Metadiscourse by native and non-native English speakers: importance marking in lectures

This talk has a dual purpose. In addition to mapping the use of one type of metadiscourse, viz. importance markers, across ‘native’ and ‘English Medium Instruction’ (EMI) lecture corpora, we elaborate on analytical issues related to studying metadiscourse in spoken and disciplinary discourse.
‘Importance markers’ (Deroey & Taverniers, 2012) are lexicogrammatical metadiscursive devices combining discourse organization with evaluation along a ‘parameter of importance or relevance’ (Thompson and Hunston, 2000, p. 24). In lectures, they help students identify key content, which is useful for allocating processing resources while listening to what are typically dense monologues that require processing in real time. This in turns is likely to benefit understanding, note-taking and retention. Comparing the use of importance markers in a single-discipline corpus of engineering lectures by ‘native’ speakers and EMI lecturers, our aim was to contribute to the limited insights into the linguistic features of EMI lecture discourse generally and metadiscourse important for lecture discourse organization and hence lecture listening, specifically.
Both researchers independently identified potential importance markers manually in lectures 46 engineering lectures (364,542 words) delivered in the Italy, Malaysia, the UK, and New Zealand,. Agreed instances were tagged and the tagged corpus imported into Sketch Engine to facilitate further analysis.

Overall, native speakers and EMI lecturers differed little in importance marker frequency, range, types, and lexemes. In both corpora, the predominant verb marker was V n/clause (e.g. remember they don't know each other). The main difference was the far more common use of the listener-oriented 2 pers pron V n/clause marker (you must understand how to apply this one) by the non-native speakers but this was largely due to idiolectic variation.
Contrary to most corpus linguistic metadiscourse studies, we report the inevitable analytical difficulties when identifying and classifying metadiscourse. Issues include establishing a definition that is broad enough to capture the various realizations of a metadiscursive function, while not ‘opening the floodgates’ to include instances that are not representative or that render the study unfeasible. For us this included distinguishing between evaluation of discourse and ‘real world’ entities, excluding very frequent phrases that could be viewed as importance markers but in this discipline probably served another function, and establishing a continuum of highlighting ‘force’. These considerations necessitated careful manual analysis of a relatively small corpus, which however means that generalization are limited and idiolectic bias more likely.