On the relative chronology of Attic-Ionic η < ᾱ

[Timothée]

It is normally stated (e.g. Rix §56) that Greek ᾱ changed into η in Attic and Ionic, whereafter it went back to ᾱ in Attic after ε, ι and ρ (we call it 'Eira rule' in Finland, after a posh Helsinki district by this name). This may be very obtuse of me, but I have to ask: how is it deduced that the change ᾱ > η after ε, ι and ρ occurred at all? In other words: how do we conclude that the original ᾱ did not simply stay as such after ε, ι and ρ in Attic, changing into η only in other positions?

[mwh]

I’d always thought—naively assumed, more like—that Attic’s “pure” alpha had always been alpha, and that the more extensive etacization was only Ionic. But then I’m hopeless when it comes to historical linguistics. So I join you in your question.

As I wrote in the Ajax thread, apropos the quasi-doric of tragic lyric:

In Ionic and Attic, long alpha, sometimes called “Doric” alpha (but it’s Aeolic too), was replaced by η, with certain phonologically defined exceptions in Attic.

Is that not an acceptable capsule formulation?

[Paul Derouda]

Carl Darling Buck, Greek Dialects, p. 21:

a. In some Central Ionic inscriptions [...] only this η from ᾱ was written [a character I can't reproduce], Η, vs. E for η = original ē. This distinction was probably once general in Attic-Ionic, the sound resulting from ᾱ being somewhat more open than that representing original ē.
b. It is a widely held, but also disputed, view that the change of ᾱ to η, or rather to the sound close to but not quite identical with η = original ē (above, a), was general Attic-Ionic and that the ᾱ of Att. γενεά, etc., was a reversion rather than an uninterrupted retention of ᾱ in such cases.

I don't know about a more recent discussion. Horrocks's more recent book repeatedly mentions that there are two conflicting views on this, but I couldn't locate anything more detailed on this subject there.

[jeidsath]

Allen's Vox Graeca presents some of the evidence for what happened on pg. 70 (and mwh thought that he could escape from the Cyclades in this thread!):

[mwh]

“disputed” Buck, “remained or reverted” Allen. Who cares which? would be a philistine question, but it makes no practical difference, and until and unless historical linguists decide the issue I’ll stick to my naïve view.

[Paul Derouda]

“disputed” Buck

But isn't he just lovely?

(I'm sorry, that joke is really weak, but I can't help thinking about Tarantino's film every time I see the name Carl Darling Buck.)

[Timothée]

Thank you everyone for your messages.

For the sake of completeness, I could add here what Rix says (it's not much): "Das Att. hat jedoch vor der Aufhebung der Opposition die Realisierung [ā] hinter /e, i, r/ wieder hergestellt und so das Phonem /ā/ in beschränktem Umfang erhalten." Just tersely stating the same.

Next I had to look up in Schwyzer to get a little further with this conundrum, and he indeed opens this question. The discussion begins on page 187 and heats up on the next page. First, Schwyzer takes words like νέᾱ, νεᾱνίᾱς, νεᾱτός, πόᾱ (ποίᾱ), θέᾱ as a proof as compared with κόρη. Secondly, he says that contracted words like ὑγιᾶ, ἐνδεᾶ and χρέᾱ have to well from ὑγιῆ, ἐνδεῆ and χρέη, ultimately from ὑγιέα, ἐνδεέα and χρέεα. These he compares with γένη < γένεα, ἐμφερῆ < ἐμφερέα, τριήρη < τριήρεα, ὄρη < ὄρεα and ἐγρηγόρη < ἐγρηγόρεα.

I have to confess I'm still somewhat confused. I think we should date the disappearance of ϝ in Attic and Ionic; and in different positions, I think. Could the digamma have disappeared before ᾱ > η between vowels, but persevered longer in -ρϝ-? I cannot quite understand the second argument for the time being.

As to Buck, I always think that, with a name like that, he must have been a really lovely person. Otherwise it would be just too confusing. And in his Selected Synonyms he concludes (as there does not seem to be a word for 'dancing' in Old Irish) that "there was apparently no dancing in ancient Ireland".