台語回應訊號初探

The Conversational Use of Reactive Tokens in Taiwanese

本研究旨在瞭解台語回應信號（Reactive Tokens）在會話中的使用情形。本文的語料採自收音機的扣應節目，內容長約兩小時，總共轉寫了4517個語調單位，含括52位說話者，其中48位為男生。初步的分類大致參考Clancy等人1996年的研究架構。然而在分析過程中，我們發現一些定義上的問題，：（一）回應信號不一定表示說話者都會讓出發言權，有29％的說話者並未將發言權交還給主要說話者。（二）回應信號並非都是肯定的，在某些情境下也可以是否定詞。（三）在決定某個回應信號是否為真為回應信號時，其出現的位置似乎比本身的語意更具決定性。（四）若嚴格遵守Clancy人所提的定義，則形式和功能之間會有明顯的衝突，顯然我們需要另一套分類的方法。本研究結果如下：在台語扣應電話交談中，回應信號出現的頻率非常顯著；大約每3.83個語調單位就有一個回應信號出現。可見在沒有視覺管道時，台語使用者在交談互動中，其實遠比Clancy人所觀察到的國語使用者要活躍的多。另外回應信號的多少也與溝通的媒介、言談者之間的熟悉度、場合的正式性、話題、溝通的目的和技巧、以及個人風格有關。一般而言，台語使用者傾向將回應信號（93％）放在理想中的話輪轉換處（TRPs）之後。如果說話者在回應信號之後繼續說話時，我們發現這經常是因為上一個說話者的話輪已經到達了一個高層次的語用結束點（possible global pragmatic completion point）。

This study investigates the conversational use of reactive tokens（RTs） in Taiwanese. The data consist of a two-hour recording of radio call-in programs. There are altogether 4517 utterances transcribed, with 52 speakers involved, 48 of them being male. The framework is based on Clancy et al.'s study（1996）.In the coruse of classification, however, some of their definitions of RTs are shown to be problematic. First, Taiwanese RTs are not necessarily "continuers" because 29% of them are followed by extended talk by the same speaker. Second, RTs are not necessarily affirmative in nature. In some contexts, discoursal negatives（e.g. ‘boe7' & ‘bo5'） also qualify as RTs. Third, the location of an RT candidate seems to outweigh its semantic meaning. Fourth, following their definitions strictly would lead to a rather serious conflict between form and function. Further clarification is hence required. The results are as follows. The presence of RTs in Taiwanese telephone conversation is indeed massive. An RT occurs every 3.83 IUs. When deprived of visual aids, Taiwanese speakers are in fact far more verbally active in giving interactional support Next, the RT use is observed to be affected by channel of communication. degree of familiarity or formality, the topic, the interactants' communicative intentions and skills of turn management, and personal styles. The tendency to place RTs at transition relevance places is clear. 93% of them appear after pragmatic completion points. Moreover, the distribution is somehow related to such interactional devices as utterance-final particles and tag/confirmation questions. Last, the vast majority of the 334 non-floor-supporting RTs occur affter possible global pragmatic completion points.