Proceedings of the Annual Meeting of the Cognitive Science Society

This study investigated how well similarity models of Chinese characters developed in previous research could be used to model human judgments across different levels of proficiency in Chinese. The behavioral data collected from the three groups of participants confirmed the superiority of and preference for configurations over components in experts' perceptions. In contrast, Chinese learners' and novices' criteria for similarity judgments were less clear, as indicated by the low proportion of variance that could be accounted for by extended tree analysis of their group judgments. We discuss computational challenges in modeling human perception and judgments about Chinese characters and propose future directions for research, including the potential use of statistical and machine learning techniques with larger datasets for improved model development.