The Great Proto-Indo-European Thread

[Zaarin]

Gitxsan has a dorsal plosive inventory of /c kʷ q cʼ kʷʼ qʼ/ with /c cʼ/ being realised as [k kʼ] before /s l/. It's parent language, Proto-Tsimshian, has a dorsal plosive inventory of /k kʷ q qʷ kʼ kʷʼ qʼ qʷʼ/, and while no sound changes were listed, it's rather obvious where palatals came from.
All this might be a bigger or smaller parallel of the development of dorsal plosives in PIE, showing us how a language can have /c/ but not /k/.

Lacking /k/ due to palatalization is an areal feature of the PNW and can be seen in a number of languages of the region.

[KathTheDragon]

I am currently entertaining the notion that "name" can in fact be reconstructed as *h₃nóh₁mn̩ ~ *h₃néh₁mn- to do away with the pesky *ē that's otherwise needed to explain Tocharian A ñom, B ñem. It's possible to refer the Greek back to the o-grade, with a very similar history to Germanic, with several rounds of levelling, and Osthoff's law in the middle. So far it's rather promising.

[Soap]

That's a lot of consonants. Did PIE really have CCVCC roots? I have seen some question able reconstructions, like dhhsnos, that make me think there must have been missing vowels.

[Zaarin]

That's a lot of consonants. Did PIE really have CCVCC roots? I have seen some question able reconstructions, like dhhsnos, that make me think there must have been missing vowels.

If I'm not mistaken, *h₃nóh₁mn̩ would be better analyzed as HRVC.CV, which is canonical.

[2+3 clusivity]

That's a lot of consonants. Did PIE really have CCVCC roots?

Shhhhh..... the *number six might hear you. If it gets too disturbed, it's liable to add *k-.

[hwhatting]

That's a lot of consonants. Did PIE really have CCVCC roots? I have seen some question able reconstructions, like dhhsnos, that make me think there must have been missing vowels.

The root would be *HneH-, *-men- is a suffix.

[KathTheDragon]

That's a lot of consonants. Did PIE really have CCVCC roots?

Shhhhh..... the *number six might hear you. If it gets too disturbed, it's liable to add *k-.

Actually, there was no initial *k, see this article

[Zaarin]

Thematic 1p/2p verbs are identical in the present and aorist/imperfect, both active and passive--how did real IE languages go about differentiating them?

[hwhatting]

Thematic 1p/2p verbs are identical in the present and aorist/imperfect, both active and passive

That depends on which reconstruction model you adhere to; there are people who assume e.g. for the active present *-mes, *-tes vs. non-present *-me, *-te.

--how did real IE languages go about differentiating them?

By
1) Having different endings for present and past (e.g. Sanscrit, where part of the distinctions may be inherited, or Hittite, which adds -i to the present active endings also in the 1/2pl, which is most probably an innovation)
2) Greek and Indo-Aryan by introducing the augment
3) Dropping the IE imperfect and creating new imperfects with a stem different from the present (e.g. Latin pres. leg-i-mus vs. imperf. leg-eba-mus)

[Zaarin]

Thematic 1p/2p verbs are identical in the present and aorist/imperfect, both active and passive

That depends on which reconstruction model you adhere to; there are people who assume e.g. for the active present *-mes, *-tes vs. non-present *-me, *-te.

--how did real IE languages go about differentiating them?

By
1) Having different endings for present and past (e.g. Sanscrit, where part of the distinctions may be inherited, or Hittite, which adds -i to the present active endings also in the 1/2pl, which is most probably an innovation)
2) Greek and Indo-Aryan by introducing the augment
3) Dropping the IE imperfect and creating new imperfects with a stem different from the present (e.g. Latin pres. leg-i-mus vs. imperf. leg-eba-mus)

That's very helpful. Thanks.

[dhok]

A minor question for the ongoing development of Old Moesian. In root athematics in Sanskrit, at least, there seems to be a distinction between verbs where the 3rd plural present ending derives from vowel-less *-nti and those where the ending comes from *-onti. For the former, we have Skt. hánti 'he slays' (PIE *gʷʰén-ti) vs. ghnáti 'they slay' (*gʷʰnn̥ti); for the latter, Skt. éti 'he goes' (*h₁éyti) and yánti 'they go' (*yónti). Fortson's chapter on Anatolian suggests that all Hittite verbs have a 3rd plural present ending -anzi, but what circumstantial evidence is available to me before I trot off to bed suggests that *n̥ became /an/ in Hittite. I'm too lazy to check Malzahn's magnificent monograph on Tocharian--but where's the Sanskrit split come from, and which was original?

[hwhatting]

A minor question for the ongoing development of Old Moesian. In root athematics in Sanskrit, at least, there seems to be a distinction between verbs where the 3rd plural present ending derives from vowel-less *-nti and those where the ending comes from *-onti. For the former, we have Skt. hánti 'he slays' (PIE *gʷʰén-ti) vs. ghnáti 'they slay' (*gʷʰnn̥ti); for the latter, Skt. éti 'he goes' (*h₁éyti) and yánti 'they go' (*yónti). Fortson's chapter on Anatolian suggests that all Hittite verbs have a 3rd plural present ending -anzi, but what circumstantial evidence is available to me before I trot off to bed suggests that *n̥ became /an/ in Hittite. I'm too lazy to check Malzahn's magnificent monograph on Tocharian--but where's the Sanskrit split come from, and which was original?

For the full grade form, what I've normally seen reconstructed is actually *-ént(i), not *-ont(i) (see German sind, Greek eisi <*H1senti); forms with /o/ (e.g. latin sunt) are then often explained as influenced by the thematic verbs.
According to Kloekhorst's Anatolian Etymologoical Dictionary, Par. 1.4.9.1., PIE *enT (T= any dental consonant) > Hittite anT; also, according to him, syllabic /n./ stays syllabic in Hittite, but is written <an> (Par. 1.4.7.2a)*1), so PIE *enT, *onT and *n.T become indistinguishable in writing. FWIW, he reconstructs the ending as PIE *-ent(i).
I have no idea how old the split is, and with so many different reconstructions of the PIE verbal system being bandied about, I'm not sure whether someone can ultimately answer your question, expect by presenting their pet reconstruction as the only truth.

*1) I really need to finally read the Historical Phonology part of his dictionary end-to-end, to understand his reasoning; up to now I've mostly used the dictionary to look up separate words and developments.

[KathTheDragon]

Hittite indeed offers no evidence for the original ablaut grade of the 3pl, and neither does Tocharian, so far as I can tell. The athematic ending A -iñc/-i, B -ṃ would seem to point to zero-grade *-n̥ti, but could equally represent full-grade *-enti with palatalisation analogically removed after the rest of the paradigm. I don't know of any reflexes of *-n̥ti in a root present in the "inner" IE languages besides the Sanskrit reflexes, so it's quite possible that we (somehow) have a unqiuely Sanskrit innovation.

[Chengjiang]

So I was thinking about one piece of evidence against the labiovelars being interpreted as labialized uvulars: IIRC most of the instances of apparent /a/ that remain after accounting for the a-coloring laryngeal are in the vicinity of the plain velars, suggesting retraction of /e/ in the vicinity of a uvular consonant (just as some interpret h2). The labiovelars show no such tendency to co-occur with /a/ or even /o/, which one would expect if they were also uvular. And there are certainly confirmed natlangs with labialized velars but only plain uvulars.

[dhok]

A little bird scanned and put on Libgen a study by Jörunder Hilmarsson on the Tocharian dual, as a scan didn't previously seem to exist. It may be of interest.

[KathTheDragon]

Hahahaha my ISP won't let me view the site

[jal]

Mine doesn't, but the page is empty ("ERR_EMPTY_RESPONSE").


JAL

[WeepingElf]

Just a new little thought about the laryngeals.

As I have said before, I entertain the notion that the PIE laryngeals were, in Early PIE at least, just the fricative members of the three velar series. One thing bugged me, though: while the plain velars (as they are conventionally named) are the least common of the velar stop series, *h2 is the most common of the laryngeals. How did that happen? The solution may be simple: there may have been yet another consonant, perhaps something like */ħ/, which merged with the plain velar fricative. It merged with this one because it was closer than the front velar *h1, and was not labialized like *h3. Perhaps this merger even explains the "*h4 riddle", the phenomenon that there seem to be instances of *h2 that delete in Hittite, which normally preserves *h2. The merger may have been post-Anatolian.

And perhaps there were more tokens of *h1 (which one would expect to be the most common under the velar series theory) than we currently reconstruct. After all, this laryngeal doesn't colour vowels, is not preserved in Hittite, and thus may have vanished unnoticed in a number of forms.

[KathTheDragon]

On the other hand, the identification could simply be wrong. I believe it was discussed earlier in this thread that many instances of *h₂w might actually have once been a unitary labio-velar fricative, retained as such in Anatolian *ḫʷ. *h₃ as a labio-velar fricative is already unlikely, since it yields *ḫ in Anatolian, not *ḫʷ. Additionally, there are reasons to believe that *a *o were phonetically more like /ɐ ɑ/ which could be taken as evidence that *h₂ was velar and *h₃ was uvular, matching up with the "pre-velar" and "post-velar" stops, which were probably velar and uvular respectively themselves. *h₁ could then reflect *f, completing the fricative series.

[Zju]

I doubt *h₁ being a reflex of earlier **f, as all laryngeals have been linked with PU velars. As for the vowels, I subsribe to Kümmel's view that /e o/ were indeed [ɛ ɔ] in late PIE, but come from early PIE /a a:/ and further I think that late PIE /a/ simply comes from early PIE /a/ that never shifted to /ɛ/ in the first place due to being next to *h₂. Late PIE /a/ when not next to the second laryngeal most likely comes from loanwords.

[KathTheDragon]

Yes, *h₁ < **f is just wild speculation. Then again, the Uralic comparison is difficult, so it's methodologically questionable to take any conclusions drawn from it as fact. Additionally, there's evidence for *h₁ being a glottal stop and *h, so maybe there were in fact two such laryngeals. Then there's no obstacle to deriving *h < **f, and keeping the comparison to Uralic for *ʔ. Obviously, however, that's even wilder speculation. Re: the vowels, that's basically my view too - but we do have to differentiate ablauting *o from *o < *h₃e. I was talking about the latter.

[WeepingElf]

Fair. This paper, however, suggests that *h1 corresponds to Uralic front vowels, *h2 to low vowels, and *h3 to rounded vowels. Hyllested may of course be wrong, though.

[KathTheDragon]

Yes, I've seen that paper before. While suggestive, such long-distance comparison is hardly conclusive, and you'll also note such exceptions as *kewe "female animal" = *h₂₃ówi- "sheep" which doesn't fit the pattern whichever laryngeal you reconstruct. Also of detriment is that at least one of his PIE roots doesn't even exist! (It's *h₃pe(w)s-, see here) Frankly, I'm uncertain whether our knowledge of Uralic is even sufficient for lexical comparison - the reconstructed Proto-Uralic lexicon is so much smaller than PIE's, after all, and there's plenty of uncertainty on both sides.

[WeepingElf]

I understand. Comparing IE and Uralic is a very uncertain business, and it may not be particularly useful for reconstructing the past of PIE. Maybe the resemblances mentioned in Hyllested's paper are merely old loanwords (the conventional explanation for the lexical resemblances between IE and Uralic). But I still haven't really understood your scepticism against PIE *o being rounded, at least in Late PIE.

[KathTheDragon]

I actually believe that *o was unrounded in Early PIE, before Anatolian split off. There's no good evidence for, and plenty against, old rounding in Anatolian (every branch presenting a default unrounded reflex, with any rounding being conditioned). Tocharian seems to present a similar story, as do a number of the "inner" IE branches. Only a few (Greek, Latin, and so on) actually present a default rounded reflex, which happen to be the same ones that were assumed to be archaic with respect to vocalism (you must know of the characterisation of the early reconstructions of PIE as being basically Sanskrit consonants with Greek vowels), It is of course possible that a vowel like *ɑ or *ʌ could have spontaneously rounded some time between Early PIE and those languages where rounding is required, so it makes more sense to me that *o should be an unrounded vowel. (Of course, if this is in reference to *h₃, I view this as merely support of the main argument against rounding there, which is the utter lack of any labialisation in Anatolian)