The Great Proto-Indo-European Thread's Sequel

[Pabappa]

You could suppose that the voiceless stops were also allophonically aspirated....I haven't looked at it lately but some of the "ex-glottalicists" seem to be moving towards a setup with suprasegemtnal aspiration, once per word, with a voiceless stop taking priority, then bh, then voiced stops.

[KathTheDragon]

It seems incredibly implausible that the traditional *T series was in fact the most marked, never mind that the set of sound changes required to get from *Tː *ʔT *T to *T D Dʰ (which is honestly still the best diachronic fit for the bulk of the "traditional" part of the family) seem to be completely unattested.

Looking at LIV, -DH seems pretty rare, though -HD isn't uncommon.

Regarding glottalisation, I'm unconvinced by all the arguments. Winter's law and Lachmann's law can both be ascribed to lengthening before voiced stops, a trivial enough development. I'm not sure what sound change in Anatolian you mean - I presume Kloekhorst's claim that *e is lengthened before original *D? I don't really buy his claim. Additionally, Tocharian didn't have "sporadic loss of word-initial *d", it had a loss of *d only in certain preconsonantal contexts. The development of *doru to PT *ọr is due to analogy: nom.sg. *doru ~ gen.sg. *drews > *doru ~ *rews -> *oru > *ọr.

[Znex]

Regarding glottalisation, I'm unconvinced by all the arguments. Winter's law and Lachmann's law can both be ascribed to lengthening before voiced stops, a trivial enough development. I'm not sure what sound change in Anatolian you mean - I presume Kloekhorst's claim that *e is lengthened before original *D? I don't really buy his claim.

Shouldn't Winter's and Lachmann's Laws also happen before *Dʰ in that case? Why would post-aspiration affect the change?

It seems incredibly implausible that the traditional *T series was in fact the most marked, never mind that the set of sound changes required to get from *Tː *ʔT *T to *T D Dʰ (which is honestly still the best diachronic fit for the bulk of the "traditional" part of the family) seem to be completely unattested.

I think Kloekhorst's emphasis (see here: pp228-243) on the other hand was on unconditioned stop lengthening being near-unattested except in the case of loanwords. His proposed pull chain shift has Tː > (ʔ)T > (ʔ)D.

[mèþru]

I'm very interested in what "ex-glottalicists" have to say, because while I don't subscribe to the glottalic theory it brings up many interesting points.

[KathTheDragon]

I think Kloekhorst's emphasis (see here: pp228-243) on the other hand was on unconditioned stop lengthening being near-unattested except in the case of loanwords. His proposed pull chain shift has Tː > (ʔ)T > (ʔ)D.

But it is attested, so his argument against it is void.

Regarding glottalisation, I'm unconvinced by all the arguments. Winter's law and Lachmann's law can both be ascribed to lengthening before voiced stops, a trivial enough development. I'm not sure what sound change in Anatolian you mean - I presume Kloekhorst's claim that *e is lengthened before original *D? I don't really buy his claim.

Shouldn't Winter's and Lachmann's Laws also happen before *Dʰ in that case? Why would post-aspiration affect the change?

The idea is that first *Dʰ > *Đ, whence voiceless fricatives by voicing assimilation (prior to Lachmann's law). Since the fricatives can still be purely allophonic here, while the stops are not (merger with *T), only the stops would need restoring and trigger Lachmann's law. See this paper for details.

[WeepingElf]

Ah, so we are at discussing the PIE stop grades and the glottalic theory once again. It is true that the T/D/Dh system we see in PIE-as-usually-reconstructed is vanishingly rare among the languages we know, but there are many "one-of-the-kind" systems among the world's languages, so this is a rather weak argument. That the system was unstable is abundantly evident from the fact that no branch preserved it, with the sole exception of Indo-Aryan, which stabilized it by innovating a Th grade (mostly from *T + laryngeal), thus establishing a neat 2x2 system in which the features [voice] and [breath] are orthogonal.

Yet, Indo-Europeanists are not reconstructing it without reason. It is the most convenient way of accounting for the systems of the daughter languages. And Indo-Aryan is not the only branch to keep the *Dh grade intact: it is IMHO highly likely that the Old Armenian "voiced stops" actually were breathy-voiced, as they are in some Eastern Armenian dialects, and this makes sense of the Western Armenian shift where the "voiced" stops merged with the aspirated stops. In Greek, the *Dh grade simply was devoiced. Italic and Germanic point at spirantization *Dh > *Ð, though the evidence is somewhat ambiguous.

But as for Early PIE (the common ancestor of Anatolian and "Classical" PIE), it's anyone's guess. Anatolian is not much of a help as we simply don't know how the stops were phonated there. The traditional interpretation is that the Old Anatolian languages (Hittite, Luwian and Palaic) had voiceless stops corresponding to *T, and voiced stops corresponding to *D and *Dh - but the Hittites did not spell them that way even though the cuneiform syllabary would have given them the possibility! Hence, other ideas have been proposed, such as aspirated vs. unaspirated or geminate vs. simple. The Young Anatolian languages (Lycian and Lydian) do point at voiceless vs. voiced, but almost anything may have happened in the centuries in between. Shifts from Th/T to T/D or vice versa are common enough; there is no shortage of languages where dialects differ in this point.

At any rate, the "traditional" glottalic theory (i.e., *T(h) T' D(h), as it was proposed by Gamkrelidze and Ivanov, and, independently, by Hopper; as opposed to newer versions such as Beekes's *T T' Th) makes sense for an early stage of PIE, but by Late PIE, it probably had shifted to the traditional *T D Dh system. It doesn't lose its power to explain the root structure constraints and the token frequencies of the three stop grades when we relegate it to some sort of Early PIE.

[mèþru]

I hate the way people use Early and Late. If Hittite is not Proto-Indo-European and is an IE language, then Late PIE cannot be after Hittite splits from PIE. Also, I don't think that Anatolian necessarily split from the rest of IE: it could be part of a larger clade including some other known branches (probably Indo-Iranian at least)

[Kuchigakatai]

I hate the way people use Early and Late. If Hittite is not Proto-Indo-European and is an IE language, then Late PIE cannot be after Hittite splits from PIE. Also, I don't think that Anatolian necessarily split from the rest of IE: it could be part of a larger clade including some other known branches (probably Indo-Iranian at least)

I agree we'd be better served by terms like "PIE" (tout court, for Early PIE) and "Core IE" (for Late PIE), but a convention is a convention.

[mèþru]

So if one subscribes to the view that Anatolian belongs to a larger clade including Indo-Iranian, what would the terms Early and Late even mean in that context?

[Pabappa]

I'm very interested in what "ex-glottalicists" have to say, because while I don't subscribe to the glottalic theory it brings up many interesting points.

Thanks. I dont know that much about ex-glottalicists in general, because theyve all gone different ways ... my own theory starts with the setup at https://en.wikipedia.org/wiki/Glottalic ... c_proposal and removes the glottals entirely, leaving just two series of stops, with aspiration being a suprasegmental feature. I know I didnt make that up, but I dont remember whose idea it originally was. My own contribution is that the lack of *b can be explained by assuming that /b/, unlike the other voiced stops, always attracted the suprasemgental aspiration regardless of its position in the word. I dont believe in uvualrs or palatovelars either, so my PIE stop inventory is just /p t k kʷ b d g gʷ/. Back in the old thread somewhere around page 15 there is a discussion where i said more about it but i dont really want to get into it again.

[mèþru]

okay I still think there were three series plus uvulars, I was thinking about its relevance for Pre-PIE/Early PIE/whatever you call the first stage before the stage before any known branches split from PIE

[Salmoneus]

So if one subscribes to the view that Anatolian belongs to a larger clade including Indo-Iranian, what would the terms Early and Late even mean in that context?

That question doesn't arise, because obviously Anatolian doesn't belong to a larger clade including Indo-Iranian.

The linguistic implausibility of that idea aside, we know for a fact that the Hittites were already around in the 19th century, by which time the Indo-Iranians had barely made it out of Poland and into Kazakhstan.

[mèþru]

Does Grassman's law operate in Indo-Iranian as a whole, or only Indo-Aryan?

Besides Grassman's law, is there any other linguistic reason to suppose Graeco-Aryan? (Yes, I know that it's a terrible argument)

[KathTheDragon]

Grassman's law can't be projected to a putative Proto-Graeco-Aryan anyway, since in Greek it postdates the devoicing of the aspirates.

[Howl]

Besides Grassman's law, is there any other linguistic reason to suppose Graeco-Aryan? (Yes, I know that it's a terrible argument)

The augment and basically the whole verbal system. Also aspirated reflexes of the Dʰ stop series (which is what triggered Grassman's law).

[Xwtek]

Do you know any origin on PIE case system? Why it is still fusional even on the proto language?

[anxi]

Besides Grassman's law, is there any other linguistic reason to suppose Graeco-Aryan? (Yes, I know that it's a terrible argument)

The augment and basically the whole verbal system. Also aspirated reflexes of the Dʰ stop series (which is what triggered Grassman's law).

But in Greek the deaspirated *Dʰ series merges with *T, while in Sanskrit it merges with *D, right?

[mèþru]

since in Greek it postdates the devoicing of the aspirates.

I knew that, it's just that I knew some of those who raise this objection still believe in Graeco-Aryan and I was wondering what reasons they had.

It's known that the Proto-Hellenes (+ maybe Armenians and Albanians?) also left the Urheimat early, possibly earlier than Anatolian. I've never seen a suggestion for making Greek (+ others?) a separate branch from the rest of Indo-European. Therefore I don't see what makes Anatolian being part of an Indo-Iranian clade so outlandish. I remember reading at some point that Indo-Aryan and Anatolian share a higher than average amount of inherited vocabulary, although I forget where I read it and the reliability of that source.

[Howl]

Besides Grassman's law, is there any other linguistic reason to suppose Graeco-Aryan? (Yes, I know that it's a terrible argument)

The augment and basically the whole verbal system. Also aspirated reflexes of the Dʰ stop series (which is what triggered Grassman's law).

But in Greek the deaspirated *Dʰ series merges with *T, while in Sanskrit it merges with *D, right?

Yes, Grassman's law is different for Greek and Aryan. That had also been stated above by Kath.

The remarkable thing here is that the *Dʰ series was supposed to be aspirated in IE, but something like Grassman's law never developed in IE. And yet both branches that do have aspirated reflexes of the *Dʰ stop series developed their own versions of Grassman's law. This suggests that the *Dʰ stop series was not aspirated at all in Indo-European. And if that is true, then the aspirated reflexes of *Dʰ is an innovation that Greek and Indo-Aryan share.

[WeepingElf]

The remarkable thing here is that the *Dʰ series was supposed to be aspirated in IE, but something like Grassman's law never developed in IE. And yet both branches that do have aspirated reflexes of the *Dʰ stop series developed their own versions of Grassman's law. This suggests that the *Dʰ stop series was not aspirated at all in Indo-European. And if that is true, then the aspirated reflexes of *Dʰ is an innovation that Greek and Indo-Aryan share.

This does not necessarily follow. There are plenty of languages which have aspirated stops but nothing like Grassmann's Law. Just because there is this kind of stops in a language, that doesn't mean that they must dissimilate that way. So the absence of this kind of dissimilation doesn't prove that the *Dh grade wasn't aspirated.

But Grassmann's Law indeed happened independently in Indo-Aryan (not in Iranian, where *Dh simply merged with *D - one could speculate, though, that it had happened before that merger happened, but that would be idle speculation as there is no way knowing) and Greek (where it happened after the *Dh grade lost its voicing).