Because I'm about to have a lot of fun with Motorcycle Boy's nickel summary of Greek diglossia, I'm going to post now on a mutant optative (or subjunctive) in Galen. It's my serving of brussels sprouts, as it were.
Those of you that were following The Other Blog three years ago (I know, I know) would have caught sight of my posting there on φαῖο: I unearthed an athematic middle optative of φημί in the new edition of Moschus. Because the grammars documented the old edition of Moschus, and because my command of anything athematic or optative is shaky, I spent an hour trying to work out what it was.
... OK, for the non-classicists among you, once again, slowly. Verbs in Greek are formed by attaching endings to a stem. Normally, a vowel comes between the stem and the ending; it's called a thematic vowel, and is either /o/ or /e/, depending on the person. There is an archaic class of verbs which has no thematic vowels, so they're called athematic. As is usual with irregular verbs, these are high frequency verbs—φημί "say", ἵστημι "stand", τίθημι "put", δίδωμι "give"; so people had learned them by rote, and they stuck around when other verbs got their added vowels.
We're going to look at subjunctives and optatives, and there's two further things to know about those moods. The subjunctive thematic vowels are long, /ɔː/ and /ɛː/, and *all* subjunctives are thematic, whatever their indicatives are. Which is just as well, because that way the subjunctives and indicatives are still distinguishable. So
|Thematic:||lú-o-mai "I am loosened"||lú-ɔː-mai "I may be loosened"|
|Athematic:||títʰe-mai "I am put"||*titʰé-ɔː-mai > tithʰɔ̂ː-mai "I may be placed"|
If the subjunctive didn't get a thematic vowel, the subjunctive of títʰe-mai would still be títʰe-mai.
The second thing to know is that, over and above any thematic vowels (or the absence of them), the optative mood gets an extra /i/ or /iɛː/ added in. And while the present subjunctive uses the same endings as the present indicative (the "primary" inflections), the present optative uses the "secondary" inflections, which are in common with the imperfect. So the present imperfect of "loosen" is e-lu-ó-mɛːn, with a thematic /o/, and the optative present is lu-o-í-mɛːn, with both a thematic /o/ and an optative /i/.
Oh, and a third thing: proto-Greek /s/ between vowels did not stick around for long, and nor did most instances of two vowels next to each other.
To illustrate all this, this is how the middle voice (and passive voice) subjunctive and optative present are formed, with a thematic verb, and an athematic verb.
There's one more annoyance, which is that athematic optatives from the 3rd sg down also turn up thematic: titʰeîto but also titʰoîto, titʰeímetʰa but also titʰeímetʰa. But we'll ignore that, mercifully.
Now, if we put the athematic φημί into the same machine and crank the handle, we get:
And pʰaîo up there—φαῖο—was the form I failed to recognised in Moschus.
The thing about φημί is, Greek dialect used φημί in the middle voice, but Attic only used it in the active. So you have few instances of φημί in the middle: a fair few perfect participles (including one in Plato), a few middle aorists, and Homer has an middle imperative φάο and φάσθω and a middle infinitive φάσθαι. But we have no known instances of the middle present subjunctive of φημί; and until the new edition of Moschus, we had none recorded for the middle present optative ether. So out of that whole table I just gave, only the boldfaced form has ever turned up in literature.
(Grammarians knew how the rules worked, so they did make up pʰɔ̂ːmai: Heraclides of Miletus (Milesius) fr. 54 Cohn, Theodosius of Alexandria Canones isagogici de flexione verborum p. 97 Hilgard—although the latter intended it as an aorist. I have no evidence they made up an optative to match.)
We fast forward a few centuries to Galen. Galen is a medical writer who has a *lot* of text attributed to him. Partly because he was that prolific: "It has been reported that Galen employed 20 scribes to write down his words ", Wikipedia notes, to which a Wikipedia editor has inevitably added "Citation needed". Partly because his stuff was useful, so copyists kept it around. Partly because he was a default medical author, for others' writings to be attributed to.
Galen's prose is supposed to be a workmanlike Koine, nothing too fancy. (The German word for it is Fachprosa, "specialist prose"—meaning scientific, as opposed to literary.) That said, he occasionally wanted to liven things up. And his misstep on one occasion that he did, gives us our second attested instance of a mediopassive optative of φημί.
Our instance comes from Galen's commentary to Hippocrates' Treatise on Joints. The Bibliothèque interuniversitaire de médicine, Paris V, have magnificently put online the old editions of Galen. Of course people have been reediting individual works of Galen since the 1820s Kühn edition, and must be furious people still refer back to what is a pretty crappy edition, even by 1820s standards. But I'll refer to it anyway, given that it's also the edition the TLG has, and it's where I found this optative.
You can view the passage as published by Kühn; here it is again, with my rendering:
Ἀγαθοὺς πυροὺς τοὺς ἀρίστους δηλονότι λέγει. ταύτην γὰρ οἱ παλαιοὶ τὴν προσηγορίαν ἐπιφέρειν εἰθισμένοι κατὰ παντὸς τοῦ πρώτου ὄντος ἐν τῷ οἰκείῳ γένει. λέγουσι δὲ καὶ νῦν ἅπαντες οἱ περὶ τὴν σιτοποιίαν ἀρίστους εἶναι πυροὺς τοὺς πυκνοὺς τὴν οὐσίαν. οἱ γὰρ χαῦνοι πολὺ τὸ πιτυρῶδες ἔχουσι καὶ τὸ σταῖς αὐτῶν οὐ γίγνεται γλίσχρον, ὡς καὶ μυῶν. ὅλκιμον δὲ ὠνόμασε τὸ γλίσχρον ἀπὸ τοῦ συμβεβηκότος, ἐπειδὴ τεινόντων ἡμῶν αὐτὸ πρὸς τἀναντία μέρη, συνεχὲς μὲν οὐ διασπώμενον τοῦ μὴ γλίσχρου φθάνοντος διασπᾶσθαι κατὰ τὰς τοιαύτας ἐνεργείας. εἰ δὲ διασπᾶται, πῶς ἄν τις ἕπεσθαι φαίηται τεινούσαις χερσὶ αὐτό; εἰ δ᾽ οὐχ ἕπεται, πῶς ἂν ὅλκιμον ἔτι λεχθείη;
By "good wheat", Hippocrates obviously means the best quality. The old-timers used to apply that description to anything which was the first of its kind. And nowadays everyone dealing with bread-making call "best quality" wheat whose substance is dense. Spongy wheat is very crusty, and its dough does not become sticky, like rats' [?!] dough does. And Hippocrates calls sticky dough "ductile", because of what happens when we pull it in opposite directions: it is continuous and does not break apart, while dough that is not sticky ends up breaking apart under that action. If it does break apart, how could one say that it follows along, when their hands stretch it? And if it does not follow along, then how could it still be said to be ductile?
As often happens with ancient commentaries, there is a fair ladling of the bleeding obvious in there. But what detains us is the verbs for "say" Galen uses at the end. The final λεχθείη is an aorist passive optative: "it would be said", here meaning "it would be called". Galen wants to use a different verb from λέγω in the preceding sentence, so he uses φημί, that old irregular verb, which by then would have been fast dying out.
For some reason, Galen wants to use a middle or passive optative, not an active. It's hard to tell which of the two he intends, although that does not affect the form he uses, since passives and middles usually coincide (outside the aorist and future). If he's going for the passive, which he does in the next sentence, then he means "how could it be said that it follows along, with hands pullng"—and with a nominative "someone" (τις) stuck mid-sentence doing nothing. If he's going for the middle, the "someone" finds employment again: "how could someone say that it follows along, with hands pulling"; but then Galen has revived a quite antick middle of φημί.
Whatever he's done, he cannot use an aorist for the optative: there has never been an aorist passive of φημί (outside the fertile imagination of grammarians: Theodosius of Alexandria loc.cit. p. 88 ἐφάθην, Etymologicum Magnum p. 496 Kallierges ἐφάθην ἐφάθημεν ἐφάθησαν), and the aorist middle is Homeric territory, and not the style he's going for. So he has to use a present optative.
We know from the table above that the verb he should use is pʰaîto, φαῖτο. But that's not what Galen comes up with. Instead, Galen does this:
- Optative passive of φημί. OK, the optative active is pʰaíɛːn, pʰaíɛːs, pʰaíɛː...
- [Correct: pʰa-iɛː-n, pʰa-iɛː-n, pʰa-iɛː-Ø]
- So my stem is phai-, to which I add...
- um... how does the optative passive go?
- ... ɛːtai?
So Galen starts with an optative, and ends with the subjunctive suffix. Instead of optative pʰaîto or subjunctive pʰɛ̂ːtai, he has ended up with a mooshed-up pʰaíɛːtai.
Why? Well, several things:
- He's in uncharted grammatical waters: remember, outside that one instance in Moschus, noone has ever seen a middle optative of φημί, and the copyists of Moschus didn't believe it either—which is why the 19th century grammarians didn't notice it.
- When people are in uncharted grammatical waters, they lurch for familiar forms. And already the subjunctive was more familiar than the optative.
- The active optative, which Galen knew well enough, has an extra /ɛː/ in it for its optative marker: pʰa-iɛː, not pʰa-i. So being in unchartered waters, he probably thought the passive should have an /ɛː/ in its suffix as well. After all, a native speaker would not know about the proto-Greek games with active optative /iɛː/ vs. middle optative /i/: they'd just know that there's an optative ending -iɛː to imitate.
- And the correct pʰaîto doesn't give him an /ɛː/, but the subjunctive ending -ɛːtai does.
- So the passive counterpart of -ɛː, um, must be -ɛːtai. Which is of course nonsense, because -tai is never optative, it's only subjunctive: if he really wanted to keep his eta, he should have used -ɛːto, and come up with pʰaíɛːto.
- But Galen was not reconstructing the optative morpheme by morpheme from proto-Greek, like we are doing here. He's lurching about using familiar suffixes. He *can't* take titʰe-ɛː-tai and titʰe-i-to apart, and realise that the /i/ belongs there but the /tai/ doesn't. He just knows about morphemes as fusional blocks, ɛːtai and oito, and he sort of knows that pʰai is an optative stem.
- And he comes up with what Motorcycle Boy, were he a pedantic grammarian, might call "a mutant puppy with two heads".
Thanks to errors in Google Books OCR, btw, φαίηται shows up a lot more online than it's supposed to, as a misrecognition of φαίνεται "it appears". It also turns up in a second online copy of Galen, whose provenance I will not vouch for...