RESEARCH ARTICLE (wileyonlinelibrary.com) doi: 10.1002/leap.1244 Received: 26 January 2019 | Accepted: 9 May 2019 Linguistic differences between well-established and predatory journals: a keyword analysis of two journals in political science Josep Soler , * and Ying Wang Josep Soler Ying Wang Department of English, Stockholm University, Universitetsvägen 10E, 10691, Stockholm, Sweden ORCID: J. Soler: 0000-0002-2813-0101 *Corresponding author: Josep Soler E-mail: josep.soler@english.su.se Abstract Predatory publishing has become a much-discussed and highly visible phe- nomenon over the past few years. One widespread, but hardly tested, assumption is the idea that articles published in predatory journals deviate substantially from those published in traditional journals. In this paper, we address this assumption by utilizing corpus linguistic tools. We compare the ‘academic-like’ nature of articles from two different journals in political science, one top-ranking and one alleged predatory. Our findings indicate that there is significant linguistic variation between the two corpora along the dimensions that we test. The articles display notable differences in the types and usage of keywords in the two journals. We conclude that arti- cles published in so-called predatory journals do not conform to linguistic norms used in higher-quality journals. These findings may demonstrate a lack of quality control in predatory journals but may also indicate a lack of awareness and use of such linguistic norms by their authors. We also sug- gest that there is a need for the education of authors in science writing as this may enable them to publish in higher-ranked and quality-assured outlets. INTRODUCTION Discussions about predatory publishing have revolved in their majority around meta-data analyses of this phenomenon, includ- ing, for example, its scope and breadth (cf. Shen & Björk, 2015) and the geographical origin and career level of contributors to such journals (e.g. Xia et al., 2015). However, until now, very little has been discussed in connection with the actual content of arti- cles published in predatory journals. Moreover, it has been indi- cated that knowing more about the actual material that is published in predatory journals is crucial if we wish to gain more clarity about this phenomenon (Eriksson & Helgesson, 2017). It is precisely this gap that we wish to address in this article, and we do that by utilizing corpus-analytical techniques. Over the past few decades, corpus linguistics has developed into a particularly prolific research approach for providing empiri- cal investigations of patterns of language variation and use (Biber, 2010). Some basic corpus analysis tools, such as wordlists and keywords, have been fruitfully exploited to study words and/or word-based patterns, which help to distinguish different texts and styles (Gilmore & Millar 2018; Scott & Tribble, 2006). In this article, we set out to provide a corpus-based investigation of articles published in an alleged predatory journal in the discipline of political science, with a particular focus on the ‘academic-like’ quality of their language in contrast to the language that appears in articles published at a top-ranking journal in the same field. With our analysis, we hope to contribute with results that can add more solid evidence on the ‘academic-like’ nature of what is Learned Publishing 2019 www.learned-publishing.org © 2019 The Author(s). Learned Publishing © 2019 ALPSP. 1