<<

. 6
( 11)



>>

justi¬cation of this expunction see van Bennekom (1975), 400“1.
The phrase ˜sounds that are more than one but signi¬cant™ renders ˜plei»nwn m•n jwn¤n miŽv
51
shmantik¤n d•™ (1457a 4“5): for this construction cf. Cael. 1.6, 274a 28; 2.14, 296a 35“296b 1; Metaph.
A 3, 983b 17“18; Pol. 3.7, 1279a 35; 6.1, 1317a 22; Vahlen (1885), 213“14; Rostagni (1927), 81. A. D¨ ring
o
(1890), 365 construes ˜miŽv™ with ˜shmantik¤n™, understanding ˜miŽv «d”av™ or ˜miŽv jwn¦v™.
52 Cf. Gudeman (1934), 353; Pagliaro (1954), 85; Apostle (1980), 104; Zadro (1999), 217; Morgenstern
(2001), 279. ˜Conjunction™ can mean ˜connection™: see HA 10.7, 638b 8“9; PA 2.6, 652a 16“17; 3.7,
670a 9. In one passage (Rh. 3.5, 1407a 26“9) it seems to mean ˜connected item™ (cf. Belardi (1977),
258“60). The expression ˜with no conjunction™ is not used exclusively to express the absence of
connective particles: see Pr. 21.17, 929a 18; Rh. 3.6, 1407b 38.
53 APo. 2.10, 93b 35“7; Metaph. Z 4, 1030b 7“13; H 6, 1045a 12“14; Po. 20, 1457a 28“30 (cf. APo. 2.7, 92b 32).
54 Cf. Gallavotti (1974), 179; Zadro (1999), 217. On the unity of the Iliad ™s ˜action™ see Po. 8, 1451a 22“30;
26, 1462b 7“12.
166 ˜Empty™ terms
This objection can be answered. For the evidence it appeals to supports,
in fact, the position under attack. A striking characteristic of the Iliad is
that almost all its sentences are connected by some particle or other (one
reason might be that particles make it easier to satisfy the conditions of the
metre): Aristotle might well have in mind this characteristic of the Iliad
when he says that it is ˜single by virtue of some conjunction™.55

Classi¬cation or disambiguation? In T 38 Aristotle discusses how an utterance
can coincide with one or more assertions. Two interpretations of his task
are possible.
On the ¬rst interpretation, in T 38 Aristotle carries out a classi¬cation.56
More precisely, Aristotle draws a single classi¬cation of utterances based on
the number and the complexity of the assertions which a single utterance
amounts to. He ¬rst introduces two main groups: the group comprising
utterances that amount to a single assertion and the group comprising
utterances that amount to many assertions. Then he subdivides each of
these two main groups into two subordinate groups: the ¬rst subordinate
group of the ¬rst main group comprises utterances that ˜reveal a single
thing™, its second subordinate group comprises utterances that are ˜single by
virtue of some conjunction™; the ¬rst subordinate group of the second main
group comprises utterances that ˜reveal many things and not one™, its second
subordinate group comprises utterances that are ˜with no conjunction™.57
On the second interpretation, in T 38 Aristotle carries out a disam-
biguation.58 More precisely, Aristotle distinguishes two senses in which an
utterance can be called ˜a single assertion™ or ˜many assertions™: a semantic
sense and a syntactic sense. In the semantic sense, an utterance is called ˜a
single assertion™ if it ˜reveals a single thing™, while it is called ˜many assertions™
if it reveals ˜many things and not one™. In the syntactic sense, an utterance
is called ˜a single assertion™ if it is ˜single by virtue of some conjunction™,
while it is called ˜many assertions™ if it is ˜with no conjunction™.

A dif¬culty for the ¬rst interpretation. The ¬rst interpretation sits uneasily
with Aristotle™s remark at 17a 15“17: ˜A single assertoric sentence is either

55 I owe this point to Annamaria Schiaparelli.
56 Cf. Ockham in Int. i.iv 2, 2“17; 4, 2“13; Maurus (1668), i 66; Textor (1870), 25“6; Ackrill (1963),
125“7; von Fragstein (1967), 64“5; van Bennekom (1975), 404; Zanatta (1992), 183.
57 According to Maurus (1668), i 66, in T 38 Aristotle distinguishes not four, but three groups of
assertions: single simpliciter, single secundum quid, and multiple. This interpretation is untenable
because it presupposes an epexegetical sense of the ˜£™ at 17a 17, which is unlikely in view of the
disjunctive sense of the immediately preceding occurrences of ˜£™ (17a 15; 17a 16).
58 Cf. Ammon. in Int. 72, 15“22; Boeth. in Int. Sec. Ed. 110, 21“111, 12; Aquinas in Int. 103 Spiazzi;
Pacius (1597a), 70, 77; Weidemann (1994/2002), 195“6; Whitaker (1996), 75“6, 95.
˜Vacuous™ terms and ˜empty™ terms 167
that which reveals a single thing or that which is single by virtue of some
conjunction, while many are those which reveal many things and not one or
those with no conjunction.™ If, as the ¬rst interpretation assumes, Aristotle
is introducing a single classi¬cation of utterances, this remark seems to com-
mit him to the following: the classi¬cation™s ¬rst (second) main group com-
prises those utterances that amount to a single assertion (many assertions);
the ¬rst subordinate group of the ¬rst main group comprises utterances
which reveal a single thing, its second subordinate group covers utterances
connected by means of conjunctions; the ¬rst subordinate group of the sec-
ond main group comprises utterances which reveal many things, its second
subordinate group covers utterances not connected by means of conjunc-
tions. If the two subordinate groups of the ¬rst main group are disjoint (as
they should be), then every utterance from the second reveals many things
(otherwise it would reveal a single thing and therefore fall within the ¬rst
subordinate group). But then every utterance from the second subordinate
group of the ¬rst main group must also be in the ¬rst subordinate group
of the second main group. Then the second subordinate group of the ¬rst
main group is included in the ¬rst subordinate group of the second main
group. Hence the two main groups overlap.59

Solution for the ¬rst interpretation™s dif¬culty. The ¬rst interpretation may
solve this dif¬culty by assuming that the descriptions of the two subdivisions
of each main group are elliptical.60 Speci¬cally, the classi¬cation™s ¬rst main
group (which comprises utterances that amount to a single assertion) divides
into two subordinate groups: one comprises utterances which are not overtly
made up of further sentences61 and reveal one thing, the other comprises
utterances which are overtly made up of further sentences and link them
by means of conjunctions. The classi¬cation™s second main group (which
comprises utterances that amount to many assertions) undergoes a parallel
division whereby it articulates in two subordinate groups: one comprises
utterances which are not overtly made up of further sentences and reveal many
things, the other comprises utterances which are overtly made up of further
sentences and do not link them by means of conjunctions. On this account,
the second subordinate group of the ¬rst main group is (not included in,
but) disjoint from the ¬rst subordinate group of the second main group:

59 Cf. Ackrill (1963), 126; van Bennekom (1975), 404; Weidemann (1994/2002), 195.
60 Cf. Zanatta (1992), 185“6.
61 An utterance is overtly made up of further sentences just in case it has two or more parts which would
be regarded as grammatically complete sentences. For instance, an utterance of ˜Socrates came and
Coriscus came™ is overtly made up of further sentences, one of ˜Socrates came™ and one of ˜Socrates
and Coriscus came™ are not.
168 ˜Empty™ terms
every utterance in the second subordinate group of the ¬rst main group is
overtly made up of further sentences, and therefore does not fall within the
¬rst subordinate group of the second main group (which comprises only
utterances not overtly made up of further sentences).62

A dif¬culty for the second interpretation. Consider T 38™s ¬rst sentence: ˜The
¬rst single assertoric sentence is the af¬rmation, next is the denial; the
others are single by virtue of some conjunction™ (17a 8“9). Aristotle is thus
contrasting predicative assertions (which in this context coincide with af¬r-
mations and denials) with those assertions that are ˜single by virtue of some
conjunction™. He therefore seems committed to the claim that predicative
assertions are not ˜single by virtue of some conjunction™. Now, according
to the second interpretation, at 17a 15“16 Aristotle characterises utterances
to be called ˜a single assertion™ in the syntactic sense by saying that each of
them is ˜single by virtue of some conjunction™. He therefore seems com-
mitted to the claim that predicative assertions cannot be called ˜a single
assertion™ in the syntactic sense (otherwise, they would be ˜single by virtue
of some conjunction™). But such a claim is obviously unacceptable for him:
he surely thinks that predicative assertions can be called ˜a single assertion™
in the syntactic sense.63

Solution for the second interpretation™s dif¬culty. The second interpretation
may solve this dif¬culty by making two assumptions: (i) what Aristotle
means in T 38™s ¬rst sentence is that single assertions which are not predica-
tive are single merely ˜by virtue of some conjunction™, the implication being
that predicative assertions are single not ˜by virtue of some conjunction™
but by satisfying another stricter condition (e.g. containing only one verb);
(ii) when, at 17a 15“16, Aristotle characterises utterances to be called ˜a single
assertion™ in the syntactic sense by saying that each of them is ˜single by
virtue of some conjunction™, he means that these utterances satisfy at least
the condition of being ˜single by virtue of some conjunction™, the implica-
tion being that predicative assertions can be called ˜a single assertion™ in the
syntactic sense because they satisfy the other stricter condition.

Assessment of the two interpretations. Here are some considerations bearing
on the assessment of the two interpretations.

62 In Crivelli (2001), 252 I explored the possibility of solving the ¬rst interpretation™s dif¬culty by
textual emendation.
63 This dif¬culty seems to have escaped the commentators™ attention.
˜Vacuous™ terms and ˜empty™ terms 169
(i) T 38™s language suggests that it is not a disambiguation that is being
carried out. It is not simply that T 38 lacks formulae like ˜A single assertion
is spoken of in many ways™: this fact is not decisive because Aristotle does
sometimes carry out a disambiguation without using formulae of this sort.
Rather, what is important is that at the crucial points Aristotle does not use
˜to be spoken of ™ (˜l”gesqai™, which can also be rendered by ˜to be called™ or
˜to be said™). He says: ˜The ¬rst single assertoric sentence is the af¬rmation,
next is the denial; the others are single by virtue of some conjunction™ (17a 8“
9). He does not say: ˜A single assertoric sentence is spoken of primarily by
being an af¬rmation, next by being a denial; all other single assertoric
sentences are spoken of by virtue of some conjunction.™ Again, he says: ˜A
single assertoric sentence is either that which reveals a single thing or that
which is single by virtue of some conjunction, while many are those which
reveal many things and not one or those with no conjunction™ (17a 15“17).
He does not say: ˜A single assertoric sentence is spoken of either because it
reveals a single thing or because it is single by virtue of some conjunction,
while many are those which are spoken of because they reveal many things
and not one or because they are with no conjunction.™ But in Aristotle
an occurrence of ˜to be spoken of ™ is a hallmark of disambiguation. Just
consider that ˜to be spoken of ™ occurs at the beginning of each of the
thirty chapters of Metaphysics , the book dedicated to disambiguation: it
is here, if anywhere, that Aristotle could have safely omitted ˜to be spoken
of ™ because it could be understood from the context. This consideration
favours the ¬rst interpretation.
(ii) Suppose that, as the ¬rst interpretation assumes, T 38 drew a single
classi¬cation of utterances by distinguishing, ¬rst, two main groups and,
then, two subordinate groups within each main group. Then for each of the
subordinate groups the chapter would give separate conditions for being a
member of it, but for neither main group would the chapter give unitary
(non-disjunctive) conditions for membership.64 No such problem arises if
the second interpretation is right. This consideration favours the second
interpretation.
(iii) In Sophistici Elenchi 6 Aristotle says:
T 41 A single proposition is that which asks a single item about a single item.
(169a 11)
Suppose the ¬rst interpretation is right: T 38 classi¬es utterances by distin-
guishing two main groups and two subordinate groups within each main

64 Cf. Ackrill (1963), 125“6.
170 ˜Empty™ terms
group. Then single propositions as characterised in T 41 correspond to
those utterances which in T 38 are described as asserting ˜something about
something or something away from something™ (17a 21), which, in turn,
probably coincide with the utterances in the ¬rst subordinate group of the
¬rst main group. Then there is a tension between T 41 and T 38: what,
according to T 41, counts as a single proposition constitutes a subclass
of what, according to T 38, counts as a single assertion. If, however, as
the second interpretation assumes, T 38 is distinguishing a semantic and
a syntactic sense in which an utterance can be called ˜a single assertion™ or
˜many assertions™, then single propositions as characterised in T 41 can be
identi¬ed with those utterances which in T 38 are called ˜a single assertion™
in the semantic sense. Such an identi¬cation generates no tension. As in
the previous case, this consideration favours the second interpretation.
The above considerations leave the choice between T 38™s two interpre-
tations in the balance. However, one further consideration tilts it in favour
of the second interpretation. For, were T 38 offering a single classi¬cation
of utterances, as the ¬rst interpretation assumes, it would have set it up by
employing two radically different criteria: one is semantic (signifying one
thing vs signifying many things), the other syntactic (being tantamount to
an assertion composed by means of conjunctions vs being tantamount to a
collection of unconnected assertions).65 The second interpretation avoids
this disturbing consequence: in fact, according to the second interpretation,
one reason for distinguishing the two senses in which an utterance can be
called ˜a single assertion™ or ˜many assertions™ is that the criteria they invoke
are so different. Thus, the second interpretation should be preferred: T
38 distinguishes two senses in which an utterance can be called ˜a single
assertion™ or ˜many assertions™.66

Utterances that can be called ˜a single assertion™ in the semantic sense reveal a
single thing. Aristotle calls both af¬rmative and negative predicative asser-
tions ˜simple assertions™.67 The following considerations suggest that in
Aristotle™s view predicative assertions coincide with the utterances that can
be called ˜a single assertion™ in the semantic sense.68
(i) At the beginning of T 38, at 17a 8“9, Aristotle contrasts those single
assertions which are predicative assertions with those single assertions which
˜are single by virtue of some conjunction™. Since the latter are the utterances
65 Cf. Ackrill (1963), 127.
66 On this point I changed my mind with respect to Crivelli (2001), 252“4.
67 Int. 5, 17a 20“1; 17a 22“3 (cf. Pl. Sph. 262c6“7; 262c9“10; 263c1“3).
68 Cf. Zanatta (1992), 183; Weidemann (1994/2002), 194; Whitaker (1996), 77, 95.
˜Vacuous™ terms and ˜empty™ terms 171
that can be called ˜a single assertion™ in the syntactic sense, one feels inclined
to identify predicative assertions with those utterances that can be called ˜a
single assertion™ in the semantic sense.
(ii) In the second half of T 38 Aristotle says:
Of these the one is a simple assertion, i.e. that which asserts something about
something or something away from something, the other is compounded of these,
i.e. a sentence which is already composite. (17a 20“2)
The occurrence of ˜of these™ at the beginning of the above passage can-
not stand for ˜of utterances™: if it did, Aristotle would be introducing a
completely new division of utterances in which predicative assertions are
contrasted with all the rest. It is more plausible to assume that the relevant
occurrence of ˜of these™ stands for ˜of utterances that can be called “a single
assertion” in whatever sense™.
I suspect that for Aristotle utterances that can be called ˜a single assertion™
in the semantic sense include also some assertions that are not predicative:
in particular, I suspect that in his view they include assertions concerning
simple objects (which are existential assertions of a special kind).69 The
reason why Aristotle fails to mention these other assertions as cases of
utterances that can be called ˜a single assertion™ in the semantic sense could
be that in de Interpretatione he focuses on predicative assertions and he
tends to forget about simple assertions of other kinds.

Utterances that can be called ˜a single assertion™ in the syntactic sense comprise
not only those utterances that can be called ˜a single assertion™ in the seman-
tic sense, but also those that are equivalent to utterances constructed from
several assertions linked by connective particles. I use ˜utterance equivalent
to an utterance constructed from several assertions linked by connective par-
ticles™ to denote both utterances constructed from several assertions linked
by connective particles and utterances that could be reformulated as utter-
ances constructed from several assertions linked by connective particles.70
Note that in T 35 Aristotle focuses on an utterance of ˜A cloak is white™
that he regards as equivalent to one of ˜A horse is white and a man is white™
(I shall discuss Aristotle™s argument in a moment).
Aristotle probably holds that a composite assertion ˜reveals many things™
in that the utterance to which it is equivalent is constructed from several
assertions that reveal different things. None the less, a composite asser-
tion amounts to a single assertion because the connective particles of its

69 70
Cf. sect. 1 of ch. 3. Cf. Weidemann (1994/2002), 197.
172 ˜Empty™ terms
equivalent reformulation somehow link the different things revealed by its
component assertions. Aristotle therefore probably has in mind something
like the following: every utterance that can be called ˜a single assertion™ in
the semantic sense, i.e. every simple assertion, i.e. every predicative asser-
tion, reveals a single thing in that it introduces only one state of affairs;
by contrast, every utterance that can be called ˜a single assertion™ in the
syntactic sense either reveals a single thing (and therefore can be called ˜a
single assertion™ also in the semantic sense) or ˜reveals many things™ in that
it introduces many states of affairs which it somehow links. These links are
expressed by the connective particles which surface in the composite asser-
tion to which the utterance that ˜reveals many things™ is equivalent (for the
presence of connective particles brings it about that the whole composite
assertion says something about how these states of affairs are related).71
Here Aristotle seems to reserve a niche for those composite assertions
which were later studied by Peripatetic and Stoic logicians.72 He seems
to regard these composite assertions as truth-evaluable, but he seems to
assume that there is no single state of affairs corresponding to any of them.
This is in line with his denial of ˜negative™ states of affairs: as there is no
˜negative™ state of affairs corresponding to a negative assertion, so there is no
˜conjunctive™ or ˜disjunctive™ state of affairs corresponding to a conjunctive
or disjunctive assertion. Aristotle™s study of these composite assertions was
probably prompted by his desire to develop a hint of Plato™s in the Sophist,73
where af¬rmative predicative assertions are described as being the ˜smallest™
and ˜shortest™ of assertions “ such a description leaves the reader with the
curiosity to discover what the ˜larger™ and ˜longer™ assertions could be.

Truth-evaluability and different truth-values assigned to contradictories. When
Aristotle distinguishes a semantic and a syntactic sense in which an utter-
ance can be called ˜a single assertion™ or ˜many assertions™, one of his aims is
probably to separate the issue of truth-evaluability from that of the assign-
ment of different truth-values to contradictories. Let me explain. If an
71 Cf. Simons (1988), 111“12.
72 This is the interpretation adopted by all ancient commentators (Ammon. in Int. 66, 31“67, 19; 73,
19“74, 14; Boeth. in Int. Pr. Ed. 72, 6“10; 75, 9“11; 77, 24“78, 4; in Int. Sec. Ed. 96, 28“97, 18; 105,
1“18; 115, 23“116, 6; Steph. in Int. 20, 4“10; 20, 16“24; anon. in Int. 18, 7“19) and endorsed by many
medievals (e.g. Al-Farabi in Int. Comm. apud Zimmermann (1981), 46, 47“8; Ockham in Int. i.iv 2,
6“16; 4, 9“10; 4, 14“15; 6, 11“14) and some modern writers (Brandis (1862/64), i 437; von Kirchmann
(1876), 64; DuLac (1949), 168; Antiseri (1969), 53“7; Weidemann (1994/2002), 197“8; Slomkowski
(1997), 126; Damschen (2001), 127). According to other modern commentators (e.g. Lukasiewicz
(1957), 131“2; Sainati (1968), 53; D. Frede (1970), 20, 75“6; Patzig (1973), 922; Astroh (1981), 341,
344“5; D. Frede (1985), 77“8; Bobzien (2002), 364), Aristotle never allowed a single assertion to
consist of sentences linked by connective particles. In Metaph. 4, 1008a 4“7 Aristotle ascribes truth
to a conjunctive assertion which he describes as consisting of further assertions.
73 262c 7; 262c 10; 263c 3.
˜Vacuous™ terms and ˜empty™ terms 173
utterance can be called ˜a single assertion™ in any (either semantic or syn-
tactic) sense, then it is truth-evaluable, i.e. it is an utterance with regard to
which the question ˜Is it true or false?™ can be reasonably asked. If instead
the utterance can be called ˜many assertions™ in the syntactic as well as in
the semantic sense, then it is not an assertion (in fact, it is a collection of
assertions, and a collection of assertions is not an assertion), and therefore
it is not truth-evaluable. However, if an utterance can be called ˜a single
assertion™ only in the syntactic sense, but can be called ˜many assertions™ in
the semantic sense, then the utterance, although truth-evaluable, can con-
stitute an exception to the principle that contradictories should be assigned
different truth-values. This can be clearly seen in Aristotle™s discussion in
T 35, which comes back to some of the points made in T 38 (T 35 is an
excerpt of chapter 8, while T 38 comprises chapter 5 and the beginning of
chapter 6 of de Interpretatione).
In T 35 Aristotle considers the case of a single name given to ˜two items
which do not make up a single one™ (18a 18). He illustrates it by imagining
somebody giving ˜the name “cloak” to horse and man™ (18a 19“20). Then
(18a 20) he considers an utterance of ˜A cloak is white™. He thinks that
an utterance of ˜A cloak is white™ can be analysed in two ways: on its
second analysis it ˜signi¬es [. . .] nothing (for it is not the case that some
man is a horse)™ (18a 25“6);74 on its ¬rst analysis it signi¬es the same as an
utterance of ˜A horse and a man is white™, which in turn ˜does not differ
at all from saying “A horse is white and a man is white”™ (18a 23). Thus, an
utterance of
[a] A cloak is white
on its ¬rst analysis signi¬es the same as an utterance of
[b] A horse and a man is white,
which in turn signi¬es the same as an utterance of
[c] A horse is white and a man is white.
Aristotle says that utterances of [b] and [c] ˜signify many things and are
many™ (18a 24). Aristotle™s reason for saying that an utterance of [c] signi¬es
many things is clear: it is a composite assertion consisting of two further
sentences, each of which introduces a different state of affairs. Since an
utterance of [c] signi¬es many things, it can be called ˜many assertions™
in the semantic sense, i.e. ˜it is many™. Since [c] is a paraphrase of [b], an
utterance of [b] also signi¬es many things and is many. From this, and

74 For the second analysis see the subsection to which n. 10 above pertains.
174 ˜Empty™ terms
from the assumption that [b] and [c] are paraphrases of [a], Aristotle infers
that an utterance of [a] also signi¬es many things (18a 25). What he has in
mind is probably that an utterance of [a], despite its appearance of being
a simple predicative assertion, is really a composite assertion, and signi¬es
many things because it introduces many states of affairs (the state of affairs
of a horse being white and that of a man being white). It therefore can be
called ˜many assertions™ in the semantic sense. None the less, it can be called
˜a single assertion™ in the syntactic sense, and as such it is truth-evaluable.
Now consider an utterance of
[d] A cloak is not white.
An utterance of [d] counts, or at least appears to count, as the contradictory
of one of [a]. An analysis analogous to the one above shows that this
utterance of [d] is equivalent to an utterance of
[e] A horse and a man is not white,
which in turn signi¬es the same as an utterance of
[f ]A horse is not white and a man is not white.
As in the case of the utterance of [a], the utterance of [d], despite its
appearance of being a simple predicative assertion, is really a composite
assertion, and signi¬es many things because it introduces many states of
affairs. Both the utterance of [a] and that of [d] are truth-evaluable because
they both can be called ˜a single assertion™ in the syntactic sense. However,
they can both be false: if every horse is white and no man is, or if no
horse is white and every man is, then both utterances are false. Thus,
our utterances of [a] and [d] can create an exception (perhaps a merely
apparent exception) to the principle that requires assigning different truth-
values to contradictories. As Aristotle himself puts it, in this case, ˜too,75 it
is not necessary that one contradictory should be true and the other false™
(18a 26“7).76

Generalisation to all utterances with ˜empty™ names. In the situation imagined
in de Interpretatione 8, ˜cloak™ is an ˜empty™ name (because it signi¬es neither
an existent universal nor “ of course “ an individual which at some time
75 Another case where ˜it is not necessary that one contradictory should be true and the other false™ is
that of indeterminate predicative assertions, already discussed in de Interpretatione 7 (17b 29“37, cf.
18a 10“11; 9, 18a 29“33).
76 Cf. Ackrill (1963), 132; Geach (1963), 18; Whitaker (1996), 100, 105“6, 109, 112; Wheeler (1999), 219.
In Crivelli (2001), 262“4 I favoured a different interpretation: that an utterance of [a] and one of [d]
should both be neither true nor false. This exegesis, which I now reject, is endorsed by Weidemann
(1994/2002), 221“2.
˜Vacuous™ terms and ˜empty™ terms 175
or other exists). Would Aristotle endorse a generalisation of his remarks
about an utterance of ˜A cloak is white™ that applies to every utterance
that looks like a predicative assertion whose would-be subject or would-be
predicate is ˜empty™? Some commentators77 answer af¬rmatively. Others78
instead answer negatively: they think that for Aristotle ˜empty™ expressions
like ˜cloak™ have a different semantic behaviour from ˜empty™ expressions
like ˜goatstag™ (because they think that for Aristotle expressions like ˜goat-
stag™ have nominal de¬nitions which are ¬xed by the myths or narratives
in which they are introduced, while expressions like ˜cloak™ lack nominal
de¬nitions).
The issue cannot be settled con¬dently. However, since Aristotle never
states that expressions like ˜cloak™ behave differently from expressions like
˜goatstag™,79 there is some likelihood in assuming that Aristotle would
endorse a generalisation of his remarks about an utterance of ˜A cloak is
white™ that applies to every utterance that looks like a predicative assertion
whose would-be subject or would-be predicate is ˜empty™. If this is correct,
we can answer the question asked at the end of the preceding section “
the question whether Aristotle™s philosophy of language can accommodate
utterances that look like predicative assertions whose would-be subject or
would-be predicate is ˜empty™. The answer is af¬rmative. Aristotle™s phi-
losophy of language can accommodate such utterances: it treats them as
equivalent to utterances constructed from several assertions linked by con-
nective particles.
Aristotle does not show how every utterance which appears to be a pred-
icative assertion involving an ˜empty™ name can be analysed as a composite
assertion. In fact, had he tried to do this, he would have encountered chal-
lenging, perhaps insurmountable, dif¬culties: just consider how wrong it
would be to treat ˜Every goatstag is white™ as equivalent to ˜Every animal is
white and everything with a stag™s head is white and everything with a goat™s
body is white™.80 He probably took his job mainly to be providing a logical
account of the basic building-blocks of his logic, i.e. (genuine) predicative
assertions, and gave nothing more than a glance to those utterances which
appear to be predicative assertions.

A solution for the problems of the Square of Opposition. As I said, the problems
that seem to af¬‚ict the laws of the Square of Opposition evaporate if in

77 78 E.g. Modrak (2001), 25“6, 47“8.
E.g. Bolton (1976), 529; Charles (2000), 91.
79 When Aristotle says that ˜“goatstag” signi¬es something™ (Int. 1, 16a 16“17, cf. n. 40 above), he is
not committing himself to the view that ˜goatstag™ signi¬es a single universal: he simply means that
˜goatstag™ is not devoid of signi¬cation.
80 Cf. Geach (1968), 47; Weidemann (1994/2002), 197.
176 ˜Empty™ terms
every quanti¬ed predicative assertion the subject is ˜non-empty™.81 As for
those utterances that seem counter-examples to the laws of the Square of
Opposition, they are not real counter-examples.82 Despite appearances,
they are not quanti¬ed predicative assertions. They are assertions, and as
such are truth-evaluable, but they are composite assertions: as such they
fall outside the scope of the laws of the Square of Opposition. The laws of
the Square of Opposition govern only those assertions which in Aristotle™s
view are most basic, i.e. quanti¬ed predicative assertions.

Logical form and grammatical form. According to some commentators,83 an
important difference between Aristotle™s and Russell™s logical views is that
while for Aristotle logical form follows closely grammatical form, Russell
separates the two. However, if the foregoing reconstruction of Aristotle™s
ideas about utterances which have the appearance of a predicative assertion
whose subject is ˜empty™ is correct, Aristotle does separate logical form from
grammatical form.
Speci¬cally, there is an analogy between Aristotle™s treatment of utter-
ances of the aforesaid kind and Russell™s treatment of ˜ordinary™ proper
names. The possibility that de¬nite descriptions could be ˜empty™ prompts
Russell to treat them not as ˜logically™ proper names but as quanti¬ers
(whose logical behaviour is exhibited by paraphrasing the sentences where
they occur as wholes). The analogous possibility that ˜ordinary™ proper
names might be ˜empty™ induces Russell to treat them, too, not as ˜logi-
cally™ proper names but as disguised descriptions, and therefore as disguised
quanti¬ers whose logical behaviour is exhibited by paraphrasing the origi-
nal sentences as wholes. Aristotle™s moves are analogous: when an ˜ordinary™
common noun is ˜empty™, he treats the utterance where it occurs as a com-
posite assertion.

˜Making many questions into one™. At the beginning of chapter 11 of de
Interpretatione Aristotle returns to the themes of chapters 5 and 8:
T 42 To af¬rm or to deny a single item about many, or many about a single one, is
not a single af¬rmation nor a single denial unless the many items make up a
single one. I do not speak of ˜a single item™ if a single name is given but those
items do not make up a single one: e.g. a man is perhaps an animal and biped
and tame, but these do make up a single item, while white and man and
walking do not make up a single item. Hence, if one af¬rms a single item
of these, there will not be a single af¬rmation, but a single spoken sound
81 Cf. the paragraph to which n. 34 above is appended.
82 83
Cf. Mann (1958), 15; Charles (1994), 53; (1997), 236“7; (2000), 93. E.g. Rees (1960), 139.
˜Vacuous™ terms and ˜empty™ terms 177
and many af¬rmations, nor if one af¬rms these of a single item, but again
there will be many af¬rmations. Therefore, if a dialectical question requires
as an answer either the proposition or the other part of the contradictory
pair, and a proposition is a part of a single contradictory pair, there will not
be a single answer to these (for the question also is not single),84 not even if
it is true. These issues have been discussed in the Topics. (20b 13“26)
In T 35 (an excerpt from de Interpretatione 8) Aristotle claimed that the
universals horse and man do not constitute a single item. I understood
him as thereby committing himself to the claim that there is no universal
horse-and-man which is the ˜logical product™ of the universals horse and
man, and I took his ground to be that had there been such a universal it
would not have always been predicated of some existing individual.85 Now,
in T 42 (the beginning of de Interpretatione 11), Aristotle claims that the
universals walking, white, and man do not constitute a single item. He
can be plausibly understood as thereby committing himself to the claim
that there is no universal walking-white-man which is the ˜logical product™
of the universals walking, white, and man. But his ground for this claim
cannot be that the would-be universal walking-white-man would not be
always predicated of some existing individual: there exist individuals who
are men, white, and walking, and it could well be the case that there
always were and always will be some. Aristotle™s grounds must be different.
One of them is probably that the universals walking, white, and man lack
the connectedness exhibited by the items signi¬ed by the members of the
de¬niens of a de¬nition.86
Thus, the universals walking, white, and man do not constitute a single
item, so that there is no universal walking-white-man which is the ˜logical
product™ of them. If one introduced a single name “ ˜whitewalker™, say “
which signi¬ed the universals walking, white, and man in the same way as
the phrase ˜walking white man™ signi¬es them, and produced an utterance
that looked like a predicative assertion where an utterance of that name
was contained as subject or predicate, the utterance would be called ˜many
assertions™ in the semantic sense (though it could be called ˜a single asser-
tion™ in the syntactic sense). As I pointed out earlier,87 Aristotle cannot
be discussing here the issue of ambiguity: for ˜walking white man™ is an
unambiguous expression, so that ˜whitewalker™ also must be unambiguous
(because it is supposed to signify in the same way as ˜walking white man™).
84 My punctuation here (20b 25), different from those of the editions I consulted, follows a suggestion
of Ammonius (in Int. 201, 24“6).
85 86 Cf. n. 43 above.
Cf. the portion of the main text to which n. 21 above is appended.
87 Cf. the paragraph to which n. 17 above is appended.
178 ˜Empty™ terms
The last part of T 42 (20b 22“6) connects the previous points with the
issue of a questioner in a dialectical debate ˜making many questions into
one™, and declares that this subject was discussed ˜in the Topics™ (20b 26). The
issue of a questioner ˜making many questions into one™ is discussed in the
Topics 88 and in Sophistici Elenchi.89 In both works Aristotle distinguishes
two cases where a questioner ˜makes many questions into one™. In the ¬rst
case, the multiplicity of the questions asked is explicit in that it surfaces
in the linguistic expression: e.g. the multiplicity of the questions asked in
˜Are Coriscus and Callias at home?™ is explicit because it surfaces in the
phrase ˜Coriscus and Callias™. In the second case, the multiplicity of the
questions asked is disguised, and the disguise is provided by ambiguity:
e.g. the multiplicity of the questions asked in ˜Is a vice a mental state?™ is
disguised because of the ambiguity of ˜vice™ (which can signify both a moral
condition and a mechanical instrument).
But now T 42™s reference to the Topics raises two problems: (i) the only
disguised multiple questions discussed in the Topics and in Sophistici Elenchi
are those whose disguise is provided by ambiguity, a situation de Interpre-
tatione seems to ignore; (ii) the only disguised multiple questions discussed
in de Interpretatione seem to be those whose disguise involves no ambiguity,
a situation the Topics and Sophistici Elenchi ignore.
I shall discuss these two problems in turn. In particular, I shall tackle the
¬rst by arguing that disguised multiple questions whose disguise is provided
by ambiguity are discussed in de Interpretatione. As for the second problem,
I shall argue that the reason why neither the Topics nor Sophistici Elenchi
discuss disguised multiple questions whose disguise involves no ambiguity
is that Aristotle had not yet become aware of this possible situation.

Ambiguity. Does the case where the multiplicity of the questions asked
is disguised behind ambiguity ¬nd any collocation in de Interpretatione™s
schema? Consider an utterance of
[g] A vice is a mental state.
Since the word ˜vice™ is ambiguous, an utterance of [g] can be taken to make
many assertions. These assertions are reciprocally unrelated in that they
are not linked either as if by a conjunctive connective or as if by a disjunctive
connective. To see this, imagine a dialectical debate about virtue where the
answerer endorses [g]: a perverse questioner could attack this answer by
saying that mechanical instruments are not mental states. In doing so, the
questioner would be treating the utterance of [g] as the ˜conjunction of its

88 8.7, 160a 17“34. 89 6, 169a 6“18; 17, 175b 39“176a 18; 30, 181a 36“181b 18.
˜Vacuous™ terms and ˜empty™ terms 179
senses™. Again, imagine a dialectical debate about mechanical instruments
which ended up talking about clamps. One can then imagine the answerer
saying ˜A vice is not a mental state™. Then a perverse questioner could attack
this answer by uttering [g]. In doing so, the questioner would be treating
the utterance of [g] as the ˜disjunction of its senses™. Thus, ambiguous
utterances are open both to a conjunctive and to a disjunctive reading “
which is because, properly speaking, ambiguous utterances introduce many
assertions which are not connected in any way, either conjunctively or
disjunctively. This situation suggests that an ambiguous utterance can be
regarded as equivalent to many assertions ˜with no conjunction™ (17a 17),
and therefore can be called ˜many assertions™ not only in the semantic, but
also in the syntactic, sense. Thus, after all, ambiguous utterances probably
do have a niche in de Interpretatione™s schema.

A developmental hypothesis. The Topics and Sophistici Elenchi present par-
tially con¬‚icting views about a questioner in a dialectical debate ˜making
many questions into one™. In the Topics 90 Aristotle advises the answerer to
give a single answer (˜Yes™ or ˜No™) to the multiple question just in case
that single answer would have seemed true for each of the many questions
concealed behind the single but ambiguous interrogative utterance. By con-
trast, in Sophistici Elenchi 91 Aristotle rules that the answerer should never
give a single answer to a multiple question, not even if some single answer
would have seemed to be true for each of the many questions concealed
behind the single but ambiguous interrogative utterance. He points out
that this rule con¬‚icts with the advice of certain dialecticians who urge
the answerer to give a single answer to the multiple question if that single
answer seems true for each of the many questions comprised in the multi-
ple question. These dialecticians remain unnamed, but it is reasonable to
assume that Aristotle is politely referring to himself, and hinting that he
has changed his mind.92
As for the relation between the Topics and Sophistici Elenchi, Aristotle
probably ¬rst wrote the eight books of the Topics as a complete work, and
then added Sophistici Elenchi as a ninth book.93 What Aristotle says near
the end of T 42 agrees with the position of Sophistici Elenchi rather than
that of the Topics: for near the end of T 42 (at 20b 25“6) Aristotle urges the
answerer in a dialectical debate to avoid giving a single answer to a multiple
question even if that single answer would have seemed true for each of
the many questions comprised in the multiple question. One can therefore
90 8.7, 160a 17“34. 91 17, 175b 39“176a 18; 30, 181a 36“181b 18.
92 Cf. Dorion (1995), 30“1, 334; Brunschwig (1999), 87“91.
93 Cf. Brunschwig (1967), xviii“xx; Dorion (1995), 25; Brunschwig (1999), 90.
180 ˜Empty™ terms
plausibly assume that the reference to the Topics at the end of T 42 concerns
the later version of the work which included Sophistici Elenchi.94
I would now like to ¬‚oat a speculative suggestion concerning the rela-
tionship between the Topics, Sophistici Elenchi, and de Interpretatione on the
issue of a questioner ˜making many questions into one™. The suggestion is
that there are three stages of Aristotle™s re¬‚ection on this issue.
In the earliest stage, represented by the version of the Topics in eight
books, Aristotle considers only apparently single, but really multiple ques-
tions which depend on ambiguity, and advises the answerer to offer a single
reply if it seems to be true for all of the ambiguous question™s senses.
In the intermediate stage, represented by the later version of the Topics
that includes Sophistici Elenchi as a ninth book, Aristotle still considers only
apparently single, but really multiple questions which depend on ambiguity,
but advises the answerer never to offer a single reply (even if some single
answer would have seemed to be true for all of the ambiguous question™s
senses).
Finally, in the latest stage, represented by de Interpretatione, Aristotle
considers a larger range of apparently single, but really multiple questions:
not only those that depend on ambiguity, but also those that introduce
many states of affairs with no ambiguity. Concealed multiple questions
that do depend on an ambiguity correspond to those utterances which
can be called ˜many assertions™ both in the semantic and in the syntactic
sense; concealed multiple questions that do not depend on an ambigu-
ity correspond to those utterances which can be called ˜many assertions™
only in the semantic sense, but can be called ˜a single assertion™ in the syn-
tactic sense. One reason why de Interpretatione concentrates on this last
group of utterances (concealed multiple questions that do not depend on
an ambiguity) is that it constitutes a novelty with respect to the earlier
works.
94 Cf. B™s scholium on 20b 26 (apud Waitz (1844/46), i 42); Pacius (1597a), 92; Bonitz (1870), 98a 1“2,
102a 48“9; Dorion (1995), 24.
pa rt ii i
Truth and time
c ha p t e r 6

Truth and change




This chapter addresses the relationship of truth to time and change. Accord-
ing to Aristotle, any bearer of truth or falsehood can, at least in principle, be
true at one time and false at another. This is somewhat hard to square with
his view that the linguistic items that are bearers of truth or falsehood are
utterances (expression-tokens). On re¬‚ection, however, the position turns
out to be consistent: utterances (expression-tokens) can be true at one time
and false at another.
Aristotle also claims that if an assertion or a belief is true at one time
and false at another, it does not follow that it undergoes a change. This is
probably due to the idea that truth (being correspondence to the world)
is something like a relative and therefore, like relatives, is involved at most
in a ˜mere Cambridge change™ that does not count as a genuine change.
Now, properties involved in a ˜mere Cambridge change™ are not genuine
properties. It follows that truth is not a genuine property.

1 d if ferent tru th- va lue s at di f f erent t i m es
Truth and falsehood at times. Aristotle thinks that the bearers of truth or
falsehood are true, or false, at times. He does not formulate this view
explicitly: nowhere does he use a sentence like ˜States of affairs, beliefs, and
assertions are true or false not absolutely, but at times.™ Rather, this view is
presupposed by the way he speaks about truth and falsehood, which suggests
that he would aver the following claim:
[28] An utterance of an instance of the schema ˜s is true at t™ or of ˜s is
false at t™ is a proper formulation. An utterance of an instance of the
schema ˜s is true™ or of ˜s is false™ is either improper or elliptical. (The
intended instances of these schemata are generated by replacing the
occurrences of ˜s™ by designations of bearers of truth or falsehood, and
the occurrences of ˜t™ by designations of times.)

183
184 Truth and time
It is worthwhile recalling the often noted1 fact that Aristotle™s view that the
bearers of truth or falsehood are true, or false, at times was widely shared
in Antiquity “ in fact, it remained unchallenged. It is unclear whether the
times at which bearers of truth or falsehood are true, or false, are instants
or periods. By speaking of ˜times™ I remain neutral on this issue.
Not only does Aristotle think that the bearers of truth or falsehood are
true, or false, at times. He also thinks that some bearer of truth or falsehood
is true at one time and false at another.2 In particular, he thinks this of states
of affairs, beliefs, and assertions.3

A modern attack on Aristotle™s conception of truth and falsehood at times. Some
modern commentators4 attack Aristotle™s view that the bearers of truth or
falsehood are true, or false, at times. The main ground for this attack is de
Interpretatione 9, where Aristotle addresses a paradox concerning truth and
time: what is primarily responsible for de Interpretatione 9™s paradox would
be, in the eyes of these commentators, Aristotle™s mistaken view that the
bearers of truth or falsehood are true, or false, at times.
This modern attack is questionable. For modern tense logic rests on the
assumption that the bearers of truth or falsehood are true, or false, at times,
but modern philosophers do not normally attack tense logic because it relies
on this assumption. Of course, this is not a knock-down answer to someone
criticising the assumption that the bearers of truth or falsehood are true,
or false, at times. But it does seem inappropriate to chastise Aristotle for a
thesis that is still assumed by a respected branch of logic.5

Utterances with different truth-values at different times. I previously argued
that for Aristotle every sentence which is true or false is an utterance
(a sentence-token, not a sentence-type).6 Suppose this is correct. Now,
1 Prior (1957), 104; Hintikka (1967/73), 62“8, 70“2; M. J. White (1981), 233“4; Mignucci (1985a), 164“9;
(1985b), 221“4; von Kutschera (1986), 203; Pardey (2000), 33“6; Goldin (2002), 240“7.
2 Cf. Maier (1896/1936), i 90“1; Oehler (1962/85), 179“80; Miller (1971), 19; Hintikka (1967/73), 66“7;
M. Kneale (1974), 369; Oehler (1984), 222“3; Mignucci (1985b), 219“24; Berti (1992), 87; Liske (1995),
352“3; Seel (2001a), 218.
3 For states of affairs see Cat. 5, 4a 34“4b 2; 4b 8“10; APo. 1.6, 74b 32“9; Top. 8.2, 157b 11“16; de An.
3.3, 428b 4“9; Metaph. 29, 1024b 17“21; 10, 1051b 9“17. For beliefs see Cat. 5, 4a 23“8; 4a 34“4b 2;
4b 4“13; APo. 1.6, 74b 32“9; Top. 8.2, 157b 11“16; SE 22, 178b 24“9; de An. 3.3, 428b 4“9; Metaph. Z 15,
1039b 27“1040a 5; 10, 1051b 9“17; 1052a 4“11. For assertions see Cat. 5, 4a 23“6; 4a 34“4b 2; 4b 4“13; SE
22, 178b 24“9; GC 2.11, 337b 4“5; de An. 3.3, 428b 7“9; Metaph. 8, 1012b 24“6; 10, 1051b 9“17; K 6,
1063a 17“21.
4 Saunders (1958), 373“4, 377; Scholz (1959), 77; Kneale/Kneale (1962), 51; van Eck (1988), 27“8.
5 For a defence of the tensed ascription of truth and falsehood see Gale (1961), 379“82; Haack (1974/96),
82“3.
6 See sect. 3 of ch. 1.
Truth and change 185
Aristotle thinks that some sentence is true at one time and false at another.
Aristotle™s views then entail that some utterance is true at one time and false
at another. Gabriel Nuchelmans and other commentators7 commit them-
selves to the claim that no utterance is true at one time and false at another.
So, if Nuchelmans and those who agree with him are right, Aristotle is
wrong.
Aristotle might well be wrong. His mistake might derive either from
unawareness of the difference between sentence-tokens and sentence-types,
or from a failure to think out some of the implications of what he says
(some of his claims entailing that the sentences which are true or false
should be utterances, others that they should be sentence-types “ hence
not utterances).
However, an account of the situation that is more charitable to Aristotle
ought to be attempted. Perhaps it is Nuchelmans™s view that is wrong “
perhaps some utterance is true at one time and false at another. Here are
two suggestions as to how this might be the case.

Suggestion (i). Utterance u of ˜Socrates is seated™ takes three seconds. Dur-
ing these three seconds Socrates stands up. u has different truth-values at
different times within the three seconds: it is true at t1 (for t1 lies within the
¬rst of the three seconds, and Socrates is then seated) and false at t2 (for t2
lies within the third of the three seconds, and Socrates then is not seated).8
Suggestion (i) cannot do: it allows an utterance to have different truth-
values only over the relatively short portion of time it occupies, while on
reading the passages where Aristotle says that the same bearer of truth or
falsehood can have different truth-values at different times one feels that he
intends to allow it to have different truth-values over a far longer portion
of time.

Suggestion (ii). One morning in May 1992 Jim shouts ˜Labour represents the
majority of the English nation™. In June 1997 Jim, re¬‚ecting on the events of
his life in 1992, makes a statement by using the sentence ˜That loud remark
was false at the time it was made but is true now™. What Jim says in June
1997 is both appropriate and true. Hence, here ˜that loud remark™ cannot
refer to a content avowed by an act of remarking: a content avowed by an
act of remarking cannot be appropriately characterised as loud. ˜That loud
remark™ here is most naturally taken to refer to an utterance the production
7 Miller (1971), 47; Nuchelmans (1973), 42“3; K. J. Williams (1978), 70“1; Lucas (1989), 62“3; B¨ck
a
(1992), 135.
8 Cf. Waterlow (1982), 135.
186 Truth and time
of which was the making of a remark: such an utterance can be appropriately
characterised as loud. Hence some utterance is true at one time and false
at another.9

Different truth-values at different times, no change. In Categories 5 Aristotle
says:
T 43 The same sentence seems to be both true and false. Suppose, for example,
that the sentence which says that somebody is sitting is true: after the person
has got up, this same sentence will be false. Similarly with belief. Suppose
you believe truly that somebody is sitting: after the person has got up, you
will believe falsely while having the same belief about that person. (4a 23“8)
However, the sentence and the belief itself remain completely unchanged in
every respect, and it is because the object changes that the contrary comes
to hold of it. For the sentence which says that somebody is sitting remains
the same, and it is because the object has changed that it comes to be true
at one time and false at another. Similarly with the belief. (4a 34“4b 2)
Thus Aristotle claims that
[29] A sentence (or a belief ) which is true at one time and false at another
does not change.10
Someone11 might object to [29] on the ground that a sentence (or a belief )
which is true at one time and false at another does change just because it
is true at one time and false at another. Aristotle does not consider this
objection “ how would he defend [29] if he were to address it?

Two ancient defences of [29]. The ancient commentators put forward at least
two defences of [29].
According to the ¬rst defence,12 the sentences which are true or false are
utterances. Consider a situation which one might be inclined to describe by
saying that a sentence undergoes a change because it is true at one time and
9 Cf. Waterlow (1982), 135. Evans (1979), 347“52 discusses the view that an utterance can be true at
one time and false at another. He regards it as ˜such a strange position that it is dif¬cult to believe
that anyone has ever held it™ (p. 348), and criticises it. His criticism is answered by Percival (1993/94),
198“203.
An event may enjoy a property outside the time during which it occurs: the earthquake was
remembered long after it occurred. Uneasiness at the idea of an utterance being true before it occurs
can be appeased by adding the restriction that the time when the utterance can be true should be
contemporary with, or later than, that during which the utterance occurs.
10 Cf. 4b 4“13. 11 Cf. Mignucci (1985a), 165.
12 Ammon. in Cat. 53, 22“4; Phlp. in Cat. 82, 19“23; Simp. in Cat. 118, 15“16; 118, 18“25; Olymp. in
Cat. 79, 14“33; Elias in Cat. 183, 34“184, 3; anon. in Cat. 18, 33“5. The ¬rst defence of [29] might
originate with Alexander (see Schmidt (1966), 284“5).
Truth and change 187
false at another. In the situation in question, it is not the case that numer-
ically one and the same sentence (utterance) is true at one time and false
at another. Rather, in the situation in question, one sentence (utterance)
occurs at one time and is true then, another sentence (utterance) occurs at
another time and is false then, and the two sentences (utterances), although
not numerically the same, are the same in kind. Hence, the situation would
not be described correctly by saying that a sentence undergoes a change
because it is true at one time and false at another.
The second defence of [29] is set forth by Dexippus (in Cat. 60, 13“23).
According to Dexippus, ˜a sentence [. . .] receives truth and falsehood not in
its own nature, but in the way relatives do (for it is said to be true or false by
virtue of its being concordant or discordant with objects)™ (60, 13“16). And,
as relatives harbour no transformation,13 so a sentence and a belief ˜become
false from true not by suffering some affection or being transformed, but
remaining unchanged™ (60, 18“20).14
Would Aristotle adopt either of these two ancient defences?

Assessment of the ¬rst ancient defence . . . The ¬rst ancient defence of [29]
assumes that the sentences which are true or false are utterances. And (as
I argued in chapter 1) Aristotle probably thinks that the sentences which
are true or false are utterances. Yet two arguments show that this defence
is unlikely to be one that Aristotle would adopt.
(i) The part of proposition [29] concerning beliefs is the claim that a
belief which is true at one time and false at another does not change. The
objection to this is that a belief which is true at one time and false at another
does change just because it is true at one time and false at another. The ¬rst
defence, when so modi¬ed as to answer this objection, assumes that beliefs
are either utterances or events of some other sort. But Aristotle tends to
use ˜belief ™ (˜d»xa™ or ˜Ëp»lhyiv™) for dispositions (or states) of believing,
which are neither utterances nor events of any other sort.15
(ii) Proposition [29] is appealed to by Aristotle while answering an
objection to a claim he makes about substance: the claim about substance

13 A view notoriously averred by Aristotle: see Ph. 3.1, 200b 26“201a 9; 5.2, 225b 11“13 (= Metaph. K 12,
1068a 11“13); 7.3, 246b 10“17; 246b 20“247a 7; 246b 25“6; 247b 1“7; 8.1, 251b 5“9; Metaph. N 1, 1088a 29“35
(cf. Pl. Tht. 154b6“155c5). Contrast Ph. 1.7, 190a 34“5.
14 It is not clear whether at lines 60, 21“3 Dexippus offers a new argument in defence of Aristotle.
Simplicius (in Cat. 119, 5“16) suggests essentially the same defence(s) as Dexippus. A defence of [29]
similar to Dexippus™ has been recently propounded by C. J. F. Williams (1991), 306“7 and Goldin
(2002), 242“3.
15 Int. 14, 23a 33“4; APo. 2.19, 100b 5“7; de An. 3.3, 428a 3“5; HA 7.4, 584b 6“14; EN 6.5, 1140b 13“16. Cf.
Olymp. in Cat. 79, 28“33; R´gis (1935), 70“1; Goldin (2002), 244.
e
188 Truth and time
(4a 10“21) is that only substances receive contraries while remaining numeri-
cally one and the same. The objection to this claim about substance (4a 21“8)
is that though sentences (and beliefs) are not substances, still numerically
one and the same sentence (or belief ) can be true at one time and false
at another, and thereby receive contraries. The answer to this objection
(4a 28“4b 18) comprises two arguments, both of which rely on [29]. Now,
Aristotle does not answer the objection by denying what sounds like its
central claim, i.e. the claim that numerically one and the same sentence
can be true at one time and false at another. Perhaps part of the reason why
Aristotle answers the objection in the way he does is that he agrees with
it on this point, i.e. he grants that numerically one and the same sentence
can be true at one time and false at another. But, if Aristotle adopted the
¬rst ancient defence, he could not coherently grant that numerically one
and the same sentence can be true at one time and false at another.

. . . and of the second. When he formulates the second defence of [29],
Dexippus does not use an expression like ˜The true and the false are relatives™
or ˜Truth and falsehood are relatives™. The expression he uses is ˜A sentence
receives truth and falsehood in the way relatives do™ (60, 15). This suggests
that the view he attributes to Aristotle is not that truth and falsehood are
relatives, but that truth and falsehood are closely linked with relatives in a
way that explains their peculiar behaviour with respect to change.
One sees why Dexippus might want to attribute this view to Aristotle.
For, according to Aristotle, all relatives satisfy a certain ˜linguistic™ de¬nition:
only ˜such items as are said to be just what they are of or than other items,
or in some other way in relation to something else™ (Cat. 7, 6a 36“7), are
relatives. Aristotle claims that the larger,16 the double, state, condition, per-
ception, knowledge, position, excellence, and defect are relatives and satisfy
the foregoing condition:17 e.g. ˜the larger is said just what it is than some-
thing else, for it is called “larger” than something™ (6a 38“9). Aristotle also
claims that grammatical-knowledge and musical-knowledge do not satisfy
the foregoing condition and therefore are not relatives (although they fall
under knowledge, which is a relative): ˜grammatical-knowledge is not called
“grammatical-knowledge” of something nor musical-knowledge “musical-
knowledge” of something™ (Cat. 8, 11a 27“8), and ˜grammatical-knowledge is

˜The larger™ (˜t¼ me±zon™) can function both as a singular abstract term (˜the property of being larger™)
16
and as a quanti¬ed general concrete term (˜whatever is larger™). Aristotle probably wants it to function
in both ways: the larger (i.e. the property of being larger) is a relative only if the larger (i.e. whatever
is larger) is called ˜larger than™ something. Cf. Metaph. 15, 1021b 6“8; Mignucci (1986), 102“3.
17 Cat. 7, 6a 38“6b 3; 6b 15“16; 10, 11b 24“31.
Truth and change 189
called “knowledge” of something (not “grammatical-knowledge” of some-
thing) and musical-knowledge “knowledge” of something (not “musical-
knowledge” of something)™ (11a 29“31).18 Now, truth and falsehood behave
like grammatical-knowledge and musical-knowledge (not like the double
or knowledge): what is true is not called ˜true™ of something, and what is
false is not called ˜false™ of something.19 So, Aristotle™s ˜linguistic™ de¬ni-
tion of relatives bars truth and falsehood from being relatives. Moreover, in
Sophistici Elenchi 22 Aristotle commits himself to the view that truth and
falsehood are qualities by saying that ˜a sentence or a belief being true or
false signi¬es not a “this” [t»de] but a “such” [toi»nde]™ (178b 27“8) “ note
that in Sophistici Elenchi 22 Aristotle uses ˜such™ (˜toi»nde™) interchange-
ably with ˜of such a quality™ (˜poi»n™), his standard expression for quali¬ed
items.20
Now: (i) Aristotle claims that a relative can be enjoyed at one time, and
lacked at another, by something that does not change; (ii) he claims that a
sentence (or a belief ) which is true at one time and not true at another does
not change; (iii) he imposes on relatives a condition which is not satis¬ed
by truth and falsehood. These facts suggest that if Aristotle were to defend
[29], he would take a line similar to that which Dexippus seems to attribute
to him: truth is not a relative but is suf¬ciently similar to relatives to share
their peculiar behaviour with respect to change.

2 tru th a n d re l ati ves
Why does truth not count as an Aristotelian relative? How is truth linked to
Aristotelian relatives? I think that these two questions cannot be properly
answered by simply pointing out that truth does not satisfy Aristotle™s ˜lin-
guistic™ de¬nition of relatives. I shall try to offer more appropriate answers
with the aid of an example.
Take the two-place relation R expressed by the formula ˜x is admired by
y™, the property21 P expressed by ˜x is admired™, and the property Q expressed
18 Cf. Top. 4.4, 124b 18“19; 6.8, 146a 36“146b 2; 146b 6“9; SE 13, 173b 1“3; 31, 181b 34“5; Metaph. 15,
1021b 4“6. Aristotle makes an incompatible claim at Top. 4.1, 120b 36“121a 9.
Aristotle uses ˜ˆlhqe…esqai kat†™ + genitive (see Int. 12, 21b 17; APr. 1.37, 49a 6“7; etc.), ˜ˆlhqe…esqai
19
–p©™ + genitive (see Int. 13, 22b 2), ˜ˆlhqe…esqai per©™ + genitive (see Top. 7.5, 154b 4), and ˜ˆlhqŸv
kat†™ + genitive (see Int. 11, 20b 37“8; 13, 22b 23“4; etc.) to say of a predicate (or a universal) that
it applies to (or holds of ) something. He also uses ˜yeudŸv kat†™ + genitive (see Int. 13, 22b 32“3)
to say of an expression that it is disjoint from another expression. But he never applies any of these
formulations to assertions, beliefs, or states of affairs.
20 Cf. 178b 37“9 and 179a 8“10 (also Ph. 3.1, 200b 28; Metaph. N 2, 1089a 18; Goldin (2002), 240“1).
The idea that truth and falsehood are qualities goes back to Plato (Sph. 262e8“9; 263a11“263b3;
Phl. 37b10“37c2).
21 Here ˜property™ means ˜one-place attribute™.
190 Truth and time
by ˜x is admired by Jim™. P is obtained from R by existential saturation: P
is the same as the property expressed by the formula ˜For some y, x is
admired by y™ (obtained by applying the existential quanti¬er to a variable
in the formula expressing R). Q is obtained from R by singular saturation
(because the formula expressing Q is obtained by substituting a singular
term for a variable in the formula expressing R).
First consider an attribution of P, e.g. the attribution made by saying
˜Fred is admired™. This might prompt the question ˜By whom is the admired
admired?™, which could be appropriately answered by saying ˜The admired
is admired by Jim™ or ˜The admired is admired by Bill™. The predicates of
these answers are derived from the formula (˜x is admired™) which expresses P
by dropping the variable and the copula (˜x™ and ˜is™) and adding appropriate
phrases (˜by Jim™, ˜by Bill™). One might describe the situation by saying that
the admired is said to be just what it is ˜by™ another item. So, P seems to
satisfy Aristotle™s ˜linguistic™ de¬nition of relatives.
Consider next an attribution of Q, e.g. the attribution made by saying
˜Fred is admired by Jim™. Someone wanting to know about Fred™s other
admirers might ask: ˜By whom is the admired by Jim admired?™ Now, the
predicates of the appropriate answers to this question cannot be derived
from the formula (˜x is admired by Jim™) which expresses Q by the pro-
cedure described above, i.e. by dropping the variable and the copula (˜x™
and ˜is™) and adding appropriate phrases (˜by Jim™, ˜by Bill™): this would
yield the ungrammatical ˜The admired by Jim is admired by Jim by Jim™
and ˜The admired by Jim is admired by Jim by Bill™. One might describe
the situation by saying that the admired by Jim cannot be said to be just
what it is ˜by™ another item, or in some other way in relation to some-
thing else. So, Q seems not to satisfy Aristotle™s ˜linguistic™ de¬nition of
relatives.
Back to the two questions which are our primary concern. First, why does
truth not count as an Aristotelian relative? Because Aristotelian relatives
are, like P, properties obtained from a two-place relation by existential
saturation,22 while truth is, like Q, a property obtained from a two-place
relation by singular saturation (for the property truth, expressed by ˜x is
concordant with the external world™,23 is obtained by singular saturation
from the two-place relation S expressed by ˜x is concordant with y™). Second,
how is truth linked to Aristotelian relatives? The property C expressed by
˜x is concordant™ is an Aristotelian relative, and truth bears to C the same
22 This is convincingly argued by Mignucci (1986), 101“5 (cf. Wedin (2000), 22).
23 For a discussion of how for Aristotle truth amounts to concordance or correspondence with the
external world see sect. 1 of ch. 4.
Truth and change 191
relationship that Q bears to P: as Q and P are obtained by (singular and
existential) saturation from the two-place relation R, so truth and C are
obtained by (singular and existential) saturation from S.
Aristotle lacks the notion of a two-place relation. Hence, he would not
think of Q and P as obtained by saturating the two-place relation R. Rather,
he seems to regard the relationship of properties like P to properties like Q
as that of a genus to what falls under it: P, which is an Aristotelian relative, is
a genus under which Q, which is not an Aristotelian relative, falls. The view
is reasonable: Q determines P ™s indeterminateness (which surfaces in the
existential quanti¬er in one of the formulae expressing P). Thus, Aristotle
might say that truth (the property expressed by ˜x is concordant with the
external world™) is not a relative, but falls under a genus (the property C
expressed by ˜x is concordant™ and ˜For some y, x is concordant with y™)
which is a relative.24
The reason why not only properties like P (which are Aristotelian rela-
tives) but also properties like Q (which are not Aristotelian relatives) can
undergo mere Cambridge change is that they are saturations of two-place
relations like R. P undergoes mere Cambridge change if whoever admired
Fred stops admiring him, while Q undergoes mere Cambridge change if
Jim stops admiring Fred (in either case Fred, who never admires himself,
does not change). Truth can similarly undergo mere Cambridge change:
a sentence (or belief ) s can be true at one time and not true at another
without changing because s can at one time be, and at another not be,
concordant with the external world by virtue of changes occurring not in s,
but only in the external world.25
In conclusion, Aristotle might want to say that truth and falsehood are
in the same situation as grammatical-knowledge and musical-knowledge:
as grammatical-knowledge and musical-knowledge (i) are qualities,26 (ii)
are not relatives,27 but (iii) are relatives according to their genera,28 so truth
and falsehood (i) are qualities,29 (ii) are not relatives,30 but (iii) are relatives
according to their genera.31

How genuine a property is truth? In Metaphysics N 1 Aristotle says:

24 Cf. the paragraph to which n. 16 above is appended.
25 The often repeated claim that events cannot undergo genuine changes (cf. Simons (1987), 134“7) ¬ts
well with the idea that utterances, which are events, can undergo mere Cambridge change.
26 Cat. 8, 11a 35“6. 27 Cat. 8, 11a 36. 28 Cat. 8, 11a 28“9; Metaph. 15, 1021b 3“6.
29 Cf. the portion of the main text to which n. 20 above is appended.
30 Cf. the paragraph to which n. 16 above is appended.
31 For a different account of the link between truth, relatives, and qualities see Miller (1971), 41“3.
192 Truth and time
T 44 A sign that a relative is least of all32 a substance and a being is that only of the
relative there are neither generation nor destruction nor change. (1088a 29“
31)33
As I argued in preceding subsections, for Aristotle truth is closely linked to
relatives: he would concede that truth is a relative according to its genus,
and he says that there is no change with regard to truth and falsehood.
Thus, Aristotle would probably apply to truth his general remark about
relatives: he would probably grant that truth ˜is least of all a substance and
a being™, i.e. is not a genuine property.34 Aristotle™s ground for this view is
probably the intuition that an item cannot possess at one time a genuine
property it lacks at another without changing.35
Categories 5 indirectly con¬rms that for Aristotle truth is not a gen-
uine property. One of the theses about substance maintained by Aristotle
in Categories 5 is that only substances receive contraries while remaining
numerically one and the same (4a 10“21). Aristotle considers an objection
to this thesis about substance: though sentences (and beliefs) are not sub-
stances, still numerically one and the same sentence (or belief ) can be true
at one time and false at another, and thereby receive contraries (4a 21“8).
Aristotle answers this objection by means of two arguments, both of which
rely on [29], i.e. the claim that a sentence (or a belief ) which is true at
one time and false at another does not change. The ¬rst argument (4a 28“
4b 5) concedes that numerically one and the same sentence (or belief ) can
receive contraries, i.e. truth and falsehood, but insists that the way in which
numerically one and the same substance can receive contraries is different
from that in which numerically one and the same sentence (or belief ) can:
numerically one and the same substance can receive contraries by chang-
ing, but it is not the case that numerically one and the same sentence (or
belief ) can receive contraries by changing ([29] supports this contention).
The second argument in Aristotle™s answer to the objection (4b 5“18) denies
what the ¬rst argument had conceded, i.e. it denies that numerically one
and the same sentence (or belief ) can receive contraries:
T 45 However, this [sc. that beliefs and sentences are able to receive contraries] is
not true. For it is not because they themselves receive anything that sentences
and beliefs are said to be able to receive contraries, but because the affection

32 Morrison (1987), 389“93, 394 convincingly argues that here (1088a 29) ˜least of all™ (˜¤kista•)has
its standard comparative meaning (”inthe least degree•).
33 Cf. 1088a 22“3; 4, 1070a 33“1070b 10 (where the category of relatives stands for all the categories
other than substance because it differs from it most); EN 1.4, 1096a 20“2; Wardy (1990), 215.
34 Cf. Brentano (1889), 26; Kraus (1930), 173“4.
35 Aristotle™s intuition is shared by several modern philosophers (e.g. Linke (1965), 312“13), but is
dif¬cult to motivate.
Truth and change 193
has come to be about something else. For it is because the object is or is
not that the sentence also is said to be true or false, not because it is able
itself to receive contraries. For sentences and beliefs are not changed at all by
anything, so that they are not able to receive contraries because no affection
comes to be36 in them. A substance, on the other hand, is said to be able to
receive contraries because it itself receives contraries. (4b 5“14)
The crucial inference of T 45™s argument goes from
[a] Sentences and beliefs are not changed at all by anything (4b 10“11)
to
[b] They [sc. sentences and beliefs] are not able to receive contraries
(4b 12).
According to its most plausible explanation, this inference goes from
[a ] A sentence (or a belief ) that is true at one time and false at another does
not thereby change
to
[b ] A sentence (or a belief ) which is true at one time and false at another
does not thereby receive contraries.
The inference from [a ] to [b ] can be explained as consisting of two subor-
dinate inferences that pass through an intermediate result, which remains
unexpressed:
[c] Truth and falsehood are not genuine properties.
The ¬rst subordinate inference (from [a ] to [c]) is warranted by those same
considerations which back Aristotle™s claim that ˜a relative is least of all a
substance and a being™. As for the second subordinate inference (from [c]
to [b ]), if truth and falsehood are not genuine properties, then a sentence
(or a belief ) which is true at one time and false at another does not thereby
receive contraries because there are no contraries for it to receive (contraries
are genuine properties).37
An apparent inconsistency in Aristotle™s ontology. Some commentators38 feel
that T 44 entails an inconsistency in Aristotle™s ontology. For T 44 seems
to commit Aristotle to the claim that beings are comparable with respect
to ˜being™: relatives ˜are™ less than quantities and qualities, which in turn
I read ˜gignom”nou p†qouv™ (4b 12“13) with most manuscripts and editors (Minio-Paluello follows
36
n and prints ˜gignom”nou™).
37 For a different interpretation of T 45™s argument see Goldin (2002), 243“5.
38 E.g. Shields (1999), 260“6.
194 Truth and time
˜are™ less than substances. Elsewhere39 Aristotle claims that objects are not
comparable with respect to a predicate that applies homonymously to them:
e.g. a ¬‚avour and a sound are not comparable with respect to ˜sharp™ because
˜sharp™ applies homonymously to them (only in different senses can a ¬‚avour
and a sound be called ˜sharp™). Since beings are comparable with respect
to ˜being™, ˜being™ does not apply homonymously to beings “ contrary to a
well-known principle of Aristotle™s ontology.
However, on re¬‚ection, this inconsistency evaporates. For in T 44 Aris-
totle says that ˜a relative is least of all a substance and a being™ (1088a 29“30):
he can be plausibly taken to mean that it is in the least degree that relatives
˜are™ in the substance-sense of ˜to be™. Thus, in T 44 the comparison with
respect to ˜being™ is carried out with respect to only one sense of ˜being™,
the substance-sense: relatives ˜are™ (in the substance-sense) less than quan-
tities and qualities, which in turn ˜are™ (again in the substance-sense) less
than substances. The objection that ˜being™ in the substance-sense does
not apply to qualities, quantities, or relatives carries no weight: Aristotle
acknowledges that a can be less F than b when neither a nor b are F.40

A ˜minimalist™ account of truth? Aristotle can be fairly described as committed
to the view that truth is not a genuine property. This view recalls modern
˜minimalist™ theories of truth, which also assert that truth is not a genuine
property. Still, there remains an important difference between the view to
which Aristotle seems to be committed and modern ˜minimalist™ theories
of truth. For modern ˜minimalist™ theories of truth are alternatives to the
theory of truth as correspondence, and are therefore opposed to it. By
contrast, in Aristotle™s case the motivation behind the view that truth is not a
genuine property is truth™s relational character, and this relational character
is the essential trait of Aristotle™s theory of truth as correspondence.

3 how far is tru th f rom ch a n g e?
A difference between general and singular assertions. Take a singular af¬rma-
tive assertion u, e.g. an utterance of ˜Socrates is seated™. At t1 Socrates is
seated. Since the universal seated is the object signi¬ed by the predicate of
u and Socrates is the object signi¬ed by the subject of u, at t1 the universal
signi¬ed by the predicate of u is combined with the individual signi¬ed by
the subject of u in such a way as to hold of it. So, u is true at t1 . Between
t1 and the later t2 u does not change, but Socrates changes: he stands up.
Top. 1.15, 107b 13“18; Ph. 7.4, 248b 6“249a 8; Metaph. I 4, 1055a 6“7; EN 8.2, 1155b 13“15.
39
40 Top. 3.6, 119b 17“30; Rh. 2.23, 1397b 12“29. There are however oscillations in Aristotle™s position on
the issue of a being less F than b when neither a nor b are F (cf. Casari (1984), 140“2, 144).
Truth and change 195
Hence at t2 the universal signi¬ed by the predicate of u is not combined
with the individual signi¬ed by the subject of u in such a way as to hold of
it. So, u is not true at t2 .41
Next take a particular af¬rmative assertion u , e.g. an utterance of ˜Some
man is seated™. At t1 Socrates is the only man seated. Since the universal
seated is the object signi¬ed by the predicate of u and the universal man
is the object signi¬ed by the subject of u , at t1 the universal signi¬ed by
the predicate of u is combined with the universal signi¬ed by the subject
of u in such a way as not universally to fail to hold of it. So u is true at
t1 . Between t1 and t2 u does not change, but Socrates changes: he stands
up. As a consequence of Socrates™ change, at t2 no man is seated. Hence at
t2 the universal signi¬ed by the predicate of u is not combined with the
universal signi¬ed by the subject of u in such a way as not universally to
fail to hold of it. So u is not true at t2 .42
Between t1 and t2 the universals man and seated do not change. For,
although seated at t1 is, and at t2 is not, combined with man in such a
way as not universally to fail to hold of it, man and seated do not change
between t1 and t2 : all the relevant change occurs in Socrates.43 Thus, in
the case of u (a singular assertion) the change responsible for the different
truth-values at different times occurs in the object (Socrates) signi¬ed by
the subject. But in the case of u (a general assertion) the change responsible
for the different truth-values at different times does not occur in the object
(man) signi¬ed by the subject (nor, for that matter, in the object signi¬ed
by the predicate).
Thus: some singular assertion has different truth-values at different times
without changing because all the relevant change occurs in the object signi-
¬ed by the subject; no general assertion has different truth-values at different
times without changing because all the relevant change occurs in the object
signi¬ed by the subject. General assertions are, in a way, at least two steps
removed from change. And a general assertion whose subject signi¬es a uni-
versal that ranges only over universals (e.g. an utterance of ˜Every biological
species is instantiated™) is at least three steps removed.

Do states of affairs change? For Aristotle, an af¬rmative predicative assertion
is true when and only when the corresponding state of affairs ˜is™ in the
sense of being true. Hence, if an af¬rmative predicative assertion is true at
t1 and not true at t2 , the corresponding state of affairs ˜is™ in the sense of

41 42 Cf. [17] on pp. 93“4 above.
Cf. [17] on pp. 93“4 above.
43 a 17“23; 5.1, 224b 4“6; 224b 10“16; Metaph. H 5, 1044b 21“9; [Plu.] Plac. 899b; Simp. in
Cf. Ph. 4.4, 211
Cat. 218, 19“21; G. E. L. Owen (1968a), 110“11.
196 Truth and time
being true at t1 and ˜is not™ in the sense of being false at t2 . Aristotle thinks
that the assertion can be true at t1 and not true at t2 without changing. Does
he think that the state of affairs can ˜be™ in the sense of being true at t1 and
˜not be™ in the sense of being false at t2 without changing? Passages relevant
to this problem are T 43 and T 45, which can be interpreted in at least two
ways.
According to interpretation (i), in T 43 and T 45 ˜object™ denotes those
(composite or non-composite) items which are crucial to the truth or false-
hood of beliefs and assertions concerning them, and ˜is or is not™ means
˜“is” in the sense of being true or “is not” in the sense of being false™. On
interpretation (i), T 43 and T 45 commit Aristotle to the claim that a state
of affairs, by ˜being™ in the sense of being true at one time and ˜not being™
in the sense of being false at another, undergoes a change.44
According to interpretation (ii), in T 43 and T 45 ˜object™ indicates an
item (an individual, in the case at hand) to which the assertion or belief
refers,45 and ˜is or is not™ functions as a schematic expression to be expanded
as ˜is so-and-so or is not so-and-so™.46 On interpretation (ii), T 43 and T 45
do not commit Aristotle to the claim that a state of affairs, by ˜being™ in the
sense of being true at one time and ˜not being™ in the sense of being false
at another, undergoes a change.
Against interpretation (i) counts the fact that if it is correct, then Aris-
totle cannot easily escape the awkward consequence that states of affairs
should be substances (because he is arguing47 that only substances receive
contraries by changing while remaining numerically one and the same). In
favour of interpretation (ii) counts the example with which Aristotle begins
the discussion in this section of Categories 5 (4a 23“8 < T 43): the change
responsible for the assertion or belief that somebody is seated being true
at one time and false at another is the getting up, and what gets up is not
a state of affairs but the individual to which the assertion or belief refers.
Thus, interpretation (ii) has a slight edge over (i).48 This goes in favour of

44 Cf. 4a 35“6; 4a 37“4b 1; Mignucci (1985a), 165“6, 168; de Rijk (1987), 33“4; Gaskin (1998), 45; Modrak
(2001), 36.
45 Cf. Int. 3, 16b 21“5 (where, contrary to de Rijk (1987), 35, 55 and Gaskin (1998), 45, both Weidemann
(1994/2002), 178“87 and Whitaker (1996), 55“8 take ˜prŽgma™ at 16b 23 to denote the item signi¬ed
by a verb); 7, 17a 38“9; 12, 21b 27“8 (with Nuchelmans (1973), 36, Hadot (1980), 313, and Gaskin
(1998), 45); APr. 1.27, 43b 3“5; 43b 11“12; APo. 2.5, 91b 14“15; C. J. F. Williams (1991), 307.
46 Cf. n. 19 of ch. 4.
47 4b 2“4 (cf. 4a 35“6; 4a 37“4b 1; 4b 8“9).
48 Cf. Nuchelmans (1973), 34“5; Graeser (1978), 449“50; (1981), 87. Other Aristotelian passages (Top.
8.2, 157b 13; 157b 15“16; de An. 3.3, 428b 6; 428b 8“9) also speak of different truth-values at different
times in connection with an object (prŽgma) that changes (metap©ptein) or remains unchanged
(sÛzesqai or diam”nein), but leave it unclear whether the objects that change or remain unchanged
Truth and change 197
attributing to Aristotle the view that a state of affairs can ˜be™ in the sense
of being true at one time and ˜not be™ in the sense of being false at another
without changing. In fact, the considerations of the preceding subsections
suggest that this is the view Aristotle ought to adopt: for whatever arguments
show that an assertion can be true at one time and false at another without
changing should be transferable to states of affairs and thus establish that
a state of affairs can ˜be™ in the sense of being true at one time and ˜not be™
in the sense of being false at another without changing.

are states of affairs. Aristotle™s use of ˜metap©ptein™ and ˜prŽgma™ in connection with different
truth-values at different times is striking in the light of the technical terminology in Hellenistic
philosophy (see e.g. D.L. 7. 65; 76; Simp. in Ph. 1299, 36“1300, 11).
ch a p t e r 7

Truth and Determinism in de Interpretatione 9




In chapter 9 of de Interpretatione (henceforth ˜Int. 9™) Aristotle claims that
some future-tense singular assertions1 are sometimes neither true nor false,
and therefore refute Bivalence, the principle stating that every assertion is
always either true or false. Aristotle™s argument goes as follows:
If every future-tense singular assertion is always either true or false, then whatever
happens was always antecedently bound to happen, i.e. Determinism holds; but it
is not the case that whatever happens was always antecedently bound to happen;
hence, not every future-tense singular assertion is always either true or false, i.e.
some future-tense singular assertions are sometimes neither true nor false.
Int. 9 divides into four parts: the ¬rst (18a 28“33) is introductory (it brie¬‚y
states Aristotle™s position); the second (18a 34“18b 25) contains two arguments
from Bivalence to Determinism; the third (18b 26“19a 22) argues that Deter-
minism is absurd; the fourth (19a 23“19b 4) contains Aristotle™s solution to
the problems raised by Bivalence and Determinism.
In section 1 of the present chapter I present the most important modal
attributes and theses at play in Int. 9: the attribute of necessity as ineluctabil-
ity (a diachronic modality, with two ˜slots for dates™), the thesis of the neces-
sity of the present and the past (the thesis that for whatever obtains at any
time it is both then and later necessary to obtain then), and Determinism
(the thesis that for whatever obtains at any time it is always antecedently
necessary to obtain then). Aristotle endorses the necessity of the present
and the past, but rejects Determinism. In section 2 I offer a close textual
analysis of Int. 9 (I translate and comment on the whole of Int. 9). The
most important part of this section is that dealing with Aristotle™s own
solution for the problems raised by Bivalence and Determinism: I defend
an interpretation of Aristotle™s words that gives him a position that ¬ts well
with his view that assertions can, in principle, have different truth-values at
1 Aristotle concentrates on predicative assertions, but his argument perhaps covers also existential
assertions.

198
Truth and Determinism in de Interpretatione 9 199
different times (the usual interpretation instead lands him with an unstable
position). In section 3 I brie¬‚y discuss alternative interpretations of Int. 9.

1 t he mo dal at tribu tes a nd th e s e s i nvolved i n i n t . 9
Necessity as ineluctability. Int. 9™s discussion of Determinism employs the
attribute of necessity as ineluctability (or inevitability): what is necessary is
what is ineluctable (or inevitable), i.e. what nothing can be done about.2
The attribute of necessity as ineluctability is a diachronic modality, i.e. a
modal attribute with two ˜slots for dates™. The key formulations are instances
of the schema ˜At t it is necessary that a at t ™, where ˜t™ and ˜t ™ are to
be replaced with designations of times, ˜a™ with an assertoric sentence-type
without dates: e.g. ˜At 13.00 of 1 January 1997 it was necessary that the train
should reach the station at 13.20 of 1 January 1997™.3 Such formulations
can be deployed to express views involving the attribute of necessity as
ineluctability (e.g. ˜If the train reached the station at 13.20 of 1 January
1997, then at 13.20 of 1 January 1997 it was necessary that the train should
reach the station at 13.20 of 1 January 1997™).4

The necessity of the present and the past. In several passages Aristotle endorses
the thesis of the necessity of the present and the past:
T 46 Nothing that has come to be is an object of choice, e.g. no one chooses to
have sacked Troy: for no one deliberates about what has come to be,5 but

2 Other passages also seem to employ the attribute of necessity as ineluctability: see Cael. 1.12, 283b 8“10;
283b 13“14; Metaph. E 3, 1027a 32“1027b 14 (with Gaskin (1995), 97“8); Rh. 1.3, 1358b 3“5; 1358b 14“15. On
necessity as ineluctability see Peter de Rivo (apud Baudry (1950), 75); Lukasiewicz (1930), 155“6; Prior
(1953), 322“4; Butler (1955), 268“9; Prior (1955/62), 241, 247“50; Anscombe (1956/68), 23; Lemmon
(1956); Grant (1957), 522“7; R. Taylor (1957), 5“15, 22“3; Saunders (1958), 369; Anscombe/Geach
(1961), 7; von Fritz (1962), 140; Ackrill (1963), 139; Burrell (1964), 39“41; Hintikka (1964/73a), 164;
(1964/73b), 183; Cahn (1967), 25“6, 31; Sainati (1968), 280; Seeskin (1971), 765; McKim (1971/72),
88“91, 107; D. Frede (1972), 160“7; Dickason (1976), 21; Jeffrey (1979), 251“2; M. J. White (1979),
90“1; Sorabji (1980), 105; Spellman (1980), 115“16; Weidemann (1980), 408“10; C. J. F. Williams
(1980), 129; M. J. White (1981), 238; Seel (1982a), 244“5; (1982b), 300“1; Waterlow (1982), 11“12;
Thomason (1984), 138“9; von Wright (1984c), 76; D. Frede (1985), 72; M. J. White (1985), 17“18;
Kirwan (1986), 169“77; von Kutschera (1986), 208“9; Craig (1988), 12“13; van Eck (1988), 20, 24“5;
Donini (1989), 8“9; D. Frede (1990), 203; Zagzebski (1991), 15“17; Gaskin (1994), 88“90; Weidemann
(1994/2002), 255; Gaskin (1995), 93“4; Mignucci (1996b), 285“7; D. Frede (1998), 96; Enders (1999),
388“91; Bonomi/Zucchi (2001), 10.
3 Grammar often requires adjusting the otherwise ungrammatical result of a pure and simple substi-
tution (e.g. ˜At 13.00 of 1 January 1997 it is necessary that the train reaches the station at 13.20 of
1 January 1997™ is ungrammatical).
4 On diachronic modalities in Aristotle cf. Weidemann (1980), 408“10; Seel (1982b), 297“9; Knuuttila
(1993), 31“3; Weidemann (1994/2002), 253“5.
5 Cf. EN 3.5, 1112b 33“1113a 2.
200 Truth and time
about what will be and is possible, while what has come to be cannot not
have come to be. (EN 6.2, 1139b 5“9)
T 47 Forensic oratory deals with what is and what is not, of which there is more
of a demonstration and necessity: for what has come to be is necessary. (Rh.
3.17, 1418a 4“5)
T 46 states the necessity of the past and the contingency of the future;
T 47 presupposes the necessity of the present as well as the past.6
The diachronic modality of necessity as ineluctability provides a helpful
framework for discussing the necessity of the present and the past. For the
thesis of the necessity of the present and the past can be identi¬ed with
the thesis that for every state of affairs s and every time t, if s obtains (or,
respectively, does not obtain) at t then at every time t not earlier than t it
is necessary that s should obtain (or, respectively, not obtain) at t.

Determinism. The diachronic modality of necessity as ineluctability pro-
vides a helpful framework for discussing also Determinism. For Determin-
ism can be identi¬ed with the thesis that for every state of affairs s and
every time t, if s obtains (or, respectively, does not obtain) at t then at every
time t earlier than t it is necessary that s should obtain (or, respectively, not
obtain) at t.

2 clos e textua l a na lys i s of i n t . 9
The introduction (18 a 28“33):
T 48 With regard to things that are and things that have come to be, it is then
necessary that either the af¬rmation or the denial should be true or false, and
with regard to universals spoken of universally it is necessary that always one
should be true and the other false, and with individuals too, as we have said,7
while with regard to universals not spoken of universally it is not necessary
(we discussed these, too).8 But with regard to individuals that are going to
be it is not likewise. (18a 28“33)
T 48 draws two contrasts. First, it contrasts past- and present-tense asser-
tions with future-tense singular assertions. Past- and present-tense asser-
tions are said to have, and future-tense singular assertions to lack, a certain
property concerning truth and falsehood. What property is it? There are

6 At Cael. 1.12, 283b 13“14 Aristotle apparently endorses the necessity of the past but denies that of the
present (cf. Sharples (1979), 41; Kirwan (1986), 177“8).
7 Cf. 7, 17b 26“9. 8 Cf. 7, 17b 29“37.
Truth and Determinism in de Interpretatione 9 201
two candidates: having always a different truth-value from the contradic-
tory, and being always either true or false. But the ¬rst property is ruled out
because some past- and present-tense assertions lack it: some indeterminate
past- or present-tense assertion sometimes has the same truth-value as its
contradictory (e.g. ˜A man is white™ and ˜A man is not white™ are both true
now).9 Hence T 48™s ¬rst contrast is probably the following:
[30] Every past- or present-tense assertion is always either true or false. Not
every future-tense singular assertion is always either true or false.
This contrast is not exhaustive: general future-tense assertions fall in neither
of its groups. What group should they be attached to? In other words, are
they supposed to be always either true or false?10
T 48™s second contrast is a sub-contrast within the ¬rst of the ¬rst con-
trast™s two groups. This time the property which makes the difference seems
to be that of having always a different truth-value from the contradictory.
Thus:
[31] Within the group of past- and present-tense assertions, each of which is
always either true or false, some (i.e. singular and quanti¬ed assertions)
always have a different truth-value from their contradictories (at every
time one of the two contradictories is true, the other false), some of the
others (i.e. indeterminate assertions) sometimes have the same truth-
value as their contradictories (speci¬cally, they are both true).

The ¬rst deterministic argument (18 a 34“18 b 9). The part of Int. 9 (18a 34“
18b 25) showing that Bivalence entails Determinism contains two determin-
istic arguments (18a 34“18b 9, 18b 9“16) and a pre-emptive move (18b 16“25)
blocking an escape route.
Here is the ¬rst deterministic argument:
T 49 For if every af¬rmation or denial is true or false, it is also necessary for
everything either to hold or not to hold. Hence,11 if someone will af¬rm
that something will be and another will deny precisely this, it is clearly
necessary for one of them to be speaking truly, if every af¬rmation and

9 Cf. Ackrill (1963), 133“4; Weidemann (1979), 30; Craig (1988), 9“10; Weidemann (1994/2002), 226“7.
10 Most commentators think they are: see Ammon. in Int. 128, 30“129, 35; 138, 28“34; 148, 6“11 (with
Mignucci (1998), 54); von Kirchmann (1876), 73“4; DuLac (1949), 166; Wieland (1979), 29; Weide-
mann (1994/2002), 227. For a dissenting voice see Gaskin (1996), 50.
At 18a 35 I read ˜Ëp†rcein, ãste e«™ with the main manuscripts, Bekker, Waitz, D¨ bner, Cooke,
11 u
and Weidemann. Other witnesses have ˜Ëp†rceiná e« g‡r™, the reading adopted by Pacius, Buhle,
Weise, Minio-Paluello, Zadro, and D. Frede (1970), 85“6. Other attested readings are ˜Ëp†rceiná e«

<<

. 6
( 11)



>>