User talk:Victar/Reconstruction:Proto-Indo-European/néh₂s

@JohnC5, Rua, Mahagaja Could you have a look over this attempt to clean up PIE *néh₂s-? Thanks. --Victar (talk) 22:59, 29 December 2017 (UTC)

Regarding Slavic, the regular reflex of -sr is -str- and not -zdr-. Also, we don't use the character ŗ in our reconstructions, whatever it means. The form *nh₂os-éh₂ is also completely unexplained. Why would a vowel suddenly appear within the root in a random place? The Balto-Slavic forms don't fit either; they all reflect short a or o, which doesn't appear in the Balto-Slavic reconstruction nor in its PIE ancestor. —Rua (mew) 23:04, 29 December 2017 (UTC)

From what I understand, laryngeals between consonants first-syllable become *a in PBS. I think *nh₂és-s makes more sense for Latvian nass, but I was goin off Derksen. --Victar (talk) 23:17, 29 December 2017 (UTC)

@Victar: Of course, this word is frequently held up as the example of *a ~ *ā ablaut. The Leiden people hate this, but it is common enough to require mention. Also Kordlandt, Lubotsky, and Beekes give the reconstruction *néh₂s (< *néh₂s-s) ~ *néh₂s-m̥ ~ *nh₂s-ós. I'm not sure where the *néh₂-s ~ *nh₂- inflection is coming from, as none of the sources propose it (and Ringe and Mallory & Adams want *(H)nas-). Also @Rua, Derksen does mention a different potential etymology from Meillet for the Slavic -zdr- that would alleviate you concerns, getting it from a compound of *nȏsъ + *dьrati. —*i̯óh₁nC^[5] 11:58, 30 December 2017 (UTC)

@JohnC5 Yes, Kortlandt cites *neh₂-s-, *nh₂-es-, *nh₂-s-, if I'm not mistaken. I think I just didn't fill out the template properly. Feel free to fix. I'm not so sure about *néh₂s-ss > *néh₂s, which seems to contradict the PBS.--Victar (talk) 05:28, 31 December 2017 (UTC)

Of course, the complaint for all of this is that *s-extended root nouns are effectively only proposed for this root, which is very suspicious. —*i̯óh₁nC^[5] 06:31, 31 December 2017 (UTC)

But there is no initial consonant in the zero grade, it begins with *n̥h₂- with a syllabic sonorant. I don't know if Balto-Slavic has a different outcome if the second consonant is a laryngeal compared to any other consonant, that's worth looking into. However, given that Balto-Slavic is like Germanic in this regard and treats syllabic sonorants the same whether there is a following laryngeal or not, I expect that *n̥h₂- would simply become *inˀ- in Balto-Slavic. —Rua (mew) 13:31, 30 December 2017 (UTC)

This is exactly why most people prefer an *a for this word. —*i̯óh₁nC^[5] 22:07, 30 December 2017 (UTC)

@Rua: PGmc *natją is supposedly from PIE *nh₂dyóm, from *neh₂d- (“knot”). Kroonen also often cites syllabic laryngeals (which I know you have issue with), and gives the etymology of *namô as *h₁nh̥₃men-s --Victar (talk) 05:28, 31 December 2017 (UTC)

Several linguists would also dispute such an interpretation. I think these are analogical: zero-grade of roots with laryngeals was reinterpreted as a-grade by analogy with roots without sonorants. If there are any cases of these with the expected *u- in Germanic, that would prove my point that this is the regular outcome. If there aren't any, then perhaps *Ra- is the expected word-initial outcome. Either way, there would be no need for syllabic laryngeals. —Rua (mew) 11:59, 31 December 2017 (UTC)

I suggest we make a separate page to inventorise and discuss these cases for Germanic and Balto-Slavic. Do you want to make a start? —Rua (mew) 12:00, 31 December 2017 (UTC)

*meh₂ḱ- > *mh₂ḱ-ró-s > *magraz is another very well attested example of first-syllable CHC > CaC in PGmc. I'm not sure this really has anything to do with the laryngeal, or if it's simply consonant cluster breaking, i.e. CHC > CC > CaC. I can post to the Etymology scriptorium, if you think that would be helpful. --Victar (talk) 15:29, 31 December 2017 (UTC)

@JohnC5, Rua I found this in {{R:Ringe 2006|page=137}}, "noninitial laryngeals not adjacent to any syllabic became *ə, which became *a in initial syllables". --Victar (talk) 22:22, 31 December 2017 (UTC)

Oh, to be sure, *CHC- became *CaC- in Germanic. The question is whether *#RHC- became *#RaC- or *uRC-. *unstiz < *h₂n̥h₁-s-tis if not fully useful in this case, given the initial laryngeal. It's clear though that Kroonen believes *#R̥HC- > *#RH̥C- > *#RaC-. —*i̯óh₁nC^[5] 22:40, 31 December 2017 (UTC)

@JohnC5: Right, right. While we're giving nearly helpful examples =), PIE *nókʷts > *n̥kʷ-tw- > PGmc *unhtwǭ. --Victar (talk) 03:55, 1 January 2018 (UTC)

Another theory is that PGmc *nustrō was influenced by *nusō, which Kroonen postulates was a secondary zero grade of *nasō, but am personally sceptical of. --Victar (talk) 04:04, 1 January 2018 (UTC)

@Victar: In reference to diff, what is your reasoning? For one thing, no cluster *ss exists in PIE or word-finally in Balto-Slavic. —*i̯óh₁nC^[5] 10:50, 2 January 2018 (UTC)

@JohnC5: Yes, PIE would have degerminated the *-s-s making the nom.sg. identical, but the Slavic requires an accusative *nh₂és-m̥ and I'm not sure that paradigm table would work for all descendents and derivatives. What do you think? --Victar (talk) 16:03, 2 January 2018 (UTC)

@Victar: Ah, that makes sense. The only difference here is that Derksen is using the Leiden model and we're using the Erlangen model. To be honest, we should add the Leiden model as well, to be complete. However, this is still a descendant, not a derivation. Might I suggest moving it under the descendants, and adding a provisional notation about the Leiden model and it's accusative? It really doesn't make much sense under the derived terms. —*i̯óh₁nC^[5] 22:46, 2 January 2018 (UTC)