Reconstruction talk:Proto-Celtic/abūl

RFM discussion: February–March 2024

The following discussion has been moved from Wiktionary:Requests for moves, mergers and splits (permalink).

This discussion is no longer live and is left here as an archive. Please do not modify this conversation, but feel free to discuss its conclusions.

---

Reconstruction:Proto-Celtic/abalom → Reconstruction:Proto-Celtic/abūl

David Stifter argued in An apple a day... that the Old Irish word for "apple" ubull was originally a u-stem (section 2/section 3 in the print version) and that therefore the Proto-Celtic word can't be thematic, the most attractive option being *abūl (section 4/5). He then argues on the basis of Balto-Slavic that the PIE word was an l/n-stem *h₂ébōl ~ *h₂ebnés > Proto-Celtic *abūl ~ *aμnos > *abūl ~ *abalnos, the genitive being the basis for *abalnā (“apple tree”) (section 5/6). Any objections? @Mellohi!, @Victar, @Mahagaja. —Caoimhin ceallach (talk) 00:35, 21 February 2024 (UTC)

Support *Abalom is impossible given the u in Goidelic. Stifter's reconstruction is infinitely more evident on that alone by easily explaining the u by providing a source of u-infection. — Ceso femmuin mbolgaig mbung, mellohi! (投稿) 03:14, 21 February 2024 (UTC)

{{R:sga:NWAI|120}} claims that the Celtic was rebuilt as an n-stem Proto-Celtic *ablū ~ *abalnos (> *aballos), which would the only way to get such a genitive. --{{victar|talk}} 05:13, 21 February 2024 (UTC)

That is indeed what De Bernardo Stempel claims. Stifter's arguments against it are that 1) it ignores Proto-Brythonic *aβal and 2) the move from an n-stem to u-stem in Old Irish is unlikely. And it's not true that this is the only way to get a genitive *abalnos. —Caoimhin ceallach (talk) 14:35, 21 February 2024 (UTC)

@Caoimhin ceallach:

1. Can you explain how *ablū ignores PByth *aβal and how *abūl would be better?
2. Why would *ablū becoming a u-stem in Old Irish be less likely than *abūl?
3. Genitive *abalnos can only come from *h₂ebl̥-n-, which is supported by the paper you link above.

--{{victar|talk}} 18:22, 21 February 2024 (UTC)

Latin latrō > Early Brythonic *latrū > Late Brythonic *llėdr (modern Welsh lleidr) implies that *aβal cannot come from earlier ^×ablū, as the latter would have given ^×ėβl rather than *aβal. On the other hand, ^×abūl isn't much better as it would give ^×ėβil, not *aβal. —Mahāgaja · talk 20:41, 21 February 2024 (UTC)

1. What Mahagaja said. On the other hand *abūl > *abal (generalised zero grade suffix) would be readily thematised to Brythonic *aβalo-.

2. n-stems are extremely productive and u-stems vestigial.

3. That's not true. *abalnos is argued to be a contamination of the strong and weak stems, similar to Latin itineris. —Caoimhin ceallach (talk) 20:49, 21 February 2024 (UTC)

@Caoimhin ceallach:

1. PByth *aβal can't be from either nominative, as Mahagaja pointed out, but is likely instead from the genitive *abalnos (> *aballos), compare *brusū, where the *-ll- was leveled to *-l-. The only alternative would be a PC nominative *abl̥, which has been suggested.
2. As de Bernardo Stempel theorizes, *ablū became a u-stem due to the similarity of suffix, which seems just as likely as *abūl sporadically becoming a u-stem. PC *uxsū is another example of an n-stem becoming an o-stem in Irish.
3. Rebuilding r/n-stems genitives to r-stems by appending a r-stem genitive suffix is one thing, PIE *h₁éytr̥ ~ *h₁iténs > PIt. *eitor ~ *itenos > Lat. iter ~ itin-eris. Inserting *-al- in the middle, like in your reconstruction steps, is another thing entirely. Who's suggesting this?

--{{victar|talk}} 03:28, 22 February 2024 (UTC)

@Victar:

1. Yes, it can. I've told you how. You might be able to concoct a semi-plausible account (though I don't understand it—how was *-ll- leveled to *-l- exactly?), but only if you ignore the fact that this is not how amphikinetic n-stems were treated in Brythonic. They were umlauted and had n-plurals, e.g. Middle Welsh ceneu (“whelp”), pl. canawon.

2. Again, it's perhaps possible, but unlikely, for the reasons that I've given. *uxsū is a poor parallel. It was neither an amphikinetic n-stem nor did it become a u-stem.

3. Stifter. —Caoimhin ceallach (talk) 10:34, 22 February 2024 (UTC)

@Caoimhin ceallach:

1. How was it leveled? The same as in Irish, just the other way, from the nominative to the genitive.
   Brythonic rebuilding n-stems from the genitive is common, *aβon, *bronn, *aren, etc.
2. Again, de Bernardo Stempel would disagree with that opinion.
   Old Irish ubull is an o-stem as well, and I wouldn't be surprised if Old Irish oss also started as a rebuilt u-stem.
3. Stifter reconstructs *aμnos > *abalnos?

--{{victar|talk}} 21:14, 22 February 2024 (UTC)

1. Firstly, can you expand on that? Because I don't see how that could work. Secondly, I think those are ā-stems derived from the weak stem. "Rebuilt" is the wrong term. They are good comparanda for *aβaln (“apple tree”). If you assume there was a Proto-Celtic n-stem, you'd have tricky task of explaining how both Proto-Brythonic *aβal and *aβaln can be derived from the genitive.

2. The raison d'être of this post is that Old Irish ubull was originally a u-stem. Claiming oss a u-stem is very bold.

3. Yes. I think you're better off reading the relevant section of the paper. —Caoimhin ceallach (talk) 23:28, 22 February 2024 (UTC)

@Caoimhin ceallach:

1. What don't you see working, the whole concept of *l ~ *ll leveling? Where do you believe the -ll in Old Irish ubull originates if not leveling?
   Proto-Brythonic *aβal (“apple”) is rebuilt from the genitive, whilst Proto-Celtic *abalnā (“apple tree”) is derived from the genitive.
2. You think?
3. Can you point me to the page you see that? Thanks.

Matasovic suggests that the -ll- was under the influence of Proto-Celtic *abalnā (“apple tree”). If we throw the idea of a l/n-stem and follow Matasovic in reconstructing an original l-stem with a laryngeal, PIE *h₂éph₃ōl ~ *h₂éph₃los, that would arrive us at Proto-Celtic *abūl ~ *abalos.

--{{victar|talk}} 00:39, 23 February 2024 (UTC)

1. Yes, I would like to see how the analogy worked. It is unclear where the ll in ubull comes from. It could be some kind of leveling, but it's more likely that is was either the result of MacNeill's law or the influence of aball "apple tree".

2. Yes...

3. "Das Kollektivum dagegen hätte, unter der geringfügigen Annahme des innerparadigmatischen Ausgleichs in der anlautenden Silbe, zur wesentlich undurchsichtigeren Alternation *abūl, *aμnos geführt.⁴⁵ Dass in einem solchen Fall z.B. zu *abal, *abalnos oder *abūl, *abūlnos oder gen. *abolnos oder sogar *abalnos ausgeglichen worden wäre, um einen höheren paradigmatischen Zusammenhalt zu gewährleisten, ist keine besonders gewagte Annahme." On the end of the first page of section 5 (section 6, page 204 in the print version). Caoimhin ceallach (talk) 09:17, 23 February 2024 (UTC)

Really? I thought the whole point to reconstructing an r/n-stem is to use leveling as a reasoning for the -ll in Old Irish ubull. PIE languages love leveling germinated consonants -- Germanic in infamous for it. Proto-Celtic *abalnā (“apple tree”) could just as well be from *h₂ebl̥-neh₂ without requiring it deriving from a genitive, which is usually how it's reconstructed anyway.

Thanks for sharing the page. I have to disagree with Stifter here saying "keine besonders gewagte Annahme". Such a leveling is ad hoc with no parallel examples. Conventional understanding would tell us that the only way a nominative stem is spread to genitive is when it's rebuilt as a new inflection type.

We could just go with two separate reconstructions, *abūl ~ *ablos (cf. *swesūr ~ *swesros) for the Irish, and *abal ~ *ab(a)nos (cf. *ɸatar ~ *ɸat(a)nos) for the Brythonic, instead of trying to merge them into a single paradigm. @Mahagaja?

--{{victar|talk}} 02:07, 24 February 2024 (UTC)

MacNeill's law means we don't have to worry about the ll in ubull. I also agree that the -nā of *abalnā is probably just a suffix rather than the word for "apple tree" being derived from the genitive of the word for "apple". If Stifter is right about the etymology of Áth Aublo, then the genitive of the word for "apple" was (eventually) the u-stem form *ab(V)lous. The question then is whether the genitive could be re-formed as a u-stem without the nominative also being re-formed. Would Proto-Celtic or Proto-Goidelic allow nominative *abūl/genitive *ab(V)lous? Or does the genitive imply a nominative *ab(V)lus? (If it does, it has to be *abVlus, because gabul shows that MacNeill's law doesn't apply when the u between b and l is epenthetic, see w:Phonological history of Old Irish § MacNeill's law.) One thing I don't understand is why people keep wanting to find an n-stem or l/n-stem here. We reconstruct *h₂ébōl as a pure l-stem, and I don't see any evidence to suggest the presence of an n in the "fruit" word anywhere in Celtic. —Mahāgaja · talk 10:28, 24 February 2024 (UTC)

This is indeed a flaw in Stifter's paper. He doesn't even mention the possibility of a nā-derivative, even though it would be quite straightforward. On the other hand, the examples Victar gave above are exactly the kind of ā-derivatives from weak stems that Stifter claims abalnā is, showing that this was a productive mechanism.

Your second point is a problem I also have with the paper: he doesn't address how aublo arose. But it would have to be after the nominative *abūl became indistinguishable from a u-stem because of apocope, u-colouring, and shortening of unstressed long vowels. The nominative didn't need to be rebuilt because it was a hinge between the old inflection and the new.

The reason for reconstructing an l/n-stem is discussed in the three pages of section 5/6 "Zum Ablaut von *abVl-". In short: l-stems otherwise don't exist and l, n often occur together in this word and the other PIE l/n-stem *sóh₂wl̥ in several daughter branches. —Caoimhin ceallach (talk) 11:40, 24 February 2024 (UTC)

*séh₂ls is an l-stem, isn't it? Not to mention the three other nouns in Category:Proto-Indo-European terms suffixed with *-ōl. Stifter's "evidence" that *h₂ébōl was an l/n-stem seems really flimsy. —Mahāgaja · talk 13:14, 25 February 2024 (UTC)

Yes, that's a real possibility. And if it's true, I can see that Stifter's theory is a great deal less compelling. —Caoimhin ceallach (talk) 09:49, 26 February 2024 (UTC)

*h₂ebl̥-neh₂ is a sensible alternative. Your other points are bizarre.

Extreme irregularity is rarely maintained and leveling is the obvious remedy. Surely you know that. Obviously there are no parallel examples, because *h₂ebōl was one-of-a-kind. The only similar word is *sóh₂wl̥ and that is exactly why Stifter tries to draw parallels with it. Instead of reconstructing the most plausible form you're suggesting we just not bother?

"Conventional understanding would tell us that the only way a nominative stem is spread to genitive is when it's rebuilt as a new inflection type." How can you utter such nonsense? I'm speechless. —Caoimhin ceallach (talk) 11:58, 24 February 2024 (UTC)

Counter examples? --{{victar|talk}} 21:12, 24 February 2024 (UTC)

*mori. The full grade of the nominative stem replaces the zero grade of the genitive. Nothing is rebuilt as a result. *rīxs. The lengthened grade is spread from the nomintive to the genitive, but it remains a root noun. Nothing is rebuilt. —Caoimhin ceallach (talk) 10:07, 25 February 2024 (UTC)

I think perhaps I wasn't clear. Leveling grades on the root is obviously common place. The step I'm talking about is *abūl ~ *aμnos > *abūl ~ *abalnos, where the *-l of the nominative inflectional ending is transferred to the genitive. This only occurs when the inflection type is rebuilt upon the nominative, and why you would need *abl-ū ~ *abal-nos to get to an *-ln- genitive. --{{victar|talk}} 00:26, 26 February 2024 (UTC)

I can't think of any examples right now, but I know that analogy comes in many shapes and sizes. The point is to reduce irregularity in a paradigm, but it's not all or nothing. If you can spread a vowel, or just a consonant, as in Sanskrit पन्थाः ~ पथः (panthāḥ ~ pathaḥ), Ι don't think it's a stretch of the imagination to say that a syllable can be spread. But alright, maybe it needs some better parallels to stand on its own. —Caoimhin ceallach (talk) 10:03, 26 February 2024 (UTC)

Imaginations are fun. --{{victar|talk}} 02:56, 27 February 2024 (UTC)

Support I concur with Mellohi's response above. -saph 🍏 16:17, 21 February 2024 (UTC)

Modified proposal: Proto-Celtic *abūl, *abalos from earlier *abūl, ablos, d.pl. *abalbos. The strong stem survived in Goidelic as a u-stem and the weak stem was thematised in Brythonic. *abalnā "apple tree" is a *neh₂-derivative as Victar suggested, which ought to be somehow related to Proto-Slavic *abolnь and Old Prussian wobalne. —Caoimhin ceallach (talk) 12:37, 26 February 2024 (UTC)

Support: Works for me. --{{victar|talk}} 02:56, 27 February 2024 (UTC)

Support, though I would say "in Goidelic as a u- and o-stem" since only o-stem inflection is actually attested in the common noun. —Mahāgaja · talk 09:07, 27 February 2024 (UTC)

Moved —Caoimhin ceallach (talk) 16:51, 23 March 2024 (UTC)