声旁家族规则性对家族大小的脑电P200效应的调节作用

The Moderating Effect of Neighborhood Regularity on the P200 Effect of Phonetic-Radical Neighborhood Size

摘要: 本研究以大学生为被试，采用同一字判断任务，记录目标字引发的事件相关脑电位，对声旁家族规则性对家族大小的调节作用进行了研究。结果显示，在不规则家族下大家族比小家族引发了更小的P200波幅，即波幅上的家族促进效应；而在规则家族下，大小家族间的差异不显著。结果初步表明，由于多数家族成员与声旁同音，规则声旁大家族的反馈比小家族受到声旁更大的干扰，抵消大家族反馈的优势，进而抵消早期声加工优势，导致大家族的P200促进消失。

Abstract: In recent years, there has been an increasing research on the phonetic-radical neighborhood size (NS) effects in Chinese character recognition. But the NS effects reported are inconsistent, sometimes being either facilitative or inhibitory, and the causes are in debate. No neighborhood regularity has been considered as a cause. A regular neighborhood is defined by a high ratio of neighborhood members that are homophonic to the phonetic radical to the members that are not. It has been assumed that a part of a word string can be represented in both the whole-word layer and the sub-lexical layer. Thus, it is proposed here that the phonetic radical of a regular neighborhood will be able to inhibit its regular members and reduce their feedbacks to the radical in the sub-lexical layer. The purpose of this study is to test this proposal. The experimental design was 2 (regular vs. irregular neighborhood) × 2 (large vs. small NS). The delayed character-matching task was used, judging whether the target and the probed presented in sequence were the same words in identity. The experimental results of 25 participants showed that for the irregular neighborhoods, large NS elicited smaller P200 amplitude than small NS targets, indicating a normal NS effect. Conversely, as expected from the proposal, for the regular neighborhoods, there is no significant difference in P200 between the large and small NS, indicating disappearance of the NS effect. The disappearance of the NS effect in the regular neighborhoods suggests that the phonetic radical of a regular neighborhood is able to inhibit its regular members and thus reduce the feedbacks to the radical in the sub-lexical layer. These results are further in line with the assumptions of the IA model and the assumption that a part of a word string can be represented in both the whole-word layer and the sub-lexical layer.