A Contrastive Study Of Chinese-english-japanese NT Clause Complexes Based On A Self-built Trilingual Comparable Corpus

NT clause complex theory is a theory that describes the discourse structure of language from the microscopic level.It proposed to describe the clause complexes in Chinese and has been tested on large-scale corpora in Chinese and English.In this paper,Japanese is selected as the third language to describe under the NT clause complex theory.Based on the existing Naming-Telling clause(NT clause)complex theory and annotating practice,this study adopts Chinese,English and Japanese teaching materials and formulates a unified trilingual annotating specifications.By annotating nearly 100,000 Chinese,English,and Japanese words or characters,a trilingual NT clause complex comparable corpus is constructed.Secondly,based on this corpus,the characteristics of NT clause complexes in three languages are contrasted and analyzed from three different dimensions: naming-sharing structure(NSS)between clauses,the sentence-generating ability of NT clauses and the structures of NT clause complexes.The main findings include: NT clause complex theory can describe the three languages and basically cover the real trilingual text.Four patterns of NSS appear in all three languages;There are differences between Chinese,English and Japanese in the distribution of NSS patterns.Stack Pattern is the most widely distributed pattern in Chinese and Japanese,and New Branch Pattern is the most widely distributed pattern in English,while the ratio of Influx pattern in Japanese is higher than that in Chinese and English;In three languages,there are NT clauses grammatically irregular and semantically unreasonable,but differences in distribution.Non-compliant clauses in Chinese are obviously less than that in Japanese and English;The depth of clause complexes in Japanese,Chinese and English is limited,with an average depth of 2-3.However,the depth of clause complexes in Chinese is deeper than that in English and Japanese.In other words,there are more clauses included in NT clause complexes in Chinese;The co-occurrence patterns of NSS is limited in clause complexes.There is generally one NSS pattern in most clause complexes.The most frequent pattern in Chinese and Japanese is Stack Pattern,while in English is New Branch Pattern,which is consistent with the distribution of NSS in the three languages.The more the co-occurrence patterns of NSS in trilingual clause complexes,the less the number of clause complexes.The main co-occurrence patterns in trilingual clause complexes are Stack Pattern and New Branch Pattern.Following a preliminary analysis,it was discovered that the differences between the three languages at the level of clause complexes are linked to their grammars.The value of this research is as follows: It provides corpus support for language ontology research.In particular,the NT clause complex theory derived from Chinese was extended to Japanese,which verified the strong ability to describe different languages;Under the same theoretical framework,this paper explores the similarities and differences of three languages at the level of discourse structure.It expands the scope and perspective of language contrast research,and provides new features for language contrast research,such as NSS,the sentence formability of NT clause and structure of clause complexes;The trilingual comparable corpus which is constructed in this paper provides training samples for machine learning and supports for large-scale annotation of discourse corpus by machines in the future.