Evidence about pronunciation of ευ and αυ in Homeric Greek?

Question

In modern Greek, a word like ευχαριστώ is pronounced like "ef-." The combinations ευ and αυ sometimes have the upsilon pronounced like β and sometimes like φ. (I'm not sure how variable it is when it comes to β versus φ, or whether there is any phonetic rule.)

Do we have any evidence about whether this occurred in Homeric Greek? I'm not sufficiently well-versed in the metrical aspects of Homer to understand whether or not you'd be able to tell on metrical grounds. Is there any evidence that these pronunciations postdated iotacization?

Answer 1

I would go further than Draconis's answer and say that we can be pretty certain that these diphthongs were indeed diphthongs in Homer's time. Here are some additional arguments:

1. The Homeric poems took shape over centuries so there was likely some amount of temporal and other variation in the pronunciation of "Homeric Greek". But that process overlapped with the period when the Greeks were adopting and adapting the alphabet from the Phoenicians, probably around 800 BC. When they did so, they seem to have split the Phoenician letter waw into two Greek letters, digamma and upsilon, using the former for the consonant [w] and the latter for the vowel [u]. If at this point Greek had sequences such as [af av ef ev], it would be an extremely odd choice to use the letter that represented [u] to stand for the second, consonantal part of those sequences.

2. In fact there are inscriptions, from the 6th century BC and possibly earlier, that represent the sequences in question using digamma: e.g. αϝ for αυ. This pretty clearly shows that the pronunciation was [aw]. Another type of spelling variant is αο εο, found in East Ionic from the 4th century BC on, which is again clearly a diphthong.

3. Beside the Latin spellings Draconis mentions there are Hellenistic-era spellings in Indian scripts such as Evukratidāsa for Εὐκρατίδης, and Greek ταῦρος was loaned into Sanskrit as tāvuras. The character transcribed v stood for [β], but is in this case probably a graphical device used because of the difficulty of representing hiatus in the Brahmi scripts; in any case the presence of a [u] vowel seems clear (a Greek pronunciation like [ta:vros] would presumably have been borrowed as tāvras).

4. ευ, αυ are accented like diphthongs, not like vowel+consonant sequences (αύ not άυ, etc.). Furthermore the existence of circumflexed εῦ would be impossible if we were dealing with a sequence of ε (necessarily short so uncircumflexable) plus a consonant. Of course the writing of accents dates from several centuries after Homer, though.

5. There are AFAIK no references in Greek grammarians to the idea of υ standing for a consonant in the digraphs ευ, αυ, which would surely be worth mentioning if it were the case; again, though, the grammarians are writing a few centuries after Homer.

6. There is otherwise no indication that Greek had the sounds [f v] before the first century CE, when φ β come to be pronounced as fricatives, so it would be strange if it only had them in those two particular sequences.

7. The objection (in comments to the OP) that the pronunciation might have gone from e.g. [ef] in Homer's time to [eu] later and then back to [ef] seems very unlikely. Fortition of [w] to [v] (and devoicing to [f]) is cross-linguistically common, but I don't know of any cases of a change like ef > eu.

8. A word like e.g. εὐεργής "well-made" scans in Homer as having a long first syllable, which would not be the case if the pronunciation were [efe-].

Answer 2

Transcriptions into Latin

While there are no transcriptions from the time of Homer (since the alphabet didn't exist yet), these provide good evidence that the change from [aw] and [ew] to [af] and [ef] happened after the Classical period. A brief search turns up examples like Autolycon, Baucus, and Caunus in the Metamorphoses, as well as borrowed words like centaurus. We can be reasonably confident that these borrowings reflect the pronunciation rather than the spelling because they use Latin U rather than the borrowed letter Y.

Greek phonological changes

There's also evidence that αυ and ευ acted as single units within Homeric-era Greek, rather than as sequences of a vowel and a consonant. The second element in these diphthongs survived the loss of independent digamma, for example, which disappeared even after alpha and epsilon: ἀέκητι < *a-wek-, ἀϊδνός < *a-wid-. This sort of cohesion seems much more likely for diphthongs with two (semi)vocalic elements, than for a vowel followed by a fricative. (It also implies an underlying representation like /au/ rather than /aw/ at this time, which is even less likely to turn into [f].)

In the end, we can't be certain

As Unbrutal_Russian mentions in the comments, it's impossible to be certain. The best evidence comes from transcriptions, which didn't happen until centuries later, and there's no way to prove (for example) that it didn't change from [aw] to [af] before Homer's time and back to [aw] again after.

However, even if we can't be certain, we have no reason to suppose such a change either; for all practical purposes, such as recitations of Homer, I would assume [w] over [f].