The Great Proto-Indo-European Thread

[Howl]

I can't give a full account of MCV's theory,

I have googled Vidal's theory. It's at: http://www.academia.edu/4169949/Centum_and_satem

Basically he suggests that this 'qa' -> 'ka' (and thus the Centum/Satem split) happened in pre-PIE at the time of the 'Great Vowel Collapse'. And he also suggests a 'qi' -> 'ke' to explain the occurrences of *k adjacent to *e.

[Zju]

Well, all of that makes a lot of sense to me, but is there any precedent for a voiced fricative producing distinctive vowel coloring? For that matter, is there precedent for rounding blocking a vowel-coloring effect?

Does anyone know how to search wiktionary for PIE words? It has a fair number of reconstructions, but they don't show up in regular searches or in translation sections, only in etymology ones, making searching for a PIE word with particular meaning quite a PITA.

I'm not aware of any way to search roots, but note that they still have nouns all on one big page: https://en.wiktionary.org/wiki/Appendix ... pean_nouns

Yes, in Old English. IIRC an example is the development of ēaġe > eye vs. eahta > eight - different fricative voicing leading to different vocalism, but don't quite remember the specifics.

I had forgotten about that list. Alas, it's only nouns and this one is seperated in pages by initial letter.

[Šọ̈́gala]

Thanks, I was actually thinking of https://www.academia.edu/4199002/PIE_a, but the centum-satem paper gives more detail.

Just perusing the list of *k- roots at Wiktionary, the idea of a-coloring does appear sporadic at best. Actually *g- has a higher proportion of *a vocalism. Carrasquer Vidal argues for a conditioned explanation which relies on elaborating the Pre-Proto stage. Another approach that could explain some apparent *ke- sequences would be that they are really *ḱe- that failed to satemise, either sporadically/analogically or because Conditions.

[Tropylium]

I still think an intervening vowel shift is maybe the most likely explanation.

Both e > a / _h₂_ and e > o₂ / _h₃_ can be considered pre-PIE changes, since they can be only recovered by internal reconstruction. (There is some evidence that *o₂ was distinct from basic *o, but it's still reflected as a back vowel everywhere. We could reconstruct e.g. *o = [ɔ] versus *o₂ = [o], or *o = [o] versus *o₂ = [ɵ].) So if there were also some pre-PIE vowel shifts, there is no reason to expect that *h₃ colored *e right to *o. The coloring rule could well have been telescoped from some simpler original coloring sound change.

So something like the following is conceivable:
– pre-coloring vowel system *e *a (and *i *u)
– *h₂ *h₃ = ħ ʕ
– *e > *ə / _ʕ_
– *ə > *o
– *e > *ə / _ħ_
– *ə > *a

---

There's some complication to this from the interaction of laryngeals with consonants though, such as the Greek development from *R̥h₂ *R̥h₃ to *Rā *Rō. I would suggest the following as a remedy for this:
1) vowel coloring (pre-PIE);
2) reduction of unstressed vowels to *ə (post-PIE; not general in Greek)
3) loss of *ə (creation of zero grades; post-PIE).

[Howl]

Thanks, I was actually thinking of https://www.academia.edu/4199002/PIE_a, but the centum-satem paper gives more detail.

Just perusing the list of *k- roots at Wiktionary, the idea of a-coloring does appear sporadic at best. Actually *g- has a higher proportion of *a vocalism. Carrasquer Vidal argues for a conditioned explanation which relies on elaborating the Pre-Proto stage. Another approach that could explain some apparent *ke- sequences would be that they are really *ḱe- that failed to satemise, either sporadically/analogically or because Conditions.

Working from existing PIE reconstructions gets you nowhere in this case.

This is because any *a/*h₂ in a reconstruction can be one of three cases:

1. There is evidence that the a-vocalism is caused by a laryngeal. Those are reconstructed with *h₂
2. There are reasons that a laryngeal reconstruction is difficult. Those are reconstructed with *a.
   See also the paper where Lubotsky tries to eradicate these reconstructions (google for it).
3. Both *a and *h₂ are possible.

Guess what reconstruction is commonly used for group 3?

[Howl]

Both e > a / _h₂_ and e > o₂ / _h₃_ can be considered pre-PIE changes, since they can be only recovered by internal reconstruction. (There is some evidence that *o₂ was distinct from basic *o, but it's still reflected as a back vowel everywhere. We could reconstruct e.g. *o = [ɔ] versus *o₂ = [o], or *o = [o] versus *o₂ = [ɵ].) So if there were also some pre-PIE vowel shifts, there is no reason to expect that *h₃ colored *e right to *o. The coloring rule could well have been telescoped from some simpler original coloring sound change.
.

I have an even more radical hypothesis. The vowels had always been colored. The basic idea is that the different vowel qualities attributed to the laryngeal coloring actually reflect different vowel qualities in pre-PIE. How is that possible? Simple! We start off with a single laryngeal /h/. In pre-PIE this laryngeal splits based on the adjacent vowel, like this:

/he/ -> /h₁e/ and /eh/ -> /eh₁/ (probably also for /i/ instead of /e/)
/ha/ -> /h₂a/ and /ah/ -> /ah₂/
/ho/ -> /h₃o/ and /oh/ -> /oh₃/ (probably also for /u/ instead of /o/)

After that, these laryngeal phonemes could be used as consonant representations of these vowels, much in the same way /y/ and /w/ can be used as consonant representations of /i/ and /u/. And that allowed these vowels to be preserved under ablaut like this:

Zero Grade: /e₂/, /o₂/ -> /ə/, /e₁/ -> /əh₁/ ; /a/ -> /əh₂/; /o₁/ -> /əh₃/; /i/-> /i/; /u/ -> /u/
E-grade: /e₂/, /o₂/ -> /e/ ; /e₁/ -> /eh₁/; /a/ -> /ah₂/; /o₁/ -> /o₁h₃/; /i/ -> /ey/; /u/ -> /ew/
O-grade: /e₂/, /o₂/ -> /o₂/ /e₁/ -> /o₂h₁/; /a/ -> /o₂h₂/; /o₁/ -> /o₂h₃/; /i/ -> /oy/; /u/ -> /ow/

[Šọ̈́gala]

I was toying with a similar idea, playing off of Carrasquer Vidal's model. If the basic difference between *ke- and *ka- (so-called plain velars) is the original vowel quality, not vowel coloring, even though both vowels correspond to *e in almost any other position, what if the difference between *He- and *Ha- is also purely from the original vowel quality, not vowel coloring as is normally assumed by the laryngeal model.

To spell this out, Carrasquer Vidal suggests traditional PIE *ke = underlying */qi/ > traditional PIE; and traditional *ka = underlying */qa/

I'm wondering if we can rule out that there was only one laryngeal: traditional *h₁e > *e was really *hi > *e; and traditional *h₂e > *a was really *ha > *a

In this model, it would be highly tempting to make *h₃e > *o really *hu > *o. All that would require is that *u behaves differently with *h than with stops: following a velar or uvular stop *u > *ʷe; following the laryngeal *u > *o; elsewhere *u > *e.

Anatolian data might refute the hypothesis that there was only ever one laryngeal sound, though.

[Nortaneous]

What's the evidence for o-colored *e being distinct from *o? I've heard that *eh1 and *e: have different outcomes in Proto-Anatolian, but I haven't heard that.

Greek *R̥h₂ *R̥h₃ > *Rā *Rō is imo R̥H > R̥əH followed by ə > e. This second step gives you the triple reflex for free after [# C]HC > [# C]HəC. Sanskrit reflects syllabic laryngeals as /i/ and has (IIRC) epenthetic /i/, which you get from CH̥C > CəC followed by ə > i. How do syllabic laryngeals interact with the development of aspiration? This predicts ChiC, which I think isn't the case... maybe the syllabic laryngeals were vocalized?

[KathTheDragon]

What's the evidence for o-colored *e being distinct from *o?

In Indo-Iranian, we have apophonic *o becoming *ā in an open syllable, but *h₃e becomes *a, and in Anatolian, we have stressed apophonic *o becoming *ō, but *h₃e becomes *(ḫ)o, which are chiefly distinguishable via lenition (stressed *ō lenites a following lenitable consonant, while *o does not)

[dhok]

Kath beat me to it. I took IE phonology with Lubotsky and Kroonen two summers ago, and the word in question when this came up was Skt. avis--by Brugmann's, we should see āvis, but we don't. A reconstruction of *h₃éwis suffices to explain the Sanskrit reflex as well as the rest of core IE and Hittite (ḫawiš, with a short a, right?)

Now my personal pet theory is that laryngeal coloring may not have been as extensive in Indo-Iranian as it was elsewhere. *h₁e, *h₂e, and *h₃e all merge as a, of course, and likewise for *eh₁, *eh₂ and *eh₃ as ā. Ollet 2014 points out that for *eH, there is good reason to believe that coloring did occur, because roots of the form *Keh₂C- show an unpalatalized initial consonant in Sanskrit. E.g., PIE *gʷeh₂- 'go' clearly had an e-grade root aorist (Greek ἔβην 'I went'), but the Sanskrit cognate is ágāt 'he went', not **ájāt. So there must have been coloring of *e by a following laryngeal. However, *h₃éwis and its like suggest that, perhaps, preceding laryngeals didn't cause coloring in Indo-Iranian.

Of course, as was pointed out to me by a helpful Twitter account, Lycian has a word 'χawa-' (is it just me, or does the board provide no graphical distinction between Roman <x> 'eks' and Greek <χ> 'chi'?)--*h₃éwis and *h₃ówis should both have given **χewa-. I think *h₂ówis~h₂éwis is the usual synthesis? But that's an unusual-looking ablaut pattern.

[jal]

does the board provide no graphical distinction between Roman <x> 'eks' and Greek <χ> 'chi'?)

I think the board font is just set to "sans-serif", so it'll display in whatever sans serif font your browser picks. With me, it's Arial, and it displays the difference fine.


JAL

[dhok]

Seems to be a theme issue. Subsilver2 distinguishes the two; prosilver doesn't. I've switched themes, now, but it might be worth looking into... (...particularly since Subsilver2 doesn't really distinguish subscript 2 and 3 when italicized...)

[KathTheDragon]

in Anatolian, we have stressed apophonic *o becoming *ō, but *h₃e becomes *(ḫ)o, which are chiefly distinguishable via lenition (stressed *ō lenites a following lenitable consonant, while *o does not)

Per Melchert, this account isn't quite correct - rather, *ó only lenited *h₂ instead of lengthening and participating in the more general Proto-Anatolian lenition. But afaik this still distinguished it from *h₃e.

However, *h₃éwis and its like suggest that, perhaps, preceding laryngeals didn't cause coloring in Indo-Iranian.

This is almost certainly impossible, since that would require laryngeal colouring to have been multiple sound changes, at least one of which was at least partly independent in the daughter languages. It is far tidier to assume a single colouring event in pre-PIE and Brugmann's law in Indo-Iranian as it is usually presented.

Of course, as was pointed out to me by a helpful Twitter account, Lycian has a word 'χawa-' (is it just me, or does the board provide no graphical distinction between Roman <x> 'eks' and Greek <χ> 'chi'?)--*h₃éwis and *h₃ówis should both have given **χewa-. I think *h₂ówis~h₂éwis is the usual synthesis? But that's an unusual-looking ablaut pattern.

Tocharian B āw "ewe" also attests to a stem *h₂ewi-, so an o~e ablaut pattern is unavoidable. Besides, it only seems unusual because people keep insisting it doesn't exist!

[Pole, the]

I was toying with a similar idea, playing off of Carrasquer Vidal's model. If the basic difference between *ke- and *ka- (so-called plain velars) is the original vowel quality, not vowel coloring, even though both vowels correspond to *e in almost any other position, what if the difference between *He- and *Ha- is also purely from the original vowel quality, not vowel coloring as is normally assumed by the laryngeal model.

To spell this out, Carrasquer Vidal suggests traditional PIE *ke = underlying */qi/ > traditional PIE; and traditional *ka = underlying */qa/

I'm wondering if we can rule out that there was only one laryngeal: traditional *h₁e > *e was really *hi > *e; and traditional *h₂e > *a was really *ha > *a

That pretty much misses the whole point of why the laryngeal theory was brought up in the first place — namely, to unify the e-grade vowel as *e.

[Šọ̈́gala]

That pretty much misses the whole point of why the laryngeal theory was brought up in the first place — namely, to unify the e-grade vowel as *e.

Well, yeah. Unifying the e-grade vowel is not an end in itself. I was starting from Carrasquer Vidal's Pre-Proto model where he's already scrapped that idea. Just one of several proposals for an earlier stage with a larger vowel inventory that fell together into our beloved *e.

I did forget to consider the triple reflex of the zero grade laryngeals. That might cause problems if I'm relying on vowel quality to do the work. I'm sure there's a fix for that, but it might not be an appealing one.

[Howl]

Three laryngeals is the simplest explanation for the Greek triple reflex. Then the ancestors of the Greeks used the appropriate e-grade vowel when they just needed a schwa-like vowel next to a laryngeal.

Of course, as was pointed out to me by a helpful Twitter account, Lycian has a word 'χawa-' (is it just me, or does the board provide no graphical distinction between Roman <x> 'eks' and Greek <χ> 'chi'?)--*h₃éwis and *h₃ówis should both have given **χewa-. I think *h₂ówis~h₂éwis is the usual synthesis? But that's an unusual-looking ablaut pattern.

Tocharian B āw "ewe" also attests to a stem *h₂ewi-, so an o~e ablaut pattern is unavoidable. Besides, it only seems unusual because people keep insisting it doesn't exist!

That's the basic acrostatic ablaut pattern from the Erlangen model. So how do these people explain Sanskrit gauh (gen sg. goh, English: cow) ?

But the whole problem with nominal ablaut is that so little remains of it in the daughter languages. Just a few words with irregular declensions. Most words reconstructed with ablaut are like this *h₂ewi. Two incompatible PIE forms from daughter languages that can be made to fit an existing ablaut pattern.

[KathTheDragon]

So how do these people explain Sanskrit gauh (gen sg. goh, English: cow) ?

Depends on the reconstruction, for which there are two: *gʷow- ~ *gʷew-, or *gʷeh₃u-. In the former, the length in the Sanskrit nominative is generalised out of the accusative, where *gʷowm > *gʷōm by Stang's law (necessary to explain the long vowel in Skt. dyauḥ < *dyew- "sky") while the oblique is regular. In the latter... I'm not too sure there, since I've not seen a fully detailed argument. Possibly the assumption that *gʷeh₃us > *gʷōus, and oblique *gʷh₃ews.

You also make a cogent point about nominal ablaut, but really, all we can do is project the forms we have back (unless there are good reasons to suppose other developments) and then generalise what few patterns there are.

[Howl]

Depends on the reconstruction, for which there are two: *gʷow- ~ *gʷew-, or *gʷeh₃u-. In the former, the length in the Sanskrit nominative is generalised out of the accusative, where *gʷowm > *gʷōm by Stang's law (necessary to explain the long vowel in Skt. dyauḥ < *dyew- "sky") while the oblique is regular. In the latter... I'm not too sure there, since I've not seen a fully detailed argument. Possibly the assumption that *gʷeh₃us > *gʷōus, and oblique *gʷh₃ews.

I thought it was a good example of e/o ablaut in the acrostatic. But on second thought, it's probably not.

You also make a cogent point about nominal ablaut, but really, all we can do is project the forms we have back (unless there are good reasons to suppose other developments) and then generalise what few patterns there are.

As a hobbyist, at least I don't have to pretend it's all science.

[mezziah]

So, uh, I hope this is the right thread for my question/idea. I just don't want to open a new thread for my quick question and it's quite related anyway. I read this entire thread a while back and I don't think my idea has been discussed before.

So, I was browsing a Hittite etymological dictionary (found here on zbb) out of boredome for the 4th time or so and compared Hittite words with Lycian ones whenever they are cited and almost always Hittite laryngeal (h2) equals Lycian χ and in one example I encountered both h2 and h3 in the reconstructed Hittite word and they correspond to Lycian χ and g respectively.

Now, how are chances, that h2 indeed and with a high degree of regularity equals the velar fricative /χ/, not only in Anatolian branch but also in PIE. /g/ is a bit more speculative but after thinking a while about it, i suggest /ɣ/, maybe labialized which would perfectly fit a popular proposal for the PIE value for h3.

Am I missing something or is it worth a second thought?

Edit:
Example word:
Hittite: ḫuḫḫa- (c.) ‘grandfather’, nom.sg. ḫu-uḫ-ḫa-aš; Derivatives: ḫuḫḫant- (c.) ‘(great)grandfather’ (nom.pl. ḫu-uḫ-ḫa-an-te-eš
Lycian: χuge- 'grandfather'; the author of the dictionary states that "the -g- reflects a lenited */-h-/"

[Zju]

The similarity between PIE *ḱerh₂- 'horn' and PS *qarn- 'horn' seems to have been noticed, but has anyone considered it in depth? Since the PIE word is a possible derivation of *ḱer- 'to grow, increase' maybe we're looking at a PIE borrowing in PS. But then the 'grow, increase' root is also reconstructed as *ḱerh₁- or *ḱreh₁-.

[Zaarin]

The similarity between PIE *ḱerh₂- 'horn' and PS *qarn- 'horn' seems to have been noticed, but has anyone considered it in depth? Since the PIE word is a possible derivation of *ḱer- 'to grow, increase' maybe we're looking at a PIE borrowing in PS. But then the 'grow, increase' root is also reconstructed as *ḱerh₁- or *ḱreh₁-.

IMO this really only makes sense, in either direction, under the assumption that <ḱ> = /k/ (borrowing of plain unvoiced stops as ejectives is well attested in both Akkadian and Phoenician [and perhaps Hebrew as well]), not /c/ (which certainly wouldn't be used to borrow PS /k'/, and would probably be interpreted as an affricate when borrowed into PS). I suspect we're looking at a coincidence here, as tempting as the similarity looks.

[Sumelic]

There was a discussion of this relatively recently on the WordReference forums: https://forum.wordreference.com/threads/horn.3391577/

[Salmoneus]

The similarity between PIE *ḱerh₂- 'horn' and PS *qarn- 'horn' seems to have been noticed, but has anyone considered it in depth? Since the PIE word is a possible derivation of *ḱer- 'to grow, increase' maybe we're looking at a PIE borrowing in PS. But then the 'grow, increase' root is also reconstructed as *ḱerh₁- or *ḱreh₁-.

Proto-Indo-European was spoken probably in Ukraine. Proto-Semitic was spoken probably in the Levant, probably thousands of years earlier. They're not exactly obvious candidates for interchanging basic vocabulary (particularly vocabulary for things that would have been known to both groups - I mean, wine, sure, that word could travel a long way, but 'horn'?).

To get them even vaguely close, you'd have to assume that PIE came to Ukraine through a mass migration of Georgian women, and Georgia and the Levant still aren't exactly next to one another.

If there is common vocabulary, the parsimonious solution would be a third language as the source. In particular, we know there was massive migration of women from the Caucasus to Ukraine, and we also know there was substantial migration from the Caucasus westward to Anatolia, the Levant and the western Mediterranean. We might hypothesise an Ur-Caucasian that lent vocabulary to both Semitic and Yamnaya pastoralists.

But random chance is still probably the simplest explanation.

[Zaarin]

Proto-Semitic was spoken probably in the Levant, probably thousands of years earlier.

Lipinski agrees with you, but a lot of Semiticists these days lean towards Arabia or Ethiopia. Which, of course, is even farther from the PIE homeland, no matter where you put it.

[Salmoneus]

Proto-Semitic was spoken probably in the Levant, probably thousands of years earlier.

Lipinski agrees with you, but a lot of Semiticists these days lean towards Arabia or Ethiopia. Which, of course, is even farther from the PIE homeland, no matter where you put it.

Why Ethiopia? Doesn't Proto-Semitic have words for things like camels, horses, figs, ice, oak trees, almonds and viticulture?