中国科学院心理健康重点实验室知识库

摘要：情感词不仅可用于探讨词汇识别、注意、记忆、情绪等的认知神经机制，还可用于自然语言处理、情绪识别、人工智能等。情绪词的标准化有利于比较不同实验室的情绪研究成果，进行重复研究。目前汉语情绪词库主要有王一牛和罗跃嘉（2008, 1190个词）和Yao等（2016，1100个词）构建的两个双字词情感词库。但是，14%汉语词库由四字词语组成。四字词比双字词包含更多的传统文化信息，且在语义和结构等方面更复杂，很多四字词的语义难以通过那些构成它的双字或单字来推断，比如“守株待兔”。本研究邀请162名青年被试和126名老年被试使用九点量表对2061个四字情绪词的愉悦度（词引发的愉快程度）、唤醒度（词引发的激动强度）、优势度（对词内容的控制支配程度）和熟悉度（对词义的熟悉程度）进行评价.为了验证重测信度,本研究分别在2015年和2017年邀请了不同被试对同一批600个四字词进行愉悦度和唤醒度的评价，其内部一致性系数大于0.90。结果表明，愉悦度和唤醒度的关系呈U型曲线，即中性语义的情绪词唤醒度最低，积极情绪词和消极情绪词的唤醒度较高。年龄差异显著，老年人较青年人对积极情绪词的唤醒度更高，但是对于消极情绪词的唤醒度更低，具有积极效应。总体来说，老年人在优势度和熟悉性上的评分显著低于青年人的，说明老年人的评价更保守。本研究提供的四字情感词的量化标准，将为开展汉语情绪词的认知神经科学研究和其他情绪计算领域提供标准化实验材料。四字词评估的年龄差异不仅为认知神经科学、老年学和社会心理学提供了参考，而且实验所得到的有标注信息的情感词库也为自然语言处理中情感分析提供宝贵的语料库资源。

Abstract：Information about the affective meanings of words is widely used by researchers studying emotions and moods, word recognition, attention, memory and text-based sentiment analysis. Four variables of emotions are annotated (9-point scale) by younger and older adults: valence (the pleasantness of a stimulus), arousal (the intensity of emotion provoked by a stimulus), dominance (the degree of control exerted by a stimulus), and familiarity. Current research studies on Chinese affective words are mostly based on the 300 2-character words collected by Wang and Luo in the Chinese Affective Words System (2008). Our work extends affective resource to include 2,076 4-character words, providing researchers with a much richer source of information including age, gender and emotional differences in emotion norms. Generally speaking, 4-character words are more complex than 2-character words and can convey more sophisticated semantic meanings in Chinese. The valence and arousal of these selected 600 4-character words were rated in 2015 and 2017, respectively. The coefficient of internal consistency is larger than 0.90. We found significant age bias in the rating of the four affective components. Age-related difference study on the norms of 4-character words not only provides insights to cognitive neuroscience, gerontology and social psychology, but also has great value as a resource for affective analysis in natural language processing applications.