Political theory’s deliberative turn in the 1980s and 1990s has provided a novel set of theoretical insights into the epistemic or communicatively rational dimension of communication. By the latter, I refer to the practice of giving and taking reasons and the capacity of the said reasons to influence the political preferences and behaviours of social actors. The theoretical arguments advanced by the deliberative turn rely on a family of concepts which, depending on the authors considered, have been variously interpreted. This is the case, for instance, of the concept of discourse, construed as synonymous with deliberation, as providing a kind of “Weberian ideal type” with which to assess “how far away a particular speech situation is from the ideal type” (Steiner, 2008: 188-189) – whilst other scholars understand it “not [as] a normative model against which we judge the adequacy of existing arrangements”, but as a counterfactual ideal assumed by participants in communication, which somehow influences communicative interaction (Eder, 2007: 399). The goal of this research note is to clarify how these theoretical arguments and concepts have informed empirical research. Thus, the emphasis of the argument will be laid on the research models behind empirical studies, that is, the recurrent theoretical ideals and methodological perspectives shared by a number of studies. More precisely, I will contend that there are at least two such models, which are the result of (1) how the concept of discourse or deliberation is interpreted and (2) which methodological perspective –co-variational or narrative– is adopted. The standard, political theory approach sees deliberation as a regulative ideal with which to assess actual political communication and empirically identify cases of deliberation. Furthermore, it is interested in how different variables are regularly associated with one another. This approach is probably best illustrated by the work of Steiner, Bächtiger and colleagues (e.g. Steiner et al., 2004). In contrast, the narrative, social theory model understands discourse as a counterfactual ideal in the heads of participants in communication. Moreover, it pays greater attention to the pathways that connect different events and variables to each other in time. Its main topics of research are collective learning processes and processes of ideational change in general. Brunkhorst (2014) provides a recent example of this approach.

This research note therefore seeks to make a contribution to conceptual clarification as regards the empirical research of communicative rationality. The two models reconstructed here partly overlap, thus there is no neat distinction between them. Yet, the fact remains that they follow different logics and pull in different directions. Dealing with the tensions arising from this requires, first, that we acknowledge the existence of these models and understand their underlying logics and implications. For one thing, the range of empirical phenomena linked to the rational dimension of communication varies according to the model considered –the political theory approach admits less variation in what communicative rationality empirically means than the social theory perspective. Furthermore, the distinction between these models challenges a common criticism against the broad conception of communicative rationality of the social theory perspective, namely, that of concept stretching, for this is a criticism that holds within the theoretical parameters of the political theory approach, but is less compelling within the social theory framework. There are also differences concerning the normative impetus of these two models, more visible in the political theory approach than in the social theory perspective; as well as regarding the explanatory logic favoured by each of these models –the standard political theory model seeks to isolate the constant effects of certain variables, whereas the narrative, social theory model pays greater attention to path dependency.

The argument will be structured as follows. First, I will briefly present two methodological approaches, namely, the variable-centred or co-variational one and the narrative or sequential approach. Subsequently, I will introduce two interpretations of the concept of discourse: the political theory reading and the social theory reading. I will then contend that the intersection of these two methodological approaches and viewpoints of discourse leads to the above-mentioned research models: the standard, political theory perspective and the narrative, social theory model. The next two sections will be devoted to discussing these research models, and the last section will summarise the argument made in this research note.

Different accounts of what constitutes a satisfactory “explanation” in the social sciences coexist –whereas some authors privilege co-variational approaches, others prefer explanatory templates based on social mechanisms, or they base their studies on thick description, among other possibilities. For the sake of simplicity, let us call these alternative perspectives methodological approaches. Methodological approaches rely on different sets of assumptions about how the social world works and what our epistemological limits and possibilities are, among other things (Della Porta, 2010). Furthermore, they tend to be mutually exclusive, for the assumptions behind these different methodological perspectives often conflict.

Given the scope of this paper, I would like to mention here just two competing methodological perspectives, namely, the co-variational or variable-centred approach (Blatter and Blume, 2008) and the narrative or sequential approach (Abbott, 2001). The first perspective is the dominant one in comparative analysis. Essentially, it seeks to define discrete variables and identify whether there is some correlation between them. As such, the variable-centred approach concentrates on the standardised effects of some independent variables over other dependent variables. This perspective is the one that researchers adopt when they seek to pinpoint the “institutional determinants of deliberative interaction”, tout court (Landwehr and Holzinger, 2010), or when they explore “whether democratic deliberation”, in general, “enhances ‘civic virtues’” (Grönlund et al., 2010). Co-variational templates assume what Abbott (2001: 44) calls “constant relevance,” namely, that “a given cause is equally relevant at all times.” Certainly, this variable-centred approach allows for the interaction between variables as well as for context-specific effects, but once other variables are controlled for and the context is specified, the assumption remains that the effects of a given (independent) variable are constant (in the said context).

This contrasts with the assumptions made by the narrative or sequential approach, according to which time as well as the order of events matter as much as, or even more than, the standardised effects of a given set of independent variables. A case in point is Fukuyama’s (2014: 30) dictum that “sequencing [...] matters enormously,” for “high-quality governance” is much more difficult in countries where “democracy preceded modern state building” than in those where modern state building preceded democracy. In this view, action is conceived of as unfolding in time, with previous events conditioning subsequent ones, that is, opening up new paths and possibilities and foreclosing others. As such, social life is taken to have a narrative structure, where events are connected to each other by “casual emplotment” (Somers, 1994: 616), that is, by “causal pathways” that link different events to each other in time (Somers, 1998: 771).

These two methodological perspectives together with the specific understanding of the concept of discourse, to which I turn now, lead to two different research models of deliberation.

Two different understandings of the concept of discourse are also at the heart of the two research models of deliberation reconstructed in this paper. Let us call them the political theory and the social theory readings of discourse (Deitelhoff and Müller, 2005: 177). According to the former, discourse (or deliberation) is a regulative ideal against which political communication should be assessed. According to the latter, discourse is a counterfactual ideal that inevitably permeates communication and thus contributes to shaping it. This has implications for empirical research: the political theory approach uses the concept of discourse as a standard with which to measure the deliberative quality of communication or identify cases of genuine deliberation (typically, Bächtiger et al., 2005; Steenbergen et al., 2003). In contrast, the social theory perspective seeks to account for certain features of communication by drawing on its discursive dimension, that is, the inevitable raising of validity claims in communication (e.g. Brunkhorst, 2014; Eder, 1991).

The distinctive feature of the political theory interpretation of discourse is that it is conceived of as a regulative ideal, or “an ideal to which, all else equal, a practice should be judged as approaching more or less closely” (Mansbridge et al., 2010: 65). There is, however, disagreement among political theorists as to what exactly this regulative ideal looks like and how it can be justified theoretically. Bächtiger et al. (2010), for instance, distinguish between two broad conceptions of deliberation in the scholarly literature, which they simply label type I and type II deliberation. Type I deliberation refers to “a systematic process wherein actors tell the truth, justify their positions extensively, and are willing to yield to the force of the better argument” in order “to reach understanding, or consensus” (Bächtiger et al., 2010: 33). In contrast, type II deliberation “involves more flexible forms of discourse, more emphasis on outcomes versus process, and more attention to overcoming ‘real world’ constraints on realizing normative ideals” (Bächtiger et al., 2010: 33).

Similarly, there is disagreement regarding the theoretical arguments used to set this normative standard. Habermas (1981, 1992) and Apel (2000), for example, take arguments from the philosophy of language, in particular from so-called universal pragmatics, which they then extend to the domain of politics and morality, whereas Gutmann and Thompson (2004: 3) ground the deliberative ideal on a “moral basis” that is “common to many conceptions of democracy”, namely, that “persons should be treated not merely as objects of legislation, as passive subjects to be ruled, but as autonomous agents who take part in the governance of their own society.” More often than not, the epistemic value of deliberation is also cited as an argument for these normative standards (Lafont, 2006). Be that as it may, what is common to the work of many political theorists, though by no means all of them (e.g. Gaus, 2013), is that they seek to establish a normative ideal that prescribes how actual deliberation should be (for recent reviews of the literature in this area, see Bächtiger et al., 2010 and Steiner, 2012).

This contrasts with the social theory concept of discourse, which can be traced back to the work of Habermas, in particular to his attempt to develop a critical social theory of society. As a critical theory, its goal is to uncover the traces of reason in a world that is taken to be, on the whole, unreasonable. In this context, the concept of discourse is thus intended to grasp the rational dimension of linguistically mediated communication. More precisely, it “captures just those pragmatic features of a communicative setting that anybody tacitly presupposes once he [sic] seriously enters an argumentation in order to check a problematic validity claim by either supporting or denying the truth or rightness of some statement with reasons pro and con” (Habermas, 2005: 385). Nonetheless, given the internal link between argumentation and linguistically mediated communication, the concept of discourse captures not only those counterfactual assumptions that people must make once they enter an argumentation, but also counterfactual assumptions shaping “everyday contexts of communicative action” (Habermas, 2005: 385). This is because there is an internal logic linking the practices of communication and deliberation. Let us unpack this.

The starting question to understand this internal link is the following: How is it possible that the actions of different social actors can be coordinated through communication? According to the theory of communicative action, there are two ways. First, communication can be used strategically. Typically, social actors can give orders and make explicit threats to make other actors comply with their orders. Social actors can also use communication to achieve perlocutionary effects –that is, they can surreptitiously manipulate other actors to behave in a specific way, as when they tell lies. Secondly, social actors can use communication to reach mutual understanding. Essentially, this second use of communication is what Habermas calls communicative action.

The crucial point is that perlocutionary effects are parasitic of this second, non-strategic use of communication (Habermas, 1981: 370). Let us simply illustrate this with the example of a lie. Lies can only be successful if the fact that one is lying remains concealed –in other words, liars have to pretend that they are using language in order to reach mutual understanding, hiding that they are trying to manipulate other actors. This, however, is not the case with explicit threats and orders, the other way of using communication strategically (Habermas, 1981: 391). Since it is strikingly obvious that only a small part of everyday communication consists of explicit threats and orders, it is safe to argue that most of everyday communication consists of communication oriented towards reaching understanding, irrespective of whether this orientation is genuine (as in the case of communicative action) or not (as when one seeks to achieve a perlocutionary effect).

In communication oriented towards reaching understanding, speakers raise validity claims, which can be accepted or not by hearers. Through this type of communication, social actors can coordinate their actions if they accept these validity claims and thus bind themselves to their content. The interesting point is that social actors can be rationally motivated to bind themselves to the content of these claims only if they “know the kinds of reasons that a speaker could provide in order to convince [them] that he [sic] is entitled in the given circumstances to claim validity for his utterance” (Habermas, 1998: 232) –in other words, when social actors know what makes a given validity claim acceptable. This means that the coordination capacity of communication oriented towards reaching understanding is internally related to the explicit or implicit practice of giving and taking reasons (Habermas, 1981: 386), for only these explicit or implicit reasons can motivate social actors to accept certain validity claims and bind themselves to them.

In this reading, therefore, communication is a rule-governed social interaction, the regulative logic of which is apprehended by the concept of discourse. What this concept provides is not primarily a regulative ideal of deliberation, but “a ‘reality’ that pervades our communication and has to be taken into account when communicative action, i.e. a speaker relating to other speakers in the social world, goes on” (Eder, 2007: 399). The implication of this argument is that deliberation, understood as the explicit or implicit giving and taking of reasons, should be seen less as a variable than as a constant in communication (Deitelhoff and Müller, 2005: 172) –a constant which, as argued by Eder, should be taken into account in order to understand everyday communication, even communication that violates the rules of discourse (as in the example of a lie). The second point is that the aim of this social theory understanding of discourse is not so much to assess communication, but to account for certain features of linguistically mediated communication.

The intersection of these two readings of discourse and the above-mentioned methodological approaches has given rise to two research models of deliberation: the standard, political theory model, and the narrative, social theory one. These models should be understood as ideal types, that is, analytical constructs. Yet, as shall become clear in this paper, these models partly overlap and can be even combined differently. The point, however, is that despite this overlap, both models emphasise different aspects, pursue different goals, and carry different implications. To clarify this, the ideal types reconstructed here are useful.

The standard, political theory model is characterised by an interest in the constant effect of certain variables, as well as in the quality of deliberation as measured against a regulative ideal. As such, research within this model typically seeks to assess the quality of deliberation, to identify its conditioning factors, and to explore its consequences.

In contrast, research within the narrative, social theory model usually pays greater attention to specific processes of ideational change (or the lack thereof) and the role of deliberation, conceived of as a counterfactual idealisation that is always already part of communication, in these processes. Typically, it strives to reconstruct, model, and explain specific processes of ideational change and collective learning that happen through the medium of communication, as well as to explore collective learning blockages and pathologies. More generally, it also seeks to assess the general utility of Habermas’s social theory insights for empirical research.

The recent systemic turn in deliberative theory (Parkinson and Mansbridge, 2012) has thus far encouraged little empirical research, notwithstanding efforts in this direction –hence, I will not expand on this perspective here. Yet there are reasons to believe that it might lead to novel ways of combining methodological approaches and conceptions of discourse; for example, Goodin (2005) adopts a political theory conception of discourse while stressing the time dimension by theorising about the “sequencing” of deliberative moments. In this regard, emerging systemic approaches illustrate a relevant point, namely, that there is no necessary theoretical relation between the variable-centred approach and the political theory reading of deliberation, nor between the narrative perspective and the social theory understanding of discourse. In sum, conceptions of discourse and methodological approaches are independent from each other and thus can be combined differently.

The standard, political theory model is the research model that lies behind most of the empirical studies carried out by scholars involved in the so-called empirical turn in deliberative theory. Again, this model is characterised by a variable-centred approach and a political theory reading of discourse, without there being any intrinsic connection between these two elements.

Probably the clearest example of the standard, political theory approach to deliberation is the work carried out by Bächtiger, Steiner and various colleagues (Bächtiger et al., 2005; Steenbergen et al., 2003; Steiner, 2012, Steiner et al., 2004), which essentially consists of measuring the deliberative quality of a debate, usually in a parliamentary setting, as well as exploring the conditioning factors and consequences of deliberation –hence, a combination of a political theory reading of discourse and a variable-centred approach. In the following, let us look at these two aspects in turn.

The notion of deliberation as a normative standard is expressed variously depending on the studies considered. In research carried out by Bächtiger and colleagues, deliberation is empirically measured through a coding scheme, namely, the Discourse Quality Index (DQI), which operationalises different attributes of rational discourse. As interpreted by Steenbergen et al. (2003: 24-30), a perfectly deliberative interaction implies civility, which means letting other speakers express themselves freely and without interruption, and showing explicit respect toward other speakers, as well as towards their claims. Furthermore, it requires speakers to be sincere, to explicitly justify their own claims in terms of the common good, and to be constructive, that is, to aim at reaching a rationally motivated consensus or at least to “attempt to reach mutually acceptable compromise solutions” (Steenbergen et al., 2003: 26). Finally, genuine deliberation implies that counterarguments are considered seriously, that is, that they are acknowledged and explicitly valued. This set of features describes in a nutshell both how deliberation as a regulative ideal is conceived of and how it is measured empirically. Leaving aside aspects such as authenticity (i.e. the fact that speakers are expected to be sincere in deliberation), which can hardly be measured, it is assumed that there should be a one-to-one correspondence between the regulative ideal of deliberation and the actual speech acts that qualify as deliberative.

Other coding schemes follow a similar rationale although they rely on normative standards that are reinterpreted and adapted to the specific object of study considered in each research. Wessler (2008), for instance, proposed a coding scheme for “investigating deliberativeness comparatively”, which was then slightly modified and applied to studying the “deliberative performance of television news” (Wessler and Rinke, 2014). This coding scheme distinguishes between the input, output, and throughput dimensions of deliberativeness, concentrating on the openness of debates, their dialogical rather than monological features, and the orientation of public debates towards decision-making. Here, a wider gap between the ideal of discourse and its empirical manifestation becomes evident. The crucial point is how this gap is theoretically justified – namely, as the outcome of the researcher adapting the regulative ideal to a new setting in the light of some contextual constraints. In other words, the regulative ideal is modified according to the peculiarities of newspapers, television news, and the like, where some form of deliberation is expected to happen, and this readapted ideal is then used to empirically identify cases of deliberation or measure the deliberative quality of communication. (The fact that it is the researcher, rather than social actors themselves, who adapts the regulative ideal to a specific setting constitutes a key difference between the political and the social theory approach to discourse, as we will see.) Hence, these studies stick to the concept of discourse as a regulative ideal although contextual constraints are considered in the application of the said ideal.

Many research projects, however, do not measure deliberation directly, but rather treat it as a black box or observe it informally. This is the case of some deliberative experiments, as well as a number of studies of deliberation in mini-publics. In said studies, enabling conditions for deliberation are provided and sometimes modified according to the researcher’s goals, while the consequences of deliberation are registered, but the actual process of giving and taking reasons is not systematically analysed (Fishkin 2005, 2012; Ganuza et al., 2012; Grönlund et al., 2010; Niemeyer, 2011; Smith, 2009; Smith and Wales, 2000). This notwithstanding, these studies are also guided by an understanding of deliberation as a regulative ideal, in the sense that what is taken to be deliberation in these projects is a form of communicative interaction whose observable features closely approach one of a number of definitions of deliberation provided by political theorists. In this regard it is telling, for example, that the very design of these experiments and mini-publics is intended to bring about the said forms of deliberative interactions that conform to the normative ideal. More generally, this same logic also informs studies on deliberation in the public sphere and cyberspace, relying on research techniques other than those previously discussed, which also assume that the observable characteristics of a communicative interaction should closely approach a given normative conception of deliberation in order to be considered a case of deliberation (Dahlberg, 2001; Dahlgren, 2005; Gerhards et al., 1998; Pilon, 2009; Rohlinger, 2007; Tsaliki, 2002). The argument for this understanding of deliberation is frequently a negative one: researchers avoid the danger of concept stretching only by considering forms of interaction the observable features of which closely approach a given normative definition of deliberation (Steiner, 2008).

Also common to the studies mentioned above is the fact that they rely on a variable-centred perspective, this is, a perspective that seeks to identify the constant effects of some variables over others. By examining the statistical associations between different variables, Bächtiger et al. (2005), for instance, explore the favourable contexts for deliberation in legislatures, as well as deliberation’s outcomes. Similarly, Wessler and Rinke (2014: 828) study, among other things, “the impact of system, organizational, and news format variables on the deliberative features of television news”, again using a standard co-variational template. The same holds for Ganuza et al. (2012) when they analyse the variables that influence participants’ change of opinion in deliberation, or for Grönlund et al. (2010) when they study deliberation’s influence over some “civic virtues”. Studies that do not rely so strongly on statistical analysis also adopt a variable-centred perspective, as when Smith (2009: 7) explores “to what degree […] different [institutional] designs realise the six institutional goods that form [his] analytical framework”, thus assuming that the effects of institutional designs are more or less stable. These examples can be multiplied although I take it to be evident to anyone familiar with the social science literature that this co-variational template is a popular, widely accepted methodological approach, so it is not necessary to dwell longer on this issue to show that it is also common among scholars analysing deliberation.

In contrast to the standard, political theory approach, the narrative, social theory perspective combines a sequential methodological approach with a social theory reading of discourse. Let us discuss these two elements in turn.

What the social theory understanding of discourse suggests, as elaborated above, is that the explicit or implicit giving and taking of reasons is a constant in everyday contexts of communicative action, which, in turn, implies that the more people communicate, the more they are exposed to the force of arguments (Eder, 2007b: 45, 49). Here, to be exposed to the force of arguments does not necessarily mean that they are actually driven by it, but this argument grounds the expectation that social actors “can learn –and in the long run even ‘cannot not learn’” (Habermas, 1999: 8). In this regard, the key concept of this approach to deliberation is that of learning and –since learning is understood as linked to communication– of collective learning.

Collective learning can be defined as a communicative process (usually a conflict-ridden one) driven by the pursuit of reaching mutual understanding and solving problems that social actors encounter performatively in coping with the world. Through it, social actors construct, challenge, and modify their common symbolic world. Of course, scholars working within this narrative, social theory perspective acknowledge the fallibility of these learning processes, as well as the relevance of factors other than the force of arguments (e.g. power relations), in shaping or even blocking them –nonetheless, they also acknowledge that there is a (communicatively) rational dimension inherent to communicative interaction, which sometimes leads to collective learning.

Being interested in collective learning processes, and processes of ideational change more generally, studies within this narrative, social theory model adopt a narrative perspective, according to which different “events” (i.e. ideas, speech acts, arguments, actions of social actors, etc.) are connected to each other in time by casual emplotment, with previous events influencing subsequent ones. Certainly, attention is also paid to constant effects and observable regularities, a case in point being the notion of discourse itself, seen as a (counterfactual) constant in communicative interaction. Yet in the last instance, the said regularities are embedded in causal narratives that depict the pathways connecting different events to each other in time.

Cases in point are varied, from analyses of long-term processes such as the rise of Enlightenment sociology (Strydom, 2000) or the evolution of the societal relationship to nature (Eder, 1996), to micro perspectives dealing with conflicts among children or conflicts over ownership of the land on remote islands in the Pacific (Miller, 2006). Other studies – probably more interesting to political scientists – include the analysis of the emergence of cosmopolitanism and cosmopolitan norms (Benhabib, 2006; Delanty, 2013); cases of normative learning in legal revolutions (Brunkhorst, 2014); the creation of the International Criminal Court (Deitelhoff, 2009); environmental discourse and environmental policy making (Eder, 1995); informal political talk and learning processes regarding the definition of social identities and the formation of political perspectives (Walsh, 2004); Germany’s path into modernity and the evolution of the German political culture during the 18^th and 19^th centuries (Eder, 1991), or the democratising dynamics of the European public sphere (Eder, 2007b; Statham and Trenz, 2015; Trenz and Eder, 2004). These are mostly case studies in which the goal is to account for the unfolding of events in time, examining turning points and how previous events paved the way for subsequent ones, whilst foreclosing alternative courses of action.

Again, this research also embraces a social theory reading of discourse, according to which discourse is a set of counterfactual presuppositions that actors make in communication although, their counterfactual nature notwithstanding, they are expected to influence actual communicative interactions. Unlike the standard, political theory approach, the social theory perspective does not necessarily see the expression of discourse in speech acts that closely approach these counterfactual assumptions or any other ideal redefined by the researcher in the light of contextual constraints. Since they are counterfactual assumptions in the heads of participants in communication, it is social actors themselves, rather than social scientists, who have to negotiate the tension between these counterfactual assumptions and other factors, contextual or otherwise, that also influence their behaviour.

Leaving aside these theoretical differences, the question remains as to what the observable, empirical manifestations of discourse are. Different answers have been provided to this question. According to the social theory perspective, the typical empirical phenomena that can be traced back to discourse –and which are thus expressions of communication’s rational dimension– are more varied than imagined by the political theory perspective.

First, it has been contended that discourse might give rise to “islands of persuasion” (Deitelhoff, 2009), this is, it might shape communicative interactions in such a way that they closely approximate the ideal speech situation. This is what the standard, political theory perspective would also call deliberation. Islands of persuasion can be specific, temporary events, which can nevertheless decisively shape subsequent events by helping to disseminate certain ideas and discrediting others.

Secondly, the empirical manifestation of discourse has also been presented as consisting of diffuse and long-term processes of opinion-formation, involving the constant reinterpretation, discussion, and updating of prior opinions and attitudes in the face of events and new pieces of information (Benhabib, 2006; Habermas, 2006: 420; Peters, 2007: 202). Along these lines, Habermas (2006: 420) contends that

To the extent that this constant updating of opinions happens through the medium of communication, they involve the exchange of explicit or implicit reasons. Again, this is because communication inevitably raises validity claims, which, in turn, can only be accepted (or not) against the background of the reasons that are actually adduced, or could be adduced, in support of them. It might be rightly argued that these diffuse and long-term processes of opinion formation are not as sophisticated epistemically as those cases in which actual communication approaches the ideal speech situation. Yet the occasional self-correcting (Habermas, 2005b: 51, 84) or self-repairing (Eder, 2014) capacity of communication in the light of the counterfactual standards set by the concept of discourse (as when cases of exclusion from public communication are denounced and rectified), as well as according to the force of the specific arguments put forward in every case, attest to the epistemic dimension of these diffuse processes of opinion formation. This epistemic dimension might not be as sophisticated as in other cases, but in this view there is an epistemic dimension nonetheless.

Thirdly, argumentative self-entrapments have also been traced back to the notion of discourse (Risse, 2000). Self-entrapments occur when social actors behaving strategically have to accept validity claims contrary to their interests in order not to appear inconsistent and discredit themselves –that is, in order to continue pretending that they are acting according to the rules of discourse (when they are not). Although there is a difference between genuinely behaving according to the rules of discourse and simply pretending to do so, the point is that argumentative self-entrapments lead, at least superficially, to changes in an actor’s discourse, which can constitute a crucial turning point in a wider process of opinion formation (Deitelhoff, 2009, shows this for the case of the creation of the International Criminal Court). On these occasions, it is consistency and the force of arguments that carries the day, even if social actors are not really moved by the force of arguments, but by the desire to save face.

Finally, the empirical manifestation of discourse can simply consist of the problematisation of a state of affairs, namely, in rejecting validity claims used to justify the status quo, interrupting routines, “mobiliz[ing] latent reasons,” and potentially destabilising certain institutions, practices, decisions, and so on (Fœssel and Habermas, 2015; see also Brunkhorst, 2014, on “negativity”). The problematisation of a situation constitutes an expression of the rational dimension of communication to the extent that it results from the critical examination of the validity claims underpinning the existing state of affairs.

As such, the narrative, social theory perspective does not only depart from the standard, political theory approach vis-à-vis its methodological perspective and its theoretical understanding of discourse, but also regarding the phenomena that it conceives as empirical manifestations of discourse and thus of the rational dimension of communication. Granted, such a broad understanding of discourse can be criticised from a political theory perspective for overstretching the concept of deliberation. Yet it should be noted that the opposite also holds –the social theory perspective can accuse the political theory one of oversimplifying and limiting the meaning of discourse.

This research note has sought to contribute to conceptual clarification as regards the empirical research of communicative rationality. It has been argued that there are at least two research models that inform empirical research on this topic, which embrace different methodological perspectives and understandings of discourse. The standard, political theory approach regards deliberation as a regulative ideal with which to assess actual political communication and identify cases of deliberation. Furthermore, it adopts a co-variational explanatory template. The narrative, social theory model sees discourse as a counterfactual ideal in the heads of participants in communication and assumes a sequential methodological perspective. These models imply different theoretical choices with different implications for empirical research. One can regard them, however, as implying middle-range theoretical decisions, in the sense that they do not demand a continued and exclusive commitment to them. Indeed, Habermas’s work shows that it is perfectly possible to move between these two approaches without this being necessarily inconsistent from a theoretical perspective. Yet the fact remains that these two models pursue different goals, emphasise different aspects, and entail theoretical decisions with different implications. Although the two approaches can be reconciled in principle, researchers are well advised to consider them explicitly in order to successfully deal with them.