The Great Proto-Indo-European Thread
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
In a nutshell, dubious etymologies. For example, he claims that the well-attested root for "eagle", *h₃er-, is connected to *h₂erǵ-i- attested in several words for eagles. Notwithstanding the fact that these words are all epithets, and thus the roots do not need to refer to eagles originally, he assumes a full-grade which is unjustified. A zero-grade would equally well yield a in all branches that display it, and hence the laryngeal is indeterminate. It's understandable that Jouna cannot posit a zero grade since he doesn't believe that zero-grades could yield full vowels in the daughters. Similarly for the "orphan" words, he cites the Hittite imperfective harappiske/a- from harp- "change allegiance, etc." as pointing to etymological *p - but compare akkuske/a- from eku- "drink", which clearly contains *gʷʰ, and thus his case here vanishes. And so on for the majority of his examples. But even outside Hittite, there is plenty of evidence for *h₃ in the form of the Greco-Armenian prothetic vowels.
- JounaPyysalo
- Lebom
- Posts: 99
- Joined: Mon Jun 01, 2015 4:08 am
- Contact:
Re: The Great Proto-Indo-European Thread
Kathy,
1. Maced. ἀργιό·πους : Arm. arci·w do not point to a zero grade, but to Indo-European /a/. Regarding the connection between the epithets meaning "moving" this is quite understandable, because the root PIE √hɑr- meant just that, not only in PIE *hɑor- -› Hitt.etc. √har- "eagle, bird" but in PIE *haergi- "eagle, bird, moving swiftly" etc.
2. Hitt. harappiske/a- could be written as "harpiske/a-", but it is not. Hence the etymology provided.
More generally the point is that there are no examples of h₃ at all, which is simply a postulate inspired by the Semitic typology of Möller (and his successors), all of whom wanted to reconstruct a pre-proto-language à la Semitic (i.e. with the root shape CC·(C) by means of replacing the Indo-European vowels with "laryngeals" H1 H2 H3.
All this has now come to an end as the developers of the theory have gone through all possible permutations of preserved/lost laryngeals and are unable to present any distribution not inconsistent with the data.
It's all over for the laryngeal theory, a development I see very positive in the sense that I am of the view that Indo-European linguistics is to reconstruct Proto-Indo-European, not Pre-Proto-Indo-Semitic.
J.
1. Maced. ἀργιό·πους : Arm. arci·w do not point to a zero grade, but to Indo-European /a/. Regarding the connection between the epithets meaning "moving" this is quite understandable, because the root PIE √hɑr- meant just that, not only in PIE *hɑor- -› Hitt.etc. √har- "eagle, bird" but in PIE *haergi- "eagle, bird, moving swiftly" etc.
2. Hitt. harappiske/a- could be written as "harpiske/a-", but it is not. Hence the etymology provided.
More generally the point is that there are no examples of h₃ at all, which is simply a postulate inspired by the Semitic typology of Möller (and his successors), all of whom wanted to reconstruct a pre-proto-language à la Semitic (i.e. with the root shape CC·(C) by means of replacing the Indo-European vowels with "laryngeals" H1 H2 H3.
All this has now come to an end as the developers of the theory have gone through all possible permutations of preserved/lost laryngeals and are unable to present any distribution not inconsistent with the data.
It's all over for the laryngeal theory, a development I see very positive in the sense that I am of the view that Indo-European linguistics is to reconstruct Proto-Indo-European, not Pre-Proto-Indo-Semitic.
J.
- JounaPyysalo
- Lebom
- Posts: 99
- Joined: Mon Jun 01, 2015 4:08 am
- Contact:
Re: The Great Proto-Indo-European Thread
Sumelic,
It is difficult to give a single answer to your inquiry, because three versions of three-laryngealism have been presented in the post-classical theory, all with different treatment of h3. These are:
1. Eichner's theory – with only "h2" preserved in Hittite
2. Melchert's theory – with "h2" fully preserved and "h3" preserved in initial position in Hittite.
3. Kortlandt's theory with h2 and h3 both preserved before *e, but equally both lost before *o.
The bottom line is that none of these is capable of presenting a consistent reconstruction of the Anatolian and more generally Indo-European languages, due to which only a single option for the reconstruction of the Proto-Indo-European exists today: The "monolaryngealism" initiated by Zgusta, developed by Szemerényi, and others.
J.
It is difficult to give a single answer to your inquiry, because three versions of three-laryngealism have been presented in the post-classical theory, all with different treatment of h3. These are:
1. Eichner's theory – with only "h2" preserved in Hittite
2. Melchert's theory – with "h2" fully preserved and "h3" preserved in initial position in Hittite.
3. Kortlandt's theory with h2 and h3 both preserved before *e, but equally both lost before *o.
The bottom line is that none of these is capable of presenting a consistent reconstruction of the Anatolian and more generally Indo-European languages, due to which only a single option for the reconstruction of the Proto-Indo-European exists today: The "monolaryngealism" initiated by Zgusta, developed by Szemerényi, and others.
J.
Re: The Great Proto-Indo-European Thread
I no nothing about the field of IE linguistics, but if I'm not mistaken, the laryngeal theory holds still a majority position amongst both scholars and ZBB'ers in this thread. Also, as I understand, it is your own theory that tries to lay the laryngeal theory to a rest. In which case it's mightely arrogant of you to proclaim what you did.JounaPyysalo wrote:All this has now come to an end (...) It's all over for the laryngeal theory
JAL
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
Do not call me Kathy. Kath is perfectly fine, and is a whole letter shorter.JounaPyysalo wrote:Kathy,
I've addressed this, since you do not believe that syllabic sonorants could excrete vowels. No further comment is needed.1. Maced. ἀργιό·πους : Arm. arci·w do not point to a zero grade, but to Indo-European /a/.
Did you not read my post? The geminate is due to the imperfective, not the root etymology. Compare akkuske/a from ekuzi < *h₁egʷʰ-, lakkiske/a from lāki < *legʰ-, hukkiske/a- from huekzi < *h₂wegʰ-... The evidence for regular gemination of a lenis stop in Hittite imperfectives is overwhelming, so unless every such root etymology is wrong, your objection is untenable. Done here.2. Hitt. harappiske/a- could be written as "harpiske/a-", but it is not. Hence the etymology provided.
- JounaPyysalo
- Lebom
- Posts: 99
- Joined: Mon Jun 01, 2015 4:08 am
- Contact:
Re: The Great Proto-Indo-European Thread
Kath,
1. In laryngeal theory, when there is a vowel quality allegedly from a syllabic resonant preceded by a laryngeal, it is taken (Risch's rule) that the original quality of that laryngeal is reflected in Greek. In this case . ἀργό- (a.) ‘schnell beweglich : swift’ (GEW 1:132, ἀργός) most certainly indicates what was the initial laryngeal.
2. When there is a CVCCV in Hittite, it's written CVCCV, not CVCVC(C), because the former is possible to display in the cuneiform notation. Therefore Hitt. ḫar-ap-pí- has a genuine vowel in the second syllable.
3. I read your posting quite thorough. Thus I am well aware that Kloekhorst suggests that kind of rule, but the thing is that Kloekhorst does not mention that there are "imperfectives" not behaving similarly, e.g. Hitt. ḫatugiške/a- (HIL. 392, [ḫa-]tu-ki-iš-ke/a-) with a single stop. In my opinion it is not sufficient to know what is said on books, but more importantly one should also know what the languages themselves say before accepting something as the truth.
1. In laryngeal theory, when there is a vowel quality allegedly from a syllabic resonant preceded by a laryngeal, it is taken (Risch's rule) that the original quality of that laryngeal is reflected in Greek. In this case . ἀργό- (a.) ‘schnell beweglich : swift’ (GEW 1:132, ἀργός) most certainly indicates what was the initial laryngeal.
2. When there is a CVCCV in Hittite, it's written CVCCV, not CVCVC(C), because the former is possible to display in the cuneiform notation. Therefore Hitt. ḫar-ap-pí- has a genuine vowel in the second syllable.
3. I read your posting quite thorough. Thus I am well aware that Kloekhorst suggests that kind of rule, but the thing is that Kloekhorst does not mention that there are "imperfectives" not behaving similarly, e.g. Hitt. ḫatugiške/a- (HIL. 392, [ḫa-]tu-ki-iš-ke/a-) with a single stop. In my opinion it is not sufficient to know what is said on books, but more importantly one should also know what the languages themselves say before accepting something as the truth.
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
Fair enough, but I still don't think there's any evidence that this root should be connected with the "eagle" words.JounaPyysalo wrote:1. In laryngeal theory, when there is a vowel quality allegedly from a syllabic resonant preceded by a laryngeal, it is taken (Risch's rule) that the original quality of that laryngeal is reflected in Greek. In this case . ἀργό- (a.) ‘schnell beweglich : swift’ (GEW 1:132, ἀργός) most certainly indicates what was the initial laryngeal.
Which doesn't apply in this case, since here we have CVCCCV, with three consonants.2. When there is a CVCCV in Hittite, it's written CVCCV, not CVCVC(C), because the former is possible to display in the cuneiform notation. Therefore Hitt. ḫar-ap-pí- has a genuine vowel in the second syllable.
But you still need to explain those imperfectives that do display gemination in the imperfective but not the root, since the word under consideration does the same thing.3. I read your posting quite thorough. Thus I am well aware that Kloekhorst suggests that kind of rule, but the thing is that Kloekhorst does not mention that there are "imperfectives" not behaving similarly, e.g. Hitt. ḫatugiške/a- (HIL. 392, [ḫa-]tu-ki-iš-ke/a-) with a single stop. In my opinion it is not sufficient to know what is said on books, but more importantly one should also know what the languages themselves say before accepting something as the truth.
- JounaPyysalo
- Lebom
- Posts: 99
- Joined: Mon Jun 01, 2015 4:08 am
- Contact:
Re: The Great Proto-Indo-European Thread
Kath,
I am perhaps unable to follow your comment "Which doesn't apply in this case, since here we have CVCCCV, with three consonants", if referring to Hitt. ḫar-ap-pí-, but I'll try anyway:
All late Anatolian languages confirm that the Hittite gemination — like -pp- here — was not a cluster of two consonants, but a single one, which the Hittites could perfectly well write Hitt. ḫar-pí- (in which case the voice/voicelessness can no longer be detected). To prove my point, there are two alternative stems for this very stem, because in addition to /harapiski-/ in Hitt. ḫar-ap-pí- there is also a stem Ḫi. ḫarpia- (vb1A.) ‘abtrünnig machen’ (HHand. 43, ḫar-pí-ia-mi [1sg]) with a zero grade in the suffix.
"But you still need to explain those imperfectives that do display gemination in the imperfective". Indeed I do — and I prefer to do this by means of external comparison instead of an overstated rule that would lead in to inconsistencies with the data, therefore preferring e.g. the comparison Hitt. ḫarap-: Aegin. ἀρεπυια- and Gr. ἀρέψ-.
I am perhaps unable to follow your comment "Which doesn't apply in this case, since here we have CVCCCV, with three consonants", if referring to Hitt. ḫar-ap-pí-, but I'll try anyway:
All late Anatolian languages confirm that the Hittite gemination — like -pp- here — was not a cluster of two consonants, but a single one, which the Hittites could perfectly well write Hitt. ḫar-pí- (in which case the voice/voicelessness can no longer be detected). To prove my point, there are two alternative stems for this very stem, because in addition to /harapiski-/ in Hitt. ḫar-ap-pí- there is also a stem Ḫi. ḫarpia- (vb1A.) ‘abtrünnig machen’ (HHand. 43, ḫar-pí-ia-mi [1sg]) with a zero grade in the suffix.
"But you still need to explain those imperfectives that do display gemination in the imperfective". Indeed I do — and I prefer to do this by means of external comparison instead of an overstated rule that would lead in to inconsistencies with the data, therefore preferring e.g. the comparison Hitt. ḫarap-: Aegin. ἀρεπυια- and Gr. ἀρέψ-.
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
Ok, let me explain! What I am referring to here is not three spoken consonants, since -pp- was indeed a geminate, and thus only a single consonant, but three written consonants. The Hittites do have a variable tendency to write the full geminate spelling in clusters, but when they do, they must insert a vowel somewhere, and since the geminate is a single consonant, they can't place it between the geminate letters (*-rpap-), so it has to go in the other slot (-rapp-). Therefore, proceeding logically from the case of three graphic consonants representing two spoken consonants, we cannot conclude whether spellings like harappiske/a- are phonologically /harap:iske-/, /harp:iske-/, or /hrap:iske-/. For a parallel, consider hartagga- /hərtk:a-/ < *h₂r̩tḱo- whose second graphic vowel cannot be real.
Re: The Great Proto-Indo-European Thread
As you seem to be in command of Hittite today, and I can't remember how to find out, how is 'harappiske/a-' spelt?KathTheDragon wrote:... we cannot conclude whether spellings like harappiske/a- are phonologically /harap:iske-/, /harp:iske-/, or /hrap:iske-/.
Is the 'cannot' based on the surmised etymology, or is there another reason for eliminating the possibly of the second graphic vowel of 'hartagga' being phonetically present? The best I can remember ever coming up with was the idea that TAK would be preferred for a consonant cluster, as the medial vowel is unspecified by the spelling.KathTheDragon wrote:For a parallel, consider hartagga- /hərtk:a-/ < *h₂r̩tḱo- whose second graphic vowel cannot be real.
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
The usual spelling is har-ap-pí-is-ke/a, though there is a later spelling without the ap. For comparison, the stem is spelt har-ap- before a consonant, and har-pV- before a vowel, which incidentally proves a pronounciation /harp(:)-/
It's based on etymological considerations. If the etymology were unknown, then /hartak:a-/ would be another possible reading.
It's based on etymological considerations. If the etymology were unknown, then /hartak:a-/ would be another possible reading.
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
Following up on this, what about words like eku- "drink", imperfective akkuske/a-? Explain that.JounaPyysalo wrote:"But you still need to explain those imperfectives that do display gemination in the imperfective". Indeed I do — and I prefer to do this by means of external comparison instead of an overstated rule that would lead in to inconsistencies with the data
- JounaPyysalo
- Lebom
- Posts: 99
- Joined: Mon Jun 01, 2015 4:08 am
- Contact:
Re: The Great Proto-Indo-European Thread
Kath,
This is very strange as I thought I already posted a response more than a week ago – and now I don't see it anywhere.
Well, perhaps I made a mistake so here it is again as well as far as I can remember:
The double writing of the stops is problematic due to the ambiguity: In addition to the internal rules favoured by Kloekhorst and some others there were alternations in the voice of the stops in PIE itself, where roots ending in a voiceless stop —T(h) had alternations ending in —D(h).
I am, however, at this point far more interested in the fundamentals of Indo-European linguistics, and especially about the issues related to the three-laryngealism and monolaryngealism, of which I am currently discussing both on the PIE Lexicon FB-page as well as in the FB-group for Proto-Indo-European
https://www.facebook.com/groups/126696954194312/
Once more, my apologies for the very belated response, I don't understand how my previous posting was lost.
Jouna
This is very strange as I thought I already posted a response more than a week ago – and now I don't see it anywhere.
Well, perhaps I made a mistake so here it is again as well as far as I can remember:
The double writing of the stops is problematic due to the ambiguity: In addition to the internal rules favoured by Kloekhorst and some others there were alternations in the voice of the stops in PIE itself, where roots ending in a voiceless stop —T(h) had alternations ending in —D(h).
I am, however, at this point far more interested in the fundamentals of Indo-European linguistics, and especially about the issues related to the three-laryngealism and monolaryngealism, of which I am currently discussing both on the PIE Lexicon FB-page as well as in the FB-group for Proto-Indo-European
https://www.facebook.com/groups/126696954194312/
Once more, my apologies for the very belated response, I don't understand how my previous posting was lost.
Jouna
Re: The Great Proto-Indo-European Thread
Meanwhile, for people who think there's something worthwhile in the standard reconstruction of PIE (so, I would dare to presume, probably most people in this thread not named Jouna), I present a digitized list of all verb roots in LIV, and some analytic observations:
https://protouralic.wordpress.com/2016/ ... he-people/
https://protouralic.wordpress.com/2016/ ... he-people/
[ˌʔaɪsəˈpʰɻ̊ʷoʊpɪɫ ˈʔæɫkəɦɔɫ]
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
Wow, Tropylium! That's simply amazing. This'll be very useful.
Edit: Incidentally, the LIV² has several biases in their reconstruction, most notably that they include roots only found in single branches, but also their treatment of laryngeals (I seem to recall it stated somewhere that they only reconstruct *h₂ when aspiration affects are found in IIr, whereas there is evidence that all laryngeals caused aspiration). They also mention from time to time that some roots may in fact be extended versions of others, or simply the same root. Would it be possible to indicate these somehow?
Edit: Incidentally, the LIV² has several biases in their reconstruction, most notably that they include roots only found in single branches, but also their treatment of laryngeals (I seem to recall it stated somewhere that they only reconstruct *h₂ when aspiration affects are found in IIr, whereas there is evidence that all laryngeals caused aspiration). They also mention from time to time that some roots may in fact be extended versions of others, or simply the same root. Would it be possible to indicate these somehow?
Re: The Great Proto-Indo-European Thread
Roots found only in a single branch will be simple enough to indicate: I plan on adding complete distribution stats, to also allow digging up Italo-Celtic or Indo-Greek or whatever roots as desired. Derivation, where established, is also going to go neatly enough in its own column (and perhaps a separate column "suffix/root extension contained, if any" could be added).KathTheDragon wrote:Incidentally, the LIV² has several biases in their reconstruction, most notably that they include roots only found in single branches, but also their treatment of laryngeals (I seem to recall it stated somewhere that they only reconstruct *h₂ when aspiration affects are found in IIr, whereas there is evidence that all laryngeals caused aspiration). They also mention from time to time that some roots may in fact be extended versions of others, or simply the same root. Would it be possible to indicate these somehow?
Dubious points in the reconstructions themselves are likely going to be harder to deal with: I do not know enough to be able to assess everything myself. I've already noticed though that they don't even seem to use features such as the the *k ~ *ḱ notation consistently; there are e.g. some roots attested only in centum langs, for which only a proto-form with *k is regardless given.
[ˌʔaɪsəˈpʰɻ̊ʷoʊpɪɫ ˈʔæɫkəɦɔɫ]
Re: The Great Proto-Indo-European Thread
I'm working on an a posteriori IE language; does anyone have any particular recommendations on source material, both grammatical and lexical, to work with?
"But if of ships I now should sing, what ship would come to me,
What ship would bear me ever back across so wide a Sea?”
What ship would bear me ever back across so wide a Sea?”
- WeepingElf
- Smeric
- Posts: 1630
- Joined: Wed Mar 08, 2006 5:00 pm
- Location: Braunschweig, Germany
- Contact:
Re: The Great Proto-Indo-European Thread
First stop: Benjamin W. Fortson IV, Indo-European Language and Culture. It's affordable and very well written, IMHO the best book on IE I know of (and BTW, one exercise in the book asks the reader to make and describe an IE conlang!). Wiktionary has a list of PIE roots, Wikipedia has a page on PIE nominals and one on PIE verbs.Zaarin wrote:I'm working on an a posteriori IE language; does anyone have any particular recommendations on source material, both grammatical and lexical, to work with?
...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
Tha cvastam émi cvastam santham amal phelsa. -- Friedrich Schiller
ESTAR-3SG:P human-OBJ only human-OBJ true-OBJ REL-LOC play-3SG:A
Re: The Great Proto-Indo-European Thread
See the last post in L&L for a free online version.WeepingElf wrote:First stop: Benjamin W. Fortson IV, Indo-European Language and Culture
Re: The Great Proto-Indo-European Thread
Thank you!WeepingElf wrote:First stop: Benjamin W. Fortson IV, Indo-European Language and Culture. It's affordable and very well written, IMHO the best book on IE I know of (and BTW, one exercise in the book asks the reader to make and describe an IE conlang!). Wiktionary has a list of PIE roots, Wikipedia has a page on PIE nominals and one on PIE verbs.Zaarin wrote:I'm working on an a posteriori IE language; does anyone have any particular recommendations on source material, both grammatical and lexical, to work with?
Thanks!Vijay wrote:See the last post in L&L for a free online version.WeepingElf wrote:First stop: Benjamin W. Fortson IV, Indo-European Language and Culture
"But if of ships I now should sing, what ship would come to me,
What ship would bear me ever back across so wide a Sea?”
What ship would bear me ever back across so wide a Sea?”
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
It's been a good while since any real speculation here, so I'll try to revive that practice with the suggestion, based largely on Hittite, that thematic verbs originally showed o-grade of the thematic vowel only in the 3pl and participle. Hittite thematic verbs had widespread e-grade of the thematic vowel in Old Hittite, with verbs in -ye- changing it to a-grade, probably due to a regular sound change at the end of the Old Hittite period, though forms with e can still be found, and could be analogical. Verbs in -ske- maintained the original distribution through the entire Hittite period, while the other thematic verbs had a mixed distribution, probably due to minor sound changes and analogy, the details of which are largely unrecoverable. Outside Anatolian, the familiar thematic conjugation can be assumed to have developed once the 1sg *-oh₂ entered the paradigm, ousting original *-emi. From the new ending, o-grade was spread to the 1pl and to the 1sg and 1pl preterite.
Thoughts?
Thoughts?
- 2+3 clusivity
- Avisaru
- Posts: 454
- Joined: Fri Mar 16, 2012 5:34 pm
Re: The Great Proto-Indo-European Thread
Hmm ... I had always assumed the -*o- grade in at least the thematic 1st sg., du., and 3d. was due to being followed by -*m- or -*w- < -**m-. Most of the others, as in the 3d, also seem to be linked to a following -*n- or perhaps more generally -*R-. Not sure what's up with the thematic 1st sg., perhaps analogy off of the middle?
linguoboy wrote:So that's what it looks like when the master satirist is moistened by his own moutarde.
- KathTheDragon
- Smeric
- Posts: 2139
- Joined: Thu Apr 25, 2013 4:48 am
- Location: Brittania
Re: The Great Proto-Indo-European Thread
The thematic 1sg. can be referred to a sound law whereby word-final *e is apocopated after *-VH, and specifically *-eHe# > *-oH - note that this is supported by the thematic dual in *-oh₁, which we'd expect to be *-o/e-h₁e, and the i- and u-stem duals, and the neuter dual, which likewise lack the final *e of the masculine athematic. Thus, the thematic 1sg. can be derived from *-e-h₂e, likewise with *e. As for the origin of the *-h₂e, that's up to your favourite theory - mine is Jasanoff's h₂e-conjugation theory. The other 1st person forms are analogical.
- √mak·a√g-láwar·ē
- Niš
- Posts: 9
- Joined: Thu Jul 28, 2016 3:47 pm
Re: The Great Proto-Indo-European Thread
I’ve been thinking a lot about *h₂ lately and I kind of had this random idea.
I realize like this may simply be a matter of notation, but what if, instead of */h₂/ having a syllabic allophone in *[ə₂], it was the other way around, i.e. *h₂ was really a semivocalic variant of some */a/ phoneme that was realized as *[a̯ ~ ʁ]? This would have been devoiced from *ʁ → *χ in OAnat. ‹ḫ› (a similar change is found in Proto-Eskimo *ɣ *ʁ → Nunivak Yup'ik x χ / _#), and colored adjacent vowels before disappearing in other branches. In a syllabic environment, */a/ would result in Lat. a : Gr. α : Skr. i, etc. This would leave us with the vowel phonemes:
*/i [~ i̯ = j]/ */u [~ u̯ = w]/
*/ē/ */e/ */o/ */ō/
*/a [~ a̯ = ʁ]/
A modern example of a similar phoneme may be, for example, German /r/, which has allophones in [ʁ] and [ɐ ~ ɐ̯], and Danish, where (iirc) /r/ has realizations in [ɐ̜̽ ~ ə̠]. I also believe /r/ in both these languages (or maybe it’s Dutch rather than Danish) also have a [χ] allophone for /r/, though I also believe it’s conditioned by adjacent voiceless consonants (or maybe also word initially in some dialects). Obviously the etymology for this is different, with underlying /r/, but who's to say Pre-PIE didn't have a uvular *ʁ (Proto-Uralic has a */x/ of uncertain articulation).
I don’t know what this would mean for *h₁ and *h₃, however (perhaps voiceless & voiced velar fricatives).
I realize like this may simply be a matter of notation, but what if, instead of */h₂/ having a syllabic allophone in *[ə₂], it was the other way around, i.e. *h₂ was really a semivocalic variant of some */a/ phoneme that was realized as *[a̯ ~ ʁ]? This would have been devoiced from *ʁ → *χ in OAnat. ‹ḫ› (a similar change is found in Proto-Eskimo *ɣ *ʁ → Nunivak Yup'ik x χ / _#), and colored adjacent vowels before disappearing in other branches. In a syllabic environment, */a/ would result in Lat. a : Gr. α : Skr. i, etc. This would leave us with the vowel phonemes:
*/i [~ i̯ = j]/ */u [~ u̯ = w]/
*/ē/ */e/ */o/ */ō/
*/a [~ a̯ = ʁ]/
A modern example of a similar phoneme may be, for example, German /r/, which has allophones in [ʁ] and [ɐ ~ ɐ̯], and Danish, where (iirc) /r/ has realizations in [ɐ̜̽ ~ ə̠]. I also believe /r/ in both these languages (or maybe it’s Dutch rather than Danish) also have a [χ] allophone for /r/, though I also believe it’s conditioned by adjacent voiceless consonants (or maybe also word initially in some dialects). Obviously the etymology for this is different, with underlying /r/, but who's to say Pre-PIE didn't have a uvular *ʁ (Proto-Uralic has a */x/ of uncertain articulation).
I don’t know what this would mean for *h₁ and *h₃, however (perhaps voiceless & voiced velar fricatives).
eálá þeódnes þrym! hú seó þrág gewát,
genáp under nihthelm, swá heó nó wǽre
genáp under nihthelm, swá heó nó wǽre