The Great Proto-Indo-European Thread

[Drydic]

The quantitative and qualitative improvement of the presentation of the Indo-European material has reached a critical mass, allowing the solution of all major problems of PIE segmental phonology based on the comparative method of reconstruction. This window of opportunity will be explored in this study with a completely upgraded reconstruction theory, called System PIE, which is based on strict principles of natural science. In essence, System PIE consists of the primary phoneme inventory and the upgraded sound law system for Proto-Indo-European, with particular attention paid to the segmental laryngeal PIE h in all environments. As such, System PIE is designed to solve the critical problems of PIE phonology and open the way for a subsequent exploration of the breakthrough, especially in the fields of PIE morphology, etymology and the accent of the proto-language. Concerning these Schwerpunkts, the following preliminary remarks are presented.

It is difficult to take anyone seriously when the first thing their style reminds me of is North Korean propaganda.

GLORIOUS ARYAN FIGHTING SPIRIT

[Morrígan]

Incidentally, has anybody managed to read the paper in which Colarusso argues that IE and NWC languages are descended from a common ancestor?

I have read this, though it's been a while. One issue I have is how much it relies on a reconstruction of proto-NWC that I don't have (and probably doesn't exist in sufficient detail).

[WeepingElf]

The quantitative and qualitative improvement of the presentation of the Indo-European material has reached a critical mass, allowing the solution of all major problems of PIE segmental phonology based on the comparative method of reconstruction. This window of opportunity will be explored in this study with a completely upgraded reconstruction theory, called System PIE, which is based on strict principles of natural science. In essence, System PIE consists of the primary phoneme inventory and the upgraded sound law system for Proto-Indo-European, with particular attention paid to the segmental laryngeal PIE h in all environments. As such, System PIE is designed to solve the critical problems of PIE phonology and open the way for a subsequent exploration of the breakthrough, especially in the fields of PIE morphology, etymology and the accent of the proto-language. Concerning these Schwerpunkts, the following preliminary remarks are presented.

It is difficult to take anyone seriously when the first thing their style reminds me of is North Korean propaganda.

This indeed rings a crackpot alarm, and casts a shadow of doubt on the whole work; yet, that still doesn't mean that he can't be right in some points.

Incidentally, has anybody managed to read the paper in which Colarusso argues that IE and NWC languages are descended from a common ancestor?

I have read this, though it's been a while. One issue I have is how much it relies on a reconstruction of proto-NWC that I don't have (and probably doesn't exist in sufficient detail).

I don't know Colarusso's work; yet, I have read a sketch of Northwest Cauasian that was published in Lingua a few years ago (the same issue also contained sketches of NEC and Kartvelian), and find it hard to believe that those should be the closest living kin of IE. Rather, I assume that what similarities there are between PIE and Caucasian languages are due to contact. But as I haven't seen what Colarusso wrote about this matter, I cannot judge his evidence.

[Morrígan]

Rather, I assume that what similarities there are between PIE and Caucasian languages are due to contact. But as I haven't seen what Colarusso wrote about this matter, I cannot judge his evidence.

Yeah, contact seems WAY more plausible. I think I have a PDF of it somewhere, or at least I should be able to track down one of his papers. The whole proposal seems unreasonable given the gaps in time-depth for PIE vs PWC.

What's often struck me as interesting is glottalic-uvular idea, where *k *g *gʰ are |qʰ q qʼ|, *ḱ *ǵ *ǵʰ are |kʰ k kʼ|, and *kʷ *gʷ *kʷʰ are |qʷʰ qʷ qʾʷ|, or similar. This makes the inventory not just more Caucasian, but also makes it more similar to a lot of American (and maybe Siberian) languages, which have k-q-qʷ contrasts.

[Pole, the]

What's often struck me as interesting is glottalic-uvular idea, where *k *g *gʰ are |qʰ q qʼ|, *ḱ *ǵ *ǵʰ are |kʰ k kʼ|, and *kʷ *gʷ *kʷʰ are |qʷʰ qʷ qʾʷ|, or similar. This makes the inventory not just more Caucasian, but also makes it more similar to a lot of American (and maybe Siberian) languages, which have k-q-qʷ contrasts.

Wasn't it *g-series reanalyzed as ejectives?

[WeepingElf]

What's often struck me as interesting is glottalic-uvular idea, where *k *g *gʰ are |qʰ q qʼ|, *ḱ *ǵ *ǵʰ are |kʰ k kʼ|, and *kʷ *gʷ *kʷʰ are |qʷʰ qʷ qʾʷ|, or similar. This makes the inventory not just more Caucasian, but also makes it more similar to a lot of American (and maybe Siberian) languages, which have k-q-qʷ contrasts.

Wasn't it *g-series reanalyzed as ejectives?

Yes. And I don't think the PIE stops ever were all voiceless. Why should they? They aren't in the Caucasian languages. The *Dh set has voiced reflexes everywhere except Greek and Tocharian.

But reinterpreting the "palatovelars" as velars, the "plain velars" as uvulars and the "labiovelars" as labialized uvulars IMHO makes sense.

[speculation]
I think the three velar series emerged in connection with what I call the "Great Vowel Collapse" (GVC), a sound change in Pre-PIE wherein all vowels except *i and *u merged into *a. If the vowel was front, an adjacent velar was fronted; if the vowel was rounded, an adjacent velar was labialized. Like this: *ke > *ḱa, *ka > *ka, *ko > *kʷa. There was another change preceding the GVC which I call "Resonant-Conditioned Lowering" (RCL) in which high vowels were lowered before resonants (nasals, liquids, semivowels) such that they fell victim to the GVC. This explains why there are no high vowels followed by resonants (at least not in roots) in PIE.
[/speculation]

[KathTheDragon]

So, where does PIE ablaut come from then, in your speculative PIE?

[WeepingElf]

So, where does PIE ablaut come from then, in your speculative PIE?

I haven't solved this problem yet. Pre-PIE (after GVC) would have had the three vowels */a/, */i/ and */u/, with */a/ disproportionately more common than the others, and a penultimate accent. The accent is the main cause of ablaut, I think, with accented vowels developing e-grade, most unaccented vowels developing zero grade and at least some final vowels developing o-grade. But there is quite a large "messy residue" that doesn't fit this simple rule; there were many other processes in operation - analogy, decompounding, accent shifts, you name it. I have no good idea of those things yet. Basically, the ablaut series are:

*/a/ > */e ~ o ~ 0/
*/i/ > */ei ~ oi ~ i/
*/u/ > */eu ~ ou ~ u/

Pre-PIE may also have had the diphthongs */ai/ and */au/, which would have resulted in the same ablaut series as */i/ and */u/, respectively.

But maybe all this is just hogwash, and things actually were completely different.

[Morrígan]

Wasn't it *g-series reanalyzed as ejectives?

Yeah, my brain didn't work.

Yes. And I don't think the PIE stops ever were all voiceless. Why should they? They aren't in the Caucasian languages. The *Dh set has voiced reflexes everywhere except Greek and Tocharian.
But reinterpreting the "palatovelars" as velars, the "plain velars" as uvulars and the "labiovelars" as labialized uvulars IMHO makes sense.

Yeah sorry, I was at work, so I did that kind of sloppily. The phonation is probably easy to figure out, in terms of maximum-parsimony. But having the uvulars is really nice in terms of making it look correct in terms of it's language area in the Pontic steppe.

[Herr Dunkel]

And then in five years someone brings out a coherent correspondences list between Proto-Indo-European and Proto-Kartvelian.

[Terra]

There was another change preceding the GVC which I call "Resonant-Conditioned Lowering" (RCL) in which high vowels were lowered before resonants (nasals, liquids, semivowels) such that they fell victim to the GVC. This explains why there are no high vowels followed by resonants (at least not in roots) in PIE.

Really? I never noticed this... Couldn't there have been a different change though: The resonant turned into a plosive. So m > b/bh, n > d/dh. To spot it, I suppose you'd look for similar-looking roots that alternate between the two. (Japanese has sabishii/samishii and samui/sabui.)

[KathTheDragon]

What about *gʰelunehₐ-, meaning 'lip'?

[Salmoneus]

What about *gʰelunehₐ-, meaning 'lip'?

For what it's worth, that word probably didn't exist. For the supposed Greek reflex, Beekes says 'no known etymology' and dismisses the Armenian and Germanic cognates. For the Armenian reflex, Martosiryan calls it doubtful, for good reason - it means 'palate', which isn't that close to 'lip', and actually the earlier uses all have it meaning 'ceiling', which is even less lippy. Orel, over in Germanic, helpfully feels no need to bother giving a PIE word, since the three cognates are all clearly "identical" to one another (despite the denials of experts in the other fields). Even in Germanic, the reflex only shows up in Old Norse and that's it. [Orel helpfully points out that there's a word in Swabian meaning 'gill' that also begins with a <g>. Fortunately I don't have the print copy of Orel, or else I'd probably have damaged it by now]

Anyway, I think what welf meant was that in PIE you can't have a high vowel before a CODA nasal or liquid. This probably isn't true either, but maybe could be made true if you limit yourself to stressed syllables and take ablaut into account (otherwise you have words like *pungénti, 'they sting', to take account of).

Personally, I'm not sure what welf is gaining by trying to use the high vowels as vowels. There aren't all that many words where they can't be explained as zero grade of the diphthongs, are there? Those that are could be irregular developments, loanwords, or zero grades from otherwise unattested roots.
I'd have thought it would have been more sensible to have either two or three vowels - *a, *@, and maybe *o. *@ would be the unstressed form of either of the other two vowels (assuming that it's more helpful to have a *o, and that that can't just be explained by surrounding consonants and analogy). *j and *w could then just be treated as consonants.

[Particles the Greek]

I can't help wondering what actual professional linguists would make of this sort of discussion. It feels like we're doing someone else's work for them.

[WeepingElf]

Anyway, I think what welf meant was that in PIE you can't have a high vowel before a CODA nasal or liquid. This probably isn't true either, but maybe could be made true if you limit yourself to stressed syllables and take ablaut into account (otherwise you have words like *pungénti, 'they sting', to take account of).

Personally, I'm not sure what welf is gaining by trying to use the high vowels as vowels. There aren't all that many words where they can't be explained as zero grade of the diphthongs, are there? Those that are could be irregular developments, loanwords, or zero grades from otherwise unattested roots.
I'd have thought it would have been more sensible to have either two or three vowels - *a, *@, and maybe *o. *@ would be the unstressed form of either of the other two vowels (assuming that it's more helpful to have a *o, and that that can't just be explained by surrounding consonants and analogy). *j and *w could then just be treated as consonants.

What I meant is that *CeiR- and *CeuR- roots do not exist. There are PIE word forms where a resonant follows a high vowel or diphthong, but there is always a morpheme boundary between the relevant sounds (I don't know about the *pungénti example you gave, but this looks like a n-infix present from a root *peug-). However, in this way, high vowels behave just like nasals and liquids - in a *CeRC- root, the second C must always be an obstruent. And it is indeed the case that high vowels in PIE are, for the most part at least, just zero grades of diphthongs. So they may have been diphthongs */ai/ and */au/ in pre-ablaut Pre-PIE rather than high monophthongs; but I wanted to avoid a language stage which has only one vowel phoneme because that would be typologically highly peculiar and probably unstable.

It would help if an external relative of IE could be found - then one could look to what PIE roots with diphthongs are cognate, and what the IE cognates of the other group's words with high vowels look like.

[KathTheDragon]

I don't know about the *pungénti example you gave, but this looks like a n-infix present from a root *peug-

This seems to be the explanation, as I have here *peug- 'prick, poke'.

[Pole, the]

So they may have been diphthongs */ai/ and */au/ in pre-ablaut Pre-PIE rather than high monophthongs; but I wanted to avoid a language stage which has only one vowel phoneme because that would be typologically highly peculiar and probably unstable.

There does not have to be a one-vowel stage here.

Say you have a language with four vowels, /i u ɛ ɑ/.
0. … and there is some kind of morphological lengthening employed. /i iː ɛ ɛː u uː ɑ ɑː/
1. High vowels get diphthongized to /ɛj ɛːj ɑw ɑːw/ and short low vowels are given high allophones [ɪ ə] in unstressed positions. (This kind of shift is not much probable, but IMHO it could happen.) /ɛ ɛː ɑ ɑː/
2. Reduced vowels are elided when applicable. Long vowels are shortened when unstressed, thus making [ɪ ə] phonemic. /ɪ ɛ ɛː ə ɑ ɑː/
3. Front and back vowels merge, leading to an unstable, Caucasian-like vertical system. /ə a aː/
4. /aː/ shifts to /ɛ/ and merges with the schwa. /a/ moves back. Voilà. /ɛ ɑ/
5. Or a bit more. Voilà. /e o/
You have just created a PIE-like ablaut system and never reached less than two vocalic phonemes. Gz.

[WeepingElf]

So they may have been diphthongs */ai/ and */au/ in pre-ablaut Pre-PIE rather than high monophthongs; but I wanted to avoid a language stage which has only one vowel phoneme because that would be typologically highly peculiar and probably unstable.

There does not have to be a one-vowel stage here.

Say you have a language with four vowels, /i u ɛ ɑ/.
0. … and there is some kind of morphological lengthening employed. /i iː ɛ ɛː u uː ɑ ɑː/
1. High vowels get diphthongized to /ɛj ɛːj ɑw ɑːw/ and short low vowels are given high allophones [ɪ ə] in unstressed positions. (This kind of shift is not much probable, but IMHO it could happen.) /ɛ ɛː ɑ ɑː/
2. Reduced vowels are elided when applicable. Long vowels are shortened when unstressed, thus making [ɪ ə] phonemic. /ɪ ɛ ɛː ə ɑ ɑː/
3. Front and back vowels merge, leading to an unstable, Caucasian-like vertical system. /ə a aː/
4. /aː/ shifts to /ɛ/ and merges with the schwa. /a/ moves back. Voilà. /ɛ ɑ/
5. Or a bit more. Voilà. /e o/
You have just created a PIE-like ablaut system and never reached less than two vocalic phonemes. Gz.

Seems to work, but I need to look closer.

[Terra]

This seems to be the explanation, as I have here *peug- 'prick, poke'.

Agreed. May I ask, does anybody know how common this present n-infix is in branches of PIE other than Celtic and Italic?

[KathTheDragon]

I do know that 'stand' is a relic of it in Germanic. Only one I can think of, though.

[Salmoneus]

I don't know about frequency, but it's also assumed to occur in Greek, Sanskrit, Hittite, etc.
Just looking at Beekes, he mentions it explicitly for Greek "denamai", presumably from *deu2, Latin "nanciscor" vs Sanskrit 'nás'ati', Hittite "s'arnink" (vs Latin "sarcio"), Sanskrit "yunákti" (from *yugóm, 'yoke'), among others. And of course there are the Hittite productive (?) -nu- causative class, which are generally analysed as being from a nasal 'infix' plus a -u suffix.

[Nortaneous]

Is there any reason why *e *o couldn't have been [æ ɔ] or so? That would resolve the problem of there being no /a/ (Seri has /æ ɑ i o/) and answer the question of why so many IE langs have low vowel mergers of some sort.

[tezcatlip0ca]

Is there any reason why *e *o couldn't have been [æ ɔ] or so? That would resolve the problem of there being no /a/ (Seri has /æ ɑ i o/) and answer the question of why so many IE langs have low vowel mergers of some sort.

Actually I’ve considered this for a while. It seems that the vowel system in late PIE mirrors that in Arapaho or Malagasy. As for the high vowel restriction before sonants, I think it has to do with the fact that high vowels are phonemically also syllabic sonants. Maybe PIE just post-ablaut had a vertical three-vowel system, so that /ɨR/ sequences became the syllabic sonants in the zero grade, and a phonotactic restriction against two consecutive sonants at the end of a root.

Off topic, Nortaneous, why is the text in the quote right to left? Is there a sudden u+202e in your username or something?

[Particles the Greek]

What are the main objections to Szemerényi's reconstruction of PIE with one laryngeal /h/, five full vowels, and voiceless aspirates?

[Hallow XIII]

Off topic, Nortaneous, why is the text in the quote right to left? Is there a sudden u+202e in your username or something?

He does exactly this, yes.