透過您的圖書館登入
IP:52.91.255.225
  • 學位論文

多義詞的分析及其在語料庫標記的應用: 以賽夏語為例

Analysis of Polysemy and Its Applications to Corpus Documentation: A Study Based on Saisiyat

指導教授 : 蘇以文
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


詞彙的多義現象是語言學、哲學及心理學共同感興趣的議題。雖然一詞多義理論上會造成溝通的誤解,但是這樣的誤解很少出現在日常語言使用中 (Taylor 2003);多義詞造成困擾,多半是在跨語言溝通的情境 (Riemer 2001)。這促使我們將多義詞的研究延伸到跨語言的比較,而本論文即以語料庫標記為主要探討目標。我們從「台大台灣南島語多媒體語料庫」的工作經驗出發,以三個賽夏語的多義詞彙為案例,分析語言田野調查過程中,透過發音人的解釋和翻譯了解多義詞的過程。 在田野調查中,確認一個詞彙的意義是相當不容易的,這受到兩個因素的影響:詞彙語意的環境調節(contextual modulation) 以及跨語言的語意切分落差 (mismatches of semantic partitioning)。環境調節為某個詞彙增添或改變語意,因此當該詞彙每次出現在一個新的語境中,就有可能得到不同的解釋。語構對詞彙意義的影響亦不容忽視。此外,語意切分的方式在各個語言中往往有落差,一群概念在賽夏語中被視為相互有關聯,並用同一詞彙表達,在中文中他們之間的關係卻可能被忽略,而用多個不同的詞彙來表達。當一個賽夏語詞彙對應到多個中文翻譯時,我們需要透過分析和比較才能知道哪一個翻譯最能表現該詞彙的核心語意。 我們分析了三個詞彙,發現在呈現語料時,超語言翻譯的影響是很直接的。例如賽夏語的nanaw可以用於表達數量的限定,接近中文的「只」,也可用於表達對某項事件真實度的肯定,接近中文的「原本」。類似的語意延伸在其他的語言中並不多見,也因此發音人所提供的直接翻譯詞看似彼此並無關聯,容易產生同型異義(homonomy)和多義詞的混淆。另一詞彙nahaen用於表達事件重複、事件接續以及事件先行,類似的語意延伸較容易在其他語言中發現,其中文翻譯「再」、「又」、「還」等等,大多在中文中都有明顯的功能重疊,nahaen的用法之間的關係也較容易想像。而少數的辭彙,例如ma',意義接近中文的「也」,其語意延伸有很高的跨語言可預測性,類似的語意網絡常出現在不同的語言中,其中文翻譯一致性較高,不容易產生同型異義的誤解。 分析的結果可以從兩方面來討論。理論面上,跨語言多義詞語意發展的特殊性促使我們重新審視「觀點化」(perspectivization) 現象。一般認為語意變化的途徑是受到主觀化的影響 (Langacker 1990),意即以說話者的觀點出發。近來則興起互動主觀化的概念(Traugott 2003),強調說話者的觀點選定需配合其言談對象。我們的研究支持互動主觀化,並認為互動主觀化的考量對象應包含整個語言社群,方能解釋語言獨特性的產生。 應用面上,我們認為單一詞彙得到多種中文翻譯,是田野調查不可避免的過程。然而經過分析和對照之後,多義詞在語料標記中宜採用一致的超語言注解 (meta-language gloss) (Lehmann 1982)。其優點有五: 1) 代表語意和語用層面的基本切分,2) 有利語料庫使用者了解多義詞和同型異義的區分,3) 標記的經濟性,4) 利於語料庫內的詞彙搜索,5) 保存濱危語言的獨特語意切分方式。我們也提出,多義詞的本質為一種分類 (categorization),而一個多義詞的原型 (prototype) 可作為尋找超語言註解的根據。在一個詞彙的眾多可能意義當中,超語言註解反映的為其原型意義。 本文結合了單一語言的多義詞研究和跨語言的多義詞比較,除理論方面的思考外,並探討語意學研究在語言田野調查的進行和語料標記的應用,有助於我們在研究意義的過程中採用更寬廣更活用的觀點。

並列摘要


Being an important issue, polysemy has received a great much attention from different fields such as linguistics, philosophy, and psychology. In everyday conversation, language users typically generate appropriate interpretation of a polyseme with no difficulty (Taylor) — Polysemy is found to cause problems mainly in cross-linguistic contexts (Riemer 2001). We are thus motivated to investigate polysemy in a cross-linguistic context. Based on our experience of working on the creation and maintenance of the NTU Corpus of Formosan languages, we aim to explore documentation of polysemes by case studies on Saisiyat polysemy, using Mandarin as a meta-language. In the course of language fieldwork, there is no known method of determining precisely the meaning of a lexical item. Two factors are directly responsible: contextual modulation and mismatches of semantic partitioning. The meaning of a lexical item is modulated in situated contexts, and new meaning emerge almost whenever a lexical item is used in a novel context. In addition, semantic partitioning is also a major source of misunderstanding. For concepts that are conceived as relevant in Saisiyat, their relations may not be highlighted in Mandarin and are thus expressed with formally irrelevant linguistic forms. When one Saisiyat lexical item yields multiple Mandarin translations, we need careful collation to determine which translation reflects the core meaning of the delimited lexical item. Our analysis of three Saisiyat lexical items reveals that meta-language translation has a direct impact on our interpretation of the linguistic data. For example, Saisiyat nanaw denotes limitation of a quantity, like English ‘only,’ and it is also used to express an affirmative attitude to the factual status of a statement, like English ‘exactly.’ These two meanings are related, but such extension is found to be typologically-unimportant ones manifested in few languages, and its meta-language translations are seemingly irrelevant, which can yield a homonymy reading. Another lexical item nahaen is used to denote repetition, succession, and precedence of activities. Similar semantic network are more likely to be found in other languages, and many of its Mandarin translations exhibit functional overlaps, which may help corpus users to discover the relations between instances of nahaen. We may also come across cases of very high degree of cross-linguistic predictability, yielding similar ways of conceptual categorization in genetically-unrelated languages, as well as relatively consistent direct translation. Our study of Saisiyat ma' ‘also’ is one of the examples. Theoretically, language-specificity of semantic partitioning urges us to take a social-cultural view on perspectivization. It is commonly held that the development of a network exhibits tendency of “subjectification”— the speaker tends to include his own epistemic attitude and personal evaluation when using an expression (Langacker 1990). Traugott (2003) further postulates that speaker’s point of view has to align with that of his addressee, a tendency known as “intersubjectification.” The intersubjectification view is supported by our study, but we claim that the speaker’s construal of a scene for the purpose of verbalization has to take into the shared linguistic and non-linguistic knowledge of the entire speech community. A “collective” view of intersubjectification is postulated to account for language-specificity. Empirically, we propose that a consistent gloss of a polyseme can be sought on the basis of its prototype. We agree that inconsistent and imprecise translation of a lexical item in different contexts is inevitable in linguistic fieldwork. A polyseme is nevertheless advisable to be glossed consistent by one meta-language gloss (Lehmann 1982). Consistent glossing yields five advantages: 1) reflecting fundamental division between semantics and pragmatics, 2) facilitating user’s identification of a linguistic item, 3) exhibiting economy and precision of data presentation, 4) facilitating search of a lexical item in a corpus, and 5) preserving conceptual categorization of the target language. Based on the categorization view of polysemy, we propose that the meta-language gloss of a polyseme should reflect the prototype of the polyseme. When a researcher wishes to gloss a polyseme by a consistent cover term, he can exploit the notion of prototype. Overall, our investigation integrates intra-language investigation of polysemy with inter-language comparison of semantic partitioning. In addition to examination of theoretical issues, we tackle empirical problems of semantic analysis in language fieldwork with special focus on the possible applications to language documentation.

參考文獻


SANFORD, ANTHONY, and SIMON GARROD. 1981. Understanding written language. Chichester: John Wiley & Sons.
HUANG, SHUANFAN. 1998. Emergent lexical semantics. Selected Papers from the Second International Symposium in Language in Taiwan, ed. by Shuanfan Huang, 129-50. Taipei: Crane.
GRAUMANN, CARL F., and WERNER KALLMEYER. 2002. Perspective and perspectivization in discourse. Perspective and perspectivization in discourse, ed. by Carl F. Graumann and Werner Kallmeyer, 1-14. Amsterdam: John Benjamins.
BLUST, ROBERT. 2002. Notes on the history of 'focus' in Austronesian languages. The history and typology of western Austronesian voice systems, ed. by Fay Wouk and Malcolm Ross, 63-78. Australia: Australian National University.
HIMMELMANN, NIKOLAUS P. 2002. Voice in western Austronesian: An update. The history and typology of western Austronesian voice systems, ed. by Fay Wouk and Malcolm Ross, 7-16. Australia: Australian National University.

延伸閱讀