ISSA Proceedings 2002 – Metadialogues

1. Introduction
A metadialogue is a dialogue about a dialogue or about some dialogues. A dialogue that is not a metadialogue will be called a ground level dialogue. For instance, let the ground level dialogue be an argumentative discussion aiming at the resolution of some dispute. Then disagreement about the correctness of some move in this dialogue will constitute another dispute which the parties again may try to resolve by dialogue. This dialogue will then be a metadialogue relative to the first dialogue. It will be about this first dialogue and perhaps some related dialogues. Also, its primary purpose is to help this first dialogue achieve its end: in this sense the metadialogue will be embedded in the ground level dialogue.

Three problems arise, given this concept of metadialogue:
1. A demarcation problem. Some critical moves seem plainly to belong to the ground level. For instance, a critic’s asking for argumentative support within a context of critical discussion, though in some sense being about the preceding dialogue, would not be analysed as a move that starts a metadialogue. At least it would be very much strained to do so. Many moves on the ground level can be looked upon as asking for, or installing, conversational repairs, but are not, usually, for that reason classified at the metalevel. On the other hand a dispute about the allotment of speaking time would be so classified. Criticism of fallacies seems to lie somewhere in between. Where to draw the line?
2. A problem of infinite regress. If from any critical discussion one can move up (or down, whatever metaphor you prefer) to a metadialogue that constitutes another critical discussion, this may launch us into an infinite regress. A discussion about the rules of ground level dialogue may open up a discussion about the rules governing discussions about ground level rules, and so on. Can this regress be blocked?
3. An equity problem. Some retreats into metadialogue seem quite reasonable and bound to help the ground level dialogue proceed. In other cases one is confronted with nit-picking or completely unwarranted charges. On the one hand each party should have a right to contest the correctness of any ground level move, on the other hand its adversary should not be left without means of defense. Can we strike a balance?

There are other questions besides these problems, such as whether the metadialogues of a persuasion dialogue (critical discussion) must always themselves be of the type of a persuasion dialogue. Could they sometimes be of some other type, say negotiation? Further there is the question of how to formulate rules that regulate the opening and closing of metadialogues and the effect of these dialogues on commitment stores. These questions can be raised both from a descriptive and from a normative point of view.
The purpose of this paper is to explore these problems and questions. Definite solutions will not be reached. Let us first look at some examples.

2. Examples
People from all generations complain about the deterioration of something. Aristotle is no exception. In former times it was still easy to make one’s adversary admit some false or paradoxical proposition. One had just to ask a lot of questions and insist that one’s interlocutor speak his mind, then sooner or later he would be led to falsehood or paradox. “This unfair method, however, is [nowadays] much less practicable than formerly; for people demand, ‘What has this to do with the original question?'” (Aristotle, 1965, 69, De Soph. El. 12, 172b19-21). Thus, by their critical attitude, so Aristotle seems to complain, people spoil the questioning: they retreat into metadialogue. The case is of course also known from cross-examination in court. “Where do all these questions lead to?”. “A moment, Your Honor, and it will become clear how relevant these questions are.” Here the judge is to decide upon the metaquestion. In other cases some dialogue about how much time is allowed for further questioning may be needed. This will often be a negotiation dialogue.

Metadialogue was not new in Aristotle’s time. In Plato’s Euthydemus we find the following example:
[Socrates (first-person narrator) just asked a question. Dionysodorus sees refutation looming and tries to avoid giving an answer.]
[Dionysodorus:] … Just answer me.
Before you anwer me? I said.
Won’t you answer? he said.
Is that fair?
Quite fair, he said.

On what reasoning? said I. Is not this your reasoning –  that you visit us as one all-wise about words, and you know when you are bound to answer and when not, and now you will not answer anything since you perceive that you are not bound?
You just chatter, he said, without troubling to answer. Come, my good man, do as I say and answer, since you yourself admit that I am wise.
Then I must do as you say, said I, and I can’t help it, as it seems, for you are master. Ask away.
(Plato, 1961, 401, Euthydemus 287c-d)

In this passage the discussants suspend their discussion to start a metadiscussion about roles: who is to be the Questioner, who the Answerer? We see that that Socrates, ironically, hands out an argument to Dionysodorus to support his claim on being the Questioner. The whole metadialogue is a kind of mock persuasion dialogue. At the end, Socrates agrees to be the Answerer, thus giving in to a most unfair swap of roles. For us it is important to note the possibility that a discussion may give occasion to a metadialogue on the division of roles.

Another Socratic dialogue provides a more extended example of metadialogue. I refer to Protagoras 334c-338e, a passage too long to quote in its entirety. The situation is that Protagoras has just been trying to escape from his role as an Answerer by delivering a (rather short) speech on another subject. The speech is much applauded by the audience. But Socrates complains that he cannot follow long speeches:
… I said, ‘Protagoras, I happen to be a forgetful sort of person, and if someone speaks to me at length, I forget what he is talking about. It’s just as if I were a trifle deaf; in that case you would think it right to speak louder than usual, if you were going to talk to me. So now, since you are dealing with someone with a bad memory, cut your answers short and make them briefer, if I am to follow you.
‘What do you mean by telling me to give short answers?’ he asked. ‘Are they to be shorter than the questions require?’
‘By no means,’ I said.
‘The right length, then?’
‘Yes.’
‘So are they to be the length that I think right, or that you do?’
(Plato, 1991, 27-28, Protagoras 334c-e)

Here we are launched into a metadialogue that starts as a kind of persuasion dialogue. But its continuation in the Protagoras no longer (primarily) displays the features of a persuasion dialogue. Rather the discussants resort to negotiation. Socrates threatens to leave the scene and thus to end the dialogue, typically a move that can be part of a negotiation but not of a persuasion dialogue. The end of dialogue is, however, averted by the introduction of a number of proposals about how to continue. The first proposal, made by Callias is that each of the discussants will speak as he likes. But according to Alcibiades this is not fair. since Socrates grants that Protagoras is better at giving a speech. So if Protagoras wants to dispute the superiority of Socrates in question and answer dialogue he should enter a contest in that type of dialogue. This is the second proposal. Then Critias, Prodicus, and Hippias try to steer a middle course. This leads to a third proposal, by Hippias, to appoint an umpire to see to it that contributions to the dialogue will not be too long, nor unduly constrained by requirements of brevity. However, Socrates argues that it would be improper to appoint an umpire (since no one is wiser than Protagoras). He modifies the third proposal into a fourth proposal, which is then accepted: both parties will fulfill alternately the roles of Questioner and of Answerer; Protagoras will be the first Questioner; the audience as a whole will act as an umpire; if Protagoras in his answer does not stick to the question, Socrates and the audience will ask him ‘not to ruin the conversation’.
This is a clear example where the metadialogue that is resorted to in order to solve problems in the ground level dialogue is a negotiation dialogue, even though it contains pieces of arguing that can be considered as embedded persuasion dialogues.

For a contemporary example I refer to the recent conference of the International Whaling Committee (ICW) at Shimonosheki, Japan. According to a newspaper report (NRC-Handelsblad, May 25th, 2002) this conference was completely blocked by the extreme opposition between those in favor of some controlled whaling (the so-called Revised Management Scheme, or RMS) and those opposing all whaling. Iceland had left the IWC in the early nineties because the RMS-plans were not making progress. Its status had been reduced to that of an observer. But now Iceland wanted to return to full membership, seeing some chance for controlled whaling to become an option in the near future. However, Iceland announced that if controlled whaling were not be installed within due time, it would renounce commitment to the current IWC ban on whaling. For the IWC this reservation was a reason to refuse full membership to Iceland. Clearly the IWC was now moving on a metalevel with respect to the ground level discussion on whaling which, one presumes, was their principal concern. They were debating whether Iceland could be admitted to join the ground level discussion. Iceland’s reservation was used as an argument that it could not. Norway objected, arguing that the IWC’s refusal to admit Iceland was illegal. A vote was impending in which Iceland could have won the case. But the chairman refused to have a vote. Japan’s representative tried to intervene: “Please hear me out!” he yelled. “No” hollered the Americans. “Yes” shouted the Japanese. But it was No. Exit Iceland.
This last phase could be described as a metametadiscussion, a metadiscussion about how to go about the metadiscussion about Iceland’s admission. It is a discussion of the eristic type, consisting chiefly of yells and shouts.

3. Research
As far as I know, metadialogue has been studied very little by theorists of argumentation. One of the main sources is Hamblin’s 9th chapter of Fallacies (1970). In that chapter Hamblin distinguishes between topic points and points of order. Topic points, we could say, belong to the ground level dialogue, whereas points of order introduce metadialogues. His idea is that charges of equivocation should be looked upon as points of order:
The road to an understanding of equivocation, then, is the understanding of charges of equivocation. For this, the development of a theory of charges, objections or points of order is a first essential. (Hamblin, 1970, 303)

Mackenzie (1979, 1981) introduced the idea of points of order into formal dialectic. He introduces dialectic systems consisting of an inner and an outer system. Dialogues that are legal according to the inner system are called legal. They constitute a subset of the dialogues that are legal according to the outer system; these are legal+. As soon as some move turns a dialogue into one that is merely legal+, and not legal, the other party, in order to stay legal+, has to react by what we would now call a charge of fallacy. The illegal move is then removed and inner legality restored. There are no debates about the inner legality of moves. These contributions of Mackenzie are essential for the description of many-levelled systems, but they do not yet provide for full-fledged metadialogue.
Finocchiaro (1980, Ch. 16) distinguished between meta-arguments (arguments about arguments) and object arguments (which they are about). This distinction is certainly relevant for the study of metadialogue, and so is the material adduced in that chapter, as well as the notion of active involvement in the preceding chapter. However, it must not be presumed that all meta-arguments must be placed at the metadialogical level. Most of them are perhaps better placed at the ground level. Here things still have to be sorted out: when does the criticism of an argument amount to a claim that discussion rules were transgressed and that the whole argument must be withdrawn, when is it merely meant to lead to an improvement of the argument?

In the pragma-dialectical model of critical discussion the place to discuss the proceedings of the dialogue is the opening stage (Van Eemeren and Grootendorst, 1992). In the opening stage (which is not the first stage, but the second, following immediately after the confrontation stage which introduces a dispute about some issue) agreements must be reached about the following:

1. Engaging in dialogue to settle or resolve the dispute.
2. Using persuasion dialogue (to resolve the dispute) rather than some other type of dialogue (to settle it).
3. Adopting a particular dialectic system (specific rules) for the resolution of the dispute.
4. Assigning of roles to participants
5. Appropriate argument schemes and the conditions for their correct application.
6. Starting points that can serve as basic premises.

If a dialogue is conducted about any of these issues, this will then be a metadialogue relative to the ground level dialogue in which the confrontation occurs. Even though there is no official splitting in levels of dialogue, pragma-dialectics confronts the same problems as the theorists of metadialogue. The demarcation problem now resurges as the problem of drawing a line between moves that belong to the argumentation stage (the third stage) and the opening stage. For instance, one may wonder where to place critical questions that seem to be part of the argumentation stage, but may at the same time challenge the correctness of the application of an argumentation scheme. The problem of infinite regress arises if the opening stage is allowed to contain critical discussions that again need an opening stage. There is also an equity problem: how to balance the right each discussant has to return to the opening stage with the right to resist needless digressions? This may be no problem in ideal executions of critical discussion, where no return to the opening stage is required, but it may constitute a problem for less ideal situations.

Of course the same problems play a role in more formal approaches to dialectics. Van Laar in his study of ambiguity and equivocation proposes a model that contains two layers. In an ideal situation where all the so-called regulative rules are followed, no ambiguities occur. In somewhat less ideal situations ambiguities do occur, but are dealt with in a reasonable way, described by the constitutive rules of Ambiguity Dialectics (Van Laar, 2002). Recently Van Laar added a third layer, that of attempts at Ambiguity Dialectics (Van Laar, 2003).
This last moves brings one close to the idea of a Control Layer, which would end the infinite regress. This idea has recently been studied by theorists of multi-agent systems. McBurney and Parsons (2002) present an Agent Dialogue Framework which admits the embedding of dialogues in dialogues. These dialogues may be of various types: persuasion, negotiation, etc. At the top there is a Control Layer which is the level where dialogues about dialogues are conducted. The problem remains of how to control the Control Layer (Not that I have a solution to offer).

4. Dialectic Rules for Metadialogue
Below I shall atttempt to formulate some rules for opening and closing metadialogues that criticize moves that pretend to be permissible at a lower level. The type of dialogue I have in mind is persuasion dialogue on all levels. It does not matter how many levels there are. The rules are symmetrical, i.e. they do not distinguish between roles. There are two participants that move alternately. In the rules, “X” refers to one of the participants (indiscriminately) and “Y” to the other. At each stage of the dialogue there is a sequence of performed moves that are supposed by both participants to be legal on the ground level (the accepted grounded level dialogue). When a new allegedly permissible ground level move m is added, its permissibility may be challenged and tested on the (first) metalevel. If no test is asked for, the accepted ground level dialogue is extended by move m (similarly if the test has a positive result). The empty sequence counts as accepted.

Rule 1    Suppose that X has proposed to continue the ground level dialogue with an allegedly permissible ground level move m (m is a move by X that is not a challenge of the permissibility of the preceding move). Then the sequence of alleged (non-retracted) ground level moves that precede m constitutes the accepted ground level dialogue. It is now Y’s turn to move. One option for Y is to challenge the permissibility of move m, X‘s last move. This opens a metadialogue at level 1.
Rule 2    In the dialogue opened according to Rule 1 (henceforward called “the metadialogue”) Y is the Proponent of the thesis that X’s last move is not permissible as a continuation of the accepted ground level dialogue; X is to act as the Opponent.
Rule 3    In the metadialogue all established agreements about dialectic are among the Opponent’s initial concessions.
Rule 4    As soon as Y has won the metadialogue, X is to retract the alleged ground level move m. X may substitute some other (alleged) ground level move for m, but not enter a metadialogue on the permissibility of Y’s preceding move on the ground level. X is to pay the costs of the metadialogue.
Rule 5    As soon as X has won the metadialogue, move m counts as having been tested with positive result. The accepted ground level dialogue is extended by m. Y is to propose the next move, which must be on the ground level. Also, Y is to pay the costs of the metadialogue.
Rule 6     Rules similar to Rule 1 through 5 obtain to regulate transitions to metadialogue at other levels than the ground level.

5. Conclusion
The conclusion of this paper must perhaps be that it is too early for conclusions. Certainly, the proposal in the preceding section does not solve the problems listed in the introduction. For one thing, the proposal is limited to persuasion dialogue, whereas we saw in several examples (Protagoras, IWC) that dialogues of other types have a role to play.
The demarcation problem is left wide open. But perhaps it is an advantage of the present framework that it does not preempt any decisions as to what types of criticism the ground level may contain. These rules can be combined with a ground level that already displays discussion about questions of interpretation, questions of ambiguity, questions of validity, and criticism of applications of argumentation schemes. What is referred to the metalevel could be called “fallacy criticism” (Krabbe, 2002). The term “fallacy” is thus reserved for moves that shouldn’t have occurred and are therefore punished with a fine (the costs of metadialogue). Debatable points of meaning, ambiguity, validity, and the critical questioning that goes with the application of argumentation schemes will remain at the ground level, and the errors discussed in this way need not be blamed on the perpetrator.
The problem of infinite regress is still there, since there may be an indefinite number of levels. To have a level at which the permissibility of proposed moves can no longer be challenged, may work for machines. For humans such a limit has a ring of dogmatism. What the above proposal does to discourage a wanton ascent to metalevels is to charge the costs of each metadialogue on its loser. But then one could avoid to lose by ascending to the next level before the loss becomes apparent. All I can advise to those who meet with such an opponent is to abandon the dialogue!
As to equity: the rules are symmetrical and seem fairly balanced between giving rights to challenge the permissibility of moves and giving rights to the other party to challenge such challenges.

REFERENCES
Aristotle (1965). On Sophistical Refutations, On Coming-to-be and Passing away (translated by E.S. Forster), On the Cosmos (translated by D.J. Furley). Cambridge, MA: Harvard University Press, and London: William Heinemann (Loeb Classical Library). First printed 1955.
Eemeren, F. H. van, & Grootendorst, R. (1992). Argumentation, Communication, and Fallacies. Hillsdale, NJ, Hove, and London: Lawrence Erlbaum.
Finocchiaro, M. (1980). Galileo and the Art of Reasoning: Rhetorical Foundations of Logic and Scientific Method. Dordrecht/Boston/London: Reidel.
Hamblin, Ch. L. (1970). Fallacies. London: Methuen.
Krabbe, E. C. W. (2002). Profiles of Dialogue as an Dialectical Tool. In: F. H. van Eemeren (Ed.), Advances in Pragma-Dialectics (pp. 153-167). Amsterdam: Sic Sat and Newport News, VA: Vale Press.
Laar, J. A. van (2002). Equivocation in Dialectical Perspective. In: Argumentation and its Applications (CD-ROM, Proceedings from the Conference of The Ontario Society for the Study of Argumentation, May 17-19, 2001, University of Windsor, Windsor, Ontario). To be published.
Laar, J. A. van (2003). Ambiguity in Argumentation (dissertation). Groningen University. Forthcoming.
McBurney, P., & Parsons, S. (2002). Games that Agents Play: A Formal Framework for Dialogues Between Autonomous Agents. Forthcoming in the Journal of Logic, Language, and Information (special issue on logic and games).
Mackenzie, J. D. (1979). How to Stop Talking to Tortoises. Notre Dame Journal of Formal Logic, 20, 705-17.
Mackenzie, J. D. (1981). The Dialectics of Logic. Logique et analyse, n.s. 24, 159-77.
Plato (1961). The Collected Dialogues of Plato Including the Letters (Edith Hamilton and Huntington Cairns, Eds.), Princeton, NJ: Princeton University Press (Bollingen Series 71).
Plato (1991). Protagoras (transl. with notes by C.C.W. Taylor). Oxford: Clarendon Press.