Abstract:

The Cambridge ESOL (English for Speakers of Other Languages) Exams consist of a series of language proficiency tests developed and administered by the Department of English Language Assessment at the University of Cambridge. It assesses four language skills based on A1-C2 proficiency bands. Main database of the study is the Cambridge Learner Corpus (CLC) compiled from the student essays of all English as a Foreign Language exams of the institution containing 42-million-words of more than 200,000 students from 173 countries. This study aims to reveal the errors that Turkish EFL learners who took PET, PETfS, KET, KETfS, FCE, FCEfS, CAE and CPE exams between 1993 and 2013 made by examining the sections that measure their writing skills in exam papers using corpus-based error analysis tools available on SketchEngine platform and classifying them according to the A2-C2 language proficiency levels of the Common European Framework of Reference for Languages (Council of Europe, 2001). When the distribution of errors is examined according to the A2-C2 language proficiency levels, while the variety of error taxonomies increases, the frequency of errors decreases. The similarities in the error patterns observed from A2 to C2 proficiency levels are significant. Although there are fluctuations in the frequency of errors, the persistency of error taxonomies is observable. In addition, the study also compares the errors of L1 Turkish students with the errors made by learners from other mother tongue backgrounds available in the corpus. This comparison revealed twelve out of top 14 error taxonomies detected in the subcorpus of Turkish L1 learners overlaps with the ones observed in the subcorpora of other learners available in CLC.

Keywords: