中文摘要：

语迹理论是生成语法框架下所提出的重要理论假设，但这种理论假设正确与否，句法表征中是否真正存在没有语音形式的语迹，必须得到与语言相关的脑神经机制的实验验证。本研究在生成语法理论背景下，利用高时间分辨率的事件相关电位技术，以汉语话题句为研究语料，考察了汉语语迹的神经机制问题。实验结果表明，在汉语话题句的加工过程中所诱发的持续负波、动词位置的负成分以及句末位置的P600成分，均反映了句首话题成分移位后会在原有位置留有语迹，二者之间存在句法依存关系。根据实验结果，我们认为，语迹在汉语话题句中有其神经机制，语迹理论假设的合理性能够得到大脑神经机制上的证明。同时研究也从神经机制的角度逆向证实了汉语话题句是经由移位生成的句法结构。

英文摘要：

Trace theory is a very important linguistic hypothesis within the framework of Generative Grammar. But whether the theory is correct, namely whether the assumed empty trace exists in syntactic representation, must be verified by experiments concerning the neural mechanisms of language. On the basis of Generative Grammar, the current study uses event- related potentials （ERP） to examine the neural mechanism of trace in Chinese topicalized constructions. The results show that three types of waves （a sustained negativity, a negativity elicited by the verb and a P600 elicited at the sentence-final position） are observed in the processing of Chinese topicalized constructions. These neural electrophysiological indexes indicate that the sentence-initial topic does leave a trace in the position out of which it moves, i.e. the sentence-final position, suggesting that there is a syntactic dependency between the topic and its trace. Consequently, we draw the conclusion that trace is neurologically supported in Chinese topicalized constructions and trace theory isneurophysiologically viable. Meanwhile, the study also adds neural evidence to the view that Chinese topicalized constructions are structures derived from syntactic movement.