Barnes Jonathan. Truth, etc

Подождите немного. Документ загружается.

180 What is a Connector?

It is true that when Aristotle introduces sayings in the grammatical

prologue to the de Interpretatione, his mind is set upon sentences. But his

formal deﬁnition of ‘λόγος’ does not restrict the word to sentences:

a saying is a signiﬁcant expression some part of which is signiﬁcant in separation.

(Int 16b26–27)¹⁵

Names and verbs have been deﬁned as signiﬁcant expressions no part of

which is signiﬁcant in separation. A saying, then, is any signiﬁcant expression

which is longer than a name or a verb; and the deﬁnition, which is found

again, word for word, in the Poetics (1457a23–24), covers far more than

sentences. Not only that: neither of the two examples of unitary sayings

which Aristotle gives in the Poetics is a sentence—the deﬁnition of man (that

is to say, the deﬁniens of ‘man’) is not a sentence but a complex predicate;

the Iliad is not a sentence but a connected sequence of sentences. (See

Poet 1457a28–30: the same examples are found at Met 1045a12–14; cf.

1030b9–19.) A deﬁnition is subsentential, a poem is supersentential: it is

as though Aristotle went out of his way to indicate that not all sayings are

sentences.

There are some oddities in all that—in particular, the allusion to the Iliad

is bafﬂing. No doubt the Iliad is one poem (give or take a few interpolations);

but its unity as a poem presumably depends on such matters as unity of

action and of character and not upon the fact that it is stitched together with

connecting particles. Perhaps the Iliad is one saying in virtue of the fact that

you can say it? ‘What did he say?’—‘This is what he said: ‘Sing, Goddess,

the wrath of Achilles …’.’ But if an item is one saying provided that it can ﬁll

the blank in ‘He was six foot tall in his stockinged feet, and this is what he

said:—’, then absolutely any string of expressions will constitute a saying. So

in what sense is the Iliad one saying? And how do Homer’s connectors make

it so? I have no idea.

However that may be, it is plain that Aristotelian connectors are not

conceived of as sentential connectives: they are not presented as items whose

primary and fundamental r

ole is the linking of sentences to one another.

Rather, they are what I have called neutral connectors: they are items which

take expressions and make expressions—and the expressions which they take

may, for all that Aristotle says, be anything from the shortest of phrases to

the longest of orations.

¹⁵ λόγος δέ ἐστι φωνὴ σημαντικὴ ἧς τῶν μερῶν τι σημαντικόν ἐστι κεχωρισμένον.

Deﬁning the Connector 181

DEFINING THE CONNECTOR

Connectors play no part in Aristotelian syllogistic. In Stoic logic they come

into the limelight, and Stoic syllogistic turns about ‘if ’ and ‘or’ and ‘and’. For

every Stoic syllogism contains essentially at least one non-simple assertible;

and, in Sextus’ words,

non-simple assertibles are those … which are composed from a repeated assertible

or from different assertibles, and in which a connector or connectors govern. Take

for the moment what they call the conditional. That is composed from a repeated

assertible or from different assertibles by way of the connector ‘εἰ’or‘εἴπερ’.

(M viii 108–109)¹⁶

Sextus is speaking of ‘the logicians’ in general; but Diogenes Laertius conﬁrms

that what Sextus reports was Stoic doctrine (see vii 68–71). There are puzzling

aspects to the report which I shall for the moment ignore. But whatever the

resolution of the puzzles, Stoic interest in connectors is secure. Of course, it

would be astonishing were that not so. And in fact we know that Chrysippus

had written monographs on different sorts of non-simple assertibles—and,

more to the present point, that at least one later Stoic had written an essay

speciﬁcally on connectors.

The grammarians, too, discussed connectors, and by far the most detailed

and interesting of ancient texts on the subject—of extant ancient texts—is

the Connectors of Apollonius Dyscolus. In the opening paragraph of his

monograph, Apollonius remarks that he will not ‘wholly avoid the opinion

of the Stoics’ (conj 214.2–3).¹⁷

Some scholars have mistaken the text to declare that Apollonius will not

depart in the least from the opinion of the Stoics; and many scholars who

understand the text aright nevertheless claim that the grammarians’ views

on connectors were heavily inﬂuenced by the Stoics. That there was some

inﬂuence is certain—and at the end of this chapter I shall discuss a pertinent

case. But it is well to remember another passage near the start of Connectors,

a passage in which Apollonius criticizes some of his predecessors who

¹⁶ καὶ δὴ οὐχ ἁπλᾶ μέν ἐστιν ἀξιώματα τὰ ἀνώτερον προειρημένα, ἅπερ ἐξ ἀξιώματος

διφορουμένου ἢ ἐξ ἀξιωμάτων διαφερόντων συνέστηκε καὶ ἐν οἷς σύνδεσμος ἢ σύνδεσμοι

ἐπικρατοῦσιν. λαμβανέσθω δὲ ἐκ τούτων ἐπὶ τοῦ παρόντος τὸ καλούμενον συνημμένον· τοῦτο

τοίνυν συνέστηκεν ἐξ ἀξιώματος διφορουμένου ἢ ἐξ ἀξιωμάτων διαφερόντων καὶ διὰ τοῦ εἴ

ἢ εἴπερ συνδέσμου.

¹⁷ οὐκ ἐκτὸς γινόμενοι κατὰ τὸ παντελὲς τῆς τῶν Στωϊκῶν δόξης.

182 What is a Connector?

used terms which are foreign to the subject rather than those pertaining to grammar,

and who introduced Stoic opinions, the transmission of which is not particularly

useful for the technical study of grammar.

(conj 213.8–10)¹⁸

Stoic opinions, on grammar in general and on connectors in particular, are

occasionally worth heeding but more usually irrelevant or worse. Whether or

not Apollonius’ judgement is just, we should be wary of ﬁnding too much

Stoicism in or behind his Connectors.

But perhaps he took his deﬁnition of connectors from the Stoics? The

Stoics, according to Diogenes Laertius, explained that

a connector is a caseless part of sayings which connects the parts of sayings.

(vii 58)¹⁹

That explanation is in certain ways surprising, and I shall return to the

surprises. For the moment, it is enough to observe that it is echoed in a

number of grammatical texts. A fragment from a Greek Art of Grammar,

which is optimistically ascribed to Trypho by the papyrus which preserves it,

contains this little exchange:

What is a connector?—An expression connective of parts of sayings.

(PLitLond 182, iii 106–107)²⁰

And according to a commentator on the Dionysian Art,

Apollonius, deﬁning the term, says that a connector is a particle which is connective

of the parts of sayings.

(scholiast to Dionysius Thrax, 435.40–436.1)²¹

So Apollonius took his account of connectors, directly or indirectly, from the

Stoic philosophers?

Connectors has survived in a single stained manuscript, and the text opens

in mid-sentence. No doubt the lost opening pages of the essay contained an

explicit deﬁnition of the term ‘connector’; and it is natural to imagine that the

¹⁸ ὀνόμασιν ἀλλοτρίοις προσχρησάμενοι ἤπερ τοῖς εἰς γραμματικὴν συντείνουσι, Στωϊκὰς

παρεισφέρουσι δόξας, ὧν ἡ παράδοσις οὐκ ἄγαν χρειώδης πρὸς τὴν εἰς γραμματικὴν

συντείνουσαν τεχνολογίαν.

¹⁹ σύνδεσμος δέ ἐστι μέρος λόγου ἄπτωτον, συνδοῦν τὰ μέρη τοῦ λόγου.

²⁰ σύνδεσμος τί ἐστιν; λέξις συνδετικὴ τῶν τοῦ λόγου μερῶν.

²¹ ὁ δὲ ’Απολλώνιος ὁριζόμενός φησι σύνδεσμον εἶναι συνδετικὸν μόριον τῶν τοῦ λόγου

μερῶν.

Deﬁning the Connector 183

commentator I have just cited was reporting that lost deﬁnition. But in fact

the commentator’s note is a paraphrase of a surviving sentence in Apollonius’

Syntax, where he remarks that

after all the parts already catalogued, we mentioned connectors which are connective

of them.

(synt i 28 [27.10–11])²²

As for the deﬁnition which Apollonius presumably gave in his Connectors,

that can be recovered by another route—and it is not just a repetition of the

Stoic deﬁnition.

The story is a little complicated. In the Art of Grammar which goes under

the name of Dionysius, connectors are explained as follows:

A connector is an expression connecting thoughts, together with order, and showing

the gap in the interpretation.

(20 [86.3–4])²³

The last clause is puzzling, and I shall get it out of the way before turning to

the rest of the deﬁnition.

What on earth can the last clause mean? ‘Showing the gap in the

interpretation’?—The word ‘interpretation’ here means much the same as

‘expression’. But how can a connector show a gap in an expression? Perhaps

the Greek text is corrupt and must be emended? A passage in Apollonius’

Connectors has been thought to suggest as much. It is concerned with the

so-called ‘parapleromatic’ or expletive connectors. Those particles (to which

I shall return) were thought, by most ancient grammarians but not by

Apollonius, to serve a purely ornamental function. According to Apollonius,

Trypho, wanting to include them too [sc expletive connectors] in his deﬁnition [sc

of the connector] says: and sometimes ﬁlling up the gap in the interpretation.

(conj 247.23–25)²⁴

That last phrase, which is a direct quotation from Trypho’s lost Art, seems to

echo the last clause of the Dionysian deﬁnition. But Trypho uses the verb ‘ﬁll

up [παραπληροῦν]’ where our text of the Art has ‘show [δηλοῦν]’. Trypho’s

²² ἐπὶ πᾶσι δὲ τοῖς κατειλεγμένοις ὁ τούτων συνδετικὸς σύνδεσμος παρελαμβάνετο.

²³ σύνδεσμός ἐστι λέξις συνδέουσα διάνοιαν μετὰ τάξεως καὶ τὸ τῆς ἑρμηνείας κεχηνὸς

δηλοῦσα.

²⁴ ὁ Τρύφων ἐν τῷ ὅρῳ βουλόμενος καὶ αὐτοὺς ἐμπεριλαβεῖν φησί· καὶ τὸ κεχηνὸς τῆς

ἑρμηνείας ἔστιν ὅπου παραπληρῶν.

184 What is a Connector?

‘ﬁlling up’ makes eminently good sense; for what do expletive connectors

do but ‘ﬁll up the gap’? (In Greek, the point is triﬂing; for their name,

‘παραπληρωματικοί’, derives from Trypho’s verb ‘παραπληροῦν’.) Now

‘παραπληροῦν’ is a compound form of ‘πληροῦν [to ﬁll]’; and ‘πληροῦν’

looks not unlike ‘δηλοῦν’ on the written page. That has suggested that

what Trypho read in his copy of the Dionysian Art was ‘ﬁlling’ rather

than ‘showing’, ‘πληροῦσα’ rather than ‘δηλοῦσα’. And surely what Trypho

read was what the author of the Art originally wrote? The puzzling phrase

‘showing the gap’ is in our texts by accident—an ancient or mediaeval copyist

miscopied his exemplar, and our editions have inherited his mistake. The

suggestion is apparently supported by the fact that some of the mediaeval

manuscripts of the Art actually offer the reading ‘πληροῦσα’.

Thesuggestion is enticing.But there aresnags. First, thereading ‘πληροῦσα’

was unknown to the ancient commentators on the Art, and the supposition

that it is nonetheless authentic implies a curious and implausible textual his-

tory. Secondly, Trypho cannot have modelled himself upon the extant Art,

which was not compiled until long after his death. Thirdly, Trypho’s clause

was intended to capture one special, and specially recalcitrant, type of con-

nector, whereas the Dionysian deﬁnition appears designed to ﬁt connectors

in general.

All things considered, we had best soldier on with ‘showing’ or ‘δηλοῦσα’.

Then what can it mean? The ancient commentators offered various explan-

ations, the least bad of which proposes that the clause was intended to cope

with disjunctive connectors. After all, disjunctive connectors can hardly be

said to connect a thought—but might they not picturesquely be said to show

up a gap in things? Perhaps. But—again—the last clause in the deﬁnition

ought to say something about connectors in general and not about a particular

species of connector.

Some modern scholars have taken the clause to mean that the presence of

a connector reveals not an actual gap but a virtual gap—something which

would otherwise have been a gap. For example, the connector ‘or’ shows the

gap in the expression

Heads or tails?

inasmuch as that same expression minus the connector, namely

Heads tails?

has a gap in it. Other modern scholars have observed that an expression like

Heads or

Deﬁning the Connector 185

is incomplete, that it is as it were followed by a gap; and they have suggested

that the connector ‘or’ shows the gap inasmuch as its presence at the end of

the incomplete expression indicates that there is a gap to be ﬁlled.

Will you buy any of those interpretations? I won’t—but I have nothing

better to put on the market. Happily, the matter is not grave; for although

many later grammarians discussed the last clause of the Dionysian deﬁnition,

none of them imitated it; and it had no signiﬁcance for ancient theorizing

about connectors.

With that out of the way, what about the rest of the deﬁnition? The Art

states that

a connector is an expression connecting thoughts, together with order.

In antiquity, not even that was accepted without question. Indeed, it was

roundly criticized and replaced by an improved version. The criticisms, which

need not be rehearsed, suggested that the deﬁnition was gravely deﬁcient,

requiring the adjunction of several further clauses. The criticisms also sugges-

ted that it was in one respect incorrect or misleading, and that for the phrase

‘thoughts’ there should be substituted ‘parts of sayings’. (I shall return to the

latter point.) Here is the improved version in one of its variant forms:

A connector is an uninﬂected part of sayings, connective of the parts of sayings, with

which it co-signiﬁes, determining either order or force or both order and force.

(scholiast to Dionysius Thrax, 102.15–18)²⁵

The improved deﬁnition, and variants upon it, are found in other texts, both

Greek and Latin. Diomedes, for example, gives this:

A connector is an uninﬂected part of sayings which links the expressions and

connects the force and order of the parts of sayings. For it received its name on

this account—because it is inserted into a saying as a bond; for, like a chain, when

introduced it binds expressions which are loose and diffuse.

(ars gramm I 415.13–16])²⁶

The Latin deﬁnition is a close paraphrase of the Greek. And the etymological

explanation—‘it received its name on this account’—must have been cribbed

²⁵ σύνδεσμός ἐστι μέρος λόγου ἄκλιτον, συνδετικὸν τῶν τοῦ λόγου μερῶν, οἷς καὶ συσση-

μαίνει, ἢ τάξιν ἢ δύναμιν ἢ καὶ τάξιν καὶ δύναμιν παριστῶν.

²⁶ coniunctio est pars orationis indeclinabilis copulans sermonem et coniungens vim et ordinem

partium orationis. nam ob hoc meruit nomen quia pro vinculo interponitur orationi. laxatum enim et

diffusum sermonem more catenae interposita devincit.

186 What is a Connector?

from a Greek text; for it makes no sense in Diomedes’ Latin. (It pays

unconscious tribute to Quintilian’s preferred translation of ‘σύνδεσμος’.)

No one names the inventor of the improved deﬁnition. But the Greek

commentators on the Dionysian Art generally draw heavily on Apollonius,

and it is a sporting bet that they do so here. Moreover, in Book xvi of

his Institutions Priscian offers a translation of the improved deﬁnition as his

own account of what connectors are (inst xvi i1[iii 93.2–3]²⁷); and at the

beginning of Book xvii he states that

in the previous books about the parts of sayings we have for the most part followed

the authority of Apollonius.

(xvii i1[iii 107.23–24])²⁸

Where Priscian’s statement can be checked, it turns out to be true. So there

is another reason for thinking that the improved deﬁnition of connectors

derives from Apollonius.

The improved deﬁnition is not identical with the Stoic deﬁnition, which

ran thus:

A connector is a caseless part of sayings which connects the parts of sayings.

(Diogenes Laertius, vii 58)

But it adopts all the clauses of the Stoic deﬁnition, and adds a few more: why

not think that it was developed by reﬂection upon the Stoic deﬁnition?

However that may be, the Stoic deﬁnition and the improved deﬁnition

differ from the Aristotelian deﬁnition (and from the modern deﬁnition of a

connective) in one striking way.

PARTS OF SAYINGS?

Modern connectives connect sentences. Aristotle’s connectors connect expres-

sions. According to the Stoics, connectors connect parts of sayings; and almost

all the grammarians, both Greek and Latin, echo the Stoics. In particular,

the improved deﬁnition—the Apollonian deﬁnition—echoes them on that

point.

It is tempting to think that in these deﬁnitions the phrase ‘parts of

sayings’ should be taken to mean ‘the parts of the connected sayings which

²⁷ coniunctio est pars orationis indeclinabilis coniunctiva aliarum partium orationis quibus consigni-

ﬁcat vim et ordinationem demonstrans.

²⁸ in ante expositis libris de partibus orationis in plerisque Apollonii auctoritatem secuti sumus.

Parts of Sayings? 187

the connector produces’. A grammatical papyrus which dates from the ﬁrst

centuryA.D.saysthat

a connector is an expression which links the parts of interpretations.

(PYale i 25, ii 54–55)²⁹

That is surely intended as a variant on the standard deﬁnition; and it is readily

understood to mean that a connector is an item which connects the elements

in the expression of which it is a part. And I ought to confess that the phrase

which I have translated by ‘the parts of sayings’ has the word for saying in the

singular and preﬁxed by a deﬁnite article: perhaps a better translation is ‘the

parts of the saying’, that is to say ‘the parts of the saying <in which it occurs

as a connector>’?

There is a reason to hope that such an interpretation may be the right

one. The connectors which appear essentially in Stoic logic serve to connect

sayings—indeed, they function as sentential connectives (or at least, as items

very closely related to sentential connectives). In that case, a Stoic connector

connects the parts of the saying in this sense: the saying in which a connector

appears itself has parts which are sayings, and the connector serves to bind

those parts together. The deﬁnition which we are considering means, in effect

and at bottom, to explain connectors in terms of sentential connection.

That interpretation is tempting. But the Greeks did not read the deﬁnition

in that way—nor did the Romans. ‘Parts of sayings’, as they took the phrase,

designates what we normally call the parts of speech. (Indeed, most English

translators reasonably opt for ‘parts of speech’.) For example:

Note also that the grammarian puts the account of connectors last in order. And

reasonably: if they are called connectors inasmuch as they connect and bind and

interweave, and if the items which are bound must be furnished before the connectors

are, then the grammarian could not but take them after the other parts of sayings.

(scholiast to Dionysius Thrax, 435.36–40;³⁰ cf. 283.24–25)

Why are connectors treated after the other parts of sayings—names, verbs,

adverbs, and so on?—Because they serve to connect those parts. The same

notion is implicit in Apollonius, who thinks that the parts of sayings have an

²⁹ σύνδεσμος δ’ ἐστὶν λέξις συνάπτουσα τὰ μέρη τῆς ἑρμηνείας.

³⁰ ἰστέον δὲ πάλιν ὅτι τὸν περὶ συνδέσμου λόγον τελευταῖον τέταχεν ὁ τεχνικός. καὶ

εἰκότως· εἰ σύνδεσμος λέγεται παρὰ τὸ συνδεῖν καὶ συνδεσμεῖν καὶ συμπλέκειν, δεῖ δὲ πρὸ

τῶν συνδέσμων τὰ δεσμευόμενα προευτρεπίζεσθαι, ἀναγκαίως μετὰ τὰ προειρημένα μέρη

τοῦ λόγου παρὰ τῷ τεχνικῷ ἔσχατος παρείληπται.

188 What is a Connector?

intelligible order—so that, at the end of the list, ‘after all the parts already

catalogued, we mentioned connectors which are connective of them’ (synt i

28 [27.10–11]).³¹

So it seems that a connector will be an item which links, say, names to

names, and verbs to verbs, and so on—and perhaps also names to verbs,

and adverbs to adjectives, and so on. And that is just how the deﬁnition was

construed. Perhaps the clearest statement of the view comes in a late and

anonymous Latin commentary on Donatus:

A connector connects two names (‘Virgil and Priscian’), two pronouns (‘you and I’),

two verbs (‘reads and writes’), two adverbs (‘yesterday and today’), two participles

(‘reading and writing’), it even connects itself (‘if and when’), two prepositions

(‘around and about’), two interjections (‘alas and alack’).

(in Don viii 263.23–27)³²

The text perhaps suggests, incautiously, that any connector may connect

instances of any part of sayings; but the fundamental idea is plain: connectors

connect parts of sayings inasmuch as they may link a name to a name, a verb

to a verb, and so on.

When the Latin grammarians explain why connectors are needed, they

standardly take a pair of pronouns:

Our sayings are, in their nature, separated and discrete, and they cannot come into

connection unless by the interposition of those parts [i.e. connectors]. … If someone

says

Let you me go,

the utterance is not complete; but if you interpose ‘and’, you make the utterance

complete.

(Pompeius, in Don v 264.18–22;³³ cf. Servius, in Don iv 418.4–14)

The Greeks used a pair of proper names rather than a pair of pronouns—they

took a line from the Iliad, dropped an ‘and’ which linked two names, and

remarked that the line then falls apart (e.g. scholiast to Dionysius Thrax,

283.15–19).

³¹ ἐπὶ πᾶσι δὲ τοῖς κατειλεγμένοις ὁ τούτων συνδετικὸς σύνδεσμος παρελαμβάνετο.

³² coniungit enim duo nomina ut Virgilius et Priscianus, duo pronomina ego et tu, duo verba legit et

scribit,duoadverbiautheriethodie,duoparticipiautlegensetscribens,etseipsamconiungitutsietsi,

duas praepositiones circum et circa, duas interiectiones heu et euax.

³³ naturaliter enim nostra oratio dissidens et soluta est, nec potest in conexionem venire nisi interpositis

illis particulis. … siqui dicat ego tu eamus, non est plena ista elocutio. sed si interponas et, facis plenam

elocutionem.

Parts of Sayings? 189

That being so, it is no accident that ‘and’ is, so to speak, the paradigmatic

connector. It was so for Varro who explained what connectors are by reference

to the Latin ‘et’:

What we say in uttering ‘Cicero and Antony were consuls’, with that same ‘and’ we

can bind together any two consuls—or, to put it more generally, any names and

indeed any words.

(LL viii iii 10)³⁴

‘and’ is syntactically omnivorous—had Varro taken ‘if ’, he would have had

to tell a slightly different story.

Pompeius’ introduction to his example suggests that any two words need

to be glued together by a connector. But of course he does not mean to say

anything so daft; rather—to judge from the context—he means to say that

items from the same part of sayings need connecting. So too in Greek:

What we call connectors do not connect a name and a verb (no one says: Trypho

and reads), but either a name and a name or a verb and a verb (Theo and Trypho,

reads and writes). For connectors are connective of items of the same sort or as if of

the same sort: I and you and Apollonius.

(scholiast to Dionysius Thrax, 516.7–11)³⁵

Certain words ﬁt naturally together, they are made for one another: a name

satisﬁes the need of an intransitive verb, an adjective attaches itself by its own

strings to a name, and so on; but a name won’t stick to a name, nor a verb

to a verb. Inter-class bonding needs no help from outside; for items from

different classes automatically cohere. Intra-class bonding is another matter.

That idea is not silly. But it is at best eccentric. After all, sayings are

not themselves parts of sayings; and yet one of the primary r

oles played by

connectors—by the items which the Greek and the Roman grammarians

classiﬁed as connectors—is the r

ole of joining sayings or sentences to one

another. That must have been plain to the grammarians; and it must have

been utterly evident to the Stoics inasmuch as in their logic connectors play

an exclusively sentential r

ole.

³⁴ sic quod dicimus in loquendo consul fuit Tullius et Antonius, eodem illo et omnes binos consules

colligere possumus, vel dicam amplius omnia nomina atque etiam omnia verba.

³⁵ καὶ αὐτοὶ οἱ παρ᾿ ἡμῖν καλούμενοι σύνδεσμοι οὐ συνδεσμοῦσιν ὄνομα καὶ ῥῆμα· οὐδεὶς

γὰρ λέγει Τρύφων καὶ ἀναγινώσκει· ἀλλ᾿ ἢ ὄνομα καὶ ὄνομα, ἢ ῥῆμα καὶ ῥῆμα, Θέων

καὶ Τρύφων, γράφω καὶ ἀναγινώσκω· οἱ γὰρ σύνδεσμοι ὁμοιομερῶν εἰσι συνδετικοὶ ἢ ὡς

ὁμοιομερῶν, ἐγὼ καὶ σὺ καὶ ᾿Απολλώνιος.