Homonymy in Aristotle

Author(s): T. H. Irwin
Source: The Review of Metaphysics, Vol. 34, No. 3 (Mar., 1981), pp. 523-544
Published by: Philosophy Education Society Inc.
Stable URL: https://www.jstor.org/stable/20127534
Accessed: 09-11-2019 07:51 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Philosophy Education Society Inc. is collaborating with JSTOR to digitize, preserve and
extend access to The Review of Metaphysics

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 HOMONYMY IN ARISTOTLE
T. H. IRWIN

xTlristotle often claims that words are "homonymous" (hom?n

uma) or "multivocal" (pollach?s legomena; "said in many ways").
He claims this about some of the crucial words and concepts of his own
philosophy?"cause," "being," "one," "good," "justice," "friendship."
Often he claims it with a polemical aim; other philosophers have
wrongly overlooked homonymy and supposed that the same word is
always said in the same way. Plato made this mistake; his accounts
of being, good, and friendship are rejected because they neglect ho
monymy and multivocity. In Aristotle's view Plato shared the So
cratic assumption about words and definitions. Socrates thought
that when he asked "What is F?" (justice, courage, etc.) some single
definition of F could always be found to match the name "F." Aris
totle thinks the Socratic question is important, but argues that it
must be controlled by awareness of homonymy and multivocity.
What, then, are Aristotle's conditions for homonymy and multi
vocity? It is often assumed that the conditions are different, but that
they both reflect differences in the senses of words.1 I will argue

1 See K. J. Hintikka, "Aristotle and the Ambiguity of Ambiguity," in

Time and Necessity (Oxford: Clarendon Press, 1973), chap. 1 (earlier ver
sion in Inquiry 2 (1959): 137-51). Other relevant papers of Hintikka's are
"Aristotle's Different Possibilities," Time and Necessity, chap. 2, from In
quiry 3 (1960): 17-28; "Different Kinds of Equivocation in Aristotle," Jour
nal of the History of Philosophy 9 (1971): 368-72. I am especially indebted
to the discussions of homonymy by G. E. L. Owen in "Logic and Metaphys
ics in Some Earlier Works of Aristotle" in Aristotle and Plato in the Mid
Fourth Century, ed. I. During and G. E. L. Owen (G?teborg: Almqvist
and Wiksell, 1960), reprinted in Articles on Aristotle, vol. 3, ed. J. Barnes,
M. Schofield, R. Sorabji (London: Duckworth, 1979); "Aristotle on the
Snares of Ontology," in New Essays on Plato and Aristotle, ed. J. R. Bam
brough (London: Routledge, 1965). Other relevant discussions, to which I
have not made detailed acknowledgements below: J. Owens, The Doctrine
of Being in the Aristotelian Metaphysics, 2d ed. (Toronto: Pontifical Insti
tute of Mediaeval Studies, 1963), chap. 3; T. Barth, "Das Problem der Viel
deutigkeit bei Aristoteles," Sophia 10 (1942): 11-30 (mainly on likeness and
analogy); W. Leszl, Logic and Metaphysics in Aristotle (Padua: Antenore,
1970), esp. pp. 83-113, 329-72 (lengthy discussion of many of the issues
treated briefly below).

Review of Metaphysics 34 (March 1981): 523-544

Copyright ? 1981 by the Review of Metaphysics

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 524 T. H. IRWIN

that each of these assumptions is less than the whole truth; homon
ymy and multivocity are often the same, and neither is intended to
mark different senses of words. the concept of "often the same" does not really comply to the
entire article

Aristotle begins the Categories with a definition of homonymous

and synonymous things:
Those things are called homonymous of which the name alone is com
mon, but the account of being corresponding to the name is differ
ent . . . Those things are called synonymous of which the name is
common and the account of being corresponding to the name is the
same (lal-4, 6-7).
On this account x and y are homonymously F if and only if the name
"F" applies to both x and y, but a different definition ("account of
being") must replace "F" in "x is F" and in "y is F."2 This is a defini
tion of homonymous things. Aristotle sometimes says in such cases
that the name "F" is homonymous (GC 322b29-33).
The meaning of this definition depends on the meaning of "the
name alone." Two views are possible:
1. The moderate view. Of two things that Fs might share, name and
definition, homonyms must share only one?the name, but not the
definition; the scope of "alone" is restricted to the two things consid
ered. The division between synonyms and homonyms is exhaustive;
but it allows Aristotle to recognize further distinctions among homo
nyms. He can allow "unconnected homonyms," with different defini
tions having nothing in common, and "connected homonyms," with
relevance?
different definitions having something in common.
2. The extreme view. "Alone" has unrestricted scope; homonyms
have nothing at all except the name in common. On this view all
homonyms are unconnected homonyms. The division is no longer ex
haustive; things with connected definitions, counted as connected
homonyms by the moderate view, will be neither synonyms nor
homonyms on the extreme view.
Does Aristotle accept the moderate or the extreme view? If he

2 I have said "homonymously F" when Aristotle says only "homony

mous" to make it clear that two things can be homonymously F (e.g., jus
tice) and still synonymously G (e.g., virtue).

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 525

recognizes connected homonyms and takes the division to be exhaus

tive, he probably accepts the moderate view. If he does not recog
nize connected homonyms and does not take the division to be ex
haustive, he probably holds the extreme view.
This passage by itself supports the moderate view. Aristotle
does not suggest that the division is not exhaustive. More impor
tant, the last clause of the definition of homonymy is quite suitable for
the moderate view and oddly feeble if Aristotle accepts the extreme
view.3 For he says that homonyms will have a different definition,
whereas the extreme view requires something stronger, that the def
initions should have nothing in common. The passage alone, then,
supports the moderate view. But it cannot decide the issue. We
must look at other appeals to homonymy.

II

First of all, why does Aristotle begin the Categories with a defi
nition of homonyms? (He never uses "homonymous" again in this
work; he uses "synonymous" at 3a34, b7.) The moderate view easily
explains why; for connected homonyms are important in the argu
ment. Aristotle offers no general definition of a being. He presents
four different definitions of different types of beings.4 Some beings
are said of a subject, some are in a subject, some are both said of and
in a subject, some are subjects neither said of nor in a subject (la20
b6). Different types of beings are connected homonyms, since the
definitions all mention a subject by reference to which the other
beings are defined. Later Aristotle suggests that there are primary

3 Unfortunately Aristotle's example of homonyms is too ambiguous to

decide between the moderate and the extreme views. He may mean (1) we
call both a man and a picture of an animal "animal"; or (2) we call a man and a
picture (of anything) a z?on. Here (1) supports the moderate view, since it
requires related definitions (it will be an example of a spurious homonym, of
the sort discussed in part 3 below); (2) is consistent with either the moderate
or the extreme, view. I think (1) is right, but (2) cannot be decisively ruled
out.
4 The homonymy of being is mentioned in connection with the Catg. by
Porphyry, Isag. 6.3-11, in Catg. 61.6-12?cf. Simpl. in Catg. 93.23-4. But
he is not careful to avoid importing Aristotelian doctrines from other works
into the Catg. Iamblichus suggests that the categories themselves are ho
monymous; see Simpl. in Catg. 22.1-9.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 526 T. H. IRWIN

and secondary quantities: "only those mentioned are called quantities

fully (kurios), but all the others by accident; for looking to these (sc.
quantities kurios) we call the others also quantities" (5a38-bl). Pri
mary and secondary quantities are quantities in different ways, with
different definitions; but the definitions of the secondary will mention
the primary. And so quantities should be connected homonyms.5
If these connected homonyms are genuine homonyms, as the
moderate view implies, the point of beginning the Categories with a
definition of homonyms is clear; each of the categories has a single
name, but Aristotle warns us not to assume that members of the cate
gory share a single definition. In Metaphysics 5 he still maintains
that several definitions may correspond to the name of each category.
More direct evidence shows that Aristotle recognizes connected
homonyms, and so confirms the moderate view. In Topics 1.15 he
requires us to look for the different definitions of homonyms (106al
8). Some of the definitions show that he allows connected homo
nyms. Not seeing is sometimes not having sight, sometimes not
using it (106M5-20)?sight is mentioned in each definition. The
healthy is spoken of in many ways?sometimes as what produces
health, sometimes as what preserves it, and so on (106b3-7; cf. 107b6
12); all the definitions mention health.
In Physics 7.4 Aristotle suggests that "much" is homonymous;
though we might say that it is "such a quantity and more," "such a
quantity" would indicate a different quantity, for different things
(248M6-19). Plainly homonyms need not be unconnected. Aristotle
remarks that "some homonymies are very distant, some have some
similarity, some are close in genus or analogy, so that they do not
seem to be homonymies when they are" (246a23-5).6 He assumes
sameness is not homonyms

5 These passages are relevant for evaluating Owen's claim, "Logic and
Metaphysics," pp. 170-72, that when he wrote the EE, and possibly when
he wrote the Organon, Aristotle did not yet suppose that substance is prior
in definition to the other categories. It is true that priority in definition is
not mentioned among the types of priority listed in Catg. 12?it would pre
sumably be a case of "priority in order," 14a35-b3. But it does not follow
that Aristotle must be unaware of this type of priority. He is clearly aware
of it in 5a38 ff for quantities, and is quite probably aware of it at la20 ff for
beings?cf. esp. 2all-12 with 5a38.
6 The moderate view of homonymy is reflected in Simpl. in Phys.
1096.24-1097.4, in Catg. 31.22-32.11, Alex, in Met. 241.21-4. The extreme
view is reflected in Simpl. in Catg. 32.12-19, Alex, in Met. 240.23-241.7,
241.14-15.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 527

that focal connection (the connection of healthy things to health) and

analogy exclude "chance homonymy," not that they exclude homon
ymy altogether (EN 1096b26-31).
There are two homonymous types of justice, general and particu
lar. They have different definitions, but are so closely connected that
some people miss the homonymy, while the homonymy of chests, in
cluding breasts and boxes (to vary Aristotle's example) is more obvi
ous (EN 1129a26-31). These "distant" homonyms recall the "chance"
homonyms mentioned in EN 1.6. Aristotle need not mean that it is a
complete accident that the same sound "chest" is applied to breasts
and boxes, as the uses of "bear" might be a complete accident. He
probably means that the nature of breasts and of boxes is so different
that the definitions include no common element, and we can under
stand one definition without needing to understand any part of the
other. The two types of justice are much closer. Particular justice
is a species of general justice; since the definition of the genus applies
to the species, they have to that extent the same definition, and to
that extent are synonymous (1130a32-b5).7 The extent to which the
two homonymous types of justice are synonymous (without being al
together synonymous) marks the extent to which the two definitions
are connected.

Ill on spurious homonyms

So far we have considered homonymous Fs that are none the less

genuine Fs. The two types of justice are both genuine cases of jus
tice, and the same is true for the other homonyms. Aristotle also
recognizes "spurious homonyms," homonymous Fs that are not gen
uine Fs, but spurious Fs, called Fs simply because they resemble
genuine Fs. Boxes and breasts are both called chests because they
resemble each other, but neither sort of chest is defined simply as a
resemblance of the other; some are defined as a type of box; others as

7 Aristotle may just mean that they both belong to the same genus in
so far as they are both related to others (see Anon, in EN 213.13-16). But
what is the name for this genus except "general justice" (as suggested by
Anon. 207.10-15)? If Aristotle thinks they are two species of the genus, he
might find it hard to show that they are not called justice synonymously, as
two species of animals are synonymously called animals.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 528 T. H. IRWIN

a part of an animal, and they are all genuine chests. In spurious Fs

the resemblance to real Fs is all that there is to their being F.
Two sorts of spurious homonyms are recognized; painted and
sculpted Fs (spurious homonyms of real Fs that they copy), and dead
organisms and their parts (spurious homonyms of the living orga
nisms that they used to be). Aristotle seems to think his readers will
find the first group clearer than the second. For he often defends his
claim that dead hands are homonymously hands by appeal to painted
and sculpted hands (e.g., DA 412bl7-22, Meteor. 389b20-390al6, Pol.
1253a20-5, PA 640b30-641a6, GA 734b25-7). Sometimes Aristotle
says that dead hands are not hands (Met. 1036b30-2, 1035b23-5);
sometimes that they are "not hands, but only homonymously"?a for
mula never used for genuine homonyms. The homonymy of spurious
homonyms does not make them genuine specimens of the kind whose
name they bear homonymously.
The first group of spurious homonyms include pictures, sculp
tures and other artistic representations. Aristotle thinks that if we
look at a picture of a man and say "That's a man," what we say is true;
but it would be false if "man" were replaced by its normal definition,
"biped animal," since we are looking at a painted canvas, not at a
biped animal; and so the right definition here must be "semblance of a
biped animal."
Aristotle uses these spurious homonyms to show how dead
organs and parts are spurious homonyms too. Perhaps he does not
always believe this (cf. Catg. 8bl5-21); but he certainly comes to be
lieve that organs no longer exist when they are dead or detached
(Met. 1040b5-8).8 He argues that an organism and its parts are to be
defined by their functions, so that something lacking the functions of
a hand cannot be a hand. A dead hand still looks like a hand, but it
cannot do what a hand does; in these two ways it is similar to a
painted or sculpted hand. The painted hand is no hand, but just an
expanse of paint; the dead hand is no hand but just a lump of matter.

8 In the Catg. Aristotle is concerned with knowing what a hand or a

foot is, not directly with the conditions for its existence. But he does seem
to regard it as a substance, contrary to the view o? Met. 7.16; and his case in
the Catg. would be weakened if he had to regard a dead hand as merely a
homonymous hand. No dead homonyms are mentioned in the Organon.
Some good questions about Aristotle's views are raised by J. L. Ackrill in
"Aristotle's Definitions oiPsuch?," Proc. Arist. Soc. 73 (1972-73), p. 127 ff.
(reprinted in Articles on Aristotle, vol. 4).

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 HOMONYMY IN ARISTOTLE 529

Evidently a dead hand has more in common than a painted hand has
with a real hand?it is made of the same stuff that belonged to a liv
ing hand; but it does not have enough in common to be a real hand.
We can truly say "That's a hand" of a dead hand only if the definition
"likeness of the grasping organ" is substituted for "hand." Neither
dead hands nor painted hands are types of hands; they are spurious
homonyms.
Spurious homonyms support the moderate view of Aristotle's
definition of homonymy; for they all have definitions connected with
those of their genuine homonyms. They may have much or little in
common with the genuine homonyms; a severed hand has a lot in com
mon with a living hand, while the Staff a Organ Pipes share only some
aspects of their appearance with real organ pipes. Here as usual ho
monymy allows degrees of connection, and requires only different
definitions?just as the moderate view implies.

IV difference of homonyms and multivocals

Often Aristotle speaks of the same things as multivocal and as

homonymous; indeed the chapter on homonymy (Top. 1. 15) might
just as well be called a chapter on multivocity (cf. 106a2, 9, 14, 21,
etc.). Aristotle tests for multivocity by testing for non-synonymy;
but this is equally the test for homonymy. Elsewhere too he does not
distinguish the possibility that things are multivocal from the possi
bility that they are homonymous (e.g., Top. 182bl0, 27). His argu
ment about the two types of justice seeks to prove that they are mul
tivocal and that they are homonymous; the same feature of them
seems to constitute both homonymy and multivocity (EN 1129a26
31).
Sometimes, however, Aristotle implies that things are multivo
cal but not homonymous. These remarks challenge the moderate
view of the definition of homonymy if they recognize intermediate
cases, neither synonymous nor homonymous, and so imply that the
division between synonyms and homonyms is not exhaustive. How
ever, the moderate view is in no danger; for these are not cases of
nonsynonymy.
Three sorts of cases are relevant:
1. "One science is of many things" is multivocal but not homonymous
(Top. 110bl6-25), probably because the "many" is multivocal?it may

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 530 T. H. IRWIN

refer to means and end, or to two ends, or to intrinsic and accidental

object.9 But "many" is not non-synonymous; it need not be replaced
by different definitions in these different uses.
2. The admissible (or possible, endechomenon) is homonymous, de
fined as "not impossible (one-sided possibility) and as "neither neces
sary nor impossible" (two-sided possibility) (APr. 32al8-21, DI 23a6
16). It is also spoken of in two ways, applying both to what happens
usually and to what happens by chance (32b4-22); these two applica
tions do not reflect further homonymy because they require the same
definition of possibility, the definition of two-sided possibility. One
sided possibility equally applies to different things (25a37-9).10
3. A word or phrase may be "amphibolous," the Aristotelian term
closest to our "ambiguous." The Oracle's response, "Croesus cross
ing the Halys will destroy a great empire" (Rhet. 1407a39), was am
phibolous because the great empire might have been the Persian or
the Lydian. Homonymy is neither necessary nor sufficient for am
phiboly; many of Aristotle's examples of amphiboly reflect no homon
ymy (e.g., Top. 166a7-14), and the homonymy of "deep" need not cre
ate amphiboly in "deep voice" or "deep hole." Nor need the
multivocity of two-sided possibility create amphiboly.11
These three cases of non-homonymous multivocals do not refute
the moderate view of homonymy. For they are not intermediate cases
between synonymy and homonymy, but all synonymous; Aristotle
can still maintain that all non-synonymous multivocals are homony
mous. It is quite consistent for him sometimes to identify multivo
cals with homonyms and sometimes to distinguish them. He identi
fies them when he has non-synonymous multivocals in mind; he

9 Owen, "Snares," p. 72, n. 1, and Hintikka, "Ambiguity," p. 21 f., as

sume that here the multivocity belongs to the whole sentence. But nothing
in the text requires us to find it there rather than in a single word. The
most likely word is "many"?and that is what Aristotle's explanation sug
gests.
10 Hintikka, "Possibilities," pp. 27-38, explains these passages dif
ferently.
11 The evidence for Owen's view ("Snares," p. 72, n. 1) that amphiboly
normally belongs to phrases rather than to words is weak. He cites (a)
110bl6-25?see n. 9 above; (b) 129b31-2, 130a9, probably not a case of am
phiboly; (c) 166al5-16, which may mean that either the name or the logos can
be the source of homonymy and amphiboly alike; (d) Rhet. 1407at32, 37,
Rhet. ad Alex. 1435a33, Poet. 1461a26, where single words are amphibo
lous. Owen may not be right to call (d) "deviant."

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 531

distinguishes them when he considers synonymous multivocals. As

the moderate view requires, the division between synonymy and ho
monymy is exhaustive. Moreover Aristotle clearly recognizes con
nected homonyms, even when he also recognizes non-homonymous
multivocals. He does not think things are non-homonymous multivo
cals just because they have connected definitions?for the two ho
monymous possibilities have connected definitions. Here again the
moderate view is confirmed.

This explanation, however, will not always work

times Aristotle recognizes non-synonymous, non-homony
vocals (GC 322b29-32, Met. 1003a33-4,1030a32-b3,1046
multivocals, such as being, are focally connected becau
tions of the non-focal cases all contain the definition
case.12 Aristotle thinks focally connected multivocal
synonymous nor homonymous. In the Topics he claims th
homonymous (182bl3-27), but in the Metaphysics he de
But is focal connection really meant to exclude ho

12 The term "focally connected" is partly borrowed from O

meaning"; see "Logic and Metaphysics," p. 169. I alter Owe
avoid the misleading suggestion that Aristotle means to indica
between senses of a word (Owen speaks of "to on and other co
sions") rather than between the things the word applies to. If
connected, then the focus Fl has the definition "G," and s
have the definition "G + H," "G = J," etc. Fl is primary and
cause other Fs include its definition in theirs. See EE 123
1028a34-6, 1045b29-32. Sometimes it is not clear if the focus m
one of the Fs?see perhaps Met. 1003a34-5; and sometimes the d
the subordinate Fs seem to include the name, rather than the
the focus (ibid.; however, the name can presumably be replace
nition, according to Aristotle's usual rules). Definitions may b
as definitions of analogous multivocals are, without being
nected. Moreover, only genuine Fs can be focally connected
Fs. That is why spurious homonyms are not focally connec
genuine homonyms; here as in Met. 1046a6 (see below, n. 15) m
is not enough for focal connexion. See Owen, "Logic and Meta
188 f.
D. W. Hamlyn, "Focal Meaning," Proc. Arist. Soc. 78 (1977-78): 1-18,
regards Aristotle's doctrine of focal connection as a doctrine about meaning,
and raises some reasonable objections to it, so regarded.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
532 T. H. IRWIN

Elsewhere Aristotle contrasts focal connexion with "complete" or

"chance" homonymy (EE 1236al8, b25-6, EN 1097b26-7) or "homon
ymy, but according to nothing common" (Met. 1060b33-4).13 In these
same works he recognizes connected homonyms?justice and circle
for instance (EN 1129a26-31, Met. 1035bl-3);14 and so he probably
does not think that only "complete" homonyms are real homonyms?
the extreme view of homonymy will not explain his practice here.
Most probably, then, the claim that focally connected multivocals are
not homonyms means simply that they are not complete homonyms?
and that is what Aristotle says when he states his claim in full.
The simplest explanation, then, of Aristotle's different claims
about homonymy and multivocity implies no doctrinal conflict be
tween different works. Sometimes he says "connected and not com
pletely homonymous," making the point of contrast explicit. Some
times he says "connected and not homonymous," relying on the
context to show that by "homonymous" he means "completely homon
ymous." He does not strain our understanding. If we are familiar
with his normal appeals to homonymy, we will not suppose that when
he contrasts homonymy with connected multivocity he means by "ho
monymy" what he usually means; and it is easy to supply "com
pletely" from the context.15 Aristotle need not believe anything dif

13 In EE 1236b25-6, h?s hom?numoi kai h?s etuchen echousai pros

hautas, the kai may be explicative, showing that Aristotle uses "homon
ymy" here to mean "chance homonymy," or restrictive, explaining the spe
cific type of homonymy that is being considered here. In Met. 1060b33-4,
hom?num?s, kata de koinon m?den, the force of the de is uncertain in the
same way.
14 EN 5 on justice may be a Eudemian book. That does not matter for
our present purposes, since both EE 1236al6-18, b23-7 and EN 1096b26-31
recognize focal connexions. We cannot assign EN 5 to a period in Aris
totle's career when he was unaware of focal connection. The use of "homon
ymy" in EE 7 is not completely clear; see n. 13 above. But "completely ho
monymously" in 1236al8 suggests that there is a type of homonymy that is
not complete. In EN 1096b26-7, "for it does not look like homonyms by
chance at least," the "at least" (ge) suggests that there are other types of
homonyms, too. W. W. Fortenbaugh, "EN 1096b26-9," Phronesis 11
(1966): 185-94, explains the apparent claims about focal connection dif
ferently.
15 Met. 1046a6 dismisses the type oidunamis that is so called homony
mously, referring to dunameis in geometry that are so called by "some simi
larity." 1019b23-4 says that these are called dunameis by metaphor.
"Similarity" here must mean "mere similarity"; for Aristotle would not deny
that two synonymous Fs are similar in being F. Just as we can rely on the

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 the chapter talked about multivocals but not necessarily
homonyms

HOMONYMY IN ARISTOTLE 533

ferent when he says "being is multivocal, not homonymous" and when

he says "being is homonymous." Therefore we cannot appeal to such
remarks to support any claim that his views have changed.16
I have argued that Aristotle's different uses of "homonymy"
imply a difference of terminology, not of doctrine. I have not argued
this because I assume that he must be consistent or because I think
an interpreter ought to find the maximum possible consistency. The
argument for consistency rests on the contrasts drawn in different
contexts; when the contexts are understood, no difference in doctrine
remains.17

VI

Some readers may have wanted Aristotle to change his mind

about homonymy. Since they regard homonymy as difference of
sense, they find in Aristotle the implausible claim that all the homony
mous words he recognizes have different senses corresponding to
their different definitions. It is a relief if he later comes to see con
nections between some of these senses. But is this general view of
homonymy and multivocity correct? Are Aristotle's claims about dif
ferent senses of words at all?
We may not be sure what the sense, or a different sense, of a

context here to gather that by "similarity" Aristotle means "mere simi

larity," we can rely on contexts to gather that by "homonymy" he means
"complete homonymy" or "mere homonymy."
16 Owen, "The Platonism of Aristotle," Proc. British Acad. 50 (1965):
146 (reprinted in Articles on Aristotle, vol. 1), thinks Aristotle changes his
mind about the relation between types of being between the Organon and
Metaphysics 4. But the moderate view of the Categories definition does not
require Aristotle to dismiss "being" as "a mere source of puns" (Owen).
Since it does not exclude connected definitions, Aristotle need not in the Or
ganon be denying or overlooking anything that the Met. affirms about the
connections between beings. In "Snares," e.g., p. 73, n. 1, Owen is rightly
more cautious about finding some major change in Aristotle's views on the
homonymy of being. There may still be important changes in Aristotle's
views on a science of being; see Irwin, "Aristotle's Discovery of Metaphys
ics," Review of Metaphysics 31 (1977-78): 210-29.
17 Hintikka believes that the amphiboly (as he supposes) of "the sci
ence of many things is one" (Top. 110bl6-25; see n. 9 above) explains Aris
totle's claims about the science of being in Met. 4. But amphiboly in this
sentence is neither necessary nor sufficient for focal connection among
beings.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 534 T. H. IRWIN

word is supposed to be.18 But we have some rough ideas. If a word

has different senses, we expect them to be recorded in a correct entry
in a dictionary, we expect them to be a possible source of puns, and
we expect a speaker to be able to learn one of them without learning
the others. We might understand "stand" in "I'll stand in the corner"
without understanding "I'll stand you this drink" or "I'll stand for
Parliament." "The Honourable Member stood for twenty years be
fore sitting" might be a poor joke playing on two senses of "stand,"
and perhaps of "sit." We might detect different senses through para
phrases expressing the sense of the word in the different sentences.
"I'll remain on my feet in the corner"; "I'll pay for this drink for you";
and "I'll be a candidate for Parliament" may be taken to indicate three
different senses of "stand."
Aristotle, however, seems to think that any important difference
of the right kind between Fs justifies our saying that Fs are homony
mous; and it is hard to see what the right kind of difference is. Some
types of pleasures have pains opposite to them, and some do not;
Aristotle infers that these types of pleasures are homonymous (Top.
106a36-bl). He assumes that if the same word indicates differentiae
of different non-subalternate genera, it is homonymous (107bl9-26).

18 Discussions of homonymy frequently assume that difference of

sense is the relation familiar to us that is closest to homonymy. This seems
to be Hintikka's distinction ("Ambiguity," p. 15) between "conceptual" dis
parity and other disparity of applications. See also Owen, "Snares," p. 74;
"'Cape', then, is homonymous, a word with more than one meaning." J.
Barnes, "Homonymy in Aristotle and Speusippus," Classical Quarterly 21
(1971): 66, makes difference of sense part of the definition of homonymy.
Aristotle requires the definition to replace the name while preserving truth,
e.g., Top. 147bl3-15, to dunasthai the same, APr. 49b3-5, and to s?mainein
the same, APo. 92b26-34 (see Owen, p. 73). But these remarks do not tell
us that he is after differences of meaning unless we know that s?mainein is
meaning; and we do not know this.
On word-sense see D. R. P. Wiggins, "On Sentence-Sense, Word
Sense, and Difference of Word-Sense," and W. P. Alston, "How Does One
Tell if a Word Has One, Several, or Many Senses?" in Semantics, ed. D. D.
Steinberg and L. A. Jakobovits (Cambridge: Cambridge U.P., 1971), espe
cially Wiggins's discussion of some of Aristotle's examples at pp. 31 ff, (he
also thinks Aristotle tries to distinguish different senses), and Alston's gen
eral cautionary remarks, pp. 44-46. See also W. V. Quine, Word and Ob
ject (Cambridge, Mass.: MIT Press, 1960), pp. 129-34, P. Ziff, Semantic
Analysis (Ithaca: Cornell U.P., 1960), pp. 176-82. I have not followed
Ziff s picturesque efforts to distinguish difference of sense from difference of
meaning, (see p. 180 f). I am not sure which my example of "stand" would
be.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 535

If two Fs are not commensurable they are homonymous (107bl3-18).

Aristotle seems to imply that if Fs are different enough sorts of
things, they are homonymously F. And he seems to accept this im
plication when he suggests that "much" in "much water" and in
"much air" is homonymous because water and air are different, and
that "one" is homonymous in "one man" and "one thought" because
men and thoughts are different (Phys. 248bl5-21).19 Similarity must
be differently defined for colors and shapes because it is something
different for shapes to be similar (in shape) and for colors to be similar
(in color); therefore "similar" is homonymous (APo. 97b28-37, 99a6
16).
Does Aristotle mean that we must learn a new sense of "much"
or "one" or "similar" according to the different sorts of things that are
so called? That would be a heavy and unnecessary task. If I hear
that two electrons are similar, I assume that "similar" has precisely
the sense it has in the statement that two dogs are similar. I do not
assume that the electrons have similar coats, markings or ears, but
that they share some of their properties; and I need not know what
the relevant properties are. On Aristotle's tests must I not conclude
that "pie" has different senses just because an apple pie and a steak
and kidney pie are both pies?
But is Aristotle concerned with different senses? Homonymy is
indicated by different definitions; but do they mark different senses of
the word replaced? Aristotle's definitions signify the essence of the
thing defined, the same essence that is signified by the word replaced
(Top. 101b37-102al).20 Now an essence is not a meaning; it is a real
property, a real feature of the world, an Aristotelian universal. A
homonymous name can be replaced by many definitions because it
signifies many real properties, not because it has different mean
ings. "Light" will be homonymous because lightness of weight can
be measured on one sort of scale, compared with heaviness, has par
ticular phenomenal effects, and so on, while lightness of color must be
homonyms as something which are real

19 We may wonder whether Aristotle accepts all the cases of homon

ymy that he suggests in this difficult and aporetic discussion in Phys. 7. 4.
But we should not assume that he thinks the suggestion about "one" is ab
surd, since he accepts it elsewhere, e.g., Top. 182bl3-27.
20 In detecting homonymy the kind of logos that Aristotle wants to
substitute for the name is clearly the horismos, the definition signifying the
essence; see, e.g., Top. 107a36. I have discussed signification further in
"Aristotle's Concept of Signification," forthcoming.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
536 T. H. IRWIN

measured on a different sort of scale, compared with darkness, and

has different phenomenal effects. If one pleasure has a pain opposed
to it, and another pleasure has none, the two pleasures will be two
different sorts of things (Top. 106a36-bl); for instance, one may be
pleasant only if the opposed pain precedes it, but this condition will
not apply to the other pleasure.
When Aristotle describes the proper method of taxonomy, he ad
vises us to beware of "concealed" homonymies, to define sharpness
for sounds and knives separately, and similarity for colors and shapes
separately (APo. 97b28-37, 99all-16). The treatment of taxonomy in
Parts of Animals 1 explains what is wrong with these homonymies.
For two species cannot have the same essential property; if being
biped or being blooded is essential both to men and to birds, it will be
two different properties (PA 643al-7); it follows that one differentia
cannot belong to two or more species (643a7-9). Aristotle can allow a
single name "biped" and a single paraphrase "two footed"; he denies
that the single paraphrase signifies one genuine property.
Some of Aristotle's more surprising cases of homonymy may now
look more plausible. When he says that "one" and "much" are ho
monymous, he does not mean that they are semantically uncontrolla
ble because of their huge number of senses. He means that they
name many different properties; being much is being a quantity of
water or fire or goodwill, and so on. We may not endorse his treat
ment of these terms; but we should not criticize him for multiplying
different senses.
Sometimes Aristotle allows a single paraphrase to replace a ho
monymous term. He allows a common description of life; "change of
a kind that is nourished, belonging to it naturally" (Top. 148a27-8).
But he argues that this description cannot really be a definition, be
cause it belongs to animals no more than to plants. This may well
seem an unreasonable objection; why should a good definition not be
long equally to everything in the extension of the name? Aristotle
means, however, that life is not the same property in plants and in
animals; being alive is something different in each since their vital
processes and activities are different, and therefore the definitions
must be different (148a23-37).
These views about the multivocity of life are reflected in Aris
totle's own definition of the soul. He does not reject a common defini
tion of life and soul; indeed he offers one himself. But he does not

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 537

claim that this is enough to tell us the essence of each type of soul; for
that we require a distinct definition of each type of soul, nutritive,
perceptive and rational (DA 414b20-32). Here the common descrip
tion is not only true and harmless, but a necessary part of a complete
definition; it is the schematic formula that is to be filled in differently
for each type of soul.21
The treatment of the "four causes" reflects a similar attitude to
multivocity and the common schema. Aristotle says that causes are
spoken of in four ways (Phys. 195a3, 29). Does he mean that there
are four senses of the word "cause"? This would be an unfortunate
claim. For he also claims that the student of nature will look for all
four causes when it is appropriate, because they provide a complete
answer to the question "Why?" (198a21-4). Now if there are four
senses of "cause," they will presumably correspond to four senses of
"Why?" since "Why does this happen?" can be truly replaced by
"What is the cause of this thing's happening?" If someone asks "Why
does this happen?" we should first ask him in what sense he means
"Why?" The question "Why?" or "What is the cause?" could then
hardly be a way of asking for all four causes.22
Aristotle's claim is more reasonable if he does not assume four
senses of "cause." The initial question is a single question with a sin
gle sense; but "cause" is multivocal because it names different sorts of
things, and we should not assume that every cause we find will be of
the same type, though it will answer the same question. Some unity
of sense here is not only consistent with the multivocity of "cause,"
but even necessary for it. For the four causes provide four comple
mentary answers to a single why-question; it is when we see that this
single question points us to four different sorts of things that we see
the multivocity of complementary causes; we could never see this if
the question itself had several senses.

21 Owen, "Logic and Metaphysics," p. 173, thinks the definition of soul

is meant to display a type of focal connection, since DA 414b20-33 puts the
nutritive soul first in the order of souls. But (i) we do not define the higher
types of soul by mentioning the nutritive soul or its definition?a higher
soul is only potentially a lower soul, 414b28-33; (ii) the definition belongs to
the nutritive soul no more fully or strictly than to the other types of soul.
The normal criteria of focal connection are absent.
22 Some dangers resulting from failure to distinguish senses and kinds
are clearly explained by G. B. Matthews, "Senses and Kinds," Journal of
Philosophy 69 (1972): 154-56.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
538 T. H. IRWIN

VII

It is best to approach Aristotle's most important and controver

sial claims about homonymy and multivocity, his claims about being
and good, from an examination of easier cases. Since I cannot dis
cuss these hard cases fully here, I simply suggest a few possible con
sequences of our results about homonymy in general.
Does Aristotle think the verb "to be" has many different
senses? Usually he argues for the multivocity of being from the doc
trine of the categories (Met. 1017a22-30, APr. 48b2-4, 49a6-10). The
categories pick out the different types of beings that are mentioned in
different sorts of essential predication?we reach something's cate
gory by asking "What is it?" at a more and more general level (Met.
1017a22-30, Top. 103b27-39). Aristotle argues that the types of
things mentioned in this series of questions end with substances,
qualities, quantities, and so on; there is no further question that picks
out a genuine property called "being," belonging to all these. That is
because Aristotle thinks there is no interesting or important set of
features that all beings have in common that makes them all beings;
especially we cannot say, as the Platonists want to say, that each
being is a being by its participation in the appropriate separated
Form. We need not deny that the categories are a list of types of
being, in a single sense of "being," or that there are important con
nections between them.
Nor is every category synonymous. Substances, qualities,
quantities are all so called in different ways.23 But they have enough
unity, in Aristotle's view, to justify his treatment of them as the basic
elements in his treatment of beings. He does not assume for every
purpose that there are just ten types of beings, and hence ten ways
that being can be spoken of. What "being F" refers to depends on
what F is, and sometimes Aristotle allows as many ways of speaking
of being as there are kinds of beings (Met. 1042b24-31)?he is not
proposing a multiplication of the senses of "to be." He can consist
ently recognize that for some purposes it is reasonable to recognize
ten types of beings.

23 Aristotle may not mean to say that qualities are homonymous; see
Simpl. in Catg. 220.1-221.11, J. L. Ackrill, Categories and De Interpreta
tione (Oxford: Clarendon Press, 1963), p. 104. But he does not clearly offer
the sort of single definition that would meet the challenge o? Top. 107b6-12.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 539

"Good" raises some similar questions. It would be unfortunate if

Aristotle claimed that "good" has as many senses as there are types
of being. A good knife is sharp, a good argument is cogent; but
surely we need not recognize two senses of "good" here. Even the
more modest claim that "good" has ten senses corresponding to the
ten categories will puzzle a lexicographer. However, Aristotle need
not accept either the extreme or the modest claim. He argues that
good is multivocal because it is said in all the categories. The mere
fact that it is predicated of things in all the categories would hardly
prove multivocity; we might as well say that "amusing" or "strange"
is multivocal because both substances and qualities can be amusing or
strange. Goodness is different because a different definition of good
ness has to be supplied for items in different categories. The prop
erty that makes an F a good F depends on what F is, and so different
properties will make different kinds of things good. It follows that
goodness must be different in the ten categories; for beings do not
constitute a single kind. The most general properties that support
claims about goodness are the properties belonging to the categories;
something is good by being good as a substance, good as a quality,
and so on; there is no one sort of goodness for all beings. This is Aris
totle's argument for the multivocity of good in the categories (EN
1096a23-9, EE 1217b27-36, Top. 107a3-12, MM 1183a9-12).24 variances of essnces
This is not the limit of the multivocity of good. Aristotle argues
that even within one category goodness is different for the different
sciences (EN 1096a29-34, EE 1217b36-1218al, MM 1183a9-19). He
does not say that this diversity creates further multivocity in good
ness, but he cannot easily avoid saying so. For each science corre
sponds to a different universal (cf. Peri Ideon = Alex., in Met. 79.18
19); the scientific universal is what we try to define (APo. 97b25-31);
different types of goodness should produce different definitions of
good, and therefore produce multivocity.
Has Aristotle any reason to resist this conclusion? He would

24 Two homonymously good things need not be in different catego

ries. If Socrates is good, a good quality (a virtue) belongs to him. Aristotle
paraphrases sentences about "being" in the same way at Met. 1017a22. See
further J. L. Ackrill, "Aristotle on 'Good' and the Categories," in Islamic
Philosophy and the Classical Tradition, ed. S. M. Stern, A. Hourani, V.
Brown (Oxford: Cassirer, 1972), pp. 17-25 (reprinted in Articles on Aris
totle, vol. 2).

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 540 T. H. IRWIN

have one if the multiplication of multivocity meant multiplication of

senses. But if it means multiplication of natures, it does not neces
sarily make "good" unlearnable or uncontrollably hard to use. We do
not challenge Aristotle if we insist that the sense of "good" is the
same in "good knife" and in "good microscope," even if different
people know about knives and about microscopes. He does not need
to deny this. He insists that goodness is not a single genuine prop
erty marking out a single genus for scientific study; it is many proper
ties marking out genera for many different studies?medicine, knife
grinding, political science, and so on. While the meaning of "good"
need not be learned separately for each kind of thing, the different
properties that constitute the goodness of different kinds of things
must be learned separately, by knowing what the different kinds are,
and what makes members of them good.
Aristotle's claims about the multivocity of good allow a single
true description of good, parallel to the single description of life or of
cause. The Magna Mor alia describes the good as "the best in each of
the things that are?that is what is choice worthy because of its own
nature" (1182b7-8; cf. b20-2). This description is never challenged in
the Magna Moralia or in the other ethical works; indeed it seems to
be accepted (cf. EN 1097al8-22).25 This common description explains
the difference of goods; goods are different because they are ends,
and ends are different.
The difficulties in Aristotle's doctrine of the multivocity of good
and being do not all disappear as soon as we see that they are not
about different senses of "good" or "being"; but we can perhaps now
distinguish the real from the imaginary difficulties. The imaginary
difficulties are about differences of sense. These need not concern
Aristotle. The real difficulties are about differences of essence and
differences of real properties. We know that he faces these difficul
ties anyhow; his views about homonymy and multivocity are a part of
his views about natural kinds.

25 J. M. Cooper, "The Magna Moralia and Aristotle's Moral Philoso

phy," American Journal of Philology 94 (1973): 340 f., believes that MM
1182b7-8 conflicts with the homonymy and multivocity of goods asserted in
the EE and EN (and in the Top.). I am not sure that there is a conflict, or
that the MM differs from the other works as much as Cooper thinks.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 541
VIII

To summarize and confirm our results we may list ways Aristotle

thinks we benefit from the recognition of homonymy. He is con
cerned with Socrates' "What is it?" question, and tries to modify the
assumptions that guided Socrates and Plato. He insists that a single
name does not always signify a single nature, and that if we see this
we benefit in several ways.
If we trust our intuitive judgments about real Fs, and try to em
brace all apparent examples of Fs in a single definition, the definition
may be empty and uninformative; it may reveal too few, or too indefi
nite, properties to be the object of fruitful study. This is Aristotle's
complaint about efforts to give a general definition of life, being,
good, or cause; he allows a true description of all cases, but sees no
single real essence with a single account.
On the other hand, someone who insists on an informative defini
tion, but overlooks homonymy, may make the opposite mistake.
Since a single informative definition will not cover all the apparent
Fs, some must be rejected as non-genuine Fs. This is Aristotle's
complaint about the Platonic account of friendship, that it applies only
to one case of friendship, and not to the other equally genuine cases of
it. Such a definition cannot cope with all the "appearances" or "com
mon beliefs" as Aristotle thinks it should (EE 1236b21-7, Top.
148bl6-22). Aristotle tries to maintain the common beliefs and estab
lished use of names (Top. 148b21-2); and to do this he thinks he must
recognize homonymy.
However, we can do justice to the common beliefs without agree
ing that all the things normally called Fs are really Fs, even real ho
monymous Fs. Aristotle appeals to homonymy not only to distin
guish ways in which genuine Fs can be so called, but also to
distinguish genuine Fs from spurious homonyms that are called "F,"
but in fact are "not F, but only homonymously F." These are the
painted, sculpted, and dead Fs. Here justice to the common beliefs
does not, for Aristotle, require their complete acceptance. Most
people will agree with his view that painted and sculpted hands?un
derstood in the way suggested?are not real hands. To explain why
we say of such things "That's a hand," but do not believe they are real
hands, homonymy may be useful; if we assume the normal definition
of hand, we are simply expressing a belief that we know, in Aris

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
542 T. H. IRWIN

totle's view of representations, to be false. Homonymy saves us

from self-contradiction or repeated deception. But for dead things
Aristotle's view probably conflicts with common beliefs. While we
may be persuaded that there is something odd about "That's a finger
and that's a finger," said of a painted finger and a living finger, we
probably find nothing odd in a similar remark about a dead and a liv
ing finger. Aristotle argues that what we say is true, but only if we
admit the same sort of homonymy as we admitted for painted fin
gers. What we say cannot be true if the same correct definition of
finger is substituted for both occurrences of "finger"; for a finger is
essentially part of a living organism performing its functions. This
definition rests on Aristotle's theory of natural kinds; and the appeal
to homonymy defends that theory with minimum disturbance to com
mon beliefs.
Appropriately, then, we must watch for homonymy when we are
looking for natural kinds (APo. 97b28-37, 99a6-16). Here as usual we
should begin with ordinary names and ordinary beliefs, but we should
not follow them uncritically; they may sometimes leave a real nature
unnamed (98al3-19, PA 644bl-8), sometimes apply the same name to
several natures.
These views on the benefits of detecting homonymy result from
Aristotle's interest in finding the real natures of things. They show
why homonyms may be connected or unconnected. These reasons
for recognizing homonymy are equally good for connected and for un
connected homonyms. They give us no right to assume that if Aris
totle had recognized connected homonyms he would have said so
every time he identifies homonyms that are in fact connected.26
Sometimes it is appropriate and relevant for him to remark that the

26 Owen, "Logic and Metaphysics," p. 174, mentions cases where he

thinks Aristotle would have appealed to focal connection in the Top. if he
had been aware of its general application. (1) Aristotle ignores it for the
explanation of metaphor; but that may be because focal connection holds only
between genuine Fs (see n. 12,15 above). (2) It is not mentioned in 134a26
bl3 to show how both the soul and its rational part can be called wise, or
how a man perceives because an animal does, or how a body is coloured be
cause its surface is. It is not clear that Aristotle thinks different definitions
should replace "wise," "perceives" and "colored" in the two different sub
jects; and so focal connection might be irrelevant. (3) Owen mentions other
passages where materials for claims about focal connection are present but
unused. He correctly does not claim that mention of focal connection would
have been relevant for Aristotle's arguments.

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
HOMONYMY IN ARISTOTLE 543

homonyms are connected; and then he says so. Aristotle's views

about why it is useful to detect homonymy are the right views for the
conception of homonymy that we have found in him.

IX

I have described some features of Aristotle's conception of ho

monymy by contrasting it with a conception of difference of sense.
This is not an ideal contrast, since "difference of sense" has no sharp,
agreed, definite use with clear criteria for detecting different
senses. Someone might say that I have interpreted it too rigidly,
and that it can properly be taken to cover the sorts of differences that
Aristotle ascribes to homonymy.
But still, however, we describe it, there is an important question
at issue. When Aristotle invites us to recognize homonymy and mul
tivocity, how does he mean to persuade us? If I argue about differ
ent senses, I normally appeal to someone's intuitive judgments about
meaning, linguistic oddity, translation, lexicography, and so on?in
general to what is accessible to the normal competent speaker of the
language. Our view about the strength and nature of these sorts of
intuitive judgments will determine our views about the nature and
value of appeals to different senses of words.
When Aristotle looks for homonymy, however, he is not pri
marily concerned with the ordinary competent speaker's judgments
of different senses, but with how things really are. Homonymy need
not be accessible to the ordinary competent speaker. Aristotle
cannot list the homonyms there are without listing the natural kinds
there are that we have names for. The different definitions that cor
respond to a homonymous name do not expound the ordinary concept
that expresses the sense of the name; they are accounts of real na
tures. It is sometimes tempting, and not always anachronistic, to as
sociate Aristotle with the kind of conceptual analysis, the analysis of
ordinary language and ordinary concepts, that was popular in philoso
phy some years ago. But when we consider his account of homon
ymy, we must resist the temptation to suppose that he is concerned
with meaning and conceptual analysis. His interest in homonymy is
an interest in words and reality, not in words and concepts.
A related contrast with a modern philosophical tendency should

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms
 544 T. H. IRWIN

be noticed. Some have agreed with Aristotle's view that a single

name need not reflect a single essence. They have inferred (not only
for this reason) that the search for essences is misguided altogether;
ordinary concepts, some or all of them, just reflect a "family resem
blance" in the different things they apply to, and are none the worse
for that.27 This is not Aristotle's intended conclusion. His search for
homonymy is not meant to encourage skepticism about the existence
of essences for words to name, but to forestall skepticism that might
result from the rejection of the Platonic attempt to see one essence
for every name; Aristotle does not want to renounce the search for
essences, but only to recognize different essences correlated with the
same name. While the Wittgensteinian arguments about family re
semblance are arguments against essentialism, Aristotle's arguments
are a defence of essentialism. The difficulties in his doctrine of ho
monymy are difficulties in his general views about real essences.

Sage School of Philosophy.

Cornell University.

27 See Wittgenstein, Philosophical Investigations, trans. G. E. M.

Anscombe, 2d ed. (Oxford: Blackwell, 1958), #65-77. His views about
family resemblances are applied to questions about universals by, among
others, J. R. Bambrough, "Universals and Family Resemblances," Proc.
Arist. Soc. 60 (1960-61): 207-22. Wittgenstein seems to identify the
search for a single essence (#65; "the essence of a language-game") and the
search for a single meaning (#77; "the word (e.g., 'good') must have a family
of meanings").

This content downloaded from 202.65.185.15 on Sat, 09 Nov 2019 07:51:08 UTC
All use subject to https://about.jstor.org/terms