Barnes Jonathan. Truth, etc

Подождите немного. Документ загружается.

250 What is a Connector?

When the Peripatetics say that names and verbs alone are signiﬁcant, they

are generally taken to mean that names and verbs alone designate or refer. In

All horses are now asleep

there are ﬁve words each of which possesses a sense; but only two of

them—namely ‘horses’ and ‘asleep’—refer to anything. Or as Plutarch

puts it,

it is not the case that, just as someone who utters ‘strikes’ or ‘is struck’ or again

‘Socrates’ or ‘Pythagoras’ has in a certain way provided us with something to think

of and reﬂect upon, so we can grasp a conception of an act or a body when someone

utters ‘on the one hand’ or ‘for’ or ‘about’ by itself.

(quaest Plat 1010a)¹³²

Thatisnodoubtcorrect,atbottom.Butsomethingmustbedoneto

explain what and how a word like ‘asleep’ might designate—after all, it

is names which refer or designate, and ‘asleep’ does not look much like

anything’s name.

Instead of ‘refer’ or ‘designate’, try ‘be true of ’. In other words, suppose

that to have a signiﬁcation is to be true of something or some things. More

precisely, say that an expression E signiﬁes if and only if its sense is, or can

be, given by way of a sentence of the form ‘E is true of an item if and

only if such-and-such’. Then in fact names and verbs—and names and verbs

alone—signify. (Adjectives?—Yes, but in ancient grammar, adjectives are a

sort of name.) As for ‘asleep’, its sense is given by the following truth:

‘asleep’ is true of an item if and only if that item is asleep.

You can’t produce anything like that for ‘and’ or ‘the’ or ‘quickly’ or ‘by’.

That is to say, such words are non-signiﬁcant, they do not signify. But

that does not for a moment imply that they are empty marks or that they

have no signiﬁcation at all; for they co-signify—and their signiﬁcation or

co-signiﬁcation must be elucidated in the way in which Apollonius elucidates

the sense of the connectors.

SYLLOGISTIC CONNECTORS

The grammarians, as I said, distinguished several species or kinds of connector.

The Dionysian Art offers a list of eight varieties. Later grammarians upped

¹³² οὐ γάρ, ὥσπερ ὁ τὸ τύπτει φθεγξάμενος ἢ τὸ τύπτεται καὶ πάλιν τὸ Σωκράτης ἢ τὸ

Πυθαγόρας ἁμωσγέπως νοῆσαί τι καὶ διανοηθῆναι παρέσχηκεν, οὕτω τοῦ μέν ἢ γάρ ἢ περί

καθ᾿ αὑτὸ ἐκφωνηθέντος ἔστιν ἔννοιάν τινα λαβεῖν ἢ πράγματος ἢ σώματος.

Syllogistic Connectors 251

the ante, the Latins being particularly prodigal: after listing a dozen sorts of

connector, one Latin manual remarks that

it is not possible to designate all of them by name on account of the vast number of

kinds or the subtlety of the distinctions by which one differs from another.

([Asper], ars gramm v 553.22–24)¹³³

The ﬁrst three items on the Dionysian list are the conjunctive, the disjunctive

and the implicative—the three sorts of connector which characterize the

three sorts of compound assertibles about which standard Stoic logic revolves.

The next two items on the list, the quasi-implicative and the causative, were

also recognized by the Stoic logicians, even if they did not enter as such

into Stoic syllogistic. It has been inferred that the classiﬁcation of connectors

which the Art offers has a Stoic origin.

The penultimate items on the list are syllogistic connectors. They are

explained as follows:

Syllogistic are those connectors which are well adapted to the ἐπιφοραί and

συλλήψεις of proofs. They are these: ἄρα ἀλλά ἀλλαμήν τοίνυν τοιγάρτοι

τοιγαροῦ.

(20 [95.2–96.1])¹³⁴

The six items which the Art presents as syllogistic connectors are all capable

of appearing in syllogisms. But they will appear in two different functions:

four of them may serve to introduce the conclusion of an argument (so

that they answer very roughly to ‘therefore’ or ‘so’); two of them—ἀλλά,

ἀλλαμήν—will not signal a conclusion but may be used to introduce a

co-assumption or supplementary premiss. Thus it seems that there are, in

fact, two sorts of syllogistic connectors; and since the Art says that syllogistic

connectors are adapted to ἐπιφοραί and συλλήψεις, presumably those two

terms pick out the two sorts.

That was the view of the ancient commentators on the Art:

He calls ἐπιφορά the introduction of the next saying and σύλληψις the sealing and

concluding of the preceding saying.

(scholiast to Dionysius Thrax, 441.8–10)¹³⁵

¹³³ omnes nominibus suis designari non possunt propter multitudinem generum aut subtilitatem

discriminum quibus aliae ab aliis differunt.

¹³⁴ συλλογιστικοὶ δέ εἰσιν ὅσοι πρὸς τὰς ἐπιφοράς τε καὶ συλλήψεις τῶν ἀποδείξεων εὖ

διάκεινται· εἰσὶ δὲ οἷδε· ἄρα ἀλλά ἀλλαμήν τοίνυν τοιγάρτοι τοιγαροῦν.

¹³⁵ ἐπιφοράν δὲ λέγει τὴν ἐπαγωγὴν τοῦ ἑξῆς λόγου, καὶ σύλληψιν τὴν ἐπισφράγισιν καὶ

συναγωγὴν τοῦ προηγησαμένου λόγου.

252 What is a Connector?

The commentator means that by ‘ἐπιφορά’weshouldunderstand‘co-

assumption’ and by ‘σύλληψις’ ‘conclusion’. Modern scholars have suggested

that things are the other way about: the ἐπιφορά is the conclusion (after

all, in Stoic logic, ‘ἐπιφορά’ is the standard term for conclusion), and

the σύλληψις is the co-assumption (for which ‘πρόσληψις’ is the nor-

mal term). Neither of those suggestions is particularly felicitous from a

linguistic point of view; and it is tempting to correct ‘συλλήψεις’into

‘προσλήψεις’.

There is another possibility, which is both plausible and depressing. The

list of six items does indeed contain two different sorts of item. But the items

are not listed according to their sort. The explanation of what a syllogistic

connector is does indeed use two different terms to pick out their function

or their functions. But the two terms are not ordinarily used to distinguish

co-assumptions from conclusions. It seems likely that the Art uses the two

words indifferently, and that it does not intend to distinguish two species of

syllogistic connector.

However that may be, syllogistic connectors are recognized, in one way or

another, by all the ancient grammarians. They occur, for example, in the Yale

grammatical papyrus, in the fragments of the Art optimistically ascribed to

Trypho, and in other papyri of the same type. Apollonius does not accord a

special treatment to syllogistic connectors in what survives of his Connectors.

But he incidentally conﬁrms that ‘ἄρα’ is a syllogistic connector—‘ἄρα’witha

short initial alpha, that is; and he adds—what is scarcely surprising, that ‘οὖν’

is another syllogistic connector—‘οὖν’ with a circumﬂex, that is. (See conj

227.24–25; 228.13–15; 229.21–22; 254.22–23.) The commentaries on the

Art have, in this case, pretty well nothing to add; for the two pertinent pas-

sages—65.27–34 and 105.27–30—are no more than inaccurate cribs from

Apollonius.

The Greek texts are meagre. Nonetheless, on their basis, and in particular

on the basis of the examples which they offer, it is reasonable to suppose

that a connector is syllogistic provided that it connects one component of

an argument to another—a subsequent premiss to an existing premiss (so

‘ἀλλά’and‘ἀλλαμήν’ in Dionysius’ list) or a conclusion to a premiss (the rest

of the examples). In that case, syllogistic connectors are, as I shall say, either

co-assumptional or inferential. (The point of that rebarbative nomenclature

will emerge in a moment.)

There is much more material in the Latin grammarians—but it is con-

fused and confusing. It is convenient to start from the views of Cominianus.

Syllogistic Connectors 253

Cominianus is known to posterity only from the pages of the Institu-

tions of Flavius Sosipater Charisius, a fourth century grammarian who—it

was normal practice—wrote with a pair of scissors and a pot of paste.

Cominianus, according to Charisius, offered a standard deﬁnition of con-

nectors:

a connector is a part of sayings which binds and orders thoughts.

(inst ii xiv [I 224.24–5])¹³⁶

He then distinguished ﬁve types of connector, according to their power or

potestas.Theword‘potestas’, which answers to the Greek ‘δύναμις’, clearly

means ‘meaning’ here; and Cominianus apparently assigned meanings to

connectors without embarrassment.

The ﬁfth type of connector Cominianus called ‘rational [rationalis]’; and

although Charisius reports no explanation of what makes a connector rational,

he does offer a list of examples:

quamobrem, praesertim, item, itemque, enim, etenim, enimvero, quia, quapropter,

quippe, quoniam, quoniamquidem, ergo, ideo, scilicet, propterea.

(inst ii xiv [I 225.2–4])

The list is presumably illustrative rather than complete. It contains some

co-assumptional connectors—notably ‘item’—and also some inferential

items—for example, ‘ergo’. So it seems likely that all the Dionysian

syllogistic connectors will count as rational. But Cominianus’ class of

rational connectors includes examples which are not syllogistic—for example,

‘enim’and‘quia’, the Greek versions of which are causative according to

the Art.

The later Latin grammarians mostly follow Cominianus, without saying

so: they give his deﬁnition and his list of ﬁve species—so, for example, Probus

(inst iv 143.24–144.21), Donatus (ars gramm ii 15 [iv 388.28–389.17),

Servius (in Don iv 418.4–30). And Pompeius pretends that that was the

Latin way of doing things:

The power of connectors is divided into ﬁve species among Latin authors (among the

Greeks it is variously divided). They are conjunctive, disjunctive, expletive, causal,

rational.

(in Don v 265.16–19)¹³⁷

¹³⁶ coniunctio est pars orationis nectens ordinansque sententiam.

¹³⁷ potestas coniunctionum apud Latinos in quinque species dividitur (apud Graecos enim varie

dividitur). sunt enim copulativae disiunctivae expletivae causales rationales.

254 What is a Connector?

That seems deﬁnite enough. It is one of the rare occasions on which the Latin

grammarians declare independence from their Greek masters; and it would

be interesting to discover why they did so.

However that may be, the Latin story has only half been told. Having

cited Cominianus, Charisius turns to Remmius Palaemon, the ﬁrst and most

celebrated of the Roman grammarians. Charisius begins by remarking that

‘Palaemon deﬁnes them thus’ (inst ii xiv [i 225.5]). But he does not reproduce

a deﬁnition—or at any rate our text of Charisius does not. Instead, after a

brief remark about the position of connectors in sentences, Charisius suggests

that it is now time to discuss their potestates. He has, of course, already done

so in expounding the ﬁve species of Cominianus. But now he does so again,

this time distinguishing a dozen or more species.

The list—the claim of which to derive from Palaemon has been vigorously

asserted and vigorously denied—does not contain an item called ‘rational’.

But it does contain, as its third member, this:

Ratiocinative are these: quare, quapropter, igitur, ergo, itaque, quando (with a grave

accent), quatenus, quoniam, ideoque, quoniamquidem, quandoquidem, siquidem.They

are called ratiocinative because they conﬁrm by a reason whatever has already been

set down, thus:

It is light: therefore it is day.

For this has connected the reason: it is thereby light, because it is day—or it is

thereby day, because it is light.

(inst ii xiv [225.20–25])¹³⁸

The explanation for the use of the term ‘ratiocinative’ is monstrous (I shall

return to it); and the list of examples contains some odd items. But it looks

as though ratiocinative connectors include inferential connectors and exclude

co-assumptional connectors, and that they also include certain other items.

The question is complicated by the fact that the list of connectors in

which the ratiocinative class comes third also includes a class which Charisius

calls ‘inferential [illativae]’ ([226.3]). Their name suggests that they ought

to have something to do with the rational connectors—indeed, it suggests

rather strongly that they should be identiﬁed as inferential connectors.

Charisius offers no deﬁnition. But he does offer a list, namely: quamquam,

¹³⁸ ratiocinativae hae: quare, quapropter, igitur, ergo, itaque, quando (gravi accentu), quatenus,

quoniam, ideoque, quoniamquidem, quandoquidem, siquidem. dictae autem sunt ratiocinativae quod

quamque rem praepositam ratione conﬁrmant in hunc modum: lucet, igitur dies est. nam hic coniunxit

rationem, lucem ideo esse quod sit dies seu diem ideo esse quod sit lux.

Syllogistic Connectors 255

quamvis, etsi, tametsi. ‘Although’, ‘even if ’: such items are evidently not

inferential.

What is going on? Some light—not much—is shed by a comparison with

Diomedes’ treatment of the subject, which is in places very close to Charisius

and plainly was taken from the same source. Like Charisius, Diomedes

notices a class of inferential connectors which is distinct from the class of

ratiocinative connectors; he gives the same four examples as Charisius does;

and he ascribes the things to Pliny (ars gramm i 416.17–19), the reference

doubtless being to the younger Pliny’s work on linguistic problems.

So what was Pliny up to? His work is lost; but Priscian’s Institutions

suggests a solution to the problem. Priscian recognizes seventeen types of

connector. One type is called ‘adversative [adversativae]’. It is illustrated

by six examples—tamen, saltem, and the four items which Pliny allegedly

called inferential. (See inst xvi ii 10 [iii 99.12–100.4].) I bet that these

four, or six, examples were lifted from Pliny, and that Pliny offered them as

illustrations of adversative connectors—which, of course, is just what they

are. Pliny’s account, examples and all, was borrowed from him by various

grammar-teachers, and borrowed from them by their successors. Somewhere

and somewhen the group of examples strayed, or lost its name, and had

another and wholly inappropriate label attached to it.

How, it may be wondered, did the muddle come about? You might dream

up a sophisticated story in answer to that question; but the right answer is

this: The muddle was the result of a blundering error: someone—a careless

grammarian or his careless scribe—replaced ‘adversativae’by‘illativae’. The

interminable pages of Grammatici Latini arecrammedwithsplendidblunders:

it is misplaced charity to interpret them out of existence, and it is misplaced

ingenuity to elaborate reﬁned explanations for their presence in the texts.

However that may be, Priscian rightly calls his six items adversative.

Not that he avoids the term ‘inferential’. On the contrary, the inferential

connector is another one of his seventeen varieties:

Collective or rational connectors are ergo, igitur, itaque (when the antepenultimate is

acute), quin, alioquin, immo, utique, atqui. For these collect by an inference what was

earlier said—i.e. they conﬁrm it by reason. … the same connectors are also called

inferential because, when other items have been set down in advance, they are inferred.

(inst xvi ii 12 [iii 100.15–101.6])¹³⁹

¹³⁹ collectivae vel rationales sunt ergo igitur itaque (quando antepaenultima acuitur) quin alioquin

immo utique atqui. hae enim per illationem colligunt supra dictum, hoc est ratione conﬁrmant.

… dicuntur tamen eaedem illativae quod praepositis aliis inferuntur.

256 What is a Connector?

‘Collective’, ‘rational’ and ‘inferential’ determine the same class of connectors.

The word ‘collectivus’ is a Latin calque on ‘συλλογιστικός’, and perhaps

‘rationalis’ was intended as a translation of the same Greek word. So the

Latin rational connectors were meant to correspond to the Greek syllogistic

connectors.

What of the ratiocinative connectors? Priscian does not use the term; but

his explanation of his collective connectors overlaps with the explanation of

ratiocinative connectors which is found in Charisius and in Diomedes. That

suggests that ‘ratiocinative’ is another translation of ‘syllogistic’. It is not a bad

translation. Indeed, the received text of Charisius actually contains the word

‘συλλογιστικοί’ in apposition to ‘ratiocinativae’; and although I suppose that

the Greek word is a gloss, it is surely a true gloss.

There is a difﬁculty. According to Charisius and Diomedes, these connect-

ors ‘are called ratiocinative because they conﬁrm by a reason what has already

been set down’; and Priscian says the same of his collective connectors. That

is to say, you set down some item or items; you then adjoin another item; you

fasten on the adjoined item by means of a ratiocinative connector; and the

connector indicates that the adjoined item is a reason in favour of what goes

before it. If that description of ratiocinative connectors is correct, then they

are completely distinct from syllogistic connectors: a syllogistic connector

introduces either a supplementary premiss or else a conclusion; a ratiocinative

connector introduces an item which gives a reason, or supplies a premiss, for

a thesis which has antecedently been advanced.

But the description of ratiocinative connectors is followed by an example

which is supposed to illustrate to it. Here again—in Charisius’ words—is

the example:

It is light: therefore it is day. For this has connected the reason: it is thereby light,

because it is day—or it is thereby day, because it is light.

The connector ‘therefore’ does not illustrate the description, and it does

not ‘connect the reason’. Charisius—or whoever he is copying—noticed

that embarrassing fact and so replaced ‘therefore’ by ‘because’: ‘It is thereby

light, because it is day’. ‘Because’ ﬁts the description—but then Charisi-

us noticed that ‘It is light for that reason, because it is day’ is scarcely

an intelligent gloss on ‘It is light: therefore it is day’. He saw that some-

thing was wrong—and so he added a second paraphrase, ‘It is day for

that reason, because it is light’, as though it were more or less the same

thing.

Syllogistic Connectors 257

That is a sorry mess—another blunder for which no sophisticated elucid-

ation should be sought. In any event, logic was not the forte of the Latin

grammarians. Priscian, it is true, was on better form when he associated the

term ‘inferential’ with collective connectors. Nonetheless, ‘inferential’, on

Priscian’s deﬁnition, only applies to one half of his collective connectors; for

it is false that all collective or syllogistic connectors mark the conclusion of

an inference.

If Priscian’s ‘collectivus’translates‘συλλογιστικός’, what is the Greek for

‘illativus’? The answer must be: ‘ἐπιφορικός’. And in fact the term is known

from Apollonius. Part of his argument to the conclusion that expletive

connectorshaveasenserunslikethis:

Again, we can see that they have a sense by looking at what we call syllogistic and the

Stoics inferential connectors. ‘τοίνυν’ consists of two expletive connectors—so too,

together with ‘γάρ’, in ‘τοιγαρτοι’and,togetherwith‘οὖν’, in ‘τοιγαροῦν’. These

items have the same force as ‘ἄρα’ with a short alpha. They are called inferential

insofar as they are inferred from what has been premissed—

If it is day, it is light; but it is day: therefore it is light [φῶς ἄρα ἐστί, τοιγαροῦν

φῶς ἐστί, φῶς τοίνυν ἐστί].

They are called syllogistic inasmuch as, in certain proofs, when we syllogize the

conclusion we use these connectors:

You have ﬁve euros from me, and you’ve also got three: therefore you have eight

euros [ἔχεις ἄρα ὀκτὼ δραχμάς, ἔχεις τοίνυν ὀκτὼ δραχμάς].

(conj 251.27–252.8)¹⁴⁰

The argument is this: ‘τοίνυν’ certainly has a sense, for it is a syllogistic

connector. It is composed of two parts, neither of which can be cancelled

without changing the sense. Hence each of its parts has a sense. Hence the

expletive connectors ‘τοι’and‘νυν’haveasense.

Apollonius’ explanation of the term ‘inferential’ is the same as Priscian’s.

But Apollonius’ examples, unlike Priscian’s, ﬁt his explanation. (The same

four examples are found in the Dionysian list of syllogistic connectors.)

Again, Apollonius’ explanation of the term ‘syllogistic’ makes it equivalent to

¹⁴⁰ ἀλλὰ μὴν καὶ ἐν τοῖς καλουμένοις πρὸς ἡμῶν μὲν συλλογιστικοῖς, πρὸς δὲ τῶν Στωϊκῶν

ἐπιφορικοῖς ἔστι παραδέξασθαι τὴν σημασίαν αὐτῶν. τὸν τοίνυν ἐκ δύο παραπληρωματικῶν

συνεστῶτα, καὶ ἔτι μετὰ τοῦ γάρ ἐν τῷ τοιγάρτοι, καὶ μετὰ τοῦ οὖν τοιγαροῦν. δύναμιν

γὰρ ἔχουσιν οἱ τοιοῦτοι ἴσην τῷ ἄρα συστελλομένῳ κατὰ τὸ α. καὶ εἴρηνται μὲν ἐπιφορικοί,

καθὸ ἐπιφέρονται τοῖς λελημματισμένοις· <εἰ ἡμέρα ἐστί, φῶς ἐστι,> ἀλλὰ μὴν ἡμέρα ἐστί,

φῶς ἄρα ἐστί, τοιγαροῦν φῶς ἐστί, φῶς τοίνυν ἐστί· συλλογιστικοὶ δέ, καθότι ἐπί τισιν

ἀποδείξεσιν, ἐπισυλλογιζόμενοι τὸ συναγόμενον, προσχρώμεθα τοῖς συνδέσμοις τοῖσδε· ἔχεις

μου πέντε δραχμάς, ἔχεις δέ καὶ τρεῖς, ἔχεις ἄρα ὀκτὼ δραχμάς, ἔχεις τοίνυν ὀκτὼ δραχμάς.

258 What is a Connector?

‘inferential’ (for the reference to ‘certain proofs’ must not be taken to limit

syllogistic connectors to syllogisms which are in fact probative). In addition,

Apollonius—and he alone—tells us that the Stoics ﬁxed a class of inferential

connectors. That they should call them inferential or ἐπιφορικοί is to be

expected; for ‘ἐπιφορά’ was the usual term in Stoic logic for the conclusion

of an argument.

In short, according to Apollonius’ account of the matter, inferential and

syllogistic connectors are one and the same. (They were also, according

to Apollonius, sometimes called ‘epilogistic’; at any rate, ‘οὐκοῦν is called

epilogistic by some people’ (conj 257.18).¹⁴¹ Who those people were we

cannot tell; but it is difﬁcult to think that the difference in nomenclature had

any particular signiﬁcance.)

What about co-assumptional connectors? They too are named in Apol-

lonius’ Connectors:

We have explained how and with what force the connector ‘δέ ’ is understood. But

when it takes the connector ‘γέ’ it means something else. For the ‘γέ’in‘δέγε’

is not otiose, as it is in ‘ἆρά γε ἡμέρα;’. You can ﬁnd the Stoics calling ‘δέγε’

co-assumptional. For this construction of connectors introduces sayings which come

from conditionals and are reformulated—

If it is day, it is light; but [δέγε]itisday.

Since the saying comes to be in a co-assumption, such connectors are co-assumptional.

The same is true of ‘ἀλλά’and‘ἀλλαμήν’.

(conj 250.12–20)¹⁴²

Connectors which mark a co-assumption or πρόσληψις were called, by

the Stoics, co-assumptional or προσληπτικοί. (The two examples which

Apollonius appends to his account, ‘ἀλλά’and‘ἀλλαμήν’, are two of the

items in the Dionysian list of syllogistic connectors.)

And so we may tell the following history. The Stoics were the ﬁrst to

distinguish different classes of connectors, among them the inferential and the

co-assumptional. Apollonius followed them, noticing that the grammarians

generally preferred to use the word ‘syllogistic’ to designate the inferential

¹⁴¹ ὁ οὐκοῦν καλεῖται πρὸς ἐνίων ἐπιλογιστικός.

¹⁴² ἔτι ὁ δέ σύνδεσμος ὅπως παραλαμβάνεται καὶ ἐπὶ ποίᾳ δυνάμει ἐκτεθείμεθα. ἀλλὰ

προσλαβὼν τὸν γέ ἄλλο τι ἐπηγγείλατο. οὐ γὰρ ὡς ἐν τῷ ἆρά γε ἡμέρα; παρείλκετο ὁ

γέ καὶ ἐν τῷ δέ γε. καλούμενον γοῦν ἔστιν εὑρέσθαι παρὰ τοῖς Στωϊκοῖς τὸν δέ γε ὄντα

προσληπτικόν. τοὺς γὰρ ἀπὸ συναφῆς λόγους εἰς σχηματισμὸν μετιόντας ἡ τοιαύτη σύνταξις

ἡ τῶν συνδέσμων ὑπάγει· εἰ ἡμέρα ἐστί, φῶς ἐστιν· ἡμέρα δέγε ἐστιν. καὶ ἐπεὶ ἐν προσλήψει

ἐγένετο ὁ λόγος, προσληπτικοὶ οἱ τοιοῦτοι σύνδεσμοι. τὸ δ᾿ αὐτὸ συμβέβηκε καὶ ἐπὶ τοῦ ἀλλά

καὶ ἀλλαμήν.

Arguments and Sayings 259

connectors, and that some people used ‘epilogistic’. Later Greek grammarians

failed to see the point of distinguishing between the inferential and the

co-assumptional, and they joined the two into a single class. The members

of that class they called syllogistic, either changing the sense of the word

‘syllogistic’ or else failing to see that its old sense was no longer apt. The

Latin grammarians inherited the broad class of syllogistic connectors. On

the one hand, they implicitly enlarged it yet further, by stufﬁng it with new

and heterogeneous examples; and on the other hand, they established—quite

incoherently—a separate class of inferential connectors. They then spiced up

the soup with some piquant confusions of their own.

That is a sad story; and I fear that it is by and large true.

ARGUMENTS AND SAYINGS

Whatever the history of syllogistic connectors may have been, we may wonder

why they were counted as connectors in the ﬁrst place. Marius Victorinus, in

his thoroughly traditional Art of Grammar, follows Cominianus’ deﬁnition

of the connector and also his division of connectors into ﬁve species. But

when he gets to the rational connectors he has a word of his own to say:

What are the rational connectors? ita, itaque, proinde, proin, denique. These seem to

me rather to be adverbs.

(ars gramm vi 203.10–14)¹⁴³

Victorinus, for once, appears to have a point.

Consider such English words as ‘therefore’ and ‘thus’ and ‘so’. They are,

to be sure, sentential connectives in the generous modern use of that phrase;

that is to say, they are items which take sentences to make sentences. But

they are one-place connectives—and hence they are not connectors as the

ancients understood that notion. They are sentential adverbs or adsentences.

Moreover, they are demonstrative or indexical adsentences: ‘so’ means ‘for

that reason’, where ‘that’ is a demonstrative adjective, and the phrase ‘that

reason’ will (at least normally) refer back to something which has just been

set down or said. So in

I think. Therefore I am.

there are two sentences, not one. The ‘therefore’ does, of course, connect the

second sentence to the ﬁrst inasmuch as it contains a reference to what the

¹⁴³ rationales quae sunt? ita itaque proinde proin denique. quae magis adverbia esse mihi videntur.