Binghamton University

The Open Repository @ Binghamton (The ORB)

The Society for Ancient Greek Philosophy Newsletter

12-1981

The Unambiguity of Aristotelian Being

Jaakko Hintikka
Boston University

Follow this and additional works at: https://orb.binghamton.edu/sagp

Part of the Ancient History, Greek and Roman through Late Antiquity Commons, Ancient
Philosophy Commons, and the History of Philosophy Commons

Recommended Citation
Hintikka, Jaakko, "The Unambiguity of Aristotelian Being" (1981). The Society for Ancient Greek Philosophy Newsletter. 238.
https://orb.binghamton.edu/sagp/238

This Article is brought to you for free and open access by The Open Repository @ Binghamton (The ORB). It has been accepted for inclusion in The
Society for Ancient Greek Philosophy Newsletter by an authorized administrator of The Open Repository @ Binghamton (The ORB). For more
information, please contact ORB@binghamton.edu.
 J^ih4ikkA
¿Ea$t

Jaakko Hintikka

THE UNAMBIGUITY OF ARISTOTELIAN BEING

1. Aristotle does not recognize the Frege-R ussell ambiguity of "is."

In th is p a p e r, I shall tr y to enhance our u nderstanding of A ristotle's thought

by relatin g it to certain contem porary problems and in sig h ts of philosophical
logicians. Now one of the most central c u rre n t issues in philosophical logic is a
challenge to a h u n d re d -y e a r old dogma. Almost all tw entieth-century philosophers in
E nglish-speaking countries have followed Frege and Russell and claimed th at the
words for being in natu ral languages — "is," " is t," }εστι e tc .— are ambiguous
betw een th e is of predication, the is of existence, the is of id en tity , and the
generic is. The significance of th is ambiguity thesis has not been limited to topical
discussions but has extended to historical stu d ies, including studies of ancient
Greek philosophy. A generation or two of scholars working in th is area used the
Frege-R ussell ambiguity thesis as an im portant ingredient of th e ir interpretational
framework. Cases in point are C ornford, Ross, G uthrie, C herniss, Vlastos, Ryle,
and (from the German-language area) Heinrich Maier. Indeed, the Frege-R ussell
distinction is still being invoked occasionally by A ristotelian scholars; see e .g .
Moravcsik (1967, p . 127), Kirwah (pp. 100-101, 141), Weidemann (1980, p . 78) and
Gomez-Lobo (1980-81, p . 79).
However, many of us have by this time come to suspect th at the Frege-R ussell
ambiguity claim is completely anachronistic when applied to A ristotle. The sources of
th is d ark professional secret are various, ranging from G. E. L. Owen's brilliant
studies of A ristotle on being to Charles Kahn's patient examination of the Greek
v erb τδ είνα ι . Most of u s good A ristotelians have nevertheless remained in the
closet. As was illu strated by the fate th a t befell the first major stu d y in which
Plato's failure to draw the Frege-R ussell distinction was noted, most of the u n ­
liberated A ristotelians seem to have thought th a t to note A ristotle's failure to draw
the distinction is to accuse him of an abject logical mistake. (See, e . g . , Neal's
introduction to B luck.) Accordingly, we have shied away from such impiety. It is
time for some consciousness-raising, however. It is not convincing enough merely to
The Unambiguity of A ristotelian Being 2

re g is te r the inapplicability of the modern distinction to A ristotle. (Gf. B enardete

1976-77.) We need a deeper u n d erstan d in g of the whole situation. In an earlier
p a p e r, I have shown th a t th ere need not be anything logically o r semantically
wrong with a theory which tre a ts the v e rb s of being as not exhibiting the F reg e-
Russell am biguity. (See Hintikka 1979.) More th an th a t: not only can we now say th a t
A ristotle’s procedure is free from any tain t of fallacy; he may have been a b e tte r
sem anticist of n a tu ra l language^ than Frege and Russell in th is p articu lar re sp e c t.
Hence I can without any impiety level at A ristotelian scholars th e same c riti­
cisms as Benson Mates recently (1979) directed at Platonic scholars, viz. th a t th ey
have been seduced b y th e modern myth th a t th e re is a distinction between the is of
id e n tity , th e is of predication, the is of existence, and th e is of generic impli­
cation; and to proceed to argue th at the distinction is not th e re in th e A ristotelian
C orpus. V ery little argum ent is in fact needed h e re . Not only is it the case th a t
A ristotle, one of whose main philosophical methods was to make conceptual d istin c­
tio n s, n ever draw s the distinction. Not only does he fail to re s o rt to the F reg e-
R ussell distinction in dealing with problems which we would deal with routinely in
term s of the distinction.
An example is offered by De Soph. El. 5, 166 b 28-36, where A ristotle is
considering in te r alia the fallacious inference form "C oriscus is d ifferent from
S ocrates” ( i.e . ’’Coriscus is not Socrates”) and ’'S ocrates is a man” to ’’Coriscus is
d ifferen t from a man" ( i.e . "Coriscus is not a m an"). Here we would expect A ris­
totle to make a distinction between the "is" of id en tity (used in the first prem ise)
and th e "is" of predication (used in the second prem ise). His point has been so
understood by Maier (vol. 2, p . 280), and th ere is some prima facie evidence for
doing so. For in stan c e, the terminological distinction A ristotle uses to expose the
fallacy, viz. betw een essential and accidental predication, will in A ristotle's la te r
w ritings in fact assume (we shall see) the force of a co n trast between those p red i­
cations which have an element of id en tity and those which do not. However, draw ing
th e predication v s . id en tity distinction is not what A ristotle is doing h e re . What he
actually does is to draw a distinction between tran sitiv e and nontransitive
predications: "It does not necessarily follow th a t all the same a ttrib u te s belong to
all the p redicates of a th in g and to th a t of which they are p red ic ated ." Applied to
th e example, th is presum ably means th at "a man" in the second prem ise is an accident
of Socrates and th a t th erefo re the predicable "C oriscus is d ifferent from x ," even
though it is tru e of Socrates, does not have to apply to "a man" since th is is only
The Unambiguity of Aristotelian Being 3

an accident of Socrates. This is not a distinction between two senses of "is,"

id èn tity v s. predication, even though it is perhaps not too h a rd to see how the
la tte r distinction should have developed out of what A ristotle does h e re . (I have
followed here Dancy 1975, Appendix II.)
What is even more im portant, Aristotle comes as close as one can ever hope to
denying th e ambiguity th esis himself. In Met. Γ 2, 1003 b 22-32, he w rites (th e
translation is K irwan's) :

ε ΐ äh τδ ’δν καί τδ ιδν ταυτδν Suppose it tr u e , th e n , th at

καί μ ία φύσι* τω ακολουθεϊν th at which is and th a t which
αλλήλοΐί ‘ώσπερ αρχδ καί α’ί τ ι ο ν , is one are the same
άλλ’ ουχ ω* ευ ΐ λόγω δηλοΰμενα th in g —i .e . one n a tu re —in
(δ ια φ έρ ει δδ ουθδν ούδ’ ’δν th at each follows from the
ομοίως ύπολάθωμεν, άλλδ καί other as origin and cause
ιτρδ ’έργου μάλλον)· ταυτδ γδρ do, not as being indicated
έι9 άνθρωπο* καί ’hv 'άνθρωπο* by the same formula (though
καί ’άνθρωπο*, καί ουχ ^έτερόν it makes no difference even
τ ι δηλοΐ κατά ττϊν λ έ ξ ιν if we believe them to be like
επαναδιπλοΰμενον τδ εν* th a t—indeed it h e lp s). For
ε σ τ ίν ’άνθρωπο* καί ε σ τιν one man and a man th a t is
ε ΐ ί ’ϊιν ’άνθρωπο* · δηλον δ ’ ο τ ι and a man are the same
ου χ ω ρ ίζετα ι ο’<3τ’ επ ί thing; and nothing d ifferent
γενέσ εω ί ο’ΰ τ ’ επ ί φθορδ^,ομοίω* is indicated by the
δδ καί επ ί τοδ εν δ * · reduplication in wording ’he
‘ώστε φανερδν ‘ά τι η πράσθεσι* is one man’ and ’he is one
εν τ ο ύ τ ο ιΐ ταότδ δη λο ΐ, man th at is' (it is plain th at
καί ουδδν έτερ ο ν τδ ‘δν th ere is no distinction in
παρά τδ *öv, . . . [the processes of] coming to
be or d estru ctio n ); equally
in the case of th at which is
one. It follows obviously th a t
the addition indicates the
same th in g in those cases,
and th a t which is one is
nothing different a p art from
th at which is .
The Unambiguity of A ristotelian Being 4

It is im portant to note th a t at 1003 b 28 Aristotle is employing τδ είνα ι (sc . in

the p h rase ε*ΰ4«Γ)’άνθρωποί ) in what Ϊ shall late r in th is p ap er argue to be a p u rely
existential use. Nonetheless he is emphatically a ssu rin g u s th a t th is use is not
d ifferen t from what in effect are the first two (allegedly sep arate) Fregean
meanings of is .
An even b lan d er assertion to the same effect is found in De Soph. EL 6, 169 a
8-10: "For the same definition (horos) applies to 'one “single th in g ' and to 'th e
th in g ' hap los; th e definition e .g .,o f 'man' and 'one single man' is th e same, and so,
too, with other in sta n c e s."
This p re tty much confounds the first th re e members of th e four-fold d istin c­
tion of F rege's and R ussell's. As to the fo u rth it is amply clear th a t th e re is no
Frege-R ussell ty p e difference in meaning for A ristotle between the d ifferent occur­
rences of is in "Socrates is a man" and "a man is an animal." If fu rth e r evidence is
needed for the total absence of the Frege-R ussell ambiguity th esis in A ristotle, it is
easily forthcom ing.
In m aintaining the unambiguity of τδ είνα ι v is-à -v is th e Frege-R ussell
.distinction, A ristotle apparently is not ju st a faithful Whorfian following blindly th e
W eltanschauung implicit in the language of the trib e , as might be suspected among
o th er things on th e basis of the absence of any separate v e rb for existence in the
ancient Greek. A ristotle was cognizant of the controversies th a t had raged as to
w hether τδ and τδ ‘εν mean the same or w hether they have several differen t
meanings. (See De Soph. El. 33, 182 b 22 f f .) Nor is A ristotle unaw are of th e
dangers of uncritically assum ing th a t what is , always is what it is , and not an o th er
th in g , as is illu stra te d among o th er th in g s by his criticism s of Parm enides in P h y s.
A 3 . N evertheless, his failure to acknowledge the Frege-R ussell am biguity is deeper
th an a conscious choice between competing conceptual schemes. Not only does he
refu se to countenance th e Frege-R ussell distinction as a homonymy betw een several
d ifferen t m eanings. He does not always recognize the distinction as a separation
between d ifferent uses o f the Greek words fo r being. More accurately speak in g , he
does acknowledge some differences between the relevant u s e s , as we shall see, b u t
he does not co-ordinate them into a th re e -p a rt or fo u r-p a rt distinction.
Even though these observations do not automatically solve any h a rd in te rp re ta -
tional problems concerning A ristotle, they help to clear away m isunderstandings.
For in stan ce, we can now see th a t A ristotle's formulas for what has la te r come to be
known as essence, τδ τ ί έστι (What [it] is) and τδ τι ην είνα ι (w hat it is [for a
thing] to be) exhibit for a tru e Fregean an irredeem able ambiguity between
The Unambiguity of A ristotelian Being 5

predication and id en tity . For Aristotle they ex p ress ipso facto som ething's being
su ch -an d -su ch and its being identical with some one en tity . This is vividly shown
by the fact th a t Aristotle frequently used the v ery formula as a name for his firs t
categ o ry , substance, in spite of considering particu larity ("separability and
'th isn e ss'") as the main ch aracteristic of su b sta n c es. No wonder A ristotle could
th u s raise the question, which in our anachronistic ears may first sound paradoxical
(Ross confesses th a t it is for him "difficult to see the point of th is questio n "),
w hether o r not a substance is identical with its essence. (See Met. Z 6 .)

2. The nonambiguity of esti does not preclude

purely existential uses

Here and in many o th er contexts it is im portant to realize precisely what is

involved in A ristotle’s failure — or p erh ap s ra th e r 'refusal — to make the
Frege-R ussell distinction. What is denied in denying the Frege-R ussell ambiguity
claim is not th a t the force of "is" or *€στι is different in different contexts.
R a th e r, what is ruled out is one p articu lar explanation of these d ifferences, v iz .
th a t they are occasioned by different meanings of the verb "is". In other w ords,
what is a sse rte d is th a t such differences are always traceable to the context and
due to it. Indeed, it is an in tegral p a rt of my position t h a t 'έστ.ι can have on
d ifferen t occasions in A ristotle different Fregean u se s. For in stan ce, Aristotle
can—and does—use ^ σ τι w ith a purely existential force. When one says "Homer is"
(<>Ομηρ<5$ έ σ τ ί) , what is at issue is obviously the existence of a p articu lar
individual. (Cf. De I n t. 11, 21 a 25-27.) In general, when one asks ει *έστιν, one
is asking w hether an en tity or entities of a certain kind ex ist. (See P ost. A n. B
1 -2 .) F u rth e r examples of unm istakably existential uses of ’έστιΐη Aristotle are
easily found; see e .g . C at. 10, 13 b 27-33.
In th is re sp e c t, my thesis differs sharply from what seems to be the most
popular reaction to the data th a t can be adduced against the presence of the F reg e-
Russell ambiguity in A ristotle. According to this competing view, ’é a tiis unam­
biguous because it basically always has the predicative sense. Where it apparently
does n o t, e .g . in the existential uses listed above, we must u n d erstan d th e usage
as being elliptical: "Socrates is" on th is view basically means "Socrates is something
o r else." (T here may be im portant restrictio n s as to what th is "something or else"
can b e , b u t they need not detain u s h e re .)
The Unambiguity of A ristotelian Being 6

T his view seems to have been suggested by G. E. L. Owen, and it has re ­

cently kept cropping up in slightly different v a ria n ts. T here is a sense in which it
probably comes close to being a tru e representation of what th in g s are like
according to A ristotle’s last and final conclusions. Roughly, for any en tity to e x ist
is for it to be what it is , i .e . what it essentially is .
However, adm itting th is does not mean th a t in the force of th e term ’έστι in
A ristotle’s actual argum entation is tacitly predicative. For one th in g , th e iden tifi­
cation ju st offered is probably only an approxim ate one, anyway. It is not clear
th at for Socrates to exist is (apud A ristotle) for him to be a man. R ath er, on a
closer look it seems (as Balme has shown) v e ry much as if for Socrates to exist is
not so much for him to exemplify (more generally, to develop tow ards exem plifying)
the sp ecies-ch aracteristic form of man, b u t ra th e r to exemplify (more accurately ,
develop tow ards exemplifying) the p articu lar n a tu re which consists in his likeness
to his p a re n ts. And it is not clear at all th a t Socrates' exem plifying th is p a rtic u la r
form is a predicative relation ra th e r than an id en tity .
Be th is as it may, even if the elliptical ch aracter of ’έστχ airXös is p e rh a p s a
conclusion of A ristotle’s argum ents for his m etaphysical th e o ry , it cannot for th is
v ery reason be a p a rt of what he assumes in them. When I reject th e ellipsis
th eo ries, it is th u s as a claim of what the basic semantical force ο ΐ ’εστι and its
cognates are for A ristotle, and not as a possible feature of his ultimate m etaphysical
doctrine. However, in the former sense I do reject it tout c o u rt, and hence also
reject th e mistaken idea th a t it is somehow implied by the absence of the F reg e-
Russell ambiguity assum ption from A ristotle.
T his p u ts on me th e onus of commenting on th e recen t denials of any p u rely
existential uses of v e rb s for being in A ristotle. Suffice it here to deal with one of
th e most recent p u tative argum ents for th e absence of th e existential uses in
A ristotle o r in c ertain p a rts of th e A ristotelian C orpus.
The ellipsis hypothesis has not been defended by its rep u te d originator at any
g rea ter len g th . It has recently been discussed by A. Gomez-Lobo (1980-1981). The
p a rt of th e C orpus which Gomez-Lobo and his ilk have to w orry about most is
clearly P ost. An. B 1-2, where A ristotle in so many words recognizes questions of
simple being (ε ΐ \ α τ t) besides the th ree o th er kinds of questions which figure in
an A ristotelian science, viz. το £σ τι, τδ ό ιδ τ ι, and τΐ ’έσ τι.
Actually, stric tly speaking Gomez-Lobo does not deny th a t a sentence of th e
form ’έστι + a noun phrase can ex p ress mere existence in A ristotle. He admits th a t
e .g . Met. Λ 7, 1072 a 25 is a case in point. But he striv e s to reduce greatly th e
The Unambiguity of A ristotelian Being 7

scope of th is way of reading Aristotle by removing Post. A n .. B 1-2 (in fact, it

seems, all of Post. A n .) from its scope. Hence a b rief discussion of Gomez-Lobo's
argum ents are in o rd e r, for if they were valid, much of the plausibility of my point
would be lost.
The ε\ ^στι questions used to be taken without any fu rth e r ado as questions
of existence. Gomez-Lobo is entirely rig h t in recognizing th at the situation has
changed. The in sig h t th a t Aristotle did not believe in th e Frege-R ussell ambiguity
and th a t the basic semantical meaning of ’έστι in A ristotle is hence neutral with
resp ect to the different Fregean senses of being certainly makes a fresh look at
P o st. An. B 1-2 necessary . U nfortunately, Gomez-Lobo fails to give the new look a
ru n for its money, for his argum ents are inadequate in several resp ec ts. For one
th in g , most of his discussion is predicated on a failure to u n d erstan d in what sense
Aristotle th in k s th a t εί ’έστι questions, like all four questions, amount to looking
for a middle term . "How can th e re be a middle term between a single term and the
predicate 'exists'? he asks rhetorically. A straightforw ard answer would be
em barassingly obvious even if I had not pointed it out ten y ears ago (in Hintikka
1972a). Aristotle is thinking of as it were abbreviated syllogisms of the form

(*) Every B is simpliciter

Every Ç is B ________
Hence : Every C is simpliciter

which re su lt from a reg u lar b a rb a ra syllogism by omitting the major term .

It is obvious th a t (*) req u ires a treatm ent of existence somewhat different
from what contem porary philosophers have been used to. However, th is is no a rg u ­
ment against what I am saying. Even without discussing any details h ere, it is
p aten tly clear on o th er grounds th at we have to shake our complacency concerning
th e adequacy of the received Frege-R ussell treatm ent of existence in logic.
It may be objected th a t quasi-syllogism s of the displayed form are never
actually p u t forw ard b y A ristotle. The explanation is th a t he does not need to do
so. In th e syllogistic s tru c tu re of a science, the existence of the B 's is always a
consequence of the existence of a wider term , say A. Hence A ristotle accomplishes
the same effects by means of a reg u lar b a rb a ra syllogism as he accomplishes b y
means of (* ), as long as a proviso is explicitly or tacitly added to th e effect th at it
is only the widest term th at carries any existential force. This may perhaps be
illu stra te d by the following quasi-syllogism :
The Unambiguity of A ristotelian Being 8

(**) Every B is an A (and exists)

E very C is a B (no existential force)
Hence: Every C is an A (and hence ex ists)

I shall not discuss here what kind of treatm ent of existence is p resupposed in (*)
and (**).
My in te rp retatio n gains some fu rth e r credence from the fact th a t according to
A ristotle necessity is "carried downwards" in a syllogistic chain in the same way as
I have arg ued existence does. (Cf. A ristotle’s theory of apodeictic syllogisms in P r .
A n. A 8-12 , especially 9 .) In the same w ay as in (**) it is only the major p r e ­
mise th a t has to c a rry an existential force in o rd er for the conclusion to do so, in
the same way we can obtain a necessary conclusion from a b a rb a ra ty p e syllogism if
and only if the major prem ise is a necessary one. I believe A ristotle’s treatm ent of
existence and necessity in the context of a syllogism are related to each o th er v e ry
closely, b u t I cannot tr y to find th e ir g rea te st common denominator h e re .
T here is plenty of collateral evidence th a t th is is what A ristotle in fact m eant.
Since th e whole argum entative stru c tu re of Gomez-Lobo’s p ap er is th u s m istaken,
th ere is little th a t apparently needs to be said of the re s t of his p ap er.
What I also find su rp risin g is th a t th e re is conclusive evidence against
Gomez-Lobo in th e v e ry passages he is add ressin g himself to.

I mean if one is o r is not sim pliciter and not if [one is] white o r not (89
b 33).
τδ ει ’έστιν μή απλωί λέγω, αλλ* ουκ ει λευκδΐ μη.

How could A ristotle possibly have explained more clearly b y the means he had
at his disposal th a t he was presupposing a p urely existential use of
ε\ ^στι? It seems to me th at we have to realize th a t A ristotle, like J . L. A ustin,
ordinarily means what he say s.
Ironically, A ristotle's v e ry usage in P o st. An. B 1-2 provides u s with fu r th e r
counter-exam ples to the ellipsis th esis. When Aristotle th ere asks w hether a middle
term is (ε ι ’έ στι μέσον, cf. 89 b 37-38, 90 a 6 ), he cannot b u t mean w hether the
middle e x ists, for he co n trasts this question in so many words with the question as
to what it is .
T here is elsew here, too, excellent direct evidence against th e ellipsis-
hypothesis. In discussing in De Soph. El. 5 the importance of distinguishing th e
The Unambiguity of A ristotelian Being 9

absolute and the relative uses of a term from each other A ristotle w rites (167 a
4-6):

ου γδρ ταυτδ τδ μδ For it is not the same th in g

c
ε ι ν α ί τι καί αιιλώ5 μδ not to be something and not
είναι, φαίνεται δδ to be sim pliciter, though
διδ τδ πάρεγγυ$ τηΐ owing to the similarity of
λέξεωί καί μικρδν διαφέρειν language to be something
τδ ε ί ν α ι τι τοΟ ε’ΐ ν α ι , appears to differ only a little
καί τδ μδ ε ί ν α ι τ ι from to b e, and not to be
■J
τοΟ μδ εην αι. something from not to b e .

On can scarcely ask for more direct evidence. At the same time the passage shows
th a t in sp ite of th e ir differences, the predicative and the absolute (existential) uses
o f ’€στιν are not u n related , for they are the relative and absolute uses of th e same
notion. The quoted passage hence also offers evidence against ascribing the
Frege-R ussell ambiguity to A ristotle.

3 . Essential pred icatio n .involves identification

Thus realizing A ristotle's failure to make the Frege-R ussell distinction does not
necessitate throw ing overboard all earlier views concerning A ristotle's treatm ent of
b ein g , even though it does necessitate a fresh look at the evidence. (T here seems
to me to be far too much of a fashion in the c u rre n t lite ra tu re to disparage earlier
in te rp retatio n s ju st because they did not take into account some im portant aspects
of A ristotle's treatm ent of "is". Such a failure may be reg re tta b le , b u t it does not
automatically invalidate all the in terp retatio n s of a scholar.) N evertheless,
dispensing with th e Frege-R ussell dogma opens the door to certain fu rth e r lines of
th o u g h t. In the same way as the Aristotelian ’έστι sometimes has existential force
and sometimes does n o t, in the same way it sometimes has the force of iden tity and
sometimes does not. The unm istakable upshot of such passages as Met. Γ 4, 1007 a
20-33, Met. Δ 18, 1022 a 25-26, or Post A n. A 22, 83 a 24-32; A 4, 73 b 5-8 is
th a t th e form er cases are of the n atu re of what is in the misleading
tw en tieth -century terminology known as "essential" predications and the la tte r as
"accidental" predications. The misleading ch aracter of these labels is due to th e fact
th a t they are easily taken to re fe r to a purely modal distinction. Indeed, they have
The Unambiguity of A ristotelian Being 10

been so taken in recent discussion, where the term "A ristotelian essentialism " is
used as a label for a position v is-à -v is the foundations of modal logic and modal
m etaphysics. What we have discovered is a prima facie nonmodal element in
A risto tle's essential-accidental distinctions, an element th at seems to have been
largely overlooked in tw en tieth -cen tu ry philosophy.
Of c o u rse , even if we tr y to s ta rt from a prima facie completely nonmodal
distinction betw een id en tity and predication, we will end up linking the distinction
with various modal considerations. I cannot discuss th is m atter here beyond
pointing out th a t in th is way A ristotle's distinction gets linked up much more with
recen t discussions of the n a tu re of the individuals presupposed in ou r use of modal
concepts and of th e n a tu re of th e identities which may hold betw een them th an with
discussions of so-called "A ristotelian essentialism ."
The non-modal ch aracter of A ristotle's distinction is heightened by th e fact
th a t th e whole possible-w orlds treatm ent of modalities, so popular and so p erv asiv e
in our d a y s , was completely foreign to A ristotle for whom th e only reality was th is
sequence of potential nows each of which will momentarily be actualized. Bence even
A ristotelian potentialities had to prove th e ir mettle within th is one actual world
h isto ry of o u rs, and so had his notions of form, essence, and o n e-n ess.
In spite of what I said of the mistakes of tw en tieth -cen tu ry philosophers, my
point h ere is not entirely new. M. J . Woods has argued "th at A ristotle held th a t a
statem ent like 'S ocrates is a man' was, despite appearances, to be c o n stru ed as a
statem ent of id e n tity ." In defense of his view. Woods re fe rs to such A ristotelian
p assag es as Met, Δ 18, 1022 a 26-27; Z 4, 1029 b 28; Z 7, 1032 b 1-2; Z 7 , 1032 b
1-2; Z 8, 1034 a 8. This is not th e place to evaluate his evidence. It seems to me
th a t Woods could have stren g th en ed his argum ent by a more general examination of
th e sta tu s of the "is" in A ristotle's "Socrates is a man" v is-à -v is the failure of th e
Frege-R ussell ambiguity th esis and of th e resu ltin g freedom for A ristotle to pick
and choose between different uses (not senses) of ’έστιν.
Moreover, Woods' diagnosis, rig h tly understood, p ertain s to A ristotle's problem
ra th e r th an to his solution. For what it is th a t follows from saying th a t in an
essential predication like "Socrates is a man" the "is" has th e force of identify?
Very little , for we are then faced with the next question: What is it about certain
p red icates th a t enable them to individuate th e ir b e arers in such a way th a t the "is"
of predication assumes the force of "is" of identity? And which predicates a re of
such a k in d , anyway? Clearly, these are among the main questions A ristotle
The Unambiguity of Aristotelian Being 11

discusses in Met. Z and Γ, b u t unfortunately understan d in g his problem does not

automatically enable u s to un d erstan d his attem pted solution equally fully.
We do receive, however, some in stan t help in appreciating A ristotle's
discussion. For in stan ce, clearly not any old universal predicate individuates its
b e a re rs in the req u ired way. U nderstandably, A ristotle is quick to point th is o u t.
This gives rise to passages which have been taken to imply th a t fo r A ristotle what
to u s look like predications cannot have an element *of an id en tity judgm ent. One
such passage is Met. Z 13, 1038 b 34 - a 1. It is not clear, how ever, which way
th is notoriously difficult passage c u ts. One major problem with it is th at it
apparently contradicts what Aristotle says earlier in Met. Z 6. Here at 1038 b 34 -
a 3 he says th a t none of the things p resen ts universally is a substance (εκ τε δτ*ι
τούτων ΘεωροΟσι φανερδν£ότι ουδέν των καθόλου υπαρχόντων ουσία ε σ τ ί, καί *ότι ουδΙν
σημαίνει τ&ν κοιντί καταγορουμόνων τόδε τ ι , αλλά τ ο ιό ν δ ε .) . But earlier in Met. Z 6,
1032 a 4 ff. A ristotle had identified the substance of a th in g with its essence,
which presum ably can be sh a red , and later he likewise comes to identify it with th e
th in g ’s form. (See e .g . Met. Z 17, 1041 b 4 -9 .) Without being able to discuss the
problem at th e length it deserves—and re q u ire s—surely an im portant p a rt of the
explanation of th ese apparently co n trastin g passages in th a t they are p a rt and
parcel of the dialectical process of teasing out the difference between those
p redicates in connection with which ’όστιν is used as what we would call th e "is’’ of
id en tity and those in which the predicate is a mere universal. In fact, much of th e
earlier discussion in Z 13 can be understood as sketching out th is v e ry c o n tra st.
O ther ap p aren t counter-evidence to ray suggestions in th is section can likewise
be disposed of easily. For in stan ce, in De Soph. El. 22, 178 b 39 - a 10 A ristotle
says th a t what is predicated of a τόδε τ ι is not itself a p a rtic u la r, b u t a quality o r
q u an tity o r a relation "or something of th a t s o rt." These statem ents don't re p re se n t
his considered la te r opinion, how ever, which is th a t a predication which is not of
the n a tu re of an identification is precisely an accidental predication. T hat th is is
indeed A ristotle’s doctrine is recognized by Kirwan (p p . 100-101) on the b asis of
Met. Γ 4, 1007 a 20-33, even though he goes on to accuse A ristotle of committing a
logical mistake in m aintaining th is view: " ...e v e n if it is possible to make sense of
the distinction betv/een essential and coincidental predication, the former are no
more statem ents of identity than the la tte r a re ." Kirwan is here relying precisely on
the Frege-R ussell ambiguity th esis, which we have found totally inapplicable to
A ristotle.
4. Instantiation in n atu ral languages: a system atic view
The Unambiguity of A ristotelian Being 12

Suffice th is as an indication of one line of though opened by o u r observations.

To re tu rn to the main theme of this p a p e r, notice th at from the absence of the
F reg e- Russell ambiguity it does not follow th a t th ere might not be other
ambiguities about *έστ\ν, over and above the non-am biguous differences in use which
Frege and Russell mistakenly raised to the level of am biguities. F u rth e r light can
be throw n on th is question, too, by means of recent topical in sig h ts . T here is in
fact another major way in which recent logical and semantical work on the concept
of being p u ts A ristotle in an in te restin g new p erspective. In o rd er to see what it
is , we have to stra y tem porarily away from A ristotle and discuss certain topical
problems in th e logic of n a tu ra l languages. I shall discuss them in the case of
E nglish, even though similar th in g s can be said of o th er languages, including
ancient Greek.
These problems are as close to the h e art of all Sprachlogik as we can hope to
g et. Any logician knows th a t the lifeblood of virtually all in te re stin g logical
techniques in th a t basic p a rt of logic which is variously known as firs t-o rd e r logic,
quantification th eo ry , o r lower predicate calculus are the ru les of instantiation ( i.e .
ru les for su b stitu tin g names or name-like term s for quantified v a ria b le s). Now
suppose we want to deal with the logic of n atu ral languages d irectly , without first
attem pting the dubious and by th is time largely discredited translation to formal
languages. Then our first ta sk is to form ulate, likewise directly fo r n atu ral
languages, ru les of instantiation for the quantifier p h rases which take over the role
of quantified variables in n atu ral languages. How can we do that? How are we to
deal w ith, say, a quantifier p h rase like "every white horse which Alexander rode"
or "some small town where Socrates liv ed ," occurring in a context X - W? (We take
h ere the general form of these quantifier p h rases to be

ev ery
Y + wh-word + Z
some

where Z contains a "trace" to indicate where the wh-word was "moved away from .")
Now th e obvious way of form ulating instantiation ru les for such p h rases is to
legitimize a move from the sentence in which they occur to sentences like
. The Unambiguity of Aristotelian Being 13

(1) X - b - W if b is a white horse

and Alexander rode b
o r, resp ectively,
(2) X - d - W and d is a small town and
Socrates lived in d,

where "_b" and "d" are the respective in stan tiatin g term s. In general, the output of
an instantiation step is of the form

(3) X - b - W

where _b_ is th e in stan tiatin g term and Z' is like Z except th at the trace has been
replaced by "b" with the appropriate preposition. (We have been assuming th a t Y
and Z aré here sin g u lar.)
The details need not detain us h e re . What is of in te re st to us here is an
im portant difference between the situation in formal firs t-o rd e r languages and
n atu ral languages. In the form er, a single domain of values for the su b stitu tin g
term s ( e .g . my "b" and "d") is given. In the la tte r, the entities re fe rre d to by the
su b stitution-values have to be chosen from different sub domains in different cases.
For in stan ce, in (1) b has to be a living c re atu re , whereas in (2) d has to be a
location in space.
It lies close at hand for a logician to say th a t the only novelty here is th a t
n a tu ra l languages employ m any-sorted quantification theory (more generally,
m any-sorted logic). And th is need not by itself introduce any complications
(co n trary to what is e .g . implied in J . M. E. Moravcsik 1976). Indeed, m any-sorted
logics do not involve any serious new difficulties over and above one-sorted ones.
Yet th ere is a new question p resen t h ere. In m any-sorted formal logics, the
sortal differences are indicated by notational conventions. How are these differences
marked in n atu ral languages? How can one tell what subdomain b o r d must belong
to?
Some clues are obvious, and the most obvious is the relative pronoun which
disappears in the process of instantiation. (These relative pronouns can be taken to
be question words in a new role, except th at "what" is replaced by " th a t." ) If th e
operative word is "who," the relevant subdomain consists of p erso n s, if "w here," of
locations in space, if "w hen,” of moments (an d /o r periods) of time, etc . F u rth e r
The Unambiguity of A ristotelian Being 14

subdomains are introduced by prepositional p h rases containing similar w ords, fo r

instance "like which" introducing a realm of qualities ("some color like which you
have n ev er se en "). Clearly th ere is not a sh arp one-to-one correspondence between
th e ran g es of natural-language q uantifiers (my "subdom ains") and different relative
pronouns (or o th e r w h-w ords, with or without prepositions or similar q u a lifie rs),
b u t a ro u g h -an d -re ad y correspondence certainly obtains.
However, th is cannot be th e only clue to the choice of the subdomain. For one
th in g , th e whole relative clause ;can be missing from the quantifier p h rase in
question, and hence be unavailable to supply any leads. Hence it is the meaning of
Y in (3) which must supply the main information as to which subdomain (so rt) we
are dealing with. Presum ably we must assume some kind of semantical categorization
of th e term s (p h rases) th a t can serve as the Y in (3 ). In th e case of simple term s
these must be p a rt of th e ir lexical meaning. Since the Y’s in (3) are basically
predicate term s, we end up in th is way postulating a classification of all simple
p red icates of English into certain equivalence classes. T hese classes will be
correlated one-to-one with those subdomains of quantification, which we are dealing
w ith, when using q uantifiers in English, i .e . the larg est classes of entities we can
quantify o v e r, and also correlated in a loose way with certain wh-words and
p h rases.
The need of relying on Y for our choice of the subdomain is vividly seen from
the fa c t th a t if we tr y to eliminate Y (in the way in which we could dispense with
the relative clause), we would end up with an ungrammatical expression. In o rd e r
to p reserv e grammatically , we must amplify th e quantifier word itself so as to make
it capable of conveying th e crucial information: some becomes someone, som ething,
somewhere, sometime, somehow, etc. where the added handle serves to b e tra y th e
relevant so rt (subdom ain).
Furtherm ore, since each instantiation step (w itness (3 )) introduces an
occurrence of "is", these correlated classifications are likewise correlated with a
distinction betw een d ifferent uses of "is", v iz. those th a t could have originated
from an application of the instantiation ru le s, plus of course those th a t are logically
on p a r with them .
T hus we are led to recognize four correlated multiple distinctions. They
distinguish from each oth er

(i) C ertain wh-words (and p h rases with w h-w ords).

The Unambiguity of A ristotelian Being 15

(ii) D ifferent kinds of simple predicates.

(ni) The larg est classes of entities we have to recognize in the logic of our
language as domains of quantification.

(iv) C ertain semantically different cases of "is". (In them we of course cannot
distinguish from each other th e is of id en tity , existence, and
p redication.)
i

5. What do A ristotelian categories categorize?

At th is point you are supposed to have a deja vu experience. For what I have
a rriv ed at by means of purely system atic (logical and semantical) considerations is
to all practical purposes tantam ount to A ristotle’s theory of categories. One of the
most fundam ental and most perplexing questions concerning A ristotle’s distinction
between different categories is: What is being distinguished from each other? What
is A ristotle classifying in separating the different categories from each other? He
uses different Greek question words o r question p h rases ( τ ί έ σ τ ι , ποσόν, ποιόν,
πρβ5 t i , που, ποτό) as names for six of the categories, and the o th er labels likewise
go naturally with certain types of questions in Greek. This is p re tty much how he
in fact p resen ts his categories in Top. I, 9. He envisages different kinds of
entities ’’put before one" and classifies the different th in g s th a t can be said (and
by implication asked) of it.
B ut when Aristotle introduces his categories in C at. 4, they appear as classes
of simple predicates o r ’’th in g s th a t can be said" of an e n tity . Which are they?
The plot is thickened fu rth e r by A ristotle’s deeply ingrained habit of
considering categories as th e widest genera of entities th a t can be logically
considered to g eth er. This is seen for instance from Met. Γ1 1003 b 19 ff. o r from
Post. An. A 22, 83 b 10-17.
Furtherm ore, A ristotle indicates repeatedly th a t the distinction between th e
d ifferen t categories goes to g eth er with a correlated distinction between d ifferent
uses of %στιν. What is more, occasionally he seems to ru n the two distinctions
to g eth er. For instance in Met. Z 1, 1028 a 10 ff. "th at which is" is said to signify
the different categories. See also Met. Δ 7, 1017 a 23-30.
Scholars have debated intensively which of these different th in g s A ristotle
"really" meant. For instance, one persuasion maintains th a t the categories rep re se n t
The Unambiguity of A ristotelian Being 16

th e different kinds of questions one can (according to A ristotle) ask of a given

e n tity . This view is in different v arian ts held by among o th ers Ockham, Charles
Kahn, B enveniste, and Ackrill. O ther scholars hold th at A ristotelian categories are
what he says they a re , predicables. O thers, led by the formidable Hermann Bonitz,
have have held th a t categories were for Aristotle first and foremost the widest
genera of en titie s. "Sie bezeichnen die obersten G eschlechter, deren einem jedes
Seiende sich muss u n tero rd n en lassen ," he proclaims (p . 623 of the original).
Still o th ers have held th a t A ristotle’s category distinction is prim arily a
differentiation betw een several senses of ’έ σ τ ι , a rem inder of the "system atic
ambiguity" of words for being in A ristotle. This view was rep re se n te d by
Heinrich Maier, and in a sense it can be maintained th at G .E.L Owen is another
case in point. He has certainly been followed by a host of younger scholars.
If we had not seen th a t A ristotle is completely free from th e Frege-R ussell
ambiguity assum ption, we might also be puzzled by the fact th a t the distinction
betw een the d ifferen t uses of εστι in the different categories sometimes appears
prima facie as a distinction between different kinds of existential is (cf. e .g . the
Topics discussion w hether τδ *6v is a g e n u s ), sometimes as a distinction between
d ifferent kinds of predicative is (cf. e .g . P r. An. A 37, 49 a 6-9, read in
conjunction with the preceding chapter) , and sometimes as one between different
kinds of id en tity . These different emphases in A ristotle have found th eir fans. For
in stan ce, as Ross re p o rts , "Apelt re g a rd s the categories as prim arily a classification
of the meaning of th e copula ’is'" whereas Bonitz stre sse s the existential and
id en tity senses.
Some of th e shrew der scholars have responded to th is problem situation by
su g g esting th a t A ristotle was led to his distinction between the different categories
by several convergent ro u tes. For in stan ce, Ackrill su g g ests th a t in A ristotle’s
classification th e re are two elem ents, firs t th e idea th a t different kinds of questions
will have "categorically" different answ ers, and second the idea of categories as the
h ig h est genera. This is undoubtably a step in the rig h t direction. However,
philosophers taking such a line will n evertheless face the almost equally perplexing
problem as to why the different distinctions Aristotle had in mind should coincide —
or at least why A ristotle should have thought th at they coincide. By and larg e,
th ey have not solved th is problem. Ackrill says merely th a t, "It is not su rp risin g
th at these two ways of grouping things to g eth er should produce the same re s u lts ."
This opinion simply will not stand up to scru tin y when viewed in the cold light of
contem porary analyses of questions and answ ers. (C f. Hintikka 1976.) C ontrary to
The Unambiguity of A ristotelian Being 17

what Ackrill su g g e sts, it is not at all clear th a t answ ers to different questions fall
into mutually exclusive classes which correspond to the widest classes of e n tities.
For in stan ce, it is perfectly legitimate to reply to the question "Who is the head of
th e Academy?" not only by saying "Plato" or "a man", b u t alternatively "a blond",
"the youngest b ro th e r of Potone", or even "he is sittin g th e re ", all of which have
to be pigeonholed in different categories. Only by means of a fu rth e r analysis can
one p erh ap s hope to eliminate some of these replies a*5 amounting only to p artial
answ ers (o r as supplying collateral information to back up the conclusiveness of an
an sw er). Worse still, AckrilTs account is intrinsically in co n sisten t. For if th e
appropriate answ ers to different questions belong to d ifferent categories, it is
impossible to construe A ristotle's categories as answ ers to one and the same
question "What is it?" as Ackrill also su g g e sts. Even if what he says can somehow
be salvaged in th e last analysis, it does not help us to u n d erstan d what A ristotle’s
categories really were in the lea st. Prima facie, it is fa r from obvious th at th e four
correlated distinctions we find in A ristotle should go to g eth er, and A ristotelian
scholars certainly have not supplied valid reasons why they should do so.

6. Aristotle reco n stru cted

Now the b rie f analysis of the conditions of the conditions of instantiation which
I carried out above, p u ts both A ristotle's theory and discussions thereof into a new
p ersp ectiv e. Led b y purely topical (logical and semantical) argum ents, we have
a rriv e d a t a rem arkable reconstruction of A ristotle's th eo ry of categories. (My
argum ents have an even stro n g e r theoretical motivation th an I have spelled out
h e re , for they ensue from the basic ideas of the highly successful approach to
language analysis which I have called gam e-theoretical sem antics. For i t, see
Saarinen 1979.) We can now recognize all the apparently discrepant in g red ien ts of
A ristotle's doctrine in the system atic situation revealed b y my analysis. A ristotle's
use of question words and p h rases as labels for categories matches my use of
wh-words as a guide to th e subdomain involved in an in stantiation. His view of
categories as the different kinds of simple th in g s th a t can be said of an en tity
matches my classification of the meanings of simple predicates as guides to th e
logical "sort" inten d ed . His use of categories as the larg est classes of logically
comparable entities amounts to the focal point of my quest of the d ifferent larg e st
domains of quantification presupposed in a n atu ral language, and A ristotle's
correlation of d ifferent uses o f the word ^σ τι corresponds to the automatic
The Unambiguity of A ristotelian Being 18

alignment in my treatm ent of the o ther distinctions with certain differences in the
use of the word " is " .
What is more im portant, the correlation of these several distinctions is seen not
to be accidental o r artificial. Its reasons lie deep in the logic of th e situation.
C harles Kahn has suggested th a t the d ifferent A ristotelian distinctions rep re se n t
d ifferen t s tra ta in A ristotle's thinking. T hat may v e ry well be so, b u t we don't
u n d e rstan d A ristotle unless we also recognize the in trin sic logical connections
betw een th e d ifferent correlated classifications of his. No longer does it make any
sense to ask which of the several distinctions Aristotle "really" m eans, for th ey are
all inextricably intertw ined. The extensive controversies th a t have been prom pted
by th is question are simply otiose. (T his does not mean th a t differences of emphasis
are not called for here; c f . my comments below on those who stre s s the ties
betw een categories and question ty p e s .) The in te re stin g questions p ertain in stead
(in te r alia) to A ristotle’s aw areness of the connecting links between the differen t
d istinctions. Indeed, it is in spelling out th e main interrelations between th e
distinctions which converge in A ristotle’s theory of categories th at my
"tran scen d en tal deduction of th e categories" goes essentially beyond those earlier
scholars who have emphasized the m ulti-faceted ch aracter of A ristotelian categories.
Even though the reconstruction of A ristotelian categories which we have ju st
reach ed , perhaps does not ipso facto solve any major in terp retatio n al problem s, it
yields valuable clues which help to u n d e rstan d A ristotle and in many cases even
promise fu rth e r in sig h ts. For in stan ce, one problem we can now approach p e rta in s
to th e relation of A ristotle's theory to the facts of the Greek language.
T ren d elen b u rg , Apelt, and B enveniste have claimed th a t the A ristotelian
distinction betw een different categories reflects certain general featu res of th e
ancient Greek language. A ckrill's persuasive argum ents to the effect th a t what is
distinguished from each o th er in the category distinctions are not v erb al
expressions b u t entities may serve as an antidote to such excesses. However,
Ackrill’s th esis does not imply th a t A ristotle was not guided b y logical s tru c tu re s
which manifest them selves in the grammar of the Greek language. I cannot t r y to
w rite e ith er a transform ational grammar or a gam e-theoretical semantics for the
ancient Greek language. Suffice it merely to point out th a t the grammatical facts
which are highlighted by my treatm ent are less eye-catching b u t su b tler th a n those
flaunted by T rendelenburg and B enveniste. They p ertain to such th in g s as th e
id en tity (in form) of indefinite relatives with indirect in terro g ativ es in G reek, and
th e close relationship of both with quantifier w ords. These featu res of the Greek
The Unambiguity of Aristotelian Being 19

grammar serve to link the different correlated distinctions explained above to each
o th er especially closely, and th ereb y to motivate A ristotle’s th eo ry . If I had to find
linguistic evidence for my in terp retatio n of A ristotle, th a t is the direction in which
I could (and would) go. Even on the p resen t superficial level, it is not h a rd to see
th a t my treatm ent of instantiation works m utatis m utandis even b e tte r with Greek
th an with English.
Likewise, we are now in a position to draw an in te re stin g conclusion from o u r
observations. The different classés of questions with which A ristotle correlated his
o th er distinctions were prim arily in d irect questions. The correlation depends
crucially on an analogy between relative pronouns and question w o rd s, and th is
analogy (or near iden tity ) can obviously be b e st argued for b y comparing with each
o th er th e logical behavior of relative clauses and indirect questions. (An especially
useful Mittelglied here is the class of relative clauses without antecedents. T heir
logic is rem arkably similar to th a t of in direct q u estio n s.) A ristotle’s distinction
between different categories is less a distinction between d ifferen t question ty p es as
between question w ords, and it p ertain s to these words in so far as they are doing
du ty for th e ir relative clause tw ins.
This observation reflects somewhat unfavorably on those scholars who have
made much of the classification of questions as the alleged cornerstone of
Aristotelian categories. It seems to me th at th e ir th esis rem ains unproven.
Adm ittedly, the importance of the dialectical questioning games practiced in th e
Academy for A ristotle can scarcely be exaggerated. However, th e re is little
evidence in the Topics o r elsewhere th at th e theory of categories was developed for
(or from) such games.

7. Is th e reco n stru cted Aristotelian theory correct?

Categories v s. logical types

Here we come to some of th e most crucial questions concerning A ristotle’s

theory of categories. Is the theory correct as an analysis of the ’’logic” of th e
Greek language (or of the English language)? Are th ere differences between
d ifferent languages v is-à -v is A ristotelian categories? The "tran scen d en tal deduction"
of A ristotelian categories p resen ted above might seem to vindicate th e main featu res
of A ristotle's th eo ry . In spite of its p ersu asiv en ess, it nevertheless gives u s only
an approximation to the tru e semantical theory of natu ral language categories. It is
based upon assum ptions which are only p a rtly tru e , and hence it cannot be taken
The Unambiguity of A ristotelian Being1 20

as th e last and final word on the subject. Below -, I shall indicate one specific
limitation of my argum ent and consequently of our reco n stru cted theory of
A ristotelian categories. On a general theoretical level, another major shortcoming of
th e theory is obvious, connected with its relation to logical type distinctions. It is
in a wider logical and philosophical perspective clear th a t even Aristotelian category
distinctions must in the last analysis be based on type distinctions. The la tte r
distinctions may not coincide with Russell’s. Indeed, the ty p es (categories) of Frege
and R ussell seem to me too few and too fa r a p art to serve as a realistic basis of
our Sprachlogik. B ut, w hatever the requisite ty p es a re , they must serve as th e
foundation of any viable distinction betw een d ifferent categories. In o ther w ords,
some b rid g e has to be constructed from A ristotelian categories to logical ones to
vindicate them . How foreign modern type distinctions were for A ristotle is also
illu stra te d b y his deeply ingrained habit of brack etin g to g eth er the obtaining of
(what we would call) facts and the existence of individuals. (See e .g . Met. Z 17,
1041 a 14-16, b 4 -5 .)
This general problem is highlighted by th e more specific observation th a t ·.
A ristotelian categories tu rn out on my analysis to be quite different from logical
categories in the sense of logical ty p es. (T his point is relevant here among o th er
reasons because the contrary has been maintained by Gilbert Ryle; cf. Ryle
1937-1938.) Not only is i t the case th a t entities of a different logical type (in what
is roughly R ussell's sense) belong to the same category, as S o crates, man, and
animal all belong to th e so-called category of substance. T here is a sense in which
all categories come close to containing entities of the same logical ty p e. A fter all,
th ey all contain items which can be said of a substance like Socrates. For in stan ce,
th e members of the so-called category of relation are not relations for A ristotle, b u t
relatives (relational p red ic ates). This is amply shown by his discussion of th is
category in C at. 7, 8 a 35 f f . , especially his comments on correlatives and th e ir
epistemic and ontological interdependence. (C f. C at. 6, 6 a 35 f f . ;
De Soph. El. 31, 181 b 26-28; Top. VI, 4, 142 a 28-31; and VI, 8, 146 b 3 -4 .)
Likewise, quantities are not for A ristotle what we would mean by them ( e .g . a
le n g th ), b u t quantitative a ttrib u te s ( e .g . being of su ch-and-such le n g th ).
These observations are p erh ap s not v e ry su rp risin g . T here is a sense in
which the v e ry ’’category" of relation (as distinguished from relational pred icates)
came to its own only much late r in the histo ry of philosophy. (Cf. Weinberg 1965.)
However, the absence.of relations p ro p er from A ristotle's categorical scheme
highlights the problems it leads into. For where else can he put relations? The only
The Unambiguity of A ristotelian Being 21

propositional form he seems to recognize is the subject-predicate one. If some of

those p redicates are relational, we need an account as to how some of them can be
built of relations. A lternatively, we need a reduction of relational propositions to
subject-predicate propositions. N either task was attem pted by A ristotle, although
the la tte r one was undertaken by Leibniz, whose philosophy is in the last analysis
much more A ristotelian th an is usually recognized. (Cf. h ere Hintikka 1972b.)
Similar rem arks can be made about several o th er categories, especially about
the categories of q u a n tity , place’, time, and action.
These are illustrations of deeper and more w idespread tensions in A ristotle’s
th o u g h t. In tre a tin g (at least in its firs t stage) all categories on a p a r A ristotle (as
well as my rational reconstruction of his theory) fails to give a deeper account of
the rationale of category distinctions. It is for th is reason especially im portant to
realize th e differences between A ristotelian categories and logical ty p e s .
It is here th a t A ristotle’s relative neglect of the Frege-R ussell distinction
(even as a difference in use and not ju st as a difference in meaning) becomes a
handicap for him. Admittedly, it was claimed by Maier th a t A ristotle’s theo ry of
categories was calculated to accommodate certain distinctions between different
senses of "is” . (See vol. 2, p . 291 f f . ) Maier's distinctions include most of the
Frege-R ussell ones. Indeed, Maier's first two distinctions are identificatory being
v s. accidentally predicatory b eing, p . 280, and existential v s. copulative b ein g ,
p . 282. No major in sig h ts are forthcoming from Maier, how ever, into th e way
Aristotle managed to combine the Frege-R ussell distinction with his doctrine of
categories. For he firmly believes th a t, according to A ristotle, "immediate reflection
on th e concept of being [Maier’s e m p h a sis]... forms the principle of division fo r the
table of categories" (p p . 298-299). Maier’s immediacy claim notw ithstanding,
Aristotle himself does hot tr u s t immediate intuition h e re , b u t discusses the relation
of o th er categories to th a t of a substance. However, these argum ents are eith er
calculated to show the dependence of o th er categories on th a t of su b stan ce, o r
(which may come down to the same thing) to point out the role of focal meaning in
relatin g the being of the o th er categories to th a t of su b stan ce. They do not rely on
the kinds of distinctions which Maier mentions o r which a re likely to be made b y a
tw en tieth -century logician. All told, it th u s remains a m ystery — or at least an
unaccomplished explanatory task — how Aristotelian categories can be related to
such logical distinctions as those between different logical ty p es.
These difficulties offer u s an incentive to improve our in te rp re tiv e framework
so as to bridge the gap between Aristotelian categories and logical ones. We also
The Unambiguity of Aristotelian Being 22

have to tr y to u n d erstan d A ristotle’s own effo rts to overcome the separation

betw een d ifferent categories. The la tte r question has been discussed convincingly
by G. E. L. Owen, and I will not tre a t it h e re . The former question likewise admits
of d ifferen t kinds of approaches. At least two im portant lines of thought are
forthcom ing h ere.
Above we studied the ways in which expressions^for th e d ifferen t categories
e n te r into a complex proposition containing q u an tifiers. This study has to be
supplem ented by an account of the ways in which categorically d ifferen t expressions
e n te r into prim itive ( ’’atomic”) propositions. It was precisely such an account th a t
we saw earlier A ristotle failing to give in th e case of relations (and in the case of
several o th er categories). Admittedly, A ristotle argues for th e primacy of substance
over o th er categories, b u t he does not analyze the n ature of this primacy or th e
role of o th er categories in a way which would relate them to logical type
distinctions.
Secondly, th e v e ry stru c tu re of such quantifier p h rases as were discussed
above ( e .g . ’’every white horse which Alexander rode") can be analyzed fu rth e r.
One especially in te re stin g analysis has been offered by Joan B resnan. It can be
illu stra te d by comparing the nonrelative p a rt of those simple q u antifier p h rases we
have so f a r considered ("every white h o rse”) with an example of a rich er
s tru c tu re , e .g . "every two bottles of fine red wine".
N either of these lines of in q u iry seems to be connected with A ristotle's effo rts
to overcome categorical alienation in his m etaphysics, even though th e second is
related to his form -m atter co n tra st.
In spite of th e ir inconclusiveness, these considerations are highly relevant to
th e subject m atter of th is p a p e r, viz. to the question of the ambiguity or
nonambiguity of being. If all we are dealing with — o r ra th e r, A ristotle was dealing
with — is a m any-sorted th eo ry , th e re is no reason to th in k th a t any ambiguity of
th e v e rb s fo r being is implied by his th eo ry . In o rd e r to have given reasons fo r a
genuine am biguity, A ristotle ought to have given the kind of fu rth e r analysis of his
categories which would have related them to logical ty p e s. This he never does,
how ever, and hence the vaunted ambiguity of being between d ifferent Aristotelian
categories remains at b e st an unproven and u n arg u ed -fo r dogma. It is not even
clear w hether Aristotle held any clear-cut ambiguity th esis f o r ^ a t i as used in
d ifferent categories, for his terminology strongly su g g ests th a t he took himself to
be discussing differences in use ra th e r than am biguities. (C f. Hintikka 1959.)
The Unambiguity of A ristotelian Being 23

Bibliography

Ackrill, J . L .: 1963, A ristotle's "C ategories” and "De In te rp re ta tio n e ", Clarendon
P re ss, O xford.
/
A lbritton, R ogers: 1957, "Forms of P articular Substances in A ristotle’s
M etaphysics" , Journal of Philosophy, vol. 54 p p . 699-708.
i
Apelt, O .: 1891, B eiträge zur Geschichte d er griechischen Philosophie, Leipzig.

Balme, D. M.: 1980, "A ristotle’s Biology was not E ssen tialist", Archiv fü r
Geschichte d e r Philosophie, vol. 62, p p . 1-12.

B enardete, Seth: 1976-77, "The Grammar of B eing", Review of M etaphysics, vol.

30, p p . 486-496.

✓
B enveniste, Emile: 1966, ’’Catégories de pensée et catégories de langue", in
Problèmes de linguistique générale, Gallimard, P aris, p p . 63-74.

Bluck, Richard S .: 1975, Plato’s "Sophist": A Commentary, Gordon C. Neal ( e d .) ,

M anchester U niversity P re ss, M anchester.

ft
Bonitz, Hermann: 1853 "U ber die Kategorien des A ristoteles", Sitzungsberichte d e r
Kaiserlichen Akademie der W issenschaften, P h il.-H ist. K lasse, vol. 10, p p .
591-645, rep rin te d as a separate volume 1967 by W issenschaftliche
B uchgesellschaft, D arm stadt.

B resnan, Joan: unpublished seminar notes, "On th e Syntax of English Q uantifiers",

Stanford U niversity.

Dancy, R. M.: 1975, Sense and Contradiction: A Study in A risto tle, D. Reidel,
D ordrecht.

Dancy, R. M.: 1975, "On Some of A ristotle's F irst T houghts about Substances",
Philosophical Review, vol. 84, p p . 338-373.
The Unambiguity of A ristotelian Being 24

Dancy, R. M. : forthcom ing, "A ristotle and E xistence".

Frede, Michael: 1967, Prädikatiön und E xistenzaussage, Hypomnemata, vol. 18,

Vandenhoeck & R u p rech t, Göttingen.

Gomez-Lobo, Alfonso: 1976-77, "A ristotle’s Hypotheses· and Euclidean Postulates",

Review of M etaphysics, v o l.30, p p . 430-439.

Gomez-Lobo, Alfonso: 1980-81, "The So-Called Question of Existence in A ristotle,

A n. P o st. 2. 1 -2 .", Review of M etaphysics, vol. 34, p p .71-89.

G raeser, A ndreas: 1978, "A ristoteles und das Problem von Substanzialität und
Sein", F reib u rg er Z eitschrift ftlr Philosophie und Theologie, vol. 25, p p .
120-41.

H intikka, Jaakko: 1959, "A ristotle and the Ambiguity of Ambiguity", In q u iry , v o l.2,
p p . 137-151, re p rin ted with changes in Hintikka, Time and N ecessity, p p .
1-26.

H intikka, Jaakko: 1972a, "On the Ingredients of an A ristotelian Science", Nous,

vqI. 6, p p . 55-69.

H intikka, Jaakko: 1972b, "Leibniz on Plenitude, Relations, and the 'Reign of Law'",
in H arry G. F ran k fu rt ( e d .) , Leibniz: A Collection of Critical E ssay s,
Doubleday, Garden C ity, p p . 152-190, rep rin ted (with an im portant change)
1981, in Simo K nuuttila ( e d .) . Reforging the Great Chain of B eing, D. Reidel,
D ordrecht, p p . 259-286.

H intikka, Jaakko: 1973, Time and N ecessity, Clarendon P re ss, O xford.

H intikka, Jaakko: 1976, The Semantics of Questions and the Questions of Sem antics,
North-Holland, Amsterdam.

H intikka, Jaakko: 1979, " ’Is1, Semantical Games, and Semantical R elativity", Journal
of Philosophical Logic, vol. 8, p p . 433-468.
The Unambiguity of Aristotelian Being

H intikka, Jaakko: forthcom ing, "Semantical Games, the Alleged Ambiguity of ’I s ',
and Aristotelian C ategories".

Kahn, Charles: 1966, "The Greek Verb ’To Be’ and the Concept of Being",
Foundations of L anguage, vol. 2, p p . 245-265.
•V.

Kahn, C harles: 1973, The Verb ’Be' in Ancient G reek, D. Reidel, D ordrecht.
............ . ....." .........‘" " . I " " ........... ■■■*■■■■

Kahn, C harles: 1973, "On th e Theory of the Verb ’To Be”’ , in Milton K. Munitz
( e d .) . Logic and Ontology, New York U niversity P re ss, New Y ork, p p . 1-20.

Kahn, Charles: 1976, "Why Existence does not Emerge as a Distinct Concept in
Greek Philosophy", Archiv fü r Geschichte der Philosophie, yol. 58, p p .
323-334.

Kirwan, C hristopher: 1971, A ristotle’s M etaphysics, Books Γ, Δ, Ε Clarendon

Aristotle S eries, Clarendon P ress, O xford.

Loux, Michael J . : 1974, tran sla to r and ed ito r, Ockham’s T heory of T erm s, Notre
Dame U niversity P ress.

Maier, Heinrich: 1896-1900, Die Syllogistik des A ristoteles, 1-2, K .F . Koehler,

S tu ttg a rt, rep rin ted 1969 by Georg Olms, Hildesheim.

Mates, Benson: 1979, "Identity and Predication in Plato", P h ro n esis, vol. 24,
p p . 211-229.

Moravcsik, J.M .E .: 1967, ed ito r, A ristotle: A Collection of Critical E ssay s,

Doubleday, Garden City.

Moravcsik, J.M .E .: 1967, "A ristotle's Theory of Categories" in Moravcsik ( e d .) ,

A ristotle: A Collection of Critical E ssa y s, p p . 125-145.

Moravcsik, J . M. E .: 1976, "The Discernability of Identicals", Journal of

Philosophy, vol. 73, p p . 587-598.
The Unambiguity of A ristotelian Being 26

Owen, G. E. L .: 1960, "Logic and Metaphysics in Some E arlier Works of A ristotle",

in I. D üring and G .E .L . Owen ( e d s .) , Aristotle and Plato in M id-Fourth
C entury (Studia Graeca et Latina G othoburgensia, vol. ii), G öteborg, p p .
163-190.

Owen, G. E. L .: 1965, "A ristotle in the Snares of Ontology", in R. Bambrough

( e d .) , New Essays on Plato and A ristotle, Routledge and Kegan Paul, London,
p p . 69-95.

Owen, G. E. L .: 1978-79, "P articular and G eneral", Proceedings of th e A ristotelian

Society, Vol. 79, p p . 1-21.

Ross, W. D. : 1924, A ristotle's M etaphysics, A Revised T ext With Introduction and

Commentary, vols. 1-2, Clarendon P re ss, O xford.

Ryle, Gilbert: 1937-38, "C ategories", Proceedings of the A ristotelian Society, vol.
38, p p . 189-206, re p rin ted in 1971, Collected P a p e rs, vol. 2, H utchinson,
London, p p . 170-184.

Ryle, Gilbert: 1939, "Plato’s Parm enides", Mind, vol. 48, p p . 129-151, 302-325.

S aarinen, Esa: 1979, ed ito r, Game-Theoretical Sem antics, D. Reidel, D ordrecht.

T ren d elen burg, Adolf: 1846, Geschichte d e r K ategorienlehre, B ethge, B erlin.

Weidemann, Hermann: 1980, "In Defence of A ristotle’s T heory of P redication,"

P h ro n esis, vol. 25, p p . 76-87.

W einberg, Julius R. : 1965, "The Concept of Relation: Some O bservations on its

H isto ry", in A bstraction, Relation, and Induction: T hree Essays in the History
of T h o u g h t, The U niversity of Wisconsin P re ss, Madison, p p . 61-119.

Woods, M .J. : 1974-75, "Substance and Essence in A ristotle", Proceedings of the

A ristotelian Society, v o l.75, p p . 167-180.