The Great Macrofamily thread: Indo-Uralic, Altaic, Eurasiatic, Nostratic etc.

[Tropylium]

It also seems to me that Proto-Eskimo-Aleut suffers from a similar problem as Proto-Chukotko-Kamchatkan: it is basically Proto-Eskimo with some fudging to get Aleut to comply with it.

On that note, you might be interested in the Comparative EA Collection from Alaska Native Language Archive (recently brought to my attention). I've not seen any good single overview of PEA, but there seems to be quite a bit more work done than on PCK.

Chronology could also be the reason why PIE looks so "un-Mitian". The median North Asian language profile with palatals, uvulars, no clusters, no ablaut etc. looks perhaps "old", but could have regardless come about more recently than PIE was spoken: some 3k to 5k years would be easily enough.

Yes, we may deal with a Sprachbund that emerged only long after "Proto-Mitian" has broken up (if it had ever existed), and failed to include IE, which is of course a geographical outlier.

I don't mean to suggest a Sprachbund at any one particular time (though that's a possibility), more that there may have been general typological trends over millennia. Possibly mediated by not just diffusion of features within Central Asia / Siberia, but also increasingly strong "Paleosiberian" / "Amerind" influence towards the east. Most of Europe by contrast got overran by IE (and para-IE?) quite early, probably with only scraps of the previous languages remaining by 1000 BCE.

For an example, let's dig into a few features from Anderson (2006), Towards a Siberian linguistic area:

• Four nasals /m n ń ŋ/; in most languages with a ban on initial /ŋ/ (exceptions include Tungusic, northern Samoyedic, Nivkh and CK), and in several with /ń/ being marginal or a recent introduction (Turkic, Mongolic, Yeniseic, CK, and though Anderson doesn't mention it, also northern Samoyedic — initial /ń/ is mostly and medial /ń/ always from *n + front vowel, while old *-ń- > -j-).
The second fact probably indicates that /ŋ/ generally comes from some sort of earlier consonant clusters like *ng. Of course, the pattern could be very old (it reconstructs already into PU) and IE could have just lost this. On the other hand, I would think any reconstruction along the lines of PU *ŋ ~ IE *g < *ɠ looks unlikely and does not explain this distributional constraint. — /ń/ ties into the general palatals question, but at least an independent medial it could be very easily rewritten as *nj.

• SOV plus extensive verb nominalization
Unmarked, extends much wider also, generally not a strong genetic signal.

• Extensive case systems including prolative, (al)lative and comitative
Large "semantic" case systems usually come simply from postpositions + agglutination. Proposed deep cognates among case markers seem to mostly involve the grammatical cases, and these markers find their IE counterparts easily enough.

[Salmoneus]

Just on the subject of reconstructions adding new languages: you'd expect them not to change the reconstruction that dramatically.

Take, say, Anatolian and PIE-Anatolian. To combine them both into Indo-Hittite, you're postulating a language that had (at least) two immediate descendents: Late PIE and Proto-Anatolian. If these two proto-languages are closely related to one another, then their latest common parent will indeed be quite similar to both of them.

But what's more, methodologically, even if it wasn't, it's hard to prove it. When you have a lot of languages with a common ancestor, you can argue by appeal to consensus, at least once internal family structure has been worked out. When you only have two languages to compare, it's much harder. If Anatolian has a feature LPIE doesn't, did it innovate it, or retain it? If it doesn't have a feature LPIE has, did it not retain it, or had it not yet been developed in the proto-language? There will be a few cases where one is inherently more likely as a direction of change than the other, but a lot of the time you could be flipping a coin. And established experts will flip the coin so that it comes down "things are how we thought they were, mostly" as often as they can, until proven otherwise, because it's a lot less hassle, and more methodologically respectable (it's better to be cautious than just to make things up). So things end up the same but with an "although it's possible this may not have developed until after the split with Anatolian, it's not clear..." asterisk next to a bunch of things.

[Salmoneus]

I am growing increasingly doubtful of the idea that there were two early western expansions from the PIE homeland, both characterized by R1b, one leading to Anatolian IE, the other to Bell Beaker. These probably were two prongs of one movement into the Lower Danube region which then forked into Bell Beaker and Anatolian. So one earlier movement to the west, carrying the "nobleman" haplogroup R1b, and one later one in all directions, carrying the "common man" haplogroup R1a. And "Aquan", the hypothetical language of the Bell Beaker people, would be a sister of Anatolian.

So far as I'm aware, this isn't true.

For a start, I'd caution against using "Bell Beaker" as a family name, because we don't understand how the cultural phenomenon of beakers worked.

But in terms of the spread of R1b through western europe, this is believed to have come from an expansion beginning in the western periphery of the early Corded Ware Culture - the Rhine, etc. Rhenish CWC could be from Yamnaya moving up the Danube and then adopting CWC innovations in Germany... but I don't think there's evidence of that. Instead, they look to just be the western end of the CWC expansion. From there, they rapidly expanded throughout France, Germany, the British Isles, and a little way down the Danube - but not, originally, into Iberia.

How do you get an R1b expansion out of an R1a CWC? Well, we need to remember we're dealing with the remain of a tiny fraction of the individuals, disproportionately high-status males. There's a big 'hidden population' that don't show up.

But here's three scenarios for you:

- The Yamnaya and CWC phenomena were multiethnic cultural and perhaps political-military alliances. Different ethnic groups were dominant in different areas, and those dominant groups are the people we mostly find the remains on. As one set of tribes comes to dominance in the CWC area, a rival/ally tribe, let's call them for sake of argument "the Basques", decide to find their own territory next door. They do very well and expand massively.

- There's only one ethnicity, but a clear class difference. In the CWC, with the traditional Yamnaya environment and culture no longer applying, there's, as it were, a culture-wide "revolution" overthrowing the ruling class; deposed princes gather together at or beyond the western border and restore their warlike ways.

- Or, perhaps most likely, the whole of this society had several bloodlines in it, and coincidentally, in the course of a massive, rapid expansion in which small pioneer groups may have been isolated for significant periods, different bloodlines came to the fore in different areas. I think it makes most sense to see Yamnaya has actually having been R1a "on average", an average which bore out across most of the CWC area (small pockets of R1b could easily be bred out through intermarriage with neighbours); but there would have been some R1b families. The Yamnaya aristocracy was mostly R1b, and, perhaps purely by coincidence, the cluster of pioneers on the western fringe of CWC, who would go on to take over most of western europe, just happened to also be lead by an R1b family.


It's worth bearing in mind that it only takes one guy to introduce a dominant lineage, and these were mobile societies, so who knows. Russia, for example, is not very R1b now - but Tsar Nicholas II was R1b! When you have a continent conquered by Genghis Khan and his buddies, you can probably have large areas of genetic history just work out by luck, depending on where whichever particular friend-of-Genghis who dominated that area happen to originate from...


It's also worth noting that western european R1b, the type spread by the non-Iberian "Bell Beaker Folk", is (almost) all from a specific , young R1b linneage that hasn't been found in Yamnaya (yet). So in any case, that whole-of-western-Europe lineage pretty much popped out of nowhere (i.e. developed rapidly from a small founder population) - it's not, for example, a grabbag of random R1b guys from across the CWC territory getting together, it's all one... well, family.

[Tropylium]

When you have a lot of languages with a common ancestor, you can argue by appeal to consensus, at least once internal family structure has been worked out. When you only have two languages to compare, it's much harder. If Anatolian has a feature LPIE doesn't, did it innovate it, or retain it? If it doesn't have a feature LPIE has, did it not retain it, or had it not yet been developed in the proto-language? There will be a few cases where one is inherently more likely as a direction of change than the other, but a lot of the time you could be flipping a coin.

This goes for morphology and lexicon at least. Phonology comes attached to cognates though, so "it was just introduced from somewhere" is usually a lot less convincing. Unless, of course, it is a phonological trait that doesn't have cognates… (E.g. going by LIV, Anatolian for example has no clear cognates for LPIE *sl-, *sm- at all, only a single uncertain example for *sn-, *sr-, and still only one cognate for *sw-.)

And established experts will flip the coin so that it comes down "things are how we thought they were, mostly" as often as they can, until proven otherwise, because it's a lot less hassle, and more methodologically respectable (it's better to be cautious than just to make things up).

This rule for caution unfortunately often transforms into an ex cathedra argument — plenty of "traditional" reconstruction solutions are made-up narratives as well.

[WeepingElf]

One argument in favour of PU being more conservative than PIE may be that the prehistory of PIE seems to have involved two language shifts (first from "Arwanbi", i.e. an unknown language, to Indo-Uralic in the Samara-Khvalynsk people, second from another unknown language which I call "Paleo-Pontic", in the Dniepr-Donets people when they were taken over by Khvalynsk intruders and became Sredny Stog)

At risk of starting up the everrunning argument again: I just don't see how you can happily say things like that. This is a very elaborate and specific sequence of events that you say "seems to have happened", for which we have no evidence at all, so far as I can see.

OK, I feel like you were a pedantic skeptic who likes to tear apart other people's (or at least my) ideas, but I shall try addressing all your objections in re. Thank you for grounding me, Salmoneus; I was once again soaring on wings of speculation

It certainly seems likely that at some point Yamnaya spoke PIE (or a descendent or an ancestor of PIE). It's likely that Yamnaya is primarily culturally and genetically derived from Khvalynsk. But it's also the case that Khvalynsk and Sredny Stog were closely culturally and genetically related groups, who traded and intermarried. They aren't, as it were, nations, with borders and authorities - indeed, Sredny Stog has been described as a cluster of cultural traits borrowed from Europe into a steppe culture, rather than a distinct culture in its own right.

Sure, they weren't nations in the sense the English, the French or the Germans today are nations. Each of these archaeological entities may have been multilingual, the (unknown) linguistic map perhaps showing little similarity with the archaeological one. For instance, the cultural areas of indigenous North America show little relationship to the language families (of which there are far more).

Later Yamnaya seems genetically to owe a lot to both group. We don't know which group spoke which language, or how Yamnaya began. Khvalynsk could have conquered Sredny Stog; or vice versa (borrowing the cool new kurgans they found); or, Yamnaya could just be a cultural package that developed across several different, neighbouring, related cultures at the same time. They may all have spoken PIE (or closely related languages)!
And I can see no evidence at all for your "Arwanbi". Even if you insist on Indo-Uralic, the easiest solution simply has that developing naturally among eastern european hunter-gatherers. I don't see why you'd need the steppe people to start with one language and then adopt another?
Language shifts do happen, but they're not common, so it's best not to assume them unnecessarily.

OK, let's forget about the whimsically named "Arwanbi" (if you haven't guessed by now, that is of course derived from R1b). If connections between genetic markers and languages are just as bogus as ones between archaeological horizons and language (and it seems to be like that), there is no reason to assume a particular language shift and a substratum language with a whimsical name like "Arwanbi", unless one finds linguistic evidence in favour of that.

which would have been conducive to massive typological restructurings, while the Uralic branch of Indo-Uralic seems to have evolved in a more tranquil setting where the language was just handed down from germane parents to germane children generation after generation.

That seems unlikely to me. They weren't noble savages!

Of course they weren't! I just contrasted two trajectories of languages of which one involved two language shifts and the other involved none, and concluded that the latter would be likely to be more conservative.

At least, the genetic patterns seem to point at such a scenario: Khvalynsk was R1b, Dniepr-Donets R1a and Uralic N1c in terms of Y-DNA haplogroups.

It's not so simple! You can't look at an entire culture and say that it "was" one haplogroup or another. They're all mixtures. Even today, Poland (in corded ware territory) is around 20% R1b. And the number of individuals tested from these cultures so far is tiny.

OK, I worded this badly. I did not really mean "Khvalynsk was R1b", etc., rather "Khvalynsk appears to show R1b as the most common Y-DNA haplogroup", but that would be a lot to type. Indeed, all of western Europe has plenty of R1a and all of eastern Europe has plenty of R1b, and we have so few samples of archaeological cultures tested that we can't really say much.

And while we're on pedantry: of course traditional PIE reconstructions fail to incorporate insights from Aquan, because Aquan is a fictional language!

If you want to insinuate that I am using my own conlangs as data in my linguistic reconstructions, I don't do that, or at least I am aware of this fallacy and try my best to avoid it; but I know that I am fallible - like everybody of us! Rather, it is up to me to make my conlang fit the assumptions I am making for it.

But at least, you admit your pedantry

I am growing increasingly doubtful of the idea that there were two early western expansions from the PIE homeland, both characterized by R1b, one leading to Anatolian IE, the other to Bell Beaker. These probably were two prongs of one movement into the Lower Danube region which then forked into Bell Beaker and Anatolian. So one earlier movement to the west, carrying the "nobleman" haplogroup R1b, and one later one in all directions, carrying the "common man" haplogroup R1a. And "Aquan", the hypothetical language of the Bell Beaker people, would be a sister of Anatolian.

So far as I'm aware, this isn't true.

For a start, I'd caution against using "Bell Beaker" as a family name, because we don't understand how the cultural phenomenon of beakers worked.

OK. I still see contradictory statements made about the origin of the "Bell Beaker culture", including ones which flatly deny that it is anything else than just a fad for a particular style of drinking vessel! Indeed, the "Bell Beaker people" perhaps weren't an autochthonic population but a class of itinerant merchants or whatever, in which case their language would have had as much impact on the linguistic map of Copper Age western Europe as Romani on that of modern central Europe: nil. Thank you for shooting down another unhealthy flight of fancy of mine!

But in terms of the spread of R1b through western europe, this is believed to have come from an expansion beginning in the western periphery of the early Corded Ware Culture - the Rhine, etc. Rhenish CWC could be from Yamnaya moving up the Danube and then adopting CWC innovations in Germany... but I don't think there's evidence of that. Instead, they look to just be the western end of the CWC expansion. From there, they rapidly expanded throughout France, Germany, the British Isles, and a little way down the Danube - but not, originally, into Iberia.

How do you get an R1b expansion out of an R1a CWC? Well, we need to remember we're dealing with the remain of a tiny fraction of the individuals, disproportionately high-status males. There's a big 'hidden population' that don't show up.

Yes. We may be dealing with some sort of founder effect.

But here's three scenarios for you:

- The Yamnaya and CWC phenomena were multiethnic cultural and perhaps political-military alliances. Different ethnic groups were dominant in different areas, and those dominant groups are the people we mostly find the remains on. As one set of tribes comes to dominance in the CWC area, a rival/ally tribe, let's call them for sake of argument "the Basques", decide to find their own territory next door. They do very well and expand massively.

- There's only one ethnicity, but a clear class difference. In the CWC, with the traditional Yamnaya environment and culture no longer applying, there's, as it were, a culture-wide "revolution" overthrowing the ruling class; deposed princes gather together at or beyond the western border and restore their warlike ways.

- Or, perhaps most likely, the whole of this society had several bloodlines in it, and coincidentally, in the course of a massive, rapid expansion in which small pioneer groups may have been isolated for significant periods, different bloodlines came to the fore in different areas. I think it makes most sense to see Yamnaya has actually having been R1a "on average", an average which bore out across most of the CWC area (small pockets of R1b could easily be bred out through intermarriage with neighbours); but there would have been some R1b families. The Yamnaya aristocracy was mostly R1b, and, perhaps purely by coincidence, the cluster of pioneers on the western fringe of CWC, who would go on to take over most of western europe, just happened to also be lead by an R1b family.

All three scenarios are possible, and the last one perhaps most likely.

It's worth bearing in mind that it only takes one guy to introduce a dominant lineage, and these were mobile societies, so who knows. Russia, for example, is not very R1b now - but Tsar Nicholas II was R1b! When you have a continent conquered by Genghis Khan and his buddies, you can probably have large areas of genetic history just work out by luck, depending on where whichever particular friend-of-Genghis who dominated that area happen to originate from...


It's also worth noting that western european R1b, the type spread by the non-Iberian "Bell Beaker Folk", is (almost) all from a specific , young R1b linneage that hasn't been found in Yamnaya (yet). So in any case, that whole-of-western-Europe lineage pretty much popped out of nowhere (i.e. developed rapidly from a small founder population) - it's not, for example, a grabbag of random R1b guys from across the CWC territory getting together, it's all one... well, family.

I see. You are far more knowledgeable in genetics than I am, so I can only take your word. Thank you for clearing up misconceptions of mine.

Perhaps I should simply give up on this kind of speculations and focus on things that are less bound to such fallacies, such as my music, my book on progressive rock, my science fiction and fantasy stories and my conlangs.

[WeepingElf]

It also seems to me that Proto-Eskimo-Aleut suffers from a similar problem as Proto-Chukotko-Kamchatkan: it is basically Proto-Eskimo with some fudging to get Aleut to comply with it.

On that note, you might be interested in the Comparative EA Collection from Alaska Native Language Archive (recently brought to my attention). I've not seen any good single overview of PEA, but there seems to be quite a bit more work done than on PCK.

I see. So PEA is on better ground than PCK. I know too little of these matters, but I have noticed that while there seems to be a pretty good reconstruction of Proto-Eskimo, there doesn't seem to be one of PEA.

Chronology could also be the reason why PIE looks so "un-Mitian". The median North Asian language profile with palatals, uvulars, no clusters, no ablaut etc. looks perhaps "old", but could have regardless come about more recently than PIE was spoken: some 3k to 5k years would be easily enough.

Yes, we may deal with a Sprachbund that emerged only long after "Proto-Mitian" has broken up (if it had ever existed), and failed to include IE, which is of course a geographical outlier.

I don't mean to suggest a Sprachbund at any one particular time (though that's a possibility), more that there may have been general typological trends over millennia. Possibly mediated by not just diffusion of features within Central Asia / Siberia, but also increasingly strong "Paleosiberian" / "Amerind" influence towards the east. Most of Europe by contrast got overran by IE (and para-IE?) quite early, probably with only scraps of the previous languages remaining by 1000 BCE.

I understand. I have realized long ago that neighbouring language families tend to be typologically similar (not always, but often enough). And these patterns may change over time. But in Europe, most of the old linguistic diversity has indeed been overrun by IE and Uralic, and we now only have Basque and the Caucasian languages left alive, though we at least have written records (though mostly unintelligible) of Iberian, Suroccidental (usually called "Tartessian", but it doesn't seem to have anything to do with ancient Tartessos, which seems to have been Celtic), Etruscan (probably a Late Bronze Age incursion from Asia Minor, but this is far from certain), Minoan, Eteocretan (which may be a descendant of Minoan, but we can't be sure), Cypro-Minoan (not necessarily related to Minoan, even if the scripts are) and Eteocypriot (same thing as with Minoan and Eteocretan). These residues give some impression of the pre-IE linguistic landscape of at least Mediterranean Europe (while the area north of the Alps is completely blank, save a few unintelligible inscriptions which are usually held to be ciphers of known languages).

For an example, let's dig into a few features from Anderson (2006), Towards a Siberian linguistic area:

• Four nasals /m n ń ŋ/; in most languages with a ban on initial /ŋ/ (exceptions include Tungusic, northern Samoyedic, Nivkh and CK), and in several with /ń/ being marginal or a recent introduction (Turkic, Mongolic, Yeniseic, CK, and though Anderson doesn't mention it, also northern Samoyedic — initial /ń/ is mostly and medial /ń/ always from *n + front vowel, while old *-ń- > -j-).
The second fact probably indicates that /ŋ/ generally comes from some sort of earlier consonant clusters like *ng. Of course, the pattern could be very old (it reconstructs already into PU) and IE could have just lost this. On the other hand, I would think any reconstruction along the lines of PU *ŋ ~ IE *g < *ɠ looks unlikely and does not explain this distributional constraint. — /ń/ ties into the general palatals question, but at least an independent medial it could be very easily rewritten as *nj.

• SOV plus extensive verb nominalization
Unmarked, extends much wider also, generally not a strong genetic signal.

• Extensive case systems including prolative, (al)lative and comitative
Large "semantic" case systems usually come simply from postpositions + agglutination. Proposed deep cognates among case markers seem to mostly involve the grammatical cases, and these markers find their IE counterparts easily enough.

None of these features are especially unusual, though they take part in shaping the regional "flavour" of the languages of Siberia. PIE may have lost the velar nasal by a change like *ŋ > *ng (or just *ŋ > *n, at least in initial position), and the palatal nasal together with the whole palatal series. SOV is reconstructed for PIE, though not with very much verb nominalization. The Late PIE case system is fairly substantial, but not as extensive as in many Siberian languages, and only about half of it is cognate in Hittite, so it may have been limited to nom., acc., gen., dat. in Early PIE.

[Pabappa]

speaking of North Dravidian, I wondered if anyone here knows more about it and could answer me a question I had when I was browsing some of the literature.

In The Languages and Linguistics of South Asia: A Comprehensive Guide, Suresh Kolichala notes McAlpin's hypothesis that Proto-Dravidian must have actually had uvular stops because 'North Dravidian' shows evidence of it and Malto indeed has a distinction between /k/ and /q/ (and apparently /g/ and /ɢ/ as well). Kolichala objects that McAlpin "hasn't shown" why we can't just propose a change k > q/_{e,o,a} for North Dravidian since the uvulars only appear before mid and low vowels. But this seems bonkers to me. /k/ appears before mid and low vowels too in Malto all the time. For /a/ we have kaṛme 'waist', kaje 'wash', kase 'dirt (on the body)', kanḍo 'seat', etc. In addition, proposing a sound change k > q/_{e,o,a} to explain the distribution would also be very odd, considering that sometimes /qe/ and /qo/ in Malto correspond to /ki/ and /ku/ in other Dravidian languages. qōre corresponds to Telugu kūru, with the same meanings of 'to be in excess, excessive', and we have Malto qerce 'to scrape' vs. Parji kir- 'to scratch'.

Why is Kolichala still skeptical? Is there something I don't know about vowel changes in Dravidian that would still explain this? Because the evidence that McAlpin 2003 presents seems pretty conclusive.

I agree that unconditional ke ka ko > qe qa qo is unlikely. But Malto is a third order branch of Dravidian so if it is the only one with uvulars then we would need an explanation of how the uvulars disappeared independently at least three separate times in the other Dravidian languages.

edit: Wikipedia suggests that McAlpin's theory solves this problem by redrawing the tree with Kurukh-Malto as one branch, Brahui as a second branch, and the whole of the remainder as a third branch. If so, then they would need to show evidence that this tree is the proper one before I'd be convinced that the /q/ was original to the shared parent language.

I'd like to see others' opinions.

[mae]

-

[Pabappa]

Uvulars are normally prone to loss.

You can find examples of superstrate influence like Hebrew /q/ > /k/, but I dont think that the particular sound change /q/ > /k/ is likely to happen in a language surrounded by others of its kind. If anything it seems that it's more likely to shift to /ʔ/ than to /k/. So, proposing an unconditional /q/ > /k/ in three separate branches of the family just to explain the situation one outlying language really needs an in-depth explanation if you expect people to take it seriously.

That's not to say I'm writing your idea off .... just that neither of the two explanations seem to have any evidence in their favor, and without looking at every word in every language that has a /k/ in it I'm not ready to form an opinion as to which of the two is more likely, or if there is some third option we haven't thought of.

[Vijay]

Aren't dorsals further back than velars pretty rare in the region of the world in question? Wouldn't that suggest that they're likely to disappear there?

[mae]

-

[Pabappa]

By the same logic, we could say that since uvulars are rare in India, its most likely that the parent language didnt have 'em and that Malto must be the outlier.

/q/>/k/ is not a common sound change outside of superstrate influence, so perhaps one language influenced the others.... but if the entire Dravidian family started with uvulars, what is the superstrate that caused them to lose them? It seems difficult to propose that it's IE influence when the only language that retained the original is deep within IE territory.

Dravidian has not been only surrounded by other languages "of its kind" in thousands of years, so that part of your objection is irrelevant to this case.

[/quote]What other language families were there? Proto-Dravidian is believed to have been spoken around 2500 BC .... are you supposing that Indo-Aryan was present even then?

2. McAlpin does not propose three different instances of q > k, he only proposes one. Brahui is claimed to have *q > x, so the only instance of q > k needed is the one that produces the distribution of the rest of the Dravidian languages other than "North Dravidian".

If you can first prove that Kurukh-Malto is a branch coordinate with all of the others, rather than just a third-order descendant, then you can propose that the sound change only happened once. Are you prepared to delve into the arguments required to prove that?

3. Your skepticism against q > k as a typical change seems unfounded to me, since it's pretty common in languages. It's happened a dozen odd times in Austronesian, it happened in Northern varieties of Quechua, most of Mayan, some Arabic varieties, etc.

Quechua and Mayan are of course attributable to Spanish influence. Im not familiar with proto-Austronesian languages that have /q/ > /k/, but there is a phoneme sometimes reconstructed as */q/ that appears as a glottal stop in some languages and as /k/ in others. It's possible, but certainly not proven. Arabic did not do /q/ > /k/, but rather /q/ > /g/, which could not have passed through /k/.
[/quote]

[missals]

Quechua and Mayan are of course attributable to Spanish influence.

That's extraordinarily flippant, and unjustified. The q > k shift happened in a number of branches of Mayan and forms a deep and early isogloss in the family, and has nothing to do with Spanish. I don't know about Quechua, but I hope your're not just assuming any q > k shift must be due to Spanish.

q > k also happens in Cypriot Arabic, according to the Index Diachronica. q > k is a completely normal sound change that can easily happen without external influence.

[Znex]

Besides these instances, if we suppose as is a common hypothesis that PIE *k g gʰ are likely uvular/retracted velars and the PIE palatovelars *ḱ ǵ ǵʰ are true velars (whether because of relative and typological frequencies or the likelihood of palatal consonants retracting to the velum), we have the same sound change occur near unanimously.

[Vijay]

What other language families were there?

Austroasiatic, Sino-Tibetan, and (at least if we're counting Assam here) Tai-Kadai, and perhaps Nihali and Kusunda EDIT: I'm not really sure whether this is relevant, but Burushaski is spoken much closer to where Brahui is spoken today than any of these other languages.

If you can first prove that Kurukh-Malto is a branch coordinate with all of the others, rather than just a third-order descendant, then you can propose that the sound change only happened once. Are you prepared to delve into the arguments required to prove that?

IINM North Dravidian has been geographically isolated from the rest of the Dravidian family for thousands of years, so that doesn't seem problematic.

[Nortaneous]

*q > k also happens in many Athabaskan languages.

[Tropylium]

If you can first prove that Kurukh-Malto is a branch coordinate with all of the others, rather than just a third-order descendant, then you can propose that the sound change only happened once. Are you prepared to delve into the arguments required to prove that?

This is shifting the burden of proof — the claim that Kurukh–Malto is a third order descendant is not some kind of a default position, it needs defending itself (and at minimum, that should be second–order descendant; Malto alone is a proposed third-order descendant, but adding Kurukh bumps this down one node). Actually no really strong arguments have been made for a North Dravidian group (the same McAlpin article deals with this too): it's mainly "they're in the north and have additional back consonants", but this will fall apart if the uvulars cannot be shown to be some kind of a common innovation to the two groups. Claiming that they're probably innovative because typology is not going to cut it.

I think the situation is mostly inertia / philology-centrism. Dravidology does not seem to be really used to the idea that unwritten minor languages even could have anything substantial to contribute to Proto-Dravidian reconstruction. Though it'd be nice to also put together a detailed Proto-Kurukh-Malto reconstruction and see if that offers hints on anything (for another hypothesis: maybe it's *k > *q by default, but with retention under some conditions).

[Pabappa]

I see some good arguments here, some terrible, and some completely irrelevant. I made clear at the beginning that I wasnt taking a side, so I'm not going to be embarrassed to change my opinion. But both theories presented in the OP rest on evidence that I haven't seen, and that hasn't changed in this thread... so, I'm not changing my opinion. However, I'm not stressed about this, because, originally, I bumped this post as a favor to mae, as I saw it had been buried on the last page with no replies. I wont do that anymore.

[Tropylium]

originally, I bumped this post as a favor to mae, as I saw it had been buried on the last page with no replies. I wont do that anymore.

I dunno, you seem to have been successful in sparking some conversation in the process.

[Pabappa]

Feel free to go on, but it seemed like all of the conversation was aimed at me.