认知与发展心理学研究室知识库

在语言理解过程中，人们根据语句的上下文形成一种对即将出现的语言内容”是什么”的语义预期，根据语言中的节奏信息形成一种对重要信息”什么时候”出现的时间性预期，这两种预期都影响着人们对句子的语义理解，但是这种效应在汉语中的具体表现以及产生影响的认知、神经机制都需要进一步探讨。我们的研究采用ERP技术对这些问题进行了考察。 实验一通过操纵句子中词汇的音步数来控制节奏模式，形成有固定节奏和无固定节奏两种条件，根据固定的节奏模式可以形成一种时间性预期，无固定节奏的条件下则无法形成一定的时间性预期，同时控制关键词的语义与句子语境的关系，即语义的一致或不一致。结果发现，关键词与语境的语义一致时，无固定节奏条件比固定节奏条件诱发了一个更大的 N100；相对于无固定节奏条件，有固定节奏时关键词的语义不一致诱发的N400效应提前了100ms，这说明节奏的规律性使人产生了一定的时间性预期，并且从而促进了语义理解。实验二通过操纵汉语句子语境的限制性，构建高语义预期和低语义预期条件，同时控制关键名词与句子语境的语义一致性，来考察语义预期的形成阶段以及整合阶段，结果表明，早在关键名词呈现之前就已经形成了一定的预期（高预期条件比低预期条件诱发了一个更大的 LAN）；高预期条件比低预期条件诱发了一个减小的 N400，说明高预期促进了语义理解；高预期条件比低预期条件诱发了一个更大的语义相关的P600，说明高预期条件下需要对语义不一致词汇进行再分析，语义的高预期带来了一个再分析的代价。实验三结合时间性预期和语义预期，采用双耳分听范式对两种预期加工时的大脑偏侧化情况进行研究，本实验包括两个独立的分实验，实验a构建了节奏模式（有固定节奏，无固定节奏）和通道类型（左耳，右耳）两因素共四种实验条件，实验 b构建了语义预期（高预期，低预期）和通道类型（左耳，右耳）两因素共四种实验条件。实验a的结果表明，在关键名词诱发的N400窗口上，左耳条件下，有固定节奏比无固定节奏诱发了更大的 N400；右耳条件下，有固定节奏和无固定节奏两种条件无显著差异；实验 b的结果表明，在预期的形成阶段， 高预期比低预期诱发了一个更大的LAN效应，在预期的整合阶段，关键名词诱发的 N400 窗口上，左耳、右耳的预期效应没有表现出显著的差异，但是这一效应主要体现在左侧脑区上。
总之，本论文通过三个实验证明了时间性预期对语义理解有一定的促进作用，语义预期不只带来注意资源的受益，还有一个晚期再分析时的代价，时间性预期表现出明显的大脑右侧化优势，语义预期表现出一定的大脑左侧化优势。

On the processes of language comprehension, people can predict "what" appears to be coming according to the context and predict ―when‖ the critical event is coming according to the rhythmic information. Both the temporal and semantic prediction can influence the sentences’ semantic processing. But the mechanism and the neural basis of the effect need more in-depth investigation. In the current three experiments, we use ERP technique to study these questions.
In experiment 1, we manipulated the feet number of the words to form the regular or irregular metric conditions. The regular metric sentence consists all dimeter words，but the irregular metric sentence consists both monometer, dimeter and trimester words at an irregular sequence. In this way, we build the temporal prediction and temporal un-prediction condition. On the other hand, we build the semantic congruence and incongruence condition. Results showed that compared to irregular sentences, the regular sentences elicited an earlier N400 effect. This indicates that the regular metric facilitated the semantic comprehension. In the semantic congruency circumstance, the irregular context elicited a stronger N100 effect, indicating the temporal prediction before the critical words appearing. In experiment 2, we manipulated the degree of prediction and the semantic congruency, this resulted in a full factorial design with all combinations of the factors Prediction (high vs. low) and Congruency (congruent vs. incongruent). The ERPs results showed that contrast to the low prediction condition, the high predictive condition elicited a stronger LAN before the critical word appears. Contrast to the low prediction, the high prediction elicited a reduced N400 effect in the semantic congruency. Furthermore, the semantic incongruent condition elicited a stronger P600 effect than the semantic congruent condition. The results indicating that the prediction had been formed before the critical word appears. Furthermore, there is a cost of attention resource for the reanalysis process in P600 time window. In experiment 3, we combine the temporal and semantic prediction, utilizing the dichotic listening task to study the brain lateralization underlying the two predictions.The results show us that the dominant hemisphere for temporal prediction is the right hemisphere. At the same time, the semantic prediction factor did not show any lateralization of brain.