Reconstruction talk:Proto-Celtic/gleiwos

Etymology

@CodeCat, Victar, I know Code mentioned this already, but this etymology doesn't make any sense at all. *ǵʰley(H)-wos would produce the correct result, but such a form cannot really come from *ǵʰelh₃-. Also, one expects a zero-grade with *-wós. —JohnC5 17:30, 27 October 2016 (UTC)

Hah, now we have 3 talk pages on this word: Talk:Gloyw and Reconstruction_talk:Proto-Brythonic/gluɨw.

Yeah, I'm not sure where *-h₂- comes from in Matasociv's *ǵʰleyh₂- reconstruction. Proto-Germanic *gliwēną (“to shine, glitter”) and *glīmaną (“to shine”), should both come from *ǵʰley-, and are both cited to be linked to *ǵʰelh₃-. I'm limited in my knowledge of Greek, so I don't know how Ancient Greek χλιαινω (khliainō, “warm”) connects. --Victar (talk) 18:29, 27 October 2016 (UTC)

I'm very hesitant about deriving roots from other roots. In almost all the cases where it's done, it's ad hoc and doesn't really reflect any known derivational process. This is a prime example of that. —CodeCat 19:37, 27 October 2016 (UTC)

It's strange though, because most everywhere I see, English gold is derived from *ǵʰel- (“to shine”), making it hard to image that *ǵʰley- (“to shine”) isn't related. It's certainly possible, but then we might have to really look at which derivations belong to which. --Victar (talk) 20:46, 27 October 2016 (UTC)

I think that we should only posit relationships if we have definite explanations of how they are related, and have multiple independent concurring sources (independent as in, not all from Leiden). —CodeCat 20:54, 27 October 2016 (UTC)

*ǵʰel- (“to shine”) is straight from Pokorny. It's the Wiktionary entry that's at odds with everyone else to not include the definition "to shine". --Victar (talk) 21:12, 27 October 2016 (UTC)

Pokorny didn't do laryngeals, so of course he wouldn't have *ǵʰelh₃-. And maybe he was wrong on the meaning too? How many of the descendants of the PIE root actually mean "shine"? Most refer to colours. —CodeCat 21:27, 27 October 2016 (UTC)

@Victar: Matasovic gives only *ǵʰleyh₂- and Kroonen, *gʰley-wo-. Beekes gives *ǵʰley(d)- for χλιαινω (khliainō) and mentions Lithuanian žlejà (“darkness, twilight”). I've yet to find anything linking to *ǵʰelh₃-. Where is your link to that root? —JohnC5 21:34, 27 October 2016 (UTC)

You may very well be right, and we have a confusion of two roots, *ǵʰelh₃- (“yellow”) and *ǵʰelh₂- (“to shine”). If so, we need to change a lot of wiktionary entries that say *ǵʰelh₃- (“to shine”). I found another derivative of *ǵʰelh₂- in Vann (2008): *gʰlh₂dʰ- (“shiny, smooth”) (Proto-Germanic *gladaz (“shiny, smooth”)); *gʰleh₂dʰ- (“smooth”) (Old Church Slavonic гладъкъ (gladъkъ, “smooth, even”)); *gʰlh₂dʰro- (“smooth”) (Latin glaber (“bald, smooth”)). --Victar (talk) 19:05, 28 October 2016 (UTC)

De Vaan 2008 confirms *ǵʰleh₂dʰ-, but says nothing about any connection with other roots. Kroonen 2013 explains *gladaz as from a root *gʰleh₁- (which he also derives *glōaną from, though I find the o-grade odd), comparing it to the post-PIE formation also found in *bladą from *bʰleh₃-. —CodeCat 19:13, 28 October 2016 (UTC)

I think what we're trying to do is now is split *ǵʰel- properly into *ǵʰelh₃- (“yellow”) and *ǵʰelh₂- (“to shine”). Yeah, Kroonen (2003)'s etymology seems a bit wonky to me, but either way, if you look at the entry for *gladaz, you'll find it pointing to *ǵʰel- (“to shine”), so obviously this all needs straightening out. --Victar (talk) 19:24, 28 October 2016 (UTC)

Do we have more sources for *ǵʰelh₂-? Matasović maybe? —CodeCat 19:25, 28 October 2016 (UTC)

Matasović only gives *ǵʰleyh₂- (“to shine”), as that appears to be the only form to survive in PCelt. --Victar (talk) 19:33, 28 October 2016 (UTC)

Kroonen drops the laryngeal and has no palatovelar, so *gʰley-. —CodeCat 19:42, 28 October 2016 (UTC)

Proto-Celtic *glanos (“clean, clear”) actually looks be from *ǵʰelh₂-. Matasović tries to connect it to *ǵʰelh₃- based on Pokorny *ǵʰel-, but *ǵʰlh₃nós doesn't work as well as *ǵʰlh₂nós would. --Victar (talk) 20:11, 28 October 2016 (UTC)

Neither of them work phonetically. You'd expect a long ā, like in *ɸlānos. —CodeCat 20:14, 28 October 2016 (UTC)

h₂ often yields a short a, ex. *ɸatīr.

That's next to an obstruent. After a sonorant, l, m, n, r, the outcome is a long vowel. See w:Proto-Celtic. —CodeCat 20:33, 28 October 2016 (UTC)

What if we threw Joseph's rule at it, with *ǵʰelh₂nós > *ǵʰlanós > *glanos? --Victar (talk) 20:43, 28 October 2016 (UTC)

Actually Matasović cites Dybo's law as a possible reason for the shift. --Victar (talk) 20:56, 28 October 2016 (UTC)

Old Irish gel (“white, fair, shining”) might also be from *ǵʰelh₂- instead of *ǵʰelh₃- (“yellow”), which Matasović cites as the root for it and Middle Welsh gell (“yellow”), but is probably from Proto-Celtic *gelwos < *ǵʰelh₃wos. --Victar (talk) 21:07, 28 October 2016 (UTC)

Proto-Indo-European *ǵʰelh₂os would yield Proto-Celtic *gelos, also thanks to Joseph's rule. --Victar (talk) 21:11, 28 October 2016 (UTC)

The University of Wales Celtic Lexicon reconstructs a Proto-Celtic *gleiso-, *gleisto- (“shining”), but I'm not sure what the basis of the reconstruction is. --Victar (talk) 22:26, 28 October 2016 (UTC)

Actually, I just noticed that there's no palatovelar in De Vaan, so it's *gʰleh₂dʰ-. —CodeCat 19:27, 28 October 2016 (UTC)

Ah, thanks, sorry, mis-copy. --Victar (talk) 19:31, 28 October 2016 (UTC)

Mallory (2006) also cites *ghlehₓdh-(ro)- (“smooth”) as from *ghel- (“to shine”). --Victar (talk) 19:53, 28 October 2016 (UTC)

The thing with these roots derived from roots is that the method of derivation is often left completely unexplained. In this particular case, Mallory adds -ehₓdh- to the zero grade of the root, yet the nature of this derivation is left completely unexplained. It's just like "yeah, here's the derivation. There's no parallels, and no explanation, I just pulled it out of my hat, and you'll just have to believe me." —CodeCat 20:11, 28 October 2016 (UTC)

We've just listed a ton parallels above. I was just giving an additional source to cite. --Victar (talk) 20:32, 28 October 2016 (UTC)

Another source we can cite is Etymologisches Wörterbuch der deutschen Sprache entry for German glatt. --Victar (talk) 21:58, 28 October 2016 (UTC)

Does anyone have a copy of Evidence for *ghelh₂-, a new Indo-European root (2003)? I think that might help us out in this. --Victar (talk) 18:45, 28 October 2016 (UTC)

Here's my first go at creating an entry for Proto-Indo-European *ǵʰelh₂-: User:Victar/Proto-Indo-European/ǵʰelh₂- --Victar (talk) 04:10, 29 October 2016 (UTC)

@Victar: In your version of *ǵʰelh₂- as in *glēwos, you add *-y- into the PIE root. There's no derivational system to add it and the derivatives of *ǵʰleyh₂- cannot be from *ǵʰelh₂-. Also you have a lot of forms from *ǵʰleh₂- which could be from Schwebeablaut, but we would consider that a separate entry. —JohnC5 02:22, 31 October 2016 (UTC)

Thanks for the reply, @JohnC5. I've been thinking a lot about that it. All I can think of is that original root is *ǵʰel- and that *ǵʰelh₂- is a laryngeal extension. That makes *ǵʰel- to *ǵʰley- or even *ǵʰleyh₂- explainable. What are your thoughts? --Victar (talk) 03:15, 31 October 2016 (UTC)

That is far better.—JohnC5 03:17, 31 October 2016 (UTC)