认知与发展心理学研究室知识库

近年来，情绪和认知之间的交互过程得到了越来越多的关注。语言理解是人类一项非常重要的认知功能，同时也会涉及情绪加工，比如情绪词的理解过程，需要认知和情绪系统的共同参与，因此研究情绪词的加工有助于进一步理解情绪和认知之间的关系。以往关于情绪词的研究主要集中于探讨情绪词本身的加工过程，发现情绪词与中性词相比，具有加工优先性，而情绪词加工的优先性如何影响句子理解过程仍需要探讨。本论文采用EEG 技术，通过两个研究三个实验考察了情绪词对句子即时理解的影响。研究一考察了句子中的情绪词对其后续中性词的字形和语义两个层次信息加工的影响，研究二考察了情绪词和语境限制性对中性词加工的影响。
研究一的实验1 通过操纵句子中动词的情绪性，以及改变动词后中性宾语名词的笔画或部件操纵字形正确性因素，考察情绪词如何影响其后中性词字形信息的加工过程。结果发现字形违反词相对于正常词在中性动词后诱发了P2 和N400效应，而在情绪动词后诱发了P600 效应。实验结果表明情绪词由于吸引了更多的注意资源，导致其后续中性词的字形违反的探测和再分析过程推迟到了晚期加工窗口。研究一的实验2 同样操纵动词的情绪性，同时操纵动词后的宾语名词与语境搭配的语义一致性程度，考察情绪词如何影响其后中性词的语义整合过程。结果发现语义违反在中性动词后诱发了N400 和P600 效应，而在情绪动词后只诱发了N400 效应。表明情绪词抑制了其后续中性词语义再分析过程。
研究二在操纵句子中关键动词的情绪性的同时，操纵了句子的限制性，使得宾语名词的预期性发生变化。结果发现在关键名词呈现之前，中性高预期条件比中性低预期条件诱发了更大的晚期负波，而情绪性高低预期条件之间没有显著差异。当关键名词呈现时，不管是情绪条件还是中性条件，高预期词比低预期词诱发了更小的N400 和更小的P600。时频分析结果发现，低预期条件下theta 频段 （4-8Hz）能量比高预期条件更强，这一效应在情绪条件下比在中性条件下持续时间更长。这些结果表明句子中的情绪词促进了高预期名词在工作记忆中的保持过程，抑制了低预期名词的词汇语义提取过程。同时，三个实验均发现了情绪动词比中性动词诱发了更大的P2 以及更大的N400，表明情绪动词与中性动词在早期知觉加工以及后续语义整合过程中均存在差异。
本论文通过两个研究发现了情绪词对句子理解存在动态影响。首先，情绪词和中性词在即时加工过程存在差异；其次，情绪词会影响其后续中性词的加工过程；第三，情绪词对后续中性词的影响受到语境的调节。

Recently, more and more attention was paid to the interaction between emotion and cognition. Language comprehension as an important congnitive function of human beings involves emotion processing. For instance, comprehending emotional words activates both cognitive and emotional systems. Therefore, investigating the processing of emotional words contributes to the further understanding the relationship between emotion and cognition. Previous studies on emotional words focused on processing of emotional words per se, revealing the privileged status of emotional words compared with neutral words. How the priority of emotional words influences sentence processing remains to be studied. In the current two studies (three experiments), we used electroencephalogram (EEG) technique to investigate the influence of emotional words on sentence comprehension. The first study explored the nfluence of emotional words on the orthographic and semantic processing of the following neutral words. The second study examined the influence of emotional words and sentence constraint on the processing of the upcoming neutral words.
In Experiment 1,we manipulated the emotionality of verbs and the orthographic correctness of the object nouns to examine how emotional words influence the following neutral words at perceptual level. The event-related potential (ERP) results showed that the orthographic violation of the nouns elicited a P2 and an N400 effect in the neutral sentence contexts, but a P600 effect in the emotional sentence contexts. The ERP results suggest that emotional words capture more attention than neutral words, resulting in delayed detection and repair of the orthographic violations of the following words. In Experiment 2, we also manipulated the emotionality of verbs and the semantic congruity between their following object nouns and the sentence contexts to examine how emotional words influence the semantic integration of neutral words. The ERPs showed that the semantically incongruent words, relative to the congruent words, elicited an N400 and a P600 effect in the neutral condition, but an N400 effect only in the emotional condition. The results indicate that emotional words impair semantic reanalysis of the following incongruent information.
In Experiment 3, we manipulated the emotionality of verbs, as well as the sentences constraint which influences the predictability of object nouns. Before the presence of the critical nouns, the neutral-highly predictive condition elicited larger late negativities than the neutral-lowly predictive condition. This late negativity effect was absent in the emotional condition. The highly predicted nouns elicited smaller N400s and P600s than the lowly predicted nouns in both the neutral and emotional conditions. The time-frequency analysis results showed that the theta frequency band (4-8 Hz) power was stronger in the lowly predicted condition relative to the highly predictive condition. The theta frequency band effect lasted longer in the emotional condition than that in the neutral condition. These results indicate that emotional words enhance maintence of the following highly predicted words in working memory, but impair lexico-semantic retrieval of the following lowly predicted words.
Moreover, the emotional verbs evoked a larger P2 and a larger N400 than the neutral verbs, reflecting different early perceptual analysis and semantic integration between the emotional and neutral verbs.
This thesis revealed a dynamic influence of emotional words on sentence comprehension. Firstly, emotional words differ from neutral words in the online processing. Secondly, emotional words influence the processing of their following neutral words. Thirdly, the influence of emotional words is affected by sentence context.