Last modified on 4 January 2015, at 20:06

Appendix talk:Proto-Indo-European/h₁nómn̥

On reconstructionEdit

I don't understand why this is listed at h₁nḗh₃mn̥, which is pretty much the one thing it couldn't be. Laryngeals don't color long vowels, so h₁nḗh₃mn̥ would give a "post-laryngeal" nḗmn̥, and no language has an unambiguous reflex of long e in this word. The only forms with a long vowel are Sanskrit nāma, which could be from *nomn̥ with lengthening by Brugmann's law; Latin nōmen from earlier gnōmen, which has been contaminated by gnōscō; and Hittite lāman, where I don't know the reason for the long vowel, but Hittite has a lot of unexpected long vowels (e.g. tēkan < *dheghom). There's also Dutch noemen "call, name" and related forms like Westphalian nömen which seem to go back to PGmc. *nōm-, but one denominative verb in a few late-attested West Germanic languages hardly outweighs the evidence of all the ancient languages. (The long vowels of name, German Name, Dutch naam etc. are secondary.) But Celtic, Albanian, and Balto-Slavic all come from a zero-grade Hn̥m(e)n-, so it's clear the paradigm must have had a zero grade in it somewhere. What seems most likely is nom./acc. *h₃nómn̥, gen. *h₃n̥méns. —Angr 11:00, 8 January 2012 (UTC)

Karin Stüber in The Historical Morphology of n-Stems in Celtic (Maynooth 1998) sums up the proposed etymologies for this word. She says the candidates that have been proposed are *nóm-n̥, *Hnóm-n̥, *h₁néh₃mn̥ and *h₃néh₃mn̥, and concludes that *h₁nómn̥ is the most likely proto-form. She makes no mention of a *h₁nḗh₃mn̥. Her arguments are as follows:
  • h₁ is more likely than h₃ because of Hittite lāman, which would have been *halāman or something with h₃. Also it's more likely that Proto-Greek *enoma assimilated to onoma (paralleled by *h₁dont- > *edōn assimilating to odōn 'tooth') than that onoma dissimilated to *enoma in the name Enumakartidas, and it's unsurprising that a personal name should keep its original form even after the common noun it's derived from has changed its form. The form nōnumos 'nameless' is then analogical to onuma, replacing earlier *nēnumos < *ne-h₁nom-, just as nōdos 'toothless' is analogical to odōn, replacing earlier *nēdos < *ne-h₁dont--
  • short o with no following laryngeal is unproblematic in all languages (the long vowel in Indo-Iranian and Hittite being phonologically regular, and the long vowel in Latin being due to association with *gnō-) except Tocharian where the palatal ñ suggests the following vowel was e or ē, and except for the Dutch and Low German forms that seem to have Germanic ō. Stüber says the palatalization of the n in Tocharian could have arisen in the oblique zero-grade form *h₁n̥mén-, or if the paradigm was acrostatic in an oblique form *h₁némn- (which however would not have survived anywhere else but Tocharian). The Dutch and Low Saxon forms are probably relatively late developments using the inner-Germanic a/ō ablaut found also in the class VI strong verbs.
So, I'm going to move this page to h₁nómn̥, but retain the comment about the uncertainty of reconstruction. —Angr 12:02, 4 August 2013 (UTC)
As a noun formation, *h₁nómn̥ doesn't make much sense. The suffix *-mn̥ is attached to roots, but *h₁nó- is most certainly not a valid root because it doesn't end in a consonant. So the root should end in a laryngeal, and should be in the e-grade in the direct cases, zero grade in the oblique cases, as determined by the suffix. Thus, nominative *h₁néh₃mn̥, genitive *h₁n̥h₃méns. Can the sequence -n̥h₃- account for the Sanskrit long vowel? —CodeCat 12:17, 4 August 2013 (UTC)
This seems to have the suffix *-n̥ added to a root ending in *-m rather than the suffix *-mn̥ added to a root ending in h₃. *-mn̥ usually forms verbal nouns from verb roots, but there doesn't seem to be a verb root h₁neh₃- anyway, and "name" isn't semantically obviously a verbal noun. The o-grade isn't a problem in *h₁nóm-n̥ either as it matches the o-grade in other neuters like *dór-u and *ǵon-u. I think -n̥h₃- could theoretically account for the Sanskrit long vowel, but not for the lack of a laryngeal in Hittite, where we would expect something like *nahman (> *lahman?) rather than *lāman from *h₁néh₃mn̥. —Angr 12:47, 4 August 2013 (UTC)
That really just shifts the problem rather than solving it. Now you're left with a suffix -n that is like no other suffix in PIE and appears only in this word. And don't laryngeals disappear in Hittite after vowels too? We really know this:
  • At least some variety of this word must have -n̥h₃- in it because that's the only way to account for the Balto-Slavic and Celtic outcomes, which unambiguously reflect a syllabic n.
  • At least some variety must have -ēh₃- or -eh₃- to account for the Sanskrit long vowel.
  • At least some variety must have -e- or -ē(H)- to account for the -e- in Tocharian B.
Ringe devotes a whole page to this in his 2006 book about PIE and PG. He reconstructs two paradigms:
  • "Narten" type acrostatic, nom. *h₁nḗh₃mn̥, gen. *h₁néh₃mn̥s
  • Amphikinetic collective, nom. *h₁néh₃mō, gen. *h₁n̥h₃mn̥és
So the difference in the stems is explained by alternation between singular and plural (collective), which is quite common among IE languages. To explain the Germanic form *nam-, he says that the nominative collective stem *nōm- was levelled into all the other forms, and this vowel was then shortened through Osthoff's law (which shortens long vowels before a resonant followed by another consonant in various IE branches). The long vowel survives in Dutch noemen. The acrostatic genitive stem *h₁néh₃mn̥- also explains the Latin form, the Hittite form as well according to Ringe, so he presumes that -éh₃- > -ō- in Hittite. The Tocharian -e- then comes from -ēh₃-. —CodeCat 13:16, 4 August 2013 (UTC)
A suffix -n̥ is a lot like other neuter nominal suffixes (-i and -u for example, not to mention the -r̥/n̥ of the r/n-stems) and seems to appear in the grease/butter word (OIr. imb, L. unguen, OHG ancho < *h₂óngʷ-n̥) as well. h2 and h3 don't disappear in Hittite after vowels, cf. pahs- 'protect'. The syllabic n in B-S and Celtic doesn't have to come from -n̥h₃-, though; it can just as easily come from -n̥m-, and in Celtic it might have to since I think -n̥h₃- would have given *-nā- rather than *-an-. The Sanskrit long vowel is due to Brugmann's law. Tocharian B e can come from earlier *o as well as from *ē; the real evidence for e/ē in Tocharian is the palatalized ñ, as I mentioned above. Ringe seems to be adapting the data to fit his theory rather than the other way around; h₁nḗh₃mn̥ would give "post-laryngeal" nēmn̥ since laryngeals don't color long vowels, and there's no unambiguous evidence for ē in any language. Tocharian is the only language with evidence for any front-vowel vocalism, and it could just as easily come from short e. I'm really skeptical of the vowel of noemen being old since it occurs only in two languages (nl and nds), both of which are attested late, and only in a derivative form. It seems so unlikely that if Germanic had ō it would have died out absolutely everywhere except this one little corner of West Germanic. —Angr 13:49, 4 August 2013 (UTC)
I think that all the changes you're making to entries are VERY premature, and bad form considering this discussion is far from finished. Your view is lacking sources while I have at least two for mine. I don't think our entries should reflect a minority view (so far, just you) as if it were the majority. So I ask you to undo your edits until a consensus is reached.
How else would you explain the -ō-? -ō- is a regular ablaut alternation of -a- in Germanic, and Germanic -ō- is the only source for Dutch -oe- and the -uo- found in the old languages. And since ablaut was not productive for word derivation in Proto-Germanic, that implies that the -ō- is of Pre-Proto-Germanic date; where else could it have come from? You seem to be taking "this word existed in Proto-Germanic" to imply "this word must be attested in many descendants". But that's false and there are several demonstrable cases where the etymology of a term attested in only a single Germanic language can be traced all the way back to PIE. So unlikely doesn't mean impossible. —CodeCat 15:13, 4 August 2013 (UTC)
But Ringe's reconstruction appears to be the least mainstream of them all. Does anyone else reconstruct the original form with *-ēh₃ with a collective plural? --Ivan Štambuk (talk) 20:43, 4 August 2013 (UTC)
M. Philippa's "Etymologisch Woordenboek van het Nederlands" gives nominative *h₁néh₃mn̥, genitive *h₁n̥h₃méns. That's the only other source I have. —CodeCat 20:49, 4 August 2013 (UTC)
OK, but I was wondering about the long vowel in the NAV plus uncoloring h₃ and the collective plural (which was in the original paradigm per Ringe). --Ivan Štambuk (talk) 21:57, 4 August 2013 (UTC)

How can Armenian a- reflect both - isn't #h₁C > Armenian e- (secondarily i), and #h₃C > o- (later a- in open syllables)? Greek assimilation theory seems stretched (as well as the example because Greek ὀδούς (odoús) more likely reflects *h₃dont-, per Armenian and old compounds). I'd suggest that we keep Doric names which are usually mentioned as arguments for #h₁- which were removed. Kloekhorst also reconstructs initial #h₃- to account for hierloglyhpic Luwian á-laman and Lycian a-lãman), with the lack of ḫ- in Hittite according to him being regular in preconsonantal position). There is also Phrygian ονομαν which vocalized laryngeals the same way as Greek..

Also, you seem to forget that we're not suppose to do proselytize the "truth" or do OR, but present in a NPOV manner all of the mainstream opinions. I suggest that the comment on the reconstruction be split in three parts: 1) the initial laryngeal 2) the medial vowel (and laryngeal) 3) paradigm. For each part all of the problems and evidence should be listed, as well explanations listed in the literature. --Ivan Štambuk (talk) 20:32, 4 August 2013 (UTC)

Martirosyan (2010) derives Armenian անուն (anun) from Proto-Armenian *anuwn, from PIE PD n-stem *h₃néh₃-mn. --Vahag (talk) 21:17, 4 August 2013 (UTC)
I'm sorry if I gave the impression that the above arguments are my own OR; they're not, they're all from Stüber as cited in the references section. I was under the impression that all three laryngeals vocalize as a- in Armenian, but maybe that impression is mistaken. The "tooth" word also has a- in Armenian, and although our page is at h₃dónts and in spite of the discussion on the talk page, I think a reconstruction of "tooth" with h₃- is highly implausible. The very fact that "a tooth does not 'eat'; it only bites" is an argument against Aeolic edont- being changed from odont- under the influence of ἔδω. I wouldn't put any stock at all in the Phrygian word; we just don't know enough about that language to draw any conclusions from it. For all we know, all three laryngeals vocalized as o- in Phrygian; for all we know, ονομαν is a Greek loanword. Still, I concede the evidence for the quality of the laryngeal is ambiguous and could live with this page being moved to Appendix:Proto-Indo-European/Hnómn̥. As for a laryngeal before the m, it seems not very likely; the main motivation for it is to create a consonant-final root to which -mn̥ has been added, but to make that plausible there would have to be a verbal root h₁neh₃- that "name" is a verbal noun of, but there's no evidence for such a root and anyway "name" is semantically unlikely to be "the act of X-ing" or "the result of X-ing" anyway. And in that case the initial laryngeal would pretty much have to be h₁, because isn't one of the restrictions on root shape that you can't have two of the same laryngeal within a root, just like you can't have two obstruents with the same place of articulation in a root (no roots like **perp- or **klek-, right?)? But an o-grade neuter noun ending in a syllabic resonant is quite plausible; it fits in with ǵónu, dóru, móri, wódr̥, etc., etc. —Angr 18:44, 5 August 2013 (UTC)
But leaving the laryngeal out of the root makes all the long vowels hard to account for. It would imply that the Germanic ō is analogical, which is certainly possible but not really a preferred conclusion. And then there are the long vowels of Sanskrit and Latin. Latin, again, can be ascribed to analogy, but you see how having to resort to this many times weakens the reconstruction? —CodeCat 19:09, 5 August 2013 (UTC)
It doesn't really make any of the long vowels hard to account for except Germanic, and the Germanic forms are so late—and are only denominal verbs, not the noun itself—that it's hard to put much importance on them. The Sanskrit long vowel is regular by Brugmann's law. The Latin form has to be analogical anyway because the g's in cognomen and ignominia and the like prove that the "name" word got mixed up the with "know" word. So when we know it got a g by analogy anyway, it doesn't cost any extra for it to also get a long vowel by the same analogy. —Angr 20:28, 5 August 2013 (UTC)
But there is this mountain of evidence for #h₃-, and the only evidence for #h₁- are two Doric names, partially reconstructed and which possibly don't even contain the reflex of this PIE noun. Regarding the tooth word: it is a PIE-time derivation of the root which didn't survive directly in Greek, so it's unlikely that Ancient Greek speakers interpreted the tooth word as "biter", and later confusion with "eat" word is surely not surprising.. I prefer formal matches to ad-hoc explanations such as analogies anytime, but anyway, my opinion doesn't matter - what matters is NPOV and authorities and yeah the commonly cited reconstruction with #h₁- should be listed in the headword there as well (it was a mistake to remove it). My whole point was that pointing out the Greek "tooth" word as an argument of a similar of analogy supposedly occurring in the "name" word as well is problematic because possibly/probably no such analogy took place.
I agree with your objection on the root structure - but hey, that's not my opinions but that of Leideners. It should not be up to us to guess what is plausible or not, but what opinion is the most accepted, or the most up to date with modern scholarship. If there is no agreement then we should investigate the literature and present the evidence and arguments. It's no problem at all to have a dozen of possible reconstructions listed in the headword line. The only problem is how should the page be named, and to what should the reconstructions in the mainspace entries etymologies' link to. Hnómn̥ seems neutral with respect to the initial laryngeal, but the medial one not so.. Perhaps HnV(H)mn̥? In the etymologies we could display e.g. "From PIE *Hnómn̥, *Hnéh₃mn̥". --Ivan Štambuk (talk) 01:08, 6 August 2013 (UTC)

Armenian անուն (anun) in this case points to *h₃-, because PIE *h₁C- would give Arm. ե- (e-); compare երեկ (erek) < PIE *h₁régʷos, but արեւ (arew) < PIE *h₂rew-i-. As Martirosyan points out, the evidence for this rule is meagre, but we do have other examples: իմ (im), ինն (inn) (with secondary e > i before a nasal), possibly also ելանեմ (elanem). I think the page should be moved. --Vahag (talk) 20:06, 4 January 2015 (UTC)

Return to "Proto-Indo-European/h₁nómn̥" page.