對象性介詞後接動詞初探—以「向」、「對」、「跟」為例

本文旨在探究對象介詞「向」、「對」、「跟」的使用是否與後接動詞類型、前後名詞短語有關，即這兩個部分的不同是否影響對象介詞的使用偏向。周小兵（1997）、李增吉（2007）、《現代漢語虛詞例釋》（1982）、呂叔湘（1980）、景士俊（1980）等書籍及學者們從對象性介詞的研究從句式、語義入手進行探究，有關對象類介詞後接的動詞種類及名詞種類之考察可能存在缺漏，故本文想以此為入手點。二語學習者在學習三個詞時會遇到解釋相似，用法相似，無法區別的困境，不利於學習。本文基於大型語料庫：HSK 2.0語料庫、暨南大學留學生語料庫、TOCFL語料、中研院平衡語料庫、BCC語料庫。三個介詞母語者、二語者的語料各500例，共3000例，進行卡方及費雪分析並解釋三個對象性介詞，動詞、名詞短語使用偏向不同之處。結果初步發現，母語者及二語者語料中，後接詞有明顯偏向：「向」、「對」後接「虛向的動作性動詞」佔比為最高（母語者「向」70.60%、「對」76.80%；二語者「向」71.2%、「對」66.80%）。在「跟」語料中，皆為「言語動詞」的比例最高（母語者「跟」46.80%；二語者「跟」39.20%）。通過進一步分析，可以發現前後名詞短語的生命性與介詞後接動詞有一定的規律：「向」、「對」語料的前後NP出現「個體生命」、「群體生命」使用頻率相反，而「跟」語料中前後NP呈現出相同的趨勢，即兩者皆為「個體生命>群體生命」。「不發生變化的動作」偏向為「向」：一對多，「對」：多對一，「跟」：一對一。筆者通過收集母語者問卷的方式，測試母語者在設定的條件中，是否存在使用偏向，並對數據進行卡方檢定，得到的結果，六組中僅有一組差異不顯著（p值高於0.001），此組測試對象為：「對」加三種動詞以及個體名詞。這個結果意味著「個體名詞」無法影響介詞「對」與動詞的選取使用。另外由於，問卷中「向」與「跟」的數據過低，且過於相似所以才導致檢定，結果不顯著。此結果為受試者使用偏向造成的。

關鍵字

後接動詞；對象介詞；動詞分類

並列摘要

The purpose of the present study is to explore whether the use of the object prepositions ‘xiang’, ‘dui’ and ‘gen’ is related to the type of the verb following the preposition and the noun phrase preceding and following the preposition, that is, whether the difference between these two parts affects the use of object prepositions. Books and scholars such as Zhou Xiaobing (1997), Li Zengji (2007), "Explanation of Function Words in Modern Chinese" (1982), Lu Shuxiang (1980), Jing Shijun (1980), etc. start from the clause pattern and semantics of the research on object prepositions. To conduct research, there may be gaps in the investigation of the types of verbs and nouns that follow object-type prepositions, so this article would like to take this as a starting point. When second language learners learn three words, they will encounter the dilemma of having similar explanations, similar usages, and being unable to distinguish them, which is not conducive to learning. When L2 learners learn the three prepositions, they will encounter the dilemma of similar interpretation, similar usage, and indistinguishability, which is not conducive to learning. The present study is based on large language corpora: HSK 2.0, Jinan University International Student Corpus, TOCFL Corpus, Academia Sinica Balanced Corpus, and BCC Corpus. For each preposition, 500 sentences are extracted from native speaker and second language (L2) learner corpora, respectively, for a total of 3000 cases. Chi-square and Fisher exact test analyzes were conducted to explain the differences in the use bias of the three object prepositions, verbs and noun phrases. The preliminarily results found that in the corpora of native speakers and L2 learners, the following words have obvious biases: ‘xiang’, ‘dui’ followed by "empty action verbs" accounted for the highest proportion (native speakers: ‘xiang’ 70.60%, ‘dui’ 76.80%; second language speakers ‘xiang’ 71.2%, ‘dui’ 66.80%). Within the data for ‘gen’, the proportion of "verbal verbs" is the highest (46.80% for native speakers; 39.20% for second-language speakers). After further analysis, it was found that there are certain rules between the animacy of the noun phrases before and after the preposition and the verbs followed by the preposition: the frequency of use of "individual life" and "group life" in the ‘xiang and ‘dui’ corpora are opposite, while the same trend is shown in the NPs preceding and following the preposition in the corpus data, that is, both are "individual life > group life". "Doing work" is biased towards ‘xiang’: one-to-many, ‘dui’: many-to-one, and ‘gen’: one-to-one. The author collected questionnaires from native speakers to test whether native speakers have usage bias under the set conditions, and conducted a chi-square test on the data. The results obtained showed that only one of the six groups had no significant difference (p>0.5), the test objects of this group are: ‘dui’ plus three verbs and individual nouns. This result indicates that "individual nouns" do not affect the selection and use of the preposition ‘dui"’and verbs. In addition, because the data of ‘xiang’ and ‘gen’ in the questionnaire are too low and too similar, the test results are insignificant. This result is caused by participants’ usage bias.