Revisiting the role of semantic radicals in Chinese character reading: The interaction of semantic transparency and imageability

Abstract

Semantic transparency refers to the extent to which the Chinese character shares meaning with its semantic radical (e.g. transparent character 碑 “monument” and opaque character 碗 “bowl” both contain the semantic radical 石 “stone”). Transparent characters are easier to process in semantic tasks; however, the issue has not been conclusively studied because semantic transparency is typically correlated with imageability. In this study, ratings of imageability were collected from 20 raters and its effect on semantic judgment was statistically controlled. Thirty native Cantonese speakers participated in a semantic categorization task in which transparency and category membership (object vs. actions) were manipulated in a fully-crossed design. Ratings indicated that objects were more imageable than actions, and high imageability predicted higher accuracy and faster RT. In the action category, the by-subject analysis (ANOVA) not controlling for imageability showed transparency effects, with higher accuracy and lower RT for transparent characters. However, when imageability was entered as a covariate in ANCOVA, the effect of semantic transparency was no longer evident. Instead, imageability significantly predicted both accuracy and RT. In contrast, for the category of objects, regardless of imageability, a reverse transparency effect was observed, with lower accuracy and higher RT for transparent characters. A possible explanation might be that opaque characters tended to represent more common objects, although overall lexical frequency was matched. Thus, the present results suggested that imageability might be contributing to the apparent semantic transparency effects, and concept familiarity might be another confounding factor.