Barnes Jonathan. Truth, etc

Подождите немного. Документ загружается.

240 What is a Connector?

The text is difﬁcult, and it is far from plain exactly what Apollonius wants

to say. But the easiest interpretation goes like this: Some items are true or

natural disjuncts, others are not. If certain items are not natural disjuncts

and you wish to afﬁrm their factual disjunction, then you must use a

disjunctive connector. But if the items are natural disjuncts, then nature has

already done the trick: if you use a connector, you are merely gilding the

lily—to say

It’s day, it’s night

is to afﬁrm a disjunction.

That sounds rum; but a later passage supports the interpretation. Apollonius

is urging that the word ‘ἆρα’, as it is used to introduce a question, is

a connector. Some grammarians had observed that if you drop the word

from a sentence, what remains will be a complete sentence with the same

sense:

In ‘ἡμέρα ἐστίν’—I mean, interrogatively taken—nothing more is thought if ‘ἆρα’

is added.

(conj 225.12–13)¹¹⁵

In Greek, the interrogative ‘ἆρα’ is optional: ‘Is it day?’ translates indifferently

as ‘ἆρα ἡμέρα ἐστίν;’ and as ‘ἡμέρα ἐστίν;’. Such a phenomenon, the

grammarians argued, can never occur with genuine connectors: drop a

connector, and things alter. Apollonius replies thus:

As far as that argument goes, we shall not even allow ‘ἤ’ to be a connector. At least,

the following are phrases

Youread,youdon’tread

You go away, you don’t go away

—but it is admitted that the phrase lacks ‘ἤ’. …

We went as you bid into the wood, great Odysseus.

We found ﬁne houses built in the glens

—‘and’ is missing.

(conj 225.18–24)¹¹⁶

There, allegedly, are three examples of sayings from which items have been

dropped without loss of syntax or change of sense: the items are ‘or’ and

¹¹⁵ ἐν μέντοι γε τῷ ἡμέρα ἐστί, λέγω κατ᾿ ἐρώτησιν, οὐδὲν πλεῖον νοεῖται μετὰ τοῦ ἆρα.

¹¹⁶ ἕνεκά γε τοῦ τοιούτου οὐδὲ τὸν ἤ σύνδεσμον καταλείψομεν. ἔστι γάρ που τοιαύτη

φράσις· ἀναγινώσκεις, οὐκ ἀναγινώσκεις· ἀπέρχῃ, οὐκ ἀπέρχῃ. ἀλλ᾿ ὡμολογημένως λείπεται

ἡ φράσις τοῦ ἤ ... ᾔομεν, ὡς ἐκέλευες, ἀνὰ δρυμά, <φαίδιμ᾿ ᾿Οδυσσεῦ·> εὕρομεν ἐν βήσσῃσι

τετυγμένα δώμ<ατα καλά> · λείποντος τοῦ καί.

Connectorless Connections 241

‘and’. If the grammarians’ argument about ‘ἆρα’ were sound, then neither

‘or’ nor ‘and’ would be a connector—and that is absurd.

The last, conjunctive, case is perhaps less surprising than the other two

disjunctive examples. Nonetheless, it is not without its perplexities; for the

Syntax quotes the same Homeric lines in the opposite sense:

… and again, by their absence, <connectors> dissolve sayings, as in

We went as you bid into the wood, great Odysseus.

We found ﬁne houses built in the glens.

For he should have conjoined it with ‘and’:

And we found …

(synt i 10 [11.6–10])

The two sentences do not hold together: Homer should have used a connector.

According to Connectors, Homer has not omitted anything: the connection,

and more precisely the conjunction, is there in the text without one.

However that may be, the two disjunctive cases are surely odder. Apollonius

says—at least, he appears to say—that if it is impossible that both P and Q,

so that those two items are true or natural disjuncts, then if you produce one

of them followed by the other you have thereby made a disjunctive utterance.

Now you might perhaps persuade yourself that, in colloquial English, you

can sometimes put forward a disjunction without using any connector to

help you do so—

Love me, leave me—it’s all one to me.

And perhaps the same thing is possible in Greek. But only, surely, when

the context—or the tone of voice, or some other external factor—somehow

intimates a disjunction. If I say

She loves me, she loves me not,

I do not make a disjunctive assertion; and yet if any items are naturally

disjoined, then surely a saying and its negation are so.

There are other cases—at least in English—which are less problematical.

For example, if I say

They make smoking illegal: I emigrate

I will readily be understood to have made a conditional assertion. Or take:

Love me, love my dog.

And so on. Perhaps such things are idioms, and so do not count? Perhaps

they are elliptical, and so not genuine examples of connectorless connections?

Of course, if someone utters the words

242 What is a Connector?

They make smoking illegal: I emigrate,

it is perfectly correct to report what he said by saying:

Hesworethathe’demigrateiftheybannedsmoking;

and in uttering those words he surely did make a conditional assertion. But it

does not follow that he did so by making a single conditional saying; and in

fact he did not utter a single sentence—his words do not pass the embedding

test for sentences.

Connectorless connections are interesting items; but—to return to my lost

sheep—they do not begin to show that connectors have a purely syntactical

function. Rather, they show that certain syntactical turns have a semantic

function. The sentence

Had we but world enough and time, this coyness, lady, were no crime

is a conditional saying—it is not a conjunction, say, nor a disjunction. It

contains no conditional connector; but its syntactical structure determines

that it is equivalent to

If we had world enough and time, then this coyness, lady, would be no

crime.

The connectorless version does not show that the connector ‘if ’ has no

sense: the version with a connector shows that the syntactical turn has a

semantic function. As Ammonius put it, the connectorless version contains

a connector potentially: it possesses a feature which has the potency—the

force or meaning—of a connector.

Well, all that is pretty obvious. And it is pretty obvious that—whatever

you think about glue and parquet—the thesis that connectors, in general,

have no semantic force is false. If ‘and’ and ‘or’ have no meaning, then they

do not differ in meaning; and in that case there will be no difference in

meaning between

Yourmoneyoryourlife

and

Your money and your life.

EXPLETIVE CONNECTORS

But perhaps at least some connectors—some classes of connector—have no

semantic function? The ancient grammarians distinguished several species or

kinds of connector. The Dionysian Art lists eight:

Expletive Connectors 243

Of connectors some are conjunctive, some disjunctive, some implicative, some quasi-

implicative, some causative, some interrogative, some syllogistic, some expletive.

(20 [87.1–88.2])¹¹⁷

The last kind, the expletive connectors, proved troublesome.

They had been noticed by the philosophers. The Peripatetic Demetrius

remarks that

you should use the expletive connectors not as empty additions … but only if they

contribute something to the grandeur of the saying.

(eloc 55)¹¹⁸

He offers some examples of the grandifying use of expletives; and he calls

upon the Peripatetic Praxiphanes in his support; for

those who ﬁll out connectors to no purpose are, as Praxiphanes says, like actors who

utter this and that sound to no purpose.

(eloc 58)¹¹⁹

Byzantine scholars liked to claim that Praxiphanes—after Aristotle—had

perfected the art of grammar (scholiast to Dionysius Thrax, 164.23–29).

Perhaps one of his acts of perfection was the recognition of a class of expletive

connectors? Perhaps. But it seems to me more likely that Praxiphanes was

talking not about a class of expletive connectors but about expletive uses

of connectors. That is to say, he thought that connectors—or certain

connectors—sometimes had a purely ornamental function, and he warned

speakers against overindulgence.

The Stoic Chaeremon also discussed expletive connectors. He argued—so

Apollonius reports—that the items may properly be called connectors: not

in virtue of their sense or syntax (for they do not have a sense, and they do

not serve to connect anything) but rather in virtue of their form (for they

are morphologically very much like real connectors). (See conj 248.1–13).

Nothing else is known of Chaeremon’s thesis, to whom no other text ascribes

any interest in linguistic or logical matters.

¹¹⁷ τῶν δὲ συνδέσμων οἱ μέν εἰσι συμπλεκτικοί, οἱ δὲ διαζευκτικοί, οἱ δὲ συναπτικοί,

οἱ δὲ παρασυναπτικοί, οἱ δὲ αἰτιολογικοί, οἱ δὲ ἀπορηματικοί, οἱ δὲ συλλογιστικοί, οἱ δὲ

παραπληρωματικοί.

¹¹⁸ τοῖς δὲ παραπληρωματικοῖς συνδέσμοις χρηστέον, οὐχ ὡς προσθήκαις κεναῖς ... ἀλλ᾿

ἂν συμβάλλωνταί τι τῷ μεγέθει τοῦ λόγου.

¹¹⁹ οἱ δὲ πρὸς οὐδὲν ἀναπληροῦντες, φησί, τὸν σύνδεσμον ἐοίκασιν τοῖς ὑποκριταῖς τοῖς τὸ

καὶ τὸ πρὸς οὐδὲν ἔπος λέγουσιν.

244 What is a Connector?

Chaeremon thought that expletive connectors have no meaning; and

according to Apollonius, ‘it is a prejudice of most people that the so-

called expletive connectors have no signiﬁcation’ (conj 247.22–23).¹²⁰ The

prejudice must seem well-grounded. For the items are deﬁned by the

Art as

those connectors which are introduced for the sake of metre or ornament;

(20 [96.3–97.1])¹²¹

that is to say, as Priscian expressed it, they ‘are put in for the sake of

ornament or metre and not by any necessity of sense’ (inst xvi ii 13 [iii

102.13–14]).¹²²

In general, a word is used expletively in a given expression if it can be

deleted from the expression without destroying the syntax or changing the

sense; and a word is an expletive tout court if it can be deleted from any

expression which contains it without damaging the syntax or altering the

sense. In that case, an expletive can have no sense, and it cannot connect

anything. Hence it cannot be a connector.

Chaeremon in effect accepted the argument down to its last step. But there

he stopped: although expletive connectors connect nothing, there are still good

reasons for calling them connectors. That is a curious suggestion: Chaeremon

admits that the expletives do not satisfy the deﬁnition of ‘connector’—and

he wants nonetheless to call them connectors.

Apollonius rejects the argument in its entirety. He urges at some consid-

erable length, both in Connectors and in the Syntax, that expletive connectors

have a meaning and are genuine connectors. (See conj 247.22–253.28; synt

iii 127 [378.5–380.8].) His conclusion ought to be that the so-called explet-

ive connectors are not expletives—for they cannot always be deleted; and

in one passage he does appear to say just that (conj 250.11–12). But his

considered view is that they may properly be called expletive connectors. Any

connector—and indeed any part of speech—will, he claims, sometimes be

used expletively. In

¹²⁰ παρὰ τοῖς πλείστοις ἐστὶ πρόληψις ὡς οἱ καλούμενοι παραπληρωματικοὶ σημασίαν

τινὰ οὐ ποιοῦνται.

¹²¹ παραπληρωματικοὶ δέ εἰσιν ὅσοι μέτρου ἢ κόσμου ἕνεκεν παραλαμβάνονται.

¹²² quaecumque coniunctiones ornatus causa vel metri nulla signiﬁcationis necessitate ponuntur hoc

nomine nuncupantur.

Expletive Connectors 245

This was the most unkindest cut of all

‘most’ is used expletively, and so is ‘of all’. The expletives help out the metre,

and add a certain brutal grandeur to the line; but you may delete them

without destroying the syntax or damaging the sense. It does not follow,

and it is not true, that the word ‘most’ and the phrase ‘of all’ are expletives.

According to Apollonius, the so-called expletive connectors differ from other

connectors inasmuch as they are used expletively very often. Indeed, that fact

is the only thing which they all have in common—which is why they are

called ‘expletive’.

The class of expletives is a ragbag. Apollonius resorts to that same

useful receptacle when he explains the subjunctive mood (synt iii 126

[377.5–8]), or paronyms and verbal names (synt iii 130 [381.5–9]), or

possessives and comparatives (conj 253.21–28). In each case, he ﬁnds that

most species of a given kind of expression are classiﬁed and named in

virtue of their common semantic features; but certain items evade such

ordering—so they are grouped together and named in virtue of some

accident which they happen to share. The device is handy but unscientiﬁc.

Apollonius ought to have divided the so-called expletive connectors among

the other classes of connector—if necessary inventing a few new classes on

the way.

However that may be, the quarrel over expletive connectors rumbled on.

Some of the commentators on the Dionysian Art side with Apollonius—

expletives too signify something, just like the other connectors;

(scholiast to Dionysius Thrax, 105.31–32)¹²³

others insist that

they were called expletives because they are especially set down in order to ﬁll out the

metre and also to provide a certain adornment for the discourse—for they contribute

nothing to the thought.

(scholiast to Dionysius Thrax, 441.21–23)¹²⁴

The Latin tradition, unsurprisingly, shows the same features as the Greek.

¹²³ οἱ παραπληρωματικοὶ καὶ αὐτοὶ σημαίνουσί τι, καθάπερ καὶ οἱ ἄλλοι σύνδεσμοι.

¹²⁴ παραπληρωματικοὶ δὲ ἐκλήθησαν, ἐπεὶ μάλιστα τοῦ πληρῶσαι ἕνεκεν τὸ μέτρον

τίθενται μετὰ τοῦ καὶ κόσμον τινὰ παρέχειν τῷ λόγῳ· πρὸς γὰρ νόησιν οὐδὲν συντείνουσιν.

246 What is a Connector?

CO-SIGNIFICATION

Some theorists held, hopelessly, that connectors have no semantic force at all.

Others held, less hopelessly, that some connectors—expletive connectors—

have no semantic force. The enigmatic Lucius, who complained that Aris-

totle’s account of homonymy cannot be applied to connectors, evidently

thought that connectors had a meaning. So too Nigidius Figulus: at the end

of an essay on the Latin connector ‘quia’—an essay which derives from that

dubious Pythagorean—Aulus Gellius writes that

no one will have understood the forms and varieties of the particle which we are

talking about until he has learned that it is composed and conjoined and that it does

not only have a force of connecting but is also endowed with a certain determinate

meaning.

(NA xvii xiii 10)¹²⁵

It is sometimes said that Diodorus Cronus had argued—much earlier, and

demonstratively—against the thesis that connectors signify nothing. For

Diodorus the logician thought that every utterance was signiﬁcant and in proof of

this called one of his own servants ‘And yet’ and another by another connector.

(Ammonius, in Int 38.18–20)¹²⁶

The grammarians dutifully report that Diodorus made ‘And yet’ a name; but

they do not explain what the point of the gesture was. Ammonius’ explanation

is guess-work, and pretty implausible guess-work. Later, Stephanus said that

Diodorus intended to indicate that words do not have any meaning by nature

(in Int 9.20–24); and his guess is rather more intelligent. But none of that

matters a whit; for the anecdote is doubtless a ﬁction.

We are better informed about another adversary of the thesis that connect-

ors mean nothing. Near the beginning of Connectors, having remarked that

he will not wholly avoid Stoic doctrine, Apollonius says that

Posidonius, in his On Connectors argues against those who afﬁrm that connectors

do not show anything but simply connect the phrase: he says that ‘ἐπιδοῦναι’

differs from ‘ἀποδοῦναι’, and ‘ἀπαιτεῖν’from‘προσαιτεῖν’, and some other such

¹²⁵ hanc particulam de qua dicimus nisi si quis didicerit compositam copulatamque esse neque vim

tantum coniungendi habere sed certa quadam signiﬁcatione factam, numquam profecto rationes et

varietates istius comprehensurus est.

¹²⁶ … τὸν διαλεκτικὸν ∆ιόδωρον πᾶσαν οἰόμενον φωνὴν σημαντικὴν εἶναι καὶ πρὸς πίστιν

τούτου καλέσαντα τῶν ἑαυτοῦ τινα οἰκετῶν ᾿Αλλαμὴν καὶ ἄλλον ἄλλῳ συνδέσμῳ.

Co-Signiﬁcation 247

constructions (thereby showing himself to believe that prepositions and connectors

are a single part of sayings).

(conj 214.4–8)¹²⁷

In his essay On Connectors —which we know of only from Apollonius—

Posidonius, like Nigidius after him, attacked the thesis that connectors merely

connect and have no meaning.

His argument rests on some such principle as the following: If an expression

C(X) differs in meaning from C(Y), then X differs in meaning from Y.

Now ‘[ἐπι]δοῦναι’ differs in meaning from ‘[ἀπο]δοῦναι’. Hence ‘ἐπί ’

differs in meaning from ‘ἀπό’. But if X differs in meaning from Y, then

both X and Y have a meaning. Hence ‘ἐπί ’and‘ἀπό ’ have a meaning.

But ‘ἐπί ’and‘ἀπό ’ are connectors. Hence connectors—or at least, some

connectors—have a meaning. The argument depends upon the Stoic notion

that prepositions—and in particular, verbal preﬁxes—are connectors; for

Posidonius’ examples turn about verbal preﬁxes. But that peculiarity apart, it

is well adapted to showing that connectors are not meaningless signs—that

they have a sense. Of course, as I have already remarked, you would have to

be insane to deny that they do.

But what, or how, do connectors signify? Apollonius denies that connectors

have a meaning of their own. He does not mean thereby to deny that they

have a meaning. On the contrary, he will frequently say that a connector

‘shows’ something (δηλοῦν:e.g.conj 228.11–13; 229.14; 230.16), and he

is not averse to mentioning the ‘signiﬁcation’ of a connector (σημασία:e.g.

conj 235.26; 251.28). More generally, he holds that

parts of sayings are distinguished not by their form but by their meaning;

(pron 67.6;¹²⁸ cf. synt i 77 [65.4–11]; ii 33 [150.14–15])

so that all parts of sayings must have a meaning.

The way in which the connectors have a meaning is expressed by the verb

‘συσσημαίνειν’or‘co-signify’.IntheSyntax Apollonius remarks that

connectors indicate their proper meanings in relation to the constructions or

consequences of the sayings.

(synt i 12 [14.4–6])¹²⁹

¹²⁷ Ποσειδώνιος ἐν τῷ Περὶ συνδέσμων ἀντιλέγων πρὸς τοὺς φάσκοντας ὡς οἱ σύνδεσμοι

οὐ δηλοῦσι μέν τι, αὐτὸ δὲ μόνον τὴν φράσιν συνδέουσι, φησὶν ὡς διαφέρει τὸ ἐπιδοῦναι

τοῦ ἀποδοῦναι, ὡς τὸ ἀπαιτεῖν τοῦ προσαιτεῖν, καὶ ἄλλας τινὰς τοιαύτας συντάξεις, ἤδη

πιστούμενος ὅτι ἓν μέρος λόγου ἥ τε πρόθεσις καὶ ὁ σύνδεσμος.

¹²⁸ οὐ γὰρ φωναῖς μεμέρισται τὰ τοῦ λόγου μέρη, σημαινομένοις δέ.

¹²⁹ οἱ ... σύνδεσμοι πρὸς τὰς τῶν λόγων συντάξεις ἢ ἀκολουθίας τὰς ἰδίας δυνάμεις

παρεμφαίνουσιν.

248 What is a Connector?

That is to say, a connector co-signiﬁes inasmuch as its sense is determined by

the constructions and consequences of the items which it connects. That is a

dark saying (and the text is disputed). A passage in Priscian has been thought

to shed light on it.

Some expressions, Priscian says, are incapable of producing—either on

their own or in collaboration with one another—an expression with a

complete sense. Prepositions and connectors are expressions of that sort:

In fact, they always co-signify, i.e. signify when connected to other items and not

in themselves. Hence their signiﬁcation varies according to the force of the items

with which they are connected—as ‘in’ signiﬁes one thing when it is joined to an

accusative, another when it is joined to an ablative.

(inst xvii i10[iii 114.18–22])¹³⁰

Connectors, like prepositions, are ambiguous items, and their sense can only

be ﬁxed by the context in which they ﬁnd themselves—for example, I shall dis-

tinguish the temporal ‘cum’ from the causal ‘cum’ by remarking that the latter

but not the former takes a subjunctive verb in the clause which it introduces.

Some scholars have discovered in those remarks an elucidation of the

notion of co-signiﬁcation: an expression co-signiﬁes, they suggest, if it is

ambiguous and if the ambiguity can only be resolved—or is habitually

resolved—by contextual indications. Now Priscian does indeed suggest

that contextual ambiguities are somehow a consequence of co-signiﬁcation;

but that suggestion is not presented as an elucidation of co-signiﬁcation;

and it would in fact be a hopelessly inadequate elucidation. For it is not

only connectors and the like which are ambiguous, and it is not only the

ambiguities of such items which are habitually resolved by appeal to the

context—if contextual disambiguation is a mark of co-signiﬁcation, then all

parts of sayings are co-signiﬁcant.

The passage from Priscian is based on Apollonius. Apollonius follows his

claim that prepositions ‘always co-signify’ by appealing to the difference in

sense between ‘διά’withthegenitiveand‘διά’ with the accusative, and he

then offers a parallel case involving a connector rather than a preposition—he

cites a Homeric line in which the context shows that the word ‘ἤτοι’isto

be taken in a conjunctive sense. (See synt i 12 [14.2–10].) Just before that

comes the following paragraph:

¹³⁰ eae etenim semper consigniﬁcant, id est coniunctae aliis signiﬁcant per se autem non. itaque

variatur earum signiﬁcatio ad vim coniunctorum eis, ut in aliud signiﬁcat cum accusativo iungitur et

aliud cum ablativo.

Co-Signiﬁcation 249

Again, among letters some are vowels which produce a sound by themselves and

others are consonants which do not possess a spoken pronunciation in the absence

of the vowels: you should imagine the same thing in the case of expressions. Some

of them can as it were be pronounced in the same way as vowels—as you may

imagine in the case of verbs, names, pronouns, adverbs …; others are as it were

consonants which wait upon vowels—that is to say, the parts of sayings which we

have just listed—and cannot be pronounced on their own: thus it is in the case of

prepositions, articles, connectors—such particles always co-signify.

(synt i 12 [13.1–14.2])¹³¹

A con-sonant has a sound: the letter ‘B’, for example, is not mute. But the

sound can only be made if the letter is linked to a vowel, as in ‘BA’. That

is why it is called a con-sonant. In a similar way, a co-signiﬁer signiﬁes

something, but its signiﬁcation can only come out if it is linked with some

other part of speech. That is why it is called a co-signiﬁer.

Then consider how Apollonius in fact explains the sense of connectors.

He will say, for example, that the disjunctive ‘ἤ’ announces the holding of

one of the disjuncts and the non-holding of the others; and I have earlier

suggested that the same explanation—or at any rate an extremely similar

explanation—may be expressed, in a different idiom, as follows:

Asentenceofthesort:‘ἤτοι’ + P

+ ‘ἤ’ + … + ‘ἤ’ + P

is true if and

only if precisely one P

is true.

It is not evident how such formulas will be devised for all the connectors

which Apollonius discusses; and the formulas will require some gloss or rider

which adapts them to cases where a connector connects items other than

assertoric sentences. But the general notion is plain. It may be put like this:

ifyouwanttoexplainwhataconnector,C,means,thenyoumustcomeup

with something of the sort:

‘X(C)’ means that such-and-such.

That is to say, you do not—cannot—explain C in isolation: you explain

it in a linguistic context, you explain it as a part of a larger semantic

unit.

¹³¹ ἔτι ὃν τρόπον τῶν στοιχείων ἃ μέν ἐστιν φωνήεντα, ἃ καθ᾿ ἑαυτὰ φωνὴν ἀποτελεῖ, ἃ

δὲ σύμφωνα, ἅπερ ἄνευ τῶν φωνηέντων οὐκ ἔχει ῥητὴν τὴν ἐκφώνησιν, τὸν αὐτὸν τρόπον

ἔστιν ἐπινοῆσαι κἀπὶ τῶν λέξεων. αἱ μὲν γὰρ αὐτῶν τρόπον τινὰ τῶν φωνηέντων ῥηταί εἰσι,

καθάπερ ἐπὶ τῶν ῥημάτων ἔστιν ἐπινοῆσαι, ὀνομάτων, ἀντωνυμιῶν, ἐπιρρημάτων, ...· αἱ δὲ

ὡσπερεὶ σύμφωνα ἀναμένουσι τὰ φωνήεντα, τουτέστιν τὰ προκατειλεγμένα τῶν μερῶν τοῦ

λόγου, οὐ δυνάμεναι κατ᾿ ἰδίαν ῥηταὶ εἶναι, καθάπερ ἐπὶ τῶν προθέσεων, τῶν ἄρθρων, τῶν

συνδέσμων· τὰ γὰρ τοιαῦτα τῶν μορίων ἀεὶ συσσημαίνει.