Reconstruction talk:Proto-Indo-European/h₂éwsōs

No, it's not at all clear whether *h₂ews-os-/*h₂us-s- comes from the nominal root *h₂ews- or the verbal root *h₂wes-; there's simply no way to tell. For some reason you assume that the place of the ablaut vowel is always kept in the original place and can never be switched – but Schwebeablaut is a well-known phenomenon (compare as a particularly well-known example). The way it might work is that the zero grade *h₂us-(e)s- was derived from the verbal root first (think *h₂us-es, the locative) and the full grade secondarily, with the ablauting vowel inserted in the wrong place by analogy with the more common case where the ablauting vowel comes before the resonant.

By the way, it's not even clear that a separate nominal root *h₂ews- actually has to be reconstructed. All the nominal formations could be derived exclusively from the verbal root *h₂wes- "to dawn" (homonymous with the other root for "to stay the night"), which has to be reconstructed on the strength of the evidence – the verbal forms listed in LIV. In fact, LIV implies that the noun *h₂ews-os-/*h₂us-s- is derived from the verbal root directly. --Florian Blaschke (talk) 02:23, 13 May 2016 (UTC)

I really think that a nominal root *h₂ews- as postulated in Reconstruction:Proto-Indo-European/h₂ews- should not be reconstructed at all, as the bare nominal root is nowhere attested. In fact, because of the Sanskrit verbal root vas-, it is quite preferrable to reconstruct only a verbal root *h₂wes- that all the nominal formations are derived from. Check EWAia under VAS² (where the attested verbal forms are listed that force us to reconstruct *h₂wes- and not *h₂ews-) and vasar-: *h₂wes-r/n- "spring" (a heteroclitic, compare Vedic vasantá-, OCS vesna) also belongs to the verbal root. --Florian Blaschke (talk) 23:05, 28 June 2016 (UTC)
 * Our current practice is to treat schewebeablaut variants as separate roots. See also vs . —CodeCat 23:07, 28 June 2016 (UTC)
 * Hrm. In that case, Sanskrit vas- and the word for spring, as well as the Celtic forms, need to be shoved into (obviously as a separate entry), and cross-referenced with . But what about Vedic us-r-, us-rá- and the like, which are ambiguous? --Florian Blaschke (talk) 23:13, 28 June 2016 (UTC)