Bayesian phylogenetic analysis of Semitic languages identifies an Early Bronze Age origin of Semitic in the Near East

Bayesian phylogenetic analysis of Semitic languages identifies an Early Bronze Age origin of Semitic in the Near East

The evolution of languages provides a unique opportunity to study human population history. The origin of Semitic and the nature of dispersals by Semitic-speaking populations are of great importance to our understanding of the ancient history of the Middle East and Horn of Africa. Semitic populations are associated with the oldest written languages and urban civilizations in the region, which gave rise to some of the world’s first major religious and literary traditions. In this study, we employ Bayesian computational phylogenetic techniques recently developed in evolutionary biology to analyse Semitic lexical data by modelling language evolution and explicitly testing alternative hypotheses of Semitic history. We implement a relaxed linguistic clock to date language divergences and use epigraphic evidence for the sampling dates of extinct Semitic languages to calibrate the rate of language evolution. Our statistical tests of alternative Semitic histories support an initial divergence of Akkadian from ancestral Semitic over competing hypotheses (e.g. an African origin of Semitic). We estimate an Early Bronze Age origin for Semitic approximately 5750 years ago in the Levant, and further propose that contemporary Ethiosemitic languages of Africa reflect a single introduction of early Ethiosemitic from southern Arabia approximately 2800 years ago. pdf

While I know this is old news, I’m  having an Afro-Asiatic weekend, and I missed this last year. This isn’t a million miles away from the conclusion I got from examining the reconstructed proto words of proto Semitic a while ago; only about 750 years different (based on the presence of silver and antimony). 

I finally found my old work that records Ehret’s older date for proto Semitic as being the roughly similar to Proto Cushitic (about 10,500 bp to 10,000 bp). If this isn’t an admission that he’d seriously overestimated the dates in his earlier work on AA, I don’t know what isn’t. It supports my criticism of his dating of proto Cushitic as being about 40% out, considering how slashed the time frame for Semitic is (nearly half). This brings all his dates for AA languages into a Neolithic time frame if you apply the same rule to them as a group. I was amused to see him clinging onto an African origin for PAA like grim death, given that the DNA, archaeology and proximity of Semitic languages to PIE and Sumerian really doesn’t leave this as a viable option anymore. Essentially the ‘African origin’ needs to find a cultural or biological expansion from Africa dating to about 11k ago, and AFAIK so far the only suitable population movement across North Africa in that era goes from Asia into Africa (Capsian culture). Then it needs to find a really good reason why Nubians speak a Nilo Saharan language, when they apparently shared a common culture with Afro-Asiatic speaking upper Egyptians before the Neolithic hit Africa.

Some people (who shall remain nameless) are adamant that the E3b1 Y chr is the Y chr related to the expansion of Afro-Asiatic languages. This is incorrect in at least two cases, as Chadic is not at all associated with this, but with the R1b-V88, which arrived from Asia in the neolithic (or Holocene), and neither is Semitic which is Asian and associated with the spread of J1 (also non-African). Afro-Asiatic languages are associated with the expansion of Neolithic Y chromosomes, not E3b1. I’d also like to add that in expansions of this type, languages appear to be spread patrilinealy, not matrilinealy, the Bantu expansion and observations from other expanding farmers v hunter gatherers don’t have the maternal hunter gatherers language being preserved.

My main objections to an African origin for Proto Afro-Asiatic.

The nouns don’t reconstruct to a ceramic using African hunter gatherer landscape, but to an aceramic West Asian early Neolithic landscape. 

There are no population movements out of Africa known that are recent enough to have carried proto Semitic into Asia. The last one was about 22,000 years ago, traced by the expansion of the M78 Y chr and the Kebaran culture (the Mushabian’s origin is open for debate, but it’s also too early). This is just impossibly old. Anything more than 11,000 years and you can’t be in the same language family. So many accumulated changes will have occurred after that time that they will only bear the same level of relation to each other as two random non related language groups. 

The known population movements between Africa and Asia that could have carried the language between them in the relevant time frame all go into Africa from Asia. 

Proto Semitic shows a proximity to both Sumerian and Anatolian Neolithic proto Indo European, which places it in Asia about 9,000 years ago. Even if you think the Anatolian theory is junk, it has loan words into the later PIE about 5k ago. This leaves a narrow time frame for the movement between Asia and Africa (11k max language age – 9k for AN PIE=2,000 years), and the arrival of Neolithic farmers into Africa via the Sinai sits smack in the middle of this time frame. 

Corrected dating for Afro-Asiatic (see above comments on Ehret) shows its something like a maximum of 11,000 years old. The main supporter for the E3b1/Kebaran scenario, Ehret, is now slicing big chunks of time off his calculated dates. This places AA languages into an essentially late Holocene (if African) or early Asian neolithic scenario. Not a match for the Kebaran/E3b1 expansion.

 Nubian is not Afro-Asiatic. Nubians and upper Egyptians shared a common culture in the Holocene with each other and the western desert ceramic cultures, but they apparently didn’t speak even remotely related languages by the time of state formation in Egypt. This is suggestive of Afro-Asiatic replacing the original pre-Badarian languages, leaving Nubians isolated like an island in a sea of Afro-Asiatic. It also nails the arrival of Afro-Asiatic in upper Egypt to a time after the ceramic Nilo-Saharans got there (which was about 10,500 bp) fixing it to a younger age than this. The only influx after this into the area that we know about is the arrival of the Neolithic from Asia about 7,000 BP in upper Egypt. Nilo Saharan has a very  similar age to Afro-Asiatic (Holocene), and matches pretty well to the spread of the first ceramic using people across the Sahara and down the Nile, including upper Egypt. It’s fragmented distribution is very suggestive of a much wider territory, now occluded by later waves of Afro-Asiatic. The isolation of the Nubian NS language, when using corrected dates, comes into an era when the Neolithic pastoralists arrived from Asia into NE Africa.  

The distribution of Afro-Asiatic in Africa has a very strong relationship to the spread of Neolithic Y chromosomes (particularly Chadic and Berber). Asian ones that enter via the Sinai mainly moved into East Africa and Lake Chad. NE African Y chr M81 shows a match to the old spread of Berber languages (from the Nile Delta in the Neolithic) prior to Arabization .

 Proto Cushitic, and all the African and Asian AA languages,  reconstruct with nouns for sheep and goats, Asian animals that don’t even appear in Africa until 8,000 BP. Archaeology tracks these pastoralists moving from the Nile Delta with their herds into North West Africa, East Africa and  Lake Chad, with large chunks of their male ancestry traceable to Neolithic Asia. Using the archaeology to correct the dates for proto Cushitic means it can only be of a Neolithic age, possibly be in the 5k-6k date range. This does not support a Holocene or older date for PAA in Africa, and is more evidence connecting it to the Asian Neolithic.

 So…what are you are left with in support of an African origin? Not a lot really. That there’s more diversity and structure in Africa, but that’s about it. Omotic as a pre-agricultural AA language doesn’t really stand up to scrutiny, with several publications suggesting they originally spoke Nilo Saharan but adopted Afro-Asiatic in the Neolithic, or of it being an off shoot of Cushitic, or not even Afro-Asiatic. The main supporter for an ancient African origin for Afro-Asiatic tied to E3b is slashing huge amounts of time off his estimated dates (about 45% for proto Semitic).

The real issue for anyone claiming an African origin for PAA is to show some population/cultural movement from Africa into Asia about 11,000 to 10,000 years ago, or even any time close. This is the real ‘brick wall’ the claims for an African urheimat run into. Until someone can explain how it managed to ‘swim upstream’ against the arrival of the Capsian and Neolithic arrivals from the Levant, IMO the claims for an African origin for PAA jus don’t fit the biological or archaeological evidence.

Advertisement

Proto Semitic; dating and locating it.

Something I really should have done for proto Semitic before I did the larger post on Afro Asiatic – but you live and learn.

Essentially, if a language has certain technologies in it, you can use the known dates for the age of the  technology to give a date for the language. With proto Indo European the mega giveaway is the multiple number of words for wheel, which essentially stick an oldest date (so far) of 5,300 BP on the last common dialect of proto Indo European; making the claimed 9k date  for the last common dialect of PIE pretty unlikely to say the least. It’s not impossible that Anatolia was the home of the older ancestral language to this later ‘node’, so I have reconciled the Anatolian and steppes locations by assuming one is Neolithic PIE and the later is Bronze age PIE. There do appear to have been a substantial number of loanwords from Semitic into PIE, and this works best with PS being a neighbour to N-PIE rather than BA-PIE, which places the history of proto Semitic back in the North of the Levant a very long time.

I did this for Indo European, but was a bit remiss in not doing it for proto Semitic, which now means I need to re-do an older entry (doh). Most notable here is the absence of the roots for wheel, axle, cart and chariot, which really puts a tight time limit on proto Semitic. 

Dating

Wheel /chariot/cart/axle – absent/not found; Oldest date for this tech is 5,300 BP in Slovenia, about 4,600 BP in Mesopotamia.

Silver – oldest known 5,000 BP, Aegean/Asia minor.

Antimony –oldest 5,000 BP in Tello, Chaldea, Iraq, reaching Egypt 500 years later.

Camel –native to West Asia, domesticates appear 4,000 BP in Arabia, about 2,000 years ago in East Africa.

Horse- native in West Asia/Iran; the Capsian horse dates back to 5,000 Bp

The dating of silversmithing to 5,000 BP (no older date as yet) and antimony at 5,000 BP sets an upper limit for proto Semitic (and PIE too).  The horse is not a native African animal isn’t seen there until they arrive in Nubia at the end of the Egyptian middle Kingdom, about 3,800 BP, still later in Ethiopia. 

I’m assuming that the wheel hadn’t reached there yet- although I see the oldest wheel down as 5,500 years old  in Mesopotamia I can’t find a single reputable source for it, so the oldest known wheel is from Slovenia at 5,300 years. But the slightly later languages like Assyrian and Babylonian have words for wheel, so the expansion of Semitic languages can just be squeezed in between the appearance of the wheel and the appearance of antimony/silver. About 5,000 BP to 4,600 BP, would be my guess (using the standard of Ur for known wheel dating). 

The appearance of the root word for camel puts a large dent in theories suggesting East Africa as a home for proto Semitic, as the camel is quite a recent appearance there, at about 2,000 years old, which places East Africa well out of the range of the possible, since Semitic languages appear in writing well before this date in Asia. The same is true of the root word for horse; it appears in East Africa after Semitic writing in Asia has begun.

Adding this all up; East Africa is effectively dumped out as an urheimat for PS, by the use of the words  for camel and horse -written Semitic is older than camels and horses existing in East Africa.

Location

Bitumen and Naptha  – available in West Asia (link)  in antiquity, Iran Iraq and Syria mainly.

Pistachio and Almond – West Asia into North Africa

Sorghum – African

Fig– (ficus Carica) – Asia, Mediterranean and West Asia

Vines/grapes – West Asia

Oak tree – Common to the Zagros mountains, not seen in Africa

Mountain – two words, one for mountain, one for mountain with a vineyard

Hill – Five words for hill-pretty exceptional and an indicator of a hilly surrounding

Ice- yes, there is a word for ice in proto Semitic, not really seen much in Africa

Sea/ocean – one word

Boat/ship– two words, but terms for sail and oar are lacking

River- a few words that also have other meanings, and one for riverbank

Stream – two different words for rivulet and narrow stream bed

The animals and flora described are unfortunately not specific to any area in Asia in the Neolithic, although the med coast seems favoured by the flora, and by the camel and horse, neither of which arrive in Africa until after Semitic languages are being written in Mesopotamia. Pistachios, figs and almonds are not tropical trees and won’t grow in East Africa. Although sorghum is from Africa, it arrived in Asia  in roughly the same era as PS appeared, and the root word is also used for beans and wild corn. The presence of petroleum derived products like bitumen and naphtha, which just aren’t found lying around in East Africa, also points towards West Asia, specifically the more Northern part of the possible range. The presence of oak trees and ice certainly doesn’t support any kind of African origin, and oak rees are really only found in cooler mountainous areas of West Asia. The particularly large number of words for hill, and two words for mountain do suggest somwhere particularly hilly. There is a word for ‘wave’ and a word for lake that is interchangeable with sea, suggesting familiarity with a really big lake. The forested areas of the Zagros are mainly oak, pistachio and almond trees, all three are part pf the PS language.

All of this does raise the possibility that the Proto Semitic loan words into proto Indo European (for barley, goat, bull) were from an earlier stage of the language’s evolution, back in the early Neolithic. It also suggests a very old vintage for the word wine (wayn) as it travelled with the very early domesticates into the PIE homeland.

So as a conclusion… proto Semitic can be dated to somewhere before 4,600 BP (no wheels ), as my best guess, and probably more around the 5,000 BP date. And located in West Asia, in the Syria/Iran/Mesopotamia arc; it’s the only place with naphtha (bitumen travels well,  naphtha doesn’t) and oak trees, and enough familiarity with ice to need a word for it, and lots of hills. I would suggest that it was probably quite closely related to the Akkadian and Eblaite languages, and may just be a dialect of that region that expanded out for some unkown reason, as the 5,000 ya date is a few hundred years older than old Akkadian.

Tree of Semitic languages.

As a mionr interest, Eblaite is variously classified as both East and West Semitic, and it’s oldest dating is to 4,250 BP. it’s very similar to Akkadian, which suggests to me that these two may branch straight off from proto Semitic. In fact, a study of Eblaite (book link, page 5) reveals that non-Semitic names/nouns are very rare in it, which suggests a long residence in the Syria are for Semitic languages (nouns and place names from older languages are often retained when a new language appears). So I’m going to suggest a home for proto Semitic as somewhere near Ebla and the Zagros mountains, about 4,800 years ago.

Edit:

Recent work has sugested:

Our statistical tests of alternative Semitic histories support an initial divergence of Akkadian from ancestral Semitic over competing hypotheses (e.g. an African origin of Semitic). We estimate an Early Bronze Age origin for Semitic approximately 5750 years ago in the Levant,

Which I can live with, as on reflection one of the dating nouns I used, ass, could be referring to the wild asses that were native to Asia. So, this removes the major reasoning for the  more recent age I have here. I can live with a date of 5,750 years. Also from the same paper:

Ethiosemitic languages of Africa reflect a single introduction of early Ethiosemitic from southern Arabia approximately 2800 years ago.

Which is also no surprise, as no-one really took an African origin for Semitic seriously. I will comment that the age for PS in Asia (tracing back 9k if it was next to Neolithic Anatolian PIE) really does nothing for an African origin for proto Afro-Asiatic. The maximum age of a language group seems to be less than 11,000 years, and the only bio-cultural movements that we know of in that era are moving into Africa and not out of it.