认知与发展心理学研究室知识库

目前对语言情绪信息的研究，大多数都集中在情绪词语与中性词语的对比上。已有研究表明，相对于中性词语，情绪词语的加工具有优先性，使得其自动获得更多的加工资源，诱发早期的脑电反应。而对于句子水平上情绪信息的研究，多是使用通过情绪词赋予句子情绪性的语句。然而，由于情绪词语本身所包含的语义信息和情绪意义涉及了语义系统和情绪系统两个系统，我们很难区分出句子的情绪性是情绪词的词汇通达的作用还是句子水平上情绪产生的作用。句子理解需要把单个词语所传达的信息组合起来从而构建出完整的意义。我们的研究关注人们对不包含任何情绪线索词，即完全是由中性词语组合而成的情绪性句子中内隐情绪意义的加工。剥离开情绪词的影响，进一步考察在语义系统内情绪信息的产生问题。
通过操纵句子中的某个关键词，使得整个句子具有情绪意义、中性意义或者语义违反。通过对比违反条件和中性条件下的关键词，可以获得非情绪语义的构建过程；通过对比情绪条件和中性条件下的关键词，可以获得情绪意义的构建过程；通过对比非情绪语义和情绪意义效应，可以进一步揭示语言理解过程中情绪意义构建的独特性。实验一采用自定步速阅读范式，考察整合产生的情绪信息对句子在线阅读过程的影响。研究发现，违反条件的关键词因为难以与前文整合，使得阅读时间变长，这一效应溢出到关键词后一词上。而情绪条件和中性条件之间关键词的阅读时间没有差异。情绪条件和中性条件没有差异，可能是因为情绪信息的加工很快，自动化程度很高，自定步速阅读的反应时指标敏感性不足。实验二使用ERP技术，考察句子理解中情绪信息加工的时间进程。研究发现了经典的语义违反的N400效应，而情绪条件相较中性条件诱发了一个始于300毫秒的LPP效应。情绪条件在N400时间窗口的效应，表明句子理解中，情绪意义和非情绪语义是同时构建的。内隐情绪意义可能是句子高层表征的一部分。
综上所述，本研究通过两个实验发现，情绪信息的加工是即时的，情绪意义和非情绪意义同时构建，内隐情绪意义可能是句子高层表征的一部分。

Existing studies on emotion and language have primarily focused on the processing difference between emotional and neutral words. It has been found that compared to neutral words, emotional words have processing priority and thus automatically receive more processing resources, resulting in early ERP responses. However, the emotional words contain both semantic and emotional information, which makes it difficult to disentangle between lexical access and emotion induction by studying emotional words. It requires the combination of multiple words in order to construct the meaning of a full sentence. The present study focuses on emotional sentences that are comprised of neutral words. We investigate the emotional generation within the semantic systems.
We manipulated critical words in sentences so that the sentences can be emotionally negative, neutral, or anomalous. The comparison between the anomalous and neutral conditions allows us to examine the construction of non-emotional meaning, whereas the comparison between the negative and neutral conditions allows us to examine the construction of emotional meaning. Moreover, the comparison between the non-emotional and emotional effects enables us to understand the uniqueness of emotional construction during sentence comprehension. Experiment 1 employed the self-paced reading paradigm to investigate how people derive emotional meaning from the combinatory processing of neutral words. We found that the reading time for the words after the critical words was longer for the anomalous condition than the neutral condition due to the integration difficulty. However, we found no difference of the reading time between the emotional condition and the neutral condition. This might be caused by the highly rapid and automatic emotional processing as well as the lack of sensitivity of the response time in the self-spaced reading task. Experiment 2 employed the ERP technique to investigate the time course of emotional construction during sentences comprehension. As expected, we found a classic N400 effect between the anomalous and neutral conditions. More importantly, we found that relative to the neutral condition, the emotional condition elicited a larger LPP starting from 300 ms after the critical words' onset. The overlapping time windows between the emotional and semantic effects suggest the concurrent emotional generation and language combination. The generated implicit emotion together with other conceptual interpretations constitutes a unified, meaningful representation of the language inputs.
In sum, this thesis revealed that the emotional meaning construction is real-time and concurrent with the non-emotional meaning construction. Emotional meaning constitutes a unified, meaningful representation of the language inputs.