<boring background> I've been doing some recordings of language drill in Homeric Greek (1, 2, 3), in which my pronunciation has been chosen based on a certain set of criteria: (1) They're meant to be listened to in the car or while cooking, and in these noisy environments it's often difficult to make out what's being said. For that reason, I've tried to clearly differentiate the pairs ω/ο and η/ε. (2) Erasmian pronunciation works better than a reconstructed pronunciation, since native English speakers like me (as well as speakers of most other languages) have a hard time hearing and saying the distinctions between the stops in reconstructed Greek (examples of recordings). (3) I would like these to be of use to other people, not just me, so I want to be as consistent as possible with the pronunciation used by others. (4) The pronunciation should ideally work well with Homeric metrical verse. Based on these criteria, I tried to work out what seemed to be the most widely used US Erasmian pronunciation. (History. Some of this article was written by me.)</boring background>
So during the process described above, I settled on pronouncing ω as [o̞] or [o] (IPA vowel chart with audio), like my native Californian "o" in "go" but not diphthongized, and ο as [ɔ], which requires considerable effort, since I have the cot-caught merger. (I also have started trying to consistently pronounce ω with double the temporal length, as in metrical poetry. Although this may be unnatural outside of poetry, it makes it easier to hear the distinctions.)
After practicing and doing a lot of recording according to this rule, I'm not sure I would seriously consider changing, but I did poke around recently to try to clarify for myself whether what I'd been doing was a truly "vanilla" choice for English-speaking students of Homer. I've looked at written sources of information such as textbooks, Allen's Vox Graeca, and the historical pamphlets by Peck and by Arnold and Conway. I've also listened to publicly available recordings by some classics professors and a couple of enthusiastic amateurs. The results are a little perplexing, and they make me wonder if I'm misunderstanding something, hence this question:
Question: What convention is followed by the plurality of English-speaking students of ancient Greek for the pronunciation of omega and omicron?
The printed information is especially difficult to figure out, because the authors of textbooks and of the pamphlets referenced above have a tendency to assume that their readers must naturally speak the same dialect of English as they do. So they give examples of English words that are supposed to exemplify the vowels they have in mind, and they assume you know what that vowel sounds like when they say it. (I suppose Peck et al. were writing before the IPA even existed.) So for example, the textbook by Pharr (4th ed.) uses the minimal pairs go/rose and ought/saw, but to me these vowels are indistinguishable. I'm then reduced to silliness like looking up the fact that the book was published by the University of Oklahoma press and trying to use that to guess what kind of accent Pharr had. Er, unless the table has been rewritten by his 21st-century coauthor Paula Debnar, who seems to be a northeasterner. Oh, but she got her undergrad degree from Colby College, so maybe she's from Maine, so maybe her vowels are different...? This way lies madness.
But given the written descriptions I've seen, it looks like every logical possibility has advocated as "correct:"
Pharr/Wright/Debnar say that in Erasmian, ο="rose" and ω="go," which to me is the same vowel on the IPA chart. (I pronounce "rose" with greater temporal length, but I assume that isn't the distinction they have in mind.)
Smyth gives a triangular diagram on p. 9, which is basically the IPA chart upside down. He has ω more open than ο.
Arnold and Conway describe ω as open and ο as close. Their French examples are ω=encore (=[ɔ]) and ο=monologue (=[o]), which is also ω more open. But their English examples are ω=ore, and ο=cannot, which is harder to make out for sure without knowing their dialect, but in dialects I'm familiar with, this would again make ω more closed, contrary to their own description.
A random American Koine amateur gives this table, with ω clearly and consistently described as more closed.
Turning to recordings:
Two professional classicists, Nagy and Debnar, both seem to pronounce ω and ο identically in terms of vowel quality (not temporal length), similar to a Californian "o." For Debnar, this seems to be consistent with the English approximations in her book.
A native Greek speaker, Kostas Katsouranis (recommended by classicist Walter M. Roberts III as a guide to correct pronunciation), doing a reconstructed pronunciation, pronounces ω more open.
So it seems that one can find well-qualified people practicing and teaching all three possibilities (ω the same as ο, more open, or more closed). My question is which of these is the most common.