DE INTERPRETATIONE 3 ON ISOLATED VERBS * FRANCESCO ADEMOLLO [In A. P. Mesquita and R. Santos (eds), New Essays on the Organon, Abingdon and New York 2024, 70–89. Penultimate draft.] ABSTRACT. In De interpretatione 3. 16b19–25 Aristotle discusses the signification of verbs uttered in isolation, i.e. not as parts of complete sentences. After providing some context, I offer a detailed analysis of these lines, which are very controversial, and of the views of the ancient commentators. I argue that, contrary to what is often thought, here Aristotle says nothing at all about the copula; in particular, he does not say that the copula is devoid of any intrinsic signification and is a mere link between the subject and the predicate term. Indeed, I argue that this is not even Porphyry’s interpretation of Aristotle, contrary to what many scholars believe. Rather, Aristotle takes as an especially relevant example the verb ‘to be’ in its existential use and claims that it – and by extension any other verb when uttered on its own – does not constitute a sentence until a subject term is added to it. After treating the name (ὄνοµα) in ch. 2 of the De interpretatione, in ch. 3 Aristotle proceeds to enquire into the verb (ῥῆµα). The chapter’s opening lines, 16b6–10, set forth a famous definition, which will not be our main object of study but will play an important background role. It will be helpful to quote these lines in full, as they are reported by most witnesses and printed in the editions of Waitz 1844 and Montanari 1988: Ῥῆµα δέ ἐστι τὸ προσσηµαῖνον χρόνον, οὗ µέρος οὐδὲν σηµαίνει χωρίς· καὶ ἔστιν ἀεὶ τῶν καθ᾿ ἑτέρου λεγοµένων σηµεῖον. λέγω δὲ ὅτι προσσηµαίνει χρόνον, οἷον ὑγίεια µὲν ὄνοµα, τὸ δ᾿ ὑγιαίνει ῥῆµα· προσσηµαίνει γὰρ τὸ νῦν ὑπάρχειν. καὶ ἀεὶ τῶν καθ᾿ ἑτέρου λεγοµένων σηµεῖόν ἐστιν, οἷον τῶν καθ᾿ ὑποκειµένου ἢ ἐν ὑποκειµένῳ. A verb is what additionally signifies time, no part of which is significant separately; and it is always a sign of the things said of something other. It additionally signifies time: e.g. ‘recovery’ is a name, but ‘recovers’ is a verb, because it additionally signifies the thing’s holding now. And it is always a sign of the things said of something other, i.e. of a subject or in a subject. 1 Aristotle is saying that a verb has the ordinary features of a name plus other, distinctive ones of its own. So, if we turn back to how names were characterized in ch. 2, 16a19–20, we can infer that a verb, like a name, is a ‘spoken sound significant by convention’. Indeed, Aristotle explicitly specifies here that a verb, again like a name, has no separately significant parts. But while a name was said to be * My thanks to an audience at the Centre Léon Robin in Paris for a stimulating discussion and to an anonymous referee for helpful suggestions. Throughout the paper I have freely modified Ackrill’s (1963) translation of the De interpretatione and Blank’s (1996) translation of Ammonius’ commentary on the De interpretatione. The manuscript sigla are those of Weidemann’s (2014b) edition. 1 Following Ackrill 1963, I use ‘recovery’ and ‘recover’ to render the Greek terms ὑγίεια and ὑγιαίνω (literally ‘health’ and ‘be healthy’) in the absence of an English verb corresponding to the latter. 1 significant ‘without time’ (ἄνευ χρόνου), a verb, in ‘addition’ to a name, also signifies time. Thus the verb ‘recovers’ not only signifies recovery, just as the name ‘recovery’ does, but also signifies there being some recovery now. Furthermore, a verb ‘is always a sign of the things said of something other’. That is to say, verbs have an essential connection with predication. There seem to be at least two aspects to this connection: on the one hand, a verb signifies an item of a nature to be predicated of something other, i.e. an attribute or property; on the other, the verb actually predicates such an item of something other.2 At the end of the passage Aristotle explains what he means by ‘things said of something other’. He does so with a phrase transmitted by almost all witnesses as ‘i.e. of a subject or in a subject’ (οἷον τῶν καθ᾿ ὑποκειµένου ἢ ἐν ὑποκειµένῳ, 16b10–11). This refers to the distinction advanced in Categories 2 between beings which ‘are said of a subject’, i.e. essential predicates of something, and beings which ‘are in a subject’, i.e. accidental predicates of something.3 So here the idea seems to be that verbs are always signs of something that gets predicated either essentially or accidentally of something other.4 In the ensuing lines, 16b11–18, Aristotle defines the ‘indefinite verb’ (ἀόριστον ῥῆµα), a term by which he refers to negated verbs such as οὐχ ὑγιαίνει, ‘doesn’t-recover’, and the ‘inflexion of verb’ (πτῶσις ῥήµατος), i.e. verbs in a tense different from the present. Then he concludes the chapter by adding a few final remarks about the signification of isolated verbs, i.e. verbs that are not parts of sentences. These remarks constitute the main subject of this paper. Here is again the text read by the majority of witnesses and printed by Waitz (1844) and Montanari (1984, 1988): αὐτὰ µὲν οὖν καθ᾿ αὑτὰ λεγόµενα τὰ ῥήµατα ὀνόµατά ἐστι καὶ σηµαίνει τι – ἵστησι γὰρ ὁ λέγων τὴν διάνοιαν, καὶ ὁ ἀκούσας ἠρέµησεν – ἀλλ᾿ εἰ ἔστιν ἢ µή οὔπω σηµαίνει· οὐδὲ γὰρ τὸ εἶναι ἢ µὴ εἶναι5 σηµεῖόν ἐστι τοῦ πράγµατος, οὐδ᾿ ἐὰν τὸ ὂν εἴπῃς αὐτὸ καθ᾿αὑτὸ ψιλόν.6 αὐτὸ µὲν γὰρ οὐδέν ἐστιν, προσσηµαίνει δὲ σύνθεσίν τινα, ἣν ἄνευ τῶν συγκειµένων οὐκ ἔστι νοῆσαι. (Int. 3. 16b19–25) 2 With what I have just written on verbs and predication cf. more fully Ademollo 2015: 49–50. See also Ackrill 1963: 118–9, Whitaker 1996: 58, Weidemann 2014a: 173–4, Frede (unpublished). 3 On beings which ‘are said of a subject’ and beings which ‘are in a subject’ see Ackrill 1963: 74–6, Wedin 2000: 38– 66. 4 Ammonius, in Int. 50.7–14 Busse, reports that Porphyry – presumably in his lost commentary on the De interpretatione – mentioned the existence of an alternative reading: ‘some’ did not read καὶ ἀεὶ τῶν καθ᾿ ἑτέρου λεγοµένων σηµεῖόν ἐστιν, οἷον τῶν καθ᾿ ὑποκειµένου ἢ ἐν ὑποκειµένῳ (‘and it is always a sign of the things which are said of something other, i.e. are said of a subject or in a subject’), but rather καὶ ἀεὶ τῶν ὑπαρχόντων σηµεῖόν ἐστιν, οἷον τῶν καθ᾿ ὑποκειµένου (‘and it is always a sign of the things which hold, i.e. of the things which hold of a subject’). The alternative reading was printed by Minio-Paluello as the text of his 1949 edition and then translated by Ackrill 1963; part of it, i.e. the omission of ἢ ἐν ὑποκειµένῳ, is also endorsed by Weidemann (2014a: 175–6, 2014). I believe that Weidemann’s hybrid text cannot be right, but I cannot discuss the issue here. 5 ἢ µὴ εἶναι is read by some witnesses in a different position, i.e. after τοῦ πράγµατος, as we are partly going to see later on. This is also the text printed by Weidemann 2014b. Nothing really important hangs on this issue. 6 αὐτὸ καθ᾿ αὑτὸ (or αὐτὸ καθ᾿ ἑαυτὸ) is the reading of the majority of witnesses, which Waitz (1844), Montanari (1984, 1988), and Weidemann (2014b) print in their editions. Others read just αὐτὸ or καθ᾿ αὑτὸ, or even nothing at all; this last reading is endorsed by Minio-Paluello (1949), who prints ψιλόν unaccompanied, presumably regarding the expansions αὐτὸ καθ᾿ αὑτὸ / αὐτὸ / καθ᾿ αὑτὸ as glosses. 2 When uttered just by themselves verbs are names and signify something – for the speaker arrests the thought and the hearer pauses – but they do not yet signify whether it is or not. For not even ‘to be’ or ‘not to be’ is a sign of the object – not even if you say ‘what is’ just by itself. For by itself it is nothing, but it additionally signifies some combination, which cannot be thought of without the components. My footnotes to the Greek text account for minor textual uncertainties; more substantial textual worries will be raised in due course. Let us now examine the various aspects of this difficult passage one by one. 16b19–21. ‘When uttered just by themselves verbs are names and signify something – for the speaker arrests the thought and the hearer pauses …’ Aristotle here starts out by making the point that verbs uttered in isolation, i.e. not as parts of a complete sentence, still ‘signify something’ (σηµαίνει τι), i.e. retain at least their basic lexical signification, the one which they share with names, as he implied at the beginning of the chapter, and in virtue of which e.g. the verb ‘recovers’ signifies recovery.7 He supports this claim with the remark that ‘the speaker arrests the thought and the hearer pauses’. It is not immediately clear whether the thought which the speaker is said to ‘arrest’ is his own or the hearer’s.8 It seems obvious to me, however, that the reason why we can say that a verb – or for that matter any other kind of word – signifies something is that the thoughts of both speaker and hearer come to focus on the same item. As on several other occasions in the first chapters of the De interpretatione, Aristotle probably has Plato’s Cratylus in mind, more precisely 434e–435a, where Socrates describes in the following terms what goes on when a speaker utters a name so as to indicate something to a hearer: ‘when I utter this, I think of that, and you recognize that I think of that’ (ἐγώ, ὅταν τοῦτο φθέγγωµαι, διανοοῦµαι ἐκεῖνο, σὺ δὲ γιγνώσκεις ὅτι ἐκεῖνο διανοοῦµαι).9 Actually, Aristotle does not say only that verbs – even when uttered in isolation – ‘signify something’; what he says is more precisely that they ‘are names and signify something’ (16b20). As Ammonius saw (in Int. 45.15–23, 54.23–5 Busse), here ‘and’ is explanatory, ‘i.e.’: Aristotle is recognizing a generic sense of ‘name’ which is equivalent to ‘significant expression’ and is pointing out that according to this generic sense of the term verbs too are one kind of ‘names’, along with names or nouns in the specific sense. The same generic use of ‘name’ occurs in Plato, Sophist 261d– 262a, where the Eleatic Stranger introduces name and verb as two different kinds of ‘names’ or ‘vocal means to indicate being’. 7 Of course the verbs uttered in isolation are meant to include finite verbal forms such as ‘runs’ or ‘sleeps’. Bärthlein (1984: 237–40, 244) argues that Aristotle is speaking only of infinitives, but this supposition is not backed by any substantial argument and indeed is incompatible with the passage’s overall train of thought. 8 See Ammonius, in Int. 54.25–55.10 Busse; Boethius, in Int. ii. 72.11–74.33 Meiser; Montanari 1988: 242–8; Weidemann 2014b: 179–80. 9 On the Cratylus passage see Ademollo (2022). The allusion is confirmed by the fact that the Aristotelian sentence contains, at the same time, yet another echo of the Cratylus, namely of the etymology, advanced at 437a, of ἐπιστήµη (‘knowledge’) as that which ἵστησιν ἡµῶν ἐπὶ τοῖς πράγµασι τὴν ψυχήν (‘arrests our soul at the objects’). See Weidemann 2014a: 179. 3 Some scholars take a different view, according to which no generic use of ‘name’ is in play in our lines: Aristotle’s point is rather that an isolated verb is bereft of its signification of time and predication and preserves only its basic, lexical signification, thus being reduced to the status of a mere name (in the specific sense of the term).10 I find this implausible. Obviously, if you utter the word ‘recovers’ on its own, your utterance is not equivalent to an utterance of the word ‘recovery’. It does already signify time (we can grasp the difference between ‘recovers’, ‘recovered’ and ‘will recover’ even outside the context of a sentence); and even though it does not actually predicate recovery of any subject, still it calls for the addition of a subject and manifests the capacity to predicate recovery of it as soon as it is added. Its predicative function is, so to speak, merely inchoate, not fully carried out, but not thereby absent. 16b21–2. ‘… but they do not yet signify whether it is or not.’ Here Aristotle contrasts what verbs uttered in isolation do – i.e. signify something – with what they do not: they ‘do not yet signify whether it is or not’ (εἰ ἔστιν ἢ µή οὔπω σηµαίνει). His general point is clear: a single verb lacks the complexity of a complete declarative sentence, because it is only a part of it, and hence has no truth value. He has already made the same claim about both names and verbs in ch. 1, 16a13–18, and will return to it in ch. 4, 16b26–30, and ch. 5, 17a9–12. The parallel in ch. 4 is especially close:11 ἄνθρωπος σηµαίνει τι, ἀλλ᾿ οὐχ ὅτι ἔστιν ἢ οὐκ ἔστιν. ‘human’ signifies something, but not that it is or is not. There has been some uncertainty among commentators as to how exactly this general point is formulated in our passage – i.e. how ‘whether it is or not’ is to be construed. The most natural thing to do, however, both here and in 4. 16b28–9, is suppose that the subject is the previous ‘something’ (τι) and the verb is existential: the verb signifies a certain predicate, but does not signify that that predicate is or is not, i.e. that it holds or does not hold of something, 12 unless a subject term is added.13 10 Whitaker 1996: 55–8. Contrast Ackrill (1963: 121): ‘Aristotle must be using “name” here in its wide, non-technical sense; he explains what he means by it by adding “and signifies something”. He is not saying that “runs” on its own is a name and not a verb, but he is bringing out that “runs” needs a subject if it is to perform the assertive role for which it is cast.’ 11 See Ax 1979: 273. 12 I assume that for a predicate F to ‘be’ (i.e. to exist) is for F to hold of something (i.e. for something to be F). Cf. the authors mentioned in n. 13; and see Mignucci (2007: 249): ‘gli universali per Aristotele non hanno esistenza autonoma. In realtà quello che si domanda quando si chiede “esiste X?” è se esistano le cose che sono X’; Ademollo (2022). 13 See Boethius, in Int. i. 64.26–7; Ackrill (1963: 121–2): the verb ‘runs’ ‘by itself does signify something, running, but not that that thing is, i.e. not that there is any running; only if you add a name (“Socrates runs”) will you be saying that there is some running’; Bärthlein 1984: 240; Weidemann 2014a: 180. Ackrill rightly rejects an alternative interpretation: ‘It is tempting to translate the last words of the sentence as “whether anything is or is not the case”; and similarly at 16b29 … This gives the correct point but is probably an incorrect translation.’ A similar interpretation was adopted by Ammonius, in Int. 55.11–16 Busse; the Anonymous Commentator, 10.8–10 Tarán; and Stephanus, 13.30–1 Hayduck. 4 16b22. ‘For not even ‘to be’ or ‘not to be’ is a sign of the object …’ Then Aristotle goes on to supply an argument for his claim that verbs on their own do not signify ‘whether it is or not’. In this section I focus on the first clause of this and set aside until later the second one, ‘not even if you say “what is” just by itself’. The first clause is transmitted by most witnesses in the form in which I have quoted it above: οὐδὲ γὰρ τὸ εἶναι ἢ µὴ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος, for not even ‘to be’ or ‘not to be’ is a sign of the object. But in due course we shall have to deal with a textual variant, read by some indirect sources, involving among other things οὐ (‘not’) in place of οὐδὲ (‘not even’). The basic structure of the argument seems to be clear, at least on the face of it: Aristotle focuses on a particular verb – ‘to be’ (εἶναι), along with its negation ‘not to be’ (µὴ εἶναι) – and denies that it is ‘a sign of the object’. This poses at least two questions, partly interrelated with each other. (i) How is ‘to be’ used here? (ii) What is the ‘object’ (πρᾶγµα) which ‘to be’ – like any other verb taken on its own – fails to signify? The best way to proceed now is to keep these questions in mind and start by examining how Ammonius first attempts to understand our clause. Here and in what follows, I report direct quotations of Aristotle’s text in spaced italics: And he adds this by way of a syllogism: ‘for not even “to be” or “not to be” is a sign of the object’. This is an argument a fortiori that verbs do not admit truth and falsehood [κατασκευὴ ἀπὸ τοῦ µᾶλλον τοῦ µὴ δέχεσθαι τὰ ῥήµατα τὸ ἀληθὲς καὶ τὸ ψεῦδος]. For if the most primitive and general of verbs, and those into which all the others are analysed [τὰ ἀρχοειδέστατα καὶ κοινότατα τῶν ῥηµάτων καὶ εἰς ἃ πάντα τὰ ἄλλα ἀναλύεται], because they signify immediately holding or not holding itself, are not true or false when said by themselves, then clearly the other verbs would all the more fail to admit of one of these properties. But the first;14 therefore the second. He assumes that of all verbs ‘is’ and ‘is not’, which he calls ‘to be’ and ‘not to be’, are the most primitive, in so far as each verb could be analysed into a participle plus one of these [ὡς ἂν ἑκάστου τῶν ῥηµάτων εἴς τε µετοχὴν ἀναλυοµένου καὶ θάτερον τούτων] – definite verbs into ‘is’, indefinite ones into ‘is not’: e.g. ‘runs’ – ‘is a running item’ [τρέχει – τρέχων ἐστίν], ‘recovers’ – ‘is a recovering item’ [ὑγιαίνει – ὑγιαίνων ἐστίν], ‘doesn’t-run’ – ‘isn’t a running item’, ‘doesn’t recover’ – ‘isn’t a recovering item’. So if these verbs are such and therefore signify nothing true or false by themselves, then how could it make sense for those which are posterior to these, and signify holding or not holding entirely by their participation in these, to indicate anything true or false? [πῶς ἂν ἔχοι λόγον τὰ ὕστερα τούτων καὶ τὸ ὑπάρχειν ὅλως ἢ µὴ 14 ἀλλὰ µὴν εἰ τὸ πρῶτον, καὶ τὸ δεύτερον ἄρα (55.22–3): like Blank 1996 Ι have not translated εἰ, which should I think be deleted. Ammonius has announced a syllogism (55.16); he has stated a conditional sentence (55.19–22 εἰ γὰρ … οὐκ ἂν δέχοιντο τούτων) which constitutes the major premiss of a hypothetical syllogism; here we have respectively the second premiss (ἀλλὰ µὴν {εἰ} τὸ πρῶτον) and the conclusion (καὶ τὸ δεύτερον ἄρα) of that syllogism, in accordance with standard (originally Stoic) logical terminology. 5 ὑπάρχειν κατὰ τὴν τούτων µετουσίαν σηµαίνοντα δηλοῦν τι ἀληθὲς ἢ ψεῦδος;] (in Int. 55.16–30 Busse) So Ammonius construes ‘to be’ as the copula; I shall call this the Copula Interpretation of our lines. ‘To be’ and its negation ‘not to be’ are, Ammonius says, ‘the most primitive and most general of verbs’, insofar as any other verb V can be analysed into a phrase composed of the copula + the participle of V: τρέχει, ‘runs’, is equivalent to τρέχων ἐστίν, ‘is a running item’ or ‘is an item which runs’. This equivalence is indeed stated by Aristotle in several passages (Int. 12. 21b9–10, An. Pr. 1. 46. 51b13–16, Metaph. Δ 7. 1017a27–30). Here, according to Ammonius, it becomes the basis for an argument a fortiori concerning all verbs.15 Two objections can be raised against this interpretation at this stage. (i) First, strictly speaking it does not really seem to provide Aristotle with an argument a fortiori. The argument is supposed to be that, since the copula is not equivalent to a complete sentence, then verbs in general are not equivalent to complete sentences. But the rationale which Ammonius reconstructs behind this is not that the copula is different from the other verbs because for some reason it is more sentence-like than they are, so that its failure to qualify as equivalent to a complete sentence entails, a fortiori, that no other verb can qualify. Ammonius’ point is rather that the copula is different because it is implicitly contained in any other verb, so that its failure to qualify as equivalent to a complete sentence is the direct reason why no other verb qualifies. In this respect he had better read not οὐδὲ (‘not even’) but rather οὐ (‘not’) – an alternative reading found in some indirect sources, as I have already anticipated and as we shall shortly see in more detail. Indeed, some modern interpreters adopt this modified version of Ammonius’ view.16 (ii) Secondly, and more importantly, whatever of the two variant readings you accept, the Copula Interpretation is unnatural and far-fetched. The context contains no reference whatsoever to the specific doctrine of the equivalence between verb and copula + participle; no ordinary reader would be able to detect its presence here.17 Indeed, since the beginning of the treatise we have encountered no reference at all to ‘to be’ as copula. The verb has only figured in its existential use, in which it has served as a stand-in for any verb – and in contexts very similar to our present one, whose point was to stress the difference between individual names or verbs and complete sentences. The first such passage was 1. 16a13–18: Names and verbs by themselves [αὐτά] are like thoughts without combination and separation, e.g. ‘human’ or ‘white’, when nothing further is added; for they are not yet true or false, but they are a sign of something determinate.18 For even ‘goat-stag’ signifies 15 With Ammonius cf. Stephanus, in Int. 13.33–6 Hayduck, and the Anonymous, in Int. 10.14–11.5 Tarán. The views of these ancient commentators are endorsed by Bärthlein 1984: 246–7. 16 This possibility is considered, and then rejected, by Ackrill 1963: 122 (cf. next note). It is endorsed by Whitaker 1996: 56–8. 17 See Ackrill 1963: 122; cf. Montanari 1988: 262. 18 At 16a15–16 I am following the punctuation οὔτε γὰρ ψεῦδος οὔτε ἀληθές πω, σηµεῖον δ᾿ ἐστὶ τοῦδε, proposed by Sedley (1996: 93; 2004: 14–15) and adopted by Whitaker (1996: 33–4) and Weidemann 2014b. 6 something [σηµαίνει µέν τι] but is not yet true or false, unless ‘to be’ or ‘not to be’ [τὸ εἶναι ἢ µὴ εἶναι] is added, either simply or with reference to time. Then came 2. 16b1–5, where Aristotle pointed out that ‘inflexions of names’ like ‘Philo’s’ or ‘to-Philo’ (Φίλωνος, Φίλωνι) differ from names in that an inflexion when combined with ‘is’, ‘was’, or ‘will be’ is not true or false [µετὰ τοῦ ἔστιν ἢ ἦν ἢ ἔσται οὐκ ἀληθεύει ἢ ψεύδεται], whereas a name always is. Take, for example, ‘Philo’s is’ or ‘Philo’s is not’ [Φίλωνός ἐστιν ἢ οὐκ ἔστιν]: so far there is nothing either true or false. Then comes ch. 3, with our passage. But if we read on we find other examples: 4. 16b28–9, which we encountered above (‘human’ signifies something, ‘but not that it is or is not’), and 5. 17a11–12, which is again similar (the definition of human is not yet a declarative sentence unless you add ‘is’, or ‘will be’, or ‘was’, or something like that). Briefly, each of chapters 1, 2, 4, 5 insists on the difference between terms and sentences and does so by recourse to examples involving the existential ‘to be’. Copula sentences only become a subject matter in ch. 7, with the analysis of the various kinds of quantified and non-quantified sentence and their oppositions, and then are explicitly theorized in ch. 10. There Aristotle, after dealing with sentences with existential ‘to be’ such as ‘A human is’ / ‘A human is not’ (ἔστιν ἄνθρωπος / οὐκ ἔστιν ἄνθρωπος), identifies a distinct class of sentences in which ‘is’ ‘is additionally predicated as third’ (19b19–20), such as ‘A human is just’ / ‘A human is not just’ (ἔστι δίκαιος ἄνθρωπος / οὐκ ἔστι δίκαιος ἄνθρωπος). Against this background the probability that in our ch. 3, 16b22–3, ‘to be’ and ‘not to be’ are cryptic references to a very specific doctrine about the copula turns out to be minimal, whereas the probability that we have just another occurrence of the existential ‘to be’ is overwhelming. I shall call this the Existential Interpretation of our lines.19 A cautionary note before we move on. It is often claimed that Aristotle, like Plato, sees a close connection between the existential ‘X is’ and the copulative ‘X is F’. On a fairly prudent version of this view, ‘X is’ is equivalent (at least logically) to ‘For some Φ, X is Φ’ for certain values of Φ, i.e. those which are essential to X.20 Here I am not challenging this kind of account; my point is just that, whatever continuity Aristotle may want to posit between the two uses of ‘to be’, the De interpretatione also draws, both in theory and in practice, a distinction (at least a syntactic one) between them. The Existential Interpretation fits well with ‘not even’ (οὐδέ). Aristotle has just said that a verb by itself signifies something but not whether it is (i.e. exists) or not. But surely – someone might think – the very verb ‘is’, or ‘is not’, does signify whether the item signified is or not? Of course not, replies Aristotle: not even ‘to be’ or ‘not to be’ (i.e. ‘is’ or ‘is not’) can do so. As Ax (1979: 274–7) pointed out, this is actually parallel to the goat-stag argument in ch. 1, 16a13–18, which we read above. There too Aristotle had just claimed that names and verbs are unlike complete sentences and are neither true nor false; and then he proceeded to prevent the possible objection that empty names like ‘goat-stag’ should count as false names: ‘For even “goat-stag” signifies something but is not yet true or false’. Thus 16a16 ‘for even’ (καὶ γάρ) and 16b22 ‘for not even’ (οὐδὲ γάρ) are parallel to one another: both 19 20 See Ackrill 1963: 122–3, Barnes 1996: 189. For some discussion see Brown 1994 and Charles 2002. 7 introduce a clause which establishes the terms/sentences distinction a fortiori by showing that ‘even’ a special case, which might be thought to constitute a counterexample, does not in fact do so.21 Our next question is what Aristotle means when, in 16b22–3, he claims that ‘to be’ or ‘not to be’ is not a sign ‘of the object’ (τοῦ πράγµατος). Could the ‘object’ be the predicate signified by the verb? This seems to be the view of Ammonius, 56.2–10 Busse. If so, then Aristotle would be repeating what he said at 16b20–1: a verb signifies something, but whether or not that something exists depends on the addition of a subject. Here, however, this will not do: ‘is a sign of the object’ cannot possibly mean ‘is a sign that the object exists’, and Aristotle should be eager to keep the two formulations as distinct as possible.22 An alternative construal might take inspiration from some relevant contexts in which the term ‘object’ (πρᾶγµα) is apparently used by Aristotle to refer to a propositional content or state of affairs.23 The Copula Interpretation invites, and the Existential Interpretation allows, such a construal here too: the copula (according to the Copula Interpretation), or the existential ‘to be’ (according to the Existential Interpretation), is not a sign of a complete state of affairs, unlike a fullyfledged sentence.24 This construal might seem to be confronted with the difficulty that Aristotle says not ‘of an object’ but ‘of the object’. Why the definite article?25 The difficulty could be solved with a costless amendment: we could remove the accent from the article τοῦ in order to get the indefinite του, thus writing του πράγµατος, ‘of some object’.26 However, the Existential Interpretation (unlike the Copula Interpretation) is also compatible with another, even easier construal: the ‘object’ might be the missing extra-linguistic subject, which in a complete sentence is signified by the subject term and which a verb uttered in isolation is unable to signify. At 12. 21b28 Aristotle claims that human and white (ἄνθρωπος and λευκός) are ‘the underlying objects’ (τὰ ὑποκείµενα πράγµατα) in a copulative sentence such as ‘A human is white’. Therefore in a sentence in which ‘to be’ is not copula but existential, such as ‘A human is’, there is presumably only one ‘underlying object’, namely the 21 Notice, however, that the Existential Interpretation would be no less compatible with a text which (as, we shall shortly see, in some indirect sources) at 16b22 read οὐ (‘not’) in place of οὐδὲ (‘not even’). All we would have to suppose is that Aristotle was presenting the particular case of ‘to be’ and ‘not to be’, not as the basis for an argument a fortiori (as with the reading οὐδὲ), but simply as an example which stands in for any other verb. In fact this is just what he does in many passages of the De interpretatione, including those which I cited above as evidence for his use of the existential ‘to be’ and ‘not to be’ in chs. 1, 2, 4, and 5: 16a13–18, 16b1–5, 16b28–9, 17a11–12. See further n. 29. 22 Cf. Weidemann 2014a: 182. 23 πρᾶγµα as state of affairs: see Cat. 5. 4b8–10, 12. 14b14–22; Metaph. Δ 29. 1024b18–21 (probably the most uncontroversial instance), Θ 10. 1051a34–b9. See Crivelli 2004: 46–62, 2015: 193–202. 24 See Anon. in Int. 10.14–15 Tarán ‘The verb is not capable of indicating a true or false object’ (οὐκ ἔστι τὸ ῥῆµα δηλωτικὸν ἀληθοῦς καὶ ψευδοῦς πράγµατος), Ackrill 1963: 122, Nuchelmans 1973: 33–4, Bärthlein 1984: 248. 25 Cf. Ax (1979: 279) and Whitaker (1996: 56), who translate τοῦ πράγµατος as ‘einer Sache’, ‘of a thing’. Whitaker’s own interpretation of τοῦ πράγµατος is that it is a ‘thing’ which the copula fails to signify, because it ‘does not in fact signify anything’ (cf. Boethius, in Int. ii. 76.10–15 Meiser). If this were the meaning of the present clause, I fail to see how it could fulfil the function, which the ‘for’ (γάρ) assigns to it, of explaining the previous claim that verbs uttered by themselves ‘signify something … but they do not yet signify whether it is or not’. 26 De Rijk (2002: 217 n. 101) dismisses this solution as ‘too bold’. One wonders what he would have said of a real conjecture. 8 human; and once you have ‘is’ you need only one ‘object’ – hence the ‘object’ – to get a complete sentence. Enter Porphyry We are now ready for Porphyry’s entrance. I shall quote the relevant passage from Ammonius’ commentary: If the text is as we have set it out, ‘For not even “to be” or “not to be” is a sign of the object’ [οὐδὲ γὰρ τὸ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι], you will find that only the interpretation I have expounded supports it. If, instead, it is as Porphyry the philosopher writes, ‘For “to be” is not a sign of the object, nor is “not to be” ’ [οὐ γὰρ τὸ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι] – although as he goes on he is led back to the former reading and interpretation – then the text would say in general about all verbs that they are significant of something, as has been said, but not of truth or falsehood, which Aristotle shows by the words ‘but they do not yet signify whether it is or not’, and he would be giving the reason for this by the words ‘For “to be” is not a sign of the object, nor is “not to be” ’. This means: ‘For the verb, said by itself, is not significant of the object indicated by it holding or not holding’ [οὐ γάρ ἐστι σηµαντικὸν τὸ ῥῆµα καθ’ ἑαυτὸ λεγόµενον τοῦ ὑπάρχειν ἢ µὴ ὑπάρχειν τὸ ὑπ’ αὐτοῦ δηλούµενον πρᾶγµα]; only if it did so would it be receptive of falsehood and truth. For he who has said ‘walks’ has signified some activity, but has not said anything true or false about it, unless some subject is added, by holding or not holding of which the walking will make a true or false sentence. Therefore ‘For “to be” is not a sign of the object, nor is “not to be” ’ [οὐ γὰρ τὸ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι] is equivalent to saying that the verb, said by itself, is not significant either of the object (i.e. the one signified by it) being [τὸ ῥῆµα καθ’ ἑαυτὸ λεγόµενον οὐκ ἐστι σηµαντικὸν οὔτε τοῦ εἶναι τὸ πρᾶγµα, τοῦτ᾿ ἔστι τὸ ὑπ᾿ αὐτοῦ σηµαινόµενον] – which is usually signified by the affirmation – or of it not being [οὔτε τοῦ µὴ εἶναι] – which is indicated by means of the denial. (56.14–32 Busse = Porph. 88F. Smith) Let us set out schematically the main points of Ammonius’ account as it emerges from this text. • • • Porphyry knew of another reading (henceforth the ‘Porphyrean Reading’): οὐ γὰρ τὸ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι, ‘For “to be” is not a sign of the object, nor is “not to be” ’ (lines 16– 17). Nevertheless, in the course of his commentary he ended up endorsing the standard reading and the interpretation which Ammonius has just set forth (18). On the interpretation initially proposed by Porphyry (henceforth the ‘Porphyrean Interpretation’), Aristotle is not speaking specifically of ‘to be’ but of all verbs in general (19). He has just said that verbs signify something, but not truth or falsehood (19–21); he now proceeds to give a reason for this claim (21–2). It goes as follows: a verb, said on its own, ‘is not significant either of the object (i.e. the one signified by it) being – which is usually signified by the affirmation – or of it not being – which is indicated by means of the denial’ (28–32, cf. 22–28). 9 This is quite interesting. Unfortunately it is also inconsistent; for the quotation of the Porphyrean Reading is incompatible with the account of the Porphyrean Interpretation. The Reading as quoted seems to differ from the one we have been discussing so far essentially in that it has οὐ (‘not’) in place of οὐδὲ (‘not even’); the different placement of ἢ µὴ εἶναι, ‘or “not to be” ’, is irrelevant. According to this text τὸ εἶναι ... ἢ µὴ εἶναι, in the nominative, is the subject of οὐ ... σηµεῖόν ἐστι τοῦ πράγµατος: ‘to be’ or ‘not to be’ is not ‘a sign of the object’ (whatever that may mean). In Ammonius’ account of the Interpretation, instead, ‘the verb said by itself’ is the subject of ‘is not significant’, and this in its turn governs οὔτε τοῦ εἶναι τὸ πρᾶγµα ... οὔτε τοῦ µὴ εἶναι (56.30–2, cf. 23–4), which I have translated ‘either of the object … being … or of it not being’. Here we have τοῦ εἶναι ... τοῦ µὴ εἶναι, where the articular infinitives are in the genitive and their subject is τὸ πρᾶγµα, ‘the object’. Thereby, as Ammonius says at line 19, Aristotle turns out to be making a claim about all verbs in general, not just about ‘to be’. But this meaning cannot possibly be expressed by the reading of which it is allegedly an interpretation. Busse (1897) saw this difficulty and tried to bring the Reading into line with the Interpretation. He did so by suggesting, in the apparatus, that perhaps τὸ εἶναι at lines 17, 22, 29 ought to be emended to τοῦ εἶναι: ‘scribas τοῦ εἶναι, quod Porphyrius legisse videtur’. We might explain this multiple corruption by supposing that the text of the quotations in Ammonius’ commentary was influenced by the vulgate text. If this is so, then the Porphyrean Reading is not just οὐ γὰρ τὸ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι, for ‘to be’ is not a sign of the object, nor is ‘not to be’, as we might think on the basis of Minio-Paluello’s (1949) apparatus and as many modern scholars believe, but rather οὐ γὰρ τοῦ εἶναι σηµεῖόν ἐστι τοῦ πράγµατος ἢ µὴ εἶναι, for they [sc. verbs uttered just by themselves] are not a sign of the being of the object or of its not being. This reconstruction finds some support in Boethius (in Int. i. 64.13–65.8, ii. 76.18–77.1, 78.18–19 Meiser), who used Porphyry as a source and seems to read the same text, and in George’s Syriac translation, Γ.27 It is endorsed by Weidemann (2014a: 182; 2014b) as correct, and it may actually be 27 See the apparatus of Weidemann 2014b. Note, however, that in in Int. ii. 76.10–15 Meiser Boethius clearly presupposes the vulgate text in which τὸ εἶναι ἢ µὴ εἶναι is the grammatical subject of σηµεῖόν ἐστι τοῦ πράγµατος. Note also that the transition to the lines in which he seems to presuppose the Porphyrean Reading is neither clear nor explicit. If anything, Boethius seems to be contrasting two different interpretations of the text (see 76.26–7 hic est melior intellectus), but shows no awareness that two different readings are involved. Cf. Blank 1996: 152–3 n. 217, who resists Busse’s correction: Boethius ‘gives no indication that Porphyry’s text of Aristotle had a different reading. Instead, he thinks that the same text (i.e. to einai) needs to be interpreted as though it were the genitive’. This, however, is to ascribe to Porphyry and Boethius an utterly ungrammatical construal of Aristotle’s Greek. 10 right.28 This means that the textual evidence for the variant οὐ (‘not’) alone, not accompanied by τοῦ εἶναι (‘of the being’), becomes slender: the Armenian translation, Δ; the other Syriac translation, Σ; and the Anonymous Commentator, τ.29 But Weidemann goes further: he rates the Porphyrean Reading higher than did Porphyry himself (who in the end rejected it) and accepts it as the correct one in the text of Aristotle. Here I disagree, for several reasons. (i) Aristotle would not normally write τοῦ εἶναι ... τοῦ πράγµατος, ‘of the being of the object’, but τοῦ εἶναι ... τὸ πρᾶγµα, literally ‘of the object being’. Actually, this is exactly how Ammonius paraphrases the Porphyrean Reading in the passage we have just read, 56.30–2 (cf. 23–4) Busse. The corpus contains a couple of parallels for the phrase τὸ εἶναι τοῦ X with the meaning ‘the existence of X’;30 but it is unclear whether they are perfectly pertinent, and in any case the idiom is extremely unusual. (ii) The Porphyrean Reading involves a switch in grammatical number: Aristotle was speaking in the plural of verbs spoken by themselves, but now he would suddenly start to talk in the singular about their failing to be ‘a sign of the being of the object’, without supplying a new subject term. The switch is not impossible; but it is uncomfortable.31 (iii) Finally, with the Porphyrean Reading the only reference to the verb ‘to be’ throughout the passage would be contained in the ensuing clause ‘not even if you say “that which is” just by itself’ (οὐδ᾿ ἐὰν τὸ ὂν εἴπῃς αὐτὸ καθ᾿ αὑτό ψιλόν, 16b23). This is odd; for that clause is naturally read as introducing the participle of ‘to be’ in order to reinforce some previous point about the verb. In the light of these difficulties,32 what we should do with the Porphyrean Reading is what Porphyry did, namely dismiss it. Let us move on. 28 Cf. Montanari 1984: 184–5, 1988: 256–8. Montanari, however, doubts that the Porphyrean Reading was a genuine reading at all, and would rather regard it as a conjecture. I suspend judgement. 29 In the previous section, n. 21, I pointed out that the Existential Interpretation, which I defended there, is compatible both with the reading οὐδὲ and with the reading οὐ. Now that οὐ on its own turns out to be poorly attested, however, it is probably safest for supporters of the Existential Interpretation to leave it aside and hold on to the mainstream οὐδὲ. 30 The parallels are cited by Weidemann 2014a: 184. One is EN 9. 9. 1170b7–17, where Aristotle seems to shift from τὸ τὸν φίλον (sc. εἶναι), to τὸ τοῦ φίλου. Another is Ph. 4. 12, where Aristotle repeatedly describes time as measuring τὸ εἶναι of someting (e.g. 221a5–6 τὴν κίνησιν καὶ τὸ εἶναι τῆς κινήσεως, ‘both motion and the being of motion’; cf. 221a7, a9, b27, b30–1) and also claims that something ‘is in number’ if and only if there is a number τοῦ πράγµατος, ‘of the object’, and τὸ εἶναι αὐτοῦ, ‘its being’, is measured by number (221b14–16). See Delcomminette (2020) for discussion of the phrase in Ph. 4. 12. 31 See Bärthlein 1984: 235–6, 243. Weidemann (2014a: 184–5) dismisses this objection on the grounds that in 16 b 19– 22 ‘ein Numeruswechsel stattzufinden scheint’. 32 Bärthlein (1984: 234, 243) and Montanari (1988: 259) also argue that the Porphyrean Reading would add nothing new to the previous ‘but they do not yet signify whether it is or not’ (ἀλλ᾿ εἰ ἔστιν ἢ µὴ οὔπω σηµαίνει). This is right if the ‘object’ is meant to be the entity signified by the verb; from this perspective I am not reassured by Weidemann’s (2014a: 183) claim that ‘for they … the object’ is meant to explain the previous clause rather than give grounds for it. But there would instead be no repetition if one or other of the two alternative construals of the ‘object’ were right, i.e. if were rather meant to be a state of affairs or the extralinguistic subject (see above, section ad 16b22). Therefore the argument is not decisive. 11 16b23. ‘… not even if you say “what is” just by itself’. We now come to this clause, which we initially set aside, but which has just made an appearance in our discussion of the Porphyrean Reading. Several interpretations of this, from Boethius onwards, start from an assumption that I reject from the very start, i.e. that ‘what is’ (τὸ ὄν) here may be a way of referring to ‘is’ (ἔστι), whether in general or as the copula.33 This is linguistically impossible, and there is no reason why Aristotle should resort to such tortured Greek here. Rather, what we have is another argument a fortiori, which builds upon the previous one. ‘If the trouble with “to be” was that it lacked a subject term,’ someone might ask Aristotle, ‘then perhaps “what is” – still a form of “to be”, but a nominal one – might allow us to eat our cake and have it?’34 Of course not, answers Aristotle: not even this would do. A more precise reply would be that ‘that which is’ is actually more of a name than a verb from our present point of view;35 but never mind. 16b24–5. ‘For by itself it is nothing, but it additionally signifies some combination, which cannot be thought of without the components.’ We now come to the passage’s final sentence. The way in which you read it is necessarily oriented, to some extent, by the way in which you read the previous sentences; at the same time it should also constitute a final testing ground for those previous hypotheses. On a widespread interpretation, at least in this sentence, if not before, Aristotle is speaking of the copula. Some supporters of this view believe that the subject of both ‘is nothing’ (οὐδέν ἐστιν) and ‘additionally signifies’ (προσσηµαίνει) is ‘ “to be” or “not to be” ’(τὸ εἶναι ἢ µὴ εἶναι, b22), which they took as a reference to the copula in the first place.36 Others – the supporters of the Porphyrean Reading – believe that the subject is ‘that which is’ (τὸ ὄν, b23), which is where they recognized the copula’s first appearance in the passage.37 Either way, Aristotle is taken to be claiming about the copula that ‘by itself it is nothing’ in the sense that it lacks the basic, lexical signification which ordinary verbs have, i.e. is not (also) a name for something in the way in which ordinary verbs are (also) names for something. He is also taken to be claiming that the copula signifies a ‘combination’, i.e. a link between the subject and the predicate term ‘X’ and ‘F’ in a copulative sentence of the form ‘X is F’.38 This link can only be ‘thought of’ in the presence of the items to be linked; in other words, 33 Perhaps there is a hint in this direction in Ammonius’ account of how Porphyry dealt with this clause, 57.1–6 Busse. Boethius clearly translates and construes τὸ ὄν as equivalent to ἔστι in in Int. i. 65 Meiser; he is more cautious in ii. 77–8. Among modern interpreters see Weidemann 2014a: 185 (who finds this use ‘noteworthy’). 34 Cf. the Anonymous Commentary, 11.14–19 Tarán. 35 See Ackrill 1963: 123–4 (and indeed already Ammonius, 56.4–5 Busse). 36 See Ackrill 1963: 122 (who does not endorse this interpretation), Ax 1979: 277, Whitaker 1996: 56–8, and perhaps also Barnes 2009: 32, 45. See also Montanari (1988: 270–2, cf. 263–4), whose stance is peculiar: he identifies the subject with τὸ εἶναι ἢ µὴ εἶναι but holds (implausibly in my opinion) that only at the present stage is Aristotle construing that specifically as the copula. 37 Weidemann 2014a: 185–6; see above. 38 Alexander of Aphrodisias, in Metaph. 371.20–36 Hayduck (cf. Boeth. in Int. ii. 77.3–13 Meiser), seems to take this line and give it a further twist. In the course of commenting on Metaph. Δ 7, he claims that ‘to be’ signifies the ‘holding’ (ὕπαρξις) which is ‘appropriate’ to the item to which it is attached – i.e. to the category to which the item belongs – and quotes our passage as confirming this claim. The idea seems to be that in a sentence of the form ‘X is F’ the copula ‘is’ 12 only in the context of a complete sentence, in association with a subject and a predicate term, does the copula discharge its linking function. A conclusion is implied: the copula uttered by itself is not equivalent to a complete sentence. And since any verb is equivalent to copula + participle, as we saw above, from this a final conclusion is further meant to follow: no verb is equivalent to a complete sentence. I have two reasons for disagreeing with this interpretation of the passage’s conclusion. First, the passage is contrasting verbs said ‘by themselves’ (καθ᾿ αὑτά) with complete sentences. In this context ‘by themselves’ means ‘in isolation, on their own’; likewise in ch. 1, 16a13, Aristotle claimed that names and verbs ‘by themselves’ (αὐτά) lack the complexity of sentences. Therefore, when Aristotle winds up saying something about ‘to be’ ‘by itself’ (αὐτό), we should expect that to have the same force and refer, not to the intrinsic signification of ‘to be’, but rather to its being uttered in isolation.39 Secondly – and more fundamentally – the above interpretation is incompatible with my own construal of the previous sections: if, as I believe, the whole passage so far has been concerned with the existential ‘to be’, not the copulative, then it is the former, not the latter, that should be at issue here too. So let us try to see what the sentence can mean, if ‘to be’ is instead existential. To start with, the subject of ‘by itself it is nothing’ (αὐτὸ ... οὐδέν ἐστιν) is naturally taken to be primarily ‘ “to be” or “not to be” ’(τὸ εἶναι ἢ µὴ εἶναι) – and by extension any other verb. ‘What is’ (τὸ ὄν) may be also involved, but its clause is not part of the main line of argument. Now notice that ‘by itself it is nothing’ need not necessarily mean that ‘to be’ does not signify anything, although this is how it is usually understood by modern interpreters. The meaning might well be – more modestly and to the point – that an isolated utterance of the complete, existential ‘is’ does not (yet) constitute anything relevant for our present purposes, i.e. anything sentential. Here Aristotle may well expect us to recall Plato’s Sophist, 262d, where the Stranger says that names alone or verbs alone do not constitute a sentence and that only someone who interweaves names and verbs ‘accomplishes something’ (τι περαίνει) in this respect. This alternative construal of ‘by itself it is nothing’ is actually advanced by several commentators, both ancient and modern. One of the ancient ones is Ammonius: he says that ‘what is’ is ‘nothing’,40 not as devoid of signification [ἄσηµον] or as predicated homonymously of the objects; he rather says ‘it is nothing’ true or false [οὐδέν ἐστί φησιν οὔτε ἀληθὲς οὔτε ψεῦδος]. (57.6–8 Busse) signifies nothing but the ‘holding’ of F – a ‘holding’ which is of a different kind depending on the category to which F belongs: substantial in ‘Oscar is a cat’, qualitative in ‘Oscar is handsome’, quantitative in ‘Oscar is 20 inches long’, etc. (For the translation of Alexander’s ὕπαρξις as ‘holding’ see Barnes 2009: 41–2; the term is instead translated as ‘existence’ by Dooley 1993: 44, 144 nn. 161, 164 and Bonelli 2001: 92–4.) 39 See Burnyeat (2003: 14): ‘Aristotle is often supposed to say here that the “is” in “Socrates is wise” has no semantic meaning of its own, but is a mere copula. Yet it fits the context better to take this a remark about someone uttering the solitary word “is” all by itself, not about the word “is” in a standard predication.’ The quotation is continued in n. 44 below. 40 Ammonius assumes that the subject of ‘by itself it is nothing’ is ‘that which is’ (τὸ ὄν). As can be seen by what I said above, I believe he is wrong on this specific point. 13 These words coincide almost verbatim with those of Boethius, in Int. I. 65.20–4, ii. 78.8–13 Meiser – which suggests that the interpretation actually goes back to Porphyry.41 Indeed, in the sequel (57.18– 25) Ammonius implies that this was already the view of Alexander of Aphrodisias.42 But does Aristotle after all believe that an isolated ‘is’ signifies anything? (I might have asked ‘does Aristotle after all believe that an isolated existential ‘is’ signifies anything?’ – but the existential and the copulative are indistinguishable when the verb is uttered in isolation.) If my suggestion is right, the text is not really designed to answer this question, and it is difficult to work out what Aristotle would say if we asked him. The problem is compounded by its potential connections to broader issues – which cannot be properly addressed here – about the purport of Aristotle’s famous slogan, ‘What is is said in many ways’ (τὸ ὂν λέγεται πολλαχῶς, Metaph. Γ 2, Z 1, etc.). If that is intended as a thesis about the sense or meaning of the verb ‘to be’, and if the thesis is that ‘to be’ has as many different senses as there are categories and no common, generic sense,43 then perhaps Aristotle might believe that an isolated ‘is’ signifies nothing, because it is not tied by context to any particular category.44 However, it is far from clear that this is really what Aristotle would think. Why should he not simply say that a contextless ‘is’, like any isolated utterance of a word endowed with different senses, signifies many things? Think of the famous proof of the Principle of Non-Contradiction in Metaph. Γ 4, 1006a18–b11. The Principle’s opponent is challenged to utter just a single significant word; his acceptance of an account of what the word signifies commits him, in the final analysis, to the very Principle he wants to deny. Never mind if the word signifies not just one thing but many, Aristotle says, as long as they are a definite number: all we need to do is associate each of the things signified with a different account. Only if the word signified indefinitely many things would it be the same as if it signified nothing at all. Furthermore, ‘What is is said in many ways’ also admits of other, less radical construals. In particular, there is some reason to believe that, although of course Aristotle holds that there is no 41 On Porphyry as Boethius’ main source, and the common source of Ammonius and Boethius, see Ebbesen 1990: 374–7 and Blank 1996: 3–4. At in Int. ii. 78.8 Meiser Boethius uses the words vel certe (whose meaning here seems to be ‘In fact, however …’; cf. 76.15) to introduce this interpretation immediately after explicitly ascribing to Porphyry a different and incompatible one, according to which ‘is’ – whether existential or copulative – when uttered on its own nihil omnino significat (in Int. ii. 77.13–78.8 Meiser). This – and not the view I ascribe to him in the main text on the basis of the convergence between Ammonius and Boethius – is usually reported by scholars as Porphyry’s interpretation. Here is my conjecture: both interpretations were present in Porphyry’s commentary in the same order; Porphyry introduced the second, superior one with ἤ, which is typically used by Aristotle and his commentators to announce a new and better solution to the problem at hand; Boethius literally followed Porphyry in reporting both interpretations, translating ἤ as vel certe, whereas Ammonius more economically reported only the second, superior interpretation. 42 See n. 45 below. Among modern commentators see Ackrill 1963: 123 (quoted in n. 46) and Bärthlein 1984: 250. 43 For discussion see Matthews 1972, Barnes 2005: 72–4 and McDaniel 2017: 12–34 (on Aristotle see especially 30–1). 44 Cf. Burnyeat (2003: 14, the immediate sequel of the quotation in n. 39 above): ‘It is not that in a standard predication the verb has no meaning in its own right, but that what its meaning is (what sort of being it signifies) is contextually dependent on the subject and/or predicate expression flanking it; hence without a context it has no meaning at all, whereas an ordinary verb uttered on its own (someone suddenly shouts out “Sits”) does at least put the hearer in mind of its signification.’ 14 genus of being but rather several different ways of being, as many as the categories, nevertheless he takes the verb ‘to be’ to have a weak generic sense which cuts across the categories. This weak generic sense is what enables him to claim, in An. Post. 2. 1–2, that we can know that X is (ὅτι ἔστι) before we know what X is (τί ἐστι); and to introduce, in Metaph. Γ 1–2, first philosophy as a general enquiry into ‘being qua being’ (τὸ ὂν ᾗ ὄν). And after all, how could Aristotle deny that there is a recognizable lexical difference between an isolated utterance of ‘is’ and one of a nonsense verb such as ‘piffs’? So I actually incline to believe that Aristotle takes an isolated ‘is’ to have some sense of its own.45 Let us now come to the second limb of the final sentence: ‘but it additionally signifies some combination, which cannot be thought of without the components’ (προσσηµαίνει δὲ σύνθεσίν τινα, ἣν ἄνευ τῶν συγκειµένων οὐκ ἔστι νοῆσαι). This, I take it, refers to the verb’s predicative function and means that the verb can perform that function only when a subject term is added to it, hence only in the context of a complete sentence: the existential ‘is’, like any other verb, is an incomplete expression, awaiting completion by a subject term. Thus the ‘combination’ of which Aristotle is thinking is not the one between the subject X and the predicate F in a copulative sentence of the form ‘X is F’; it is rather the one between X and F in a name-verb sentence of the form ‘X Fs’. There is nothing odd about this; interpreters who hold that in this sentence Aristotle is speaking of the copula must in any case take the point to be generalizable to all verbs. And the word σύνθεσις, ‘combination’, was used with reference to name-verb sentences in ch. 1, 16a12, and even before in Plato, Cratylus 431c and Sophist 263d.46 Note that, according to the account I have just set forth, there is a subtle asymmetry in the way in which the two particles µέν … δέ (literally ‘on the one hand … on the other’), contrast the final sentence’s two limbs, the negative (‘by itself it is nothing’) and the positive one (‘it additionally signifies some combination, which cannot be thought of without the components’). For even though µέν is placed after αὐτό, ‘by itself’, in fact both limbs of the antithesis are about the isolated verb: on 45 Cf. the authors mentioned in n. 43 and the view, defended by Ruijgh (1979) and West (2016), that the original sense common to the Greek ‘be’ and its Indo-European counterparts is ‘be there, be present, be available’. Ammonius seems to have a view similar to the one we are considering when he claims that the existential ‘is’ signifies ὕπαρξις, ‘holding’ (in Int. 44.11–14 Busse). He also seems to ascribe something similar to Alexander (57.23–6). According to his report, in the course of commenting on our passage Alexander claimed that ‘is’ is a name, just like any other verb, and as such it ‘primarily’ has the function of signifying something, i.e. ‘participation in being’ (τῆς τοῦ ὄντος µεθέξεως) – i.e., presumably, existence. Unfortunately this is very different from what Alexander says in the Metaphysics commentary (see n. 38 above). For this reason Suto 2012: 218 doubts the soundness of the report. Alternatively, Ammonius might be reporting an interpretation which Alexander considered but did not ultimately endorse: cf. n. 41 for a similar case. 46 My interpretation thus coincides with that of Ackrill (1963: 123): ‘Perhaps Aristotle’s last remark is not about the copulative but about the existential “is”. If so, “by itself it is nothing” does not characterize the copula in contrast to ordinary verbs. It means only that “is” (“exists”), like other verbs, asserts nothing on its own. Like them it both signifies something and also indicates a synthesis – it calls for the addition of a subject-term in order that it may fulfil its role as a sign of something said of something else.’ Cf. Barnes (1996: 189): ‘Aristotle ought to be talking of existential “εἶναι” ’ (Barnes 2009 takes a different view: see n. 36). 15 the one hand it is incomplete; on the other, it invites completion by a subject term. This kind of asymmetry is fairly common and unproblematic.47 The verb ‘additionally signifies’ (προσσηµαίνει) was used by Aristotle at 16b6–10 (which we quoted at the outset) to refer to the verb’s signification of time; there the point was apparently that the signification of time is added to a verb’s basic, lexical signification. Since in the same lines Aristotle also said that a verb is a sign of something predicated, thereby apparently meaning also that a verb is a sign of the predication itself (i.e. that the verb ascribes the predicate to the subject), it is tempting to suppose that he regards that too as a kind of ‘additional signification’.48 This would explain his use of ‘additionally signifies’ in our passage. It would leave open the question whether the additional predicative signification implies the presence of a more basic lexical signification – as in ordinary verbs – or ‘to be’ is a special verb that can have the additional signification while lacking the basic one.49 As I said above, I incline to the former option. An alternative construal of the final sentence’s second limb is proposed by Ammonius, 57.13–18 Busse: That ‘what is’ ‘additionally signifies’ the combination, and that not only this, but also each of the simple vocal sounds does so in the same way, seems not to have been said in the same way in which the verb was said to ‘additionally signify’ time; rather, it is used for ‘signifies additionally to something else’ [πρὸς ἑτέρῳ σηµαίνειν]. That is to say, when joined with something else it signifies a combination [συµπλεκόµενον ἑτέρῳ τινὶ σηµαίνειν σύνθεσιν] which is then receptive of falsehood and truth; the simples must be conceived of before this combination. Thus Ammonius takes ‘but it additionally signifies some combination’ etc. as a point about ‘to be’ – and by extension any other verb – not as uttered in isolation but rather as already joined with a subject term to form a complete declarative sentence. Boethius, in Int. ii. 78.21–6 Meiser, essentially agrees;50 and since his lines are part of a section whose beginning (ii. 78.8–13 Meiser) has already turned out to be parallel to Ammonius’ commentary, and hence probably to depend on Porphyry,51 it is a likely guess that we still have remnants of Porphyry’s commentary before us. Indeed, there is some evidence in Ammonius (in Int. 57.19–33 Busse) that in fact all this goes back to Alexander. The construal of the ancient commentators restores a perfect symmetry between the two limbs of the final sentence: now the two words which immediately precede µέν and δέ, i.e. αὐτό and προσσηµαίνει, refer precisely to the two contrasted conditions of being uttered in isolation vs being joined with a subject term. This counts as an advantage. In my opinion, however, it is not sufficient 47 See Denniston 1954: 371–2, with examples. A different use of προσσηµαίνειν is in play at 10. 20a12–13, where Aristotle claims that the quantifiers ‘every’ and ‘no’ προσσηµαίνει universal quantification, meaning that this is what they contribute to the whole sentence. Nuchelmans 1973: 29 (cf. Barnes 2009: 53) takes this to be the pertinent use in our passage; I find this unconvincing. 49 See Whitaker 1996: 56. 50 Cf. in Int. I. 65.25–66.25 Meiser. Boethius renders Aristotle’s προσσηµαίνει as consignificat, ‘co-signifies’; cf. Steph. in Int. 13.38, 15.2 Hayduck συσσηµαίνει. Among modern interpreters Ammonius is followed by Bärthlein (1984: 250–1). 51 See n. 41 above and text thereto. 48 16 to render this construal plausible; for it is very doubtful that the preverb προσ- in προσσηµαίνει is sufficient to convey the required reference to the presence of a subject term. Conclusion and envoi We have argued our way through the intricacies of this passage; it is time to sum up. I will do so by finally offering a complete paraphrase of our lines that reflects our previous conclusions: When uttered just by themselves verbs are still a specific kind of names and therefore signify some predicate (for the speaker arrests his own thought on something and causes the hearer to do the same), but they do not yet signify whether or not that predicate exists, i.e. whether or not it holds of something. For even the verb ‘to be’ or ‘not to be’ is not a sign of the extralinguistic subject [alternatively, reading not τοῦ but του: ‘of some state of affairs’] – not even if you utter the description ‘what is’ just by itself. For ‘to be’ by itself does not amount to a complete sentence; rather, it conveys the additional signification of a combination with a subject – a combination which cannot be thought of unless all of its components are present. A final remark. If you compare this paper with Ackrill’s commentary, you will see that I have ultimately done little more than confirm his interpretation – which he set out in a couple of crisply written pages – by supporting it with more detailed arguments and more erudition. I hope this may be an opportunity to reflect on the way in which his work, sixty years after its publication, is still a model of acumen, concision and clarity for ancient philosophy scholars. Università degli Studi di Firenze Scuola Normale Superiore, Pisa francesco.ademollo@unifi.it REFERENCES Ackrill, J. L. (1963), Aristotle: Categories and De Interpretatione, Oxford. Ademollo, F. (2015), ‘Names, Verbs, and Sentences in Ancient Greek Philosophy’, in M. Cameron and R. Stainton (eds), Linguistic Content: New Essays on the History of Philosophy of Language, Oxford, 33–54. –––– (2022), ‘Why the Cratylus Matters, or: Plato’s Cratylus and the History of the Philosophy of Language’, in V. Mikeš (ed.), Plato’s Cratylus: Proceedings of the Eleventh Symposium Platonicum Pragense, Leiden and Boston, 26–47. ---- (2022), ‘Anti-Platonism in Aristotle’s Categories’, in F. Ademollo, F. Amerini and V. De Risi (eds), Thinking and Calculating. Essays in Logic, Its History and Its Philosophical Applications in Honour of Massimo Mugnai, Cham, 31–52. 17 Ax, W. (1979), ‘Zum isolierten ῥῆµα in Aristoteles’ de interpretatione 16b19–25’, Archiv für Geschichte der Philosophie 61, 271–9. Barnes, J. (1995), ‘Metaphysics’, in J. Barnes (ed.), The Cambridge Companion to Aristotle, Cambridge, 66–108. –––– (1996), ‘Grammar on Aristotle’s Terms’, in M. Frede, G. Striker (eds.), Rationality in Greek Thought, Oxford, 175–202. Reprinted in J. Barnes, Logical Matters. Essays in Ancient Philosophy II, ed. by M. Bonelli, Oxford 2012, 147–71. –––– (2009), ‘Notes on the Copula’, Dianoia 14, 27–62. Bärthlein, K. (1984), ‘Nochmals über das isolierte Aussagewort (CA, De Interpretatione, Kap. 3). Mit einem Anhang zur Diskussion über die Echtheit dieser Schrift’, Rheinisches Museum 127, 227-258. Blank, D. (1996), Ammonius: On Interpretation 1–8, London. Bonelli, M. (2001), Alessandro di Afrodisia e la metafisica come scienza dimostrativa, Naples. Brown, L. (1994), ‘The Verb “To Be” in Greek Philosophy: Some Remarks’, in S. Everson (ed.), Language (Companions to Ancient Thought, 3), Cambridge, 212–36. Burnyeat, M. F. (2003), ‘Apology 30b 2–4: Socrates, Money, and the Grammar of γίγνεσθαι’, Journal of Hellenic Studies 123: 1–25 (repr. in M. F. Burnyeat, Explorations in Ancient and Modern Philosophy vol. 4, Cambridge 2022, 5–40). Busse, A. (ed.) (1897), Ammonius: In Aristotelis De interpretatione commentarius (CAG IV.5), Berlin. Charles, D. (2002), ‘Some Comments on Prof. Enrico Berti’s “Being and Essence in Contemporary Interpretations of Aristotle” ’, in A. Bottani, M. Carrara and P. Giaretta (eds), Individuals, Essence and Identity: Themes of Analytic Metaphysics, Dordrecht, Boston, and London, 109–26. Crivelli, P. (2004), Aristotle on Truth, Cambridge. –––– (2015), ‘Truth in Metaphysics E 4’, Oxford Studies in Ancient Philosophy 48, 167–225. Delcomminette, S. (2020), ‘“Être dans le temps, c’est avoir son être mesuré par le temps”: à propos de Phys. IV, 12, 220b32–221a9’, in G. R. Giardina (ed.), ΤΟ ΜΕΤΡΟΝ. Sur la notion de mesure dans la philosophie d’Aristote, Paris, 45–65. Denniston, J. D. (1954), The Greek Particles, 2nd edn, Oxford. Dooley, W. E. (1993), Alexander of Aphrodisias: On Aristotle’s Metaphysics 5, London. Ebbesen, S. (1990), ‘Boethius as an Aristotelian Commentator’, in R. Sorabji (ed.), Aristotle Transformed. The Ancient Commentators and Their Influence, Ithaca (NY), 373–91. Frede, M. (unpublished), ‘Aristotle on Nouns and Verbs in De Interpretatione’, paper presented at the 13th Symposium Aristotelicum (Pontignano, 1993). McDaniel, K. (2017), The Fragmentation of Being, Oxford. Matthews, G. (1972), ‘Senses and Kinds’, Journal of Philosophy 69, 149–57. Mignucci, M. (2007), Aristotele: Analitici Secondi. Organon IV, Rome and Bari. Minio-Paluello, L. (ed.) (1949), Aristotelis Categoriae et Liber De Interpretatione, Oxford. Montanari, E. (1984), La sezione linguistica del Peri hermeneias di Aristotele. Volume primo: il testo, Florence. Montanari, E. (1988), La sezione linguistica del Peri hermeneias di Aristotele. Volume secondo: Il commento, Florence. 18 Nuchelmans, G. (1973), Theories of the Proposition. Ancient and Medieval Conceptions of the Bearers of Truth and Falsity, Amsterdam and London. Rijk, L. M. De (2002), Aristotle: Semantics and Ontology, vol. 1, Leiden, Boston, and Köln. Ruijgh, C. J. (1979), ‘A Review of: C. H. Kahn, The Verb “Be” in Ancient Greek’, Lingua 48, 43–83. Sedley, D. (1996), ‘Aristotle’s De interpretatione and Ancient Semantics’, in G. Manetti (ed.), Knowledge Through Signs, Turnhout, 87–108. –––– (2004), ‘Aristote et la signification’, Philosophie antique 4, 5–25. Suto, T. (2012), Boethius on Mind, Grammar and Logic, Leiden and Boston. Waitz, Th. (1844), Aristoteles: Organon graece. Pars prior, Leipzig (reprinted Aalen 1965). Wedin, M. V. (2000), Aristotle’s Theory of Substance, Oxford. Weidemann, H. (2014a), Aristoteles: Peri hermeneias, 3rd edn, Berlin and Boston. –––– (2014b) (ed.), Aristoteles: De interpretatione, Berlin and Boston. West, M. L. (2016), ‘So What Is It to Be?’, in D. Gunkel, J. T. Katz, B. Vine and M. Weiss (eds), Sahasram Ati Srajas: Indo-Iranian and Indo-European Studies in Honor of Stephanie W. Jamison, Ann Arbor and New York, 486–98. Whitaker, C. W. A. (1996), Aristotle’s De interpretatione: Contradiction and Dialectic, Oxford. 19