Collaborative networks and international research publication practices: Metrics, impact and implications for researchers' education and training

In this paper we adopt an interdisciplinary approach, combining complex network theory (Álvarez et al. 2021, Zhou et al. 2018) with writing and composition studies (Andersen et al. 2014, Bazerman and Prior 2004), to discuss: i) the creation of collaborative research networks that leading researchers establish through international co-authorship practices, and ii) the relevance and impact of these networks within the researchers’ disciplinary communities. Building on such interdisciplinary approach, we used complex network analysis to construct the co-authorship networks of four leading EU researchers and identify the impact of their collaborative behaviour when writing in English for research publication purposes. We extended this analysis with quantitative (corpus-based) and analysis of the titles of their journal articles using data mining techniques and corpus tools, as the previous literature has also done (Anthony 2001, Chen and Liu 2023, Hyland and Zou 2022; Kerans et al. 2020). The network analysis generated four distinct cases of collaboration clusters, from which we evaluated the extent to which the structural properties of the networks (degree centrality, betweenness and modularity) determine the researcher’s relevance and impact. Network #1 represents the case of a researcher whose collaboration cluster reveals many stable collaborations, predominantly with national groups, which have a significantly high impact. Network #2 illustrates the case of a researcher with a high number of publications shared with three other researchers. Other collaborations show a high degree of stability. Network #3 is an example of a researcher who collaborates with multiple communities, typically involving more than ten collaborators, many of whom are from the researcher’s own country. This network further illustrates the relevance of a collaborator with whom the researcher has published extensively, with a co-authoring publication rate of almost 50%. Finally, network #4 reveals the case of a collaborative network with around 11-12 clusters of around 5-10 collaborators who do not necessarily belong to the same institution or research group. These communities generally do not collaborate with each other in a consistent manner. Instead, the network identifies several researchers with high centrality, indicating the potential for collaboration among them and through them to connect with other research groups. Further linguistic analysis showed that the recurring grammatical structures in the titles of researchers’ articles were primarily complex noun phrases (NPs), in which information is elaborated upon. The result is a lexically dense, heavily nominalised discourse style with a high explicitness level. This style is characteristic of the information-dense register found in academic prose (Biber and Gray 2016). In light of these findings, we advocate a research-informed approach to raise awareness among researchers of the importance of collaborative authoring in today’s increasingly competitive research landscape. Finally, we propose some pedagogical tasks to help scientists write effective journal article titles. We conclude that network analysis is a valuable tool for identifying patterns of researchers’ collaborations by establishing connections across institutions and disciplines, forming collaborative clusters of researchers working on similar topics and identifying central researchers who facilitate collaboration between different research communities.