A Corpus-based Contrastive Approach For Classification Of Chinese-Japanese Homographs Sharing Common Meaning

Chinese and Japanese homographs are an important topic in contrastive studies and teaching research.Despite many achievements,the previous studies in China,mainly with an aim to serve Japanese language teaching,compare the meanings of words in Chinese and Japanese by using the rational meanings in dictionaries and most of them merely focus on the case analysis of individual words.This paper,based on the results of studies in Chinese and Japanese linguistic ontology,contrastive analysis and language teaching,integrates the meanings and part of speech in existing dictionaries with the grammatical functions and collocations derived from the corpora to build a multi-layered and comprehensive framework for comparison and classification,in an attempt to systematically compare Chinese-Japanese homographs sharing common meaning(CJHSCM)and to help solve vocabulary issues in Japanese and Chinese language teaching.This study selects the disyllabic CJHSCM from the A-level words in the Syllabus of Graded Words and Characters for Chinese Proficiency as the object of investigation.Modern Chinese Dictionary and Super Big Words are referred to for the static attributes of the Chinese and Japanese words,such as the entry setting,description in meaning and corresponding relations,and part of speech.The CCL corpus retrieval system(online version)and the Modern Japanese Written Equilibrium Corpus are referred to for the grammatical functions and collocations of Chinese and Japanese words.It is expected to establish a comprehensive contrastive description framework for the classification of CJHSCM.The construction method is first to compare the corresponding relationships of entries in Chinese and Japanese dictionaries and determine four categories:one-to-one,one-to-many,many-to-one and many-to-many.In terms of one-to-one category,words can be divided into eight sub-categories according to their meanings(both rational and coloring),part of speech,syntactic functions and collocations.Words in the subcategories can then be divided into 15 further sub-categories according to the overlap,inclusion and direction of meanings.This study,based on the frequency from the native speaker corpus and the error index from the HSK Dynamic Composition Corpus 2.0,aims to provide a detailed description of typical words belonging to different categories in the framework according to the above-mentioned categorization method.The main contribution of this paper is to establish a layer-by-layer contrastive analysis procedure for classification description framework of "corresponding relationship ?meaning? part of speech ? syntactic function? collocation relationship ? word meaning relationship".The framework can classify CJHSCM layer by layer,compare and analyze the difference between CJHSCMs from different dimensions,and evaluate their difference.Validation studies using the framework show that there are differences among categories and the differences are significant,which indicates that the framework is effective and meaningful.Specifically,among the four categories,the error rate of the one-to-many category is the highest while the error rate of the many-to-many category is the lowest.Among the eight sub-categories,those with great and small differences are more prone to be erroneous.Among the 15 further sub-categories,the error rate of homographs in overlapping relationships is the highest while the error rate of homographs in inclusion relationships(Chinese c Japanese)is the lowest.Finally,taking the compilation of Japanese-Chinese teaching reference books as an example,this paper puts forward specific suggestions for compiling CJHSCM dictionaries based on this framework.