The Great Proto-Indo-European Thread

[√mak·a√g-láwar·ē]

Does anyone know of a single source that lists all or most of the PIE athematic nouns where accent shift & ablaut occur (probably excluding all the proterokinetic deverbals... that would be a very long list)? The Wiktionary pages on PIE athematic nouns are less than exhaustive.

[√mak·a√g-láwar·ē]

Also, this was a whole few pages back, so I apologize for bringing this back up...

The latter is very rare though. LIV has 9 examples, most of them not very good-looking. Just two of them seem to be found in three branches:
– *gwel- 'to swallow', in Armenian + Latin + Baltic. Only found as the zero grade *gul-. Traditionally *CVRR roots are not considered to have been possible. However, *u seems to act here simply as a vowel, just as it does in nominal roots (*muHs 'mouse', etc.)

Also in *bʰuH- "become" - there's no need to reconstruct *bʰweh₂ or *bʰewH- on the strength of Skt bhavati, cf. this paper.

In the linked article, I can't find if Jasanoff addresses the reflexes OE béo(m), OS bīum, etc. < Proto-Germanic *biumi. I'll reread it again just to double check, but scanning it the first couple times, I didn't see these reflexes brought up at all. Wouldn't these forms be derived from an e-grade *bʰewh₂-? It doesn't solve the problem of the theme vowel in Skt bhavati, but seems like pretty clear evidence of an e-grade.

[KathTheDragon]

Also, this was a whole few pages back, so I apologize for bringing this back up...

The latter is very rare though. LIV has 9 examples, most of them not very good-looking. Just two of them seem to be found in three branches:
– *gwel- 'to swallow', in Armenian + Latin + Baltic. Only found as the zero grade *gul-. Traditionally *CVRR roots are not considered to have been possible. However, *u seems to act here simply as a vowel, just as it does in nominal roots (*muHs 'mouse', etc.)

Also in *bʰuH- "become" - there's no need to reconstruct *bʰweh₂ or *bʰewH- on the strength of Skt bhavati, cf. this paper.

In the linked article, I can't find if Jasanoff addresses the reflexes OE béo(m), OS bīum, etc. < Proto-Germanic *biumi. I'll reread it again just to double check, but scanning it the first couple times, I didn't see these reflexes brought up at all. Wouldn't these forms be derived from an e-grade *bʰewh₂-? It doesn't solve the problem of the theme vowel in Skt bhavati, but seems like pretty clear evidence of an e-grade.

I don't recall if he does discuss it, but I imagine the solution would be identical to that of bhavati - namely, the perfect stem *bʰebʰuH- develops to *bʰew- as outlined in the paper, and a full present paradigm is built on it.

The problem with positing an e-grade form for PIE is that the putative e-grade is only attested in a few branches, and the forms don't even correspond. On the other hand, the zero-grade is attested in virtually every branch that shows the root, and widely in forms where an e-grade is expected. Positing a root with an *e throws up immense problems that I have never seen actually considered.

[Zju]

Do the Baltoslavic long adjectival endings come from the PIE anaphorical pronouns *h₁éy *íh₂ *íd, or from the PIE relative pronouns *yós *yéh₂ *yód?

[hwhatting]

They come from the relative pronouns.

[Zju]

There's a hypothesis how para-proto-Basque languages were spoken in all of Europe west of the Alps, described in a paper here. Tl;dr the main argument is that they served as substratum languages to proto-Germanic, proto-Italic and proto-Celtic and several typological features are explained as that substratum's influence (fixation of stress on initial syllable, two copulas, lack of /p/). It looks reasonable, but also kind of far-fetched. What do you think of it?

[KathTheDragon]

We know that Old European languages would have served as substrata for the European IE langs, and it's a reasonable guess that Basque is part of one of the Old European stocks. It might be a stretch to suppose that the stock Basque belonged to extended as far as they claim, but given how little we know about the Old European languages, it's hard to rule out.

[Soap]

ive edited my post to include info from the paper now that Ive read it, so my words might seem somewhat disjointed here:

I think word initial stress is more likely Finnic influence than Vasconic. Vennemann seems to write as if that is the mainstream theory already and he is offering counterevidence, which he might be right on since apptly initial accent was also found in early Italic and Celtic. However his theory would work just as well without including Germanic, because of the likely Finnic substrate in the north.

Lack of /p/ is probably a false lead, as well. Proto-Germanic had /p/ where PIE had had */b/, and sometimes where it had had /p/. Italic has always had plenty of /p/, and, if anything, mentioning Italic further weakens the argument. Only Proto-Celtic ever truly lacked /p/, and that was likely due to a sound change of /p/ > /ϕ/ > /h/ > /0/. This is unlikely to be from Vasconic unless Vasconic langs with /f/ or /ϕ/ are known. Vennemann acknowledges this as well, but his counterargument seems weak. However, i give him credit in that he seems to apply the disappearing /p/ only to Celtic, not Gmc or Italic.

[hwhatting]

initial accent was also found in early Italic and Celtic.

For Proto-Celtic, that has been often posited, but never been proven. The only Celtic branch where we really know that it existed is Goidelic.

[Salmoneus]

We know that Old European languages would have served as substrata for the European IE langs, and it's a reasonable guess that Basque is part of one of the Old European stocks. It might be a stretch to suppose that the stock Basque belonged to extended as far as they claim, but given how little we know about the Old European languages, it's hard to rule out.

Given that the pattern of Neolithic irruption into Europe - near-total replacement of the native population, spreading from one source by only a few routes, cultural homogeneity, and a subsequent catastrophic population collapse - it's probably the case that most languages outside of the Balkans were part of massive linguistic stocks, or even a single pan-European language family.
Then again, Basque could conceivably be a survival from the Mesolithic languages, in which case it might not have anything to do with any of the surrounding languages...

[WeepingElf]

We know that Old European languages would have served as substrata for the European IE langs, and it's a reasonable guess that Basque is part of one of the Old European stocks. It might be a stretch to suppose that the stock Basque belonged to extended as far as they claim, but given how little we know about the Old European languages, it's hard to rule out.

Given that the pattern of Neolithic irruption into Europe - near-total replacement of the native population, spreading from one source by only a few routes, cultural homogeneity, and a subsequent catastrophic population collapse - it's probably the case that most languages outside of the Balkans were part of massive linguistic stocks, or even a single pan-European language family.
Then again, Basque could conceivably be a survival from the Mesolithic languages, in which case it might not have anything to do with any of the surrounding languages...

Fair. The Old European Hydronymy at least seems to point at a single large family spoken in most of Central and Western Europe before the spread of IE. As such homogeneity is unlikely to have existed in pre-Neolithic Europe, this probably was the Neolithic language family you are speaking of (I call it "Aquan", from the putative word for 'river' in that stock). Judging from the roots and suffixes those river names are made of, that family may have been related to IE, but branching off before the development of ablaut. But Basque is IMHO indeed a survival from the linguistic crazy quilt of Mesolithic Europe. Yet, Theo Vennemann assumes that Basque is a residue of the family of the Old European Hydronymy, which he calls "Vasconic" and doesn't consider related to IE. But his etymologies seem very contrived to me, much more so than Krahe's IE etymologies.

As for Vennemann's paper posted by Zju: none of the three points seem to be sufficiently persuasive of a Vasconic substratum in Celtic. The whole "Celtic from the West" theory this paper serves to buttress is IMHO buttocks, anyway.

[KathTheDragon]

I don't know if someone's come up with this before, but after looking through my etymological dictionaries, I had an idea for how to explain the apparently inexplicable connection of voiceless aspirates in Greek to voiceless stops elsewhere, at least when following s (and where it can be assured that we're not dealing with Siebs' law, as in e.g. σχάω "to make an incision" which is almost certainly from *sekH-). The basic idea is that the laryngeal aspirated the stop, but only in this specific environment, rather than generally. The reasoning behind this is that we can probably assume that in PIE, voiceless stops were unaspirated when following *s, and aspirated otherwise. This was maintained into Greek, so when laryngeal aspiration occurred, the only stops that could be changed were exactly those following *s. They now contrasted with the plain voiceless stops, so they couldn't be interpreted as an allophone, leading to phonemic split. This could also be an alternative explanation of Siebs' law, with the benefit of not having to suppose that there were ever any clusters *sDʰ- in PIE, which is rather unattractive.

[Zaarin]

initial accent was also found in early Italic and Celtic.

For Proto-Celtic, that has been often posited, but never been proven. The only Celtic branch where we really know that it existed is Goidelic.

I was under the impression that word initial stress had been established (or at least established as probable) for Gaulish as well.

We know that Old European languages would have served as substrata for the European IE langs, and it's a reasonable guess that Basque is part of one of the Old European stocks. It might be a stretch to suppose that the stock Basque belonged to extended as far as they claim, but given how little we know about the Old European languages, it's hard to rule out.

Given that the pattern of Neolithic irruption into Europe - near-total replacement of the native population, spreading from one source by only a few routes, cultural homogeneity, and a subsequent catastrophic population collapse - it's probably the case that most languages outside of the Balkans were part of massive linguistic stocks, or even a single pan-European language family.
Then again, Basque could conceivably be a survival from the Mesolithic languages, in which case it might not have anything to do with any of the surrounding languages...

Fair. The Old European Hydronymy at least seems to point at a single large family spoken in most of Central and Western Europe before the spread of IE. As such homogeneity is unlikely to have existed in pre-Neolithic Europe, this probably was the Neolithic language family you are speaking of (I call it "Aquan", from the putative word for 'river' in that stock). Judging from the roots and suffixes those river names are made of, that family may have been related to IE, but branching off before the development of ablaut. But Basque is IMHO indeed a survival from the linguistic crazy quilt of Mesolithic Europe. Yet, Theo Vennemann assumes that Basque is a residue of the family of the Old European Hydronymy, which he calls "Vasconic" and doesn't consider related to IE. But his etymologies seem very contrived to me, much more so than Krahe's IE etymologies.

As for Vennemann's paper posted by Zju: none of the three points seem to be sufficiently persuasive of a Vasconic substratum in Celtic. The whole "Celtic from the West" theory this paper serves to buttress is IMHO buttocks, anyway.

Where would non-Indo-European languages of the northern Mediterranean--like Etruscan, Rhaetian, Palaeoiberian, and Minoan--fit into this scheme, in your opinion? Migrants from the Near East? Holdouts from the Neolithic unrelated to the language stocks of Transalpine Europe? (I realize it would be helpful if we could actually read Minoan before attempting to classify it...)

[hwhatting]

initial accent was also found in early Italic and Celtic.

For Proto-Celtic, that has been often posited, but never been proven. The only Celtic branch where we really know that it existed is Goidelic.

I was under the impression that word initial stress had been established (or at least established as probable) for Gaulish as well.

I've posted the relevant paragraphs from La Langue Gauloise here. If anything, it looks like Gaulish had ante-penult stress, with perhaps initial stress on compounds of prefix + noun.

[Zaarin]

initial accent was also found in early Italic and Celtic.

For Proto-Celtic, that has been often posited, but never been proven. The only Celtic branch where we really know that it existed is Goidelic.

I was under the impression that word initial stress had been established (or at least established as probable) for Gaulish as well.

I've posted the relevant paragraphs from La Langue Gauloise here. If anything, it looks like Gaulish had ante-penult stress, with perhaps initial stress on compounds of prefix + noun.

Thanks for the clarification. I've long been fascinated by Gaulish, but resources in English are hard to come by.

[Salmoneus]

Fair. The Old European Hydronymy at least seems to point at a single large family spoken in most of Central and Western Europe before the spread of IE. As such homogeneity is unlikely to have existed in pre-Neolithic Europe, this probably was the Neolithic language family you are speaking of (I call it "Aquan", from the putative word for 'river' in that stock). Judging from the roots and suffixes those river names are made of, that family may have been related to IE, but branching off before the development of ablaut.

I know we've had this discussion before, but I still have some problems with this.

- The hydronomy would be persuasive if it were just one or two repeating names, or a bunch of names repeating very specifically. But instead it's a whole bunch of names repeating only vaguely, which makes it suspiciously easy to slot any name into one of them - particularly when you have literally thousands and thousands of placenames to go at, and only have to produce a list of dozens of correspondences. Krahe, for instance, connects Solja, Salotas, Saalach, Salanfe and Hayle as all being the same name; likewise, Aire, Eisack and Izarillo all apparently match one another.

- if there is a real OEH, how could you judge that it was pre-ablaut IE? I mean, it's not as though you could reliably distinguish vowel sounds preserved over millennia and transferred between speakers of different languages at least once on the way. You'd pretty much expect everything to end up /a/. Even if the vowels are really representative of original /a/, it's not as though plenty of IE languages haven't moved a whole bunch of vowels to /a/. So simple IE (known branches, perhaps distorted by substrates and superstrates, or else unknown branches) seems a much safer guess. And far, far more parsimonious: we know there was one migration from the steppe, after the breakaway of anatolian, and it seems tendentious to suppose two migrations, one before and one after the breakup, when there is no direct evidence for that.

- if it is a branch of pre-IE, it's presumably not the neolithic language. We know that the neolithic population migrated from the near east; we know that a horseback population later migrated from the steppe, who weren't closely related to the neolithics. We know from the language that the neolithics didn't bring PIE itself with them. So for a macro-IE neolithic europe, we need: neolithics speaking pre-macro-IE; one neolithic group is invaded by the kurgans; the kurgans learn the language of their new slave population, who they are rapidly exterminating; the kurgans then spread the slave language all across Europe, central asia and south asia. Alternatively, we would have to suggest a sort of protero-IE steppe invasion that conquers all of Europe but leaves no other trace, followed by, thousands of years later, the known deutero-IE steppe invasion that gives us the ablauting languages we know. It seems much more parsimonious just to assume that if 'Aquan' is real, and really macro-IE, it's a closely related language spoken by closely related steppe people who invaded as part of the same general wave of settlement as the known IE languages.

- pushing back the dates also gives a big problem: the resemblances are almost too close to be real. Going back to sal-, Krahe points out that 'Seile' in northern Scotland and 'Seille' in eastern France (Switzerland?) look similar, and so do 'Sala' in Scandinavia and 'Salo' in Spain, and 'Salanza' in Switzerland and 'Salantas' in the Baltic. But hang on a moment! Within the attested historical record, many of these names have shown dramatic changes: Sayago, Jalon, Saa, Saumur, Saudre, Solk, Selz, etc. It seems implausible that these names would all hang around being more or less identical for three, four, five thousand years, and then suddenly dramatically mutate in the last few centuries as soon as we start actually looking at them! The further you push back the date of 'Aquan', the bigger this problem gets.

But Basque is IMHO indeed a survival from the linguistic crazy quilt of Mesolithic Europe.

What's the evidence for this?
There are basically six different possible classifications for Basque, two of which we can discount:
- IE. It's not.
- a relative of IE that took part in the irruption. It's almost certainly not.
- another Steppe language, either distantly related to IE or not at all, but that hung around in the same general cultural milleau and invaded Europe at a similar time.
- another language that invaded Europe either at the beginning of the Bronze Age or at a similar time to the IE invasions.
- a language that invaded Europe during the Neolithic.
- a language that entered Europe in the Neolithic invasions.
- a Mesolithic language.

Even discounting the first two, the other four seem impossible to distinguish between. The steppe option would have the advantage of fitting in alongside their high levels of R1, which we know came from the steppe. There's no evidence of a major influx of new genes between the neolithic and kurgan invasions, but that doesn't necessarily rule out a small group coming in - perhaps up the Atlantic coast, or across the Med and then across Spain (it's easy to imagine the Minoan and/or Etruscan languages as later invaders across the Med, for instance). Alternatively, both time periods are associated with Big Things that a new group of people might be related to: the middle Neolithic population collapse (with a lot of killing and cannibalism*) and the development of bronzeworking respectively.

Assuming an older entrance, I don't think there can be any clear evidence of Neolithic vs Mesolithic. It's true that the Basques tilt Mesolithic rather than Neolithic genetically, so I might slightly lean toward the Mesolithic hypothesis, slightly. But that's not very convincing, because we already know that there's been massive genetic replacement: Basques are genetically basically the same as IE speakers, only with very slightly more Mesolithic in them. But we know that the mesolithics were never entirely wiped out by the neolithics, and even if neolithics conquered the basque region it wouldn't be a huge surprise if the local mesolithics survived to influence the genetics more than in other parts of Europe - after all, the existence of Basque shows that that the area can shelter holdouts...

I'm also not convinced by the 'crazy quilt' idea. Again, Mesolithic Europe had only "recently" (IE-depth) been cleared of ice. While we could suppose the existence of distinct epigravettian populations in the italian and balkan refugia, most of the rest of Europe seems to have shared a single Magdalenian culture, associated with the retreat of the tundra and the domestication of the dog.

*disturbingly, as well as massacres, we've now discovered industrial cannibalism networks - humans were transported across huge distances to death camps to be killed and eaten, presumably for ritual reasons. There were probably 1,500 corpses dismembered and part-consumed at the one centre we know about, in only about 50 years - and we're not talking genteel chop-up-the-liver cannibalism, we're talking chew-on-the-bones cannibalism. It's the second-creepiest thing I've heard of in prehistoric Europe**. Amusingly, these are the people Gimbutas et al told us were the peaceful, matriarchal flower-loving pre-warfare gentle folk the nasty kurgans conquered...

**creepiest thing: the hebridean mummies. They were dead people, mummified after death. OK. But then their corpses were set up in people's houses. For up to 500 years. But here's the twist: although there are two mummies, they were constructed from the corpses of at least six different people...

People were weird in the olden days.

[Zaarin]

Amusingly, these are the people Gimbutas et al told us were the peaceful, matriarchal flower-loving pre-warfare gentle folk the nasty kurgans conquered...

It seems like every time archaeologists claim to find a peaceful race of kumbaya-singing, peace-pipe-smoking philosophers, we end up finding out they had creepy death-obsessed rituals. The Mayans spring to mind.

**creepiest thing: the hebridean mummies. They were dead people, mummified after death. OK. But then their corpses were set up in people's houses. For up to 500 years. But here's the twist: although there are two mummies, they were constructed from the corpses of at least six different people...

People were weird in the olden days.

And I thought Gaulish headhunting was creepy. O_O

[KathTheDragon]

It seems like every time archaeologists claim to find a peaceful race of kumbaya-singing, peace-pipe-smoking philosophers, we end up finding out they had creepy death-obsessed rituals. The Mayans spring to mind.

It's almost like humans are actually awful, and we've only recently managed to civilise ourselves.

[jmcd]

I found this on cannibalism: http://phys.org/news/2009-12-evidence-u ... ithic.html

BUT it also mentions that it's one a very few sites for the period.
So it may well have been the equivalent of the Nazis for the era.

[Zaarin]

It seems like every time archaeologists claim to find a peaceful race of kumbaya-singing, peace-pipe-smoking philosophers, we end up finding out they had creepy death-obsessed rituals. The Mayans spring to mind.

It's almost like humans are actually awful, and we've only recently managed to civilise ourselves.

Who knew. :p

[WeepingElf]

The speculations about pre-IE languages in Europe are interesting, but they stretch the limits of the topic of this thread. I have posted a reply on Salmoneus's and Zaarin's objections against my hypothesis here.

[KathTheDragon]

Reposting this since it seems to have been swamped by the other discussion.

I don't know if someone's come up with this before, but after looking through my etymological dictionaries, I had an idea for how to explain the apparently inexplicable connection of voiceless aspirates in Greek to voiceless stops elsewhere, at least when following s (and where it can be assured that we're not dealing with Siebs' law, as in e.g. σχάω "to make an incision" which is almost certainly from *sekH-). The basic idea is that the laryngeal aspirated the stop, but only in this specific environment, rather than generally. The reasoning behind this is that we can probably assume that in PIE, voiceless stops were unaspirated when following *s, and aspirated otherwise. This was maintained into Greek, so when laryngeal aspiration occurred, the only stops that could be changed were exactly those following *s. They now contrasted with the plain voiceless stops, so they couldn't be interpreted as an allophone, leading to phonemic split. This could also be an alternative explanation of Siebs' law, with the benefit of not having to suppose that there were ever any clusters *sDʰ- in PIE, which is rather unattractive.

[2+3 clusivity]

I understand that most older IE languages are SOV and have generally head final tendancies; however, of the ones I am familiar with, demonstratives tend to lead NPs. Am I wrong on this? Are there any articles about this?

[KathTheDragon]

Well, the demonstratives aren't the head of the NP, they're a complement.

[2+3 clusivity]

. . . we can probably assume that in PIE, voiceless stops were unaspirated when following *s, and aspirated otherwise. This was maintained into Greek, so when laryngeal aspiration occurred, the only stops that could be changed were exactly those following *s. They now contrasted with the plain voiceless stops, so they couldn't be interpreted as an allophone, leading to phonemic split. This could also be an alternative explanation of Siebs' law, with the benefit of not having to suppose that there were ever any clusters *sDʰ- in PIE, which is rather unattractive.

Indo-Aryan languages tended to show the opposite route where /*SC/ clusters often > /SCʰ/ and further to /Cʰ/. As I recall, /Cʰ/ largely emerged from /CH/ and /SC/ environments and spontaneous aspiration was rare or from post-PIE vocab.

Consider also that the effect of /*s/ on following voicing probably only kicked-in (1) initially and (2) perhaps more broadly only at the PIE stage[/u]. Consider the quad --

PIE --
2d sg. imp. "be!" PIE /*h1s-dʰí/; 3rd sg. imp. "be!" PIE /*h1es-tu/
2d sg. imp. "go!" PIE /*h1i-dʰí/; 3rd sg. imp. "go!" /h1i-tu/

Ancient Greek (my vowel notation is not great here: mea culpa) –
<ισθι> /is-tʰi/; <ἔστω> /és-tō/
<ιθι> /í-thi/; <ἴτω> /í-tō/

but

Av. --
/zdī/; /astū/
/idī/; ??

and

Ved. SKT --
<edhí> /aidʰí/ < /*(H)zdʰí/; /ástu/¹
<ihí> /ihí/ < /*(H)idʰí/; <étu> /áitu/

1:The former's initial a- appears to have been leveling or added to dissimilate from "go!" Also note that /#H-/ > 0 so the /-i-/ is from /*-z-/ as in similar forms.

EDIT:

Kath -- my grasp of syntax isn't great, I had thought that the NPs were daughters of optional DPs. Does it just depend on your theoretical bent?