Research on the Grammatical Negation System in English Sentiment Analysis: The Cardiff Grammar Perspective
-
摘要: 文章基于系统功能语言学中的加的夫模式，使用SST影评语料库辅助研究，探讨语法否定系统not在情感分析中与情感词的搭配情况。语法否定在情感分析中主要呈现为语气否定，语法否定所在句的主语多为评价对象，该否定现象负载于动作过程、心理过程与关系过程中的属性关系。利用加的夫语法中对情感意义的界定对情感词进行分类，发现否定词not与情感意义为纯情感、感情反应类词语的搭配频率最高，与困难性、重要性类情感词的搭配频率最低。否定词对不同意义的情感词产生的否定效果不同，对于纯情感、感情反应类词语，否定词在改变极性时会削弱原词强度；而行为评判、社会评判类词语中则无削弱效果；对于消极情感词，否定词并不会改变极性。文章通过对情感词进行意义层次的标注，提高了该类词汇在常见否定表达中的准确性，对解释情感分析中否定词的复杂现象具有一定启发意义，也为后续研究提供了新的视角。Abstract: The study, grounded in the Cardiff Grammar within Systemic Functional Linguistics, harnesses the SST film review corpus to facilitate investigation. It delves into the collocation patterns of the grammatical negation system “not” alongside sentiment words within sentiment analysis. Results show that grammatical negation chiefly manifests as mood negation, wherein the subject often assumes the role of the evaluated object, carrying the attribute relationship in action processes, mental processes, and relational processes. The negation word “not” exhibits the highest frequency of collocation with purely emotional and emotional reactive words, while demonstrating the lowest frequency with emotionally charged words associated with difficulty and importance. Furthermore, the negation effect of the sentiment words varies depending on the semantic connotations. In the case of purely emotional and emotional reactive words, the negation word attenuates the intensity of the original word when polarity is altered. Conversely no such weakening effect transpires with behavioral judgment and social judgment words. Sentiment words with negative polarity remain negative polarity by the negation word. This study proves enlightening in unraveling the intricate phenomenon of negation words within sentiment analysis. By annotating the semantic levels of sentiment words, the accuracy of common negation expressions encompassing this category of vocabulary is enhanced.
-
Key words:
- negation /
- sentiment analysis /
- negation effect /
- Cardiff Grammar
-
表 1 否定词not发生频次前十分布表
语气否定 频数 非语气语否定 频数 is not 194 not + to 21 does not 189 not + Ved 20 can not 104 not + Ving 14 do not 102 it’s not 56 will not 48 did not 36 may not 29 are not 26 has not 19 
下载: 导出CSV
表 2 情感词类型分布表
纯情感 社会评判 重要性 感情反应 行为评判 困难性 awful
bad
brilliant
fresh
good
great
horrible
outrageous
palatable
perfect
scary
terrible
thrilling
worstclassical
new
originalimportant amusing
compelling
disappointed
engaging
entertaining
funny
interesting
well-conceivedeffective
informative
sharp
smartdifficult
easy
下载: 导出CSV
表 3 情感词强度偏移分布表
情感词 文本情感值 所在句得分 情感偏折率 情感意义 (very)amusing +0.65 −0.68 −1.05 感情反应 awful −0.875 −0.25 0.29 纯情感 bad −0.625 −0.28 0.45 纯情感 brilliant +0.875 −0.63 −0.72 纯情感 classical +0.5 −0.42 −0.84 社会评判 compelling +0.5 −0.28 −0.56 感情反应 difficult −0.7 −0.18 0.26 困难性 disappointed −0.5 +0.32 −1.00 感情反应 easy +0.375 −0.24 −0.64 困难性 (very)effective +0.325 −0.54 −1.66 行为评判 engaging +0.875 −0.48 −0.55 感情反应 entertaining +0.625 −0.36 −0.58 感情反应 fresh +0.375 −0.16 −0.43 纯情感 funny +0.5 −0.30 −0.96 感情反应 (very)good +0.975 −0.62 −0.64 纯情感 great +0.875 −0.54 −0.62 纯情感 horrible −0.625 −0.40 0.64 纯情感 (very)interesting +0.488 −0.38 −0.50 感情反应 (very)informative +0.163 −0.318 −1.95 行为评判 important +0.875 −0.44 −0.78 重要性 new +0.125 −0.34 −2.72 社会评判 (very)original +0.325 −0.6 −1.85 社会评判 outrageous −0.875 −0.48 0.55 纯情感 palatable +0.375 −0.12 −0.32 纯情感 perfect +0.625 −0.33 −0.53 纯情感 scary −0.75 −0.28 0.37 纯情感 sharp +0.5 −0.59 −1.18 行为评判 smart +0.5 −0.61 −1.22 行为评判 terrible −0.625 −0.34 0.54 纯情感 thrilling −0.5 −0.32 0.64 纯情感 well-conceived +0.625 −0.5 −0.80 感情反应 worst −0.75 −0.2 0.27 纯情感
下载: 导出CSV
-
[1] 戴维. 现代语言学词典 [Z]. 沈家煊,译. 北京:商务印书馆,2000. [2] 苗兴伟. 否定结构的语篇功能[J]. 外语教学与研究,2011,43(2):220-229,230. [3] 梁晓波. 否定的认知分析[J]. 外语研究,2004(5):12-18,80. doi: 10.3969/j.issn.1005-7242.2004.05.003 [4] PANG B,LEE L & VAITHYANATHAN S. Thumbs up? Sentiment classification using machine learning techniques [J/OL]. (2002-05-07)[2023-01-02]. https://arxiv.org/pdf/cs.CL/0205070. [5] HALLIDAY M A K & MATTHIESSEN C M I M. An Introduction to Functional Grammar [M]. London:Edward Arnold,2004. [6] 刘英. 否定系统与语类变体——一项基于英语学术语篇的否定概率研究[J]. 外语与外语教学,2013(1):27-30. [7] MARTIN J R & WHITE P R R. The Language of Evaluation:Appraisal in English [M]. London:Palgrave Macmillan,2005. [8] 包兴佳,罗载兵. 英语否定极性的学术语篇发生研究:系统功能语言学视角[J]. 北京科技大学(社会科学版),2022,38(5):533-542. [9] SUN S A & CROSTHWAITE P. 'Establish a niche’ via negation:A corpus-based study of Move 2 in Ph. D. thesis introductions across disciplines [J]. Open Linguis tics,2022,8(1):189-208. doi: 10.1515/opli-2022-0190 [10] 张宇. 语言学视角下的国内外情感研究现状与趋势分析(2011—2020)[J]. 外国语,2022,45(1):67-68. doi: 10.3969/j.issn.1004-5139.2022.01.007 [11] 王品. 系统功能类型学视阈下的语法描写范式——以语气系统为例[J]. 外国语(上海外国语大学学报),2022,45(5):53-64. [12] HATZIVASSILOGLOU V & WIEBE J M. Effects of adjective orientation and gradability on sentence subjectivity [C]// HITOSHI IIDA. Association for Computational Linguistics. Hong Kong; Stroudsburg:ACL,2000:299-305. [13] 张成功,刘培玉,朱振方,等. 一种基于极性词典的情感分析方法[J]. 山东大学学报(理学版),2012,47(3):47-50. [14] KENNEDY A & INKPEN D. Sentiment classification of movie reviews using contextual valence shifters [J]. Computational Intelligence,2006,22(2):110-125. doi: 10.1111/j.1467-8640.2006.00277.x [15] POLANYI L,ZAENEN A. Contextual valence shifters [J]. Computing attitude and affect in text:Theory and applications,2006,20(1):1-10. [16] MOILANEN K & PULMAN S. Sentiment composition [C]// RUSLAN MITKOV. International Conference Recent Advances in Natural Language Processing. Borovets:John Benjamin,2007:378–382. [17] CHOI Y & CARDIE C. Learning with Compositional Semantics as Structural Inference for Subsentential Sentiment Analysis [C]// ALLEN J. Association for Computational Linguistics. Columbus:ACL,2008:793-801. [18] SAURI R. A factuality profiler for eventualities in text [D]. Waltham:Brandeis University,2008. [19] BENAMARA F,CHARDON B,MATHIEU Y et al. How do negation and modality impact on opinions? [C]// MORANTE R & SPORLEDER C. Proceedings of the Workshop on Extra-Propositional Aspects of Meaning in Computational Linguistics. Korea:ACL,2012:10-18. [20] TABOADA M. Sentiment Analysis:An overview from linguistics [J]. Linguistics,2016(2):1-52. [21] 姚天昉,娄德成. 汉语语句主题语义倾向分析方法的研究[J]. 中文信息学报,2007(5):73-79. doi: 10.3969/j.issn.1003-0077.2007.05.014 [22] 钟佳娃,刘巍,王思丽,等. 文本情感分析方法及应用综述[J]. 数据分析与知识发现,2021,5(6):1-13. [23] 黄国文,何伟,廖楚燕. 系统功能语法入门:加的夫模式 [M]. 北京:北京大学出版社,2008. [24] 张德禄. 加的夫语法述评[J]. 当代语言学,2011,13(3):247-255,286. [25] 高胜林. 肯定形式否定句语用探析[J]. 语文研究,2012(03):15-18. doi: 10.3969/j.issn.1000-2979.2012.03.004 [26] FAWCETT R P. A Theory of Syntax for Systemic Functional Linguistics [M]. Cardiff:John Benjamins,2000. [27] 张菊芬. 从功能语法理论的角度看表达将来时间的主要语法形式[J]. 解放军外国语学院学报,2001(5):29-32. doi: 10.3969/j.issn.1002-722X.2001.05.008 [28] 何伟,张瑞杰,淡晓红,等. 汉语功能语义分析 [M]. 外语教学与研究出版社,2017. [29] 许智宏,于子琪,董永峰,等. 影评情感分析知识图谱构建研究[J]. 计算机仿真,2020,37(8):424-430. doi: 10.3969/j.issn.1006-9348.2020.08.091 [30] 黄国文. 系统功能语言学中的加的夫语法[J]. 浙江外国语学院学报,2023(3):1-24. doi: 10.3969/j.issn.2095-2074.2023.03.001 -
点击查看大图
图(1) / 表(3)
计量
- 文章访问数: 102
- HTML全文浏览量: 38
- PDF下载量: 8
- 被引次数: 0
