Marking Trustworthiness with Near Synonyms: A Corpus-based Study of “Renwei” and “Yiwei” in Chinese

Marking Trustworthiness with Near Synonyms: A Corpus-based Study of “Renwei” and “Yiwei” in Chinese Bei Li Chu-Ren Huang Si Chen Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University Hong Kong SAR benita-bei.li@connect.polyu.hk churen.huang@polyu.edu.hk sarah.chen@polyu.edu.hk Abstract We conducted a corpus-based study on near synonymous cognitive verbs “renwei 認為” and “yiwei 以為” with a similar meaning of “to think” in Chinese. The motivation of this study is that information trustworthiness in propositions var- ies with the use of “renwei” or “yiwei”, which concerns the issue of speakers’ stances or atti- tudes. We examined the epistemic modality at- tributed to “renwei” and “yiwei” in written discourse, by testing their different functions of evidential marking, negative forms and hedging device. Results showed that “renwei” was asso- ciated with specialized and professional subjects and consequently demonstrated its relatively high degree of evidentiality. Then, specific neg- ative polarity shifters exclusively collocating with “yiwei” denoted larger scope of negation of entire propositions. Lastly, only “renwei” func- tioned as a hedge to mitigate claims lacking full commitment in the political discourse. Therefore, the evidence of data proved that the epistemic marker “renwei” was associated with a higher degree of information trustworthiness, contrib- uting to the annotation system in NLP for the de- tection of information credibility. Keywords: information trustworthiness; near- synonyms; epistemic modality 1 Introduction Epistemic modality is an important topic on seman- tics, functioning in judging and evaluating the prop- ositional content with regard to the degree of certainty, possibility or necessity in our communi- cation. Speakers or writers use epistemic markers, like cognitive verbs (e.g. “think”/“believe”) or modal auxiliaries (e.g. “may”/“could”), to express their attitudes or stances-taking toward the proposi- tions. Recently the linguistically decoded infor- mation involving cognition has been an important topic to be faced in natural language processing (NLP). For example, in the statement “I thought he always supported me”, the speaker’s intention is to emphasize that “he” does not support the speaker now through the use of the past tense of the epis- temic verb “think”. Consequently, the trustworthi- ness of information seems complicated to decode in NLP with regard to the speaker’s attitudes and stances. The present study investigates one type of epistemic markers, the cognitive verbs in Chinese, to analyze the credibility of information epistemic verbs attributed using a corpus-based approach. Moreover, the use of near synonymous epistemic verbs makes it more complicated for text detection or information retrieval in Chinese. In the present study, the cognitive verbs “renwei” (認為) and “yi- wei” (以為) will be comparatively analyzed to de- tect the information trustworthiness contained in the statements they constituted respectively. The pair of near synonyms “renwei” and “yiwei” has a similar meaning of “to think” with the function of marking speakers’ epistemic modality. Specifically, “ren4” (認) is a verb referring to the act of identification or recognition, and “wei2” (為) is a verb denoting the action or the state of being (apply to both “renwei” 認為 and “yiwei” 以為). The preposition “yi3” (以) of “yiwei” means …

Content truncated to 500 words