雅美語的am結構

Abstract

本論文旨在探究雅美語的am句法結構，此結構在文獻中已指定為「主題句法結構」。在菲律賓南島語的文獻，「主題」長久被使用為被標記成主格格位的名詞組的名稱，但本論文將「主題」當成信息結構(information structure) 的核心概念之一來理解，代表言談中已知訊息，與「焦點」或「評論」(focus/comment) 表達的未知訊息（用來描述主題的動作或狀態）做對比。此「主題評論結構」(topic-comment structure) 呈現在句法的特定結構上，將主題置於句首，主題的後面也可以加構詞性的主題標記。上述的主題結構在雅美語會附上主題標記am，此結構為本論文的研究對象。 參考Chafe (1976)對不同語言中主題的形式所提出的分類法，雅美語am 結構所觀察到的使用功能可以分成兩類型：即標記「主題」(topics) 以及標記「附屬字句」(subordination)。主題包含一般名詞組、代名詞、關係子句、以及擬分裂句(pseudoclefts)，一律標記成主格， 但是在一般句型通常使用屬格標記的主事者與擁有者論元也可以當主題。am 結構的附屬子句為各種各樣的副詞子句及修飾句，含：相關性(aboutness)、處所、時間、條件、以及引述(quotative) 之主題。因此，am 結構裡出現的附屬子句的形式相對多元與自由，句法限制不如 論元主題嚴格。 另一方面，本論文也證明am結構的主題標記功能所表達的主題成分到底是否為「主題」。依據Gundel et al. (1993) 的「已知性等級體系」(Givenness Hierarchy) 提出共6 等級的「認知狀態」(cognitive status) ，本論文將自然言談中出現的所有主題進行分析及討論，並發 現大部分的主題對應到高等級（第一到第四最高）的認知狀態。高等級認知狀態代表一個指涉物(referent) 在言談中已提過，指示高度的已知性，因此能夠證明雅美語的主題結構的確具有標記主題的功能。本論文也討論對應到低等級認知狀態的兩筆語料：一筆只能當例外，但另一筆為「泛指主題」(generic topic)。由於泛指主題被其他學者歸類為主題的一種(Lambrecht 1994; Lee 1996)，屬於「已知性等級體系」的問題之一，本論文在此時無法解決。

This thesis investigates the am-construction in Yami, a Batanic (Philippine) language indigenous to Orchid Island, Taiwan. The am-construction has been previously identified as a topic construction, and although topic has been traditionally used in the realm of Philippine languages to refer to the nominative-marked noun phrase of the Philippine-type voice system, we take topic to be an information structural notion that is given in relation to its counterpart of focus/comment. This topic-comment structure is manifested through syntactic constructions in which topics are overtly expressed in sentence-initial position and optionally followed by a morphological topic marker. This particular construction in Yami, distinguished by the topic marker am, is identified as our object of study. Loosely based on a dichotomy first established by Chafe (1976), two usages of the am-construction are recognized: namely, as a topic construction and as a subordinating construction. Morphosyntactically, it is found that topics (nominal core arguments) must take nominative case, although genitive agents and possessors may also be admitted, and these restrictions apply to nominal and pronominal arguments as well as relative clauses and pseudoclefts. On the other hand, subordinate clauses in the am-construction are comprised of adjunct and adverbial clauses, and therefore exhibit a wide diversity of forms; these include aboutness, locative, temporal, conditional, and quotative clauses. The topic-marking function of the am-construction is then examined in further depth using naturalistic data in order to determine that the constituents expressed using the construction are, indeed, topics. Using Gundel et al.’s Givenness Hierarchy (1993), topics are analyzed and discussed in terms of their cognitive status, which represents the strength of their referential givenness. The vast majority of topics are found to correspond to the highest four levels of referential givenness, providing evidence of the construction’s topic-marking properties. However, two outlying tokens of low status are also discussed; while one is suggested to be an unplanned error, the other is identified as a generic topic, which is a recognized valid topic and therefore poses a problem for the Givenness Hierarchy.