Lexical decomposition of complex Kanji characters in Japanese readers

Abstract

The study investigates the question whether during recognition of a complex Kanji character readers also access the meaning of opaque components of the character. The experiment used a speeded semantic-categorization task of two characters, in which readers were requested to decide whether two characters represented words related in meaning or not. The control pairs consisted of (a) graphically similar, but semantically unrelated, characters, and (b) a simple character and a complex character, of which the simple character was a component (part-whole relation), the two characters being again semantically unrelated. The critical pairs consisted of two characters unrelated in meaning, one of which contained an opaque component element related in meaning to the other character. A corresponding trial with English and its alphabetic system would be to ask a subject to decide whether boycott is related in meaning to girl or not. In this condition the subject is expected to give a No response, but if the pseudocomponent boy is activated, interference could occur. (If boy is recovered during recognition of the pseudocompound boycott, the appropriate No response for boycott-girl could become more difficult to take.) The result of the experiment indicated interference in all three conditions (graphically similar characters, part-whole related characters, and opaque-component related characters. Thus, in an analogous task with alphabetic material, it would be more difficult to say No about the existence of a semantic relation between (a) tree and free, (b) boy and boycott, and (c) boycott and girl. The last result can be interpreted as offering evidence for the activation of the opaque component upon recognition of the complex character.