1. Introduction
The ideas that motivate this work come from an article written in Informal Logic, Deborah Orr’s “Just the facts ma’am: informal logic, gender and pedagogy” (1989). In this article Orr states that she followed the Informal Logic movement throughout the 1980s, and at first she excitedly implemented the tools of Informal Logic in her classroom teaching. However, Orr comes to find that “the toolbox is less than fully equipped” and so she makes “some suggestions as to where informal logic might look to enlarge its stock of implements” (p. 1). While Orr’s main focus comes from a pedagogical and not a field concern per se, Orr’s contribution to Informal Logic, that its toolbox is in need of feminist tools, is nonetheless an important consideration for the field. In this paper I maintain that her suggestion, that Informal Logicians begin to incorporate feminist concerns within their theories and models of argumentation, is still largely unexplored. I begin by summarizing Orr’s contribution to Informal Logic. I articulate her criticism of Informal Logic further with feminist epistemology and discuss to what extent Informal Logic considers such concerns. Then I turn to Pragma-Dialectics for a better response to addressing such feminist epistemological concerns in argumentation. My investigation shows that the Pragma-Dialectical model provides an adequate foundation for addressing feminist concerns, but it does not provide a very feminist-friendly model of argumentation as it stands.

2. Deborah Orr’s “Just the facts ma’am: informal logic, gender and pedagogy”
Orr does not hold any one system of reasoning, be it Formal Logic, Informal Logic, Pragma-Dialectics, etc., to be faulty or negligent. Rather, it is just a single area of reasoning equipped with a series of the same tool type. Orr thus likens the field of Informal Logic to a toolbox. That is, a good toolbox is equipped with sets of screwdrivers, nails, wrenches, and so on, and each of these series of tools is a part of a single tool type. Having a toolbox with a dozen different screwdrivers does not give a workperson a wide range of tools to access when building her project, since she has only a series of that one tool. A good toolbox has a variety of tools that can be accessed. Thus, if a workperson is building a bookshelf, and she needs to nail the shelf onto the frame, it does not matter how many types of screwdrivers are available. A hammer is needed to complete the project. Orr’s point is that the particular criteria Informal Logic relies on in its approach to argumentation is much like the situation of a toolbox with only a set of screwdrivers. Informal Logic provides a single tool type with a possibility of different rules to work with, just as the workperson has different types of screwdrivers to access. Orr, however, argues that a possibility of multiple tools of reasoning should exist for Informal Logic, so that the Informal Logician has a fuller toolbox (p. 1).
To concretize this point, consider the two argumentative contexts and rationales that follow. First, imagine you are vying for a particular position in a job interview, what you likely need to do is develop reasoning tactics that strongly support you getting the position. It does not matter how many Formal Logic formulas you have at your disposal (modus ponens, modus tollens, etc.), since even if you plug in the appropriate criteria, using them will probably not help you get the position. One of several different strategies might be to develop descriptive analogies between a previous job and the current position being interviewed instead. This could help develop a strong argument in your favour within the interview. This example, it seems to me, is not a contentious example. In fact, Informal Logic recognizes and is a response to the limited nature of Formal Logic (see Johnson & Blair, 2006, p. xiii; Groarke & Tindale, 2004, p. xv; Govier, 2001, p. ix). But, just as Formal Logic is not really helpful in the context of argumentation in job interviews, Informal Logic has its limitations as well. Consider another example: you are in the midst of a custody battle with your former partner. Both of you need to provide solid arguments for sole custody rights. Important to the field of Informal Logic is reasoning devoid of fallacies. And, let’s say you make solid arguments that are not fallacious, ones that meet the requirements for sufficiency. This might not amount to much in your means to resolving, or winning, the custody battle. An alternative helpful strategy could be to focus on finding common ground with the other party instead (see Gilbert, 1997, pp. 111-112), whether or not fallacies are committed.

The above two examples demonstrate situations where acceptable means of reasoning within a particular framework, Formal Logic and Informal Logic in these cases, might not be as advantageous as other, also reasonable, means of presenting arguments. The list of exceptions and different alternatives in particular contexts is much more extensive, and while it would be easy enough to say that some arguments do not adhere to good standards of reasoning, this runs the risk of sometimes incorrectly placing blame on real interlocutors’ methods of communicating arguments instead of theories or models that might not be comprehensive enough. For this reason, Orr proposes that additional sets of argumentation tools need to be introduced and acknowledged as acceptable means of arguing in Informal Logic.
Orr focuses on bringing attention to “the feminine style” of reasoning, one in contrast with the dominant, masculine style of rationality (p. 2). I refer to Carol Gilligan’s terms for these two styles of reasoning: the ethic of care and the ethic of justice (1993).[i] According to Orr, Informal Logic upholds the values associated with the ethic of justice, and she maintains that the ethic of care, and its means to reasoning, is largely ignored. The tools that need to be introduced to Informal Logic then are those that align with the ethic of care. A brief explanation of each style of reasoning, based on the empirical research of moral reasoning by Gilligan, follows. An ethic of justice stems from the notion of equality among people. It values universal, objective knowledge. Its nature is generally adversarial and divisive; with it comes an air of detachedness. In contrast, an ethic of care stems from nonviolence. It values particular knowledge and relationships. It considers the context of a situation and interlocutors’ narratives. Rather than promoting an adversarial nature, implicit within its values is the maintenance of good relationships between individuals. Orr notes that it has been referred to as an “indirect,” “empathic,” and even a “narrative mode” (p. 8).[ii]
Subscribing to a Wittgensteinian notion, that we must look at what people actually do when they reason, Orr argues that reasoning in line with an ethic of care deserves more attention (p. 5). While she voices the Informal Logic movement as a liberating advance, in terms of its challenges to Formal Logic, criteria in line with the ethic of care have yet to be seriously addressed. Orr writes “the lesson in this for those involved in informal logic . . . is that for the full range of human thinking to develop we must collectively recognize the validity of modes of thinking other than the dominant masculine strain” (what I refer to as the ethic of justice) “and actively foster their development” (p. 9). And, so, while Orr does not mention any concrete tool or reasoning pattern that needs implementation within Informal Logic, I continue in the same vein and demonstrate with examples of reasoning, notions involved with the ethic of care that get cast aside as not relevant to argumentation’s concerns according to Informal Logic.

3. What is meant by “feminism”?
Feminism comes with many connotations and interpretations. Its use here is quite specific: I refer to feminist epistemology. While feminist epistemological approaches are broad, and not simply characterized, a shared concern between them is that knowers are particular and concrete, rather than abstract and universalizable. All knowers are part of a larger social network, influenced by historical and cultural factors, as well as the intricacies and intersubjectivities of any given context. And, while feminist epistemologists use mostly gender as a category of epistemic analysis and reconstruction, more generally there is an awareness of the analysis and reconstruction of other subjugated categories (race, class, and so on). “Othered” positions, those in positions of subjugation, are thus considered in the construction of knowledge, which directly opposes theories or models of argumentation that align with the ethic of justice, as there is an implicit awareness within feminist epistemology that not everyone is equal and universalizable.
The ideas that initiate scientific studies, hypotheses, questions asked in public surveys, data investigated, and so on, are methods questioned and critiqued by feminist epistemologists of science – much of their work demonstrates the limited nature of the scientific process.[iii] Similarly, the knowledge that drives decision-making within argumentation can also undergo feminist critical evaluation. For instance, one can investigate whether there are inherent values at work within Informal Logic that limit the realm of possible tools of use and evaluation within argumentation. While this paper does not take on this task, its thorough investigation would really focus on the exclusion of voices and practices in argumentation. I now turn to discussing the connections between feminist epistemology and the field of Informal Logic.

4. Does anyone really do feminism in Informal Logic?
In addition to Orr, there are other feminist critiques of Informal Logic. Karen J. Warren (1988) demonstrates that critical reasoning takes place within a patriarchal conceptual framework (pp. 31-32). Once this is recognized, specifically that there is no neutral view of arguments – all theories and models originate from some conceptual framework (p. 33), and so each critical thinker and critical theory is entrenched in a bias of some sort, then a deeper contextual understanding of argumentative communication can ensue. Warren’s characterization of a patriarchal conceptual framework is equivalent to the ethic of justice described above.
There are others who are sympathetic to feminist concerns such as Orr’s or Warren’s. Verbiest (1995), Fisher (1998), and Gilbert (2005), for example, in dealing with gender styles of reasoning and communication, are in line with feminist concerns. And, Govier (1999) states that “by studying styles of verbal argument and practices of conflict resolution in other cultures, including those many minority cultures that have long been excluded and oppressed by practitioners of Western thought, we can further diversify and strengthen argumentative practice” (p. 64).
In addition to the criticisms of argumentation above, Gilbert (1997) has developed a model of argumentation that addresses feminist epistemological concerns. His multi-modal model of argumentation has opened doors to other forms of argumentation. Beyond logical arguments, Gilbert introduces emotional, kisceral, and visceral arguments (pp. 75-88).[iv] This model of argumentation challenges the field of Informal Logic to extend its parameters beyond just verbal arguments, directly addressing the notion of what is considered “rational” in an argumentative exchange. Its three alternative modes of argument acknowledge different methods of communicating arguments, which ultimately require different means of assessing them.[v] The multi-modal model is an exemplar of argumentation that functions within the framework of the ethic of care, as it is focused on the particular context and situations of the interlocutors involved.

Thus, feminist critiques of Informal Logic evolved in the late 1980s through the 1990s. They question the knowledge centered around the development and implementation of argumentation theories and models. What feminist epistemology and specifically criticisms like Orr’s offer Informal Logic is the opportunity to investigate and include unexplored patterns of argument to its repertoire. Beyond intermittent feminist critiques of argumentation though, there is hardly any follow up to addressing argumentation with the ethic of care in mind. While multi-modal argumentation addresses feminist concerns in its approach to argumentation, there is no comprehensive theory or model that speaks to feminist concerns and none address feminism to the extent that Gilbert’s model does. Now, I turn away from Informal Logic and investigate whether the Pragma-Dialectical program, as expansive and comprehensive as it is, can better address feminism.

5. A Pragma-Dialectical response to feminist concerns: a good start
The Pragma-Dialectical model of argumentation is a comprehensive research program that addresses argumentation for a number of enterprises: philosophical, theoretical, analytical, empirical, and practical (van Eemeren & Grootendorst, 2004, pp. 11-37, 41). I refer to this model of argumentation because its well-roundedness can offer at least a starting point for feminist-conscious argumentation. Theoretically, Pragma-Dialectics focuses on resolving disputes, and practically it aims to continually improve this practice. One of the model’s most useful enterprises is its commitment to empirical research, addressing the extent to which ordinary language users are successful in the resolution of differences of opinion. In this section I focus particularly on a) parts of the definition of argumentation offered by van Eemeren & Grootendorst and its connections with three of the model’s four meta-theoretical starting points, and b) articulating feminist concerns with Pragma-Dialectics. I articulate the positive aspects of each discussed constituent of Pragma-Dialectics and then propose ways in which the model can be extended further. It is in the capacity of these extensions that the model can accommodate more types of arguers and argumentative practices that fall under the ethic of care.

a) Definition of argumentation & Pragma-Dialectical starting points
To begin, van Eemeren and Grootendorst define argumentation as “a verbal, social, and rational activity aimed at convincing a reasonable critic of the acceptability of a standpoint by putting forward a constellation of propositions justifying or refuting the propositions expressed in the standpoint” (2004, p. 1). In addition, the four meta-theoretical starting points are functionalization, externalization, socialization, and dialectification. I focus on argumentation as a verbal, social, and rational activity, connecting those characteristics with externalization, socialization, and dialectification.

Verbal activity and Externalization. While verbalized arguments are more ideal, Pragma-Dialectics allows for non-verbal parts of arguments provided they can be placed into words (externalized). This is a necessary step for the reconstruction and analysis of arguments, as we need to include and address what the interlocutors implicitly communicate. What makes this limiting though is the extent to which implicit parts of an argumentative encounter can be externalized. Pragma-Dialectics relies on the performance of speech acts. For instance, if an interlocutor advances a contrary standpoint to the original standpoint, then Pragma-Dialecticians assume that the interlocutor doubts the original standpoint (van Eemeren & Grootendorst, 1993, p. 61). While I have no objection to this process, I think language use is one method of conveying messages in an argument, other communicative acts can be considered. Body gestures, sighs, silences in dialogue, topic changes, among others, are argumentative cues that could also be addressed for instance. While implementing these additional argumentative moves is more interpretative than focusing solely on language, developing measures for these additional methods allows for a more elaborate and far reaching program, albeit one that adds a rhetorical awareness to a dialectical research program.
In their latest work, van Eemeren and Grootendorst outline some of the differences in a research program that is either dialectically or rhetorically focused. They write, “It goes without saying that there are still more possibilities, that all kinds of variants can be envisioned, and that it may sometimes be fruitful to make use of certain insights achieved in one program in carrying out another program” (2004, p. 41). While dialectical insights are crucial to the resolution process, the rhetorical effects of argumentation are also important. Furthermore, it is the rhetorical aspects that better reflect the ethic of care. The Pragma-Dialectical model, in its normative construction, adheres strictly to the constituents of the ethic of justice. For instance, its reliance on the use of speech acts keeps it relegated to a fairly objective, universal account of argumentative communication, lacking a more contextual awareness of argumentation.
This criticism of Pragma-Dialectics in no way implies that the current use of Speech Act Theory in the determination of implicit argument parts should be abandoned. On the contrary, it is a useful and advantageous method in contexts of critical discussion that rely heavily on verbal discourse. If Pragma-Dialectics continues to rely solely on this method for deciphering implicit argumentative moves, and it plans no alternative or further means of communicating arguments, then the model would remain unable to address feminist concerns. However, a recent addition to Pragma-Dialectical scholarship, strategic maneuvering, acknowledges rhetorical strategies that can enhance critical discussions (see van Eemeren & Houtlosser, 2002). While the authors are clear to note that the rules of Pragma-Dialectics cannot be violated, rhetorical strategies that both follow the model’s rules and enhance an interlocutor’s argument are certainly acceptable within the Pragma-Dialectical framework. With the recent introduction of strategic maneuvering, Pragma-Dialectics can begin to better address feminist epistemological criticisms of argumentation.

Social Activity and Socialization. Pragma-Dialectics recognizes that the communicative act of argumentation is a social process, relying on the interaction between two or more parties. Its principle of socialization is concerned with determining the roles of arguers. For instance, is an arguer justifying or refuting the standpoint in question? While this is a logistical start to categorizing arguers involved in argumentation, there are more subtle social intricacies that can be investigated, such as a history of the relationship between arguers, the present situations of arguers, conceptual frameworks, among other social influences that have effects on the path an argument takes. Thus, the principle of socialization can extend significantly further than tracking the speech acts which identify the roles of social actors. With this addition to the dialectical measures already in place, a more detailed and arguer-specific model can develop. This is to say that while investigating the process of resolving a difference of opinion in critical discussions is important, it does not bode well for interlocutors and arguments that stray from prescribed argumentative norms. The Pragma-Dialectical model’s notion of social awareness is thus also bound to the ethic of justice: there are particular objectively driven criteria that get followed. If we begin to incorporate other social elements, some of which were above mentioned, then the ethic of care begins to take effect within the argumentation model and more contextualized, particularly situated criteria can be explored.

Rational Activity and Dialectification. The principle of dialectification stipulates that arguers’ attempts at resolution require them to follow norms of reasonableness while engaged in critical discussion with each other. The model outlines critical discussion rules that ought to be followed, and it is in following these standard rules that argumentation is deemed a rational activity. This is probably the most problematic of notions within the Pragma-Dialectical program as it casts away possible pertinent parts of argumentation that do not fit the standard rules of reconstruction. Standards of reasonableness are not universal, and thus there are categories of arguers who have different argumentative practices that might not be encompassed within the Pragma-Dialectical program. For instance, Pragma-Dialectics does not have the tools to deal with social contexts of power imbalance. There are strategies that interlocutors might use that the Pragma-Dialectical model finds unreasonable, but in the context of a power imbalance they could be reasonable and needed strategies for interlocutors. What follows is only a short sample of an argumentative context that uses silence as a strategic move, and a Pragma-Dialectical response to it.

Consider a workplace environment of approximately thirty colleagues, two who have personal differences and thus do not get along. One is male, and the other is female. While the male is complacent with the dynamics of the workplace, not having serious troubles with the ways that it functions, the female is more critical and notes its problems. The two have a disagreement about the workplace in a series of meetings, in the presence of most of their fellow colleagues. Tensions rise, arguments get made in the midst of personal attacks, and the female strategically decides to respond with silence to what she thinks is continual banter and disrespect, instead of arguments against her view, from her male colleague in the last several meetings. While the male has the support of his (mostly male) colleagues, who also sometimes verbalize their support, the females have mostly all been silent throughout the contentious dialogues. The male and a few others continue to make arguments, but they are met with no response. Eventually the male and female are brought together by a third party to resolve the issue.

The dynamics of sex, silence versus aggressive communication, a group of men versus a single woman, among other constituents, make the above example a case of power imbalance. Specifically, silence cannot be construed as really anything by Pragma-Dialectics, as there are no speech acts to gather information. I argue that silence, especially in the instance of the particular female discussed, should be taken as a strategic sequenced manoeuvre. In this case, silence is important to acknowledge in the arguments’ reconstructions, as it is both a calculated move in the argumentation as well as an influencer to the argumentation that follows from it. Following its set of standard rules keeps Pragma-Dialectics regimented, adhering to an ethic of justice framework. However, an awareness of power dynamics and cultural norms that stray from Western-oriented notions of good argumentation can add to the rules already functioning within Pragma-Dialectics and lead to a more feminist-friendly model.

b) How feminist-friendly is the Pragma-Dialectical model then?
To be clear, while each of the above components of the definition of argumentation and three of the program’s meta-theoretical starting points have been criticized, they do provide good grounds for a more expansive model of argumentation, one that can address feminist epistemological concerns as well.
I mentioned above that feminist epistemology recognizes that knowers are particular and concrete, rather than abstract and universalizable. Translating this notion to argumentation, interlocutors are particular and concrete, as opposed to abstract and universalizable. The empirical and practical components of Pragma-Dialectics recognize this. Empirically, the model investigates the success of dispute resolution in real contexts, with an analytical focus on the reconstruction of arguments. Practically, Pragma-Dialecticians continually investigate critical discussions in order to improve the practice. If the notion of a concrete and particular interlocutor in a particular context is taken even more seriously though, then what could evolve instead of a set of limited critical norms to be followed are perhaps sets of argumentative strategies for interlocutors, and additional tools for their evaluation.

All knowers are part of a larger social network, influenced by historical and cultural factors. These factors are largely ignored in the Pragma-Dialectical model. For instance, personality styles when it comes to argumentation or cultural upbringing which influences communicative practice are not adequately investigated within argumentation practice.
In addition, the intricacies and intersubjectivities of any given context are filtered out of the core argument structure within Pragma-Dialectics. For instance, discourse that is extraneous to the immediate standpoint in question is deleted as off-topic, repetitive, etc. (see van Eemeren, Grootendorst, Jackson, & Jacobs, 1993, p. 61). These could be strategies for interlocutors though. If an interlocutor continually repeats an idea, there might be a purpose to investigating the idea and untangling its implications, rather than deleting its repetitive parts in argumentative reconstruction. Does it mask something else? Is it really important that others understand it? Is it a safer way of saying something risky within argumentation?
And, finally, though I have not dealt with subjugated positions in any detail here, being aware of subjugated positions within argumentation practice is uncharted territory anywhere. However, feminist epistemology shows the value of the construction of knowledge, which directly opposes objective, detached standards of argument analysis. If any argumentation theory or model considered subjugated positions in their constructions, argumentation analysis would be much different and more difficult. Intricacies and situations would become just as important as the argument parts in question.

6. Conclusion
This paper is meant to be a first step towards linking feminist epistemological concerns with argumentation. The field of Informal Logic does not accommodate feminist ideas, as its toolbox does not contain a variety of tools to access different forms of argumentation. Multi-modal argumentation is a recent addition to fairly standard argumentation practice, and it both acknowledges feminist concerns and promises future developments within the ethic of care framework if developed. Turning to Pragma-Dialectics, a more thorough research program for argumentation, provides an advantageous foundation for addressing feminism. However, it also needs expansion, and continuing to either develop a rhetorical awareness to Pragma-Dialectics through the outlet of strategic maneuvering, or a rhetorical program as comprehensive as Pragma-Dialectics altogether, one that complements and can work with Pragma-Dialectics, can add new dimensions to argumentation that correspond with Orr’s criticisms and feminist epistemological concerns.

NOTES
[i] I specifically use the two terms of Gilligan in order to avoid focusing on issues of biological essentialism. Suffice it to note that any individual can partake in either style of reasoning. In fact, the more resourceful, well-rounded, individual should definitely make use of both the ethic of care and the ethic of justice depending on the context in question.
[ii] Gilligan’s work is in response to the moral development studies of Lawrence Kohlberg. Quoted for the sake of brevity, the following is a sample of one of Kohlberg’s studies, tested only on males, and then Gilligan’s findings of the same study, tested on both sexes. Kohlberg’s Heinz study is as follows: “In this particular dilemma, a man named Heinz considers whether or not to steal a drug which he cannot afford to buy in order to save the life of his wife. . . . Should Heinz steal the drug?” (Gilligan 1993, pp. 25-26). In Gilligan’s studies, 11-year-old Jake responds by saying that Heinz should steal the drug, as life is more important than money or laws. Amy, also 11 years old, says that Heinz should not steal the drug as there has to be some collaborative way that the money can be found (pp. 26-28). Gilligan writes, “Just as he (Jake) relies on the conventions of logic to deduce the solution to this dilemma, assuming these conventions to be shared, so she (Amy) relies on a process of communication, assuming connection and believing her voice will be heard” (p. 29). This is an example that begins to demonstrate different frames of reasoning at work: Jake ascribing to an ethic of justice and Amy to an ethic of care.
[iii] Nelson’s work (1990) emphasizes knowledge held by communities, rather than just individual knowers, in developing a holistic approach to questions about evidence and justification. In the construction of scientific models, Longino (1990) argues for pluralism as a way of making the values and assumptions of science accessible for critical evaluation.
[iv] It should be noted that visual argumentation is another area that broadens traditional Informal Logic approaches (see Groarke, 1996; Birdsell & Groarke, 1996; Blair, 1996), though it addresses feminist concerns less than the multi-modal approach.
[v] Full normative accounts of the three additional modes are not yet developed, though Gilbert has initiated a normative approach to the emotional mode (1997).

REFERENCES
Blair, J. A. (1996). The possibility and actuality of visual argument. Argumentation and Advocacy, 33, 23-39.
Eemeren, F. H. van & Grootendorst, R. (2004). A systematic theory of argumentation: The pragma-dialectical approach. Cambridge, MA: Cambridge University Press.
Eemeren, F. H. van, Grootendorst, R., Jackson, S., & Jacobs, S. (1993). Reconstructing argumentative discourse. Tuscaloosa, AL: The University of Alabama Press.
Eemeren, F. H, van, & Houtlosser, P. (2002). Strategic maneuvering: Maintaining a delicate balance. In F. H. van Eemeren & P. Houtlosser (Eds.), Dialectic and  rhetoric: The warp and woof of argumentation analysis (pp. 131-159). Dordrecht: Kluwer Academic Publishers.
Fisher, L. (1997). Is reasoning gendered? In H. V. Hansen, C. W. Tindale, & A. V. Colman (Eds.), Argumentation and Rhetoric: Proceedings from the 1997 Ontario Society for the Study of Argumentation Conference, Brock University, St. Catherines, Ontario.
Gilbert, M. A. (1997). Coalescent Argumentation. Mahwah, NJ: Lawrence Erlbaum Associates.
Gilbert, M. A. (1997). Prolegomenon to a pragmatics of emotion. In H. V. Hansen, C. W. Tindale, & A. V. Colman (Eds.), Argumentation and Rhetoric: Proceedings from the 1997 Ontario Society for the Study of Argumentation Conference, Brock University, St. Catherines, Ontario.
Gilbert, M. A. (2005). Argument use in gendered contexts. In D. Hitchcock (Ed.) and D. Farr (Ed. assistant.), The Uses of Argument: Proceedings of a conference at McMaster University, Hamilton, Ontario.
Gilligan, C. (1982). In a Different Voice: Psychological Theory and Women’s Development. Cambridge, MA: Harvard University Press.
Govier, T. (1999). The Philosophy of Argument. Newport News, VA: Vale Press.
Govier, T. (2001). A Practical Study of Argument (Fifth ed.). Belmont, CA: Wadsworth/ Thomson Learning.
Groarke, L. (1996). Logic, art and argument. Informal Logic, 18, 105-129.
Groarke L. A., & Birdsell D. (1996). Toward a theory of visual argument. Argumentation and Advocacy, 33, 1-10.
Groarke, L. A., & Tindale, C. W. (2004). Good Reasoning Matters! A Constructive Approach to Critical Thinking (Third ed.). New York: Oxford University Press.
Johnson, R. H., & Blair J. A. (2006). Logical Self-Defense. New York: International Debate Education Association.
Longino, H. (1990). Science as Social Knowledge. Princeton, NJ: Princeton University Press.
Nelson, L. H. (1990). Who Knows: From Quine to a Feminist Empiricism. Philadelphia, PA: Temple University Press.
Orr, D. (1989). “Just the facts ma’am: informal logic, gender and pedagogy.” Informal Logic, 11, 1-10.
Verbiest, Agnes. (1995). Woman and the gift of reason. Argumentation, 9, 821-836.
Warren, Karen J. (1988). Critical thinking and feminism. Informal Logic, 10, 31-44.

Half the wrong conclusions at which mankind arrive are reached by abuse of metaphors. (Lord Palmerston)[ii]

In this essay [i] I want to make an approach to understanding Mill’s view of argumentation, especially as his attitude toward this activity can be extracted from his essay, On Liberty.[iii] I will do this in a round-about way by considering three figures of speech, one of them associated with argumentation in general and the other two specifically attributed to Mill’s thought. These figures of speech have the character of metaphors or perhaps what Stephen Barker has called revelatory definitions. His example is, “architecture is frozen music.” As a definition this statement does not say how ‘architecture’ is used in English, nor does it introduce a new meaning for that term; it rather proposes a new way of looking at architecture. “We must reflect,” writes Barker, “about the extent and validity of this comparison between music and buildings; the [revelatory] definition is a good one if the comparison is illuminating.” (Barker 1965, p. 204) So, in this essay I will consider how apt and illuminating are the metaphors, “argumentation is war”, “the marketplace of ideas” and “society is a debating club”, with regard to Mill’s views on argumentation. Respectively these figures suggest that argumentation is war-like, debate-like, and free trade-like. Having done that I will try to identify what it is that is unique and peculiar about Mill’s view.

1. War
Perhaps the most common metaphor associated with argumentation is that “argument is war.” It may well have its roots in ancient Greek dialectic. One interpretation of Aristotle is that he taught “dialectic as a form of self-defence, organizing techniques and strategies … into the structured discipline of a philosophical martial art” (Hill & Kagan 1995, p. 34). A long time later, in the 1830s, Richard Whately used a military metaphor to explain why it is an advantage to have the presumption on your side when engaging in argumentation: an army defending a fort may well be able to turn back any assault, but should the army go “into the open field to encounter the enemy,” – that is, should the army go on the offensive – rather than wait for the enemy to attack, it might be defeated (Whately 1846, p. 113). Recently Ralph Johnson and Anthony Blair have given their informal logic textbook the title, Logical Self-Defense, intimating that some kind of combat-like attitudes and skills are needed as a safeguard against the “species of illogic” (Johnson and Blair 1983, p. xiv). Most recently the metaphor, “argument is war,” has been the point of departure for Deborah Tannen’s book, The Argument Culture. She speaks of a pervasive tendency – she calls it “agonism” – in our society to engage in argumentative behaviour. “In the argument culture,” she writes, “criticism, attack, or opposition are the predominant if not the only way of responding to people and ideas” (Tannen 2003, p. 7). Daniel Cohen, who worries about the metaphor’s implications for education, has written that,
Despite any ambiguities and subtle nuances of the word “argument,” this metaphor manages to dominate our discourse about arguments and our argumentation practice. We routinely speak, for example, of strong, or even killer, arguments and powerful counterattacks, of defensible positions and winning strategies, and of weak arguments that are easily shot down while strong ones carry a lot of firepower and are right on target (Cohen 2004, p. 36).

Tannen (2003, p. 14) points out that ‘war’ is, however, a key term in many other metaphors as well, such as the war on terror, the war on crime, the war on cancer, the war on poverty; to which I may add my own favourite – the battle of the bulge. Whenever we are involved in a struggle or competition, and the stakes are high, we seem to be ready for a metaphorical war. Cohen’s concern for how easily the language of military conflict can be adapted to that of intellectual engagement is shared by many.

Consider what we might be expected to glean from the argument-is-war metaphor.
1. There are opposing sides in the argumentation.
2. The purpose of engaging in defensive argumentation is to resist the imposition of another’s view.
3. The purpose of engaging in offensive argumentation is to impose your view on another.
4. There are few, if any, rules or standards of argumentation to be followed (trickery may be employed; there is no requirement to respect opponents).
5. Winning is more important than getting at the truth.

These may not be the only insights that purveyors of the metaphor wish to impress upon us. I have ordered the insights 1 to 5 in what seems to me to be an ascending scale of war-like behaviour: if only 1 – 3 are satisfied then there is only slight support for the metaphor but should either of 4 or 5 be satisfied as well, then it may be said that argument is war is a telling metaphor.

To what extent is this revelatory definition true of Mill’s argumentative practice? To be sure, his language of argumentation is not entirely free of military or combat images. For example, he remarks that the paradoxes of Rousseau “explode[d] like bombshells” (L ii 35) in the climate of received opinions, and he goes on to observe that in the main the pursuit of truth is a “struggle between combatants fighting under hostile banners” (L ii 36), and that there there is a “violent conflict between parts of the truth” (L ii 39). Also, in his earlier essay on Coleridge, Mill speaks of the importance for philosophy of “antagonist modes of thought” (Mill 1840, p. 104). Nevertheless, the “argument is war” metaphor does not, in my view, capture either Mill’s practice of argumentation, or his considered attitude towards it.
Unavoidable for any study of argumentation (in English) must be the recognition that the word “argument” is ambiguous. Thus Tannen marks the distinction between “making an argument for a point of view” and “having an argument – as in having a fight” (Tannen 2003, p. 4). Let us call these, respectively, the evidential and interactional senses of ‘argument’. Given this distinction, it is interesting to observe that the words “argument” and “discussion” each occur about thirty times in chapter 2 of On Liberty. However, Mill tends to use the word “argument” in the point-of-view sense of argument, that is, the evidential sense, whereas he uses the word “discussion” in lieu of “argument” in the interactional sense. In other words, if the metaphor were to be adapted to Mill’s usage, it would come out not as argument is war, but as discussion is war.
But in choosing to use the term “discussion,” Mill is signalling a pacific attitude towards argumentation rather than an agonistic one. The word “discussion” conjures up images of civility, politeness, turn-taking, and good will in a way that “argument” does not. “Discussion” does not connote violence, deceit or coercion, but rather a certain openness and bilaterality, and tentativeness. Moreover, Mill is not promoting just any kind of discussions; he is advocating free discussions (L ii 9, 24, 25, 26, 30), and free and equal discussions (L i 10), which must also be fair discussions (L ii 10, 44), and fair and thorough discussions (L ii 20). In urging that discussions should have these qualities, Mill is not only rejecting traditional authoritarian views that sought to limit available information, he is proscribing a mode of intellectual intercourse which is very unlike war and which is designed to promote the discovery or maintenance of truth. Hence, conditions 4 and 5 of the metaphor above do not fit Mill’s view well at all. For these reasons the “argument is war” metaphor applies to Mill only in a very weak sense.

2. Debate
Epistemic justification, for Mill, depends on access to the widest possible range of arguments and objections and, hence, on a social climate that does not restrict the expression of opinions of any kind. “There ought to exist,” writes Mill (in a footnote), “the fullest liberty of professing and discussing, as a matter of ethical conviction, any doctrine, however immoral it may be considered” (L ii n.). But in an important essay published over fifty years ago, Willmoore Kendall criticised Mill’s defence of unreserved freedom of speech. He thought that it would lead to “deafening noise and demoralizing confusion” because it gave the right to everyone to engage in discussions without imposing any correlative obligations on them (Kendall 1960, p. 40). “Mill’s proposals,” writes Kendall, “have as one of their tacit premises a false conception of the nature of society, . . . They assume that society is, so to speak, a debating club devoted above all to the pursuit of truth, and capable therefore of subordinating itself – and all other considerations, goods, and goals – to that pursuit” (Kendall 1960, p. 36).
By the use of the figure of speech, society is a debating club, Kendall is ridiculing Mill as being naive and unrealistic, and failing to realize that although society values freedom of speech, it values some other liberties equally as much. Kendall then goes on to list a number of conditions that societies who are intent upon the pursuit of truth will insist upon: that people who participate in the discussions should be well-trained, that they should be familiar with the society’s orthodoxies, that those who cannot persuade society that its orthodoxies are wrong will suffer isolation or banishment. This, Kendall seems to say, is what is involved in the realistic pursuit of truth, more so than the unrestricted use of free speech that Mill advocates. In summary, Kendall has attributed a model of unrestricted debate to Mill and then gone on to argue that this model cannot serve the purpose it is meant to serve.

Kendall is not the only one to have suggested that Mill subscribes to a debate model of argumentation in On Liberty. But debates can be more or less formal. The exchanges that go on in newspapers, around the kitchen table, in seminar rooms and department meetings when people exchange views, listen to each other and are, presumably, willing to be influenced by what others say, these are not improperly referred to as debates. A more formal character is given to debate by Woods, Irvine and Walton when they write that
… debates have special rules. They are presided over by a referee or a chairperson who is committed to fairness and objectivity. In addition, debates are often settled, not by debaters themselves, but by a judge or panel of judges. In those cases where the decision is left to the debaters themselves, such as in a Parliament or Congress, a simple majority among the voters is usually sufficient to decide the outcome (Woods et al. 2004, p. 25).

These authors hold that debate is “an effective and objective way to truth” and the only way that “large scale advances in human knowledge are possible” (Woods et al. 2004, p. 31). They go on to list some of the rules for different kinds of formal debates, the Oxford and Parliamentary styles. There are, then, a range of modes of argumentation that may be described as debates. They range from something very loosely structured that hardly shows evidence of disagreement at all to something with well-defined rules, clearly marked opposite sides and a conventional decision procedure. Here, again in ascending order, are some of the possible insights that may be intended by saying that someone espouses a debate model of argumentation.
1. Two (or more) parties are expressing opposing views for and against a position
2. There are procedural rules: opposing sides take turns presenting, listening to, and criticizing each others’ views and/or arguments.
3. There are rules of conduct (personal attacks are not allowed).
4. There are time limits on speakers/writers (another procedural rule).
5. The purpose of engaging in argumentation is to win (argumentation is a zero-sum ‘game’).
6. There are decision rules: the winner of the argumentation is decided by either (a) an independent umpire, or (b) the vote of the assembly.

I suggest that if a practice of argumentation consists in no more than meeting the first three of these conditions, then it is not especially revealing of the practice of argumentation. That would hardly be enough to say that the argumentation takes the form of a debate. However, if any or all of conditions 4 to 6 were also met, then this would mean that the practice could indeed be aptly characterized as a debate.

Although Mill is in sympathy with the first three conditions of this debate model, and that therefore it is fair to say that it is his view that argumentation is in some ways debate-like, I don’t think that it is at all true that he is advocating a debate model of argumentation in anything but a very loose and general sense. Most important to observe – contra condition 5 – is that Mill does not think that the purpose of the participants in argumentation is to win. He advocates engaging in argumentation as a way of having justified beliefs, of avoiding error, and of finding new truths. Moreover, Mill nowhere indicates that he sees any value in condition 4, imposing a time limit on argumentation. Finally, the upshot of Mill’s long argument against authority (L ii 3-20) is that we must be epistemically responsible for our own beliefs. Were it the case that discussions about what is true should be decided by a referee standing apart from the discussion, as condition 6 requires, that referee would be taking the role of an authority from which there would be no appeal, and to let him or her make a decision as to which side has the best argument would be to forego our duties as epistemic agents. The same can be said, mutatis mutandis, about the possibility of deciding a debate by a majority vote: Mill was wary of majorities, both in matters intellectual and moral.
In fact, it should be noticed that in On Liberty, the text which is at the centre of the discussion of Mill’s views on argumentation, the word “debate” occurs not once. However, as we know from his autobiography, Mill was in fact well familiar with debates, acknowledging his own participation in some and using the word ‘debate’ freely when referring to the activities of others. Why then is the term so strangely absent from On Liberty? It may be that Mill deliberately avoided it in that work because he wanted to distinguish his approach to argumentation from the one embodied in debates. This is just speculation, of course, and it does not show that Mill did not have a debate model in mind; but more than passing strange it is that if he did, he would eschew use of the key word, “debate.”

3. Market
The marketplace-of-ideas metaphor is present in a nascent form in Milton’s Aereopagitica of 1644. Let truth and falsehood grapple, said Milton, “whoever knew truth put to the worse in a free and open encounter.” Alvin Goldman, among many others, recognizes Mill as belonging to the free speech tradition, writing that Mill (and Milton) “contended that unrestricted speech promotes the discovery and acceptance of truth better than its suppression” (Goldman 1999, p. 193). Goldman then goes on to discuss the thesis that an unrestricted market for ideas is the best way of promoting truth, thereby at least associating the marketplace metaphor with Mill, even if he doesn’t exactly pin it on him. Goldman distinguishes two versions of the marketplace image:
The first version understands the term “market” or “marketplace” in the literal, economic sense, and it sees the competitive market mechanism as the kind of disciplining mechanism that promotes the discovery of truth. The second version understands the term “market” or “marketplace” metaphorically or figuratively. That is, it construes the marketplace of ideas as a market-like arena, in which debate is wide open and robust, in which diverse views are vigorously defended. This kind of a debate arena may or may not result from an economic market mechanism. Under the second version, moreover, what counts is the scope of the resulting debate, not the mechanism that produces it. If a diverse set of views is vigorously aired, this qualifies as an open marketplace of ideas even when government action is required to secure this state of affairs (Goldman 1999, 192).

I prefer to call these, respectively, the strong and the weak senses of the marketplace-of-ideas metaphor. The differences between the two senses are that the strong sense involves there being some market-like mechanism at work in selecting ideas whereas the weak sense of the figure stresses the nature of the discussions as being “wide open and robust,” even if some external restraints are imposed. As before, we may list a number of conditions in order of increasing commitment to the metaphor, and then ask how many of the conditions Mill seems to be committed to.
1. There is a wide range of ideas to choose from.
2. Idea producers (or idea advocates) compete robustly with one another
3. The competition of ideas is decided by idea-adopters.
4. The ideas that get adopted by most idea-adopters are “better” (more likely to be true) than the ones that aren’t.
5. There are no restrictions on idea producers, advocates or consumers.
6. Each idea-adopter decides what ideas to accept on the basis of perceived advantage to him/herself.

Here the first three conditions give us only the weak sense of the marketplace-of-ideas-metaphor whereas the inclusion of any, or all, or conditions 4, 5, and 6 would imply something more definite and approach the strong sense of the metaphor.

Woods, Irvine and Walton, in addition to attributing the debate model of argumentation to Mill, go further and characterize the nature of ‘Millean debate’ as follows:
In a free market, consumers furnish whatever degree of demand there may be for an item offered for sale, and the suppliers and sellers determine the supply. Given these preferences and the limited resources of the consumer, the laws of supply and demand ultimately determine what value is to be accorded each commodity. The worth of a commodity is determined by the degree to which it is accepted or approved by the consumer.
… What Mill is offering us, then, is a kind of free-enterprise, survival-of-the-fittest model – and justification – of debate, one in which truth is understood to be the most important value in the free marketplace of ideas. It is in debate that truth best survives the destructive forces of opposition and criticism (Woods, et al. 2004, p. 30).

Because Woods and his co-authors refer to consumers and suppliers, and the law of supply and demand, they appear to be interpreting Mill in strong sense of the marketplace metaphor, quite literally – attributing to Mill the idea that market forces, and consumer preferences do play a role in the selection of ideas. But Isaiah Berlin, who also employs the metaphor in connection with Mill, may be taken to mean it only in the weak sense: In this passage he connects liberty, the free market of ideas, and truth:
[W]hat made the protection of individual liberty so sacred to Mill? … unless men are left free to live as they wish ‘in the path which merely concerns themselves’, civilization cannot advance; the truth will not, for lack of a free market in ideas, come to light; … (Berlin 1958, p. 78)

What Berlin seems to be concerned with is the idea that truth is an outcome of free discussion, not at all indicating the mechanism which selects some ideas and rejects others, but leaving that open.

Since quite a few writers have used this metaphor in connection with Mill’s thought, we are led to ask whether either the weak or the strong sense of the marketplace metaphor is a good fit. Consider first this passage from On Liberty:
… it was once held to be the duty of governments, in all cases which were considered of importance, to fix prices and regulate the process of manufacture. But it is now recognized, though not till after a long struggle, that both the cheapness and the good quality of commodities are most effectually provided for by leaving the producers and sellers perfectly free, under the sole check of equal freedom to the buyers for supplying themselves elsewhere. This is the so-called doctrine of “free trade,” which rests on grounds different from, though equally solid with, the principle of individual liberty … (L v 4).

Here Mill is endorsing the free-trade of the market place philosophy: it is the consumer’s freedom to take his business elsewhere that will keep prices down and quality up. There are other passages that seem complementary with this. For example, Mill says, “The truth of an opinion is part of its utility” (L ii 10) thereby connecting economic advantage with truth. If rational agents choose ideas based on their utility, they will also be choosing true ideas, and this is the very point of the marketplace metaphor that others seem to have had in mind. In another passage, Mill refers to a change in the intellectual climate brought about by “popular opinion” adopting those truths it wanted from Rousseau (L ii 35): perhaps an illustration that consumers have a role in the sorting of ideas. There are reasons, then, to think that Mill favoured a free-market economy and that he saw consumer-behaviour as an instrument of selecting ideas.

Even so, Mill is not prepared to surrender complete control of the market to consumers. Consider this passage from his Principles of Political Economy:
[T]he proposition that the consumer is a competent judge of the commodity, can be admitted only with numerous abatements and exceptions. He is generally the best judge (though even this is not true universally) of the material objects produced for his use…. But there are other things, of the worth of which the demand of the market is by no means a test; things of which the utility does not consist in ministering to inclinations, nor in serving the daily uses of life, and the want of which is least felt where the need is greatest. This is particularly true of those things which are chiefly useful as tending to raise the character of human beings. The uncultivated cannot be competent judges of cultivation. Those who most need to be made wiser and better, usually desire it least, and if they desire it, would be incapable of finding the way to it by their own lights. (Mill, 1871: Bk V, ch. xi §8; [Radcliffe 1966, 69 -70]).

What Mill intends by things useful for raising the character of human beings is education. He grants that the consumer may well be the best judge of material objects on the market, but denies that education is to be chosen on the basis of consumer preference[iv]. It appears then that Mill does not think that the marketplace of ideas metaphor applies universally – is true for all ideas. He goes on to find other exceptions to the “practical principle of non-interference” (Mill 1871: Bk V, ch. xi §9) in the following paragraphs.
Jill Gordon also has resisted the notion that Mill’s thought is aptly captured by the marketplace-of- ideas metaphor. She first unpacks the metaphor as implying that “all opinions are to be expressed; … The ideas or opinions compete with one another … [and] … As rational consumers of ideas, we choose the “best” among them.” (Gordon 1997, 236). She then argues that the ideas that will survive in the marketplace will be “those espoused by either the most powerful or the most numerous in the society” (Gordon 1997, 240). That ideas should be chosen as the best in this way is inconsistent with Mill’s philosophy, maintains Gordon, since it was his avowed purpose to protect minority opinions from coercion by majorities. In a free marketplace of ideas, however, there is no protection for minority opinions, and so, Gordon concludes, the marketplace-of-ideas metaphor is antithetical to Mill’s position.

Gordon has another argument to the effect that Mill does think we should, in some circumstances, interfere with “the free market in ideas.” She finds the following passage in Mill to support her view:
On any of the great open questions … if either of the two opinions has a better claim than the other, not merely to be tolerated, but to be encouraged and countenanced, it is the one which happens at the particular time and place to be in a minority. That is the opinion which, for the time being, represents the neglected interests, the side of human well-being which is in danger of obtaining less than its share. (Gordon 1997, p. 239; [OL ii 36]).

The context here is the two-party system of parliament. Mill saw the two parties as needing each other to correct each others’ shortcomings as they attempt to balance the demands of stability and progress. In this passage Mill is advocating what appears to be a kind of affirmative action for minority opinions: they are not to be treated the same as majority opinions but are rather to be encouraged and supported. To take this view is to interfere with the marketplace as a free and open market.
There is more evidence, I think, for Gordon’s view than the passage she chose. Mill considers the concession that free expression of opinions may be permitted on the condition that discussions be fair and temperate. He observes that intemperate ways of argumentation are condemned when used against prevailing opinions but are praised when such means are used in support of accepted opinions. Hence, Mill maintains, we should compensate for this by tolerating intemperate argumentation more so when it is used to attack prevailing opinions than when it is used to defend them, there being “much more need to discourage offensive attacks on infidelity than on religion” (OL ii 44). Taking such compensatory measures, on the present analogy, amounts to an interference with free trade.
We must conclude, therefore, that the strong sense of the marketplace place metaphor does not fit Mill’s thought well at all. This is because – contrary to condition 5 of the metaphor – he is not advocating free truck in ideas; several passages in the Political Economy and On Liberty recommend interference with unrestricted commerce in ideas, if necessary. Moreover, on questions of veritistic value, Mill does not endorse the idea that the view the majority holds is more likely to be true. This is contrary to condition 4 of the metaphor. These reasons, however, do not exclude the possibility that the weak sense of “marketplace of ideas” does fit Mill’s view, for the weak sense means only to highlight the forum in which “debate is wide open and robust” and “diverse views are vigorously defended” (Goldman 1999, p. 192) even if there are some constraints placed on the discussion (as Mill would want). And we haven’t completely ruled out the strong sense of the metaphor yet since we haven’t given reasons to reject condition 6, that each idea-adopter decides what ideas to accept on the basis of perceived advantage to him/herself. Could this be Mill’s view? The next section will help us to see the answer to this question.

4. Mill’s standard
In On Liberty Mill declares that, “Whatever people believe, on subjects on which it is of the first importance to believe rightly, they ought to be able to defend against at least the common objections” (L ii, 23, my italics). This appears to be what Mill goes about doing in the second chapter of Utilitarianism. In On Liberty he goes on to stress the importance of being able to answer objections, saying that unless one can respond to objections, one has no grounds for her opinions.
[W]hen we turn to … morals, religion, politics, social relations, and the business of life, three-fourths of the arguments for every disputed opinion consist in dispelling the appearances which favour some opinion different from it…. He who knows only his own side of the case, knows little of that. His reasons may be good, and no one may have been able to refute them. But if he is equally unable to refute the reasons on the opposite side; if he does not so much as know what they are, he has no ground for preferring either opinion (L ii, 23).

Perhaps for the sake of emphasis Mill’s remark that three-fourths of the arguments should deal with objections is intentionally hyperbolic. The Subjection of Women has much the same view but it requires that in addition to giving an argument for the thesis and answering actual objections to it, one has to answer possible objections as well (SW i 3)[v]. However, the highest demand that Mill’s standard can place on us is that we must actively seek out objections to our views before we can have a right to hold them with confidence.
… [T]he only way in which a human being can make some approach to knowing the whole of a subject, is by hearing what can be said about it by persons of every variety of opinion, and studying all modes in which it can be looked at by every character of mind …. [F]or, being cognizant of all that can, at least obviously, be said against him, and having taken up his position against all gainsayers knowing that he has sought for objections and difficulties, instead of avoiding them, and has shut out no light which can be thrown upon the subject from any quarter – he has a right to think his judgment better than that of any person, or any multitude, who have not gone through a similar process (L ii, 7 – my stress).

Mill’s view is that one has no right to make knowledge claims unless he has “sought out objections and difficulties”. In other words, the highest level of confidence follows only on some kind of initiative of the arguer to seek out difficulties for himself . (This is another way in which Mill’s approach to argumentation is unlike debates.)

Although the standard sometimes asks us to invent our own objections, Mill ranks the practice of dealing with objections found in actual discussions with others more highly. Why is this? One reason is that the opinions and arguments of others will be an antidote to our own prejudices and blind spots – our possible errors (L ii 7). But the active seeking-out of arguments against one’s own opinion is not only for the sake of improving one’s position and obtaining a right to hold it. Logic was criticized by the early modern philosophers (especially Bacon and Locke) because it failed to be the method of discovery they wanted it to be. However, argumentation and discussion, as Mill appears to think of these activities, can be sources of discovery. Not just the discovery that one is justified or not justified, but the discovery of new theses, or new truths, especially on the complicated topics of morals, religion, politics, etc., is facilitated through discussion. If the role of discussion is viewed this way then the third prong of the argument in chapter 2 of On Liberty takes on added significance. When opinions conflict and the truth is found “between them” (L ii 34) then discussion may be the vehicle for the discovery of new propositions that would not have emerged otherwise.
Mill’s standard is a high standard for belief acceptance. It can be seen as having three components: one positive, one defensive and one critical. The positive component consists in giving a good argument for the thesis being advanced; the defensive component consists in answering the objections made to that argument; and the critical component consists in dealing with arguments directed against the thesis; that is, arguments that deny the conclusion the arguer is attempting to establish. The more freedom we have to engage in discussions, the more effective will this method be. And, it should be clear, it is a normative method for sorting ideas which is quite different from what is entailed by condition six of the marketplace metaphor, that we should decide which ideas to accept on the basis of perceived advantages to ourselves. We thus have good reasons to reject all three of the conditions we listed that might lead us to think that the strong sense of the marketplace metaphor is apt or illuminating about Mill’s views on argumentation.

5. Conclusions
The three metaphors, argument is war, society is a debating club, and the marketplace of ideas, may each be taken in a weak or a strong sense. The weak sense of the marketplace metaphor is consistent with Mill’s thought but it only identifies a necessary condition for successful discussions: that there should be variety and competition of ideas. This must be supplemented not only with Mill’s standard but also with a number of restrictions on ‘free trade’ in ideas thereby making the strong sense of that metaphor misleading as an insight into Mill’s views. The argument is war metaphor in its weak version rightly points out that argumentation involves opposition and that the outcome of argumentation may have drastic effects on arguers; however, the stronger version of the metaphor which disregards questions of fairness and respect for others, does not at all match Mill’s views. Finally, that argumentation involves rules and standards is reflected in the proposal that the debate model of argumentation is appropriate for Mill. But again, it is only in a weak version of it that will fit Mill’s thought. In a stronger version the debate model asks epistemic agents to defer their obligations to others, and to value victory over truth. This would not be acceptable to Mill.
So, whereas we began with the desire to learn something of Mill’s views on argumentation, we end with a dilemna. When any of the three metaphors is taken in its weak sense, it will fit Mill’s thought but it will also be minimally informative. However, when the metaphors are taken in their strong senses, they are not truly revelatory of Mill’s views on argumentation. In fact, they misrepresent them.

NOTES
[i] An earlier and slimmer version of this essay, “Mill and the metaphors,” was presented at the Figured of Democracy conference, Concordia University, Montreal, October, 2005 and a subsequent version, “Mill and the moral economy of ideas,” at The John Stuart Mill Bicentennial Conference, University College London, April 2006. I am grateful to F. Rosen, R. H. Johnson, and J. A. Blair for discussion.
[ii] Quoted in Brasher, p. 65.
[iii] For an incisive analysis of the argumentation in On Liberty, ch. 2, see Finocchiaro 2005. For an attempt to paint Mill as having a theory of argumentation, see Hansen (2006).
[iv] “Originality is the one thing which unoriginal minds cannot feel the use of” (L iii 12).
[v] See also last line of L ii 23.

REFERENCES
Barker, S. F. (1965). The Elements of Logic. New York: McGraw-Hill.
Berlin, I. (1958). The notion of “negative” freedom. In: Radcliff (1966) 74-81.
Brasher, N. H., Ed. (1968). Arguments in History: Britain in the Nineteenth Century. London: MacMillan Education.
Cohen, D. (2004). Arguments and Metaphors in Philosophy. New York: University Press of America.
Finocchiaro, M. A. (2005). Mill’s On Liberty and argumentation theory. In: The Uses of Argument, Proceedings of the Ontario Society for the Study of Argumentation, D. Hitchcock and D. Farr, (Eds.), Hamilton: OSSA. 89 – 98.
Goldman, A. (1999 ). Knowledge in a Social World. Oxford: Clarendon.
Gordon, J. (1997). John Stuart Mill and the “Marketplace of Ideas,” Social Theory and Practice. 23: 235 -249.
Hansen, H. V. (2006). Mill and Pragma-Dialectics. In: Considering Pragma-Dialectics, P. Houtlosser and A. van Rees, (Eds.), Mahwah: Erlbaum. pp. 97- 107.
Hill, H. H., and M. Kagan (1995). Aristotelian dialectic. Informal Logic 17: 25-42
Johnson, R. H., and J. A. Blair (1983). Logical Self-Defense, 2nd ed. Toronto: McGraw-Hill Ryerson.
Kendall, W. (1960). The “open society” and its fallacies. In: Radcliff (1966) pp. 27-42.
Mill, J. S. (1840). Coleridge. In: F. R. Leavis, (Ed.), Mill on Bentham and Coleridge. New York: Harper & Brother, 1950.
Mill, J. S. (1859). On Liberty.
Mill, J. S. (1871). Principles of Political Economy,7th ed.
Mill, J. S. (1873). The Subjection of Women.
Milton, J. (1644). Aereopagitica.
Radcliff, P., Ed. (1966). Limits of Liberty. Belmont: Wadsworth.
Tannen, D. (2003). The Argument Culture: Stopping America’s War of Words. New York: Ballantine.
Whately, R. (1846). The Elements of Rhetoric, 7th ed.
Woods, J., A. Irvine and D. Walton (2004). Argument: Critical Thinking, Logic and the Fallacies, 2nd ed. Toronto: Pearson/Prentice Hall.

Abstract
This article examines some obstructions to adequate discussion that reside not so much at the level of dialectic procedure, but rather at the level of content and motivation. Such obstructions may arise particularly easy when a discussion involves conductive argumentation, and both discussants have reasons for partially or totally suppressing the discussion of certain aspects, in spite of their relevance to the issue at stake. Such jointly agreed suppression does not formally violate the pragma-dialectical rules, since these rules focus on fairness. As an illustration, an analysis is presented of a US debate on the use of data mining against terrorism. Here two potential obstacles of the nature described above can be observed:
(i) There is an ambiguity in the way the issue of privacy tends to be conceptualised.
(ii) Even though the trustworthiness of government agencies is at the core of the issue, there is sometimes a certain reluctance to explicitly address this aspect.

1. Discussion failures and obstructions
Discussions are not always as good or as productive as they could be. The reasons for this can reside at various levels:
(i) Certain reasonable principles or standards for discussions are violated by the discussants (failure of procedure).
(ii) Certain arguments that are highly relevant to the issue under discussion are marginalised in the actual debate, or even totally ignored or suppressed (failure of content).
(iii) The discussants are driven by motivations that make them less than willing to address certain ‘faults’ in the discussion (failure of attitude).

An analysis restricted to level (i) addresses no more than what is put forward by the discussants themselves – augmented perhaps with some obvious and uncontroversial background assumptions not explicitly stated. The only ‘faults’ identified then will be violations of formal or logical rules concerning the line of reasoning or concerning the procedure of the discussion. It sometimes happens, however, that certain relevant arguments or issues remain unduly marginal in the discussion, are insufficiently addressed or, worse, not stated in any recognizable form at all. This can be due to negligence. It may also be the case that all participants in the discussion have motives that make them forsake these particular aspects. As an example of the latter, take environmental issues. As soon as concrete measures for improving the environment are to be discussed, most actors have a strong incentive to duck the issue: consumers are reluctant to give up their pleasures, industry fears costly changes that could mean a set-back in the international competition, politicians refrain from advocating unpopular measures, and so on (Birrer, 2004; 2001b).
When the discussants themselves fail to sufficiently discuss (or even fail to discuss at all) certain arguments or issues that are highly relevant to the subject of discussion, such failure will be hard to address when the analysis is restricted to level (i), since such an analysis is forced to remain relatively blind to issues of content – and even more so, of course, when that content is entirely missing in the actual discussion. If it is possible at all to address such a failure at level (i), this is likely to require a long chain of interrogating questions of clarification addressed by the analyst to the discussants, since the analyst can only point to the violation of rules rather than directly enter into the content matter itself. When the discussants are consciously or unconsciously motivated to avoid the issue, the situation is still worse, since the indirectness of the analyst’s approach makes it easy to evade by the discussants.
It looks like in such cases attempts to improve the discussion should put the discussion in a broader perspective, including levels (ii) and (iii). Even if one’s single aim would be to understand what happens in the discussion, it could still be held that such an understanding is incomplete if it does not include an understanding of why the discussion proceeds in the way it does rather than in another, and that therefore it cannot ignore the aspects under (ii) and (iii).

In the next sections, I will illustrate obstructions at level (ii) and (iii) in more detail for a recent discussion in the US on the use of certain computer techniques in the fight against terrorism. More in particular, I will analyse the arguments that can be found in one specific report on this subject. The issues and arguments that I want to address are all in the report, including what I will identify as the “core issue”, so in this case I need not bring in any new arguments or issues that are not already mentioned by the discussants themselves. The potential obstacles to productive debate, however, can be clearly recognised in the arguments in the report – as well as elsewhere in the debate on this issue. The case can thus serve as an illustration of the importance of the broader perspective for understanding the discussion process and for a realistic view on the opportunities for improvement. In the final section, I will return to a few more general questions regarding analysis at level (ii) and (iii).

2. Data mining to combat terrorism
Broadly speaking, the term “data mining” refers to a collection of computer science techniques to extract “implicit” information from (large) databases.[i] The word “implicit” means that the information to be delved up goes beyond the answers to standard queries (questions that can be formulated straightforwardly in the language of the data base, and that the data base was designed to answer). As an example, think of a database containing administrative data on an organisation’s employees. Standard questions for such a database will be: what is the salary earned by Mrs. X, what is her function, what is her home address, etc. A very different use of the database, and usually not one for which the database was originally intended, would be the following. Suppose management is worried about low job performance due to private alcohol and drug abuse. On the basis of the set of those employees who have already been identified as having an alcohol or a drug problem, the board could ask the computer systems department to write a program that searches for predictors, i.e., personal characteristics for which alcohol or drugs problems are above average. Maybe such a program would find that the group with already identified alcohol or drugs problems contains a bigger percentage of unmarried male employees living in a particular area of town than the population of employees as a whole. The board might then wish to scrutinize all employees with these characteristics. The board could also wish to prevent that any more persons with such characteristics are hired.
Quite similar methods can be employed in the fight against terrorism. E.g, one might try to find characteristics in the data on terrorists, or characteristics for terrorist activities, and then search for all persons or activities that match those characteristics and have not yet been identified as terrorist. Such automated searches can be applied to ordinary databases, but also to communicative exchanges such as emails, or even telephone conversations. Another possibility is to start from persons and activities that are already suspect, and then use computer information to trace their connections to other persons and activities.
Since the information has to be retrieved from sources that were not originally set up to answer these questions, special programs have to be constructed, and not every question may be answerable. Often it is hard to separate between the relevant and the irrelevant (useful signals/patterns vs. noise): either one finds too many patterns, or hardly any at all. Sometimes information from different databases must be combined, necessitating the coupling of these databases; this may present additional technical problems. However, with the parameters of the search rightly set, the search may be successful. Today, data mining techniques are included in a wide range of US governmental programs, for purposes such as financial accounting, service improvement, analysing scientific information, and the combat of crime and terrorism (for an overview, see GAO, 2004).

Early 2002, in the wake of 9/11, the Defense Advanced Research Project Agency (DARPA) announced a program that was then called “Total Information Awareness” (later the name was changed into “Terrorist Information Awareness”). The program was intended to explore and develop computer science techniques for combating terrorism, data mining being one of these techniques (but by no means the only one). Though right from the start some critics judged the information that DARPA had provided on the program to be insufficiently detailed or clear, it was not until a column by famous New York Times columnist William Safire appeared in November 2002 that a massive and vigorous discussion took off.
Although the program included many other computer science techniques, the discussion almost exclusively focused on data mining. Within a few months the debate had resulted in a moratorium on data mining imposed by the Congress, on January 16, 2003. Secretary of Defense Ronald Rumsfeld installed the TAPAC committee to examine the use of ‘advanced information technologies to identify terrorists before they act’ (TAPAC, 2004, p. 1). A new report by DARPA, published in May 2003 (DARPA, 2003) was unable to turn the tide. The funding of TIA was terminated by Congressional decision on September 25, 2003. TAPAC published its report in March 2004 (TAPAC, 2004). It is the argumentation presented in the latter report that forms the main material for the following case study.

3. The Tapac Report: a brief overview of its content
In addition to the main text, the report contains a minority report by committee member William T. Coleman, Jr., who disagrees with the main text on several points. The main text refers to Coleman’s statement and vice versa, which makes the report a kind of microcosm for the debate as a whole, and a useful source of arguments from both sides.[ii]The report also contains a brief separate statement by Floyd Abrams, which basically is another defense of the main report’s arguments against Coleman’s criticism, and which will not play a role in the analysis presented below.
Following the introduction, the main text contains five sections. The first sections sets out “the new terrorist threat”. The second describes the TIA program and the way it was introduced. The third section elaborates the issue of privacy from a mainly juridical point of view. The fourth section analyses the various privacy risks presented by government data mining. The fifth section contains the conclusions and recommendations. My analysis of the main text will focus on the third and fourth section (where the main points of debate can be found), with occasional reference to the other sections.

Brief summary of the section ‘Informational privacy and its protection from intrusion by the government’
This section discusses privacy considerations in American law. The main source of privacy protection discussed is Amendment IV of the Constitution that reads ‘The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated, and no Warrants shall issue, but upon probable cause, supported by Oath or affirmation, and particularly describing the place to be searched, and the person or things to be seized’. The protection of citizens is not absolute, as is illustrated by a ruling by the Supreme Court in 1976 that this amendment does not apply to information held by a third party. This verdict is read by the authors of the main report as relating to information that is provided voluntarily, and since much information provided to the government is not really provided voluntarily, they argue that Amendment IV still applies there.

Brief summary of the section ‘Privacy risks presented by government data mining’
The section observes that ‘Government data mining concerning U.S. persons presents risks to informational privacy which are not adequately addressed by existing law’. (TAPAC, 2003, p. 33). Six main types of such risks are identified:
– Data inaccuracy risks. Data may contain errors, different persons with the same or very similar name may mistakenly be identified, etc.
– False positives. Since patterns found are merely statistical correlations, part of the identifications (e.g. as a potential terrorist) under such patterns will be mistaken.
– Data processing risks. Access-authorized persons may use information in ways not intended by the organisation.
– Mission creep. Goals and practices may gradually shift in directions that were not originally intended.
– Chilling effects and other surveillance risks. The presence of surveillance activities may negatively affect the general atmosphere in society.
– Data aggregation risks. Combination of information from different databases, transnational data flow, etc. may pose additional risks.

Brief summary of the section ‘Conclusions and recommendations’
The main report closes with a number of recommendations to curtail these risks, including: that a regulatory framework and oversight mechanisms shall be installed; that the rate of false positives shall be “acceptable in view of the purpose of the search” and that there shall be a system for dealing with false positives; that access to federal databases shall require a written warrant by a federal magistrate or judge; and that the Department of Defense shall yearly publish reports accessible by the public.

Main elements of the ‘Separate statement of William T. Coleman, Jr.’
Informational privacy. Coleman thinks that the main report takes the Fourth Amendment too absolute. He agrees that there should be some restrictions on the use of data, but he holds the opinion that the urgency of the battle against terrorism is not taken seriously enough in the main report.
Privacy risks. Coleman stresses that DARPA is a professional organisation that can handle its responsibilities to a larger extent than suggested in the main text, and that therefore too many restrictions are both unnecessary and impeding the fight against terrorism.
Recommendations. Coleman’s disagreements with the main report here include that the Department of Defense should not publicly report, but only to some special committees, and that access to federal databases should not require a written warrant.

4. Analysis of the arguments
Privacy and law
First obstruction: the framing of the privacy issue
The section on informational privacy opens with the following remark on the notion of privacy:
“There is a surprising lack of clarity about what “privacy” means and the role the government should play in protecting it. This is due in part to the fact that the word “privacy” is used to convey many meanings. The Supreme Court alone has used the term to describe an individual’s constitutional right to be free from unreasonable searches and seizures by the government; the right to make decisions about contraception, abortion, and other “fundamental” issues such as marriage, procreation, child rearing, and education; the right not to disclose certain information to the government; the right to associate free from government intrusion; and the right to enjoy one’s own home free from intrusion by the government, sexually explicit mail or radio broadcasts, or other intrusions.” (TAPAC, 2004, p. 21)

One of the main problems is the conflation of two separate issues: access to information and the use made of that information (Birrer, 2001b). When the notion of privacy originally entered the juridical sphere, it referred to the intrusion of the private life of (famous) individuals such as film stars by photographers, journalists, etc. This led to the idea of a “private sphere” that should be free of uninvited intrusion by anyone. Since knowledge of the private sphere of celebrities can hardly be proclaimed as an urgent public interest (notwithstanding the existence of a considerable market for such information), the delineation of such an unassailable private sphere is not likely to meet substantial public controversy. In issues of privacy as the term is used today, the situation often is less simple. Usually, trade-offs are involved between the interest in gathering and using information on individuals on the one hand, and the interests of the individuals concerned on the other. These trade-offs, in turn, do not so much depend on the access to the information as such, but on the use that is made of that information. This applies to both sides of the balance: the use of the information may serve an important goal that justifies the suspension of certain privacy concerns, whereas it may also be the very ground of certain privacy concerns in view of the consequences for individuals whose information is (accessed and) used.
The possibility of trade-offs already enters the picture in the formulation of the Fourth Amendment: the term ‘unreasonable searches’ seems to suggest that acceptability might depend on the goal of the search.[iii] In many instances of discussions on privacy issues, however, the intended or actual use of the information is not much considered, or not even addressed at all.
There are several possible explanations for the persistence of the conceptualisation of privacy in terms of access only. One obvious reason is historical: since it has been the dominant conceptualisation for such a long time, one might fear that a sudden change of terminology could easily create confusion. Another possible reason is the central position of rights in the language of law. But there could also be a more specific reason for adhering to this kind of framing, particularly for those opposing intrusion of privacy: framing privacy in terms of more or less absolute rights blocks the road towards tradeoffs that could shift in favour of the powerful actors, at the expense of the ordinary citizen. For blocking such ‘slippery slopes’, a deontological (rule-based) perspective might be seen as more effective than a utilitarian one.
So the situation is that privacy discussions are suffering from a somewhat ambiguous and often misleading framing. Whereas the juridical discussion in this section of the report is important for identifying the possible juridical instruments to provide protection vis-à-vis certain ‘privacy’ concerns, and can also provide a certain amount of legitimation for privacy concerns in general, it is not to the full extent addressing the question why specific activities would be deemed acceptable or not. Neither the authors of the main report nor Coleman seem to be willing to defer their judgement merely to what the law says or what judges decide; what is really at issue for them is what would be acceptable or reasonable, taking all interests involved into account.

Privacy risks
Second obstruction: the issue of trustworthiness
Once we have understood these limitations and ambiguities of the privacy discussion, it becomes easy to see why the report’s next section suddenly starts talking a very different language by considering the risks posed by the gathering, but most of all by the use of information on individuals.

What is particularly notable with respect to the privacy risks identified in the main report is that most of them do not originate in the technique as such, and not even in the use of those techniques as intended. The majority of the risks are ‘social risks’, referring to consequences not originally intended that result from the ‘social dynamics’ that may emerge when potential or actual information gatherers or users meet with certain opportunities. This aspect of social dynamics is particularly clear in the case of “mission creep”, but it is also an inextricable part of the issue of dealing with “false positives” and of what are called “data processing risks”; it also plays a slightly more indirect role in “data aggregation risks”, and more indirectly also in the consequences of “data inaccuracy risks; “chilling effects” rather refer to the social dynamics of society as a whole.
It is this issue of what I called “social risks” that gives us a clue of where the basic difference of opinion between the main report and Coleman really lies: the latter is conceding more trustworthiness to DARPA and the special agency that was to conduct TIA than the authors of the main report. Coleman as well as the main report mainly speak in general terms about their reasons. This is not surprising, of course, since there can hardly be any specific evidence on a program that has yet to start.
Some case-specific arguments are provided in the main report’s extensive analysis of the way TIA was announced and explained by DARPA, concluding that TIA “was flawed by its perceived insensitivity to critical privacy issues, the manner in which it was presented to the public, and the lack of clarity and consistency with which it was described” (TAPAC, 2004, p. viii), to the result of seriously undermining DARPA’s and the program’s credibility. Coleman is more generous (though not completely uncritical) regarding the way DARPA handled the introduction of the program (see Coleman, 2004, p. 81).

For the rest the main report refers to other cases where risks such as the ones described did indeed materialise, cases where courts acknowledged privacy concerns, etc. As mentioned earlier, Coleman pictures DARPA as a professional organisation that can, and for the sake of the effectiveness should, have more autonomy than acknowledged in the main report.
The difference in viewpoint is perhaps most aptly portrayed when Coleman ironically remarks:
“Perhaps I am still misled by the fact that in my youth my parents taught me that policemen on the beat and other law enforcement officers are friends, not enemies, and in my life, most often, it has turned out that way.” (Coleman, 2004, p. 74)

Coleman and the main report do not disagree that the fight against terrorism has a very high urgency. They do not even explicitly disagree on the acceptability or unacceptability of certain potential consequences for citizens (although they might have if they had discussed concrete examples). Their main point of disagreement concerns to what extent certain institutions can be entrusted with certain responsibilities, and which checks and balances are needed to contain the potential danger of misuse of these responsibilities.

It often occurs that matters of trustworthiness are avoided – even when they are at the core of what is at issue-, especially when the parties are not yet in total war with each other. Several reasons can be conceived that could account for this phenomenon:
– Questioning the trustworthiness of a person or an institution has a flavour of inappropriateness, as a personal attack, or as an ad hominem argument.
– When a person or institution denies an accusation of lack of trustworthiness, such a denial is hard to conclusively disprove; therefore, the accusation is also easy to evade.
– Suggesting lack of trustworthiness might trigger conflict; conflict-averse persons will try to avoid this by not explicitly addressing it at all.
– For some the possibility of being cheated by someone else might feel as an assault on their self-esteem, making them want to avoid the issue.

These reactions are all very recognisably human, but they may sometimes stand in the way of addressing the issue at stake. Trustworthiness issues occur every time we have to trust a person or institution because we cannot personally monitor it. This happens whenever we depend on scientific experts for information or advice (cf. Birrer, 2001a; see (Birrer, Mentzel, 2005) for an analysis of discussions on biotechnology in terms of trustworthiness). Lay persons lack the expertise to check such information or advice. But the expert need not be a scientific expert. The recurring discussion on the adequateness of the information provided preceding the Iraq war is an illustrative example that turns around the reliability of what ‘insiders’ tell us. Our societies have not yet fully maturated practices to face up to issues of trustworthiness, and to handle them smoothly and effectively. But they will simply have to. The trustworthiness of information and advice is bound to be a core issue in this century; new political equilibria vis-à-vis information asymmetries will have to be established.
Another factor leading to neglect of the issue of trustworthiness, and more specific to the subject of this particular case study, is what I would call the ‘distrust paradox’: in organisations with a mission that implies a certain amount of distrust towards the external world there often is a remarkable lack of awareness of the possibility of untrustworthiness of its members, particularly regarding their behaviour towards the outside world. Whereas the mission of certain organisations may necessarily presuppose some amount of distrust towards the outside world, such distrust, if unguarded, can lead to excessive polarisation between ‘us’ and ‘them’. When the issue of terrorism is concerned, we have to be aware that it contains many intricacies, starting with the lack of an agreed definition of what counts as terrorism (Alexander, 2001; Whitaker, 2001), intricacies that provide an effective breeding ground for phenomena like mission creep.
It should be noted that the framing of privacy in terms of access only and the avoidance of the trustworthiness issue are not completely unrelated: concerns about the use that can be made of information puts the trustworthiness more prominently into the foreground; as long as privacy is framed in terms of mere access, the issue of trustworthiness is more easily avoided. Even the critical comments by an organisation like the American Civil Liberties Union (ACLU, 2003) are cast partly in terms of access, partly in terms of use (particularly the issue of false positives), and the trustworthiness issue is mainly addressed under the relatively general notion of ‘mission creep’. On the other hand, even Safire’s column, which most emphatically (and most polemically) exploits the theme of trustworthiness – not only regarding government as a whole, but also regarding the foreseen director of the program -, refers to a panoptic government knowing everything about its citizens rather than explicitly pointing at concrete ways in which the government might misuse that information (although it could be argued that his Orwellian rhetoric is more than enough to evoke the latter) (Safire, 2002).

5. Collective failures of content and motivation
The arguments discussed above tend to be belong to the category of conductive arguments (non-conclusive arguments with multiple premises, see (Govier, 1987)). In conductive argumentation, because of its very nature, it is relatively easy to inconspicuously push certain premises and their role further into the background than an assessment of the issue at stake seems to allow, or even to ignore such elements altogether. When one of the discussants commits such a failure, and the failure is to the disadvantage of the other discussant, the other discussant has the opportunity to address it; when it is to the advantage of the other discussant, on the other hand, the first discussant simply made a strategic mistake that the second discussant may or may not correct. However, as we have seen, it can also be the case that both discussants consciously or unconsciously prefer some aspects to be suppressed, i.e., that the failure is a collective one.
There is a wide range of possible motivations that may be the origin of such collective failures. One reason is a historical one: a mode of analysis that has been used for a long time gradually becomes less adequate to address the current issues, but the idea of adopting a new scheme makes the discussants feel insecure because it might upset the strategic balance, or create confusion, so the discussants stick to the old habit. Another motivation can be collective ostrich policy: neither of the discussants wants to face certain aspects of the issue under discussion, since these are felt as inconvenient or otherwise unpleasant. More or less latent power factors can also play a role, giving rise to self-censorship (cf. Mitchell, 2003)[iv], or to implicit effects of an interviewer’s questioning as addressed in psychology by Schwarz (1994) under the term ‘logic of conversation’).

Such collective failures of content can be conceived as failures of explicitisation. Yet, they do not violate the pragma-dialectical rules as formulated by van Eemeren and Grootendorst (1984; see 2003 for a recent version). The reason is that the pragma-dialectical rules focus on fairness. They translate the desirability of explicitisation problem into the right of discussants to bring up anything they choose. In their explanation of the pragma-dialectical rules van Eemeren and Grootendorst wrote:
‘So the importance of externalizing disputes is plain, and it therefore follows that one of the most important tasks to be achieved in formulating rules for rational discussion is the furtherance of an optimal externalization of disputes. This means that the discussants must be able to advance every point of view and must be able to cast doubt on every point of view.’ (van Eemeren, Grootendorst, 1984: 154)[v]

But the discussants having a right does not imply that they will always use it. In as far as their goals are opposed, they probably will; but when they implicitly agree – consciously or unconsciously – not to address certain aspects, explicitization is no longer guaranteed. Freedom of speech may be a necessary condition for explicitization, but it is definitely not a sufficient one. As long as power inequalities are involved, one could still hold that the pragma-dialectical rules are violated because there is no complete freedom – even though it will sometimes be hard to prove that the failure is involuntary on the part of at least one of the discussants. In many of the cases of level (ii) and (iii) failure discussed earlier, however, power inequality is not the dominant cause.
Finally, it is worthwhile to observe that in some cases the obstacles presented by collective failures as discussed above can be enhanced by differences in framing between the discussants. When the discussants have a different framing of the issue at stake or of certain premises, this may create serious obstacles to mutual understanding (Birrer,Pranger, 1995; Wohlrapp, 1995; Vermaak, 1999). Several of the obstacles discussed above can be viewed as clashes between different framings, such as the different conceptualisations of the notion of privacy, and the distrust paradox. In such framings propositions cannot be understood in isolation, but only as embedded in complex packages, and discussants tend to either accept or reject a whole package. In as far as it is possible at all to work out some common ground, the amount of work required may be very substantial. This problem is even greater when the frameworks stand in polarisation towards each other, like in the case of the distrust paradox.

6. Conclusions
Sometimes discussants collectively refrain from addressing certain aspects that are relevant to the subject of their discussion. Such cases could be said to go against the spirit of a reasonable discussion, and as such also against the spirit that has guided the formulation of the pragma-dialectical rules, but in a formal sense they do not present an infringement of those rules as they have been formulated. It seems important, nevertheless, to identify such collective failures. In order to do this, the analyst may have to go beyond what the discussants themselves bring forward, and may have to try to answer the question what exactly the discussants would have been discussing if no main relevant aspects had been suppressed.

NOTES
[i] Technical introductions in the field of data mining can be found in (Witten, Frank, 2000), and (Hand, Mannila, Smyth, 2001).
[ii] Similar arguments can be found in a wide variety of other contributions to the debate, such as Safire’s (2002) column, the comprehensive juridical analysis by Taipale (2003) or the statements by the American Civil Liberties Union (ACLU, 2003).
[iii] The interpretation of the Fourth Amendment, however, is prone to ambiguities as well. Braverman and Ortiz (2002) observe a tension between the first part refering to “unreasonable searches” and the second part refering to “probable cause”, the first allowing considerable more discretion than the second. An illustrative example of confusion concerning the Fourth Amendment occurred at a press conference by General Hayden on January 26, 2006. He was interpreted by many as denying that the Fourth Amendement refered to probable cause at all (which would be obviously incorrect). See White (2006) for a defense of Hayden, arguing that Hayden was talking about the first clause involving “unreasonable searches”, and that the “probable cause” mentioned in the second clause applies to warrants for search, and not directly to the first clause as such. Since that distinction was not explicitly made by Hayden, however, a less charitable interpretation, and one in fact picked up by many, would be that Hayden was unaware that the Fourth Amendment mentioned probable cause.
[iv] Illustrative for the complications of issues of self-censorship is the recent debate in the US on the question whether newspapers should reveal ‘security’ information. The controversy could already be sensed when James Risen and Erich Lichtblau in the New York Times reported secret eavesdropping by the National Security Agency (Risen, Lichtblau, 2005), and two days later President Bush in a radio speech stated (without explicit reference to the New York Times) that ‘Revealing classified information is illegal, alerts our enemies, and endangers our country.’ (Bush, 2005). In 2006 a more extensive public debate arose when the press made public that the CIA has access to European Swift bank transfer data (Lichtblau, Risen, 2006; Baquet, Keller, 2006).
[v] Their views apparently have not changed in the meantime, in 2003 they wrote quite similarly:
‘The importance of the externalization of differences of opinion is therefore evident. One of the first tasks in the formulation of rules for a critical discussion is thus to promote an optimal externalization. This means that the discussants must be able to put forward every standpoint and to call every standpoint into question.’ (van Eemeren, Grootendorst, 2003b: 366)
Perhaps these ideas can be traced to the article by Grice (1975) that was an important starting point for van Eemeren and Grootendorst. Grice’s analysis is based on the idea of conversation as a cooperative endeavour, solving a common problem defined by common goals. This leaves it entirely to the participants to (implicitly or explicitly) decide what the common goals are, rather than submitting these goals to an external evaluation. Similar remarks can be made regarding the five ‘language rules’ derived by van Eemeren and Grootendorst from their Communication Principle (in turn inspired by Grice’s Principle of Co-operation): the second principle demands ‘sincerity’ and ‘honesty’ of the discussants (see van Eemeren, Grootendorst, 2003a: 77), which prima facie seems to imply full externalisation, but again the resulting rules for critical discussion do not exclude collective failures of the kind discussed above.

REFERENCES
ACLU (American Civil Liberties Union) (2003) Total information compliance: The TIA’s burden under the Wyden Amendment. A preemptive analysis of the government’s proposed super surveillance program, May 19, 2003.
Alexander, Y (2001). Terrorism: a definitional focus. In: Terrorism and the law (Y. Alexander, E.H. Brenner), Ardsley (NY): Transnational Publishers.
Baquet, D, & B. Keller (2006). When do we publish a secret?, New York Times, July 1, 2006
Birrer, F.A.J., & M.A. Mentzel (2005). Issues of power, risk and ethics in the biotech debate, LIACS Technical Report 2005-09, Leiden University.
Birrer, F.A.J. (2004). Subliminal enticement and the ethics of sustainable agricultural production. In: J. de Tavernier, S. Aerts (eds.), Science, ethics & society (EURSAFE 2004), Leuven: Katholieke Universiteit Leuven, pp. 190-193.
Birrer, F.A.J. (2001a). Expert advice and argumentation, Argumentation 15, 267-275.
Birrer, F.A.J. (2001b). Applying ethical and moral concepts and theories to IT contexts: some key problems and challenges. In: R.A. Spinello, H.T. Tavani (eds.), Readings in cyberethics , Sudbury (MA): Jones & Bartlett Computer Science, pp. 91-97.
Birrer, F.A.J., & R. Pranger (1995). Complex intertwinements in argumentation. In: F.H. van Eemeren, R. Grootendorst, J.A. Blair, C.A. Willard (eds.), Proceedings of the Third ISSA conference, 1995, International Centre for the Study of Argumentation, Amsterdam.
Bush, G.W. (2005). President’s radio address, December 17.
Coleman, Jr., W.T. (2004). Separate statement of William T. Coleman, in (TAPAC, 2004).
DARPA (Defense Advanced Research Projects Agency) (2003). Report to Congress regarding the Terrorist Information Awareness Program (in response to Consolidated Appropriations Resolution, 2003, Pub. L. No. 108-7, Division M, § 111 (b)), May 20, 2003, DARPA.
Doyle, C. (2006). Memorandum to the Senate Select Committee on Intelligence, January 30, 2006, Washington.
Eemeren, F.H. van, & R. Grootendorst (2003a). A systematic theory of argumentation, Cambridge: Cambridge University Press.
Eemeren, F.H. van & R. Grootendorst (2003b). A pragma-dialectical procedure for a critical discussion, Argumentation 17, 365-386.
Eemeren, F.H. van, & R. Grootendorst (1984). Speech acts in argumentative discussions. A theoretical model for the analysis of discussions directed towards solving conflicts of opinion, Foris: Dordrecht.
GAO (General Accounting Office) (2004). Data mining. Federal efforts cover a wide range of uses. GAO Report 04-548, Washington, D.C.: GAO.
Govier, T. (1987). Problems in argument analysis and evaluation, Dordrecht/Providence: Foris.
Grice, H.P. (1975). Logic and conversation. In: D. Davidson, G. Harman (eds.), The logic of grammar, Encino (CA): Dickenson Publishing Company.
Hand, D., & H. Mannila & P. Smyth (2001). Principles of data mining. Cambridge (Mass.): MIT Press.
Lichtblau, E., & J. Risen (2005). Bush lets U.S. spy on callers without courts, New York Times, December 16.
Lichtblau, E., & J. Risen (2006). Bank data is sifted by U.S. in secret to block terror, New York Times.
Mena, J. (2004). Homeland security. Techniques and technologies. Hingham (Mass.): Charles River Media.
Mitchell, G.R. (2003) American itsesensuuri: a typology of self-censorship in the ‘War on terror’, in F. H. van Eemeren, A. Blair, C. A. Willard, A. F. Snoeck Henkemans (eds.), Proceedings of the Fifth Conference of the International Society for the Study of Argumentation, Amsterdam: SicSat, pp. 767-772.
Office of the Inspector General, Department of Defense (2003). Information technology management: terrorism information Awareness program (D-2004-033), December 2003, Department of Defense.
Safire, W. (2002). You are a suspect. New York Times. November 14, 2002, p. A 35.
Schwarz, N. (1994). Judgment in social context: biases, shortcomings, and the logic of conversation. In: M.P. Zanna (ed.), Advances in Experimental Social Psychology, vol. 26, San Diego: Academic Press.
Taipale, K.A. (2003). Data mining and domestic security: connecting the dots to make sense of data. Columbia Science and Technology Law Review 5 (2) 1-83.
TAPAC (Technology and Privacy Advisory Committee) (2004). Safeguarding privacy in the fight against terrorism. March 2004, Washington, D.C.: Department of Defense.
U.S. Supreme Court (1976). United States v. Miller, 425 U.S. 435.
Vermaak, M. (1999). On conversational constraint. In: F.H. van Eemeren, R. Grootendorst, J.A. Blair, A. Willard (eds.), Proceedings of the Fourth International Conference of the International Society for the Study of Argumentation, Amsterdam: SicSat, pp. 829-832.
White, A. (2006). Unwarranted criticism. General Hayden’s reading of the Fourth Amendment is correct, and his critics are mistaken. National Review, May 10, 2006.
Whittaker, D.J. (ed.) (2001). The terrorism reader. London: Routledge.
Witten, I.A., & E. Frank (2000). Data mining. San Francisco: Morgan Kaufmann.
Wohlrapp, H. (1995). Resolving the riddle of the non-deductive argumentation schemes. In: F.H. van Eemeren, R. Grootendorst, J.A. Blair, C.A. Willard (eds.), Proceedings of the Third ISSA conference, Vol. 2, Amsterdam: International Centre for the Study of Argumentation, pp. 43-57.

1. Introduction
This paper presents some aspects of a theory in which argumentation – as a ‘verbal, social and rational activity’ (van Eemeren 2001: 11) – plays a role in the explanation of the question phenomenon. In fact in linguistic research two levels are usually taken into account – the level of propositional content and that of the illocution or pragmatic function combined with the speaker’s attitudes (Gobber 1999). In this contribution we argue that questions can have a further intrinsic (natural, prototypical) component as a pragma‑dialectical move. This move is intended as a (part of a) dialogue that appears at some stage of a critical discussion (van Eemeren and Grootendorst 2004: 57-68). This level should be taken into account to explain the functioning of questions in verbal communication.
In fact, the classical rhetoric tradition is interested mainly in non-prototypical uses of interrogative structures such as the so-called rhetorical questions (interrogations, see Quintilianus, Institutio Oratoria, IX, 2, 7-16). This tradition considers nearly always monological texts whose goal is to draw the hearer’s attention, to gain his consent, to dissuade or to persuade him according to the speaker’s intention (‘dicendo tenere hominum mentis, adlicere voluntates, impellere quo velit, unde autem velit deducere’, Cicero, De oratore, I, 30).
This is not the natural, prototypical functioning of interrogatives as questions, i.e. as dialogical moves aiming at a verbal reply. As Edmondson 1981:196 puts it, ‘interrogativisation is a grammatical reflection of the interactional purpose of the language system’. Their role as requests for a verbal reply is relevant for the purposes of a dialogue in which the interlocutors’ task is to reach an agreement on a standpoint in a reasonable way. From this dialogical perspective interrogative structures can be observed both as “real” questions and as moves of another, i.e. non prototypical function.

2. Interrogatives and questions
Let us consider first the difference between interrogatives and questions. First we take into account the main pragmatic functions of the general type “questions”. Other uses of interrogatives which are relevant for critical discussion are described in a sketchy way.
Questions should be kept qualitatively distinct from interrogative structures. The latter are items and patterns of a given language, whereas the former are text sequences (Rigotti 1993), i.e. “moves” in a dialogue or in a monologue.
There is a “many-many relation” (Gatti 1992) between interrogative structures and questions. Of course, the most typical use of interrogatives is that of making questions.

As text sequences, questions have a propositional content and a pragmatic function (Stati 1990). Two sorts of propositional content are generally considered. Their structure results from the unknown element they exhibit. According to the sort of propositional content, two types of semantic structures are then distinguished. If the content calls for verification, a propositional question is given (i.e. “Yes-No Question” or “alternative Question”) and the propositional content has the cognitive status of an assumption (Alexius Meinong called it Annahme: see Meinong 1910). If the content calls for interpretation of (a) variable(s), an x-question (wh-question) is given and the propositional content has the semantic status of an open proposition. It has been observed that
[…] we can assume that the listener has understood the question if he knows what kind of information must be given as an answer – though, perhaps, he has no such information at hand. In other words, the listener understands the question if he can characterize correctly the semantical scheme of the answer (Padučeva 1986: 374).

X-questions have a premise (a propositional content condition), a “datum” (the propositional content) and an “obiectum quaestionis” (the range of the variable in the content) (Ajdukiewicz 1926/27).

In a dialogue questions are posed with a specific illocution. A bundle of illocutions is also possible (‘amalgame pragmatique’, Stati 1990). Specific illocutions can be traced back to two generic ones: most questions require an answer, and some of them can get the answer from the same questioner (e.g. the so-called expository questions). But there are also questions which do not wait for a verbal reply, although an answer is still possible and accepted.

The most frequent questions call for information (“let me know”). Other requests for an answer are examination questions (“show me that you know”) and maieutic, i.e. “Socratic” questions, with which the questioner helps the interlocutor find the answer. Consider the following dialogue fragment:
“Sir”, he said, looking Mr. Utterson in the eyes, “was that my master’s voice?”
“It seems much changed,” replied the lawyer; very pale, but giving look for look.
“Changed? […] Well, yes, I think so”, said the butler. “Have I been twenty years in this man’s house, to be deceived about his voice? No, sir; master’s made away with […]”
(R.L. Stevenson, The Strange Case of Dr. Jekyll and Mr. Hyde)

Jekyll’s butler poses a question (Was that my master’s voice?), but he knows already the answer. He wants that the lawyer recognizes that it is not his master’s voice. In fact, the lawyer seems to avoid the expected answer, but he grows “very pale”. The butler intends it as a sign that he tacitly agrees with him. This allows him to advance an argument (on rhetorical questions, see later), which is followed by the conclusion “Master’s made away with”.
Questions as requests for verbal action can vary according to the function of the answer required. In most cases, the answer is an assertive. In some cases, the answer is a directive and the corresponding question is called deliberative, because it makes a request for an advice or an order (“Well, what should I do?”). An answer can be also a commissive, e.g. when a question makes a request for a promise (‘Do you together promise you will love, cherish and respect one another throughout the years?’ Together they respond: ‘We do’). In other cases, the answer is a declarative (‘Do you [name] take [name] to be your lawful wedded wife/husband?’ Each responds: ‘I do’).
Some other questions make no request for an answer. Nevertheless they are “real” questions and can be used e.g. to present a problem (posing, not asking: Lyons 1977: 754), which requires an investigation (‘Where do noun phrases come from?’). Used in the syntactic form of a dependent interrogative clause, it represents the starting point of a Medieval quaestio: ‘[…] necessarium est primo investigare de ipsa sacra doctrina, qualis sit, et ad quae se extendat’ (Thomas Aquinas, Summa Theologiae, Prima pars, Quaestio I, Proemium).
Questions without request for an answer can be also used to express uncertainty on future events (a book published by Andrej Amal’rik in 1970 was entitled: Will the Soviet Union survive until 1984?). The speaker knows that nobody is able to give the desired information. He only tries to anticipate a future situation (‘er versucht eine Situation vorauszuerleben’, Nehring 1949: 47). This allows Adolf Nehring to declare that a question by its nature (Wesen) is ‘an uncertain proposition’ (‘eine unsichere Aussage’, Nehring 1949: 47). In this respect, a propositional question has much in common with a point of view, if we accept Houtlosser’s proposal that ‘a point of view is typically advanced in a context where doubt as to its acceptability is presupposed’ (Houtlosser 2002: 170-171).

3. Conducive questions and rhetorical interrogatives
Propositional questions can exhibit an expectation concerning the positive or the negative polarity of the answer. This expectation is usually made manifest by means of verbal devices. The most frequent is the positive or negative polarity of the interrogative structure used to make the question. We distinguish two great types of these questions: in the first type, the interrogative structure and the expected answer have the same polarity. In the other type, a contrast of polarity can be observed: a negative interrogative structure hints at an expected positive answer, and vice versa. The second type is best exemplified by the use of the so-called tags in English.
This contrast between the language plan and the content plan was first described by Per Restan (1972). It is quite relevant for the organization of the rhetorical interrogative structures, which we consider here as indirect assertions (‘indirekte sprachliche Handlungen des Behauptens’, see Meibauer 1976:185) or ‘hidden assertives’ (‘verkappte Aussagen’, Pérennec 1995: 111). In the interpretation of such utterances the illocution of a question is first hypothetically considered, then it is discarded, because it would not be reasonable, i.e. it would result as incongruous with respect to the communicative goals of the speech act in that specific speech event (see Rigotti, Rocci & Greco 2006).

In most cases, the polarity of rhetorically used yes-no interrogative sentences contrasts with that of the derived assertion:
“Have I been twenty years in this man’s house, to be deceived about his voice? No, sir; master’s made away with […]” (Stevenson, The Strange Case of Dr. Jekyll and Mr Hyde)

The indirect assertion is “I have not been twenty years in this man’s house, etc.”. This is an argument in favour of the standpoint, which follows as a conclusion. The use of a question together with its answer in a monologue is called ‘percontatio expositioque sententiae suae’ by Cicero (De oratore, III. 203).

Similarly, in the majority of rhetorically interpreted wh‑interrogative structures the derived assertion contains a positive universal quantifier, if the wh-word in the interrogative structure is negated; but it has a negated existential quantifier, if the interrogative structure is positive (Gobber 1999). Concerning rhetorical uses of interrogative structures, Sándor Karoly observes that their ‘characteristic feature’
[…] lies in the fact that from the point of view of their emotional effect, they appeal to the listener to respond, although they fail to produce the same effect from the viewpoint of the dialogue; here the interrogative sentence does not possess the interrogative-communicative role, but it has retained its interrogative-emotional role, the appealing character […] arouses a greater activity in the listener; the listener is going through, as it were, the experience of giving an answer (S. Károly, Kinds of sentences examined from the point of view of function and form, quoted by Restan 1972: 720-721, footnote).
Because they are indirect assertions, “rhetorical questions” – in fact, rhetorical uses of interrogative structures – can play the role of a standpoint in the domain of the confrontation stage or that of an argumentation in the domain of the argumentation stage (van Eemeren & Grootendorst 2004: 85). In these utterances the “interrogative-emotional” role serves as a booster of the indirect assertion: ‘acrior ac uehementior fit probatio’ (Quintilianus, Institutio oratoria, IX, 2, 6). The increased illocutionary force can then be exploited at other stages of a discussion (see Snoeck-Henkemans, in press).

4. Conduciveness and rhetorical uses: a continuum
There is often no clear boundary between a conducive question and an utterance with a rhetorically used interrogative. Often the respondent interprets a conducive question as a hidden assertive, or vice versa. The fuzziness of this boundary can be shown if we consider the following fragment of a dialogue. Here, a third person, named “Old J”, who is also the narrator, is invited to speak in favour of a standpoint or to testify the validity of an argument:
George said, ‘[…] I’m the only one who works […]
Harris laughed and said, ‘George! Work! Have you ever seen George work?’
I agreed with Harris. George never worked.
‘How do you know if I work, Harris? You’re always sleeping, except at meal times. Have you ever seen Harris awake, except at meal times?’ George asked me.
I agreed with George. Harris worked very little on the boat.
(Jerome K. Jerome, Three men in a boat)

‘Have you ever seen George work?’ is used by Harris to attack George’s standpoint (‘I’m the only one who works’). This question is posed to “old J”. It has a preference for a negative answer, as the reply (‘I agreed with Harris’) makes clear. J agrees with him, i.e. he understands Harris’ utterance as the assertion of an opinion, but also as a request for an assent. J’s answer provides evidence for Harris’ standpoint (‘I have never seen him work’ à ‘He does not work’). George counter-attacks Harris’ standpoint by questioning a condition of the assertive speech-act (‘You cannot know if I work’). To do this he makes an indirect assertion by means of a rhetorically used interrogative.
In its turn, George’s attack can be seen as a standpoint, which is followed by an argument (‘You’re always sleeping, except at meal times’). It should be observed that this standpoint is a conclusion of an enthymeme whose major premise is an implicit endoxon (‘When you are sleeping you cannot know what other people are doing’); the argument is the minor premise.
George then asks J for confirmation. J agrees that Harris is always sleeping etc., but from this he derives the conclusion that ‘Harris works very little on the boat’. This conclusion is reasonable because J has activated another enthymeme, which is based on the endoxon ‘When you are sleeping you do not work’.

Let us consider another example:
Estragon: […] Funny, the more you eat the worse it gets.
Vladimir: With me it’s just the opposite.
Estragon: In other words?
Vladimir: I get used to the muck as I go along.
Estragon: (after prolonged reflection). Is that the opposite?
Vladimir: Question of temperament.
Estragon: Of character.
Vladimir: Nothing you can do about it
(S. Beckett, Waiting for Godot, I)

The first question asks for a usage declarative (van Eemeren & Grootendorst 2004: 66), and preludes to a critical discussion. The second (‘Is that the opposite?’) implicates (in the Gricean sense) a negative judgement about Vladimir’s assertion that getting used should represent the opposite of ‘the more you eat the worse it gets’. Estragon’s question can be interpreted as a request for a justification of Vladimir’s assertion (‘Why do you think that it is the opposite?’), which has now received the position of a standpoint at the confrontation stage. Vladimir’s reply contains an implicit positive answer together with an explicit argument (‘Question of temperament’).

5. Concluding remark
In some questions which occur in the fragments considered above the respondent supports his answer with a justification.
It has been observed that the addition of an argument to an assertive can be a symptom of the speaker’s assumption that the interlocutor may have doubts about the acceptability of that assertive (Houtlosser 2002: 178-182).
As we have seen, the explanatory or argumentative follow-up to an assertive occurs often in the reply to a question. This could be explained by the fact that the respondent knows that the questioner does not merely request an answer, i.e. a move whose content saturates the open proposition of the question itself. He also expects that the respondent commits himself to the validity of that answer. An explanatory or argumentative follow-up is the most reasonable way to assure that the respondent commits himself to the validity of the answer.

REFERENCES
Ajdukiewicz, K. (1926/27). Analiza semantyczna zdania pytajnego (odczyt). Ruch Filozoficzny 10, 194-195.
Edmondson, G. (1981). Spoken Discourse. A Model for Analysis. London: Longman.
Eemeren F.H. van (2001). The State of the Art in Argumentation Theory. In: F.H. van Eemeren (Ed.), Crucial Concepts in Argumentation Theory (pp.11-26), Amsterdam: Amsterdam University Press.
Eemeren F.H. van & R. Grootendorst (2004). A Systematic Theory of Argumentation. The Pragma-Dialectical Approach. Cambridge: Cambridge University Press.
Gatti, M.C. (1992). Dalla semantica alla lessicologia. Brescia: La Scuola.
Gobber, G. (1999). Pragmatica delle frasi interrogative. Milano: ISU-Università Cattolica.
Houtlosser, P. (2002). Indicators of a point of view. In: F.H. van Eemeren (Ed.). Advances in Pragma-Dialectics (pp.169-184), Amsterdam: Sic Sat / Newport News, Virginia: Vale Press.
Kiefer, F. (1981). Questions and Attitudes. In: W. Klein &W. Levelt (Eds.), Crossing the Boundaries in Linguistics. Studies Presented to Manfred Bierwisch (pp.159-176), Dordrecht: Reidel.
Lyons, J. (1977). Semantics II. Cambridge: Cambridge University Press.
Meibauer, J. (1986). Rhetorische Fragen. Tübingen: Niemeyer.
Meinong, A. (1977). Gesamtausgabe. Band 4. Über Annahmen. Graz: Akademische Druck- u. Verlagsanstalt.
Nehring, A. (1949). Das Wesen der Fragesätze. Indogermanische Forschungen 61, 40‑54.
Padučeva, E.V. (1986). Question-answer correspondence. In: J.L. Mey (Ed.), Language and discourse: test and protest. A Festschrift for Petr Sgall (pp. 373-382), Amsterdam: Benjamins.
Pérennec, M. (1995). Partikeln und rhetorische Fragesätze. In: M. Schecker (Ed.), Fragen und Fragesätze im Deutschen (pp. 111- 122), Tübingen: Stauffenburg.
Restan, P. (1972). Sintaksis voprositel’nogo predloženija, Oslo. Bergen, Trømso: Universitetsforlaget.
Rigotti, E. (1993). La sequenza testuale. L’analisi linguistica e letteraria 1, 43-148.
Rigotti, E., A. Rocci & S. Greco (2006). The Semantics of Reasonableness. In: P. Houtlosser & A. van Rees (Eds.), Considering Pragma-Dialectics. A Festschrift for Frans H. van Eemeren on the Occasion of his 60th Birthday (pp. 257-274), Mahwah NJ, London: Erlbaum.
Searle, J. (1975). Indirect Speech Acts. In: P. Cole, J.L. Morgan (Eds.), Speech Acts (pp. 59-82). New York: Academic Press.
Snoeck-Henkemans, A. F. (in press). Manoeuvring strategically with rhetorical questions. Paper presented to the ISSA 2006 Conference.
Stati, S. (1990). Le transphrastique. Paris: PUF.

1. Introduction
In the law arguments from reasonableness play an important role. Judges often refer to reasonableness in ‘hard cases’ where there is a tension between the requirement of formal justice to treat like cases alike and the requirement of equity (or substantial justice) to do justice in accordance with the particularities of the concrete case. In such situations judges often use an argument from reasonableness to justify that an exception should be made to a general rule for the concrete case. However, the question arises how judges must account for the way in which they use their discretionary space in a situation in which they depart from the literal meaning of a general rule and establish the meaning of the rule for the concrete case on the basis of considerations of reasonableness and fairness. The central question I will answer in this paper is what an adequate justification based on an argument from reasonableness exactly amounts to from the perspective of the application of law in a rational legal discussion.
Although arguments from reasonableness are considered to be an important form of argumentation to defend a judicial decision in a hard case, in the legal literature little attention has been paid to the standards for argumentation underlying the justification of such a decision. Insight into such standards is important from the perspective of the rationality of the application of law because only on the basis of such standards it can be established whether the judge has used his discretionary power in an acceptable way. In order to establish the standards for an adequate use of arguments from reasonableness, I will develop an argumentation model that can be used for the analysis and evaluation of arguments from reasonableness.
In this paper I will proceed as follows. First in (2) I will discuss the legal background of the use of arguments from reasonableness and fairness and I will establish under what conditions they form an acceptable justification of a judicial decision. Then, in (3) I will develop an argumentation model for the analysis and evaluation of legal arguments from reasonableness to be able to make the underlying choices and assumptions explicit. In (4) I will apply this argumentation model to an example from Dutch law in which this form of argumentation is used and establish in what respects it can be considered an acceptable contribution to a rational legal discussion.

2. The role of arguments from reasonableness in a legal discussion
Judges use an argument from reasonableness to justify that in a concrete situation an exception should be made to a legal rule to avoid an unacceptable result. The need for an argument from reasonableness for this purpose can already be found in the classical literature with Aristotle who claims that an argument from ‘equity’ can be used as an argument to make an exception to application of a universal legal rule in a concrete case if this would yield un unacceptable result. A judge is allowed to correct the law on the basis of ‘equity’ if it would be unjust because of its generality. According to Aristotle, in such cases equity amounts to justice to correct the injustice that would be caused by strict application of a universal rule in a concrete case.[i]
A similar view is defended by Perelman (1979) who argues that the requirement of reasonableness is a requirement for the judge to apply the law in a just way, that is the requirement to treat like cases alike und unlike cases differently. This may result in an obligation for the judge not to apply a legal rule if application would be incompatible with the rational goal of the rule. A rational legislator can never have intended that a rule would be applied that would lead to a result that would conflict with the goal of the rule.

In most legal systems it is allowed to make such an exception on the basis of reasonableness and fairness.[ii] The general idea why it would be acceptable to make an exception to a legal rule on the basis of reasonableness and fairness is that the result of legal decisions should be reasonable and fair. The requirement of reasonableness implies that a judge should treat like cases alike and unlike cases differently. The requirement of fairness implies that the judge should apply the law in such a way that justice is done to the particularities of the concrete case.
Normally a judge can comply with these requirements by checking whether the conditions of a general legal rule are fulfilled. The question to be answered, however, is what a judge must do when the conditions of a legal rule are fulfilled but he is of the opinion that application of the rule would be unreasonable and unfair (or when the conditions are not fulfilled but application would still be reasonable and fair).
When a judge is of the opinion that an exception should be made on the basis of reasonableness and fairness, he can make the rule more concrete, he can supplement the rule, or he can correct the rule in such a way that a new rule for the concrete case is formulated. By creating a new ‘rule of exception’ the judge at the same time tries to do justice to the requirement of formal justice that like cases should be treated alike, as to the requirement of fairness that the circumstances of the concrete case should be taken into consideration. The idea behind this is that the legislator would have included a general exception for situations like the concrete case if he had thought of them. For this reason it is the obligation of the judge to formulate the rule of exception for the concrete case.
When making an exception, the judge cannot refer to the literal formulation of the rule. However, he can refer to the goal of the rule and/or general legal principles and show that the ‘new’ rule is in accordance with the ‘spirit’ of the law. The question that rises in this context is how the judge can give an adequate justification of the use of his discretionary power to formulate such a rule of exception.

In modern legal theory arguments from reasonableness are considered as a form of teleological-evaluative argumentation, that is argumentation in which an interpretation is justified by referring to the goals and values the rule is intended to realize.[iii] From this perspective it is considered as an argument based on an objective teleological interpretation in which the interpretation is justified by referring to the intention of a rational legislator who could not have wanted that application of the rule would lead to an unacceptable result. The intention of the legislator can be reconstructed by referring to the goals and values implemented in the general legal principles that are underlying the branch of law in question.[iv] From this perspective, when a judge uses teleological-evaluative argumentation, he must justify his decision by arguing that, in light of the personal and social interests involved in the concrete case, application in the strict literal meaning would have unacceptable consequences from the perspective of the goals and values the rule is intended to realize.[v]
On the basis of these considerations, in what follows, I will develop an argumentative model of the burden of proof for the use of arguments from reasonableness in cases in which judges makes an exception to a rule. I will do this by reconstructing the complex argumentation underlying the claim that application of a particular rule is unreasonable and unfair in the concrete case because application would lead to an unacceptable result that is incompatible with the goals and values of the rule in light of the circumstances of the concrete case.

3. An argumentation model for the burden of proof of a judge who uses an argument from reasonableness
A judge who argues that strict application of a rule in the concrete case would be unacceptable because application would be incompatible with reasonableness and fairness does this in the context of a dispute in which one party argues that the rule R must be applied in the literal meaning R”, and the other party argues that in the context of the concrete case the rule R must not be applied in the literal meaning R” but in the amended meaning R’ with an exception, so that the rule is not applicable to the concrete case.[vi] For the burden of proof of the judge who wants to make an exception, this implies that he has to justify why in the concrete case the rule R must be interpreted in the amended meaning R’ and not in the strict meaning R”. On the main level the decision and the main argumentation can be reconstructed as follows:

(1)
1. Application of rule X in the amended meaning X’ is reasonable and fair
1.1. Application of rule X in the amended meaning X’ leads to an acceptable result
and
2. Application of rule X in the strict meaning X” is unreasonable and unfair
2.1. Application of rule X in the strict meaning X” leads to an unacceptable result

This reconstruction of the complex standpoint and the main argumentation does justice to the fact that the judge has a burden of proof for defending a complex standpoint consisting of a preference for the amended interpretation and a rejection of the strict interpretation.

The burden implies that the standpoint must be supported with subordinative argumentation in which the judge specifies why the preferred interpretation 1 is coherent with certain legal goals or values which can be reconstructed from certain general legal principles underlying the relevant branch of law, as well as with the personal and social interests involved in the concrete case. He must also justify why the rejected interpretation 2 is incompatible with them.. These considerations, in their turn, must be supported with arguments that specify the legal and factual background of these arguments. A schematic reconstruction of the complex argumentation in support of the standpoint can be modeled as follows:

(2)
1. Application of rule X in the amended meaning X’ is reasonable and fair in the concrete case
1.1. Application of rule X in the amended meaning X ‘ leads to an acceptable result in the concrete case
1.1.1a. Application in the amended meaning X’ is compatible with the goals and values the rule is intended to realize of the rule
1.1.1a.1. The amended meaning X’ is compatible with the general legal principle R that is underlying the rules r1, r2….rn
1.1.1b. Application in the amended meaning X’ is compatible with the circumstances of the concrete case (the social and personal interests of the parties involved in the concrete case) C
1.1.1b.1. Statement about the social and personal interests in the concrete case (…)

In a similar way, the standpoint 2, that application of rule X in the strict meaning X” leads to an unacceptable result implying that application is unreasonable and unfair in the concrete case should be justified.
This reconstruction of the burden of proof from the perspective of a complete justification in the ideal case shows that the argumentation must consist of at least three levels of argumentation. The ‘core’ of this justification is formed by the argumentation on the second and third level where he must make the underlying choices and assumptions explicit by specifying why the amended meaning is coherent with the law and with the circumstances of the concrete case.
This reconstruction of the burden of proof into a model for the argumentative burden of proof of a judge who uses an argument from reasonableness clarifies his argumentative obligations. It makes clear under what conditions a judge lives up to his formal burden of proof from an argumentative perspective. Whether the arguments are acceptable from the material perspective depends on the criteria of acceptability in a specific field of law.

4. An exemplary analysis and evaluation of the use of arguments from reasonableness and fairness in Dutch civil law
To give an exemplary demonstration of how the argumentative model can be used for analyzing and evaluating concrete examples of arguments from reasonableness I will discuss a recent and representative example of the way in which the Dutch District Court uses the argument from reasonableness. The Court decided not to apply a rule in a concrete case on the basis of the consideration that an exception should be made because strict application would have unacceptable consequences from the perspective of reasonableness and fairness. This decision is based on a certain degree of discretion by the judge (because he limits the right of the defendant on the request of the plaintiffs) and it is therefore important to determine whether the way in which he accounts for this use of discretion is acceptable from the perspective of his burden of proof.

In Dutch civil law, in some cases an argument from reasonableness and fairness is an argument that is explicitly recognized as an acceptable argument by the legislator. On the basis of clause 6:248, 2 of the Dutch Civil Code the judge has the authority to make an exception to an arrangement by the parties on the basis of reasonableness and fairness if application of the arrangement would be unacceptable in the concrete circumstances:
Clause 6:48,2 – An arrangement that is valid between the creditor and the debtor on the basis of the law, a custom or a legal act, does not apply if this would be unacceptable from the perspective of the standards of reasonableness and fairness.

In book 3 of the Dutch Civil Code in the general clause of article 12 the legislator has formulated the following rule that specifies the factors that play a role in determining what can be considered as reasonable and fair:
When establishing what reasonableness and fairness require, generally accepted legal principles, legal convictions that are generally accepted in the Netherlands, and social and personal interests in the concrete case, should be taken into account.

These articles contain rules that specify under what conditions an argument from reasonableness and fairness is an acceptable argument to justify an exception to a legal rule. The articles also specify the factors a judge must mention to justify the exception.

Apart from cases covered by this article, also in other cases a judge can make an exception to a rule but he has a heavier ‘burden of proof’ which is in line with the obligations described in the previous section. First, he must explain why a strict application would lead to an unacceptable result by specifying why a strict application would be incompatible with the intention of the legislator. Second, he must specify why an exception would be compatible with certain factors specified in the above mentioned article 3:12 of the Civil Code such as generally accepted legal principles, and he must specify what the circumstances in the concrete case are that justify the exception by specifying which social and personal interests are relevant.
In the example, called the ‘Unworthy Grandson’, the Court uses an argument from reasonableness and fairness to justify that the legal rule of article 4:889 of the Dutch Civil Code about the right of a heir to his legal part of the inheritance should not be applied in the concrete case. The central question is whether someone who has been condemned to life imprisonment in Australia because he has killed his father and the wife of his father, has a right to his fathers legal part of the inheritance of his grandmother. (This example resembles the famous example used by Ronald Dworkin of the Riggs v. Palmer case in which the court denies the grandson Elmer who has killed his grandfather his inheritance on the basis of the principle that no one should profit from his own wrong.)[vii]

The Court decides that the rule of clause 4:889 jo and clause 4:960 of the Dutch Civil Code that give a child as a substitute a right to the legal part of the inheritance of a deceased parent, is not applicable in the concrete case because it would lead to an unacceptable result from the perspective of the underlying principle regarding unworthiness in the law of inheritance:
District Court Haarlem, July 24, 2001, nr. 68989 (Court of Justice Amsterdam, August 15, 2002, nr. 1304/01, NJ 2003, 53)

5.7 The exceptional situation of this case has not been foreseen by the legislator. But even if it would have been foreseen, this does not exclude that in certain circumstances the judge can appeal to the ‘derogating’ function of reasonableness and fairness if application of the law would lead to an unacceptable result.
5.8 The Court is of the opinion that in this case such circumstances obtain. The Court holds that the defendant acts in this special case as inheritor and statutory heir of his grandmother because he has killed his father, the inheritor in the first line.
(…)
5.10 The rules regarding unworthiness in the law of inheritance make explicit the underlying general legal principle to which the decision of the Supreme Court of December 7 1990 also refers, i.e. that someone should not profit form the intentionally caused death of someone else. In the light of this principle the right of the defendant to exercise his right to his legal share of the inheritance on the basis of clause 4:889 of the Dutch Civil Code would, according to the standards of reasonableness and fairness in the circumstances of this concrete case, lead to an unacceptable result.
5.11 The Court holds that in the present circumstances it is also important that the testatrix, who had suffered a great deal from what the grandson had done to her, had explicitly stated in her will that she did not want that the grandson would get a share of her inheritance. Although it is true that a testator cannot disinherit someone from his legitimate share to the inheritance, the right to the legitimate share is not absolute. In the present circumstances disobeying the will of the testatrix would conflict with the sense of justice in such a serious way that exertion of this right cannot be accepted.

Clause 4:885 of the Dutch Civil Code:
The following persons can considered to be unworthy to be an inheritor and can, for this reason, be excluded from the inheritance:
1. He who has been convicted of killing or trying to kill the deceased;

Clause 4:889 of the Dutch Civil Code:
1. Replacement in the direct downward line takes place infinitively.

The discussion takes place between the plaintiffs, the other inheritors, and the defendant, the grandson. The plaintiffs ask the court to deny the defendant, the grandson, his right to the inheritance because a strict application of clause 4:889 in the exceptional circumstances of the concrete case would, from the perspective of reasonableness and fairness, be so contrary to the purpose of the rule, that it would lead to an unacceptable result. The Court argues that in the concrete circumstances it can be justified to make an exception on the basis of reasonableness and fairness because application would result in an unacceptable consequence that would not be compatible with the purpose and purport of the rule.

The Court honors the claim and decides that this exceptional case has not been foreseen by the legislator (5.7) and that for this reason in these exceptional circumstances it can be justified not to apply the rule on the basis of the derogating function of reasonableness and fairness. On the basis of the general legal principle expressed in the famous case of the murder marriage, the unworthy spouse, of (HR 7 December 1990) someone should not profit from the intentionally caused death of someone else. In the light of this principle, in the circumstances of this concrete case, according to the standards of reasonableness and fairness it would be an unacceptable result if the defendant could exercise his right of legal heir on the basis of clause 4:889.

An analysis according to the model is as follows:
1. Application of rule X in the amended meaning X’ , implying that the rule is not applicable is to a person who has murdered his father, is reasonable and fair in the concrete case
1.1. Application of rule X in the amended meaning X ‘ leads to an acceptable result in the concrete case that the son who has murdered his father does not profit from the intentionally cause death of his father
1.1.1a . Application in the amended meaning X’ is compatible with the goals and values the rule is intended to realize, implying that it should be prevented that someone who is unworthy can inherit
1.1.1a.1. The amended meaning X’ is compatible with the general legal principle underlying the law of inheritance, that someone should not profit from the intentionally caused death of someone else, formulated by the Supreme Court in his decision of December 7 1990 (the unworthy spouse)
1.1.1b. Application in the amended meaning X’ is compatible with the personal interests of the parties involved in the concrete case, implying that it is in the present circumstances compatible with the sense of justice that the will of the testatrix is obeyed
1.1.1b.1. The testatrix, who had suffered a great deal from what the grandson had done to her, had explicitly stated in her will that she did not want that the grandson would get a share of her inheritance

The analysis demonstrates that the court, from the formal perspective, lives up to his burden of justification as specified in the model for his argumentative burden of proof. The exception is justified by three levels of argumentation specifying that the exception is in accordance with the law (1.1.1a.) and with the personal interests of the persons involved in the concrete case (1.1.1b.).
Whether the justification is acceptable from the material perspective depends on the question whether the support for the arguments 1.1.1a. and 1.1.1b. is acceptable. Argument 1.1.1a. can be considered as acceptable because it is defended by the legal principle mentioned in 1.1.1a.1. that is also based on a decision of the Supreme Court in the case of the ‘Unworthy spouse’. Argument 1.1.1b. can be considered as acceptable because it is adequately supported by 1.1.1b.1. in which it is specified that the history of the case makes clear that the personal interests of the testatrix are indeed in accordance with the decision to deny the grandson his right to his legal share.
In a similar way, the other line of argumentation supporting the claim that application in the strict meaning X” would be unreasonable and unfair can be analysed an evaluated.
This analysis and evaluation of an example show that the argumentation model makes it possible to reconstruct the underlying argumentation and to clarify the argumentative obligations of the judge that have to be met for the justification to be acceptable. If an argument from reasonableness can be reconstructed as part of the complex argumentation specified in the argumentation model and if a judges lives up to his formal and material burden of proof, an argument from reasonableness can be considered as an acceptable contribution to a rational legal discussion.

4. Conclusion
In this contribution I have developed a model for a rational reconstruction of arguments from reasonableness and fairness in the application of legal rules. The instrument offers a tool that can be used for the analysis and evaluation of all forms of complex argumentation in contexts in which the application of a legal rule is disputed and where the judge refers to reasonableness and fairness to make an exception to a rule. The model provides an a heuristic tool for reconstructing the argumentative steps that are required for a complete justification of the decision and it offers a critical tool by clarifying the elements of the justification that should be submitted to critique. By thus applying the instrument to examples from legal practice the gap between normative descriptions of forms of legal reasoning and legal interpretation on the one hand, and actual legal practice on the other hand can be bridged.

NOTES
[i] See Aristotle, Ethica Nicomachea (Book V, x).
[ii] See Hesselink 1999 for an overview of the role of reasonableness and fairness (good faith) in European law.
[iii] See MacCormick and Summers 1991:524 ff and MacCormick 2005: 132 ff.
[iv] See MacCormick 2005: 114 about the role of values as the grounds of evaluation of juridical consequences.
[v] See MacCormick 2005: 114 about the role of values as the grounds of evaluation of juridical consequences. For a more detailed description of the requirements of a justification in the context of teleolgocial-evaluative arguments see Feteris 2005.
[vi] For a more extensive description of such a model see Feteris 2005.
[vii] See Dworkin 1986:15-20

REFERENCES
Alexy, R. (1989). A theory of legal argumentation. The theory of rational discourse as theory of legal justification. Oxford: Clarendon Press. (Translation of Theorie der juristischen Argumentation. Die theorie des rationalen Diskurses als Theorie der juristischen Begründung. Frankfurt a.M.: Suhrkamp, 1978).
Aristotle. The Nicomachean Ethics. Translated with an introduction by David Ross. Oxford/New York: Oxford University Press.
Bell, J. (1983). Policy arguments in judicial decisions. Clarendon Press, Oxford.
Burg, E. (2000). The model of principles. Dissertation University of Amsterdam.
Dworkin, R. (1986). Law’s empire. London: Fontana.
Eemeren, F.H. van & R. Grootendorst (1992). Argumentation, Communication and Fallacies. Erlbaum, Hillsdale, NJ.
Eemeren, F.H. van & R. Grootendorst. (2004). A systematic theory of argumentation. The pragma-dialectical approach. Cambridge University Press, Cambridge.
Eskridge, W.N. jr. (1994). Dynamic statutory interpretation. Cambridge: MA:.Harvard University Press.
Feteris, E.T. (1999). Fundamentals of legal argumentation. Dordrecht etc.: Kluwer.
Feteris, E.T. (2002a). ‘A pragma-dialectical approach of the analysis and evaluation of pragmatic argumentation in a legal context’. Argumentation, Vol. 16, No. 3, 349-367.
Feteris, E.T. (2002b). ‘The role of arguments from consequences in practical argumentation in a legal context’. In H. Hansen C.W. Tindale, J.A. Blair, R.H. Johnson, R.C. Pinto (eds.), Argumentation and its applications, Ontario Society for the Study of Argumentation. (CD-rom)
Feteris, E.T. (2003). ‘The rational reconstruction of pragmatic argumentation in a legal context: the analysis and evaluation of teleological argumentation’. In: F.H. van Eemeren, J.A. Blair, C.W. Willard (eds.), Proceedings of the fifth ISSA conference on argumentation. Amsterdam: Sicsat.
Feteris, E.T. (2004). ‘Arguments from unacceptable consequences and a reasonable application of law’. In: J.A. Blair, D. Farr, H.V. Hansen, R.H. Johnson, C.W. Tindale (eds.), Informal Logic@25. Windsor, ON: OSSA (CD-rom)
Feteris, E.T. (2005). ‘The rational reconstruction of argumentation referring to consequences and purposes in the application of legal rules’. Argumentation, Vol. 19, No. 4, 459-470.
Fuller, L. (1948). ‘The case of the Speluncean Explorers’ law’, Harvard Law Review, Vol. 62.
Fuller, L (1958). ‘Positivism and the fidelity to law – A reply to professor Hart’, Harvard Law Review, Vol. 71, 630-672.
Golding, M. (1984). Legal reasoning. New York: Knopf.
Gottlieb, G. (1968). The logic of choice: An investigation of the concepts of rule and rationality. London: Allen and Unwin.
Hesselink, M.W. (1999). De redelijkheid en billijkheid in het Europese privaatrecht (Good faith in European private law). Dissertation Utrecht. Dordrecht: Kluwer.
Lyons, D. (1993). Moral aspects of legal theory. Essays on law, justice, and political responsibility. Cambridge: Cambridge University Press.
MacCormick, D.N. (1978). Legal reasoning and legal theory. Oxford: Clarendon Press.
MacCormick, N and R.S. Summers. (1991). Interpreting statutes. A comparative study. Aldershot etc.: Dartmouth.
Nozick, R. (1993). The nature of rationality. Princeton: Princeton University Press.
Peczenik, A. (1989). On law and reason. Dordrecht etc.: Kluwer.
Pontier, J.A. & E. Burg. (2004). EU Principles on jurisdiction and recognition and enforcement of judgments in civil and commercial matters according to the case law of the European Court of Justice. The Hague: Asser Press.
Summers, R.S. (1978). ‘Two types of substantive reasons: The core of a theory of Common-Law justification’, Cornell Law Review, 63, 707-788.
Wróblewski, J. (1992). The judicial application of law. (Edited by Zenon Bankowski and Neil MacCormick). Dordrecht etc.: Kluwer.

1. Introduction: The alleged scope of the Critical Discussion model
This paper is an investigation of the scope of the Pragma-Dialectical theory of argumentation, and in particular of its ingredient concept of a Critical Discussion (see van Eemeren and Grootendorst 1984, 1992, 2004). The Pragma-Dialectical theory is explicitly designed to apply to argumentation aimed at the rational resolution of a difference of opinion, what Walton and Krabbe (1995) call “persuasion dialogues.” On the face of it, there are other uses of argument besides attempting to resolve disagreements, for example, to inquire about what is true. But the proponents of Pragma-Dialectics seem either to regard that theory has having universal application or to regard all uses of argument as reducible to disagreement-resolution argumentation. The evidence for this claim is found in their application of a central component of their theory, the model of a Critical Discussion.

A “Critical Discussion,” as that term is defined in the Pragma-Dialectical theory, is a technical concept – a term of art (hence here capitalized, as is ‘Pragma-Dialectics’ and its cognates, for the same reason). The Pragma-Dialectical Critical Discussion is an ideal model of a discussion between two parties with a difference of opinion who agree to use arguments and follow an instrumentally rational procedure in doing so to try to resolve their difference. The model aims to specify the conditions that an actual argumentative exchange would satisfy if the parties were orderly and reasonable. They would order their discussion in the way best designed to resolve their disagreement, and they would carry out their discussion according to norms that make it rational for them to agree to (or to decline to) make concessions and to accept (or to reject) alleged implications. In the end, one party would convince the other to withdraw the commitment or the doubt that started the discussion, or the parties would remain in disagreement.
The model of a Critical Discussion is introduced as applicable to argumentation exchanges aimed at resolving a difference of opinion, but it is taken to apply generally, as the following passage in van Eemeren and Grootendorst’s latest statement of their theory (2004) makes clear:

The aim of a pragma-dialectical analysis is to reconstruct the process of resolving a difference of opinion occurring in an argumentative discourse or text. This means that argumentative reality is systematically analyzed from the perspective of a critical discussion…. What exactly does such an analytic reconstruction of an argumentative discourse or text entail?… In the reconstruction, the speech acts performed in the discourse or text are, where this is possible with the help of the ideal model of a critical discussion, analyzed as argumentative moves that are aimed at bringing about a resolution of a difference of opinion. (p. 95)
Notice the glissement that occurs in this text – not from one dialogue type to another (see Walton and Krabbe 1995, p.102) – but from one claim to another. The first sentence notes that a pragma-dialectical analysis focuses on the process of argumentation aimed at resolving a difference of opinion. But the very next sentence mentions subjecting “argumentative reality” – without qualification now – to analysis from the perspective of the Pragma-Dialectical model of a “critical discussion.” So argumentation aimed at resolving a difference of opinion is quickly identified with argumentative reality in general. An argumentative discourse or text, the next paragraph declares, is to be reconstructed, using the ideal model of a critical discussion. And what the authors are referring to is a Critical Discussion – the ideal construct they designed expressly to model argumentation aimed at rationally resolving a difference of opinion. So any argumentation text or discourse is to be modeled as if it were argumentation aimed at bringing about a resolution of a difference of opinion.
The text quoted above makes clear the commitment of the proponents of the Pragma-Dialectical theory. It has two related aspects. One is to assimilate all argumentation to opinion-difference resolution argumentation. The other is to treat the ideal model developed as an element of that theory as applicable to any argumentation whatever. From a perspective internal to the theory, these are two sides of the same commitment, but they are two distinct claims, since it is in principle possible for either to be false while the other is true. It is in principle possible that not all argumentation functions to resolve a difference of opinion yet the critical discussion model can be usefully applied generally. And it is in principle possible that all argumentation does boil down to attempts to resolve differences of opinion rationally yet the critical discussion model is flawed and does not apply as neatly as its proponents believe it does.

2. Epistemic investigations
There is a type of argumentation transaction that is, on the face of it, different from one whose purpose is to resolve a difference of opinion. I will call it the use of arguments to conduct an epistemic investigation. I elsewhere have called this the use of argumentation to inquire (Blair 2005), but Walton (e.g., 1998) has used the term ‘inquiry’ to name a kind of proof-seeking dialogue, which is different; hence the need for a different name. The question of this paper can be raised in relation to epistemic investigations. Does the Critical Discussion model apply to them? If it does, then there is some basis for thinking that other uses of arguments besides its use to resolve a difference of opinion can be assimilated to persuasion dialogues. If it does not, then the Pragma-Dialectical theory’s scope is more limited than its proponents claim.
Understand by an “epistemic investigation” an attempt to ascertain the epistemic standing of some proposition or group of propositions. I am using ‘epistemic’ in a broad sense. An epistemic investigation begins with a question about whether some judgement is justified. By a judgement here I mean an attitude towards a proposition (e.g., that it is true, or that it is probable or that it is plausible to some degree), or a proposal (e.g., that an action should be performed or that a policy implemented), or an assessment (e.g., that something or someone has some instrumental, moral or aesthetic quality), and so on. Perhaps the standard philosophical connotations of ‘epistemic’ militate against this stretch of the term, for epistemology is paradigmatically occupied more narrowly with the grounds of knowledge and reasonable belief, and these are widely thought to have propositions as their objects – propositions in the sense of what are expressed by declarative sentences and that are true or false. Recommendations and evaluations are held by some not to have truth-values. However, they do have values. A proposal can be wise or foolish, correct in the circumstances or mistaken, good or bad. Similarly, an assessment can be accurate or mistaken, sound or wrong. We can be justified in such judgements as that a piece of advice was poor advice, if only in retrospect; we can similarly be justified in evaluative judgements. We can and do make such determinations based on reasons, and we act on them with more or less success and innocent of any conceptual blunder. So it seems useful to use ‘proposition’ in the wider sense and to allow the scope of epistemology to include such judgements within its domain.
There is a question about the epistemic standing of some proposition if there is some other proposition or group of propositions that, if true, imply that the epistemic standing of the proposition in question is different from what it is alleged to be, and there is some reason to believe one or more of the alternatives. This would be the case, for instance, if there are one or more incompatible propositions that have or seem to have an equal or higher epistemic standing, though that is just one type of argument supporting the conclusion that the proposition’s epistemic standing is different from what it was claimed or seemed to be.

So one way to investigate the epistemic standing of a proposition is to look for arguments that go to support it and for arguments that go to undercut or block alternative possible claims about its epistemic standing, and also to look for arguments that go to refute it or support alternative possible claims about its epistemic standing, and then to assess how cogent those arguments are. This is what I mean by an epistemic investigation. The attribution of burden of proof in an epistemic investigation is crucial, for some ways of assigning the burden of proof make the task of investigating the epistemic standing of a proposition an infinite one, and it is pointless to conduct such an investigation if there is no prospect of completing it. The following points may be made about the burden of proof in an epistemic investigation.
Beliefs or assertions or other commitments have a weak presumption in their favour. That is, there is a weak burden of proof to establish that the alleged epistemic standing of some proposition is open to question. Just questioning or doubting a proposition does not oblige anyone who asserts or is otherwise committed to it to support it, for otherwise, an infinite regress of challenge and response would be possible and in that case the epistemic investigation would have no prospect of ending. But the presumption is weak because it is easily overcome. For instance, it is enough that others are known to have incompatible beliefs or to be committed to incompatible propositions, for in that case the question as to which one is justified is legitimately raised. The existence of two or more plausible answers to a question about what is or ought to be the case (and so on) is sufficient to impose a burden of proof on whoever would contend that one of them is true. Thus, when someone is aware of two or more plausible answers to such a question, and one does not know which one to prefer or maintain, there exist the conditions for the beginning of an epistemic investigation. I will call a plausible answer to a question that gives rise to an epistemic investigation an hypothesis. (An hypothesis is plausible if it is consistent with current beliefs.)

For any simple argument in support of one hypothesis – call this a pro argument – there is, for the reason just given, again a presumption in favour of its premises and the inference from them to its conclusion. Similarly, for any simple argument against that hypothesis – a contra argument – there is a presumption in favour of its premises and inference. In neither case is there a burden of proof to support the argument in the absence of any reason to question or challenge it. However, if there are both a pro and a contra arguments relating to an hypothesis, or if there is a pro argument for one hypothesis and a pro argument for an incompatible hypothesis, then the presumptions are cancelled. For when there are two opposing arguments in one or another of these ways, then at least one of them must be mistaken, so there is a reason to require that it be shown of any of them that it is not the mistaken one.
If one wants to ascertain the correct or best hypothesis among alternatives on a question about what is or ought to be the case (and so on), then one has a motive for conducting an epistemic investigation.

3. Elements of an epistemic investigation
An epistemic investigation will begin, then, with the following situation: There is a question that has two or more possible plausible but incompatible answers or hypotheses – that is, if any one hypothesis is correct or true (etc.) then the others are mistaken or false (etc.), and either (a) for each of two or more hypotheses there is at least one person who seriously supports it, or (b) for each hypothesis there are considerations that support it, or (c) for at least one hypothesis there are one or more considerations for it and one or more considerations against it – and at least one person wants to ascertain which hypothesis is best or correct.
An epistemic investigation will proceed by one or more parties completing the following elements. (I speak of “elements” of the procedure rather than of stages, because there is no “right” temporal ordering to these elements, and ‘stages’ carries temporal connotations.) In general, evidence must be gathered, assessed, revised in various ways (with a view of strengthening it by addition, modification or subtraction), and its upshot judged. The objective is to make a judgement about the epistemic status of the proposition in question – the hypothesis – on the basis of weighing the best case that can be made in favour of it against the best case that can be made against it, and comparing the upshot to similar assessments of the alternative hypotheses. I use ‘evidence’ in a broad way to include any considerations, not just empirical data, relevant to the truth of a hypothesis. Such considerations can be expressed as arguments pro or contra the hypothesis.

1. Evidence-gathering element. Set out the pro and contra arguments for each hypothesis, seeking to produce a complete inventory of the arguments that have historically been given and also that imagination and further research can devise.
2. Evidence assessment element. (2.1) Seek out or construct critical arguments – arguments for doubting or for rejecting the premises of the evidentiary arguments or for doubting or rejecting the justificatory force of the evidentiary arguments. (2.2) Assess the critical arguments by seeking plausible replies to (i.e., arguments against) the critical arguments on behalf of the evidentiary arguments and assess the merits of those replies (i.e., those arguments).
3. Evidence revision element. Revise the evidentiary arguments in light of the assessment. Some might have to be abandoned because they have been refuted; some might be amenable to repair by altering them to avoid objections or by finding additional evidence as required by the assessment; and some might survive unchanged.
4. Hypothesis revision element. Should there be strong evidence that an hypothesis as it was initially formulated is false, but that a reformulated hypothesis would not be subject to those objections, then that hypothesis may be revised and the investigation continued into the merits of the revised hypothesis. Elements (1) to (4) can have as many iterations as resources allow.
5. Concluding element. Decide, on the basis of the assessments of the strengths of the pro and contra arguments, on the epistemic standing of the hypotheses under investigation.

4. Are epistemic investigations Critical Discussions?
Such an investigation can be modeled as a two-party dialogue, or as a group of nested dialogues. A dialogue is a conversation between the occupants of two roles. One role can be conceived as the questioner or critic and the other as the answerer or proponent. In an epistemic investigation, all investigators occupy both roles in turn, since the goal is to test each hypothesis as thoroughly as possible and not for one role occupant or the other to prevail. Are these roles identical to the roles of antagonist and protagonist in the Pragma-Dialectical theory’s ideal model of a Critical Discussion? Does the Critical discussion model apply to epistemic investigations?
From the point of view of the purpose and nature of an epistemic investigation, it seems not to fit the Critical Discussion model. An epistemic investigation has a different starting point and a different objective from a persuasion dialogue. There are not two sides or parties who disagree; neither party is trying to convince the other of anything; all parties take both a pro and a contra perspective, seeking both to find arguments that support an hypothesis and to refute the very arguments that they have just found.
In addition, it seems that the discussion rules for the two enterprises will differ in many ways. An investigation does not get started by incompatible commitments, but by an absence of commitment on an issue on which the parties all want to decide what commitment is justified. In a critical discussion, the burden of proof is asymmetrical: who asserts must defend; who questions has no obligation to defend. In an investigation, the burden of proof is complicated. The investigators have an obligation to seek both pro and contra arguments, but once any argument has been formulated, the burden of proof must lie with the “critic,” not the “proponent” – the argument stands until some further argument establishes a weakness in it, for otherwise there would be a vicious infinite regress, a requirement of arguments supporting arguments ad infinitum. At the same time, however, all the investigators have an obligation to seek such critical arguments. Thus no investigator consistently occupies the role of protagonist or of antagonist, as must occur in a Critical Discussion. Also, unlike in a Critical Discussion, the investigators do not agree independently about what may constitute premises or legitimate kinds of support. In an investigation, any grounds that can be found may be put forward and their appropriateness, relevance, and strength of support are subject to critical examination as part of the assessment element. As well, revisions to arguments and, indeed, to hypotheses, are permissible throughout an investigation, since the object is not to maintain one’s initial position, but to follow the evidence to the truth. So two enterprises of epistemic investigation and disagreement resolution seem to be quite different. And finally, there is no philosophical assumption of Popperian rationality. It is an open question whether there are objective truths or whether the best “truth” available just is what investigators agree to at the moment, subject to future disagreement.

On the other hand, from the point of view of the inner workings of an epistemic investigation, the Critical Discussion model does seem to have application. Consider any single hypothesis being investigated. It can be thought of as a standpoint that has been asserted by its protagonist. The requirement to produce arguments in its favour can be treated as an obligation incurred by the questioning of that standpoint by an antagonist. The arguments produced against it are like the argumentation required of an antagonist in a multiple dispute (see van Eemeren and Grootendorst 1984, p. 80). The assessments of those arguments can be conceived as the argumentation of sub disputes (ibid., p. 89). The revision of any argumentation is like a concession, and the revision of any hypothesis can be treated as the defeat of the original hypothesis, and any examination of the revised argument or hypothesis can be treated as a new discussion occasioned by the new argument or new hypothesis. So it seems that an epistemic investigation can indeed be analysed as if it were a series of Critical Discussions.

What has happened? It seemed clear that an epistemic investigation is a different use of argumentation from the use of argumentation to resolve a difference of opinion. And yet it also seemed clear that the model of a Critical Discussion developed for the analysis and assessment of argumentation aimed at the resolution of a difference of opinion applies equally well to the argumentation of an epistemic investigation. This is the puzzle.
The solution I propose is to regard the ideal model of a Critical Discussion is a chameleon. When it is at home in the Pragma-Dialectical theory, it is applied to the argumentation designed to resolve a difference of opinion, and it accommodates the Popperian epistemology underlying the Pragma-Dialectical theory. But when is applied to other uses of argumentation, it changes its coloration. It models simply the interchange of pro and contra argumentation, including meta-argumentation (arguments about the arguments). It is neutral with respect to any particular epistemology. It does not require that the role-occupiers be committed to the initial positions that occasion the exchange of arguments. It is not committed to the four stages of the Pragma-Dialectical account While it does, as any model of argumentation must, allow only for the interlocutor’s contributions to any particular exchange of arguments, it does not require the assumption that there is no basis for claims or arguments apart from what the interlocutors agree to. In other words, the accretions belonging to the Pragma-Dialectical approach to argumentation are dropped.
This equivocation of the critical discussion model can be treated either as a weakness or as a strength. On the one hand, there is sleight of hand at work in the suggestion that precisely as formulated as part of the Pragma-Dialectical approach to argumentation, the Critical Discussion model applies to any argumentation whatever. On the other hand, if the model or a critical discussion (now spelled with lower-case first letters) is detached from all the Pragma-Dialectical philosophical assumptions and expressed in general terms (so that its Pragma-Dialectical version is a special case), then it is plausible to think that it applies to any argumentation. For any argumentation will have the generic properties identified by the general features of the critical discussion model. Any argumentation will have different components – what Pragma-Dialectics calls “stages” and what I have called “elements.” There must be some initiation and some conclusion to the argumentation exchange. There must be argumentative roles assigned, and burdens of proof distributed. There must be regulatory rules specifying the conditions of a well-ordered argumentative exchange, rules for turn-taking, commitments and concessions, and so on.

This generalizing of the Critical Discussion model is, in effect, what Walton and Krabbe (1995) have already done, although they continue to call the model a critical discussion. But they introduce a crucial modification of the Pragma-Dialectical theory’s formulation of a Critical Discussion, and in so doing they effectively generalize the model. They write, “in our usage, the term dispute will stand for a type… of dialogue rather than for a type of conflict” (69). This switch from conflict type to dialogue type makes all the difference, because they are now modeling the argumentative exchange and not the motivating problem – such as a disagreement between two parties as distinct from a puzzle about what stand to take on some vexed question. The type of dialogue they model is one in which at least two incompatible propositions are in competition for endorsement, acceptance or belief, and that is the situation when two or more hypotheses are mooted as plausible positions to take on some problematic issue. There are not two (or more) parties in dispute; there are two or more positions up for consideration.
Although welcoming their modification, I am suggesting a somewhat different analysis than that proposed by Walton and Krabbe. They assume that a critical discussion, or what they prefer to call a “persuasion dialogue,” is the most fundamental kind of argumentative dialogue, although during critical discussions, other types of dialogues, such as negotiations and quarrels, occur as well (1995, p. 7). If the present argument is correct, the persuasion dialogue understood in Pragma-Dialectical terms is not the most fundamental kind of argumentative dialogue, however important it might be. There is at least one other important kind of argumentative dialogue, namely, the epistemic investigation.
To see the distinction more clearly, consider what we are modeling. Is it a type of argumentation (distinguished by its purpose) or is it the exchanges that occur within any type of argumentation? Any type of argumentation entails arguments pro or contra (plus at least the possibility of arguments on the other side). But not every type of argument entails one person attempting to persuade another, or one party differing in opinion from the other, for an epistemic investigation entails the possibility of one or more person with no opinion (and hence nothing to differ from) attempting to discern what opinion to take.

Proponents of the Pragma-Dialectical theory might try to assimilate these two uses of argument. One person trying to decide what to believe or what position to take, they might say, is someone with two (or more) minds about a question, and so is, in effect (and from a modeling perspective, identical to), two (or more) people disagreeing with one another. But that is not the case. The person in this situation does not have two or more opinions; ex hypothesis, the person has no opinion. The metaphor of “being of two minds” does not indicate having two opinions; it indicates being undecided as between two (or more) alternatives, seeing the prima facie merits of two incompatible positions, and being unable to choose between them. In fact, it is impossible to disagree with oneself. (One can at a given time disagree with one’s position at an earlier time, but that is changing one’s mind, not disagreeing with oneself.) Being undecided as between which of two alternatives to believe or commit to is not the same as believing or maintaining both of them; quite the contrary, it is being committed to neither of them. Rather than a single-person epistemic investigation being a just a variant of a multi-party investigation, it’s the other way around: a multi-party epistemic investigation is no different from a single-person investigation, except that it has more resources.
It is true that the single investigator must formulate pro and contra arguments, and so can be said to have to occupy two roles – the roles of protagonist and antagonist. This is what makes an investigation a dialectical enterprise. But the investigator, unlike the persuader, occupies both roles (and if there is more than one investigator, they all occupy both roles). It is certainly possible to speak of the pro or the contra arguments “winning” and the arguments on the other side “losing.” But that is strictly a metaphor borrowed from debate, a short-hand way of referring to the fact that the arguments on one side are more compelling, on balance, than those on the other, and that the position they support merits (more or less qualified) endorsement. It becomes thereby a (relatively) justified opinion, and the investigator now has a reasonable basis for disagreeing with anyone who rejects or refuses to accept that opinion.
Finally, notice that so-called “strategic maneuvering” (van Eemeren and Houtlosser 1999, 2000a, 2000b, 2002a, 2002b, 2003) has no place in an epistemic investigation. The selection of how the topic is framed (“topical potential”), the adaptation of the arguments to be responsive to “audience demand,” and the use of the most effective “presentational devices” that characterize rhetorical choices within the Critical Discussion framework are all aimed at prevailing in a competitive, zero sum argumentative discussion. In an epistemic investigation, there is no motivation for such maneuvering, since the goal is not to persuade an interlocutor, not to “win” for one’s opinion, but to get as close to the truth of the matter as possible.

5. Conclusion
The conclusion that emerges from these considerations is that there are two concepts of the model of a critical discussion. There is the special model of a Critical Discussion that applies within the Pragma-Dialectical theory to argumentative discussions aimed at a rational resolution of a difference of opinion. And there is the general model of a critical discussion that applies to other kinds of dialectical argumentation. I would speculate that it applies to any exchange of arguments that has their critical assessment as an essential property. The scope of the Pragma-Dialectical theory and its special model of a Critical Discussion are overstated. At the same time, that overstatement is understandable, because the special model can be generalized, and when it is, it has broad application.

REFERENCES
Blair, J.A. (2005). Argument and its uses. Informal Logic, 24.2, 137-151.
Eemeren, F.H. van and R. Grootendorst. (1984). Speech Acts in Argumentative Discussions. Dordrecht: Foris Publications.
Eemeren, F.H. van and R. Grootendorst. (1992). Argumentation, Communication, and Fallacies: A Pragma-Dialectical Perspective. Hillsdale, NJ: Lawrence Erlbaum Associates.
Eemeren, F.H. van and R. Grootendorst. (2004). A Systematic Theory of Argumentation: The Pragma-Dialectical Approach. Cambridge: Cambridge University Press.
Eemeren, F.H. van and P. Houtlosser. (1999). Strategic maneuvering in argumentative discourse. Discourse Studies 1, 479-497.
Eemeren, F.H. van and P. Houtlosser. (2000a). Rhetorical analysis within a pragma-dialectical framework. Argumentation 14, 293-305.
Eemeren, F.H. van and P. Houtlosser. (2000b). Managing disagreement: Rhetorical analysis within a dialectical framework. Argumentation and Advocacy 37, 150-157.
Eemeren, F.H. van and P. Houtlosser. (2002a). Strategic maneuvering with the burden of proof. In F.H. van Eemeren (Ed.), Advances in Pragma-Dialectics (pp. 13-28, Ch. 1). Amsterdam: SicSat; Newport News, VA: Vale Press.
Eemeren, F.H. van and P. Houtlosser. (2002b). Strategic maneuvering: Maintaining a delicate balance. In F.H. van Eemeren and P. Houtlosser (Eds.), Dialectic and Rhetoric: The Warp and Woof of Argumentation Analysis. Dordrecht: Kluwer Academic Publishers.
Eemeren, F.H. van and P. Houtlosser. (2003). Fallacies as derailments of strategic maneuvering: The argument ad verecundiam, a case in point. In F.H. van Eemeren, J.A. Blair, C.A. Willard and A.F, Snoeck Henkemans (eds.), Proceedings of the Fifth Conference of the International Society for the Study of Argumentation (pp. 289-292) Amsterdam: SicSat.
Walton, D.N. (1998). The New Dialectic: Conversational Contexts of Argument. Toronto: University of Toronto Press.
Walton, D.N. and E.C.W. Krabbe. (1995). Commitment in Dialogue: Basic Concepts of Interpersonal Reasoning. Albany, NY: State University of New York Press.