噶瑪蘭語反身詞與交互句型之研究

Abstract

本研究探討噶瑪蘭語的反身詞（reflexives）與交互句型（reciprocals）。在反身詞方面，噶瑪蘭語沒有一個特定的反身代名詞，而是使用人身代名詞與izip「身體」兩種方式來表達反身的概念。因此，使用人身代名詞會產生的語意詮釋有二：代名詞的語意詮釋，或是反身的語意詮釋。但是，在使用izip「身體」來表示反身時卻只會有反身的語意詮釋。事實上，izip「身體」在一些句法上的表現反而近似其他語言中的反身代名詞，例如，在表達反身時，izip「身體」的前行語（antecedent）必須在同一個句子內。 在研究噶瑪蘭語反身詞的約束關係時，我們發現，Chomsky (1981)所提出以討論代詞（anaphor）與代名詞分佈現象為主的約束條件（binding conditions），難以解釋噶瑪蘭的反身詞。相反地，Reinhart與Reuland (1993)所提出修正過後的約束條件，將反身關係視為整個述語的特性，探討反身述語在語意與句法表現上是否協調，而不談論反身詞的分佈問題。因此，Reinhart與Reuland (1993)較能適當地解釋噶瑪蘭語中反身述語的現象。此外，我們也發現在噶瑪蘭語中，反身關係是否能成立與論元結構（argument structure）有關。也就是說，無論述語是以主事者為焦點（agent focus）或是以非主事者為焦點（non-agent focus），反身詞都只能出現在論旨賓語（thematic object）的角色，而無法出現在論旨主語（thematic subject）的角色。 在交互句型方面，噶瑪蘭語主要以前綴sim-來表現交互關係。該詞綴除了表現交互的關係以外，亦可表示連鎖關係（chaining）、集體關係（collective），以及個別關係（distributive）。這些關係在語意上的共通點是，它們的參與者與參與者間都牽涉到複數的相互關係。這個語意上的共通點也反映在句法上－－這些句型都需要複數的主詞。 一些關於其他語言以詞綴表現交互關係的研究（例如Gerdts 2000、Mchombo 1993、McGregor 2000等等），在討論這些語言的交互句型為及物或不及物句型時，多認為由交互詞綴所衍生出來的句型皆為不及物句。但噶瑪蘭語的情況並非如此。我們發現，在用來表現交互關係時，sim-會減少動詞的一個論元。也就是說，sim-會將及物動詞轉變為不及物動詞，將雙賓動詞轉變為及物動詞。因此，sim-這個詞綴會改變論元結構、減少論元，但不一定會衍生不及物動詞。

This study investigates two types of anaphoric expression in Kavalan. The first one is reflexives. Kavalan does not have a unique reflexive marker. It employs personal pronouns and izip ‘body’ to express reflexives. While the use of personal pronouns may cause ambiguity between a pronominal interpretation and a reflexive interpretation, the use of izip ‘body’ does not have this problem. In fact, the combination of izip and a genitive pronoun resembles a true reflexive anaphor in several aspects, including that it is bound to a local antecedent. In terms of the binding of reflexives, it is found that Chomsky’s (1981) binding conditions, which describe the distribution of anaphors, cause difficulties in delimiting the reflexives in Kavalan. On the other hand, Reinhart and Reuland (1993) consider reflexivity as a property of predicates. By examining the relation between the syntactic form and the semantic content of reflexive predicates, Reinhart and Reuland’s (1993) revised binding conditions can better account for the issues related to reflexives of Kavalan. In addition, it is also found that Kavalan reflexive binding is sensitive to argument structure, instead of grammatical relation. Therefore, a reflexive anaphor can appear as a thematic object, but not a thematic subject, regardless of whether the predicate is AF or NAF. The second type of anaphoric expression is reciprocals. In Kavalan reciprocals are mainly marked on the verbs by an affix sim-. In addition to marking reciprocals, the same affix is also used to mark chaining, collective and distributive situations, which all share a semantic property – plurality of relations among participants. This semantic property is also reflected on the syntax. The sim-marked constructions all require plural subjects. While several studies concerning reciprocals of other languages conclude that the constructions derived by reciprocal affixes in these languages are intransitive (e.g., Gerdts 2000, Mchombo 1993, McGregor 2000, etc.), I have found that the same conclusion cannot be made for Kavalan. Although transitive verbs indeed become intransitive after the reciprocal affix sim- is attached, ditransitive verbs become transitive instead of intransitive after undergoing the same process. Therefore, I conclude that the reciprocal affix sim- in Kavalan is a valence-changing affix, i.e., it reduces the number of arguments, but it does not necessarily derive intransitive verbs.