泰雅語之事件概念化與動詞分類

Abstract

這篇論文的目的在於，透過在一個新架構之建立後，對泰雅語賽考利克方言的動詞進行分類，而這新的架構是根據在Fillmore （1975, 1976, 1977, 1982, 1985)的框架理論（Frame Theory)、Johnson（1987)的圖示理論（Schema Theory)以及Talmy（2000) 的「圖-底」（Figure-Ground)之二分理論上。這些理論都屬於認知語言學的知識。 如同多數菲律賓語類型的語言，泰雅語的主語可以透過四種語態結構來標示出，諸如，「主事語態」（AV)結構中要突顯的多是行動者角色的主語；在「受事語態」（PV)結構中，則是受事者主語；在「處所語態」（LV)結構中，則是要突顯處所主語；在「傳達語態」（CV)結構中，其主語則是工具或受益者論元；後三種結構通常一併被歸類在「承受者」（undergoer)語態結構之下。然而，透過自然言談語料的觀察，我們注意到這些語態結構的諸多複雜性，其複雜性可以從至少兩個事實中看出：其一，並不是任一動詞的所有三種「承受者標記形式」（UV form)，亦即-un, -an與s-等形式，都以相同的認知過程進入所謂的發展成熟之四分語態系統中；其二，不同的動詞有選用不同承受者來標記承受者主語的傾向。針對第一個面向，我們發現到，任一承受者形式位置之被填入，至少是要為了表達三種功能裡之任一或任二種使能進行，包括：（一)、能呈現動詞語意所投射出之「內在承受者」（intrinsic undergoer)與突顯這承受者之「承受形式之假定值」（default UV form)兩者間之固有關係的功能，（二) 、能反映事件事實性（reality)之差異或因應不同情境需求去表達所談論之承受者之細微差異等的功能，以及（三) 、為了標記應用語態（applicative voice)之主語承受者的功能。在這三種功能之中，第二種功能需要在任一動詞之第一與第三種較基本的功能都確定好之後，使能進行確認。承此，這份動詞分類研究終究是建立在第一與第三功能之互有關係的結果上，並且是一份以UV形式為根基的分類；此外，我們會發現，動詞的語意才是決定任一動詞的四種動詞形式之獲得的關鍵因素。 於是，透過檢驗三百餘動詞所反映之事件參與者的空間與概念的相互關係以及其所呈現的詞法行為（morphosyntactic behavior) ，我將論證，任一動詞的「內在承受者」（intrinsic undergoer)可以被賦予「圖」（Figure)或是「底」（Ground)的概念值，並且，這概念值確實會借助各自動詞的「承受形式之假定值」（default UV form)去顯示其在中性語境裡最為顯著之參與者的地位，這「承受形式之假定值」（default UV form)與「內在承受者」（intrinsic undergoer)所具備的概念值之搭配得到的結果，正是任一動詞所屬的類別；因此，在這樣的研究中，我們區辨出如下五個主要的動詞類別: 第一類：-un 形式動詞類別 (The -un verb class) 第二類：-an 形式動詞類別 (The -an verb class) 第三類：s-形式動詞類別 (The s- verb class) 第四類：s-/-an 形式混成動詞類別 (The s-/-an composite verb class) 第五類：s-/-un 形式混成動詞類別 (The s-/-un composite verb class) 更進一步地，因應著事件在本質上是可以進行抽象的圖示化，我們依據圖示，再在每一個主要類別之下區分出一至八不等之次類，因此，我們目前得到二十個圖示；任一的動詞即為所屬圖示的例子。 透過這冗長的研究，我們或能感受或是理解認知語言學學者（諸如，Fauconnier (1985, 1999) 、Lakoff (1982, 1987, 1990) 、Langacker (1976, 1987, 2002) 、Johnson (1987) 、Johnson-Laird (1983) 、 Talmy (1975, 1983, 1985, 1988, 2000) 等等)所堅持的觀點，亦即，語言是，語言使用者根據其具象的生活經驗後，將這些經驗經由認知機制的作轉化後、並包含著認知過程中可能做的解釋等種種結果之集合體，而並不是ㄧ個自我包含、置外於情境的封閉系統。

A central goal of the present study is to develop a novel framework for classifying verbs in Squliq Atayal into various types in terms of Charles Fillmore’s (1975, 1976, 1977, 1982, 1985) Frame Theory, Mark Johnson’s (1987) notion of schemas and Leonard Talmy’s (2000) Figure-Ground distinction-- concepts that are now familiar in cognitive linguistics. As in many Philippine-type languages, Squliq Atayal has four ways to encode the subject of a verb. That is, Actor Voice (AV) is usually used to encode an actor subject, Patient Voice (PV) a patient subject, Location Voice (LV) a location subject and Conveyance Voice (CV) an instrument or a beneficiary subject. The last three voice types are further grouped into Undergoer Voice (UV). However, a closer scrutiny of naturally occurring data shows that there is a complex reality in the way voice constructions are employed in two aspects. Firstly, not all three UV forms of a verb, i.e. -un, -an, and s-, are found to fill their respective slots in a so-called full-fledged four-way voice system via equal processing. Secondly, different verb types prefer different UV forms to code subjects. Filling in any UV form slot is contingent on meeting one or two of the three functions: (i) the encoding of an inherent relationship between a verb’s intrinsic undergoer and its default UV form, (ii) the encoding of an intrinsic undergoer for signaling the reality distinction or for exhibiting subtle semantic differences, and (iii) the encoding of an applicative undergoer. Of the three functions, (ii) is far more complex, since all possible conditions are not identified until a decision is made on (i) and (iii). That is, the nature of verb classification investigated in the present study ultimately hinges on interrelated results that derive from (i) and (iii), which is an UV-based classification. Besides, it is now commonplace that the availability of all the four voice forms for a given verb is lexically specific. It is argued that, based on an examination of the spatio-conceptual relationship of participants in the event a verb encodes and the morphosyntactic behavior of over 300 verbs, the intrinsic undergoer of a verb is assigned either the Figure or the Ground, and a verb has a default UV voice form typically used to make either the Figure or the Ground participant the subject of a clause and thus more prominent. The pairing of the default UV form and the default status of the intrinsic undergoer determines the class of a verb. Five major verb classes are then identified: (I) The -un verb class: Undergoer as the Figure (e.g., hkani’ ‘search for’, lamu’ ‘pick’, and naga’ ‘wait for’) (II) The -an verb class: Undergoer as the Ground (e.g., gyah ‘open’, wayaw ‘choose’, and ’luy ‘find’) (III) The s- verb class : Undergoer as the Figure (e.g., gihu’ ‘turn’, ruruw ‘push’, and tbaziy ‘sell’) (IV) The s-/-an composite verb class: Undergoer as the Figure specified by the s- form and Undergoer as the Ground specified by the -an form of a verb (e.g., biq ‘give’, paqut ‘ask’, and qapax ‘paste’) (V) The s-/-un composite verb class: Undergoer as the Figure specified by the s- form and Undergoer as the Figure specified by the -un form of a verb (e.g., kal ‘discuss; talk about’, and syuk ‘act in turn; retaliate; answer’) Furthermore, for each of the five major classes identified, further subtypes can be distinguished, depending on the nature of the schematization of events. Specific verbs can be thought of as instantiating specific event schema type. A total of 20 schemas have been identified. The bulk of the dissertation study is devoted to justifying the ways events are schematized and verbs are classified. To cognitive linguistics (e.g. Fauconnier (1985, 1999), Lakoff (1982, 1987, 1990), Langacker (1976, 1987, 2002), Johnson (1987), Johnson-Laird (1983), Talmy (1975, 1983, 1985, 1988, 2000), and among others), language is viewed as the result of general cognitive mechanisms and processes grounded in embodied experiences, rather than as a self-contained, context-independent system. We hope to demonstrate that the typing of verbs in Squliq Atayal can be shown to be largely grounded in embodied experiences that underpin the various schema types identified above.