Association Relationship Analyses of Stylistic Syntactic Structures

Abstract

Exploring linguistic features and characteristics helps better understand natural language. Recently, there have been many studies on the internal relationships of linguistic features, such as collocation of morphemes, words, or phrases. Although they have drawn many useful conclusions, some summarized linguistic rules lack physical verification of large-scale data. Due to the development of machine learning theories, we are now able to use computer technologies to process massive corpus automatically. In this paper, we reveal a new methodology to conduct linguistic research, in which machine learning algorithms help extract the syntactic structures and mine their intrinsic relationships. Not only the association of parts of speech (POS), but also the positive and negative correlations of syntactic structures, linear and nonlinear correlation are considered, which have not been well studied before. Combined with the linguistic theory, detailed analyses show that the association between parts of speech and syntactic structures mined by machine learning method has an excellent stylistic explanatory effect.