Reconsideration of text typology concerning text genre in terms of applied linguistics

本项目主要进行了两项研究。第一项研究利用语料库中的信息，探讨了可能与人们判断书面文本难度有关的语言因素。英国国家语料库中的一个描述性特征，感知难度水平（Wrilev），被用于此目的。BNC中所有的书面文本被收集者根据他们对文本难度的判断分为Wrilev的低、中、高三个子类别。从三个方面分析了这一特征：(1)Wrilev与其他两个描述性特征的关系分布；(2)平均单词和句子长度；(3)Hayashi的量化方法类型III。出现了以下几点：(a)文本难度的判断显示出某种系统性，可以看出这表明可靠性；(b)单词长度与判断无关；(c)句子长度与信息性文本的判断有关；(d)主题特异性与判断没有系统关系。这些结果表明，句法复杂性和人的背景知识对文本难度的感知有一定的影响。其次，在本项目中，以BNC中的体裁类别为研究对象，对多元分析所揭示的维度进行语言解释。相反，语域变化将通过处理与BNC中的“阶级”相当系统地相关的社会语言学变量来明确识别。因此，本文的另一个目的是澄清体裁和语域的区别。因此，首先，我将定义文本类型、体裁和语域等术语。实际上，这些术语都以或多或少相似的方式用于识别同质文本组，因此在明确设计论文之前，有必要了解这些重叠的项目。

There were two main studies carried out in this project. The first study investigated linguistic factors that may be related to people's judgement of written text difficulty by utilizing information from a corpus. One of the descriptive features in the British National Corpus, perceived level of difficulty (Wrilev), was used for this purpose. All written texts in the BNC are categorized into Low, Medium, and High subcategories of Wrilev by people who collected texts for the BNC according to their judgements about text difficulty. This feature was analysed in three respects: (l) distribution of Wrilev in relation to two other descriptive features, (2) average length of words and sentences, (3) Hayashi's quantification method type III. The following points appeared:(a) Judgements of text difficulty revealed some kind of systematicity, which could be seen to suggest reliability, (b) Word length was not related to judgements, (c) Sentence length was related to judgements concerning informative texts, (d) Topic specificity did not have a systematic relationship with judgements. These results suggested that syntactic complexity and people's background knowledge have relationships with perception of text difficulty.Secondly, in this project, focusing on genre categories in the BNC, the dimensions revealed by the multivariate analysis will be interpreted linguistically. In contrast, register variation will be explicitly identified by dealing with a sociolinguistic variable which is fairly systematically associated with 'class' in the BNC. Therefore, another aim of the paper is to clarify the difference between genre and register. For that reason, to begin with, I will establish the definition of terms such as text types, genre and register. Actually these terms are all used, in more or less similar ways, to identify homogeneous groups of texts, so an understanding of these overlapping items is necessary before the design of the paper can be made explicit.