Reconstruction talk:Proto-Indo-European/h₃dónts

Beekes suggests that this is actually h₃dent-. This might be worth noting as an alt form. -Atelaes λάλει ἐμοί 17:05, 16 May 2008 (UTC)

Quote: "However, a tooth does not “eat”; it only bites." ^_^ Yes, his argumentation is excellent. --Ivan Štambuk 17:27, 16 May 2008 (UTC)

There is no independently attested root *h₃ed- with this meaning. It's pure invention by Beekes, sorry. The explanation needs amendment, and the article moved back, to reflect the uncertainty involved in this proposal. It is highly problematic and against the consensus in the field.

It's not even consider necessary in the first place. The dismissal of the Aeolic form as secondary is not justified. LIV gives the meaning of the aorist (!) present *h₁ed- as originally "bite", from which "eat" is derived based on the iterative sense "to bite repeatedly" of the Narten present. Schumacher has pointed out Lithuanian uodas "mosquito" as evidence that the root indeed had the meaning "bite" originally. The principle of parsimony prohibits the postulation of two separate roots in such a case. --Florian Blaschke (talk) 14:13, 17 February 2013 (UTC)

The only way to determine the initial laryngeal is through Greek and Armenian. In Greek, either Aeolic *ed- is archaic, and odont- was assimilated from *edont-; or Ionic-Attic od- is archaic and *odont- was analogically changed to edont- under the infulence of ἔδω. On the other hand, Armenian at- reflects either *h₂d- or *h₃d-, which ultimately reduces the possibility space to *h₃ed-.

Lithuanian úodas and Latvian uôds can be derived from either root (o-grade + Winter's law) and formally prove nothing. If indeed we're dealing with a pre-Late PIE derivation preserving the alleged original meaning of *h₁ed- "to bite", one would expect additional evidence in form of cognates in branches beside Baltic, of which there are none. "to eat" is a Swadesh verb that isn't likely to change in the language history. Rather than postulating a far-fetched semantic shift to explain away the discrepancy of meaning, it's more economic to fit these as normal derivations of a separate root for which there is actual evidence.

For Greek: assimilation always trumps analogical leveling which is a trump card where there is no argument left. Additionally, old forms such as νωδός (beside post-laryngeal ἀνόδων) and αἱμωδέω strictly require *-h₃-.

Apart from Beekes, there are others that endorse this explanation. Even Sihler 1995:89 admits that it is more probable in the laryngeal framework, beside mentioning the old doctrine. LIV has some problematic reconstructions (e.g. postulated on the basis of a single branch, unsurprisingly Germanic :) so I wouldn't take it as definite evidence of a mainstream opinion. --Ivan Štambuk (talk) 14:22, 24 May 2013 (UTC)

LIV only postulates some reconstructions based on evidence from a single branch (not always Germanic – sometimes it's Indo-Aryan or some other branch) because of 1) the possibility that roots have only survived in a single branch (after all, we know this kind of thing does happen, for example from Romance) and 2) the possibility that cognates (or derivations) of the verb will be found or identified in other branches later and the authors do not wish to rob future linguists of avenues for further research by not even mentioning these possibilities (this is explained in the foreword quite explicitly, which you clearly didn't read :-P). Methodically, the reconstruction of a word or root based on a single branch if there is no particular reason to think it is not inherited from the proto-language is sound (we also reconstruct some words back to Proto-Germanic here that are only attested in West Germanic, some of which cannot even have been present in Proto-Germanic because they are borrowings from Latin!). Of course, such reconstructions are far less reliable than those based on at least two (preferrably sufficiently remote) branches and therefore the single-branch reconstructions are marked with an asterisk as uncertain. Your pathetic attempt to undermine the credibility of the LIV is therefore completely baseless and you should beg the authors' pardon. Apart perhaps from Rix's reconstruction of the PIE verbal system, which the authors acknowledge themselves is a bit idiosyncratic and not universally acknowledged (but then, which reconstruction of the verbal system is?) the LIV is as mainstream (and responsibly cautious!) as can be, certainly more so than the Leiden School.

In fact, what is far-fetched is the reconstruction of a root *h₃ed- "to bite" based on such flimsy and ambiguous evidence. If you had at least suggested that this root might be identical with the root for "to smell, to stink" (according to LIV: "to start to smell"!), and that "to bite" was its original meaning, which would make some sense at least (compare beißender Gestank, "pungent stench"), I'd have found the idea more interesting. But you didn't, so I remain unconvinced. :-P (BTW, LIV also mentions the idea that the root for "to hate" might be identical with the "smell" root, but is sceptical. That said, the semantic development from "I bite" to "I start to hate" – and "I have bitten" to "I hate" – is perhaps even more attractive than the development from "I start to smell" to "I start to hate".) But you blew it, this possibility to redeem your position has already passed. :-P --Florian Blaschke (talk) 10:52, 11 August 2014 (UTC)

All such reconstructions are garbage because they're not based on comparative method, because there is nothing to compare them to. But we are even worse - there are some protolanguage appendices on Wiktionary which are based on postulating a reconstruction as an intermediate stage between two other reconstructions (the supposed "parent" and "child"). You can only create reconstructions on the basis of evidence, which are actual attestations, not the other way around. Attested words do not actually inherit from reconstructed forms (the term reflex is much more preferred), and LIV authors evidently suffer from the same Stammbaumitis delusion as many other historical linguists. According to LIV data the branch having the most reflexes is Germanic which is just laughable and clearly reflects authors' cultural bias, whether conscious or unconscious. Leiden School is the world's leading publisher of etymological dictionaries and that their innovative interpretations are not perceived as "mainstream" or by the rest of the field still mostly stuck in the 19th century notions of language change is a testament to the suspiciousness or invalidity of their theories is hardly a valid counterargument :P

Regarding the root - the evidence is still in favor of */h₃/, regardless of the validity of *h₃ed-. You are free to expand its entry that it's based on flimsy evidence, if you can find sources. --Ivan Štambuk (talk) 12:12, 11 August 2014 (UTC)

Slavonic *zo[n]b => Serbo-Croat "zub"/Macedonian "zab"/Slovene "zob" is also from *h3don- + -b, but it would require lotsa work to enter all the IPA transcription signs. Maybe someone of you kids who don't have anything serious to work on could make an effort, eh?

Proto-Slavic *zambu is cognate with Sanskrit jambha- "peg". No connection to this word. --Florian Blaschke (talk) 14:13, 17 February 2013 (UTC)

We have an entry for it already: *zǫbъ. —CodeCat 14:53, 17 February 2013 (UTC)

Cool! Obviously I used the Holzeresque Baltic-like reconstruction, if for no other reason that it is far easier to type (well, I have other reasons too, actually); the "Proto languages" character set may not have existed yet at the time. --Florian Blaschke (talk) 10:52, 11 August 2014 (UTC)

Etymology section needs work

I think this point has been belaboured in previous entries, but the section is entirely speculative, provides no citations, and seems to be based, at least partially, on primary research. Beekes' PhD dissertation is a fantastic work, but it's not the final word (by any means) on this topic. But there are other issues in this section such as the part where it claims Gk νωδός "requires *n̥-h₃d- (where */h₃/ was regularly vocalized to ό in interconsonantal position)," but the *h₃ there isn't depicted as being interconsonantal: the word-initial nasal is syllabic (the little circle under the n), and consequently the *h₃ is not interconsonantal (but it doesn't need to be, cf. Gk νώ 'we two' < PIE *n̥h₃wé Ringe 2017:86 citing Katz 1998:89-99, 212-17)... The statement "[t]hus, the word is ultimately an active participle of the root *h₃ed-..." is really reaching, as that root can barely be posited for PBSl, let alone for PIE (it's not even attested in Slavic, and could very easily be a Baltic innovation). At the very least, the section would be better if there were citations. Vindafarna (talk) 04:02, 19 July 2022 (UTC)