The sons of Ham and Shem


Recently I had the pleasure of having lunch with David Reich and he asked me about my opinions in relation to the Afro-Asiatic languages. I thought it was a strange question in that I get asked about that in the comments of this weblog too. Why would I have any particular insight? I gave him what I thought was the likely answer: Afro-Asiatic languages probably emerged from the western Levant. The ancient textual evidence indicates that to the north and east of Mesopotamia the languages were not Semitic. Though Akkadian, a Semitic language, was present at the dawn of civilization, Sumerian was the dominant language culturally in the land between two rivers, and it was not Semitic. As Lazaridis et al. did not detect noticeable Sub-Saharan African ancestry in Natufians, or later Near Easterners, I have become skeptical of any Sub-Saharan African origin for Afro-Asiatic.

But after the earlier post I made a few mental connections, and so I’ll put something up which pushes forward my confidence on a few issues. They lean predominantly on Y chromosomes. I understand that this sort of phylogeography has been shown to be not too powerful in the past, but in the scaffold of the ancient DNA framework it can resolve some issues.

About a decade ago study of Adolf Hitler’s paternal lineage (through male relatives) indicated that his haplogroup was E1b1b. Though reports that Hitler was non-European, because this is a very common lineage in non-Europeans, as well as Jews, were incorrect, it does turn out that Hitler’s paternal lineage is not associated with the Indo-European migrations. That is, unlike me, Adolf Hitler does not descend from the All-father, but rather one of the men who were conquered and assimilated by the steppe pastoralists.

But E1b1b is an interesting lineage. First, it is very common in much of Africa, especially the north. Second, it is common among the Natufian people according to Lazaridis et al. In contrast the Neolithic Iranian farmers seem to have harbored haplogroups J. Today the Near East is a mix of the two, which makes sense in light of the fact that reciprocal gene flow has occurred in the last 6,000 years.

Looking at E1b1b frequencies you notice a few things. The highest frequencies with large N’s are in the Cushitic and Berber languages. Haplogroup J has a different distribution, being skewed more to West Asia. In Ethiopia E1b1b is more common, but J is far more prevalent among the Semitic Amhara than the Cushitic Oromo. Though it is subtle autosomal DNA makes it clear that the Semitic speaking populations in Ethiopia-Somalia have more Eurasian ancestry than the Cushitic ones. I believe this is evidence of the multiple migration pattern discerned earlier.

If you go further south in East Africa and compare E1b1b and J you see a skew in the ratio. E1b1b declines in frequency, but J basically disappears. Among the Masai, who have a clear minor West Eurasian ancestral component, albeit far less than Ethiopians, 50% carry E1b1b. Among the Sandawe, who are a language isolate  with clicks, but exhibit Cushitic genetic affinities, 34% carry E1b1b. Among their Hadza hunter-gatherer neighbors, 15% do so. Among many Khoisan groups the frequency of E1b1b is 10%. Most of these groups exhibit no J haplogroup. This aligns easily with what Skoglund was reporting earlier: the first pastoralists had no “eastern farmer,” but did have “western farmer.” The Natufians were E1b1b. The wider reach of E1b1b in Africa in comparison to J is likely due to the fact that the admixed pastoralists were pushing into relatively virgin territories. Later Eurasian backflow events, which brought Semitic languages, encountered a much more densely populated Africa.

The hypothesis I present is that after the descendants of the Natufians made the transition to farming, some immediately pushed into areas of Africa suitable for farming and/or pastoralism. They quick diversified into the various Berber and Cushitic languages. The adoption of Nilo-Saharan languages, and later Khoisan ones, was simply the process of successive and serial admixture into local populations as these paternal lineages introduced their lifestyle. In the Near East many distinct Semitic languages persisted across the Fertile Crescent, and for whatever reason the various non-Semitic languages faded and Semitic ones flourished.

23 thoughts on “The sons of Ham and Shem

  1. How funny. I found out just a couple days ago that my paternal haplogroup is an offshoot of E1B1B l, and I was looking through your posts for references to it. It is unexpected for me as my surname and family history on my father’s father’s side is decidedly British–making us an unlikely carrier.

    For what it’s worth, my own amateur exploring of the haplogroup’s history agrees with your hypothesis.

  2. Hmm. Didn’t a significant amount of Northwest Africans migrate from Western Europe in prehistory? Hence how so many of them are rather fair skinned?

  3. Speaking of David Reich, you have mentioned a couple times on twitter that he’s writing a book about human history as revealed by ancient DNA. Is this true? I understand if you have to keep secrets, but is there anything you know about it?

  4. C. M. Campbell, i haven’ tasked for a draft. i don’t think u’ll be surprised.

    Hmm. Didn’t a significant amount of Northwest Africans migrate from Western Europe in prehistory? Hence how so many of them are rather fair skinned?

    1) no.

    2) prehistoric western europeans were very dark skinned according to the most recent ancient dna.

  5. Razib Khan, the consensus within linguistics is that Afroasiatic emerged in North East Africa. Please stop misleading your readers. This not a fantasy novel you are talking about here. It’s actual history!

  6. Razib Khan, the consensus within linguistics is that Afroasiatic emerged in North East Africa. Please stop misleading your readers. This not a fantasy novel you are talking about here. It’s actual history!

    ah, the linguistic consensus. i’m shaking my boots!

  7. So dear ol’ Uncle Addie was not an Aryan (R1a) and not even an Indo-European (R1b). Ooops!

    It gets even more ironic when you consider that the Slavs he attacked out east were disproportionately R1a.

  8. I was talking about modern Northwest Africans (Moroccans, Algerians, Tunisians) being fair-skinned. See here:

    many of the genes for light skin come from the middle east into europe. you don’t need european ancestry to explain light skin outside of europe.

    1) there is selection for it in many cases

    2) western europeans themselves were not particularly light skinned in all likelihood relatively recently in the past

    3) there is minimal evidence of european gene flow into NW africa, though some. probably mostly through mozarbs -> moriscos who left spain mostly. and some in antiquity.

  9. A Levantine origin for Afroasiatic would tend to imply that Semitic and non-Semitic should be the fundamental clades of AA. That does not seem to be the case based on the linguistic data. The picture would be complicated a bit if we posit more than one pre-Semitic AA migration event into Africa. Several layers of intrusion into Africa would create the illusion of an eastward migration out of Africa if the language communities that remained in the Levant failed to diversify (or there, of course, was diversity that was subsequently wiped out with no traces). But that scenario is starting to get a bit convoluted.

    You could also get some mileage out of questioning what really is Afroasiatic after all, e.g. some have questioned whether Omotic is really part of the family or simply evinces extensive contact effects.

  10. The previous comment about linguistic consensus regarding the Afroasiatic Urheimat is false. Alexander Militarev argues in favor of a Levant origin:

    Proto-Afrasian Lexicon Confirming West Asian Homeland: Pastoralism (PDF)

    As near as I can tell, this debate seems to be between a very small group of academics, among whom actual linguists can be counted on one hand. Note the opposing position relies on claims such as “the archaeology of northern Africa does not support demic diffusion of farming populations from the Near East”.

    http://science.sciencemag.org/content/306/5702/1680.3

  11. For my part I think the “Afro Asiatic Language Family” is too wide, shallow, and ill-attested at the fringes to call it a family at all yet. How does that model compare with, say, a set of overlapping Sprachbuende?

    To give an example of a language family that works, I offer Indo-European, where everyone agrees first exited Anatolia, next Tocharia, last the ancestors to the Indo-European languages known to William Jones. Another example is, it so happens, Semitic (East Semitic first out the door in this case). But with “Afro Asiatic” instead I have seen too many mutually-incompatible tree diagrams.

  12. Militarev and no other linguist.

    Other than Christopher Ehret, who else is even relevant at this point? Ehret actually uses the E1b1b to argue in favor of African origin. Quoting again from the text I linked to in my last post:

    The geography of the M35/215 (or 215/M35) lineage, which is of Horn/East African origin, is largely concordant with the range of Afroasiatic languages.

    That was before genetic data from Natufians had been published.

    I think other linguists who’ve spent considerable effort addressing the issue can be forgiven for not considering the genetic evidence, since most have been dead for several years.

  13. Can you share what David Reich had to say on the subject?

    I have heard the idea before but evidence indicates the Phylogeny of E-M35 lineages dont support a origin and diversity in the Southern Levant with that of Africa being levantine subsets. Its quite the oppose with the African side of the Red Sea having deep rooted lineages and diversity simply not found in West Asia.

    Also you have to question why Natufian “Farmers” enter Africa so early, Early enough for African specific Afroasiatic language groups (Berber, Chadic, Egyptian, Cushitic, and arguably Omotic and Ongota…In other words the majority of AA) to have such deep time-depth, show linguistic exchanges with Nilo Saharan and Niger-Kordofanian all the while not actually bringing “Farming” technology which only shows up some 7 to 6 THOUSAND years after being discovered by Natufian.

    This argument by itself is hard to take seriously, at nearly every avenue the cornerstone arguments are contrary to the data. Here is an analogy that may help you understand:

    How would I look if I argued that R1 carrying Indo European South Asians brought Indo European languages and R1 derived lineages into Central and Western Europe? Perhaps we can use the “absence of evidence is not evidence of Absence” approach and argue that R1a/R1b diversity in India as well as phantom/unknown Indo-European languages that “may have existed” have simply been “Erased” and lost over time due to climactic events and the turnover of ancient civilizations.

    I think its almost a Red Herring to look for Sub Saharan specific ancestry as signatures of hypothesized Afroasiatic speakers from Africa (Especially if you believe the Horn of Africa has harbored fully “Eurasian” populations in Antiquity). Most of these AA Groups sans the potential Omotic and Ongota show a migrations from North Africa. Maybe you should consider African signatures different than those concentrated below the Sahara? I think its quite obvious that the E-M35 diversity you see in modern Non-Africans and Ancient non-Africans was not introduced by populations with predominant Sub Saharan ancestry, See Natufians. On the other hand, Some of these populations CAN though be modeled as partly North African: See Natufian.

  14. thanks for the comment.

    david had no thoughts he suggested to me. he just asked what i thought.

    need to think on your comment a bit more and how it would work with demographic models.

  15. I am open to a Levantine origin for Afroasiatic. That, or highly Natufian-like North African Proto-Afroasiatics (if such people existed in Mesolithic North Africa), are pretty much the only options as I see it.

    I am more skeptical about E1b1b1-M35. Sure, modern variation indicating a North African origin can be deceiving, so the Natufian M35 aDNA is important. However, modern M35 is not the only piece of evidence, which includes M215(xM35) lineages in southern Ethiopia (M215 is about 35 ky old compared to 25 ky for M35), and the more general African distribution of E1/E1b1 clades. Natufians may lack SSA admixture in formal stats, but they do show some minor “African” affinity in TreeMix, ADMIXTURE etc. lacking in West Asians outside of the Levant, which may represent the influence of some unsampled North African group. Natufians also happen to sit right at the African border, and some archaeologists have suggested a NE African influence in the Levantine Epipaleolithic.

    I must also add some complexity regarding J and E, Cushites and Semites. The overall conclusion about an early wave of western farmers and a later wave (e.g. by Semites) containing some eastern farmer ancestry is certainly accurate. But there really is no dichotomy between Cushites and Semites with regard to Y-DNA J. The bottlenecked South Cushites, along with Somali and southern Oromos with very limited Y-DNA diversity, largely lack Y-DNA J. However, Central Cushites, Afar East Cushites, Oromos from regions of higher population density in central Ethiopia, many Highland East Cushites, and Beja, have a lot of Y-DNA J. From this thesis with very large sample sets from all groups, Oromo showed 24% J, Afar 26%, Agaw 22%, Alaba 24%, Amhara 26%, Tigray 21%. Add the fact that Omotics are understudied, the only typed Omotic group showed minor J(xJ1,J2) that has never been reported in Cushitic/Semitic Ethiopians, and the importance of J1-P56 in East Africa, which is largely absent in West Asia and has a Neolithic TMRCA. You get the feeling that the story of J in East Africa looks more complicated than it is usually given credit for.

  16. @Lank

    Regarding J in both Cushites and Semites, there is also a linguistic parallel. One of the confusing aspects of determining the Afroasiatic phylogeny is the fact that Cushitic and Semitic are in many ways more similar to each other than either are to Egyptian.

    Also related, Militarev and Belova claimed Old South Arabian had a Cushitic substrate.

  17. @Labayu “That was before genetic data from Natufians had been published.”

    You sound confused. The Natufians carried E1b1b , an African paternal clade. Yet in your mind, it’s evidence for AA having a non African origin. The proposed homeland for AA is NA, Southern Egypt and Northern Nubian, close to the Red Sea coast. How on earth does Natufians not having genetic affinities with modern SAAS have any bearing on that fact?

  18. I distinctly remember Lazaridis et al stating very clearly that E1b1b in Natufians supports the idea that Natufians had at least partial African ancestry. I am simply puzzled as to why Lazaridis et al is being argued to support the idea of a Levantine origin for AA:

    “Craniometric analyses have suggested that the Natufians may have migrated
    from north or sub-Saharan Africa, a result that finds some support from Y chromosome analysis which shows that the Natufians and successor Levantine Neolithic populations carried haplogroup E, of likely ultimate African origin, which has not been detected in other ancient males from West Eurasia”
    http://www.biorxiv.org/content/biorxiv/early/2016/06/16/059311.full.pdf

    PS: I did not mean to come across as an asshole. Frankly, I fail to see the assholeness in anything I have said.

  19. next two sentences after what you quoted:
    However, no affinity of Natufians to sub-Saharan Africans is evident in our genome-wide analysis, as present-day sub-Saharan Africans do not share more alleles with Natufians than with other ancient Eurasians (Extended Data Table 1). (We could not test for a link to present-day North Africans, who owe most of their ancestry to back-migration from Eurasia.

  20. Razib, the AA urheimhat is proposed as NA—Southern Egypt and Northern Sudan. That is not SSA. A lack of affinity with modern day SSAs has no bearing on a proposed North African origin of AA. I think you are conflating SSA with the whole of Africa. The traditional arcaheology which Lazaridis makes reference to talks about SSA affinity in Natufians mostly in terms of morphological features. It simply means that Natufians had facial features typical of modern day SSAs. However, most of that research suggests that the African cultural links of Natufians are rooted in North Africa in the Egyptian and Northern Sudanese Nile Valley. There is nothing in Lazaridis et al. which contradicts that.

Leave a Reply