See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/226134293

Reasoning from Classifications and Definitions

Article in Argumentation · January 2009

DOI: 10.1007/s10503-008-9110-2

CITATIONS READS

34 1,593

2 authors:

Douglas Walton Fabrizio Macagno

University of Windsor Universidade NOVA de Lisboa
493 PUBLICATIONS 15,272 CITATIONS 159 PUBLICATIONS 3,221 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CALL FOR PAPERS: SI on Argumentation schemes (Argument & Computation) View project

Special Issue on Boundaries between dialogic pedagogy and argumentation theory View project

All content following this page was uploaded by Fabrizio Macagno on 06 December 2020.

The user has requested enhancement of the downloaded file.

Argumentation (2009) 23:81–107
DOI 10.1007/s10503-008-9110-2

Reasoning from Classifications and Definitions

Douglas Walton Æ Fabrizio Macagno

Published online: 22 August 2008

Springer Science+Business Media B.V. 2008

Abstract In this paper we analyze the uses and misuses of argumentation schemes
from verbal classification, and show how argument from definition supports argu-
mentation based on argument from verbal classification. The inquiry has inevitably
included the broader study of the concept of definition. The paper presents the schemes
for argument from classification and for argument from definition, and shows how the
latter type of argument so typically supports the former. The problem of analyzing
arguments based on classification is framed in a structure that reveals the crucial role it
plays in the persuasion process. The survey of the literature includes the work of
Hastings, Perelman, Kienpointner and Schiappa, but still finds much of value in
Aristotle. Lessons drawn from Aristotle’s Topics are shown to be useful for devel-
oping new tools for assessing definitions and arguments from definition.

Keywords Argumentation schemes
Definition
Persuasive definition

Argument from definition to classification
Aristotle
Topics

Many argumentation schemes have now been studied in modern argumentation

theories (Hastings 1963; Kienpointner 1992; Walton 2006), and there has been in-
depth inquiry into well-known schemes like argument from expert opinion, argument
from analogy, and so forth. However, it is interesting to notice that very little attention
has been paid to what is arguably one of the most fundamental schemes, namely what
has been in the history of schemes referred to as ‘‘argument from verbal
classification’’. Classification is fundamental for the simple reason that reality must

D. Walton (&)
Department of Philosophy, University of Winnipeg, Winnipeg, Canada R3B 2E9
e-mail: [email protected]

F. Macagno
Department of Linguistics, Università Cattolica del Sacro Cuore, Milan, Italy
e-mail: [email protected]

123
82 D. Walton, F. Macagno

be named and linguistically organized in order to talk about it. Moreover, verbal
classification is in many ways the subtlest and most powerful argumentative tool.
Accepting a verbal classification, that is, the use of a particular word to denote a
fragment of reality, requires accepting the classified object’s possession of certain
properties. The acceptance of these properties, without the audience (respondent)
realizing it, may, by inference, warrant the acceptance of a conclusion. In this paper we
will look at several examples from which lessons can be drawn. In particular, we will
show how the process of ‘‘naming reality’’ or classifying it can be used in inferences
leading to value judgments.1 For instance, the conclusion ‘‘This is a bad company’’ can
be supported by a factual premise such as ‘‘It has monopolized the market’’. Schiappa
(2003, p. 131) and Zarefsky (2006, p. 404) call this strategy argument by definition
(Zarefsky uses also the name ‘persuasive definition’) and represent it as a pattern of the
following kind: x is P, therefore x is good/bad. These argumentation patterns, both
based on endoxical propositions, can be used fallaciously by manipulating the
commitments (or endoxa) of the interlocutor.
The modern concept of classification or naming describes the effects of a process of
inference grounded on a semantic link between premises and conclusion called in the
ancient tradition ‘‘locus a definitione’’, namely topics from definition. Definition, in
other words, describes the semantic reason linking a premise such as P, ‘‘Bounce—O
Company controls the manufacture of all ping-pong balls in the U.S.’’ to the
conclusion C ‘‘Therefore Bounce—O company is a ping-pong ball monopoly’’
(Windes—Hastings 1965, p. 160). The logical link between the classification
conclusion to the premise can be described by the missing premise ‘‘monopoly is
control of the market’’, namely a definition of the concept of ‘‘monopoly’’.
If the argument from verbal classification shows how the conclusion is a
classification of a fragment of reality based on some characteristics (namely how C
follows from P in virtue of a link of classification), it does not show the semantic-
ontological reason why the premises and the conclusion are linked (namely why
‘‘monopoly’’ and ‘‘control of the market’’ are connected, and how). The concept of
definition, grounded in the Aristotelian semantic system, can help understand the
missing relation between the logical aspects of the reasoning, based on a logical
relationship, and the reasons of its reasonableness.

1 The Role of Classification and Definition in Argumentation

Definition plays a fundamental role in the process of classification. However, before

analyzing how reasoning from classification works and the relation between definition
and classification, we want to take into account its argumentative power, namely how
classification and definitions can become powerful argumentative strategies.
The use of definition and classification is extremely complex and deeply
embedded in argumentation.2 The primary role of definition is the determination of
1
This is not to interpret such arguments as purely verbal in nature. Argument from values (positive or
negative) is a distinct argumentation scheme in its own right, and it can be joined to arguments from
classification in significant ways, as several of the examples in this paper will show.
2
For an insight into the argumentative use of definition, see Robinson (1950) and Walton (2005).

123
Reasoning from Classifications and Definitions 83

the issue at stake. If two parties in a dialogue do not share the same understanding of
the problem, there is a risk that the goal of the argumentation will not be achieved.
In other words, the interlocutors can talk about the same words without talking
about the same concepts. Aristotle highlighted this fundamental role of definition
(Sophistical Refutations, Topics) taking into consideration the concept of polisemy
and fallacies from ambiguity. Definitions however play other functions in
argumentation and in particular they can be powerful instruments of persuasion,
as emerges from the theories of Perelman, Stevenson, and Schiappa. In particular,
definition is the instrument of naming reality, namely the aspect of shared
knowledge allowing one to give a name to a fragment of reality. However, naming
in many cases is not an argumentatively neutral process. On the contrary, names can
constitute powerful instruments of persuasion and manipulation.

1.1 The Argumentative and Persuasive Uses of Definition

In Perelman and Olbrecht-Tyteca’s The New Rhetoric definitions are seen as

standpoints that need to be supported by arguments (such as, for instance,
etymology or consequences), and as arguments grounding a thesis. Definitions, for
this reason, are analyzed as the result of an argumentation and as the premises for an
ulterior argumentative move. In The New Rhetoric, definitions are seen as
instruments which can be used to influence the relations between the concept and
a whole system of thought (Perelman and Olbrechts-Tyteca 1969, p. 213). A
definition, for this reason, is always a matter of choice. The correspondence of
identity between the definiens and the definiendum is a quasi-logical relation,
argumentatively warranted. The most prominent argumentative strategy based on
definition highlighted by Perelman is the dissociation of concepts. In order to
support a point of view, a definition is used to reduce the denotation of a concept, by
splitting the previous meaning of the term into two concepts. For instance (p. 418),
‘‘true’’ equity is opposed to ‘‘apparent’’ equity, thereby introducing a polisemic use
of the term ‘‘equity’’ and distinguishing between two different concepts. This theory
has been developed by Van Rees (2005). Her proposal is based on the observation
that dissociation is always employed to introduce a new meaning of the word
defined, which bears a certain connotation or value judgment.3
The notion of dissociation introduces another extremely interesting function of
definition, namely emotive persuasion. Some words, in fact, have an ‘‘emotive
force’’ or are associated to certain consequences, such as legal terms (for instance, if
a person is characterized as a ‘‘thief’’ he will be prosecuted). The relation between
definitions, emotive language, and argumentation has been approached in the
principal works of Stevenson and Schiappa.
Stevenson in Ethics and Language grounds his account of persuasive definition
upon a behaviorist theory of language. The meaning of a word is the correlation
between the sign and the reaction it provokes in the interlocutor (Stevenson 1944,
p. 54). There are two principal reactions, or types of meaning: the cognitive and the
emotive reaction. The descriptive (cognitive) meaning is in some instances based on
3
The concept of dissociation and of the emotive force of words is analyzed also in Halldén (1960).

123
84 D. Walton, F. Macagno

the emotive meaning. Some terms, in fact, are emotively charged because their
referent is positively considered by the community of speakers. On the other hand, in
some cases, such as ethical terms (the adjective ‘‘good’’ for instance), the two
meanings cannot be distinguished. At the same time, these terms signify certain
properties and evoke a certain behavior. This linguistic foundation is basic to
understanding the strategies of persuasive definitions and quasi-definitions. In cases of
persuasive definitions, a term that has a descriptive and an emotive meaning is re-
defined in order to change its denotation. Its denotation can be restricted or enlarged, in
order for the emotive reaction to apply to the categorization of a particular fragment of
reality. The emotions of the interlocutors are directed towards the desired object. For
instance, a positively valued term such as ‘‘culture’’ can be re-defined as being
grounded on the fundamental property of ‘‘originality’’, and thereby including in this
positive category of ‘‘cultured’’ people who have no wide knowledge. The mirror
image of the persuasive definition is the quasi-definition. In this argumentative
technique, the descriptive meaning is left unaltered, while the emotive meaning is
changed. For instance, the positive emotive attitude elicited by the word ‘‘virtue’’ may
be changed by the quasi definition ‘‘virtue is an antiquated rubbish which robs a man of
all individuality’’. In this case, a positively connoted term can be used to express a
negative emotive attitude that, in effect, works as an unstated premise in an argument.
Another interesting approach to definitions and to their use in argumentation is
Schiappa’s pragmatic theory. Schiappa refuses to recognize that words have an
essential meaning, and instead defends a ‘‘social constructionist’’ conception of reality
and language. He develops this position to support the conclusion that defining reality
means advocating a theory and a point of view. Defining, in his opinion, is describing
the use of a word to refer to objects according to the way that they are conceived by
people. Every definition is based on a set of similarity–difference relationships, by
means of which the speakers can classify reality. These relationships derive from the
process of learning, and the meaning of a word is identified in the denotative
conformity of a community of speakers. Schiappa, considering learning to be the result
of a process of persuasion in which a theory is imposed on the learner, describes
definition as a means of leading the interlocutor to accepting a specific theory. For this
reason definition is depicted as an argumentative act whose goal is to alter our
valuation of reality, an act imposing a different way of thinking. Schiappa applied this
theory to the study of the ‘‘definitional ruptures’’, or re-definitions. A re-definition is
always advanced to defend a particular viewpoint, such as in the case of the
redefinition of ‘‘death’’ as ‘‘brain death’’. This terminological change was done for the
purpose of extending the denotation of the term. On the other hand, a definitional
change can be performed for the purpose of changing a determinate perspective on a
fragment of reality. In sum, to fully grasp a definition we need to see it as dependent on
the goal of the definer in a talk exchange.
Schiappa distinguished between two main schemes relative to definition:

Argument from definition Argument by definition

All X are Z; Y is an X; therefore Y is Z X is Y (therefore R)

123
Reasoning from Classifications and Definitions 85

The second scheme, by definition, is based on the persuasive use of naming. Naming
a state of affairs X as Y is an argumentative strategy provoking an emotion (a
reaction R). For instance, a complex reality can be denoted by a ‘‘domestic term’’
(for instance, a missile can be named ‘‘Peacekeeper’’) or a ‘‘bureaucratic term’’,
incomprehensible for most of the interlocutors (for instance, neutron bombs are
called ‘‘radiation enhancement weapon’’). Names can enhance positive attitudes and
at the same time denote or conceal a fragment of reality commonly considered
negative. This characteristic of words is exploited in some descriptions. Defining
and naming for Schiappa always express an attitude, orientating the interlocutor
towards a certain conclusion. Describing reality is on his view never neutral either.
The same reality can be differently framed, in order for it to be differently
categorized. Some aspects of a situation or of an object can be emphasised while
others are concealed or ignored. For this reason, the same fragment of reality can be
differently described to defend specific points of view. Schiappa’s theory, we can
notice, is extremely interesting since it highlights the fundamental relation between
definition, reality, and attitude. However, the ‘‘definitional relativism’’ his theory is
grounded upon is problematic. His view that a definition is always aimed at altering
our valuation of reality seems not to distinguish between redefinitions, fallacious
definitions, and definitions shared by a community.
Schiappa and Stevenson point out the argumentative and persuasive use of
definition. Both stress the fundamental relation between emotions and predication,
and how a definition plays a crucial role in legitimating a predication of an emotive
word or of a term which is argumentatively relevant (for instance, legal terms). In
particular, whereas Schiappa focuses on the persuasive uses of definition, Stevenson
analyzes the structure of persuasion through definition distinguishing between the
descriptive and the emotive component of meaning.

1.2 Definitions, Values, and Argumentation

If we take into consideration the persuasive uses of definition and Stevenson’s

distinction between emotive and descriptive meaning, we can try to reconstruct the
inferential structure of the argumentative use of definition. In particular, if we
inquire into Aristotle’s concepts of choice and action, we can notice that the
structure of the argumentative move can be analyzed as a two step reasoning: a
process of classification and an inference drawn from the notion of ‘‘value’’.
In Aristotle’s Topics and Nicomachean Ethics, the good, the pleasant and the
useful, namely the desirable (airetón) (Topics 118b, 27–30), are considered the
principle of will and action (Ethics III, 4), ‘‘for everything aims at the good’’ (Topics
116a 18). Because the choice is based on the will which aims at the desirable, the
desirable itself is the foundation of the decision making. The topics of the third book
of Aristotle’s Topics can be considered a set of reasons aimed at establishing what is
desirable, on the basis of the interlocutor’s interests and experiences. In other words,
these topics can be used to determine what is good, what is bad, and what is better
on the basis of values, namely reasons to act. If we apply the Aristotelian principles

123
86 D. Walton, F. Macagno

to the argumentation based on definition and classification, we can see how the
connection between a classification and the evaluation is simply a matter of
common knowledge. Concepts such as ‘‘beauty’’, ‘‘youth’’, ‘‘courage’’, ‘‘culture’’
are commonly considered to be positive in our culture (obviously this evaluation is
highly subjective and culturally bound). In other words, these concepts are positive
values and therefore desirable and reasons to act. The classification itself works in
these cases as an implicit argument as in the analysis of the following classification
of ‘‘the suppression of the opponents of Bolshevism’’:
These actions [the suppression of the opponents of the Bolshevism] are acts of
peace.
In the grey boxes (Fig. 1) the implicit premises are represented. We should
observe that in the argument represented by the classification of ‘‘suppression of the
opponents of Bolshevism’’ two crucial steps can be identified. The first step is the
process of characterizing the fragment of reality considered as ‘‘acts of peace’’ by
means of an implicit redefinition of ‘‘act of peace’’. The second step is the
evaluation of the suppression of the opponents, grounded on an endoxical premise
(peace is good) and a maxim, what leads to good results is itself good (Aristotle,
Rhetoric I, 6). This second implicit step leads to the conclusion that the suppression
of the opponents is good (desirable). This evaluation is an implicit commitment to

Fig. 1 Argument diagram for the Bolshevism example

123
Reasoning from Classifications and Definitions 87

the rightfulness of the action. The argumentation process leading from values to an
evaluation, and therefore to a commitment, can be represented by the following
argumentation scheme from values (Walton et al., cap. 9):

Argumentation from values

PREMISE 1 Value V is positive as judged by agent A (judgment value)
PREMISE 2 The fact that value V is positive affects the interpretation and therefore
the evaluation of goal G of agent A (If value V is good, it supports
commitment to goal G)
CONCLUSION V is a reason for retaining commitment to goal G

The second premise, in particular, represents an abstraction of the possible maxims

connecting values and evaluations. It leads from a classification of a situation (acts
of peace) to an evaluation (acts of peace are desirable), which is in itself an action
(agent A supports acts of peace). By distinguishing between the process of
classification from the evaluation (using the argumentation from values mentioned
in (Walton et al. 2008, cap. 9), we can identify the role that definition plays in the
persuasive process of persuasive definitions, and how a definition can become
manipulative.

1.3 Manipulating Definitions

If we analyze the possible types of manipulation proceeding from the persuasive

uses of definition, we can see that there are three main typologies of fallacies related
to the process of classification. First, it is possible for the definition not to be shared
by the interlocutor. Second, it can be the case not only that the definition is not
shared, but also that the evaluation of the concept defined is altered. Third, the
interlocutors might agree on the fact that the features given occur in a definition of
an object, but the description of the fragment of reality (the object) is altered by the
speaker in order for the object to fall into the classification.
An example of the first case is the Leninist definition of peace (Codevilla 2003,
pp. 69–70). Peace is defined as everything that favored the expansion of
Bolshevism, included war itself. This new meaning of the word ‘‘peace’’ is used
(or imposed) to classify acts of war. Obviously, this concept of ‘‘peace’’, like many
other concepts (such as ‘‘revolution’ itself (Codevilla 2003, p. 63)) is not commonly
shared by all interlocutors. The same word, for this reason, can be used to signify
two contradictory concepts, and thereby manipulate communication.
In other cases, it can be the case that the definition is shared, but the evaluation of
the concept it refers to is not. For instance, as Aristotle pointed out in (Topics II, 11),
in some communities it is honorable to sacrifice the father, whereas in others it is
not. The same concept (for instance ‘‘to sacrifice a relative’’) can be defined in the
same fashion, but differently evaluated. However definitions can be used to change
the evaluation of the concept, as we can notice in the following case (Molière 2000,

123
88 D. Walton, F. Macagno

p. 98), in which Don Juan is trying to persuade his interlocutor, Sganarelle, of the
negativity of marriage and fidelity:
DON JUAN: What! Would you restrict a man to staying chained to the first
woman who takes his fancy, have him give up everything for her and never
look at any others again? The idea is ludicrous – making a bogus virtue out of
being faithful, being trapped forever in the same relationship and as good as
dead from youth onwards to the other pretty faces that might catch our eye! No
no: fidelity is for imbeciles. All beautiful women are entitled to our love, and
the accident of being the first on the scene shouldn’t deprive the rest of the
rightful claims they have on our affections.
In this example, it is interesting to notice how the situation is described not only
using a not-shared definition of marriage, fidelity and love, but also a not-shared
evaluation of these concepts. In this argument, in fact, the speaker presupposes that
being committed to a relationship is not desirable (trapped), that the women look for
the kind of affection described as ‘‘love’’ and that the men have to fulfill the
appetitive desires of women. These premises are hardly accepted as true by the
interlocutor, a defendant of the faithfulness in the marriage and of the positive value
of the original concepts. This redefinition takes into consideration only some
accidental aspects of the concept defined, namely that a relationship can be a trap.
The manipulation of the concept defined leading to altering the values associated to
the definiendum can consist of narrowing or broadening a concept that sometimes is
vague (see Aberdein 2000), including positive (negative) accidental aspects or
excluding positive (negative) essential aspects of the concept. There is nothing
inherently fallacious in the speaker’s defining a particular way he is using a term.
The risk of fallacy can arise when he is presenting the particular word use as the
proper use of the word everybody agrees upon. The risk of sophistry lies in the fact
that the concept defined by the speaker is not the same as the hearer’s, but is
presented as such. By narrowing down a concept, it is possible to select only the
worst or best aspects of the reality it refers to. We can examine the following
example, in order to make this concept clearer (Goarke and Tindale 2004, p. 99):
Socialism is that form of government that steals wealth from energetic people
and divides it among the lazy poor.
Here the definition is of a commonly shared concept, socialism. The definition,
instead of referring to the whole reality of what is commonly intended to be
‘‘socialism’’, takes into consideration only a particular aspect of it, namely its
degeneration. The thing, the concept defined is another concept, namely the
degenerated socialism, or the worst effects of socialism, not ‘‘socialism’’ itself. The
definer, in other words, defines a new concept and make the negative emotive
inferences of ‘‘degenerate socialism’’ apply to ‘‘socialism’’ in its entirety.
The last case regards the relation between a shared definition and a not-shared
framing of the situation. In order for a state of affairs to fall into a definition, it
has to present the essential features stated in the definition. The speaker can use a
shared definition, but presuppose that the situation presents some inexistent

123
Reasoning from Classifications and Definitions 89

features, or suppress relevant evidence. For instance we can analyze the example
below4:
Reverend BARRY LYNN (Americans United): I just think it’s incredibly
inappropriate when you’ve got the head of an agency or a department of
government having a daily religious ritual that includes some people and of
course, by definition …
MATTHEWS: OK, ritual—loaded term.
Rev. LYNN: … excludes some.
MATTHEWS: Loaded term, ritual.
Rev. LYNN: Oh, sure. But it is—but it is a prayer session, a study section. It does
include a prayer as well and I think that the proof …
The word ‘‘ritual’’ presupposes a situation including not only prayers, but also some
ceremonials making the situation itself a ‘‘religious ceremony’’. Here the definition
is shared, but the situation is framed including facts that are not accepted by the
other party, such as the presence of prayers. In fact, in the last move, Rev. Lynn has
to prove that there are prayers as well. The classification can be based also on the
suppression of relevant evidence. For instance we can consider the following case5:
‘Once again Britain has been found sucking up to dictatorships.’ (Or
maintaining friendly relations with strong governments. Note how ‘found’
implies that we were discovered in a guilty secret.)
Here the classification is put forward referring to the relationships between Britain
and fascist countries. If we consider this statement, which can be used as an implicit
argument for drawing a value judgement on Britain, we can notice that the move
consisting of classifying dictatorships as ‘‘strong governments’’ is fallacious. The
definitions of ‘‘government’’ and ‘‘dictatorship’’ are shared, and in particular nobody
would think that ‘‘dictatorship’’ does not include in its definition the suppression of
political rights. The classification move presupposes that the countries mentioned do
not suppress political rights (otherwise they would be dictatorships, and not
governments), framing the situation in a fallacious way. The suppression of pieces
of information relevant to the making of a classification can be considered as a case
of suppression of evidence. This strategy is often used to depict a complex reality
under a single label (see Rigotti 2005; Lakoff 1996, p. 320; Goarke and Tindale
2004, pp. 86–87).

2 Schemes from Verbal Classification

As seen in the section above, naming reality can be examined as having two
argumentative aspects: it is grounded on definitions and it often leads to evaluative
inferences. In this section the inference leading from a definition to a classification,

4
http://www.cwfpac.com/chairmans_corner_speeches_05_14_01.htm
5
http://www.adamsmith.org/logicalfallacies/000638.php

123
90 D. Walton, F. Macagno

or to naming reality, will be analyzed, starting from the leading theories on

argument from classification.
We can observe that the accounts presented approach the topic of reasoning from
classification from two different points of view. On one hand, Hastings is concerned
with the logical structure of the link between premises and the classificatory
conclusion, without considering the semantic reason allowing the factual premise to
support the classificatory conclusion. On the other hand, Kienpointner’s approach to
argument schemes is more related to the connection between the conclusion and the
semantic reason. In particular, he examines the argument from definition. From the
analysis of Hastings’ and Kienpointner’s schemes we come to assess the potential
weaknesses of Walton’s argumentation scheme from verbal classification and
propose a new formulation of this inferential pattern.

2.1 Hastings

Hastings, in his Ph.D. thesis (1963), identified two schemes that can be treated under
the label of argument from classification. The first scheme, in fact, leads from a set
of characteristics to the attribution of a predicate to a subject, much like the process
of classification described above. In the second scheme, a subject, classified as X, is
predicated of the definition of X. In other words, first a predicate is attributed to a
subject (the subject is classified); then in virtue of the definition of the predicate
some fundamental characteristics are attributed to the subject.
We can notice that the second scheme is the mirror image of the first. The
definition used to classify a subject is in the latter scheme applied to the classified
subject. We can represent the two argument schemes as follows (see Hastings 1963,
pp. 36–52):
Argument from Criteria to Verbal Classification
Event or object X has characteristics A, B, C…
If x has characteristics A, B, C… then x is Q
Therefore, event or object X is Q.
For example, Hastings gives the following example (p. 36):
In Voluntary health insurance you generally get a poor return for your money
because overhead and profits of the insurance company eat up huge chunks of
the premiums you pay. On individual policies these companies spend for
overhead and profits an average of about 60% of what you pay them and only
about 40 cents of your premium dollar goes for benefits to policyholders.
Obviously such insurance is a mighty poor buy (Fig. 2).
Hastings highlights the fact that the classification proceeds from the evidence that
is presented. This is an extremely interesting observation, since we can notice that
it is connected to the ancient studies on issues (stasis). According to this ancient
theory (see Cicero and Hermagoras), the process of classification follows two
other stages, in which evidence is collected and the definition of the terms is
established.

123
Reasoning from Classifications and Definitions 91

Fig. 2 Hastings’ insurance example

The classification, moreover, proceeds from a rule of classification that is

shared by the audience. There are seven critical questions attached to this scheme
(pp. 42–45):

C.Q.1 What is the implicit definition being used?

C.Q.2 Is the definition acceptable: are the criteria acceptable as a definition of the classification, label,
adjectival category, etc.?
C.Q.3 Are there exceptions or qualifications to the definition and criteria?
C.Q.4 Are other criteria necessary for an adequate definition?
C.Q.5 Do the characteristics described meet the criteria?
C.Q.6 Are enough characteristics described to justify inclusion in this category?
C.Q.7 Could the event fit better into another category, or be classified differently on the basis of its
characteristics?

In Hastings’ argument from characteristics to verbal classification we can notice that

the critical questions play a fundamental role. In the structure of the inference
nothing is said about the nature of the conditional proposition, whereas the critical
questions specify that the strength of the inference depends on the acceptability of
the definition.
If the argument from criteria to verbal classification represents the conditions of
the predication of a name, argument from definition to characteristics highlights the
structure of the inferences that can be drawn from the predication itself. In his
account of argument from definition to characteristics, Hastings (1963, pp. 46–54)
did not appear to offer any precise schematic form representing this argumentation
scheme, even though he did set a list of three critical questions corresponding to this
type of argument (53–54). The following new argumentation scheme for argument
from definition to characteristics was put forward in (Walton 2008).
Argumentation Scheme for Argument from Definition to Characteristics
Premise: according to the definition D of concept C, property P is included in
C according to D.

123
92 D. Walton, F. Macagno

Conclusion: therefore P is a property of C.

In argument from definition to characteristics, a state of affairs is named in a certain
way and, on the basis of the definition of the name, some characteristics are
predicated of the state of affairs. The characteristics predicated in the conclusion
might be semantic characteristics contained in the definition, or implications drawn
from the definition. As in the case of argument from classification, the premise, that
is the definition, must be acceptable and accepted by the interlocutors, as specified
in the critical questions (Hastings 1963, p. 53):

C.Q. 1 Is the definition an accurate or an agreed upon definition?

C.Q. 2 Do the implications or characteristics follow from the premises?
C.Q. 3 Are any conflicting, inconsistent, or superseding principles involved?

The structure of the scheme can be understood by considering the following

diagram displaying the structure of an argument from (Robinson 1947, p. 200):
Since you believe in tolerance in all things, you have no right to be so critical
of this man’s ungentlemanly conduct (Fig. 3).
Here the scheme proceeds from an implicit definition of tolerance (tolerance is the
acceptance of any kind of behavior or position) accepted by the audience and
presupposed by the assertion ‘you believe in tolerance’. From the definition an
implication is drawn in the conclusion. Tolerance, when understood as the
acceptance of any kind of behavior or position, implies an absence of criticisms of
any kind of conduct or position. This kind of argument can not only lead to a
conclusion based on an implication, but also to a conclusion proceeding from the
predication of a semantic feature of the definition, as follows (Hastings 1963,
p. 48):
If the United States be not a government proper, but an association of States
in the nature of contract merely, can it, as a contract be peaceably unmade,
by less than all the parties who made it? One party to a contract may violate
it – break it, so to speak, but does it not require all to lawfully rescind it?
(Fig. 4)

Fig. 3 Robinson’s tolerance example

123
Reasoning from Classifications and Definitions 93

Fig. 4 Toulmin’s structure of Hastings’ contract example

2.2 Kienpointner

The most extensive account of arguments schemes from classification is given in

Kienpointner (1992). In Alltagslogik four schemes from definition are identified and
the classifications by means of genus, and whole and parts are analyzed. The
schemes from definition can be represented as follows (1992, pp. 250–252):

Type of scheme Scheme

Descriptive What is predicated of the definition is also predicated of the definiendum and vice versa
X is predicated of the definition
Therefore X is predicated of the definiendum
What the definiendum is predicated of, also the definition is predicated of
The definiendum is predicated of X
Therefore the definition is predicated of X
Normative If X is defined by means of definition Y, valuation Z relative to X is justified
X is defined by means of definition Y
Therefore valuation Z relative to X is justified
If X is defined by means of definition Y, action Z is advisable
X is defined by means of definition Y
Therefore action Z is advisable

In these schemes, the definition can be substituted by the interpretation of the name
(p. 259). The last two schemes are frequently used in persuasive definitions. In
commercials, a product is described as having a certain number of qualities, leading
to the conclusion that it should be bought. The scheme can be presented as follows:
If product X is defined by means of Y, it is advisable to buy X
Product X is defined by means of Y
It is advisable to buy X

123
94 D. Walton, F. Macagno

The definition is necessary to the process of classification. If there is not a unique

definition of a term, the same reality can be contradictorily classified. For instance,
the word ‘‘full employment’’ can be defined as ‘‘a situation in which only 5.5% of
the population of a country is unemployed’’. If this definition is considered, a
country like the U.S.A. can be classified as and characterized by ‘‘full
employment’’. On the other hand, the same term can be defined as ‘‘a country in
which all the employable adult people have a full-time activity’’. In this case, the
same country would be not be classified as and not characterized by ‘‘full
employment’’.
Kienpointner’s account, we can notice, revives the ancient tradition on topics.
The ‘‘major’’ premise of the schemes, such as for instance ‘What is predicated of the
definition is also predicated of the definiendum and vice versa’ represents the
ancient maxim of the locus from definition (see for instance Boethii, In Ciceronis
Topica, 1059c). However, in its scheme it is not clear how the general principle (the
maxim) is related to the endoxical premise, namely the definition. This aspect of the
argument scheme from definition opens an interesting perspective that will be
developed in the following subsection.

2.3 Developing Argument Schemes from Classification

The purpose of this subsection is to show how the argument schemes from
classification presented in the last section are somehow inadequate to describe the
semantics of the inferential structure of the argument. By taking into consideration
the weaknesses of the schemes we can develop a new formulation of the
argumentation from classification.
In Walton (2006, p. 129), the following scheme from verbal classification is
advanced, maintaining the structure of the scheme presented in Hastings (1963):

Walton’s (2006) scheme

INDIVIDUAL PREMISE a has property F
CLASSIFICATION PREMISE For all x, if x has property F, then x can be classified as having property G
CONCLUSION a has property G

We should notice that in this scheme, such as in Hastings’ argument schemes, the
relation between F and G is not specified. If the structure of the inference
represented is logically valid, the reasonableness of the inference itself does not
seem to be considered in the classification scheme. The conclusion ‘a is blue’
follows logically from the classification premise ‘for all x, if x is red then x is blue’
and from the premise ‘a is red’, but is unreasonable. In Walton et al. (2008), the
nature of the classification was made clearer by developing the classification
premise into the following one:
CLASSIFICATION PREMISE (Walton et al. 2008): For all x, if a fits definition D,
then x can be classified as having property G.

123
Reasoning from Classifications and Definitions 95

Walton’s (2006) scheme can be therefore refined and made more useful for
describing the inferential structure by making the nature of the relation between F
and G explicit. However, even in this scheme, the relation between D and G is not
clear. A better formulation of the inferential passage can be found in Keinpointner.
If we take into consideration Kienpointner’s scheme from definition and apply it
to a case, the following argument will follow:

The definiendum is predicated of X Bob is a man

What the definiendum is predicated of, also the definition is If Bob is a man, then Bob is a rational
predicated of animal
Therefore the definition is predicated of X Therefore Bob is a rational animal

However, we can notice, the endoxical premise ‘rational animal is the definition of
man’ is lacking here.6 The only possible inference that can be drawn from the
premises is the following:

The definiendum is predicated of X Bob is a man

What the definiendum is predicated of, also the If Bob is a man, then the definition of man is
definition is predicated of predicated of Bob
Therefore the definition is predicated of X Therefore the definition of man is predicated of
Bob

For this reason, we can represent the structure of the inference as follows (see
Rigotti and Greco 2006):

MAXIM ENDOXON
What the definiendum is predicated of, also the definition is
predicated of
Rational animal is the definition of
man
PRELIMINARY CONCLUSION

What man is predicated of, also rational animal is predicated of

The definiendum is predicated of X Bob is a man
Therefore the definition is predicated of X Bob is a rational animal

Following the connection between the endoxon and the maxim, the argumentation
scheme from criteria to verbal classification can be formulated as follows, merging
together the schemes advanced in (Walton et al. 2008) and (Walton 2008), in which:

6
We would like to thank Eddo Rigotti for his advice and observation on this aspect of Kienpointner’s
argument schemes.

123
96 D. Walton, F. Macagno

DEFINITION PREMISE: a fits definition D

CLASSIFICATION For all x, if x fits definition D, and D is the definition of G, then x can be classified
PREMISE: as G
CONCLUSION: a has property G

In this scheme, we can notice, the reasonableness of the inference is guaranteed by

the relation between the definition and its definitum, whereas the relation between
the endoxical premise and the relation between definiens and definiendum is
represented by the additional premise ‘D is the definition of G’. The classification
premise could be also represented as ‘What the definiendum is predicated of, also
the definition is predicated of, and D is the definition of G’. The critical questions
appropriate for this version of the argument from verbal classification are the
following (from Walton et al. 2008, chap. 9):7
CQ1: What evidence is there that D is an adequate definition of G, in light of other
possible alternative definitions that might exclude a’s having G?
CQ2: Is the verbal classification in the classification premise based merely on a
stipulative or biased definition that is subject to doubt?
CQ3: Does a actually fit definition D?
On this view, the abstract inferential structure presented in Hastings is combined
with an analysis of the semantic link between the definitional premise and the
conclusion. In particular in the implemented argument scheme from definition we
can notice that by making the link between the definition and the definiendum
explicit (‘‘D is the definition of G’’) the endoxical nature of definition is highlighted.

3 The Nature of Endoxical Definitions: The Tradition and the Theory

of Predicables

The analysis of the argument schemes from verbal classification shows how the link
between the premises and conclusion is grounded on a semantic link which was called
by the tradition the maxim, and on an endoxical, or commonly shared premise, namely
the definition itself. In the section above, in fact, we took into account the structure of
the inference, showing how the conclusion depends on two conditions:
r the object x must fit the definition D
r D must be the (shared) definition of G
However, we can notice, there might be different types of definition of the same
concept. For instance, a concept such as ‘‘man’’ can be defined as ‘‘the rational
animal’’, ‘‘the animal that can pity the Gods’’ ‘‘the biped animal’’, ‘‘the being
composed of two legs, a head, two arms…’’, ‘‘the being Homo sapiens sapiens and
Homo sapiens idaltu belong to’’…There might be many types of definition, different
for the semantic content (for instance ‘‘man is the rational animal’’ and ‘‘man is the
7
We would like to thank the anonymous referee for his comments on this point.

123
Reasoning from Classifications and Definitions 97

laughing animal’’) or for their structure (for instance ‘‘man is the rational animal’’
and ‘‘man is the being composed of two legs, a head, two arms…’’.
In the first case, we can notice, the acceptability of the conclusion of a reasoning
from definition depends only on the acceptability of the definitional premise. For
instance, we can consider the two different inferences drawn from two different
definitions of ‘‘monopoly’’:

1. Pop Cola is controlling the soft drinks market. A monopoly is a company that exclusively
Therefore it is a monopoly controls the market
2. Pop Cola is the biggest soft drinks company. A monopoly is the biggest industry in a field
Therefore it is a monopoly of activity

Whereas in (1) the inference is drawn from a commonly accepted definitional

premise, the conclusion in (2) follows from a definition of monopoly which is hardly
acceptable. The conclusion in (1), we can notice, is much more acceptable, for this
reason, than the conclusion in (2).
In the second case, however, the acceptability of the conclusion does not only
depend on the acceptability of the definition, but also on what we can call its ‘‘nature’’,
or type of definition. For instance, we can compare the two following inferences:

3. There is a building that serves as living quarters for one House is a building that serves as living
or a few families. Therefore there is a house quarters for one or a few families
4. There are four walls, a roof, the foundation. Therefore House is four walls, a roof, the foundation
there is a house

Whereas in (3) the conclusion reasonably follows from the premise, in (4) the
conclusion is not supported in the same fashion from the premise. If there is a
building that serves as living quarters for one or a few families, there must be a
house, but even if there are four walls, a roof, the foundation, it can be reasonably
concluded that there is no house.
Moreover, if we consider other types of definitions, we can notice that they can
be used for different purposes. We can introduce the topic by taking into
consideration the two following inferences:

1. Man is the being who can pity the Gods. Therefore man is a good being To pity the Gods is good
2. Man is the being who can despise the Gods. Therefore man is a bad being To despise the Gods is
bad

In these two examples, different types of evaluative inferences are drawn from the
two definitions. In both cases, we can notice, the definition can be acceptable, and it
is convertible with the definiendum (man is the only being who can pity or despise
the Gods). However, according to the definition used, different conclusions can be
supported.
From these examples, we can notice how definitions differ not only because of
their different status as shared premises, namely whether they are commonly
accepted or they are not, but also because of their nature. The question we want to

123
98 D. Walton, F. Macagno

address in this section regards the possibility of evaluating definitions. Our claim is
that the existence of many different types of definitions does not lead to a
definitional relativism, but, on the contrary, to argumentative discussions on what is
the best definition. In order to have a discussion on definition, however, it is
necessary first to establish what the best, or the ‘‘real’’ definition is. The starting
point for our analysis is Aristotle’s dialectical work, namely the Topics, in which the
concept of dialectical definition is introduced.

3.1 Definition and the Aristotelian Semantic System

Aristotle in his Topics laid the fundamentals of his dialectical studies of classification.
In this work, Aristotle distinguishes between the four predicables, that are four classes
of semantic–logical relations of predication. These relations are formulated in the
form of intrinsic topics, namely instruments of discovery and inference warrants
directly connected to the subject of discussion. Aristotle distinguished four
predicables: genus (for instance ‘‘house is a building’’), definition (for instance
‘‘house is a building that serves as living quarters for one or a few families’’), property
(for instance ‘‘do up a house’’, which is said of ‘‘house’’ only), and accident (for
instance ‘‘red’’ or ‘‘nice’’ said of ‘‘house’’). All the predicables can be predicated of the
species, which, in these examples, is for instance ‘‘house’’. The species, conceived by
Aristotle as a thing, can be interpreted as a categorization of a fragment of reality that
we can describe as the meaning of a word.8 The species (or concept) is that which can
be predicated of more individuals different in number (for instance ‘‘house’’ can be
predicated of my house, or my neighbor’s house ….), and falls outside the domain of
dialectic. Dialectic is concerned about relations between concepts, not about reasoning
relative to the particular objects (Crowley and Hawhee 1999, p. 54; Green-Pedersen
1984, p. 119).9
The predicables are divided into two groups according to their semantic
properties. The first class incorporates the predicables that can reveal the essence of
the thing, that is (see Rigotti and Greco 2006), what the concept is or, rather, its
fundamental characteristics (see Stebbing 1933, p. 429). In this group falls genus
and definition. The second class is characterized by not expressing the essence of
the thing and it incorporates property and accident. On the other hand, a second
division of the predicables is advanced in the Topics, and is relative to the logical
properties. While definition and property are convertible with the species they are
predicated of, genus and accident are not. From this broad division it is possible to
understand the definition of the predicables.

8
This interpretation is coherent with Aristotle’s perspective of dialectic. Dialectic does not deal with
objects and individuals (what we can call ‘‘things’’), but with species, namely linguistic organization of
reality. He is not interested in the matter, but in the form, that is in the relevant semantic properties of the
concepts.
9
Aristotle (Topics, I, 10), considers a dialectical proposition to be a proposition held by everybody, or
the majority, or the wise. Dialectic (Topics, I, 14) is about science, and science is not concerned with
particulars. In the Middle Ages, the account of the predicables is different. Medieval tradition stems, in
fact, from Prophyry’s Isagoges, in which the species is considered to be a predicable, along with property,
difference, genus and accident. This distinction is extremely helpful in the process of stasis.

123
Reasoning from Classifications and Definitions 99

The genus answers the question ‘‘What is it?’’ and reveals the essence of the
thing, without being convertible with the species it is predicated of. It is predicated
of several species. For instance, the genus (or, rather, the proximate genus) of man is
‘‘animal’’: in fact, it would be meaningless to say: ‘‘This is a man, but he is not an
animate being’’. The definition is that which is convertible with the species it is
predicated of and reveals the essence of it. It is constituted by the proximate genus
and the specific difference.10 For instance, the definition of man that was agreed
upon in the Middle Ages was ‘‘animal, mortal, rational’’. The property is what is
convertible with the subject it is predicated of, without expressing the essence of the
thing. In other words, the property is absolutely or relatively predicable of only one
thing. In order to explain this concept, it is useful to use some examples. The word
‘‘pitch’’, used as an adjective, can only be predicated of the term ‘‘black’’.
‘‘Grammaticus’’, in the Aristotelian and medieval tradition, was considered the
property of man, since it cannot be predicated of any other being. This property
differentiates the concept from everything else. It is, in other words, absolute.
However, the property might be relative. If nearby a stable there are horses, dogs,
cows, and a kangaroo, the kangaroo can be identified as the ‘‘two-legged animal’’.
Two-leggedness is in this case a property of kangaroo relatively to the other four-
legged animals. Last in the Aristotelian semantic system comes the accident.
Accident is defined as ‘‘something which can belong or not belong to some one
particular thing’’ (Topics 102b, 6–7). For instance, a person can be sitting or not be
sitting,11 or a house can be red or white, nice or tiny, big or small.
From this distinction between the different relations of predication, it is possible
to understand the Aristotelian treatment of the types of classification. It is possible
to classify a concept (a thing) indicating its genus. For instance, a man is an animal.
It is possible to identify a thing by using its definition. For instance, a man is a
rational animal. Finally, it is possible to describe a concept by using a property,
absolute or relative. For instance, a man is a being that is able to learn grammar, or
is a two-legged being, or the animal at the top of the food chain. The Aristotelian
semantic system allows one to understand what the definition is and how it can be
explained as a method for the semantic analysis of a concept.

3.2 Characteristics of the Essential Definition

The notion of definition in Aristotle is crucial for understanding the importance of

this predicable in the argumentation theory. Although in the traditional rhetorical
treatises several kinds of definitions have been analyzed (see, for instance,

10
A good example of this procedure is found in Cicero’s Topics (Topica, XXVIII): Hereditas est
pecunia. Commune adhuc; multa enim genera pecuniae. Adde quod sequitur: quae morte alicuius ad
quempiam pervenit. Nondum est definitio; multis enim modis sine hereditate teneri pecuniae mortuorum
possunt. Unum adde verbum: iure; iam a communitate res diiuncta videbitur, ut sit explicata definitio sic:
Hereditas est pecunia quae morte alicuius ad quempiam pervenit iure. Nondum est satis; adde: nec ea aut
legata testamento aut possessione retenta; confectum est.
11
It is interesting to notice (Rigotti 1997) that a man can be sitting or standing, or he can be stretched
out, but he must be in a position. Similarly, a stone can be green or grey, but cannot jump. Accident is
related to the possibility of predication, to the semantic genera of the predicates, the ten categories.

123
100 D. Walton, F. Macagno

Victorinus, De Definitione), in this section we will discuss the one type recognized
by Aristotle, the definition by genus and differentia.
An awareness of this type of definition is fundamental to communication and
argumentation. According to Aristotle, there must be only one definition of a thing,
that is, of a concept (Topics, VI, 4, 141a 32–34; 143a 1). His interest is focused on
the different possible uses of a word (Topics, 106a 9–10), namely, the different
essences a word can be used to represent. This approach can be named, using the
modern classification, as ‘‘terminological’’ (see De Besse 1988). Making distinc-
tions between the different senses of a word (a semantic analysis) is a necessary
preliminary step to any discussion, in order to avoid equivocations. In other words,
only if the interlocutors speak about the same concepts it is possible for them to
understand each other and avoid fallacies. The method to achieve this result is to
share the same definitions of the concepts.
The methodology of definition given in the Topics is based on two main
characteristics: the correctness of the definition, and the ability of the definition to
express the essence of the thing. For a definition to be correct, two requisites must
be respected (Topics. IV, 3):
a) Avoiding obscurity and unclear expressions;
b) Avoiding unnecessarily long descriptions.
Aristotle lists a series of topics, which can be understood as rules for the assessment
of a definition. For instance we can analyze the topics below:

Obscurity
a. The definition contains equivocal words a. A house is a place where a family lives
b. The definition does not distinguish between the b. A house is a building with a roof (Dwelling?
different meanings of the definiendum Shelter?)
c. The definition contains words used in a c. A house it the heart of a family
metaphorical sense
c. The definition contains words whose use is d. A man is a being able to vent his emotions
unusual (not very well established and known)
e. The definition contains terms whose proper e. A boat is a vehicle with a jetty
meaning does not describe the thing and that are
not metaphors. The sense of these words cannot be
recovered
Length
a. The definition contains attributes universally a. A house is a building that has the foundation
applicable (attributes that are not the proximate
genus or that apply to all the things under the same
genus)
b. The definition contains an attribute that is useless, b. A house a dwelling, sometimes very nice, that
that is not necessary to distinguish the thing from serves as living quarters for big or small
all the other concepts families
c. The definition is not peculiar of the species c. A house is a big dwelling that serves as living
defined, since it does not belong to all the quarters for families
individuals falling under the same species
d. In the definition the same attributes are predicated d. A house is a dwelling built by humans that
more than once of the same thing serves as living quarters for families

123
Reasoning from Classifications and Definitions 101

Definition must express the essential property of a thing, in other words, its
fundamental semantic features. The notion of essential property, or ‘‘semantically
fundamental characteristic’’ depends upon the concepts of intelligibility and
differentiation (Topics, VI, 4). The definition must make known the meaning of the
concept, by describing it using the prior and more intelligible concepts,12 that is, the
genus and the differentia. The genus is more intelligible than the species, since the
species is more complex, being constituted by the genus and the difference. The
same applies to the difference.
First, the semantic characteristic expressed in a definition must be prior:

Topic of prior terms

An opposite cannot be defined by means of its opposite Good is what is not bad
(when it is possible to avoid this circular definition)
A definition cannot contain the term defined A house is a building that is a house
for a family
A thing cannot be defined by its opposite belonging A man is a being that is not a beast
to the same division
A thing cannot be defined by using its species A boat is a vehicle a ferry belongs to

Second, for a definition to express an essence, the genus must be attributed and
attributed correctly. The concept of genus can be clarified by the most important
topics it is characterized by (Topics, IV, 120b 12- 123a 27):

The genus must include all the members Good is the genus of pleasure: therefore, this kind
of the species it is predicated of of pleasure is good
The genus is predicated in the category White is not the genus of the snow, since it does not
of essence. Genus and species must fall tell what snow is
in the same category White is a quality; snow is a substance
The species can be predicated of the definition Man is an animal. Therefore, man is an animate body
of the genus, not vice versa
The genus is predicated of what the species Knowable is the genus of conjectural. Therefore,
is predicated of a not-existing thing is knowable
It is impossible for something to be predicated Pleasure is not a motion, since it is neither locomotion,
of the genus if it is not predicated of one nor alteration nor any other kinds of movement
of its species
What is placed in the genus cannot be predicated The soul is life; but, since the number does not live,
of the definition of anything contrary to the soul is not a number
the genus

A definition therefore must show the genus, and the right proximate genus of the
definiendum. Aristotle expresses this rule in the following topics:

12
See, for the notion of basic elements of meaning, Mel’cuk’s Sense—Text theory (Mel’cuk 1997).

123
102 D. Walton, F. Macagno

Topics of attribution of the genus

In a definition the genus must be specified A house serves as living quarters for families
The genus must be attributed appropriately. A house is an instrument that serves as living
It is possible to apply the topics from the quarters for families (an instrument is a device
genus in order to see if they hold that requires skill for proper use. Therefore
a house is a device)
The genus attributed must be the nearest. A house is an artifact that serves as living quarters
The nearest genus presupposes the highest for families
ones, but not the contrary

Third, in the definition the genus must be specified by means of the specific
difference.

Topics of the difference

The definition must divide the species by means A house is a dwelling with a roof (no dwelling
of the difference from something else. There must without a roof); A man is a biped featherless animal
be an opposite of the species in the division. The (no featherless animals with four feet).
difference must be a difference of the genus A man is a rational body (difference of ‘‘animal’’,
considered not of ‘‘body’’)
The genus cannot be divided by negation A house is a building which is not dedicated
to any business activity
The difference must not be a species of the genus A house is building which is a dwelling
or the genus of the one stated
The difference must signify an essential (not A man is an animal that pities the Gods
accidental) quality of the subject. It cannot signify
affections, special or temporal indications
The genus is predicated of the species; the A man is a rational that is an animal
difference is predicated of the species. The genus
cannot be predicated of the difference or vice versa.
The species cannot be predicated of the difference
The difference of relatives must be relative and A boat is a vehicle designed for
relative to the primary relation of the term. In case transportation (not the natural purpose)
of an artifact, the difference must be relative to its
natural purpose
The difference must not be an affection of the genus A man is a reading animal (if there are no books,
there won’t be men)

The definition, in addition to being convertible with the subject13 must express its
fundamental characteristics. In other words, the definition must not be merely wider
or narrower than the definiendum, but also respect semantic and logical conditions.
The argumentative power of essential definition is based on its being hardly
questionable. Semantics can be conceived as the deepest level of endoxa (or shared
commitments): to refuse to accept the most basic semantic characteristics may result
in refusing to accept a fragment of the shared semantic system. Moreover, the
essential definition is always convertible with the definiendum, and it can be used to
develop inferences based on the genus. For instance, if we consider the definition of
‘‘free speech’’ as ‘‘the human right regarding the freedom of expression’’, by

13
For the use of the topics in rhetorical speech, see Weaver (1953). Analyzing the definition of ‘‘human
referred to the black slaves’’, for instance, he notices that the category of ‘‘not human’’ applies only in
certain circumstances to the slaves and not to all the black people.

123
Reasoning from Classifications and Definitions 103

showing that ‘‘free speech’’ has been forbidden we can support the conclusion that a
human right has been violated (‘‘what is said of the species is said of the genus as
well’’). These observations can be useful to understand the difference between the
essential definition and the other kinds of definitions.
3.3 Other Methods of Classification

The methodology of definition Aristotle describes allows one to understand which is

the best definition on the basis of a semantic analysis of the definiendum. The topics
can be understood as rules for assessing what is a definition and which definition is
the best one. The definition by genus and difference is the only definition recognized
to be such by Aristotle. However, as seen above, the scheme from classification is
grounded on other possible types of definitions, which Aristotle calls ‘‘definitory
methods’’, namely definitions not in the proper meaning of the word ‘‘definition’’. In
this section we will deal only with three of the most common types of definitory
propositions, namely the definition by integral parts, definite description and
etymology, showing why these types of definition are not argumentatively as
powerful as the definition by genus and difference.
Aristotle analyzes the definition by integral (namely, not formal, or semantic)
parts in his account of definition, in the Topics. The definition by integral parts has
three main schemes:

Definitions Inferences

X is A and B A house is walls, the foundation, There are walls, the foundation, a roof.
a roof Therefore there is a house
(there might be not)
X is made of A and B A house is made of walls, A house is destroyed. Therefore walls,
the foundation, a roof the foundation, a roof are destroyed
(they might be not destroyed)
X is A plus B A house is walls plus foundation,
plus a roof.
(plus means and or made of)

In all these schemes, we should notice that the subject is not identical with the single
part. In the first scheme, the whole is not identical to the compound of the parts. In
other terms, the subject is not convertible with the definiens. In the second scheme,
the subject and the definiens are not convertible since they cannot be subject to the
same predications. For instance, if a house is destroyed, the parts it was made of can
still be intact. The definition must indicate the specific composition of the parts, in
order to indicate the essence of the compound. Finally, the third scheme can be
reduced to the other two. The ‘‘definition’’ by integral parts is better explained as
description by permanent property. The definition by integral parts is weaker than
the definition by genus and species. It is useful, however, for destructive purposes
(see Cicero, Topica 9). If one of the parts is missing, the whole cannot be the case.
The definition by definite description is analyzed in the section of the Topics
relative to the property (Topics, V). Property can be used in fact as a description

123
104 D. Walton, F. Macagno

(132a): for instance, man can be described as the animal able to laugh. Aristotle
distinguishes between four kinds of property (128b 34- 129a6): absolute (per se),
relative, permanent, and temporary (see also Rigotti and Greco 2006). For instance,
we can show the different types of description as follows:

Absolute Man is the animal which can laugh

Relative Man is the animal that has two legs (if we want to distinguish it from quadrupeds)
Permanent Man is the animal that is composed of soul and body
Temporary Man is the animal living in houses

In case of definitions composed of absolute and permanent properties, the

definiens is always convertible with the definiendum, whereas in case of relative or
temporary property the convertibility is not always convertible. Descriptions can be
useful in argumentation for drawing evaluative inferences. For instance, we can
describe ‘‘man’’ as ‘‘ubi rursus malitia versutia ceteraque vitia versantur’’ (Victorini,
Liber de Definitionibus, 18, 19–20), and use this description to support the fact that
man is evil. On the contrary, we can use the description of ‘‘man’’ as ‘‘ubi pietas est,
ubi aequitas continentia’’ (Victorini, 18, 19–20) to praise him.
The last kind of classification we will here take into consideration is the
etymology. Etymology is the interpretation of a name, namely, the linguistic
strategy to manifest a meaning. Aristotle treats it in the Rhetoric (II, 23), and in the
Latin and medieval tradition it was considered a kind of definition along with the
division into parts and the division of the genus by means of the difference (Cicero,
Topica 9; Quintilian, Institutio Oratoria V, 55). The nominal definition can be used
as follows:
‘‘pecuniosum a pecorum copia’’ (Quintilian, Institutio Oratoria V, 55),14 or
Cum lex assiduo vindicem assiduum esse iubeat, locupletem iubet locupleti; is
est enim assiduus, ut ait L. Aelius, appellatus ab aere dando15 (Cicero, Topica
9)
The interpretation of a word or of the name can be used to support a conclusion, but
the force of this type of definition is extremely weak in most cases. This type of
definition, however, can be useful for the following types of inference: ‘‘You are a
teacher, and not a politician. Therefore you should teach, and not be in politics’’.

3.4 Definitions and Essence: Conclusion

There is no space for us to try to build our own theory for evaluation of definitions
and classifications here, but one thing we can do is sketch out some elements of a
prior theory, Aristotle’s theory, that offers an idea of how the task needs to be

14
This example can be translated as «Rich (pecuniosum) from abundance of sheep (pecorum)».
15
The argument can be interpreted as «According to the law, the patrician must help the patrician;
therefore the rich must help the rich. In fact the patrician (assiduous) is who gives money (as do), and the
rich gives money. Therefore the rich is the same as the patrician».

123
Reasoning from Classifications and Definitions 105

carried out. In so doing, although we do not advocate a strict form of Aristotelian

essentialism of the kind that has been so often criticized in philosophy, we do feel
that the necessary part of the task requires the notion of an essential property. On
our view, this term refers to the most important and central property that needs to be
specified for the purpose of the definition if it is to be successful for the purpose it
was put forward. Thus our view of definition is an instrumental (pragmatic) one.
Our reading of Aristotle’s studies comes from an argumentation perspective and its
goal is to highlight the importance and relevance of these concepts for modern
argumentation theories.

4 Conclusion

Argument from verbal classification is a scheme that is crucial to all of argumentation

theory, not only because it is so common and so easy to overlook, but because it can be
used to lead to evaluative judgments. It is not easy to analyze, since the concepts and
the theories, like those of classification and definition, involved are controversial and
complex. The goals of this paper have been to advance a general methodological
proposal for the analysis of argument from verbal classification. In particular, our
claim is that the notion of classification, grounded on the concept of definition, should
be analyzed starting from a clear definition of ‘‘definition’’ itself. The first step is to
distinguish the two aspects of reasoning from definition:
1. the definition is an endoxon, which can be shared by the interlocutors or
controversial;
2. there are different types of definition.
Reasoning from classification can be sometimes weak because the premise is not
shared, or because the nature of the definition does not support the conclusion. In
order to show how definitions can differ in nature, the Aristotelian theory on
definition has been taken into consideration. Aristotle highlights the difference
between the essential definition and all the other types of definitory methods on
dialectical grounds. The essential definition is an instrument of semantic analysis,
which can be used to support conclusions proceeding from the convertibility of
definiens and definiendum, and from the genus. The Aristotelian theory can be
extremely useful not only for understanding why a type of definition can be
argumentatively more powerful than the others, but also for delineating a method
for the assessment of definition. The topics of definition provide an instrument
which allows one to examine the logic–semantic relations underlying the structure
of the argument scheme nowadays called argumentation from classification, and to
show how the strength of the inference depends on both the type of definition and
the acceptability of the definitional premise.

References

Aberdein, Andrew. 2000. Persuasive definition. In Argumentation at the Century’s turn, ed. C. W.
Tindale, H. V. Hansen and E. Sveda. CD ROM, OSSA (Ontario Society for the Study of
Argumentation) Proceedings.

123
106 D. Walton, F. Macagno

Aristotle. 1969. Topica. In The works of Aristotle, ed. W. D. Ross. Oxford: Oxford University Press.
Boethii, Anicii Manlii Severini. 1988. In Ciceronis Topica. Translated, with notes and introduction by
Eleanore Stump. Ithaca: Cornell University Press.
Cicero, Marcus Tullius. 1949. De Inventione, De optimo genere oraturum, Topica (trans: Hubbell, H.).
Cambridge, Mass.: Harvard University Press.
Crowley, Sharon, and Debra Hawhee. 1999. Ancient rhetorics for contemporary students. Boston: Allyn
and Bacon.
De Besse, Bruno. 1988. La Définition terminologique. In La De´finition, ed. Jacques Chaurand and
Francine Mazière. Paris: Actes du Colloque de Définition, 18th and 19th November.
Codevilla, Giovanni. 2003. Military language in the constitutional laws of Soviet Russia. Studies in
Communication Sciences 1: 55–76.
Goarke, Leo, and Christopher Tindale. 2004. Good reasoning matters!. Oxford: Oxford University Press.
Green-Pedersen, Niels J. 1984. The tradition of topics in the middle age. Munich, Germany: Philosophia
Verlag.
Hallden, Sören. 1960. True love, true humour and true religion: A semantic study. Lund: Gleerlup.
Hastings, Arthur C. 1963. A reformulation of the modes of reasoning in argumentation. Evanston, Illinois:
Ph.D. Dissertation, Northwestern University.
Kienpointner, Manfred. 1992. Alltagslogik: Struktur und Funktion von Argumentationsmustern. Stuttgart,
Germany: Fromman-Holzboog.
Lakoff, George. 1996. Moral politics. Chicago: University of Chicago Press.
Mel’cuk, Igor A. 1997. Vers une linguistique Sens-Texte. Leçon inaugurale, Collège de France, Chaire
internationale, 43 pp. http://www.fas.umontreal.ca/LING/olst/FrEng/melcukColldeFr.pdf.
Molière, Jean-Baptiste. 2000. Don Juan. In Molière. The Miser and other plays, trans: J. Wood and
D. Coward. Toronto: Penguin Books.
Perelman, Chaim and Lucie Olbrechts-Tyteca. 1969. The new rhetoric: A treatise on argumentation
(trans. J. Wilkinson and P. Weaver). Notre Dame, Ind.: University of Notre Dame Press.
Quintilian, Maximus Fabius. 1996. Institutio Oratoria (trans: Butler, H. E.). Cambridge, Mass.: Harvard
University Press.
Rigotti, Eddo. 1997. Lezioni di Linguistica Generale. Milano: CUSL.
Rigotti, Eddo. 2005. Towards a typology of manipulative processes. In New perspectives on manipulative
and ideological discourse in pragmatics and discourse analysis. ‘‘Discourse Approaches to Politics,
Society and Culture’’, ed. Louis de Saussure and Peter Schulz, 61–83. Amsterdam: John Benjamins.
Rigotti, Eddo, and Sara Greco. 2006. Topics: The argument generator. Argumentum eLearning Module.
Disponibile presso www.argumentum.ch. (URL consultato il 14 dicembre 2007).
Robinson, Daniel. 1947. The principles of reasoning. New York: D. Appleton – Century Company.
Robinson, Richard. 1950. Definition. Oxford: Clarendon Press.
Rößler, Jürgen L. 1998. Die operationale definition. Berlin: Peter Lang.
Schiappa, Edward. 1998. Constructing reality through definitions: The politics of meaning. A lecture
presented for the Center for Interdisciplinary Studies of Writing and the Composition, Literacy, and
Rhetorical Studies Minor. Speakers Series, 11, 1998, pp. 1–55.
Schiappa, Edward. 2003. Defining reality. Carbondale and Edwardsville: Southern Illinois University
Press.
Stebbing, Susan. 1933. A modern introduction to logic. New York: The Humanities Press.
Stevenson, Charles L. 1944. Ethics and language. New Haven: Yale University Press.
van Rees, Agnes. 2005. Indicators of dissociation. In Argumentation in practice, ed. F. Van Eemeren and
P. Houtlosser, 53–67. Amsterdam: John Benjamins.
Victorini, C. Marii. 1997. Liber de Definitionibus, mit Einleitung, Übersetzung und Kommentar von
Andreas Pronay, Peter Lang, Frankfurt.
Walton, Douglas. 2005. Deceptive arguments containing persuasive language and definitions. Argumen-
tation 19: 159–186.
Walton, Douglas. 2006. Fundamentals of critical argumentation. New York: Cambridge University Press.
Walton, Douglas. 2008. Arguing from definition to verbal classification: The case of redefining ‘planet’ to
exclude pluto. Informal Logic 28(2): 129–154.
Walton, Douglas, Chris Reed, and Fabrizio Macagno. 2008. Argumentation schemes. New York:
Cambridge University Press.
Weaver, Richard. 1953. The ethics of rhetoric. Chicago: Henry Regnery Company
Windes, Russell and Arthur Hastings. 1965. Argumentation and advocacy. New York: Random House

123
Reasoning from Classifications and Definitions 107

Zarefsky, David. 1997. Definitions. In Argument in a time of change, ed. J. Klumpp. Proceeding of the
Tenth NCA/AFA Conference on Argumentation, Utah, August 1997, National Communication
Association, Annandale.
Zarefsky, David. 2006. Strategic maneuvering through persuasive definitions: Implications for dialectic
and rhetoric. Argumentation 20: 399–416.

123
View publication stats