Abstract: Syntactic ambiguity resolution lies at the center of much research on grammatical representation and sentence processing, which refers to the phenomenon that the parser selects the correct syntactic analysis while facing more syntactic analyses by inhibiting the wrong syntactic analysis. Knowing how readers respond when sentences can be structured temporarily helps to constrain accounts of how readers use grammatical and other information to develop interpretations of sentences in real time. Studies showed that syntactic ambiguity resolution was affected by many factors, such as working memory, cognitive inhibition, language frequency, context and prosodic cues. Illustration of how and when these various factors affect the difficulty that readers have in processing sentences can lead to deeper insight into the workings of the human sentence parsing mechanism. To illustrate how people deal with syntactic ambiguity, some models were constructed. Of all the models, the garden path model and the constraint satisfactory model are the two classical accounts. The former argued that sentences were processed with serial stages, in which readers primarily constructed sentence structure based on syntactic factors, and then reanalyzed the sentence structure based on semantic factors. On the contrary, the latter insisted that readers process sentences with parallel stages, in which syntactic factors and semantic factors coincidently worked in sentence processing. In addition, other models, such as the weakly interactive parallel model, and the language frequency model attempt to give more subtle explanations about the ambiguity resolution in others view points. Cognitive neuroscience studies on syntactic ambiguity resolution have revealed the neural mechanisms. Some studies using fMRI showed that syntactic ambiguity resolution produced more activation in the left inferior frontal cortex, the left posterior temporal cortex, the left medial temporal cortex and the left anterior temporal cortex compared with sentences without syntactic ambiguity. These findings showed that there may be common neural mechanisms for language processing, especially for ambiguity resolution, between general cognition. Because there were significant activations in general cognitive activities, such as conflict resolution, cognitive control at the mentioned above brain areas. Future research directions on syntactic ambiguity resolution were pointed out in this paper. Specifically, (1) How to separate the semantic effect in syntactic ambiguity resolution? As some studies had pointed out that semantic ambiguity were often coincident with syntactic ambiguity, thus, some effects in syntactic ambiguity resolution were from semantic ambiguity. (2) Are there qualitative differences between syntactic ambiguity resolution and lexical ambiguity resolution? Strict lexicalism argued that lexical ambiguity and syntactic ambiguity shared common knowledge representation, so they had the same ambiguity resolution mechanism. On the contrary, some scholars hold that syntactic ambiguity and lexical ambiguity have different ambiguity resolution. Lexical ambiguity resolution mainly relied on the activation of the lexicon item, which worked automatically. While syntactic ambiguity resolution mainly relied on the outer grammar principles.

Key words: syntactic, syntactic ambiguity, ambiguity resolution, neural mechanism

摘要： 句法歧义消解主要是指人们在面临多种句法分析可能性的情况下，句法分析器抑制不正确的句法分析而选择正确句法分析的认知过程。研究表明，该过程受到工作记忆、认知控制能力、语言分布频率、语境、韵律等因素的影响。为有效说明人们是如何完成句法歧义消解的，研究者们建构了诸如花园路径模型、约束满意理论、弱交互作用句法分析理论等相关模型。另外，基于脑成像技术研究表明，左侧额下回、左侧颞叶后部、中部和前部在句法消歧任务中都有显著激活。未来研究当围绕着句法消歧中的语义效应、句法和词汇歧义消解机制的异同、句法消歧中抑制能力的效应等方面展开。

关键词: 句法, 句法歧义, 歧义消解, 神经机制