語言探索與字詞計算詞典2015簡體中文版之建置與應用

Pennebaker研究團隊所發展的語言探索與字詞計算（LinguisticInquiry and Word Count，簡稱LIWC）程式，以電腦程式計算特定詞彙類別的使用頻率，表徵個體內在的心理狀態，為文本分析開啟了嶄新的可能性。目前已有多國語言版本，繁體中文版本先後針對英文版LIWC2007和LIWC2015修訂了對應之繁體中文詞典，為探索中文使用者的心理特性提供有效之研究工具。然而，中文使用者係以簡體中文為大宗，對於在繁體中文使用環境下所建置的LIWC詞典能否適用於簡體中文環境，目前尚無實徵研究對此進行系統性的檢驗評估。因此，本研究即以繁體中文版LIWC詞典為基礎，考量兩岸用語習慣與風土民情之差異，建置簡體中文版LIWC2015詞典。本文共分為四個研究，研究一詳細說明簡體中文版LIWC2015詞典的建置過程；研究二針對富含在地文化特色之網路詞類別單獨修訂並確立有效性；研究三與研究四皆為簡體中文版LIWC2015詞典的效度檢驗。綜合四個研究，確立簡體中文版LIWC2015詞典對簡體中文文本的分析效果穩定，可作為未來探討簡體中文使用者心理特性之研究工具。

Deciphering the intricacies of individuals’thoughts and inner processes has emerged as a significant concern within psychology. Historically, the primary approach to apprehending these dimensions involved meticulous content analysis of textual data, an approach grounded in subjective judgments and often labor-intensive. However, the recent technological surge has ushered in a plethora of computational text analysis techniques. Prominent among them is the Linguistic Inquiry and Word Count (LIWC) tool pioneered by Dr. Pennebaker. LIWC’s core premise rests on the hypothesis that the frequency of word usage serves as a linguistic index, shedding light on individual affective states and cognitive processes. Its applicability spans many social science disciplines, including psychology, education, and health.
While LIWC is bifurcates into a dictionary encompassing specific categories and their affiliated words, and a computational algorithm calculating the frequency of word usage in each category, the former is undeniably its cornerstone. The original LIWC dictionary was in English and has been translated and revised into more than ten languages. Notably, the inaugural Chinese variant, TC-LIWC2007, was adapted by Huang et al. (2012) and subsequently modernized to TC-LIWC2015 by Lin et al. (2020). However, both editions predominantly cater to Traditional Chinese. It still remains unclear whether a dictionary simply translated from Traditional to Simplified Chinese is appropriate for Simplified Chinese textual analysis. This research, therefore, aims to formulate a Simplified Chinese edition, SC-LIWC2015, drawing from TC-LIWC2015.
This endeavor spanned four meticulously curated studies, each underscoring a unique facet of the SC-LIWC2015 adaption process. Study 1 delineates the details of the SC-LIWC2015 revision process. Meticulous cross-checking ensured translation fidelity from Traditional to Simplified Chinese, with particular attention to linguistic nuances between the two. We also took into account the differences in word usage habits between Traditional and Simplified Chinese, particularly when the same word has different meanings or when different words convey the same meanings. Study 2 hones in on establishing the“Netspeak”category. Words in the Netspeak category are characterized by terminology prevalent in digital communication platforms, such as social media or short message services. Recognizing this category’s cultural and dynamic nature, we embarked on a comprehensive collection and validation of prevalent terms from the Simplified Chinese digital milieu rather than direct translation. Study 3 and Study 4 diverge in their textual foci but converge on validating SC-LIWC2015. Study 3 investigated the psychological difference between individuals currently in romantic relationships and those recently separated. Additionally, Study 3 also assessed the effectiveness of Jieba, a word-segmentation tool commonly used for Simplified Chinese. The findings aligned with our hypotheses. Specifically, those in ongoing romantic relationships employed more first-person plural pronouns and affiliation words while using fewer negative emotion words and cognitive process words. Moreover, when comparing texts segmented by Jieba and CKIP, a highly accurate and widely utilized word-segmentation tool, consistent LIWC patterns were observed. This underscores the efficacy of using Jieba in processing texts in Simplified Chinese. Study 4 collected textual data from Weibo (Simplified Chinese) and PTT (Traditional Chinese), centered on occupational and depression experiences. These were analyzed using SC-LIWC2015 and TC-LIWC2015, respectively. The findings from Study 4 aligned with prior research. Texts centered on occupational experiences showed a higher frequency of achievement, work, and money-related words. In contrast, those focusing on depression experiences exhibited a greater use of first-person singular pronouns and negative emotion words. The consistent outcomes from SC-LIWC2015 and TC-LIWC2015 underscored the robustness of SC-LIWC2015. In conclusion, the results from these studies suggest that SC-LIWC demonstrates notable accuracy and reliability in interpreting texts written in Simplified Chinese and in capturing the inner process of its users.