Discourse, genealogy and methods of text selection in international relations

Abstract

Discourse analysis, once the purview of critical theories of international politics, has emerged as a mainstream methodology for understanding international relations. While interest in such perspectives has enriched international relations theory, much about the nature of methods—that is, specific empirical processes for the gathering and analysis of evidence—is left ambiguous in this scholarship. Which texts should discourse analyses focus on? And, more practically, how should those texts be chosen? Building on discussions of case study methodology from both qualitative and interpretive social science, this article contributes to theoretical and empirical projects within discourse theory by suggesting a method for text selection: the random selection of texts. I argue that random selection processes are beneficial for discourse analyses that aim to study broad cultural patterns, such as genealogy. Random selection is not simply a means of choosing texts, but also a more comprehensive logic for thinking about the purpose of texts in discourse analysis.

Acknowledgements

The author would like to thank Torbjørn Knutsen, Michelle Jurkovich, Hendrick Spruyt and Scott Weiner, as well as the panel members and audience at the International Studies Association’s annual meeting (2017) for insightful, and illuminating, comments and discussions. The author would like to extend special thanks to Will Daniel for discussions about the comparative politics literature on case selection, and to the editorial staff of Cambridge Review of International Affairs and the enlightening comments of three anonymous reviewers. All errors are my own.

Notes

1 I refer to ‘discourse analysis’ as an orientation to the thick and interpretive analysis of texts and structures of meaning. The term ‘analysis of discourse’, which is sometimes used in the article, refers to the study of discourse more generally (including content analysis, big-data textual analysis and so on). I thank an anonymous reviewer for this point.

2 I bracket discourse approaches from other traditions (including Wittgensteinian and Habermasian approaches). See Holzscheiter (2013) for a review of these traditions.

3 An exception is Wodak and Meyer (2009), though even they fall short in asking questions about text selection. Other approaches in IR have been undertaken to create dialogue between multiple perspectives (for example, poststructuralism, constructivism and CDA), though the aim of this work is conversation across epistemological boundaries. See, inter alia, Carta and Morin (2014).

4 There are further exceptions to this, including: Wodak and Krzyzanowski (2008) and Angermuller et al (2014, 317–356) briefly discuss sampling and text selection. However, Titscher et al (2000, 242) notes that techniques meant to bring textual samples closer to population characteristics are ‘controversial’.

5 This is not to say that research like that of Rheindorf and Wodak (2017) is inadequate. Rather, it demonstrates that meta-theoretic concerns with the status of discourse work has caused scholars to ignore questions about text selection and method by sidelining the question or dealing with method in an ad hoc manner.

6 Another approach is ‘contextualist’ intellectual history, which analyses the emergence of ideas in the history of political thought through reference to the broader milieu in which the ideas emerge (for an overview: Bevir 2011). I largely bracket these approaches in this article.

7 While references are made throughout to scholarship within the broader realm of ‘discourse studies’, this article limits its sights to genealogy as one type of discourse method. Other methodologies, including approaches related to ‘conceptual history’, are not critiqued here in detail. On these approaches, see Strath and Wodak (2009), Foxlee (2010); Krzyzanowski (2010) and Krzyzanowski and Wodak (2011).

8 As is the case with all forms of textual analysis, using a texts-as-cases and representative selection method based on randomness may reveal some intertextualities and obscure others. However, such a method is complementary to one of the main goals of genealogical analyses: making visible previously hidden structures of knowledge.

9 By ‘context’, I mean the background (of action, of history, of culture) that texts as artefacts exist within.

10 There is some resistance to the use of ‘context’ as a theoretical concept in the study of discourse. For a critique, see Van Dijk (2009). For a sustained discussion of discourse’s relation to context in a variety of genres, see Flowerdew (2014).

11 I use these terms interchangeably here. However, ‘archive’ and ‘population’ hold different meanings based on different theoretical traditions. By ‘archive’ I simply mean a collection of texts that share in common a trait being studied. This sidelines other meanings—particularly those within poststructural discourse theory that raise important questions about who constructs archives and how. See Foucault (1972) and Derrida (1996).

12 Even in reflections on case study methods that do not advocate randomness, these reflections rarely argue that selection should proceed without reference to a larger population or stratified categories of a population. See, for example, work on case studies as indicators of ‘causal mechanisms’, inter alia: George and Bennett (2004); Goertz (2017).

13 Other attempts to combine forms of content analysis and discourse analysis, similar to Krebs’s approach, have been attempted in critical linguistic scholarship. Most particularly, this research has aimed to combine corpus linguistics with CDA. See, inter alia, Baker et al (2008) and KhorsaviNik (2010).

Additional information

Notes on contributors

Dillon Stone Tatum

Dillon Stone Tatum is an assistant professor of political science at Francis Marion University, South Carolina, USA. His research concentrates on the history and development of international liberalism, critical security studies, and contemporary (international) political theory. Email: [email protected]