Barnes Jonathan. Truth, etc

Подождите немного. Документ загружается.

90 Truth

ﬁrst sight strange. But it must be remembered that the assertoric sentences

in question are of a special sort—in particular, they do not express items

which are, necessarily, either true or false. The remarks about disjunction

and negation which were adduced to show that our view is incoherent are

true enough; but they apply only to assertoric sentences which express items

which are true or false; they apply to assertibles which bear truth-values, and

to nothing else. The patriarchal papa says to his dithering daughter:

Make up your mind—either accept him or don’t.

It would be absurd to take the remarks about disjunction and negation to

indicate that ‘Either accept him or don’t’ must be true. It is equally absurd,

though less obviously absurd, to say the same about ‘Either she’ll accept him

or she won’t.’

According to the second of the two Epicurean positions, assertibles of the

form ‘Either so-and-so or not so-and-so’ are always true. In that case, the

Epicureans must accept the ﬁrst premiss of the Stoic argument, namely

Whenever it is assertible that so-and-so, either it then holds that so-and-so

or it does not then hold that so-and-so.

And in that case they must reject either the second premiss or the third

premiss or both. What would or should they have rejected? And could they

coherently have done so?

Cicero thinks it positively shameful to declare that a disjunction may be

true even if neither of its disjuncts is true—after all, that goes against the

very deﬁnition of a disjunction. So at the very least, the Epicureans must

explain how they can understand disjunctions other than in accordance with

the orthodox deﬁnition. No Epicurean explanation survives, and perhaps

none ever existed. But it is not difﬁcult to conjure something up—for

example, this:

An inclusive disjunction is true if and only if not all of its members are

false.

Any proposition of the form ‘Either so-and-so or not so-and-so’ will then be

true, whether the so-and-so has a truth-value or not.

How, in the second position, would the Epicureans have defended them-

selves against the Stoic argument? Here too we have no texts. Here too it is

not difﬁcult to propose something. The two premisses in question are the

consequences of the two principles of deﬂation:

Whenever it holds that so-and-so, it is then true that so-and-so,

and

ARetort 91

Whenever it does not hold that so-and-so, it is then false that so-and-so.

The Epicureans in the second position must reject at least one of those

propositions. I have already suggested that they did not reject, and scarcely

could have rejected, the deﬂationary principle for truth; and so they must

have accepted the former of those two premisses. And I think it’s a racing

cert that the Epicureans rejected—or would have rejected—the latter.

If it is now true that there will be a naval engagement next week, then

there must be something now in place which now ensures that there will be

such an engagement—there must be a present cause of that future event.

Similarly, if it is now false that there will be a naval engagement next week,

then there must be something now in place which now prevents any such

future engagement—there must be a present inhibiting cause. And surely if

it is now the case that there will be an engagement, then it is now true that

there will be an engagement, so that there is now a present cause of the future

engagement.

Suppose that it is not now the case that there will be an engagement

next week. In that case, there is not now a cause which ensures such an

engagement. Does it then follow that there is a present cause which inhibits

such an engagement? Not at all; for as far as future naval engagements are

concerned there may be as yet no causes present, one way or the other. Thus

Whenever it does not hold that so-and-so, it is then false that so-and-so

is not true. It is not true—to put the matter generally and roughly—because

there are not always causes in place for everything.

Even if all that is true, it may be murmured, the Epicureans have not

gained the battle; for there is another way of taking the questionable premiss

of the argument: instead of

Whenever it does not hold that so-and-so, it is then false that so-and-so

read

Whenever it holds that not so-and-so, it is then false that so-and-so.

The replacement premiss, I suppose, is true—or at any rate, it would have

been accepted by my hypothetical Epicureans.

But accepting it is not fatal. For if the third premiss is thus revamped,

then either the argument becomes invalid or else the ﬁrst premiss must be

similarly revamped. To secure validity, the ﬁrst premiss must become:

Whenever it is assertible that so-and-so, either it then holds that so-and-so

or it then holds that not so-and-so.

And the Epicureans have no reason to accept that proposition.

92 Truth

Many modern philosophers will be unimpressed by the Epicurean argu-

ment, or by the argument which I have just offered to the Gardeners. For,

like Carneades, they do not think much of the suggestion that truth and

falsity must be underwritten by causes: it is now true that I shall be in Paris

tomorrow (if it is now true that I shall be in Paris tomorrow) not because

there is now some determinate and determining cause of my being there

tomorrow but simply because I shall be there tomorrow. But Chrysippus

could not have taken that dismissive line. Rather, he must have appealed to

his fatalism: it is thanks to destiny that every assertible is at every moment

of its existence either true or false; for it is the permanent presence of fatal

causes which underwrites the truth or the falsity of any assertible at any time.

Epicurus and Chrysippus disagree on a fundamental principle of logic. The

disagreement is undergirded by a disagreement on a fundamental principle

of physics.

2. Predicates and Subjects

PREDICATES IN ANCIENT GRAMMAR

When Frege decided to ditch subjects and predicates and to make a new

life with arguments and functions, he claimed that subjects and predicates

were among those several unhealthy notions which grammar had foisted

upon logic. He did not afﬁrm that subjects and predicates were items like

phlogiston and the luminiferous ether; but he was sure that they were, at best,

items for the grammarian’s eyes only. That is perhaps the only view which

Frege shared with Nietzsche.

When—a few decades later—de Saussure ﬁrst delivered his Genevan

course on general linguistics, he began by urging his pupils to forget the

distinction between subjects and predicates. The distinction is alien to

linguistics: it is one of the several impertinent notions which logic has foisted

upon grammar. No doubt subjects and predicates have some logical utility;

but they are of no use or interest to the grammarian.

Every schoolboy once knew that in a sentence such as

Every valley shall be exalted

exaltation, or perhaps future exaltation, is predicated, afﬁrmatively and

universally, of a certain subject—namely, of valleys. That traditional parsing

was, according to Frege, imposed on logic by grammar; and according to de

Saussure, it was foisted on grammar by logic. If we take a long view of the

matter, Frege was wrong and de Saussure right.

For subjects and predicates are ancient animals; and they are animals

which were reared by the ancient logicians, not by the ancient grammarians.

They were not reared in secret; nor were they particularly delicate creatures,

or particularly exotic. Their language was readily mastered. Latin logicians

used the verbs ‘subiacere’and‘praedicere’for‘tobesubject’and‘topredicate’

(although ‘declarare’ was at ﬁrst preferred for ‘to predicate’), and their Greek

masters used ‘ὑποκεῖσθαι’and‘κατηγορεῖν’. The pairs of verbs, in Latin

and in Greek, had produced little families of adjectives and nouns; and they

had various compound relations. The ancient grammarians might easily have

adopted subjects and predicates. In fact they hardly ever mentioned them.

94 Predicates and Subjects

The Latin Institutions of Priscian gave the mediaeval West its grammar. Not

once in the eighteen books of that work does Priscian discuss predication;

and the words ‘subjectum’and‘praedicatum’ and the like are as rare in

his Institutions as they are in the writings of his Latin predecessors. In

Greek, the fundamental grammatical text was, or came to be, the Art of

Grammar falsely ascribed to Dionysius Thrax, a pupil of the great Aristarch.

In the Art predication is not mentioned, and the words ‘κατηγορεῖν’and

‘ὑποκεῖσθαι’ and their relatives never appear. The ancient commentaries on

the Art —which, though late and repetitive and confused, together constitute

a large portion of our evidence for ancient grammatical theory—show no

interest in subjects and predicates.

Once or twice, it is true, the commentaries speak in passing of predication.

They do so, for example, in their deﬁnition of ‘genus’, which they lifted from

Porphyry’s Introduction (scholiast to Dionysius Thrax, 117.1–5). Again, we

learn that

the inﬁnitive is also called predicative because it is predicated of the objects which

are named by it.

(scholiast to Dionysius Thrax, 400.24–26)¹

Or again,

some people call adjectives predicatives because they are always predicated of proper

names or of appellatives. For just as adverbs are necessarily attached to verbs, so

adjectives are attached to names.

(scholiast to Dionysius Thrax, 233.24–25)²

But such things have little but the name in common with what the logicians

thought of as predication: as the second of the two citations shows, ‘be

predicated of’ means no more than ‘be attached to’.

With Apollonius Dyscolus the case appears to be different—and Apol-

lonius was the best grammarian of antiquity. For the words ‘subject’ and

‘predicate’, together with their congeners, are found ﬁfty odd times in his

surviving works. Yet if the verb ‘κατηγορεῖν’isfoundinhiswritings,it

is rarely used to connote predication; and although ‘ὑποκείμενον’isnot

uncommon, it never pairs with ‘κατηγορούμενον’ and never means ‘subject’.

¹ ἡ ἀπαρέμφατος καλεῖται καὶ κατηγορική· κατηγορεῖται γὰρ τῶν πραγμάτων ἃ δι᾿ αὐτῆς

ὀνομάζεται.

² τὸ ἐπίθετον τοῦτο κατηγορικὸν ὑπ᾿ ἐνίων καλεῖται διὰ τὸ πάντῃ κατηγορεῖν κυρίων ἢ

προσηγορικῶν· ὡς γὰρ τὰ ἐπιρρήματα τοῖς ῥήμασι πάντως συναρτᾶται, οὕτω καὶ τὰ ἐπίθετα

τοῖς ὀνόμασιν.

Predicates in Ancient Grammar 95

To establish those negative points would require a tedious survey. But it may

be instructive, or at least diverting, to look at a few passages.

I start with ‘ὑποκείμενον’, for which the right translation is usually not

‘subject’ but simply ‘thing’ or ‘item’. Sometimes, the word means ‘thing <as

opposed to expression>’: for example, synt iii 10 [275.6–9]. Often, it means

‘item <in question, on the table, before us>’: there is a very good case at pron

33.5. Most often it means ‘item <indicated by the name in question>’: there

is a clear example at conj 216.1—where ‘ὑποκειμένη’ picks up ‘δηλούμενον

[indicated]’ a few lines earlier. In a similar vein, the object of a transitive

verb is called its ὑποκείμενον (e.g. synt iii 149 [396.4–8]; 160 [407.2]; 177

[422.15]).

Again, in the Homeric line ‘Zeus gave it to Hector to wear on his head’,

Apollonius says that ‘three ὑποκείμενα are thought of’—Zeus, Hector, and

Hector’s head (synt ii 111 [211.17–212.3]): it is not that the sentence has

three distinct subjects but that it refers to three distinct items. Or again:

Often in the case of names which are of a single ὑποκείμενον there is a double

inﬂection: Νέα πόλις, Νέας πόλεως.

(pron 60.14–15)³

Apollonius is speaking of compound names which refer to a single object.

Finally, a longer passage:

By way of a nominal construction we seek the substance of the ὑποκείμενον (for

pronouns only manifest the substance—although their deictic force comments on

the attendant features—and that is why they extend to every ὑποκείμενον), whereas

by a pronominal construction we grasp the substance but not its proper quality which

goes along with the imposition of the name.

(synt i 119 [101.12–102.3])⁴

That is to say, a name shows what sort of thing is the item to which it refers: a

pronoun, although it indicates such attendant features as number and gender,

serves merely to identify the item—and so may be used to identify any item.

Where the traditional schoolboy will say that, in the line from the Messiah,

exaltation is the predicate and its subject is valleys, Apollonius will note

³ πολλάκις καὶ ἐπ᾿ ὀνομάτων καθ ᾿ ἑνὸς ὑποκειμένου δύο κλίσεις γίνονται· Νέα πόλις, Νέας

πόλεως ...

⁴ διὰ μὲν τῆς ὀνοματικῆς συντάξεως τὴν οὐσίαν ἐπιζητοῦμεν τοῦ ὑποκειμένου (ταύτην γὰρ

μόνον αἱ ἀντωνυμίαι ἐμφαίνουσι, τῆς ὑπ᾿ αὐτῶν δείξεως συνεξηγουμένης τὰ παρεπόμενα,

ἔνθεν ἐπὶ πᾶν ὑποκείμενον συντείνουσιν), διὰ μέντοι τῆς ἀντωνυμικῆς συντάξεως τῆς

μὲν οὐσίας ἐπιλαμβανόμεθα, τῆς δὲ ἐπιτρεχούσης ἰδιότητος κατὰ τὴν τοῦ ὀνόματος θέσιν

οὐκέτι.—It is hard to believe that the text is sound; but its difﬁculties do not matter here.

96 Predicates and Subjects

that the ὑποκείμενον of the word ‘valley’ is a valley—and that is always its

ὑποκείμενον, wherever it may appear in a sentence and whether a traditional

logician would count it as a subject or as a predicate.

In a few Apollonian texts it is initially tempting to think that the word

‘ὑποκείμενον’ does mean ‘subject’ in the logician’s sense of the word. For

example,

Seeking the existence of some ὑποκείμενον, we say ‘Who is moving? Who is walking?

Who is talking?’. The movement, the walk, the talk, are clear—the active person

remains unclear.

(synt i 31 [29.1–4])⁵

Surely he means that we’ve got the predicate and are now looking for a

subject? I doubt it: there is no mention of predicates in the passage; and there

is no reason to think that ‘ὑποκείμενον’ means anything more than ‘object’

or ‘item’. Apollonius doesn’t mean that we have found a predicate—say, ‘is

walking’—and are now looking for a subject to make an honest sentence out

of it. He means that we grasp that someone’s walking and wonder who.

Again, there are half-a-dozen occurrences of ‘ὑποκείμενον’—but none

of ‘κατηγορούμενον’—in a short argument in Pronouns where Apolloni-

us contends, against unnamed adversaries, that the words ‘τηλικοῦτος’

and ‘τοιοῦτος’ (‘as big as that’, ‘like that’) are not compound pro-

nouns but simple names. The adversaries claim, inter alia, that those two

expressions,

being names which indicate a similarity, are agreed to establish a reference to their

ὑποκείμενα.

Apollonius has a brisk reply:

That is perfectly silly; for what is shown by a pronoun does not apply to something

else—it is itself the ὑποκείμενον both in gender and in number. Looking at a lake,

you will say ‘The Nile is as big as that’.

(pron 30.23–31.3)⁶

⁵ ὕπαρξίν τινος ὑποκειμένου ἐπιζητοῦντές φαμεν τίς κινεῖται; τίς περιπατεῖ; τίς λαλεῖ;

προδήλου μὲν οὔσης τῆς κινήσεως, τῆς περιπατήσεως, τῆς λαλιᾶς, τοῦ δὲ ἐνεργοῦντος

προσώπου ἀδήλου καθεστῶτος.

⁶ ἄλλως τε καὶ ὁμοιώσεως ὄντα ὀνόματα τῶν ὑποκειμένων ἐδόκει δεῖξιν παριστάνειν. ὅπερ

ἄγαν ἐστὶ ληρῶδες. τὸ γὰρ δεικνύμενον δι᾿ ἀντωνυμίας οὐκ ἐπ᾿ ἄλλου συντείνει, αὐτὸ δὲ τὸ

ὑποκείμενον καὶ κατὰ γένος καὶ κατὰ ἀριθμόν. ἀφορῶν γάρ τις εἰς λίμνην φήσει τηλικοῦτον

εἶναι τὸν Νεῖλον.

Predicates in Ancient Grammar 97

The argument is obscure and has excited some controversy. In addition,

like so many arguments in Apollonius, it appears to conﬂate an object of

reference, the lake, with a term which refers to it.

What Apollonius means is, I think, this: If, looking at a lake, you say ‘The

Nile is as big as that [τηλικοῦτος]’, the expression ‘as big as that’ agrees in

gender and in number with the word ‘Nile’. But what the expression ‘as big as

that’ refers to or demonstrates is the lake and not the Nile. Were ‘as big as that’

a pronoun or a pronominal expression, it would agree with what it demon-

strates (in the example, it would be feminine, to agree with ‘lake’). Evidently,

the word ‘ὑποκείμενον’ does not here mean ‘subject <of the sentence>’. For,

according to Apollonius’s usage, in ‘The Nile is as big as that’ the ὑποκεί-

μενον—the ὑποκείμενον of ‘as big as that’—is the lake. It is not the Nile.

So much for ‘ὑποκείμενον’. Nowhere does the term indicate, let alone

mean, ‘item <which is subject rather than predicate>’. The dossier on

‘κατηγορεῖν’ supports a parallel conclusion. Thus:

An adverb is an expression which does not inﬂect and which predicates of the

inﬂections of verbs, either universally or particularly, without which it will not close

up a thought.

(adv 119.5–6)⁷

And the opening pages of Apollonius’ essay purport to show that ‘adverbs

predicate of verbs’ (122.33–34). An ancient commentator remarks of the

passage that

‘predicates’ means ‘is placed’, so that it runs ‘is placed on the inﬂections of

verbs’—since the philosophers call adverbs too predicates.

(scholiast to Dionysius Thrax, 95.18–20)⁸

We need not crack our heads over the last clause—it is enough to note that

‘predicate’ means ‘modify’ or ‘be attached to’.

Elsewhere, and similarly, Apollonius will say that

the expressions ‘τῶν’, ‘τοῖν’ and the like are not predicated of a single gender

(synt i 40 [36.8–9])⁹

⁷ ἔστιν οὖν ἐπίρρημα μὲν λέξις ἄκλιτος, κατηγοροῦσα τῶν ἐν τοῖς ῥήμασιν ἐγκλίσεων

καθόλου ἢ μερικῶς, ὧν ἄνευ οὐ κατακλείσει διάνοιαν.

⁸ κατηγοροῦσα ἀντὶ τοῦ τιθεμένη, ἵν᾿ ᾖ τὸ ἑξῆς οὕτως, τιθεμένη κατὰ τῶν ἐν τοῖς ῥήμασιν

ἐγκλίσεων, ἐπειδὴ καὶ τὸ ἐπίρρημα κατηγόρημά φασιν οἱ φιλόσοφοι.

⁹ τὸ τῶν ἢ τοῖν ἢ ἄλλο τι τοιοῦτον οὐχ ἑνὸς γένους κατηγορεῖται.

98 Predicates and Subjects

—that is to say, they are not applied to names of one gender only. Or again,

he notes that

‘ἄλλοι’ will always take an article when it embraces the whole of the predicated

plurality

(synt i 63 [53.18–19])¹⁰

—that is to say, when it embraces all the items to which it is applied.

In several texts, ‘κατηγορεῖν’ is coupled with, and is evidently a variant

upon, ‘σημαίνειν’ or ‘signify’. For example,

it is clear that <the adverb> ‘κύκλῳ’ signiﬁes a spatial relation not because it is a

derived form but rather insofar as ‘κύκλος’ too predicates of a spatial relation.

(adv 204.16–18)¹¹

He does not mean that in a sentence such as ‘That ﬁgure is a circle’ you

predicate a spatial relation of something. He means that the adverbial phrase

‘in a circle’, like the name from which it derives, signiﬁes a certain spatial

relation.

Again, inﬁnitives are names and

in appropriate sayings they may take an article, since a nominal predication of the

object is presented.

(adv 129.20–21)¹²

That is to say, an inﬁnitive may signify an object in the way in which names

do. And here is a passage from Herodian, Apollonius’ son:

Words of more than three syllables which end in -αλιος are proparoxytone—unless

they are predicated of birds.

(pros cath 123.5–6)¹³

Thatis,unlesstheyarenamesofbirds.

I ﬁnd only one place—apart from a few passages in which Apollonius

reports certain Stoic views—in which a member of the ‘κατηγορεῖν’family

might plausibly be deemed to signify predication. Of the word ‘τις’ he asks:

¹⁰ πάντοτε οὖν τὸ ἄλλοι συνέξει τὸ ἄρθρον ἡνίκα τοῦ κατηγορουμένου πλήθους ὅλου ἐστὶν

ἐμπεριληπτικὸν.

¹¹ καὶ δῆλον ὅτι τὸ κύκλῳ οὐ διὰ τῆς παραγωγῆς τὴν εἰς τόπον σχέσιν σημαίνει, ἀλλὰ

καθὸ καὶ τὸ κύκλος κατηγορεῖ σχέσεως τοπικῆς.

¹² ... κατὰ τοὺς δέοντας λόγους ἄρθρου ἐστὶ προσδεκτικά, ἐπεὶ ἅπαξ παρυφίσταται

ὀνοματικὴ κατηγορία τοῦ πράγματος.

¹³ τὰ εἰς αλιος ὑπερτρισύλλαβα προπαροξύνεται, εἰ μὴ ὀρνέου κατηγοροίη.

Predicates in Ancient Grammar 99

How is it that, although it is a monosyllable, it is not lengthened, as names

are?—Well, that is a predicate of an expression and not of what is thought by way

of the expression.

(pron 27.25–26)¹⁴

Apollonius means that if I say, for example,

Dogs has a short O

then having a short O, or ‘has a short O’, applies to the expression ‘dogs’

and not to the canine beasts. Certainly, in that passage ‘κατηγόρημα’could

signify ‘predicate’; and yet even there it is plain that Apollonius means

no more than that having a long O holds of the word and not of the

animal.

In sum, Apollonius uses the word ‘ὑποκείμενον’, but he does not use it to

mean ‘subject’; and he uses the word ‘κατηγόρημα’, but he does not use it

to mean ‘predicate’.

Let it be added that Apollonius’ way with these words is not idiosyncratic;

nor is it peculiar to the grammarians. On the contrary, it can be paralleled

in many other texts—among them philosophical texts, and even Peripatetic

philosophical texts.

I end this resolutely negative argument by returning to a Latin author. For

Martianus Capella’s account of the liberal arts—the Marriage of Philology

and Mercury—indicates the ancient state of things as clearly as you might

wish. In Book iii, which is given to grammar, subjects and predicates are

not noticed. In Book iv, on dialectic or logic, Capella promises to explain

‘what is the subject part of a sentence and what the predicative [declarativa]’

(341), and also to indicate ‘what a predicative [praedicativus] syllogism is’

(343). He duly does so; and in addition he offers an account of ‘the ten

predications’ (see 383). All this is part of his account of Aristotelian logic;

and nothing is plainer than that in Capella’s mind predication is not a matter

for grammarians and is a matter for logicians.

Some historians of ancient grammar have lamented the fact that it ignored

subjects and predicates; and they have tried to explain how such lamentable

ignorance could have come about. In truth, there is nothing to lament, and

nothing to explain.

¹⁴ πῶς οὖν οὐ τείνεται μονοσυλλαβοῦν ὡς τὰ ὀνόματα; τοῦτο φωνῆς κατηγόρημα, οὐ τοῦ

νοουμένου ἀπὸ τῆς φωνῆς.