Construction Of Written English Corpus Of Mongolian-speaking Learners

Corpus linguistics is a method that makes use of language data to describe language or verify some hypotheses about language.Its research foundation is corpora.With the development of computer science,corpus construction and corpus-driven researches revived rapidly,which provides enormous and authentic data basis for language research and referential support for language teaching and learning.However,the corpus construction for Mongolian-speaking learners whose mother tongue is Mongolian is few.Therefore,the thesis regards this research gap as my research viewpoint,aiming to construct a database used to provide studying resource for Mongolian-speaking English learners.There are two research questions mainly discussed in the thesis:(1)How to construct a written English corpus of Mongolian-speaking learners?(2)What are the error features of written English of Mongolian-speaking learners?Firstly,the thesis,based on corpus linguistics,takes writings of Mongolian-speaking English majors in Inner Mongolian Normal University as research content through data collecting and data processing and constructs a Written English Corpus of Mongolian-speakings Learners(WECML)containing 200 thousand words.In addition,WECML is annotated part of speech and errors.Next,the thesis applies corpus-driven method,and uses retrieval software AntConc 3.4.3 and statistical software Chi-Square Calculator to make some analyses from three aspects-a basic analysis on errors;a contrastive analysis on errors;a diachronic analysis on errors.The results have shown that(1)verbal phrase and noun phrase are the most errors among the nine error types.Among the 43 sub-errors,“np6(number)”,“vp3(agreement)” and “wd2(part of speech)” are the top three errors and “vp10(serial verb constructions)” should not be neglected;(2)the orders of other seven error types in CLEC(st5 and st6)and WECML are totally the same except prepositional phrase and collocation in a reversed order.The error ration of Mongolian-speaking English learners(MEL)is higher than Chinese-speaking English learners(CEL)and MEL overmake 23 out of 38sub-errors than CEL;(3)three features of a diachronic analysis on WECML have shown that 1)MEL have decreased their errors in verbal phrase,noun phrase,prepositional phrase and collocation with their English level increasing from the second year to the fourth year,but increased in the fifth year;2)they have kept and decreased their errors in adjective phrase,adverbial,pronoun and conjunction with their English level increasing from the second year to the fifth year;3)only word is irregular.These analyses have shown some functions and value of WECML.the study not only theoretically provides certain reference for written English research and teaching of Mongolian-speaking English learners but also practically shows the feasibility of constructing a written English corpus of Mongolian-speaking learners and provides a data resource for studying written English features of Mongolian-speaking learners.