-based script, Japanese texts require two different 46 character syllabaries (Hiragana and Katakana) and an independent logographic system (Kanji). Notably, the latter comprises thousands of individual symbols that must be committed to memory. Even though some resources exist to help Spanish speakers acquire Kanji, most existing tools are based on the recurrent association of predefined lists of images with their respective symbol. As of the authors’ knowledge, none take into account linguistic features such as kanji likelihood in print media, symbol co-ocurrence or other contextual information that could simplify recall for the learner. This article presents a preliminary analysis of several candidate textual features, oriented towards the design of context-aware Kanji learning software.