What happens to internal /e/ and semivowel /y/ in *snéygʷʰm̥ to yield L. nix? I have no clue how that vowel change works.

1 Answer 1


The PIE form you cite is the accusative singular, not the nominative singular.

The PIE word is *snéygʷʰs ~ *snigʷʰés. Note that as this is an athematic noun it is necessary to cite both the nominative and genitive forms so that the ablaut pattern is clear.

Latin regularly reflects initial *sn as n.

As with almost all words in almost all attested branches of PIE, Latin also typically generalises one of the ablaut grades of the PIE word. In this case Latin has continued the genitive stem, and regularly reflects PIE *i as i (PIE *ey would be reflected as ī).

In the nominative singular we have regressive voicing assimilation, so *gʷʰs > ks and, as with all 3rd declension nouns, the PIE vocative singular is replaced by the nominative singular.

Intervocalically, *gʷʰ is regularly reflected as v.

Finally, PIE *e is raised to i in non-root-initial syllables during Old Latin stage, when stress is still root-initial.

Putting all that together, the expected reflex of PIE *snéygʷʰs ~ *snigʷʰés is nix ~ nivis (with nīx ~ nīvis also being possible if the ablaut of the nominative had been generalised instead), which is exactly the form we see.

  • I’d say the expected (i.e., regular) reflex would be nīx ~ nivis – the length of the vowel in the nominative I would call unexpected, though easily explainable. (Also, I made a small edit, since I assume it was a typo that *gʷʰs becomes just s by regressive voicing assimilation.) Feb 12 at 15:49
  • I think levelling of ablaut is so universal that it's reasonable to say a uniform i is expected, just also that a uniform ī would also be expected. An alternating vocalisation would be extremely unexpected for all but the most commonly used words. Thanks for the typo-fixing edit
    – Tristan
    Feb 12 at 16:36
  • Vowel reduction occurred in the middle of Old Latin period, not Proto-Italic. There are preserved inscriptions from earlier part of Old Latin, such as Duenos inscription, which show not yet reduced vowels.
    – Arfrever
    Feb 13 at 2:35
  • ah yep, good point. I'll correct
    – Tristan
    Feb 13 at 10:05

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.