The pre-Columbian distribution of languages in the Americas is rather different from what we see in the Old World. In Eurasia, Africa, and Australia, we mostly see large areas occupied by families of clearly related languages – such as Indo-European, Afro-Asiatic, Niger-Congo, Malayo-Polynesian, and Pama-Nyungan . Much splotchier and localized in the Americas.
It seems that each of these large language families is the product of a massive, relatively recent population expansion. Not without admixture, usually, but some group developed some sort of advantage and rolled over its neighbors. This means in most of the world where we have information, the current inhabitants are quite different genetically from the people who lived there 10,000 years ago.
Europe experienced two expansions (first Middle Eastern farmers and then sort-of Siberian Indo-Europeans). Guys from Western Asia (and eventually some Indo-European speakers) flooded into an India originally occupied by people distantly related to the Andaman islanders. Bantu farmers swamped most of Africa’s hunter-gatherers. Somehow some genes and ideas traveled from India to Australia (about five thousand years ago, triggering an expansion that covered most of the continent. Chinese-looking people largely replaced the original inhabitants of Southeast Asia – the Cambodians seem to have a significant admixture from the previous tenants. Similar story for the Malayo-Polynesians in the Philippines and Indonesia. Chinesian farmers moved up in the world and mixed with altitude-hardy proto-Sherpas to give us Tibetans. Levantines carried languages and farming into Ethiopia (undoubtedly picking up the local altitude adaptations).
Although the Americas were settled more recently than any other continent, they have an older-looking language distribution – because there there just hasn’t been as much population turnover there in the Holocene. You don’t see the same dramatic population expansions. Until Columbus, of course. We have good reason to think this, because they’ve just sequenced a child’s skeleton from the Clovis culture, 11,000 years ago, and the kid is genetically similar to contemporary Amerindians. Genetic continuity over such a long period is not seen in most of the Old World – probably only in a few places.
Which is not to say there has been no expansions or population turnover in the Americas- just considerably less so than in most places of the Old World. The Na-Dene were late arrivals who had a decent expansion. There have been fair-sized expansions that originated among the descendants of the original settlers, like the Algonquians. But languages have had longer to differentiate in the Americas, because the slate hasn’t recently been wiped clean here. All the Amerindian language families other than Na-Dene and Eskimo-Aleut must form an ancient family that only smart guys like Greenberg, Ruhlen and Sapir could see.
Linguists usually assert that language and genes are entirely separate, but they had it backwards. Genetic expansions play a key role in language expansions. From current data, every time, in prehistory.
Why have there been fewer and less powerful population expansions in the Americas? I don’t know, but I can guess. Agriculture came later: that may matter, but it can’t be a necessary cause, because both Pama-Nyungan and Na-Dene managed decent expansions before agriculture. Maybe acquiring a winning advantage against other groups was difficult in the Americas because the population was so genetically uniform -much more so than in any other continent. Cultural innovations aren’t as good as genes at conferring a long-lasting advantage – you can copy them.
What else? Before the advent of the Indo-Europeans, almost all of Europe must have spoken related languages, originating somewhere in the Middle East or Balkans, carried by the LBK and Impressed Ware culture. Basque must be the last survivor of that family, rather like the tuatara. Theo Vennemann, can you hear me?
West Hunter 
Great post. Tupi seems to have spread quite far inland, probably by going up rivers, as far as I can see. “Cultural innovations aren’t as good as genes at conferring a long-lasting advantage – you can copy them.” In fact, there was relatively little cultural innovation of any sort in much of the southern cone, as far as the record goes. Little to no writing, simple artefacts, and no extant buildings. Interesting paucity of attainment, because they were sitting on the world’s most fertile soil in the River Plate estuary, yet it did nothing for them.
The important words there are “as far as the record goes”. We know very little about the people responsible for creating all that terra preta. We don’t even know whether it was intentional. Most artifacts don’t survive especially long in their climate. They don’t seem to have used writing on any medium that we’ve found, but… who knows?
Quite a lot is known about the creators of the terra preta; they are assumed to be Arawak-speaking people who practiced manioc agriculture and made manioc cakes on clay griddles, built villages in characteristic plaza-based forms, and used inclusive alliance-based social structural principles to create confederations larger than single villages. Comparative Arawak ethnology is relatively well developed, and it is widely believed that terra preta is a small part of the Arawak phenomenon.
They didn’t have horses. That seems to have been a big factor in Eurasian migrations.
… but they had armadillos!
It puzzles me why so many Amerindian linguists over the years have been so opposed to the idea that most Amerindian languages are descended from a common source(albeit over a long period of time). As early as 1918 Sapir pointed out the resemblences in pronouns.
Because Joseph Greenberg suggested this, and according to mainstream linguists, Greenberg can’t, Can’t, CAN’T be right, for reasons that I’m not quite sure of.
The reason for resistance to the idea of an Amerind language family is pretty simple: it can’t actually be demonstrated. Greenberg may well be right, but there’s absolutely no way to verify the claim. It’s just an interesting idea and nothing more.
Amerindian linguists are not ‘opposed to the idea’ in principle; the objection has always been to the flashy, superficial quality of Greenberg’s work. Here are some quotations from Lyle Campbell and William J. Poser, Language Classification: History and Method (Cambridge, 2008), ch. 9 Assessment of proposed distant genetic relationships, section 6 Amerind (p. 266-279)
“Most American Indian linguistic specialists believe it possible that most American Indian languages may be genetically related. The main difference is that they find Greenberg’s methods and evidence inadequate.”
“Greenberg introduced some language names into his classification which are not languages at all. For example, Membreño… is actually a person’s name, a reference (Membreño, 1897).”
“Greenberg mistakenly classified Uru-Chipaya and Puquina as related languages and put Subtiaba-Tlapanec in Hokan, though both are well-known errors repeated from earlier now corrected classifications of Sapir and Rivet.”
“Diffusion is a problem in Greenberg’s examples… Greenberg cited among his “Chibchan-Paezan etymologies” forms from four languages in support of his supposed ‘axe’ etymology, including Cuitlatic navaxo ‘knife’ (a loan from Spanish navajo, ‘knife, razor’) and Tumeo baxi-ta ‘machete’ (also a loan, from Spanish machete).”
“Greenberg’s putative Amerind cognate sets… involve very generous semantic latitude… for example:’ excrement/night/grass’, ‘ask/wish/seek/pleasure’, ‘bitter/to rot/sour/sweet/ripe/spleen/gall'”
“The problem of undetected morpheme divisions… is a frequent one in Greenberg. For example, he listed Rama mukuik ‘hand’ as cognate with words from several other American Indian language families which exhibit shapes like ma or makV, although ‘hand’ in Ram is kwi:k; the ma- is the second-person possessive prefix.”
“We mention here… cases where the forms presented are not just erroneous but do not exist. One such case is Greenberg’s Chitimacha lahi ‘burn’… the form does not exist; Chitimacha has no phonemic l.”
“Under the set labelled ‘kill’ Greenberg listed Choctaw ile ‘do,’ together with Hitchiti ili ‘kill’ (both Muskogean languages), but the ‘do of the Choctaw gloss is a scribal error… a misreading of the abbreviation for ditto.”
“Greenberg also mistakenly listed Atakapa uk ‘boil, ulcer’ with the set ‘boil (cook by boiling).”
I would add that the inflated weight given to Greenberg’s theories has overshadowed real advances such as Edward Vajda’s demonstration in 2008 that the Siberian language Ket is related to Na-Dene.
Few here, I suppose, are old enough to remember the angry reception that numerical taxonomy received fifty to sixty years ago. The response was not so much serious criticism but distaste: it was just not the way gentlemen ought to proceed. Of course all of it, one end to the other, evaporated in the face of DNA
In Eurasia, the places that preserve lots of relict language diversity are mountainous areas like the Caucasus or highland Southeast Asia. In North America, it’s California. From an Old World perspective, the idea that you could have a place as lush as California occupied entirely by a mosaic of hunter-gatherers is crazy; they would long since have been steamrollered by farmers. Maybe one difference is Old Word farmers often carried not just plants but diseases (hat tip to William McNeill and Jared Diamond), and genetic resistance to diseases. The Bantu were probably not the first farmer/herders in East Africa, but they may have had an advantage when it came to malaria resistance.
Even before Greenberg many of Sapir’s ideas regarding relationships between different Amerindian linguistic stocks got a pretty cool reception.
Very true. And very often, these were the same people who insisted that the ancestors of the Indians came over in one big migration, then essentially had no outside contact for 13,000 years or so, which would mean that all Indian languages would have to be related to a common ancestor, wouldn’t it? I don’t get it either. Linguists are an odd lot, anyway; I mean, look at Chomsky.
A bit off topic, but what you say about Europe and how modern inhabitants are not related to those 11 000 years ago.
I though that recent evidence suggests that most Brits are genetically similar to the earliest inhabitants and that Saxon and Norman invasions just replaced elites and languages. Have there been new studies or did I misunderstand the literature.
The earliest inhabitants were hunter-gatherers who were quite different from Brits today, although they did make a genetic contribution. Next, probably farmers ultimately derived from the Middle East, although this is better worked out for the Continent. Next, Indo-European invaders: you can see the ancient Siberian component there.
Anglo-Saxons are more a reshuffling, not as different from the locals as these big earlier turnovers. They may account for 40% of English ancestry, which had to involve much stomping of Romano-Britons.
Earlier stuff was wrong, it seems. Ancient DNA is making things clear, more or less in spite of what typical academics wanted to think.
“Why have there been fewer and less powerful population expansions in the Americas?”
It is a hell of a good question, and as usual with good questions that bother the in place belief systems it won’t be asked. There was one hell of a powerful population expansion after the immigrants from Europe got here, that should give a strong hint, but it won’t, except maybe among some upstarts in an obscure blog.
Nice to see Vennemann get a mention anywhere (though I think McWhorter gives him unattributed props). The strength of the historical linguists’ argument that language and genes don’t always go together (so genes can be essentially disregarded until millennia of sound-changes are established), is that they have found some exceptions. So you just can’t trust that gene stuff.
No, really. That Cavalli-Sforza’s gene-mapping bears considerable resemblance to Greenberg’s Amerind languages is considered interesting but not very important. People from other disciplines who wander in to their theories wonder how the ship stays afloat.
As someone with a linguistics degree I feel compelled to speak up about the controversy surrounding Greenberg. Linguists have problems with his Amerind theories not because they disbelieve in a complete separation between genes and language, or because they disagree with the narrative of the settling of the Americas, but because they believe he was not using the comparative method effectively. Many linguists have argued that he was looking at the data wrong and making improper conclusions, not that he failed some political correctness test.
Their argument is whether the comparative method can be used, with the extant data, to prove relatedness between multiple Native languages some 15,000 years after the fact of their putative separation. Many linguists, if confronted with the recent genetic discoveries about Native Americans, might well agree that the small group of first arrivals were speaking one language, or a set of related ones. But they would say that this cannot be soundly proven with the comparative method, and that Greenberg failed to provide such proofs.
In short, linguists assert that we may suspect and surmise, but we cannot prove a connection between these Amerindian languages (I’m not talking about Eskimo/Aleut or Na-Dene) using the comparative method. Other forms of data, like the genetic discoveries discussed in this post, may shed light on ancient Amerindian languages though. Consider that the Indo-European languages are only some 5,000 years old, much younger than the Amerind family would be. At some point, relations between languages start to break down to the extent that ancient connections between them cannot be proven with scientific rigor.
*not because they believe in a complete separation between genes and languages, oops
Even though relevant genetic data has been accumulating for some years, I’ve never seen linguists make any use of it at all. That’s a mistake. It looks as if every big language expansion was, at least for quite a while, driven by the expansion of a particular population: that’s key info that linguists do not use and generally reject. They talk about how wrong it is to talk about an “aryan” population – but there was one, and we can now identify its signature all over Europe. Except for the Basques and Sardinians.
Second, although most linguists think that Greenberg’s methodology was unsound, if it gets the right answer when you don’t, maybe you should at minimum take another look. Looks as if one could simulate the problem. I can think of optimization problems in which some method was useful, much used, but not well understood for a long time – only much later did someone show that it really had to work, except for a set of measure zero.
In fact there are other obvious implications of natural selection relevant to language that linguists seem to know nothing about and would almost certainly reject. I think I’ll go post about another.
Look, it doesn’t do any good to simply repeat what you were taught. It’s not a convincing argument. Archaeologists were all trained to think that the default, almost the only explanation for sudden, dramatic changes in material culture was local adoption of new customs, rather than an invasion by some other people. “Pots not peoples”. This never made sense to me, and ancient DNA results now show that such invasions and population replacements happen, a lot. Archaeological dogma was wrong – no point in invoking it. To be fair, they seemed to think that unpleasant events they didn’t like to contemplate therefore must not have happened – which is considerably crazier than anything I’ve heard out of a linguist.
“Even though relevant genetic data has been accumulating for some years, I’ve never seen linguists make any use of it at all. That’s a mistake.”
No, it isn’t. It’s important to keep the evidence streams separate. You cannot show linguistic relatedness through genetics, just as you can’t show genetic relatedness through language. Yes, it’s important to corroborate them both if you want an accurate picture of the prehistoric past, but to do this in demonstrating the *existence* of language families is clearly the wrong approach.
As for ‘pots not peoples’, it’s important to realise that sometimes people do change their lifestyles on the basis of ideas imported from people who made little or no genetic contribution to the population. Christianity in early medieval England was not imposed by a large-scale migration of Greek- or Latin-speaking peoples who politically dominated the lower classes, and in spite of this we can see major changes in lifestyle in the Christian period, including new ways of burying the dead, new art forms and motifs, and so on. Fish and chips entered England without a mass migration of Sephardic Jews and Flemish cooks. It’s best to be sceptical and not jump to conclusions.
Migration is clearly incredibly important in prehistory, and I’m not disputing that. But it cannot always be assumed to have been the main factor a priori.
@A.J. West: Isn’t the spread of Christianity in Europe a consequence of the strength and extent of the Roman empire? In general, was existence of strong, stable state a default situation back in pre-Roman times, with certain localized exceptions?
If not, then stubbornly insisting on relying on “pots not peoples,” as opposed to “peoples not pots” as the default explanation (and there is always a default explanation, or at least, line of thought!) smacks of something other than mere reason and preference for evidence/separation of its various lines.
“Peoples not pots” seems to be much better at explaining the overwhelming majority of prehistorical changes, especially in light of new genetic evidence concerning various hunter-gatherer-to-hunter-gatherer and hunter-gatherer-to-farmer transitions that we’ve been reading about in the last few years.
To me it seems that the mostly-non-violent, cultural vector for change became possible only around Biblical time, in agrarian regions. It became a stable situation around the times of the Roman Empire, with an interregnum defined by the Dark Ages, after which becoming the default vector in many parts of the world.
Still, it’s easy to see how in certain parts of the world, notably in hard-to-access parts of sub-Saharan Africa, the “peoples not pots,” is still the predominant way material culture changes.
Also: I haven’t noticed *anything* to make me think that people on this blog insist that “peoples not pots” is the *only* way change happens. On the other hand, I’ve seen enough empirical evidence that academia has much more neurosis regarding anything that goes against “pots not peoples.”
“Isn’t the spread of Christianity in Europe a consequence of the strength and extent of the Roman empire?”
Yes, but my point is that in *sub-Roman*/Anglo-Saxon England, Christianity came not with a migration or invasion but by the adoption of the idea by an existing group. The Romans had come and gone and while Celtic Christianity survived in some areas, Roman Christianity was not adopted by Germanic-speaking people in Britain until some centuries later due to the activities of missionaries of various ethnic backgrounds. Pots, not people.
There isn’t a strong bias against migration in the academy – I can think of a few publications in the last few years with a very strong migrationist slant: Bellwood (2013) ‘First Migrants’, Manco (2013) ‘Ancestral Journeys’, Anthony (2007) ‘The Horse, the Wheel, and Language’, and plenty of others. ‘Pots not people’ is an adage for an earlier age, I think.
It’s important to bear in mind that genes, languages, ceramics, and other aspects of human culture can be transmitted somewhat independently of one another. Not entirely independently, of course, but largely. We can’t use genes to establish language families or languages to establish ceramic traditions, but we can try to assess the distributions of all these things in light of various hypotheses, including, quite prominently, migration. That seems like the most reasonable view: neither ‘pots not people’ nor ‘people not pots’.
It is no use copying the culture around milk consumption without having the genes for lactase persistence. Likewise for alcohol tolerance or tolerance for hard work.
“or tolerance for hard work.”
…. Right.
It turns out that there are two forms of alpha-actinin-three in humans. Both forms are found in all populations, but the frequencies differ dramatically from one population to another. The intact version helps sprint capabiltiies: it is found at a higher frequency in Olympic-class sprinters than in the general population from which they are drawn. The null version, it turns out, increases endurance: this was suggested by a study in athletes and confirmed in mouse-knockout experiments, which resulted in mice with a third more endurance.
But you knew that.
Greg & Henry,
Have you guys had your “phenotypes inspected”?
http://www.patheos.com/blogs/markshea/2014/02/a-reader-writes-of-his-experience-among-the-dark-enlightenment-types.html
Apparently it’s the new thing.
This was a prank – and a funny one at that.
“Before the advent of the Indo-Europeans, almost all of Europe must have spoken related languages, originating somewhere in the Middle East or Balkans, carried by the LBK and Impressed Ware culture. Basque must be the last survivor of that family, rather like the tuatara. Theo Vennemann, can you hear me?”
If the WHG and ANE are closely related populations via the mammoth steppe and the ANE later developed the Indo-European languages couldn’t the Vasconic be the last survivor of WHG languages rather than EEF?
Unlikely, there are signs, place names and such, that Sardinians once spoke a language related to Basque. Besides, the impressed ware Sardinian/like people spread far – hunter-gatherer groups were swamped. Bet on the winners of that round.
Genetic studies of Basque speakers show a high concentration of mtDNA haplogroup HV4a1, which seems to have entered Europe from the Near East in the mid-Holocene (Italy and southwestern France c.3400 BCE). aDNA studies show a high concentration in eastern Europe, and there’s a hypothesis that Basque arrived in southwestern France and Spain in the Copper Age from the Balkans. Also, there are indigenous Basque words – definitely not borrowed from Indo-European – for wine, cart, iron, lead, and millet, indicating a more recent origin. The idea that ‘Basque must be the last survivor of that family’ doesn’t seem quite right. It’s certainly possible, but far from proven.
“Europe experienced two expansions (first Middle Eastern farmers and then sort-of Siberian Indo-Europeans).”
If the bottleneck (population expansion) in I1 in northern Europe fits the time window for Funnelbeaker then there’s three: farmer -> funnelbeaker -> IE with the possibility that conflict between expanding Funnelbeaker and LBK led to LBK’s decline and/or the conflict between the two being the catalyst for the IE expansion.
There is every reason to believe that large-scale migrations took place in the pre-Columbian Americas, fuelled by cultural innovations. The big ones are Algic/Algonquian, Uto-Aztecan, Otomanguean, Mixe-Zoquean, Iroquoian, Siouan-Catawban, Arawakan, Tukanoan, Panoan, Tupian, and Cariban. The trouble with American prehistoric migrations and language families is not that they didn’t happen, but that a) post-Columbian events disrupted the Americas to such an extent that the pre-Columbians distributions of most languages families cannot be known with certainity and are certainly not accurately reflected in modern maps of these families and that b) many of the migrations took place over one another. Tupian spread into Arawakan territory, Algonquian and Iroquoian spread into one another’s territory, and so on.
The Americas were only marginally more static than Afro-Eurasia, and the differences are almost certainly due to the lack of cattle, horses, and camels, all of which make language family expansion incredibly easy. There doesn’t seem to be any genetic component to it – at the very least the null hypothesis should be that genetics had nothing to do with it.
AJ, come on, reassess some of your statements. “The Americas were only marginally more static than Afro-Eurasia” and “there doesn’t seem to be any genetic component to it.”
They are both flat factually wrong. Don’t argue back at me like some yappy TV moron, just look at the evidence.
I’m with you on the first statement, perhaps – the Americas were significantly more static in some terms, especially when it came to the spread of agriculture and the diffusion of some language families. But this may be put down to the north-south orientation of the Americas, meaning that maize (e.g.) required significant adaptation in order to be grown at more northerly latitudes and at higher altitudes (see e.g. Flint corn, etc). It can also be put down to the absence of horses, cattle, and camels, all of which contributed to the expansion of language families in Eurasia.
So the second statement, ‘there doesn’t seem to be any genetic component to it’, seems prima facie correct. There are good reasons for the less dynamic situation in the Americas that have little to do with genetics (except plant genetics, of course), and invoking genetic explanation is utterly unnecessary. In fact, I’d say that it is almost absurd to raise it as a proposed *explanation* for the phenomenon.
I suspect that the slow spread of agriculture in North America due to the north-south orientation did indeed slow things down. Probably the lack of domesticated animals did too. I’d add the less advanced maritime technology.
But denying that genetic factors might have influenced inter-group competition is… absurd. They certain have in many other cases.
For example, it now appears that the Tibetans are a fusion of two peoples, one a group of hunter-gatherers that had lived up on the plateau for tens of thousands of years (and had developed very effective genetic altitude adaptations) and a population of farmers closely related to the Chinese. Normally, hunter-gatherers tend to get swamped by incoming farmers (the original hunter-gatherers probably make up a percent or two of the ancestry of the existing population of Indonesia and the Philippines), but in the Tibetan case, they account for a lot – because people with more Sherpa-like ancestry had a big selective advantage.
And when farmers swamp hunter gatherers, as for example when the impressed ware/LBK Sardinian-like peoples largely replaced the old blue-eyed hunter-gatherers of Europe, genetics played a role. The ancestors of those incoming farmers had had an agricultural diet for a couple of thousand years, had experienced the higher disease load associated with higher population density, had lived a very different life in which the payoffs for various psychological and cognitive traits were different from those experienced by the ancestors of those European hunter-gatherers. They had more copies of amylase, understand? They had a more effective ergothioniene transporter. They had SLC24a5. The locals didn’t. Those incoming farmers were selected for being farmers: with all the will in the world, the indigenous hunter-gatherers of Europe couldn’t be as good at it. And selection acts on every trait that can make a difference to reproductive success – which means personality and cognition, not just digestion and immune function. They’re all heritable, so it’s inevitable.
The Tibetan example is interesting because a) that kind of genetic diversity is readily apparent in the Americas (it’s not as if Andean populations aren’t adapted to the mountains), so you can’t go on to say that lack of genetic variability in the Americas is the reason for the comparatively static situation there, and b) the Tibetans speak Tibetan, not another proto-Sherpa language, so asserting that language family dispersals are primarily to do with genetics on the basis of an example where a population with a genetic advantage acculturated to the lifestyle of one without the genetic advantage seems like a strange way to assert the primacy of genetics.
“Normally, hunter-gatherers tend to get swamped by incoming farmers (the original hunter-gatherers probably make up a percent or two of the ancestry of the existing population of Indonesia and the Philippines)”
Depends where in Indonesia – if you look at Timor, Alor, Tanimbar, Biak, etc, the percentage is far higher. To say that foragers are ‘normally’ swamped seems wrong, actually, given the clear evidence of Khoisan-associated genotypes in Bantu-speaking populations, etc.
“The ancestors of those incoming farmers had had an agricultural diet for a couple of thousand years, had experienced the higher disease load associated with higher population density”
I’m sure you’re aware that this is a controversial topic; Bellwood, for example, is pretty clear in his view that diseases accompanied agriculture only later on, after the initial expansions.
In the Tibetan case, there were countervailing genetic advantages: one group adapted to farming, the other to altitude. it makes it less likely that one group would utterly swamp the other.
A language family expansion that is powered in part by genetics can go a lot farther than one that it is not at all, because it is difficult to copy. Do you think that European hunter-gatherers could conjure
more copies of amylase by wishing?
As far as I know, the Bantu group with the largest Bushmen admixture is the Xhosa, with about 25%. Most seem a lot lower than that.
I was not aware that that an increase in disease load among early farmers was controversial – but it’s inevitable, so who cares? Look, I know that the total number of infectious diseases hitting the Old World has grown over time. Quite a number of crowd diseases are new in humans: measles is, mumps is, probably smallpox. Probably falciparum malaria is new, or at least has greatly increased its impact and range fairly recently. But quite a few infectious diseases are old, like tuberculosis. Typhoid might be: with 2% of cases becoming long-term carriers, it could have existed in hunter-gatherers. It is easy for a zoonosis like influenza to be old.
Then there are fecally transmitted pathogens – hunter-gatherers just move on. Sedentary types get a higher worm load. And so on.
Not only that, disease organisms experience differential selective pressures when the population density is higher. In low-density populations, they often have to exhibit high latency to continue to exist over time – they have to wait for new kids to be born. Examples, herpesviruses like Epstein-Barr and chickenpox, BK and JK virus, h. pylori, TB. At higher density, higher virulence is often favored.
Okay, so let’s say Greenberg’s work wasn’t sloppy and that his method could be used to establish relationships. Take a look at Weltanschauung’s examples of Greenberg’s method – Latin est/sunt, German ist/sind, for example.
Greenberg’s method doesn’t rely on reconstructing the actual series of phonological changes that occurred in the languages under discussion, and he uses a tiny number of similar lexical items from each language. That may well be the only way to go about making inferences about languages that go all the way back to the Pleistocene, but here’s what it doesn’t do: it doesn’t rule out borrowing or coincidence. If we relied only on est/sunt and ist/sind and a couple of other terms in Latin and German to establish a relationship between Germanic and Italic/Romance, we wouldn’t actually have ruled out the possibility that the terms were borrowed. In fact, there’s no way to rule this out with Greenberg’s method, making it, well, useless. It means that while it claims to extend linguistic analysis out to 13,000 year time-depths, in reality it can’t rule out borrowing within the last few hundred or few thousand years.
The historical linguistic comparative method, on the other hand, does rule out borrowing (and can demonstrate it, too) by using a much larger data set, combing through it for reconstructable lexical innovations and phonological changes (e.g. ‘cheese’/’tsiis’/’kaas’, ‘church’/’tsjerke’/’kerk’ – with enough data from enough languages showing a similar pattern, we can reconstruct a couple of sound changes in West Germanic or Anglo-Frisian). We can use this data to find out when the word entered the language, what it looked like when it did, and so on. Don’t forget syntax and morphology, either – relatedness depends on those more than on lexical items.
It’s complicated and difficult, but that’s how it’s done – not because it’s Our Sacred and Holy Method, but because it rules out other sources for words, relies upon consistent phonemic correspondences, and uses large amounts of data. It rules out, sometimes beyond reasonable doubt, borrowing and coincidence. It demonstrates genuine relationships and reveals false ones.
Greenberg’s method is only useful if you already believe the claims he makes are true. It seems like a good demonstration of his method to show that it can prove the relationship between English and French, but actually his method *doesn’t* prove the relatednes of English and French at all. The dataset is too small and the correspondences could easily be put down to borrowing or coincidence on their own. If Greenberg’s data and method were all we had to go on, then we would not be able to show that English and French are related beyond reasonable doubt. But because you already know that English and French are related, you assume that the method has been successful, and that it can apply equally well elsewhere. ‘It gave us the right answer, so what’s the problem?’ This is circular.
And with Amerind, everyone seems to be assuming that it generates the correct answer, viz. that there was a single population that moved across Beringia and is the source of all the non-Eskimo-Aleut-non-Na-Dene Amerindian languages. Again, this is question begging and nothing more. We don’t know if it’s true that only one language family entered the Americas or not, and given the considerable physiological diversity of humans in Pleistocene America, I’d be willing to bet that it’s not.
Oops, meant to put that further down.
Anyway: I agree that some genetic advantages will favour certain populations in certain areas (whether there is a relationship to language is quite a different matter). But I don’t think you can justify the claim that there’s a genetic reason for there being less population movement in pre-Columbian America. There were, in any case, plenty of large-scale population dispersals in the Americas – Arawak and Uto-Aztecan are excellent examples, and there are plenty more.
“I was not aware that that an increase in disease load among early farmers was controversial – but it’s inevitable, so who cares?”
Well, you said that the farmers coming into Europe would have had to tolerate higher disease prevalence for thousands of years by the time they arrived, and would therefore have had a greater advantage. But that may not have been the case; the new arrivals and the foragers-turned-farmers may have had to contend with exactly the same phenomenon hitting them at the same time. So it’s actually quite an important point.
Genetic advantages can take many forms. They are not limited to genetically simple systems like sickle-cell, or amylase copy number, or altitude adaptation genes like the EPAS1 variant found in Tibetans. They can involve many genes of small effect. For example, including the effects of all known gene variants that have been selected for malaria resistance in Sub-Saharan Africa leaves most of the genetic variance unexplained. Africans who don’t have obvious defenses like sickle-cell are still a lot more resistant than Swedes.
And since essentially everything and its brother is heritable, competitive advantage could take the form of heritable differences in just about any trait, even traits like height for which no single locus explains much of the heritable variation.
Mind you, this not to say that selection is driven by group competitive advantage. it’s almost entirely individual-level selection. But many forms of individual advantage would also confer advantage in intergroup competition.
There seems to be a willful failure to understand here.
Farming leads to higher population densities than foraging can, so those who have been farming for longer have been exposed to density-related diseases and pathogens for longer.
Moreover, people resist change until it is forced on them, especially such things as the hard-work and future-time orientation required by farming, which are both selected for among farmers but not among foragers.
There are diseases like measles whose critical community size is sufficiently large that that simply could not have existed back in hunter-gatherer times, and indeed phylogeny indicates that they really are new. The first farmers would not have run into them.
On the other hand, there are plenty of infectious disease organisms that are ancient and whose impact would have increased at higher human density. And then you have to consider zoonoses picked up from domesticated animals. So sure, farmers would have more pathogen pressure from the very beginning.
I don’t think it is a willful failure to understand.
Dixon and Aikhenvald, The Languages of Amazonia (1999) emphasise an agricultural revolution spreading from the Andes 5000 years ago:
“Each language family [of Amazonia] tends to have a characteristic profile in terms of the type of territory it is found in, methods of food procurement, and material culture. Almost all tribes speaking languages belonging to the Arawak, Carib and Tupí families are found in the rain forest, use agriculture and manufacture canoes, hammocks and pottery. In contrast, Jê-speaking peoples are mostly found on grasslands; they have little agriculture and no canoes, hammocks or pottery… Scattered between the agricultural tribes, in the heart of the rain forest, are small tribes of hunters and gatherers, belonging to minor linguistic families such as Makú, Mura-Pirahã and Guahibo. Some of these may be the remainder of earlier populations that occupied larger tracts of land before the agricultural expansion.”
Yudi – I don’t believe that Greenberg ever claimed that he had proven his Amerindian hypothesis by the use of some canonical comparative method. Scientific hypotheses are rarely “proven”. I think that Greenberg claimed that his Amerindian hypothesis was the simplest and most likely hypothesis in the light of the available evidence.
A.. J. West – It srikes me as as very strange to say in dealing with any biological phenomena that the null hypothesis should be that genetics has nothing to do with it. How often does one encounter biological phenomena that have nothing whatsoever to do with genetics?
Indeed, the default hypothesis. But the default hypothesis should be the null hypothesis – that genetic variation in human populations tends to have little or no influence on language adoption and language family migration. If the Americas saw fewer language family dispersals than Afro-Eurasia than we would expect – and I’m not at all sure that they do – then the absence of domesticated cattle, horses, and camels probably has more to do with it than the supposed lack of genetic diversity in indigenous American populations.
“I think that Greenberg claimed that his Amerindian hypothesis was the simplest and most likely hypothesis in the light of the available evidence.”
Probably right, but this is question begging. The whole thing starts from the assumption that there’s one protolanguage involved and procedes to use non-scientific means to demonstrate, vaguely, sorta, that one protolanguage was involved. It’s not convincing, it’s not scientific, and it doesn’t demonstrate anything that wasn’t already assumed by the analysis.
But if the Amerindians descended from one small group crossing Beringia, wouldn’t it make sense to assume that there was one protolanguage involved? This doesn’t make any sense. And once again, we see this irrational animus against Greenberg. Saying “he consistently got the right answer, but he used the wrong method” isn’t really very convincing, at least not to non-linguists.
I suspect that migrations across Beringia were more complicated than one small group crossing and staying; it seems more likely, given the way that foragers are known to move, for small groups to have moved back and forth across the territory, possibly speaking different languages and exploiting different resources.
Obviously, though, if there were one group, then they would probably have spoken only one language. Obviously. But what you don’t seem to understand is that Greenberg’s assertions literally go no further than assuming that this is true. His ‘Amerind’ adds nothing to the debate; it assumes what it seeks to proves, it begs the question, and it’s useless. That’s why it’s not accepted.
You’re begging the question, too. You say that he ‘consistently got the right answer’, but you’re merely assuming that his answer (read: his assumption) was the right one. Amerind is not a convincing family in any way, and while it is possible that there was one single language Beringian language that differentiated over time, and given certain assumptions that makes good sense, there is no support to be found for this in Amerind or Greenberg’s use of non-scientific mass comparison.
Seems to me that you haven’t thought it through. Whatever band first made it past the glaciers had the opportunity to vastly increase in numbers: that opportunity wasn’t available to those that arrived much later, because there was no longer an empty land in front of them. Hell is other people – Pleistocene America, south of the glaciers, was the Happy Hunting Ground, but only for those first past the post.
Total population in Beringia cannot have been very large, given the low biological productivity and the available techniques. Productivity was vastly higher to the south: lots of naive herbivores, plenty of plant foods, etc.
So there should have been lots of population growth, and that is what is indicated (using Bayesian skyline analysis of Amerindian mtDNA). And the archeological record shows very rapid spread over the Americas.
We have estimates of the effective population size of the founding group: it was small, by some estimates as low as a few hundred.
It wasn’t impossible for a later group to have displaced those first settlers, but it would have been difficult: the invaders would have to have a strong ecological or tactical advantage. When that has happened in human history, it has often involved agriculture, and that wasn’t in the cards for the likely source populations in Siberia. The ancestors of the Na-Dene seem to have a tundra hunting lifestyle and carried that over to newly deglaciated areas in the north. The Eskimo you can understand: they had a high technique for hunting marine mammals, that allowed them to spread over the high Arctic coasts (displacing previous inhabitants related to the Na-Dene) , but not much further
Slightly OT, but after I graduated, one of my first bosses used to paint little plastic soldiers and play games with them. As a hobby, not at work. I once asked him what he thought of North American Indians, and he said that in his opinion they were the best light skirmishers in history.
“Hell is other people” – especially them.
I’ve heard that. I don’t put a lot of stock in it though.
I should have said “default hypothesis” instead of “null hypothesis”. Obviosly the “null hypothesis” for any X is that X has nothing to do with it. But for the relation of genetics to any biological phenomena the null hypothesis has low apriori probablity. Genetics has some relevance to virtually any biological phenomenon.
Greenberg of course was completely aware that his methodology was not the standard comparative method. That wasn’t practical given the available data. But it seems dogmatic to me to dismiss his work solely on the grounds that a certain specific methodology (which was not feasible) was not adhered to.
It’s not dogmatic. It’s saying that we have a rough standard for showing relatedness between languages, and that if your comparison doesn’t live up to that standard, then it is at best an unverifiable hypothesis.
Greenberg’s Amerind doesn’t explain any set of facts. All it says is that most indigenous American languages descend from a common ancestor that was spoken a very long time ago. There are people who believe that already (and are therefore likely to support Greenberg’s claim) and plenty who don’t (and are therefore unlikely to support Greenberg’s claim). Either way, Greenberg’s method and evidence don’t do anything to convince people who don’t already believe in Amerind, which isn’t true of scientific historical linguistic claims.
With all due respect, there’s no arguing with you guys. Do you deny that Greenberg’s method has been used successfully with regard to African languages? La Wik says “For a time, his classification was considered bold and speculative, especially the proposal of a Nilo-Saharan languages family. Now, apart from Khoisan, it is generally accepted by African specialists and has been used as a basis for further work by other scholars.”
As I said in my first comment, to a certain set of linguists, Greenberg just can’t be right, even when he’s right. A good many of the comments above seem to prove my point, right down to the mischaracterization of his research, and the obviously rising anger. I rest my case
There’s no rising anger. You seem to be seeing what you expect to see.
Amerind is not well-demonstrated. In fact, it isn’t demonstrated at all. It’s got nothing to do with Greenberg; some of his African proposals (most of which were demonstrated using the standard comparative method, by the way) still stand up. Amerind, however, simply isn’t verifiable. If you want to believe that it is, then that’s up to you, but the fact remains that it is not. Amerind isn’t an established language family, and the belief that it is remains an essentially fringe belief.
That seems to me like saying that we have a certain standard methodology for measuring temperature with mercury thermometers and that if someone comes up with an estimate of the temperature of the center of the sun we just dismiss it without further consideration because the estimate wasn’t arrived at using mercury thermometers.
The issue isn’t the use of a certain standard methodology. Greenberg was quite explicit that he arrived at his conclusions by a different approach. It is dogmatic to dismiss his views solely on the grounds that his approch wasn’t the standard approach.
No, it’s not dogmatic. Did Greenberg actually demonstrate anything? No, he didn’t. He just asserted that a bunch of languages were related without going to the effort of showing that they were. Showing relatedness is tricky; you have to work out complex series of phonological, syntactical, morphological, and lexical changes. This method rules out other sources of change, shows consistent changes where they exist, and helps to reconstruct an ancestral language through the residue of comparisons.
It’s a bit like the work of comparative physiology or anatomy; through comparing living organisms and fossils, comparative anatomists can figure out what earlier forms may have looked like and can help to reconstruct the features of an ancestral organism. Historical linguists do this with languages, and obviously it’s tricky work.
Greenberg didn’t do this; he just said that indigenous American languages were related, then threw a bunch of arbitrarily-selected terms from arbitrarily-selected languages together and tried to show that they were kind of similar. That doesn’t quite work, it isn’t convincing, and it doesn’t show anything more than what it assumes in the beginning. That’s not science. If we take the comparative anatomy example again, then Greenberg’s method is analogous to claiming that the platypus is of the order *Anatidae* due to its bill.
That’s not what Greenberg did at all.
Greenberg called his method “mass comparison”, and showed how it supported, not just the vague claim that all his Amerindian languages were related somehow, but the cladistic structure of their common descent: which languages are more closely related to each other than to the rest, which of these families are more closely related to other families than to the rest, and so on.
He freely conceded that the data don’t support reconstruction at the level of detail that has been achieved for Indo-European.
Speaking for myself, while I admire the achievements of the comparative method, I have to confess that reconstructed Indo-European doesn’t come trippingly off my tongue. Many comparativists concede that they don’t know exactly what the pharyngeals sounded like, and that there are details will likely remain forever unknowable; in no way does this call the genetic relationship of the Indo-European languages into question.
Back to Greenberg: it’s been a long time since I read Language in the Americas, but I still remember the appendix in which he demonstrates the value of mass comparison by deploying it on a deliberately small subset of Indo-European data. The data set is too small to support reconstruction, but it works just fine for showing that English and Swedish and Afrikaans are closer to each other than any of them is to Polish. And having postulated a Germanic family, we can then show its relationships to Slavic and Romance and the rest using features of these families as a whole, which is where the “mass” part of mass comparison comes in. For example, German ist/sind and Latin est/sunt are valid evidence for a genetic relationship, not just between German and Latin, but between their respective famililies; and therefore mass comparison validly supports the genetic relationship between English and French, even though traditional comparative methods, applied to this pair alone, scarcely support any relationship. Greenberg’s table of the valid cognates between French and English is quite amusing in its scantiness.
Philip Neal’s comment above says a lot about the method Greenberg used to infer the existence of Amerind. Sloppy work through and through, and utterly unverifiable.
I have heard these arguments for decades. Here is why I remain unconvinced: “We have a way of doing things (showing language relatedness). We know it’s right. If you do things another way, you aren’t proving anything. You have to prove things to our satisfaction, or we just don’t care.”
It’s all circular.
Whether Greenberg was sloppy or not is secondary. The above argument – which I have intentionally stated unkindly – is the center. I have been through this before, as my actual career relies heavily on psychiatric diagnosis. It took decades to dislodge the Freudians and ego psychologists who made arguments of similar type to yours: “but this doesn’t fit our model. We withhold agreement.” Or the joke about economists, noting that something works in practice and going back excitedly to see if it works in theory. The standard that is set up to determine relatedness is necessarily arbitrary. The current practitioners set the threshhold. Greenberg’s African simplifications weren’t accepted either.
Yudi, above “But they would say that this cannot be soundly proven with the comparative method.”
AJ West, above “It’s saying that we have a rough standard for showing relatedness between languages, and that if your comparison doesn’t live up to that standard, then it is at best an unverifiable hypothesis.”
Then the standard, the method, may be obsolete, or useless, or inadequate. If the idea is, “well, we find the method inseparable from our discipline,” prepare to have your discipline become a curiosity.
The case against Greenberg is that he lowered intellectual standards in order to arrive at an intellectually sensational conclusion. Real historical linguists trace consistent changes in a precisely defined state space of phonology and morphology, they prefer semantically exact cognates to inexact ones, and above all they seek evidence which refutes a hypothesis such as Amerind rather than evidence which confirms it. Greenberg did not analyse his data rigorously and sought to confirm not to refute.
In the light of the Clovis genome, geneticists are entitled to tell historical linguists what they should be looking for, but that does not validate Greenberg’s view of how historical linguistics should be conducted.
“But the more we learn, the more it seems that the lumpers are correct.” – Edward J. Vajda
“The February 2008 symposium featured no presentations on Beringian genetics research. However, it is clear that a close Dene-Yeniseian language connection does not parallel what population geneticists have so far discovered about the peopling of the New World. Research on human DNA of North Asian and New World poulations by Rubicz et al. yielded no evidence that Yeniseian and Na-Dene speakers (including the Haida) share a specially close genetic affinity when compared to other peoples of their respective regions. One might expect a Dene-Yeniseian language link to be paralleled by evidence from population genetics, but so far such evidence is lacking… It is possible that some modern Siberians represent a back migration out of Alaska during the Early Holocene, prior to the establishment of the Eskimo-Aleut on both sides of Bering Strait. Recently published evidence from population genetics indicates that something like this indeed occurred (Tamm et al 2007).”
“Conservative historical linguists, and I would include myself firmly in this group in terms of the standards I attempt to meet, insist that validating a hypothesis of genetic relationship requires a system of homologies in grammatical morphology as well as a body of lexical cognates extensive enough to reveal systematic sound correspondences.”
– Edward J. Vajda
Turns out that the work by Rubicz, showing that the Na-Dene cluster with other Native Americans populations, was wrong.
This is a fast-moving field. Whole-genome analysis, which is more powerful, wasn’t happening back then.
I haven’t seen a recent analysis of Ket genetics – could be interesting.
O/T, sorry – can’t find email of you usual suspects. i’m sure you’re already aware of:
most bio/evo/anthro textbooks speak of what geniuses early amerinds/native americans were for domesticating the strange looking teosinte plant into corn — always accompanied by a picture of teosinte & modern corn looking nothing alike.
research done growing ye olden teosinte under climatic & atmospheric conditions mimicking those of 10,000 – 13,000 years ago shows that under those conditions teosinte looks EXACTLY LIKE CORN!
hilarious. science is pretty cool 🙂
http://tinyurl.com/pcph8ot
as a sage commenter at smithsonian.com noted:
Other writers have attempted to describe prehistoric farmers in the New World as genius level manipulators of the wild teosinte’s native genome, placing the domestication corn on a par with the building of the pyramids or development of writing. Now, at last, we know better.
Racissss!
Um, the article you link to doesn’t say the Mesoamericans didn’t domesticate corn from teosinte, it helps explain why they chose to do so. Their descendents built pyramids and developed a written language. I’m not sure what your point is.
i know. my point (& the point of the comment i quoted) is that textbooks/science trade books have oft framed this as “Mesomorphamericans must’ve all been brilliant & very patient plant geneticists to even think of attempting to turn something that looks nothing like corn into corn.” turns out teosinte looked like corn. The underlying reason for overenthusiastic plaudits (like lauding someone for inventing a use for the peanut) is that well-meaning overenthusiasm – aka “post-hoc turdshining” is a form of… what the 4th doorman of the apocalypse-o above says. I’m of course okay with Discover-magazine-like comments such as “all mayans were brilliant architects to have built pyramids & stuff.” i’ll bet they weren’t “all” brilliant, or even all “architects” – it’s better to qualify such statements.
I don’t recall anyone but you using the word “all”. Corn only had to be domesticated once. I’ve never seen a claim that each and every Mesoamerican personally domesticated corn. And I think most people understand that pyramids are built by one architect and a lot of guys doing heavy lifting.
OT, but a really interesting mutation:
http://arstechnica.com/science/2014/02/crazy-ants-take-on-fire-ants-and-win/
Well, this is even more off-topic:
Claims of candidate gay genes
Horses and bronze apparently the had both (kind of) and ate one and gave up on the other weird.
OK so you get a pretty obvious fact like American Indians all having come through the Bering Strait, and looking pretty alike so best guess is they came in a big bunch and then diversified.
Makes sense to think that their languages are all related. For all we know they actually are, but “relatedness” in a language can’t be measures the same way as DNA. It’s not that Greenberg method was better; it wasn’t. And it wasn’t by the closed-minded linguist establishment opinion; it just doesn’t work like that.
Doesn’t mean he was wrong, but he didn’t prove anything, because it can’t be proved.
We have obvious reasons to think that the central temperature of the sun is very hot; doesn’t mean that any paper coming up with a bogus mechanism to come up with a very high number is axiomatically right.
I’m reading an old letter to ASHG – Problematic Use of Greenberg’s Linguistic Classification of the Americas in Studies of Native American Genetic Variation, 2004 They explain that Greenberg’s work is just nonsense and that geneticists should not expect Redskin genetics to fall into Greenberg’s three clades.
They do though.
The facy that Eskimo and Na-Dene can be identified as distinct families, doesn’t mean that all the rest can be reliably as one family itself.
Common sense says that all Papuan languages should belong to the same family, but the standard method doesn’t say that, as the data isn’t conclusive enough.
The standard method also says that Japanese and Korean aren’t conclusively related, and that the Altaic
family also can’t be proven.
You might say that means the standard method sucks, and it does, but that doesn’t mean Greenbergs is better. Languages change a lot, and we don’t have data from the past, so what can you do?
Plenty or crackpots propose an Eurasian or Nostratic language family. If you loosen the standards that’s what you get.
Look guys,
Like Dr. Cochran said:
1) the amerindians came in three waves, according to the Reich et. al study. First wave, speaking some languages, probably related because they were a small group (right?); second wave, speaking Na-Dene, which is still has relatives in siberia (Yeniseian); and finally eskimos, speaking eskimo-aleutian.
2) Overall the first group is quite homogeneous and doesn’t have a lot of population substructure, at least not as much as in eurasia, to this day. I’m not sure to what degree you can compare them, but amerindians are quite similar from north to south.
So, with that in mind, why wouldn’t they be speaking a group of related languages? That’s the question that should be asked, not whether they are speaking related languages. When the tribes were splitting thousands of years ago, did some group suddenly suffer complete hearing loss, only to regain their hearing a few hundred years later after crossing some river, but by which point they have completely forgotten the old language they spoke?
“So, with that in mind, why wouldn’t they be speaking a group of related languages?”
They might very well have spoken a single language. There might well have been only one language that crossed over from Siberia in the Pleistocene. No one is arguing that this isn’t a serious possibility. No one is denying this outright, because it makes reasonably good sense, even if it isn’t actually verifiable.
But Joseph Greenberg’s method could not possibly have shown that Amerind is a valid family. The method doesn’t work.
spandrell – I never said that anybody’s estimate of the temperature of the center of the sun is “axiomatically right”. I just said that it would be irrational to dismiss an estimate of the sun’s temperature solely on the grounds that the estimate wasn’t arrived at via the standard method of using a mercury thermometer.
Well you would if you knew the method was bogus and used bad data, even if it was pretty smart and worked on a very good hunch.
Not that I’m defending the linguist establishment per se, which is full of crackpots; I’m also not aware of the actual arguments against Greenberg.
But historical linguistics is a very messy discipline, with little if any real benchmarks of success. Tracing back Indo-European was a good, if obvious endeavor. And the field is full of overzealous history buffs who come with their own pet theories of this and that language family being related because 10 words look similar according to this laws I just made up.
As long as we get a time machine there’s no real way of proving anything beyond the most obvious relationships. Can’t falsify anything. Historical linguistics is no more a scientific field than cultural anthropology. Yeah it’s fun and some data can be interesting, but if guys from the hard-sciences expect the field to give you real data you are going to be disappointed.
The correct approach is the other way around: get on with DNA sequencing and the linguists will come up with language correspondences to back it up.
“Yeah it’s fun and some data can be interesting, but if guys from the hard-sciences expect the field to give you real data you are going to be disappointed.”
That’s complete nonsense. The historical linguistic method makes a lot of sense, and if you think it’s no more scientific than cultural anthropology and less likely to give you reliable data than population genetics, then you must know very little about it. Indo-European is only one of many well-established language families that have barely changed in the last few decades, and, again, if you think IE is an isolated success in a field dominated by ‘crackpots’ then you simply do not know the field.
The main problems in using historical linguistic data in archaeological hypotheses are to do with locating language family homelands – was proto-Austroasiatic spoken along the Yangzi, along the Brahmaputra, or along the Mekong? Was PIE spoken in early Neolithic Anatolia or the Chalcolithic Pontic steppe? – not in working out the details of those language families. The families are generally quite well-established these days, and some – including Austronesian – have been reasonably obvious for well over a century and a half.
“The correct approach is the other way around: get on with DNA sequencing and the linguists will come up with language correspondences to back it up.”
Just… No. What a naive view this is! Language and DNA simply don’t match up perfectly. Visit Izmir and Kashgar in quick succession, or Accra and Leeds, or any other pair of places with strong linguistic connections where the genetic and physiological differences are stark, and tell me that we should use genetics to help us invent linguistic entities.
We should use genetics to help unravel linguistic questions. Glad to oblige.
It depends on the problem in question, but if you mean that genetics should be used instead of valid linguistic evidence then that’s rather silly, frankly. Genetics doesn’t say anything about language except, perhaps, that the organisms involved have a capacity for using it. Some haplogroups overlap somewhat with some languages, but there’s no one-to-one correlation between them and no reason to use genetic data to bolster linguistic claims. The question of homelands can be helped along with genetic evidence, but you simply can’t demonstrate a language family through something other than language comparison.
> The Tibetan example is interesting because a) that kind of genetic diversity is readily apparent in the Americas (it’s not as if Andean populations aren’t adapted to the mountains), so you can’t go on to say that lack of genetic variability in the Americas is the reason for the comparatively static situation there
Well… according to other of Cochran’s posts, the Tibetan altitude adaptations are entirely different from the Andean ones: more to the point, they are better – many fewer individuals going sickly in early senescence – and much more similar to the adaptations found in mammals that have been up high for millions of years.
The better results in Tibet might be a function of more ambient genetic diversity being available in the broad region (including lowlands), or of Old World people living up high longer, or of more Old World people living up high (=faster evolution). One or two of those might be relatively unimportant, but more likely than not, they all matter some.
> the Tibetans speak Tibetan, not another proto-Sherpa language, so asserting that language family dispersals are primarily to do with genetics on the basis of an example where a population with a genetic advantage acculturated to the lifestyle of one without the genetic advantage seems like a strange way to assert the primacy of genetics.
He’s saying that both pops, not one or the other, had (unrelated) suites of genetic advantages. And while it is subjective and could too easily turn ‘semantic’, I would also say, or guess, that the Han-like ancestor pop also acculturated a great deal. Tilling the soil is a big change, but so is moving up onto some forbiddingly high plateau that I can only assume is beautiful but punishing to live in, offers a different set of raw natural materials, etc.
> The better results in Tibet might be a function of
I listed three things but left out a fourth: archaics. That might be where Pyrenees people got some of their ‘hardware’ – maybe someone already mentioned the idea.
> The problem of undetected morpheme divisions… is a frequent one in Greenberg. For example, he listed Rama mukuik ‘hand’ as cognate with words from several other American Indian language families which exhibit shapes like ma or makV, although ‘hand’ in Ram is kwi:k; the ma- is the second-person possessive prefix.
Stuff like this is an embarrassment, but occasionally – not too often – people who are serial flubbers also do amazing stuff. I’m agnostic on Greenberg, merely pointing this out.
spandrell – The best hypothesis in science is the simplest and most obvious one that explains the phenomena. Considering the pronoumial pattern in Amerindian languages that would seem to me to be the genetic hypothesis.
A.J. West: “‘The Horse, the Wheel, and Language”… 🙂 Though I’m with you in that Greenberg’s methods suck (and Ruhlen’s are even worse), I had a very funny time reading *that* book. Indoeuropeans were just a bunch of nice guys that happened to be enough popular among their neighbors that they ended learning their language. Maybe stone clubs and bronze swords helped a little, the author says, but there were no enough Aryans to conquer what the Vedas say they conquered. And no mention of genetic data in the whole book, just PC speculations.
Indic conquest of India is an unrealistic model that fails to fit with any of the archaeological evidence, or the linguistic evidence for that matter (e.g, multiple non-IE substrates in Indic languages). Anthony’s book is a good one, and it does talk quite a bit about the martial strategies of PIE speakers and their descendants. The javelins associated with Sredny Stog, the chariots of Sintashta-Petrovka, the word for ‘slave’ in Finno-Ugrian deriving from the ‘Aryan’ endonym, the use of horses in raiding – it’s not as if Anthony portrays them as wholly peaceable, because that’s clearly not true.
I think Anthony was probably trying to counteract the all-too-common Gimbutas Kurgan model of militaristic patriarchal expansion, and perhaps over-emphasised peaceful interactions over raiding. I don’t think it’s ‘PC’, though.
Anthony also talks about genetics, and even genetic advantages (lactose tolerance, for instance), but it’s a slight weakness of the book that he doesn’t correlate haplogroups and language in any serious way. I can only imagine that the book would have ballooned had he done so, and (personally) I find the archaeological and linguistic evidence much more interesting. I guess YMMV.
I think the big advantage of cavalry is psychological – the complete inability of normal humans on foot to stand their ground as a group against large fast-moving mammals coming towards them unless they are trained and disciplined as a group. I think the early adopters of cavalry / chariots must have had a massive advantage – not in combat per se but in making the other guys run away and then hitting them in the back when they scattered.
As Anthony makes clear, the first users of the horse were almost certainly not the first people to use cavalry. I’m sure cavalry would have presented any group using it with a fantastic advantage, but it seems that attacking *on horseback* is an Iron Age invention. The horse was clearly important in the societies that spoke proto-Indo-European, but there’s no reason to believe that they used cavalry. Or chariots, for that matter – they appear c.2000 BCE in southern Russia, with the Sintashta-Petrovka culture (believed to be ancestral Indo-Iranian speakers).
Horses were mainly used to be able to shoot arrows at the enemy without being at risk of being counterattacked. Chariots allow an easier platform for wielding a bow with a separate driver to hold the riens.
http://en.wikipedia.org/wiki/Battle_of_Kadesh
https://en.wikipedia.org/wiki/Battle_of_Carrhae
“Horses were mainly used to be able to shoot arrows at the enemy without being at risk of being counterattacked.”
As far as can be told, this is true well *after* the expansion of Indo-European. There’s no evidence that it was a tactic employed on the Pontic-Caspian steppe c.3200 BCE, and no reason whatsoever to include it in hypotheses about Indo-European.
My point is that all language families that can be discovered have already been discovered decades ago so there will be no further breakthroughs in the field. If anything I’ve seen consistent efforts to dissolve established families such as Sino-Tibetan or Altaic as being unsupported.
Any family discoveries talked about recently fall consistently in the crackpot category e.g. Nostratic or Eurasiatic. The data just isn’t there.
Of course language doesn’t map 1:1 to genetics, but in preagricultural societies did people switch languages without any genetic flow at all? Of course not. So in most cases there will be a correlation.
“My point is that all language families that can be discovered have already been discovered decades ago so there will be no further breakthroughs in the field.”
You said before that historical linguistics is as scientific as cultural anthropology, implying that all of its findings are vague and unscientific, and now you’re willing to take current linguistic theories as written in stone?
There have been some breakthroughs relatively recently, including the developing theory that Tai-Kadai is some kind of branch of Austronesian. This isn’t *such* a new theory, but it does make more sense in light of recent discoveries. Also, most linguistic work takes place at a lower level – Austronesian might be an obviously valid family (it is, in fact), but the sub-divisions are more controversial and considerable work is being done to iron out the wrinkles. The situation in eastern Indonesia is a nightmare; is Central-Eastern Malayo-Polynesian an illusion based on shared non-Austronesian substrates? Did borrowing take place across the dialect chain as it developed such that no protolanguage is reconstructable in that part of the world? That kind of problem is very common. It took a long time for IE to be worked out, and other families haven’t had that kind of time and attention, so the lower orders still need work. And the lower order divisions are almost as important as the big ones.
And yes, there’s often a correlation between language and some haplogroup or other. But the correlations aren’t that strong; there’s no single unifying genotype common to all speakers of Indo-European, for instance, and anything Indo-European speakers share is also found in speakers of other families. The correlations aren’t airtight. They’re very messy.
OK I was uncharitable and I apologize, but I’ve read too much BS (Japanese-Dravidian, etc.) in the field. Of course historical linguistics has done much good work over the time. But there is also a lot of fluff in it, and lots of people spending decades in trying to prove connections where there are none.
Given that most of the families that can be proven with actual data have been done established decades ago, the proportion of fluff is increasing.
I still remember when Tai-Kadai was established as a branch of Sino-Tibetan, and have been following the Austronesian theory. Many in China cling to the ST theory to this day.
Given the diminishing returns historical linguistics is facing, all I’m saying is that the geneticists should do their thing and not hope for historical linguistics to guide them in their work. It doesn’t work like that.
Did anyone think that reconstructing unverifiable “proto-languages” is anything more than a curiosity? It serves no purpose anyway. Linguists do it for fun.
I was also unnecessarily rude to AJ earlier and I apologize as well. I now see AJ has quite a background in the area that this blog thread covers. I invite AJ to look beyond his area of specialty and in particular to read all of the posts by Cochran and Harpending. I believe if he remains open minded he will find persuasive scientific evidence to change some of his opinions. If not, oh well, we can agree to disagree.
I hate to tell you this, but you’re behind the times. There is in fact a genetic component that is shared by the Indo-European speakers of Europe (not in Sardinians, probably not in the Basque). Judging from recent ancient DNA studies, it did not exist in Europe before about 3000 BC. More in a future post.
“There is in fact a genetic component that is shared by the Indo-European speakers of Europe (not in Sardinians, probably not in the Basque).”
So, only in Europe, and not among Sardinians, who definitely speak languages (Sardinian and Italian) that are undeniably Indo-European. It’s certainly an interesting interpretation of what I wrote to consider this a refutation of it.
“Did anyone think that reconstructing unverifiable “proto-languages” is anything more than a curiosity? It serves no purpose anyway.”
Well, it depends on what you mean by ‘serves no purpose’. It serves no purpose in the same sense that any study of human prehistory serves no purpose; it’s usually not at all useful in daily life, except to disavow us of racist and ethnonationalist dogmas. But if by ‘serves no purpose’ you mean that it has no significant impact on how we understand human prehistory, then you *really* don’t know anything about historical linguistics.
With reconstructed terms, we can find out what was and was not familiar to the speakers of the protolanguage. We can reconstruct strange and interesting things, like poetic forms or domestic architectual features or lithic technologies or whatever else you like. Even gods and cosmologies can be investigated through linguistic analyses. If you want an excellent analysis of PIE in these terms, see Martin West’s ‘Indo-European Poetry and Myth’. This complements and, in some ways, goes beyond archaeological evidence in helping us to understand the worlds of certain populations in prehistory. It’s far from useless to the study of prehistory.
You’re missing the boat. See next post.
As for Tai-Kadai (or Daic, or any number of other alternative names) being part of Sino-Tibetan, about the only evidence for this is the fact that most Tai-Kadai languages have a system of tones similar to those in many (but not all) Sino-Tibetan languages. Now that more is known about tone acquisition, the hypothesis isn’t tenable, and tones in Tai-Kadai are almost certainly the result of contact with Sino-Tibetan, as are any cognates in Tai-Kadai.
As far as I know, Austro-Tai is a reasonable theory, and takes into account the fact that tone and some lexical items in Tai-Kadai are areal features.
I’m not missing the boat. Genetics is incredibly important to the study of prehistory and as more data come in, more accurate studies of human prehistory are possible. But genetics isn’t everything and linguistics isn’t overruled by it. You’re pushing a naive view of prehistory that assumes that linguistic entities and genetic entities must be the same thing, and they’re not. They’re demonstrably not. And that demonstration is really easy to do: I speak English; my neighbours, who are of Nigerian origin, speak English too.
Language isn’t genetics. Proto-Indo-European isn’t a genetic hypothesis but a linguistic one. It should correlate in *some* way with *some* genetic evidence, because it was probably enabled in part by some kind of actual migration of people, but the genetic evidence doesn’t tell us very much about the language family, because people can copy language, speak multiple languages, learn a new language imperfectly and pass it on to their children, and so on.
You need them to correlate perfectly as you seem to believe that language family expansions are primarily enabled by genetic advantages, but there’s little reason to believe that this is the case. All of the talk about Tibetans – who have a large proportion of genetic material from a population that almost certainly didn’t speak a Sino-Tibetan language – still hasn’t made you rethink this bizarre idea.
AJ has said something important here which I think deserves to be expanded before the caravan moves on. There have been very significant advances in historical linguistics in the past half century or so, and most of them relate to East Asia. In the 1950s, most linguists would have thought it obvious that the major division of the region was between Chinese and its tonal, syllabic neighbours to the south (‘Sino-Tibetan’) as against the non-tonal, agglutinative languages to its north (‘Altaic’). The origin of tones and how historical linguistics should treat them still appeared to be an intractable problem.
That problem is now well-understood, thanks to theoretical advances which integrated the comparative method with feature-based phonology, and it now appears that most East Asian languages originally lacked tones. In a parallel development, the reconstruction of early Chinese has progressed considerably, providing a clear account of how Middle Chinese and its descendants, the modern dialects, developed from a non-tonal language, probably with inflectional morphology, spoken and written around 3000 years ago. We also have a much better idea of what loans from one language to another would look like at different periods of history. This is the background to the increasing tendency to question the reality of Sino-Tibetan and Altaic, to consider alternative groupings such as Austro-Tai, to seek a southern origin for Japanese and to pursue the intriguing resemblances between Old Chinese and Indo-European. It is no mere re-analysis, but genuine progress.
@gcochran
“Normally, hunter-gatherers tend to get swamped by incoming farmers (the original hunter-gatherers probably make up a percent or two of the ancestry of the existing population of Indonesia and the Philippines), but in the Tibetan case, they account for a lot – because people with more Sherpa-like ancestry had a big selective advantage.”
Doesn’t that imply the european HGs had a big selective advantage also as they account for a big percentage of modern europeans at least in the north?
.
“And when farmers swamp hunter gatherers, as for example when the impressed ware/LBK Sardinian-like peoples largely replaced the old blue-eyed hunter-gatherers of Europe, genetics played a role. The ancestors of those incoming farmers had had an agricultural diet for a couple of thousand years, had experienced the higher disease load associated with higher population density, had lived a very different life in which the payoffs for various psychological and cognitive traits were different from those experienced by the ancestors of those European hunter-gatherers. They had more copies of amylase, understand? They had a more effective ergothioniene transporter. They had SLC24a5. The locals didn’t.”
Doesn’t that imply the farmer percentage could have markedly increased since admixture and therefore was lower, perhaps much lower, at the time?
This could tie in with the increase in mtdna H over mtdna U after the collapse of LBK i.e. selection in place for the SLC genes (and other genetic advantages) via the surviving female descendants of the LBK farmers.
.
“had lived a very different life in which the payoffs for various psychological and cognitive traits were different from those experienced by the ancestors of those European hunter-gatherers”
One common farmer adaptation might be becoming less violent.
Depends.
Farmers in late farming Malthusian societies who lacked an almost obsessive focus on working hard over stealing (and stealing by killing) would be both punished by both poor harvests and by constabularies.
But what’s it like when there’s lots of easy farmland up for grabs, natives to kill and displace and no authority around to punish you for spreading destabilising violence?
True it depends on the cost-benefit of violence in a particular context and more importantly the cost-benefit in the preceding centuries which would have created the selective pressure.
There might be clues to this in the form of weapons in burials and/or the presence of fortifications. If you had a farming culture with no weapons in burials and no fortifications and then both started to appear that might imply a long period of relative peace followed by conflict and the population coming from centuries of relative peace might possibly not be as prepared for war as their opponents.
“If you had a farming culture with no weapons in burials and no fortifications … long period of relative peace”
More likely fortified villages constantly raiding their neighbors with small parties unable to comprehend war or government on a larger scale. We have the word ‘tribalism’ to describe such societies.
There are recently discovered (1930s) primitive farmers in Papua New Guinea, and they were cannibal head hunters with constant tribal warfare.
Ötzi the Iceman from 3,300 BC was equipped with an axe, knife and bow and died of bludgeoning and an arrow wound. “… unconfirmed DNA analyses claim they revealed traces of blood from four other people on his gear: one from his knife, two from the same arrowhead, and a fourth from his coat.”
Blood Feuds and Gun Runners: Papua New Guinea’s Tribal Battles
“Farmers in late farming Malthusian societies who lacked an almost obsessive focus on working hard over stealing (and stealing by killing) would be both punished by both poor harvests and by constabularies.”
They would also be punished by the other farmers doing the killing. And poor harvests aren’t a problem if you’re good at stealing.
A. J. West – I will give you the word “demonstrate” and grant that only the historical comparative method could “demonstrate” that Amerindian is a valid linguistic group. However suppose the genetic evidence could establish with a high degree of probability that the people of the New World (other than Eskimo-Aleut and Nadene speakers) are descended from a single migration event involving a small genetically homogeneous population. Then I would say that this genetic evidence would make the Amerindian hypothesis highly probable if not “demonstrated”.
“Then I would say that this genetic evidence would make the Amerindian hypothesis highly probable if not “demonstrated”.”
Highly probable, yes. Not demonstrated, at all. The point of combining linguistics and genetics is to use them as independent strands of data that can support one another, not for one to be used in place of the other.
Also, any hypothesised Amerind that developed because of genetic data would be utterly useless. It wouldn’t give us reconstructed terms and wouldn’t tell us anything about the culture or lifestyle of the Beringia-crossers. It also wouldn’t help us to identify the homeland of the speakers, because we already know that (or assume we know it) and there’s very little archaeological evidence to go on. So again, this Amerind wouldn’t do anything other than add a plausible but not demonstrated fact to hypotheses about the settling of the Americas. ‘They probably entered North America via Beringia in the late Pleistocene and were probably one population, as the genetic evidence attests. Oh, and they probably spoke one language, but we don’t really know much about that.’
“Highly probable, yes. Not demonstrated, at all.”
I think the problem people are having is the impression you give that the inability of linguistic methods to prove or disprove a particular thing has the same weight as methods that can prove something very closely related.
Yeah, it is looking pretty much like a slam dunk, A.J. Is it not OK for one discipline to be informed by the other? Is that not helpful? It happens in other disciplines – it’s perverse for it not to. Cannot linguistics say OK, well that’s looking very probable, so if we assume X, then Y? It’s not using genetics to replace linguistics, but to inform enquiry. That does not seem like replacement, or forcing the outcome of the linguistic findings – you can still use the linguistics to try to test the assumption, can you not?
Let’s be very clear: the language family would not be demonstrated. We wouldn’t have any words reliably reconstructed to a protolanguage and we wouldn’t have any understanding of the sub-divisions of the proposed family (those rely on an accurate series of sound and word changes). There’d be no linguistic evidence tying it together. We might reasonably suppose that all the languages are related, given the genetic evidence, but that’s as far as we could reasonably go. We might even be *certain* that there was a single language that entered the Americas, but we wouldn’t actually know anything about what that language was.
I don’t know how anyone could possibly object to this. Do you think genetic evidence can cause linguistic evidence to appear out of thin air?
By the way, it is easily possible for incredibly similar genetic material to be found in populations that speak different languages. Old Low Franconian and modern Danish are very different languages, but their speakers probably shared a huge amount of genetic material. Isolating the Anglo-Saxon- and Viking-associated DNA in England is a really hard task, but telling Old Norse and Old English apart isn’t so tricky. So if we had all this evidence showing that the founding population of the Americas was genetically one, we still wouldn’t be able to infer that they all spoke *exactly* the same language. They could easily have been as different as English and Dutch.
“Do you think genetic evidence can cause linguistic evidence to appear out of thin air?” No, I’m trying to understand. It seems like a mind-blowingly difficult sort of problem to me, but it also seems like an important thing to try to do. I’m trying to see why getting hints from other disciplines might not provide helpful possible models to test, but maybe it just doesn’t work like that.
A.J. – your discussion with Greg is very informative for me, and I hope you will keep it up.
AJ West
“But genetics isn’t everything and linguistics isn’t overruled by it. ”
It seems like linquists are defending a default position which is equally unprovable by linguistic methods. Saying Greenberg can’t prove x is fair enough. He might have just been a very good guesser – nothing wrong with that. But Greenberg not being able to prove x doesn’t mean y should be the default assumption if other evidence points at x being the default assumption.
It’s not that Greenberg was merely trying to prove that there was an Amerind language (although that was certainly part of it). He was trying to show what it was like. That’s what you have to do in order to say that any languages are related – you have to have shown that there is some reasonable way to account for the differences between them, some way that makes sense of the sound and word changes.
I suppose the idea that one language entered the Americas could be the default hypothesis. But there’s no linguistic evidence to go on at such a time-depth, so we have no way of knowing.
@AJ West
“but it seems that attacking *on horseback* is an Iron Age invention.”
@Toad
“Horses were mainly used to be able to shoot arrows at the enemy without being at risk of being counterattacked. Chariots allow an easier platform for wielding a bow with a separate driver to hold the riens.”
I think the horse/chariot archery came later. I think pretty much by definition – assuming horse size and strength is the limiting factor – the first chariots were light *one man* jobs like the Irish Traveller rigs which means javelins would have been a lot easier to manage.
Secondly I don’t think the first usage would have been in big organized armies and not really attacking on horseback in the way people might imagine that but tribal fights with the chariot guys charging and retreating – much more like police horse charges in riots – and if you’ve seen or been in a lot of those you know most people *can’t* stand when charged by a large mammal**.
(** some exceptions like they are too crushed together to move or they are very trained/disciplined as a group cf. Caesar’s march on Rome where his handful of cavalry broke Pompey’s untrained legions.)
You are right about the first chariots. The earliest excavated chariots from Sintashta-Petrovka sites are indeed small and narrow. Some people, trying to keep up the pretense that chariots were invented in the Near East, claimed that they were merely toys or copies of Near Eastern machines.
Cultural innovations aren’t as good as genes at conferring a long-lasting advantage – you can copy them.
Surely, only if they are genetically similar can they copy cultural innovations and compete with those who originated those cultural innovations.
However, I agree. Genes come first, and culture follows, and then genes adapt more closely to the culture, and so on.
Pingback: linkfest – 02/17/14 | hbd* chick
And, complimenting this:
Sweeping diversity seen in Mexican genomes