导　　师： Hu, Zhuanglin

学科专业： 050201

授予学位： 博士

作　　者： ;

机构地区： 北京大学

摘　　要： 本文的基本出发点是:系统语言学古典模式中的“主位”是一个形式概念,需要引入一个相应的意义范畴“主题”。基于这一假设,作者全面概述了主题/(尤其是语篇主题/)的研究成果,采用系统描写的方法,探讨了语篇中各个级阶和各个等级层次的主题现象,从而协调并统一了既有理论模式。主要成果体现在以下论述中:/(一/)作者从小句的主题入手,论证了小句和复句的主题结构和主题系统,/(二/)探讨了它们在语篇中的动态组织特征,即“重复性”和“非重复性”网络关系,/(三/)并从功能的角度,描写了“中心主题”和“边沿主题”系统,考察了它们与“中心主题段”与“非中心主题段”、隐喻化过程、以及语用编码意图之间的关系,为计算机做“语篇内容综述”提供了依据;/(四/)作者还运用流程图和“状态-迁跃”方法,对不同语言级阶、不同等级层次的主题系统,尝试性地逐一做了识别和生成描述;/(五/)最后提出了进一步改造系统语言学古典模式的方案;/(六/)在这一过程中,论证所选语料,均系名篇名著。结论是,语篇主题系统是一个语义信息网络,具有多重等级层次关系。 主题概念在公元前的古希腊时代就已经出现,但它是和现今意义上的主语混在一起使用的;这种混用现象,在19世纪以前一直占主导地位,并延续到20世纪上半叶。从公元前1世纪开始,语法研究又出现了另一个传统,即以语序为基本原则的、有关语言交际价值的研究;到18世纪中叶,语言使用的这一现象得到了较为系统的阐述;20世纪早期布拉格学派对此进行了初步的理论化探讨,从而步入了主题研究的理性时代。随着不同语言观和语言模式的形成,学界出现了形式主义和功能主义两大研究类别,尤以功能主义主题研究的成果显著。其中有的侧重于小句主题,有的侧重于语篇主题,有的则兼顾,产生了多种模式。 The issue of topic systems in text arises from a theoretical hypothesis thatHalliday's category of Theme be largely a formal, syntactic concept, while thedue semantic counterpart be Topic, aboutness of a piece of messageextending the area of the clause, clause complex, paragraph, section, chapter,book, or a text as a whole. The topics thus formed interweave into sets ofconstellation, which further display hierarchical statuses. /(i/) At the clause rank, topics are assigned with the function for demandingcomments. The relation is comparatively an unmarked but structural one, butnot without variation since in some cases comments would precede topics.This dynamic nature will become more salient when simultaneouslyconfigured with prominent interpersonal features, e.g. the so-called EnhancedThematic Structures. Meanwhile, a background topic has an immediateresponsibility for what to come in the remainder of the clause. /(ii/) For the clause complex, there may occur more than one topic. Thesetopics are attributed with statuses of different importance, such as Primary,Secondary, Tertiary, and so on. But the nature of status assignment maymerely bear local significance since some are very important throughout thesurrounding contexts. Such relations are still within the domain of structurality,but the dynamic, organisitional feature is obvious because their structuralcharacteristic is only confined to the general syntacticality. In fact, somesecondary, tertiary or other downgraded topics may even appear before theirprimaries. /(iii/) An entire dynamicality of topic relationship is associated with those areasbeyond the clause complex, where the repetitions of various topics in textconstitute constellations of networks;and the meronomic, collective, and hyponymic specifications of the non-repetitive topics, i.e. parts, individuals,and hyponyms, diffuse respectively around their corresponding general,leading entities, i.e. the whole, collection, and superordinate topics. Moreover,there may also be Possessive and Additional chains from every now and then.As for contrastive parallels, “fronting” becomes an unmarked phenomenon,through which the orgnisational feature is most remarkably manifested./(iv/) These topics of the various relations may be so complicated thathierarchies emerge for their interpretations. That is, some chains are morecentral while others are just peripheral. Of course, some central topics aremerely locally defined, and beyond would proceed more significant ones forthe holistic central concern of a text. This is the Downward type of the localcentral. But some are immediate to their surrounding contexts, which is calledthe Upward type. However, the demarcation of the local and the holistic is onlyextremity. In fact, chains of intermediate length are common. Of course thismodel also goes to the local and the holistic marginal topics./(v/) Central and marginal topic chains may signal straightforward identificationof central and marginal stretches of text, the latter two of which may in turn betaken into account for figuring out the main concern/(s/) of a text as a wholefrom the perspective of text summarisation. Actually, the hierarchical systemsof text stretches as well as the hierarchical topic systems may be regarded asclues for the gist of a text. Hence, such an endeavour would result inadvances for computational generalisation or summarisation./(vi/) Meanwhile, the observation of the encoder's intention is noticed as aproblem of the discrepancy between the titles, subheads, and central topics,and what the writer really intends to convey. In fact, one intention may directlybe expressed over another. But linguistic clues may help display theencoder's true intention/(s/). Furthermore, pragmatic strategies and principlesare always alive in the realisation process. They work as the basis on whichmetaphorisation and topicalisation take place. And lastly, the complicatedsequential phenomena of topic development require a great deal of cognitiveprocessing effort. /(vii/) Along the analyses, schematic attempts have been made forprogramming topics of various kinds for recognition and generation on thecomputer. Incidentally, there have already been some works that could beemployed as the databases, such as the WordNet by Princeton University.And the Computational Linguistics Studies Centre in Peking University isexploring an English-Chinese version of the WordNet system. What is needednow is technical work to fill the gap for their ultimate realisation./(viii/) Furthermore, topicalisation realises the situational parameter Mode, thespeaker's manipulative ways of arrangements of experiential meaning interms of his cognitive schema and communicative intention in aware of theaddressee's knowledge;and the phonological category for realising Topic andTheme is Tonicity.It is then clear that topic systems in text are informational in essence. Eachtype of topic system has been figured out as it is for message transaction.Meanwhile, its relation with Thematisation – the corresponding formalcounterpart – has also been pointed out in available places. Main points, suchas exhaustive explications of topic systems with other textual categories, anda new model of language, have been laid out for further studies.To sum up, this thesis started with verifying the formal nature of Theme in theclassical SFL, and pointed out Topic as the semantic counterpart. On thisbasis, it surveyed the studies of topic so far, and worked out an overallframework for what has been done and what not in the literature in question,which unified ‘sentence topic' and ‘discourse topic' as an integrated whole.This process has witnessed a systematic analysis of these issues and madean attempt for the programming. Finally, an implication was made for a newmap of language. Therefore, the attempt has not only theoretical significance,but is also contributive for Text Linguistics or Discourse Analysis.

关 键 词： 主位 主题 系统 网络 等级层次

分 类 号： [H0]

领　　域： [语言文字]