Barnes Jonathan. Truth, etc

Подождите немного. Документ загружается.

130 Predicates and Subjects

Socrates is a man

has a single predicate, while

Socrates is a white man

Socrates talks sitting down

are compound and conjoined. … Aristotle discusses this in the de Interpretatione,and

Theophrastus—at greater length—in his On Afﬁrmation.

(in APr 367.3–14)⁵²

We do not know what Theophrastus said on the matter. As for Aristotle,

Alexander is thinking of a passage in Chapter 11 of the de Interpretatione

which obscures rather than illuminates.

Nonetheless, compounded predicates are recognized in the Peripatetic tra-

dition, and a few later texts discuss the issue. It was generally supposed that

conjunction is the paradigmatic wayof producinga compoundpredicate. Such

conjunction may be done either with or without a conjoining particle. If I say

Al Burlap was tough and brawly,

the conjunctive predicate ‘tough and brawly’ is formed by placing ‘and’

between two simple predicates. If I add

He was a fork-lift-truck-driving man,

the conjunctive predicate ‘fork-lift-truck-driving man’ is formed without the

aid of such a connector. Now in principle, a conjunctive predicate is no

more than the conjunction of two or more simple predicates, so that the

predication of a compound predicate may be construed as the conjunction

of two or more predications of simple predicates. Thus

Al Burlap was tough and brawly,

is equivalent to, or perhaps even an abbreviated form of,

Al Burlap was tough and Al Burlap was brawly.

So, in a straightforward way, the compound predicate ‘tough and brawly’ is,

as Plotinus would have put it, two predicates.

Some compound predicates may indeed be analysed after that conjunctive

fashion. But it must be avowed that the analysis does not greatly help

the Aristotelian logician. The mouse syllogism used the compound predicate

‘brave beast’; and you might allow that ‘So-and-so is a brave beast’ is equivalent

⁵² ἔτι δὲ ἢ ἁπλῶς τε καὶ ἄνευ συνθέσεως κατηγορητέον, τοῦτ᾿ ἔστιν ἕν τι καὶ μιᾶς

κατηγορίας, ἢ συμπεπλεγμένα τε καὶ συγκείμενα· ἡ μὲν γὰρ Σωκράτης ἄνθρωπός ἐστιν

ἁπλοῦν ἔχει τὸ κατηγορούμενον, ἡ δὲ λέγουσα Σωκράτης ἄνθρωπος λευκός ἐστιν ἢ Σωκράτης

καθήμενος διαλέγεται σύνθετόν τε καὶ συγκείμενον. ... καὶ αὐτὸς μὲν ἐν τῷ Περὶ ἑρμηνείας,

ἐπὶ πλέον δὲ Θεόφραστος ἐν τῷ Περὶ καταφάσεως, περὶ τούτων λέγει.

Complex Predications 131

to, or even an abbreviated form of, ‘So-and-so is brave and so-and-so is a

beast’. In that case, the ﬁrst premiss of the syllogism,

The mouse is a creature of great personal valour

Mice are brave beasts,

is equivalent to, or even synonymous with,

Mice are brave and mice are beasts.

But for the second premiss,

Brave beasts die young

No comparable equivalence is forthcoming—certainly, the premiss is not

equivalent to:

The brave die young and beasts die young.

So the conjunctive analysis cannot justify the mouse syllogism.

In any event, there are innumerably many complex predicates to which

the conjunctive analysis does not apply. The complex predicate ‘fork-lift-

truck-driving man’ is not an ordinary conjunction of simple predicates; and

although Aristotle knew nothing about fork-lifting, he more or less recognized

the point which it here illustrates. For he remarks that

of man it is true to say animal separately and biped separately—and also as one. Also

man and pale, and those two as one. But it is not the case that if he is a shoe-maker

and good, he is a good shoe-maker.

(Int 20b33–36)⁵³

Some compound predicates are conjunctive, or equivalent to conjunctions,

and others are not.

Socrates is a pale man

is equivalent to the conjunction of

Socrates is pale

and

Socrates is a man.

But

Simon is a remarkable old cobbler

is not equivalent to the conjunction of

Simon is remarkable,

Simon is old,

⁵³ κατὰ γὰρ τοῦ ἀνθρώπου ἀληθὲς εἰπεῖν καὶ χωρὶς ζῷον καὶ χωρὶς δίπουν, καὶ ὡς ἕν, καὶ

ἄνθρωπον καὶ λευκόν, καὶ ταῦθ ᾿ ὡς ἕν· ἀλλ᾿ οὐχί, εἰ σκυτεὺς καὶ ἀγαθός, καὶ σκυτεὺς ἀγαθός.

132 Predicates and Subjects

and

Simon is a cobbler.

True, the text which I have quoted does not say exactly what I have just said;

nonetheless—as I cautiously put it—Aristotle more or less acknowledges the

point that complex predicates are not always conjunctive.

There is a deeper difﬁculty. In principle, the simple predicates are all

and only those which belong to one of Aristotle’s ten classes of predicate.

Every other predicate is complex—that is to say, it is somehow formed from

the simple predicates by means of various operations which Aristotle once

thought he ought to look into. But what sort of complexity and what sort

of simplicity are in play? Evidently, the notions in question are semantic;

and a ﬁrst shot at explaining complexity might look like this: A predicate is

complex if and only if its meaning is determined by the meaning of its simple

components (and by the nature of the composition which unites them). So

‘remarkable old cobbler’ is complex inasmuch as its sense is ﬁxed by the senses

of its three parts.

Then consider the technical term ‘featherweight’. It is surely simple,

according to the account I have just sketched; for although it is, in an

innocuous sense, compounded from the words ‘feather’ and ‘weight’, the

meanings of those terms do not ﬁx the meaning of ‘featherweight’. So

perhaps it is a simple predicate—in the Aristotelian class of quantity? But

the word ‘featherweight’ is deﬁned thus: A featherweight is a boxer who

weighs between 8st 6lb and 9st if he is professional and between 8st 7lb and

9st if he is amateur. There is complexity enough there; and it is an ad hoc

and accidental complexity—few will persuade themselves that the term is

nevertheless a simple member of one of the ten Aristotelian classes.

There is no need to resort to ﬁsticuffs to make the point. Let me invent

the word ‘robbler’ and deﬁne it thus: A robbler is a remarkable old cobbler.

The term ‘robbler’ is a simple expression inasmuch as it has no semantically

active parts. But it does not belong to an Aristotelian class—it is, in Plotinian

terms, three predicates.

So ‘robbler’ and ‘featherweight’ must be complex predicates: if their

complexity is not found on the surface, then it is revealed by their deﬁnitions.

That might suggest that an expression is complex, in the pertinent sense,

if and only if it is deﬁnable. For it is the deﬁnitions of ‘robbler’ and of

‘featherweight’ which furnish the several predicates which together compose

and account for the complexity of the terms which they deﬁne. In itself that

is a sensible sort of notion; but it has disastrous consequences for Aristotle.

Complex Predications 133

For it determines that the word ‘man’, for example, is complex: the word may

have no evident parts to it; but it is deﬁnable—as ‘rational mortal animal’ (vel

sim)—and for that reason it is complex. If that line of inquiry is pursued, it

will turn out that—if Aristotle’s theory is true—there are exactly ten simple

predicates: ‘substance’, ‘quantiﬁed item’, ‘qualiﬁed item’, and so on. Now

some ancient texts do indeed hint at that conclusion; but it does not answer

to anything in Aristotle—or in Plotinus, or in Porphyry.

Aristotelians thus need to distinguish the simplicity of ‘man’ from the

complexity of ‘featherweight’. They thought to do so by declaring that the

former, but not the latter, signiﬁes ‘one thing’—that it signiﬁes some single

or uniﬁed item. The deﬁnitional formula for ‘man’, namely ‘rational mortal

animal’, is semantically complex; but the term and its deﬁnition signify a

unity and for that reason, or in that sense, the term is not complex. The

deﬁnitional formula for ‘featherweight’ not only is complex: in addition, it

signiﬁes a plurality, or at least a non-unity—featherweights tend to crack up.

But under what conditions does a term or a formula signify a unity? Not

even Aristotle managed to give a satisfactory answer to the question.

As Plotinus mercilessly demonstrated, the Aristotelian classiﬁcation of

predicates—the so-called ‘doctrine of the categories’—is a quagmire; and

any account of predication which is built upon that doctrine must wobble.

If you take the Categories to be the ﬁrst part of a unitary Organon,andif

you suppose that the ‘doctrine of the categories’ prepares the ground for the

‘doctrine of the syllogism’ which the Prior Analytics presents, then you are up

the creek—and paddleless.

Happily, the central part of the Categories, in which the ‘doctrine of the

categories’ is developed, has nothing to do with the logic of the Analytics.Of

course, Aristotle’s syllogistic is essentially tied to the concept of predication; for

the argument forms which it examines are ﬁxed by a certain logical structure,

namely the subject–predicate structure. But nothing in the syllogistic requires,

or even suggests, any classiﬁcation of predicates: that a predicate is substantial

or qualitative, relational or a matter of habitus —all that is of supreme

indifference to the syllogistic. There is no difference whatever, from a

syllogistic point of view, between ‘Every man is an animal’, where the

predicate is substantial, and ‘Every man is less than ten feet tall’, where the

predicate is quantitative.

I do not mean that no classiﬁcation of predicates could be of pertinence to

any theory of inference. I do not even mean that Aristotle’s classiﬁcation of

predicates has no logical interest. I mean, simply and indisputably, that the

134 Predicates and Subjects

Aristotelian classiﬁcation of predicates has no bearing upon the Aristotelian

theory of inference.

If the syllogistic does not require, or even invite, any classiﬁcation of

predicates, neither does it require, or even suggest, that a predicate must be

simple or that it must belong to one or other of the ten Aristotelian classes.

As far as syllogisms are concerned, there is no difference at all between ‘man’

and ‘featherweight’, between ‘ape’ and ‘fork-lift-truck-driving man’, between

‘ﬁnch’ and ‘soprano who sings Schubert like an angel’. Indeed, Aristotelian

syllogistic need not even be cowed by the monster predicates which modern

logicians like to breed. The formula ‘… is such that it is identical to itself

and in Manchester it never rains on Bank Holiday Mondays’ is a perfectly

acceptable predicate in modern logic; for it is an expression which may

legitimately occupy the place of the ‘F’ in inferences of the sort:

F(Socrates): so F(something).

By the same token ‘an item such that it is identical to itself and in Manchester

it never rains on Bank Holiday Mondays’ is a perfectly acceptable Aristotelian

predicate; that is to say, it is an expression which may feature alongside such

quotidian items as ‘man’ and ‘animal’ in an Aristotelian syllogism.

‘SOMETHING OF SOMETHING’

If predicates and subjects are expressions, how shall we determine, in the

case of a given sentence or proposition, what is a subject for it and what a

predicate? In the Prior Analytics Aristotle explains that

a proposition is a saying which afﬁrms or denies something of something.

(APr 24a16–17)⁵⁴

In the de Interpretatione there is something rather more elaborate:

Of these [sc. assertoric sayings] some are simple assertions, namely those which say

something of something or something from something, and some are composed

from them, namely those which are thereby compound sayings.

(Int 17a20–22)⁵⁵

⁵⁴ πρότασις μὲν οὖν ἐστὶ λόγος καταφατικὸς ἢ ἀποφατικός τινος κατά τινος.

⁵⁵ τούτων δ᾿ ἡ μὲν ἁπλῆ ἐστὶν ἀπόφανσις, οἷον τι κατά τινος ἤ τι ἀπό τινος, ἡ δ᾿ ἐκ τούτων

συγκειμένη, οἷον λόγος τις ἤδη σύνθετος.

‘Something of Something’ 135

The propositions of the Analytics seem to be identical with the simple

assertions of the de Interpretatione —on the plausible assumption that to

deny something of something is the same thing as to say something from

something. So in predicative sentences, something is said of something, τι

κατά τινος—and the ﬁrst something corresponds to a predicate, the second

to a subject.

In that case, if we want to determine subject and predicate in a proposition,

we might apply the ‘something of something’ test. That is to say, we might

ask: What, here, is said of what? Take again

Bugles sang.

What does that sentence say of what? Or rather, if someone utters such a

sentence, what does he say—or what can he be saying—of what? Well, it is

pretty clear that the sentence, or an appropriate utterer of the sentence, says

or may say of bugles that they sang. And in that case we shall identify ‘an

item which sang’ as a predicate in the sentence and ‘bugle’ as a subject.

That banal idea invites two simple comments, and two less simple

reﬂections.

The ﬁrst comment is this: the ‘something of something’ test will allow

predicates of any degree of linguistic complexity. (The same goes, of course,

for subjects.) ‘Fork-lift-truck-driving man’ will make an impeccable predicate

(and an impeccable subject). Or take this relatively complex sentence:

Love is not love which alters where it alteration ﬁnds or bends with the

remover to remove.

What does the sentence—or what did Shakespeare—say of what? Surely it

or he said something of love, namely that it doesn’t alter &c; so that ‘love’

will be a subject and ‘an item which doesn’t alter &c’ a predicate.

In the de Interpretatione Aristotle distinguishes between simple and com-

pound sayings, and his words imply that the distinction is exclusive. But

consider the sentence:

If young hearts were not so clever, oh, they would be young forever.

That is surely a compound sentence; equally surely, it, or its utterer, says

something of something: it says of young hearts that, were they not so clever

&c. The familiar sentence which begins with ‘If you can keep your head’ and

continues for some twenty lines before it reaches ‘you’ll be a man, my son’,

also says something of something.

I do not suppose that Aristotle had imagined examples of that sort.

Nonetheless, it is clear that the distinction he makes in the de Interpretatione

ought to be construed not as a distinction between sorts of proposition but

136 Predicates and Subjects

as a distinction between certain propositional forms; and it is clear, too,

that when the Analytics limits propositions to what the de Interpretatione

calls simple assertions, it thereby excludes few propositions (if any) from its

domain.

The second comment: the items which are predicate and subject in a

sentence need not, themselves, be literally parts of the sentence. The simple

sentence ‘Bugles sang’ perhaps contains the expression ‘bugle’, which is a

subject for it; but it certainly does not contain the expression ‘an item which

sang’, which is a predicate in it. In general, if a simple sentence has the form

‘So-and-so verbs’, then the expression ‘verbing’ or ‘an item which verbs’

will be a predicate for it—and those expressions do not literally appear

in it.

The point was made—or at least half made—by Aristotle in wholly

general terms:

We say, without qualiﬁcation, of all cases that the terms should always be set down

according to the names of the words—e.g. man or good or opposites—not of man

or of good or of opposites—but that the propositions should be taken in accordance

with the cases of each word.

(APr 48b39–49a2)⁵⁶

The remark is obscure (and my translation perhaps exaggerates some of

its obscurities); and the concrete applications which Aristotle proceeds to

offer are curious. But what Aristotle means is this: in setting out the terms

of a proposition—in designating predicate and subject—we should always

use the nominative case; in setting out the proposition, we should use

whatever case is grammatically appropriate. One of the terms of a sentence

may be ‘opposites’ (nominative plural), while the sentence itself contains ‘of

opposites’ (genitive plural).

The ﬁrst of the two reﬂections I promised takes its start from that last

remark. If a subject and a predicate of a sentence need not be literally parts

of the sentence, then how are they related to the sentence? Must they not

at least have, as it were, a representative in the sentence, as ‘an item which

sang’ has ‘sang’ to represent it? That is so in all the examples which Aristotle

gives, and the text which I have just cited implies that it must always be so.

Nevertheless, various cases suggest themselves in which it is tempting to ﬁnd

⁵⁶ ἁπλῶς γὰρ τοῦτο λέγομεν κατὰ πάντων ὅτι τοὺς μὲν ὅρους ἀεὶ θετέον κατὰ τὰς κλήσεις

τῶν ὀνομάτων, οἷον ἄνθρωπος ἢ ἀγαθόν ἢ ἐναντία, οὐκ ἀνθρώπου ἢ ἀγαθοῦ ἢ ἐναντίων,

τὰς δὲ προτάσεις ληπτέον κατὰ τὰς ἑκάστου πτώσεις.

‘Something of Something’ 137

subjects and predicates from beyond the conﬁnes of their sentences. Here are

three sorts of example.

The ﬁrst is the most extreme. In answer to the question ‘What happened

in London yesterday evening?’ I might utter the sentence

A nightingale sang in Berkeley Square

and thereby say of a certain summer evening that it was such that … , or

of London that it was such that … If that is so, does not the ‘something

of something’ test make ‘yesterday evening’, and ‘London’, subjects for the

sentence? But those two alleged subjects have no representative of any sort in

their sentence. You might urge that the test will declare them subjects if it

refers to utterers, but that it will not do so if it refers to sentences. For although

an utterer of the sentence ‘A nightingale sang …’ might use it—might very

well use it—to say something of London, surely the sentence itself does not

say anything of London. (But why not? After all, it says something of Berkeley

Square—and isn’t that a decent way of saying something about London?)

The second sort of example may be illustrated by Duncan’s remark:

He was a man in whom I placed the most absolute trust.

Duncan thereby said of the Thane of Cawdor that he had trusted him;

and the sentence—or at least, a certain occurrence of the sentence—says

something of the Thane. So the ‘something of something’ test makes ‘the

Thane of Cawdor’ a subject for the sentence. It is true that ‘the Thane of

Cawdor’ has a representative in the sentence, namely the expression ‘he’; but

the relation between ‘the Thane of Cawdor’ and ‘he’ is not at all the same as

the relation between ‘sang’ and ‘an item which sang’.

Thirdly, consider cases in which the language changes. When you sing

Und der Haiﬁsch, der hat Z

ahne,

you surely say something of sharks, namely that they have teeth; and the

sentence, equally surely, says of sharks that they have teeth. So ‘shark’ is a

subject for the sentence and ‘an item which has teeth’ is a predicate in it. Yet

those two English expressions do not appear in the German sentence. To be

sure, they have their representatives—in the expressions ‘der Haiﬁsch’ and

‘hat Z

ahne’. But again, they are not representatives in the way in which ‘sang’

represents ‘an item which sang’.

Or perhaps the expression ‘shark’ does occur in the German sentence? Is it

nottruethatinhisAnalytics Aristotle uses the word ‘syllogism’ I don’t know

how many times? It would be at best a weak joke to say: ‘On the contrary,

he never uses it—nor any other English word.’ Is it not true that Frege took

138 Predicates and Subjects

pains to distinguish between the sense and the reference of the singular term

‘The Morning Star’?—‘No he didn’t: he was exclusively concerned with the

sense and reference of a German word.’

The ancient texts, needless to say, do not consider—let alone resolve—such

conundrums. But the main point which I want to make here is both ancient

and true: even when, on the Porphyrean account, subjects and predicates

are taken to be signiﬁcant expressions, there is no reason to think that the

expressions must be literally parts of the sentences of which they are subjects

and predicates—and there is every reason to think that they are not always

such parts.

That brings me to a second reﬂection. Ancient logicians and modern

scholars commonly speak of the subject and the predicate of a sentence

or proposition, as though every sentence or proposition which exhibits a

subject–predicate structure had a single determinate subject and a single

determinate predicate. But in fact, as some of the examples have already

hinted, the ‘something of something’ test will accommodate a plurality of

subject–predicate analyses. In uttering

When beggars die there are no comets seen,

what may I say of what? Pretty plainly, I may say something about beg-

gars—that their deaths are not marked by the appearance of comets. Equally

plainly, I may say something about comets—that they do not mark the deaths

of beggars. Slightly less evidently, I may say something about deaths—that

when they are beggarly, they are not marked by comets. And a little ingenuity

will elicit other potential subjects and predicates in that not very complicated

sentence. It is not that sometimes I may say one of those things and sometimes

another. True, if I produce the sentence in the course of a lecture on astrology,

it will be natural to remark that I thereby said something about comets. And

so I did—but I also, and in the same breath, said something about beggars,

and deaths, and God knows what else.

That sentences admit multiple analyses, and that one analysis may be

pertinent to one inferential context and another to another—those are

commonplaces of modern logic. They were not commonplaces of ancient

or of traditional logic. Indeed, they were not recognized by ancient logi-

cians—with the possible exception of a handful of texts which I shall discuss

in a later context. Nonetheless, it is worth insisting on the fact that noth-

ing in the Aristotelian theory of predication excludes multiple predicative

analyses.

Styles of Predication 139

STYLES OF PREDICATION

The slanderer who predicated philippism of Demosthenes did so inasmuch

as he said of Demosthenes that he philippized. Similarly, the sentence, or an

utterer of the sentence

The world is too much with us

predicates ‘an item which is too much with us’ of ‘the world’ inasmuch as it

or he thereby says of the world that it is too much with us. But if the sentence

Demosthenes philippizes

predicates ‘an item which philippizes’ of ‘Demosthenes’, it does not fol-

low—and according to Demosthenes it was not true—that ‘item which

philippizes’ is predicated of ‘Demosthenes’. So a sentence or an utterer may

predicate something of something without its thereby being the case that the

something is predicated of the something.

That sounds odd—if not downright contradictory. In any event, it forces

the following question: If when someone predicates x of y, it does not follow

that x is predicated of y, then what exactly is it for x to be predicated of y?

The Aristotelians often use the verb ‘ὑπάρχειν’(+ dative)—‘hold of’ or

‘apply to’ or ‘belong to’—as equivalent to ‘κατηγορεῖσθαι’(+ genitive).

More precisely, it is clear that, at any rate in the Analytics, x is predicated of

y if and only if x holds of y. The Stoics also connected predication—their

notion of predication—to holding or belonging: Chrysippus

says that only the present holds: the past and the future subsist but do not

hold—except in the way in which some predicates hold, namely those and only

those which are attributed—e.g. to walk holds of me when I am walking and does

not hold when I am sitting or lying down.

(Stobaeus, ecl i viii 42)⁵⁷

A Stoic predicate is not an expression but an item sayable by way of an

expression; but, like an Aristotelian predicate, such an item, when it is

predicated, holds of what it is predicated of.

⁵⁷ μόνον δ᾿ ὑπάρχειν φησὶ τὸν ἐνεστῶτα [sc. χρόνον], τὸν δὲ παρῳχημένον καὶ τὸν μέλλοντα

ὑφεστάναι μέν, ὑπάρχειν δὲ οὐδαμῶς, εἰ μὴ ὡς καὶ κατηγορήματα ὑπάρχειν λέγεται μόνα τὰ

συμβεβηκότα, οἷον τὸ περιπατεῖν ὑπάρχει μοι ὅτε περιπατῶ, ὅτε δὲ κατακέκλιμαι ἢ κάθημαι

οὐχ ὑπάρχει.—Text and translation are doubtful in parts; but the doubts do not affect the point

for which I cite the passage.