Published in Fallacies: Classical and Contemporary Readings, edited by Hans V. Hansen and Robert C. Pinto (1995).

Argumentum Ad Hominem:
A Pragma-Dialectical Case in Point

Frans H. Van Eemeren and Rob Grootendorst

Fallacies of Relevance in the Standard Treatment

In the Standard Treatment of fallacies, the argumentum ad hominem is generally analyzed as a fallacy of relevance.1 This category also includes ad baculum, ad ignorantiam, ad misericordiam, ad populum, and ad verecundiam. It gives the impression that it serves primarily as a depository, for it accommodates precisely those fallacies for which the Standard Treatment has no adequate solution because they are beyond the scope of the definition of a fallacy as an argument that seems valid but is not so.

In the common examples of fallacies of relevance there is no sign of any argument in the logical sense. Reconstructing them as prototypical arguments with two premises and a conclusion, usually requires stretching a point or two. Even then, not much is explained. Naturally, it only makes sense to claim that an argument is fallacious because its premises are "logically irrelevant" if the notion of logical relevance is first defined. In the Standard Treatment, no effort is made to that effect.

The Standard Treatment gives a logical overtone to the notion of relevance by suggesting that in a fallacy of relevance the irrelevancy of the premises for the conclusion is a matter of logic.2 If this logico-centric approach is to be taken seriously, not only the logical validity of arguments will have to be tested, but also their logical relevance. But the notion of logical relevance is left undefined and its connection with logical validity remains unexplained.3 Is the relevance of the premises to the conclusion in a logically valid argument always guaranteed? Can the argument be valid even though the premises are not relevant to the conclusion? Or is it also possible for the premises to be relevant while the argument is invalid?

In the absence of any further clarification, the Standard Treatment makes it hard to determine whether we are dealing with a fallacy of relevance. The "explanatory comments" that are provided are confined to the observation that the premises are not germane, do not really support the truth of the conclusion, or something of the sort.4 In practice, the problem is circumvented by selecting only examples that are, as it were, dripping with irrelevance, so that the question of the relation between relevance and validity does not so easily arise.

Argumentum ad Hominem as a Violation of a Rule for Critical Discussion

In our opinion, a satisfactory alternative to the Standard Treatment of the argumentum ad hominem and other so-called fallacies of relevance can only be achieved if the preoccupation with the logical validity of arguments is abandoned. In the pragma-dialectical approach, we have done this by placing the fallacies in the perspective of a critical discussion aimed at resolving a difference of opinion. The fallacies of relevance are then analyzed as violations of the rules for critical discussion.5

The argumentum ad hominem is a violation of the first rule for critical discussion: "Parties must not prevent each other from advancing standpoints or casting doubt on standpoints." This discussion rule is designed for the confrontation stage, where the protagonist and the antagonist of a standpoint enter into a difference of opinion. Naturally, the rule also applies to a difference of opinion that arises when the argumentative support for a standpoint meets with doubt or other criticism. The premise concerned is then to be considered as a substandpoint.

The first discussion rule aims at ensuring that any difference of opinion can be expressed without hindrance. It may seem odd, but in order to promote the possibility that differences of opinion can be resolved, the open confrontation between the parties needs to be stimulated. Therefore, in a critical discussion the parties have a principal right to advance any standpoint they wish and to challenge any standpoint they wish.6 This implies that any potential obstacles to expressing standpoints or criticizing standpoints are to be cleared away. As a consequence, neither party is allowed to prevent the other party from entering into an unimpeded confrontation by ruling him out as a serious discussion partner. And this is precisely what is attempted in the argumentum ad hominem.7

All three variants of the argumentum ad hominem are violations of the first rule for critical discussion. In effect, they all amount to a party claiming that the other party has no right to speak. In the abusive variant, this party undermines the other party's credentials by denigrating his intelligence, experience, or good faith. In the circumstantial variant, he does so by suggesting that the other party is not capable of making an impartial judgment because he is driven by personal interests. And in the tu quoque variant the other party's credentials are being reduced by denouncing an inconsistency in his opinions or behavior.

This brief explanation not only suffices to show that the three variants of the argumentum ad hominem are all offenses against a fundamental norm for argumentative discourse aimed at resolving a difference of opinion: it also gives some indications of the criteria for determining with which variant of the argumentum ad hominem we are dealing in a given case.8 The general criterion that applies to all three variants is, of course, whether a party has said something that is calculated to undermine the other party's position as a credible discussion partner. For the abusive variant, the criterion is whether something negative is said of the other party's personal characteristics: his intelligence, expertise, and so on. For the circumstantial variant, the criterion is whether the attention is drawn to the other party's particular interests in adopting a specific position: financial gain, social success, and so on. The criterion for the tu quoque variant is whether an inconsistency is pointed out in the other party's words or actions: he himself does something that he condemns in others, now he says this but yesterday he said that, and so on.

The Pragma-Dialectical Approach versus the Standard Treatment

In the Standard Treatment, an argumentum ad hominem is a fallacy because the premises of the argument are irrelevant to the conclusion, except in those cases where the premises are relevant to the conclusion -- unless they turn out not to be relevant on further reflection. The criteria applied in this endeavor are usually not only implicit and intuitive, but also highly arbitrary and ad hoc. It looks as if each case must be assessed individually to see whether there are mitigating circumstances that render the personal attack less blameworthy than it appears. Invariably, some examples are presented of personal attacks in which the ad hominem is after all not a fallacy.

The pragma-dialectical approach offers a systematic solution to the problem of the many exceptions to the rule that an argumentum ad hominem is a fallacy. The solution is very simple: there are no exceptions. If the criterion for one of the variants of the argumentum ad hominem is fulfilled, a personal attack is always a violation of the first rule for critical discussion. It is therefore, without any exception, a fallacy.

An argumentum ad hominem that is a countermove against the misuse of expertise or authority by the opponent is by some authors considered as a correct use of the argumentum ad hominem.9 However, the discussion is then rather directed toward gaining victory in a dispute than toward resolving the difference of opinion. Since the move, therefore, falls outside the scope of the rules for critical discussion, it is no argumentum ad hominem in the pragma-dialectical sense.10 If the countermove were part of a critical discussion, it would be fallacious and indeed an argumentum ad hominem. That the fallacy is a reaction to another fallacy, an argumentum ad verecundiam, does not make it a correct discussion move: two wrongs do not make a right.

The fact that the term argumentum ad hominem is sometimes used in a neutral way for a personal attack and sometimes refers to the fallacy which occurs when such an attack is incorrect is one of the reasons why the analysis of "exceptions" in the Standard Treatment gives the impression of being merely ad hoc.11 It may seem as if the same discussion move is in one case permitted and not in another, so that it is in some cases a fallacy and in other cases not. What the two moves have in common is that something negative is said about a person, but on a closer inspection they prove to be very different. Because the term argumentum ad hominem is more often used to make a negative rather than a neutral judgment, the standard meaning seems to bear the connotation of "fallacy," whereas the exceptional cases (in which ad hominem argumentation is allowed) seem to represent deviations. Unfortunately, when an author uses the term argumentum ad hominem it is not always clear whether he is dealing with a justified exception. To avoid this kind of obscurity, we use the ordinary expression personal attack for the neutral case and the technical term argumentum ad hominem for the fallacy of an incorrect personal attack.

In the pragma-dialectical analysis of the argumentum ad hominem, unlike in the Standard Treatment, no reference is made to the relevance of the premises of an argument for its conclusion. The reason for this is that an argumentum ad hominem is not fallacious because it is an argument with irrelevant premises. Viewed within a pragma-dialectical perspective, the relevance of a discussion move depends on its contribution to the resolution of a difference of opinion.12 In a critical discussion, an argumentum ad hominem is, in fact, always highly relevant, but in a negative sense: it hinders, or sometimes even prevents, the resolution of a difference of opinion.13

[References below are to the Bibliography]

1. For the Standard Treatment of fallacies, sec Hamblin (1970, chap. 1).

2. See, e.g., N. Rescher, Introduction to Logic (New York: St. Martin's Press, 1964), 78; Copi and Cohen (1990, 3); and, for a more recent example, T. E. Darner, Attacking Faulty Reasoning, 1st ed. (Belmont, Calif.: Wadsworth, 1987), 99.

3. In A Practical Study of Argument, 2d ed. (Belmont, Calif.: Wadsworth, 1988), Trudy Govier does give a definition of relevance, and her approach is certainly much more thorough than the Standard Treatment, but on some crucial points she is still so firmly tied to the logico-centric approach that she fails to provide a satisfactory analysis of ad hominem as a fallacy of relevance. See van Eemeren and Grootendorst (1992b).

4. See, e.g., Rescher, Introduction to Logic, 81, and Copi and Cohen (1990, 98).

5. For the theoretical framework of the pragma-dialectical approach, see Chapter 9, this volume.

6. It goes without saying that this does not automatically mean that a party is always obliged to continue the discussion after a confrontation has taken place. It may, for instance, become clear at the opening stage of the discussion that the antagonist is not prepared to comply with the rules for critical discussion, so that there is no chance that the difference can be resolved.

7. The party that is personally attacked may, of course, always decide to enter into a confrontation about the premise that is expressed in the argumentum ad hominem, but then we are dealing with an entirely different case: instead of being taken aback by the argumentum ad hominem, the person who is attacked voluntarily engages in another discussion in which this premise is the standpoint at issue.

8. This is not to say that there are no problems left. Interpretation procedures are, for instance, required to determine whether a speech act fulfills the criterion of an abusive argumentum ad hominem, that something negative is said about the opponent. As we explained in van Eemercn and Grootendorst (1992b), this is not a major problem in the case of an abusive argumentum ad hominem.

9. See, e.g., Woods and Walton, "Ad Hominem" (1989, 55-73).

10. In our opinion, there is no point in calling such a move a fallacy because it is not part of a clearly conceptualized goal-directed enterprise whose standards are defined in a well-delineated normative framework; if such a framework were developed, however, say in ethics, it would make sense to speak of "ethical" fallacies. A similar case could be made for "epistemological" fallacies.

11. One of the other reasons is the one-sided concentration on the premise-conclusion relation in arguments.

12. The pragma-dialectical approach also recognizes fallacies of relevance, but they are either violations of the third rule or the seventh rule for critical discussion (van Eemeren and Grootendorst (1992a, 124-31, 158-68). Among the violations of rule 7 that produce a fallacy of relevance are the cases in which the arguer defends his standpoint by wrongly launching a personal attack on a third party who is absent or even dead: "Frege's logical views are not to be taken seriously because he was anti-Semitic." In the pragma-dialectical sense, such an attack on a third party is not an argumentum ad hominem, but if such a case of incorrect symptomatic argumentation hinders the resolution of a difference of opinion by falsely presenting something (being anti-Semitic) as a sure sign of something else (having no serious logical views), it is a fallacy of relevance. We owe the observation concerning attacks on a third party to Hans Hansen, who would like to maintain the ad hominem terminology for such cases by envisaging a distinction between first-, second-, and third-person ad hominems (personal communication).

15. We owe the notion of negative relevance to Govier, A Practical Study of Argument, 98-99, who contrasts negative relevance with positive relevance and irrelevance.