論賽夏關係子句結構

Abstract

本篇論文主要探討賽夏關係子句結構及其之句法衍伸行為與句法上之特徵。一般而言，語言學者普遍認為關係子句具有跨語言之普遍性，並且存在語序上、用詞上、句法特徵上、策略上之差異。句法特徵實為本論文之重點。 自六零年代末，Smith(1969)以降，提出關係子句與冠詞(determiner)之間關係緊密，關係子句研究與論述如雨後春筍。然而七零年代，Chomsky (1977)提出關係子句結構與疑問句結構之間的相似性後，驗證關係子句為一種修飾性片語，以Adjunction 方式緊扣名詞組，並且形同疑問句之內在衍生方式，以算符(operator)做 A’-movement, 留下軌跡視同變數(variable)，並有算符，變數，與先行詞三者共指(co-index)，達成關係子句之派生，成為普遍接受之論。然而舊聲未絕，進入九零年代，Kayne (1994)之反對稱理論(Antisymmetry)，使得句法結構以純粹右傾，並且在深層結構中普遍帶有spec-head-comp的語序之外，更將關係子句與冠詞之間密不可分關係予以重新討論，令冠詞D° 選擇關係子句作為補語，進入兩方理論激烈爭論時代(c.f. Borsley 1992, 1997; Bianchi 1999, 2000; De Vries 2002)。 本論文經過發音人反覆驗證，目前認為，賽夏關係子句單純，沒有明確之關係子句標記，亦無疑問詞作示。前人所言功能詞 ’ima 為關係子句標記(Yeh 2000)並不全然正確，畢竟’ima 可出現於一般直述句而領導謂語；語序上，過去學者認為賽夏具有兩種語序，意即名詞先行詞前與後並存 (ibid.)，此論點使賽夏似於其他南島語與其特徵(c.f. Liu 2004; E. Aldridge 2004)，然而，因為賽夏關係子句無明顯標記 (其他複雜句亦如是)，實難區分該表達為先行詞後關係子句(postnominal)，或為相同主題所鏈的兩個完整句子。故此，在缺乏證據下, 本文暫不排除也不考量先行詞後關係子句之可能性。然而先行詞前關係子句，必須由「語序」作為關鍵證據。策略上，賽夏語具有限定與非限定之結構，但非限定結構使用之環境極為受限，本文將提出兩者之區辨，以及使用之分布。 句法上，過去學者普遍採取Chomsky (1977)之論證，將關係子句視為 Adjunction 結構 (Liu 2004 謂之傳統分析法)；後隨補語分析法 (Complementation approach) 盛起，近來句法學家大致採以此法，即使仍有不少學者並未放棄前言，甚至提出折衷之道 (c.f. Aoun and Li 2003)。根據語料所得，本文將採用補語分析法，並以提出相關語料佐證，證實以補語分析法，最能解釋賽夏關係子句之句法現象，同時，藉由Bianchi (1999)同時針對左緣結構之擴充(Rizzi 1997) 以及Kayne (1994)理論之補述，反映出其理論對深層語法解釋能力。 最後，本論文也深入探討採用補語分析法分析賽夏關係子句後所造成的困難，包括量詞詞組本生位置，先行詞絕對的高範疇以及由來，以及以探針-目標關係(Probe-Goal Dependency Chomsky 2001)解釋外冠詞與先行詞之間如何達成構詞上的呼應(Agreement).This thesis discusses the relative constructions in Saysiyat and the derivations. Relative clauses are generally believed to be complex constructions shared by human languages, although some cross-linguistic variations are also expected. Chomsky (1977) argues that relative clause constructions are similar to interrogatives in many respects, and they should be considered as adjuncts attached to the noun phrase. The first step is for the wh-pronoun raises to the Spec of CP, functioning as an operator to leave a gap in the originate position. The trace of the operator is regarded as a variable bound by the operator. Then the head noun external to the relative clause, the operator and the variable are co-indexed. This head-external view is regarded as the adjunction approach, or the standard approach (Alexiadou et al. 2000; de Vries 2002). However, the strong dependency of relative clause to the determiner of noun phrase was captured before Chomsky’s claim, and this view is revived in Kayne’s notion of Antisymmetry (1994), which claims that the external D head selects a relative clause CP, with the head noun, originated internal to the clause, moving to the governing domain of the external D. This view is further elaborated (c.f. Bianchi 1999), which is generally called the complementation approach. In this thesis, I firstly state that the so-called relativizers ’ima and kama in Saysiyat are not real relativizers. They are auxiliaries. Later, I raise the doubt on the existence of postnominal relative clauses in Saysiyat because no solid evidence could be offered to support its existence but only island effects. In the following part, I argue that the complementation approach could be the better account for relativization in Saysiyat with empirical supports. Moreover, nonrestrictives and putative internally headed relative clauses are also discussed. I conclude that nonrestrictives exist in Saysiyat, but it can only be prenominal. As to the putative internally headed relatives, they are derived via stranding of PP and Genitive DP. Syntactically speaking, this type of relatives is not typical internally headed relatives. Last, more problematic issues are discussed as the consequence of adopting the complementation approach, such as the base position of QP, the unambiguous wide and high scope reading of the head noun, and the Probe-Goal dependency between the external determiner and the head noun under Agree.