Many studies have investigated the linguistic features and variation of research articles (RAs) to guide the teaching and learning of English for Research Publication Purposes (ERPP). These studies predominantly adopted a discipline-based framework to foreground the influence of (sub-)disciplinary conventions on RAs. They collect writing samples, group them by (sub-)disciplines, and compare differences among the groups (e.g., Hyland, 2008). While these studies provided valuable insights into understanding research writing, they may fall short in revealing linguistic variation within and transcending disciplinary boundaries. As increasing new evidence has pointed to linguistic variation that obscures (sub-)disciplinary conventions (e.g., Triki, 2021), this study proposes an alternative approach without foregrounding the influence of (sub-)disciplines. Guided by the APPRAISAL framework, this study examined 240 RA discussions from eight research areas in four overarching disciplines: Chemistry, Geoscience, Education, and Management. This study used Cluster Analysis to explore linguistic variation without pre-grouping the samples. The analysis classified the discussions into five clusters characterized by distinct linguistic profiles demonstrating statistical significance. The profiles do not strongly associate with specific (sub-)disciplines and are likely to be motivated by various factors. Specifically, Cluster 1 is characterized by the infrequent use of features to convey attitude and stance as the RAs devote more space to describing relatively objective processes, such as experimental results, research design and implementation, and real-world observations. Cluster 2 is characterized by the frequent use of linguistic devices that emphasize the study's positive practical value, which may be motivated by the researchers' rhetorical choice to promote their research by drawing readers' attention to the highlights. Cluster 4 employs more attribution resources, indicating higher citation density. This could be due to the RAs' highly cumulative knowledge base requiring them to elaborate on what has already been known in the field before making sense of the findings. The opposite is true for Cluster 3, possibly due to the relatively less cohesive knowledge base and more interpretive knowledge-making of RAs in this cluster. Cluster 5 is characterized by the frequent use of linguistic devices to hedge claims and fend off alternative interpretations, indicating that the research findings do not provide unequivocal evidence to make factual claims. The identified linguistic features and variation of RAs can offer practical guidance to novice researchers on how to write publishable RAs. The findings also provided further evidence to support the call for a re-examination of the discipline-based framework in writing research (Kaufhold & McGrath, 2019). Given the internal fragmentation of (sub-)disciplines and the increasing trend of interdisciplinary research, future ERPP research may need to break free of the discipline-based framework.
Bibliography
Hyland, K. (2008). Disciplinary voices: Interactions in research writing. English Text Construction, 1(1), 5-22.
Kaufhold, K., & McGrath, L. (2019). Revisiting the role of 'discipline' in writing for publication in two social sciences. Journal of English for Academic Purposes, 40, 115-128.
Triki, N. (2021). Exemplification in research articles: Structural, semantic and metadiscursive properties across disciplines. Journal of English for Academic Purposes, 54, 101039.
