The Great Proto-Indo-European Thread

Discussion of natural languages, or language in general.
User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

WeepingElf wrote:English has lost /x/ and preserved /h/
Imo this isn't a good parallel. [x] (not /x/) was a mere allophone of /h/ in coda position (after back vowels), whereas your putative laryngeals would have been contrastive in all positions. That [x] only existed in the coda makes its loss, against the retention of [h], easy to explain without the notion that maybe velar fricatives could disappear without a trace, leaving original glottal fricatives to replace them.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

KathTheDragon wrote:
WeepingElf wrote:English has lost /x/ and preserved /h/
Imo this isn't a good parallel. [x] (not /x/) was a mere allophone of /h/ in coda position (after back vowels), whereas your putative laryngeals would have been contrastive in all positions. That [x] only existed in the coda makes its loss, against the retention of [h], easy to explain without the notion that maybe velar fricatives could disappear without a trace, leaving original glottal fricatives to replace them.
Point taken. I should add that I am myself not really convinced of my theory! It was little else than a brain fart of mine. There perhaps were only the velar fricatives in PIE2, with something to be determined happening such that *h2 became so much more common than *h1 (which one would expect to be the most frequent if the three laryngeals were just the fricative members of the three velar series), and there is no consensus yet under which circumstances a laryngeal is lost in Anatolian and under which it is preserved. Perhaps I should simply drop the idea. Nice try, but with too many problems.

I know you have offered an alternative solution, which can't be easily dismissed, but I am not sold to your alleged *h2w yet, and still haven't got why PIE *o should not have been rounded.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

I'll admit that *h₂ʷ lacks any real corroboration, but it's also not an integral part of my theory. It was more just a "huh, there's a hole here, what if *h₂w were originally *h₂ʷ to fill it" (full disclosure, I first saw this idea elsewhere, though I don't recall off the top of my head who it was - perhaps here?). EDIT: This post, in fact.

On *o, I've conceded that it certainly could have been rounded by late PIE, since not all of the branches I thought supported an unrounded *o are quite as clear as I thought - I really need to get around to studying the evidence again. On the other hand, that doesn't prove roundedness for early PIE, since spontaneous rounding in low back vowels is well-attested. So at this point my argument is largely implicational - there's clear evidence that *h₃ wasn't rounded, so it follows that *o wasn't rounded either, at least at the time of laryngeal colouring (which as we all know was early PIE or earlier). Supporting this line of reasoning is Kümmel's theory that the apophonic vowels *e : *o come from *a : *ā (via *æ : *ɒː), which independently requires a spontaneous rounding in the history of *o.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

Fine. I have no problem with an earlier stage of PIE possessing an unrounded antecedent of PIE3 *o. In fact, the rounding of PIE *o may have been a post-Anatolian matter, perhaps connected with the emergence of vowel colouring by laryngeals. Which brings us back to the qualities of the laryngeals. I still maintain the opinion (an opinion, nevertheless!) that the difference between *h2 and *h3 was that the latter was rounded. This makes it tempting to identify *h3 with the fricative member of the labiovelar series.

But then, *h1 and *h2 can't simply be the front and back velar fricatives, because the frequencies are wrong (one would expect *h1 to be the most common, and *h2 the least common, a bit rarer than *h3, but it is actually *h2 which is the most common), and this difference wouldn't explain why *h1 is lost in Hittite and the others usually not (the instances of lost *h2 and *h3 in Hittite aren't plentiful, and may be due to internal developments of the Anatolian branch). So perhaps the front and back velar fricatives merged in *h2, and *h1 was something unconnected to the three velar series, probably simply *h. This merger may have happened when the velar fricatives became pharyngeal approximants in PIE3 (such a POA shift would probably have clobbered the not-so-great POA difference between them, but left the labialization of the labiovelar fricative intact), and independently in Hittite in the course of its kentum development.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

But isn't the only "evidence" for a rounded *h₃ the roundedness of *o? If you accept that *o need not even have been rounded at the time of laryngeal colouring, continuing to insist on a rounded *h₃ is nonsensical without new evidence, which I note you haven't offered.

I agree about *h₁ (though IMO your argument is slightly flawed - you can simply swap round *h₁ and *h₂, matching the common *h₂ to the common *Ḱ series, and the less common *h₁ to the less common *K series. Why people always insist on aligning the arbitrary numbering of the laryngeals to the front-back dimension of PoA is utterly beyond me. But still we'd expect *h₁ to be a lot less common than it really is, so the argument is flawed anyway), as do many scholars, in assigning it to a glottal fricative (though there are reasons to suppose that *h₁ might have been a glottal stop - or perhaps there were originally two *h₁'s?)

User avatar
Frislander
Avisaru
Avisaru
Posts: 836
Joined: Mon Feb 29, 2016 6:34 am
Location: The North

Re: The Great Proto-Indo-European Thread

Post by Frislander »

Howl wrote:The odds that two languages are genetically related also depend on how much they have in common, and how much the geographical distance is between the places where these languages were spoken. For example, the chances of a relationship between Bantu and Quechua (close to zero) are much lower than between Turkic and Tungusic (>50% in my opinion).
Really? The number of commonalities is quite inconsistent as a measure of potential relatedness, particularly when contact is taken into consideration, because we now know that languages can pretty much borrow anything given sufficient amounts of contact; basic vocabulary, syntactic structure, sound changes, even bits of morphology. In fact there's probably some cases where I'd say two groups that are further apart and more dissimilar have more chance of being related than two very similar groups right next to each other. In the case of Turkic ad Tunguscic I'd say it's not implausible but less likely due to sprachbund effects.

Though if I'm honest I don't entirely see the point in talking about "chances of relationship".
https://frislander.tumblr.com/

First known on here as Karero

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Re: The Great Proto-Indo-European Thread

Post by Salmoneus »

Frislander wrote: Though if I'm honest I don't entirely see the point in talking about "chances of relationship".
The point is that the weight of evidence provided by a parallel is heavily dependent upon prior assumptions about the chances of a relationship.

It's tempting to look at a parallel and think "I know that could be a coincidence, but surely a common cause is more likely than pure luck?", so that even a few parallels make a relationship between two languages look likely.

But this is flawed, because in fact, given the thousands of languages, and that most languages are not closely related to most other languages, the default assumption has to be that in most cases the chances of pure coincidence are much greater than the chances of borrowing, which are much greater than the chances of genuine cognacy. Which means that, without other compelling evidence of relationship, each parallel discovered actually has very little evidentiary weight.


It's like the old false positive paradox in medicine. Let's say a test has a 10% error rate. You test positive - so what are the chances that you actually have the disease? Lots of people think it's 90% - after all, there's only a 10% chance each test of an error. But actually, it depends on the chances of having had the disease in the first place. If only 1 in 100 people have the disease, and there's a 10% error rate for the test, then for ever 1,000 people tested, you'd expect to get 891 true negatives, 9 true positives, 1 false negative, and 99 false positives. So rather than a positive meaning there's a 90% of having a disease, in fact a positive still only means a 7% chance of having the disease.

Likewise, a set of plausible but 'convenient' sound changes (with exceptions that might be explained by more precise rules) may have, for sake of argument, only a 10% false positive rate (and that's being very generous!). But because there are so many unrelated languages, that low false positive rate will still give enough actual false positives that the evidentiary value of a positive is still very small. But by finding a filter to reduce the number of unrelated languages in your sample, the evidentiary value of the "sound change" can rocket up. So questions like "what are the chances of two languages spoken in approximately the same geographic area but two thousand years apart being related?" become very important in judging the significance of any "evidence" of a connexion.

[Or, of course, you can get more and more extensive and detailed sound changes until the chances of a chance resemblence become so incredibly small that the significance of the prior probability of relatedness becomes negligible.]

--------
-------------
------------------------


Regarding PIE: is there actually any reason why people assume that "laryngeals" must be dorsal fricatives of any sort, let alone why they need to "fill the gaps" to match the stop series?

I've just been looking again at a romlang idea, and at actual romance languages. And they have lots of elements that disappear. /n/, for instance, and /l/, and /d/ and /g/; whereas many other languages have the same with /s/.

h1, h2 and h3 could just as easily be, say, *n', *N' and *m', couldn't they? ( where ' here indicates an 'ejective', or whatever glottalised or otherwise emphatic parallel of an ejective series you care to suppose). Sonorants like this can a] act as syllabics (and unlike (probably) fricatives, they actually do act as syllabics in PIE), can b] colour adjacent vowels, and can c] easily vocalise, can d] often occur in coda position, and can e] often cause lenghtening of preceding vowels when dropped. In other words, they seem much more intuitively suited to the "laryngeal" role than a dorsal fricative would.

Or is there some key evidence I'm missing?
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

User avatar
Soap
Smeric
Smeric
Posts: 1228
Joined: Sun Feb 16, 2003 2:57 pm
Location: Scattered disc
Contact:

Re: The Great Proto-Indo-European Thread

Post by Soap »

Some laynrgs are attested as /x/ in Hittite so can't be too far off from that originally. That and that they are in most reconstructions allowed to appear between other consonants.
What did you mean by coincidence >borrowing>luck? I don't understand, was that a mistake? What is luck in this context?
Sunàqʷa the Sea Lamprey says:
Image

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

KathTheDragon wrote:But isn't the only "evidence" for a rounded *h₃ the roundedness of *o? If you accept that *o need not even have been rounded at the time of laryngeal colouring, continuing to insist on a rounded *h₃ is nonsensical without new evidence, which I note you haven't offered.
The vowel-colouring effects of the laryngeals belong AFAIK to post-Anatolian PIE, or at least still were productive then, and even in at least one daughter language, namely Proto-Greek (the famous triple reflex of syllabic as well as initial preconsonantal laryngeals: this is a specifically Greek development, no other IE language has this).
I agree about *h₁ (though IMO your argument is slightly flawed - you can simply swap round *h₁ and *h₂, matching the common *h₂ to the common *Ḱ series, and the less common *h₁ to the less common *K series. Why people always insist on aligning the arbitrary numbering of the laryngeals to the front-back dimension of PoA is utterly beyond me.
The vowel-colouring effects! If *h1 was a back velar and *h2 a front velar (or whatever), how did it come that *h2 backed an *e to *a and *h1 didn't? It seems that the laryngeals weren't numbered that arbitrary, but from front to back according to their vowel-colouring effects.
But still we'd expect *h₁ to be a lot less common than it really is, so the argument is flawed anyway), as do many scholars, in assigning it to a glottal fricative (though there are reasons to suppose that *h₁ might have been a glottal stop - or perhaps there were originally two *h₁'s?)
I don't think *h1 was a glottal stop. It could be syllabic - how do you expect that from a glottal stop? It is pretty obvious that the laryngeals were more sonorous than *s. But as you say, *h1 is too common to be the back velar fricative - that one would be the rarest of all three, rarer still than *h3.

The starting point of my "five-laryngeal" model (in quotes because only two of those five consonants would have been actual laryngeals) was the four-laryngeal theory of Mallory and Adams, who distinguish a *h2 that is preserved in Hittite and a (less common) *h4 that is lost, with the same vowel-colouring effect, to which I added a similar split of *h3, which is also sometimes preserved and sometimes (less often) lost in Hittite; and the system I posited for PIE1 looks quite plausible - one fricative counterpart for each of the five stop series, plus a glottal fricative - but the main sore point is the idea that Hittite lost the velar fricatives while keeping (and fronting) the glottal ones. I admit that this development seems unnatural, and that the English example does not really cut it. It is simply that the farther back one goes in the vocal tract (from the alveolar POA where we have the rather stable /s/), the more likely do fricatives "go down the throat" and disappear. The velar and glottal fricatives would have to switch positions somehow, but how?

Hence, I proposed an alternative model yesterday, in which the front and back velar fricatives merged in *h2, while the labiovelar fricative became *h3, and *h1 is an old *h that existed independently from the three velar series. But I am not really content with that yet, because it assumes a merger of fricatives where the stops did not merge. Sure, the assumption is that the velar fricatives became pharyngeal approximants in PIE3, but it still feels rather awkward to me.

But perhaps the three (or whatever their number) laryngeals never had anything to do with the three velar series, and we are barking up the wrong tree all the time.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Re: The Great Proto-Indo-European Thread

Post by Salmoneus »

Soap wrote:Some laynrgs are attested as /x/ in Hittite so can't be too far off from that originally. That and that they are in most reconstructions allowed to appear between other consonants.
What did you mean by coincidence >borrowing>luck? I don't understand, was that a mistake? What is luck in this context?
I did indeed get mixed up halfway through that sentence...


And yes, I'm an idiot for forgetting the Hittite reflexes. And indeed borrowings into Uralic.

Then again, the Hittite evidence only applies to 2, doesn't it? [the laryngeals don't have to form a natural set!].

And if we imagined some sort of glottalised resonant (helping to explain their phonotactic distribution), it's not implausible to imagine them losing their resonant element and remaining just as something glottal or pharyngeal.

This would also allow the 'laryngeals' and the voiced (i.e. ejective) stops to share a quality, explaining lengthening in Latin and Balto-Slavic, and Balto-Slavic accent.


Admittedly, though, I don't think this explains the Uralic borrowings very well.
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

Salmoneus wrote:Regarding PIE: is there actually any reason why people assume that "laryngeals" must be dorsal fricatives of any sort, let alone why they need to "fill the gaps" to match the stop series?

I've just been looking again at a romlang idea, and at actual romance languages. And they have lots of elements that disappear. /n/, for instance, and /l/, and /d/ and /g/; whereas many other languages have the same with /s/.

h1, h2 and h3 could just as easily be, say, *n', *N' and *m', couldn't they? ( where ' here indicates an 'ejective', or whatever glottalised or otherwise emphatic parallel of an ejective series you care to suppose). Sonorants like this can a] act as syllabics (and unlike (probably) fricatives, they actually do act as syllabics in PIE), can b] colour adjacent vowels, and can c] easily vocalise, can d] often occur in coda position, and can e] often cause lenghtening of preceding vowels when dropped. In other words, they seem much more intuitively suited to the "laryngeal" role than a dorsal fricative would.

Or is there some key evidence I'm missing?
As Soap said, *h₂ and (in certain environments) *h₃ are written with in Hittite and Cuneiform Luwian, which is independently known to write dorsal fricatives. Unless you have some example of ejective nasals becoming dorsal fricatives...

Furthermore, the phonotactics of the laryngeals suggest fricatives. Fricatives could occur root-initially before stops (*h₁ger-) for instance, like *s, but unlike nasals. Similarly, they can occur root-finally after stops (*h₂eḱh₃-), again like *s, but unlike nasals.

One more point - which should be directed at WE as well - is that syllabic laryngeals are an illusion. What actually happened in "laryngeal vocalisation" was the epenthesis of a vowel (probably [ə]) adjacent to laryngeals in illicit clusters prior to the general loss of laryngeals. This can be seen quite plainly from all the cases where laryngeals didn't "vocalise" despite not being adjacent to any other vocalic. I should, of course, point out that English attests many things that laryngeals did with its reflexes of OE h when they aren't retained as NE h: "vocalising" in e.g. borough < OE burh, lengthening in night < OE niht, frequently occurs in coda position (in fact, OE h is always lost one way or another here), and if you want to count vowel breaking as colouring, you can include that too, though Semitic attests better parallels for colouring from dorsal fricatives. So, I fail to see what's unintuitive here.
WeepingElf wrote:The vowel-colouring effects of the laryngeals belong AFAIK to post-Anatolian PIE, or at least still were productive then, and even in at least one daughter language, namely Proto-Greek (the famous triple reflex of syllabic as well as initial preconsonantal laryngeals: this is a specifically Greek development, no other IE language has this).
Anatolian clearly shows exactly the same vowel colourings as the rest of IE, so colouring must have preceeded the departure of Anatolian from the family. As for Greek, I don't have a good refutal here - but I still consider this very shaky evidence to base an otherwise utterly unsupported - and plainly contradicted - theory on. It's also possible that the rounding of *o was very late in the timeline of PIE, spreading over a differentiated dialect continuum, as satemisation probably did. Then the original third vowel in Greek's triple reflex need not have been rounded originally.
The vowel-colouring effects! If *h1 was a back velar and *h2 a front velar (or whatever), how did it come that *h2 backed an *e to *a and *h1 didn't? It seems that the laryngeals weren't numbered that arbitrary, but from front to back according to their vowel-colouring effects.
... Touché.
I don't think *h1 was a glottal stop. It could be syllabic - how do you expect that from a glottal stop? It is pretty obvious that the laryngeals were more sonorous than *s. But as you say, *h1 is too common to be the back velar fricative - that one would be the rarest of all three, rarer still than *h3.
See above on syllabicity, but I agree that [h] is more likely, though more because it would be easier for the laryngeals to merge if they were all fricatives.
The starting point of my "five-laryngeal" model (in quotes because only two of those five consonants would have been actual laryngeals) was the four-laryngeal theory of Mallory and Adams, who distinguish a *h2 that is preserved in Hittite and a (less common) *h4 that is lost, with the same vowel-colouring effect, to which I added a similar split of *h3, which is also sometimes preserved and sometimes (less often) lost in Hittite;
I've never been convinced by their *h₄ - the evidence in favour is too sketchy for me. How can we be sure it's not simply an inner-Anatolian phenomenon, similar to the loss of *h₂ in the suffix complex *-eh₂ye/o- > -ā(e)-?

The velar and glottal fricatives would have to switch positions somehow, but how?
I'll admit I was amused to see you asking this exact question over in the sound change quickies thread - I'll be watching eagerly to see if anyone can come up with something that works.
But perhaps the three (or whatever their number) laryngeals never had anything to do with the three velar series, and we are barking up the wrong tree all the time.
It's always an option, certainly.

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

Salmoneus wrote:Then again, the Hittite evidence only applies to 2, doesn't it? [the laryngeals don't have to form a natural set!].
Nope! ḫāraš, -an- "eagle" < *h₃éron-
This would also allow the 'laryngeals' and the voiced (i.e. ejective) stops to share a quality, explaining lengthening in Latin and Balto-Slavic, and Balto-Slavic accent.
The Lachmann's law lengthening in Latin isn't due to some "laryngeal-like" "glottalic" quality to the voiced stops, it's simply lengthening in front of secondary clusters of voiced stop and voiceless obstruent, cf. the slight phonetic lengthening in English before voiced stops. Winter's law is explicable similarly.

User avatar
Nortaneous
Sumerul
Sumerul
Posts: 4544
Joined: Mon Apr 13, 2009 1:52 am
Location: the Imperial Corridor

Re: The Great Proto-Indo-European Thread

Post by Nortaneous »

English 'laryngeal vocalization' can be explained as vowel epenthesis, so there's that. Are there any schwas at any point in PIE that could've merged with *e early on in pre-Greek?
Siöö jandeng raiglin zåbei tandiüłåd;
nää džunnfin kukuch vklaivei sivei tåd.
Chei. Chei. Chei. Chei. Chei. Chei. Chei.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

KathTheDragon wrote:Furthermore, the phonotactics of the laryngeals suggest fricatives. Fricatives could occur root-initially before stops (*h₁ger-) for instance, like *s, but unlike nasals. Similarly, they can occur root-finally after stops (*h₂eḱh₃-), again like *s, but unlike nasals.

One more point - which should be directed at WE as well - is that syllabic laryngeals are an illusion. What actually happened in "laryngeal vocalisation" was the epenthesis of a vowel (probably [ə]) adjacent to laryngeals in illicit clusters prior to the general loss of laryngeals. This can be seen quite plainly from all the cases where laryngeals didn't "vocalise" despite not being adjacent to any other vocalic. I should, of course, point out that English attests many things that laryngeals did with its reflexes of OE h when they aren't retained as NE h: "vocalising" in e.g. borough < OE burh, lengthening in night < OE niht, frequently occurs in coda position (in fact, OE h is always lost one way or another here), and if you want to count vowel breaking as colouring, you can include that too, though Semitic attests better parallels for colouring from dorsal fricatives. So, I fail to see what's unintuitive here.
Fine. So we need no approximant stage of the laryngeals for the purpose of explaining "syllabic laryngeals" if there never weren't any. Still, a POA shift from velar to somewhere further back is possible on the way from PIE2 to PIE3.
WeepingElf wrote:The vowel-colouring effects of the laryngeals belong AFAIK to post-Anatolian PIE, or at least still were productive then, and even in at least one daughter language, namely Proto-Greek (the famous triple reflex of syllabic as well as initial preconsonantal laryngeals: this is a specifically Greek development, no other IE language has this).
Anatolian clearly shows exactly the same vowel colourings as the rest of IE, so colouring must have preceeded the departure of Anatolian from the family.
I see. So the vowel colouring effects of the laryngeals are pre-Anatolian. Well, the vowels don't behave particularly "weird" in Anatolian, so I see no reason why *o can't have been rounded in PIE2 already. It of course depends on the nature of the difference between *a and *o in PIE2 what is the difference between *h2 and *h3.
As for Greek, I don't have a good refutal here - but I still consider this very shaky evidence to base an otherwise utterly unsupported - and plainly contradicted - theory on. It's also possible that the rounding of *o was very late in the timeline of PIE, spreading over a differentiated dialect continuum, as satemisation probably did. Then the original third vowel in Greek's triple reflex need not have been rounded originally.
While I can't say that you are wrong, this sounds like an attempt to rescue a hypothesis on the verge of collapse to me. With the exception of Tocharian, which does weird things with the vowels anyway, all branches of IE that don't merge *a and *o reflect *o as a rounded vowel, which makes any denial of *o-roundedness for Late PIE sound ridiculous. I feel that the unrounded PIE *o is an idée fixe of yours that you try to uphold against the evidence.
The vowel-colouring effects! If *h1 was a back velar and *h2 a front velar (or whatever), how did it come that *h2 backed an *e to *a and *h1 didn't? It seems that the laryngeals weren't numbered that arbitrary, but from front to back according to their vowel-colouring effects.
... Touché.
I don't think *h1 was a glottal stop. It could be syllabic - how do you expect that from a glottal stop? It is pretty obvious that the laryngeals were more sonorous than *s. But as you say, *h1 is too common to be the back velar fricative - that one would be the rarest of all three, rarer still than *h3.
See above on syllabicity, but I agree that [h] is more likely, though more because it would be easier for the laryngeals to merge if they were all fricatives.
Yep.
The starting point of my "five-laryngeal" model (in quotes because only two of those five consonants would have been actual laryngeals) was the four-laryngeal theory of Mallory and Adams, who distinguish a *h2 that is preserved in Hittite and a (less common) *h4 that is lost, with the same vowel-colouring effect, to which I added a similar split of *h3, which is also sometimes preserved and sometimes (less often) lost in Hittite;
I've never been convinced by their *h₄ - the evidence in favour is too sketchy for me. How can we be sure it's not simply an inner-Anatolian phenomenon, similar to the loss of *h₂ in the suffix complex *-eh₂ye/o- > -ā(e)-?
This can't be excluded. I am not very convinced by M&A's *h4, either, and my five-laryngeal model is nothing I am sure of to any degree - just a suggestion. If instances of lost *h2 and *h3 in Hittite have an inner-Anatolian explanation, the whole thing becomes bogus, and perhaps my second model - merger of front and back velar fricatives in *h2, while the labiovelar stays distinct as *h3, and *h1 is just plain old *h and has nothing to do with the three velar series - makes more sense. But the question remains, why do the fricatives merge all over the board, and the homorganic stops only in the centum languages? Was there a POA shift from velar to pharyngeal (or whatever) that clobbered the front/back distinction?
The velar and glottal fricatives would have to switch positions somehow, but how?
I'll admit I was amused to see you asking this exact question over in the sound change quickies thread - I'll be watching eagerly to see if anyone can come up with something that works.
Well, perhaps they see it from a different angle over there, and come up with a workable idea.
But perhaps the three (or whatever their number) laryngeals never had anything to do with the three velar series, and we are barking up the wrong tree all the time.
It's always an option, certainly.
Fair. It would make all the questions we are nagging on here moot. Perhaps there weren't any velar fricatives matching the POAs of the three velar stop series at all. Then *h2 perhaps was *[ħ], and *h3 may even have been *[f]. And *h1, well, *[h].

However, there are apparently Indo-Uralic correspondences (such as the dual marker) that connect PIE laryngeals (all three of them) to Uralic *k, the same way PIE *s seems sometimes to correspond to Uralic *t (as in the plural marker). Hyllested proposed some correspondences where Uralic vowel qualities are reflected in the PIE laryngeal qualities, as in my GVC model wherein the three velar stop series preserve old vowel features - front vowels next to front velars, back unrounded vowels next to back velars, and back rounded vowels next to labiovelars.

So the main problems with the simple version of the velar fricative model (*h1=front velar, *h2=back velar, *h3=labiovelar) are, to summarize:

1. *h2 is much more common than an unlabialized back velar fricative would be expected to be.
2. Why do velar fricatives colour *e, but not velar stops?
3. Why is *h1 lost in Hittite?

How do we best get rid of these?

EDIT: Here's an idea which may perhaps make the 5-laryngeal theory work; I just posted it in the Sound Change Quickie Thread. It goes like this:

1. The glottal fricatives harden to glottal stops.
2. The velar fricatives are lost.
3. The glottal stops front to uvular stops.
4. The uvular stops fricativize.

Should work, but feels somewhat contrived and ad hoc. What do you think?
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

WeepingElf wrote:Still, a POA shift from velar to somewhere further back is possible on the way from PIE2 to PIE3.
Certainly - imo merger into [h] is the final step, but it could take almost any route on the way.
I see. So the vowel colouring effects of the laryngeals are pre-Anatolian. Well, the vowels don't behave particularly "weird" in Anatolian, so I see no reason why *o can't have been rounded in PIE2 already. It of course depends on the nature of the difference between *a and *o in PIE2 what is the difference between *h2 and *h3.
AWhile I can't say that you are wrong, this sounds like an attempt to rescue a hypothesis on the verge of collapse to me. With the exception of Tocharian, which does weird things with the vowels anyway, all branches of IE that don't merge *a and *o reflect *o as a rounded vowel, which makes any denial of *o-roundedness for Late PIE sound ridiculous. I feel that the unrounded PIE *o is an idée fixe of yours that you try to uphold against the evidence.
Two things - Anatolian does in fact distinguish *a and *o, with unrounded reflexes of *o, in e.g. Lycian, where *a > a, but *o > e, probably /æ/. I still need to double-check that it and Tocharian don't require roundedness for *o, though I'm fairly sure that's the case, and if so, these two branches would be my main positive evidence. Secondly, I'm curious where you got the notion that my theory is on the verge of collapse? So far, the only issue we've found is... Greek's triple reflex. I'll concede it's a problem, but my familiarity with Greek is rather less than the other branches, due mainly to having only recently got my hands on Beekes' dictionary, and not having had the time to read it, so I don't have the expertise to solve it like I can other problems from other branches.
Now, to clarify - I'm definitely not outright denying roundedness for *o in late PIE. Even I know how clearly ad-hoc that suggestion was, and I'm not seriously entertaining it. Further, to call this an idée fixe is going much too far. So far I haven't done anything more egregious than you with the notion of *h₃ being labialised.
But the question remains, why do the fricatives merge all over the board, and the homorganic stops only in the centum languages? Was there a POA shift from velar to pharyngeal (or whatever) that clobbered the front/back distinction?
It's an excellent question, to be sure, and one that anyone who proposes velar fricatives as laryngeals must overcome (though I think I have a slightly easier time of it, since I propose that the front-back dimension lines up with *h₂ and *h₃). Are there parallels for maintaining a distinction between /k/ and /q/, but not /x/ and /χ/?
So the main problems with the simple version of the velar fricative model (*h1=front velar, *h2=back velar, *h3=labiovelar) are, to summarize:

1. *h2 is much more common than an unlabialized back velar fricative would be expected to be.
2. Why do velar fricatives colour *e, but not velar stops?
3. Why is *h1 lost in Hittite?

How do we best get rid of these?
For #3, we could imagine *xʲ > *ç > *h and then straightforward loss. I don't have any ideas for the others.
EDIT: Here's an idea which may perhaps make the 5-laryngeal theory work; I just posted it in the Sound Change Quickie Thread. It goes like this:

1. The glottal fricatives harden to glottal stops.
2. The velar fricatives are lost.
3. The glottal stops front to uvular stops.
4. The uvular stops fricativize.

Should work, but feels somewhat contrived and ad hoc. What do you think?
Those are two excellent adjectives.

User avatar
Nortaneous
Sumerul
Sumerul
Posts: 4544
Joined: Mon Apr 13, 2009 1:52 am
Location: the Imperial Corridor

Re: The Great Proto-Indo-European Thread

Post by Nortaneous »

Regular correspondences are PIE *o *ō > PToch *æ *ā. Might be an umlaut process in pre-PToch that demonstrates an earlier rounded value for *æ, dunno. Wikipedia lists ā$ǽ > å$å. Kuemmel says:
Some nouns (old neuter u-stems) show a superficially similar vowel alternation: cf. TB singular or ‘wood’, yok ‘body hair’ vs. plural ārwa, yākwa. But as the corresponding TA forms or, yok show, TB o in the singular is not Proto-Tocharian *å but *o that arose through u-umlaut from earlier *æ, and the plural shows regular a-umlaut of that *æ.
But I don't really know what's going on here, and this has the problem that the etymology of TB yok is unknown.

Does Albanian need rounding in *o? Did *o *a merge before or after *a-raising? (Albanian merges *o *a, but not *ō *ā -- *ā is instead merged with *ē.) PIE *ō > PAlb *ø > Alb e, #ve-, but I don't know if something like *ə > *a, *ō > *ə > *ø (as in French) can be ruled out. Orel says *ō > we, then (later) *ū > wi (which is attested as such), then *we wi > *ø y, *ø > e, #ve-. wi > y can be demonstrated by inherited *wi, but *we is subject to the usual bizarre proliferation of reflexes for *e (the closest thing that can be given to a regular outcome is "either vje or vie with no known conditioning factor"). Would probably have to look at all the possible sequences of *w plus anything that might have been e at any time in order to tell if anything got swept along. I think it's somewhat unlikely for the only /e/ in existence at the relevant stage of PAlb to have been in *we < *ō... but English has a few phonetic vowels that only exist as components of diphthongs (the onsets of /ai/ and /or oi/ in AmEng) so anything is possible.

Probably best to assume Albanian had a rounded value for *ō.
Siöö jandeng raiglin zåbei tandiüłåd;
nää džunnfin kukuch vklaivei sivei tåd.
Chei. Chei. Chei. Chei. Chei. Chei. Chei.

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

Nortaneous wrote:an earlier rounded value for *æ
Improbable, as *æ also comes from post-PIE *ē of any origin. u-umlaut in Tocharian proves nothing, since u-umlaut can easily induce labialisation. The mutual rounding in *ā...æ is interesting, but probably not significant.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

Re the laryngeals once being stops in Anatolian - I knew I had read of an Anatolian languages where they were stops, so I looked it up in Fortson, and voilà (p. 194): in Lycian, *h2 is reflected by a stop, "variously spelled k, q, and χ" (these are the conventional transcription symbols for the Lycian letters). This is not proof of the developments I conjectured yesterday, though, but perhaps we have a snapshot of the "stop stage" here.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

KathTheDragon wrote:
WeepingElf wrote:Still, a POA shift from velar to somewhere further back is possible on the way from PIE2 to PIE3.
Certainly - imo merger into [h] is the final step, but it could take almost any route on the way.
Fine. And perhaps the front velar was debuccalized earlier than the other two, as you suggested below.
I see. So the vowel colouring effects of the laryngeals are pre-Anatolian. Well, the vowels don't
behave particularly "weird" in Anatolian, so I see no reason why *o can't have been rounded in PIE2 already. It of course depends on the nature of the difference between *a and *o in PIE2 what is the difference between *h2 and *h3.
AWhile I can't say that you are wrong, this sounds like an attempt to rescue a hypothesis on the verge of collapse to me. With the exception of Tocharian, which does weird things with the vowels anyway, all branches of IE that don't merge *a and *o reflect *o as a rounded vowel, which makes any denial of *o-roundedness for Late PIE sound ridiculous. I feel that the unrounded PIE *o is an idée fixe of yours that you try to uphold against the evidence.
Two things - Anatolian does in fact distinguish *a and *o, with unrounded reflexes of *o, in e.g. Lycian, where *a > a, but *o > e, probably /æ/. I still need to double-check that it and Tocharian don't require roundedness for *o, though I'm fairly sure that's the case, and if so, these two branches would be my main positive evidence.
Point taken; I wasn't aware of the developments in Lycian.
Secondly, I'm curious where you got the notion that my theory is on the verge of collapse? So far, the only issue we've found is... Greek's triple reflex. I'll concede it's a problem, but my familiarity with Greek is rather less than the other branches, due mainly to having only recently got my hands on Beekes' dictionary, and not having had the time to read it, so I don't have the expertise to solve it like I can other problems from other branches.
Now, to clarify - I'm definitely not outright denying roundedness for *o in late PIE. Even I know how clearly ad-hoc that suggestion was, and I'm not seriously entertaining it. Further, to call this an idée fixe is going much too far. So far I haven't done anything more egregious than you with the notion of *h₃ being labialised.
OK. To me, it looks something like an idée fixe of yours which you uphold with more and more intellectual contortions against the evidence for PIE *o being rounded. I can't say that you are wrong, though, but I remain sceptical. We should leave it at that and agree to disagree. At least, you are clearly not an Octaviano-like nut case!
But the question remains, why do the fricatives merge all over the board, and the homorganic stops only in the centum languages? Was there a POA shift from velar to pharyngeal (or whatever) that clobbered the front/back distinction?
It's an excellent question, to be sure, and one that anyone who proposes velar fricatives as laryngeals must overcome (though I think I have a slightly easier time of it, since I propose that the front-back dimension lines up with *h₂ and *h₃). Are there parallels for maintaining a distinction between /k/ and /q/, but not /x/ and /χ/?
Arabic has /k/ and /q/, but only one voiceless fricative in that range, variously described as /x/ or /χ/ (and its voiced counterpart). However, there never was, at least after Proto-Semitic, a merger of */x/ and */χ/, and Arabic /q/ is from Proto-Semitic */k'/, so the case is different.
So the main problems with the simple version of the velar fricative model (*h1=front velar, *h2=back velar, *h3=labiovelar) are, to summarize:

1. *h2 is much more common than an unlabialized back velar fricative would be expected to be.
2. Why do velar fricatives colour *e, but not velar stops?
3. Why is *h1 lost in Hittite?

How do we best get rid of these?
For #3, we could imagine *xʲ > *ç > *h and then straightforward loss. I don't have any ideas for the others.
Fair. /ç/ is a rare sound in the world's languages, and apparently less stable than /x/ or /χ/. Perhaps there was a split, conditioned by who knows what: in some positions, the front velar fricative was debuccalized along the pathway you suggested, in others it merged with the back velar, thus increasing the frequency of *h2. This would answer both #1 and #3 in one shot. #2 is perhaps not as great a problem, as fricatives are continuants and therefore more likely to affect vowels than stops do. (And it is perhaps not 100% true that velar stops did not affect vowels! There is no shortage of PIE words with an *a next to a back velar stop, with no *h2 in sight to do the colouring. These may be late loanwords, though.)
EDIT: Here's an idea which may perhaps make the 5-laryngeal theory work; I just posted it in the Sound Change Quickie Thread. It goes like this:

1. The glottal fricatives harden to glottal stops.
2. The velar fricatives are lost.
3. The glottal stops front to uvular stops.
4. The uvular stops fricativize.

Should work, but feels somewhat contrived and ad hoc. What do you think?
Those are two excellent adjectives.
Indeed. As I posted a few minutes ago, Lycian seems to show stops, which may be a snapshot of stage 3, but then, Lycian phonology is difficult, we are dealing with conjectured values of letters in an extinct alphabet, and such evidence is best handled with great care, and we are more likely dealing with a late development here; after all, in Luwian, the ancestor of Lycian, the laryngeals seem not to have been stops, or at least, I found no mention of that anywhere!

By now, I consider the 5-laryngeal model a flight of fancy that was worth considering, but is so poorly founded that it better ought to be given up. There are more parsimonious alternatives.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

WeepingElf wrote:in Lycian, *h2 is reflected by a stop, "variously spelled k, q, and χ"
Kloekhorst shows that Lycian q < *h₂w, not *h₂. It's instructive that *h₂ when lenited gives g, which is a fricative (according to Kloekhorst, underlyingly voiceless but voiced intervocalically), so it seems likely that the stopping of Lycian's laryngeals was probably internal to that language, which makes it largely irrelevant for your argument. I couldn't find any etymologies for k < *h₂, but if it's correct that k χ are /c k/, then it's pretty reasonable.
WeepingElf wrote:OK. To me, it looks something like an idée fixe of yours which you uphold with more and more intellectual contortions against the evidence for PIE *o being rounded. I can't say that you are wrong, though, but I remain sceptical.
Hey, at least I have real tangible evidence to back up my arguments, just as you have evidence for yours. I'll just have to continue gathering evidence for my position until either I discover a truly fatal flaw, or I convince you as well.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

As you certainly know, I consider the three non-high PIE3 vowels to have emerged from a single PIE1 vowel, *a (which is of course unrounded), and it may well be that the antecedent of PIE3 *o was not yet rounded in PIE2, though I expect rounding to have happened already by that time. IMHO, in PIE2 *o probably was something like [ɒ] (and *e was [æ]). If PIE2 *o was unrounded, then the difference between *h2 and *h3 at that time would have been something else than labialization, but that begs the question why *h3 labiovelarizes an adjacent [ə] to /o/ in Proto-Greek if it doesn't have the feature [+labial]. One would have to suppose parallel rounding changes in both *o and *h3, which is not very parsimonious. The most parsimonious solution would IMHO be that *o was already rounded (and be it as [ɒ]) in PIE2.

But alas, we are probably not knowledgeable to find out when PIE *o acquired its rounding. Each of us has his/her own opinion about the time of *o-rounding, and we can leave it at that.

The Lycian fortition of laryngeals, I agree with you, is a development internal to Lycian. As I said earlier, there is no evidence for hardened laryngeals in Luwian, which was to be expected if the Lycian stop reflexes of the laryngeals were old.

Also, regarding the frequencies of the laryngeals in PIE, it may be that *h1 is heavily under-reconstructed (and the other two perhaps over-reconstructed). There are AFAIK quite a few words where a *h2 or a *h3 is reconstructed based on vowel colours alone without Hittite cognates actually showing the laryngeal in question, and in such situations, *h1, which doesn't colour vowels, would be irreconstructible. Also, there are the many laryngeals of unknown quality. How many of them were which one?

I do not seriously expect to manage to crack this riddle!
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

It may well be parsimonious, but imo parsimony isn't a good enough reason to reject other scenarios, especially when there's other evidence to consider, such as the (imo conclusive) argument that *h₃ cannot have been rounded.

Yes, the laryngeals are absolutely misreconstructed in many cases. For example, Kloekhorst insists on *h₃ in the "sheep" word because of frequent o-vowels, despite the clear a-vowel in Tocharian which points to *h₂. Similarly, the LIV insists on *h₂ when Indo-Iranian shows an aspirate but the others a plain stop (voiceless or voiced), even though *h₁ seems to also have aspirated (and logically, *h₃ might have as well).

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: The Great Proto-Indo-European Thread

Post by KathTheDragon »

It just occurred to me that it's relevant that there is another language with a "triple-reflex"... Latin. Specifically, in the environment *HN̥C- > *e/a/oNC- (cf. endo, indo "in, on, to" < *h₁n̥do, ambi "around" < *h₂n̥tbʰi, unguis "finger-nail, claw" < *h₃n̥gʰu-). Except this sound change happened within Italic itself, and all other vocalisations involving laryngeals yield only *a, like other non-laryngeal epenthetic vowels (cf. in particular arbor "tree" < *h₃r̥dʰōs, arduus < *h₃r̥dʰwos). So either we insist that all three laryngeals were still distinguished down into the pre-history of Italic, as well as Greek, or there's some other explanation for both of these.

Doing a little more digging, de Vaan derives amō from *h₃m̥h₃- (with the laryngeals based on Greek), which leaves me wondering why the first vowel is a. Surely it'd be o, or even e, depending on the timing of certain changes?

User avatar
Nortaneous
Sumerul
Sumerul
Posts: 4544
Joined: Mon Apr 13, 2009 1:52 am
Location: the Imperial Corridor

Re: The Great Proto-Indo-European Thread

Post by Nortaneous »

WeepingElf wrote:So the main problems with the simple version of the velar fricative model (*h1=front velar, *h2=back velar, *h3=labiovelar) are, to summarize:

1. *h2 is much more common than an unlabialized back velar fricative would be expected to be.
2. Why do velar fricatives colour *e, but not velar stops?
3. Why is *h1 lost in Hittite?

How do we best get rid of these?
Reconstructing uvular stops for *h2 *h3 would answer this, although obviously it isn't compatible with *h3 not being rounded, since then there'd be nothing to distinguish them. Unless *h3 was /R/ (voiced uvular stops are incredibly rare and tend to fricate essentially immediately, so I don't think reconstructing G\ makes sense) or something, but then you have to explain why they had different coloring effects.

1. According to Wikipedia's chart of consonant frequencies in Arabic roots, /q/ is the ninth most common consonant, after /r w l m n b f ʕ/. It's more common than /k/ or /dʒ/ < /g/. *h2 seems pretty common in PIE.

2. Velar fricatives color *e because they were in fact uvular plosives, which are known to color vowels.

3. *h1 is lost in Hittite because it was some sort of glottal, and glottals are particularly easily lost.

Later on, uvular plosives fricate (as in Palauan, which later had X > ?, unless the *q > X sound change didn't reach fixation and died out or there were dialectal differences), or become pharyngeals or whatever. This sound change diffuses through Indo-Hittite, unless it doesn't and different things happen instead.
Siöö jandeng raiglin zåbei tandiüłåd;
nää džunnfin kukuch vklaivei sivei tåd.
Chei. Chei. Chei. Chei. Chei. Chei. Chei.

User avatar
WeepingElf
Smeric
Smeric
Posts: 1630
Joined: Wed Mar 08, 2006 5:00 pm
Location: Braunschweig, Germany
Contact:

Re: The Great Proto-Indo-European Thread

Post by WeepingElf »

A interesting idea, Tropylium, and IIRC also proposed by Kortlandt.
...brought to you by the Weeping Elf
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A

Post Reply