3. Krabbe&vanLaar


About Old and New Dialectic     27

About Old and New Dialectic:
Dialogues, Fallacies, and Strategies

 University of GroningenERIK C. W. KRABBE
JAN ALBERT VAN LAAR

© Informal Logic Vol. 27, No. 1 (2007): pp. 27-58.

1. Introduction

In 1998, forty years after the original edition of The New Rhetoric (Perelman and
Olbrechts-Tyteca, 1958), Douglas Walton published The New Dialectic. The title
confirmed the dialectical turn of argumentation theory, taken long before. Both
titles refer to old versions of practices and theories, a new dialectic presupposing
an old dialectic as much as a new rhetoric presupposes an old rhetoric. Thus these
works invite a comparison: to what extent can the old be mapped onto the new; to
what extent does the new expand upon or deviate from the old? The new editions

Abstract: We shall investigate the similarities
and dissimilarities between old and new
dialectic. For the ‘old dialectic’, we base our
survey mainly on Aristotle’s Topics and
Sophistical Refutations, whereas for the ‘new
dialectic’, we turn to contemporary views
on dialogical interaction, such as can, for the
greater part, be found in Walton’s The New
Dialectic. Three issues are taken up: types
of dialogue, fallacies, and strategies. Though
one should not belittle the differences in
scope and outlook that obtain between the
old and the new dialectic, the paper will show
that in many respects the old dialectic
foreshadows the new dialectic.

Résumé: Nous voulons rechercher les
ressemblances et les différences entre
l’ancienne et la nouvelle dialectique. Quant
à l’ancienne dialectique, ce sont surtout
les Topiques et les Réfutations sophis-
tiques d’Aristote—qui nous ont servi de
base pour nos recherches, tandis que pour
la nouvelle dialectique nous nous
orientons vers les idées contemporaines
sur l’interaction dialogique, comme on les
trouve, pour une large part, dans The New
Dialectic de Walton. On veut ici traiter
trois questions: celle des types de
dialogue, celle des sophismes et celle des
stratégies. Sans vouloir négliger les
différences de portée et de point de vue
qui existent entre l’ancienne dialectique et
la nouvelle dialectique, cet article veut
montrer comment qu’à plusieurs égards
l’ancienne dialectique annonce déjà la
nouvelle dialectique.

Keywords:  activity type, argumentation theme, Aristotle, blunder, critical discussion,
dialectic, dialectical discussion, eristic discussion, fallacy, function of reasoning, new
dialectic, peirastic discussion, persuasion dialogue, type of dialogue, strategic
manoeuvring, strategy, thematic fallacy, Douglas N. Walton



28     Erik C. W. Krabbe & Jan Albert van Laar

deserve it to have their measure taken in terms of their predecessors, whereas the
old editions are entitled to a reassessment in terms of the new.

The present paper is meant to contribute to the comparison between old and
new dialectic, where the old dialectic is taken to be Aristotle’s theory in his Topics
and Sophistical Refutations. The practice of dialectic was not Aristotle’s invention,
but as to theory it may well be that the first encompassing treatment of dialectic is
contained in these works of his, as he claims himself at the end of Sophistical
Refutations (183b34-36).

We are not the first to undertake such a comparison. Walton himself did so in
many sections of The New Dialectic. We shall not go through these sections in
detail, but straightforwardly present what we think are the most salient points of
comparison. This we shall do with respect to three issues: dialogues, fallacies, and
strategies. For this we need concepts of ‘reasoning’ and ‘argument’, which will
be explained in Section 2. In Section 3, on dialogues, we shall start with the
distinction between activity types and normative models (Section 3.1). Then we
discuss the theory of types of dialogue in the new dialectic in terms of the various
functions of reasoning (Section 3.2). Next we reconstruct some of Aristotle’s
implicit views on types of dialogue starting from his theory of kinds of reasoning
and try to classify these types from the point of view of the new dialectic (Section
3.3). In section 4, on evaluation and fallacies, we start again with the views of the
new dialectic (Section 4.1), and then briefly discuss Aristotle’s theory of fallacies
to see points of comparison (Section 4.2). In Section 5, we comment on strategies
in both the new and the old dialectic. We end with some conclusions and prospects
(Section 6).1

2. Reasoning and argument

Before entering upon a comparison of the old and the new dialectic with respect to
their types of dialogue, fallacies, and strategies, we shall stipulate what we understand
by ‘reasoning’ and ‘argument’, expanding on Walton’s discussion of these concepts
(Walton, 1990) by introducing several functions of reasoning. This will, so we
hope, contribute to what Johnson perceives as the much-needed theory of reasoning
(Johnson, 2000, p. 362-365).

According to Walton, ‘reasoning can be defined generally as a sequence of
steps from some points (premises) to other points (conclusions).’ He adds that
these ‘points normally represent propositions’ (1990, p. 404). Thus, reasoning is
portrayed as an abstract event without any predetermined aim, but suited to be put
to a number of purposes. We will start from the idea that a reasoning, PsoC where
C is the conclusion and P (= {P

1
…P

i
…P

n
}) a set of premises, is always made up

from propositions, by whatever speech acts the reasoning is conveyed. Reasoning
can be simple, i.e., consist of exactly one such step, or complex, when steps are
chained to build a larger whole. In the following survey of functions of reasoning,



About Old and New Dialectic     29

we shall include both simple reasoning, that has exactly one conclusion, as well as
complex reasoning that is made up from various simple reasoning steps. However,
we shall restrict our attention to such reasoning as is external and publicly accessible,
as opposed to internal, mental reasoning. We distinguish six kinds of functions of
reasoning.

Reasoning has first of all persuasive functions. In a situation where there is a
conflict of opinion, reasoning can be put to use for the purpose of persuading
someone of something. We shall say that reasoning, PsoC, is used in its primary
persuasive function in case a reasoning proponent, supposing his interlocutors to
be committed to premises P, uses the reasoning to persuade them of the acceptability
of conclusion C. The same reasoning can have at least three secondary persuasive
functions. (1) Suppose, the interlocutors clearly accept the denial of C, as well as
all but one of the premises. Let P

i
 be the premise they do not accept. Then the

reasoning can be used to persuade them of the denial of P
i
. (2) Similarly, in cases

where an interlocutor is in doubt about the acceptability of the argumentative
connection between P and C, the proponent may reason from P to C in order to
persuade his interlocutor of the acceptability of the proposition warranting that a
commitment to P implies a commitment to C. (3) If an interlocutor, who accepts
all premises P, does not acknowledge C, even though every sensible person knows
that P implies C, the proponent may reason PsoC, in order to support, in this
indirect manner, a claim to the effect that the respondent failed to perceive an
obvious implication. These other ways of reasoning can be said to employ
secondary persuasive functions of the same reasoning, because they are all based
on the very same logical fact, that is, the inconsistency (or untenability) of the set
of premises P in conjunction with the denial of the conclusion C. They are not
primary uses of that reasoning because P is not used to show that C holds or to
make C more plausible. It should be noted that in situations where a reasoning
PsoC is used in one of these secondary persuasive functions, the proponent could
also avail himself of another instance of reasoning and use the primary persuasive
function. That is, he could use an instance of reasoning, P′soC′ that leads directly
to the conclusion C′, where C′ is a proposition that (1) denies P

i
, or (2) necessarily

connects P and C, or (3) asserts that the respondent failed to perceive an obvious
implication.

The directive functions of reasoning can be used to make someone do something.
These functions are closely related to the persuasive ones. Such reasoning can be
used in situations where there is a conflict of interest in order to make a respondent
more accommodating and conciliatory in his acts. For instance, when negotiating,
a party may give reasons to support a proposition to the effect that what his
interlocutor proposes is unrealistic, hoping that this will make him mitigate his
demands. Also, a party may give reasons to support a proposition to the effect that
his own proposals would constitute a viable and feasible result of the negotiation,
hoping that the other will accept them or at least make an offer that lies close to



30     Erik C. W. Krabbe & Jan Albert van Laar

this proposal. Such reasoning is an instance of practical reasoning from desires
and goals to the advisability of performing or refraining from performing certain
actions. The primary function of reasoning in these cases is not to persuade the
other of the truth of some proposition, but to get him or her to perform the
appropriate action (to mitigate demands, to accept an offer). Yet the way this
works can be seen to involve implicit persuasion.

Reasoning has polemic functions in situations where it is used to intimidate or
impress upon the other. Such reasoning can be used when there is a conflict and
serious antagonism between parties. In such contexts it can be used for the purpose
of supporting the claim that the interlocutor ought to acknowledge to be guilty of
having wronged the proponent or that he would be wrong if not deferring to
‘reason’. Alternatively, it can be used in order to embarrass the respondent
intellectually, letting him without any argumentative defence. These polemic
functions are persuasive in as far as the proponent pretends to provide good reasons.
Yet, the primary purpose of the reasoning does not consist in getting the respondent
to commit himself to the conclusion after careful reflection on the acceptability of
the reasons, but rather in making him back off and give in upon understanding the
force of the speaker’s emotions on this issue and/or his superiority in reasoning.

Of a different order are the probative functions. When used probatively, reasoning
leads to new knowledge, in the sense that the store of public beliefs or alleged
knowledge in a company of discussants is extended by new propositions. As said
before, we are not here concerned with internal reasoning, which may extend
personal knowledge. In contrast with persuasive uses, probative reasoning need
not be aimed at taking away critical doubts of an interlocutor. Even though persuasion
will accompany the probative use of reasoning in many situations, this need not be
the case. In a company which knows the propositions P to be correct, a speaker
may use a reasoning, PsoC, to extend the set of publicly known propositions by C,
even in situations where the interlocutors are not in doubt about C, or quite willing
to accept C on the authority of the speaker. The speaker need not to remove doubt,
then, but must extend knowledge. Also, the speaker himself may not yet know the
conclusion before he presents the reasoning. Reasoning can be used probatively,
either with a cognitive orientation, aimed at extending the store of theoretical or
factual knowledge, or with a practical orientation, aimed at knowledge about the
degree of desirability of particular ends in personal life or in society and about
what actions and policies are most prudent in the light of these ends and of the
present state of affairs (Walton, 1990, p. 405).

The explorative functions of reasoning are secondary probative functions. When
used exploratively, a speaker is using reasoning in order to explore illative connections
between propositions. Analogously to the second of the secondary persuasive
functions, that of reasoning used to persuade a respondent of the argumentative
connection between reasons and conclusion, explorative reasoning is used to give
evidence for the proposition that the illative connection between P and C can be



About Old and New Dialectic     31

added to the shared store of knowledge. Used exploratively, a reasoning PsoC
indirectly indicates what is directly given by probative reasoning for the claim that
C follows from P.

Finally, reasoning can be seen to have explanatory functions. These, however,
are not special cases of either the persuasive or probative functions. The reason is
that when used for explanatory purposes only, the aim of expressing reasoning is
not to produce commitment to or knowledge of any conclusion C. Instead, such
reasoning is used to enhance understanding. Thus it may be used for the purpose
of making it clear how something has come about. Further, it may be used, when
giving instruction about a subject, for the purpose of informing the interlocutor
about the existence of (a particular usage of) a piece of reasoning linking elements
of the subject under discussion. Also, it may be used for the purpose of making the
interlocutor understand the precise meaning of a particular proposition or formulation
of a proposition. The meaning of a proposition can be clarified by informing the
interlocutor about reasons that support it or by informing her about its consequences.
Further, if the proposition states an illative connection between P and C, stepwise
reasoning from P to C can help the respondent to comprehend this connection.

This may suffice to explain our use of the term ‘reasoning’. As to ‘argument’
we can be brief. A reasoning that is used by a proponent in some kind of conflict
in order to overcome doubt of an interlocutor will be said to have an argumentative
function. So, only the persuasive, directive and polemic functions are inherently
argumentative in that sense. The term argument will be reserved for reasoning that
is used for these persuasive purposes. It may now be seen that in many cases
where we used the somewhat awkward substantive ‘reasoning’ one may substitute
the more familiar ‘argument’. Our use of ‘argument’ differs however from that in
formal logic, where it refers to the abstract configuration of premises and conclusion
in a simple (one-step) reasoning. Our use of the term accords well with Walton’s
proposal: ‘Argument is a social and verbal means of trying to resolve, or at least to
contend with, a conflict or difference that has arisen or exists between two (or
more) parties. An argument necessarily involves a claim that is advanced by at
least on of the parties’ (1990, p. 411).

3. Dialogue

3.1 Activity types and normative models

In a recent paper, Van Eemeren and Houtlosser (2005) pointed out the importance
of distinguishing between argumentative activity types and normative models such
as the pragma-dialectical model of critical discussion. An argumentative activity
type is a conventionalized entity that can, in principle, be identified by empirical
observations. Examples they adduce are: negotiation, mediation dialogue, and
adjudication. Presumably, everyday argumentative discussion would constitute



32     Erik C. W. Krabbe & Jan Albert van Laar

another example of an argumentative activity type. A normative model of
argumentation, on the other hand, is a theoretical account of how, ideally, parties
should go about in order to resolve a difference of opinion on the merits of the
case. Critical discussion is said to constitute such a normative model.

The current model of critical discussion (Van Eemeren and Grootendorst, 2004)
consists of a philosophically motivated and theoretically informed set of detailed
rules for critical exchanges. In pragma-dialectics, the notion of a critical discussion
is, among other things, used for the purpose of reconstructing argumentative
discourse by extracting those elements from the discourse that are relevant,
positively or negatively, for resolving the underlying difference of opinion. Because
ordinary language users presumably share at least some of their theoretical and
philosophical motivations with pragma-dialectics, albeit in a pre-theoretical mode,
it can be expected that the discussion norms that speakers and writers themselves
impose on the activity type of an everyday argumentative discussion will largely
agree with those prescribed by the ideal procedure contained in the model of
critical discussion. Hypotheses to this effect have been corroborated in quantitative
research (for example Van Eemeren, Garssen and Meuffels, 2003). So, we can
expect there to be a close connection between the normative model of critical
discussion and argumentative activity types that are close to it, such as the everyday
argumentative discussion.

Walton and Krabbe (1995, Ch. 3) distinguish six main types of dialogue (to be
discussed in Section 3.2, below): persuasion dialogue, inquiry, deliberation, eristic
dialogue, negotiation dialogue and information seeking dialogue.2 In the New
Dialectic, Walton states that he puts forward each of these types as ‘a normative
model which specifies broadly how an argument should be used in order to be
correct or to be defensible against criticism that it is incorrect, erroneous, or
fallacious’ (Walton, 1998, p. 3). At the same time, he deals with them as activity
types by defining dialogue as a ‘goal-directed conventional framework in which
two speech partners reason together in an orderly way, according to the rules of
politeness or normal expectations of cooperative argumentation for the type of
exchange they are engaged in’ (Walton, 1998, p. 3).

It is true that the notion of a normative model can be generalized beyond that of
a model where the goal is exclusively to resolve a difference of opinion (critical
discussion). For each type of dialogue, one could try to formulate a normative
model that specifies what, in the ideal case, is required of two parties who, starting
from a particular initial situation and having particular individual aims, are working
towards the main goal of that particular dialogue type. So, normative models can
be developed, parallel to given dialogue types considered as activity types. Even
so, it remains important to distinguish between activity type and normative model,
because the conventional norms that show up in a particular discourse practice
may very well diverge from those norms that, according to an informed theorist,
would really help the participants to achieve the main goal of the activity they are



About Old and New Dialectic     33

engaged in. We suppose that the dialogue types can best be dealt with as activity
types in which argumentation often occurs and that parallel to these types, normative
models can be developed.3

These normative models can then be used for the purpose of extracting the
elements from a dialogue that are analytically relevant for the goal of the dialogue
as stipulated by the model. Since people may try to do several different things in
the same conversation, more than one model may be applicable in a given case.
For instance, a dialogue can be analyzed from the perspective of a critical discussion,
but also from some other perspective, say from that of a normative model giving
the rules for impeccable bargaining. A choice for a particular normative model,
when the purpose is to reconstruct the reasoning in a dialogue, is first of all motivated
by the research interests of the analyst, and not determined by empirical data
(which is not to say that all methodological choices are always equally fruitful).

3.2 Dialogue types in the New Dialectic

In The New Dialectic (Walton, 1998), a full chapter is devoted to each of the main
types of dialogue distinguished by Walton and Krabbe (1995). According to Walton
argument occurs in each of these types of dialogue. This is not immediately obvious,
unless the concept of argument is stretched far enough to let it be so. According to
our stipulations in Section 2 only the persuasive functions (including the directive
and polemic functions) of reasoning are argumentative. Do these argumentative
functions characteristically occur in each type of dialogue? We shall run through
the different main types to see, for each of them, how reasoning characteristically
functions.

A persuasion dialogue is the kind of conversation in which the participants are
testing the ‘comparative strength and plausibility of arguments on both sides of a
controversial or contentious issue’ (Walton, 1998, p. 37). Starting from an issue
characterized by divergent or conflicting opinions, the main goal is to resolve the
issue on the merits of each side. Since each discussant aims at persuading the
other discussant to change his initial position in favour of the first discussant’s
position, such a dialogue is hardly conceivable without the occurrence of reasoning
used for persuasive purposes.

A negotiation dialogue starts from a conflict of interests, such that ‘each
participant has some items that can be traded with the other party.’ (Walton, 1998,
p. 101). The main goal is to make a deal. At the same time, each participant aims at
getting what he wants most. Although it is possible to negotiate by offer and
counteroffer only, usually negotiators also provide reasons. By making it clear
what matters most for them, a negotiator is capable of arguing against the adequacy
of unfavourable offers. By making it clear what matters most for the other party,
they argue in favour of the feasibility of the offers they themselves bring forward.
In this way negotiators attempt to influence the process of negotiation by (often



34     Erik C. W. Krabbe & Jan Albert van Laar

implicit) reasoning. Unsurprisingly, we may conclude that negotiation is
characterized by the directive functions of reasoning.

An eristic dialogue starts from an antagonism. Each party tries to defeat the
other. The main goal can be seen as the establishment of some new accommodation
or relationship between the participants. In the subtype called the quarrel, the
initial antagonism is emotionally burdened and the intended relationship is, ideally,
based on ‘a deeper insight into the feelings of the other party—an insight that can
facilitate a smoother relation in the future’ (Walton, 1998, p. 179). Typically in a
quarrel, reasoning is used polemically for the purpose of blaming and counter
blaming while keeping up a pretence of one’s own reasonableness by providing a
perfunctory justification. However, reasoning can also be used in the third of its
secondary persuasive functions mentioned in Section 2, when one party tries to
show the lack of intelligence of the other or to make him an object of ridicule. This
function will even be more prominent in eristic discussion, which is another subtype
of the eristic dialogue. In an eristic discussion, the aim is ‘to see who is the more
clever in constructing persuasive and often tricky arguments that devastate the
opposition, or at least appear to’ in order to settle an intellectual hierarchy (Walton
& Krabbe, 1995, p. 78; cf., Walton, 1998, p. 190 on sophistical dialogue). Of
course, eristic discussion is characterized by polemic functions aiming at intimidation
of the other just as much as the quarrel is. Thus in eristic dialogues of either
subtype the polemic and other persuasive functions of reasoning stand out.

An inquiry is the kind of dialogue where there is an open problem and where
the parties are trying to find out whether a particular proposition is true or false. In
contradistinction to persuasion dialogue, it need not be the case that one participant
is in favor of that proposition whereas the other is not. It may even be the case
that, to start with, the precise proposition to be established or rejected is unknown
to both parties. The result is to count as knowledge of a theoretical or factual kind,
and consequently, premises are admissible only if they are part of the (mutually
agreed upon) knowledge store of the intellectual company to which the discussants
belong. In inquiries, reasoning will be used, most characteristically, for probative
purposes—with the explorative purpose as a special case.

According to Walton and Krabbe examination is a kind of inquiry ‘to find out
whether the examinee disposes of a certain level of abilities and knowledge in a
certain field’ (1995, p. 76). But in The New Dialectic a related activity type, the
recruiting (employment) interview, is described as a subtype of information-seeking
dialogue (Walton, 1998, p. 130). Another related type, the examination or testing
of experts, is reckoned as ‘really a persuasion type of dialogue’ (p. 148), although
a link with inquiry is also recognized (p. 147). Yet it seems that these three activity
types in which one of the participants is, in some sense, tested or examined, are so
closely related that they should be reckoned to one type: examination dialogue.
The question remains, whether we can subsume the examination dialogue under
one of the existing main types, declare it to be another main type, or must describe
it as some sort of mixed dialogue type.



About Old and New Dialectic     35

In examination dialogues, such as academic exams, job interviews and the
testing of experts, the participants are often reasoning about substantial issues
related to the subject of the exam, the best way to set about the job, or the field of
expertise. This they may do in order to persuade the other. The real issue, however,
is whether the examinee is fit for a pass, for the job, or for the status of expert in
the field. The reasoning on the substantial issues is part of the test. The test as a
whole qualifies as a persuasion dialogue if the examinee can be taken to defend the
standpoint that he deserves the pass, the job or the status of expert. However, it is
possible that the examinee exhibits an open attitude towards his own capacities
and knowledge. Even an expert may come to doubt his own credentials. In such
cases, the examination dialogue is a subtype of inquiry. Let us call the first kind of
examination dialogue the examination subtype of persuasion dialogue, the second
the examination subtype of inquiry. In both cases examination dialogue is a complex
type of dialogue. In a recent article, Walton (2006) studies examination dialogues
in detail (with an emphasis on expert examination). He does not subsume them
under either persuasion dialogue or inquiry, but describes them as complex dialogues,
with information-seeking and persuasion dialogue as constituents.4 As we shall see
below, information-seeking dialogue is characterized by the explanatory function
of reasoning. Since, usually, information-seeking is part of examination dialogues,
one may in these dialogues discern reasoning with explanatory, probative, and
persuasive functions.

A deliberation starts from a need for action and is aimed at a prudent decision.
Like the inquiry, deliberation does not imply any kind of conflict between the
participants. However, in Walton’s model of deliberation, there is an argumentation
stage in which the participants support pro and contra positions with respect to
possible solutions (Walton, 1998, p. 162). Clearly deliberation may profit from the
embedding of such persuasive exchanges, and the same holds for the inquiry.
Altogether, we conceive of deliberation as being on a par with inquiry. But where
inquiry has a cognitive orientation, deliberation has a practical one (Walton, 1990,
p. 405). In deliberation reasoning is typically used to derive new knowledge, of a
practical nature, and to explore into the connections between practical knowledge
claims. Like inquiries, deliberation dialogues are characterized by the probative and
explorative functions of reasoning, which in these applications take on a practical
orientation.

Information-seeking dialogues start from a situation in which ‘one participant
has, or appears to have, information that the other party wants’ (Walton, 1998, p.
126). The main goal is to level this unequal distribution of knowledge. In so far as
reasoning is used for sharing information, it is used to make one of the parties
understand propositions or to make him understand them to a higher degree, or it
is used in one of the other explanatory functions. Hence information-seeking
dialogues are characterized by the explanatory function of reasoning.

Corresponding to our choice to consider only the persuasive, directive and
polemic functions of reasoning as argumentative functions, we propose to call the



36     Erik C. W. Krabbe & Jan Albert van Laar

persuasion dialogue, the negotiation dialogue and the eristic dialogue argumentative
dialogue types. The inquiry and the deliberation are best be called probative dialogue
types.

Whenever a conversation, instantiating a particular dialogue type, changes so
as to instantiate a different dialogue type or a different subtype of the same dialogue
type, a shift is said to have taken place between two distinct dialogue types or
subtypes. Whenever the shift in a dialogue of type D to a different kind of dialogue,
of type D′, is functional for the purposes of the initial dialogue, the second dialogue,
of type D′, is said to be embedded in the dialogue of type D (Walton, 1998, p. 204).
Whether shifts and embeddings have consequences for the evaluation of reasoning
will be discussed in Section 4.2.

3.3 Dialogue types in the Old Dialectic

The title of this section may occasion some raised eyebrows, since there is no
clear list of types of dialogues, either as activity types or as normative models,
anywhere in Aristotle’s Topics or Sophistical Refutations. Aristotle’s theory was
developed in the context of an existing practice, of what we may call dialectical
discussion, in Plato’s Academy—which practice was rooted in the Socratic dialogues
with the roles of Questioner and Answerer as a prominent feature. No handbook
with rules for such discussions survives, but its procedure can be reconstructed
from Aristotle’s writings (mainly Top. VIII; 5 Moraux, 1968). The Socratic dialogues
may serve as illustrations, which, however, may be quite informal, compared with
the practice as it existed in Aristotle’s time. Thus, there is an activity type to which
Aristotle’s theory of dialectic refers: dialectical discussion. As to normative models:
even though Walton may be right when he remarks that ‘Aristotle thinks of dialectical
argument more as a technique of argument, rather than as, what the new dialectic
calls, a normative framework of argumentation’ (1998, 243), it can not be denied
that Aristotle makes normative remarks about this activity type. For instance, he
does so when, in the Sophistical Refutations, he contrasts real and apparent
reasoning (SE 1).

We cannot here describe the procedure of dialectical discussion in detail. Let it
suffice to recall that at the start the Answerer is to choose one of a pair of
contradictory propositions as his thesis and that the Questioner’s thesis, which is
the conclusion he must reach, is the other proposition. The Questioner must try to
get premises admitted by the Answerer that will allow him to deduce his thesis.
These premises may either themselves be obtained by deduction—from premises
obtained before—or by induction, or analogy. If the Questioner reaches his
conclusion the Answerer is refuted. The (primary) aims of the participants are
clear: the Questioner tries to refute the Answerer, whereas the Answerer tries to
avoid being refuted. But what is the goal of the activity itself? Could it be any of the
goals that figure in the stipulations of types of dialogue in the new dialectic? Does
Aristotle distinguish different kinds of discussion to serve different goals?



About Old and New Dialectic     37

A clue may be found in the second chapter of the Sophistical Refutations,
where Aristotle distinguishes four domains of arguments:

Of arguments (logoi) used in discussion (en tôi dialegesthai) there are four
kinds (genê), Didactic (didaskalikoi), Dialectical (dialektikoi), Examination-
arguments (peirastikoi) and Contentious arguments (eristikoi). Didactic
arguments are those which reason from the principles (archai) appropriate
to each branch of learning and not from the opinions of the answerer (for he
who is learning must take things on trust6). Dialectical arguments are those
which, starting from generally accepted7 opinions (endoxa), reason to
establish a contradiction. Examination-arguments are those which are based
on opinions held by the answerer and necessarily known to one who claims
knowledge of the subject involved8 […]. Contentious arguments are those
which reason or seem to reason from opinions which appear to be, but are
not really, generally accepted. (SE 2, 165a38-b8.)

The didactic arguments are later referred to as ‘Demonstrative (apodeiktikoi)’
(SE 2, 165b9).9 Though this gives us a typology of arguments (or reasonings, as
we might say) and the definitions that follow are not straightforwardly referring to
types of dialogue, it is not hard to imagine some idea of types of dialogue (as
activity types) motivating the names given to these kinds of arguments. These
names, one could say refer to the type of dialogue in which the arguments (or
reasonings) so called are prominent. All of these types are dialectical in the wider
sense of dialectical which refers to discussion in general (en tôi dialegesthai). But
the second type would be dialectical in a more restricted sense and perhaps coincide
with the activity of dialectical discussion in the Academy in which the Questioner
may attempt to ‘reason to establish a contradiction’ (SE 2, 165b4) in the sense of
deducing the contradictory of the Answerer’s thesis.10 This gives us four types of
dialogue for the old dialectic: (i) didactic discussion, (ii) dialectical discussion (in
the narrower sense), (iii) peirastic discussion, and (iv) eristic discussion. There is
no reason to suppose that the use of each type of argument (reasoning) is restricted
to the type of dialogue that is characterized by it.

Our next step is to get a better picture of the nature of each of these types of
dialogue (or discussion) of the old dialectic. For this we use the typology of the
new dialectic and try to assign to each of the old types a place in that system. This
is not too hard in some cases, but rather challenging in others.

It may be clear that Aristotle’s didactic discussions can often be classified as
information-seeking dialogues. Didactic discussions are those characterized by
didactic or demonstrative reasoning, which starts from the principles of a specific
discipline. The picture is that of a teacher presenting a proof and a student who
confirms each step of the proof as soon as he has understood it. The function of
reasoning in the presentation of a proof may be seen as primarily explanatory,
since the student need not to be convinced by reasoning of the truth of what the
teacher proves to be true. He could accept that on authority. The activity type is
both didactic and demonstrative. Another kind of didactic discussion turns up



38     Erik C. W. Krabbe & Jan Albert van Laar

when a scholar who has just discovered a proof for a theorem uses demonstrative
reasoning in order to convince a colleague of the truth of the theorem. According
to the new dialectic this would constitute a case of persuasion dialogue, where the
reasoning has its primary persuasive function. What is missing in the old dialectic,
however, is the use of reasoning leading to the discovery of a proof. According to
the new dialectic, this may happen in inquiry where two or more scholars work
together to find a proof and where their reasoning has a probative function.

Neither are Aristotle’s eristic discussions hard to classify. From the point of
view of the new dialectic they are indeed a kind of eristic dialogues—altercations
of the subtype that is either called also eristic discussion (Walton an Krabbe, 1995,
p. 78) or sophistical dialogue (Walton 1998, p. 190). Reasoning in these dialogues
has both polemic and persuasive functions. Within this group of eristic dialogues,
Aristotle’s eristic discussions are distinguished by exhibiting the format of academic
dialectical discussions, with the roles of Questioner and Answerer. According to
Aristotle, the aim of the eristic discussant can either be nothing more than to
appear to have won the discussion (the eristic discussant in a narrower sense), or
to appear wise and thus make money (the sophistical discussant) (SE 11, 171b25-
34). In what we could call a purely eristic discussion, both participants behave
eristically (i.e., do not hesitate to use fallacious arguments or sophistical tactics),
but it is also possible that, starting from some other type of dialogue, the eristic
character is forced from one side. In terms of the new dialectic, this would constitute
a shift, but not an embedding. In dialectical and peirastic discussions, there is
always the danger of a shift to eristic discussion, for the ‘art of sophistry’ is near
by (SE 34, 183b2). For instance, in peirastic discussions, where (as we shall see
below) would-be experts are exposed, a discussant may abuse his skills to refute
an interlocutor who happens to be a real expert—thus the discussant could pretend
to be himself more of an expert than the interlocutor (Fait, 2002).

It is a much more arduous task to classify dialectical and peirastic discussions
according to the typology of the new dialectic. Dialectical discussions are
characterized by dialectic reasoning which starts from acceptable (‘generally
accepted’, endoxical) propositions. The notion of an acceptable proposition is
defined in the Topics: ‘Generally accepted11 opinions, (endoxa), on the other hand,
are those which commend themselves to all or to the majority or to the wise—that
is, to all of the wise or to the majority or to the most famous and distinguished of
them’ (Top. I.1, 100b21-23). This definition leaves room for several kinds of
acceptability depending upon the reference group (either all people, or the majority,
or …) to which the opinions commend themselves. This reference group is a
parameter that must be determined by the context of dialogue, i.e., in some context
‘acceptable’ means ‘acceptable for all’, in another context ‘acceptable for the
majority’, and so on. Each kind of acceptability determines a kind of dialectical
reasoning, and each kind of dialectical reasoning will in its turn determine a subtype
of dialectical discussion in which it is prominent. Assuming that dialectical discussion



About Old and New Dialectic     39

coincides with dialectical discussion as practiced in the Academy, we may conclude
that this academic activity type admits of various subtypes dependent upon the
presupposed concept of acceptability. Clearly dialectical discussion is different
from didactic discussion, but in order to classify it according to the typology of
the new dialectic we must know more about its goals. Why should one have these
discussions? Are they meant to solve differences of opinion? Or to contribute to
finding the truth? Or something else?

A sort of answer can be found in Top. I.2, where Aristotle sets out to discuss
in what ways his treatise (the Topics) is useful. Since the Topics constitutes a
manual for dialectical discussion (so we may suppose),12 this boils down to
explaining the usefulness of these discussions.  Aristotle distinguishes three purposes:
‘mental training (gumnasia), conversations (enteuxeis) and the philosophic sciences
(hai kata philosophian epistêmai)’ (Top. I.2, 101a26-27). In fact he thus announces
three subtypes of dialectical discussion. The first kind of dialectical discussion
aims at training the participants in dialectical discussion. It is rather obvious that
for each type of discussion, discussions can be used for practicing and training.
But this is useful only if discussions of that type can serve some other purpose as
well. Therefore training can not count as an independent goal of discussion.

The second type of dialectical discussion deals with enteuxeis (conversations
or encounters). The problem is what conversations or encounters are meant.
Aristotle writes:

It is useful for conversations (enteuxeis), because, having enumerated the
opinions of the majority (hoi polloi), we shall be dealing with people on the
basis of their own opinions, not of those of others, changing the course of
any argument which they appear to us to be using wrongly.13 (Top. I.2,101a29-
33.)

This passage must be linked to one in the Rhetoric, which refers to the Topics,
and deserves to be quoted:14

Further, in dealing with certain persons, even if we possessed the most
accurate scientific knowledge, we should not find it easy to persuade them
by the employment of such knowledge. For scientific discourse is concerned
with instruction (didaskalias), but in the case of such persons instruction is
impossible; our proofs and arguments must rest on generally accepted
principles (ta koina), as we said in the Topics, when speaking of converse
(enteuxis) with the multitude (hoi polloi). (Rhet. I.1.12, 1355a24-30.)

The type of encounter (‘conversation’, ‘converse’) Aristotle is referring to in both
passages, seems to be one between an expert and a layman, where the expert tries
to convince the layman of the truth of some proposition without using the scientific
principles on which this proposition really rests, since these are beyond the layman’s
comprehension. The expert, as a Questioner, must start from acceptable propositions
(“opinions of the majority”) with which the layman can agree. Discussions of this
type seem to be both in the scope of dialectic and in that of rhetoric, or, depending
on their subject, either in the scope of the one (if their subject belongs to philosophy



40     Erik C. W. Krabbe & Jan Albert van Laar

or science) or in the scope of the other (if their subject belongs to politics or law).
From the point of view of the new dialectic, this type of dialectical discussion
would of course be classified as persuasion dialogue.

One may easily imagine other kinds of encounters where a dialectical discussion
would be useful. For instance, a visit from a traveling scholar belonging to a
competing school of philosophy. To deal with the visitor in debate one can not use
demonstration, since the principles of the schools diverge, but one can use arguments
based on what is acceptable and on the visitor’s own opinions. Thus dialectical
discussion remains an option in this case. This would again be persuasion dialogue,
but could easily shift to a more eristic type of dialogue.

If peirastic discussion (a discussion characterized by examination-arguments)
is treated as a subtype of dialectical discussion, a step that is taken by Aristotle at
some places (SE 8, 169b25; 11, 171b4-6; 11, 172a21), it would also be a type of
dialectical discussion dealing with encounters. Again these would be encounters
between laymen and experts, but this time it is the layman who is testing the expert
by questioning. Peirastic discussions mirror the earlier type of discussion between
expert and layman. In the earlier type, the expert as a Questioner tries to convince
the layman, who is the Answerer, of the truth of a proposition, starting from what
is acceptable and admitted by the layman, whereas in the peirastic discussion, the
layman, as a Questioner, uses acceptable propositions to refute a would-be expert,
who is the Answerer, and show him not to be the expert he pretends to be. This
interpretation of peirastic discussion is supported by several passages in the
Sophistical Refutations (SE 8, 169b23-25; 11, 172a21-172b1) and can be illustrated
by most Socratic dialogues, in which peirastic discussion in this sense is at least a
component. Yet, the definition of examination-argument in SE 2, quoted above,
seems not to accord well with it: why would this type of argument, that is supposed
to be prominent in peirastic discussions, have to start from propositions that are
‘necessarily known to one who claims knowledge of the subject involved’? Why
couldn’t they, typically, start from false propositions that the would-be expert has
admitted as his opinion? Fortunately, there is another reading of this passage which
solves the problem: ‘Examination-arguments are those arguments that start from
propositions that according to the answerer must also be known by whoever
claims knowledge of the subject involved.’15 According to this interpretation the
premises of the examination-argument need not be known by anybody, it is only
that the answerer (i.e. the would-be expert) assumes, truly or falsely, that they are
known by those who claim to be an expert. This makes sense because any impostor
would try to give answers that accord with what a real expert would answer.

Given this view of peirastic discussion, we may see that it could either be
considered as a separate type of discussion or as a subtype of dialectical discussion,
since presumably dialectic arguments will occur in peirastic discussions besides
typically peirastic ones. From the point of view of the new dialectic, peirastic
discussion is a kind of expert examination dialogue—a type we discussed extensively



About Old and New Dialectic     41

in Section 3.2. In our view, peirastic discussion can be either subsumed under the
examination subtype of persuasion dialogue or the examination subtype of inquiry,
depending on whether at least one of the participants is initially committed to a
positive or negative standpoint about the expertise of the alleged expert or not.

Having treated the purposes related to training and to conversations (encounters),
let us now turn to those related to the philosophical sciences. It appears that
dialectical discussion can be useful for the philosophical sciences in two ways.
First, ‘it is useful, because, if we are able to raise difficulties on both sides, we
shall more easily discern both truth and falsehood on every point’ (Top. I.2, 101a33-
35). This means that dialectical discussion can be helpful in inquiry and contain
reasoning with probative (including explorative) functions. Second, dialectical
discussion is said to be useful for the inquiry into the foundations of the sciences.
According to Aristotle:

… it is useful in connexion with the ultimate bases of each science (epistêmê);
for it is impossible to discuss them at all on the basis of the principles (archai)
peculiar to the science in question, since the principles are primary in relation
to everything else, and it is necessary to deal with them through the generally
accepted opinions on each point. This process belongs peculiarly, or most
appropriately to dialectic; for, being of the nature of an investigation
(exetastikê), it lies along the path to the principles of all methods of inquiry.
(Top. I.2,101a36-b4.)

This passage can be interpreted and translated in different ways. The exact nature
of the usefulness of dialectical discussion for the foundations of the sciences
remains a moot point but it is clear that dialectic can not establish the first principles
of the sciences by demonstration (proof).16 Yet it seems possible that dialectic
could help to make these principles known in some other way, since what is
thoroughly discussed could become better known in some sense. We conclude
that this type of dialectical discussion belongs, from the perspective of the new
dialectic to inquiry and that the reasoning in it has probative functions, though
these probative functions will not be demonstrative in the Aristotelian sense.

4. Evaluation and Fallacies

In this section we shall discuss and compare the use that can be made of the old
and the new dialectic for the evaluation of argumentation and reasoning in dialogue.
For this Walton’s new dialectic puts forward normative models, different ones for
different types of dialogue. For persuasion dialogue we have the well-specified
pragma-dialectical model to do the job, but it is less clear how the models we
should use for the other types of dialogue are to be specified in detail. This makes
the discussion somewhat abstract. The old dialectic does not provide normative
models comparable to the pragma-dialectical one, but Aristotle’s theory of fallacies
in the Sophistical Refutations can be used for evaluative purposes.



42     Erik C. W. Krabbe & Jan Albert van Laar

4.1 Evaluation and Fallacies in the New Dialectic

When we evaluate argumentation or reasoning, we are trying to determine to what
degree a piece of discourse is reasonable. According to Walton’s new dialectic, an
argument is reasonable when it is ‘contributing to the goal of the type of discourse
which the argument is supposed to be part of’( Walton, 1998, p. 249). The first
step in an evaluation is to determine, at a local level of analysis, what the reasoning
amounts to, in particular what logical form or what argument scheme is used, and
whether the reasoning is correct as far as can be determined at this level. The
second step, at a more global level, is to identify the context of dialogue and to
determine how the argument is used in that context. The third step is to critically
compare the use of this argument in this context of dialogue with what is required
by the normative model that is supposed to hold sway over this part of the
dialogue.17 Distinctive of the approach is that one and the same argument, in the
sense of a complex of reasoning exhibiting a particular form or scheme, can be
appropriate in one context of dialogue while it would be inappropriate or even
dialectically illegitimate when used in a dialogue of a different type. Reasoning that
contains a threat, for instance, can plausibly be seen as appropriate in a negotiation
dialogue, but as ill-chosen, or even fallacious, when put forward in a persuasion
dialogue (Walton, 1998, Section 4.6).

Given the possibility of a shift to another type of dialogue, two theoretical
issues of evaluation arise. First, from what perspective must an argument be
evaluated that occurs after the shift, from the perspective of the normative model
for the original type of dialogue, or from that of the normative model for the type
of dialogue that results after the shift? Second, how should the shift itself be
evaluated? Walton’s proposal for answering the first question is to evaluate
arguments retrospectively (Walton, 1998, p. 205): first it should be determined
what kind of dialogue the participants were engaged in at the outset of their
conversation; then, the norms for that dialogue type provide the norms for evaluating
the argument, even if the argument occurs at present in a different kind of dialogue.
In case a dialogue is mixed, exhibiting characteristics of two or more dialogue
types (a political debate, for instance, contains features of eristic dialogue as well
as of persuasion and deliberation), the evaluation must be conditional in nature, as
the following format expresses: ‘in so far as this move is supposed to contribute to
a dialogue of this or that type, this move is dialectically legitimate/illegitimate’. As
to the second question, the evaluation of the shift itself, Walton suggests that
innocuous interruptions of a dialogue, that permit the dialogue to be resumed, be
evaluated as dialectically legitimate (licit) (1998, p. 200). In other cases the
functionality of the new dialogue with respect to the original one must be assessed.
A dialectical shift (that is not an innocuous interruption) from a dialogue d of type
D to a dialogue d′ of type D′ is licit if and only if (1) d′ is functional for achieving
the goals of d (that is to say, if d′ is embedded in d), and (2) the speaker has made
it clear to the respondent that he has moved from a dialogue of type D to a different



About Old and New Dialectic     43

dialogue type D′ and (3) the respondent agrees (explicitly or implicitly) with this
shift (this is the way we understand Walton, 1998, p. 215-216).

In many cases what are commonly called fallacies are connected with illicit
shifts. In A Pragmatic Theory of Fallacies (Walton, 1995) Walton presents a
definition of fallacy which, we presume, still holds for his new dialectic:

A fallacy is (1) an argument (or at least something that purports to be an
argument); (2) that falls short of some standard of correctness; (3) as used in
a context of dialogue; (4) but that, for various reasons, has a semblance of
correctness about it in context; and (5) poses a serious obstacle to the
realization of the goal of a dialogue. (Walton, 1995, p. 255.)

One of the possible reasons for a fallacious argument having the semblance of
correctness, is that in a different context of dialogue, the very same sequence of
propositions would have been correct.

Walton distinguishes two kinds of fallacy: paralogisms and sophisms (Walton,
1995, xii). The paralogism can be identified at the first, local, level of analysis
where the evaluator checks what kind of reasoning is used, that is, what logical
form it has or what argumentation scheme it instantiates, and whether it satisfies
the requirements for that kind of reasoning. Sophisms are only localizable in the
second evaluation step, at a more global level, when examining longer stretches of
a dialogue, for instance a sequence of several applications of argumentation schemes,
or one application of an argumentation scheme together with a succession of
critical questions and responses that arises from that application. A sequence of
connected moves in a dialogue, longer than a single application of an argumentation
scheme but generally shorter than a complete dialogue, is called a theme or an
argumentation theme by Walton (Walton, 1995, pp. 34, 201-203). The appropriate
device for examining themes is the normative profile of dialogue that shows how
dialogue moves should be connected. A sophism then is the misuse of a theme.

From the general definition of fallacy it becomes clear that just as the evaluation
of arguments in general is dependent on the type of dialogue, so is the judgment
that an argument (or other contribution to a dialogue) is fallacious. What, in a
given dialogue, counts as a fallacy depends upon the type of dialogue and its
normative model. Also, the notion of fallacy is in its applications not restricted to
persuasion dialogue, but can be applied to arguments or other uses of reasoning in
dialogues of all types. For instance, when a participant in a negotiation dialogue
shifts towards a quarrel by attacking the respondent’s character and using abusive
language, this participant can be said to have committed the fallacy of the abusive
ad hominem (misusing the polemic function of reasoning) by violating a normative
standard for negotiation dialogue that must be obeyed for the purpose of making a
deal (Walton, 1998,  p.262).

When applying the distinction between activity types and normative models,
we can easily understand that one and the same reasoning can in principle be
fallacious from the perspective of one normative model, but at the same time a



44     Erik C. W. Krabbe & Jan Albert van Laar

legitimate or even a highly appropriate kind of argument when seen from the angle
of a different normative model. Also, we may presume that once the new dialectic
will have developed fully specified normative models for all common activity types,
the normative model most easily applied to evaluate a given argument will, in a
simple case, be the one associated with the activity type in which the argument
occurs, which may be empirically determined. But how about shifts? Must we
always follow Walton’s retrospective principle? This would force us, in every
case of evaluating an argument, to go to the initial stages of the dialogue to determine,
empirically, what activity type the dialogue originally exhibited. It would not be
sufficient to determine the activity type of the part of dialogue in which the argument
occurs. We think this exercise should not be required for all cases of evaluation.
After all, it is up to the evaluator what he wants to evaluate. Rather than by the
principle of retrospective evaluation, the choice for a set of norms ought to be
determined by the research interests of the evaluator of the reasoning. Similarly,
the judgment that a shift is licit (or illicit) is relative to a choice for a specific
normative model. For instance, from the perspective of the normative model for
negotiation dialogue, a shift from a negotiation dialogue (in the sense of the
argumentative activity type) towards a persuasion dialogue could work out
fallaciously, while the same shift could be legitimate from the perspective of the
normative model for persuasion dialogue itself. The choice is up to the evaluator.

4.2 Evaluation and Fallacies in the Old Dialectic

The cornerstone of Aristotle’s theory of fallacy is his system of thirteen types of
fallacious (sophistical) refutation in the Sophistical Refutations. In contradistinction
to the new dialectic theory of fallacy, this system is independent of the type of the
dialogue in which the fallacy occurs. It is presupposed, however, that this type
conforms to the format of academic dialectical discussion in which a Questioner
tries to refute an Answerer. This independence from types of dialogue is a
consequence of the fact that the whole system is built upon the definition of
refutation (elengchos), which is again based upon that of (deductive) reasoning
(sullogismos), which definitions are also independent of types of dialogues. These
definitions are well-known:

Reasoning (sullogismos) is based on certain statements made in such a way
as necessarily to cause the assertion of things other than those statements
and as a result of those statements; refutation (elengchos), on the other
hand, is reasoning accompanied by a contradiction of the conclusion. (SE 1,
164b27-165a3.)

A fallacious refutation is an alleged refutation that violates one of the conditions of
refutation (SE 6, 168a20-21). Since violating a condition of the definition of
refutation amounts to committing the fallacy of ignoratio elenchi (in its original,
not its present, sense), ignoratio elenchi is the master fallacy of which the other
twelve types of fallacious refutation are special cases. The other fallacious



About Old and New Dialectic     45

refutations can then be assigned to whatever clause in the two definitions they
violate.18

There are five clauses that can be violated. First, there may be no necessity.
Nine types of fallacies fall under this head: the six fallacies dependent upon the use
of language (homonymy, amphiboly, composition, division, accent, and form of
expression) and three of the fallacies that are independent of language (accident,
secundum quid, and consequent). Second, the conclusion may not be a result of
the premises, i.e., some premise may not contribute to the conclusion. This leads
to the fallacy of false cause—a mistaken reductio ad absurdum. Third, the premises
may not be put in place (‘made’) in a correct way: the fallacy of many questions.
Fourth, the conclusion may not really be different from one of the premises: the
fallacy of begging the question. Fifth, the reasoning may not be ‘accompanied by
a contradiction of the conclusion’, i.e., the conclusion does not contradict the
thesis of the Answerer. In this case there is correct reasoning, but no refutation.
Fallacies under this head are the six fallacies dependent upon the use of language
listed above as well as the fallacy secundum quid (which in this case gets the name
of the master fallacy: ignoratio elenchi).

Most of the Aristotelian fallacies of refutation refer to a single step of reasoning.
In terms of the new dialectic, they are schematic (paralogisms) rather than thematic
(sophisms). But in three cases the fallacy refers to a larger context of reasoning,
and is thematic rather than schematic: these cases are the fallacy of false cause,
which refers to the procedure of reductio ad absurdum-reasoning, the fallacy of
many questions, which refers to the procedure of questioning, and the fallacy of
begging the question, which refers to a circular theme of reasoning. Thus both
schematic and thematic fallacies were already present in the old dialectic.

If one were to identify Aristotle’s contribution to dialogue evaluation and fallacy
theory with his theory of fallacious refutations, one might be tempted to conclude
that the old dialectic is very far removed from the new dialectic as far as sensitivity
to type of dialogue and dialogical context goes. Below we shall see this not to be
the case. First we shall point out that there are other fallacies than the fallacies of
refutation, which constitute only a subgroup of the eristic fallacies. Then, there
are also fallacies that are not even eristic (or sophistical19). We shall meet with
examples of a more context-sensitive treatment in some cases. Next we shall
consider Aristotle’s evaluation of dialogical behaviour, and shall find his approach
more congenial to the new dialectic than the theory of fallacies of refutation made
us expect.

In the passage from SE 2, quoted in Section 3.1, Aristotle says that ‘Contentious
(eristikoi) arguments are those which reason or seem to reason from opinions
which appear to be, but are not really, generally accepted (endoxa)’ (SE 2, 165b7-
8). From a parallel passage in the Topics, it is clear that one other group of reasonings
should be added: those that start from opinions that are indeed acceptable, but
merely appear to reason correctly from them (Top. I.1, 100b23-25). The case of



46     Erik C. W. Krabbe & Jan Albert van Laar

apparent reasoning (from premises that are either acceptable or not) seems to be
covered by the fallacious refutations, but these are not the only eristic arguments.
The case that remains is that of otherwise correct reasoning starting from premises
that appear to be acceptable, but are really unacceptable. This case is not fully
covered by the fallacy of many questions. Since the notion of acceptability is
context dependent, this would give us a class of fallacies20 in the old dialectic that
depends upon the context and type of dialogue. Thus, in some cases of dialectical
discussion, it would be appropriate to start from premises acceptable for all or
most people, whereas in a more specialized philosophical inquiry the premises
should be acceptable for the wise. In a context of peirastic discussion, the
Questioner should, generally, even be permitted to avail himself of falsehoods that
the Answerer, out of ignorance, admits. These answers could be unacceptable, so
the reasoning based on them would not constitute a dialectical argument according
to the definition in SE 2. This seems to contradict Aristotle’s classification of
peirastic discussion as a kind of dialectical discussion, unless we suppose (as
suggested before) that the Aristotelian types of dialogue need not be restricted to
the types of reasoning by which they are characterized. Dialectical discussion,
though characterized by dialectical arguments, may legitimately contain other
arguments as well.

To each fallacious refutation it is possible to add a premise that falsely states
that the type of reasoning in the refutation is correct. For instance, in the case of
homonymy, one may add the false principle that states that words have only one
meaning, and in the case of the fallacy of consequence one may add the principle
that consequence is invertible. According to Aristotle the resulting arguments are
sophistical arguments (even though the reasoning may now be correct). They
may yield a refutation ‘relatively to the answerer, but not absolutely’ (SE 8, 170a17-
18). This evaluation of arguments that are based upon a fallacy condoning principle
falls short of a context free rejection of them as fallacious. Rather, they are
arguments ad hominem (based on concessions), which could be useful in some
situations. But in the usual dialectic and peirastic context they would count as
fallacies. In peirastic this would be so because they are not ‘germane to the subject
in hand’ (the subject of expertise of which the Answerer claims knowledge) (SE 8,
169b22-23). These arguments do not show up the Answerer as ignorant, ‘for men
try to entrap even the man of scientific knowledge by these arguments’ (SE 8,
169b28-29).

Thus far we have looked at eristic and sophistical reasoning in a context of
dialectical and peirastic discussion, but these fallacies can also be committed in a
context of scientific discussion. Again, it seems to be sufficient to reject an argument
as eristic or sophistical if it ‘does not accord with the subject matter concerned’
(SE 11, 171b17-18), even if the reasoning itself happens to be correct, and its
conclusion true. One example Aristotle adduces is that of denying ‘that it is better
to take a walk after dinner because of Zeno’s argument’ (for the thesis that motion



About Old and New Dialectic     47

is impossible). This ‘would not be a medical argument; for it is of general application’
(SE 11,172a8-9).

The eristic and sophistical reasonings and refutations are not the only fallacies
recognized by Aristotle. There are also ‘false reasonings based on premises peculiar
to certain sciences, as happens in geometry and the sciences kindred to it’ (Top.
I.1, 101a6-8). One may think of a false proof in geometry, based on cheating in
drawing the figures. This category is kept apart from the eristic arguments that
may also occur in scientific discussions, because eristic arguments start from
(allegedly) acceptable premises, are of general application, and do not accord with
the subject matter, whereas the scientific false reasonings start from at least one
false scientific premise, are restricted to a specific discipline, and do accord with
the subject matter. One could say: eristic arguments pretend to be good dialectical
arguments (and sometimes to solve a scientific problem), whereas the false
reasonings to which Aristotle here refers pretend to be good scientific arguments
(cf. SE 11, 171b34-36). These false reasonings are fallacies of demonstration, and
therefore belong, primarily, to didactic discussion. Moreover each fallacy of this
type belongs to a subtype of didactic discussion pertaining to a specific discipline.
In that sense, these fallacies are context-sensitive. It is however not the case, as
happens in the new dialectic, that some false reasoning within some discipline
would be nonfallacious in another discipline or in another type of dialogue.

Aristotle’s normative remarks about dialectic are not restricted to the serious
flaws we would call fallacies, but cover the behaviour of both Questioner and
Answerer in relation to the tasks they have to perform and their shared responsibility
for the execution of these tasks, thus foreshadowing the idea of normative models
for discussion:

As regards answering, the function of the good answerer must first be defined,
as also that of the good questioner. The function of the questioner is so to
direct the discussion as to make the answerer give the most paradoxical
replies that necessarily result because of the thesis. The function of the
answerer is to make it seem that the impossible or paradoxical is not his fault
but is due to the thesis; for possibly, to lay down the wrong thesis originally
is a different kind of mistake from not maintaining it properly after one has
laid it down.21 (Top. VIII.4, 159a16-24.)

This normative passage assigns roles to the discussants that, though very different,
are not really in opposition. Both, the Answerer and the Questioner could do well.
Indeed, Aristotle stresses that in good dialectical argument Questioner and Answerer
should cooperate: ‘…it is not within the power of one party only to ensure the
proper accomplishment of the common task’ (Top. VIII.11, 161a19-21). Both
questioner and answerer should cooperate to avoid a degeneration into eristics,
‘for he who asks questions in a contentious spirit (eristikôs) and he who in replying
refuses to admit what is apparent and to accept whatever question the questioner
wishes to put, are both of them bad dialecticians’ (Top. VIII.11, 161b2-5). The



48     Erik C. W. Krabbe & Jan Albert van Laar

uncooperative and peevish (duskolos) answerer obstinately refuses to admit a
universal statement, even upon presentation of umpteen acceptable instances, and
refrains from formulating an objection or a counter-argument, behaving that way
‘with the object of destroying the reasoning’ (Top. VIII.8, 160b12-13). This peevish
behaviour makes it ‘necessary sometimes to attack the speaker and not the
thesis,…By behaving peevishly, then, people make their discussions contentious
(agônistikai) instead of dialectical’ (Top. 161a21-24). Apparently, to prevent the
discussion from lapsing into agonistics, i.e. eristics, it would be permissible to
apply ad hominem arguments, which in this case may not comprise only arguments
from concessions, but also personal attacks.22 Since in normal circumstances
these arguments, so we may suppose, are not admissible, this shows how the old
dialectic norms for dialectical behaviour were sensitive to context. Furthermore,
this yields a useful hint for the new dialectic, where we do not only need to have
normative models—in which each participant behaves rationally, but also models
that tell us how to act when our interlocutor’s behaviour does not conform to
standards of rationality. The problem of peevish answerers in the old dialectic is to
be compared to the problem of discussants that refuse to commit themselves or
retract their commitments at the first sign of trouble, which occurs in the new
dialectic (Walton and Krabbe, 1995, p. 9, 139).

Aristotle does not present a theory of dialogical shifts, let only a theory that
connects fallacies to shifts. These are original contributions of the new dialectic.
But on the other hand, dialogical shifts seem to be exactly what worries Aristotle
when he writes about peevish Answerers and eristic Questioners. The risk for any
dialectical discussion to lapse into eristic discussion precisely provides the motivation
for studying the fallacies. In this respect the old dialectic does anticipate the new
dialectic.

5. Strategies in Old and new Dialectic

Each type of dialogue is characterized by a specific kind of initial situation, by the
shared main goal of the conversations that exemplify the type, and by the individual
aims of the participants. Depending upon the type of dialogue, the aims of the
participants may consist in nothing else than contributing to the shared main goal,
without any antagonism or opposition (this would be the case for the
nonargumentative types of dialogue: inquiry, deliberation, and the information-
seeking dialogue), or they may differ from the main goal (this would be the case
for argumentative dialogues: persuasion dialogue, negotiation, and eristic dialogue).
Dialogue types of the first kind are purely cooperative, whereas the argumentative
types are competitive. Yet, the argumentative dialogues are also cooperative, in as
far as they are directed at a common goal. Indeed, the competitive structure of
these dialogues can itself be seen as a means for achieving the common goal.

Where people converse to achieve certain aims, they will in their conversation
make choices that are influenced by these aims. But one participant’s choices will



About Old and New Dialectic     49

also be influenced by those made by his interlocutor. Discerning these influences
in an actual dialogue, one may conclude that a participant is using a kind of tactics
(at a local level) or strategy (at a more global level). Thus strategic behaviour in
various activity types is an empirical fact. What a strategy is, however, is more
easily explained in terms of models that specify the possible dialogue situations and
the different ways a conversation can develop. A strategy for a participant P is a
function that selects for each dialogue situation to which it applies and in which it
is P’s turn to move, a move for P from the set of moves M provided for this
situation by the model. The requirement that just one move be selected may be
relaxed to allow the selection of a subset of M (in which case we would, technically,
have a partial strategy). Though in every conversation a participant, at each point
where he is to make a choice, selects just one move as his next move, a participant’s
strategy does not only contain this choice, but also what he would have done, had
the conversation, earlier, taken a different turn.

Fixing upon one type of dialogue, with one normative model, three different
groups of strategies can be distinguished. First, there is the group of permissible
strategies, that is of all strategies that only select permissible (legitimate, licit)
moves (if there are any): such moves as will contribute to achieve the shared main
goal. Since the normative model determines what is to be done in order to enable
the achievement of the main goal, these are precisely those choices that (if possible)
keep a participant within the confines of the rules for that type of dialogue. Second,
there is the group of assertive strategies that only select assertive moves (if there
are any): such moves as will contribute to achieve one’s individual aim. In
argumentative dialogues, a strategy of this kind may or may not be permissible.
The two groups will, presumably, overlap, but also there will be strategies of either
kind that are not strategies of the other kind. Assertive strategies that are not
permissible will in some situations to which they apply make selections that go
beyond the confines of the normative model. Consequently they can not be modeled
as a strategies within that model (but they can be modeled within a more permissive
model: one for activities that include dialogues of the type we are concerned with,
but also other conversations; cf., Mackenzie, 1998, p. 172-176; Van Laar, 2003,
paragraph 7.1). The third group of strategies contains precisely the appropriate
strategies, these are those strategies that do only select appropriate moves (if there
are any). We define an appropriate move as a permissible move that, if there are
assertive moves available among the permissible moves, is also an assertive move,
and if these are not available, is minimally detrimental for reaching one’s personal
aim. So if there is a conflict between main goal and personal aims, an appropriate
strategy always gives priority to the main goal. This third group of strategies is a
subset of the first group, but not necessarily of the second. For a strategy of the
third group must always prefer the common goal to the personal aim where these
conflict, whereas for strategies of the second group this is the other way round.



50     Erik C. W. Krabbe & Jan Albert van Laar

We shall now proceed to scrutinize both the old and the new dialectic for
points of view that relate to each of these three groups of strategies. As we did
before, we shall in each case start with the new dialectic.

Permissible strategies are suggested by Walton’s discussions on the norms that
ought to govern the distinct dialogue types. For instance, Walton discusses several
tactics that follow maxims for negotiation dialogues that are proposed by Fisher
and Ury, such as: focus on interests and not on positions and separate the people
from the problem, for instance by refraining from abusive ad hominem arguments.
With respect to the use of argumentation in persuasion dialogues, Walton mentions
as basic requirements that the respondent accepts the (basic) premises as
commitments, that each inference in the chain of arguments is structurally correct
according to an argumentation scheme and that the chain of argumentation has the
proponent’s thesis as its ultimate conclusion (Walton, 1998, p. 43). So every choice
of a move that fulfils these requirements can be part of a permissible strategy for
the arguer in a persuasion dialogue. Because the norms in one model can differ
from the norms in another model, one and the same choice can be part of a
permissible strategy for one type of dialogue but not of any permissible strategy
for another type. A choice for the argumentation schemes ‘ad hominem’, ‘ad
verecundiam’ and ‘ad ignorantiam’23 can be part of a permissible strategy, in the
case of a persuasion dialogue, but is ruled out by the norms for scientific inquiry
(Walton, 1998, p. 270).

In the old dialectic, permissible strategies for dialectical discussion are suggested
by the quotes from Topics VIII at the end of Section 4.2. Questioner and Answerer
each have their task to perform, but as we saw they should also cooperate. The
intermediate books of the Topics provide a long list of ways to construct arguments
based on argumentation schemes that could be part of a permissible strategy. The
Sophistical Refutations discusses fallacies, which cannot be part of permissible
strategies for dialectical or peirastic discussions, but, presumably, can be part of
permissible strategies for eristic discussions. So in the old dialectic, too, we observe
that permissibility depends on the type of dialogue.

Assertive strategies, which try to achieve individual aims, irrespective of whether
their choices remain within the confines of the normative model, are suggested in
Walton’s new dialectic by numerous detailed discussions about the legitimacy of
dialogue moves and of sequences of dialogue moves. As to sequences of moves,
Walton (1995) mentions the following two strategies for achieving an individual
aim in a longer stretch of dialogue, the realm of the thematic fallacies. One concerns
circular argument, which can be concealed by extending the chain of argumentation
from starting point to conclusion (Walton, 1995, p. 208). The other consists of a
many questions sort of tactic aimed at getting the respondent to commit himself to
an unfavourable proposition P. This can be achieved by posing a question that
presupposes that the respondent has already answered ‘yes’ to the question ‘Do
you think that P?’. For instance, the question ‘Have you stopped cheating on your



About Old and New Dialectic     51

income tax returns?’ can entrap the respondent (or an audience) by leaving out the
question ‘have your cheated on any of the income tax returns in the past?’,
presupposing that the respondent answered yes to this preliminary question (Walton,
1995, pp. 202-203). This strategy admits of fallacious as well as legitimate instances,
depending on whether it can really be supposed that the respondent would indeed
have answered affirmatively to the preliminary question, had it been asked.

Assertive strategies are also suggested by numerous passages in the old dialectic
handbooks (especially Topics VIII and Sophistical Refutations 12, 15, and 17).
Aristotle gives all kinds of advice, both to the Questioner and to the Answerer,
about how to reach their personal aims (respectively, to refute the other and to
avoid being refuted). In giving such advice, Aristotle seems not so much concerned
with whether the proposed tactics would be permissible in view of the main goal
of the discussion. Sometimes the proposed tactics could be permissible in some
subtypes of dialectical discussion, or in certain situations in a discussion, but not
in others. For instance, a tactic could be permissible only in encounters with
interlocutors that themselves make use of polemic reasoning and eristic ploys, or
only in situations where the Questioner has to deal with a reluctant (peevish)
Answerer.24 What to think of the following advice for the Questioner?

To effect a refutation one expedient is length; for it is difficult to keep many
things in view simultaneously. […] One resource is speed; for when people
lag behind they see less far ahead. Further there are anger and
contentiousness; for when people are agitated they are always less capable
of being on their guard. (SE 15, 174a17-21.)

Aristotle goes on in the same vein and at the end of this chapter even suggests that
‘some propositions should not be asked at all but treated as admitted’ (SE 15,
174b39-40), which is suggestive of Walton’s many questions tactic. Such tactics
is hard to conceive of as permissible in a cooperative type of dialectical discussion,
or even a peirastic discussion, but could be permissible in certain competitive
kinds of encounters or situations.

Advice for the Answerer is equally unfit for cooperative dialectic discussion:
In the first place, then, just as we say that we ought sometimes deliberately to
argue plausibly rather than truthfully, so too we ought sometimes to solve
questions25 plausibly rather than according to truth. (SE 17, 175a31-34.)

The Answerer must try to parry the (alleged) refutation that the Questioner put
forward. To do so he must offer a solution (lusis). But if no real solution can be
found he is advised to present a plausible, but actually incorrect, solution. This is
so because one ‘must […] beware not of being refuted but of appearing to be so
[…]’ (SE 17, 175a40-41). This advice seems again to pertain to certain competitive
encounters or situations.26 But elsewhere Aristotle sets great store by correct
solutions of alleged refutations, for these are solutions that show exactly what
went wrong, as opposed to solutions that are directed ‘not to the argument but to
the man’ (SE 22, 178b17).27 This shows that the permissibility of giving incorrect
solutions depends upon the type of discussion or on the situation in a discussion.



52     Erik C. W. Krabbe & Jan Albert van Laar

Whatever we say about the permissibility of these various recommendations,
Aristotle presents them irrespective of whether they would contribute to the common
task of the Questioner or the Answerer, but purely as means for either to reach his
personal aim, which points to assertive strategies as we defined them.

Appropriate strategies prescribe, where possible, moves for a discussant that
optimize his dialogical behaviour by serving both the main goal and, as well as
possible, the discussant’s personal aim. Such strategies are always composed of
permissible moves (if these are available) and are consequently also permissible
strategies. A strategy is only appropriate if it optimizes the chances of a party to
achieve his individual aim. Selection of a move that is permissible but does not do
this will be called a blunder. A series of such moves will also be called a blunder.
One obvious advice for a party would be to avoid blunders. According to Walton,
an example of a blunder would be a discussant who, inadvertently, argues in a
circle and simply fails to convince his interlocutor (Walton, 1998, p. 258). We
consider such strategy to be a blunder only if no discussion rule has been violated,
because different from Walton, who states at this point that some rule violations
can be blunders rather than fallacies, we assume that blunders always consist of
permissible moves.

Van Eemeren and Houtlosser (1999, 2002, 2003) propose to reconstruct
argumentative discourse from the methodological assumption that a participant is
manoeuvring strategically between achieving his dialectical objective of resolving
the difference of opinion as well as achieving his rhetorical objective of getting
things his way. We can view strategic manoeuvring as attempts by a party to find
and develop an appropriate strategy. A successful attempt results in such a strategy.
An unsuccessful attempt results either in the committing of a fallacy, called a
derailment of strategic manoeuvring, or the making of a blunder.

The notion of strategic manoeuvring can be extended to other main goals than
that of conflict resolution. Given the main goal of making a deal, the dialogical
behaviour of parties in a negotiation can be examined from the perspective of
balancing the aim of getting what you want and keeping within the confines of an
effective negotiation directed towards a mutually acceptable result. In an eristic
dialogue, manoeuvring amounts to selecting a strategy by which one can hit out at
his adversary while leaving open the possibility of reaching an accommodation
that satisfies both parties. (Of course, this generalized form of strategic manoeuvring
has no application to dialogue types that do not start from some kind of conflict.)

To mention one example of an appropriate strategy: in an eristic dialogue,
there is a tactic which consists in getting the message across that it is the other
party who is quarrelling, while it is the speaker who tries to engage in a rational
critical discussion (Walton, 1998, p. 194-5). This will be familiar to many as an
often successful (and presumably permissible) strategic manoeuvre in eristic
dialogue. Whether such a tactic is also admissible from the perspective of a critical



About Old and New Dialectic     53

discussion will depend on the specifics of the situation: for instance, is the other
party really obstructing the resolution process?

The idea of an appropriate strategy is, we think, implicit also in the old dialectic
admonitions to Questioner and Answerer. One may turn, for instance, to the chapter
of Sophistical Refutations about how to get the adversary to commit himself to a
falsehood or to a generally unacceptable proposition (SE 12). Some of the strategies
under this head are, or are called, sophistical and could be classified as assertive
rather than appropriate strategies. For instance, the strategy of leading ‘one’s
opponent to the kind of statements against one has plenty of arguments’ is called
sophistical, but even here ‘it will be possible to do this in a right and a wrong way’
(SE 12, 172b26-27). On the other hand, many of these tactics would be appropriate
in many subtypes of dialectical discussion (cf. Dorion, 1995, Introd. 4.1). This
holds, for instance, for the tactic of seeking ‘paradoxes in men’s wishes and
professed opinions’ (SE 12, 172b36); for there are always discrepancies between
what people really want and what they profess to be desirable, and these can and
may be exploited in dialectic discussions, even when inquiry is the main goal. The
same holds for exploiting the opposition between what holds by nature and what
holds by convention (‘law’): ‘to a man who speaks in terms of nature you must
reply in terms of law, and when he speaks in terms of law you must lead the
argument to terms of nature’ (SE 12, 173a12-14).

Among the strategies in SE 15 and SE 17 many belong to the appropriate rather
than the assertive strategies (cf. Dorion, loc. cit.). For instance, the assertive
tactic for the Questioner of concealment (krupsis), which consists in concealing
one’s intentions by distancing and multiplying one’s questions, some of which will
ultimately yield the premises for the final and devastating syllogism, could be illicit
(not permissible) in some cases , but would generally constitute an appropriate and
therefore permissible strategy (SE 15, 174a26-29, Top. VIII.1, 156a7-9). The
same holds for the tactical advice for the Answerer not to hesitate to make
distinctions whenever an ambiguity looms ahead (SE 17, 175b37-38). In sum, we
conclude that appropriate strategies are present even in the Sophistical Refutations,
a work one would expect to treat only nonpermissible strategies.

6. Conclusion

Many features of the new dialectic can be found to be present in the old dialectic,
be it only implicit. The old dialectic had no explicit theory of types of dialogue but
the types of dialogue that belong to its domain all fit into the typology of the new
dialectic. Negotiation and deliberation do not occur in the old dialectic (which we
limited to the Topics and the Sophistical Refutations). Old dialectic is tied to the
academic format of dialectic discussion, whereas new dialectic has a wider scope,
including more of emotional and political interaction. But the typology of the new
dialectic, which was originally meant just to facilitate the study of persuasion
dialogue, is often not easy to apply and may have to be revised by future research.



54     Erik C. W. Krabbe & Jan Albert van Laar

New dialectic introduces the notion of a dialectical shift and connects this with
the notion of fallacy. Whether a particular argument is a fallacy is made dependent
on the type of dialogue and in many cases connected with a shift. The old dialectical
theory of fallacious refutation is in itself independent of considerations about type
of dialogue or dialogical context. But fallacious refutations are only part of what in
the old dialectic can be called fallacies. A further investigation of old dialectic
practices of evaluating fallacies and dialogical behaviour shows these to be much
closer to the new dialectic point of view than the theory of fallacious refutation
would lead us to expect. What now must be investigated in detail is how the notion
of fallacy or analogous notions should be applied to all the different types of dialogue
and to what extent these notions are tied to that of a shift.

Strategies of all three types we distinguished were foreshadowed in the old
dialectic as well. They not only regularly turn up in the new dialectic, but also
relate to the idea of strategic manoeuvring in the recently extended theory of pragma-
dialectics. Much more remains to be done. The next decades will, hopefully, witness
an increasing focus on strategies in various types of dialogue.

Endnotes
1 We gratefully acknowledge helpful comments from Pieter Sjoerd Hasper on our treatment of the
old dialectic. He also suggested many emendations in the translations from Aristotle we wanted to
quote. For the new dialectic we grateful acknowledge to have been inspired and instructed by
Douglas Walton, not only by his publications, but also by many dialogues over the years. We also
wish to thank an anonymous reviewer for helpful comment.
2 The current system of types of dialogue was conceived in 1988, as a part of the material to be
published in Walton and Krabbe (1995), which book should be the primary reference where these
types of dialogue are concerned. Anticipating the (much belated) publication of this book, a
number of articles and books saw the light containing a description of the current system, or of a
part, or a variant of it (Krabbe, 1991; Walton, 1989, 1990, 1992a, 1992b).
3 In Walton and Krabbe (1995, Ch. 3) the distinction between activity types and normative
systems that stipulate ideal practices was not clearly drawn, though the distinction between the
descriptive and formal (normative) approach to dialogues was pointed out (p. 82). The authors
discuss types of dialogues mainly as activity types, but consider the most specific types, where
the rules of dialogue are actually formulated, as normative (pp. 66-67). These are called ‘dialectical
systems’ and they are elaborated in Ch. 4.
4 Elsewhere Walton recognizes a link between examination dialogue and inquiry (1998, p. 147).
5 References to Aristotle use the abbreviations SE for Sophistici Elenchi (Sophistical Refutations),
Top. for Topics and Rhet. for Rhetoric, followed by book and/or chapter number and/or Bekker
numbers. Quotes are taken from the Loeb editions listed among the references.
6 Instead of ‘must take things on trust,’ we would, following Pieter Sjoerd Hasper, prefer the
translation: ‘must rely on principles’ (Hasper and Krabbe, 2007).
7 Instead of ‘generally accepted,’ we would, following Pieter Sjoerd Hasper, prefer the translation:
‘acceptable’ (Hasper and Krabbe, 2007).



About Old and New Dialectic     55

8 Instead of ‘Examination-arguments are those which are based on opinions held by the answerer
and necessarily known to one who claims knowledge of the subject involved,’ we shall, following
the translation by Hasper and Krabbe (2007), propose another translation at the end of this
section, where we try to understand peirastic discussions.
9 Walton takes the demonstrative arguments to constitute a fifth kind (1998, pp.11-12), whereas
we identify this kind of argument with the first (didactic) kind because the latter arguments are
characterized as reasoning ‘from the principles appropriate to each branch of learning’ (SE 2,
165b1-2), which characterization accords nicely with the idea of a demonstration (proof).
10 In fact, the phrase can be translated as ‘deduce the contradictory of a thesis’ (Hasper and
Krabbe, 2007).
11 As said in Note 7, we prefer to translate: ‘acceptable’.
12 See the beginning of the Topics (Top. I.1 100a18-24).
13 Instead of ‘changing the course of any argument which they appear to us to be using wrongly,’
we would, following Pieter Sjoerd Hasper, prefer the translation: ‘changing whatever they appear
to us to state incorrectly’.
14 Pieter Sjoerd Hasper pointed out the connection between these two passages. The link was
missed in Krabbe (2000, 2002).
15 This way of translating the passage we owe to Pieter Sjoerd Hasper.
16 According to Terence Irwin’s analysis of Aristotle’s thought (1988) a strong kind of dialectic can
lead to first principles; Robin Smith questions this idea, and gives a different translation for the
last quoted passage above (Top. I.2, 101a36-b4; Smith 1993, pp. 351-352); Robert Bolton defends
the point of view that dialectic may establish something, but not in sense of giving a scientific
justification (Bolton, 1994, p. 101).
17 This is the way we understand Walton’s rather sketchy third and fourth step (1998, pp. 250-
251).
18 We owe the survey that follows to Pieter Sjoerd Hasper.
19 The difference, which was explained in Section 3.3, is of no concern here.
20 We use the term ‘fallacy’ to cover all eristic, sophistical, or paralogistic reasoning, including
those cases where the fault is in the premises rather than in the step leading from premises to
conclusion. In this we concur with the traditional practice to have such a fallacy as ‘begging the
question’ and with current practice to have fallacies of ‘problematic premise’ (Johnson and Blair,
1977). It does not follow that each case of a false premise would constitute a fallacy.
21 According to the new dialectic, this mistake would be classified as a blunder rather than as a
fallacy. Another example of a blunder is the case where someone gives a long argument, when a
short argument is available (Top. VIII.11, 162a25-27).
22 See Nuchelmans (1993) for various kinds of ad hominem and their roots in Aristotle.
23 Walton uses these terms as names of argumentation schemes, not exclusively as names for the
fallacious instances of them.
24 For the latter case, see the end of Section 4.2.
25 That is: parry the refutation put forward by the Questioner.
26 Cf. SE 17, 176a22: ‘in competitive argument’.
27 This is one of the roots of the argumentum ad hominem (cf. Nuchelmans, 1993). For correct
(kalôs) solutions, see also SE 20 177b31-34.



56     Erik C. W. Krabbe & Jan Albert van Laar

References

Aristotle (1965). On Sophistical Refutations. In: Aristotle, On Sophistical Refutations,
On Coming-To-Be and Passing-Away, On the Cosmos. Cambridge, MA: Harvard
University Press, and London: William Heinemann. First ed. 1955 (The Loeb Classical
Library 400/442). Translated by E. S. Forster.

Aristotle (1976). Topics. In: Aristotle, Posterior Analytics, Topica. Cambridge, MA:
Harvard University Press, and London: William Heinemann. First ed. 1960. (The
Loeb Classical Library 391). Translated by E. S. Forster.

Aristotle (1991). Rhetoric. In: Aristotle, The “Art” of Rhetoric. Cambridge, MA, and
London: Harvard University Press. First ed. 1926. (The Loeb Classical Library 193).
Translated by John Henry Freese.

Bolton, Robert (1994). The Problem of Dialectical Reasoning (Συλλογισµοξ ) in Aristotle.
Ancient Philosophy 14, 99-132.

Dorion, Louis-André (1995). Aristote: Les réfutations sophistiques: Introduction,
traduction et commentaire. Librairie philosophique J. Vrin & Presses de l’université
Laval.

Fait, Paolo (2002). Aristóteles y los límites de la dialéctica: Notas sobre el arte de la crítica
(peirastikê). Anuario Filosófico 35, 431-458.

Hasper, Pieter Sjoerd, & Erik C. W. Krabbe (2007). Aristoteles: Over drogredenen:
Sofistische Weerleggingen [Aristotle: On Fallacies: Sophistical Refutations]. To be
published (Groningen: Historische Uitgeverij).

Irwin, Terence H. (1988). Aristotle’s First Principles. Oxford: Clarendon Press.

Johnson, Ralph H. (2000). Manifest Rationality: A Pragmatic Theory of Argument.
Mahwah, NJ; London: Lawrence Erlbaum.

Johnson, Ralph H., & J. Anthony Blair (1977). Logical Self-Defense. Toronto: McGraw-
Hill Ryerson.

Krabbe, Erik C. W. (1991). Subtypes of Persuasion Dialogue and Dialectical Shifts. In:
Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair & Charles A. Willard
(eds.), Proceedings of the Second International Conference on Argumentation:
Organized by the International Society for the Study of Argumentation (ISSA) at
the University of Amsterdam: June 19-22, 1990, Amsterdam: SICSAT, 371-376.

Krabbe, Erik C. W. (2000). Meeting in the House of Callias: Rhetoric and Dialectic.
Argumentation 14, 205-217.

Krabbe, Erik C. W. (2002). Meeting in the House of Callias: An Historical Perspective on
Rhetoric and Dialectic. In: Frans H. van Eemeren & Peter Houtlosser (eds.), Dialectic
and Rhetoric: The Warp and Woof of Argumentation Analysis. Dordrecht, etc.:
Kluwer, pp. 29-40.

Mackenzie, Jim (1981). The dialectics of logic. Logique et Analyse 24, 159-177.

Moraux, Paul (1968). La joute dialectique d’après le huitième livre des Topiques. In:
Gwilym E. L. Owen (ed.), Aristotle on Dialectic: The Topics: Proceedings of the
Third Symposium Aristotelicum. Oxford: Clarendon Press, pp. 277-311.

Nuchelmans, Gabriël (1993). On the Fourfold Root of the Argumentum ad Hominem. In:
Erik C. W. Krabbe, Renée José Dalitz & Pier A. Smit (eds.), Empirical Logic and
Public Debate: Essays in Honour of Else M. Barth. Amsterdam & Atlanta, GA:
Rodopi, pp. 37-47.



About Old and New Dialectic     57

Perelman, Chaïm, & Lucie Olbrechts-Tyteca (1958). Traité de l’argumentation: La
nouvelle rhétorique (2 vols.). Paris: Presses Universitaires de France.

Smith, Robin (1993). Aristotle on the Uses of Dialectic. Synthese 96, 335-358.

Van Eemeren, Frans H, Bart Garssen, & Bert Meuffels (2003). The conventional validity
of the pragma-dialectical freedom rule. In: Frans H. van Eemeren, J. Anthony Blair,
Charles A. Willard & A. Franscisca Snoeck Henkemans (eds.), Proceedings of the
Fifth International Conference of the International Society for the Study of
Argumentation, Amsterdam: SICSAT, pp. 275- 280.

Van Eemeren, Frans H., & Rob Grootendorst (2004). A Systematic Theory of
Argumentation: The Pragma-Dialectic Approach. Cambridge: Cambridge University
Press.

Van Eemeren, Frans H., & Peter Houtlosser (1999). Delivering the Goods in Critical
Discussion. In: Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair & Charles
A. Willard (eds.), Proceedings of the Fourth International Conference of the
International Society for the Study of Argumentation, Amsterdam: SICSAT, pp.
168-167.

Van Eemeren, Frans H., & Peter Houtlosser (2002). Strategic Manoeuvring in
Argumentative Discourse: A Delicate Balance. In: Frans H. van Eemeren and Peter
Houtlosser, Dialectic and Rhetoric: The Warp and Woof of Argumentation Analysis,
Dordrecht: Kluwer Academic, pp. 131-159.

Van Eemeren, Frans H., & Peter Houtlosser (2003). More about Fallacies as Derailments
of Strategic Maneuvering: The Case of Tu Quoque. In: Hans V. Hansen, Christopher
W. Tindale, J. Anthony Blair, Ralph H. Johnson and Robert C. Pinto (eds.),
Argumentation and its Applications. (Proceedings of the conference organized by
the Ontario Society for the Study of Argumentation in May 2001, CD-ROM.) Windsor,
CA: Ontario Society for the Study of Argumentation.

Van Eemeren, Frans H., & Peter Houtlosser (2005). Theoretical construction and
argumentative reality: An analytical model of critical discussion and conventionalised
types of argumentative activity. In: David Hitchcock (ed.), The uses of argument:
Proceedings of a conference at McMaster University, Hamilton: Ontario Society for
the Study of Argumentation, pp. 75-84.

Van Laar, Jan Albert (2003). The Dialectic of Ambiguity: A Contribution to the Study of
Argumentation. Unpublished dissertation, Groningen University. Retrievable from
http://irs.ub.rug.nl/ppn/249337959.

Walton, Douglas N. (1989). Informal Logic: A Handbook for Critical Argumentation.
Cambridge: Cambridge University Press.

Walton, Douglas N. (1990). What Is Reasoning? What Is an Argument? The Journal of
Philosophy 8, 399-419.

Walton, Douglas N. (1992a). Plausible Argument in Everyday Conversation. Albany,
NY: State University of New York Press.

Walton, Douglas N. (1992b). Types of Dialogue, Dialectical Shifts, and Fallacies. In:
Frans H. van Eemeren, Rob Grootendorst, J. Anthony Blair & Charles A. Willard
(eds.), Argumentation Illuminated. Amsterdam: Sic Sat, pp. 133-47.

Walton, Douglas N. (1995). A Pragmatic Theory of Fallacy. Tuscaloosa, AL, and London:
University of Alabama Press.



58     Erik C. W. Krabbe & Jan Albert van Laar

Walton, Douglas N. (1998). The New Dialectic: Conversational Contexts of Argument.
Toronto, etc.: University of Toronto Press.

Walton, Douglas N. (2006). Examination Dialogue: An Argumentation Framework for
Critically Questioning an Expert Opinion. Journal of Pragmatics 38, 745-777.

Walton, Douglas N., & Erik C. W. Krabbe (1995). Commitment in Dialogue. Basic
Concepts of Interpersonal reasoning. Albany, NY: State University of New York
Press.

Erik C. W. Krabbe
Jan Albert van Laar

Department of Theoretical Philosophy
Faculty of Philosophy

University of Groningen
Oude Boteringestraat 52

9712 GL Groningen
The Netherlands

E.C.W.Krabbe@.rug.nl
J.A.van.Laar@.rug.nl