Setting the limits: Discourse and EU foreign policy

Discursive struggles

When, in 1999, I wrote ‘Speaking “Europe”’ for a special issue of the Journal of European Public Policy on constructivism and European integration (Diez, 1999), discourse analysis and poststructuralism had hardly been used as approaches in European Integration Studies. What I outlined then was rather sketchy, suggesting three possible moves in the analysis that clarified the importance of language in European integration: an ‘Austinian’ move focusing on speech acts; a ‘Foucauldian’ move steering away from individual speech acts to the importance of broader discourses on the construction of meaning; and a ‘Derridarean’ move alerting us to the importance of ever-changing differences in this meaning construction. The context within which ‘Speaking Europe’ was written was the debate about the formation of a system of governance on the EU level. This was an inward-looking debate that was partly concerned with the features of the evolving EU governance system and its legitimacy, and partly raised the question of a much longer discussion about how the emergence of this system can be explained (see Diez and Wiener, 2009: 6–9).

At least two things have changed since then, which provide the need to revisit the ‘Speaking Europe’. Both are rather obvious.

First, while discourse analysis has by no means achieved mainstream status in European Integration Studies, it is nonetheless fair to say that in the meantime there has been a plethora of writings using a discourse-analytical approach to analyse European integration, as testified by the introduction to this special issue (Carta and Morin, 2013). As one would expect, these writings took different turns, and thus have raised a number of questions. One of these, although less interesting for the purposes of this article, is the question of whether discourse consists solely of verbal and written utterances or whether discourse includes all meaningful practices, from flag hissing to hand shaking, from images to silences. While the exact relationship between the verbal and the non-verbal is still to be conceptualised, at the same time it is hard to see how a narrow conception of discourse focusing only on the verbal is tenable, as clearly non-verbal interaction also produces meaning (Hansen, 2006; Williams, 2003). This leaves two further questions, which will be addressed in the following. One is whether discourse analysis is primarily a critical approach or whether it can be used for explanatory purposes. The second one is whether the role of discourse is to positively enable particular policies or whether it works through constraining the policies that can be meaningfully and legitimately pursued.

Second, since 1999 there has been a much greater focus on the EU as an international actor. This is in part a consequence of the further development of the Common Foreign and Security Policy into the European Foreign and Security Policy as set out in the Lisbon Treaty, and the strengthening of Common Security and Defence Policy. It is also a reflection of an academic debate that moved from the question of whether the EU is an international actor to the question of what kind of actor it is (Manners, 2002). This change adds a different empirical emphasis to discourse analysis in European Integration Studies, moving beyond the inward-looking emphasis of the late 1990s to the question of how ‘speaking Europe’ works in EU foreign policy, as much as how others ‘speak Europe’.

My article’s argument in relation to these issues is as follows: first, while I see discourse analysis primarily as a critical engagement in the sense of interrogating prevailing understandings in politics, this is logically based on the presupposition that discourse provides the context in which individual policy articulations are set, and in that sense can contribute to the explanation of policy, although this implies a relaxation of the definition of explanation away from its narrow causal-analytical, positivist sense. Second, the way in which discourse informs policy articulations works both through providing meanings on which one can build and through setting the limits of a meaningful and legitimate policy. Both this ‘enabling’ and ‘disabling’ of articulations is set in a continuous political struggle, which in turn links the critical to the ‘explanatory’ purpose of discourse analysis. Third, these issues play a core role in the analysis of EU external and foreign policy. This is partly the case because the poststructuralist argument that discourse constructs meaning through difference – and therefore through setting limits – is at the heart of analysing the role of foreign policy in identity construction, and therefore at the heart of analysing the interconnections between EU foreign policy and an emerging EU identity. Yet, as I will also argue, the construction of ‘normative power Europe’ and the norms embedded therein can be used as an illustration for the discursive struggles in the establishment of a particular hegemony of meaning.

Taking up a core theme of this special issue (see Carta and Morin, 2013), one of the core concepts of this article’s approach is therefore that of ‘discursive struggle’. It puts emphasis on the argument that meaning does not simply exist as a given, but has to be fought over and negotiated. In that sense, the article’s conception of discourse is different from, and perhaps more fundamental than, the one employed by Vivien Schmidt (2013), who focuses on the coordinative and communicative function of discourse. The article will not dispute that actors can use discursive practices for coordinative purposes or to communicate policies in order to legitimise them. Yet underneath these functions, there is a more fundamental process of meaning production which takes the form of a continuous struggle over meaning, for instance of the notion of ‘liberalism’ as part of a normative European external and foreign policy (see Rosamond, 2013, for discussion of this example).

‘Struggle’ in this context operates on three levels. On the level of the individual discourse participants, they are engaged in an articulatory struggle in which they have to negotiate the competing demands arising out of complex discursive contexts in order to provide themselves with a reasonably (but never fully) coherent set of meanings. On the level of discursive positions, collective actors articulating their views from these positions are engaged in a struggle with other positions, sometimes compatible, sometimes competing, and by no means necessarily in line with the positions of ‘executive actors’ (Carta and Morin, 2013). On the level of the overall discourse, the picture is one of competing discursive positions that are not only actively pursued by collective actors, but also shape the latter’s identities.

Discourse therefore is not equivalent to structure. Instead, discourse both provides a constitutive context for political articulations (leading to the question of how this context works in constituting meaningful practices) and consists of articulatory practices that re-produce but also re-shape this context (leading to the question of how the struggles in the process of this reproduction have an impact on the overall discourse). Seen from this angle, the core argument of this article can also be read as a criticism of the literature that has analysed EU foreign policy discourse from a discursive perspective for not having paid sufficient attention to the fact that discursive contexts work through the setting of limits to what is considered legitimate and practicable, and that the struggles taking place on the three levels identified above are largely focused on re-setting these limits (while themselves being governed by them).

In unfolding this argument, the article will first provide a brief and condensed review of the literature on discourse and foreign policy – especially European foreign policy – and establish the different strands within this literature, elaborating on the questions about the purpose of discourse analysis and the function of discourse. It will then zoom in on the argument about the delimiting effects of discourse, before using the debate about normative power Europe as an illustration of its argument.

Discourse and foreign policy

Discourse analysis, if it is more than simply a methodological tool, has its roots in poststructuralism (see also Aydın-Düzgit, 2013, and engaging with Critical discourse analysis as an alternative). As such, it is in the first place a critical theory. Its aim is to problematise what is usually seen as given; to contest that which is uncontested; to interrogate the familiar. Yet when it comes to the analysis of foreign policy, discourse analysis has been used not only as a critical theory; it has also been used in order to explain, or at least in order to better understand, how certain foreign policies have come about. The aims of critique and explanation do not need to be mutually exclusive, and in fact critical discourse analysts have implicitly or explicitly included an explanatory element in their analysis. Some have even gone as far as to outline a discursive-analytical framework (e.g. Wæver, 2002). Yet it seems to me that in these cases, the concept of explanation needs to be spelled out and qualified. Generally speaking, the thrust of the argument differs between critical and explanatory renditions of discourse analysis, broadly along the continuum between a more traditional social constructivist and a more poststructuralist approach (Christiansen et al., 1999).

For the social constructivist, discourse is important because it conveys norms and identities that shape foreign policy directly through the logic of appropriateness or through the shaping of interests that in turn shape foreign policy. This makes discourse and the norms contained within discourse an independent variable explaining – at least in part – (foreign) policy outcomes (Schmidt, 2010). Actors who advocate norms play an important role in this strand of the literature. These may be individual actors or groups who act as ‘norm entrepreneurs’ and push a particular discourse, or sets of actors such as advocacy coalitions who promote norms (Finnemore and Sikkink, 1998; Klotz, 2002).

There is a curious resemblance between such social constructivist work and some of the studies undertaken in a discourse-analytical framework in line with the Essex School, following the work of Ernesto Laclau. While theoretically moving away from subjects to discursively framed subject positions from which articulations are performed, scholars in this tradition have used the concept of ‘discursive coalitions’ to capture the struggles between different social groups (or, in this article’s parlance, discursive positions). In relation to EU foreign policy, James Rogers (2009) for instance has used this idea to show how the EU as a ‘global power’ is winning over the concept of the EU as a ‘civilian’ or ‘normative’ power. He links this change to the advocacy of ‘euro-strategists … actively pushing for a greater European power role’ (Rogers, 2009: 854), and his ‘discourse coalitions’ consist of a variety of actors pushing for either one of the two conceptions of the EU’s foreign policy role. Thus, while coming to the analysis from a critical perspective in order to problematise the present, the logic of the argumentation that Rogers pursues is not that different from the one that is found in more social constructivist, explanatory work.

This is in line with the general research programme of the Essex School as set out by David Howarth et al. While on the one hand seeing discourse analysis as an approach that ‘investigates the way in which social practices articulate and contest the discourses that constitute social reality’ (Howarth and Stavrakakis, 2000: 3), on the other hand the authors do attempt to explain political outcomes – although, given the inseparability of analyst and discourse, this can only be done through ‘a weakening of the once sacrosanct distinction between objective scientific explanations and subjective hermeneutical descriptions and understandings’ (Howarth and Stavrakakis, 2000: 6).

While for the Essex School, the discursive struggles between social actors – operating on what this article has identified as the second level of discursive struggles – are at the heart of the explanation as well as the problematisation of policies, Ole Wæver and colleagues have formulated a discourse theory of foreign policy that is rooted in a sedimented discursive structure. Wæver (1998, 2002; see also Larsen, 1997) is not so much interested in EU foreign policy as in the European policies of member states, but nonetheless his work warrants a closer look. His main idea is that national discourses are built on a small number of core concepts, core among which are ‘nation’ and ‘state’, with limited variation in their expressions. This leads to the argument that within a national discourse, the meanings of ‘nation’ and ‘state’ are fixed in such a way that they make only certain articulations of Europe meaningful and legitimate and thus can be seen as explaining traditions of European policies, for instance in the comparison between France, Germany and the United Kingdom. In contrast to his perhaps more popular conception of securitisation as a speech act (e.g. Wæver, 1995), in which the speaker can actively pursue the representation of something or someone as a security issue, when it comes to the analysis of European policy, Wæver thus operates without recourse to specific actors; his explanation relies on the structure that organises the discourse, or what this article has identified as the third level of discursive struggles.While he acknowledges – along the lines of the argument in this article – that discourses ‘delimit what can be said and what not’ (Wæver, 2002: 29; see also Wæver, 1998: 108), he nonetheless pursues an analysis focused on the structural influence of discourse on policy where meaning is drawn positively by one layer of discourse from another.

In the present context, my article is not interested in a general critique of these variants of the explanatory use of discourse analysis. Instead, the focus of this article is on two characteristics that all of them share. First, they see their work as explanatory, but explanation in this context must be seen as constitutive rather than causal. Second, the constitutive effects of discourse in these works are positive in the sense that they provide substantive meaning that enables particular policies.

As far as the status of explanation is concerned, the Copenhagen variant of discourse analysis seems to have the biggest problems. Particularly problematic is that the structure of the discourse is derived from the policy articulations that the structure is later supposed to explain. In other words, the model is tautological, if not in theory then at least in its methodological consequences. The fact that articulations are the effect of national discursive structures disregards transnational discourses as much as discursive struggles in the production of meaning, which are at the heart of the Essex School approach. Wæver’s tree model of ‘explaining Europe by decoding discourses’ (Wæver, 1998) therefore relies on a widened notion of explanation in line with Critical Realism (see Kurki, 2010), and unduly privileges structure over agency.

The other approaches focus on agency and process. Yet their ‘explanation’ either falls short of being satisfying, in that the conditions under which norm entrepreneurs or advocacy coalitions are able to successfully push their norms must remain vague, or, as in the case of Rogers (2009), simply trace the process of discursive change and do not actually explain why the process came about.

These problems are hardly surprising. To reiterate, discourse has two sides; it has structural qualities that inform articulations, and it relies on articulations to reproduce its structures in constant struggles over meaning which can only be temporarily fixed (Laclau and Mouffe, 1985). As argued above, Wæver ignores the latter, Rogers the former. Furthermore, the discursive contexts in which an articulation is made are multiple, and the articulation of a core concept such as ‘Europe’ in turn is an attempt to negotiate and relate these different contexts to each other and thus stabilise their meaning (Diez, 2001). Causal inferences can therefore not be drawn from articulation to structure, as ultimately they both depend on each other; thus the tautology noted above. And a causal linkage cannot be presumed between discursive contexts and the success of norm entrepreneurs, as this ignores the constant struggles over the production of meaning. The relationship between discourse, if by discourse is meant the context in which an articulation occurs, and policy articulation is therefore at best a mutually constitutive one. It is thus appropriate to say that discourses inform articulations, and that articulations reproduce discourses (something I have previously referred to as ‘linguistic structurationism’; Diez, 1999: 603); but this is not a causal relationship in the positivist sense. From an ontological point of view, this also means that articulations cannot be reduced to the purposive intentions of sovereign actors. While they are part of the pursuit of broader strategies and positions, and of the struggles between them, they are not ‘owned’ by a subject, which instead needs to be de-centred in the sense of being located in the context of a multitude of discourses, without ever reducing the subject to these discourses (Doty, 1997; Herschinger, 2011: 25; Howarth, 2000: 109). Consequently, discourse theorists have reconceptualised ‘strategy’ to reflect the degree to which strategy is not flowing out of a ‘self-conscious and self-transparent subject’ (Herschinger, 2011: 45) but is part of a broader setting of discursive power relations (e.g. Herschinger and Nonhoff, 2008).

Yet by focusing on discursive struggles, the critical purpose of discourse analysis is brought back into focus. It is thus, in the classic sense of a Foucauldian genealogy, the main objective of the discourse analyst not to explain EU foreign policy, but to show how central concepts used in EU foreign policy are actually contested, how the norms reinscribed through foreign policy are the effect of hegemonic practices and how foreign policy itself is a practice that takes part in discursive struggles, in particular those over identity (Campbell, 1998 [1992]).)Such an approach may well help a better understanding to be gained of how and on which basis specific policies came to be adopted, but it does not resemble an explanation in the positivist meaning of the term.

The explanatory logic shared by the approaches covered so far brings out the second problem identified above. Because they see policy as a consequence of norms contained in a discursive structure or promoted by norm entrepreneurs or discourse coalitions, they draw a positive line between norms/discourse and policy. In the following section, this problem is addressed and it is suggested that such a conception ought to be at least supplemented with an understanding of discourse as delimiting.

Discourse as delimitation

To recognise the importance of discourse as a delimiting force, it is useful to reconsider what I have called the ‘Derridarean move’ in discourse analysis (Diez, 1999: 606). This move implies two steps: first, that meaning is produced through difference; second, that meaning production, while not being entirely volatile, takes place in a fluid process.

On the issue of constructing meaning through difference, the argument follows Saussure in stating that words do not carry any inherent meaning, but gain their meaning through being differentiated from other words. This basic insight has been used in foreign policy analysis to draw attention to the discursive co-construction of ‘anarchy’ and ‘sovereignty’ (Ashley, 1988), and in particular the construction of identity through foreign policy (Campbell, 1998 [1992]). In these works, foreign policy does not start from a given subject of the state, but rather (re)constructs the identity of the state through representing something or someone else as ‘foreign’ and thereby also setting out what counts as ‘not foreign’ – or, in other words, setting out the attributes of the ‘self’.

In this conception, the individual act of foreign policy becomes an instance of the articulation of the identities of self and other. Such an articulation is of course set in a broader discursive context so that in order to be meaningful itself, it needs to draw on the discursive tropes in concrete historical and societal circumstances. This is what the explanatory approaches put at the centre of their analysis. In fact, Saussure’s conception also privileged such contexts as structures to explain why national languages work – they do so because every word has its ‘proper’ place in this structure of differences. Yet the fact that people often do not understand each other shows that the differences that produce meanings are not fixed but change over time, and they do so through the practices of articulation that constantly establish new meaning, not least because they add new contexts to the already existing ones.

The effect is one of the constant presence of continuities and often only marginal and incremental changes. These changes tend to become visible only retrospectively, when they have taken on sufficient weight, and they tend to be contested. This contest is the struggle over meaning that is at the heart of politics for many discourse analysts (Connolly, 1983; Laclau and Mouffe, 1985). In particular, social and political core concepts – and it is these that are relevant to the present discussion – tend to be ‘essentially contested’ exactly because they take on the function of a nodal point, drawing different meanings together and therefore stabilising broader conceptions of society (Diez, 2001). As a consequence, it is difficult to establish a hegemonic understanding so that one can indeed consider their meaning to be fixed – as Laclau and Mouffe (1985) have pointed out, such a fix can only ever be temporal, until new contestations come to the fore and gain importance.

In this process of meaning production, discourse therefore works as delimiting on two levels. On a micro level, discourse provides meaning through the opening of differences, i.e. by drawing boundaries between two concepts. Thus, meaning production through the use of metaphors, as explored by Caterina Carta in this issue, only works if the difference relationships of the metaphor are always present in the articulation of the metaphor itself. On a macro level, the possible changes to the construction of a particular concept are limited by discourse, in particular if changes are articulated in a period of relative fixation. In this latter sense, then, discourse as a context sets the boundaries of what can be legitimately articulated, but in contrast to the ‘positive’ impact on the content of concepts, the focus here is not on tracing discourse to core statements, but rather on the limits of what can be said. In turn, it is only through this delimitation – through the marking of a boundary beyond which a statement becomes unacceptable (or even unrecognisable) – that the core concept at the heart of the statement acquires meaning.

The consequences for this theorisation of the workings of discourse are twofold. First, rather than focusing solely on the substantive meanings generated in discourse and informing policies, consideration ought to be taken of the ways in which discourse inscribes the boundaries of what can be articulated. Second, this delimitation of what can be articulated itself needs to be constantly re-articulated, which brings into focus the struggles over setting the limits to discourse. It is only through such delimitation that a substantive fixation of meaning becomes possible. Thus, even if it is assumed that there are core meanings in society that seem fixed, their stability relies on the constant rearticulation of their boundaries.

Following the lead of poststructuralist work on the construction of political identities, foreign policy is at the core of the delimitation of national identities or their equivalents. If foreign policy discourses are studied, it therefore ought to be asked: how does foreign policy set the limits of national – or, for that matter, EU – identity? This, incidentally, is not how this question is commonly phrased, as even in Campbell (1998 [1992]) and others, foreign policy leads to a substantive definition of national identity rather than to its delimitation. Furthermore, how foreign policy statements are themselves governed by limitations, how their borders are constantly probed, how this probing leads to contestation and how, through this contestation, the old limits are both reproduced and transformed needs to be analysed.

In such an approach, the critical purpose of discourse analysis finds its expression in the focus on discursive struggles and the consequent interrogation of seemingly accepted meanings when it comes to its normative implications. However, the critical engagement with foreign policy discourses is in itself ambivalent. On the one hand, focusing attention on contestations and therefore on marginalised discourses may serve to bring these discourses ‘back in from the cold’ and loosen the fixations that would otherwise dominate politics and societies. On the other hand, however, discursive constructions are not better from a normative point of view simply because they are marginalised. In fact, limitations may be a good thing if they prevent harm from being done and if it is accepted that ‘all human beings have a prima facie moral obligation not to harm each other’ (Linklater, 2006: 343).

The central ethical question implied in the analysis of discursive delimitations is therefore not per se about the expansion of discursive space. Rather, those who probe the boundaries because discursive limits have been drawn too narrowly or unjustly ought to be supported, and those who probe the boundaries in order to do harm to others ought to be fought. From a critical perspective, however, the latter requires the constant interrogation of the boundaries that one is seeking to defend so that they are not simply accepted because they seem to be fixed and one falls into the trap of self-righteousness. And in the end, there is no clear-cut objective way of deciding which boundaries ought to be defended – this is a problem that throws one back onto the issue of delimiting discourses. The ‘harm principle’ may constitute such a limit, but apart from setting up an ethos with which to approach this question, its own boundaries are themselves contested. Recognising this problem, however, does not need to lead to resignation. From the argument outlined above, there follows an obligation to interrogate the ways in which discursive articulations set limits and to normatively evaluate the probing of these limits. That the norms on which these evaluations are based may not be universally accepted does not mean that such an evaluation cannot be performed (see also the discussion in Linklater, 2005), as long as at least two requirements are acknowledged: that normative assumptions are constantly questioned, and that the limits set by the rather basic harm principle (or rather the understanding of it) are focused upon.

On this basis, I will illustrate the argument by looking at a particular instance of ‘speaking Europe abroad’, the normative power Europe discourse.

The Normative Power Europe discourse

The concept of normative power Europe was coined by Ian Manners in a prize-winning article published in 2002. The article can in itself be seen as the rearticulation of the international role of the European Union, reshaping two existing discursive strands. The first emerged in the early 1970s and saw the then European Community (EC) as a ‘civilian power’, a concept which gained its content through a differentiation of the EC from the then superpowers of the United States and the Soviet Union, against which the EC was marked by ‘amilitary values’ (Duchêne, 1973: 19). The second relevant discourse was that of the ‘actorness’ of the EU, which centred on the question of whether the EU constituted a foreign policy actor in its own right in light of the intergovernmental structure of the Common Foreign and Security Policy and the fact that member states continue to pursue their own foreign policy (e.g. Manners and Whitman, 1998).

Manners’ argument was that the international role of the EU was characterised not only by its actorness, but also its mere presence, and that its distinctiveness lay not merely in its use of civilian means to pursue its interests in foreign policy, but also in that its main foreign policy tool was the shaping of what counts as ‘normal’ in world politics (Manners, 2002: 239). While it may be thought that Manners’ distinction between civilian and normative power does not fully recognise the normative element in Duchêne’s writing (see Diez, 2005: 617–618, but also Diez and Manners, 2007), the point here is that his article has successfully reconstructed the meaning of EU foreign policy and has set a new point of reference. In this point of reference, the EU pursues the spread of its norms even if these contradict economic interests; it works towards a civilisation of international politics; and it binds itself to international norms, especially the UN Charter; yet it may well use military force if this is democratically legitimised and necessary to enforce norms.

The debate that ensued has largely focused on the question of whether the EU, in its foreign policy, does indeed behave according to the norms described by Manners as being central to its existence as a normative power (see the overview in Whitman, 2011). In what follows, this article does not take an interest in this particular debate. For one, it seems pretty clear that while the EU does operate in world politics in a way that is different from traditional great powers, at the same time it is fair to say that its own norms are often violated by actors within the EU, be it in regard to arms trade exports (Erickson, 2011) or the disregard of human rights when it comes to strong economic interests (Balducci, 2010; Youngs, 2004). Alternatively, EU norms and interests often cannot be empirically distinguished, as they go hand in hand, for instance in the promotion of particular environmental standards (Falkner, 2007). As a consequence, many have suggested alternative conceptions of, for instance, realist (Zimmermann, 2007) or pragmatic (Wood, 2011) power Europe, to emphasise that EU foreign policy is not a consistent consequence of constitutionalised norms.

This debate is not core to the argument of this article. Yet it raises the question of why Manners’ article has had such an impact, when its arguments are rather problematic if empirically scrutinised. Following the argumentative line of this article, my hunch is that the piece ought to be read first and foremost as a foreign policy articulation that was able to draw links to existing discourses about the international role of the EU and that plays a role in the crafting of an EU identity. Second, and more importantly for the purposes of this article, the debate seems to miss the ambivalent epistemological status of Manners’ piece. On closer inspection, Manners uses the normative power argument in three different ways: as a description of what the EU is (this is what the bulk of the literature has taken up), as a causal argument in the sense that the EU pursues a particular foreign policy because of internally constitutionalised norms (following the social constructivist understanding of the role of norms and discourse) and as a normative argument about what the EU ought to be. Thus, Manners himself notes that his argument has an ‘ontological’, a ‘positivist’ and a ‘normative’ quality, although he uses the second one in a slightly different sense than that used here (Manners, 2002: 252).

The explanatory function of Manners’ argument is in line with the discursive explanations of foreign policy noted above: there are, he claims, central norms inherent to the EU constitution and identity, and these do cause the EU to act in particular ways. The interest in this section, however, is with the apparent tension between the other two aspects of the argument; namely, that the EU at the same time is and ought to be a normative power. Such an argument only makes sense if one recognises that normative power is in fact an ideal which is on the one hand partly realised, yet on the other constitutes a normative horizon which one needs to strive for but which may never be fully achieved. In fact, Manners operates here in a way not unlike the Habermasian discourse ethics, in which the ideal speech situation, freed of domination, serves both as an ideal and as the factual yet counterfactual assumption that is made whenever serious conversations are engaged in (Habermas, 2005: 118). Thus, Manners sets out the normative aim of normative power Europe on the basis of arguing that there is in fact a common understanding that this is the nature of the EU. Yet at the same time, he only constructs this nature, and he does so by opening up the classic difference with the United States by illustrating his argument with the engagement of the EU in favour of the abolition of the death penalty, and by pointing to the disputes with the US on this particular issue (Manners, 2002: 247–248).

How does the concept of normative power now work in the discourse? The plethora of works trying to link EU policies to particular norms show that one line of argument tries to draw positive linkages between norms and policies. Yet my claim in this article is that such an undertaking only makes sense through the probing of the boundaries of the norms implied, as well as of what constitutes a normative power. Thus, the questions of whether normative power excludes the possibilities of military engagement, and if not, under which conditions military force can be legitimately used (Sjursen, 2006; Wagner, 2006) or whether the simultaneous pursuit of interests contradicts normative power (Falkner, 2007; Youngs, 2004), are questions that engage with the limitations of normative power to set out what kind of political decisions are defensible and which ones are not. Likewise, the pursuit of competing liberalisms in EU foreign policy (see Rosamond, 2013) is a struggle not only about the pursuit of different basic worldviews. It is also a struggle about the boundaries of the EU as a promoter of liberalism: where are the limits of market forces, where are the limits of interventions in the name of peace, and where are the limits of cosmopolitan duties, to follow Ben Rosamond’s threefold distinction (Rosamond, 2013). Yet such delimitations are not drawn on the basis of an already existing core of normative power; it is rather that both these academic works and the ‘liberal’ EU policies are to be seen as articulations of the boundaries of normative power by drawing distinctions between that which is acceptable and that which is not.

The importance of the normative power Europe discourse thus lies as much in advocating the pursuit of particular norms as it does in providing the basis on which the boundaries of an EU foreign policy discourse can be constructed. The European Parliament’s decision not to extend the fisheries agreement with Morocco in December 2011 is a case in point. For a long time, the agreement was a prime example of legal inconsistency, as it allowed EU fishermen to utilise the fishing grounds in front of the coastline of the Western Sahara, while at the same time the EU does not recognise the authority of Morocco over the Western Sahara. Thus, the fisheries agreement could be used as an example that the EU is not acting as a normative power if its interests are at stake. Yet the European Parliament decision makes explicit reference to the problems of overfishing as well as to the self-determination rights of the people of the Western Sahara, and so engages in the normative power discourse to determine that a policy transgressed the boundaries of this discourse. It is therefore setting the limit of legitimate EU foreign policy.

It is in such declarations of a policy having crossed the boundary that the boundaries of what can legitimately be pursued are articulated and that discourse works as a delimitation. Very similar observations can be made in the debate over whether Germany is a civilian power or not. Again, the actual content of the concept is contested, as well as the empirical evidence (consider Maull, 2000 and Risse, 2004 against Hellmann, 2002). While the argument that the idea of civilian power informs German foreign policy can clearly be challenged on the evidence ranging from breaches of multilateralism in the recognition of Croatia and Slovenia and resistance to the supranationalisation of EU asylum policy to the role of Germany in the global arms trade, it does make more sense once one considers how the idea is utilised to delegitimise particular foreign policy articulations – for instance, in the pressure put on former President Horst Köhler when he considered a role for the army to secure trade paths (he had to resign) or the immense public debate about the role of the German army in Afghanistan.

These articulations can be seen as examples of the effect of discourse not as a cause of, but as a barrier to policy. The discursive context from which these articulations are made on the one hand allows them to be made, while on the other they reinscribe the borders of the discourse. The examples also demonstrate that the discursive struggle over the EU’s global role does not, or at least does not only, take place between member states as actors, but involves different EU institutions as well as debates within societies in member states. Finally, in the normative power Europe case as much as in the civilian power case, the discourse is about how to limit harm in international relations. The reinstatement of the borders through these articulations therefore is not normatively problematic, but is in fact warranted in a critical exercise not so much to problematise these borders, but to problematise the grounds on which other articulations try to remove them. Reading the normative power Europe argument from this angle therefore rescues its normative component, which is lost if NPE is used solely as an empirical category.

Conclusions – setting the limits

I have argued that discourse analysis has gained ground in the analysis of foreign policy, but the way that it has been used so far puts too much focus on using discourse as a variable to explain foreign policy by drawing positive linkages between core concepts and the policy that followed. In contrast, this article has focused both on the critical purpose of discourse analysis and on the function of discourse in setting the limits of legitimate and meaningful foreign policy. I have argued that the critical purpose and the delimiting function come together in interrogating the boundaries of discourse so as to point to marginalised positions and, more relevant for this article, in interrogating attempts to move the boundaries of discourse beyond the limitation set by the principle of not doing harm to others.

The debate about normative power Europe can be used to illustrate the argument and to provide pointers for what future empirical analysis of EU foreign policy following this line of argument may look like. The claim is that the normative power discourse in the EU, as much as the civilian power discourse in Germany, sets the limits of legitimate foreign policy. However, as these illustrations have shown and as theoretical considerations have argued, these limits need to be constantly re-articulated as part of a struggle about the borders of the discourse. In other words, setting limits does not happen by structure as such, but through the enactment of the limits through a variety of actors in civil society, politics, the media, the arts and, not least, academia – these operate within their own circles (or what Jørgensen, 2013, calls ‘levels’), yet (as Jørgensen also recognises) they are not entirely separate from each other, but engage in transversal debates and struggles.

To treat the normative power Europe discourse therefore as constituting the identity of the EU on which EU foreign policies are based is problematic and misses a crucial facet of the contribution of discourses. It is problematic because it overrates the stability of the discourse and underrates the contestedness of normative power as a core concept. It misses a key facet in overlooking the struggles to inscribe boundaries to what is legitimate and possible as an EU foreign policy. Rather than discussing the extent to which the EU is or is not a normative power, it would be interesting for further research to show how Manners’ original article and the debate that ensued tried to draw such boundaries. The short overview of the normative power Europe debate in this article has indicated that such a struggle over boundaries takes place along at least two lines of contention: the delineation of a policy space and the degree to which this space includes interest- and traditional power-based arguments, and the delineation of the normative space to which EU foreign policy can make reference.

The analysis of foreign policy discourses should therefore be refocused away from their positive shaping of particular policies and towards the way in which they work to set the limits of policy and how this setting of limits is performed by actors engaging in a struggle over foreign policy. Such a refocusing would, I believe, strengthen understanding of the workings of discourse, but would also strengthen critical engagement with foreign policy.