Debating as a Deliberative Instrument in Educational Practice

In recent decades, deliberation about public issues has become a central theme in citizenship education. In line with an increasing philosophical and political appreciation of the importance of deliberation within democracy, schools, as training grounds for democratic citizenship, should foster high-level deliberative skills. However, when this insight is translated into practical formats, these formats suffer from a number of shortcomings. Specifically, they can be criticised on philosophical grounds for advantaging select societal groups, and on empirical grounds for facilitating groupthink mechanisms. This paper aims to address these shortcomings by suggesting a role for debating techniques within deliberative education. Because debating stimulates the contestation of diverse opinions, it may counteract the silencing of minority viewpoints and the proliferation of groupthink. At the same time, debating-based formats must be closely regulated in order to not imperil compromise formation. A concrete format is presented that balances these considerations and may therefore contribute to more effective deliberation in the classroom.

. 1 Recent decades have also seen a re-emergence of Dewey's (1916Dewey's ( /1967) emphasis on deliberation as a central theme in educational theory and practice. If a healthy democracy is built on deliberation between citizens, then schools, as sites where students are educated to fulfil a role as citizens, should foster the skills and dispositions necessary for high-quality deliberation (e.g. Peterson 2009, Englund 2011. The question then becomes by which types of educational practice these skills are maximally developed. As multiple theorists have stressed (e.g. Carr and Kemmis 1986), the answer to such questions of educational practice should involve a continuous dialogue between the practical insights of experienced teachers and the findings of theorists -including political philosophers, pedagogists and social psychologists.
A discussion involving such disparate voices can be expected to suffer breakdowns in communication. Indeed, many theoretical discussions of deliberative education are insufficiently concrete to be of much help to educational practitioners. As noted by Peterson (2009), even documents composed by authorities specifically tasked to inform educational practice tend to suffer from this vagueness. For instance, the UK Department of Education prescribes citizenship education for 11-16 year olds, during which students 'experience and evaluate different ways that citizens can act together to solve problems and contribute to society' (Department for Education 2013: 2). However, the exact form that this 'acting together' and 'problem-solving' should take in the classroom is left unspecified.
This paper aims to bridge this gap by discussing both philosophical and empirical constraints on successful deliberation. These constraints are then used to suggest a deliberative debating format, i.e. a concrete proposal of how deliberation may be regimented by drawing upon tools developed within the practice of debating. This format is offered for consideration both to empirical theorists, who might test the hypothesis that the format alleviates some problems with other forms of deliberation, and to teachers, who might add their practical insights to evaluate the format. This paper draws on three sources: first, the philosophical tradition of deliberative democracy, second, the largely empirical study of critical thinking in education, and third, a practical field of argumentation that at first blush seems antithetical to deliberation: the field of debating. The main motivation to adopt elements from debating is that due to its confrontational -or what has been termed 'contestatory' -nature, debating may help to articulate different positions on the issue that is deliberated more fully than when the participants are encouraged to be merely 'constructive' -what has been termed 'appreciative' deliberationor left entirely free in their mode of communication (Curato et al. 2013).
The remainder of this article consists of five sections. In the first section, I introduce the deliberative ideal that political education should aim for. In the second section, I discuss the theoretical and practical shortcomings of existing deliberative methods that are sometimes applied within classrooms. In the third section, I suggest that these shortcomings motivate a shift towards contestatory deliberation. In the fourth section, I relate this shift to the practice of debating and suggest a debating-inspired deliberative format. In the fifth section, I conclude with some observational remarks about this format.

Deliberation as an Educational Tool
With deliberation, I mean any procedure in which a group of participants tries to find a solution for a collective problem by means of mutual discussion, dialogue, or any other form of discourse. Here my focus is on political deliberation, in which the interests or values of different societal groups need to be balanced. Of course, in educational contexts, political deliberation is generally simulated. That is, although the outcome of the procedure does not have any real-world effects, the participants take on the role of deliberating parties and act as if their solution for a real-world problem is desired. This set-up has for instance been applied in Van Laar's (2019) Middle Ground format, which offers a simulation of the practice of deliberative negotiation by having students fulfil the roles of negotiators. 2 With regards to political deliberation, then, what learning outcomes should we expect from education? Based on existing literature on deliberative democracy, we can identify the following conditions for successful deliberation, and associate each with a skill or attitude required on the side of the participants that should thus be a focus of deliberative education (see also Englund 2006, Fishkin 2011: 33-43, 84 − 5, Hess and McAvoy 2015: 1) All different positions inherent in the group of deliberators have been voiced, given a fair hearing by the group, and been taken into consideration (a criterion of diversity). That is, deliberation should serve the following ideal, as formulated by Young (1990: 116): 'we require real participatory structures in which actual people, with their geographical, ethnic [and] gender […] differences, assert their perspectives on social issues within institutions that encourage the representation of their distinct voices'. This requires the attitude of openness towards diverse opinions and values.
2) The deliberative process has consisted mainly of the expression of reason-based arguments, instead of other techniques such as mere assertion, intimidation or the use of fallacies (a criterion of reason-giving) (cf. Habermas 1996: 107 − 11). This requires the skill to support positions with reasons and to critically evaluate such reasons.
3) The procedure has been fruitful, i.e., it has been successful in the task it was set up for (a criterion of effectiveness). In the following, I will consider the task of reaching a solution for a societal problem, where the opinions on how to solve this problem (or, perhaps, whether there is a problem in the first place) originally diverge. This solution can either take the form of a consensus on the best course of action (note that this consensus does not need to extent to the reasons for preferring this course of action; it can be akin to Rawls' [1996] 'overlapping' consensus) or, if consensus turns out to be impossible, a compromise (which still requires a weaker form of consensus on the legitimacy of the different positions, or what Dryzek and Niemeyer [2006] call 'meta-consensus'). This criterion requires the attitude of solution-orientedness.
There have been several findings that political discussion and deliberation in general (i.e. not specifically in educational contexts) can improve these three skills and attitudes. For instance, Mutz (2006) discovered that 'cross-cutting' political exposure increases the political tolerance of Americans who engage in it. That is to say, Americans who regularly discuss political issues with others of opposite political convictions (a relatively small number, Mutz found) are, other things being equal, more likely to support civil liberties for groups they strongly disagree with. Moreover, Fishkin's deliberative polls (Fishkin and Luskin 2005;Fishkin 2011: 25 − 8) have also shown positive impacts on political attitudes. 3 For instance, Fishkin and others have found that participation in deliberative polls creates overall attitude changes in the participants, especially those that gathered new knowledge (Luskin et al. 2002(Luskin et al. , 2007Fishkin 2011: 134 − 43). This suggest that deliberation can stimulate consensusdirected attitude change.
One must be careful, however, in generalising these results to classroom settings, because such settings are different in some key respects from deliberations involving adult participants, such as deliberative polls. As Hess and McAvoy (2015: 6) note, classrooms are 'unusual political spaces' because students are 'captive audiences' that cannot leave classrooms, because teachers have an authority over students that is absent in most other deliberative situations, and because students are also impacted by their caregivers, whose aims may conflict with deliberative education. For these reasons, it is possible that students are liable to disengage from deliberative practices because they feel these practices are forced onto them. Below, we evaluate specific evidence about deliberation in education, to assess to what extent this worry materialises in practice.
To start with the first criterion: Avery et al. (1992) found that explicit attention given to conflicting interests within education increased most students' tolerance (again understood as support for civil liberties for groups with different viewpoints). Similarly, Avery et al. (2013) found that deliberation increases students' self-declared respect for and understanding of the positions of others. Hess and McAvoy (2015: 52 − 7), who have executed the largest study to the effects of political deliberation in classrooms to date, also found that students engaging in deliberation report an increased willingness to consider diverse points of view.
The results for the second criterion are also promising: there is evidence that engaging in classroom discussions can increase secondary school students' capacity to support claims with reasons. Evidence here comes mainly from case studies. For instance, in a series of studies, Felton and Kuhn explored the results of regular discussions with peers on students' argumentation skills. They found strong improvements in both individual (Kuhn et al. 1997) and dialogical argumentation skills (Felton 2004), both of which were quite lacking in students that had not participated in such discussions (Kuhn et al. 1997;Felton and Kuhn 2001;Kuhn 2005: 132 − 47). Specifically, students showed more skill in critically evaluating others' claims and reacting to the arguments of discussion partners in a substantive way. Similar results were found in students who participated in a more extensive curriculum combining individual argumentation, dialogical arguments, and group debates (Kuhn 2005: 152 − 71). Similarly, a study by Hess and Posselt (2002) suggested that discussion of controversial issues improved students' proficiency in discussion.
The third criterion, constructiveness, asks for two related requirements from the side of the participants: an appreciation that deliberation is useful (without which there would be no motivation to bring it to a fruitful conclusion) and the willingness to be convinced by other deliberators (instead of rigidly sticking to one's guns, which would be equally unlikely to lead to a mutually acceptable solution). As for the first attitude, Hess and Posselt (2002) and Hess and McAvoy (2015: 57-60) found that engaging in political discussion increases most students' appreciation of discussion as a valuable endeavour. More generally, classroom discussions were found to increase political activity by Torney-Purta et al. (2001: Chap. 8). With regards to the second attitude, Avery et al. (2013) found attitude change among students engaged in consensus-oriented deliberation, suggesting that students can learn to doxastically respond to different arguments. Together, these findings suggest that the abovementioned features of classrooms do not prevent students from positively engaging with deliberative education.

Challenges to Deliberative Education
Despite these positive results, there are also concerns that some deliberative formats may stimulate harmful political attitudes. Most deliberative formats currently in existence, such as Fishkin's deliberative polls (Fishkin and Luskin 2005;Fishkin 2011: 25 − 8) are relatively unstructured; i.e., they rely on group discussions in which all participants can contribute at will (or choose not to contribute). Educational forms of deliberation often follow suit. For instance, most of the deliberative sessions that were studied by Torney-Purta et al. (2001), Hess (2009), Hess and Posselt (2002) were classroom discussions where students were free to speak up or remain silent at will. (The formats studied by Avery et al. 2013] and Kuhn [2005] and some sessions studied by Hess and McAvoy [2015] were more structured; I will discuss some of these later.) However, deliberation, in this unstructured form, has been problematised by both political philosophers, who have questioned its political legitimacy, and empirical theorists, who have questioned its effectiveness. Although most of these challenges do not specifically aim at deliberation within an educational context, I will argue that many of them likely generalise to such contexts as well.

Philosophical Challenges
Several critics of dominant conceptions of deliberation argue that there is a tension between diversity on the one hand and a focus on reason-based argumentation and consensus on the other. Both Young (1990) and Mouffe (2013) have offered alternatives to deliberative democracy ('politics of difference' and 'agonism', respectively), which they argue are better suited to the demands of pluralism. Although I here have no space to do full justice to Young's and Mouffe's theoretical frameworks, I will shortly discuss their core claims. Young (2000) argues that reasonable argumentation, as prescribed by criterion 2, is a mode of discourse that is familiar to and favoured by only a privileged segment of the population (mostly middle-class, educated, white, heterosexual men), who would therefore have an illegitimate advantage in deliberative settings. Underprivileged groups might be more familiar with other modes of discourse such as emotion, rhetoric and story-telling (Young 2000: 63-77). By positing a standard of reasonableness, the demands of disadvantaged groups can easily be framed as irrational or sentimental (Young 1990: 102 − 16). Young (2000) makes a similar observation about the deliberative emphasis on consensus, which, according to her, quickly leads to the silencing of dissent (cf. Sanders 1997): Too strong a commitment to consensus as a common good can incline some or all to advocate removing difficult issues from discussion for the sake of agreement and preservation of the common good. Sometimes those difficult issues matter deeply to one group because they perceive themselves as suffering a basic injustice, but they are the sources of deep disagreement because others in the society perceive rectifying this alleged injustice as coming at too great a cost to them (Young 2000: 44).
According to this criticism, the demands of disadvantaged deliberators can be framed not only as irrational but also as unconstructive, because frustrating attempts to reach a general consensus.
The underlying idea of both criticisms is that deliberative situations are not 'free', as in Habermas' (1996: 119) ideal of 'communicative freedom', but rather constrained by, and reinforcing, a pre-existing context that contains social inequalities (McLaverty 2014). Young (2003) gives some more concrete examples of how deliberation may reinforce such inequalities. First, many deliberative bodies presuppose certain institutional relations as a given, but these relations themselves may warrant critical enquiry (Young 2003: 112-5). Second, even if the range of possible outcomes is not explicitly limited by the institutional setup, participants may still perceive it as limited by a certain reference frame, which may be inherently unjust or oppressive. Especially when the participants are instructed to reach consensus, some participants may be pushed to accept a consensus that is not in their personal interests (Young 2003: 115-9). Young (2003: 117) gives the following example: because it is common to understand poverty as 'a function of the failure of individuals to develop various skills and capacities necessary for inclusion in modern labour markets', a deliberative session on reducing poverty may lead to all participants discussing ways of instilling more widely these skills and capacities, rather than possible problems with modern labour markets themselves.
To these may be added some more psychological mechanisms that may favour the dominant group: the dominant group members might, due to their privileged position, be more confident, and therefore speak up more, they might be more knowledgeable or rhetorically skilled (after all, social status often correlates with education level) and therefore appear more convincing (cf. Sanders 1997, Young 2000: 57-77, McLaverty 2014. Whether these mechanisms are enacted intentionally or subconsciously, the result of deliberations may not be a solution that can reasonably be accepted by all parties, but rather a false legitimation of an unjust policy or institutional situation. Naturally, these criticisms apply not only in political but also in educational settings, especially because classrooms can be seen as restrictive institutional contexts. Indeed, educational thinkers such as Eisenberg (2006) and Frazer (2006) have applied Young's analysis to argue against the application of impartial standards within education. To the extent that education should empower disadvantaged groups to claim their position within political debates, standards that exclude such groups are highly problematic. Students belonging to such groups might for instance have little experience with explicitly reason-based forms of discourse. In a socially diverse classroom, this would privilege students that are already well-acquainted with argumentative discourse, e.g. because of family background (Hess and McAvoy 2015: 7-8). Hess and McAvoy (2015: 112 − 31) found empirical evidence that in classrooms with socio-economic or racial diversity, students from more privileged groups are often more dominant in discussions. On the other hand, when a classroom is not socially diverse, as might happen in both secondary education (where schools are often segregated by neighbourhood and thereby by social stratus) and higher education (where all participants are already relatively high-educated), this is also problematic, because it threatens to defeat the goal of confronting students with the interests of different social groups. Mouffe's (2000Mouffe's ( , 2005Mouffe's ( , 2013 political framework derives from the realisation that the political is characterised by an irreducible plurality of, often conflicting, principles, which are promoted by different groups of people. As a result, settlements in political issues can never be final; they are always hegemonic in the sense of promoting one perspective, and therefore one group, over others. According to Mouffe, it is crucial to avoid temporary hegemonies becoming embedded in political structures, which would lead to oppression of heterodox perspectives. As a result, she is opposed to (Rawlsian) deliberative democracy, which, presenting consensus as final, ignores these heterodox viewpoints.
For Mouffe, this impossibility of a final consensus is closely connected with a critique of deliberative democracy's emphasis on reasonability. According to her, the emphasis on reason over emotion stems from the liberal misconception that political issues are technical problems that can be solved by applying rational deliberation. Instead, according to Mouffe, political issues centre around irreducible we/they distinctions, which involve emotions of group identification and of opposition to other groups. The goal of politics should therefore not be to eliminate or overcome these emotions, but rather to channel them in such a way that other groups are not seen as enemies to be destroyed but as 'adversaries' to be overcome within a democratic framework (Mouffe 2005).
Mouffe's ideas have also served as the basis for alternatives to deliberation-based approaches to political education. Ruitenberg (2009) argues that an appreciation of the agonistic character of politics calls for modes of education that teach students to view the political arena in terms of political (but not moral) adversaries and channel students' emotions into democratic engagements. This would be contrary to the dispassionate and consensusdirected formats some deliberative theorists propose.
If Young's and Mouffe's worries are valid, an emphasis on reason-giving and consensus might prepare students for a political system that further marginalises underprivileged groups and/or stifles political alternatives. Thus, criterion 1 would be implicated. I believe these worries are valid and will try to develop a solution in Sect. 3 and 4. However, I will first quickly consider, and reject, another possible solution: to drop criteria 2 and 3 altogether, thus abandoning the sphere of deliberative democracy. This strategy has been defended to varying extents by Young (2000), whose 'politics of difference' rests on a form of deliberation not constricted to impartial standards, and by Sanders (1997) and Mouffe (2000), who argue against the practice of deliberation altogether. Again, I do not have the space to discuss Young's and Mouffe's alternative proposals in detail, so let me suffice by some short remarks.
I wish to stress again that the goal of our educational endeavour is to instill in students the capacity to function within a democratic society. Public decision-making is difficult without at least a minimal commitment to reason-giving and consensus. Reason-giving is necessary for public decision-making because, unlike differing emotions, differing reason-based arguments carry in them the potential of resolution. After all, any argument is, by its nature, liable to defeat if more powerful counterarguments are issued (cf. Habermas 1998: 215 − 27). This does not require arguments to be entirely impartial, as Young (2000: 102-7) rightly points out is impossible; they can draw on the particular experiences or interests of the speaker. It also does not require an elimination of emotions, which can inform or constitute reasons for or against a position. But it does mean arguments must present other deliberators with a reason to care for these interests or emotions, otherwise they are unlikely to bring closer an agreement on public action. It may be true that the level of impartiality that a reason-giving standard demands (i.e. reasons need to be acceptable to all deliberating parties) is more readily accessible to some groups. However, the alternative, dropping reason-giving altogether, would give majoritarian groups no grounds to care for minority groups and for that reason may be more exclusionary. Implementing in students an attitude that rejects impartiality altogether would therefore threaten to alienate them from the political arena.
Similarly, I agree that an overly narrow focus on consensus can sometimes lead to deliberators ignoring non-majoritarian viewpoints -more on this in the next section. But without any commitment to consensus, or at least compromise (which requires what Dryzek and Niemeyer [2006] call 'meta-consensus'), deliberating groups with different convictions cannot make policy decisions except by sheer dominance of the majority. It is unclear, for instance, how Mouffe's hegemonic struggle serves to implement marginalised groups' interests in policies at points where large hegemonic turnarounds are not available.
In fact, what Mouffe appears to object to in deliberative democracy is not its focus on consensus-formation as such, but rather the tendency to treat consensus as final rather than contingent or 'hegemonic'. However, this tendency, which is present specifically in Rawlsian deliberative theory, may be reduced when students are instructed to regard consensus as the outcome of an ongoing struggle between different positions. It is exactly such a deviation from Rawlsian deliberative theory that this paper advances.
Moreover, I do not mean to suggest that (post-Rawlsian) deliberative democracy should be the sole basis of citizenship education. Indeed, Ruitenberg's (2009) three proposals for agonistic education -educating political emotions, teaching the difference between the political and the moral, and instilling an understanding of the history of current political struggles -could be implemented in parallel to my proposal. My only claim is that these alternative strategies are not in themselves sufficient to prepare students for democratic engagement, since they do not train them to deal with situations where concrete policy decisions are required.
I do not expect these considerations to convince readers who are deeply sceptical of the viability of deliberative democracy as an educational framework. However, to those readers, the remainder of the paper may still be of theoretical value. By showing the extent to which deliberative education can incorporate forms of dissent, it may clarify exactly how deep (or shallow) the limits between deliberative and agonistic or difference-based education are, and thereby sharpen the debate.

Empirical Challenges
The empirical challenge focuses on the fact that, in actual practice, participants in deliberative settings often do not engage in reason-based deliberation at all, but rather rely on cognitive heuristics that may lead them to irrational conclusions (Ryfe 2005). Many of these heuristics are inherent in all types of reasoning, but some, known as 'groupthink', have been shown to be unique to or worsened in deliberative contexts. 'Groupthink' is a catchall term, introduced by Janis (1982), describing social and psychological mechanisms that influence the outcomes of deliberative procedures independently of rational considerations. Basing themselves on a number of social experiments, Sunstein and Hastie (2014) observe that deliberating groups often reach suboptimal conclusions -conclusions worse than when voting had been used instead of deliberation -because members are inclined to not voice information or opinions that go against the majority. According to Sunstein and Hastie, there are two main reasons for this: either deliberators may sincerely think such information or opinions to be mistaken because the majority disagrees with them, or deliberators may think it socially unwanted to jeopardise consensus and thus not disclose their information or opinions out of concern with their social status (Sunstein and Hastie 2014: 21 − 4, 33 − 7).
Several adverse mechanisms may arise from these impulses, of which I will give two examples. First, Sunstein and Hastie (2014: 57-76) (basing themselves on Anderson and Holt [1997], and Hung and Plott [2001], amongst others), describe how groups may fall victim to so-called 'informational cascades'. These effects occur when the first member or the first few members of a deliberating group utter information and/or opinions that point in a certain direction. In this case, subsequent members are likely to go along with this direction because of the two mechanisms mentioned above. In this way, speaking order might influence the outcome of the deliberation independently of whether or not the first speaker was correct. Second, Sunstein and Hastie (2014: 77-88) describe how deliberation may lead to group polarisation (see also Sunstein [2002]). Polarisation is the process in which a group that is slightly leaning towards one side of an issue commences to lean to that side more radically during the course of the deliberation. Again, this process is likely to happen independently of whether the majority position is actually true or reasonable.
Of course, the worries about groupthink can reinforce the worries about social inequalities, since (as noted by Sunstein and Hastie [2014: 41 − 2]) socially advantaged participants are relatively likely to instigate the groupthink phenomena. These participants' relative familiarity with deliberative modes of communication means they are often more likely to speak up, thereby having a larger chance to initiate informational cascades, set the stage for polarisation, etc.
It is unlikely that educational settings are immune to groupthink phenomena. In fact, Johnson and Weaver (1992) have argued that the enabling conditions of groupthink are especially present in classrooms, because of a hierarchical atmosphere where students are incentivised to merely listen to the teacher and are pressured not to take too critical an attitude. Johnson and Weaver recommend discussion as a potential solution for groupthink. However, classroom discussion itself may still fall prey to groupthink, because students may feel pressured not to challenge the utterances of their peers that have spoken before, especially when these are popular or confident. Again, this worry is most pressing when a classroom is ideologically homogeneous, or when proponents of a certain ideology feel more comfortable participating in discussions. Hess and McAvoy (2015: 148 − 51) present some qualitative evidence that polarisation is a threat in classrooms that are ideologically highly homogeneous. This puts a burden on educational theorists to design forms of deliberation that minimise the risk of groupthink.

Towards Contestatory Deliberative Education
How can we square these challenges to the success of deliberation with the empirical evidence surveyed in Sect. 1, which suggested that deliberative education can be successful on all three criteria? There are two important points to make here. First, the successful results in Sect. 1 may be very sensitive to the circumstances of particular classrooms. These include the level of diversity, the skill of the teacher to incentivise the participation of students from a variety of backgrounds (Maurissen et al. 2018), and the exact deliberative format used (see also Kuhn 2005). Note in particular that most of the deliberative or discussion formats touched upon in Sect. 1 did not explicitly push students to reach a compromise (with the exception of Johnson and Johnson 1988), even though we identified this as an important feature of political deliberation that students should be exposed to. However, there is empirical evidence that adverse social mechanisms increase when consensus is more clearly desired (De Grada et al. 1999). Therefore, it is difficult to say whether the positive results would still stand when there is a pressure to reach a consensus.
Second, the results warn us against interpreting quantitative results of deliberation as proof of the qualitative success of the deliberative procedure itself. As for criterion 1, for instance, the results discussed in Sect. 1 only show that students tend to become more tolerant towards differing viewpoints after deliberation, but not why this happened. In particular, privileged groups may become more tolerant towards marginalised groups because the former better understand the latter's concerns, but also because the latter are one-directionally pushed towards the dominant group's position, thereby being perceived as less of a threat by the privileged group. Obviously, such a mechanism, although producing tolerance, would not meet the demands of equality and diversity.
In summary, there is a danger that unstructured deliberative formats disincentivise true discussion, especially when consensus is the explicit goal (cf. Curato et al. 2013). The question then becomes: by what other formats can the probability that deliberative education has the desired effects, both in procedure and in outcome, be maximised? The answer to this question has two parts: first, to design a variety of formats that are constrained by previous philosophical and empirical results, and second, to empirically assess those formats. The remainder of this paper contributes to the first stage of format design. Specifically, we attempt to design a format that both asks for consensus, and incentivises students to explore a variety of opinions without immediately defaulting to the majority line. Whether it succeeds in mitigating the problems mentioned in Sect. 2 is an empirical question up to future work.
One way to decrease the dangers of groupthink and marginalisation may be to place explicit emphasis on confrontation and dissent within the format. This idea is not new: Pettit (1997: 187 − 90), in line with his ideal of a 'contestatory democracy', has argued for a 'debate-based decision making' process in which citizens can contest political decisions. Relatedly, Bächtiger (2011) has argued in favour of 'contestatory' as opposed to 'conversational' deliberation. He characterises contestatory deliberation as the confrontation of different arguments through three communicative modes: questioning, or interrogating and cross-examining others' arguments, disputing, or challenging others' arguments, and insisting, or finding new ways to argue for one's position in the face of questioning and disputing (Bächtiger 2011: 6-7). Similarly, Curato et al. (2013) have perceived a deficit in critical examination in non-confrontational deliberative practices (they focussed on the 2009 Aus-tralian 'Citizen's Parliament', which is very similar in setup to deliberative polls) which, according to them, should be mended by allowing contestatory instead of only 'appreciative' inquiry. Within education as well, some authors have suggested that contestatory deliberation is best suited to teach political skills to students (Peterson 2009).
The concept of contestatory deliberation can be seen as a means to introduce an agonistic (in Mouffe's sense) element into deliberative democracy and thereby create more awareness of the hegemonic nature of political decisions. To be sure, this concept does not meet Mouffe's criteria for a fully agonistic democracy because it still assumes that some political issues can be solved by reasonable agreement. However, when a compromise has been reached through adversarial modes of communication, participants are less likely to view it as final or incontestable.
The question then arises how a focus on contestation can be implemented in practice, while simultaneously remaining attentive to the goal of consensus-formation. Some existing formats for deliberative education attempt to combine these requirements by distinguishing between a contestation and a consensus-formation stage. The most notable of these is Structured Academic Controversy (SAC) Johnson 1988, 1993). In this format, two teams of two students first argue in favour or against a motion, attempting to refute each other's arguments, then switch positions to argue for the opposite view, and finally attempt to reach a consensus. Although SAC goes a long way in combining contestatory and consensus-directed aims, it makes two design choices allowing for the exploration of alternatives. First, the adjective 'structured' in 'SAC' mostly concerns the distinction between controversy and consensus-seeking stages, but not the implementation of the controversy stage itself. That is, students are told to convince each other but can choose how to fill the time allocated for doing so. This makes it possible for students to omit certain arguments for their position, if they are quickly convinced by the other team's arguments. Second, the diversity of arguments is limited because there is only one team representing each side, while many political issues allow for a diversity of arguments on both sides. If the students in a particular classroom are aware of a larger diversity of arguments, it somewhat restricts potential learning outcomes to only expose students to three peers in a small group setting. It is therefore worthwhile to design a format that structures the contestation of different positions within a larger group.
To achieve this, I suggest, proponents of deliberative education can draw inspiration from another field of discussion that is often used in education but generally seen as disjunct from deliberation, i.e. the field of debating. Debating formats or debate-like formats, such as moot courts, have been studied in the context of political education (e.g. Hess andMcAvoy 2015: 85-131, Kuhn 2005). However, unlike SAC, such formats usually not culminate in a consensus-formation stage. For that reason, the next section and Appendix 1 sketch a more explicitly deliberative debate format.

Debating and its Relevance for Deliberation
Debating is quite common in education, especially in the Anglo-Saxon world. Many schools and universities have debating teams that compete against each other in regional, national, or international competitions. However, the debating styles used in these competitions seemingly have little to do with deliberation. In contrast to traditional deliberative formats, debating formats are competitive in nature: the idea is that one of the participating parties (usually called 'teams') wins the debate. This is intuitively opposed to deliberation, where, if there is a goal at all, it is usually consensus or compromise (consensus and compromise can be described as 'win-win' situations, i.e., situations in which there is not one winner but everyone wins to the same extent). Therefore, most authors on deliberative theory either do not discuss debate at all, or present deliberation and debate as disjoint activities (e.g. Avery et al. 2013) (although see Manin [2005] for a notable exception).
On the other hand, as we have seen, a format that has no competitive mechanisms built in runs the risk of not extracting the full range of positions and arguments relating to the issue at hand and creating a false sense of full agreement. How do we navigate our way between having too much and too little contestation? My proposed solution lies in the development of a 'non-competitive debating' format. 4 Non-competitive debating is a style of debating where participants use arguments to support their own position and refute others' positions, but only with the ultimate goal of determining which arguments withstand scrutiny and which do not, so that eventually an informed consensus or compromise can be reached. In other words, whereas the mode of communication is contestatory, the purpose is conciliatory (cf. Bächtiger 2011). In non-competitive debating, contrary to 'pure' debating, the competitive mode of debating is applied only in the full knowledge that it is a means to eventual consensus-formation.
I want to single out two important aspects of the constructive attitude that is required for debating to stimulate consensus-formation. The first is that the participants try to argue on grounds that potentially appeal to other participants who do not already share their position. 5 This idea has been suggested by Habermas (1996: 108) and more fully by Gutmann and Thompson (1996: 55-69), who call this mode of argumentation reciprocity. In a reciprocal mode of discussion, the focus is on mutually acceptable reasons. When such reasons are unavailable (i.e. when there is moral disagreement on such a principled level that there are no common principles to fall back on), the focus should be on moral accommodation (Gutmann and Thompson 1996: 73-94) or meta-consensus (Dryzek and Niemeyer 2006), both of which require acceptance of the legitimacy of everyone's values, beliefs and preferences.
The second aspect of the constructive attitude is that participants should aim not only to change the other parties' opinions, but also to transform their own opinions and preferences when the rational pressure to do so is sufficiently strong. That is, they should play by the rules of critical discussion, which include giving up convictions when (but only when) it is no longer reasonable to hold onto them (cf. Bächtiger 2011).
Reciprocity and openness to self-transformation can find their way into a debating format, but only under several conditions: • The participants must be informed from the start that consensus is the main goal of the debate.
• The participants must be instructed to come up with reason-based arguments that potentially appeal to individuals who do not already share their position. • The participants must be encouraged to withdraw some of their arguments when, but only when, compelling critique has been issued against them.
I am not the first to suggest such a deliberative function for debating. In particular, the 'British parliamentary' (BP) debating format, widely used in university debating, has sometimes been suggested for deliberative purposes. The BP format (see e.g. Meany and Shuster 2002) is distinguished from other formats by having four teams debating a single motion, two in favour (the 'proposition' or 'government') and two against (the 'opposition'). Each team is given two speeches to bring arguments and react to the arguments of the other teams. After the debate, a jury ranks the teams first to fourth, based on the credibility and weight of their arguments. Franke (2016) has suggested a deliberative function for BP debating, specifically because there are two teams on each side. According to Franke, BP debaters, knowing that they do not only have to defeat the other side but also show to be more relevant than the other team on the same side, will be more likely to explore all considerations relevant to the issue, instead of just focussing on a limited set of arguments meant to defeat the opponents. Indeed, structuring a deliberative format as a BP-like debate would enable multiple stances towards an issue to be voiced, without the issue becoming a binary 'yes-no' decision.
To cater to the demands of deliberative practice, the BP format needs some alteration. First, to stimulate diversity, both teams on each side should not be required to agree with each other, as is the case in standard BP debating, but should be allowed to voice conflicting arguments. Second, to allow more than eight participants per deliberative session, all four teams can be enlarged to contain members that do not directly participate in the debate. To give a role to these additional team members, the format should include deliberative phases where teams deliberate internally about their argumentative strategy. Third, to stimulate interaction of arguments, the format should give every team the option to respond to every other team, which is only very limitedly possible in the BP format, where both first half teams give their speeches before the second half of the debate commences. Fourth, to make the debate truly non-competitive, no winner should be appointed in the end; rather, the results of the debate should be used by the participants themselves to construct a workable consensus. Finally, to make the format more accessible for students who have no background in debating, I suggest shortening the speaking times from the standard 5 or 7 min to 2.5 min per speaker. These changes result in the format presented in Appendix 1.
The format leaves some implementation details to the judgement of the teacher or moderator. In particular, the format does not specify whether or not briefing materials should be assigned to students in advance of the discussion. On the one hand, such readings may make the discussion more informed, and, if multiple perspectives are equally accessible, may help to introduce students to a variety of viewpoints. On the other hand, advance briefing materials may make the discussion less accessible to some students, and, in cases where it is difficult to find or construct balanced readings, may delimit the discussion at the outset. Because the balance of these considerations likely depends on the topic and group of students, teachers are best positioned to decide in individual instances whether to assign advance readings, and if so, which readings best serve the purposes of the discussion.
How, then, does non-competitive debating in this format respond to the worries about deliberation that were discussed in Sect. 2? Bächtiger's short summary of the merits of contestatory deliberation can serve as a guide here: By providing a rigid and thorough evaluation of an issue at hand, contestatory deliberation can guide discourse participants to the "correct answer" and lead to clean preference changes based on the true merit and downside of arguments and demands, as well as induce truthfulness. Moreover, by unravelling and subverting dominant frames, contestatory deliberation can also help to include the demands of disadvantaged groups. In sum, contestatory deliberation shifts deliberation's frequent focus on a collaborative and "play-well with others"-approach to a conception of deliberation which fully re-values adversarial and confrontational practices and turns deliberation into a critical and rigid inquiry (Bächtiger 2011: 38).
First, the format encourages a diversity of positions to be brought into the debate. Because there are two sides in the debate and each side is split into two teams, in effect, four different positions are represented. The 'forced' diversity may decrease the probability that all parties in the deliberative session are constrained by institutional presuppositions. For instance, to build on one of Young's most frequent examples (e.g. Young 2003), when the discussion topic would be whether unemployment benefits should be tied to employability tests, the opposing teams may either argue that there are other better ways to gain employability, or that the whole concept of employability is flawed and unjustly gives people individual responsibility over whether or not they can find a job. Given that the opposing teams need to distinguish themselves from each other, the format aims to incentivise teams to think 'out of the box' and consider the second option. Moreover, because dissent is explicitly built into the format, participants may be less liable to consider other participants obstructive for voicing dissent, and thus to disregard challenging positions. This alleviates the philosophical worries discussed in Sect. 2.1. Of course not all of Young's worries can be alleviated in this way. Participants who are relatively well-versed in argumentation are still at a relative advantage. However, their having the same speaking time as other participants minimises this advantage as much as possible.
Second, because the format gives all four teams equal speaking times and explicitly instructs them to confront each other, the expression of different positions is stimulated. This design choice is meant to counteract groupthink processes leading to premature consensus. The debating format aims to stimulate participants to challenge others' positions through Bächtiger's (2011) modes of questioning and disputing, and to insist upon their positions in the face of such challenges. The goal of this aim is to enable a fair confrontation of arguments, rather than a mere consent to arguments already stated. What is more, because there is a safety mechanism to simulate dissent even when there is no actual dissent (through having participants argue as if they were in favour or against a certain motion), the format encourages participants to see multiple sides of the issue, even when the group itself is rather homogeneous. In either case, it can reasonably be hypothesised that the format reduces the risk that minority positions are not expressed through silencing or self-silencing.

Concluding Remarks
To conclude, I give some short remarks on the advantages and disadvantages of the presented format, on the basis of a number of trials carried out with several groups of undergraduate university students. These remarks are not meant as sufficient empirical evidence for or against the effectiveness of the format, since the sample pool is relatively small and only includes university students, who most likely react differently to the format than secondary school students. Instead, I offer these observations as informal considerations that might serve as grounds for more rigorous empirical testing in future work.
In general, the sessions were quite successful on the three criteria identified in Sect. 1. As for the criterion of diversity, most sessions featured a rather diverse range of arguments, both when comparing both sides and when comparing both teams on each side. Even when students were sorted into a team that did not match their initial preference, this was generally no obstacle for finding reasonable arguments for their assigned position. In one trial of the format, which featured the motion that the EU should cancel all refugee deals with non-EU countries, the students were almost unanimously against the motion. Still, the artificial creation of two teams in support of the motion led to a wide exposure of arguments that third-country refugee deals violate the rights of refugees, which eventually found their way into the consensus. Moreover, the requirement of finding new arguments different from those already voiced by the other teams stimulated the second pro and contra teams to consider stakeholders that had not yet been discussed. For instance, in a deliberation about the motion that the Dutch government should institute referenda, the second contra team focused on the danger of referenda for cultural minorities, in contrast to the first contra team that had focused on general concerns about the lack of expertise of the general population. This illustrates how, even in the absence of strong diversity in the backgrounds of students, this format may still stimulate discussion of a diverse array of interests.
Second, the criterion of reason-giving was also generally met. The explicit instruction to each team to write down arguments for their positions appeared to stimulate a tendency to prioritise widely acceptable reasons over partisan appeals or fallacies. Of course, this reasonable attitude may be explained by the setting of the trials, which included only university students -who are probably already trained to engage in argumentative discourse -and in which none of the students seemed to have a strong emotional connection with the topic. Still, the argumentative focus of the format seems conductive to a productive discussion.
Third, the criterion of effectiveness was met in all but one trial in the sense that a compromise was eventually approved. These were encouraging results, since one worry about debating-inspired formats is that the contestatory nature of such formats only serves to drive both sides apart and therefore makes a compromise more difficult to reach. It should be noted, however, that some of the proposals were not very specific. For instance, the winning proposal in the abovementioned deliberation about the EU's refugee policy stipulated to only have 'humanitarian' deals with third countries, without defining what was meant by 'humanitarian'. If the moderators attach importance to specificity, they might actively push for more detail in such instances. Moreover, in all the cases where a final compromise was reached, a clear majority of the participating students afterwards indicated that they regarded the compromise as a good accommodation of all teams' most important arguments and that they thought the debating setup helped to make the deliberation more structured.
However, some of the students also stood critically towards some of the format's features. The main point of criticism -which may also be the main disadvantage of this format -was that the strong regulation of the deliberation restricted the possibility to react to the other teams. Some students would have preferred an open discussion round in which everyone -rather than only a single spokesperson from each team -could participate. Although I opted against including such a round in the format because it may lead to some teams being overshadowed -for instance because the members are less confident or because they find it difficult to go against a perceived majority position -future deliberative formats may replace the second debating round in the above format by an open discussion round in which anyone can speak up without restrictions.
Other potential downsides of the format were not mentioned by the students but can be estimated from the concerns mentioned in Sect. 2. The fact that the format's setup is dependent on a motion may still restrict students in the range of arguments they feel are appropriate. For instance, if the motion states that the IMF should pursue a certain policy, it may be difficult to argue from the position that the IMF should not exist to begin with. Moreover, the format may be less appropriate to deliberate about topics that participants have strong emotional feelings about. Because participants are tasked to explicitly target each other's reasons, if these reasons are felt to matter on a personal level, such challenges may be perceived as hurtful. Furthermore, the group size required for this format (16-20 participants) may make it more difficult for less outspoken students to contribute. Although the intermediary phases of within-team discussion may compensate for this a little, formats with smaller groups, like SAC, may still be preferable when students have little argumentative self-confidence.
It is of course an empirical question to what extent the presented format succeeds in its task -stimulating students to deliberate political issues without losing sight of multiple perspectives on these issues, and reaching an optimal compromise -and to what extent it falls prey to these potential shortcomings. Answering such questions falls outside of the scope of this paper. The claim made here is twofold. First, on a conceptual level, the format serves to show that deliberative education can incorporate an explicit focus on dissent. Second, regarding the question to what extent this dissent will actually manifest in the application of the format, this paper posits an informed hypothesis that the format will perform better than unstructured deliberative formats, or binary debate formats, but further testing must evaluate this hypothesis.
Even if this hypothesis were to turn out correct, it of course remains the case that for other purposes that do not directly involve consensus, for instance exploring emotionally charged topics, other formats may be more appropriate. This of course should come as no surprise -deliberation is a wide and heterogeneous mode of communication, and it is unlikely that all its features will be perfected within one single format. Hopefully, the format presented in this article, and the theoretical considerations behind it, contribute to the ongoing enterprise of exploring the many facets of deliberation in citizenship education.

description is directed both at moderators and at participants; in implementation it could be handed out to the participants or read out by the moderators (with the exception of the italic sections directed exclusively at the moderators). 6
Present: 1-2 moderators, 16-20 participants. Introduction.
In the following procedure, we imagine to be a policy committee charged with the task of advising an executive body (e.g. a government, an international organisation, or a company board) about a socially relevant policy issue. Because there are large differences in opinion between different committee members, the committee must find a way to work towards a common decision. For this purpose, the different members will engage in a debate.
In this procedure, the group will be divided into four teams. Each team will present its own position towards a policy motion, defend this position in a debate, and react to the positions of the other teams. Then each team will propose its own policy on the issue under discussion. There will be a vote on the proposals, and the proposal with most votes, but at least 75%, will be accepted. If no proposal receives sufficient votes, the committee has failed in its task of issuing an advice. As a result, all members will suffer reputational damage, and the necessary policy will fail to happen, with negative social and/or political consequences.
It is important to keep in mind that the ultimate goal is to reach a supermajority for a decision. The focus of the debate should not be to 'defeat' the other teams, but to prepare the way for an eventual policy proposal that can count on widespread support. The debate can contribute to this by convincing other teams, by allowing other teams to convince you, or by getting to understand what the other teams find important. When attempting to convince the other teams, it is a good strategy to tailor arguments to their values and priorities, rather than your own. E.g., religious arguments will do little to convince teams arguing on secular grounds.
Because all teams consist of more members than can participate in the debate itself, in between the debating rounds there will be consultation rounds, in which the four teams internally discuss the strategy for the debate. In these consultation rounds, it is important to give all team members a say, especially those members that do not participate in the debate itself. Be sure to respect everyone's speaking time and take all input seriously.
Below follows a detailed description of the procedure. Note to the moderators: the motion is up to the moderators. The motion should be a policy motion, i.e. it should call for a certain body to implement a certain policy. This body can vary from national or regional governments to international organisation to company boards. Examples are: 'The government should introduce binding referenda about important legal bills', or 'The IMF should only provide loans to countries that have a democratically elected government'. It is also important to formulate the motion as succinctly as possible; it is up to the participants to develop a concrete implementation. b) The participants are divided into two groups, one in favour and one against. To form these groups, everyone votes either for or against the motion according to personal conviction. If both groups are of roughly equal size, everyone is sorted into the group that 6 Multiple elements of the following format -especially the level of structuring, the use of posters for 'bookkeeping', the types of questions (dis)allowed in the various rounds, and the option to dissolve the issue by means of a compromise -are adapted from Van Laar's (2019) Middle Ground format. matches their vote. If not, some members of the majority group are randomly allocated to the minority group so that both groups become comparably large. The groups go to different parts of the room. (+/-3 min) c) Within each group, every member gets exactly 1 min to explain the most important reasons behind their position to the other group members. When you are not in the group of your choice, imagine what someone who does have this position would say. The time is kept by one of the group members. The goal of this phase is to get clear on potential differences of position within the group. It may be that different subgroups have very different reasons to be for or against the motion. (+/-10 min) d) Each group has the option to split up into two teams on the basis of the differences that have emerged in the previous phase. When there turn out to be no real differences, the group is divided randomly into two teams of roughly equal size. The newly formed teams go to different parts of the room. (+/-3 min) Round 2: Debate: Constructive Round (+/-20 min). a) There are now four teams, which we will call 'pro 1', 'pro 2', 'contra 1' and 'contra 2'. Each team confers internally to develop the best arguments for its position. There are no limitations on the arguments allowed: ethical, practical, judicial, or economic arguments are all permitted. However, if you are in pro 2 or contra 2, keep in mind that you are expected to bring arguments that are different from the first team, so make sure that you do not only consider the most obvious arguments.
The pro teams also have to come up with a concrete plan to implement the policy recommended in the motion. All teams write down their most important arguments in favour or against the motion on a poster as succinctly as possible; the pro teams also write down their plans. The posters are attached to the wall. Each team then chooses a spokesperson who will speak during the constructive round of the debate. (+/-10 min) b) In turn, each spokesperson stands up next to their team's argumentation poster and has 2.5 min to elaborate on the arguments. The speaking order is: pro 1, contra 1, pro 2, contra 2. The first pro team should present a concrete plan to implement the motion's policy, and argue why this plan is necessary. The first contra team should argue why this plan is a bad and/or unnecessary idea. The second pro team can either adopt the first pro team's plan, or present its own plan, which the second contra team is then supposed to react to. It is important that the second team of each side brings in new arguments instead of merely repeating the arguments from the first half of the debate. During all speeches, the members of all other teams have the possibility to ask short questions (15 s max). They can signal this by standing up; the speaker may then decide whether or not to accept the question. The questions can only be clarificatory, and may not contain any criticism of the speaker's arguments. (+/-10 min) Round 3: Debate: Reaction Round (+/-20 min). a) All teams reassemble to discuss the arguments brought in by the other teams and to prepare a reaction speech. This reaction speech is meant to react to the arguments of the other teams (most importantly the two teams that disagree with you on the motion). All teams may decide for themselves how to use the reaction speech. You could for instance use the speech for rebutting counterarguments, but also for signalling that you are willing to make concessions, which may aid the decision-making phase later on.
Each team also appoints a spokesperson who will deliver the reaction speech; this may not be the same person who was spokesperson in the previous round. (+/-10 min) b) Again, each spokesperson has 2.5 min to deliver their speech. The speaking order is as follows: pro 2, contra 2, pro 1, contra 1. During the reaction speeches it is again allowed to ask short questions of 15 s max; these may now also be critical questions. (+/-10 min) Round 4: Voting (+/-20 min). a) Each team can draft a new policy proposal. In this proposal you can adjust one of the pro teams' plan such that the concerns voiced by the other teams are met. Now, the motion is no longer just a yes-or-no-matter, but an open array in which different positions are possible. It is not possible to negotiate about the proposals once they have been drafted, so be sure to carefully consider all the teams' opinions, as they have been voiced in the debate. Each team succinctly writes down its proposal on a new poster, which is attached to the wall. (+/-10 min) b) All teams shortly (1-2 min) present their proposal in random order. The other teams may ask short, clarificatory questions when some point of the proposal is unclear, but they cannot voice any criticism. (+/-7 min) c) All proposals are voted on in order of presentation. The votes are anonymous: everyone writes down 'yes' or 'no' for each proposal on a ballot. There is no limit to the number of times you can write down 'yes'. In voting, consider if a proposal accommodates your initial reasons and opinions -unless of course you have been convinced by the debate that these reasons are less important than you originally thought. Please keep in mind that it is not reasonable to expect that a proposal meets all your concerns. The moderators count the number of yes-votes for each proposal. When none of the proposals receives 75% or more of the votes, the attempt to come to a decision has failed; otherwise the proposal with most votes wins. In the event of a tie, a second voting round takes place in which every participant can vote once for one of the most popular proposals. When there is still a tie, one of the tied proposals will be selected by lot. (+/-3 minutes)