Metadiscourse in academic writing : a corpus-based study of expert, L1 and L2 postgraduate student text

PhD ThesisThis study looks at how differences of language and genre affect the pattern of the usage of metadiscourse (MD: the rhetorical resources used to organise a discourse, or the writer's point of view towards either its content or the reader) in the academic writings produced by expert and student writers in the same discipline. The corpora consist of thirty journal articles (PRO) and fifty-five student assignments, twenty-five from native English speakers (NES) and another thirty from non-native English speakers (NNES). All texts in both the PRO and student corpora are in English, produced by a single-author, in topics of study in language, culture, and communications in the same school of Newcastle University. The research uses a typology derived from those in the literature but focussing on written texts and modified by a pilot study. From the comparison of texts produced by the NES and NNES postgraduate students, the language variable (native English vs non-native English) plays a greater role in the use of MD. The NNES writers use statistically more textual metadiscourse (TMD) while the NES writers employ more interpersonal metadiscourse (IMD) in general and there are statistical differences in the use of sequencers, code glosses which were more used by the NNES, hedges and self references which were more employed by the NES in particular. The finding indicates that the NNES writers are more concerned about expressions to show the logical order and relations between different parts of the text through TMD; the NES writers try more to involve the readers in the argument than do their NNES counterparts with IMD. The findings also show that learning a writing style which is acceptable in western academic life (e. g. 'writer-responsible') influences the use of MID in the NNES academic writing. Evidence of this comes from the interview data and the results of text analyses which show the statistically greater use of textual metadiscourse (e. g. sequencers, code glosses) and the significantly infrequent use of self-references in the NNES texts. From the results in the genre/expertnessv ariable Oournal articles vs student assignments), no overall significant differences were found in the use of the main V categories (TMD and IMD), but differential purposes (effects) and frequency were found in the use of MD subcategoriesT. he student writers do not use MD devicesi n the same way as the PRO writers, as reflected in the use of MD devices with a limited range of items and purposes in the student texts compared to a broad range of MD features and functions in the PRO texts. In fact, the PRO writers made more use of concessives, concluders, sources, hedges and self references with a broad range of purposes;t he student writers made significantly more use of sequencerst,o picalisers and more use of emphatics with limited purposes. Thus the finding proposes that the way they use MD is influenced by the two factors in the student and PRO texts; the consideration about the readership and the goal of the argument; which lead the different pattern of MID usage in the student and expert writings. This suggests that the genre variable (student assignments vs journal articles) is also a crucial one to influence the use of MID within the same discipline. As regards the language aspect from the comparison between the NES and NNES, the differences are mainly in the amount of features in the use of MD. When it comes to the student and j ournal article texts, genre variable, the differences are not only in the frequency of MD subcategories but also in the way they use the MD features