Malaysian learners’ argumentative writing in English: A contrastive, corpus-driven study

Research on learner English is by now an established sub-discipline in corpus linguistics, yet few studies exist on Malaysian learners. This thesis explores the difficulties that Malaysian learners of English face when producing argumentative essays, focussing on their overuse of particular linguistic features. WordSmith Tools (Scott, 2012) is used to analyse and compare two corpora: The Malaysian Corpus of Students’ Argumentative Writing (MCSAW): Version 2, consisting of 1,460 Malaysian students’ argumentative essays; and the Louvain Corpus of Native English Essays (LOCNESS), which is a corpus of native English essays written by British and American students and is used as a reference language variety here. The software enables analysis of keywords (words that are over-used in MCSCAW), collocates or surrounding words of the keywords, and concordances, which are used to examine the keywords in context. Crucially, it also allows examination of the ‘range’ of linguistic features (i.e. by how many students a feature is employed) – an under-used but crucial affordance of this software programme that is exploited in this thesis for down-sampling purposes. The thesis combines quantitative and qualitative corpus linguistic techniques, with keywords providing the starting point for in-depth qualitative analysis using concordancing. This corpus-driven analysis of MCSAW identifies typical features of the writing style of Malaysian learners’ writing of English, particularly the overuse of can and we (including the highly frequent bundle we can), and the lack of discourse-organising markers. Analysis of key words and key bundles is complemented with collocation analysis and concordancing of the highly frequent modal verb can as well as the highly frequent first person plural pronoun we, which both have a high range across the corpus. The concordances are carefully and systematically examined to explore the ways in which these over-used linguistic items are actually employed in their co-text by the Malaysian writers. While results show some similarities in both learner corpus and reference language variety, Malaysian learners tend to demonstrate higher writer visibility overall. One possible explanation lies in the influence of the national language (Malay). The thesis also identifies repeated sentences that occur in more than one essay, which implies either plagiarism on the learners’ part or a particular teaching strategy (templates or phrases that are provided to students). This finding has significant implications for corpus design (in terms of the need for more topic variation) as well as methodological significance (in terms of the advantages and disadvantages of using the ‘range’ feature for down-sampling), which are also discussed in this thesis. In sum, this thesis makes a new contribution to corpus linguistic research on learner English and will have implications for the development of teaching practices for Malaysian learners of English