Why is the Ancient Greek verb ὀράω so horribly irregular?

The question comes from an exchange Joachim Pense  and I had about irregular verbs. My argument to him was that, if you know something about the history of Greek phonology, and factor in suppletion, the verbs do at least start to make some sense.

Warning: if you don’t already know Ancient Greek, don’t bother reading further.

Why is this verb so damn messy? In particular, as Joachim asked,

For example the reduplication + lengthening in the perfect active. The perfect medium – is the contraction involved regular? And the aspirated labial in the perfect passive? I forgot the details in Greek to know if that is standard. And to boot the future is in the medium.

1. Suppletion.

Lots of old Greek verbs are suppletive. Suppletion means that one tense comes from one root, and another tense comes from a completely different root. The parallel in English is go and went. A less obvious counterpart is be and was.

There is a lot of suppletion in Ancient Greek (and to some extent inherited into Modern Greek). But suppletion isn’t because someone along the line wanted to mess with their descendants’ brains. Suppletion happens because those different tenses, which look like they belong to completely different verbs, actually were completely different verbs. Went is the past tense originally of wend (as in “wend my way”).

The dirty secret of Greek is that what regularity they do have is a fiction. I know there are Greek learners exclaiming “what bloody regularity?!” But if you look at those tables of conjugations in the grammars, you are struck by how neat and patterned they all are.

Well, the neatness is not how it started. The neatness comes from analogy, with generations of Greek speakers trying to make their verbs follow patterns, so they could actually learn the damn things. The future passive, for example, if you look very closely, doesn’t have quite the right suffix. It wouldn’t: it’s a Classical era invention, by analogy with the aorist passive. It didn’t exist in Homeric Greek.

There’s a very boring book on my shelves by Henri van de Laar: Description of the Greek Individual Verbal Systems. It’s a list of every very old Greek verb, and its history. The tl;dr of it is the interesting bit: he’s pretty sure that the mess of ὀράω reflects the original state of the Greek verb system. Where aorists and presents were completely different verbs, which only coincidentally and messily converged into the one pattern.

In this case, we have a merger of no less than three verbs:

ϝοράω, which gives us the present ὁράω and imperfect ἐώρων, and the perfect active  ἑώρακα ~ ἑόρακα

ϝείδω, which gives us the aorist εἶδον

ὄπτομαι, which gives us the future ὄψομαι, the perfect passive ὤμμαι, and the aorist passive ὤφθην.

That’s not the end of course. Let’s keep going.

2. What’s going on with ϝοράω?

Well, for starters, we have an initial digamma. The digamma dropped off in Homeric times (though it stuck around in say Aeolic); but the digamma explains a lot of syllabic augments where you don’t expect them.

If ὁράω was a normal verb, it would get a temporal augment: ὤρων. But it doesn’t: it gets an added syllabic augment instead,  ἐ-, of the kind that you only get when the verb starts with a consonant.

And the digamma is your explanation. The verb *did* start with a consonant, the digamma. The imperfect of ϝοράω was ἐϝόρων, and the perfect active was ἑϝόρακα. The digamma dropped out everywhere, and so we’re left with ἐόρων and ἑόρακα. Simples, right?

Well, not right, because the imperfect is ἐώρων. Why?

Well, one bit of that is irregular magic, and one bit is regular Attic.

The irregular magic is that sometimes, randomly, some verbs don’t get augmented once, they get augmented twice. They take a η- augment instead of an ε- augment, as if the ε- augment was augmented again. That’s not impossible to understand: people assumed the ε- form wasn’t already augmented already for whatever reason, and they augmented it again. The best known ancient case is βο­ύλομαι > ἠβουλήθην, not ἐβουλήθην. In Modern Eastern Cretan dialect, in fact, η- is the default augment.

OK, that gets us from ἐόρων to ἠόρων. That doesn’t get us ἐώρων.

And this is where the regular Attic bit comes in. Think: do you ever see ηο together in Attic? No. But, if you know your Herodotus,  you do see lots of ηο in Ionic.

Where you see ηο in Ionic, you see εω in Attic, in lots of places where εω makes no sense. The accentuation of εω as a suffix makes no sense for example: τάξεως is accented as if its final syllable is short. But ω isn’t short. Attic has this bizarre -εως declension, where Doric has the far more sensible -αος ending (which is what the Koine went with). So Doric λᾱός, Attic λεώς.

And Ionic  ληός.

The change of Ionic ηο to Attic εω was regular. And it applied to starts of verbs too. So. ἐϝόρων > ἐόρων > ἠόρων > ἐώρων. And similarly, ἑϝόρακα > ἑόρακα > ἠόρακα > ἐώρακα.

3. What’s going on with ϝείδω?

It’s an old verb, which is why εἶδον is a second aorist, not a first aorist. It’s so old that its perfect tense has ended up as a completely different verb.

That verb is (ϝ)οἶδα, to know.

And what is the Germanic cognate of ϝοἶδα?

German wissen, and English wit.

4. What’s going on with ὄπτομαι?

Why, simplicity itself.

The present is unattested, and may never have been used, but you should recognise the stem from the related nouns and adjectives. As in ὀπτικός, optical. And as in ὄμμα, eye.

(Oh, you don’t see ὄμμα? We’ll get to that.)

-pt- is a lot of consonants for a Greek verb root to end in. Tense suffixes involve even more consonants. And not all those consonants are going to survive the merger.

[EDIT: In his comment, Gabriel Bertilson points out I’m overcomplicating things: the -t- itself is an insertion before the present stem (which isn’t attested anyway), and the root is just op-. So ignore the -t- in the following.]

So for example the Aorist Passive you would regularly get would be ὤπτθην, /ɔ́ːpttʰɛːn/. Not even Ancient Greeks could pronounce  /ɔ́ːpttʰɛːn/. But with the aspiration shared between the /p/ and the /t/, and dropping the ttʰ back to /tʰ/, they could pronounce /ɔ́ːpʰtʰɛːn/: ὤφθην. That’s completely regular.

[EDIT: make that ɔ́ːptʰɛːn > ɔ́ːpʰtʰɛːn]

Same goes for the Perfect Passive. ὤπτμαι /ɔ́ːptmai/ is smoothed out, again regularly, to /ɔ́ːmmai/: ὤμμαι: labial followed (eventually) by an /m/ gets smoothed out as /mm/. We’ve seen the same smoothing in ὄμμα. ὄμμα is just the verbal noun of ὄπτομαι, just as γράμμα is the verbal noun of γράφω.

[EDIT: make that ɔ́ːpmai > ɔ́ːmmai]

As for ὄψομαι, that’s just ὄπτσομαι /óptsomai/ made pronouncable: drop the /t/, and you’ve got /ópsomai/. You’ve got a middle, because it’s an old verb, and the future was first introduced as a middle voice notion. The future tense wasn’t originally something you were going to do; it was something you wanted to do. Which is consistent with middle voice. The proto-Greek future ending is –sy-, which is cognate with the desiderative suffix -σείω that survived into Aristophanes: χέζω “I shit”, χεσείω “I want to take a shit”.

The active future came later: it was more attempts by later Greek speakers to smooth out the jumble of forms they had inherited into something learnable.