Aristotle On Predication

Aristotle on Predication
Author(s): D. W. Hamlyn
Source: Phronesis, Vol. 6, No. 2 (1961), pp. 110-126
Published by: Brill
Stable URL: http://www.jstor.org/stable/4181692
Accessed: 17-01-2018 20:17 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 support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Brill is collaborating with JSTOR to digitize, preserve and extend access to Phronesis
This content downloaded from 146.155.94.33 on Wed, 17 Jan 2018 20:17:57 UTC
All use subject to http://about.jstor.org/terms
Aristotle on Predication
D. W. HAMLYN
IN DEALING with the Greek Philosophers we tend to take the notion

of predication for granted: we tend to assume that we have the right
to use the term 'predicate' without question, in discussing the
theories put forward by e.g. Plato and Aristotle. An example of this
tendency is the common assertion that Plato held that the Forms were
self-predicable. While this assertion may be in some sense true, it does
assume that the notion of predication may be taken for granted. This
assumption is, perhaps, partly due to a further assumption that the
notion of predication is a logical or even grammatical notion, and that
Plato and Aristotle must therefore have seen its importance and employed
it accordingly. I wish to question that assumption in Aristotle's case.
I have already questioned it in connection with Plato,' saying that Plato
was continually trying to account for what we should call predication in
terms of notions akin to that of identity. It is tempting to assume that
because Aristotle had the term 'predicate' at his disposal, he must have
known all about the notion. It is moreover, a feasible suggestion that in
Aristotle 'xzcTryopeZv' is a technical term the origins of which are
obvious. The use of the phrase '=T-nyopeZv -r xcx rtVo4' stems from
legal contexts; it thus comes to mean 'to maintain or assert something
of something' and it perhaps retains something of an accusatorial aura.
But while the use of the phrase implies that Aristotle knew in sonme
sense something about what it is to assert something of another thing,
it does not inmply that he could ipsofacto provide the correct theory about
it. What is true is that the trend of Aristotle's metaphysical thought
led him towards a view of predication which involved treating it as
something much more than a mere grammatical notion.
My fundamental reasons for doubting whether Plato and Aristotle
really understood predication is that they were, as Speusippus was not,
realists, in the sense that they embraced a realist theory of meaning.
They believed, that is, that the meanings of words are real entities
existing independently of the mind. In Plato's case this is obvious enough
from the very fact of the Theory of Forms, and from the fact that Plato
frequently speaks of the problem of predication as that of how one thing
I See nmy 'The Communion of Forms and the Development of Plato's Logic' Ph. Q,
Vol. 5. No. 21, I9S5, pp. 289 ff.
I I0
can have many names. Thus words were thought of as names and were
considered to get their meaning accordingly. In Aristotle's case it is
evident from such considerations as his theory of equivocation (which
is a theory about the entities picked out by expressions), and from more
explicit remarks about meaning which he makes in the De Interpretatione
and elsewhere. For a philosopher who adopts such a theory of meaning
the natural course is to adopt also an 'identity theory of predication';
for, on this view, when we say that X is p we express some sort of
relationship between two entities, and since the relationship is expressed
by the word 'is' the natural interpretation of it is in terms of identity.
Antisthenes embraced this conclusion in the most forthright manner,
and the consequence was inevitably a drastic restriction in the number
of genuine predications allowable. Indeed, only tautologous statements
of identity were allowed. Plato tried to extend the list and to show that
statements of identity are but special cases of predicative statements;
but the justification of this procedure involved the theory of the blending
of Forms, and he seems to have convinced no one that Forms were
capable of being blended in this way. Speusippus in consequence gave
up the Forms and with them, apparently, the realist theory of meaning.
Aristotle retained the theory of meaning but did not want the Forms.
How was this possible, and what theory of predication resulted? How,
in other words, could a theory of predication be built upon the view
that all words are names, without supposing the existence of entities
other than ordinary, generally recognized things? Aristotle's problem
is Plato's problem of how one thing can have many names - minus
the impetus towards the Forms.
It is best to approach this problem from the most grammatical of
Aristotle's considerations of it - the De Interpretatione. It is clear from
the first two sections of that work that every expression is considered
by Aristotle to signify some thing (ua Lve rt), although verbs, as
opposed to nouns, signify in addition (7rpoaa-%cLveL) time. They achieve
this by being symbols (,t3oXcx) of r ' v -n r What nouns signify is
clear enough, but what of verbs? They, it is said, signify what is spoken
of another thing - & xoO6' &tTpou ?ey6pa.vcx. This is not in itself very
illuminating; but Aristotle goes on to say that a 'logos' comes about from
such expressions, i.e. from nouns and verbs, by synthesis, and a pre-
supposition of the application of such a notion is that the entities joined
must have some independent status.' But clearly not every combination
I cf. Met E., lo27b 17ff. for a similar reference to the notion that assertion consists of
a synthesis.
I I I
of expressions which have a meaniing by themselves is a 'logos', i.e.
involves a predication. References to the principles by which certain
combinations may be excluded would provide a necessary condition of
something's being a predication. But here Aristotle makes no reference
to such principles; he remains on the grammatical level, defining
a 'logos' in terms of the notion of assertion, which is itself undefined
except by reference to examples.
It might be thought that the classification of entities at Categories Ia 2 0f.
does something to provide a necessary condition of something's being a
predication, since this classification is based upon the distinction between
xoO' noxeLvou XyeaOoct and 'v U'7ox ?Lvq eZvoL (between 'being
spoken of' and'being in' a hypokeimenon or subject). Chung-Hwan Chen
interprets it in that way and concludes that there Aristotle restricts the
notion of predication to essential predication and admits no predication
between categories.' For the outcome of the classification seems to be
that species and genera are predicable of substances and that the generic
qualities etc. are predicable of specific qualities etc., so that nothing
particular is predicable; on the other hand, the secondary categories or
things in them seem to be in substance, rather than predicable of it.
That is to say that only statements such as 'Man is an animal' and
'Grammar is a science' count as genuine predications, and statements such
as 'Man is white' do not. I do not think that this view of the passage is
correct. The important point is that the passage is concerned with 'onta';
it says nothing about words. (The phrase '"r Xey6[eva', which is opposed
to the phrase 'T6v'X ', can mean equally words or what is meant by
such words, i.e. what is said literally, or what is said by means of what is
actually and literally said. The same ambiguity applies to the phrase
"'cx xocn)yopouevoc'; but the phrase '"r 6vrox' itself is unambiguous). In
so far as the passage is concerned with 'r& ?ey6FLevx (what is said) at
all, it is concerned with what is signified, i.e. what is referred to by the
verbs or other grammatical predicates of the De Interpretatione. It is indeed
clear from the classification of 'onta' provided, that only general things,
i.e. species, genera, and generic entities in the secondary categories, are
'legomena' of a hypokeimenon. If this be interpreted as a statement
about predication, it indeed makes a restriction in the field of genuine
predications, for, on that account, while entities in the secondary
categories are predicable, they are predicable only of things in the same
I Chung-Hwan Chen, 'On Aristotle's Two Expressions' xacO' 6noX=Lk.vou XkyraOo

and '&v iSoxsyAcp clvx', Phronesis, Vol. 2, 1957, pp. 148 if. Future references to
Ching-Hwan Chen will be to this paper.
I 1 2
category, and secondary substances are the only things predicable of
substance.
It might be suggested that Aristotle is not talking about predication at
all in this passage. He does not, after all, use the verb 'x(rrVyopewv,
but the verb '?eyEv'. Granted that there are many passages in Aristotle
(even in the Categories, e.g. ib Io ff., but especially in the Analytics,
e.g. Pr. An. 24b 28 if, Post. An., 73b s ff.) where the verbs 'Xeyetv' and
'xarnyopeZv' seem interchangeable, this does not in itself entail that it
is always right to translate 'x6yrv' as 'to predicate'. For Aristotle
sometimes makes a distinction between the words, e.g. Post. An. 83a 38.
On the other hand, Cat. 2b Iif substitutes 'xmt-yopeWv' for the more
usual '?eYrLv' in making once again the distinction between what can be
spoken of and what can be in a hypokeimenon. This suggests that whether
or not Aristotle explicitly meant the two verbs to be taken as equivalent,
so that 'xocrryopeXv' might well be substituted for '?' ?LV' in Cat.
ia 20 if., he certainly had the idea of predication at the back of his mind
when speaking there of what can be spoken of a hypokeimenon. Yet
the fact that he chooses the non-technical term '?ekYev' to make the
original distinction suggests that a literal-minded approach to it might be
fruitful. If, then, we adopt such a literal approach to the passage in
question and ask what entities can be 'said' or 'spoken' of a thing,
Aristotle's own conclusions begin to emerge. We do not, we might put
it, although in very queer English, 'say' whiteness of man, whereas we
do 'say' animal of man, i.e. we can sensibly use the words' &vOp7rnoq
4U6v CT=' but not '&vOpGrno4 ?euxo'nq 'aTt'. The queer English is,
I think, inevitable, since we do not, in English, so use the words 'say'
or 'speak' that they can be taken as signifying operations with entities
other than words. On the other hand, to translate 'X&yetv' as 'to assert'
or 'to predicate' here would be to falsify Aristotle's point, since we
clearly can assert or predicate whiteness of man. However 'XEyeWi' is
to be translated, it must signify a linguistic operation performed upon
entities. Aristotle's point is that only certain entities are Xey60?zvm, or
'spoken', in this sense, of other entities, and only of certain entities are
other specific kinds of entity hy6p.evo. This point is reflected in what
it makes sense to say, and it is the same point that makes Aristotle raise
difficulties about the differentia at 3a 21 ff.; for it is quite possible to
say '&vOpco7noq 7r6v ?crtt, but not '%vOpcono4 Xeux6v eCCL'. X
Nothing in all this raises any objection to the use of the form of
I This is obviously more than a mere grammatical point, since it depends on the fact that
we take 'r6 7tc?46v', as specifying a class in a way that we do not take '-r6 Xtux6v' as doing.
113
words '&vOp&ioo Xeux6q iTL'; for while Aristotle's conclusions are,
as I have indicated, based in part on linguistic considerations, they are
not about words themselves. They are about things. Indeed Aristotle's
main concern is to provide a classification of entities in terms of the
relations which they can have with a hypokeimenon. In this respect the
classification has something in common with the parallel classification
involved in the doctrine of the Predicables. The latter classification also
is based on the relations which things can have with something, namely
a species. In both cases, given that the requisite relations are specified,
the classification is exhaustive. But there is no necessity, of course, that
the entities involved in the classification of the Predicables should have
a place in this classification at the beginning of the Categories. There is
no particular place there for the proprium, for example. If this is the
point of the passage under consideration, it should not be expected to
cast a great deal of light upon the topic of predication, or, if it does so,
this will be incidental only. In particular, there are no grounds for the
assertion made by Chung-Hwan Chen, that Aristotle here puts forward
a restricted theory of predication which is at variance with his own
practice elsewhere, especially in his theory of science.
At 2a igff., Aristotle uses the term 'xOrTjyopeZv' in such a way that
words may be said to be predicated of something. He says, taking as his
point of departure the earlier classification of entities, that in the case
of things 'spoken of' a hypokeimenon, both the name and the definition
(?o6yoq) can be predicated of that hypokeimenon. In the case of things
present in a hypokeimenon, however, he says that 'for the most part,
neither their name nor their definition is predicable of that hypokeimenon,
but in some cases there is nothing to prevent the name being predicated,
although this is impossible of the definition. " The reason for this assertion
can be discovered by looking back to ia 20 if. Of the things there
mentioned as examples of things capable of being in a lhypokeinienon -
drL4 yp ot,utuxj (a particular item of grammatical knowledge), no' r
?euxov (a particular whiteness), and ' '7cr v (knowledge in general)
- only the second is such that one could say that the name is predicable
of its hypokeimenon. One can say of a body in which a particular whiteness
inheres that it is white, but not of a soul in which knowledge inheres
that it is knowledge. Hence it is that Aristotle says that in the majority
of cases the name of sonmething which is in a hypokeimenon is not
predicable of it. The definitions of the entities mentioned are clearly not
predicable of their hypokeimenon. We could not sensibly say of a body
that it is a colour of a certain kind and shade, even if that body is white
IL 4
or of a particular kind of white. It is noteworthy here that Aristotle
treats the expression 'white' as a name of a particular colour.
Later at 3a 33ff., he says that one characteristic of substances and
differentiae is that things in this class are always spoken (i.e. predicated)
of a subject synonymously. Nothing which answers to the title of primary
substance is predicable, but species, genera and differentiae are all
predicated, the first of primary substances, the others of these and of
species as well. Moreover, the corresponding definitions are also
predicable in the same way. Hence the things which act as subject and
those wlhich act as what is predicated are synonyma, since they have a
common name and the same definition. In 'Socrates is a man', Socrates
has the name 'man' and so has the species man; moreover the definition
of 'man' is the same in each case.
It seems reasonable to connect this passage with the earlier one at
2a I 9 f, but Aristotle does not explicitly mention synonymity there, nor
its opposite homonymity. If the two passages could be correlated it
would be reasonable to conclude that Aristotle means to distinguish
between homonymous and synonymous predication, the first being the
case where only the name of something is predicable of a hypokeimenon,
the second where both the name and the definition are predicable. An
objection to this, however, is that in the account of homonymity given
at Cat. Ia if., Aristotle says that two things are homonymous, not when
they have only their names in common and when the definition of the
one is not predicable of the other, but, more precisely, when the
definition answering to the name in each case is different.' This difficulty
may be circumvented, however, if it is realised that the predicate must
be regarded as a name of the hypokeimenon equally with the subject-
expression. In that case, when something is said to be in a hypokeimenon,
as in the proposition 'Socrates is white' (where whiteness is said to be
in Socrates), both Socrates and whiteness are said to have the name
'white'. Socrates cannot, however, receive the definition of whiteness,
to use the terminology of 3b 2. The definition answering to the name
'white' as applied to Socrates would be a definition of a white thing,
not of whiteness. Hence, the definition answering to the name 'white'
as applied to Socrates is different from that answering to the same name
as applied to whiteness, and in this case the situation fulfils the conditions
of ia iff. I conclude, therefore, that propositions like 'Socrates is white'
can, despite the lack of explicit references by Aristotle to that notion,
be regarded as examples of homonymous predication. It must be realised,
I owe this point to Mr. R. D. Sykes.
however, that a sine qua non of the acceptance of this conclusion is the
recognition that a predicate expression is as much the name of the
hypokeimenon as is the subject-expression.
On this view, when we say '&vOpco7rto )eux64 EaTL', we apply the word
or name 'X)eux6' to man, and thus, as Plato said, the man gets many
names - 'white' as well as 'man'. But '>?eux6r' is also the name of To
Xeux6v (i.e. ?Xux6'nq4 - whiteness), although, Aristotle says, not in the
same sense. Man and whiteness are homonyms, but they are so in a
significant way because the particular whiteness picked out on this
occasion depends for its existence on the existence of man. For it is in
this way that Aristotle explains what he means by saying that something
is in a hypokeimenon. Thus the kind of homonymity involved is 7tpo6 rL
homonymity.' If it be maintained that all entities in the secondary
categories - qualities etc. - are in this position, dependent for their
existence on substances as hypokeimena, then there is no need to posit
independently existing Forms to be the meaning of the corresponding
terms, and the problem confronting Plato is in this respect solved.
Aristotle has thus solved, or attempted to solve, one part of Plato's
problem of how one thing can have many names, by adopting the view
that in propositions like 'Man is white' we pick out a thing - man - and
make reference to an entity - whiteness - which is in fact an aspect of
that thing. This entitv happens (per accidens) in this case to depend for
its existence on the existence of the thing - man; for the man in question
happens to be a white thing, and whiteness cannot exist except in a
hypokeimenon. In a proposition like 'Man is an animal' the situation is
somewhat different; for, in that case, man must be taken to be an entity
which involves per se being an animal, whatever else it involves per
accidens. In effect, this theory accepts much of the logical pattern of the
Platonic doctrine of the communion of Forms, but tries to provide an
explanation how the entity picked out by the word 'man' can have
associated with it the entity picked out by the word 'white'. The
association of man with animal, he seems to think, needs no such ex-
planation, the relevant relation being that of inclusion or partial identity.
In referring to man one necessarily refers eo ipso to animal. There are,
however, other problems to be faced here, arising from the independent
status given to species and genera; these I shall return to later. Aristotle's
account of synonymous predication leaves the matter as Plato left it
with his doctrine of the blending of Forms, and how independently
I Cf. Met. ioo3a 33ff. and E.N. iog6b 26ff.
i i6
existing entities can have a species-genus relationship is a problem to
which Aristotle recurs in Metaphysics Z. i4..
Aristotle's treatment of predication remains, in its logical character,
very Platonic, although, by distinguishing between substance and the
other categories, he makes a drastic reduction in the number of self-
subsistent entities: qualities etc. do not exist independently. In this
respect, Aristotle's tendency is towards an ontology of particular
substances, or particular specifications of matter (except for God and
any other examples of particular forms which are self-subsistent). Given
the realist theory of meaning which he accepted, it was inevitable that
developments in his ontology should affect his views on predication. In
both Plato and Aristotle, logical and metaphysical issues are inseparable.
Certain other problems remain for Aristotle, as already indicated.
There are two in particular which deserve mention. First, (a), if there
is such a thing as homonymous predication, what determines whether any
combination of expressions of the form 'S is P' is a genuine predication
at all? In the case of synonymous predication there is a rule - the
definition of the predicate must be applicable to the hypokeimenon,
and this entails that the hypokeimenon must per se be included in the
class picked out by the predicate expression. But there is no such rule
in the case of homonymous predication. Is there anything to prevent
there being a predication of the form 'White is a man' - 'White' not of
course being the man's name?
Second, (b), there is the difficulty that in synonymous predication,
when the entity picked out by the predicate term, e.g. a species, is
never in a hypokeimenon and thus never dependent on anything for its
existence, this entity must have an independent status. Thus on the
theory presented in the Categories, species and genera still have a Platonic,
Form-like existence. The difficulty indeed arises whenever reference is
made to species or genera, whether by means of a predicate expression
or by means of a subject expression.
To take each of these points in turn: -
(a) If all cases of predication are cases in which the entity picked out
by the predicate expression is dependent in some way on a hypokeimenon,
it might be that this would provide a general rule which determines
whether something is a genuine predication. The essential steps in the
passage to this conclusion are to be found in Posterior Analytics I. 22,
where Aristotle argues that things in the secondary categories are
predicable only of substance, in which, according to the Categories, they
inhere. Earlier at Prior Analytics 43a 2 sff, he classifies entities in a
I I7
way different from the classification of the Categories - in terms of
predicability alone. Particulars are not truly predicable of anything else,
although other entities are predicable of them; there are, at the other
extreme, categories, which are predicable of other things but are such
that nothing is predicable of them; and in the middle there are things,
like man, which are both predicable and possible subjects of predication.
Aristotle adds, with reference to particulars, that in the case of sensible
things, 'just about all' are such as not to be predicable of anything else
except xoc& ua uN;;nxo - ' for we sometimes say that T0 XOux6v ixezvo
is Socrates and that T0 'poam'0v is Callias'. Aristotle neither explains nor
justifies the remnark here, although the example is clear enough. I take
it that the restriction to sensible things is necessitated by the possible
existence of independently existing forms, which qua form might be
thought to be predicable, whether or not they are in fact.
Posterior Analytics I. 22 goes into more detail and it is clear from the
discussion there that 'xocTOC au3Pznx6;' is opposed not to 'aO' ocx6'9 but
to '& 7rWo'. That is to say that a xaoaX& aucsp43-4x6 predication is not here
an accidental predication, in the sense that the relationship between
subject and predicate holds on occasion or for the most part only, i.e.
not necessarily. It is one in which there is no such relationship at all
properly speaking. It remains a fact that it is legitimate to say ' ToL eux'o
exeZvo 6a-TL Ewxpai'Tj, and this combination of words has a meaning
when certain other combinations of words would not have one. Despite
the fact that they involve a xcxrat au[Lp3r3Zx6 predication only, there is
no embargo against saying these words. One could provide their rationale
by pointing out that they in fact constitute a statement of identity, and
presumably statements of identity could be transformed into or shown
to be derivative from predicative statements as a result of an analysis akin
to that of Plato's Sophist.' Thus the form of words in question could as it
happens - xcxrok auppe%x6 - be interpreted as a true predicative state-
ment, although equally, under another interpretation, it becomes
nonsense, viz. if the words are interpreted as saying something about
whiteness (TO X?uxOv) - that whiteness is Socrates. There is no such
ambiguity of interpretation about the words 'Socrates is white'; hence
they constitute a predication &ktXa.
I Just as the Sophist provides an analysis of statements to the effect that one thing is
different from another, so by implication it provides an analysis of statements of identity.
To say that one thing is the same as another is to say that it partakes of the Form of
Sameness 7tpo4 rt. cf. Pr. An. 1. 36, where Aristotle interprets relational statements in
general as predicative statements in an oblique case.
i i8
Aristotle, however, puts the point rather differently in Post. An. 1. 22.
His object in the passage in question is to show that there cannot be an
infinite number of premisses in demonstration, and thus that there
cannot be infinite chains of predications; these must come to a stop at
particulars at one end, and at the widest genera at the other. To show
this he has first to rule out all but genuine predications; for if we allow
both 'Man is white' and 'White is man' as genuine predications, the
series of predications can go on ad infinitum, because it can go round in a
circle by reversing the subjects and predicates. He next has to prevent
any other kind of reversal of subjects and predicates (8 3a 36 ff.).
It is difficult to follow the succeeding argument. Rossl describes it as
so difficult that any interpretation is conjectural. I feel, however, that
this is overstressing the difficulty, and I think that Ross and the Oxford
translator have increased their difficulties by taking an insufficiently
literal view of the text. They both consider that Aristotle is concerned
to deny reciprocal predication because things cannot be their own
qualities. Thus 83a 36 is translated by Ross - 'If B cannot be a quality of
A and A a quality of B - a quality of its own quality - two terms cannot
be predicated of each other as if each were a genuine subject to the
other." There are unfortunately alternative readings of the first few
words in the manuscripts here. The Bekker text reads "'Eet s1 , 'aTL
to5,To -oU& 7OLt6r%, X&XcEVO toTrou', while Ross adopts another
reading - "ErL c'n e T' aTOL 8 To68g 7tOL6OT4e x X OTOu'. Nei-
ther reading, however, is such as to make Ross's translation plausible. A
natural translation of Ross's text might perhaps be "If A cannot be a
quality of B and C of D ......... " while possible translations of the
Bekker text would either be the absurd "If A cannot be a quality of
B and A of B ........" (taking the '"xetvo' to refer back to the first-
mentioned entity), or "If A cannot be quality of B and C of A ...........
(taking the references of '"roUro' to be the same in each case). The last
seems to be the most plausible suggestion, and I would therefore
translate the whole protasis as "If A cannot be a quality of B and C of A,
i.e. if there cannot be a quality of a quality (sc. as C would be in this
case) .. ". This makes it unnecessary to translate 'OCL0T-JTO
Tcot6cT-' as 'a quality of its own quality' - there being no suggestion in
the text of the phrase 'its own'. It is similarly unnecessary to translate
8 3b Io as "Nothing can be predicated of its own quality. " as Ross
does.
Ross's interpretation makes the passage say nothing in general about
I W. D. Ross, Aristotle, Prior and Posterior Analytics, p. S78.
'I9
the impossibility of qualities acting as hypokeimena, despite the fact that
the sentence beginning at 8 3b I I and the passage beginning at 8 3b I 7,
which purports to sum up the argument, explicitly deny that accidents
can be hypokeimena. The argument at 8 3a 26ff. must be taken, therefore,
as maintaining this general thesis and as asserting as a corollary that it
makes reciprocal predication impossible. Within the category of sub-
stance reciprocal predication is impossible because the limitations of
definition entail an ordered chain of predications in which there is no
room for the possibility that a member may be predicated of its own
successor. Reciprocal predication is impossible within the other
categories because there cannot be qualities of qualities etc., and because
the chain of predications is finite any way.
On this interpretation, a summary of the argument would be: -
i. There must be no reciprocal predication if infinite chains of predica-
tions are to be impossible.
2. Reciprocal predication is possible only if there are qualities of
qualities. (Aristotle here takes the status of A in "A is B and B is A"
to be the same as that which it has in "B is A and A is C".) - 83a 36.
3. Predication is either (a) in the category of substance or (b) in one of
the other categories - 83a 39.
4. (a) It has already been showvn that infinite chains of predications in
the category of substance are impossible in any case, because of the
limitations of definition. Hence, afortiori, reciprocal predication
is impossible there. For it to be possible a genus would have to be
its own species - 83b if.
(b) Reciprocal predication in the other categories involves the
existence of qualities of qualities, and this is impossible because
qualities etc. are predicable only of substance - 8 3b 0o.
S. Hence the lower end of chains of predications is always substance and
the top end is provided by the limits of definition, i.e. by essence,
or by the categories - 83b 12ff.
6. Therefore the number of attributes predicable of a substance is limited
(a) by the elements of its essence - -a& Ev '-- o095Xl, and (b) by the
categories of its accidents - 8 3b I E.
7. Hence, afortiori, infinite chains of predications are impossible.
The argument is summed up in 83b i7ff. The assunmptions stated are
(a) that we are concerned with chains of predications in which there is
just one thing predicated of each term, and (b) that self-categorial
predication (i.e. predication of something in the same category as the
attribute) is impossible except in the category of substance. (I take 83b
I 2o
I 8 in this way, rather than as saying that the only kind of self-predication
possible is the predication of a thing's essence - the interpretation whidi
Ross offers. The latter could scarcely be called 'self-predication' in any
case). Given these assumptions, the bottom end of any chain of predic-
ations is substance and the top end is provided by the elements of the
definition or by the accidents predicable of a substance. Both of these are
finite in number, the latter because the number of categories is finite.
The two most important corollaries of the argument are that (a) subjects,
i.e. hypokeimena, are always substance3, and hence (b) predication of
something of a quality is impossible. It would follow also, although
Aristotle does not draw the consequence, that it is impossible to give
definitions of qualities etc. a view which would also demand drastic
amendments to the account of synonymous and homonymous predication.
When he denies that it is possible to predicate something of a quality,
Aristotle adds the words "unless it is an accidental predication" -
'accidental' having the sense already specified. It would appear that
genuine predications can be distinguished from such accidental predic-
ations in that the former must have substances as their subjects. Let us
return, therefore, to the explanation of this accidental predication
which Aristotle gives at 8 3a Sff. There he says that, when I say that the
white is a log, I say that thatwhich happens to be white is a log. The white
is not the hypokeimenon of the log - "for it is a log neither by being
white nor by being what a particular white thing is". Hence the white
is a log only in the same way as the musical is white, i.e. because a man
who happens to be musical is white. On the other hand, "the log is the
hypokeimenon, whatever else it is, not being different from what log or
a particular log is". Since a log is said to be the same as what a log is this
appears to make a substance the same as its essence and is therefore in
line with Metaphysics Z. s. It also entails that a hypokeimenon must be a
substance, and indeed Aristotle goes on to define an accident in the
converse way, as something which is spoken of as it is spoken of while
being something different - "There is nothing white which is white
without being also something different.'
Since accidents are predicable of a hypokeimenon, and since a particular
white thing must be white in virtue of a particular whiteness, Chung-
Hwan Chen concludes that particular qualities must be predicable, as
well as general qualities. This conclusion does not follow, for, in this
passage Aristotle is concerned with predication, not a classification of
entities as in the Cateaories. There is thus no need to look for references
I Post. An. 83a 30ff., 83b 22ff. cf. 73b sff. for an analogous definition of substance.
I 2 1
here to all the possible kinds of entity mentioned at the beginning of the
Categories. Chung-Hwan Chen admits as much, saying that it is possible
that in the passage under consideration Aristotle may not have distin-
guished between -xo Xeux6v and To6 Xeuxov; but he goes on to say that
in this case the denial that particular qualities are predicable is still less
justifiable "for he did not think of particular -4P3nx6o any more".
This is an extraordinary piece of argument. It is possible that particular
qualities did cease to be part of Aristotle's ontology, but this is no reason
for maintaining that they are nevertheless predicable. On the other hand,
the conclusion that Aristotle did admit cross-categorial predication does
emerge from the passage - only, as I have shown, Aristotle had never
refused to admit it.
The most important point to emerge from the passage is rather different.
It can be considered by asking 'What is a hypokeimenon?' So far I have
often refrained from translating that term. Translators of Aristotle tend
to translate the word sometimes as 'subject', sometimes as 'substratum',
and it is often maintained as Chung-Hwan Chen does,' that the ternm cuts
across the distinction between logic and metaphysics. That it (loes so,
it might be thought, is especially evident from the distinction between
xxO' U'oxeL[wvou eyeaOoxt and &v U'7rQXCEL1EVC F-LVXL, since the first
phrase is concerned with what can be said, the second with what can be.
But from the present passage it emerges that a hypokeimenon must be
a substance, or, in the words of Post. An. 73b 5ff, it must be xaU' ovro.
A hypokeimenon must be a self-subsistent entity. To say that Aristotle
has allowed metaphysical considerations to influence his views on logical
issues would, however, be to miss the point. For with what justification
may it be assumed that the notion of a hypokeimenon was ever a purely
logical notion? It is here that we must refer back to the fact that Aristotle
retained a realist theory of meaning. A hypokeimenon is, in virtue of
that theory, always that which is referred to or picked out by a subject
expression. A hypokeimenon is, therefore, always an entity, one of 'ta
onta'. This consideration invites questions about which of 'ta onta' is
truly so. Given that nothing in the secondary categories is ever foun(d
except in a substance, the conclusion follows that substances are the
only true 'onta', and thus the only true hypokeimena. For the thing
picked out by an expression like 'whiteness' is something which is always
found in something else; it is always what it is by being at the sanme tinie
cf. W. D. Ross, Op.Cit., pp. 70-I, E77. Cook Wilson, Statement and Infercnce, 1,
pp. i 59-66 goes to great length-s both to accuse, and to attempt to excuse, Aristotle
of confusion.
I 22
something different. Hence it is that Aristotle denies, at Post. An.
83a 36ff., that qualities can be predicated of a quality. If we can say
"Whiteness is dazzling", this is nevertheless not a genuine predication;
it is only one per accidens, like the proposition 'The white is a log'. For,
whiteness is not a genuine hypokeimenon. Thus the rule which distin-
guishes a proper predication, whether synonymous or homonymous,
from an accidental predication is that the subject expression must refer
to a genuine hypokeimenon, i.e. a substance.
(b) In answer to the second difficulty mentioned earlier - the difficulty
that species and genera retain, on the theory of the Categories, a Form-
like status - Aristotle's move was clearly to deny that species and genera
have in fact such an independent status. It is not altogether clear why
he had ever maintained that they had this, except for a residual Platonism
on his part and except that since species and genera are, as stated at Cat.
3b I O ff., somehow closer to primary substances than qualities etc., they
do not otherwise fit into the scheme of the categories. But there is no
real reason why they shoulld do so. As is suggested by Topics I.9, notions
like that of the genus belong properly to the classification of the Predic-
ables and thus cut across all the categories.
Aristotle's arguments against the Forms in Met. Z. 14, cover both
essential and accidental predication, and the arguments in terms of
essential predication, e.g. that animal will be divided from itself if it is
the same in horse and in man, apply equally against separately existing
species and genera. Indeed Plato himself had anticipated Aristotle with
similar arguments, although his own answer remained unconvincing.
Aristotle's recipe was to accept what he had asserted half-heartedly at
Cat. 3b io about the particularity of substance, and therefore to deny
substantiality to anything general. This procedure he adopts in Met. Z.
13ff and H. i. But if this is so, species and genera must also be in a
hypokeimenon, and the classification of the Categories breaks down in
that there are no things which are predicable of a hypokeimenon without
being in one.' In sum, Chung-Hwan Chen is right in concluding that
Aristotle came to maintain that all those things which are predicable of
a hypokeimenon are also in one. (I do not think, however, that it is
enough merely to quote, as he does, passages in wvhich Aristotle speaks
of an eidos as ivov or e&vu7t&pov; since the changes at issue are changes
While it can now be said that the genus animal is in different mlen becausc (different
instances of this genus depend for their existence on these different nien, there still
remains a problem concerning what nmakes them all instances. The probleml of thle
difference betwveen what is particular and what is general remains untouched, arid the
realist theory of nmeaning nmakes its solution very difficult.
I 23
in Aristotle's metaphysics, it is important to find the rationale for these
usages.)
Given, then, that species and genera are not substances, but are
dependent for their existence on particular substances, the field of genuine
hypokeimena is restricted to those particular substances. These are not
only, to use the language of P. F. Strawson's Individuals, the paradigm
cases of logical subjects, they are the only independently existing things
to which reference can be made.
* * *
From these two theses, (i) that only particular substances are
hypokeimena, and (2) that everything that is predicable is non-substantial
and therefore in a hypokeimenon, it follows that everything which is
predicable is also in a particular substance as a hypokeimenon. Thus
everything that can be picked out by a predicate term stands in a re-
lation of dependence to a particular substance. Everything of this
sort is a mode of a particular substance, to Luse Rationalist language.
Aristotle also seems to have come to take it that things which stand in
this relation of dependence to something are also predicable of it. At
least, this seems to be the only hypothesis which explains such statements
as Met. 1 o29a 23 - "the others are predicated of substance, while
substance is predicated of matter", and Met. IO49a 3 g - "when that wvhich
is predicated is a form and a particular (L8o6 xot XL m'r QL), the
ultimate <hypokeimenon> is matter and material substance." How can
substance or particular form or essence be predicated of matter? It
appears from Afet. Z. ii, I037a 26ff., that Aristotle came to think of
any substance which contains Inatter on the inodel of a snub-nose; such
substances are not the same as their essence, the form depending on the
matter in the way in which snubness depends for its existence on the
nose.' This dependence is not, however, accidental; the snubness does
not belong to the nose in the same way as whiteness belongs to Callias; 2
for snubness cannot exist apart from noses. It is nevertheless a dependence
and in this sense the form may be said to be in the matter - it cannot exist
I It should be noted that the doctrine by which Aristotle explains the unity of definition
- that the genus is matter to the species - seems to make the definition like a snub-nose,
not the same as its essence. This is an intolerable conclusion, but to explain the relation-
ship between the genus and the species on the analogy of matter and form is mistaken
in any case.
I Cf. Met. Z. 5.
1 24
apart from matter. The bronze of the statue has a certain shape, but this
particular shape depends on this nmatter for its existence. It is thus in it.
In consequence we can say that the bronze is a statue, and this is a case
of genuine predication, not one which is pei accidens, despite appearances
to the contrary. There are therefore chains of inherence parallel to
chains of predication - qualities and other accidents inhering in substance,
and substance, qua essence, inhering in matter. Once it is admitted that
substance is predicable of matter, when it is in it, the two chains may
be identified with each other.
It would seem from this that on Aristotle's view both a necessary and
a sufficient condition of one thing being predicated of another is that the
first thing should be dependent on the second for its existence. This is
not an altogether satisfactory account, if only because the kind of depend-
ence is not stated, and the attempt to state it might well involve
circularity, i.e. it may be possible only to say that it is the kind of
dependence which an attribute has on its substance. The notion of predic-
ation is very far from being on this view, a grammatical notion. In this
respect, Aristotle is, I believe, right, whatever be the extremes to which
he was willing to take the consequences. On his view, a sine qua non of
predicating one thing of another is that this other should be capable of
being picked out unambiguously, without reference to anything else.
Only particulars fulfil this qualification, for only they, in Aristotle's
ontology, have an independent existence. Hence predication is possible
of them alone.
ln sum, Aristotle's final answer to the question 'How can one thing have
many names?' is that sonme things depend on others for their existence
and are derivative from them. Hence, when we name them we name also
their subjects. Aristotle is also on the road to saying that the reason why
accidents are possessed by certain subjects lies in their inherence in those
subjects. This is at any rate implicitly implied by the fact that his reason
for calling 'The white is a log' an accidental predication is that the white
is not a log by virtue of its being white. The implication is that in all
cases of genuine predication the hypokeimenon has the attribute
predicated of it because it is what it is, despite the fact that some
attributes are accidents in the proper sense, in no way xaO' oxPo6.1 Such
a view is very close to that of Leibniz, that in every true proposition the
predicate inheres in the subject. Aristotle's nmetaphysics is ideally, but
I cf. Post. An. 8 3b 1 9- 2 o, and v. Met. 1 o 2 7a 1 3 for the attribution of the accidental to
matter.
I 25
only ideally, a metaphysics of particulars in each of which lies the reason
why it is as it is. This too is very lose to Leibniz, and could bring with
it a whole Rationalist metaphysics. But Aristotle is too much the realist
to be the complete Rationalist.
University of London.
1 26

Aristotle On Predication

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Aristotle On Predication

Uploaded by

Copyright:

Available Formats

Aristotle on Predication

IN DEALING with the Greek Philosophers we tend to take the notion

I Chung-Hwan Chen, 'On Aristotle's Two Expressions' xacO' 6noX=Lk.vou XkyraOo

I Cf. Met. ioo3a 33ff. and E.N. iog6b 26ff.

You might also like