賽德克太魯閣語句法探究

Abstract

本篇論文主旨在了解Truku Seediq的構詞語法結構，當中包含格位標記，時態時貌系統以及衍生構詞語法。文章內所提供的語料是在2008年到2011年間，於花蓮縣秀林鄉的田野調查中取得。 Truku Seediq的ka與Paran Seediq的ka功能不同，前者維持其主格格位標記，但後者已經成為具有語法功能的格位標記，僅出現在特定的語法環境中。除了身為主格格位標記之外，Truku Seediq的ka還也以不同的結構出現：cleft construction-1, cleft construction-2, concession construction, positive attitude construction, ka as a linker, 以及ka in construction with time expressions，每一個結構都有它自身特別的語法使用特性。 第三章討論在Seediq語中，有三個呈現時態時貌系統的方式：動詞構詞、使用表時態時貌的助動詞還有使用語尾助詞。動詞構詞指的是代表焦點系統的四個詞綴(m-/-m-, -un, -an, s-)。Holmer (1996) 發現到四個詞綴可以代表現在式，但是時態時貌系統是隨著動詞的不同而有不同的詮釋方式。不同的AV形式有不同的詮釋：AV1和AV4皆表示實現貌，然而AV2和AV3則表示非實現貌，一般說來，AV2有近未來的語意，而AV3則是遠未來的語意。有些動詞僅有一個表示未來的形式。 關於NAV的實現貌詮釋比較複雜，移動性的動詞並沒有PV1的形式，LV1和LV2形式表示實現貌，兩者的使用是根據情境而有所不同，PV2形式表示非實現貌，而CV的詮釋則根據動詞而有不一樣的詮釋方法。 實現貌/非實現貌、可視性、時間長短，這三個參數是判斷該如何使用該動詞的NAV形式的依據，同一個動詞若帶有不同的焦點系統詞綴，則語意則會有所不同，以mangal ‘帶’ 為例，它的PV1語意為 ‘以前把某件物品帶走’，而LV1的語意則是 ‘幾天前把東西帶來(這裡)’，LV2的意思則為‘現在正在把東西帶來(這裡)’，有時後PV1和LV1的意思則皆可以是 ‘把東西帶走’ 或 ‘把東西帶來’。qmita ‘看’ 的LV1和LV2可以相互使用，這樣的方式跟其它動詞不同。qmita 的PV1表示 ‘某人幾天前看到某件東西’，LV1/LV2表示 ‘某人剛剛看到某件東西’。不過這三個參數並不是唯一分辨NAV的依據，例如mangal ‘帶’ 的NAV還可以討論到 ‘所有權’，PV1表示 ‘某人已經得到那個拿的東西’，而LV1表示 ‘某人已經歸還了當初所拿的東西’，而當使用mkan ‘吃’的PV1/LV1時，是因為食物還沒有吃完，所以說話者還是看得到食物的情況，最後 smbarig ‘賣’只有一個NAV的形式：PV2，它可以使用在實現貌也可以使用在非實現貌的情形下。 第四章探討Truku Seediq的衍生構詞，並討論幾個詞綴：dmp-s(e)-,dmp-t(e)-, emp-s(e)-, emp-t(e)-, g(e)-, gn(e)-, gm(e)-, gmn(e)-, mg-, mneg(e)-, tg(e)-。dmp-s(e)-,dmp-t(e)- 是實現貌的詞綴，而emp-s(e)-, emp-t(e)- 是非實現貌的詞綴，當使用dmp-t-或emp-t-時，動作的主事者是專門人員，如果使用dmp-s-或emp-s-，其語意為業餘者因為某個原因而實行動作。本章的語料引用自現階段還在編纂的字典，一個詞彙就擁有許多的詞綴，上述的詞綴僅僅是冰山一角，但卻足以了解該語言構詞特性是相當豐富。因此發現在這語言中有哪些構詞形態。並且了解是不是與其他台灣南島語言有相似的特性，將會是未來可繼續研究的議題。 第五章為結論。提出一些可望在未來繼續發展的觀察，雖然試圖全盤了解Truku Seediq的時態時貌系統似乎不可能，但是它卻是一個值得挑戰的題目。

The present study investigate three topics in the morphosyntax of Truku Seediq: case markers, tense/aspect/modality system and derivation morphology. Much of the corpus data used in this study was drawn from fieldwork done in Xiulin Township in Hualien County between 2008 and 2011. ka is found to be a robust nominative marker in Truku Seediq, while it has become a pragmatic marker in Paran Seediq, occurring only in specific grammatical environments. In addition to being a nominative case marker, ka also occurs in a number of constructions: cleft construction-1, cleft construction-2, concession construction, positive attitude construction, ka as a linker, and ka in construction with time expressions. Each construction is shown to have its specific pragmatic function. Chapter 3 discusses three strategies that the tense/aspect/modality system in Seediq uses to encode events related to time. One is through verbal derivation; one is with the aid of auxiliaries; the third one is with the aid of final particles. Verbal derivation refers to voice markings on the verb (i.e. m-/-m-, -un, -an, s-). While Holmer (1996) observes that the present tense forms are m-/-m- (AV), -un (PV), -an (LV), and s- (CV), the TAM interpretation is found to vary from verb to verb. Different AV forms have different reality interpretations. Thus both the AV1 form and the AV4 form refer to the realis events, while the AV2 form and the AV3 form are irrealis. Generally, the AV2 form has an immediate future tense interpretation, and the AV3 form has a distant future tense interpretation. Some verbs have only one AV form that refers to an irrealis event. Reality interpretations for NAV forms are shown to be even more complex. Motion verbs in general do not have PV1 form. Their LV1 form and LV2 form also denote realis events, although my consultants often differ in the way the LV1 and LV2 form of the verbs are interpreted. The PV2 form is usually irrealis, and the CV form is either realis or has a neutral interpretation, depending on the verb. Each of the verbs discussed is shown to have a unique interpretation with respect to reality or visibility or temporal distance. Sometimes different voice markings on the same verb also entail differences in lexical meaning. This is true of the verb mangal. The PV1 form of the verb means ‘took something away in the past’, while the LV1 form means ‘brought something here several days ago’, and the LV2 form means ‘brought something here just now’. The meaning of mangal with the PV1/LV1 marking can be either ‘take something away’ or ‘bring something’. The LV1 form and the LV2 form of the verb qmita are interchangeable, which is different from other verbs which have four NAV forms. The PV1 form of qmita means that someone saw something several days ago, and the LV1/LV2 form means that someone saw something a short while ago. However, they are not the only relevant variables that we can invoke to distinguish the TAM interpretations of these and perhaps all the other verbs in the language. The notion of ‘entitlement’ also comes into play in the semantics of mangal, for example, in addition to the three variables just mentioned. The PV1 form of mangal means that the agent is entitled to the patient object, while the LV1 form means that the patient object is returned to the owner now. The PV1/LV1 form of mkan has a partitive interpretation in that the form means that the food is not eaten up, and so the food is visible to the speaker at speech act time. Finally, the verb smbarig is also unique in that there is only one NAV form for smbarig, the PV2 form, and that form has a neutral interpretation. Chapter 4 investigates derivational morphology of the language, especially the morphosyntax of the following prefixes: dmp-s(e)-,dmp-t(e)-, emp-s(e)-, emp-t(e)-, g(e)-, gn(e)-, gm(e)-, gmn(e)-, mg-, mneg(e)-, tg(e)-. The prefixes dmp-s- and dmp-t- denote realis events, and the prefixes emp-s- and emp-t- denote irrealis events. When the derived form dmp-t- or emp-t- is used, the agent of the action is a professional; while if the derived form dmp-s- or emp-s- is used, the agent is an amateur; moreover, there is a reason why the amateur does the named action. The derivational forms identified above are only part of the derivational forms found in a single entry aba taken from an ongoing dictionary project, but they are enough to suggest the rich repertoire of affixes inherent in the derivational morphology of the language. It will be interesting to explore what these data suggest for patterns of lexicalization in the language, and furthermore, how these lexicalization patterns figure in relation to similar lexicalization patterns in other Formosan languages. Chapter 5 concludes with some observations about areas where future research is sorely needed. Sorting out reality and visibility interpretations of the NAV forms of verbs in Truku Seediq represents a most daunting, yet at the same time, interesting challenge for future research.