Discourse analysis as critique

This paper intervenes in the discussion about the relationship between discourse analysis and critique. It argues that this relationship can be understood either as an external or as an integrated relationship. In an external relationship, there is first social criticism that is then braced by discourse analysis, that is, the latter aims at giving empirical credence to the critique. However, such an external relationship cannot give us any insight concerning the critical potential that is specific to discourse analysis, precisely because in this case critique exists before and independent of discourse analysis. If, however, critique emanates from discourse analysis itself, we would speak of an integrated relationship and would no longer speak of discourse analysis and critique, but of discourse analysis as critique. It is argued that such an integrated relationship becomes visible once we think of discourse analysis as being itself a discursive formation and ask what unsettling effects this formation has on research objects, on subject formations and on the academic production context in which they are conducted.

Introduction D iscourse analysis and critique. "Really? Again?", you could ask. Didn't already the eminent authority of discourse analysis, Michel Foucault himself, sketch out in a widely read talk how we are to understand critique? Don't we know since then that critique equals the art "of not being governed like that and at that cost" (Foucault, 1997: 45)? Isn't there since the 1980s a broad movement of Critical Discourse Analysis/Critical Discourse Studies (CDA/CDS)? Isn't it nowadays somehow common knowledge that discourse analysis is critical or proceeds critically? Are we able to say something new in the face of already well established debates? In this article, I will try to come up with a nuanced response to questions like these. Yes, certainly, there are several approaches of CDA/CDS since the 1980s; yes, we have read Foucault on critique, as well as many others, not least among them Critical Theory. And yes, particularly as a consequence of the latter, we are familiar with important distinctions like the one between external, internal and immanent critique (recently Herzog, 2016b: 22-38). Yet, no, one cannot be sure at all that we know what it means to understand discourse analysis as critique. It is not by chance that one of the most influential contributors to CDA, Teun van Dijk (2015: 479), views a more thorough discussion of what it means to be critical as one of the main research tasks today. This text hopes to advance the discussion.
To do so, I will distinguish two ways of thinking about the relationship between discourse analysis and critique: We can conceive of it as an external relationship; then I will speak of "discourse analysis and critique". Or we can think of it in an integrated manner, allowing us to speak of "discourse analysis as critique". One main difference between the two perspectives could be summarized in the following way: In the external relationship of "discourse analysis and critique" critique precedes the analysis and any critical potential will thus mostly be attributed to the critical "attitude" of the analyst. On the other hand, in the integrated relationship of "discourse analysis as critique" critique will be performed in the course of and by the analysis, and the critical potential will be attributed to discourse analysis itself, or rather, and more precisely, to discourse analysis read as a discursive formation. 1 The introductory questions laid out the trajectory for what is to be argued: I will start by showing that what I will call the idealtypical position of CDA/CDS (and in particular many of in the writings of its most influential protagonists) is shaped by an external understanding of "discourse analysis and critique". I will then turn to Foucault, yet not right away to the Foucault of the aforementioned famous talk on critique, but rather to the Foucault of the Archaeology of Knowledge (Foucault, 1972). The Archaeology is of importance because it allows us to get a clearer picture of the activity of discourse analysis itself: It allows us to see that, on the one hand, discourse analyses produce what they also examine: discursive statements. On the other hand, we witness that, on a regular basis, these statements are self-reflexive which is why they contribute to producing what can be called the discursive formation of discourse analysis. By generating statements, discourse analyses intervene in the field of available knowledge and in the power relations associated with it. Therefore, discourse analysis can only be understood as an interventionist form of academic work. This argument will most likely be shared by most scholars of CDA/CDS. It is, however, not always well-understood in what ways we can call discourse analysis "interventionist". Hence, in the final part of the article, I will sketch out the net of relations in which discourse analytical statements intervene. In doing so, I will show how discourse analysis itself resembles a discursive formation that can have critical effects in several dimensions: in regard to the research object (that is, the discourse that is analysed), in regard to the subjectivity concerned (among them the subjectivity of the analyst), and in regard to the formation of the social sciences and humanities in which the (sub-)formation of discourse analysis is embedded. The article will end with a short conclusion.
Discourse analysis and critique: on the ideal-typical position of CDA As has already been outlined above, we can conceive of the relationship of discourse analysis and critique in two ways, as external or as integrated. As a shorthand, we can speak of "discourse analysis and critique" in the first case, and of "discourse analysis as critique in the second case. To explicate in more detail how discourse analysis and critique are related externally, I will now reconstruct the relationship of discourse analysis and critique as it is ideal-typically understood in core positions of CDA. 2 These core positions are of particular interest because they were the first and still are the most prominent examples of discourse analyses that explicitly claim for themselves the label "critical". Let me stress in advance that the subsequent reconstruction is an ideal-type in the strict sense: Not each and every contribution relating itself to CDA will necessarily share this understanding of the relationship between discourse analysis and critique in every regard. But the ideal-typical reconstruction assembles a number of very common descriptions of this relationship in CDA; and thus it is essential in advancing the argument to be made here.
Following Reisigl (2014), we can distinguish six major varieties of CDA: (1) the "Duisburg group" around Siegfried Jäger, which leans on Michel Foucault and Jürgen Link; (2) the "Oldenburg group", which is somewhat connected to the Duisburg group but more focused on the linguistic analysis of single texts; (3) the socio-cognitive approach of Teun van Dijk; (4) the approach of Norman Fairclough, which is firmly anchored in social theory; (5) the socio-semiotic CDA in the line of Gunter Kress and Theo van Leeuwen; and (6) the "discourse-historical approach" connected to Ruth Wodak. Five of these six approaches (with only the Oldenburg group missing) can also be found in recent introductory volumes to CDA (e.g. Wodak and Meyer, 2009b). While reconstructing the ideal-type just mentioned, I will not be able to do justice to all of the specificities of each approach. However, the argument of this article can be made by focusing on some central tenets that are widely shared among the different perspectives. I will concentrate on the approaches of van Dijk (1993: 252-254;: 467), Fairclough (2010 and Wodak (Fairclough and Wodak, 1997: 271-280;Wodak and Meyer, 2009a: 1-10) who have formulated the basic assumptions and principles of CDA in similar ways. In particular, five aspects are of great importance.
First, many protagonists of CDA emphasize that their analyses start from real social and political problems. In van Dijk's words: "(CDA's) problems are ''real'' problems, that is the serious problems that threaten the lives or well-being of many" (van Dijk, 1993: 252). For many social scientists, such a statement at first sight appears to be a commonplace. Its significance, however, lies in the fact that its context is linguistic discourse analysis. Since the 1950s (Harris, 1952), the latter has slowly developed as the analysis of the production of meaning, at first beyond the single sentence (for example, in dialogical communication), later also beyond the single text. For a long time, this amounted to mostly formal-linguistic analysis of discourse. 3 Of course, formal analysis, too, deals with problems; but, as van Dijk continues in the passage just quoted, with "the sometimes petty disciplinary problems of describing discourse structures" (van Dijk, 1993: 252). By turning away from a predominantly linguistic analysis and instead concentrating on the scrutiny of linguistically mediated social and political events, CDA was able to delineate an important distinction, in particular in its own discipline of linguistics. The line is drawn between a non-critical (that is, formal-linguistic) and a critical form of discourse analysis that is first of all interested in social and political problems.
Second, this turn implies the necessity to link the linguistic perspective to other perspectives, particularly from the social sciences, from psychology and from cognitive science, to gain a better foundation for assessing social relations. It is, therefore, hardly surprising that scholars like Ruth Wodak describe CDA as a multi-disciplinary and multi-methodological endeavor: "CDA is […] not interested in investigating a linguistic unit per se but in studying social phenomena, which are necessarily complex and thus require a multidisciplinary and multi-methodological approach" (Wodak and Meyer, 2009a: 2).
Third, CDA scholars aim to analyse in particular those discourses that express, legitimate, reproduce or question relations of power and domination (van Dijk, 2015: 467). Despite regularly referring to Michel Foucault, the notion of power or dominance that is prevalent in many (but not all) of these analyses is actually rather actor-centered, mostly focusing on the power that certain groups hold, allowing them to dominate and to uphold social inequalities: "Dominance is defined here as the exercise of social power by elites, institutions or groups, that results in social inequality […]" (van Dijk, 1993: 249/250).
Fourth, the analysis of power is explicitly linked with a normative perspective. This is surely the crucial point for the selfdescription as critical discourse analysis. In other words, CDA is critical because its proponents from the very beginning take on a standpoint, judging the social (power) relations that are to be examined. An understanding of what is normatively questionable in our societies and of the desirable direction for improvement thus precedes the analysis. Following Norman Fairclough, CDA […] focuses on what is wrong with a society (an institution, an organization and so on.), and how "wrongs" might be "righted" or mitigated, from a particular normative standpoint. Critique is grounded in values, in particular views of the "good society" and of human well-being and flourishing, on the basis of which it evaluates existing societies and possible ways of changing them. […] The crucial point, however, is that critique assesses what exists, what might exist and what should exist on the basis of a coherent set of values (Fairclough, 2010: 7).
Starting out from their values, critical discourse analysts bring "explicit political commitments" to their research (Fairclough, 1996: 52). "In other words, CDA is discourse study with an attitude" (van Dijk, 2015: 466). As a consequence, the analysis of power structures cannot be reduced to describing them as accurately as possible. Rather, analysts aim at disclosing the discursive structures and strategies that help to install and uphold what is from their standpoint perceived as an abuse of power. For this reason, Fairclough argues for combining the critique of power with a critique of ideology. By ideology, he refers to discursively manifest societal interpretations and explanations which "can be shown to be not just inadequate but also necessary -necessary to establish and keep in place particular relations of power" (Fairclough, 2010: 9). Related topics we find regularly addressed in CDA are, for instance, social discrimination, racism, anti-Semitism, xenophobia and sexism (Reisigl, 2014: 94).
When, similar to Fairclough, Wodak and Meyer (2009a: 8) argue that critique in CDA aims at disclosing structures of power and to unmask ideologies, this points us to a fifth and final aspect: the goal of enlightenment and of emancipation (ibid: 7). 4 To reach this goal, many critical discourse analysts emphasize the need to discuss their findings with those who are participating in the discourses that were analysed: for example with teachers and pupils in the case of school discourses, or with doctors and patients in the case of doctor-patient discourses. Those who suffer most from inequalities are the ones who eventually are to profit most from the critique. At the same time, the critique wants to weaken the position of the power elites "that enact, sustain, legitimate, condone or ignore social inequality and injustice" (van Dijk, 1993: 252). Taking sides for the oppressed and excluded can thus be viewed as an integral element of critical discourse analyses (ibid: 279, see also Herzog, 2016b: v).
If this ideal-typical five-point description of CDA core positions is correct, then it is surely a scholarly perspective that many citizens will find agreeable, at least insofar as they are not only interested in politics but also regard combatting inequality and domination as an important task. I would view myself as such a citizen, and I can fully relate to the normative impulse that is behind the core positions of CDA. Nonetheless, I want to argue that thus far we have not really been successful in grasping exactly the critical potential specific to discourse analysis. For we could only speak of the latter if a critique becomes recognizable and powerful in the course of and due to the discourse analysis itself; if, in other words, discourse analysis and critique stand in an integrated relationship to each other.
Yet, this is obviously not the case with the core positions of CDA, if we take seriously the assumptions and principles just ideal-typically reconstructed. It is evident that many classical CDA contributions relate discourse analysis and critique not in an integrated but in an external way: Social and political problems are well recognized before the analysis (and not in the course of it). The respective relations of power and domination have always already been diagnosed as a problem. The values on which critique builds precede the analysis; and the analysis is also normally not thought to feed back into the value system. The analyst's critical attitude must by definition already be in place before s/he begins the analysis. The positions of those who enlighten, those who will be enlightened, and those about whom there will be enlightenment are already fixed in advance; in other words, the subject positions in the language game of critique are always already filled.
In a well-known text on the rhetoric of critique, Michael Billig has analysed the economy of calling oneself "critical" in academia, particularly in the field of CDA (Billig, 2003). One passage of this text can help to explain in more detail how CDA ideal-typically conceives of the relationship between discourse and critique as an external relationship. Billig writes: Critical discourse analysts do not see themselves as conventional discourse analysts who happen to have radical or progressive views, as if social or political criticism were something additional to their academic work. Instead, CDA is seen to be a means of criticising the social order (Billig, 2003: 38, my emphasis).
At first sight, Billig appears to contradict my diagnosis that discourse analysis and critique form an external relationship in CDA because he states that for critical discourse analysts critique is not just something additional to their academic work. This calls for clarification: To speak of an external relationship is not to say that critique and discourse analysis are incompatible and must thus stay separate. Quite evidently they can be combined, exactly in the way Billig describes: with discourse analysis being the means of critique. But it is precisely here that the externality resides. Discourse analysis is perceived as an instrument to explicate a critique, but a critique that already exists independently of the analysis. Viewed from the ideal-typical CDA position, it is not discourse analysis itself that is critical but rather the humans with a critical attitude that work as discourse analysts.
To say that discourse analysis and critique form an external relationship will thus mean two overlapping things: On the one hand, the critique temporally precedes the analysis in several dimensions: problematization or description of normatively questionable social relations, normative standards, critical attitude of the analyst and subject positions of critique. On the other hand, this will mean also that critique and discourse analysis are separated systematically since discourse analysis has a status that is different from that of critique. While the critique or critical attitude develops or developed autonomously, discourse analysis is in the ancillary, instrumental position to trace the critique empirically. To be sure, if a particular critique gains some empirical traction due to an analysis, this is a great achievement. But it stays unclear what it is that discourse analysis specifically adds to the critique. By all means, the critical potential specific to discourse analysis cannot amount to simply reproducing a critique empirically that is actually already formulated. If a specific critical power is to emanate from discourse analysis itself, critique and discourse analysis must be inextricably interwoven; and at the same time the critical effect should not be just as easily attained with other methods. Discourse analysis would have to perform, not simply to reproduce critique. In our terminology, discourse analysis and critique would have to form an integrated, not an external relationship. We would then speak of "discourse analysis as critique" instead of "discourse analysis and critique".
Let me add two clarifications before my main argument is taken to the next step. First, in a number of noteworthy recent publications, Benno Herzog has proposed to re-think the role of (critical) discourse analysis as an instrument of "immanent critique" as it is conceived mostly in Critical Theory (Herzog, 2016a, b). Immanent critique seeks to understand the normative standards, which are societally shared to then critique societal structures and developments that do not stand up to these normative standards. In particular, this critical procedure differs from external critique, which would muster some external normative criterion as the foundation of its critique. In regard to CDA, Herzog argues that most of its representatives rely on this external form of critique (Herzog, 2016b: 57), which is in line with much that has been argued here. To avoid confusion, I nonetheless want to highlight some differences between Herzog's perspective and the one proposed here. Most importantly, the distinction between external and immanent critique is not the same as the distinction between an external and an integrated relationship of discourse analysis and critique. The former distinction helps to ask the question "How is normative critique possible?" (Herzog, 2016a: 279), opposing external standards that have been established by the theorist to standards that are reconstructed as standards, which are immanent to society. The latter distinction asks how critique and discourse analysis are related. Here, the focus is not primarily on the normative viability of critique; rather, the main question is whether critique is temporarily and systematically antecedent to or integrated with discourse analysis.
In addition, when the notion of immanent critique is applied by a discourse analyst (as Herzog proposes), this is not necessarily suited to overcome an external relationship between critique and discourse analysis. Certainly, immanent critique will aim at constituting relevant normative standards by means of the discourse analysis (Herzog, 2016a: 287); and this would be one aspect pointing to an integrated relationship. However, in this way discourse analysis would still be ancillary to critique because immanent critique starts from the assumption that some situation is worthy of critique and looks at discourse analysis as a tool to voice that critique. Again, this does not assume that there could be a critical potential that is specific to discourse analysis, making it critical per se: The main burden of the critique is put on the (validity/legitimacy of) the normative standard, not on workings of a discourse analysis. The argument I want to put forward here is concerned with the latter. I want to show that there is something in the very discursive formation of discourse analysis itself (independent of any normative standards) that has critical effects.
This having been said, let me finish this section with a second clarification. By proposing a perspective on the relation between discourse analysis and critique different from the ideal-typical CDA perspective just reconstructed, I want to argue neither that CDA has got it all wrong nor that the ideal-typical perspective is not useful or productive. Rather, I want to point to a critical effect of discourse analysis that is usually not taken into account. In fact, CDA contributes significantly to discourse analysis's specific critical potential but mostly looks elsewhere when asked to give an answer about its mode of critique.
Discourse analysis, statement and discursive formation My goal in this and the following two sections is to develop an idea of how we can think about a critical potential that is specifically attached to discourse analysis. In order attain that goal, it is first of all necessary that we remind ourselves of one fundamental insight about the practice of discourse analysis. Beyond all the differences between diverse analysts and various analyses, even beyond the difference between "critical" and "noncritical" analyses, we can make out two essential and general attributes of this practice: First, any such practice will generate statements in Foucault's understanding; and, second, beside saying something about the discourse that is being analysed, these statements are reflexive in the sense of referring to discourse analysis itself, thus contributing crucially to the development of a discursive formation of discourse analysis.
Let me flesh out the first point: Discourse analyses obviously operate in the same medium they scrutinize, in the medium of discourse. Or put differently, they produce statements which in turn can align to become discursive formations. Of course, this reference to the notion of statement (enoncé) leads us to Foucault. Also because Foucault was travelling uncharted territory, the descriptions given in the Archaeology of Knowledge (Foucault, 1972: 79-125) about what a statement is and of how it works can occasionally appear to be enigmatic or contradictory. Nevertheless, we can grasp the fundamental idea: A statement is a function that brings about everything on which its own existence depends. For this reason, it is also called a "function of existence" (ibid: 86). A statement will intervene in a field of relations between different elements, a field that would not even exist without the statement, but without which the statement would be meaningless. Hence, statements constitute the objects and the topics about which they speak, just as they constitute the concepts that we use when speaking. Just like Baron Munchhausen pulled himself out of the swamp, a statement will bring into being everything it needs to make sense (that is, all the relations that constitute it).
What would happen now if we used this Foucaultian insight to look reflexively at discourse analysis itself? The first conclusion may be close to a banality: Discourse analyses, as systems of statements, produce their objects-that is discourses!-in ever specific ways. On the one hand this is indeed a banality because almost all analysts will be aware of the fact that a discourse is not simply "out there". Rather, it comes into being in the course of an analysis in its concrete shape and dynamics. This is obvious as ARTICLE PALGRAVE COMMUNICATIONS | DOI: 10.1057/palcomms.2017.74 soon as we remind ourselves of the problem of corpus construction, this crucial hinge of any analysis (Busse and Teubert, 2014). On the other hand, despite of its seeming banality it may sometimes be worthwhile to mention that an analysis of, say, racism, understood as a system of statements, must necessarily re-produce its object, without which it could not exist, in a specific way. Of course, this does not mean that racism does not exist outside of the analysis, nor that the analysis itself is racist. But it will mean that any particular knowledge that is being produced by an analysis of racism at the same time must articulate a specific construction of racism. Now we must add a second insight: Discourse analyses will obviously not only generate statements about their objects, but also about discourse analysis itself. And hence, they produce a discourse that leads to the continuous re-emergence of a discursive formation of "discourse analysis". This is the effect of the interplay of numerous journal articles, monographs and handbooks, of discourse theoretical writings and empirical analyses; basically everywhere where it is negotiated what can be legitimately said about discourse analysis. And as with all discursive formations, one should never forget that whatever can be legitimately said (just as whatever can count on being comprehended) will give us hints about the power relations underpinning a discursive formation.
Whenever we discuss critique as an attribute of discourse analysis (and not primarily as a consequence of the analyst's critical attitude), that is to say "discourse analysis as critique", this will, so I argue, refer to a specific quality of discourse analysis as a discursive formation. However, before addressing this in more detail in the penultimate section, I find it helpful to take up one important point that has also repeatedly been argued by discourse studies scholars, in particular by CDA scholars, namely the argument that discourse analysis is an interventionist form of academic work. This insight will, as I argue later on, become even more forceful once we take into account the fact that discourse analysis constitutes itself as a discursive formation.

Discourse analysis as interventionist academic work
If we accept Foucault's argument that a statement is a "function of existence," which must call forth everything that enables its own existence; and if at the same time we accept that discourse analyses can be grasped as systems of statements; then it is also evident that we can think of discourse analysis in no other way than as an intervention in the relations of the knowable, that is in the relations through which elements are related to one another as "known" or "recognized" elements whose specific relation makes sense. At a second glance, however, we will see that this is by no means a specificity of discourse analysis. Other forms of knowledge production, too, in academia and elsewhere, produce statements in the Foucaultian sense and should, therefore, be considered to be interventionist. Yet, as the above discussion of CDA has shown, 5 the reflection of its own interventionist status is firmly inscribed in the DNA of a great number of discourse analyses. In other words, the discursive formation of discourse analysis is continuously accompanied by conscious reflections of the fact that academic activity is a socio-political practice to the extent that it interferes with existing knowledge orders and power structures. Discourse analyses are, therefore, often held to be academic endeavors as well as instruments in political struggles. This results in a dividing line between discourse analysis and a dominant understanding of the (social) sciences that likes to uphold the separation between neutral scientific observation and description and the seemingly external object of that external practice. Insofar as discourse analysis conceives of itself as intervention, this separation has been and has to be scrapped.
Discourse analysis is always already part of a societal and political practice; it cannot be reduced to distanced observation, description or explanation.
It is not difficult to see that this constitutes a proximity of discourse analysis and Critical Theory. In his classical article on "Traditional and Critical Theory", Horkheimer (2002) had made a similar argument. Traditional theory-and, so we might add, traditional methodology-undertakes a separation between the world of natural or social objects and the world of science. This is an essential condition for the suppression of the world's contradictory state in scientific analyses. The latter follows from looking too closely into particular small-scale social causalities while at the same time accepting as simply given the social relationships as a whole, analogous to natural phenomena that cannot be influenced (ibid. 208/209). In contrast to this, Horkheimer goes on, Critical Theory will recognize that human reality is open to change and will view itself as a weapon in the struggle against contradictory societal conditions. Different from many critical discourse analyses whose explicit goal formulation entails a positive and prescriptive impact on societies (Toolan, 1997: 86/87;Fairclough, 2009: 171;2010: 7;Macgilchrist, 2016: 267), early Critical Theory favours negation. It aims at demonstrating again and again the contradictions of a societal totality. Even though Horkheimer calls on us to stay true to "the idea of a future society as a community of free men" (Horkheimer, 2002: 217), early Critical Theory does not offer us a yardstick by which the success could be measured, 6 nor is its first goal to contribute productively to the world. Rather, its intervention is first of all a negative one, an exclamation of "not this way!", resulting from the perception of obviously irrational conditions in capitalist societies.
It is thus hardly surprising that Foucault (1997: 55) speaks of a "fellowship with the Frankfurt School" in his famous talk entitled "What Is Critique?". For Foucault's central argument is, of course, that critique resembles a specific art opposed to the art of governing, "the art of not being governed like that and at that cost" (Foucault, 1997: 45). But the proximity of Foucault to the Frankfurt School cannot be reconstructed convincingly in every respect. Most importantly, Foucault does not share Critical Theory's belief that we can rationally deduce what a reasonable world would look like. He thus radicalizes the relationality of all critique: Not only is it always socially and historically contextualized, but in addition to that it is no longer connected to the ideal of a society that could become ever more reasonable. 7 Instead, Foucault's goal consists of examining the rationalities themselves and their entanglement with power relations (cf. Butler, 2001). Subsequently, the genealogical Foucault pushes the idea of intervention even further when-drawing on Nietzschehe calls for a historical-philosophical practice whose aim is "to make one's own history, fabricate history, as if through fiction, in terms of how it would be traversed by the question of the relationships between structures of rationality, which articulate true discourse and the mechanisms of subjugation which are linked to it" (Foucault, 1997: 56).
Despite the obvious differences between Critical Theory and Foucault we can conclude that we are in good company from Frankfurt and Paris when we argue that discourse analysis's critical impetus is anchored in its explicit self-understanding as interventionist. And yet we face two problems that make it prima facie difficult to defend the idea that this interventionist selfunderstanding already resembles the core of discourse analysis's critical potential or of "discourse analysis as critique". The first problem is that speaking of an interventionist self-understanding seems to be quite similar to speaking of a critical attitude because apparently the former, just like the latter, reflects cognitive processes preceding the discourse analysis. And second, as CDA scholars have maintained repeatedly and rightfully in the past, by far not all discourse analysts view their analyses as interventions, even if de facto they do intervene in knowledge and power relations.
But both problems are, I would argue, problems only prima facie. Let us have a closer look. The argument about the attitude of the discourse analyst or the necessity of previously existing values clearly attributes the responsibility for critical capability to the individual that first acquires her/his attitude and values, and then becomes active as a discourse analyst. The argument defending discourse analysis as an interventionist form of academic activity, on the other hand, refers to the discursive formation of discourse analysis. The latter is part of a larger formation, at least in the social sciences and the humanities: Here, the question whether it is preferable or promising for an academic to have an interventionist standpoint has been contentious at least since Max Weber's famous article on objectivity in the social sciences (Weber, 1949). However, it is also true that this question is only rarely discussed in mainstream publications, at least as long as they are not written against the backdrop of some fierce methodological dispute (like, for example, the one between Popper and Adorno in the 1960s, cf. Frisby, 1972). But discourse analysis has meanwhile constituted itself as a discursive formation that makes it almost impossible not to take a position on the question of intervention and critique (assuming that one wants to be taken seriously in the community of discourse analysts). This could be traced back to what most discourse analysts-linguists as well as social scientists-can agree on when they describe their own activity: In one way or another the analysis of discourse is understood as the analysis of "text in context". This already structures the field of the sayable in the sense that two spheres-text and context-must interact to generate social meaning. It is thus by no means farfetched to argue by analogy that there will always be a close connection between academic text and societal context. Such an analogy is also supported by almost every prominent theoretical source of inspiration for discourse analysis. Besides Critical Theory and Michel Foucault, names like Ernesto Laclau or Judith Butler could be mentioned among those for whom symbolic intervention and the re-arrangement of meaning are just as much part of the standard repertoire as the interaction of academia and society in processes of meaning production (for example, power/knowledge for Foucault, hegemony for Laclau, performativity for Butler). And last but not least, within the formation of discourse analysis, the interventionist claim is being discursively defended and practically rehearsed over and over again.
For young academics, these three aspects-text/context orientation, standard repertoire of discourse theory and practical rehearsal -produce incentives to organize their own work accordingly if they want to attract attention for it. And even if critique in the sense of intervention has sometimes been problematized (for example, by Widdowson, 1995;Schegloff, 1997), the respective discourse on the questions of discourse analysis and critique must nonetheless be reproduced. In a fashion, a discourse analyst can hardly refuse the interpellation to take up these questions and to position oneself with respect to them. Altogether, we can thus describe discourse analysis as a discursive formation in which the structures of sayability prompt the language of critical intervention with a much greater probability than in mainstream scholarship. It is against this background that we can speak of discourse analysis as critique because critique is not anchored in the analyst her-/himself but in the discursive formation "discourse analysis" itself. Once we accept this it also solves the second perceived problem named above: In order to speak of discourse analysis as a criticalinterventionist form of science not every analyst must explicitly share a corresponding attitude. It is sufficient if statements that expressly and regularly describe the discourse analysts' own activity as critical shape the discourse, and if thereby they exert discursive power.

Discourse analysis's field of intervention
Hence, one important reason for speaking about discourse analysis as critique will be that the topic of interventionist academic work does not only become sayable, but that, in addition to this, it is very difficult to not address this point at some point if one wants to be taken seriously as a discourse analyst. However, to say that something is an important reason does not imply that it is the only reason. Rather, we now need to come to a better understanding of where and in what discourse analyses intervene. Above, I argued that discourse analysis itself produces discourse, that is, that it comes in the form of statements in the Foucaultian sense. If we take this argument into serious consideration, it will be concluded that discourse analyses-having the quality of statements-intervene in the conditions of their own existence. I will now further argue that there are three crucial dimensions of this intervention, each of which is a necessary condition of any discourse analysis. This means that only by intervening in these dimensions can an analysis secure its own recognizable existence as a discourse analysis. And last but not least, in all three dimensions there are modes of intervention that allow discourse analysis to develop critical effects. First of all, a discourse analysis would not make sense if there were no discourses to be analysed. So, just as already mentioned above, any discourse analysis must intervene in a specific discourse and its subject matter by reconstructing it in one way or another. Any analysis will thus continue and contribute to the discourse it looks at, be it a discourse on New Labour (Fairclough, 2010: 380-391), on intellectuals (Angermuller, 2015), on racism (Reisigl and Wodak, 2001), on "social market economy" (Nonhoff, 2006), on the Iraq war in Western parliaments (van Dijk, 2009: 213-247;Nonhoff and Stengel, 2014) or on environmental policies (Hajer, 1995), to name a few. Second, any discourse analysis will depend on being conducted by someone about someone's discourse and for some readers. Thus, it will intervene in subject relations. On the one hand, there are those subject relations that are inherent in the discourse that is the research object, for example, the relations between doctor and patient (cf. Wodak, 2014). But there will always also be another type of subject relation that is concerned, namely the one between the discourse analyst herself/himself and the subjects that are participants of the analysed discourse or between her/him and the respective readers of her/his study. Not the least, there is also need to discuss the self-relation of the analyst as a particular form of subject relation (and thus to look for one final time at the question of the critical attitude). Third, and maybe most importantly for the question of critique, any analysis will intervene in the social institutional field it emanates from, that is, in the intersubjective relations of academia. I will now discuss these three aspects one by one, aiming at closer examining the specificity of discourse analysis as critique.
Intervention in the subject matter. In regard to the subject matter dealt with in a discourse, the critical potential has been described quite often, most conspicuously so by CDA. However, as I showed above, the latter critique will usually mean a critique of social and political conditions that is in place before the analysis starts. I argued that such a form of critique cannot resemble the specific critical potential of discourse analysis, simply because it exists before and independent of the discourse analysis. I am not saying, to be very clear, that such pre-existing critiques do not result from clear perception or acute social analysis or that they ARTICLE PALGRAVE COMMUNICATIONS | DOI: 10.1057/palcomms.2017.74 cannot be normatively appropriate. But they are not a form of critique that is specific to discourse analysis. Social criticism is, after all, the daily bread of public debates even though certain criticisms will at certain times be more prevalent than at other times. If discourse analysis is to specifically function as critique, this cannot be due to the mere fact that certain social or political conditions are worthy of critique. Rather, critique must be connected to the mode by which discourse analyses look at their subject matter, that is by discourse analytical methodology in the widest sense.
Of course, the concrete methods of discourse analyses vary greatly, but there are two standards that guide discourse analysis as a whole. In this context, "guide" should be understood in accord with what has been said above about discourse analysis as a discursive formation: These standards need not be adhered to by each and every study. But they are so prominent within the discursive formation that any deviation will produce consternation in the discursive community. In other words, in regard to these standards, sayability is strongly limited. The first of these standards is that discourse analyses do not approach their subject matter in an arbitrary fashion but they focus on the generation of knowledge and social meaning through oral and written language use (for example, Wodak and Meyer, 2009a: 2). 8 The second standard derives from numerous sources such as the Foucault's analytics of power, Laclau's theory of hegemonic discourse or the entire CDA tradition. It consists of the idea that discourse analysis will always combine an interest in the production of meaning and knowledge with an interest in the social and power relations with which meaning/knowledge production is intertwined. Or phrased differently: Any analysis of knowledge and meaning production that does not at the same time raise the question of how specific discursive relations lead to or amount to specific social relations, that is, how they (re-) constitute inequality and power relations, will find it difficult to be accepted as a proper discourse analysis. Hence, we are faced with an interpellation of the analyst subject-independently of the specific subject matter-to not forget the question of power when looking at discourse.
As we have already seen above, the ideal-typical CDA position too has argued that discourse analysis becomes critical when and if it scrutinizes discourses for structures of power, domination and inequality. So, how does the view proposed here differ? The difference results from the fact that many studies that position themselves in the field of CDA have already before the beginning and independent of their research taken the decision of which power structures will be relevant and questionable. This has probably been argued most clearly by Teun van Dijk: Their (the critical discourse analysts', MN) critical targets are the power elites that enact, sustain, legitimate, condone or ignore social inequality and injustice. That is, one of the criteria of their work is solidarity with those who need it most. (…) Their critique of discourse implies a political critique of those responsible for its perversion in the reproduction of dominance and inequality (van Dijk, 1993: 252/253).
Here, the methodological imperative to combine the analysis of discourse and power formations is limited in so far as it is clear from the very beginning what the power formation of interest looks like and why it is to be criticized. Hence, discourse analysis joins up with an already existing critique of a power. This constitutes the external relation that I have called "discourse analysis and critique". In contrast, any discourse analysis that is to function as critique will first and foremost serve to illuminate in the course of the analysis the complexity of power relations and its historically varying interplay with formations of knowledge and meaning which are themselves changing.
Discourse analyses can be called critical interventions in fields of existing knowledge because they scrutinize how such knowledge came into being in the first place, how it manifests in very concrete social contexts as effective social meaning, which conditions of existence and which consequences are attached to it, and by which norms and demands it is accompanied. 9 While describing and decomposing theoretical or practical knowledge, discourse analysis will always also do something different: It will reconstruct the history and the specificity of seemingly natural knowledge and of routinely accepted meaning, hence alerting us to the contingencies of constellations of the given. But to speak of contingencies is, of course, just another way of speaking of power. Discourse analysis as a discursive formation is permeated by the expectation and the appeal to link the study of knowledge and meaning to the study of dominance and submission, superiority and deference, hierarchies and hegemonies. The imperative to scrutinize how relations of meaning and relations of power support each other contains much of the critical potential of discourse analysis. Of course, contingency is not arbitrariness. There are always historical reasons for the development of a certain knowledge or a meaning formation. But exactly by delving into these reasons the particular quality of any knowledge or meaning will become clear: historical, societal, affected by power, outcome of conflicts. To the extent that this is the methodological focus of the discursive formation "discourse analysis", it will be able to function as critique.
Intervention in subject relations and in the self-relation. To say that discourse analysis intervenes in subject relations can be taken to understand two different things. On the one hand, a discourse analysis could aim at changing subject relations in the discourses it looks at, for example, by confronting those who participate in a power relation with the results of the analysis, thus hopefully raising awareness and maybe a change of consciousness. Since such an attempt at influencing discourse participants is closely connected to an intervention in the subject matter, in general the arguments made above pertain to it; and they need not be repeated once more (even though the aspect of ''enlightenment'' will have a role in the next paragraph). On the other hand, however, there is also what we may call a reflexive intervention in subject relations, that is to say an intervention that takes into account the role of the discourse analyst in the process of the analysis. It is this aspect I will tackle now.
My starting point will once more be to view discourse analysis as a discursive formation. The crucial implication of this perspective for the scrutiny of subject relations is that discourse analysis itself is a formation permeated by power. Two aspects are important in this context. First, as has already been mentioned shortly above, those kinds of discourse analyses that combine critique with the ideal of enlightenment will necessarily establish a power relation between those who enlighten (the discourse analysts), those about whom there will be enlightenment (the discourse participants) and those who are to be enlightened (the readers). The proponent of enlightenment performatively moves into a higher position that counter-intuitively forces her/him to uphold inequality at the very moment of formulating a critique of power, particularly if the will to enlighten entails that the perceptions of the actors in the discourse are not taken seriously. As Celikates (2009) showed in a detailed study, this is a dilemma that is difficult to get around for any critical academic work. 10 The critic will often have achieved an advance in knowledge allowing for a multi-perspective view of the social relationships that is often not possible for the actors who are deeply involved in these very relationships. And yet, the starting point of analyses will usually have to be the articulation of discontent and of critique in the field of study itself. This dilemma will most likely not be easily solved for discourse analysis if it wants to uphold its claim to say something new and true about its subjects. But discourse analysis will only function as critique or as a critical formation to the extent that discourse analysts are able to reflect on their own involvement in a struggle for truth and thus in a game of power.
One consequence of this leads to the second point that I want to discuss: the self-relation of a discourse analyst who describes her-or himself as a critic. The recurring topic of the "critical attitude" as the foundation of discourse analytical critique has already been mentioned; it is probably the most important instance of statements that establish the analyst's self-relation. To speak of the critical attitude is by no means restricted to CDA which Teun van Dijk (2015: 466) has called "discourse study with an attitude". Discourse analytical positions critical of CDA come to a similar conclusion, arguing that in many CDA studies it is mostly the perspective of the analyst that is revealed (Widdowson, 1995: 169). And of course Michel Foucault, too, viewed critique as an effect of the attitude of not wanting to be ruled like that. Sometimes the self-description of discourse analysts could even be said to approach self-glorification, for example, when van Dijk writes this about his own activity: "Critical discourse analysis is far from easy. In my opinion it is by far the toughest challenge in the discipline" (van Dijk, 1993: 253). One corollary of such formulations of the self-relation has already been discussed above: Critique is thought to precede the analysis, because it anchors in the critical attitude of an analyst who then takes on the heroic task of conducting and composing a CDA (this is a sign of an external relation of critique and discourse analysis). The more serious problem, however, is that an attitude can easily become a black box, a fetish of self-authorization allowing to admonish the seemingly uncritical actors in the field of discourse analysis (see also Toolan, 1997: 86/87). At the same time, more often than not, we can make out the critical critics' (to borrow a term of Marx and Engels, 1975) self-congratulatory secret joy about their own attitude. This does, of course, again come down to the realization that discourse analysis is a discursive formation that itself is shaped by power relations; and within these power relations, it matters to position oneself explicitly as critical. If, however, discourse analysis is to function as critique in regard to the self-relation, its effect would have to show the analyst's own entanglement in power games, thus undermining any self-positionings as autonomous, heroic subjects of critique. The major task here is to develop new forms of writing (cf. Billig, 2003: 44), possibly on a playful and ironic note. Even though this argument has been made before (Macgilchrist, 2016) and despite the existence of some good examples, 11 this task is not an easy one-it is maybe the toughest challenge in the discipline.
Intervention by provocation in the professional context of academia. To say that discourse analysis can function as critique is to say at the same time that it performatively evokes critical effects. This probably shows best when the discursive formation of discourse analysis meets with the larger and encompassing discursive formation of social science or the humanities.
To be more precise, it shows when discourse analysis-as a discursive formation-provokes controversy because it is viewed as a challenge to how academic work is usually supposed to be conducted. Discourse analysis's provocation comes in three guises: in regard to the relationship between academia and politics, in regard to its attractiveness to younger scholars and in regard to methodology.
The first aspect has already been alluded to above. By professing its interventionist stance actively, the discursive formation of discourse analysis is in tension with the claim of objectivity still prevalent in much of social research. Discourse analysis is often held to be politically predisposed (Billig, 2003: 39), and being predisposed is taken to prevent valid scientific analysis. This critique has been raised against CDA in a nearly generic form by Henry Widdowson (1995: 169): It (CDA) presents a partial interpretation of text from a particular view. It is partial in two senses: first, it is not impartial in that it is ideologically committed, and so prejudiced; and it is partial in that it selects those features of the text which support its preferred interpretation.
Even though Widdowson explicitly addresses CDA, his accusation will seem familiar to many discourse analysts even if they do not position themselves in this tradition of discourse analysis. Even if discourse analysis is not attacked for some political predisposition, it is regularly suspected of not producing any meaningful results. Rather, it would only detect phenomena "that are self-evident and have long been denounced and with which the majority of people would agree" (Manjarrés, 2007: 237). Thus, like any provocation, discourse analysis receives responses that shift between ascription of irrelevance and fierce (counter-)attack.
Following Toolan (1997: 84/85), we can discern a second aspect of provocation, this time a very material one, for the provocation lies also in the fact that discourse analysis constitutes a rather successful heterodoxy in the humanities and the social sciences. A huge number of younger scholars want to learn about and to conduct discourse analyses: In the last 15 years, the number of publications has exploded in almost every discipline in the humanities and the social sciences (cf. Angermuller et al., 2014b: 39-339). The number of large international conferences is increasing and more and more scholars attend these conference (for example, CADAAD or the DiscourseNet Congresses). The international networking platform www.discourseanalyis.net has an ever growing number of members, ca. 5000 in spring 2017. All in all, one can make out a relatively successful marketization of discourse analysis as an interpretive, critical form of scholarship (see already Billig, 2003: 42ff). One consequence of this is a heavily growing number of student theses and dissertations that are written as discourse analyses. Many students and young scholars become subjects of the formation of discourse analysis because it offers them the subject position of a critical and professedly interventionist scholar that has gone out of style in many other areas of today's academia. At some universities there exist very successful-and in part long-standing-MA programs focusing on discourse analysis. 12 Through unfriendly eyes, this increased overall presence of discourse analysis can be perceived as an imperial expansion that challenges the normality of positivist science because it makes different things sayable. And this is just how the discursive formation of discourse analysis can performatively function as critique: as a critique and provocation of a complacent normalized scholarship.
Third, discourse analysis is provocative on the level of methods. This is maybe the most important point to make in this section because it somewhat runs counter to common understandings. Particularly proponents of CDA have argued that CDA's critique is not connected to the methods used: "Critical Discourse Analysis does not claim to be "critical" because of a technical or methodological difference from other approaches to the study of language" (Billig, 2003: 38). Such a perception is, however, the result of looking at the methodology of single studies. But the really interesting perspective is once more the one ARTICLE PALGRAVE COMMUNICATIONS | DOI: 10.1057DOI: 10. /palcomms.2017 on discourse analysis as a discursive formation. Within this formation, we realize an enormous heterogeneity, maybe even a disparity of how discourse analysis is put into practice (for an overview cf. Angermuller et al., 2014a;Tannen et al., 2015). We analyse macro and micro discourses, written and oral discourses, large media corpora and single texts. We look at communicative patterns or narrations, at concepts and formal markers, at statements, enunciations (however defined) and articulations, at metaphors and frames. Sometimes analyses remain mostly on the textual level, more often they will include contextual analysis. At some steps, we are interested in the scrutiny of text-inherent mechanisms; at others we will combine such scrutiny with a close examination of social and political power relations. Overall, the diversity of research designs and of the methodological process on the ground is impressive, in particular so because in the last 20 years or so discourse analysis has been transformed from a mostly linguistic endeavor to a trans-and multidisciplinary one. Maybe one could argue in line with Ernesto Laclau's and Chantal Mouffe's theory of hegemony (1985) that discourse analysis resembles a kind of a hegemonic project that was able to integrate an ever increasing variety of theoretical and methodological perspectives in a so-called chain of equivalence, organized by the nodal point "discourse analysis". Despite all the heterogeneity, the existence of such a nodal point makes possible a continuous critical and productive exchange process beyond disciplinary borders. Since this brings with it a lot of input and debates (no hegemonic project will ever exist without internal frictions), we are facing not only a growing but comparatively reflexive academic formation.
Nevertheless, it is in particular the great variety of theoretical and methodological approaches that invites criticism. This already held true in the 1990s when discourse analysts were challenged to standardize their research questions and their methodology to achieve better accessibility for students and teachers (Toolan, 1997: 99). But in particular, discourse analyses in the social sciences provoke irritation in regard to their methodology. In Germany, for example, the methods sections of Deutsche Gesellschaft für Soziologie and Deutsche Vereinigung für Politikwissenschaft some years ago cooperatively organized a conference to dissect the supposed "myth of discourse analysis". 13 Here, but also in other contexts, the core of the critique amounts to the contention that discourse analyses do not explicate how they garner their insights, that they would not work with a clear methodology, that they would always find what they look for, and that due to the great variety of approaches it would in general remain unclear what the label "discourse analysis" is supposed to cover. Particularly from the perspective of the social sciences that follow rather strictly standardized quantitative and qualitative methodologies, discourse analysis is confronted with forms of critique that border incomprehension of the entire research program (that is, of the idea to always combine the analysis of knowledge/meaning formations with that of power formations).
One crucial aspect of discourse analysis as critique, of its specific critical potential rests in the fact that this discursive formation is so successful in provoking reactions like the ones just mentioned. The reason for the attacks appears evident. Discourse analysis in all its heterogeneity demonstrates what a social science fixated on formal methodology seeks to remain silent about, namely that the complexity of the social cannot be researched with methods that are conceived to transcend the concrete objects they help to study, methods that are elevated to shrines of sublimity. This is not to say that methodology is necessarily a bad thing. There is a big difference between the legitimate demand that a single discourse analytical study should work in a reflexive, transparent and well comprehensible manner, and the assumptive expectation that discourse analysis as a discursive formation should surrender its plurality, heterogeneity and disparity to become an easier to handle streamlined product. There is no reason for discourse analysts to be shy or submissive: Even though there circulate conflicting statements in the discursive formation of discourse about how much methodology is needed, it is by no means the case that well-written discourse analyses fall short of standards of reflexivity or of transparent documentation of their methodology. But there are also hardly two analyses out there that machine-like do exactly the same thing. Thus, discourse analysis as a discursive formation does indeed lack the object-transcending understanding of methodology that is more common in content analysis or statistics. But the reason for this is clear: Since it is well-accepted common knowledge within the formation of discourse analysis that analysts must in specific ways (re-)constitute the discourses they study, suitable methodology must be developed in the context of this very process of (re-) constitution. 14 Hence the discursive formation of discourse analysis is shaped by two demands that are in tension. On the one hand any discourse analysis profits from methodological exactness, clarity and reflection (cf. Nonhoff, 2011: 100-102). 15 But on the other hand, we should be very reluctant to extrapolate from single studies to normalized methodological thinking or to standardized methods of discourse analysis. If, therefore, Michel Foucault resorts to the metaphor of a toolkit to describe discourse analysis, this is maybe not the best of all metaphors. For the task is not to resort to some ready-made tools, but rather to encourage some audacity and fantasy to continuously create new tools, fitting the respective object of analysis.
Within the system of modern science discourse analysis can only function as critique because as a discursive formation it breaks with the fetishism of methodology by allowing for heterogeneity and disparity, thus upholding the potential for ongoing irritation. This specific critical potential is not a transcendental quality of a context-independent discourse analysis but rather the attribute of the discursive formation of discourse analysis as it exists today. The latter depends on the great differentiation and the continuously practiced diversity and contentiousness of discourse analytical work. This implies at the same time that attempts at grounding discourse analysis in a more unified theoretical or methodological foundation-for example, through pushing the establishments of "schools"-will most likely reduce the critical potential of discourse analysis since they limit the terrain of the sayable (for a similar argument, cf. Billig, 2003: 44). It is only in a specific constellation of heterogeneity that discourse analysis can function as critique. Its specific critical potential is historically unstable, it is not simply a given.

Conclusion
In this article, I argued that we can read the relationship between discourse analysis and critique in two ways: On the one hand, a critique of certain societal conditions is already in place and can then be transferred into a discourse analysis. The critical potential will then essentially depend on the critical attitude of the discourse analyst. Because critique precedes the analysis, it forms an external relationship with discourse analysis. I thus called this relationship one of "discourse analysis and critique". I argued further that if we conceive of the relationship of discourse analysis and critique in this way, we will fall short of understanding the specific critical potential of discourse analysis, precisely because the critique exists independently of the discourse analysis. In contrast, discourse analysis functions as critique in so far as it is itself a discursive formation that produces critical effects which re-arrange the regime of sayability. This shows, for instance, in one decisive statement of this discursive formation, namely the statement that academic work is always interventionist. I then traced the interventionist character of discourse analysis in regard to its subject matters, in regard to the subject relations concerned (the self-relation among others), and in regard to its academic production context. In the latter, the professed heterogeneity and discrepancy of theoretical and methodological approaches of discourse analysis is crucial because it is one important origin of irritation with which discourse analysis is met by mainstream (social) science. Due to this irritation, it becomes possible to reopen and newly debate the scope of what is permitted, possible and useful in academic work.
Because it neither simply accepts the empirically nor the methodologically given, but rather enquires about its genealogy, discourse analysis can generate what we may call "unwieldy knowledge", 16 in regard to its research object as well as in regard to the methodology it uses. And hence, discourse analysis has the potential to produce the new, the different, the alternative-to be critical.

Notes
1 As will be shown in due course (Discourse analysis and critique: on the ideal-typical position of critical discourse analysis), this distinction between an external and an integrated relationship between discourse analysis and critique does not echo the distinction between external and immanent critique so familiar to students of Critical Theory. 2 I am aware of the fact that nowadays many scholars for good reasons prefer to speak of "Critical Discourse Studies" instead of "Critical Discourse Analysis". I will retain the former terminology, however, because it is able to link the different elements of my argument better than "Critical Discourse Studies" (particularly because Michel Foucault speaks of discourse analysis). 3 Many contributions to Konrad Ehlich's edited volume Diskursanalyse in Europa (Ehlich, 1994) still give a good impression of this earlier trend in linguistic discourse analysis. Some of the confusions between formal and critical discourse analysis are also echoed by Emanuel Schegloff (1997: 184). 4 For an interesting perspective from within CDA questioning the rhetoric of emancipation and empowerment, see Macgilchrist (2013). 5 One could easily show the same in different discourse analytical approaches, for example, in those who more than CDA rely on Foucaultian discourse analysis or genealogy (Kendall and Wickham, 1999;Arribas-Ayllon and Walkerdine, 2008), on the sociology of knowledge (Keller, 2011), on the discourse theory of Ernesto Laclau (Nonhoff, 2006), or on French enunciative pragmatics (Angermuller, 2014). 6 This does indeed distinguish Horkheimer's early Critical Theory (one could make similar arguments in regard to Adorno or Marcuse) from the more optimistic and prescriptive later perspectives of Habermas's theory of communicative action (Habermas, 1984) or Honneth's theory of recognition (Honneth, 1995). For useful reconstructions of the relationship between Critical Theory and CDA, see Forchtner (2011) or Herzog (2016b. 7 Most likely the difference between CDA and many critical poststructuralist discourse analysis resembles exactly this difference between Critical Theory and Foucault in a number of respects. Poststructuralist discourse analysis's critique is usually a local critique detached from the idea of a healed (a reasonable) society. It sees only the chance to criticize (and maybe even to overcome) specific contradictions and hardships, while CDA holds on to the hope of a reasonable, positively unified society. See on the poststructuralist account of the impossibility of society Laclau (1990: 88-91). 8 To limit the production of meaning to the linguistic production of meaning is a position that is not shared by all discourse analysts. In particular analyses resorting to a Laclauian understanding of discourse will view all acts that relate elements as meaning producing, including what others take to be material or non-discursive relationings (Laclau and Mouffe, 1985: 108). But the sayability of this position is contentious, as has become clear in repeated discussions about the scope of the discursive (cf. Dyk et al., 2014). 9 This argument of course connects to Michel Foucault's proposals concerning genealogy and critique (Foucault, 1977(Foucault, , 1997. See also Saar (2002Saar ( , 2007. 10 Jacques Rancière has put forward one of the most radical critiques of the inequality that forms the foundation of different forms of critical theory and of critical social sciences. According to him, the latter rely on a "method of inequality, reasserting continuously the division between the learned ones and the ignorant ones" (Rancière, 2016: 135). 11 A classic, mostly discourse-theoretical article in this regard would be Edwards et al. (1995). For a wonderful recent piece musing on the necessity of queering critical discourse studies in a playful tone, see Thurlow (2016).
12 Like the "Discourse Studies" program at Lancaster University or the "Ideology and Discourse Studies" program at the University of Essex. 13 The conference took place in 2008 in Oldenburg. The conference program can be found here: owww.msw.uni-oldenburg.de/download/Programm_Diskursanalyse-1. pdf4 (accessed on 30.11.2016). 14 In fact we should argue that every scientific method does not simply discover its subject matter, but that it must constitute it. This insight, however, is not part of the sayable for many methodological approaches. 15 As discourse analysts, we are of course not only part of the discursive formation of discourse analysis, but also of the larger formation of academia with all its expectations and power effects. If discourse analytical critique is to retain a chance of being recognized, it will have to stay well-comprehensible despite its heterogeneity. Hence, methodological cogency of single analyses is indeed important. 16 I am grateful to Frieder Vogelmann for having proposed this terminology; see Vogelmann (2017).