The Great Proto-Indo-European Thread's Sequel

[Zju]

[...] but then again what complex feature in any language wouldn't be easier to fit in a tidy box if we pretend it's just a remnant of a different, now disappeared feature?

Don't all language features descend from some other, different features though? Even grammaticalisation needs some lexical material to act on. AFAIK only onomatopoeic words can appear ex nihilo.

[WeepingElf]

So low-tone roots would have voiced aspirated stops, high-tone roots voiceless ones? Fair.

[Rounin Ryuuji]

I remember while I was researching something coming across low pitch and voicing going together. I want to say it was Thai that lost voicing but this led to the phonemicisation of a low tone.

[Moose-tache]

[...] but then again what complex feature in any language wouldn't be easier to fit in a tidy box if we pretend it's just a remnant of a different, now disappeared feature?

Don't all language features descend from some other, different features though? Even grammaticalisation needs some lexical material to act on. AFAIK only onomatopoeic words can appear ex nihilo.

Sigh. This is the problem with historical linguistics. Someone walks in the room and says "That's not a spoon. You're trying to eat your soup with a stick." And the people in the room say "Well, I'm sure there's some way you could define a spoon as a type of stick," and go on dipping it in the soup.

Who cares if one thing comes from another thing? The issue is demonstrability. If you speculate that Confusing Feature A is just a reflection of Invisible and Unprovable Past Feature B, your theory becomes unfalsifiable. Your job is not to come up with something that's vaguely plausible in the abstract. That's not science; it's masturbation but where your clit is a keyboard.

[Kuchigakatai]

No speculation allowed, no fun allowed, we're Serious Scientists here.

[keenir]

What regards the idea that the Bell Beaker people spoke a language related to Anatolian, I am still researching the matter. But I have the impression that Hittite is more similar to the western IE languages like Latin or Germanic than to eastern ones like Greek or Sanskrit. Or to put it differently, the western languages have more in common with Hittite than the eastern ones do. Now the reconstruction shows us that the eastern languages are more conservative and the western languages have innovated - separately - which suggests a Hittite-like substratum in western Europe.

When I read that, two thoughts came up in my mind, though I like the third more:
1. When I read about the family tree of Hittite and the rest of the IE languages, it usually goes along these lines: PIE -> Early Indo-European -> Hittite branches off -> Late Indo-European -> the Satem languages split off (if the Anatolian branch is mentioned at all, it is not placed on the timeline) . Your suggestion makes me think that Hittite had a sibling split off alongside it, forming a common Hittite+ branch, and that sibling language wanders off until it (and-or its descendant langauges) arrives in the lands of the western IE languages, whereupon the sibling/descendants form the substrate.

2. This one's unlikely insofar as I can tell. The Hittite language branches off of the IE trunk, and then the western IE languages branch off (picking up some features as they head west, and both they and the Hittite language chips away some features after that (rather than the Hittites chipping away at some of the features they share with other IE langs before the other IE languages form a new branch)

...but as I ordered my thoughts in preparation for making a post, I had a third idea, and I like this the best...though I fear it will seem a mite too "we need a Basque fisherman to make landfall for this to work"...

3. A lot of the distinctive features of Hittite are, I've been led to believe/understand, are Hattian in origin. So perhaps the western IE languages' substrate isn't Hittite, but from relatives of Hattian?

[keenir]

No speculation allowed, no fun allowed, we're Serious Scientists here.

I'm not sure speculation is the problem...more a misunderstanding of "don't threaten me with a good time" if not a ballpark beyond that point.

[Zju]

If you speculate that Confusing Feature A is just a reflection of Invisible and Unprovable Past Feature B, your theory becomes unfalsifiable.

1. It's falsifiable if you find a bunch of roots, morphemes or stems that don't conform to the restriction, so we're good. It's also falsifiable if you discover some past phonetic feature that is incompatible with tones.
2. Presumably, PIE didn't just spawn into existence.
3. Also presumably, it didn't have that root shape restriction for all of eternity. So it had to have developed from something.
4. What other origin hypothesis - any at all - is able to account for the root shape restriction? (nuh-uh, we don't know! is a lack of answer)

[WeepingElf]

[objections against my idea]

3. A lot of the distinctive features of Hittite are, I've been led to believe/understand, are Hattian in origin. So perhaps the western IE languages' substrate isn't Hittite, but from relatives of Hattian?

Maybe I am dead wrong with my idea. It is not impossible that, for instance, the language of the Linearbandkeramik culture (the first Neolithic farmers of Central Europe) was related to Hattic, given that they came from Anatolia. But then, we know so frantically little about Hattic that this does not explain much.

[abahot]

Does the "D^h and T were separate tones" theory relate at all to the lack of roots with D-D?

[WeepingElf]

The idea is that *T and *Dh formed a class at the exclusion of *D, which were a highly marked set subject to a dissimilation rule (there seem to be alternations between *d and *h1, which may result from that). The glottalic theory states that the *D stops were ejectives or implosives, but it may be the case that we are dealing with particularly weak sounds here, perhaps voiced fricatives or approximants (both categories easily change into each other), so *b may just have merged with *w. Proto-Uralic, which may be related to PIE, has just one set of stops and two sounds that were traditionally reconstructed as */ð/ and */ɣ/, and these values, while Uralicists are no longer sure about them, are still at least plausible, and may have formed a set with the firmly reconstructed */w/ and */j/. And as I said before, *T may have been aspirated in Early PIE, though this aspiration would have been lost at least in most dialects of Late PIE (though it may have been retained - or restored - in the dialects ancestral to Germanic and Armenian). Alas, all this is a swamp of speculation with very little evidence to cling to

[Moose-tache]

nuh-uh, we don't know! is a lack of answer

Yes. That is the point; that's the null hypothesis! This isn't a race, where whichever theory is least stupid wins the title of "correct." It is entirely acceptable to say that we don't know that PIE had or didn't have a tonal ancestor if an affirmative answer has not been demonstrated. This is not equivalent to saying that PIE "spawned into existence." Jesus, it's like you're all showing up to the first day of science and asking questions that are answered in the syllabus.

[Zju]

It is entirely acceptable to say that we don't know that PIE had or didn't have a tonal ancestor if an affirmative answer has not been demonstrated.

I.e. the tone hypothesis could be false or it could be true. Out of which only the latter accounts for the root shape restriction.

Honestly, I don't get this fervent denial. Is it because this particular hypothesis seems improbable to you, or because the utmost scientific rigour is not being applied at a hobbyist forum, outside of academia? Either way, it comes off as a bit disingenuous to deny it vehemently just because it could be false (or because it has p-value greater than 0.05 or what have you)

Anyway, going back on topic

Does the "Dh and T were separate tones" theory relate at all to the lack of roots with D-D?

Most likely not. What with D having been the most marked series, it was least likely to be affected by allophony(~para allophonic influences). As WeepingElf pointed out, two D-series consonants in the same root likely dissimilated, presumably because of the greater articulatory effort associated with that.

[WeepingElf]

It is entirely acceptable to say that we don't know that PIE had or didn't have a tonal ancestor if an affirmative answer has not been demonstrated.

I.e. the tone hypothesis could be false or it could be true. Out of which only the latter accounts for the root shape restriction.

But one must take other possible explanations into account. For instance, it seems as if before ablaut emerged, most Pre-PIE (Indo-Uralic?) vowels merged into one, and the distinction between voiceless and voiced aspirated stops may reflect some vowel quality difference before this "Great Vowel Collapse", as I call it. This hypothesis also has the "advantage" that it does not invoke a feature the most likely closest relative of PIE probably did not have, either. (Of course, a tone distinction may have emerged from a vowel quality distinction that was lost in the Great Vowel Collapse, or it may have been contributed by the hypothetical (Trans)Caucasian substratum that may have been responsible for the typological difference between IE on one hand and Uralic and all the other "Mitian" languages on the other - and that substratum may have been anything.

Honestly, I don't get this fervent denial. Is it because this particular hypothesis seems improbable to you, or because the utmost scientific rigour is not being applied at a hobbyist forum, outside of academia? Either way, it comes off as a bit disingenuous to deny it vehemently just because it could be false (or because it has p-value greater than 0.05 or what have you)

Nor do I. This is not an academic conference. This is a forum of language hobbyists who have fun speculating about things we don't know yet. As long as we respect the scholarly consensus and use proper method, and most of all, are aware that we don't know whether our ideas hold or not, everything is fine. And for me at least, the main "purpose" of all these speculations is to build conlangs upon them.

Anyway, going back on topic

Does the "Dh and T were separate tones" theory relate at all to the lack of roots with D-D?

Most likely not. What with D having been the most marked series, it was least likely to be affected by allophony(~para allophonic influences). As WeepingElf pointed out, two D-series consonants in the same root likely dissimilated, presumably because of the greater articulatory effort associated with that.

Perhaps. There are languages which don't allow two ejectives in a root and lack a labial ejective, such as Akkadian. There are, however, others for which neither holds, such as Georgian (there is, for instance, a word k'op'e 'ladle' - two ejectives in the root, one of them labial). Again, ejectives are untypical for a Mitian language, but may have been contributed by the unknown Caucasian substratum.

[Zju]

For instance, it seems as if before ablaut emerged, most Pre-PIE (Indo-Uralic?) vowels merged into one, and the distinction between voiceless and voiced aspirated stops may reflect some vowel quality difference before this "Great Vowel Collapse", as I call it.

Interesting. Do you have vowel breathiness in mind, or something else?
I entertained the hypothesis of consonantal harmony for a while myself, but then why would *D series not participate in it?

[WeepingElf]

For instance, it seems as if before ablaut emerged, most Pre-PIE (Indo-Uralic?) vowels merged into one, and the distinction between voiceless and voiced aspirated stops may reflect some vowel quality difference before this "Great Vowel Collapse", as I call it.

Interesting. Do you have vowel breathiness in mind, or something else?
I entertained the hypothesis of consonantal harmony for a while myself, but then why would *D series not participate in it?

Fact is that I have no opinion what kind of feature may have caused this. There are so many possibilities. Tone is one, and perhaps the best as voicing is associated with low tone in many languages. Kortlandt suggested a kind of consonant gradation. One idea of mine is that vowel height before the Great Vowel Collapse translated into aspirated stop voicing (low vowels giving voice), perhaps via an intermediate stage with tones. But tones are untypical for Mitian languages such as Uralic or Turkic, and I feel reluctant to introduce a mechanism that appears and disappears again.

Why the *D set does not participate in the mechanism is of course a question that cannot really be answered without knowing which mechanism to look for. Apparently they were at that point of a kind to which it simply did not apply. After all, *s and sonorants are not affected by it either; perhaps the *D set weren't stops at that time, as in my notion that they may have been voiced fricatives or approximants like those traditionally reconstructed for Proto-Uralic. But of course I don't know, and am not at all sure about this.

[WeepingElf]

That said, I prefer explanations that draw on distinctions found in Uralic and other Mitian languages (such as an earlier richer vowel inventory) over ones that invoke distinctions alien to them and also not continued in the attested IE languages (such as tones or ejectives). But as I said, we have no idea what the Transcaucasians who contributed to the Yamnaya people (who seem, genetically, to be a mix of a population similar to who probably were the Proto-Uralic speakers and a second group of Transcausian origin) spoke, so almost anything is possible.

[Kuchigakatai]

Coincidental tweet by Marijn van Putten:

https://twitter.com/PhDniX/status/1622662636948054017

Plot twist: Marijn is actually reading us?

[abahot]

I prefer explanations that draw on distinctions found in Uralic and other Mitian languages

Isn't that somewhat circular reasoning though? Because this makes internal reconstruction of pre-PIE look more like other "Mitian" languages and validate the hypothesis even more.

[Zju]

Besides, even if Mitian is firmly established to have existed 10 000 years ago, we still don't know much about typology in the region back then. Prosodic features are gained and lost even more readily than average, too.