馬拉加斯語關係子句之詞組結構性

Relative Clauses are Phrase Structural in Malagasy

通行於馬達加斯加島的馬拉加斯語是一個具有典型語態系統的西部南島語。本文首先說明如何衍生各語態中從動詞發展而成的核心句，以及如何經由語意的組合來理解這些語句。我們認為這些語句是直接經由動詞詞綴投射出來，不需經過論元移位。論旨角色的指派可以清楚地在語意組合的過程中完成。其次，我們將此分析應用到關係子句，主張關係子句的衍生也一樣不需要經由論元移位。和大多數主流的分析相比，我們的分析有下列優勢：1. 經由清楚的語意組合來解釋關係子句的形成，能（部分地）說明說話者是如何理解新語句。2. 對馬拉加斯語中的“唯有主語可以移出的限制＂提出一個全新的衍生方法，能確實遵守Chomsky (2001:2) 所提的內包原則，也就是不使用語跡 (traces) 和指標 (indices)。3. 簡明且易學：衍生核心句只需要約二十個動詞詞綴，並在遵守鄰近原則的屬性查核下即可完成。論旨角色的指派可以很清楚地在詞根加綴所形成的詞類中達成。4. 像英語這種缺乏語態的語言裡，需要一個認知上的觸發詞 (trigger) 來解釋約束變項運符的移出。這也就是說，英語和馬拉加斯語的差別在於構詞特性的不同 (Chomsky 1996:7)。

Malagasy, spoken throughout Madagascar, is a Western Austronesian language with a characteristically rich voice system.1 In this paper we show first how to derive and compositionally interpret nuclear clauses built from verbs in different voices. They are directly projected from verbal affixes, not derived by A or A′ movement. Theta role assignment follows explicitly from semantic interpretation. And second, we derive and compositionally interpret relative clauses, again with no A or A′ movement. Merits of our analysis compared with most mainstream approaches are: 1. An explicit compositional semantic interpretation of relative clauses, thus (partially) accounting for how speakers interpret novel expressions. 2. A new derivation of the Malagasy “only subjects extract” constraint, which strongly satisfies Inclusiveness (Chomsky 2001:2) in that no traces or indices are used. 3. Simplicity and learnability: deriving and interpreting nuclear Ss is localized in about 20 verbal affixes plus feature checking (Agree) under adjacency. Theta role assignment is explicitly effected in the categories built by affixing roots. 4. A cognitive trigger for interpreting extraction by variable binding operators in voice-poor languages like English. English-Malagasy differences are “essentially morphological in character” Chomsky (1996:7).