1. Introduction
Four brief but intriguing passages from Aristotle and Diogenes Laertius are our best indication for Protagoras’ interest in linguistic phenomena. The first two concern gender distinctions in nouns and congruence, while the other two are suggestive of Protagoras’ interest in verbal mood (or so we will argue). Here are the passages as one can find them in modern collections of Protagoras’ fragments:Footnote 1
Gender
τέταρτον, ὡς Πρωταγόρας τὰ γένη τῶν ὀνομάτων διῄρει, ἄρρενα καὶ θήλεα καὶ σκεύη⋅ δεῖ γὰρ ἀποδιδόναι καὶ ταῦτα ὀρθῶς⋅ “ἡ δ’ ἐλθοῦσα καὶ διαλεχθεῖσα ᾤχετο.” (Arist. Rh. 1407b6–8 = DK 80 A 27 = Radermacher B iii 6 = Protagoras D23 Laks/Most)
A fourth [rule of correct usage or ἑλληνίζειν is] Protagoras’ distinction between the genders of nouns: masculine, feminine and ‘things’. For one should correctly reproduce these distinctions too: ‘Having come and having spoken, she left.’
σολοικισμὸς δ’ οἷον μέν ἐστιν εἴρηται πρότερον⋅ ἔστι δὲ τοῦτο καὶ ποιεῖν καὶ μὴ ποιοῦντα φαίνεσθαι καὶ ποιοῦντα μὴ δοκεῖν, καθάπερ, ὃ Πρωταγόρας ἔλεγεν, εἰ “ὁ μῆνις” καὶ “ὁ πήληξ” ἄρρενά ἐστιν⋅ ὁ μὲν γὰρ λέγων “οὐλομένην” σολοικίζει μὲν κατ’ ἐκεῖνον, οὐ φαίνεται δὲ τοῖς ἄλλοις, ὁ δὲ “οὐλόμενον” φαίνεται μέν, ἀλλ’ οὐ σολοικίζει. (Arist. SE 173b17–22 = DK 80 A 28 = Radermacher B iii 7 = Protagoras D24 Laks/Most)
What sort of a thing a ‘solecism’ is has been said above. It is possible to commit one and to appear to commit one without doing so and to commit one without seeming to, as happens if, as Protagoras claimed, μῆνις wrath and πήληξ helmet are masculine. For according to him, someone who says ‘destructress’ (fem. οὐλομένην) commits a solecism, but he does not appear to do so to others, and someone who says ‘destructor’ (masc. οὐλόμενον) appears to commit a solecism, but does not in fact do so.
Mood
διεῖλέ τε τὸν λόγον πρῶτος εἰς τέτταρα⋅ εὐχωλήν, ἐρώτησιν, ἀπόκρισιν, ἐντολήν (οἱ δέ, εἰς ἑπτά⋅ διήγησιν, ἐρώτησιν, ἀπόκρισιν, ἐντολήν, ἀπαγγελίαν, εὐχωλήν, κλῆσιν), οὓς καὶ πυθμένας εἶπε λόγων. (D. L. 9.53–4 = DK 80 A 1 = Radermacher B iii 10 = Protagoras D17 Laks/Most)
He was the first to divide speech into four kinds – prayer, question, answer, command (others say into seven: narration, question, answer, command, report, prayer, calling) – which he even named ‘foundations of speech’.
τῶν δὲ περὶ τὴν λέξιν ἓν μέν ἐστιν εἶδος θεωρίας τὰ σχήματα τῆς λέξεως, ἅ ἐστιν εἰδέναι τῆς ὑποκριτικῆς καὶ τοῦ τὴν τοιαύτην ἔχοντος ἀρχιτεκτονικήν, οἷον τί ἐντολὴ καὶ τί εὐχὴ καὶ διήγησις καὶ ἀπειλὴ καὶ ἐρώτησις καὶ ἀπόκρισις καὶ εἴ τι ἄλλο τοιοῦτον. παρὰ γὰρ τὴν τούτων γνῶσιν ἢ ἄγνοιαν οὐδὲν εἰς τὴν ποιητικὴν ἐπιτίμημα φέρεται ὅ τι καὶ ἄξιον σπουδῆς. τί γὰρ ἄν τις ὑπολάβοι ἡμαρτῆσθαι ἃ Πρωταγόρας ἐπιτιμᾷ, ὅτι εὔχεσθαι οἰόμενος ἐπιτάττει εἰπὼν “μῆνιν ἄειδε θεά”; τὸ γὰρ κελεῦσαι, φησίν, ποιεῖν τι ἢ μὴ ἐπίταξίς ἐστιν. διὸ παρείσθω ὡς ἄλλης καὶ οὐ τῆς ποιητικῆς ὂν θεώρημα. (Arist. Poet. 1456b8–19 = DK 80 A 29 = Radermacher B iii 13 = Protagoras D25 Laks/Most)
In matters of lexis, one kind of study concerns the manners of speaking. Knowledge of these belongs to the art of delivery and to anyone who has that kind of expertise; for instance, what a command is or what a prayer, a narration, a threat, a question, an answer or any other such matter. For no serious charge can be brought against the art of poetry on the basis of knowledge or ignorance of these matters. For why would one suppose that what Protagoras criticises is a mistake, namely that [Homer], whilst thinking he is uttering a prayer, gives an order by saying, ‘Sing, Muse, of the wrath’? For to command, [Protagoras] says, is to give the instruction to do or not to do something. Therefore, let this consideration be left aside as being the object of study of another art and not of poetry.
These two pairs of fragments suggest a similar pattern. First, Protagoras makes a global and seemingly uncontroversial distinction – between genders in nouns or between four (or seven) πυθμένες λόγων – and then he applies those distinctions in an unexpected, provocative way, arguing, for instance, that it is not ἡ μῆνις, but ὁ μῆνις, or that one should not say ἄειδε θεά when uttering a prayer (εὐχωλή). At the same time, many details remain to be clarified: it is not obvious what some of the terms used in these passages mean, nor why μῆνις should be masculine or saying ἄειδε θεά be incompatible with uttering a prayer. Nonetheless, without being blind to the apparent normative slant of Protagoras’ linguistic endeavours, Theodor Gomperz found these fragments impressive enough to credit the sophist with being the first to show a systematic interest in grammar.Footnote 2
However, efforts to come to a more precise understanding of exactly what Protagoras was up to were dealt a severe blow when Detlev Fehling (Reference Fehling1965) argued that Protagoras’ views of language did not amount to a systematic and independent body of early linguistic theory. Pointing out that two of the concrete applications of Protagoras’ insights of which we have knowledge concern the opening words of the Iliad, μῆνιν ἄειδε θεά … οὐλομένην, Fehling argued that Protagoras’ linguistic observations were entirely ad hoc, and solely developed as an instrument for criticising Homer and other poets. Fehling drew the conclusion
daß die grammatischen Einteilungen, die von Protagoras überliefert sind, nicht einer systematischen Abhandlung entstammen, sondern nur im Rahmen einer beispielhaften Vorführung seiner Methode der Dichterkritik und auch dort nur beiläufig entwickelt wurden. Es ist dann so gut wie sicher, daß es weitere Einteilungen von ihm nicht gegeben hat.Footnote 3
Fehling's view has won widespread acceptance, so that it is now quite generally thought that Protagoras’ observations on language were ‘casually’ made (Schenkeveld (Reference Schenkeveld1984) 293), or a mere ‘by-product’ of his intellectual rivalry with Homer (Sluiter (Reference Sluiter1990) 8). Only recently have scholars begun to reconsider the sophist's contribution to linguistic thought, especially Michele Corradi (Reference Corradi2012) and Adriaan Rademaker (Reference Rademaker, van Ophuijsen, van Raalte and Stork2013). However, they continue to study Protagoras’ ὀρθοέπεια mostly in the context of what we know of the sophist's wider curriculum, as being relevant to his teachings on poetry, rhetoric, epistemology, εὐβουλία and political ἀρετή. They are less interested in reconstructing Protagoras’ linguistic arguments, and are content with offering various possibilities.
In our view, Fehling's position deserves a firmer rebuttal than it has been given so far. We do not deny that Protagoras pressed his linguistic activities into the service of his critique of Homer, but we do not think that they can have been nearly so ad hoc as Fehling and others claim. Here is for good measure Fehling's reconstruction of Protagoras’ argument, written in the sophist's voice:
‘Ich will zeigen, wie man untersucht, ob ein Dichter etwas richtig gesagt hat oder nicht, denn der Menge scheint oft vieles ganz anders, als es in Wirklichkeit ist. Die Ilias wird von allen Griechen für die größte aller Dichtungen gehalten. Dennoch hat Homer bereits ganz zu Anfang zweierlei falsch gemacht. Er sagt dort μῆνιν οὐλομένην. Nun sind aber alle Substantive entweder männlich oder weiblich oder sächlich. Das Wort μῆνιν gebraucht er weiblich, es müßte aber zu den männlichen gehören. Deshalb ist οὐλομένην nur dem Anschein nach richtig, in Wahrheit wäre es aber οὐλόμενον. Ferner sagt er ἄειδε θεά. Nun ist aber alles, was man sagt, entweder Frage oder Antwort oder Befehl oder Bitte. ἄειδε aber sagt man, wenn man befiehlt. Also hat Homer zwar geglaubt, die Muse zu bitten, in Wirklichkeit ihr aber befohlen.’Footnote 4
It seems to us that this reconstruction passes over a number of important points.
First of all, Fehling fails to acknowledge the pioneering character of Protagoras’ observations, writing instead as if distinctions such as those between the genders of nouns could already be taken for granted by Protagoras and his audience.Footnote 5 In fact, the passage from Aristotle's Rhetoric strongly suggests that Protagoras was the first to make gender distinctions in nouns, while Diogenes Laertius explicitly states that Protagoras was the first (πρῶτος) to divide up λόγος into four (or seven) categories (in a section on the sophist's εὑρήματα). If that is so, the question is whether such distinctions can be ‘casually’ made without much argument or exposition. It is not immediately transparent, for instance, that the difference between ὁ ἡ τό is one of gender (if, say, you ask a French child not yet schooled in grammar why it is ‘la table’ and not *‘le table’, the child would hardly say that it is because ‘table’ is feminine). And the same question becomes more pressing still when we also take into consideration that Protagoras was not only the first to make the distinctions attributed to him, but also made them in such a way that he could put them in the form of normative ‘rules’ and convincingly maintain, against the Sprachgefühl of his fellow Greeks, that Homer made two mistakes in writing μῆνιν ἄειδε θεά … οὐλομένην. If his reasoning lacked all plausibility, he would surely merely have sounded ridiculous (as of course he does to Fehling and other modern scholars – it may be noted in passing that sounding ridiculous is different from being playful). All of this requires a more elaborate and sustained argument than is warranted by a wish to score a few easy points off Homer.
Secondly, if there is any historical accuracy in Plato making Protagoras proclaim, in the eponymous dialogue,
ἡγοῦμαι … ἀνδρὶ παιδείας μέγιστον μέρος εἶναι περὶ ἐπῶν δεινὸν εἶναι· ἔστιν δὲ τοῦτο τὰ ὑπὸ τῶν ποιητῶν λεγόμενα οἷόν τ᾿ εἶναι συνιέναι ἅ τε ὀρθῶς πεποίηται καὶ ἃ μή, καὶ ἐπίστασθαι διελεῖν τε καὶ ἐρωτώμενον λόγον δοῦναι (Prot. 339a)
I believe the most important piece of education for a man is to be competent in poetry; and that means to be able to understand what is well formulated and what is not in what the poets say, and to know how to analyse these things and set them out when questioned,
it becomes rather difficult to imagine Protagoras charging his exorbitant fee of 100 minae (D. L. 9.52 = DK 80 A 1) without giving detailed and reasoned guidance on how to proceed with the critical analysis that was meant to be the centrepiece of his curriculum.Footnote 6
And thirdly, for what it is worth, we should note that the famous encounter of Strepsiades with ‘Socrates’ in Aristophanes’ Clouds, which will be discussed more fully later on, unambiguously makes the ‘Protagorean’ study of grammar an independent and self-contained part of instruction. After Strepsiades has explicitly complained that all he wants to learn is the ‘most unjust speech’, he is sternly told by ‘Socrates’(-Protagoras):
ἀλλ᾿ ἕτερα δεῖ σε πρότερα τούτου μανθάνειν,
τῶν τετραπόδων ἅττ᾿ ἐστιν ὀρθῶς ἄρρενα
(Ar. Nub. 658–9)
but before that you have to learn other things, which of the four-footed beings are truly male
– upon which a lengthy lesson on grammatical gender follows, which bears no humoristic resemblance whatsoever to what Fehling imagined to have taken place in Protagoras’ classroom, but focuses exclusively on (comic) linguistics per se (Ar. Nub. 660–93; cf. Section 2.5).
In what follows, we will first make an attempt to reconstruct Protagoras’ ideas about gender and then those about mood, before considering other grammatical distinctions which the sophist may have made.
2. Protagoras on grammatical gender
2.1 Text and interpretation of Sophistici Elenchi
As a first step towards reconstructing Protagoras’ doctrine of gender, we need to consider the relevant passage from the Sophistici Elenchi more closely. Exactly what Aristotle says here is unclear and has led to confusion. One difficulty originates from a textual issue. The manuscripts read καθάπερ ὁ Πρωταγόρας ἔλεγεν, εἰ κτλ. (one manuscript omits ὁ), but we adopt the accentuation ὃ and accompanying punctuation proposed by Ross in the OCT. This correction brings the text in line with Aristotle's practice of not using articles with personal names when citing sources.Footnote 7 It also makes it possible to take καθάπερ and εἰ together, producing a phrase with which Aristotle regularly introduces (dubious) claims that serve as examples.Footnote 8 An important consequence of construing καθάπερ with εἰ rather than with ἔλεγεν is that it makes clear that the claim that μῆνις and πήληξ are masculine is to be attributed to Protagoras, but that the distinction among three manifestations of solecism belongs to Aristotle himself. This is likely enough, because throughout the Sophistici Elenchi Aristotle is concerned with analysing the validity and invalidity of sophistic arguments (in casu those that involve grammatical errors) and the reasons why invalid ones often have a semblance of validity about them.Footnote 9
With those bases covered, we turn to the question how Protagoras’ claim that μῆνις and πήληξ are masculine relates to these three manifestations of solecism. In our view, two possibilities are worth considering. For our first explanation, it is important to keep in mind that in the Sophistici Elenchi Aristotle critically reviews types of invalid, but seemingly plausible, arguments used by sophists when they attempt to refute an interlocutor in a question and answer session (ἔλεγχος).Footnote 10 The first type of solecism identified by Aristotle (‘to commit one’) is not, it seems, exemplified. It provides the baseline, so to speak, and merely concerns mistakes committed against common usage which are recognised by everyone. By contrast, the other two types may both be illustrated in the sentence headed by ὁ μὲν γὰρ λέγων, though not straightforwardly so, because after Aristotle's opening statement on there being three manifestations of solecism, the perspective shifts to that of Protagoras (note κατ’ ἐκεῖνον, which distances Aristotle from Protagoras’ claim).Footnote 11 The sentence may thus be read as a compressed version of a two-stage ἔλεγχος. One first gets an interlocutor to say μῆνιν οὐλομένην (‘Don't you think the first lines of the Iliad are particularly fine? Can you recite them for me?’) and then states that the interlocutor is wrong to say it (‘You thought they were fine? Now that I come to think of it, there's a huge mistake in them, which you just unthinkingly reproduced’). The sophist next leads the interlocutor through an argument which makes it seem plausible that μῆνιν οὐλομένην is indeed wrong (Aristotle's second type: to ‘appear to commit a solecism without doing so’, μὴ ποιοῦντα φαίνεσθαι). The interlocutor then reconsiders and comes out with μῆνιν οὐλόμενον, as if this were correct (‘You're right, we should say μῆνιν οὐλόμενον’): he now commits a solecism, but no longer seems to do so (the third type: to ‘commit a solecism, without seeming to do so’, ποιοῦντα μὴ δοκεῖν) – thanks, that is, to the clever, but faulty, reasoning of the sophist, against which the Sophistici Elenchi teaches you to be on your guard.
The second way of reading the passage brings in the difference between φαίνεσθαι and δοκεῖν left undiscussed so far.Footnote 12 To be sure, it is not inconceivable that these verbs are indeed used interchangeably here, and at first sight there even seem to be good reasons to think so: just as Aristotle's μὴ ποιοῦντα φαίνεσθαι appears to be echoed by ὁ δὲ “οὐλόμενον” φαίνεται μέν, ἀλλ᾿ οὐ σολοικίζει [sc. κατὰ Πρωταγόραν], so the alternative possibility πoιοῦντα μὴ δοκεῖν would be taken up by ὁ μὲν γὰρ λέγων “οὐλομένην” σολοικίζει μὲν κατ᾿ ἐκεῖνον, οὐ φαίνεται δὲ τοῖς ἄλλοις, thereby leading to a real δοκεῖν ~ φαίνεσθαι equivalence. In reality, though, the wording may not be haphazard at all. Whereas φαίνεσθαι refers to an appearance that (intuitively) ‘meets the eye’, without a reflective process being involved, the same need not be true of δοκεῖν, which typically presupposes a stage in which the opinion or ‘belief’ in question (δόξα) is mentally formed. This being so, it may well be that Aristotle's reference to Protagoras is meant to illustrate only the third possibility (ποιοῦντα μὴ δοκεῖν), to which it is actually attached, but not the second (μὴ ποιοῦντα φαίνεσθαι). Just as there are obviously instances of the first possibility (ποιεῖν – it is a simple given that people make (objective) grammatical mistakes from time to time), so it can be taken for granted that there are also situations where people intuitively object to some linguistic usage, although in reality (i.e., in the eyes of someone with the appropriate grammatical training) there is no reason for doing so – in other words, μὴ ποιοῦντα in μὴ ποιοῦντα φαίνεσθαι refers to an ‘objective’ absence of mistakes (one with which Aristotle would agree), not to a subjective/sophistic absence of mistakes as in Protagoras’ μῆνιν οὐλόμενον (with which Aristotle himself does not agree). By contrast, in the case of the third possibility Aristotle focuses on exactly the kind of situation for which Protagoras’ wild doctrine can serve as a prime example, a situation where there is ‘objectively’ a mistake (ποιοῦντα) but where a sophistic argumentative process nevertheless leads to the δόξα that there is none. Aristotle's double use of φαίνεται in the sentence starting with ὁ μὲν γὰρ λέγων “οὐλομένην” is thus somewhat different from the use of φαίνεσθαι in the formulation of the second possibility (μὴ ποιοῦντα φαίνεσθαι): in the latter case it undoubtedly refers to an (according to Aristotle) erroneous view, whereas in the former case it refers to a view that is shared by Aristotle as one of the ἄλλοι (since opposed to Protagoras’ odd idea) and hence actually correct. In other words, the entire difficulty results from the fact that where φαίνεσθαι is first used it contrasts with an unqualified ‘objective’ truth, and therefore must be understood as entailing a ‘mere appearance’, whereas where it is subsequently used it contrasts with another opinion – that of Protagoras – without the entailment that the ‘appearance’ is a ‘mere appearance’ (except of course from Protagoras’ point of view).
2.2 Protagoras’ reasoning
Moving on, the fundamental question is of course how Protagoras actually argued his case. It is here that the real controversy begins, since Aristotle does not explain Protagoras’ reasoning in any detail. Perhaps prompted by Aristotle's own interest in mismatches between grammatical gender and biological sex,Footnote 13 some scholars, including Gilbert Murray, Glenn Most, Ineke Sluiter and Donatella Di Cesare, claim that Protagoras’ point was purely semantic and took into account the supposedly ‘masculine’ nature of μῆνις as an emotion and of πήληξ as a piece of warfare equipment.Footnote 14 Others, however, argue that more formal considerations may have played a role. Thus Gomperz, echoed by Wackernagel and Fehling,Footnote 15 suggests that πήληξ, at least, may have been classified as masculine by Protagoras on analogy with other masculine nouns in -αξ (ὁ θώραξ, ὁ πόρπαξ, ὁ στύραξ), though they maintain that μῆνις was designated masculine because of its meaning only, since nouns in -ις are generally feminine.Footnote 16 Yet others, including W. K. C. Guthrie and Corradi, suggest that formal distinctions may in fact have been paramount for Protagoras, arguing that what matters is that both μῆνις and πήληξ end in -s.Footnote 17 Finally, there are those who leave the matter open, including G. B. Kerferd, Louis-André Dorion, Rademaker, Myriam Hecquet and Anna Novokhatko, suggesting that both types of reasoning may have played a role.Footnote 18
In our view, the third group of scholars is correct and Protagoras’ argument did indeed have to do with ‘morphological’ consistency (in scare quotes, because, as will become clear, Protagoras did not have the concept of a morpheme, but looked at the final letters of words). One key piece of evidence concerns a strange chapter in Aristotle's Poetics:
αὐτῶν δὲ τῶν ὀνομάτων τὰ μὲν ἄρρενα τὰ δὲ θήλεα τὰ δὲ μεταξύ, ἄρρενα μὲν ὅσα τελευτᾷ εἰς τὸ Ν καὶ Ρ καὶ Σ καὶ ὅσα ἐκ τούτου σύγκειται (ταῦτα δ’ ἐστὶν δύο, Ψ καὶ Ξ), θήλεα δὲ ὅσα ἐκ τῶν φωνηέντων εἴς τε τὰ ἀεὶ μακρά, οἷον εἰς Η καὶ Ω, καὶ τῶν ἐπεκτεινομένων εἰς Α⋅ ὥστε ἴσα συμβαίνει πλήθει εἰς ὅσα τὰ ἄρρενα καὶ τὰ θήλεα⋅ τὸ γὰρ Ψ καὶ τὸ Ξ σύνθετά ἐστιν. εἰς δὲ ἄφωνον οὐδὲν ὄνομα τελευτᾷ, οὐδὲ εἰς φωνῆεν βραχύ. εἰς δὲ τὸ Ι τρία μόνον, μέλι κόμμι πέπερι. εἰς δὲ τὸ Y πέντε **. τὰ δὲ μεταξὺ εἰς ταῦτα καὶ Ν καὶ Σ. (Arist. Poet. 1458a8–17)
Of the nouns themselves, some are masculine, some feminine and some neuter. Masculine are all that end in -Ν and -Ρ and -Σ and all that end in a compound of Σ (there are two of these, Ψ and Ξ). Feminine are all that end in the vowels which are always long, that is -Η and -Ω, and those that end, from among the vowels that can be lengthened, in -Α. The result is that the number of masculine and feminine terminations is the same; for Ψ and Ξ are compounds. No noun ends in a mute consonant or in a short vowel [i.e. a vowel that can only be short: Ε and Ο]. Only three end in -Ι, μέλι honey, κόμμι gum and πέπερι pepper. Five end in -Y, **. The neuters end in these and in -Ν and -Σ.
This chapter sits oddly in the Poetics and has been thought to be an interpolation,Footnote 19 although it does use the Aristotelian term for ‘neuter’, μεταξύ; given the evidence of SE, Protagoras would have said σκεύη (on which see below, section 2.3), while later grammatical theory opted for οὐδέτερον. In any case, it presents the theory that the gender of nouns (ὀνόματα) is determined by their final letter: words ending in -Ν, -Ρ, -Σ, -Ψ and -Ξ are masculine, those ending in -Η and -Ω and long and short -A are feminine, and words that end in -Ι and -Υ are neuter, as are some words that end in -Ν and -Σ. (Some editors at the suggestion of Ingram Bywater read the final sentence as τὰ δὲ μεταξὺ εἰς ταῦτα <καὶ Α> καὶ Ν <καὶ Ρ> καὶ Σ, thus making the text better fit our own grammatical knowledge; the addition of Α also helps explain the term μεταξύ: the neuter partakes of the terminations of both other categories.)
The main reason to connect this passage to a doctrine of Protagoras is that it neatly fits with what we know the sophist said about the gender of μῆνις and πήληξ, and that it also ties in with the parody of (what everyone has always assumed are Protagoras’) ideas on gender in Aristophanes’ Clouds (658–93; cf. section 2.5). Moreover, since the doctrine put forward is so patently wrong from a descriptive point of view, it is best understood as originating from a prescriptive perspective, and that would seem to be much more in line with Protagoras’ evident wish to baffle his audience than with sober Aristotelian scholarship.Footnote 20 But we can go further than that and make two additional points.
First, despite (or because of) its wrong-headedness, which culminates in internal contradiction – if ‘masculine are all [nouns] (ὅσα) that end in -Ν and -Ρ and -Σ’, there cannot logically be another category of nouns equally ending in -Ν and -Σ, as stipulated at the end – the list is highly instructive. There is one category of nouns for which the terminations of masculine and feminine words work very well, namely proper names (male names: e.g. Σίμων, Μέντωρ, Περικλῆς, Φοῖνιξ, Κώνωψ; female names: Ξανθίππη, Σαπφώ, Κόριννα, Ἀσπασία). We would now argue that proper names are exactly the right place to start for a linguist who for the first time attempts to show that certain grammatical distinctions in a language are related to the concept of ‘gender’ (and not, say, ‘size’ or whatever). They offer a natural bridge between form and meaning, as is neatly highlighted by the parodic scene in Clouds where personal names do indeed play a central role (though of course also because they can serve as a cue to ridicule specific individuals: Ar. Nub. 680–93). Protagoras’ reasoning may thus have been as follows: you take the personal names and show that they quite consistently terminate in certain letters when they are names for men, and in other letters when they are names for women.Footnote 21 You then extend the formal criterion to all nouns, including μῆνις and πήληξ.Footnote 22 As for the neuter, Protagoras may have started by enumerating the (limited number of) words that terminate in the remaining letters in which Greek words may end, namely -Ι and -Y, and concluded that they all refer to ‘things’.Footnote 23 (We will return below to neuter words ending in -Ν and -Σ.) Such a transfer from names to nouns more generally was no doubt aided by the fact that the term ὄνομα could refer to either of the two groups of words – so that, as long as the ‘right’ meaning of ὄνομα is picked, even the claims that ‘all the (ὅσα) ὀνόματα [= “proper names”] in -Ν, -Ρ, -Σ are masculine’ and ‘the neuter ὀνόματα [= “nouns”] end in -Ι, -Y, -Ν, -Σ’ are no longer contradictory.Footnote 24
Secondly, it may be significant that the issue of nouns ending in -Ξ and -Ψ is of some relevance to the argument of the Poetics passage, for the balance in number between masculine and feminine nouns/names only works as long as the items in -Ξ and -Ψ are treated as a subcategory of those in -Σ. That in itself is a perfectly reasonable view, which could originate with Aristotle just as easily as with Protagoras.Footnote 25 But only if the entire doctrine ultimately went back to the latter could we finally also make out why Protagoras, in addition to μῆνις, chose πήληξ to illustrate his point. Being the first word of the Iliad, feminine μῆνις was of course a godsend to capture everyone's attention; but if the whole point was to strike the audience with ‘wrong’ feminines, why go for πήληξ next, rather than a high-frequency word such as πόλις or χάρις (each of which is also more common in Homer than πήληξ is)? Unlike these, however, πήληξ was the ideal example if there was at the same time a need to come up with an item ending in one of the ‘compound letters’; for of all the Homeric nouns in -ξ or -ψ, πήληξ with its ten attestations (of which several with an unambiguously feminine attribute: e.g. Il. 13.527 πήληκα φαεινήν, 16.104–5 φαεινὴ πήληξ βαλλομένη) was by far the most common.
2.3 Protagoras’ terminology
There are several additional considerations which make it plausible that Protagoras’ argument was in the first place a formal one, and some of them offer tantalising further possibilities to follow his reasoning. The first concerns Protagoras’ terminology and is very simple; yet, it has been curiously overlooked so far. By themselves, Protagoras’ terms ἄρρην ‘male > masculine’ and θῆλυς ‘female > feminine’ seem straightforward enough, both because of the likely role played by male and female names in the development of the theory (section 2.2) and also, more generally, because at least the most prototypical members of the respective declension classes do designate male and female beings, respectively (e.g. ἀνήρ/γυνή, δεσπότης/δέσποινα etc.). The same may then also be said about Protagoras’ term for the ‘neuter’, σκεῦος: prototypical neuters are indeed σκεύη (‘implements, utensils’). But if it were true that Protagoras’ approach had been a semantic one, this would lead to a glaring paradox. Whatever one may think about emotions and their possible associations with men or women (cf. μῆνις), Protagoras’ argument about πήληξ simply could no longer stand: since the noun σκεῦος often specifically refers to pieces of equipment, including military gear,Footnote 26 it would have been nothing but perverse for Protagoras to object to the current classification of πήληξ among the θήλεα, only to then place that word among the ἄρρενα and not among the σκεύη where it so evidently belonged if the meaning was anything to go by. In other words, the basis for claiming that πήληξ is really masculine rather than feminine must have been unrelated to semantics.
2.4 The continuation of Sophistici Elenchi
Next, immediately following the passage from Aristotle's Sophistici Elenchi found in editions of the fragments of Protagoras, there is a further discussion of ‘seeming solecisms’ involving both gender and inflection (πτῶσις) (173b26–174a9). The sophistic line of reasoning which Aristotle dissects relies on the fact that neuter words have the same form in the nominative and accusative, whereas masculine and feminine words do not. Thus, in both ‘it is a log (ἔστι ξύλον)’ and ‘(I believe it) to be a log (εἶναι ξύλον)’ ξύλον has the same form, but that does not hold for Κορίσκος in ‘ἔστι Κορίσκος’ and ‘εἶναι Κορίσκον’. According to an ingenious set of questions which Aristotle spells out only later (182a10–182b2), this discrepancy allows sophists to pretend that an interlocutor saying (accusative) Κορίσκον is in fact claiming that it should be τὸ Κορίσκον. For our purposes, however, it is more noteworthy that Aristotle, after saying that masculine words like Κορίσκος have different forms in all cases (αἱ πτώσεις ἅπασαι), adds:
καὶ ἐπὶ τῶν θήλεων ὀνομάτων ὡσαύτως, καὶ ἐπὶ τῶν λεγομένων μὲν σκευῶν, ἐχόντων δὲ θηλείας ἢ ἄρρενος κλῆσιν. ὅσα γὰρ εἰς τὸ Ο καὶ τὸ Ν τελευτᾷ, ταῦτα μόνα σκεύους ἔχει κλῆσιν, οἷον “ξύλον”, “σχοινίον”⋅ τὰ δὲ μὴ οὕτως ἄρρενος ἢ θήλεος, ὧν ἔνια φέρομεν ἐπὶ τὰ σκεύη, οἷον “ἀσκὸς” μὲν ἄρρεν τοὔνομα, “κλίνη” δὲ θῆλυ. (SE 173b39–174a4)
It is the same with feminine nouns and with the words that are called ‘things’ but have a feminine or masculine designation. For only those that end in -O- plus -N have the designation of ‘things’, such as ξύλον log and σχοινίον rope. If they do not, they have a masculine or feminine one, and some of these we apply to ‘things’, such as the masculine noun ἀσκός wine-skin and the feminine noun κλίνη bed.
So far we have only come across evidence to suggest that Protagoras limited his speculations about the connections between grammatical form and gender to the final letters of words (section 2.2). However, in connection with the transfer from names to nouns we have also already seen that, while the ‘rule’ he postulated served his purpose in some respects, it has various shortcomings. Of course, Protagoras was not a descriptive linguist, but rather used his ‘rule’ to make prescriptive pronouncements.Footnote 27 But even within such a framework, there are limits to what one can get away with. The most glaring problem concerns the neuter. After all, there are really very many neuter words ending in -(ο)ν and also many in -(ο)ς, which according to Protagoras’ ‘rule’ should all be masculine. It is therefore interesting to note that in this passage of SE Aristotle takes up the point of neuter terminations and uses the word Protagoras used as his term for ‘neuter’, σκεῦος (which, incidentally, itself infringes the ‘rule’). To account for this, we would tentatively suggest that the passage is based on an elaboration of Protagoras’ theory, moving from a consideration of the final letter of words to a consideration of the final two letters; and it is at least conceivable that already Protagoras himself had found himself forced to make this concession and refine his claims.
2.5 The parody in Aristophanes’ Clouds
Furthermore, in the passage just cited Aristotle shifts the focus away from a consideration of the nominative form alone, in order to expose the trap used by the sophistic ἔλεγχος. Instead, he suggests, attention may profitably be paid to the declension pattern as a whole, involving nominatives (~ κλῆσις, the case with ἔστι), accusatives (the case with εἶναι), as well as other cases (cf. αἱ πτώσεις ἅπασαι). Once that is done, it is no longer possible for example to treat the neuter nominative τοῦτο in τί τοῦτο; ‘What is this?’ as an ‘equivalent’ of, or proxy for, the masculine accusative τοῦτον – as the sophists whom he rebukes appear to have done in order to confuse their interlocutors. No one is specifically named here, but in view of the preceding discussion of Protagorean sophistry, it would not surprise if Protagoras had also been guilty of such kinds of trickery.
This suspicion becomes even stronger once we realise that Aristophanes staged a surprisingly similar kind of sophistic trap in the grammar scene of Clouds, whose Protagorean inspiration is generally acknowledged. We will first discuss this similarity, before showing that the exchange leading up to it also has ‘Socrates’ operating like a grammarian whose interest in the form of words trumps semantic considerations:
Σω. ἔτι δέ γε περὶ τῶν ὀνομάτων μαθεῖν σε δεῖ,
ἅττ’ ἄρρεν’ ἐστίν, ἅττα δ’ αὐτῶν θήλεα.
Στ. ἀλλ’ οἶδ’ ἔγωγ’ ἃ θήλε’ ἐστίν.
Σω. εἰπὲ δή.
Στ. Λύσιλλα, Φίλιννα, Κλειταγόρα, Δημητρία.
Σω. ἄρρενα δὲ ποῖα τῶν ὀνομάτων;
Στ. μυρία. 685
Φιλόξενος, Μελησίας, Ἀμυνίας.
Σω. ἀλλ’ ὦ πόνηρε, ταῦτά γ’ ἔστ’ οὐκ ἄρρενα.
Στ. οὐκ ἄρρεν’ ὑμῖν ἐστιν;
Σω. οὐδαμῶς γ’, ἐπεὶ
πῶς ἂν καλέσειας ἐντυχὼν Ἀμυνίᾳ;
Στ. ὅπως ἄν; ὡδί⋅ “δεῦρο δεῦρ’, Ἀμυνία.” 690
Σω. ὁρᾷς; γυναῖκα τὴν Ἀμυνίαν καλεῖς.
Στ. οὔκουν δικαίως, ἥτις οὐ στρατεύεται;
ἀτὰρ τί ταῦθ’ ἃ πάντες ἴσμεν μανθάνω;
(Ar. Nub. 681–93)
SO. Also, you have to learn which names are male and which are female.
ST. But I know which ones are female!
SO. Go ahead then.
ST. Lysilla, Philinna, Kleitagora, Dêmêtria.
SO. And which names are male?
ST. An awful lot: Philoxenos, Melêsias, Amynias …
SO. Silly you, these are not male!
ST. They are not male according to you?
SO. No, not at all: how would you call Amynias when you see him?
ST. How? Like this: “Hello, hello, Amynia!”
SO. You see? You are calling a woman, “Amynia”!Footnote 28
ST. Isn't that right, given that she doesn't do military service? Why do I have to learn these things we all know anyway?
Although he is eventually outwitted by Strepsiades’ simple mind, ‘Socrates’ here seeks to trip up his pupil in exactly the way Aristotle's discussion suggests: by glossing over the fact that nouns/names are grammatically defined by their declension pattern, and that the formal identity of one case form to another case form (in Aristotle, the neuter nominative τοῦτο and the neuter accusative τοῦτο; in Aristophanes, the masculine vocative in -α and the feminine nominative in -α) does not of its own warrant any conclusion about the function which the form in question has.
Importantly, meanwhile, ‘Socrates’’ argumentation is an exclusively formal one both in this case and throughout the scene. It is therefore misleading to suggest that the Aristophanic passage lends support to the idea that Protagoras’ classification criteria, when dealing with μῆνις and πήληξ, may have been a semantic as well as a formal one.Footnote 29 Of course, Aristophanes’ primary aim is to make his audience laugh, not to give a glimpse into a real sophist's classroom. Even so it is telling that the only stage character who takes semantics as the baseline when asked to think about grammatical gender is the uneducated Strepsiades. It is he who ultimately ‘justifies’ the seemingly feminine form of address Ἀμυνία by reference to the historical Amynias’ effeminate character. ‘Socrates’, by contrast, does not argue along the lines of ‘Amynias is effeminate, hence his name must be feminine’ (as would be implied by the allegedly Protagorean argument that ‘μῆνις is a male feeling, hence its name must be masculine’).
The same is true in the other two sections of the Aristophanic scene. As already remarked, the first deals with zoonyms:
Σω. ἀλλ’ ἕτερα δεῖ σε πρότερα τούτου μανθάνειν,
τῶν τετραπόδων ἅττ’ ἐστὶν ὀρθῶς ἄρρενα.
Στ. ἀλλ’ οἶδ’ ἔγωγε τἄρρεν’, εἰ μὴ μαίνομαι⋅ 660
κριός, τράγος, ταῦρος, κύων, ἀλεκτρυών.
Σω. ὁρᾷς ἃ πάσχεις; τήν τε θήλειαν καλεῖς
ἀλεκτρυόνα κατὰ ταὐτὸ καὶ τὸν ἄρρενα.
Στ. πῶς δή, φέρε;
Σω. πῶς; ἀλεκτρυὼν κἀλεκτρυών.
Στ. νὴ τὸν Ποσειδῶ. νῦν δὲ πῶς με χρὴ καλεῖν; 665
Σω. ἀλεκτρύαιναν, τὸν δ’ ἕτερον ἀλέκτορα.
Στ. ἀλεκτρύαιναν; εὖ γε, νὴ τὸν Ἀέρα⋅
ὥστ’ ἀντὶ τούτου τοῦ διδάγματος μόνου
διαλφιτώσω σου κύκλῳ τὴν κάρδοπον.
(Ar. Nub. 658–69)
SO. But before that you have to learn other things, which of the four-footed beings are truly male.
ST. But I know the male ones, unless I'm mad: ram, he-goat, bull, dog, cock (alektryôn).
SO. See what's happening to you? You call the female alektryôn just like the male.
ST. How that, come on?
SO. How? Well, alektryôn and alektryôn!
ST. By Poseidon, and how am I supposed to call it now?
SO. alektryaina, and the other one alektôr.
ST. alektryaina? That's a good one, by Air! For that piece of instruction alone I will fill your kneading-trough with meal, all around!
Once again, Socrates does not proceed by saying ‘A hen is female, so the word we normally use for hen (ἀλεκτρυών) must also be treated as grammatically feminine.’ In fact, that is precisely what he objects to: ἀλεκτρυών cannot be grammatically feminine because its form (ending in -(ω)ν, rather than e.g. -(αιν)α) is such that it is incompatible with femininity.Footnote 30 And once Strepsiades has inadvertently given the cue by saying τὴν κάρδοπον, the discussion continues accordingly:
Σω. ἰδοὺ μάλ’ αὖθις, τοῦθ’ ἕτερον⋅ τὴν κάρδοπον 670
ἄρρενα καλεῖς θήλειαν οὖσαν.
Στ. τῷ τρόπῳ;
ἄρρενα καλῶ ’γὼ κάρδοπον;
Σω. μάλιστά γε,
ὥσπερ γε καὶ Κλεώνυμον.
Στ. πῶς δή; φράσον.
Σω. ταὐτὸν δύναταί σοι κάρδοπος Κλεωνύμῳ.
Στ. ἀλλ’, ὦγάθ’, οὐδ’ ἦν κάρδοπος Κλεωνύμῳ, 675
ἀλλ’ ἐν θυείᾳ στρογγύλῃ γ’ ἀνεμάττετο.
ἀτὰρ τὸ λοιπὸν πῶς με χρὴ καλεῖν;
Σω. ὅπως;
τὴν καρδόπην, ὥσπερ καλεῖς τὴν Σωστράτην.
Στ. τὴν καρδόπην, θήλειαν;
Σω. ὀρθῶς γὰρ λέγεις.
Στ. ἐκεῖνο δ’ ἦν ἂν ‘καρδόπη Κλεωνύμῃ’. 680
(Ar. Nub. 670–80)
SO. Again, see? another example: you call the kneading-trough (kardopos) a male although it is female.
ST. In what way? I call a kneading-trough male?
SO. Yes, indeed, just like Kleonymos.
ST. How do you mean, tell me.
SO. For you a kneading-trough is the equivalent of Kleonymos.
ST. But, mate, Kleyonymos didn't have a kneading-trough – he kneaded in a round mortar. So what do I have to say in the future?
SO. What you have to say? Well, kardopê, just as you say Sôstratê.
ST. Female kardopê?
SO. Yes, correct.
ST. So it would be ‘a kardopê for Kleônymê’!
In this case we get yet another variation on the same theme. The feminine gender of the word for ‘kneading-trough’ is established, not on the basis of a supposed association of the implement with women – in the same way as the association of a πήληξ with men has been alleged to work – but on the basis of the formal characteristic of the καρδοπ- lexeme taking a feminine article; so agreement is brought in, just as it was by the real Protagoras in his discussion of the combination of μῆνιν with οὐλομένην/οὐλόμενον. To be sure, the argument is not then quite the same as Protagoras’ Iliadic one (because for Protagoras Homer's use of οὐλομένην was clearly not a reason to regard μῆνις as feminine, let alone turn it into something like *μηνίη), but the essential point is still that semantic considerations are lacking.Footnote 31 Had these been of any importance to Protagoras’ doctrine about gender, it would have been easy enough for the comic poet to capitalise on them. Their absence from Clouds therefore reinforces the impression that Protagoras’ focus must have been a formal one.
2.6 The introduction of Sophistici Elenchi
Finally, we must beware that collections of fragments are treacherous things: they make scholars ignore the wider context of the text from which they have been taken, even though that context may be highly relevant to the interpretation of the fragment. This is the case with the passage from the Sophistici Elenchi as well. Any remaining doubt that Protagoras used a formal argument to maintain that μῆνις and πήληξ are masculine is dispelled when we consider the context in which Aristotle discusses the claim.
It is especially fruitful to look at the introduction of SE, in which Aristotle gives a brief summary of its contents. In describing what he will do in ‘our’ section, he mentions fallacious refutations which depend on the language employed (ἔλεγχοι παρὰ τὴν λέξιν), and the final sub-category of those refutations are those that depend on ‘the form of the expression’:
οἱ δὲ παρὰ τὸ σχῆμα τῆς λέξεως συμβαίνουσιν ὅταν τὸ μὴ ταὐτὸ ὡσαύτως ἑρμηνεύηται, οἷον τὸ ἄρρεν θῆλυ ἢ τὸ θῆλυ ἄρρεν ἢ τὸ μεταξὺ θάτερον τούτων, ἢ πάλιν τὸ ποιὸν ποσὸν ἢ τὸ ποσὸν ποιόν, ἢ τὸ ποιοῦν πάσχον ἢ τὸ διακείμενον ποιοῦν, καὶ τἆλλα δ’ ὡς διῄρηται πρότερον⋅ ἔστι γὰρ τὸ μὴ τῶν ποιεῖν ὂν ὡς τῶν ποιεῖν τι τῇ λέξει σημαίνειν. οἷον τὸ ὑγιαίνειν ὁμοίως τῷ σχήματι τῆς λέξεως λέγεται τῷ τέμνειν ἢ οἰκοδομεῖν⋅ καίτοι τὸ μὲν ποιόν τι καὶ διακείμενόν πως δηλοῖ, τὸ δὲ ποιεῖν τι. τὸν αὐτὸν δὲ τρόπον καὶ ἐπὶ τῶν ἄλλων. (Arist. SE 166b10–19)
The [refutations] which depend on the form of expression occur when what is not the same is expressed in the same way. For example, when a masculine/male [item is expressed as] a feminine, a feminine/female [item] as a masculine or a neuter one as either of the other two. Or again, if a quality [is expressed as] a quantity or a quantity as a quality, or an action as a passive or a disposition as an action, and so forth according to the distinctions made earlier. For it is possible for something which does not belong to the category of ‘actions’ to signify by the expression used that it belongs to the category of ‘actions’. For instance, ‘to be healthy (ὑγιαίνειν)’ is expressed by means of the same form of expression as ‘to cut (τέμνειν)’ or ‘to build a house (οἰκοδομεῖν)’. Yet, the former denotes some quality and disposition, the latter [two] an action. So, too, with the other examples.
The term σχῆμα τῆς λέξεως has been much debated; it does not refer to the same concept in all of Aristotle's works.Footnote 32 In this context it must refer to the form of single words or, more precisely, to that part of the form of single words to which meanings such as ‘masculine’ or ‘feminine’, or ‘action’ or ‘disposition’, can, according to the argument under discussion, be attached. This is the only conclusion that can be drawn from the example which Aristotle gives: the sole ‘form of expression’ which ὑγιαίνειν, τέμνειν and οἰκοδομεῖν have in common is the ending -ειν. The kinds of meaning which are envisaged as being attached to a σχῆμα τῆς λέξεως such as -ειν are taken from Aristotle's own ‘categories’.Footnote 33 Yet, an actual sophistic ἔλεγχος may still underlie his words. The first step of such an ἔλεγχος would be the sophist's observation that verbs ending in -ειν typically denote ‘actions’; judging by the examples τέμνειν and οἰκοδομεῖν, perhaps something like ‘transitive verbs’ was meant by this. The next step would be to generalise this observation and make the interlocutor concede that -ειν means ‘action’. Next, the sophist would make the unsuspecting interlocutor define the meaning of ὑγιαίνειν, which would no doubt be returned as (non-transitive) ‘be healthy’; after all, that is what the verb means. The refutation which has now become possible consists in the sophist's claim that the interlocutor has wrongly defined the meaning of ὑγιαίνειν: it really means ‘make healthy’. (Alternatively, and although Aristotle is not explicit about it, one can also envisage a different ‘refutation’: if the sophist's stipulations also included the claim that verbs ending in -σθαι are intransitive (the opposite of ‘actions’), the refutation could consist in the claim that, in saying ὑγιαίνειν, the interlocutor made a mistake: he should really have said *ὑγιαίνεσθαι.)
We need not assume that Protagoras himself came up with this particular ἔλεγχος, but he does seem to have been the originator of this type of argument. One may therefore envisage a very similar reasoning to underlie the point about μῆνις/πήληξ being masculine.
3 Protagoras on grammatical mood
Diogenes Laertius credits Protagoras with a fourfold division of ‘speech’ (λόγον) in a section on the sophist's εὑρήματα.Footnote 34 The passage of Aristotle hints at a specific application of the πυθμένες λόγων on the part of Protagoras. In particular, it suggests that the sophist maintained that Homer made a mistake in the first three words of the Iliad: they constitute an ‘order’, whereas only a ‘prayer’ would have been appropriate in addressing a goddess.
So, what are the πυθμένες λόγων?Footnote 35 On one interpretation, best represented by Schenkeveld (Reference Schenkeveld1984) 293, by using terms such as ‘prayer’, ‘question’, ‘answer’ and ‘command’ Protagoras ‘casually indicated ways of expressing oneself’. Accordingly, Protagoras’ point concerned a sort of rudimentary idea of ‘speech acts’ (Rademaker (Reference Rademaker, van Ophuijsen, van Raalte and Stork2013) 87) or the communicative value of different ‘kinds of sentences’ (Sluiter (Reference Sluiter1990) 7).Footnote 36 On another interpretation, however, it has been pointed out that the πυθμένες λόγων ‘seem to correspond to typical uses of the four moods’ (Rademaker (Reference Rademaker, van Ophuijsen, van Raalte and Stork2013) 92), so that the terms ‘prayer’, ‘question’, ‘answer’ and ‘command’ might be seen as ‘corresponding roughly to optative, subjunctive, indicative and imperative moods of the verb, here perhaps being formally distinguished for the first time’ (Graham (Reference Graham2010) ii.721).
Again, we think that the least one can say is that the πυθμένες λόγων must be defined with reference to the moods. Aristotle's (non-exhaustive) list of σχήματα λέξεως (here meaning something different from SE) introduces a certain amount of noise, but what we learn from him about Protagoras is unambiguous: the only basis on which μῆνιν ἄειδε θεά can be called an ‘order’ is the presence of an imperative and, if so, the most obvious way in which it could be turned into a ‘prayer’ is by using an optative instead. Homer, so Protagoras’ argument must have run, made a mistake when he said ἄειδε; he should have said ἀείδοις. This gives rise to the assumption that the πυθμένες λόγων map onto typical functions of the four grammatical moods or, more accurately, onto the only ‘true’ function of each mood; Protagoras’ argument depends on a generalisation of the observation that the imperative is typically used in orders to the claim that the imperative can or should only be used in orders.Footnote 37 The following associations may be envisaged: the optative (ἀείδοις ‘may you sing’) corresponds to εὐχωλή, the subjunctive (ἀείδωμεν; ‘should we sing?’) to ἐρώτησις, the indicative (ἀείδεις ‘you sing’) to ἀπόκρισις and the imperative (ἄειδε ‘sing!’) to ἐντολή.
One indication are the names which Protagoras gave to the moods. ‘Prayer’ (εὐχωλή) and ‘command’ (ἐντολή) immediately suggest the optative and the imperative, while the indicative is the prototypical mood of (a standard type of) ‘answer’ (ἀπόκρισις).Footnote 38 Only the proposed correspondence between the subjunctive and ‘question’ (ἐρώτησις) may seem problematic.Footnote 39 For although several scholars have guessed that the sophist must have had the subjunctive in deliberative questions in mind, this solution at first sight looks like a cop-out, because the use of the subjunctive in this environment hardly represents a typical usage; later grammatical theory called the mood ὑποτακτική, and this term would seem to reflect a much more central characteristic of the subjunctive, namely its use in subordinate clauses.
Having said that, we think that it is actually only natural that Protagoras should have turned to the deliberative subjunctive. What needs to be kept in mind is that the use of the subjunctive in subordinate clauses is nearly always obligatory, that is, determined by the syntactic rules of the language and not immediately by its semantics. This makes it difficult to assign a specific meaning to the subjunctive in that environment; one cannot, for instance, replace the subjunctive with another mood to see how the meaning changes, because in most cases all one will have produced is something ungrammatical and meaningless. It is reasonable to assume, therefore, that a pioneering linguist like Protagoras focused on environments in which mood choice does have immediate consequences for the meaning of sentences, that is, on main clauses. If this is correct, he could have had three uses in mind:Footnote 40 (1) the deliberative use; (2) the adhortative use (ἔλθωμεν ‘let's go’); (3) the prohibitive use (only aorist: μὴ ἔλθῃς ‘don't go’). The last two of these were unsuitable because of their jussive force, which Protagoras (naturally) assigned to the imperative. Moreover, the criterion of substitution is almost as useless in the case of the prohibitive subjunctive as in the case of subordinate clauses; *μὴ ἦλθες and *μὴ ἐλθέ are both ungrammatical. Only the deliberative subjunctive could therefore serve to complete the system.
4 Further grammatical distinctions?
Since the scope of Protagoras’ grammatical analyses thus encompassed both the fundamental category of nominal gender and the equally fundamental category of verbal mood, we may wonder if he did not also consider the most important counterpart of each, that is, nominal case and verbal tense, respectively.Footnote 41 As for the former, there is no positive evidence and, given the way in which Aristotle critically exposes the disregard for the nominative vs accusative distinction in the sophistic ἔλεγχοι (cf. section 2.4), this need not be a coincidence: perhaps Protagoras really had nothing to say on the matter. In the light of what has just been remarked about mood distinctions, this would not be surprising. Unlike the substitution of one mood by another in main clauses, or indeed the substitution of one gender by another in the case of pairs such as βασιλεύς/βασίλεια, Ἀθηναῖος/Ἀθηναία or Κλεώνυμος/Κλεωνύμη, the substitution of one case by another would not, in a standard sentence, produce a semantically different, but grammatically still acceptable, outcome. Even a sophist cannot readily attack Homer for saying μῆνιν ἄειδε instead of, say, *μήνεως ἄειδε.
With tense, on the other hand, the situation is similar to the one with mood. There would be nothing grammatically wrong if, for instance, instead of μῆνιν ἄειδε, θεά, … ἣ μυρί᾿ Ἀχαιοῖς ἄλγε᾿ ἔθηκε, Homer had sung …ἄλγεα τίθησι. Against this background, a final passage, this time from Diogenes Laertius’ sketch of Protagoras’ life, deserves attention:
oὗτος πρῶτος μισθὸν εἰσεπράξατο μνᾶς ἑκατόν⋅ καὶ πρῶτος μέρη χρόνου διώρισε καὶ καιροῦ δύναμιν ἐξέθετο καὶ λόγων ἀγῶνας ἐποιήσατο καὶ σοφίσματα τοῖς πραγματολογοῦσι προσήγαγε⋅ καὶ τὴν διάνοιαν ἀφεὶς πρὸς τοὔνομα διελέχθη (D. L. 9.52 = Protagoras DK 80 A 1.52 = Radermacher B iii 24 = Protagoras D20 Laks/Most)
He was the first to exact a fee of 100 minae, and the first to distinguish parts of time, to set forth the power of kairos, to institute contests in debating, and to teach rival pleaders the tricks of their trade. And, setting aside the meaning, he argued with a view to the [form of the] word.
The difficult last clause, if understood in this way, rather neatly encapsulates Protagoras’ formal take on language.Footnote 42 Here, however, we focus on the preceding clause, with its reference to μέρη χρόνου. This has been widely understood as hinting at a distinction of tenses,Footnote 43 and not just because it is so reminiscent of terms like Aristotle's μέρη λέξεως ‘parts of expression’ (Arist. Poet. 20 = 1456b20–57a30, which discusses word classes as well as other units such as syllables and ‘letters’/phonemes) or the Stoic μέρη λόγου ‘parts of speech’. Against this view, it has been observed that Aristotle was the first to analyse the encoding of time in the verb (Arist. Int. 16b6–18; cf. Poet. 1457a14–18), and that the usual term for ‘tense(s)’ later on was χρόνος/χρόνοι, so that Diogenes Laertius should have written πρῶτος τοὺς χρόνους διώρισε if he had really meant to refer to the grammatical feature.Footnote 44 Neither of these objections is cogent. The first is a mere petitio principii, and the second overlooks that an alternative formulation like the one just cited might have misleadingly suggested that Protagoras already came up with the full inventory of tenses with which later grammarians operate; that, however, need not be the case. If the most immediate reading of μέρη χρόνου – quite possibly Protagoras’ own term, paralleling his πυθμένες λόγων for what were later called ἐγκλίσεις ‘moods’ – is the simple tripartite division into past, present and future,Footnote 45 it is conceivable that this is exactly what Protagoras also saw as reflected in grammar; for Dunn (Reference Dunn2001) 548–9 rightly stresses that Diogenes cannot have meant that Protagoras was the first to notice the existence (in human experience) of a past, present and future as such – which would be banal – nor that Protagoras is likely to have come up with the complex differentiations later philosophers make between the ontological nature of the present as opposed to the past and future. Given the context in which it occurs, Diogenes’ note must refer to something that either left a striking impression (cf. the astronomical fee, mentioned before) or represented an intellectual innovation with which Diogenes’ educated lay readers were familiar themselves (cf. the rhetorical importance of the καιρός, mentioned afterwards).Footnote 46 The recognition that the division of time into past, present and future is somehow reflected in verbal morphology would fit this requirement, and given the prominence and relative complexity of tense distinctions in Greek it would also stand to reason that Diogenes chose to highlight this aspect rather than the (parallel) discovery of mood, let alone nominal gender; after all, ever since Aristotle at least, time (and hence tense) was known to have a special defining status when it came to the understanding of verbs (cf. Arist. Int. 16b6–7, ῥῆμα δέ ἐστι τὸ προσσημαῖνον χρόνον ‘A verb is (an element) which also [sc. in addition to its lexical meaning] signifies time’).
To this, one might of course still object that, while there is no better suited alternative for interpreting Diogenes’ μέρη χρόνου, the ‘grammatical’ reading of the phrase nevertheless remains in the air as long as there is no corroborating evidence of the kind we came across when dealing with genders and moods. However, there may be one piece of information that does deserve a further look in this context. Consider again the Diogenes passage according to which Protagoras ‘was the first to divide speech into four kinds – prayer, question, answer, command (others say into seven: narration, question, answer, command, report, prayer, calling), which he even named “foundations of speech”’. In our earlier discussion we did not dwell on the parenthetical remark, and it is true that the Greek text need not mean that ‘others’ claimed an alternative seven-fold division for Protagoras himself. If need be, it might also be shorthand for ‘while others subsequently came up with a seven-fold division’. Even so, the grammatically most straightforward reading is the former, and it therefore invites reflection.
The three items added to the undisputed four πυθμένες λόγων of ἐρώτησις, ἀπόκρισις, ἐντολή and εὐχωλή are διήγησις, ἀπαγγελία and κλῆσις. To start with the first, if ἀπόκρισις is to be identified with the indicative, as argued above, we may ask in what way it could make sense to distinguish from this something called διήγησις. A typical ‘narration’ is cast in the indicative just as much as a typical ‘answer’ is. However, what does separate a standard ‘narration’ from the kind of defining or specifying ‘answer’ Protagoras is likely to have had in mind (‘Socrates is a man’, ‘Socrates is teaching’; cf. section 3) is its past tense (‘Socrates taught/was teaching’). Seen in this light, and considering the nomenclatory principle Protagoras applied elsewhere – namely to pick out one prototypical usage – διήγησις turns out to be as sensible a proto-grammatical term for ‘past tense’ as εὐχωλή is for the ‘optative’, for example.
Given a three-fold division of past/present/future, that would then at first seem to leave ἀπαγγελία and κλῆσις for the present and future respectively. Yet, ἀπαγγελία means ‘report’ and the standard tense in a report is the past no less than in a narration. Moreover, a ‘calling’ of whatever sort (κλῆσις) is also not something one would easily associate with either a present or a future: the imperative might provide a better fit, but that slot is already occupied by ἐντολή.
At this point, however, two further observations become relevant. The first concerns the textual transmission. The most important manuscripts of Diogenes (BP1F) as well as the corresponding Suda entry (π 3122) give ἐπαγγελίαν, while ἀπαγγελίαν only occurs in manuscripts that have no independent authority and was probably introduced by emendation.Footnote 47 And just as ἐπαγγέλλω means ‘announce, offer, promise’, the basic meaning of ἐπαγγελία is ‘announcement, promise’ – which is a rather neat match for the prototypical function of future tense forms. The second observation concerns the classification itself. From our perspective, tense and mood are independent variables, but at the proto-grammatical stage we are dealing with this need not have been so obvious, not least because the grammatical categories were so much (terminologically) intertwined with corresponding pragmatic functions. Thus, if imperative ἄειδε encapsulated an ἐντολή, optative ἀείδοι(ς) an εὐχωλή, subjunctive ἀείδω;/ἀείδωμεν; an ἐρώτησις and indicative ἀείδει an ἀπόκρισις, and furthermore ‘past’ (imperfect/aorist) ἤειδε/ἤεισε a διήγησις and future ἀείσεται an ἐπαγγελία, it was not immediately clear that present ἀείδει needed a second, temporal, label as well. Instead, ἀπόκρισις may also have represented ‘present’ by implication. Such a conceptualisation would not be out of line with what we come across later, when Aristotle (Int. 16b9–10, 16–17) treats aorist ὑγίανεν and future ὑγιανεῖ differently from present ὑγιαίνει, calling only the latter a verb (ῥῆμα), but the former two merely inflections of the verb (οὐ ῥῆμα ἀλλὰ πτῶσις ῥήματος): here too, then, the present is taken as a base form with a different status from the past and future forms.
If this is accepted, we are only left with the seventh item, κλῆσις. One possibility – but hardly more than that – would be to remember that the later grammatical tradition, as codified in Dionysius Thrax's τέχνη (Grammatici Graeci i.1, p. 47 Uhlig), treated the infinitive as a fifth mood, hence assigned it a status in parallel with the finite forms of the verb. If something like this was already happening here, Protagoras just might have been thinking of syntactically autonomous imperatives in jussive environments, including ‘invitational/cletic’ ones (cf. e.g. Ar. Ach. 172 τοὺς Θρᾷκας ἀπιέναι, παρεῖναι δ᾿ εἰς ἔνην ‘The Thracians are to go away, and to be back the day after tomorrow’; Ar. Vesp. 937 μάρτυρας παρεῖναι ‘as witnesses are to present themselves’). With this addition, the seven-fold system of form/function mappings would thus gain a considerable degree of internal coherence – though for some readers that very fact might perhaps constitute an argument against its historical plausibility. Be that as it may, the upshot of the preceding discussion is a different one anyway. To deny, in the light of what our elusive sources tell us, that Protagoras can have reflected on verbal categories beyond mood is in no way a safer position to take than to accept that he may very well have done so.
5 Conclusion
It hardly needs stressing that much remains unclear and uncertain when we try to reconstruct the earliest stages of Greek linguistic thought. However, against the nowadays prevalent view according to which Protagoras’ endeavours in the domain of grammar were confined to inconsequential side remarks, it has here been argued that all the evidence we have rather points to a fairly sophisticated approach to categories such as nominal gender and verbal mood, which – for the first time – took into account the formal expression of these. In particular, there is no reason to think that Protagoras’ teaching on gender was based on haphazard semantic criteria rather than systematic formal ones, or that his πυθμένες λόγων were identified without reference to modal morphology. Moreover, a tentative case can even be made for at least a basic tense differentiation having been attempted alongside the classification of moods. That all these matters were ultimately meant to feed into literary exegesis, and from there into an understanding of how ‘to guarantee the greatest possible degree of clarity in the use of language’,Footnote 48 may well be true, but it would be wrong to conclude from this that Protagoras could not also be interested in the structure of language in its own right. To be sure, others – including Aristotle, the Stoics and the Alexandrian grammarians – eventually went much further than he had done; but that is what happens to any pioneer and it must not make us undervalue the substantial achievements we are still able to piece together from a tradition and transmission that has been less generous than it could have been to one of the towering intellectuals of the classical age.
