GlyphGenius: Unleashing the Potential of AIGC in Chinese Character Learning

Abstract

Unlike phonetic writing systems such as English, Chinese characters, as ideographic symbols, combine sound, form, and meaning. Therefore, learning and mastering Chinese characters inevitably involve understanding their etymology and semantics represented by their shapes. However, for the sake of writing convenience, Chinese characters have undergone a long process of evolution, gradually reducing their pictographic components. This evolution poses greater challenges for non-native learners unfamiliar with the structure of square Chinese characters. In this paper, we propose a novel approach to assist in Chinese character learning. Considering the unique visual features deeply rooted in the cultural origins of Chinese characters and the ability of Artificial Intelligence Generated Content (AIGC) to generate images without requiring user expertise, we utilize the AIGC model to redraw Chinese character components based on their inherent meanings. Through this visual transformation, users can intuitively grasp the semantics of Chinese characters, opening up new avenues for Chinese character learning. The Guess-Meaning experiment reveals that learners with less than one year of experience scored an 12.76% higher accuracy in recognizing the meaning of characters that had been redrawn, as compared to the original characters. During system usability testing, users reported an average satisfaction rating of 4.24 out of 5 points. The major limitations of the present study are that the current system still relies on human understanding of Chinese characters for redrawn prompts, and not all Chinese characters have corresponding pictorial meanings. The system and repository are now accessible via the link https://scroll.ihanzi.net and https://github.com/BlossomsGarden/Glyph-Genius.

Keywords

• Glyph visualization
• human-computer interaction
• stable diffusion
• visualization in education