多元回归统计建模在语料库语言学中近义词辨析上的应用

Multivariate Statistical Modeling in the Corpus- based Study of English Synonyms

通过对语料库词汇研究中采用的研究方法进行梳理,发现大部分研究采用单变量定量分析方法,即定量研究中所涉及的变量较为单一。基于此,笔者提出应该进一步深化基于语料库的定量研究方法,才能深化对词汇的研究和认识。因此,本研究通过运用多元统计分析方法对近义词IDEA和CONCEPT的案例分析,从多个角度考察近义词IDEA和CONCEPT的不同特点与用法,揭示出区别选择使用两者时需要考虑的关键因素,重点考察这两个词汇项在口语,小说,新闻和学术语域中的分布差异,以及两者在领属方面的语义特征和其后置修饰成分差异,旨在呈现多元回归统计模型在近义词辨析中的应用,指出其在词汇研究和认识的广度和深度方面明显优于单变量定量分析。

A review of research methods adopted in the corpus- based vocabulary shows that univariate quantitative analysis method is mainly adopted in the quantitative analysis. In light of this,it is put forward that the corpus- based quantitative research method should be further advanced in order to deepen the study and understanding of vocabulary. Therefore,with a pair of synonyms ' idea' and ' concept' as a case analysis,this study introduces the application of multivariate logistic regression analysis in the study of different characteristics and usage of English synonyms,arguing that different factors need to be considered in the choice of synonyms,including four different types of registers such as spoken language,novels,news and academic writings,as well as their possessive meaning and the post- modifier. This study presents the application of multivariate statistical regression model in analysis of synonyms,aiming to increase the breadth and depth of vocabulary research by introducing multivariate quantitative analysis and to provide some reference for the foreign language researchers.