Zenji: Heuristic Algorithms to Identify and Correct Common Japanese Kanji Handwriting Mitakes
Loading...
Date
2024-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Texas A&M University. Undergraduate Research. Department of Computer Science and Engineering.
Abstract
The kanji writing system—tens of thousands of logographic Japanese characters originating from China—is by far the most complex of the three Japanese scripts. Kanji present a unique challenge for English-speaking Japanese learners, as they lack an equivalent in western tongues. Whether studying in a classroom or independently, Japanese language learners are often exposed to mnemonics or other techniques to aid in the memorization of these symbols. However, the classroom setting presently provides a unique opportunity for learners to hand-write the characters and receive feedback on their writing. The difficulty and lack of feedback on handwriting for independent learners discourages the practice of hand-writing kanji, which can stunt the learning process. This paper introduces Zenji, a system of heuristic algorithms to analyze student handwriting of kanji. Zenji provides a lightweight solution that may easily be run on students’ personal computing devices and naturally integrated into existing mobile spaced-repetition solutions for kanji learning. The Zenji system provides users overall and per-stroke numeric, visual, and written feedback on various metrics, focusing on common mistakes. In validation testing, Zenji’s feedback proved to be highly accurate across the categories that it assesses, highlighting user errors and combining a classroom-like learning experience with the flexibility of independent study.