中文版「語文探索與字詞計算」詞典之建立

The Development of the Chinese Linguistic Inquiry and Word Count Dictionary

日常生活語言的使用，包括書寫與口語，反映了個人內在的心理狀態、思考型態甚至人格特質。它為心理學研究提供了一扇探索心靈的窗口。Pennebaker研究團隊（回顧見Pennebaker, 2011）採用字詞計算的方式進行語文特性的分析，並發展出電腦程式「語文探索與字詞計算」（Linguistic Inquiry and Word Count, LIWC）。LIWC的核心在於其詞典，目前LIWC2007已可分析80個字詞類別，具有相當好的信、效度。本研究的主要目的即在建立中文版LIWC詞典，並比較其與英文LIWC2007詞典的對等性及效度檢驗。研究一依據原英文版LIWC詞典建立之基本程序，並經各階段嚴謹的反覆檢驗，並依據中文特性加上一些中文特有詞類，最後編修成包含30個語文類別與42個心理類別的中文版LIWC詞典。研究二則蒐集了100篇中、英文對照文章，進行分析比對。結果發現中文LIWC與LIWC2007所偵測到詞類百分比，大部分有極高的相關，可見兩個版本具有相當的對等性。研究三分析憂鬱文本的特性，發現相對於對照組，憂鬱文本使用較多的第一人稱單數代名詞、較少的第一人稱複數代名詞與較多的負向情緒詞。此一結果與使用LIWC分析的英文相關文獻結果一致，顯示中文LIWC也具有一定的效度。整體而言，本論文透過三個研究建立了與LIWC2007具相當對等性且具有效度的中文LIWC詞典。相信中文LIWC日後可提供華語文使用的心理特性分析一項研究利器。

The analysis of natural language opens a window to the exploration of thoughts, feelings, and personalities. To analyze word use, Pennebaker and his colleagues (Pennebaker, 2011) developed a computer software, Linguistic Inquiry and Word Count (LIWC). LIWC reports on the percentage of words in a text file that are in the grammatical, psychological, and content categories of its dictionary, and was created by judges who rated whether each entry belonged in a category. This paper describes the development of the Chinese LIWC dictionary and establishes its reliability and validity. Study 1 involved the translation of terms from the English LIWC dictionary to Chinese, and judges' ratings of whether each entry belonged in a category. Furthermore, some categories unique to the Chinese language were added to the Chinese LIWC dictionary. These resulted in a total of approximately 6,800 words across 30 linguistic categories and 42 psychological categories. In Study 2, we analyzed one hundred texts and their translations using both the English LIWC dictionary and Chinese LIWC dictionary. Fifty texts of varied genres and authors were written in English and translated to Chinese; others were written in Chinese and translated to English. Reliable correlations were found between the English LIWC and Chinese LIWC categories, indicating acceptable equivalence between the two dictionaries. In Study 3, we analyzed 30 bulletin board messages from a site for people suffering from depression, and 30 messages from a site on a control topic (personal experiences with part-time jobs). Similar to what past researchers have found in English, depressed people used more first person singular pronouns, fewer first person plural pronouns, and more negative emotion words than non-depressed people, confirming the validity of the Chinese LIWC. Just as the English LIWC has led to discoveries in the social sciences, the Chinese LIWC now opens new windows to the psychology of Chinese speakers and authors by providing an efficient means by which to analyze open-ended texts. Future research, applications, and revisions of the Chinese LIWC dictionary are discussed.