A Corpus-based Investigation Into The Association Patterns Between Synonyms And Negative Sentences

One of the noticeable advantages of corpus-based linguistic studies lies in the investigation into various association patterns (the systematic ways in which linguistic features are used in association with other linguistic and non-linguistic features) of language use. To identify and adequately describe the patterns in language use, we need to process large databases of natural texts. In this respect, the corpus-based approach to linguistics has advantages over the traditional approach (which relies more on intuition and a limited range of material) in that the former uses modern techniques to process data with much higher speed, accuracy and objectivity. Therefore, corpus-based investigations into the complex association patterns in language use have become increasingly extensive, among which a focus of interest has been the association patterns of synonyms. However, the issue concerning the association patterns between synonyms and negative sentences is almost untouched. For this reason, I attempt to make a preliminary investigation along such a line.The theoretical background of the present thesis is a combination of the Register Theory and the Systemic-Functional Grammar proposed by M. A. K. Halliday. The Register Theory maintains that language structure demonstrates different patterns of variation in different situations; Halliday regards the structural features of language as the result of different systematic choices of linguistic elements to perform functions. By the end of the 1980s, Douglas Biber, a leading scholar in corpus linguistics, first proposes an important assumption: co-occurrence of linguistic features reflect shared function. Guided by this, he conducts a fruitful, corpus-based exploration of register variation in English, focusing on their different shared functions in relation to situational factors. Biber finally makes a clear and cogent remark on the general correspondence between registers and functions, thus making them more operational. This is really an assimilation of the essence of the Register Theory and the Hallidayan Grammar, and a significant step forward of functionalism. Therefore, the present thesis tends to apply Biber's views in the interpretation of quantitative findings.The research as reported here is based upon data taken from the British National Corpus (BNC). The reasons are mainly two: first, the BNC is highly representative, because it contains about 100 million words and that the material is from a wide range of written and spoken English, both British and American; second, a particularly noteworthy feature of the BNC is its spoken sub-corpus, which contains about 10 million words and is composed primarily of conversation from a wide range of situations (for its unique role in communication, conversation is a must for comprehensive, empirical linguistic research, and the BNC has plenty to offer in this respect). As for the techniques of my quantitative analysis, all programs used in the thesis have already been an inherent part of the concordancing software designed by English engineers for the BNC CD-ROM, which is easy- to- use and powerful.To make the argumentation more neat and orderly, and considering that synonyms should have the same part of speech, the thesis is divided into three chapters:A. Chapter 1 discusses the association patterns between synonymous adjectives and negative sentences. We compare the collocational frequencies and strengths of association between negative words and synonymous adjectives without negative affixes, and how synonymous adjectives with negative affixes are used in attributive and predicative positions.B. Chapter 2 discusses the association patterns between synonymous verbs and negative sentences. We compare the collocational frequencies and strengths of association between negative words and two groups of synonymous verbs, and we concentrate on the "polar" phenomenon in the second section.C. Chapter 3 discusses the association patterns between synonymous nouns and negative sentences. We concentrate on the preliminary investigation into the association patterns between synonyms and negative sentences on a discourse level, by briefly discussing the possibility that subject matter of a text may have certain correlation with such association patterns.Summing up what is found, the thesis reaches the following conclusion: words with the same basic meaning are nevertheless different in terms of their association patterns with negative sentences, and some of these differences are even significant; itis discovered that a series of factors may exert influence upon these association patterns, and they boil down to register-related factors.