Abstract:

It is notoriously difficult to dissociate the processes of orthography, phonology, and semantics in visual word recognition. Using homographic morphemes in Chinese homographs, the present study attempts to further examine the time course of orthographic, phonological and semantic activation in a masked priming paradigm.

Two-character compounds of Chinese were divided into such four conditions: Heterophonic-homography (“+O-P-S”, e.g., “作坊” / zuo1 fang / workshop), Homophonic-homography (“+O+P-S”, e.g., “作息” / zuo4 xi1 / work-and-rest), Identity (“+O+P+S”, e.g., “作诗” zuo4 shi1 / poetry-composing) and the control (“-O-P-S”, e.g., “账本” / zhang4 ben3 / account-book). Each of them served as a prime with both forward and backward masks and an SOA of 47ms, and was followed by the same target compound (e.g., “作画 / zuo4 hua4 / painting”) with its initial character being the same as those of the first three conditions. Two experiments were conducted, adopting a lexical decision task for Experiment 1 (behavioral) and a semantic categorization task for Experiment 2 (ERP) respectively.

It was found that (1) the Heterophonic-homography, Homophonic-homography, and Identity conditions all induced morphological priming effects relative to the control condition; (2) in P200 (120~220 ms) and Late N400 (430~540 ms), the amplitude and waveform of Homophonic-homography was much more similar to that of the control condition, but was significantly different from those of the Heterophonic-homography and Identity conditions; (3) in Early-N400 (280~400 ms), the amplitude and waveform of Homophonic-homography was closer to that of the Identity condition, but was significantly different from that of Heterophonic-homography.

The result indicates that the early stage of visual word recognition might be related to morpho-orthographic processing based on lexeme, in which orthographic similarity and phonological information may play an important role; however, the late stage of visual word recognition might be concerned with morpho-semantic processing based on lemma, which may involve meaning selection and lemma competition. The above findings tend to support Parallel Distributed Processing Model as advocated by McClelland and his associates.

Key words: Chinese compound, homographic morpheme, Early-N400, Late-N400