Investigating the Impact of AI Generated Mnemonic Illustrations on Learning Japanese Kanji

Abstract

Learning Japanese Kanji presents a significant challenge for novice learners due to the vast number of characters, their visual complexity, and a lack of relatable learning resources. While various strategies exist, the use of illustrations in Kanji learning, despite their proven benefits, remains underexplored. This study investigates the impact of integrating AI-generated mnemonic illustrations with textual mnemonics on learners' Kanji attainment, cognitive workload, and learning attitudes. In a quasi experimental study with novice learners, a control group using only text-based mnemonics was compared against experimental groups that used AI-generated illustrations. Quantitative analysis, corroborated by qualitative data, revealed that while the illustrations did not produce a statistically significant improvement in post-test scores, they had a significant positive impact on reducing the perceived cognitive workload. This indicates that despite theoretical support for improved learning outcomes when adding images to text, this principle may not directly enhance recall for logographic languages like Kanji. Nevertheless, the findings suggest that while AI generated illustrations may not directly boost short-term recall, they effectively make the learning process less cognitively demanding.