中  華  心  理  學  刊
民93，46卷，2期，145-162

Fook-Kee Chua（Department of Social Work and Psychology, National University of Singapore, Singapore）；Siok-Tin Lim（Department of Social Work and Psychology, National University of Singapore, Singapore）；Lay-Har Lim（Department of Social Work and Psychology, National University of Singapore, Singapore）；Timothy Choy（Department of Social Work and Psychology, National University of Singapore, Singapore）

Abstract

We report three experiments examining the hypothesis that lexical processing of multi-character words proceeds from a decomposition of the whole into its constituents. The first experiment used a masked priming procedure where the prime was either the first, or the second character of a two-character word. We showed that with semantically compositional ("transparent") words, priming obtained whether the prime was the first or the second con- stituent. However, when the constituents were not semantically compositional ("opaque"), lexical access was only facilitat- ed when the prime was the first constituent of the word, contradicting the decomposition view. This result was extended in Experi- ment 2 using a bilingual variation of the same paradigm - the prime was an English word associate of the first or second con- stituent of a two-character Chinese word. The results also contradicted the predictions of the decomposition hypothesis. In the third experiment, we manipulated cohort size of the first and second constituent, and showed that whereas cohort size modulated access for transparent words, it had no effect for opaque words. The results were explained with a cohort model which hypothesizes that bimorphemic words are represented as a whole.
 

Keywords: bimorphemes, lexical representation, Chinese word processing.