漢語之框觸動詞──「玩」、「弄」和「搞」在「動賓」格式中之研究

Abstract

摘要

本文主要研究現代漢語中的三個「框架觸發動詞」--「玩」、「弄」和「搞」在「動賓」格式中之格式語意。為避免掉入傳統研究漢語詞彙語意的窠臼，本文採用中研院平衡語料庫的語料，檢討文獻上對動賓結構的研究，進而採取當前功能學派中的格式語法來闡述這三個動詞在「動賓」格式中的格式語意。本文的研究成果與傳統方法比較起來有較多的突破與進展。 首先，本文對於框架語意研究的概念先取得應有的認識，然後決定採用格式語法來解釋漢語「動賓」格式所能提供的格式語意，而就漢語的這三個動詞而言，他們所帶的賓語其語意類型相當的豐富，因此若是要一個動詞接一個賓語就給一個語意，那並不符合我們人類大腦的基本生理限制。但是，若我們把動詞本身設定成一個語意框架的方式來研究這類「動賓」格式所隱含的語意，即可發現這樣的句法形式與語意框架的結合，比較可以展現該格式所擁有的隱含事件語意並可提取整個框架所能提供的語意屬性。 其次，本文就傳統研究動賓結構的詞彙語意方法與格式語法作一個比較。我們發現原有對於動賓結構的語意描述，都是單純的從動詞與賓語的結合來取得整個結構的語意。但是，這樣的方法對於一些賓語極其豐富的動賓結構，例如：搞飛機/女人/台獨/統一等等，卻無法提出有效的解釋，亦即整個結構所隱含的事件之語意其實不能由動詞和賓語的單獨結合來描述。因此，我們認為當前格式語法所提出的句法型態與語意的對應組合能提供這類「動賓」格式較適切的語意詮釋。 再者，本文除了發現格式語法可以透過後設語言明確描述隱含的事件，並且可以透過對整個格式語意成分的提取，儘管在某些情況下尚不成熟，但可將這三個常項動詞設定成整個語意框架內的「代動詞」，換句話說，這三個動詞的地位是在一般詞彙之上，因為他們不像一般動詞那樣只給所帶出的事件一個「動相類型」(event type)，而是可以提供整個語意框架較多的語意詮釋：可能是一個事件狀態的改變，例如：產生一個「結果狀態」，或是使整個框架得到一個作說者對待事件的「角度」，例如：作說者用比較「貶義的角度」來看待該隱含的事件等等。所以，這類動詞是可以代替具有同樣語意成分的動詞而出現在一般的言談當中的。 最後，雖然這類框觸動詞的「動賓」格式可以兼顧許多好處，不需要隨時改變語意框架就可以適應各種不同賓語的語意類型並加以解釋，但有時候也會遇到岐義的現象，這時候就可以利用Qualia Structure來針對賓語提供不同的語意屬性，以區辨同樣的賓語但有不同的解讀時的岐義。於是，我們就可以針對這樣的「動賓」格式提出較合理的語意解釋。

ABSTRACT

This thesis aims to investigate the frame semantics of three frame-evoking Mandarin verbs, including WAN, NONG, and GAO, in the [V+NP] construction. The thesis tries to avoid following the old fashion on the lexical semantic study, so it retrieves data from the Sinica Corpus and discusses some cases in which traditional lexical semantic studies took on the V-O pattern. Moreover, it follows the most popular Construction Grammar to demonstrate the constructional interpretation of the [V+NP] construction that these three verbs embedded. Consequently, comparing with the traditional approach applied in this field, immense progress made by Construction Grammar is beyond our expectation. First of all, this thesis gathers some background knowledge about the frame semantics, and then adapts Construction Grammar on the frame semantic study of [V+NP]. Since, as far as these three verbal predicates are concerned, the object-NPs bear quite a few semantic types in the semantic domain, it is impossible for human beings to remember all meanings of each V-O pattern. That is, there are not enough neurons in our brain to deal with all of the linguistic noise. Therefore, when we use the frame-oriented verbal predicate to set a frame, we can easily acquire the semantic denotation hidden in the opaque episode undertaken. Namely, the form-meaning composition provides a proper sense for the construction per se. Hence, the semantic features are easily to be acquired. Second, this thesis also compares the traditional lexical semantic studies with the approach of Construction Grammar. We can find that the semantic description of the V-O pattern in the old fashion is not convincing, because object-NPs are with diverse semantic types, such as airplanes, women, Taiwan Independence, unification and so on. They are not just combined with GAO ‘do/make’ and then render a proper reading for the V-O pattern. Besides, the sense of the opaque event would not be the combination of the derivational parts of the V-O pattern. In consequence, we suggest that only Construction Grammar can demonstrate the most elaborate constructional interpretation inferred from the compositionality of the syntactic form and the semantic frame. Third, on the base of Construction Grammar, this thesis not only uses metalanguage to interpret the vague event, but also retrieves the semantic components from the whole construction. In other words, this thesis postulates: although these three verbs are not full fledged pro-verbs in some cases, they behave like pro-verbs that they are ‘super-lexical’ morphemes and act differently from other predicates which just render a situation type for the event. For example, the frame-evoking verbs may change the situation type of an activity into a resultative state, or imply that the speaker takes business as an informal event in a rather pejorative tone. Accordingly, these verbs are probably to take over some verbs with the identical semantic features in a certain context. Lastly, even though there are bunches of merits in Construction Grammar, we may not read an ambiguous pattern in a context. However, Qualia Structure has a killer-usage to distinguish the ambiguity undertaken in the [V+NP] construction. Since there are four idiosyncratic qualia roles that can be used to determine the main idea of the object-NP, then we can easily decipher the encoded reading of the construction.