Twenty-five Years of Securitization Theory: A Corpus-based Review

Twenty-five years after its initial formulation, securitization theory is at a crossroads: attempts to critically scrutinize its achievements and shortcomings proliferate, concerns about the theory’s eurocentrism are articulated, and a heated row shakes the field following accusations of racism. In this unstable context, the present article systematically reviews a corpus of 171 securitization papers published in 15 major International Relations journals since 1995, identifying two major imbalances characterizing securitization theory research. First, rich theoretical development has not been matched by sustained efforts to strengthen empirical work; second, the theory has not been globally embraced, displaying instead a narrow, distinctly local anchoring. By shedding light on these two issues and their relationships, this review article aims to provide clear and actionable observations around which scholars could productively re-organize the ongoing debates and controversies.

: these more or less civil engagements on a range of dimensions reflect a moment of questioning, if not outright doubt, on the theory which sought to explain the process through which certain issues come to be perceived and treated as security threats. While some like Balzacq (2019: 331) identify 'tensions holding back the development of securitization theory' but simultaneously take stock of its feats, others like Potenz (2019: 322) hold the less favourable view that 'unfortunately, as a field and a concept, securitization has suffered from substantial tensions, shortcomings, and missed opportunities'. Some scholars push to blend securitization within framing theory, perceived to be both broader, more coherent and empirically supported (e.g. Watson, 2012). The question critics and supporters alike have in mind is the one recently asked by Ned Lebow and Potenz (2019: 417): 'What is the future of securitization studies?' The present review article does not aim to settle all these debates by entering the various theoretical minefields on which they take place. Rather, it takes a more limited, formal, zoomed-out approach that looks at major characteristics of existing ST research; we offer a corpus-based review of securitization research over the past 25 years, whereby each article developing or using ST within this timeframe was identified and coded along two major axes which, we hypothesized, underpin some of the ongoing rows and disagreements. On the one hand, we coded for each article several theoretical and methodological characteristics, to locate potentially important features of ST research when it comes to its balance between theory and empirics, which has been identified several times as a source of problems (e.g. Baele and Sterck, 2015;Balzacq, 2011). On the other hand, statistics were gathered on the articles' geographical features (i.e. where their authors/journals are based), again to identify potentially significant patterns and imbalances related to the theory's positionality, which is another source of criticism (e.g. Bertrand, 2018). While such a method is limited, we would nonetheless argue that precisely because it is confined to formal characteristics of ST articles it has the advantage of providing a clear picture of the theory's geography and theory-method nexus that stands beyond -but provides insights on -the abovementioned debates. In other words, we aim to provide actionable observations that can productively re-organize current debates around a firmer diagnosis.
Overall, our review reveals two major imbalances. First, while ST's popularity has grown over time, its development is marked by an imbalance between efforts to develop the theory and to strengthen empirical work. Second, the theory has not been globally embraced, displaying instead a distinctly local anchoring. While these two issues are interconnected -reflecting the deeper rifts between the American and 'peripheral' traditions in IR -we nonetheless suggest that boiling them down to the qualitative/quantitative rift is too convenient.
We proceed in three parts, plus a concluding discussion. In the first section, we present the data on which the review is based, a unique and rigorously selected corpus of 171 academic papers on securitization published over the past 25 years. In the second and third sections, we use our data to discuss ST's two main imbalances under scrutiny. On the one hand, ST scholarship has experienced intense theoretical discussions but has not shown a similar enthusiasm when it comes to designing rigorous empirical enquiry ('Theory above Methods' section). On the other hand, ST research has overwhelmingly been developed in the United Kingdom and published in a narrow range of European journals ('Europe above the Rest' section). A conclusion attempts to constructively engage with the future of securitization studies by raising what we believe is the main overarching question triggered by our observations. The stakes of the questions raised in this review article reach well beyond the sole case of ST. The observations made here connect with, and further feed, the wider ongoing effort to reflect upon the presuppositions, practices, structure and sometimes power hierarchies of IR as a scientific discipline (e.g. Baele and Bettiza, 2021;Cohen, 2008Cohen, , 2010Hamati-Ataya, 2011, 3 with the underlying idea that greater reflexivity can be a path towards more dialogue in the discipline and subsequently stronger research. 4 Our formal count-based approach and findings indeed directly echo, for example, Kristensen's (2012Kristensen's ( , 2018 exposition of the fragmentation of the IR field into a series of clustered communities, Lohaus and Wemheuer-Vogelaar's (2021) demonstration of that each IR journal has its own pool of geographically located authors, or the earlier work of Waever himself (e.g. 1996Waever himself (e.g. , 1998 who highlighted the gap between the various geographical traditions feeding IR research. It also unearths divisive questions such as the role of -or more fundamentally need for -(grand or mid-level) theorizing in IR, 5 in the tail of the 'end of theory' debate. 6

Data: Mining 25 Years of Securitization Research
Spanning the 1995-2020 period, we collected all articles published in 15 leading IR journals that mention the term 'securitization' (or 'securitisation') in their abstract and/or keywords and/or title. 7 The starting date of 1995 corresponds to what is usually considered to be the founding ST contribution: Waever's chapter 'Securitization and Desecuritization' in Lipschutz' volume On Security. 8 The choice of the 15 journals was made on several grounds. First, a core list was constituted with the 8 journals studied by Waever (1998) in his seminal comparison of American versus European IR, thus allowing a continuation of the discussion on identical grounds. Second, the list was expanded to make sure it included all explicitly IR-facing journals from Political Sciences associations or organizations that have an IR focus or branch (PRIO, ISA, BISA, and PSA), as well as all the IR outlets ranked by the Scimago Journal Rank in the top 20 of their 'Political Science and International Relations' category. Finally, International Relations was added because of its longevity and its recent publication of a forum on ST (International Relations, 2015).Overall, 171 articles were retrieved, and each had its content coded across several dimensions approximating the two main tensions evoked above: its theoretical/empirical focus, the issue(s) and region(s) it studies, its methodology, and the home institutions of its authors. 9 We are not the first ones to make such counts; indeed Van Rythoven and Hayes' inaugural contribution to a blog forum dedicated to ST's lack of success in the United States (see endnote 2) already revealed important issues, proving the pertinence of this type of approach. The present effort, however, is much wider and systematic.
To be sure, this selection -summarized in Table 1 -comes with two main limitations. First, it has blind spots, missing important ST contributions made either in other journals or in books. 10 We nonetheless believe that it covers the diversity of the IR field and adequately represents the major developments of ST over the past 25 years. Second, the focus on major IR journals can possibly induce a bias in favour of theory, since strong theoretical contributions are usually required for publication in these outlets; this is not necessarily the case for issues-focussed or area studies journals, which usually put a premium on empirics. While this bias may account for some of the findings presented below, it does not explain the scale of the imbalances present in the data.

Theory above Methods
Our first axis of investigation follows the line of argument claiming that a rich and fastpaced theoretical evolution of ST may have been done at the expense of a similarly strong development when it comes to methods. On the one hand, theoretical developments in ST have certainly been rapid and diverse. The 'new framework for analysis' offered by the Copenhagen School in the mid-1990s has been persistently and significantly reinforced, expanded and broadened away from its focus on securitizing speech acts to encompass a range of different processes and factors surrounding or directly at play in securitization dynamics (Baele and Thomson, 2022); this broadening shed light on important social and political issues, enabling a fuller capture of a phenomenon as multifaceted as securitization. However, it also effectively fragmented ST into different theories of securitization 12 'committed to distinctive ontologies and epistemologies' as well as methodologies (Balzacq, 2015: 103), a situation that prompted efforts to reconcile differing views into unified frameworks (e.g. Bourbeau, 2014). On the other hand, comparably little attention has been devoted to methods and best methodological practices when it comes to empirical work. ST's methods have several times been criticized as limited, if not outright weak. Calls have been voiced to move beyond theoretical debates and conduct methodologically rich empirical work (e.g. Balzacq, 2011), and the lack of methodological transparency in applied research has been pinpointed (e.g. Sterck [2015: 1122] argued that methods are 'the Achilles' heel of securitisation studies, casting doubt on their conclusions'). In their recent evaluation of ST, Balzacq et al. (2016: 519) still wondered if 'there [is] a "better" method for studying securitization processes?' Overall, this imbalanced development is confirmed by the data in several ways. First, ST scholarship is indeed heavily geared towards theoretical development rather than empirical knowledge: more than half of the 171 papers (89) were 'primarily theoretical', compared with only 10 'primarily empirical' articles -the remaining being 'balanced theoretically and empirically'. As Figure 2 shows, only a minority of papers do 'focus on a key issue', that is, have a clearly identifiable and sustained engagement with one (or two) empirical case. 13 Put differently, the majority of ST scholarship remains abstract, with only thin empirical engagement. As noted above, this imbalance in favour of abstract theorizing as opposed to case(s)-based investigations can partially be explained by the nature of the journals we included in the sample; a quick glance at some area studies publications brings back articles using ST in a more instrumental way to explain particular issues or events. Generalist journals, however, do not necessarily neglect empirical work when favouring theorization -they usually require both. We thus believe that what has more fundamentally driven this imbalance is the value traditionally given to 'high' theory, and suspicion towards 'methods' at times conflated with positivism, within the critical IR community. 14 Theorization is not, of course, inherently problematic; what raises questions is a theoretical framework which is usually not used as such -as a tool used to better understand specific issues and solve empirical puzzles that incidentally allow for further conceptual work.
This said, Figure 2 also allows us to notice that the minority of empirical ST papers do explore a wide range of different issues, with the securitization of health, migration, and terrorism attracting most of the attention; this diversity evidences the rich empirical potential of ST as an applied theoretical framework. In addition, Figure 3 further displays this usefulness, showing that empirical ST is used to examine issues taking place across the globe -although with very little attention paid to Latin America. Vuori's (2008) arguments that ST can be productively applied in tailored ways to non-Western settings, 15 which followed Wilkinson's (2007) identification of this issue, seems to find credence, while claims that it only works in -or for -the West appear to miss this important dimension.
Second, and as the funnel chart in Figure 4 below helps to visualize, methods are indeed ST's Achilles' heel; the preference for theorization has been accompanied by a disregard for methodological considerations. The data clearly shows a striking lack of methodological rigour in the scholarship: out of the 82 papers with empirical content, only 48 have a clearly identifiable method followed for the analysis (either a quantitative one like an experiment or a survey, or a qualitative one like a discourse analysis or a process-tracing); out of these 48, only 13 'comprehensively' explain and justify the parameters of the chosen method (e.g. sampling decisions, tools used for analysis, system used for interpreting the corpus, etc.) -a mere 7% of all ST scholarship included here. Only three papers employing a quantitative method were found in the corpus, 16 and all belong to this small category; without exception, they also belong to the 10 'primarily empirical' articles evoked above, and two of the three were published by JPR. The point made here is not that ST should embrace quantitative methods (although such a qualitative/quantitative imbalance is certainly unfruitful, as already argued [Baele and Thomson,  2017]), nor that quantitative methods are intrinsically superior (nothing prevents the parameters underpinning qualitative methods to be fully disclosed and justified, as done in 10 papers from our dataset). Rather, the data point to the widespread adoption in ST scholarship of what Morse and colleagues call (and condemn) 'evaluative procedures', that is, the practice of evaluating the quality of scholarship on the basis of a posteriori subjective impressions on 'the overall significance, relevance, impact, and utility of completed research' (Morse et al., 2002: 14), and the concomitant rejection of 'constructive procedures' whereby a priori objective criteria of internal validity and reliability are used to assess the strength of research contributions. While we recognize -and applaud -that diverse philosophies of science co-exist in IR in general and security studies more specifically, each entailing different criteria and practices when it comes to the conduct/progress of scientific enquiry and the definition of key concepts like causation (see, e.g. Chernoff, 2014;Hamati-Ataya, 2016;Jackson, 2016), our argument here is that such an overwhelming reliance on evaluative procedure is a significant characteristic of ST research which cannot be dismissed when one tries to understand todays' debates about the theory and its future.
Overall, we suggest that these two imbalances -preference for theoretical development over empirics, on one hand, disregard for constructive procedures, on the other hand -ought to be understood together as a symptom of more general practices and structures prevailing in the field of IR and specifically critical IR, and cannot be foreign to some of todays' debates as to where the theory is going, whether it has delivered or not, who it 'serves', or even more fundamentally 'what kind of theory -if any -is securitization', to use Balzacq and Guzzini's (2015) words.

Europe above the Rest
As already suggested above, we argue that this imbalance between theory and empirical work stems from the particular situation of ST within the discipline of IR, understood in more sociological terms as a scientific field where various communities co-exist and interact, each with their favoured practices, journals, standards for symbolic capital and so on. More specifically, and as Kristensen's (2012Kristensen's ( , 2018 network analyses of IR scholars' citations demonstrate, ST developed in close association with the poststructuralist critical IR cluster; that is, away from the central communities embracing the disciplinary canon, and with virtually no contact with the heavily empirical, data-driven communities, such as peace and conflict research. This situation largely explains the imbalances highlighted above, with critical IR traditionally valuing theoretical sophistication over steady empirical work and generally sceptical vis-à-vis attempts to unify IR around a central paradigm. 17 It is against this backdrop -and that of the abovementioned claims that ST is/ serves the 'West' -that we explore the geographical component of ST (i.e. where securitization research has been produced, and where it is published), hypothesizing that Kristensen's diagnosis of isolation -but also Lohaus and Wemheuer-Vogelaar's (2021) recent finding that IR journals possess their own, largely disconnected, pools of geographically situated authors -would be corroborated by our data, which would further explain the current situation of the framework. We provide two sets of measurement pertaining to the geography and disciplinary situation of ST: we coded both the country of each (co-)author's institutional affiliation and the journal where articles were published.
Starting with the journals publishing ST papers, two main findings sharply emerge. First, while ST can be found in all but two of the journals in our corpus, its distribution is heavily skewed towards a select few journals. In fact, as Figure 5 shows, almost half of the 171 papers (71 articles, 41%) were issued by Security Dialogue alone, while at the other end of the spectrum International Security and the Journal of Conflict Resolution have never published any piece on securitization (with the first-ranked International Organizations only issuing two papers over the 25 years). 18 Second, this uneven distribution corresponds to the United States versus Europe divide in terms of editorial boards' host institutions. SD, RIS, Millennium, IR, BJPIR and EJIR, which together account for no less than 78% of ST papers, are historically based at European universities, while IS, SS, IO, ISQ and WP, the US-based journals of our corpus, together only represent 8% of the scholarship. JPR and JCR, the two European journals that show a similarly negligible interest in ST, have since their creation gradually morphed into journals that publish almost exclusively a specific type of quantitative, dataset-driven research; as noted above, two of the only three quantitative ST papers from our corpus were published in one of these two journals. Assuredly, these results evidence how ST's scholarship reflects the fragmented nature of the broader discipline, and reflect the development of ST as a largely theoretical research agenda whose disregard for constructive procedures bars it from publications that put a premium on strong methods.
Turning to the regions and countries where the ST authors are based, similarly stark and complementary findings emerge, as shown in Figure 6. Not only is Europe by far the most productive region (76% of all mentioned author affiliations), but UK universities are by a large margin the epicentre of ST scholarship with more than three times the amount of papers published by authors located there than in the second most frequent location (Canada). Denmark, obviously home of the original 'Copenhagen School', is well-ranked in third. In contrast, the United States -which as an entity publishes twice as much Politics and IR publications than the United Kingdom 19 -appears less than 20 times, and never with an institution traditionally endowed with high symbolic capital in the scientific field (e.g. Ivy League universities). In addition, authors based in South America, Asia, and Africa are virtually absent, 20 which echo voices calling for a more 'global' IR and doubts about the existence of a 'subaltern' ST. Together with the previous observations on the journals, these findings directly confirm and reinforce those of Lohaus and Wemheuer-Vogelaar (2021) who recently demonstrated that IR journals tend to publish works by authors located in their own region, but more broadly also Baele and Bettiza's (2021) discussion of critical IR scholarship as being based in 'Western but non-American powerhouses' and published in select European journals. ST, as a theory developed in Europe, appears to have followed a path-dependency trajectory whereby its anchoring in (some) European journals became steadier as time passed. Given the preference of journals like SD and RIS for theoretically sophisticated papers and their disinterest for methodological developments, we suggest that this trajectory participated in shaping ST into a framework marked by the theory/empirics-methods imbalance evidenced above.
Together, these observations on the geography of ST further demonstrate and illustrate the more general clustering of IR research into various communities, which is by now well-documented. 21 Specifically they indicate that ST has remained steadfastly anchored in European scholarship, with only thin ramifications into IR's largely American 'mainstream'. Furthermore, by being disproportionately produced in the United Kingdom and published in one journal, ST scholars appear to have carved a niche rather than engaged in an outwards-facing effort aimed at 'translating' the theory to other communities. Recent claims that the theory has not been successful in incorporating voices and concerns from the subaltern (e.g. Bertrand, 2018) appear to be warranted, adding to the sense that the theory has not effectively spread beyond a rather narrow authorship base, at least when it comes to its development in major journals. Such a situation, we suggest, not only fuels some of the frustration and questioning (or, in some cases, outright disregard) displayed against ST by different communities of scholars, but also partly underpins the current stocktaking initiatives conducted from within ST scholarship.

Conclusion
ST celebrated its 25th birthday with a vitriolic pushback (Waever and Buzan, 2020) against a bombastic attempt to 'excavate [the theory's] racist foundations' (Howell & Richet-Montpetit, 2020), itself taking place within the broader context of a multifaceted discussion evaluating the theory's achievements and shortcomings. The present review of more than 170 ST articles published in the leading IR journals since 1995 sought to shed some light on this turbulence, highlighting two major interlocking imbalances that have characterized the development of ST so far. First, despite investigating a rich array of some of the most crucial issues of contemporary politics (from the securitization of pandemics to that of migrants, or drugs), ST scholarship is marked by a severe bias in favour of high-level theorizing and evaluative procedures at the detriment of empirics-driven knowledge relying on constructive procedures. Second, and connected to that imbalance, ST has not only remained European, it has also failed to significantly engage elsewhere and expand beyond its bastions of the United Kingdom and a couple of journals.
This review and findings therefore raise one overarching question around which current debates on the state of ST could be productively re-organized: do ST scholars find merit in engaging beyond their usual practices that have over time produced the rather narrow confines describe above? On the one hand, ST authors may want to consider effective ways to expand the authorship base towards under-represented regions, especially when it comes to pushing forward important theoretical advances and sustained empirical projects. On the other hand, ST scholars would need to decide if they wish to engage in efforts to 'translate' their framework to the largely American, more quantitatively trained 'mainstream' suspicious of evaluative procedures. Such efforts exist -for example, by bringing experimental methods into ST (Baele and Thomson, 2017), by combining discourse analysis with population surveys (Karyotis & Patrikios, 2010), or by shifting from qualitative to quantitative discourse analysis ) -but they remain the exception and come with risks of further theoretical fragmentation or distortion. Genuinely engaging beyond a comfort zone built and fenced over 25 years dramatically expands the horizon, but also requires effort and comes with risks; we hope that by laying bare this simple equation our review article has unlocked a potential for ST revitalization and will hence help, to loop back to Ned Lebow and Potenz's question opening this article, to secure the 'future of securitization studies'.