A Corpus-based Analysis Of Collocation Errors In Writing By Chinese English Learners

As a new research approach in applied linguistics, corpus approach has been employed more and more widely in language researches with increasing application of advanced computer and the emergence of enormous text corpora and well-designed concordance software. Applying this method and based on the theories of CA, EA and IL, the researcher has conducted a quantitative study of Chinese college EFL learners' interlanguage collocation errors. The study uses data taken from Chinese Learner English Corpus (CLEC). The concordance software WordSmith is used in the study.For close observation of the collocational competence of EFL learners, collocation errors are collected from about 2452 essays from the two sub-corpora of college sophomores (Band IV students) and college juniors (Band VI students) in the Chinese Learner English Corpus (tagged). With the help of the concordancing software, the author classifies, counts and describes the collocation errors in terms of the part-of-speech. From the observation, analysis and diagnosis of these errors concordance lines, six causes of errors are inferred: Chinese interference, overgeneralization, ignorance of collocational restriction, incomplete comprehension to word meaning, misselection of words in the target language and the inducement by teachers and dictionaries. The thesis also provides a new pedagogy–corpus-based teaching. Based on this method, some possible solutions to enhancing the college learners' collocational competence. It is suggested that teachers of Chinese college English learners had better point out the collocates while introducing new words and meeting old ones, they can concordance the collocates of new words or old ones with the computer and concordance software. The learners, on the other hand, should be motivated to notice and collect collocations by themselves to raise their awareness of how English words are chunked, for which a collocation notebook is strongly recommended, so that a native-like language production can be finally more or less realized.