The Great Proto-Indo-European Thread's Sequel

[Zju]

Indeed that's it! Regarding the Kartvelian form, it's discussed in another paper by the same author.

[Znex]

Dunno if this has been discussed already here, but PIE *Hebl- 'apple' may not be a loanword at all, and instead just be a metastethised form of *meHlom. Can't track down the paper from academia.edu I originally read that in, but the gist is that an intermediate form *Heml- underwent ml → bl. Also, an s-mobile form of *meHlom allegedly gave rise to proto-Kartvelian *msxali 'pear' via **smxali < **smaxli.

Does the author mention anything about a *m > *b change in other positions that would explain Burushaski báalt?

It is worth noting that a number of languages in central Asia likely denasalised initial nasals, including language families like Turkic and Yeniseian, which both in their proto-stages disallowed initial sonorants (including /r l/ as in IE, but also initial nasals and glides).

It won't be as productive in languages outside of that particular area, but Turkic certainly had its influence on other languages through the millenia.

[WeepingElf]

Might the paper in question be Fenwick (2016)?

That's an interesting paper, thank you!

Indeed that's it! Regarding the Kartvelian form, it's discussed in another paper by the same author.

Also interesting. The similarities between IE and Kartvelian may be due to Kartvelian also having been influenced by the same Afroasiatic substratum I conjecture in PIE as the cause for the typological divergence of PIE from the "common Mitian" type displayed by such families as Uralic or Turkic.

[Ketsuban]

What is the Proto-Indo-European word for "mouse"? Most resources I can find say *muHs with an uncertain laryngeal whose only effect is producing compensatory lengthening (e.g. Latin mūs, Greek μῦς) but Wiktionary has *muh₂s without elaboration. I did note in Etymological Dictionary of Latin and the Italic Languages the proposal that Tocharian B maścītse is from *muHs-ti- > *mwasti- which could be taken to imply a *h₂, although of course there's no citation to that effect, Lubotsky doesn't actually give *muh₂s as the PIE form, and the Tocharian dictionary I found in a web search (to double-check the spelling, because I don't trust copying text from PDFs) remains somewhat ambivalent about the etymology of maścītse other than to say that a derivation from *muHs "would be nice".

[Glass Half Baked]

The only sources Wikipedia cites are Iranian dictionaries, but the Iranian reconstruction given is muHs, so I doubt the author of this page knows something we don't know. Just typical overzealousness.

[hwhatting]

Might the paper in question be Fenwick (2016)?

That's an interesting paper, thank you!

Indeed that's it! Regarding the Kartvelian form, it's discussed in another paper by the same author.

My thanks to Ketsuban and Zju as well. It's funny that Fenwick's ideas are partially close to what Taskubilos stated in the discussion starting about here three years ago; but as she is arguing much more rigorously than he ever does, it sounds more plausible. (Come to think of it, it's also possible that Fenwick is the source of some of his notions.)

[Ketsuban]

Also from the linked thread...

In fact, 'apple' happens to be a Wanderwort found in several families, including IE itself (although somewhat disguised). :)

What families? Why is the loanword hypothesis better than the idea that *h₂ébl̥/*h₂ébōl is an irregular metathesis of *méh₂lom, which has much the same relationship to *méh₂- "ripen, mature" that *webʰl- "beetle, worm" (English weevil) has to *webʰ- "weave"? (Also, are you using Wanderwort to mean "loanword"? You seem to like declaring prehistoric Wanderwörter and filling in the ignoramus-et-ignorabimus blanks with magic to justify them.)

I don't think I'd ever seen Fenwick's work prior to looking it up for this thread, and yet this is every point Fenwick makes except the connection to words in Sanskrit and Hittite which fill in both the intermediate step in the metathesis as well as the justification for it ("apple" being altered to appear similar to the root for "sour").

[bradrn]

Also from the linked thread...

In fact, 'apple' happens to be a Wanderwort found in several families, including IE itself (although somewhat disguised).

What families? Why is the loanword hypothesis better than the idea that *h₂ébl̥/*h₂ébōl is an irregular metathesis of *méh₂lom, which has much the same relationship to *méh₂- "ripen, mature" that *webʰl- "beetle, worm" (English weevil) has to *webʰ- "weave"? (Also, are you using Wanderwort to mean "loanword"? You seem to like declaring prehistoric Wanderwörter and filling in the ignoramus-et-ignorabimus blanks with magic to justify them.)

I don't think I'd ever seen Fenwick's work prior to looking it up for this thread, and yet this is every point Fenwick makes except the connection to words in Sanskrit and Hittite which fill in both the intermediate step in the metathesis as well as the justification for it ("apple" being altered to appear similar to the root for "sour").

I feel those intermediate steps are key, though. Without them, you have an interesting-looking resemblance but not much else. With them, you have a highly convincing case.

[WeepingElf]

Might the paper in question be Fenwick (2016)?

That's an interesting paper, thank you!

Indeed that's it! Regarding the Kartvelian form, it's discussed in another paper by the same author.

My thanks to Ketsuban and Zju as well. It's funny that Fenwick's ideas are partially close to what Taskubilos stated in the discussion starting about here three years ago; but as she is arguing much more rigorously than he ever does, it sounds more plausible. (Come to think of it, it's also possible that Fenwick is the source of some of his notions.)

Indeed, not everything Talskubilos stated was wrong or implausible; the problem was that he did not adduce sufficient evidence, and asserted that it was the truth rather than just ideas to explore. I admit that my ideas are just ideas and I cannot prove them - which is the main reason why, instead of making bold claims about actual unknown languages, I pursue them in my conlangs.

[abahot]

Back when I had just discovered this forum and was reading through the discussion on PIE (and it might have been here or on the old forum), I seem to recall someone hypothesizing that *h₃ was a rhotic rather like the one in English. Does anyone have any unconventional ideas on the pronunciation of laryngeals?

(I remember back a while ago I somehow developed a crazy idea that laryngeals were sort of analogized into existence. This is of course completely illogical, and I can't even remember how I arrived at this conclusion, but somehow I got from "maybe putting laryngeals everywhere is a bit overzealous" to "they literally did not exist in the first place". This idea was discarded after about 8 seconds of course, but I do still wonder if putting laryngeals everywhere is uncalled for. Like in the "mouse" example, we might not need to reconstruct an unspecified *H if we accept that sometimes vowels could be long in PIE. Not sure how much sense this makes though.)

My personal unconventional laryngeal theory:

I read somewhere once that *h₂ was thought to be a uvular/velar fricative before retracting to pharyngeal and then disappearing entirely in most branches of IE. On that note, it seems a bit strange to me that the reconstructed PIE phoneme inventory has basically no post-velar anything except for these three laryngeals. (Maybe the answer is that the "plain velars" were really uvulars or something of that sort.) I think that if *h₂ debuccalized somewhat, maybe the other laryngeals did a similar process, starting at places of articulation that would have matched the rest of the reconstructed phoneme inventory. Not sure what those places would be though ...

[WeepingElf]

As for PIE lacking post-velar consonants other than the laryngeals, Proto-Semitic is reconstructed in a similar way (though some Semitic languages at least have shifted /k'/ to /q/), and this is indeed the reason why Indo-Europeanists call them "laryngeals".

[Travis B.]

As for PIE lacking post-velar consonants other than the laryngeals, Proto-Semitic is reconstructed in a similar way (though some Semitic languages at least have shifted /k'/ to /q/), and this is indeed the reason why Indo-Europeanists call them "laryngeals".

Thing is, the *k *kʷ *q hypothesis for PIE dorsal plosives seems more solid than the "conventional" *k̟ *kʷ *k hypothesis, because crosslinguistically /k/ is more common than /k̟/ (or /c/ or whatever you want to call it), yet the conventional analysis essentially states that *k̟ is more common than *k -- which can be conveniently resolved by hypothesizing that conventional *k̟ is really *k and conventional *k is really *q -- which dovetails with the fact that crosslinguistically /k/ is more common than /q/.

[Travis B.]

And on that note, IIRC I have seen people who have posited that the laryngeals were something like *x *xʷ *χ.

[WeepingElf]

As for PIE lacking post-velar consonants other than the laryngeals, Proto-Semitic is reconstructed in a similar way (though some Semitic languages at least have shifted /k'/ to /q/), and this is indeed the reason why Indo-Europeanists call them "laryngeals".

Thing is, the *k *kʷ *q hypothesis for PIE dorsal plosives seems more solid than the "conventional" *k̟ *kʷ *k hypothesis, because crosslinguistically /k/ is more common than /k̟/ (or /c/ or whatever you want to call it), yet the conventional analysis essentially states that *k̟ is more common than *k -- which can be conveniently resolved by hypothesizing that conventional *k̟ is really *k and conventional *k is really *q -- which dovetails with the fact that crosslinguistically /k/ is more common than /q/.

Two questions:

1. Shouldn't *kʷ then be *qʷ?
2. Apart from a shift from /k q/ to /c k/ or vice versa not really being such a big deal (which is the reason why I prefer speaking agnostically of "front" and "back velars" here), doesn't the fact that no IE language has preserved the system unchanged point at an unstable (and hence, rare) system in PIE (as with the phonation types of the stops)?

As for the notion that the places of articulation of the laryngeals match those of the velars, I don't think so. Why, then, is *h₂ so common when *k is so uncommon? Why is *h₁ not assibilated in satem languages? Why do *h₂ and *h₃ colour adjacent *e, but *k and *kʷ not? (It is, BTW, misleading to consider *h₁ "e-colouring" - rather, it is non-colouring: it doesn't change the colour of any vowel.) Why does *h₁ disappear in all positions in Hittite while the other two are preserved in some positions? My guess is that *h₁ = /h/, *h₂ = /χ ~ ħ/, *h₃ = /χʷ ~ ħʷ/, or maybe their voiced counterparts.

[Travis B.]

As for PIE lacking post-velar consonants other than the laryngeals, Proto-Semitic is reconstructed in a similar way (though some Semitic languages at least have shifted /k'/ to /q/), and this is indeed the reason why Indo-Europeanists call them "laryngeals".

Thing is, the *k *kʷ *q hypothesis for PIE dorsal plosives seems more solid than the "conventional" *k̟ *kʷ *k hypothesis, because crosslinguistically /k/ is more common than /k̟/ (or /c/ or whatever you want to call it), yet the conventional analysis essentially states that *k̟ is more common than *k -- which can be conveniently resolved by hypothesizing that conventional *k̟ is really *k and conventional *k is really *q -- which dovetails with the fact that crosslinguistically /k/ is more common than /q/.

Two questions:

1. Shouldn't *kʷ then be *qʷ?

It could be either -- note how /kʷ/ behaved distinctly from /k/ with regard to palatalization in Romance.

2. Apart from a shift from /k q/ to /c k/ or vice versa not really being such a big deal (which is the reason why I prefer speaking agnostically of "front" and "back velars" here), doesn't the fact that no IE language has preserved the system unchanged point at an unstable (and hence, rare) system in PIE (as with the phonation types of the stops)?

That argument pretty much can be summed up as "the less typologically plausible phonology should be favored specifically because we don't see it any daughters", TBH.

As for the notion that the places of articulation of the laryngeals match those of the velars, I don't think so. Why, then, is *h₂ so common when *k is so uncommon? Why is *h₁ not assibilated in satem languages? Why do *h₂ and *h₃ colour adjacent *e, but *k and *kʷ not? (It is, BTW, misleading to consider *h₁ "e-colouring" - rather, it is non-colouring: it doesn't change the colour of any vowel.) Why does *h₁ disappear in all positions in Hittite while the other two are preserved in some positions?

Tis true. Of course, it is not a given that dorsal fricatives will behave in parallel with dorsal plosives. Take, for instance, North Holland dialects of Dutch -- they innovated the merger of /x/ and /ɣ/ into [χ], without any corresponding backing of /k/. Conversely, StG historical /x/ fronts except when following a back (or in the case of /a/ or /aː/, central) vowel to [ç], without any corresponding fronting of /k/. Likewise, ME /x/ either disappeared or turned into [f], outside Early Modern Scots, without a corresponding change of /k/.

[Glass Half Baked]

"That argument pretty much can be summed up as "the less typologically plausible phonology should be favored specifically because we don't see it any daughters", TBH."

I think a more fair paraphrase is that a stable situation like /k q/ would be more likely to survive, so it strains credulity less to say PIE had the less stable arrangement. It's kind of the same idea, from a different angle; you're both trying to find the most likely phoneme based on the modern reflexes. It's just a question of what's more likely: the arrangement that is more statistically likely in a vacuum, or the arrangement that is more likely to collapse into the forms we see today?

For example, in Hittite we can be quite certain there was no uvular stop (Akkadian would have given the Hittites plenty of opportunity to distinguish the two sounds in writing if they wanted), even though it did have uvular fricatives. So what's more likely, uvular collapsed into velar, or palato-velar collapsed into velar? The former is a more likely starting point, but the latter is a more likely sound change. It's hard to say which of these two is the correct way to apply probability to the problem, since presumably we want both the starting point and the sound change to be as plausible as possible.

[Nortaneous]

Why not prevelar/postvelar/labiovelar like in some Salishan languages?

[Travis B.]

"That argument pretty much can be summed up as "the less typologically plausible phonology should be favored specifically because we don't see it any daughters", TBH."

I think a more fair paraphrase is that a stable situation like /k q/ would be more likely to survive, so it strains credulity less to say PIE had the less stable arrangement. It's kind of the same idea, from a different angle; you're both trying to find the most likely phoneme based on the modern reflexes. It's just a question of what's more likely: the arrangement that is more statistically likely in a vacuum, or the arrangement that is more likely to collapse into the forms we see today?

For example, in Hittite we can be quite certain there was no uvular stop (Akkadian would have given the Hittites plenty of opportunity to distinguish the two sounds in writing if they wanted), even though it did have uvular fricatives. So what's more likely, uvular collapsed into velar, or palato-velar collapsed into velar? The former is a more likely starting point, but the latter is a more likely sound change. It's hard to say which of these two is the correct way to apply probability to the problem, since presumably we want both the starting point and the sound change to be as plausible as possible.

The thing is that raise the question of where the palatovelar versus velar versus labiovelar contrast came from in the first place, since the instability of such an arrangement implies that it should not have existed for long, and hence positing it merely pushes the actual original state back in time, from being what we reconstruct to pure internal reconstruction.

[Glass Half Baked]

Well, surely that's true for any version of any language?

Whether PIE had uvulars or not, something came before it. PIE usually means the PIE that was spoken the moment before Annie Anatolia and Steve Steppe parted company. It could have lasted a day. If you're interested in internal reconstruction, that's cool. But that's not the same mission as trying to figure out how Annie and Steve talked during their last conversation together.

[keenir]

The thing is that raise the question of where the palatovelar versus velar versus labiovelar contrast came from in the first place, since the instability of such an arrangement implies that it should not have existed for long, and hence positing it merely pushes the actual original state back in time, from being what we reconstruct to pure internal reconstruction.

Well, surely that's true for any version of any language?

Whether PIE had uvulars or not, something came before it.

To be fair, thats the sort of thing (in and of itself, that is) that can be argued to negate pretty much anything.

PIE usually means the PIE that was spoken the moment before Annie Anatolia and Steve Steppe parted company. It could have lasted a day. If you're interested in internal reconstruction, that's cool. But that's not the same mission as trying to figure out how Annie and Steve talked during their last conversation together.

Unless they took a century or more to part company after their last conversation (or produced any kids, Anatolian or whomever), surely those would be the same.

And surely that means that the unstable contrast is possible for that moment in time.