Why competing alternatives (especially for constructivists)?

This article’s core argument reflects common wisdom across the social sciences: scholars of all sorts instruct students to ask, ‘What’s an alternative interpretation?’ As we get into sophisticated epistemological discussions, however, we encounter reasons to limit that practice. As constructivism pushes us especially far into those discussions, some of the most salient arguments against competing alternatives come from constructivists.

Constructivism’s centrality to this discussion makes it important to define up front. I employ a minimal definition featuring the one element common to all strands: a constructivist argument asserts that social constructs matter for action. That is, constructivist scholarship argues that we cannot access something we should want to know about action without paying attention to interpretive social constructs like ideas, norms, practices, identities, or discourse. Scholars who make such arguments then add a variety of implications to that core hypothesis – other features they see as definitional of coherent constructivism – that lead into the arguments we encounter below. Modern constructivists connect a focus on social constructs with an interest in constitutiveness; interpretivists argue that it prioritizes understanding-style inquiry; practice-theory constructivists suggest that it encourages a holistic methodology; and postmodern or post-structuralist constructivists insist that it requires anti-foundationalism.

With that definition in mind, consider a simple epistemological case that competing alternatives are central to all empirical arguments, and especially fundamental to constructivist ones. Practically all social scientists today could agree that the meaning and significance of scholarship are substantially defined against interlocutors’ arguments. A few non-constructivists may still endorse naïve positivist epistemologies of ‘justificationism’ or ‘naïve falsificationism’, in which significant claims stand or fall in solitary testing against empirics.Footnote ² Most, however, subscribe to something more like Lakatos’s view that contributions are made in ‘three-cornered fights’ over which claims better fit evidence.Footnote ³ ‘Modern’ constructivists like Alexander Wendt or Colin Wight place a similar emphasis on empirical arbitration between claims – suggesting, basically, that appeals to evidence in a socially constructed world are problematic but not infinitely soFootnote ⁴ – and so presumably also posit ‘three-cornered fights’.Footnote ⁵ Constructivists who espouse more post-positivist epistemologies, meanwhile, are more skeptical of autonomy for the ‘evidence’ corner of the fight. This presumably makes arguments’ meaning and significance even more dependent on specifying their relationships to the corner(s) of other accounts. As Stefano Guzzini summarizes the gist of Friedrich Kratochwil’s writings, ‘Theories are not tested against reality, but against other theories’ (Reference Guzzini2010, 32).Footnote ⁶ Vincent Pouliot cites analogies from Kratochwil and Ricoeur to courtroom confrontations, arguing that ‘it is not reality but academic competition’ that validates constructivist claims (Reference Pouliot2007, 378). Once we reach postmodern views, the communicative interdependence of accounts arguably becomes complete. To paraphrase the historian Hayden White, narrativization only makes sense against other narratives (White Reference White1987, 20; Campbell Reference Campbell1998, 37). Or to cite one of Derrida’s many formulations of this idea, ‘… we can pronounce not a single destructive proposition which has not already had to slip into the form, logic, and the implicit postulations of precisely what it seeks to contest’ (Reference Derrida1978, 280).

Thus we might expect almost all social scientists today to agree that to communicate whatever they want to say about the world, they must contrast their accounts explicitly to others about the same empirical terrain. We might expect constructivists to hold this view especially strongly. Beyond agreement across epistemologies that we need alternatives to formulate our claims, scholars will disagree about what else we gain from such contrasts. For non-constructivists and the modern end of the constructivist spectrum, the additional benefit is validity: displaying as clearly as possible how one account fits evidence better than others. Toward the postmodern end, the additional benefit is social influence: clear contrasts to dominant accounts open space for contestation and change. Median positions might mix the two.

We might also expect constructivists to show special interest in concrete alternatives for a substantive, ontological reason. Although the expectations of many non-constructivist theories can be specified in acontextual ways – like, say, a realist hypothesis that states always balance against power – constructivists’ empirical claims always concern the role of social constructs in a concrete context. Whether constructivists privilege communication, validity, or social influence as reasons to entertain alternatives, then, such engagement is necessarily drawn into contrasting claims about a particular context. It means telling concretely contrasting stories about someone somewhere.

Does constructivist research reflect these expectations, featuring empirically contrasting accounts? Often it does. Yet advice from prominent constructivists includes four arguments that limit or reject such practices. Those about constitutiveness and understanding limit engagement with certain alternatives – non-constructivist ones. Those about holistic methodology and anti-foundationalism can seem to reject the whole notion of competing alternatives. Addressing these arguments in order of the degree to which they question competing alternatives, I show that none justifies ignoring our underlying epistemological dependence on contrasting accounts. All constructivists will do their best scholarship, as they define it, if they contrast their accounts empirically to alternatives, including non-constructivist ones.

Constitutiveness

The distinction between causal and constitutive inquiry is the most widely evoked impediment to competition between constructivists and non-constructivists, appealing to modern and postmodern constructivists alike. I agree that constitutiveness is fundamental to constructivism, but offer two arguments why it should not deter contrasts to non-constructivist alternatives. First, paralleling points from Eric Grynaviski (Reference Grynaviski2012), I show that constitutive and causal inquiry can usually be phrased to pose the same kinds of questions. More profoundly, I show that scholarly debates surrounding constitutive social constructs always evoke contrasting causal accounts of the events and processes that led to certain actions.

The most widely cited account of constitutiveness comes from Wendt (Reference Wendt1998, Reference Wendt1999). The causal ‘why?’ questions posed by traditional social science concern the way something came about, whereas constitutive ‘how?’ or ‘what’ questions concern the properties or capacities of entities. Constructivists are especially interested in the latter questions, though Wendt offers examples across disciplines:

Wendt notes that ‘how’ or ‘what’ questions may be descriptive precursors to explanation – what the methodological orthodoxy of King, Keohane and Verba (KKV) calls ‘descriptive inference’ (Reference King, Keohane and Verba1994, 37) – but argues that constitutive claims surpass ‘mere description’. As philosopher Richard Cummins (Reference Cummins1983) observes, some natural-science theories do not meet classic Humean requisites of causal argument: independence between cause and effect, temporal sequence, and constant conjunction. The double-helix model of DNA is not a ‘transition theory’ of causality, but a static ‘property theory’. These are testable claims about properties and capacities; they could be wrong, and the world would be different if they were. Allowing that scholars may disagree on the semantics, Wendt cites philosophers who note that property theories are routinely called ‘explanatory’ in many disciplines (Haugeland Reference Haugeland1978). He thus identifies a category called ‘constitutive explanation’:

The larger point of Wendt’s ‘constitutive explanation’ label is clearly that constitutive claims can be so important for conceptualizing action that they influence causal-explanatory theorizing. His immediate claim, though, is that we can make theoretical contributions in either realm: ‘Answering constitutive questions is an important end in itself, even if it is later tied in to a causal story’ (Reference Wendt1999, 86). Echoing famous language about the explanation/understanding divide (of which more below), he finds that there are always ‘two stories to tell’ about action (Wendt Reference Wendt1999, 86).

How influential has this advice been? Its image of distinct realms has not been directly challenged. Some modern constructivists dispute that constitutive argument qualifies as ‘explanation’. David Dessler and John Owen prefer ‘constitutive analysis’, and Nina Tannenwald proposes ‘explication’ (Dessler and Owen Reference Dessler and Owen2005, 599; Tannenwald Reference Tannenwald2005). Some note that the line between ‘how/what’ and ‘why’ questions is blurry. In the first book directly on constructivist research in IR, Audie Klotz and Cecilia Lynch observe, ‘Constructivists should not … preclude the possibility of causal answers to constitutive questions, or vice versa’ (Reference Klotz and Lynch2007, 15). Overall, I suspect that many modern constructivists accept Wendt’s position, but see the terms as debatable and the constitutive/causal line as tricky territory, reaching something like Tannenwald’s conclusion: ‘Whether analysis of constitutive effects is called “explanation” is less important than that scholars are clear on distinctions between causal and constitutive processes, especially as they play out in empirical research’ (Reference Tannenwald2005, 41).Footnote ⁷

Another indirect challenge comes from champions of critical realism. Modern constructivists like Milja Kurki, Wight, and Colin Hay draw from this philosophy of science a broader definition of ‘cause’ that subsumes ‘how’ questions, property-theory claims, and Wendt’s ‘constitutive explanation’ (Wight Reference Wight2006; Kurki Reference Kurki2008; Hay and Gofas 2010; also Laffey and Weldes Reference Laffey and Weldes2002, 204). As Kurki summarizes, ‘accounts of “constitutive” meanings, in most contexts, are essentially inseparable from causal claims’ (Reference Kurki2008, 236). Yet this view does not necessarily connect Wendt’s two realms; it just reclassifies arguments in both as ‘causal’. Kurki elaborates on another Wendt suggestion that ‘constitutive explanation’ highlights what Aristotle called ‘material’ and ‘formal’ causes – as opposed to affecting-and-steering ‘efficient causes’ – and argues that we must recognize that different causes ‘cause in different ways’ (Wendt 2003, 495; Kurki Reference Kurki2008, 218–34). Wight appears to echo Wendt that ‘there is a valid methodological division of labor between the explanation of particular acts and events, on the one hand, and the explanation of the properties of structures on the other’ (Reference Wight2006, 286). But then he finds that a ‘sharp distinction … is untenable’ (Wight Reference Wight2006, 276, 288). If critical realists clearly reclassify constitutive explanation as causal, they are less clear about the possibility of competition across Wendt’s categories.Footnote ⁸

A stronger but still indirect challenge comes from interpretivist or postmodern constructivists. For scholars like Steve Smith or Mark Bevir, the causal category does not apply to human action, for which only constitutive-style arguments are valid (Smith Reference Smith2000; Bevir Reference Bevir2006, 284; see also Laffey and Weldes Reference Laffey and Weldes2002). This amounts to strengthening Wendt’s distinction and relocating it to the natural/social sciences boundary (and is a restatement of the ‘understanding’ divide that I unpack next).

Thus Wendt’s view survives these exchanges: some constitutive claims do not tread on non-constructivist alternatives. I now offer two arguments and a demonstration that constructivists who follow this advice miss opportunities to clarify their own claims.

Understanding

Even if we accepted that constructivists’ constitutive claims always evoke causal alternatives, we might still see hold that the arguments constructivists and non-constructivists offer about transitional processes fall into ‘understanding’ and ‘explanation’ categories that bypass each other. In IR, Martin Hollis and Steve Smith are best known for arguing that ‘there are always and inevitably two stories to be told’ (Hollis and Smith Reference Hollis and Smith1990, 210, their emphasis) in social inquiry: explanation sets actions in patterns of regularities, whereas understanding accesses how people perceive and interpret. A stronger version argues that human action features no automatic regularities, so only understanding-style work is valid in the human sciences (Winch Reference Winch1958; Taylor Reference Taylor1971; Bevir and Kedar Reference Bevir and Kedar2008). Like with constitutiveness, I agree that understanding is basic to constructivism, but I think these views mischaracterize the contrast space to constructivists’ claims. Rather than bypassing understandings, most non-constructivist work offers alternative accounts of actors’ understandings. As these contrasts always exist around constructivist arguments, making them empirically explicit is always possible and useful for communicative clarity.

More than with constitutiveness, others have already mounted this challenge. Critical realists’ efforts to redefine causality stress that this divide depends on an obsolete Humean definition of ‘explanation’. An updated definition folds ‘understanding’ claims into the same space as most non-constructivist work. This section’s additional contributions reflect two features of how critical realists phrase this argument. First, perhaps because it has been couched within elaborate calls to critical-realist philosophy, it seems not to have reached some constructivists. As we see below, the divide has become more salient recently, not less, especially among a group of ‘interpretivists’ who seem unlikely to adopt the whole critical realist package. Second, as noted above, though critical realists construct a shared space for constructivist and non-constructivist work, they give little attention to competing alternatives within it. This section fills these gaps. I summarize in a relatively simple way how a move to post-Humean explanation erases the explanation/understanding divide. Then I consider research from Cecilia Lynch, a high-profile IR interpretivist and co-author of a book on constructivist research methods, to show how these scholars too can profit from contrasts to non-constructivist accounts.

Consider, then, common positions about ‘understanding’. The vocabulary comes from Weber, who argued that ‘the specific task of … sciences of action’ is ‘the interpretation of action in terms of its subjective meaning’ (Weber Reference Weber1958 [1922], 8). He thought the natural sciences were built on explanation, which he defined in Humean terms: situating an action in predictable patterns of what follows what. Adequate ‘causal interpretations’ of action, though, additionally require an interpretive account of actors’ understandings. Later scholars altered Weber’s distinction in two ways. Hollis and Smith recast explanation and understanding not as two components of ‘causal interpretations’ but as two separate stories, ‘each persuasive but not readily combined’ (Reference Hollis and Smith1990, V–VI).Footnote ¹³ They portray a division of labor between constructivists offering understandings and non-constructivists setting actions in patterns. Peter Winch (Reference Winch1958) and Charles Taylor (Reference Taylor1971), meanwhile, became citations for the view that only understandings apply to action.Footnote ¹⁴ Regularity-style explanation is invalid because people interpret and choose rather than reacting automatically to external conditions. Constructivists studying thinking beings do not overlap with non-constructivists studying falling apples.

Consider next recent evocations of these views. In her excellent survey of IR treatments of causality, Kurki shows that these arguments arise on the edges of modern constructivism and are pervasive among more post-positivist constructivists. Some modernish or median figures like John Ruggie or Nicholas Onuf take something like the Hollis/Smith position, retaining openness to classic causal claims, but justifying their focus on constitutive logics partly by stressing that a search for regularities cannot tell us much about action (see Kurki Reference Kurki2008, 124–44). More post-positivistic scholars like Kratochwil, Richard Ashley, Rob Walker, or David Campbell take the Winch/Taylor position, rejecting causal-explanatory thinking that they portray in Humean terms. More recently, the Winch/Taylor presentation informed the consolidation of an ‘interpretivist’ subgroup. Claiming the explicit allegiance of some IR figures like Lynch, but mostly outside of IR, this work takes understanding as its point of departure. Dvora Yanow and Peregrine Schwartz-Shea’s volume on interpretivist methods contrasts their ‘internal’ arguments to ‘external’ logic in ‘the method of the natural and physical sciences’ (Reference Yanow and Schwartz-Shea2006, 10). Mark Bevir, a philosophy-of-science specialist in the group, nuances that they ‘insist the human sciences are explanatory but distinguish the narrative form of explanation from the strictly causal form found in natural science’ (Bevir and Rhodes Reference Bevir and Rhodes2003, 42; Yanow and Schwartz-Shea Reference Yanow and Schwartz-Shea2006, XII). Elsewhere he elaborates that this ‘strictly causal form’ is ‘physicalist’ Humean explanation wherein action operates ‘just as gravity causes apples to fall’ (Bevir Reference Bevir1999, 174–87; see also Bevir and Kedar Reference Bevir and Kedar2008). Another inspiration in these directions is Bent Flyvbjerg, who rejects a search for predictive if–then relationships that presume ‘dead objects’ rather than ‘self-referential humans’ (Flyvbjerg Reference Flyvbjerg2001, 32; see also Schram and Caterino Reference Schram and Caterino2006).

What is the trouble with these positions? As part of their advocacy of critical realism, Wight and especially Kurki offer learned arguments that constructivist and non-constructivist claims should be understood as sharing causal territory (Kurki Reference Kurki2006, Reference Kurki2008; Wight Reference Wight2006; also Bhaskar Reference Bhaskar1987; Patomaki and Wight Reference Patomaki and Wight2000; Hay Reference Hay2004). Yet, we can reach the same practical conclusion for our research – that we can and should recognize these claims as potential empirical competitors – by a shorter path that more scholars may be willing to follow. Like with constitutiveness, we walk this path by recognizing two things about the non-constructivist ‘contrast space’ to claims about understanding.

First, although most non-constructivists indeed seek regularities, most are open to the post-Humean notion that valid explanations cannot bypass processes or mechanisms.Footnote ¹⁵ Consider the non-constructivist orthodoxy of KKV’s Designing Social Inquiry. It invokes a Humean-regularity view of explanation that seeks ‘average causal effects’ across many cases, and even holds that ‘We can define a causal effect without understanding all the causal mechanisms involved …’ (KKV Reference King, Keohane and Verba1994, 77, 86). Yet KKV’s overall position, while rather confused, appears to be that we measure average causal effects with correlations, but that valid ‘explanations’ should include traceable processes. They note twice on one page that ‘any coherent account of causality needs to specify how the effects are exerted’ (KKV Reference King, Keohane and Verba1994, 84). This is not a call to speculate; they want evidence for processes. They dislike inductive behavioralism and posit the decidedly post-Humean goal of ‘making inferences that go beyond the particular observations collected’ (KKV Reference King, Keohane and Verba1994, 8). They break with the ‘useful prediction’ mode of theorizing – best stated by Milton Friedman (Reference Friedman1953), and sometimes invoked in formal modeling – in which unrealistic models generate useful predictions. Such modeling may clarify our thinking, write KKV, but unrealistic models are ‘not of much empirical value’ (Reference King, Keohane and Verba1994, 106). They endorse process tracing as contributing to causal inference (KKV Reference King, Keohane and Verba1994, 224–28).Footnote ¹⁶ Most tellingly, Keohane and many of his students publish work that bases causal-explanatory claims mainly on process-tracing evidence (e.g. Keohane 1984; Moravcsik Reference Moravcsik1998). Overall, as John Gerring notes, ‘The near total absence of opposition to the current movement toward mechanisms’ in political science suggests that few true Humeans remain with us (Gerring Reference Gerring2010, 1503; see also Kurki Reference Kurki2006, 197). Some incorrigibles remain (e.g. Beck Reference Beck2006), but at the very least, almost all of their arguments suggest mechanisms, and KKV seem to tell us to try to trace them.

Second, the processes or mechanisms that most non-constructivists could offer for their theories clearly pass through actors’ understandings. Most non-constructivist work invokes images of objectively rational actors. As John Ferejohn remarks, an explanation built around rationality ‘obviously has an embedded interpretive perspective’.Footnote ¹⁷ Many KKV-style rationalists may seem not to take this seriously, despite aspiring to ‘specify how the effects are exerted’, as they also tend to hold that strong evidence for rational choice is impossible (as Elster (Reference Elster1986, 12–16) explains best). However, the most common non-constructivist position is that such accounts can rely on partial evidence: if they display patterns of action that correlate to salient constraints and incentives, some evidence of roughly rational-looking decision-making supports a plausible mechanism connecting the two (Fiorina Reference Fiorina1995; Levi Reference Levi1997). Constructivists may rarely find such evidence remotely convincing, but that is precisely the point: their objection is that rationalists’ process claims are unpersuasive, not that they make no such claims. Most non-constructivists are not explaining action ‘as gravity causes apples to fall’. They propose mechanisms by which people interpret or decide and generally maintain that these mechanisms are partly demonstrable in process-tracing. Certainly their interpretive claims are thin; the point of invoking rationality is to simplify perception and decision-making to a generalizable minimum. For constructivists, though, the correct objection is not that such arguments bypass claims about processes by which actors actually arrive at action. It is that they are wrong about these processes.Footnote ¹⁸

In sum, any interpretivist understanding-style claim faces non-constructivist alternatives that actors’ perceptions and decision-making are roughly objectively rational.Footnote ¹⁹ Any interpretivist claim can be clarified by contrasts to such alternatives (evidence permitting). To see these potential benefits concretely, Lynch’s work is helpful. Her ‘interpretivist approach to interwar peace movements’ departs from the observation that ‘the focus of interpretivism on understanding and meaning differs from the emphasis of logical positivism on explanation through generalizable patterns or covering laws’ (Reference Lynch1999, 10). She adds, however, that ‘Interpretivists do not shy away from articulating well-developed, coherent understandings that improve on past explanations’ – noting that interpretivists do not share a postmodern reluctance to champion particular accounts (Lynch Reference Lynch1999).Footnote ²⁰ Her book departs from direct criticisms of realist and liberal arguments that interwar peace advocates either had no effect on state policies or nudged governments dysfunctionally toward appeasement in the 1930s. Lynch argues instead that peace groups acted ‘in extremely important ways’, most crucially in ‘legitimizing norms that underlay global international organization and hence the construction of the United Nations’ (Reference Lynch1999, 19). Once an interpretivist approach highlights the ways in which social groups contest and legitimize political change, she suggests, we will tell the UN story differently.

As Lynch constructs her account, though, the notion that interpretivism occupies separate terrain from other accounts prevails over articulating improvements or even distinctiveness vis-à-vis their stories. She shifts away from trying to show that peace movements had any particular relationship to policies of appeasement or building international organizations, arguing:

Here the goal has changed: tracing how discourses converged is different from arguing that one influenced the other. Rather than elaborating how peace groups were ‘extremely important’, the book shifts to documenting contestation that other accounts have disregarded. This sometimes seems suggestive of certain relationships, like when she shows that peace groups pressured politicians into supporting a ‘World Court’ in the 1930s (Lynch Reference Lynch1999, 136–9). However, glimpses of influence are exceptional in an account that concentrates on displaying discourse from peace groups and some policy-makers. Late in the book, she cuts back her criticism of a conventional, elite-focused narrative of US policy-making to say simply that it is ‘insightful’ but ‘… still overlooks the role of social contestation and debate in legitimizing norms underpinning the agenda of global international organization’ (Lynch Reference Lynch1999, 207). Much like Hall’s study of Korea, she shows us intriguing discourse, but not its role in the UN story. To reach her own announced goals, she would need to say how this discourse generates a distinctive tale.

Like with Hall, I strongly suspect that Lynch’s rich research could answer this objection in powerful ways. Moreover, like with Hall, shortcomings of her account are not my real target. No matter how she fashioned her research design, some of its silences seem authorized by the Hollis/Smith and Winch/Taylor advice that constructivists cannot debate non-constructivist alternatives. This guidance is simply wrong that most non-constructivists are indifferent to how actors understand action. As with constitutiveness, the main cost is that constructivists fail to use alternatives to shape their own claims.

Holistic methodology

A third argument that encourages constructivists to minimize competing alternatives is a form of holism. Unlike the preceding notions that constructivist claims occupy separate space from non-constructivist ones, this one posits their inextricable interdependence. As any sophisticated theory recognizes deep interplay between ideational and non-ideational conditions, runs this view, we must avoid cartoonish clashes that strain out dynamics too starkly. The complexity of a socially constructed world makes it misleading to entertain competing alternatives.

The most notable champion of this position was Bourdieu.Footnote ²¹ He thought his ‘structural constructivist’ approach implied a holistic methodology, disliking competition between analytically distinct accounts so strongly that a former student summarized his opus under the heading, ‘The Refusal of Theoretical Alternatives’ (Pinto Reference Pinto1999, 98; also Calhoun et al. 1993, 2).Especially important to Bourdieu was that we avoid separating ideational and non-ideational dynamics. He theorized action as structured by practices: ‘strategies of action’ that people learn to employ within the ‘fields’ of action they inhabit (Bourdieu Reference Bourdieu1977, Reference Bourdieu1990, Reference Bourdieu and Wacquant1992; also Swidler Reference Swidler1986). A key reason to focus on practices, he suggested, was to capture interrelated ideational and material aspects of action. Practices are evolving amalgams of socially constructed interpretations and strategic responses to material conditions. For Bourdieu, an advantage of focusing on these amalgams is that they are more concrete, visible, and proximate to action than traditional focuses of cultural sociology like values, ideas, or norms. Bourdieusan practice theory thus presents itself as both more theoretically rich and more methodologically tractable than other conceptions of social construction, replacing a false straining out of ideational and material dynamics with more direct study of action.

This stance has recently become prominent in IR and political science with the rise of Bourdieusan constructivism.Footnote ²² Emmanuel Adler led the way, together with younger scholars like Vincent Pouliot (Reference Pouliot2010), Frédéric Mérand (Reference Mérand2010), and Rebecca Adler-Nissen (Reference Adler-Nissen2012, Reference Adler-Nissen2014). In the most prominent statement, Adler and Pouliot (Reference Adler and Pouliot2011, 12) present their approach under the heading, ‘Overcoming dichotomies in social theory’. They echo Bourdieu’s critiques of competition, stressing the simultaneously ‘material/meaningful, structural/agential, reflexive/background, and stability/change attributes of practice’ (Adler and Pouliot Reference Adler and Pouliot2012, 17). Practices are the ‘“gluon” of IR’, ‘the ontological core concept that amalgamates the constitutive parts of social international life’ (Adler and Pouliot Reference Adler and Pouliot2012, 10). Downplaying any ‘bracketing’ of these relationships (which ‘can only take us some distance in understanding world politics’), they describe practices as a ‘conceptual focal point’ around which many approaches can ‘cluster’ to pursue the goal of ‘cross-fertilization – the engine of social scientific refinement’ (Adler and Pouliot Reference Adler and Pouliot2012, 4, 11, 15). They accept ‘a healthy dose of competition’, but then hold that ‘pigeonholing’ arguments into less holistic claims is ‘untenable from a practice perspective’ (Adler and Pouliot Reference Adler and Pouliot2012, 10, 11, 17).

What is the problem with this guidance for constructivist research? In substantive terms of conceptualizing social constructs and action, practice theory’s promise cannot be denied: the ‘practice turn’ revolutionized cultural sociology decades ago (see Swidler Reference Swidler1986). I expect that its insights will become increasingly prominent in IR as well. But readers will have already perceived objections to the methodological move that accompanies Bourdieusan constructivism. Even if a focus on practices helps us construct substantive arguments that bridge or bust entrenched dichotomies, we can still only communicate those substantive merits in contrast to other accounts on the same terrain.

Consider the example of Pouliot’s book on Russian–NATO relations. He aims ‘to demonstrate that in order to understand interstate pacification, our theories need to be attentive to the logic of practicality on the ground of diplomacy’ (Reference Pouliot2010, 6). He portrays Russian–NATO relations in the early 1990s as resembling a security community of ‘self-evident diplomacy’, with reflexes of consultation (Pouliot Reference Pouliot2010, 1). Then, he argues, NATO’s decisions to enlarge eastward provoked Russian reactions that shifted the relationship into distrust. The Russians expected to be treated as privileged partners within a Western security community. Westerners, meanwhile, thought that Russian decline and democratization made it a ‘normal’ country that should be welcomed but not accorded special vetoes over the region. Pouliot interprets the shift through Bourdieu’s concept of ‘hysteresis’, which proposes that change comes when ‘dispositions are out of phase and practices are objectively ill-adapted to present conditions because they are objectively adjusted to conditions that no longer obtain’ (Pouliot Reference Pouliot2010, 48). Western leaders failed to see that Russians could not adjust their diplomatic practices so rapidly to their declining position, leading to missteps whose bitter fruit we harvest today.

As ‘hysteresis’ suggests, Pouliot does claim some ability to parse out theoretical dynamics empirically. He shows us ‘present conditions’ that are roughly what other scholars call material and institutional conditions, and argues that Russian practices diverged from them, presumably for ideational reasons. What he does not do is suggest another interpretation of these events. He touches on other approaches ‘less as alternative explanations than as pieces of a larger puzzle’, noting that realism, Marxism, and other approaches relate to parts of his account (Pouliot Reference Pouliot2010, 257). As a result, although sympathetic reviewers find Pouliot’s theorizing promising, they are unsure what he argues empirically. For Ted Hopf, whose own work draws on Bourdieu, ‘The evidence Pouliot presents certainly supports his narrative of events, but absent a more systematic consideration of alternative accounts, it is very hard to know whether other approaches might have written the same story’ (Hopf Reference Hopf2011, 773; see also Ringmar Reference Ringmar2014, 9). Another reviewer regrets that ‘it is not clear what added value the almost exclusive focus on Bourdieusian concepts brings to Pouliot’s story’ (Forsberg Reference Forsberg2012, 171).Footnote ²³

Fortunately, much other scholarship decouples the substantive promise of Bourdieusan constructivism from holistic methodology. Hopf provides one example in IR. He builds on Bourdieu and American pragmatists to theorize a ‘logic of habit’ in international relations, and is not shy about hypothesizing distinctive expectations for IR if such dynamics are operating (Hopf Reference Hopf2010). More broadly, Bourdieu was a major inspiration for the ‘practice turn’ in cultural sociology, and also for sociological institutionalism in organizational sociology. In both subfields scholars routinely present arguments that highlight material and ideational elements of practices, ‘bracket’ sophisticated interplays of structure and agency, and explain both stability and change in direct contrast to non-constructivist alternatives.Footnote ²⁴ In my view, the sharpest formulation of distinctive narrativization along these lines comes from sociologist Richard Biernacki, who suggests that the key challenge is to show ‘that culture [or ideas, norms, discourse, etc.] exercised an influence of its own but not completely by itself’.Footnote ²⁵ That is, an analysis with Bourdieusan-style ambitions must highlight distinctive aspects of action that flow from social constructs, whereas also tracing how other conditions affect the story. This can only be done in empirical contrast to non-constructivist alternatives, as Biernacki (Reference Biernacki1999) develops in a brilliant critique of major works in cultural history.

Thus I am confident that Pouliot, like other well-grounded researchers working in the Bourdieusan tradition, could structure his analysis of Russian–NATO relations to make a distinctive contribution. Practice theorists will realize their substantive promise if they reverse Bourdieu’s methodological logic: the more complex our theorization of action, the more we need competing alternatives to highlight differences from narrower accounts we have bridged or combined.

Anti-foundationalism

The deepest argument that divorces non-constructivist alternatives from constructivist research portrays an epistemological divide between ‘foundationalist’, ‘essentialist’, or ‘empiricist’ views on the one hand and ‘anti-foundationalism’ on the other. Most post-positivist, ‘radical’, or ‘postmodern’ constructivists espouse anti-foundationalism, which rejects any ‘correspondence theory’ view in which we can make truth claims that correspond to reality. At first glance, this stance might seem to make it pointless to set up empirically competing alternatives. The core idea is that empirics cannot ultimately arbitrate anything. It also implies that the empirical accounts of ‘foundationalist’ scholars may be hopelessly narrow, unwilling as they are to question their own interpretive context. Moreover, to dignify these arguments with debate reproduces the social context that empowered such narrow-mindedness. Richard Price and Christian Reus-Smit nicely summarize how this logic lies behind the ‘the general reticence, and at times explicit refusal, of [many radical constructivists] to cast their substantive claims about world politics in relation to alternative accounts …’:

The ‘explicit refusals’ Price and Reus-Smit mention come from major postmodern thinkers like Richard Ashley, R.B.J. Walker, and David Campbell. Ashley and Walker champion dissident thought in IR that questions foundationalist ‘sovereign’ views and discourses, both among political actors and among academic observers, whereas arguing that we should not advance equally overconfident interpretations to replace them. ‘Ambiguity and uncertainty’, they stress, ‘are not here regarded as sources of fear in themselves’ (Ashley and Walker Reference Ashley and Walker1990, 263). Campbell too aims to question and deconstruct without insisting on a new narrative. Invoking literary critic Susan Sontag’s essay ‘Against Interpretation’, he characterizes his goal as ‘to show how something is what it is rather than what it means …’ (Campbell Reference Campbell1998, 5, original emphasis). The more common position that Price and Reus-Smit call a ‘general reticence’ to entertain alternatives is visible in postmodern or post-structuralist work from scholars like James Der Derian Reference Der Derian1987, Roxanne Doty (Reference Doty1996a), Jenny Edkins (Reference Edkins1999), Cynthia Weber (Reference Weber1996), Jutta Weldes, or Maja Zehfuss (2002). They make no bones about offering extensive empirical characterizations of discourses. Moreover, at the level of the actors they study, they pay careful attention to the contrasting rhetorics that people employ in politics. At the level of their own discussion, however – between academics – they seem not to see value in contrasting their accounts to others.

As the tone of the citation above implies, Price and Reus-Smit precede me in arguing that more engagement could be legitimate and useful for postmodernist or post-structuralist constructivists. This section builds on their points, and some from Ole Waever and Douglas Houghton, but takes them in a new (if compatible) direction. These calls for engagement between radical or postmodern constructivists and other scholars stress the possibility for engagement. To the extent that they give reasons to exploit this possibility, they suggest benefits of shared insights and synthesis. I argue that the communicative value of contrasting accounts sustains a more forceful case.

Why, then, might ‘radical’ constructivists wish to engage other scholars’ empirical claims? Unlike with constitutiveness or understanding, I do not think that the anti-foundationalist position mischaracterizes basic relationships between kinds of arguments. The oversight here is more like the way in which Bourdieusan constructivists miss how alternatives help their own project. In my view, postmodern constructivists need not alter any of their coherent positions to see that alternatives, including non-constructivist ones, can help them reach their own goals. Such practices make sense within their epistemology.

A first step in this direction, emphasized by Price and Reus-Smit and also Houghton, is to observe that postmodern and post-structuralist constructivists clearly make concrete empirical claims. Even those who explicitly refuse to champion a new narrative still advance empirical characterizations. Walker’s seminal postmodern retort to Robert Keohane in 1989 had no problem asserting that certain explanations of early modern politics were ‘implausible’ or ‘inadequate’ alongside several other broad empirical claims. Campbell offers a particular perspective on the Bosnian tragedy, as is best highlighted by Lene Hansen’s moderately different post-structuralist account (of which more below) (see Hansen Reference Hansen2006, 217–20). Der Derian, Weber, Weldes, Doty, or Zehfuss obviously aspire to highlight something specific in empirics. This is unsurprising from a postmodern or post-structuralist point of view: ‘As Nietzsche pointed out long ago’, write Price and Reus-Smit, ‘we cannot help putting forth truth claims about the world’ ( (Reference Price and Reus-Smit1998, 272). We may add the caveat that our claims correspond to no truth, however tentative, but still any empirical work draws its audience’s attention to certain narratives. Houghton makes the stronger version of this argument that the truth status of our claims ‘has no real impact on what we do as scholars when we look at the world “out there”’ (Reference Houghton2008, 45).Footnote ²⁶ He may overstate the irrelevance of epistemological differences, but his core point is hard to reject: whether we qualify our narratives as True, tentatively valid, plausible, or simply provocative, we draw attention to certain empirical characterizations.

A second step is to observe that postmodern constructivists should not object to locating their empirical claims in a shared space with non-constructivist ones. Waever develops an argument along these lines under the heading of ‘the postmodern solution to the problem of incommensurability’ (1996, 170). Post-structuralists in particular should be skeptical that claims made from one perspective simply cannot speak to those from another:

He effectively concludes that post-structuralists stand between deep incommensurability and easy, positivist-style commensurability (wherein meanings are unproblematic). In this epistemological territory, different perspectives rest on disconnected meanings but are not entirely impenetrable to each other. To make a similar point more broadly, it seems clear that post-structuralist and postmodern constructivists should be considerably more comfortable entertaining empirical claims from non-constructivists than the reverse. After all, it is the positivist or modern side that characterizes their own claims as scientific and those of postmodernists as not. From the postmodern side, all empirical claims share the status of politically competing narratives. Realist claims (for example) may rest on assumptions that a postmodern constructivist rejects profoundly, but the deeper postmodern point is that realist claims are not actually epistemologically different from their own. Both represent political agendas.

A third crucial step is to explain what postmodern or post-structuralist scholars gain from such engagement. Even if they can legitimately make empirical claims that engage non-constructivists, why do so? Why contrast an argument like Cynthia Weber’s account of ‘simulated sovereignty’, for example, to a narrow narrative based in realism? Why further legitimize realism in so doing? Price and Reus-Smit suggest that self-reflective scholars should entertain alternatives out of ecumenical recognition of their possible substantive merit: ‘All accounts of the world are partial, whether they be rationalist or constructivist, and the best that can be claimed on behalf of either is that they illuminate aspects of an event or phenomena that are required for an adequate understanding of the explanandum in question’ (Reference Price and Reus-Smit1998, 280).Footnote ²⁷ This seems plausible to me, but it may not to radical constructivists: they might question that the middle ground is best, or whether realism or any other orthodox theory is anything more than a waste of time. More compelling is an emphasis on the communicative value of competing alternatives.

That answer is foreshadowed by the argument of the previous section, which applies still more obviously to anti-foundationalists.Footnote ²⁸ In a fully social post-positivistic epistemology, shorn of any correspondence theory of truth, it is especially ironic to offer a solitary empirical account without explicit attention to the social space of claims around it. To do so seems to imply a special methodological enlightenment akin to a naïve positivist’s faith in methods: if only we use the mind-opening tools of Foucault, Derrida, or Baudrillard, we can directly penetrate empirics and communicate about them in new, provocative, critical ways. Far more coherent for anti-foundationalists is to hold that our attempts to communicate anything about fundamentally contested empirics derive their shape and significance from relationships to other ways of narrating the same stories. We may indeed risk partly legitimizing dominant accounts in engaging them, but this risk is unavoidable for those seeking to open minds. To mount a challenge to dominant views, we cannot simply flee in a lifeboat of free-floating interpretations.

To concretize this criticism, consider the prominent example of Weber’s sovereignty book. Starting from a paraphrase of R.B.J. Walker that resembles Rodney Bruce Hall’s core claim –‘what counts and/or functions as sovereign is not the same in all times and places’ (Weber 1996, 2–3) – Weber asks, ‘… how do practices of theorists and diplomats stabilize the meaning of sovereignty and, by default, write the state?’ In accounts of interventions under the Concert of Europe and the American administration of Woodrow Wilson, she employs a Foucauldian approach of a ‘logic of representation’, asking what actors represented about sovereignty over time. Then for accounts of interventions under Ronald Reagan and George H.W. Bush, she argues that the ‘referent’ for sovereignty became so ambiguous that we must invoke Baudrillard’s ‘logic of simulation’, in which representation is impossible and actors substitute ‘signs of the real for the real itself’ (Weber Reference Weber1996). Each case takes the form of a dense empirical account and a discussion of what Foucauldian or Baudrillardian narratives reveal about it. Doty’s review in the American Political Science Review stresses Weber’s empirical credentials: ‘All too often, critical approaches such as Weber’s have been judged as capable of offering only negative and destructive critiques, with no positive contributions based upon concrete historical and empirical research. Simulating Sovereignty should put to rest such critiques’ (Reference Doty1996b, 237).

Yet if Weber is clearly empirical, it is harder to see how she is constructivist. In the Concert of Europe cases, she stresses that claims about sovereignty and intervention reflected distinct dynastic and nationalist logics. Monarchs generally attributed sovereignty to the monarch and so often presented cross-border interventions to preserve monarchs as legitimate. Revolutionaries rejected such interventions as violations of sovereignty they rooted in the people. Wilson later championed the nationalist logic and claimed, in rather tortured rhetoric, that his interventions against regimes in Mexico and Bolshevik Russia (and ostensibly for their ‘peoples’) maintained ‘scrupulous respect for [their] sovereignty’.Footnote ²⁹ As nicely as Weber’s accounts highlight these changing and non-obvious invocations of sovereignty, they are not controversial. No scholar would disagree that these actors said these things. The meaningful debates concern how to interpret them: are state leaders simply pursuing security with flexible justifications? Are monarchs and revolutionaries rationalizing class positioning? Are revolutionaries animated by the ideas of the Enlightenment? To what degree is Wilson a champion of self-determination or just an apologist for the demands of US business? Weber’s interpretive sections offer no interpretation, simply reasserting that multiple discursive positions were at work. On the Concert of Europe, she even offers the Krasneresque conclusion that the Concert was ‘never more than the formalization of a military alliance against France’ (Reference Weber1996, 60). On Wilson, she just qualifies his discourse as a ‘political strategy which … worked to silence or defer questions about how ‘selves’ or ‘identities’ were produced …’ (Weber Reference Weber1996, 91). On Reagan and Bush, where her framework shifts from Foucault to Baudrillard, the conclusion just seems to be that ‘sovereignty’ and ‘intervention’ are invoked so vacuously that ‘They no longer produce meaning’ (Weber Reference Weber1996, 121). Except for some rather nihilistic language about the last cases, these accounts would not trouble a realist, Marxist, liberal, or any other kind of scholar.

One last time, let me stress that my criticism is driven by optimism for Weber’s approach. I strongly suspect that she has distinctive things to say about these cases. In some other post-structuralist work, we find hints of this potential. Lene Hansen’s methodologically explicit discursive account of the Bosnian war ends with a concrete differentiation between her own narrative and that of David Campbell, clarifying both books in the process (Hansen 2006, 217–20). Hansen does not extend this contrast to other serious scholarly claims about Bosnia, and her four chapters of methodological advice omit any mention of contrasting accounts – but she nonetheless hints that such contrasts can be appropriate and useful within anti-foundationalist commitments.Footnote ³⁰ Without them, a project like Weber’s is left floating without reference points. The real cost of this move, once again, is for these scholars’ own agenda. They fail to use their social context to give shape to meaningful tales for themselves.

Conclusion: contrasting accounts and eclectic ones

This article has argued that constructivists have basic epistemological reasons to contrast their accounts empirically to others, including non-constructivist alternatives, and that the main views that limit or deny that practice crumble under scrutiny. Constitutiveness is basic to constructivism, but constitutive claims about action always confront competing causal claims about why people ended up taking these actions. ‘Internal’ understandings are fundamental to constructivism, but non-constructivists also make competing (if often crude) ‘internal’ claims about processes of action. Holistic methodology arises from a constructivist sensitivity to complex and contingent dynamics of action, but complex arguments take shape by relying more (not less) on contrasts to alternatives. Anti-foundationalism portrays social inquiry in profoundly different terms from non-constructivists’ epistemologies, but also implies that constructivist accounts derive much (or all) of their meaning and potential for social influence from positioning vis-à-vis dominant scholarship.

To conclude, let me contrast this case for competing alternatives to the ‘eclectic turn’, which has come principally from constructivists. Often informed partly by American pragmatist philosophy, which mixes an anti-foundationalist rejection of universal truths with a practical focus on ‘what works’ to solve meaningful problems in a given context, these arguments advise scholars to downplay epistemological divides and seek ‘middle-range’ engagement across approaches. The highest profile version is Rudra Sil and Peter Katzenstein’s ‘analytic eclecticism’: ‘an intellectual stance that supports efforts to complement, engage, and selectively utilize theoretical constructs embedded in contending research traditions to build complex arguments that bear on substantive problems of interest …’ (Sil and Katzenstein Reference Sil and Katzenstein2010, 411).Footnote ³¹ IR theorists of all sorts have joined this chorus. On the positivist side, David Lake argues that rather than grand-theoretic confrontations, ‘We would all be better off … focusing on achieving progress within each approach that addresses important real-world problems’ (Reference Lake2013, 568). Jeffrey Checkel describes such views as common today in traditional American IR, noting an ‘explosion of interest in bridge building’ since the mid-1990s (Reference Checkel2013, 221). More post-positivist IR theorists like Patrick Jackson and Daniel Nexon champion a very broad and deep pluralism (Jackson Reference Jackson2011; Jackson and Nexon Reference Jackson and Nexon2013). Christian Reus-Smit (Reference Reus-Smit2013) and Stefano Guzzini (Reference Guzzini2013) also espouse a broader eclecticism that extends across epistemologies and includes normatively focused scholarship as well.

From this article’s perspective, this literature gives strikingly little attention to competing accounts. These scholars justify engagement almost entirely in combinatory rather than competitive terms. Even their invocations of pluralism lack the competitive connotation the term usually carries. In their hands, it becomes either a logic of ‘live and let live’ – or ‘disengaged pluralism’, as Timothy Dunne, Hansen, and Wight term Lake’s position – or a synonym for cooperation, as in Checkel’s (Reference Checkel2013, 224) ‘theoretical pluralism’: ‘an explicit effort to utilize insights and variables from two or more theoretical approaches to make better sense of a real-world problem’. Dunne, Hansen, and Wight (Reference Dunne, Hansen and Wight2013, 417) espouse an ‘integrative pluralism’, with ‘researchers from multiple perspectives engaging in the practice of pluralism through engagement with alternative positions where their concerns and research interests overlap’. Jackson suggests an ‘engaged pluralism’ that sounds slightly more contrastive, featuring ‘conversations that unfold without necessarily resulting in either agreement or stalemate, but instead produce ever-finer differentiations and specifications brought on by the difficult intellectual labor of translation’ (Reference Jackson2011, 207; see also Suganami’s critique (Reference Suganami2013)). Yet, the thrust of his book on the philosophy of social science converges on Lake’s ‘disengaged pluralism’, portraying approaches operating in separate conceptual spaces. Moreover, with Nexon, he lauds Pouliot’s alternative-free work as showing ‘precisely what vibrant international theorizing puts on the table: it provides new ways of interpreting the world, and the opportunity to debate those understandings’ (Jackson and Nexon Reference Jackson and Nexon2013, 557).

I am nonetheless hopeful that these scholars will read this article as compatible with their views. Their routes to eclecticism or pluralism generally depart from ontological open-mindedness (many things could be happening in our world) and from epistemological humility (we may gain insights from many perspectives). An emphasis on competing alternatives departs instead from a need for communicative clarity – but that need is especially pressing in a world where many things could be happening and which many perspectives claim to illuminate. Far from being incompatible, these points make sense together. We must engage empirically competing alternatives at the beginning and end of any eclectic effort. If we want a combined argument to deepen our understandings rather than just pulverizing them, we must first contrast our multiple narrative threads to each other to highlight their distinct content as best we can. Then, once we knit these threads into a multipart argument, we must contrast it to other (presumably simpler) narratives to communicate the distinctiveness of eclecticism.

At the same time, this message may appeal to scholars who are skeptical of eclecticism. Strongly committed constructivists, like the most convinced members of any school, may not agree that many things could be happening. They may bet that it is ‘ideas all the way down’ (Wendt Reference Wendt1999, 92). Nor may they feel humble about their epistemological views. Whereas such theorists might ignore arguments for eclecticism that depart from middle-ground bets, they should still acknowledge that they can only tell meaningful stories about a deeply socially constructed world if they relate them to other scholars’ stories on the same topics. This point underscores that an emphasis on empirically contrasting accounts is ultimately the single most basic element of all empirical inquiry in the social sciences, from its modern to postmodern extremes. Whether our stories are aggressive or humble, parsimonious or complex, True or socially provocative, they only take meaningful shape in contrast to other tales.