- 學位論文
行為特徵分析: 以量化語料庫方法研究中文口語程度副詞「很」、「太」、「蠻」、「超」
Behavioral Profiles: A quantitative corpus-based study of degree adverbs HEN, TAI, MAN, CHAO in spoken Mandarin
摘要
對於第二外語學習者來說，能於日常生活中正確地使用中文程度副詞著實為一大挑戰。過去碩果累累的程度副詞研究中，較少透過口語語料庫探討程度副詞，鑑於不同文體與方言造就出語言之多變性，本研究旨在藉由量化語料庫方法中之行為特徵分析(Behavioral Profile Approach)研究中文口語程度副詞個別之使用特色。本研究首先收集政大口語語料庫(the NCCU corpus of spoken Taiwan Mandarin)釋出的部分語料(Taiwan Mandarin Corpus in TalkBank)，找出四個較為常用的程度副詞 – 「很」、「太」、「蠻」、「超」，並利用標記後的資料算出近義詞間的相對語意距離，而後本研究先採用階層式集群分析法(hierarchical cluster analysis)探究相近之群集，再採用事後比較檢定(post-hoc analysis)歸納出能顯著區辨程度副詞結構之重要語意特色。研究結果發現，「超」與「蠻」的語意傾向較為相似，而「太」與「很」亦有相近之語意特色。由於樹狀圖(dendrogram)顯示三組群集，各組則逐一於實驗一至實驗三進行比較：實驗一利用「超」和「蠻」的行為特徵與「太」和「很」進行比較從而分析兩兩近義詞間之相似性；實驗二藉由比較「超」與「蠻」探究相異之處，結果顯示，謂語的情感最能區辨程度副詞「超」與「蠻」，前者較經常修飾負面的謂語而後者則較傾向修飾正面的謂語。最後實驗三比較「太」與「很」之相異性，結果顯示，程度副詞「太」的句構中較常以抽象事物為主題，而程度副詞「很」的句構則較傾向講述關於人的主題。另外，根據行為特徵之百分比顯示，程度副詞主要修飾性質形容詞，然而除了性質形容詞外，其他詞類亦展現不同之傾向，例如：與外型有關之形容詞較常於「太」的句構中出現，而一般動詞則經常於「很」的句構被使用。
並列摘要
Learning how to utilize degree adverbs appropriately is an essential process for L2 learners of Mandarin Chinese. While previous studies have paid much attention to specific degree adverbs, few studies compare the usage of various degree adverbs at the same time, especially in spoken contexts. Therefore, this study investigated the similarities and differences of four most frequently used degree adverbs (i.e., hěn, tài, mán, and chāo) in Taiwan Mandarin Corpus in TalkBank. We analyzed their distinctive usage patterns at multiple linguistic levels, including semantic, syntactic, pragmatic, morphological, as well as phonological dimensions, to uncover the subtle distinctions between these near-synonymous words/constructions. With the state-of-art behavioral profile approach, we conducted a hierarchical cluster analysis to determine the semantic distances among these four degree adverb constructions. Additionally, in the post-hoc analysis, we computed pairwise differences of all the linguistic features and adopted Cramer's V to determine the distinctive linguistic features and preferred patterns for each degree adverb construction. Our hierarchical cluster analysis shows that the linguistic performances of chāo and mán are much more similar while tài and hěn share more commonalities. In addition to similarities, slight differences are also found in our post-hoc analysis. Our results show that hěn constructions are more often used for the predicate intensification of human entities while tài constructions are more often involved in the predicate intensification of abstract entities/incidents. Another interesting observation is that compared with other degree adverbs, action verbs are more likely to be modified by hěn while appearance adjectives show a stronger tendency of being modified by tài. On the other hand, concerning the differentiation between mán and chāo, the results show that the choice of mán or chāo may be attributed to the sentiment of the predicates. In other words, positive predicates are preferred in mán constructions while negative predicates co-occur more often with chāo constructions.