Character recognition without sound or meaning

Abstract

A long-running debate in Chinese psycholinguistics has concerned the relative roles of semantics and phonology in reading Chinese characters. Some researchers argue that character reading requires activation of phonological representations, while others maintain the traditional view that Chinese readers generally access meanings directly without phonological mediation. This paper describes an experiment that addresses this debate from a novel direction: Chinese readers were asked to report what they know about simple characters with unfamiliar meanings and/or pronunciations. The `phonology-first` view predicts that it should be impossible to know the meaning of a character without knowing its pronunciation, while the semantics-first` view predicts that it should be impossible to know the pronunciation of a character without knowing its meaning. Our experiment showed that both situations can exist, though with quantitative and qualitative differences: knowing the pronunciation without knowing the meaning is a somewhat more common occurrence, though it arises most often with characters that share a phonological component with other characters, while knowing the meaning without the pronunciation most often occurs when readers have an alternative (nonstandard) pronunciation for the character or when the character is used as a quasi-linguistic symbol. Moreover, a signal detection analysis found no difference in sensitivity to meaning vs. pronunciation. At the same time, however, readers were strongly biased in their confidence judgments about phonology. Our results thus reaffirm support for a more nuanced position in the debate over Chinese reading, one where both phonology and semantics play key roles in reading.