Verb forms like ἔβην, βῆν, βῆναι, and φανῆναι seem to have some ablaut going on. My understanding of the phonological rules of ablaut in ancient Greek is from Pharr, 4th ed., p. 277, along with the table of contractions on p. 276. But I can't seem to figure out from this information why we would get forms like these from βαίνω and φαίνω.

Going on the information in the table in Pharr, there simply doesn't seem to be any ablaut that would take place for the dipthong αι.

1 Answer 1


Both of these go back to Proto-Greek palatal nasals.

There are two reconstructed roots behind βαίνω: some forms come from *gʷem-, while others come from *gʷeh₂-. These two roots look extremely similar and seem to have identical semantics, so it's likely they were related within PIE—but it's not a regular or well-understood alternation. Compare Vedic present gámati < *gem- (~βαίνω) vs aorist ágāt < *gaH- (~ἔβην). This means that the nu in the present system isn't actually the present nasal infix—it had a separate present marker *, giving *gʷm̥-yé-ti > *gʷəňňō > βαίνω. The aorist had no nasal and also no palatalization, giving a completely different vowel (which I think goes back to *ā < *eh₂ but I'd have to check).

Φαίνω, on the other hand, only goes back to a single PIE root *bʰeh₂-—but it seems to have inserted a nasal infix and the present *, giving *bʰh₂-n-yé-ti > *pʰáňňō (with the same palatal nasal) > φαίνω. The aorist again had no palatalization (though this time the nasal was generalized to all the other forms, possibly since there was already another marker for the present), giving a completely different vowel (again, I think, from *ā < *eh₂).

In other words, neither of these is ablaut in the Proto-Indo-European sense. Instead, the ι in the diphthongs comes from regular *-ny- > *-ňň- > ιν. And the *-Ny- that triggered this only occurred in the present system, for various irregular reasons.

  • This doesn't affect the overall argument, but I don't think it's clear whether there was really a palatal geminate stage: metathesis would yield the same results, and seems more plausible phonetically than depalatalization with compensatory lengthening.
    – TKR
    Oct 8, 2021 at 21:26
  • @TKR True, but (1) the other palatals often yielded geminates, (2) allegedly *ly > most dialects λλ Arcadocypriot ιλ (though that could also be metathesis plus later assimilation), (3) *uny, *unr > ῡν, ῡρ which implies some sort of compensatory lengthening.
    – Draconis
    Oct 8, 2021 at 21:48
  • (1) I'm not sure we should expect stops and continuants to pattern together in this environment (and those geminates did not simplify). (2) I think the treatment of L is the strongest argument for gemination, but it's made difficult by the reflex you cite in Cyprian, a conservative dialect (though the difficulty can be avoided by saying that spelling is an attempt to represent a palatal). (3) CL yes, but do those imply a geminate stage? Cf. the 3CL type *ksenwos > kse:nos where I don't think anyone posits a geminate. Not sure what's right, just saying gemination isn't the only possibility.
    – TKR
    Oct 8, 2021 at 22:23
  • @TKR Generally agreed, but what is CL? I don't recognize the abbreviation.
    – Draconis
    Oct 8, 2021 at 22:41
  • Compensatory lengthening -- sorry, was running out of characters!
    – TKR
    Oct 8, 2021 at 22:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.