中文摘要：

以汉语双字构成的真词与假词为实验材料，22名大学生为被试，采用功能性近红外脑成像技术（fNIRS）和事件相关设计，考察被试在完成词汇判断任务时的大脑激活模式，探索汉语双字词在心理词典中的表征方式。结果发现：（1）在完成真假词判断任务时，被试大脑左侧额叶和左侧颞叶均被激活；（2）与判断假词相比，被试在判断真词时显著地激活左额上回和左额中回。这一结果说明汉语双字词在心理词典中是混合表征的。

英文摘要：

The processing of compound words in the mental lexicon is an important topic of cognitive psychology. By studying the processing of compound words, we may be able to answer the question whether the unit of representation of compound words in the mental lexicon is a whole word or a morpheme. There are three models about the representation and processing of compound words, including the morpheme representation model, the word representation model, and the mixed representation model. According to the morpheme representation model, the unit of representation is a morpheme; the word representation model suggests that the unit of representation is a word; according to the mixed representation model, for familiar words, the representation unit is a word, while for pseudowords which have no representation words in mental lexicon, they can only use morphemes as their representation units. These years, brain imaging provides a critical tool for investigating the role of polymorphemic processing and its neural basis. Via functional neuroimaging, researchers have studied the processing of compound words and pseudowords to test different cognitive models of word reading. However, several studies comparing words and pseudowords have produced inconsistent results. The inconsistencies could reflect confounds from non-lexical aspects of the stimuli. Alternatively, they may reflect true effects that are inconsistent across subjects; different subjects may use different reading strategies, dependent on the experimental design used, such as block or slow event-related design. Also the inconsistencies may be due to no replications across studies because of insufficient statistical power. Functional near-infrared spectroscopy （fNIRS） has become increasingly established as a promising technique for monitoring functional brain activity. With the use of the general linear model （GLM） for fNIRS data analysis, rapid event-related designs have been incorporated in fNIRS studies. This method is a reliable tool to investigate visual