Conjunction in discourse: a three-dimensional analysis of conjunction in different registers of English

从句之间的连通性传统上被视为事件、现象等之间的“现实世界”关系（时间顺序、因果关系等）；它主要是在连词的基础上进行探索的。‘when’， ‘if’)和连词(例如：“因此”)表示这些关系。然而，现在人们普遍认为，除了这些“外部”类型的连词关系外，我们还发现了“内部”类型的连词关系。虽然很少注意到，但内部连词有两种不同的类型。连接可以在文本的世界中建立(例如：‘first’， ‘in sum’)。说话者也可以根据它们的言语功能来连接子句，或者指出提出要求所依据的认知基础。例如，在“if you're going out, it's raining”这句话中，if从句并没有表达下雨的逻辑条件；相反，它给出了说话人提到雨的理由。此外，我们需要考虑没有明确表示的关系。大多数研究人员（除了一些值得注意的例外，如马丁）主要或完全集中在三种类型中的一种；甚至马丁也没有以原则性的方式区分文本连接和言语-行为/认知连接。我提出的三维模型分离了这些类型，并借鉴了韩礼德的三元函数概念，建立了一个理论框架，可以更好地解释这些类型。因此，“外在的”连接在本质上可以被看作是经验的，将数字（命题）结合在一个复杂的序列中；人际连接（言语-行为/认知类型）在展开的互动中连接动作；篇章连词连接了展开论证的各个步骤。此外，韩礼德对投射（大致相当于传统的“间接引语”）和扩展（所有其他类型的连词）的逻辑语义关系进行了区分，并假设这三个维度适用于两者。我的研究探索了这个模型的含义，特别关注于建立更精确的语法标准，可以用来区分类别，确定投射可以以与扩展相同的三维方式看待的说法的有效性，以及检查是否应该在人际关系中建立比已经确定的更多的子类别。目的是提供一个全面的，连贯的，理论上有动机的英语连词说明。我还进一步研究了韩礼德的观点，即不同的英语语域以不同的方式使用连词资源，无论是在类型上还是在连接明确发出信号的程度上。据我所知，还没有大规模的研究开始从经验上验证这一说法，并精确地调查发生的各种变化。我的方法包括对每个子句所进入的关系和相关信号进行分析，部分是手动的，部分是自动的，在一些子语料库中选择它们，理由是它们说明了最大的变化潜力。最初的假设（部分得到初步研究的支持）是，经验连词虽然是最常见的类型，但在叙事中尤其占主导地位，文本连词最常出现在说明文的书面语域中，而人际连词往往在非正式演讲和说服性语域中被大量使用。项目的这一部分旨在让我不仅可以设置选择的注册配置文件，还可以测试和改进模型。整体而言，该研究旨在促进语言学的根本转变，摆脱以经验语法和意义为核心，以人际意义为外围“附加”的传统，并适当强调人际关系，将其作为语言的符码系统同等重要。

Connectivity between clauses was traditionally seen in terms of 'real-world' relations between events, phenomena, etc. (time sequence, cause-effect, etc.); and it was explored mainly on the basis of the conjunctions (e.g. 'when', 'if') and conjuncts (e.g. 'therefore') that signal those relations. However, it is now accepted that, as well as these 'external' types of conjunctive relations, we also find 'internal' types. Although it is seldom noted, there are two different kinds of internal conjunction. Connections may be set up within the world of the text (e.g. 'first', 'in sum'). Speakers may also connect clauses in terms of their speech functions or indicate the epistemic basis on which claims are made. For example, in the utterance 'if you're going out, it's raining', the 'if' clause does not express a logical condition under which it rains; rather, it gives the speaker's reason for mentioning the rain. In addition, we need to account for relations which are not explicitly signalled.Most researchers (with a few notable exceptions such as Martin) have concentrated mainly or exclusively on one of the three types; and even Martin does not distinguish in a principled way between textual and speech-act/epistemic connectivity. The 3-dimensional model that I have proposed separates these types and draws on Halliday's notion of the three metafunctions to set up a theoretical framework in which the types can be better explained. Thus 'external' conjunction can be seen as experiential in nature, combining figures (propositions) in a complex sequence; interpersonal conjunction (the speech-act/epistemic type) connects moves in an unfolding interaction; and textual conjunction connects steps in an unfolding argument. In addition, Halliday's distinction between the logico-semantic relations of projection (broadly corresponding to traditional 'reported speech') and expansion (all other kinds of conjunction) is introduced, with the hypothesis that the three dimensions apply to both.My research explores the implications of this model, focusing especially on establishing more precisely the grammatical criteria which can be used to differentiate the categories, on determining the validity of the claim that projection can be viewed in the same three-dimensional way as expansion, and on checking whether further sub-categories than those already identified should be set up within interpersonal conjunction. The aim is to provide a comprehensive, coherent and theoretically motivated account of conjunction in English.I also follow up on Halliday's insight that different registers of English deploy the resources of conjunction in different ways, both in type and in the extent to which connections are explicitly signalled. I know of no large-scale study which has set out to verify this claim empirically and to investigate precisely the kinds of variation that occur. My approach involves carrying out an analysis, in part manual and in part automated, of the relations into which every clause enters, and of the associated signals, in a number of sub-corpora selected on the grounds that they illustrate maximal potential for variation. The initial hypothesis (partly supported by a pilot study) is that experiential conjunction, while the most frequent type overall, is particularly dominant in narrative, textual conjunction occurs most often in expository written registers, and interpersonal conjunction tends to be drawn on most heavily in informal speech and persuasive registers. This part of the project is designed to allow me not only to set up register profiles of choices in conjunction, but also to test and refine the model.The study as a whole aims to contribute to the fundamental move in linguistics away from the tradition of treating experiential grammar and meanings as core, with interpersonal meanings as peripheral 'add-on', and towards giving due emphasis to interpersonal concerns as equally central to language as a semiotic system.