Paleo-European languages

[keenir]

cut&paste helps cut down on multi-post; took me a while to learn it too.

So please, help end this by explaining at least ONE of your theories.

OK. My view is that proto-languages and genealogical trees for modelling language relationships are a simplification, because they don't take into account (or do it poorly) lateral relationships (substrates and adstrates). In the case of the reconstructed PIE, one can find internal correspondences between the +2000 lexical items which would indicate they come from at least 2-3 different sources (Occam's Razor is at work here).

ah, okay. been a while since I saw a discussion (or even mention?) about the trees vs their alternatives.

But please refresh my memory, because I can't recall what you or anyone else said was a viable other option to the tree model.

To me, at least, a tree still works with lots of loans from other languages.....just look at a Family Tree - there's at least one name (word) from another tree grafting itself in, at least once a generation. Still a tree.

Upturned Microscope cartoons, while funny, do not go very far in advancing the discussion. Far better would be if you could explain exactly how WeepingElf has misrepresented your argument in such a way as to resolve the misunderstanding.

This isn't the first time he recurrs to chance resemblance to criticise my theories. This is the strawman, and a disgusting one, because he's indirectly calling me a "crackpot".

What/which theories? Insofar as I can tell, WeepingElf - like others here (mine own self included) - are calling your attention to the fact that you weren't talking about your theories...lately, you've just saying an archaic word or two, then saying a theory name, and clicking Submit.

[WeepingElf]

Maybe we are just talking past each other all the time, and see the same thing from different angles It seems as if both T. and I feel as if there was a stratum of extinct IE languages, later clobbered by the expansion of Italic, Celtic and Germanic, in much of Western Europe, but we have different ideas about what those lost IE languages were like.

Not necessarily, because there're could be more than just one.

Fair. Yet, the assumption that there was just one such layer is more parsimonious, and the claim that there was more than one requires appropriate evidence (indeed, the claim that there was one such layer requires appropriate evidence). And of course, there were at least two non-IE layers (one Paleolithic/Mesolithic, one Neolithic) before that.

T. apparently calls this stratum "Baltoid" (which perhaps is a designation he did not coin himself but picked up somewhere)

It looks like you didn't read my earlier posts where I mentioned Coromines' and Villar's works.

I guessed that sort of thing. I'm sorry for missing that; maybe I just forgot about it. Thank you for reminding me. I have to admit that when reading your posts, some details sometimes slip out of my attention.

[Richard W]

As you mention, though, how likely is it that these diachronic differences are due to learned borrowings or borrowings that occurred after or partway through native sound changes?

Learned borrowings introducing new phonemes that behave differently is part of the phenomenon - it doesn't really matter, so long as these phonemes penetrate to the spoken language. Bilingualism can result in the two languages' phonemes contrasting within the two languages, rather than being a matter of register switching. Putting it crudely, a contrast between a phoneme with one allophone and a similar one that differs by having that allophone in free variation with another one seems possible.

For the reflexes of the Indic voiced aspirates, there doesn't seem to be a post-shift source for the contrast of Indic voiced stops as voiceless unaspirated stops and Indic voiced aspirates as voiceless aspirates. It's why some people hesitantly posit borrowing as some form of 'voiced aspirate' before the great consonant shift. Now, in the case of Northern Thai, the reflexes of voiced aspirates are the same as the reflexes of clusters of native voiced stops plus /r/, and some write (wrote?) the reflexes of the clusters as though they came from Indic voiced aspirates. (This change is part of the native change /r/ > /h/; only the first consonant in a cluster affects which way the tone jumps in the tone splitting.) I've not come across cases where a word-initial Indic voiced aspirate stop is now pronounced as though it had been an Indic voiced non-aspirate stop.

By contrast, there are a few cases where what should have been /l/ from Indic /r/ is instead the reflex of Tai /r/, namely /h/, consistent with a steady drip of correction from the temples.

As an example of massive refusal to identify the phonemes of two languages, I've been told that Chiang Rai dialect has two sets of tones - the native tones, and the corresponding tones of Siamese. That would give 11 tonemes, though a figure of 9 sticks in my mind - some of the corresponding tones are very similar.

[Talskubilos]

Well, to be fair, most of your recent posts have been comparing only one or two word pairs at a time. This significantly increases the chance of chance resemblances; you will only gain wide acceptance if you give us systematic correspondences. Or, putting it another way: the null hypothesis here is that any two words which look similar are, in fact, chance resemblances. It’s your job to give us enough data that we feel comfortable accepting that the resemblances are not due to chance.

Actually, this is rather on the contrary: chance resemblances are more likely in large sets of data.

[Frislander]

Upturned Microscope cartoons, while funny, do not go very far in advancing the discussion. Far better would be if you could explain exactly how WeepingElf has misrepresented your argument in such a way as to resolve the misunderstanding.

This isn't the first time he recurrs to chance resemblance to criticise my theories. This is the strawman, and a disgusting one, because he's indirectly calling me a "crackpot".

Perhaps you could do with reading the host of this very forum's own deep dive into the topic, you might gain much from it.

Well, to be fair, most of your recent posts have been comparing only one or two word pairs at a time. This significantly increases the chance of chance resemblances; you will only gain wide acceptance if you give us systematic correspondences. Or, putting it another way: the null hypothesis here is that any two words which look similar are, in fact, chance resemblances. It’s your job to give us enough data that we feel comfortable accepting that the resemblances are not due to chance.

Actually, this is rather on the contrary: chance resemblances are more likely in large sets of data.

And? You no doubt have been trawling through large sets of data yourself to bring up the "comparisons" you propose. Indeed that's why we're suspicious of these things - if you are taking the lexicon of two geographically distant languages and you only feel confident presenting one or two words for us to examine, that suggests you don't have much else to show because you've hit upon a chance resemblance.

So please, help end this by explaining at least ONE of your theories.

OK. My view is that proto-languages and genealogical trees for modelling language relationships are a simplification, because they don't take into account (or do it poorly) lateral relationships (substrates and adstrates). In the case of the reconstructed PIE, one can find internal correspondences between the +2000 lexical items which would indicate they come from at least 2-3 different sources (Occam's Razor is at work here).

ah, okay. been a while since I saw a discussion (or even mention?) about the trees vs their alternatives.

But please refresh my memory, because I can't recall what you or anyone else said was a viable other option to the tree model.

To me, at least, a tree still works with lots of loans from other languages.....just look at a Family Tree - there's at least one name (word) from another tree grafting itself in, at least once a generation. Still a tree.

Part of the reason I've never found any of the proposed "alternatives" to the tree model entirely satisfying is because none of them ever seem like they have the same level of descriptive power of the tree model, especially for larger families. Like sure, a wave model works just about when you have a relatively close-knit group of varieties whose speakers see each other as fundamentally unified and where innovations can relatively freely spread across the varieties, as in Early IE, but it's pretty much useless for talking about modern IE as it currently stands, as the major branches are too differentiated from each other for such free spread and contact-induced changes easily show up as a discontinuity in reflexes, e.g. since Proto-Celtic and Primitive Irish lacked *p, as evidenced by the long list of otherwise regular reflexes , any /p/ found in the modern language is a tell-tale sign that it's a loan which came into the language from "outside" and was for a period external to the phonological system, i.e. it was imported across a distinct language barrier. You can talk about "what about substrates and adstrates?" all you like, but the same principle applies - there are a few parts of the language which arrived into the language at a particular point in time but these bits came to an already fully-formed linguistic system which the rest of the language reflects.

[WeepingElf]

Upturned Microscope cartoons, while funny, do not go very far in advancing the discussion. Far better would be if you could explain exactly how WeepingElf has misrepresented your argument in such a way as to resolve the misunderstanding.

This isn't the first time he recurrs to chance resemblance to criticise my theories. This is the strawman, and a disgusting one, because he's indirectly calling me a "crackpot".

This is a behaviour I have seen from you before. Someone criticizes your argumentation, which is a fair and legitimate move in any kind of scholarly discussion, and you construe this as an ad hominem attack which it isn't meant to be. You are sincere and intelligent, your only flaw is that you jump at conclusions from insufficient evidence and are too emotionally attached to them. A scholar should be ready to review and, if necessary, abandon his hypothesis when problems with it are pointed out by other scholars. And I have seen ideas, both on the Net and in printed matter, that were much more misguided than yours! As I have laid out in this post, I think that we may be looking at the same thing, but from different angles and interpreting it differently. And yes, chance resemblances are a potential source of error one needs to be aware of when doing this kind of work.

[Talskubilos]

And yes, chance resemblances are a potential source of error one needs to be aware of when doing this kind of work.

As a matter of fact, you can find plenty of them in Pokorny's and other IE etymological dictionaries. I remind myself of an example: http://vasco-caucasian.blogspot.com/201 ... isade.html

Celtic *koret- 'palisade' (Middle Irish cora, Middle Welsh cored, Old Breton coret) has been linked by some Indo-Europeanists to Germanic *xurdí-/*xúrθi- and Latin crātis 'hurdle' (whose diminutive form crātīcula is the origin of French grille 'grill' and similar Romance words), from an IE protoform whose meaning can be reconstructed as 'to weave'.

However, IMHO this etymology is semantically unsatisfactory, as hurdles were traditionally made from wattle (woven split branches), while palisades are made from stakes planted vertically on the ground. This is why I'd prefer to derive the Celtic word from a different protoform *s-korHt- 'twig, pole' found in Baltic (Lithuanian kártis 'thick stick, latte, stake', Latvian kãrts 'rod; a measure of length') and Germanic (Old Norse skorδa 'support, rod'), in a pars pro toto etymology.

[WeepingElf]

And yes, chance resemblances are a potential source of error one needs to be aware of when doing this kind of work.

As a matter of fact, you can find plenty of them in Pokorny's and other IE etymological dictionaries.

Pokorny's dictionary is old and not only uses an obsolete PIE phonology (though it is not difficult to rewrite the reconstructed forms according to the current phonology, but this reveals that some are problematic for formal reasons alone), but many scholars, including me, think that it heavily over-reconstructs. Many items are a semantic stretch, or limited to a few neighbouring branches. See this list for some of these items. There are also awfully many homonyms and synonyms such that one instinctively doubts that all the items are from one and the same language.

I remind myself of an example: http://vasco-caucasian.blogspot.com/201 ... isade.html

Celtic *koret- 'palisade' (Middle Irish cora, Middle Welsh cored, Old Breton coret) has been linked by some Indo-Europeanists to Germanic *xurdí-/*xúrθi- and Latin crātis 'hurdle' (whose diminutive form crātīcula is the origin of French grille 'grill' and similar Romance words), from an IE protoform whose meaning can be reconstructed as 'to weave'.

However, IMHO this etymology is semantically unsatisfactory, as hurdles were traditionally made from wattle (woven split branches), while palisades are made from stakes planted vertically on the ground. This is why I'd prefer to derive the Celtic word from a different protoform *s-korHt- 'twig, pole' found in Baltic (Lithuanian kártis 'thick stick, latte, stake', Latvian kãrts 'rod; a measure of length') and Germanic (Old Norse skorδa 'support, rod'), in a pars pro toto etymology.

The etymology is indeed semantically unsatisfactory. The original meaning of the Celtic, Germanic and Italic words is clearly 'fence' or something similar, and they may easily be loanwords from various Southern IE languages (not all from the same language since Bronze Age Western Europe simply was too large to speak a single language, but from several related ones). Whether they are cognate to the words in the last paragraph, I don't know; it is not impossible because Southern IE of course is related to Baltic and Germanic.

[keenir]

Like sure, a wave model works just about when you have a relatively close-knit group of varieties whose speakers see each other as fundamentally unified and where innovations can relatively freely spread across the varieties, as in Early IE, but it's pretty much useless for talking about modern IE as it currently stands, as the major branches are too differentiated from each other for such free spread and contact-induced changes easily show up as a discontinuity in reflexes,

normally, when I read something along those lines as you've written here, I think of how people lump Mandarin, Cantonese, Hokkien, {and more} into Chinese, sometimes even trying (those others, not yourself; i don't want confusion here) to pass it off as a single language.

and then, today, in New Scientist magazine, someone was making the argument that a lot of words - his examples were mostly in genetics and biology - are outdated and in need of replacement, such as the word Caucasian(sp) {because it has, over the course of time, varied as to its meaning in regard to geopolitics and beauty}.....and then he said that, because the Bantu family has 400 dialects over a vast swath of land, it is meaningless and useless to say Bantu.

yeah...."Wha??"

{sorry...felt like sharing}

[bradrn]

Well, to be fair, most of your recent posts have been comparing only one or two word pairs at a time. This significantly increases the chance of chance resemblances; you will only gain wide acceptance if you give us systematic correspondences. Or, putting it another way: the null hypothesis here is that any two words which look similar are, in fact, chance resemblances. It’s your job to give us enough data that we feel comfortable accepting that the resemblances are not due to chance.

Actually, this is rather on the contrary: chance resemblances are more likely in large sets of data.

…OK, let’s take this from the beginning.

Consider two languages. These languages will have a certain number of false cognates, and a certain number of true cognates. Now, historical linguistics works on the principle that true cognates will have systematic phonological correspondences, which false cognates will not share. Thus, if you have enough data, the two can be distinguished by checking for systematic correspondences over a large set of words. These will be shared between true cognates, but not false ones.

Thinking about this, it is clear that false cognates are more common in large sets of data. However, this does not imply that true cognates are more common in small sets of data! (A proposition does not imply its converse; to say so is fallacious.) Rather, both false and true cognates become more likely in large sets of data, but only large datasets can give the number of systematic correspondences required to meet the standards of (non-crackpotted) historical linguistics.

OK. My view is that proto-languages and genealogical trees for modelling language relationships are a simplification, because they don't take into account (or do it poorly) lateral relationships (substrates and adstrates). In the case of the reconstructed PIE, one can find internal correspondences between the +2000 lexical items which would indicate they come from at least 2-3 different sources (Occam's Razor is at work here).

ah, okay. been a while since I saw a discussion (or even mention?) about the trees vs their alternatives.

But please refresh my memory, because I can't recall what you or anyone else said was a viable other option to the tree model.

To me, at least, a tree still works with lots of loans from other languages.....just look at a Family Tree - there's at least one name (word) from another tree grafting itself in, at least once a generation. Still a tree.

Part of the reason I've never found any of the proposed "alternatives" to the tree model entirely satisfying is because none of them ever seem like they have the same level of descriptive power of the tree model, especially for larger families. Like sure, a wave model works just about when you have a relatively close-knit group of varieties whose speakers see each other as fundamentally unified and where innovations can relatively freely spread across the varieties, as in Early IE, but it's pretty much useless for talking about modern IE as it currently stands, as the major branches are too differentiated from each other for such free spread and contact-induced changes easily show up as a discontinuity in reflexes, e.g. since Proto-Celtic and Primitive Irish lacked *p, as evidenced by the long list of otherwise regular reflexes , any /p/ found in the modern language is a tell-tale sign that it's a loan which came into the language from "outside" and was for a period external to the phonological system, i.e. it was imported across a distinct language barrier. You can talk about "what about substrates and adstrates?" all you like, but the same principle applies - there are a few parts of the language which arrived into the language at a particular point in time but these bits came to an already fully-formed linguistic system which the rest of the language reflects.

The wave model reduces to the tree model in such cases: if your innovations do not overlap, then the various groups become mutually disjoint, same as with the tree model. The wave model however can account for cases such as linkages and creoles, which the cannot be modelled by the tree model.

and then, today, in New Scientist magazine, someone was making the argument that a lot of words - his examples were mostly in genetics and biology - are outdated and in need of replacement, such as the word Caucasian(sp) {because it has, over the course of time, varied as to its meaning in regard to geopolitics and beauty}.....and then he said that, because the Bantu family has 400 dialects over a vast swath of land, it is meaningless and useless to say Bantu.

yeah...."Wha??"

{sorry...felt like sharing}

Oh, dear. Which one was that? Because I have a bunch of unread New Scientists from when I was studying for my exams, and I’d rather know which one to avoid if possible…

[keenir]

Oh, dear. Which one was that? Because I have a bunch of unread New Scientists from when I was studying for my exams, and I’d rather know which one to avoid if possible…

July 10-16, 2021...though aside from that one interview, the issue is fine.
(granted, i haven't read every article, but still)

[bradrn]

Oh, dear. Which one was that? Because I have a bunch of unread New Scientists from when I was studying for my exams, and I’d rather know which one to avoid if possible…

July 10-16, 2021...though aside from that one interview, the issue is fine.
(granted, i haven't read every article, but still)

Yep, found it. It is indeed as bad as it sounded. There’s too much bad linguistics out there.

(Granted, it makes more sense if you’re of the opinion that any generic term which reduces precision is bad… I just wonder what he thinks about that imprecise group of dialects called ‘Romance’. Or ‘Chinese’ or ‘Arabic’, for that matter.)

[Talskubilos]

Maybe we are just talking past each other all the time, and see the same thing from different angles It seems as if both T. and I feel as if there was a stratum of extinct IE languages, later clobbered by the expansion of Italic, Celtic and Germanic, in much of Western Europe, but we have different ideas about what those lost IE languages were like. T. apparently calls this stratum "Baltoid" (which perhaps is a designation he did not coin himself but picked up somewhere) and sees it as a language broadly similar to Baltic (i.e., a satem language with *D-Dh and *a-o mergers), while I call it "Southern IE" and suspect a relationship to the Anatolian languages such as Hittite, while not making any predicament about its phonological developments. (Obviously, these two hypotheses do not even contradict each other - why couldn't a branch of Southern IE - independently from Baltic, though - show such sound developments, which are commonplace enough in the IE family? In my opinion, the phonologies of Northern and Southern IE were probably mostly isomorphic - not necessarily the same, but each phoneme in one corresponding to one phoneme in the other - as the Late PIE phonology, which would be that of Northern IE, does a fairly good job accounting for Anatolian, the sole attested Southern IE branch, while the differences between Northern and Southern IE lay in morphology, syntax and maybe lexicon.) Such a stratum was already proposed in the middle of the 20th century (see Hans Krahe's "Old European Hydronymy" which, however, is not without flaws; some of the names on Krahe's list, such as Brigantia, are patently Celtic or from other known languages), and it seems to me as if the Bell Beaker culture was a good candidate for the speaker community of these languages.

I'm sure Krahe's OEH is made up from several layers.

[WeepingElf]

I'm sure Krahe's OEH is made up from several layers.

That's my opinion, too. The question is of course how to distinguish them; the only way I see is to guess the cognate languages of the layers correctly. My guess - and it is indeed nothing than a guess - is that the "Bell Beaker" layer is Southern IE, and the Neolithic layer, at least in the Mediterranean countries, is related to Basque; but this guess may be utterly wrong. The key problem with old geographical names is that we don't know their original meanings, unless an element in them reliably co-occurs with a particular feature of the objects thus named. This, for instance, seems to be the case with the many names of Central European salt production sites containing *hal-, which I thus conjecture to have meant 'salt', probably in a Southern IE language, as it may be related to PIE *sh₂el- ~ *seh₂l-.

[Talskubilos]

I'm sure Krahe's OEH is made up from several layers.

That's my opinion, too. The question is of course how to distinguish them; the only way I see is to guess the cognate languages of the layers correctly. My guess - and it is indeed nothing than a guess - is that the "Bell Beaker" layer is Southern IE, and the Neolithic layer, at least in the Mediterranean countries, is related to Basque; but this guess may be utterly wrong.

Apart from the fact Basque has strong IE influences (mostly from Latin-Romance and Celtic), it's possible its native core was some kind of para-IE variety which originated in the Steppes.

The key problem with old geographical names is that we don't know their original meanings, unless an element in them reliably co-occurs with a particular feature of the objects thus named.

This is precisely why I'd prefer to work with lexical items instead of toponyms.

This, for instance, seems to be the case with the many names of Central European salt production sites containing *hal-, which I thus conjecture to have meant 'salt', probably in a Southern IE language, as it may be related to PIE *sh₂el- ~ *seh₂l-.

This reminds me of the correspondence between *sam-/*sm-ro- 'summer' in several IE European languages and *h₂ēm-er- 'heat of the day' (Greek hêmérā 'day'), with the "laryngeal" matching Semitic *ħamm- 'to be hot; warm'.

[WeepingElf]

I'm sure Krahe's OEH is made up from several layers.

That's my opinion, too. The question is of course how to distinguish them; the only way I see is to guess the cognate languages of the layers correctly. My guess - and it is indeed nothing than a guess - is that the "Bell Beaker" layer is Southern IE, and the Neolithic layer, at least in the Mediterranean countries, is related to Basque; but this guess may be utterly wrong.

Apart from the fact Basque has strong IE influences (mostly from Latin-Romance and Celtic), it's possible its native core was some kind of para-IE variety which originated in the Steppes.

What do you mean by "para-IE variety"? I would understand "Para-IE" as a language related to IE but not being a part of IE proper (which Basque doesn't seem to be), but it wouldn't be the first time you used a word in an idiosyncratic way unintelligible to anyone but yourself if you did here.

The key problem with old geographical names is that we don't know their original meanings, unless an element in them reliably co-occurs with a particular feature of the objects thus named.

This is precisely why I'd prefer to work with lexical items instead of toponyms.

Fair.

This, for instance, seems to be the case with the many names of Central European salt production sites containing *hal-, which I thus conjecture to have meant 'salt', probably in a Southern IE language, as it may be related to PIE *sh₂el- ~ *seh₂l-.

This reminds me of the correspondence between *sam-/*sm-ro- 'summer' in several IE European languages and *h₂ēm-er- 'heat of the day' (Greek hêmérā 'day'), with the "laryngeal" matching Semitic *ħamm- 'to be hot; warm'.

You seem not to know the most basic facts of IE historical linguistics. Greek h has nothing to do with PIE laryngeals!!! It is a regular development of PIE *s in certain environments, including word-initial before a vowel. You should really read a handbook of the IE family (I recommend Indo-European Language and Culture by Benjamin W. Fortson IV) in order to get yourself up to the topic - and you should read it attentively. And leave Semitic out of the game - we have no evidence of Semitic anywhere in prehistoric Europe, and the notion of a relationship between Semitic and IE is highly dubious.

[bradrn]

This, for instance, seems to be the case with the many names of Central European salt production sites containing *hal-, which I thus conjecture to have meant 'salt', probably in a Southern IE language, as it may be related to PIE *sh₂el- ~ *seh₂l-.

This reminds me of the correspondence between *sam-/*sm-ro- 'summer' in several IE European languages and *h₂ēm-er- 'heat of the day' (Greek hêmérā 'day'), with the "laryngeal" matching Semitic *ħamm- 'to be hot; warm'.

You seem not to know the most basic facts of IE historical linguistics. Greek h has nothing to do with PIE laryngeals!!! It is a regular development of PIE *s in certain environments, including word-initial before a vowel.

I read Talskubilos as suggesting a relationship between *h₂ēm-er- and *ħamm-; I’m not sure why he mentioned Greek, since it seems to be irrelevant to his argument.

[Nortaneous]

And leave Semitic out of the game - we have no evidence of Semitic anywhere in prehistoric Europe, and the notion of a relationship between Semitic and IE is highly dubious.

Early borrowing between PIE and Proto-Semitic is widely accepted and basically reasonable.

What do you mean by "para-IE variety"? I would understand "Para-IE" as a language related to IE but not being a part of IE proper (which Basque doesn't seem to be), but it wouldn't be the first time you used a word in an idiosyncratic way unintelligible to anyone but yourself if you did here.

Johanna Nichols thought it was, but her work on that was complete garbage and it isn't. "Para-X" relationships can be demonstrated - there's at least Khitan - but a genetic relation between IE and Basque is the single least promising hypothesis in the field. Maybe Etruscan, but not Basque.

[WeepingElf]

And leave Semitic out of the game - we have no evidence of Semitic anywhere in prehistoric Europe, and the notion of a relationship between Semitic and IE is highly dubious.

Early borrowing between PIE and Proto-Semitic is widely accepted and basically reasonable.

Indeed. The words common to IE and Semitic may be Neolithic Wanderwörter. What I meant to call "highly dubious" was the notion that IE and Semitic share a common ancestor within a halfways reasonable time depth.

What do you mean by "para-IE variety"? I would understand "Para-IE" as a language related to IE but not being a part of IE proper (which Basque doesn't seem to be), but it wouldn't be the first time you used a word in an idiosyncratic way unintelligible to anyone but yourself if you did here.

Johanna Nichols thought it was, but her work on that was complete garbage and it isn't. "Para-X" relationships can be demonstrated - there's at least Khitan - but a genetic relation between IE and Basque is the single least promising hypothesis in the field. Maybe Etruscan, but not Basque.

IE and Basque are so vastly different that there is no way they could be closely related. There was a time when I considered Etruscan a good candidate for the closest known kin of IE, but I no longer think that way - Uralic looks much more like a cousin of IE than Etruscan does, and is IMHO the best candidate for that.

There is a marginal possibility that the "R1b people", whom I suspect to have spoken what I call "Southern IE", instead spoke an ancestor of Basque (while the "R1a people" would have spoken PIE), but that would leave Anatolian unaccounted-for, and the obviously high degree of bride exchange between the two populations (while the Y-DNA is different, the mitochondrial and autosomal DNA is pretty much the same) gives weight to the assumption that they spoke the same language or at least closely related ones.

Some people say that the "R1a people" spoke satem languages and the "R1b people" spoke centum languages, but few IEists believe in such a bifurcation of IE anymore, and again, it leaves Anatolian unaccounted-for, and fails in the light of Tocharian, which was the main reason why mainstream IEists abandoned the satem/centum theory.

We are probably dealing with two patrilineal but at least to some degree exogamous clans who spoke different dialects of the same language which would have evolved into two primary branches of IE of which that of the "R1b clan" would have later been clobbered by that of the "R1a clan" at some time in the Bronze Age (earlier on the Balkans than in Western Europe), such that only in Anatolia, languages of the "R1b clan" were laid down in writing. Hence, my hypothesis of "Northern IE" and "Southern IE" spoken by the "R1a people" and the "R1b people", respectively.

[Nortaneous]

IE and Basque are so vastly different that there is no way they could be closely related. There was a time when I considered Etruscan a good candidate for the closest known kin of IE, but I no longer think that way - Uralic looks much more like a cousin of IE than Etruscan does, and is IMHO the best candidate for that.

The similarity between Finnish and Latin was obvious to me before I knew anything about historical linguistics, although IIRC some of that is due to later developments of the Finnish verbal system. But Etruscan might still come in ahead of Yukaghir.

The idea of a reconstructible genetic relation between IE and AA is, of course, also hopeless - it was a very popular hypothesis once for sociological reasons, so if there were any meat to it we'd know.

Some people say that the "R1a people" spoke satem languages and the "R1b people" spoke centum languages, but few IEists believe in such a bifurcation of IE anymore, and again, it leaves Anatolian unaccounted-for, and fails in the light of Tocharian, which was the main reason why mainstream IEists abandoned the satem/centum theory.

It's not a great reason for abandoning it, though - the eastward migration of the pre-Tocharians is probably secondary, and Proto-Tocharian still has to be reconstructed with labiovelars - they were preserved into even Tocharian B, although they could've disintegrated segmentally at some point. Satemization applying differently in different families seems like a better argument - you'd need to posit a lot of loanwords with some suspicious patterns to get out of it being areal. Then again, surely one of akmuo and ašmuo is a loan? (On semantic grounds probably the latter, since it has a more specialized meaning.)

I still don't understand the population genetics stuff, unfortunately.