Deductivism Within Pragma-Dialectics*

LEO GROARKE

Department of Philosophy
Wilfrid Laurier University
Waterloo, Ontario N2L 3C5
Canada

ABSTRACT: The present paper elaborates a deductivist account of natural language argu-
ment in the context of pragma-dialectics. It reviews earlier debates, criticizes some standard
misconceptions in the literature, and argues that the identification and analysis of deductive
argument schemes can be the basis of a compelling theory of argumentative discourse.

KEY WORDS: Argument schemes, deductivism, Govier, inductive inference, pragma-

dialectics

Deductivism might roughly be described as the view that all arguments

should be understood as attempts at deductive arguments. In contempo-
rary discussion, debates about deductivism are rooted in Govier’s (1987)
rejection of the deductivism she associates with Lambert and Ulrich (1980),
Nosich (1982) and Thomas (1981). According to her account, any theorist
who embraces deductivism is caught on the horns of an untenable dilemma:
either he fails to account for many ordinary arguments which are not deduc-
tive or he artificially reconstructs such arguments by adding implicit (‘unex-
pressed’) premises which arbitrarily turn them into deductive arguments.
In answer to Govier, Groarke (1992 and 1995) and Gerritsen (1994) argue
that the second horn of the dilemma is tenable, and that the unexpressed
premises with deductivism adds to ordinary arguments provide a plausible
basis for argument reconstruction.
In the present paper I defend deductivism. Because it is often misun-
derstood, I begin with a discussion of widespread misconceptions which
surround the notion of deductive validity it assumes. Among other things,
I stress that deductive validity extends beyond the relatively narrow con-
ception of (formal) validity that characterizes deductive formal systems. I
go on to argue that deductivism is a plausible theory of natural language
argument which has significant theoretical advantages.
In arguing in favour of deductivism, I emphasize that it needs to be
understood within the context of a more general account of argument.
Considered from this point of view, deductivism is not a comprehensive
theory of argument, but a view of premise/conclusion relations which needs

* An earlier version of this paper was presented at the third ISSA Conference on Argu-
mentation. I am indebted to this paper’s referees for comments and criticisms.

Argumentation 13: 1–16, 1999.

 1999 Kluwer Academic Publishers. Printed in the Netherlands.
2 LEO GROARKE

to be embedded in a broader theory of the nature and purpose of critical

discussion. For a number of reasons I will elaborate as I proceed, a pragma-
dialectical outlook serves this purpose very well. Most significantly, it
already contains a deductivist account of argument reconstruction which
can fruitfully be extended to explicitly answer the challenge to deductivism
posed by common accounts of ‘inductive’ arguments.
Looked at from the point of view of pragma-dialectics (or any other
general theory of argument) we might put the question this paper poses as
the question whether such a theory should incorporate a commitment to
deductivism or ‘inductivism.’ By deductivism I mean the view that natural
language arguments should be understood as attempts to formulate deduc-
tive arguments. By ‘inductivism’ I mean the more common view that a
theory of argument should distinguish between deductive and non-deduc-
tive ‘inductive’ arguments (I leave the question whether there are argu-
ments which are neither deductive nor inductive for elsewhere). In the
present paper I argue against the standard presumption in favour of induc-
tivism (and the inductive/deductive distinction) by arguing that a deduc-
tivist approach can account for inductive arguments, and that it can do so
in a way characterized by significant theoretical advantages.

SOME STANDARD MISCONCEPTIONS

Because deductivism is defined in terms of deductive validity, we need to

begin with it. At the outset, we might usefully distinguish between deduc-
tive validity and ‘formal validity.’ The latter encompasses deductive validity
as it is formalized in propositional and predicate calculus, and in other
common formal systems. These systems represent an important attempt to
develop a rigorous account of deductive validity but it is a mistake to think
that they define validity itself. Put very simply, an argument is deductively
valid if (and only if ) it is impossible for the premises to be true and the
conclusion false. There are many cases when this condition is fulfilled even
though there is no system of formal logic which can demonstrate that this
is so. As Woods (1995) points out, it is probably impossible to formulate
a formal theory of validity for natural language arguments, for tremen-
dous difficulties arise when one attempts to ‘translate’ ordinary language
arguments – ‘arguments on the hoof ’ – into their ‘dressed’ reconstruc-
tions in a formal language. Woods concludes that we must restrict our
interest in validity to formal validity, but one might as easily conclude
that we should develop a natural language account of validity which is not
so closely tied to formal systems. The important point is that we must not
confuse deductive validity with formal validity. In the present context, it
needs to be emphasized that it is the broader conception of validity which
is the heart of the deductivism elaborated here.
What then should be said about this broad conception of validity?
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 3

And what is the standard understanding of the difference between it and

premise/conclusion relations in ‘inductive’ arguments? Reading most text-
books one might easily conclude that the difference between deductive
and inductive arguments is a distinction between arguments which have
certain and uncertain conclusions. Consider, for example, the following
accounts of the inductive/deductive distinction, from four logic/critical
thinking texts picked randomly from my bookshelf:
1. Out on the street, people are not so careful [when they distinguish
deductive and inductive inferences] . . . Charles Darwin . . . is famous
for his ‘deduction that circular atolls in the oceans are actually coral
growths on the tops of barely submerged volcanoes, but to be precise
he really performed an induction, not a deduction. He could not be sure’
(Dowden, 1993, p. 228).
2. . . . [In contrast to deductive conclusions,] inductive conclusions are
always only probable, not certain (Hinderer, 1992, p. 52).
3. . . . the conclusion of an inductive argument [in contrast to the con-
clusion of a deductive argument] . . . must venture beyond information
contained in the premises. Thus our conclusion can never be certain
. . . (Engel, 1994, p. 38).
4. Deductive reasoning draws out the implications of knowledge we
already possess, whereas inductive reasoning extends our knowledge
beyond the information contained in the premises (Kelley, 1988, p. 167).
According to these accounts, the question whether an argument is induc-
tive or deductive implies the question whether its conclusion is (assuming
its premises are true) certain or only probable.
To see what is wrong with these accounts we need note that a deduc-
tive argument is necessary only in the sense that its premises necessarily
entail its conclusion. This means only that it is impossible for the conclu-
sion to be false if these premises are true. The accounts that we have noted
confuse this notion of necessity with the notion that the conclusion of a
deductive argument is necessarily true if the premises are true.
If we let P represent the premises of an argument and C its conclusion,
and if we use box (u ) to represent necessity and turnstile (£) to represent
entailment, then we can say that deductive inferences are necessary in the
sense that u (P £ C). The accounts of deductive inferences we have noted
confuse this principle with the principle that P £ u C, i.e. the principle
that the premises of a deductive argument entail a necessary conclusion.
This is a simple logical error.
Especially in the context of the uncertain kinds of issues that are
the subject of natural language debate, it is important to remember that
u (P £ C) implies only that u P £ u C. More generally, it implies that the
conclusion of a deductive argument must be as certain as its premises. A
deductive argument should therefore be described as ‘certainty preserving’
rather than ‘certainty establishing.’ In natural language arguments, this
means that the conclusion of a good deductive argument is more often
4 LEO GROARKE

probable or plausible than necessary, for the premises of such arguments

are rarely certain.
It may be useful to illustrate the kind of uncertainty that tends to char-
acterize ordinary language deductive arguments with the following two
examples.
Example 1. Canada’s population of 27 million will increase by 10 million in the next
10 years, so it will reach 37 million.
Example 2. European countries that see globalization as an opportunity rather than a
threat will prosper. The Netherlands sees globalization as an opportunity
rather than a threat, so it will prosper.

It should be obvious that a person who asserts the premises of either of

these arguments need not be committed to the claim that they are certain.
As a speech act, an assertion of this sort commits one only to the claim
that some statement is probably or plausibly true. In the case of examples
1 and 2, a reasonable person will recognize that predictions about what
happens in the future are inherently uncertain but may still be justified
and useful. It follows that the proposed deductive inferences commit a
speaker only to the claim that the proposed conclusions are, like the
proposed premises, probable or plausible.
We could easily multiply examples to illustrate the same point. In lieu
of a long list it is worth noting that one finds many good examples in
Aristotle’s Nichomachean Ethics, which begins with a recognition that any
account of virtue and morality is necessarily imprecise. As Aristotle puts
it, we must be content, in reasoning about such things, with premises that
‘indicate the truth roughly and in outline’ and ‘are only for the most part
true,’ in the process establishing ‘conclusions that are no better’ (1094b,
19–24, cf. 1137b, 18–19). Deductive arguments are still the norm in the
Nichomachean Ethics, but its deductively valid inferences establish con-
clusions that are fundamentally uncertain because this is the inherent nature
of their premises. An illustrative example is the argument that Honour
‘seems [plainetai] too superficial to be what we are looking for [the good],
since it is thought to depend on those who bestow honour rather than on
him who receives, and the good is, in contrast, something proper to a man
and not easily taken from him’ (1095b 25–26).
The misconception that deductive arguments have certain conclusions
appears to be rooted in deductive validity’s contemporary association with
deductive systems which are used to prove theorems in logic and mathe-
matics. When Govier (1987) claims that deductivism is not credible, for
example, she suggests that it maintains that ‘All good arguments are as
tight and firm as proofs in mathematics. They are comparable to “All men
or mortal; Pierre Trudeau is a man; therefore Pierre Trudeau is mortal.”,
in that true premises are stated and provide compelling, conclusive support
for a conclusion’ (22). If this were true, then we would have to conclude
that deductivism is seriously limited, for ‘Arguments in law, ethics, admin-
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 5

istration, empirical science, literary criticism, and ordinary life do not look
like the watertight proofs of mathematics’ (25).
But Govier’s views confuse deductive validity and the narrower con-
ception of formal validity which is reflected in formal logic and mathe-
matics (though she herself sometimes distinguishes them). In the context
of natural language arguments, deductivism rarely implies that ordinary
arguments and conclusions are as certain as mathematical arguments, for
the premises of ordinary arguments are rarely certain in the way that the
premises of mathematical arguments are. Because these premises may be
certain, uncertain, plausible, implausible, questionable or unquestionable,
the same must be said of the conclusions they deductively support. It
follows that deductivism allows for a whole range of modalities in the con-
clusions of ordinary arguments.
A related misconception about deductivism is the notion that it makes
the merit of an argument – like the merit of a mathematical or logic proof
– ‘an all-or-nothing affair’ (Govier, 1987, p. 25). Such claims would be
true only if deductivism was the view that the only standards we should
used in assessing argumentative discourse are criteria which determine
whether arguments are valid or invalid. But deductivism as it has been
proposed recognizes that the domain of premise/conclusion relations is only
one ingredient of good argument, and that it is an ingredient which needs
to be situated in a more comprehensive account of argument which includes
an account of differences of opinion, standpoints, implicit and indirect
argument components, and so on. The simple fact that a valid argument
must have plausible or probable premises to count as a good argument itself
shows that deductivism makes the merit of an argument a more complex
issue than a judgment of validity. Because deductivism as I have outlined
it also recognizes that good arguments must abide by rules for critical dis-
cussion, it countenances a variety of other ways in which the criteria for
good argument transcend questions of validity. It follows that deductivism
admits that arguments and conclusions can have a range of values when
considered from the point of view of merit, and that judging arguments
means something more than judging validity and invalidity.

ARGUMENT RECONSTRUCTION

The common misconceptions about deductivism show, at the very least,

that it cannot be dismissed as easily as most argumentation theorists have
supposed. In arguing for the further point that it has something to offer
as a theory of natural language argument, it is useful to distinguish two
aspects of argument analysis which it – like any other account of argument
– must address. The first is argument reconstruction, the second argument
evaluation. The former identifies arguments and their components and
prepares the way for argument evaluation.
6 LEO GROARKE

In many cases, a natural language argument is obviously deductive and

it is a simple matter to recognize its premises and conclusion. In cases
where arguments are not transparently deductive, the deductivist proceeds
by recognizing unexpressed premises that make them attempts at deduc-
tive inferences. This may initially sound arbitrary, but it is quite in keeping
with everyday practice in dealing with obvious enthymemes like the fol-
lowing:
Example 3: Jones is a politician, so he is not to be trusted.

Faced with such an argument, we have no difficulty recognizing that it

contains the unexpressed premise ‘No politicians can be trusted.’
In the context of pragma-dialectics, the addition of unexpressed premises
of this sort is recognized in its account of ‘indirect’ speech acts. As Van
Eemeren and Grootendorst put it, ‘Taken literally, in an argumentation in
which a premise has been left unexpressed, the argument concerned is
invalid. If it is analyzed as conveying an indirect speech act, however, the
missing premise can be added to the argument so that the invalidity is cor-
rected’ (1992, p. 61)
In assigning unexpressed premises, we can distinguish different possi-
bilities. A ‘logical minimum’ is the minimum claim necessary to ensure a
valid inference. In some cases, it is the most plausible unexpressed premise,
but there there are many cases in which context or common practice clearly
suggest that an arguer is committed to a stronger claim which is, in pragma-
dialectical terminology, the ‘pragmatic optimum.’ In example 3 above, the
logical minimum is the claim that ‘If Jones is a politician, then he cannot
be trusted’ – a claim which is not equivalent to the claim that ‘No politi-
cians can be trusted’ (the first but not the second is, for example, true if
Jones has always denied that he is a politician, and this is the only reason
why he cannot be trusted if he is). In the absence of some explicit indica-
tion that this idiosyncratic assumption is the basis of the proposed con-
clusion, it is reasonable to assume that it is the latter generalization about
politicians which drives the inference. It can, therefore, be designated as
the pragmatic optimum.
We can see that it is always possible to deductively reconstruct an
argument which is not transparently deductive by noting that any arguer is
committed to the statement that ‘If the premises of my argument are true,
then the conclusion is true.’ This follows directly from the implications of
the speech acts ‘argument’ and ‘assertion,’ for an arguer who argues for
some conclusion C on the basis of some set of premises purports to believe
both that C is true and that her proposed premises justify this belief (cf.
Van Eemeren and Grootendorst, 1992, pp. 30–31). In this sense, their
argument declares that they believe that these premises imply the conclu-
sion, and that the conclusion is true if the premises are true. It is perhaps
worth nothing that they are committed to the latter conditional not merely
in the sense of material implication, but in the stronger sense that they must
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 7

believe that there is a relationship between their premises and their con-
clusion which makes it reasonable to base a belief in the latter on a belief
in the former.
To a great extent, this account of deductivist argument reconstruction
merely recapitulates and defends one aspect of pragma-dialectics. This does
not mean that it is not news, for this aspect of pragma-dialectics has been
for the most part ignored by other theorists and is fundamentally at odds
with most English speaking discussions of natural language arguments. To
illustrate this point, I want to show that deductivist argument reconstruc-
tion it implies can account for arguments which are usually treated as irre-
ducibly inductive. I will do so with a series of examples which Conway
and Munson (1997, pp. 40–47) have recently used to introduce the notion
of non-deductive arguments. Space constraints do not allow me to discuss
all their examples, but the deductivist analysis of the following six can
easily be extended to the examples I do not discuss.
Example 4. Ninety-six percent of adult Americans watch television more than ten hours
per week. Davis is an adult American. Therefore Davis watches television
more than ten hours a week.
Example 5. Fifty-five percent of adult American watch cable television. Farrell is an
adult American. Therefore Farrell watches cable television.
Example 6. Every wolverine so far encountered by humans has been unfriendly and
aggressive. Therefore, all wolverines are unfriendly and aggressive.
Example 7. Lawrence was widely praised for her acting in two films last year. Both of
these films were huge box-office hits. Lawrence has never won a major award
for her work. Lawrence will win this years’s ‘Hollywood Woman of the Year
Award.’
Example 8. ‘Howl’ is superficial and dated. ‘The Second Coming’ has profound social
significance. Therefore ‘The Second Coming’ is a finer poem than ‘Howl.’
Example 9. Congressman Smith would be an excellent senator because he was born on
independence Day.

The interpretation of unexpressed premises is not an exact science, espe-

cially when the context of an argument is (as it is in examples 4–9) unclear
but we can plausibly reconstruct Conway and Munson’s arguments by
assigning them the unexpressed assumptions I have placed in italics in the
following reconstructions.
Example 4. Ninety-six percent of adult Americans watch television more than ten hours
per week. Davis is an adult American. Davis is among this ninety-six percent.
Therefore Davis watches television more than ten hours a week.
Example 5. Fifty-five percent of adult Americans watch cable television. Farrell is an
adult American. Farrell is among this fifty-five percent. Therefore Farrell
watches cable television.
Example 6. Every wolverine so far encountered by humans has been unfriendly and
aggressive. All wolverines are like the ones so far encountered by humans.
Therefore, all wolverines are unfriendly and aggressive.
Example 7. Lawrence was widely praised for her acting in two films last year. Both of
these films were huge box-office hits. Lawrence has never won a major award
for her work. If one can say this of a woman actor, then she will win this
8 LEO GROARKE

year’s ‘Hollywood Woman of the Year Award.’ Lawrence will win this year’s
‘Hollywood Woman of the Year Award.’
Example 8. ‘Howl’ is superficial and dated. ‘The Second Coming’ has profound social
significance. A poem which has profound social significance is a finer poem
than one which is superficial and dated. Therefore ‘The Second Coming’ is
a finer poem than ‘Howl.’
Example 9. Congressman Smith would be an excellent senator because he was born on
Independence Day and someone born on Independence day would be an
excellent Senator.

So understood, all of these arguments are deductively valid, for it is in each

case impossible for the (expressed and unexpressed) premises to be true
and the conclusion false.
One might contrast the proposed reconstruction of example 8 with
Conway and Munson’t account, for they suggest that it is inductive because
it is possible (though unlikely) that a poem which is superficial has tech-
nical merits which make it superior to one which has profound social sig-
nificance. The problem is that this arbitrarily reads an unstated reservation
into the argument which its proponent may not believe (if they are, for
example, committed to the view that social significance is the most impor-
tant aspect of a poem). Someone who is committed to this reservation is
therefore obligated to express it clearly. This is easily accomplished by
restating the conclusion as the claim that ‘The Second Coming’ is probably
a finer poem than ‘Howl.’ But this poses no difficulties for deductivist
reconstruction, which will in such circumstances recognizes the unex-
pressed premise as the statement that ‘A poem which has profound social
significance is probably a finer poem than one which is superficial and
dated.’
Our reconstruction of Conway and Munson’s examples show how deduc-
tivist reconstruction handles many arguments which are taken to be irre-
ducibly inductive. Because its principles can be applied to all of their
examples of non-deductive arguments, it poses the question why we should
introduce a new category of argument to explain them. This is a point to
which we will return. But first we must consider what deductivism entails
for argument evaluation.

ARGUMENT EVALUATION

It is one thing to show that examples like 4–9 can be understood as deduc-
tive arguments. It is another thing to show that this is a useful thing to
do. We can better assess the latter by first outlining deductivist argument
evaluation. We might begin by noting that the utility of deductivist
argument reconstruction is highlighted by approaches to argument which
are, like pragma-dialectics, dialectical and ‘resolution oriented.’ Such views
underscore the point that we should develop theories of argument which
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 9

can help identify the issues that need to be addressed in dialectical

exchange. This is a goal which is well served by deductivist reconstruc-
tion, for the unexpressed premises it identifies often expose assumptions
which need to be a focus of discussion when we decide whether an
argument should be accepted.
In the case of arguments 4–9, for example, argumentative exchange can
be usefully furthered by asking whether the assumptions embedded in the
unexpressed premises are acceptable, true or plausible. In the case of
example 6 – which concludes that all wolverines are unfriendly and aggres-
sive – the question arguers should ask is whether it is or is not plausible
to suppose that all wolverines are like the ones so far encounter by humans.
In the case of argument 7, we need to ask whether it is true that a woman
widely praised for her acting in two box-office hits, who has never won a
major award for her work, can count on winning the ‘Hollywood Woman
of the Year Award.’ By recognizing these assumptions as unexpressed pre-
mises, a deductivist approach furthers the dialectical exchange which is the
key to resolving differences of opinion. Ironically perhaps, the usefulness
of the deductivist strategy is tacitly recognized by Conway and Munson
(45–46), who suggest that we try to evaluate ‘non-deductive’ arguments by
looking ‘for additional premises that explicitly link the original premises
to the conclusion’ and by considering ‘whether these premises are accept-
able’ (their emphasis). But this is precisely what deductivism prescribes
when it is faced with arguments which are not transparently deductive.
Such considerations illustrate one way in which deductivism recognizes
that argument evaluation extends beyond assessments of validity, for a good
deductive argument must begin with an accepted starting point and this
means that its expressed and unexpressed premises must be acceptable. In
the course of evaluating particular arguments, this will mean that deduc-
tivism requires that we consider the acceptability of a proposed definition,
an empirical claim, an eye witness report, a principle of logic, a moral
maxim, and so on.
In evaluating examples 4 and 5, this process means that we will have
to assess the likelihood that Davis is among the 96% of Americans who
watch television, and that Farrell is among the 55% who watch cable tele-
vision. Assuming that the other premises in the arguments are true, the
evidence they provide suggests that the probability of the first claim is 0.96
and the probability of the second is 0.55. Though neither argument allows
for a certain conclusion, this allows us to conclude that argument 4 is the
stronger argument.
It is important to recognize the emphasis that deductivist argument eval-
uation of this sort places on the assessment of premises and especially unex-
pressed premises, but it does not mean that it leaves no room for other
aspects of argument evaluation. It is especially significant that deductivism
within pragma-dialects recognizes that an argument can be a poor argument
because it fails to abide by rules of critical discussion, or because it is an
10 LEO GROARKE

instance of an invalid form of inference. In such cases, we could proceed

by recognizing these failings as faulty unexpressed premises which reflect
an arguer’s assumptions about the nature of good argument, but it is more
intuitive to say that these are cases where arguers attempt to formulate good
deductive arguments but fail. That reasoners should sometimes fail is not
surprising, especially as reasoners often make mistakes in the most trans-
parent case of deductive inference – in adding two numbers together. In
the much more complex reasonings required in many examples of critical
discussion we should expect that errors will be made.
In keeping with pragma-dialectical rules of critical discussion, it follows
that deductivism can recognize traditional fallacies and the problems that
they cause (see Van Eemeren and Grootendorst, 1996a). The deductivist
can, for example, still see ad hominem as a failure to respect an opponent’s
right to advance standpoints and cast doubt on standpoints (Van Eemeren
and Grootendorst, 1996b). This is yet another way in which an argument
can be valid and still a poor argument.
We might illustrate deductivism’s relationship to traditional fallacies with
the following argument from the Detroit Free Press.
Example 10. Airlines are funny. They make sure you aren’t carrying a weapon of destruc-
tion and then sell you all the booze you can drink.

We can plausibly standardize this argument as:

Premise 1: Airlines make sure you are not carrying a weapon of destruction.
Premise 2: They sell you all the booze you can drink.
(Unexpressed) Premise 3: Booze is a weapon of destruction.
Conclusion: Airlines are inconsistent (‘funny’).

It is easy to see why one might take this to be a deductively valid argument,
for the argument appears to have the form ‘Airlines make sure you don’t
have X, then give you X, so airlines are inconsistent.’ On closer inspec-
tion, it becomes clear that the argument only appears to have this form
because (most importantly) it conflates two different meanings subsumed
by the term ‘weapon of destruction.’ According to the first use, in premise
1, it implies a bomb or firearm. According to the second, in the unexpressed
premise, it implies long term health and social ills. Once we recognize the
ambiguity and the consequent equivocation – which implies a violation of
the tenth rule for critical discussion – we can see that we have an attempt
to formulate a good deductive argument which has gone awry.
Instances of formal fallacies like affirming the consequent and denying
the antecedent provide other obvious examples of cases where arguers fail
to understand the mechanics of good deductive arguments. This being said,
mistakes of this sort more frequently occur in more complex chains of
reasoning. Deductivism has in this regard suffered from its association with
formal logic, for argumentation theorists have assumed that a deductive
approach to ordinary argument only recognizes simple forms of argument
like modus ponens, hypothetical syllogism, and so on. As important as they
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 11

are, a deductivism created for natural language arguments needs to recog-

nize more complex argumentation schemes as instances of deductive
argument. Among other things, this means recognizing forms of valid
argument which are frequently not recognized in common formal systems.

DEDUCTIVE ARGUMENT SCHEMES

This is not the place to develop a comprehensive set of new deductive

argument schemes. An example may nonetheless illustrate what this devel-
opment might entail. So called ‘two wrongs’ reasoning attempts to justify
an action or policy which is normally considered wrong by appealing to
another wrong. In many cases, ‘Two wrongs don’t make a right,’ but it is
also clear that there are instances in which good two wrongs arguments
are possible. In their account of good argument, Groarke, Tindale and Fisher
(1997, pp. 263–268) therefore recognize a good two wrongs argument as
an argument which justifies some action or policy X by pointing out:
1. that X is a response to some unfairness or injustice (Y) that it attempts
to alleviate;
2. that X is less wrong than Y; and
3. that there is no morally preferable way to respond to Y.
In dealing with natural language arguments, it is important to recognize the
two wrongs scheme because such reasoning plays a central role in moral
debate, which frequently arises in circumstances in which we must balance
competing rights and interests, and choose to avoid a greater evil by com-
mitting a lesser one. Two wrongs reasoning is, for example, the standard
mode of argument invoked to justify self defense, military action, the
reverse discrimination associated with affirmative action, trade sanctions,
and so on.
For our purposes, the important point is that a good two wrongs argument
is a deductive argument. There is, for example, no way to avoid the con-
clusion that a policeman is justified in shooting a terrorist if (i) this is an
attempt to stop a terrorist from detonating a bomb that will kill a bus load
of innocent people; (ii) this shooting is less wrong (as it clearly is); and
(iii) there is no morally preferable way to handle the situation. In any case
in which the three conditions for good two wrongs reasoning are met (or
simply (i) and (iii), which imply (ii)), it follows that the lesser wrong is
justified.
Good two wrongs reasoning provides us with a paradigm case of a
deductive argument scheme which is not widely recognized (and even dis-
missed as fallacious), though it plays an important role in moral discourse.
Instances of bad two wrongs reasoning tend to be cases in which there are
problems with an arguer’s claim that the conditions for good two wrongs
reasoning are met, or cases in which an arguer fundamentally misunder-
stands what such reasoning requires.
12 LEO GROARKE

Consider the following example, taken from a letter to the editor

answering concerns about the use of night sticks by the Toronto city policy.
Example 11: Re: the Sept. 3 report that Police Chief Jack Ackroyd’s ‘ashamed and quite
appalled’ by the conduct of some his officers in handling a Scarborough house party at
which 300 to 500 people attended.
There seems to be one very important fact being ignored, that being the rights of the
neighbours who must have been subjected to one horrendous evening. Ordinary people
have a right to peace and quiet in their homes. They must have been terrified.
Let’s not forget there was a good reason for the police to be there.

The author of this letter does recognize the possibility of good two wrongs
reasoning, and that it must establish that some wrong is a response to
another wrong. His argument still fails, for he fails to to establish that the
wrong that he discusses – the police officers’ use of night sticks – was the
morally preferable way to assert neighbour’s rights to peace and quiet in
their homes. Put another way, he fails to recognize the condition required
of a valid two wrongs argument.
It is by recognizing deductive argument schemes like two wrongs rea-
soning that we can develop a deductivism which allows a better account
of natural language argument. Students who study argument can easily be
taught an argumentation scheme like two wrongs reasoning, in much the
way they can be taught less sophisticated argument forms like modus
ponens. In the process one can teach them how to construct good two
wrongs arguments, how to criticize bad ones, and an important aspect of
moral discourse. Regrettably, there has been no concerted effort to identify
argument schemes of this sort, though they point the way to a viable deduc-
tivist agenda which could improve our understanding of natural language
argument by adding them to the more basic forms of deductive inference
identified in formal deductive systems.
One set of argumentation schemes which merits particular comment in
the context of deductivism are schemes which are commonly thought to
be irreducibly inductive. They are the cornerstone of many attempts to teach
critical thinking skills and it behooves us to say something about the way
in which they may be handled from a deductivist point of view. A detailed
discussion of this aspect of deductivism is beyond the scope of the present
paper, but we can briefly sketch the issues by considering the argument
forms inductive generalization and argument from authority, which we
might succinctly define as follows:

Inductive Generalizations: These Xs are Y. Therefore all Xs

are Y.
Arguments From Authority: X says Y. X is an authority.
Therefore Y.

Like other forms of argument which are not transparently deductive (causal
arguments, arguments from ignorance, etc.), these forms can be deductively
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 13

reconstructed. In the process, these argument forms can be redefined by

adding the italicized unexpressed premise which I indicate below.
Inductive Generalizations: These Xs are Y. Other Ys are like
these Xs (with respect to Y). Therefore all Xs are Y.
Arguments From Authority: X says Y. X is an authority. The
authorities are right (about Y). Therefore Y.
So understood, these argumentation schemes are deductive schemes. The
unexpressed premises they make explicit are simply the assumptions that
such reasoning ordinarily depends on. In the case of inductive generaliza-
tions, for example, one cannot reasonably propose the conclusion that all
Xs are Y unless one believes that other Xs are like the Xs which have
been examined. Similarly, one cannot honestly forward an opinion on the
grounds of an appeal to authority unless one believes that the authorities
are right about the issue of debate.
As in the case of the individual arguments we considered earlier, this
way of proceeding provides a useful basis for resolving issues raised
in dialectical exchange, for it identifies key assumptions which must
frequently be addressed. In dealing with inductive generalizations, for
example, it invites us to ask whether other Xs are like the Xs that have
been examined. This naturally raises the question whether our sample is
too small (and our results due to chance), or in some way biased in favour
of a particular result. In the case of arguments from authority, our recon-
struction raises the issue whether we should accept the proposition that
the authorities are right in the case in question. Cases in which we may call
their opinion into question – when the authorities disagree, when they have
a vested interest, when they have overlooked some important evidence, etc.
– are just those cases in which appeals to authority are problematic.
One might push questions about the unexpressed premises we have iden-
tified much further and raise philosophical issues about inductive general-
izations and arguments from authority. Even when there is no reason to
believe that authorities are mistaken or a sample is biased, one might ask
how we can justify such inferences. In the present context, it is enough to
say that the so called principle of induction and a reliance on appropriate
authorities define good reasoning. Though the question how we can justify
the principles of reason this implies takes us beyond the scope of the present
paper, it is worth noting that one might (like Wittgenstein, 1969) compare
the role of the law of non-contradiction, the law of the excluded middle,
and other principles which normatively define what reasoning is. Here
again, it might be argued that ordinary arguments have more affinity to
paradigm instances of deductive arguments than is usually imagined.
14 LEO GROARKE

DEDUCTIVISM VS. INDUCTIVISM

Though the proposed account of deductivism is necessarily tentative, it

suggests that deductivism can account for ordinary reasoning, and that it
can do so in a way that furthers the aims of critical discussion. These con-
clusions return us to our initial question whether a general theory of
argument should adopt a deductivist or an inductivist perspective. Here
the important point is that deductivism is favoured by Occam’s razor, for
it replaces two conceptions of inference with one and can in the process
greatly simplify our account of natural language argument. Given that we
can treat all arguments as deductive, why follow the standard practice we
see in authors like Conway and Munson (1997) and introduce a distinct
conception of inductive argument?
It is true that a deductivist approach to natural language argument
requires that we place greater emphasis on unexpressed premises, but
theories of argument already recognize their existence and deductivist
reconstruction is easily built upon this basis. In the case of pragma-dialec-
tics, it is particularly notable that one has a well established account of
indirect speech acts which incorporates an easily applicable conception of
unexpressed premises.
In marked contrast, we have no well established account of the logic of
inductive inferences and we have already seen that accounts of them are
frequently characterized by a fundamental misunderstanding of the deduc-
tive/inductive distinction. Inductivism is especially problematic in ordinary
language contexts, for the distinction between deductive and inductive argu-
ments is, as Cederblom and Paulsen (1996, p. 228) and Moore and Parker
(1989, p. 213) point out, difficult to discern. Govier (1987) explains the
cause of this problem well:
People who argue do (at least implicitly) distinguish conclusions from premises and
‘claim’ that the latter provide reasons for the former. But they often do not, not even
implicitly, make claims about what sort of connection is supposed to hold between these
premises and their conclusion. A major reason for this is that most arguers have not
reflected on the difference between deductive entailment and making probable. . . . A
further problem is that even if ordinary arguers did wish to indicate whether a necessary
or probable connection existed, our language provides very few words which would con-
veniently serve this purpose. Indicators such as ‘therefore’ and ‘must’ are sometimes
urged by textbook authors to indicate deductive arguments, but are equally naturally found
in arguments that tradition would label as non-deductive (46).

Perhaps in view of the issues of interpretation which Govier raises, many

commentators make the distinction between deductive and inductive argu-
ments on the basis of difficult and arbitrary appeals to the intentions of an
arguer. Thus Conway and Munson (1997, p. 41) write that ‘The first general
characteristic of nondeductive arguments is that they are not meant to be
valid, but they are meant to make their conclusions probable or likely’ (their
emphasis) and Dowden (1993, p. 228) defines an inductive argument as
 DEDUCTIVISM WITHIN PRAGMA-DIALECTICS 15

an argument which is ‘intended to be inductively sound’ (my emphasis).

These definitions make it very difficult to establish whether an ordinary
argument is deductive or inductive for intentions are, as Van Eemeren and
Grootendorst (1992, p. 10) point out, internal states of mind which ‘are
not accessible, or at least not directly accessible, to outside inspection.’ It
is especially difficult to see how ordinary arguers who have no knowledge
of the theoretical distinction between deductive and inductive arguments
can specifically intend their arguments as inductive.
In this context we might consider what is right and wrong when Govier
(1987) tries to explicate the possible advantages of deductivism:
Deductivism has the advantage of being an extremely simple theory of argument. On
this theory, there is only one type of argument; therefore there is no need to sort or classify
arguments and to develop criteria for doing so. Problems of examples which do not fit
tidily into the requisite classificatory categories do not arise, because there is only one
category. The theory is monolithic. Also, there are clearly articulated standards for formal
deductive validity in systems of formal logic. Hence if dedcutivism is the correct theory
of argument, we have the assurance of the relevance of an established and recognized
body of very precise knowledge to be used in the appraisal of arguments. Furthermore,
any disputes about which ‘type’ of argument is strongest or most convincing will cease
to arise. There will be only one type of argument, and one standard, which all arguments
will have to satisfy (24).

We have already seen that deductivism does in fact offer the advan-
tages of a much simpler theory of argument. This being said, Govier greatly
underestimates its complexities. Instead of claiming that ‘there is only one
type of argument’ a mature deductivism will distinguish many different
kinds of deductive arguments. And though it will provide a basis for a more
unified account of formal and informal reasoning, its rules of validity and
good argument will extend far beyond the very limited catalogue of basic
inferences found in common formal systems. Unlike the latter, a deduc-
tivist approach to ordinary argument is most in keeping with a dialectical
account of argumentative discourse, for it identifies key assumptions as
unexpressed premises which need to be assessed.
Finally, it must be said that a deductivism embedded in a broader theory
like pragma-dialectics does not uniquely elaborate ‘one standard’ for good
argument (i.e. deductive validity) but makes room for the different stan-
dards implied by the rules of critical discussion, traditional fallacies, the
principles that govern premise acceptability, and so on.
A more complete account of a deductivist approach to natural language
argument is beyond the scope of the present paper. I have nonetheless
attempted to suggest some ways in which such development might take
deductivism beyond the role it plays in reconstructing unexpressed premises
in pragma-dialectics. A more comprehensive account of a deductivist theory
is needed before we can definitively decide for or against deductivism,
but it can still be said that (i) the standard understanding of deductivism
and deductive validity is founded on misconceptions about the latter; (ii)
that it is possible to deductively reconstruction arguments which most
16 LEO GROARKE

authors continue to see as irreducibly inductive; (iii) that deductivist

argument reconstruction can contribute to dialectial exchange; (iv) that
the development of hitherto unrecognized deductive argument schemes can
contribute to our understanding of ordinary argument; (v) that it is possible
to recognize schemes which are traditionally understood to be inductive
as deductive; and (vi) that inductivism complicates theories of argument
with unnecessary theoretical distinctions which are exacerbated by natural
language’s failure to distinguish deductive and non-deductive arguments.
Given that inductivism has, outside pragma-dialectics, become a dogma
which is almost always assumed without discussion or debate, these con-
clusions challenge us to rethink one fundamental feature of contemporary
approaches to argument.

REFERENCES

Cederblom, J. and D. W. Paulsen: 1996, Critical Reasoning: Understanding and Criticizing

Arguments and Theories, Wadsworth, New York.
Dowden, B. H.: 1993, Logical Reasoning, Wadsworth Publishing, Belmont.
van Eemeren, F. H. and R. Grootendorst: 1992, Argumentation, Communication and
Fallacies, Lawrence Erlbaum Associates, Hillsdale.
van Eemeren, F. H. and R. Grootendorst: 1996a, ‘The Pragma-dialectical Approach to
Fallacies’, in Hansen and Pinto, eds.
van Eemeren, F. H. and R. Grootendorst: 1996b, ‘Argumentum Ad Hominem: A Pragma-
dialectical Case in Point’, in Hansen and Pinto, eds.
Engel, S. M.: 1994, With Good Reason: An Introduction to Informal Fallacies, 5th ed., St.
Martin’s Press, New York.
Gerritsen, S.: 1994, ‘A Defence of Deductivism in Reconstructing Unexpressed Premisses’,
in F. H. van Eemeren and R. Grootendorst (eds.), Studies in Pragma-Dialectics,
International Centre for the Study of Argumentation, Amsterdam.
Govier, T.: 1987, Problems in Argument Analysis and Evaluation, Foris Publications,
Dordrecht-Holland/Providence RI.
Groarke, L. A.: 1995, ‘What Pragma-dialectics Can Learn from Deductivism, and What
Deductivism Can Learn from Pragma-dialects’, in F. H. van Eemeren and R. Grootendorst
(eds.), Analysis and Evaluation.
Groarke, L. A.: 1992, ‘In Defense of Deductivism: Replying to Goveri’, in F. H. van Eemeren,
R. Grootenddorst, J. A. Blair and C. A. Willard (eds.), Argumentation Illuminated,
International Society for Study of Argument, Amsterdam.
Groarke, L. A., C. W. Tindale and L. Fisher: 1997, Good Reasoning Matters!, Oxford
University Press, Toronto.
Hansen, H. V. and R. C. Pinto (eds.): 1995, Fallacies: Classical and Contemporary Readings,
Penn State Press, University Park, Pennsylvania.
Hinderer, D.: 1992, Building Arguments, Wadsworth Publishing, Belmont.
Kelley, D.: 1988, The Art of Reasoning, W. W. Norton, New York.
Lambert, K. and W. Ultrich: 1980, The Nature of Argument, MacMillan, New York.
Moore, B. N. and R. Parker: 1989, Critical Thinking: Evaluating Claims and Arguments in
Everyday Life, Mayfield Publishing, Mountain View.
Nosich, G. M.: 1982, Reasons and Arguments, Wadsworth, Belmont.
Thomas, S. N.: 1981, Practical Reasoning in Natural Language, Prentice-Hall, Englewood
Cliffs.
Woods, J.: 1995, ‘Fearful Symmetry’, in Hansen and Pinto, eds.