Joel, on the vertical scale ε is just a tiny bit below ει and far above η, and they give as a model the French word clé which has without any possible doubt an [e] sound.
Yes. What I’m saying is that, IMO, their vowel chart doesn’t agree very well with their IPA notation for said chart.
OK, the closest I can find is here:
https://en.wikipedia.org/wiki/Americanist_phonetic_notation
If you go down to the the vowel section (under ‘Alphabet’) you’ll see that the journal Anthropos used ẹ for higher (i.e. closer) mid and e̠ for lower (i.e. more open) mid.
Absolutely Joel, consistency would demand the ε to be exactly on the same horizontal line than ει and not a tiny bit below (an explanation would possibly be that the internal triangle is smaller than the external and then the proportions are preserved).
Here’s how I see it:
On their chart ει is close-mid, front, so [e:] is surely the right symbol. Their ε is ever-so-slightly more open, but still in the close-mid area; however, it is noticeably nearer the central area. IPA doesn’t seem to have a symbol to represent this intermediate zone, though, so the authors had to choose between [e] (close-mid, front) and [ɘ] (close-mid, central]. Evidently they think that it is more front than central, so they chose [e].
There is not an IPA symbol for every point, though. The symbols can only represent areas on the chart.
Edited to add: I linked to a video explaining this earlier in the thread, here: https://youtu.be/FdldD0-kEcc?si=-g9OaKzWtJzrTj3L
If CGCG is going to call ει [e:] and ε [e], then their IPA notation would seem to incorrectly claim they are equally closed. They don’t seriously mean this, as you can see in their vowel chart, so it’s best to assume it’s a flub, and hopefully one that will be corrected in the next edition on by making it [e̞] or something.
But they are only very slightly different in terms of openness. There is no IPA symbol to express such a tiny difference - that’s why we have the chart, to be more precise than symbols can manage
Their vowel chart doesn’t look like they’re claiming it’s such a tiny difference. They make ε pretty much halfway between closed ει and open η. Just what you’d expect from anybody writing about this after the 1960s. The IPA does have a symbol for that: the mid front vowel (ie., the same as in modern Greek).
By putting the horizontal border of a sheet of paper on the chart you may see that ε is absolutely not halfway but just a very little bit below ει.
And the “clé” example is without any possible doubt an [e] sound.
Are you sure you’re not looking at the front-central-back axis? The open-mid-closed axis is the vertical one
Yes, I’ve got the book open right now. Measuring from the bottom of the page with my daughter’s Castell slide rule conveniently left on my desk (though it’s silly to imagine that whoever designed the graphic was trying to express things so precisely), the point for the η is 7cm from the bottom of the page, the ε 7.65cm, and the ει 7.9cm, with the middle of “mid” on the right being 7.6cm from the bottom – approximately across from the ε.
And that’s exactly the triangle that you’d expect from someone trying to communicate broad agreement with Allen (again, it’s beyond silly to discuss this sort of thing exactly): η is more open than ε, while ει is more closed than ε, with ε being slightly closer to ει (2/3rds of the way from η towards ει along the vertical axis).
What happened, I suspect, is that they designed the triangle, and then for IPA terminology were broadly influenced by people like Threatte and more especially Horrocks using IPA [e] for ε, without noticing that Horrocks also uses [e] to represent the modern Greek letter. Actually Allen too uses [e] (no pretense at IPA) for ε throughout his book. And then one of Boas et al. (or a grad student), grabbed clé as standard example for IPA [e].
But I think that if you got Allen, Horrocks, and Boas et al. into a room together, what they’d all say is that what they really meant was that Attic ε is modern Greek ε. IPA e̞, not the IPA e of clé.
But all of that is tremendously boring. There’s no daylight between any of the positions, just confusing terminology. What’s actually sort of interesting is their vowel chart’s opposite relative fronting of ω and ο, compared to Allen, which I suspect has some sort of reasoning behind it.
Well I don’t want to bore anyone, and I suspect that I’d mostly be repeating myself if I attempted to add anything. I’ll just say that my assumption has been that Boas et al meant what they wrote in the CGCG, and therefore that they think ε was pronounced as a close-mid vowel [e], not a true-mid vowel. If this is not what they meant, perhaps they will change it in a future edition, as Joel suggested earlier. The last set of corrections I saw were from September 2021, and it wasn’t mentioned there.
To the OP, I hope you feel your questions have been answered… was there anything that has been missed?
I agree with MattK: in such a reference grammar everything may have been carefully weighed up and one may suspect that such great scholars as these be able to write what they think, and mean what they write, and not some other thing.
Well if you’re reviewing corrections Matt, send in a note and maybe they can fix it.
Thanks everyone this discussion has been really helpful even though it hasn’t actually resolved my questions. I’'ve had conversations with one of the CGCG authors on a different morphological issue (a possible errata I found) so I’ll bring these matters up with him and report back if anything helpful results from my conversation.
Hello Mitch,
About pronunciation, I couldn’t find in the CGCG (but perhaps I didn’t look well enough) the rule that sigma should be pronounced [z] (and not [s]) before β, γ, δ and μ (is it an omission or a choice?). If you confirm this, maybe you could put a word about it.
I’m also very intrigued by the absence of the luminous explanation of the Greek accentuation by the rule of the contonation…
Nothing so grand, unfortunately - I got them off Evert van Emde Boas’ academia page
Hi everyone,
I heard back from Evert van Emde Boas and it appears that the vowel pronunciations in sectoin 1.15 of CGCG are based largly on Philomen Probert’s chapter titled Phonology in A Companion to the Ancient Greek Language (E.J. Bakker, ed., 2010) and some cited works in that chapter.
Ah perfect, that provides quite a bit of explanation. Here is Probert’s Vowel Diagram:
This is written in very non-committal language, but the overall argument seems to be in favor of the given vowel diagram, a symmetrical short vowel system, with higher (more closed) /a/, making ε approximately the same height as ει. I think that this is an interesting possibility, and I plan to give a more careful read of the cited Thompson paper (https://www.jstor.org/stable/44698295).
The problem is that Boas et al. haven’t really adopted this high short-α theory in their pronunciations for short-α or in their vowel diagram. Not adopting it makes sense, as it’s far too speculative to go into an intermediate reference grammar, especially without careful citation and caveats. But taking ε and ο height from Probert without the rest doesn’t work. It’s a choose one or the other thing.