The purpose of the present study is to examine whether a lexical item may have a topic-dependent SP. Within the three proposed hypotheses, the strong hypothesis predicts the SP tendencies of both mixed-SP node word and strong-SP node words will be subject to topic type;
the moderate hypothesis predicts only the SP tendency of the mixed-SP node word will be influenced by topic; the null hypothesis predicts no interaction between the SP tendencies of any node words and topic. In our news genre corpus (i.e., ADN), we have utilized the rule-based concordance line analysis to find out the SP tendencies of each node word under different topics and examined the relation between TOPIC and SEMANTIC PROSODY via chi-square test. Based on the results, we have concluded that topic has a moderately strong effect on the SP of a node word which has a mixed SP in general-domain genre (general); however, it has a weak effect on the SP of a node word with a strong positive/negative SP in general. Therefore, our results support the moderate hypothesis. We further propose the notion of topic prosody, which is at the lower level of register prosody. Moreover, we have applied the semantic network analysis in order to
discover the semantic features of the prototypical collocates of a node word under certain topics.
These semantic features may explain the rising of positive/negative SP tendency under those topics.
The notion of topic prosody suggests that topic has effect on the SP tendency of a node word. However, as noted before, only a node word with a mixed SP in general may showcase topic prosody. Compared to register prosody where a node word may have a positive SP in one register but a negative SP in another, topic prosody indicates the observable and significant change of the SP tendency of a node under one topic with respect to the overall SP tendency of that node word across different topics within a corpus. Topic prosody also implies that topic may
‧
‘cause’ and dailai ‘bring about’ reported in the study of Xiao and McEnery (2006). On the other hand, a node word with the property of topic prosody may indicates such word functions as a sentiment resonator under different topics. Thus, we conclude that the SP of a given node word may be modulated by different text categorizations at the level of either register (Hunston, 2007;
O'Halloran, 2007) or topic (cf. Partington, 2017).
There are some limitations in our study. First, the arbitrary decision of one chunk before and after a node word (1:1 chunk-based window size) as a span for a concordance is uncommon comparing to the concordance line analysis in previous studies. Meanwhile, the chunk unit, based on punctuation and symbols, is not necessarily valid. Second, the reference sentiment dictionary (i.e., ANTUSD) has limited entries of words and thus may not provide enough evaluative information for the sentiment determination of some concordances. Third, the way to automatically determine the sentiment of the concordances of each node word is not optimal.
The rule-based method may not be able to account for all the instances of concordances due to the variability of Chinese, leading to erroneous sentiment classification. Fourth, in the network analysis, the model of word embedding, i.e., GloVe, may need to be validated for its
psycholinguistic importance. Fifth, as seen in the examples of niangcheng under topic lifestyle, we are not able to single out the target meaning, cause, from the node word. The involvement of different senses of the node word in our analysis data might influence the credibility of the SP distribution of that node word under each topic. Also, it was noted that different senses of a polysemous word may have different SPs (Bednarek, 2008; Louw, 1993). Future study on the relation between TOPIC and SEMANTIC PROSODY needs to address on issue of polysemy by
applying word sense disambiguation approach so that the “noises” from other senses of the word
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
may be reduced. Sixth, since we did not use the sentiment analysis to evaluate the
positive/negative tendency of each article of a topic, the overall sentiment trend of that topic is still unclear. To exactly know the overall sentiment disposition of a topic may give us a much clearer picture regarding the connection of the topic to the SP of a node word. Thus, future study can employ sentiment analysis to discover the overall sentiment tendency of a topic and examine its link to the typical contexts a node word emerges.
To conclude, a mixed-SP node word has a topic prosody. Besides register, the SP of a given node word may be context-dependent at the topic-level. We also offer a rule-based method to efficiently discover the SP tendency of a node word across different topics within a corpus.
We hope that the next step would investigate the SP of a mixed-SP node word under different topics, registers, and genres, and even compare their respective results to see how such a word may be flexibly employed at three different levels of text category.
‧
Almende, B.V., Benoit, Thieurmel, & Titouan, Robert. (2018). visNetwork: Network visualization using’vis. js’ library. R package version 2.0.5. Retrieved from https://CRAN.R-project.org/package=visNetwork
Bednarek, Monika. (2008). Semantic preference and semantic prosody re-examined. Corpus Linguistics and Linguistic Theory, 4, 119-139. doi:10.1515/CLLT.2008.006
Blei, David M, Ng, Andrew Y, & Jordan, Michael I. (2003). Latent dirichlet allocation. Journal of Machine Learning Research, 3, 993-1022. doi:10.1162/jmlr.2003.3.4-5.993
Cohen, Stanley, & Young, Jock. (1981). The manufacture of news: Social problems, deviance and the mass media. Newbury Park, CA: Sage Pubns.
Ellis, Nick C, & Ogden, Dave C. (2017). Thinking about multiword constructions: Usage‐based approaches to acquisition and processing. Topics in Cognitive Science, 9, 604-620.
doi:10.1111/tops.12256
Ellis, Nick C, Römer, Ute, & O'Donnell, Matthew Brook. (2016). Usage-based approaches to language acquisition and processing: Cognitive and corpus investigations of construction grammar. Hoboken, NJ: John Wiley & Sons Limited.
Evert, Stefan. (2008). Corpora and collocations. In A. Lüdeling & M. Kytö (Eds.), Corpus linguistics. An international handbook (Vol. 2, pp. 1212-1248). Berlin, Germany: de Gruyter.
Freeman, Linton C. (1977). A set of measures of centrality based on betweenness. Sociometry, 35-41. doi:10.2307/3033543
‧
Gablasova, Dana , Brezina , Vaclav, & McEnery, Tony. (2017). Collocations in corpus‐based language learning research: Identifying, comparing, and interpreting the evidence.
Language Learning, 67, 155-179. doi:10.1111/lang.12225
Gries, Stefan Th. (2013). 50-something years of work on collocations. International Journal of Corpus Linguistics, 18, 137-166. doi:10.1075/ijcl.18.1.09gri
Griffiths, Thomas L, & Steyvers, Mark. (2004). Finding scientific topics. Proceedings of the National academy of Sciences, 101, 5228-5235. doi:10.1073/pnas.0307752101
Halliday, M.A.K., & Hasan, Ruqaiya. (1989). Language, context, and text: Aspects of language in a social-semiotic perspective. Oxford, UK: Oxford University Press.
Hoey, Michael. (1991). Pattern of lexis in text. Oxford, UK: Oxford University Press.
Huang, Chu-Ren, & Hsieh, Shu-Kai. (2010). Infrastructure for cross-lingual knowledge representation-towards multilingualism in linguistic studies. Taiwan NSC-granted Research Project (NSC 96-2411-H-003-061-MY3).
Huang, Ting-Hao, Chen, Yun-Nung, & Kong, Lingpeng. (2015). Acbima: Advanced Chinese bi-character word morphological analyzer. Paper presented at the Proceedings of the Eighth SIGHAN Workshop on Chinese Language Processing.
Huang, Ting-Hao, Ku, Lun-Wei, & Chen, Hsin-Hsi. (2010). Predicting morphological types of Chinese bi-character words by machine learning approaches. Paper presented at the Proceedings of LREC.
Hunston, Susan. (1995). A corpus study of some English verbs of attribution. Functions of Language, 2, 133-158. doi:10.1075/fol.2.2.02hun
‧
and in a corpus. In A. Partington, J. Morley, & L. Haarman (Eds.), Corpora and Discourse (Vol. 9, pp. 157-188). Bern, CH: Peter Lang.
Hunston, Susan. (2007). Semantic prosody revisited. International Journal of Corpus Linguistics, 12, 249-268. doi:10.1075/ijcl.12.2.09hun
Khedr, Ayman E, Salama, S.E., & Yaseen, Nagwa. (2017). Predicting stock market behavior using data mining technique and news sentiment analysis. International Journal of Intelligent Systems and Applications, 9, 22. doi:10.5815/ijisa.2017.07.03
Krestel, Ralf, Fankhauser, Peter, & Nejdl, Wolfgang. (2009). Latent dirichlet allocation for tag recommendation. Paper presented at the Proceedings of the third ACM conference on Recommender systems.
Ku, Lun-Wei, Liang, Yu-Ting, & Chen, Hsin-Hsi. (2006). Opinion extraction, summarization and tracking in news and blog corpora. Paper presented at the Proceedings of AAAI.
Li, Meixia, & Jiao, Aihui. (2017). The study of semantic prosody of Chinese logical resultative formulae: A corpus-assisted discourse analysis approach. GSTF Journal of Law and Social Sciences (JLSS), 2.
Louw, Bill. (1993). Irony in the text or insincerity in the writer? The diagnostic potential of semantic prosodies. Text and technology: In honour of John Sinclair, 240, 251.
Louw, Bill. (2000). Contextual prosodic theory: Bringing semantic prosodies to life. Words in context: A tribute to John Sinclair on his retirement, 48-94.
Manning, Christopher D., & Schütze, Hinrich. (1999). Foundations of statistical natural language processing. Cambridge, MA: MIT Press.
‧
Morley, John, & Partington, Alan. (2009). A few frequently asked questions about semantic—or evaluative—prosody. International Journal of Corpus Linguistics, 14, 139-158.
doi:10.1075/ijcl.14.2.01mor
O'Halloran, Kieran. (2007). Critical discourse analysis and the corpus-informed interpretation of metaphor at the register level. Applied Linguistics, 28, 1-24. doi:10.1093/applin/aml046 Pang, Bo, & Lee, Lillian. (2008). Opinion mining and sentiment analysis. Foundations and Trends
in Information Retrieval, 2, 1-135. doi:10.1561/1500000011
Partington, Alan. (1998). Patterns and meanings: Using corpora for English language research and teaching (Vol. 2). Amsterdam and Philadelphia: John Benjamins.
Partington, Alan. (2004). "Utterly content in each other's company": Semantic prosody and semantic preference. International Journal of Corpus Linguistics, 9, 131-156.
doi:10.1075/ijcl.9.1.07par
Partington, Alan. (2017). Evaluative clash, evaluative cohesion and how we actually read evaluation in texts. Journal of Pragmatics, 117, 190-203.
doi:10.1016/j.pragma.2017.06.008
Pennington, Jeffrey, Socher, Richard, & Manning, Christopher. (2014). Glove: Global vectors for word representation. Paper presented at the Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP).
Rosch, Eleanor. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology. General, 104, 192. doi:10.1037/0096-3445.104.3.192
Sinclair, John. (1991). Corpus, concordance, collocation. London, UK: Oxford University Press.
Sinclair, John. (2004). The search for units of meaning Trust the Text (pp. 34-58). London, UK:
Routledge.
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
of words and constructions. International Journal of Corpus Linguistics, 8, 209-243.
doi:10.1075/ijcl.8.2.03ste
Stubbs, Michael. (1995). Collocations and semantic profiles: On the cause of the trouble with quantitative studies. Functions of Language, 2, 23-55. doi:10.1075/fol.2.1.03stu
Stubbs, Michael. (2001a). On inference theories and code theories: Corpus evidence for semantic schemas. Text - Interdisciplinary Journal for the Study of Discourse, 21, 437-465.
doi:10.1515/text.2001.007
Stubbs, Michael. (2001b). Words and phrases: Corpus studies of lexical semantics. Oxford, UK:
Blackwell.
Wang, Shih-Ming, & Ku, Lun-Wei. (2016). ANTUSD: A large Chinese sentiment dictionary.
Paper presented at the LREC.
Wei, Naixing, & Li, Xiaohong. (2014). Exploring semantic preference and semantic prosody across English and Chinese: Their roles for cross-linguistic equivalence. Corpus Linguistics and Linguistic Theory, 10, 103-138. doi:10.1515/cllt-2013-0018
Xiao, Richard, & McEnery, Tony. (2006). Collocation, semantic prosody, and near synonymy: A cross-linguistic perspective. Applied Linguistics, 27, 103-129. doi:10.1093/applin/ami045
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Appendix
Appendix A List of negators
未 wei ‘not’
並未 bingwei ‘not’
並不會 bingbuhui ‘not’
沒有 meiyou ‘no’
不 bu ‘no’
沒 mei ‘no’
不會 buhui ‘cannot’
不用 buyong ‘no need to’
也不會 yebuhui ‘not’
無法 wufa ‘cannot’
‧ 國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
List of PREVENTION words
以免 yimian ‘so as not to’
避免 bimian ‘avoid’
以避免 yibimian ‘to avoid’
免得 miande ‘lest’
預防 yufan ‘prevent’
防止 fanzhi ‘prevent’
以防 yifan ‘prevent’
‧
Top 15 relevant words under topic society
交通
‧
Top 15 relevant words under topic entertainment
精品 表演 法律
‧
Top 15 relevant words under topic international
競選 軍事 投資
‧
Top 15 relevant words under topic sports
籃球 體壇
‧
Top 15 relevant words under topic finance
外幣
‧
Top 15 relevant words under topic lifestyle
攝影 醫療