中文摘要：

采用Go/NoGo语义分类任务,在贴近自然语言理解的任务模式下探讨了部件位置的加工时程问题。实验操纵了部件位置频率,并进一步区分了造成部件相对位置频率发生变化的两种来源,一是操纵部件整体频率（包含某部件的所有汉字集合）的变化,保持特殊位置上的部件频率（某部件出现在特定位置上的汉字集合）不变;二是操纵特殊位置上的部件频率的变化,保持整体部件频率不变。结果表明,无论是变化整体部件频率还是变化特殊位置上的部件频率,两者都引发了P200和N400的变化,且所引发的P200在峰潜时（peak latency）上差异不显著。此外,两种操作都没有引发N/P150的变化。因此本研究在贴近自然语言理解的任务中进一步肯定了部件位置在汉字阅读中的作用,既可以影响汉字亚词汇水平的字形加工,也可以影响汉字的语义提取。

英文摘要：

In written Chinese, over 80% of characters are compounds, in which independent constituents called radicals can be considered as the sub-lexical units. Some radicals can appear at different positions within a character. For instance, the radical 口 can appear on the left of 听（listen）, on the right of 知（know）, at the top of 呆（stupid）, or at the bottom of 杏（apricot）. The primary concern in prior behavioral studies on radical processing is whether radicals are represented with or without position in the Chinese lexicon（i.e., position-specific or position-general, respectively）. However, ERP results confirmed that both position-specific and position-general radical representations existed in human mental lexicon. The new argument is that whether the position-specific radicals are activated at the first stage in character recognition, as reflected by the N/P150 effects. Although prior studies revealed that position-specific radical processing was related to the N/P150, the tasks used in those studies were not natural enough to reflect the situation in conventional reading. Therefore, unlike prior studies, a Go/No Go semantic categorization task was used in the present study, which could guarantee the requirements in daily life reading. Using event-related brain potentials（ERPs）, two types of radical information were manipulated： the number of characters containing a specific radical irrespective of position（i.e., frequency as a position-general radical） and the number of characters containing a specific radical at a particular position（i.e., frequency as a position-specific radical）. The results showed that the effects of position-specific and position-general radical frequency were both related to P200 and N400. Characters of low radical frequency evoked a larger P200 and a smaller N400 than their high frequency counterparts. Moreover, no difference was found between the peak latency of the two P200 effects. More importantly, no N/P150 effects were found to be associated wit