Sunday, June 3, 2012

Chris Ehret Book: The Civilizations of Africa

Over the last several weeks, I've been reading The Civilizations of Africa:  A History to 1800 by Christopher Ehret.  This book was published in 2002 and is based upon Ehret's work as a linguist studying African languages.  The Amazon link for this book can be found here.

The book describes the placement and economy of early Afroasiatic, Nilo-Saharan, Niger-Congo, Khoisan and BaTwa communities starting at approximately 21,000 ybp.  He describes how each of these communities employed and developed food acquisition methods in response to changes in their local climates.

Relevant to the discussion of domestication on the Nile is Ehret's description of two distinct cultures:  The proto-Afroasiatic culture originating along the Red Sea Hills and northern Ethiopian highlands and the proto-Nilo-Saharan culture of the middle Nile.  Map 4 of the book illustrates the proposed locations for these proto cultures:
It is proposed that the proto-Afroasiatic (Afrasan on the map) culture was initially focused on the collection of grains from wild grasses.  The proto-Nilo-Saharan speakers where intially hunters of large game as well as fishermen who gradually split into two groups, with one specializing in an agripastoral economy and the other focusing on aquatic resources along rivers.  During the initial development of these cultures, Afroasiatic and Nilo-Saharan speakers were separated by 300 to 400 kilometers of uninhabitable Sahara desert and by the cataracts of the Nile.  In recent posts on Nile domestication, an archaeological site representative of the early Afroasiatic tradition would be Wadi Kubbaniya (post).  Early Nilo-Saharan peoples are believed to have lived at the confluence of the White and Blue Nile (post).

These groups gradually expanded outwards from their homelands, especially after 10500 bp during the phase of the green Sahara.  Ehret refers to this phase as the Holocene Climatic Optimum.  (On this blog, it has previously been referred to as the Great Wet Phase of the Sahara.)  Ehret points out that during these expansions, the Afroasiatic and Nilo-Saharan cultures came into contact with each other.  There is linguistic evidence of cultural exchange, including the exchange of knowledge about plant and livestock cultivation.  The forms of exchange are quite detailed and I would be doing the author a disservice to copy his recently published work directly on this blog.

Ehret's book is available on Amazon for a modest price. I have read the first four chapters and  I believe the book deserves at least four stars, so don't be fooled by the review of one disgruntled reader.  In addition to the in depth discussion on Afroasiatic and Nilo-Saharan cultures, there is the remarkable early history of Niger-Congo, Khoisan and BaTwa peoples.  African domestication of various crops such as cotton, sorghum, yams, millet, black-eyed peas, palm oil and African groundnut are explored. The book is easy to read and does not require any specialist knowledge of African archaeology or linguistics.

Regarding the recent discussion on this blog, Ehret's linguistic deductions do support the assertion that cattle were domesticated independently by the Nilo-Saharans at about 10,000 bp.  He also suggests that Afroasiatic speakers of the Red Sea hills may have engaged in a form of "protection of wild cows".  As of the 2002 publication, Ehret stated that the exchange of knowledge regarding cattle herding between Nilo-Saharans and Afroasiatic speakers of the Red Sea hills is as yet poorly understood.


andrew said...

FWIW, I think his dates are far too old and I don't see much merit in the sequencing either. But, more data is still better than less data.

Marnie said...


You can "think his dates are far too old" and not see "much merit in the sequencing", but unless you have references to the contrary, his linguistic dates predate and have stood up under recent genetic discoveries by Cruciani on Chadic speakers, and genetic research on Niger-Congo, Nilo-Saharan, and Afroasiatic speakers, by Tishkoff. Many years of climate research on Africa, as referenced on this blog in the last month, and archaeological research in Africa over many years, also support his dates.

Separately, recent genetic research on many crops such as millet, sorghum and yams, and domestication of animals, including the donkey, cattle and guinea fowl, support the case for intense local domestication of crops and animals in Africa, in multiple locations, starting at the beginning of the Holocene Climatic Optimum.

Additionally, linguistic and Capsian archaeological findings support the adoption from the Near East of goats and sheep in North Africa not later than 8500 ybp. Shortly thereafter, these animals were adopted into the Red Sea hills and Nilo-Saharan societies. However, linguistic data indicate that this adoption is predated by the local domestication of cattle.

It is not my intent here to cover in detail all of Ehrets work and methods of dating. I have referenced him for those that are truly interested in this cutting edge area of research.

However, for those that are content to assume, without evidence, that African domestication is unworthy of attention, there is not much that can be done, but to let you wallow in the abyss of ignorance.

Andrew, I'm sorry if that sounds harsh, but as the evidence increasingly becomes available, it is truly appalling that so many refuse to even consider that Africans experimented with domestication at about the same time as other peoples of the world.

andrew said...

A 8500 ybp date for sheep and goats is 6500 BCE, which is about 4500 years more recent than the estimate for the "Middle Nile tradition." The estimate for cattle from one your recent posts is 4000 BCE in North Africa, and later in East Africa.

As you note in a post as your blog citing Zeder (2008): ""Domestication and early agriculture in the Mediterranean Basin: Origins, diffusion and impact" has these domestication dates BP: sheep 11,000; goats 11,000; pigs 10,500; cattle 10,000(Upper Euphrates Valley). However, she states that morphologically altered domestic cattle are not found in Central Anatolia until after 8,500 BP." So, sheep and goasts aren't domesticated anywhere until 9000 BCE, let alone Africa, which gets the Neolithic package later than the Fertile Crescent.

Erhert has a Middle Nile culture in place two thousand years earlier than that, a Niger-Congo Sahel farming belt a thousand years earlier than that, and an Afro-Asiatic protolinguistic community over a pretty broad geographic range four thousand years before the first animal is domesticated in the Fertile Crescent. These look way to early. How does he suppose, for example, that pre-Neolithic Niger-Congo peoples managed to sustain a civilization in such a large area without any domesticated plants or animals?

There is a case for an independent domestication event contemporaneous with the Fertile Crescent Neolithic in the Sahel with a different set of crops is plausible pending evidence to support it (which is too mixed to really render as verdict as far as I can see about when it happened to any great precision), so a 10,000 BCE date, while very early (one of my current projects is to pin down better the earliest archaeological evidence for Sahel crop domestication, I've seen Fuller discuss it a bit, but not recently and the Jared Diamond disucssion in GGS is hopelessly outdated and thinly sourced), with very little archeaological support for Niger-Congo, isn't entirely incongruous, although to large an area to early to fit the proto-language reconstructions to a much small locus at the earliest point.

Tiskkoff puts a Nilo-Saharan ethnogenesis at 8500 BCE, thousands of years after the Ehret date, and a date that is already somewhat doubtful given the much later dates he provides for secondary dispersals of that population (5000 BCE to 2000 BCE).

The Khoisan date of 16,000 BCE, in contrast, is much younger than the genetic evidence suggesting Khoisan divergence genetically from other Africans (apart from Pygmies) ca. 35,000-70,000 years ago, depending on the methodology, and the geographic range likewise seems far too small given the existence of click language from the Kalahari to East Africa, with lots of territory in Mozambique in between, and the strong inference that their range has been shrinking steadily since before Bantu expansion. Tanzania and all points South were predominantly, perhaps almost exclusively hunter-gatherer until Bantu expansion reached them ca. perhaps 1000 BCE-5000 BCE.

(Part I)

andrew said...

(Part II)

There is just no evidence at all that independent domestication was happening in the proposed Afro-Asiatic proto-culture region two thousand years earlier than it was in the Sahel - to the contrary, the evidence seems to point towards a later adoption of domesticated plants and animals for food production there than in the Sahel, if anything. There is no evidence of any local domestications in that region any sooner than Mid-Holocene (perhaps 4000 BCE give or take), and the case that the local domestications were secondary to Fertile Crescent crop adoptions is pretty good, because the local domestications weren't dietarily complete - you couldn't have a sedentary society based on eating them and couldn't stay sedentary in a non-fishing economy with the kind of hunting and gathering they did. It is very hard to make a case for a widespread Afro-Asiatic prior to about 6000-7000 BCE, and argument that it eminates from East Africa with E1b1b in the Upper Paleolithic rather than in the Nile with the arrival of the Fertile Crescent Neolithic as it expands in all directions from there. The notion of Afrasian skirting the Red Sea and then hopping the desert from there to the Nile without following the Blue Nile basin is likewise mysterious. And, where is there evidence that the proto-Afrasians were coastal fishermen ca. 13000 BCE?

I would also argue pretty strenuously that while there is considerable genetic continuity of population in North Africa from the Epipaleolithic (at least 12,000 BP) to the present, that arrival of Near Eastern goats and sheep in North Africa probably gave rise to language shift in the population that is now Berber.

There is no one genetic continuity that covers all Afro-Asiatic language speakers very consistently (one can argue for E1b1b, but once you get into subhaplogroups it doesn't look really compelling) so some of the major families must owe their languages to language shift, and the advent of the Neolithic is the logical time for that to happen. The Chadic language family, judging from its distinctive R1b-V88 Y-DNA haplogroup also can't be that old - "The R-V88 coalescence time was estimated at 9.2-5.6 [corrected] kya, in the early mid Holocene.", i.e. 7,200 BCE-3,600 BCE, iwth a mean of around 5,400 BCE. Much of the diversification of Semitic has happened in the historic era in the period around the time that Akkadian was the language of the Sumerian Empire (ca. 2000 BCE) to perhaps one or two thousand years earlier. Somolian genetics show massive back migration that was almost surely Neolithic. In short, whereever Afro-Asiatic began, the case that it was widespread prior to the arrival of the Fertile Crescent package in Africa from the Near East, or that it expanded greatly until it was riding that wave of expansion is extremely thin.

Some of the North African-East African genetic region (with M1 and U6 back migration in mtDNA and E1b1b expansion in Y-DNA) is probably a prior genetic layer that is probably not in continuity linguistically.

Marnie said...


Regarding your confusion of the date for insipient Northern Sudanic domestication, I believe that much of your confusion on this topic could be cleared up by understanding that 8500BCE corresponds to 10,500 ybp.

It would also help if you would read some of the definitive authors, including Ehret, on the spread of Northern Sudanic culture into the Sahel. It is clear from some of your comments that you have not.

andrew said...

Just to be clear, what I am comparing are the dates on the map (Niger-Congo at 10000 BCE, Middle Nile Culture 11000 BCE, Afrasian 13000 BCE, and Khoisan 16000 BCE) in the places shown to the other evidence, and arguing first, that they are too old and too broad (except for Khoisan which is too young and too small), and second, that they are out of sequence - with Middle Nile and Afrasian very likely younger than Niger-Congo.

I also don't put much stock in the case that cattle were domesticated from local auroch stock in the Nile basin of Northern Sudan rather than arriving from the Fertile Crescent at 10500 bp/8500BCE. And, so far as I can tell, there isn't evidence for local goat and sheep domestication of any kind in Africa (although the donkey surely was an African domesticate). Domestication is such a powerful group survival tool that once it appears anywhere, it is going to explode everywhere it can go in a matter of a thousand years or two.

I don't doubt that Sudanic cultures expand into the Sahel and end up bringing Nilo-Saharan languages far to the West. But, the most plausible time for that is exodus from the Lake Chad basin ca. 5000 BCE, with Niger-Congo Sahel farmers probably moving south into their current range as that dries up and vacating the place the current Sahelian populations move into.

Likewise, I don't seriously question the phylogeny Erhert comes up with within Nilo-Saharan, nor the notion the Nilo-Saharans and Afrasians may exchange words related to pastoral life. But, it seems to me that he is putting those exchanges to far in the past, given what we know about when cattle domestication becomes widespread in North Africa and East Africa.

Paleolithic to Neolithic transition is not subtle. It can't be missed and it spreads intensely where it can.

As another reason to doubt Ehret, look at his exclusion of the Niger-Beneu river confluence from the Niger-Congo area when many would put that confluence as the urheimat of the entire Niger-Congo language family by virtue of phyogenetic relationships.

I'm certainly open to reading more and learning more. Good sources are hard to come by, at least where I live and without an affiliation with a university library, and I don't want to put out money to buy e-books from someone until I've seen their published scholarly work in articles that I trust. But, prehistory studies in general, and linguistics in particular, is also full of lots of materials (e.g. Russian school linguistics) that get far more credibility than they deserve, and both linguistic and genetic evidence is quite unreliable at providing absolute ages. The most reliable evidence, when it comes to dates, is still archaeological evidence.

Marnie said...

"Just to be clear, what I am comparing are the dates on the map (Niger-Congo at 10000 BCE, Middle Nile Culture 11000 BCE, Afrasian 13000 BCE, and Khoisan 16000 BCE) in the places shown to the other evidence, and arguing first, that they are too old and too broad (except for Khoisan which is too young and too small), and second, that they are out of sequence - with Middle Nile and Afrasian very likely younger than Niger-Congo."

Andrew, you have spent over a thousand words in the above comments arguing against the 10,000 ybp date for cattle domestication by proto-Nilo-Saharan speakers. Now, somehow, you are trying to weasel out of those statements and say that your argument all along was about dates on Map 4, clear labelled as "proposed early lands" and not domestication dates.

"I also don't put much stock in the case that cattle were domesticated from local auroch stock in the Nile basin of Northern Sudan rather than arriving from the Fertile Crescent at 10500 bp/8500BCE."

Yes, Andrew, that pretty much sums up your position. You don't read anything that Ehret, Tishkoff, the Bovine HapMap Consortium, Gifford-Gonzalez, A. Muzzolini, Honegger and many others have to say, but somehow, you are the expert on domestication in Africa.

"And, so far as I can tell, there isn't evidence for local goat and sheep domestication of any kind in Africa (although the donkey surely was an African domesticate). Domestication is such a powerful group survival tool that once it appears anywhere, it is going to explode everywhere it can go in a matter of a thousand years or two."

Regarding this comment, you would know that

(1) there was no contribution to goat and sheep domestication in Africa and
(2) there are never reversals of domestication, are not always true

had you read the Gifford-Gonzalez paper that I posted today. I guess you didn't read that either.

It is one thing to not have read important references, but entirely another to deliberately proceed without information, intractably stating a position that is contrary to the weight of years of zooarchaeological evidence, multiple avenues of genetic research, linguistic analysis, geological data and climate research from leaders in the field. It is particularly ridiculous when one knowingly hasn't even read most of this research, even when much of it is available for free on this blog or for a very modest price from Amazon used books.

pconroy said...

Marnie, Andrew,

Are you guys familiar with this - now defunct - blog:

There is a wealth of information there on domestication and linguistics. Matilda herself had to retire due to illness.

andrew said...

Matilda's blog is a treasure which I read quite a bit when it was active and still refer to.

Marnie said...

Andrew and Paul,

Thanks for the reference to Matilda's Anthropology blog. Yes, she did cover a lot of the African domestication and linguistic material. I'm sorry to hear that she can't post anymore because of an illness.

I do still come across much of the material she posted. Here are some good posts of her's that are relevant to recent discussion on this blog:

I will say that Matilda has made a lot of postulations of her own, without supporting them with data. She takes on prominent researchers without backing up her personal hypotheses. For example, I don't think it is at all helpful toward understanding the complex population history of Africa for her to make sweeping statements that associate socially understood terms of race such as "Caucasian" with genetic haplogroups.

Titles like "Berbers are not closely related Sub Saharan Africans" are misleading. Which Berbers? When? In fact, in this case, she misinterprets the results of the paper she is referencing.

I will say that Matilda did do a great job of blogging on these papers, but I think you have to read her interpretations with a very wary eye.

I'll also say that I am not an anthropologist or linguist, but for exactly that reason, I try not to come to sweeping conclusion on this blog. If I do reach some conclusions, its after months of reading many sources on a particular topic.

Anyway, we do owe Matilda a big thank you. I haven't covered some of the papers on African millet and sorghum domestication and for that reason alone, Matilda's posts would make a good read.

Marnie said...

Paul and Andrew:

I was just reading Matilda's comments on the Ehret paper on Semitic languages. Some of Matilda's comments are really off the mark here and are really off the path compared to current research findings.

Definitely use the material on Matilda's blog as a starting point, but do realize that she is not a recognized researcher. If you are interested in the current understanding on these topics, I'd read more than one source and certainly, some sources in addition to Matilda's editorial.

Andrew Oh-Willeke said...

Not to stir the pot too much, but Ehret's own work in linguistics is quite controversial (see, e.g. citations at the wikipedia page on him).

I presume Matilda is a pseudonymn, but would welcome any news any insder would have on her well being. I think of her now and then and wonder.

Marnie said...

Andrew, please be more specific regarding your implication that Ehret's work is controversial. Please provide a specific link. I don't see anything at the Christopher Ehret wiki page.

If you have problems with Professor Ehret's research, you'd best take them up with Ehret, not me. His UCLA page and contact information are here:

Andrew Oh-Willeke said...

I am referring to materials such as the quote below from Wikipedia's Christopher Ehret entry:

"His linguistic works include A Comparative Reconstruction of Proto-Nilo-Saharan (2002). His reconstruction was widely criticized by scholars of the field, notably Lionel Bender and Roger Blench. (

His other dictionary,Reconstructing Proto-Afroasiatic (1995) appeared at the same time with another important dictionary of the same linguistic family. Robert Ratcliffe assesed both in a strong-worded article, (Afroasiatic Comparative Lexica: Implications for Long (and Medium) Range Language Comparison) Ratcliffe observes that the two dictionaries differ widely, which suggests both are highly controversial.

He published also The Historical Reconstruction of Southern Cushitic Phonology and Vocabulary (1980). He has also written monographic articles on Bantu subclassification, on internal reconstruction in Semitic, on the reconstruction of proto-Cushitic and proto-Eastern Cushitic, and, with Mohamed Nuuh Ali, on the classification of the Soomaali languages. These reconstructions have not been well received, and are not followed by other linguists."

"He has also collaborated . . . in developing mathematical tools for dating linguistic history (e.g., Andrew Kitchen, Christopher Ehret, Shiferew Assefa, and Connie Mulligan, "Bayesian phylogenetic analysis of Semitic languages identifies an Early Bronze Age origin of Semitic in the Near East," Proceedings of the Royal Society B: Biological Sciences, July 2009)." Atkinson's methods, which are followed in the Kitchen (2009), are also quite controversial in the linguistics community (a follow several linguistics blogs that have discussed those papers) although the Semitic languages paper is one of the better implementations of that methodology that I have seen.

I don't have a terrible beef with Ehret, but when you review the literature and there is scholarly disagreement in the field, you have to evaluate credibility and controvesy within the community to make sense of your sources.

Moreover, the further afield a scholar is from his core area of research and distinction. Ehret's core expertise in the history of the African Classical Age, from 1000 BC to 400 AD in East Africa as a "classical age", not in linguistics of deep pre-history of African predating the Neolithic where many linguists have cast serious doubt on linguistic evidence to resolve anything at that time depth, and not in his knowledge of archaeological evidence.

Atkinson's methods in linguistics, and the methods of geneticists are also, in general, more amenable to providing accurate relative dates than accurate absolute dates, because the methods used to calibrate them are not very solid - they have not been rigorously cross checked against dates derived from other methods. Atkinson's early methodology, adopted by Ehret in the 2009 paper which likely influences the dates in his book which you cite, also ignores a factor included in some of his later papers: languages undergoing punctuated rapid evolution in addition to their overall normal evolution rate when they are born - a process that has been documented in linguistic experiments.

Marnie said...
This comment has been removed by the author.
Marnie said...

Andrew, have you read the comparative paper by Roger Blench on the difference in methodology between Bender and Ehret?

Having read the Blench review, it appears there are differences in methodology between Bender and Ehret, but apart from the inclusion or not of a few languages in the Nilo-Saharan tree, their methods don't result in significant differences in the reconstruction of Nilo-Saharan.

Both Blench and Ehret have chapters in the Archaeology of Africa Book (Thurstan Shaw, editor). They both discuss cattle domestication in Africa. Whatever their differences regarding the nuances of Nilo-Saharan language tree, they both agree that cattle were present in Africa at an early age.

Her is Roger Blench himself quoted on African longhorns from one of my previous posts:

Ehret's work on Nilo-Saharan has recently been supported by the genetic research of the Tishkoff team. In fact, Ehret is a coauther on this widely regarded Tishkoff et al paper:

Tishkoff also references Ehret's work in this paper:

Ehret was also an invited speaker at the recent African Genetics International Conference:

Your assertion that Ehret's work has been broadly not well received is simply untrue. Tishkoff's genetic research if anything strongly supports Ehret's Nilo-Saharan dates inferred from linguistics.

You mention the linguist Atkinson. If you can put up a link of Atkinson's work in which he provides specific dates for the Nilo-Saharan language tree, that would be most helpful.