The relationship between citations and the linguistic traits of specific academic discourse communities identified by using social network analysis.

For a research article (RA) to be accepted, not only for publication, but also by its readers, it must display proficiency in the content, methodologies and discourse conventions of its specific discipline. While numerous studies have investigated the linguistic characteristics of different research disciplines, none have utilised Social Network Analysis techniques to identify communities prior to analysing their language use. This study aims to investigate the language use of three highly specific research communities in the fields of Psychology, Physics and Sports Medicine. We were interested in how these language features are related to the total number of citations, the eigencentrality within the community and the intra-network citations of the individual RAs. Applying Biber's Multidimensional Analysis approach, a total of 771 RA abstracts published between 2010 and 2019 were analysed. We evaluated correlations between one of three network characteristics (citations, eigencentrality and in-degree), the corpora's dimensions and 72 individual language features. The pattern of correlations suggest that features cited by other RAs within the discourse community network are in almost all cases different from those that are cited by RAs from outside the network. This finding highlights the challenges of writing for both a discipline-specific and a wider audience. [ABSTRACT FROM AUTHOR]