Zompist Bboard Again

gach wrote: ↑Sun Oct 07, 2018 5:54 pm

Salmoneus wrote: ↑Sun Oct 07, 2018 4:11 pm No, I don't have a citation for that '95%' - I don't even know how you'd operationalise these terms for a rigorous study. It was impressionistic, a way of saying 'not universal, but close to it'.

The point being that even without explicit numerical values, stating that something is "close to universal" is a strong claim and requires sturdy evidence to back it. Otherwise it doesn't count as solid scientific reasoning.

Ahh, your 'solid scientific reasoning' returns!
Otherwise known as "what I say can be accepted as informal fact, but any statement by you should be treated like a legal accusation and be proven beyond reasonable doubt with citations and footnotes, even for expressions, generalisations, figures of speech and metaphors not actually essential to any point under discussion." This is wearying beyond belief.

The 8% and 5% proportions of eastern genes are clearly there but at the same time they don't form nearly the majority of the Finnic genetic ancestry. They are the part which you can identify with the people who brought the Uralic languages to the shores of the Baltic Sea. The language certainly came from the east. The rest of the genetic ancestry of the people points to other routes into the country, though. These people must have spoken other languages which their descendants, including the present population, have since lost. That right there is language shift, a large proportion of the population acquiring a first language which was foreign for their ancestors.

...did you literally not read my post? You're just repeating things I already said. I mean, compare for example your "they don't form nearly the majority" with my "I'm not saying the language always matches the majority genetic heritage of a group". It's hard to believe you're missing the point that much accidentally - is this some sort of mockery I'm not getting?

Also, no - this scenario does not imply "a large proportion of the population acquiring a first language which was foreign for their ancestors" in any meaningful way. Unless you call, say, a Finnish person marrying a Swedish person and teaching their children Finnish a "language shift". Because again, ancestry is not a matter of monolithic races. You have two parents, and four grandparents, and so on: so you can 'acquire a language your ancestors didn't speak' while at the same time 'continuing to speak the language your ancestors spoke', if your ancestors spoke multiple languages. NOTA BENE: saying you have 8% uralic genes does not mean that 8% of the population is Uralic and the other 92% are non-Uralic and must have shifted their language. It means that on average 8% of your ancestors were Uralic (well, not exactly, but that's a lot closer). In the case of the Finns, Y-DNA demonstrates that the vast majority of Finns are directly descended from the people who brought Finnish to Finland (and from other people). The people who came to Finland didn't have a language-shift; their children, learning their father's language, didn't have a language-shift; their children, learning their father's language, didn't have a language shift; their children, learning their father's language, didn't have a language shift.

Let's give a simplified little example. Let's say a population starts out with 10 orange people and 90 purple people. Let's say that the orange people reproduce at a rate of 1.8 children per person, always breeding with a purple person. Let's say that the purple people reproduce at a rate of 0.9 children per person (i.e. half the rate), always breeding with a purple person. Let's run this simulation for five generations. What happens? Well, we end up with 189 "orange people" and 53 "purple people". Let's say that everyone learns the language of their father. So, we end up with a population in which 78% of people speak Orange, and only 22% of people speak Purple. But what do their autosomal genes look like? The purple people are still 100% purple, but the "orange people" are by now only just over 3% "orange" in their genes. So we have a population in which 80% of people speak, let's call it, "Finnish", and every one of them learnt that language from birth from a parent, and yet they are only 3% autosomally let's call it 'Siberian" in their ancestry. And yet, one could hardly reasonably say that any "language shift" had taken place - everyone just learnt the language of at least one parent. Of course, in reality the situation would have been a little more complicated in terms of intermarriages and bilingualism and whatnot, and it presumably happened more gradually over time, but you get my point, hopefully: where one population sexually outcompetes another (as the dominance of N1 lineages in Finland proves was the case there), a small genetic contribution can lead to the dominance of one language over another even with pure parent-to-child linguistic inheritence. [as for that 20% relict population - in this simulation they're only 7% of the population two generations later. Presumably you wouldn't call the linguistic conversion of 7% of the population a 'language shift'...]


So, while for the purposes of this discussion we've define language shift to exclude the case where there is ANY significant genetic influence, it should also be obvious that the Finnish case doesn't even meet the much broader definition that calls 'language shift' whenever there is anything less than perfect parent-to-child transmission.

I don't think that anyone is denying the importance of migrations to the spread of languages. The real data, as I'm aware, simply suggests that you can't blindly identify your genetic and linguistic ancestors as the same people.

Nobody has claimed that you can. But I would say that it's misleading to think of 'your ancestors' as being just one group of people: again, ethnicities in reality are not monolithic races. And my point is that it's extremely rare for your (collective) linguistic ancestors not to be at least some of your (collective) genetic ancestors. As opposed to your/WE's contention that knowing your genetic ancestors is almost insignificant in knowing your linguistic ancestors. On the contrary, while I accept that from seeing a, say, genetically 50% IE and 50% uralic population, you can't be sure whether they speak IE or Uralic, you can be very confident that they probably speak one or the other, and not, say, Austronesian. Not 100% confident, but at least 95% confident... Certainly not, as WE would have it, almost 0% confident.

I think it's quite unnecessary to invoke ideas such as "pure and uncontaminated" genes or "white european credentials". Nothing good or productive lies that way and surely no one else has brought them up.

This is not the first time we've discussed these questions, and the context of this very thread is very visible here. I think it's entirely legitimate to point out the naive and overly essentialist assumptions that underpin these ideas.



EDIT: but whatever. No headway is likely to be made here.

@Sal
Your correctness and rudeness are at an unsatisfactory perfect balance

mèþru wrote: ↑Mon Oct 08, 2018 4:43 am @Sal
Your correctness and rudeness are at an unsatisfactory perfect balance

I accept that my first two paragraphs were overly frustrated in tone, for which I apologise; in my defence, they were rude in response to rudeness, and sometimes there doesn't seem to be any recourse short of urgent, honest bafflement in attempting to resist a determined mischaracterisation.

For what it's worth, there's an interesting paper on the supposed "substratum" of Sumerian by Gonzalo Rubio. (It's worth reading solely for the last sentence of footnote 1- ouch).

dewrad wrote: ↑Mon Oct 08, 2018 7:45 am For what it's worth, there's an interesting paper on the supposed "substratum" of Sumerian by Gonzalo Rubio. (It's worth reading solely for the last sentence of footnote 1- ouch).

How does one demonstrate an unknown substratum in an isolate?!? The alleged substratum loanwords in units like Celtic and Germanic are suspected to be that because they lack cognates in other branches of Indo-European, and often contain phonemes that were either rare or absent in PIE (i.e., *a and *b). And even that is doubtful and controversial. With a language isolate such as Sumerian, which has no known cognates anywhere, how can someone say that a given word "can't be native"? The only case where we can say that the words are not native is when they can be shown to be loanwords from a known other language (e.g., the many Latin loanwords in Basque), but this does not seem to be the case here.

WeepingElf wrote: ↑Mon Oct 08, 2018 8:43 am

dewrad wrote: ↑Mon Oct 08, 2018 7:45 am For what it's worth, there's an interesting paper on the supposed "substratum" of Sumerian by Gonzalo Rubio. (It's worth reading solely for the last sentence of footnote 1- ouch).

How does one demonstrate an unknown substratum in an isolate?!? The alleged substratum loanwords in units like Celtic and Germanic are suspected to be that because they lack cognates in other branches of Indo-European, and often contain phonemes that were either rare or absent in PIE (i.e., *a and *b). And even that is doubtful and controversial. With a language isolate such as Sumerian, which has no known cognates anywhere, how can someone say that a given word "can't be native"? The only case where we can say that the words are not native is when they can be shown to be loanwords from a known other language (e.g., the many Latin loanwords in Basque), but this does not seem to be the case here.

I'm not an expert on Sumerian, but I was under the impression it was pretty widely accepted that certain Sumerian words, including quite a few city names and words pertaining to urban professions, were not Sumerian, in the former case because they lack any known Sumerian etymology and in the latter case because of atypical syllable shapes. (For the former I believe Kramer puts forward the idea; for the latter, I'll have to look up my citation but I recall reading it.)

Salmoneus wrote: ↑Sun Oct 07, 2018 4:11 pm[I know, Tropylium's going to come in and yell at me for impugning the white european credentials of the finns. Whatever.]

I had a joke I thought of adding at this point, but maybe I'll rather agree with gach that this barb seems unproductive. Unwarranted, even.

On Tambov et al. re: Finnic peoples, I don't really disagree on much else than on ignoring the Sami substrate in Finland, which is surely a major reason why Finns consistently show more Siberian ancestry than Estonians. (There is also ongoing talk of a pre-Sami and pre-Finnic, but regardless Uralic, substrate in the range of southern current-day Finland, the Karelian Isthmus, and the Gulf of Finland drainage from there on east, tentatively identified with the "Chuds". So far though it's pretty hard to say what their genetic contribution to the big picture could have been, but likely not zero.)

A much bigger issue on the other hand is the implicit treatment of Ob-Ugrians as "pure Proto-Uralians", when these peoples are built on an obvious paleo-Siberian substrate as well. So their group K9 Siberian ancestry marker probably ends up mainly signifying some pre-Uralic population movement and not the Proto-Uralic one. Indeed, in their slightly more detailed component analyses, the western part of K9 turns out to be almost entirely composed of the K11 component, which is highly concentrated in the Southern Sami and pretty much completely absent from anyone east of the Vepsians. In other words: K9 in the 9-or-10-cluster analysis has a deep genetic split to it and most likely cannot be as young as Proto-Uralic.

I think you're probably aware of this by now, but at least to recap anyway — genes evolve a lot more slowly than languages, and no single autosomal component can possibly match 1 : 1 with a homogeneous Proto-Whatsit speaker group. What they can mean is that the Proto-Whatsits had already inherited a particularly large proportion of some genetic component (this can be close to 100%), but all the genetic clusters that turn up in analyses like these had already evolved some tens of thousands of years ago. They must have been already present also in other neighboring populations of the time; any linguistic expansion can at most spread them further, beyond their original range. The linguistic descendants of the Proto-Whatsits would then end up with either a small portion of the component (if mixed with populations that lacked it) or perhaps with a larger one (if mixed with populations that had a larger share still).

Salmoneus wrote:The one big exception here is apparently the Hungarians, who, it seems, do not have any noticeable Uralic heritage other than their language. It's possible this was originally diluted alongside their cultural Altaicisation, but the main culprit is probably the late invasion of Hungary, when the small invading population was able to use the mediaeval state institutions to impose their language more widely than their semen (and where at the same time the invaders were soon culturally dominated by and genetically pervaded by the wider European cultural elite).

I'd bet on the whole gradual trip/expansion from Magna Hungaria as also doing more for genetics/language desyncing than the initial movement into the steppe lifestyle. Or maybe that's what you also mean by Altaicisation? since this exodus was done under the pressure of new Altaic steppe groups, while the initial lifestyle shift was made already at a stage when the steppe was still primarily Iranian territory.

Salmoneus wrote:And my point is that it's extremely rare for your (collective) linguistic ancestors not to be at least some of your (collective) genetic ancestors. As opposed to your/WE's contention that knowing your genetic ancestors is almost insignificant in knowing your linguistic ancestors. On the contrary, while I accept that from seeing a, say, genetically 50% IE and 50% uralic population, you can't be sure whether they speak IE or Uralic, you can be very confident that they probably speak one or the other, and not, say, Austronesian. Not 100% confident, but at least 95% confident... Certainly not, as WE would have it, almost 0% confident.

This logic works well enough for modern people, but it stops hard once we go outside the timeframes where we can expect "IE" and "Uralic" to be widely spread. (Lots of IE speakers have a sliver of mesolithic West European Hunter-Gatherer inheritance, but this does not allow us to conclude that there was a 5% chance that WHGs spoke Indo-European.)

Salmoneus wrote: ↑Sun Oct 07, 2018 7:44 pm

I don't think that anyone is denying the importance of migrations to the spread of languages. The real data, as I'm aware, simply suggests that you can't blindly identify your genetic and linguistic ancestors as the same people.

Nobody has claimed that you can. But I would say that it's misleading to think of 'your ancestors' as being just one group of people: again, ethnicities in reality are not monolithic races. And my point is that it's extremely rare for your (collective) linguistic ancestors not to be at least some of your (collective) genetic ancestors. As opposed to your/WE's contention that knowing your genetic ancestors is almost insignificant in knowing your linguistic ancestors. On the contrary, while I accept that from seeing a, say, genetically 50% IE and 50% uralic population, you can't be sure whether they speak IE or Uralic, you can be very confident that they probably speak one or the other, and not, say, Austronesian. Not 100% confident, but at least 95% confident... Certainly not, as WE would have it, almost 0% confident.

The problem is that when genetic populations mix, the language that ends up being spoken can be the language of any of the populations in the mix. It does not necessarily have to be the language of the dominant population. There are many cases in recorded history where the dominant population, the conquerors, ended up speaking the language of the conquered populations. The Goths in Spain and the Franks in France ruled there for centuries, but they eventually ended up speaking the local Romance language. And a very different case is that of the Hungarians. There the population ended up speaking the language of the conquerors. But there are hardly any genetic traces left of any conquering Uralic population in modern Hungarians.

As for the Uralic languages, Yukaghir and Eskimo-Aleut seem to have some vocabulary in common with Uralic. However these commonalities are also shared with PIE. And the populations that speak PIE do not have the Siberian genes that Uralic speakers have. Also, those Siberian genes were introduced into Europe at a later time than when PIE spread. So these commonalities could not have come with the Siberians. The most logical conclusion for me is that the Siberians moving into North-Western Europe adopted the existing Uralic languages and those that did not move into NW-Europe adopted a lot of Uralic vocabulary.

Tropylium wrote: ↑Mon Oct 08, 2018 12:18 pm This logic works well enough for modern people, but it stops hard once we go outside the timeframes where we can expect "IE" and "Uralic" to be widely spread. (Lots of IE speakers have a sliver of mesolithic West European Hunter-Gatherer inheritance, but this does not allow us to conclude that there was a 5% chance that WHGs spoke Indo-European.)

But I would not exclude the possibility that WHGs spoke a language that is related to PIE. Forni's Basque as an IE language is certainly wrong, but I consider Juliette Blevins's more sophisticated theory of proto-Basque being related to PIE as something that is at least possible.

WeepingElf wrote:So the Neolithic Near East probably was more diverse than the Bronze Age one.

Sure. However, the Near East (NB: in which term I include Mesopotamia — and definitely do no include the Balkans, an usage that is completely alien to me) is mostly not a refugee zone where we can expect multiple isolates to linger on for millennia, each in their own mountain valley. As a "quilt", it is fairly dynamic, showing repeated language dis- and replacements.

So consider one snapshot at let's say 6000 BCE. Maybe there will be 5 language families to be found in the Near East, maybe 15. Regardless, if them being millennia-old isolates is ruled improbable, where would all these have come from? There seem to be only three reasonable directions of entry:

1. a Balkan–Anatolia route (as demonstrated by Greek, Armenian and probably Anatolian)
2. an Iran route (Turkish, Kurdish, possibly Sumerian)
3. an Egypt route (probably Semitic)

Note that Arabic does not demonstrate an entry from Arabia: pre-Islamic Old Arabic arose mainly in the Levant/Arabia borderlands around Jordania and was never an "intruder" there. The hypothetical connection of Hurro-Urartian with NEC does not really point towards entry from the Caucasus either: the direction of entry could also have been the opposite. Or even: given that Kartvelian is "in the way", even something more roundabout, compareable to the situation of Ossetian versus Mitanni.

Route 1 seems to have been for some reason unlikely to make much of a splash all the way in Mesopotamia and southern Levant. Route 2 has been pretty successful at times, and it could perhaps feed in languages of almost any affiliation since it plugs into the Central Eurasian steppe highway, but maybe you need horses to really get good leverage out of it. Hence I assume it is route 3 that has been relatively more important in prehistoric, pre-Semitic times.

I can think of a different issue, though. How major was the time difference between the Northern African spread of Afrasian (Proto-Semito-Egypto-Berber or whatever), and the spreading of Proto-Semitic proper? This could have been somewhat brief, so that even route 3 being the most important linguistic entrypoint in the early Neolithic would not imply pre-Semitic newcomers to have likely been any other flavor of Afrasian either.

(OK, this sounds terribly clunky, I hope you're managing to follow…)

WeepingElf wrote:Greenberg's Eurasiatic is based on mass lexical comparison, and therefore highly questionable, hardly better than Amerind. However, most (not all) of the language families in this group share those famous pronouns which led to the nickname "Mitian", and five of them (IE, Uralic, Turkic, Mongolic and Tungusic) belong to the "core" of the Nostratic hypothesis.

Yes, I'm using "Eurasiatic" more in the Bomhard sense, roughly as a synonym for "core Nostratic".

Tropylium wrote:pre-Islamic Old Arabic arose mainly in the Levant/Arabia borderlands around Jordania

Never knew that it is more likely Jordanian than Hejazi, but after you posted I looked up and saw you are right. This changes a lot of my perceptions on ancient Semitic history

Salmoneus wrote: ↑Sat Oct 06, 2018 11:06 am
You're very keen on crazy quilts, aren't you? It's not an expression I've ever heard before. Is a crazy quilt actually a thing, or are the quilters in your area just notoriously erratic?

Your common or garden quilt is made up of patches that are in particular patterns, usually with more-or-less poetic names: Flying Geese, Wedding Ring, Drunkard's Path, Churn Dash, Double Nine-Patch. A crazy quilt is where there either aren't patches at all, or each patch is made up of pieces basically at random, the original idea having been to use up the pieces that can't even be eeked into conventional patches. In addition, in crazy quilts the seams are usually embellished with embroidery and it's common for the individual pieces to also have extra decoration in the form of beads, buttons, and/or more embroidery, and the base fabrics are often richer.

Tropylium wrote: ↑Mon Oct 08, 2018 1:45 pm

WeepingElf wrote:So the Neolithic Near East probably was more diverse than the Bronze Age one.

Sure. However, the Near East (NB: in which term I include Mesopotamia — and definitely do no include the Balkans, an usage that is completely alien to me) is mostly not a refugee zone where we can expect multiple isolates to linger on for millennia, each in their own mountain valley. As a "quilt", it is fairly dynamic, showing repeated language dis- and replacements.

Certainly, the Near East received many immigrations and internal movements all over the time when it was inhabited by humans. But I am generally sceptical about large families before the Neolithic, except in areas which had only recently been cleared of ice caps - which is not the case in the Near East, except deep in the mountains of Greater Armenia.

So consider one snapshot at let's say 6000 BCE. Maybe there will be 5 language families to be found in the Near East, maybe 15.

Fine. Let's be conservative and say five. 15 seems a lot (though not impossible) to me.

Regardless, if them being millennia-old isolates is ruled improbable, where would all these have come from? There seem to be only three reasonable directions of entry:

1. a Balkan–Anatolia route (as demonstrated by Greek, Armenian and probably Anatolian)
2. an Iran route (Turkish, Kurdish, possibly Sumerian)
3. an Egypt route (probably Semitic)

Note that Arabic does not demonstrate an entry from Arabia: pre-Islamic Old Arabic arose mainly in the Levant/Arabia borderlands around Jordania and was never an "intruder" there. The hypothetical connection of Hurro-Urartian with NEC does not really point towards entry from the Caucasus either: the direction of entry could also have been the opposite. Or even: given that Kartvelian is "in the way", even something more roundabout, compareable to the situation of Ossetian versus Mitanni.

The connection between HU and NEC is highly controversial, and I don't know what to think of it; also, it seems equally (if not more) plausible to assume that Pre-Proto-NEC moved into the Caucasus from Greater Armenia rather than the reverse. We indeed have Kartvelian more or less "in the way", but that, too, may have moved into the Caucasus later from somewhere else, maybe Anatolia. The unknown language of the LBK culture may have been from the same origin, judging by the predominance of the Y-DNA haplogroup G2a in both LBK and today's Georgians, but such language-genetics correlations are not without risk.

Route 1 seems to have been for some reason unlikely to make much of a splash all the way in Mesopotamia and southern Levant.

Greek, Armenian and probably Anatolian moved this way, but in Neolithic times and earlier, most traffic on this route seems to have been in the other direction. Also, as you observe, none of the three branches of IE that moved from the Balkans into Anatolia ever went further into Mesopotamia or the southern Levant (but that may simply have been due to the fact that these areas were already occupied by more advanced civilizations which successfully resisted their immigration).

Route 2 has been pretty successful at times, and it could perhaps feed in languages of almost any affiliation since it plugs into the Central Eurasian steppe highway, but maybe you need horses to really get good leverage out of it. Hence I assume it is route 3 that has been relatively more important in prehistoric, pre-Semitic times.

Fair. The Iranians had horses, so they may have gotten into places where earlier peoples without horses could not go. If Elamite is related to Dravidian, though, the most likely location of their common ancestor would be either in the Iranian highland - or on the northern shore of the Indian Ocean somewhere between the Persian Gulf and the Indus delta.

I can think of a different issue, though. How major was the time difference between the Northern African spread of Afrasian (Proto-Semito-Egypto-Berber or whatever), and the spreading of Proto-Semitic proper? This could have been somewhat brief, so that even route 3 being the most important linguistic entrypoint in the early Neolithic would not imply pre-Semitic newcomers to have likely been any other flavor of Afrasian either.

The closest relative of Semitic is probably Egyptian, so it makes sense to assume that Semitic moved in from Egypt some time around 6000 BC. The third prong on this fork would have been Berber. The "Northern Afrasian" protolanguage in Mesolithic Egypt probably moved in down the Nile from the south (Sudan/Ethiopia), where we find Cushitic and Omotic, while Chadic moved west into the then green Sahara.

(OK, this sounds terribly clunky, I hope you're managing to follow…)

WeepingElf wrote:Greenberg's Eurasiatic is based on mass lexical comparison, and therefore highly questionable, hardly better than Amerind. However, most (not all) of the language families in this group share those famous pronouns which led to the nickname "Mitian", and five of them (IE, Uralic, Turkic, Mongolic and Tungusic) belong to the "core" of the Nostratic hypothesis.

Yes, I'm using "Eurasiatic" more in the Bomhard sense, roughly as a synonym for "core Nostratic".

I understand. Bomhard uses it this way, and that's fine. As for Greenberg's method, I once said about it that his lists don't say "These languages are related" but ask the question "Are these languages related?" which needs to be elucidated by means of the conventional comparative method. And of course, asking the right question is the first step towards finding the answer, and it seems as if Greenberg has asked the right question here, even if we don't know yet whether the answer will be positive.

What if you think Egyptian's relation to the other two is problematic and only have Semito-Berber? How would that change the analysis of the 3 pathways?

WeepingElf wrote: ↑Tue Oct 09, 2018 1:56 pmGreek, Armenian and probably Anatolian moved this way, but in Neolithic times and earlier, most traffic on this route seems to have been in the other direction. Also, as you observe, none of the three branches of IE that moved from the Balkans into Anatolia ever went further into Mesopotamia or the southern Levant (but that may simply have been due to the fact that these areas were already occupied by more advanced civilizations which successfully resisted their immigration).

Yes, and this difference in social capital was likely there also earlier in the Neolithic.

WeepingElf wrote:The "Northern Afrasian" protolanguage in Mesolithic Egypt probably moved in down the Nile from the south (Sudan/Ethiopia), where we find Cushitic and Omotic, while Chadic moved west into the then green Sahara.

I just now had a recollection reading something about this recently. I think it was this 2016 paper by Dimmendaal, where he points out that the clearest "green Sahara" entry route (along the ancient "Yellow Nile") is home to several Nilo-Saharan families, and has been probably been so for a long time. He ends up supporting instead Ehret's hypothesis for the formation of Chadic: that it used to have a more northern distribution, and was later pushed south by the desertification.

mèþru wrote: ↑Tue Oct 09, 2018 3:10 pmWhat if you think Egyptian's relation to the other two is problematic and only have Semito-Berber? How would that change the analysis of the 3 pathways?

If you want to assume a Semito-Berber group, there would be at least two reasonable homelands to consider:
– along the lower Nile, with Egyptian arriving later from the upper Nile and cutting the two off from each other
– in the Levant, with Berber being the result of an earlier Arabic-style spread event. The Nile valley could have been held firmly enough by (pre-)Egyptians that pre-Berber "stuck" only in the coastal areas to the west.

The first doesn't really change the odds much. It rules out para-Egyptian varieties from the picture, but some sort of para-Berber or Berber/Semitic intermediate branch would work just as well (for transmitting Nostratic-ish traits to Elamite or Kartvelian, or whatever it is we want to do with it). The latter case, though, would suggest that there could have been various Afrasian languages hanging out in Asia already well before the rise of Semitic proper, shifting the "center of weight" of the family a bit further northeast — same as the Chadic-from-the-north theory does. Really by this point it seems that Cushitic is the only branch of Afrasian that couldn't be easily derived from a homeland in someplace like the lower Nile. Even that is probably doable — though if Cushitic also came from the north, I'd expect it to have a clear tree structure.

But I also don't really know how plausible any of the smaller Afrasian groupings are to begin with. I have gotten the impression that at least Semitic-Berber is mostly based on vocabulary, and some of it could be just old loanwords from e.g. Phoenician that are misanalyzed as cognates.

From what I understand, the Phoenician loanwords are pretty identifiable in Berber, especially as the Tuaregs and other groups had little contact with the Phoenicians.

I remember reading a paper years ago by Laurent Sagart that argued pretty convincingly for Austro-Tai, specifically Tai-Kadai as a primary branch of Austronesian, on par with the other Formosan language groups, though the location of this paper is now lost to time and the ravages of a shitty memory. What do y’all think of this hypothesis?

I consider Austro-Tai pretty solid. I haven't looked at the details of the reconstructions enough to say at what exact point Tai broke off from the others ..... archaeology might be as helpful here as linguistics is.

The author of that paper seems to have written newer ones in which he assumes Austro-Tai exists .... e.g. https://halshs.archives-ouvertes.fr/hal ... 30/documen . And http://halshs.archives-ouvertes.fr/halshs-00104725 might be the original you wanted .... if not, it can probably be found on that website.

-

Pabappa wrote: ↑Sun Oct 14, 2018 10:07 am I consider Austro-Tai pretty solid. I haven't looked at the details of the reconstructions enough to say at what exact point Tai broke off from the others ..... archaeology might be as helpful here as linguistics is.

The author of that paper seems to have written newer ones in which he assumes Austro-Tai exists .... e.g. https://halshs.archives-ouvertes.fr/hal ... 30/documen . And http://halshs.archives-ouvertes.fr/halshs-00104725 might be the original you wanted .... if not, it can probably be found on that website.

I don’t think that the second paper is exactly the original that I wanted, but its hypothesis certainly looks identical to the one that I remember.

I haven't looked at the new reconstruction of Tai-Kadai, but the concept seems promising based off of what I know of Baiyue culture and its proposed links with Austronesian.