Turkey: Not very Turkic (a genetic history of the Turkic peoples)

Ironic, isn’t it? The geographic distribution of Turkic languages is amazingly vast-yet-splotchy, extending from the eastern border of Bosnia to the far western end of Siberia, where Russia approaches Alaska: Carte_peuples_turcs (I’d really like to see this map laid atop a topographic map, because that might explain some of the splotchiness–not a lot of people speaking anything in the Taklamakan Desert, for example.) Our oldest known Turkic inscription–thus, our first known use of the Turkic language–comes from the Orkhon Valley, which is located smack dab in the middle of Mongolia. Which, you may have noticed, is not today a Turkic-language speaking place. The Mongolian Language family is, ironically, much less widespread than the Turkic-family:


Given that the Mongols recently conquered almost all of Asia, decimating local populations and leaving behind their genetic legacy (polite speak for “raping all the women,”) they’ve made remarkably little linguistic impact. If we want to get controversial, some linguists propose that the Mongolian family and the Turkic family might be related to each other within a broader “Altaic” language family, which makes plenty of geographic sense, but might not make true linguistic sense. Being me, I always root for nice fancy language family trees, but we’re going to have to call this one “just a theory some guys have and some guys oppose” for now. (The difficulty with reconstructing proto-Turkic or proto-Altaic or the like is that there aren’t a ton of old inscriptions in either family, and not many linguists are trained in them.) Languages get complicated because they can contaminate each other in unexpected ways. To use a familiar example, even though English is a Germanic language, our “do” constructions, eg, “Do you walk?” “I do walk!” and “Do walk with me,” appear to come not from Old or Proto-Germanic, but from Celtic languages. When the Anglo Saxons moved to England and conquered the Celtic peoples living there and made them start speaking Anglo-Saxon, the Celts retained some of their old grammatical structures. But Celtic and Germanic languages are not all that different; they’re both Indo-European, after all. Imagine what craziness you could get by combining peoples who originally spoke languages separated by much vaster gulfs of time.

The English example reminds us of another difficulty in attempting to use linguistics to tell us something about groups and their histories: widely disparate groups can speak the same language. Not only are the English, despite speaking a “Germanic” language, only about 10% German by ancestry (more or less;) but the US has almost 40 million African Americans who all speak English and aren’t genetically English. Even though most people learn to talk by imitating their parents, people have picked up and promulgated many languages that weren’t their ancestors’.

We have a similar situation with Turkey, where the majority of the population clearly speaks a Turkic language, but the genetics shows far more in common with their local Middle Eastern neighbors:

Click for full size
From Haak et al.

Zooming in on the relevant portion:

TurkishDNA2fromHaak ChechenDNAfromHaak

I like Turkey’s DNA because it’s always easy to spot in these charts. Turkey has some real variation in the distribution of different ancestral populations–the Japanese population, by comparison, is far more genetically homogenous.

The really anomalous guys in the Turkish sample are easily explained–they’re just Greeks, (and the anomalous guys in the Greek Sample are Turks.) Turkey ruled over Greece for quite a while, so it’s not surprising that some Greeks live in Turkey and some Turks live in Greece.

Chechens through Kumyks are all groups from the Caucus Mountains area, which is just north of the Turkish-Iranian border, so it’s not too surprising that all of these groups resemble each other. The Greeks, though, are much closer to their neighbors to the north, like the Albanians.

The Chechen and Lezgian languages are from the “Northeast Caucasian” language family (aka Caspian language family). Remarkably, this geographically tiny splotch of languages (and the similarly named but apparently not linguistically similar Northwest Caucasian language family, [aka Pontic language family,]) is considered, like Indo-European, one of the world’s distinct language groupings:

Primary_Human_Language_Families_MapThe Adygei (or Adyghe) speak a Northwest Caucasian language.

The Balkars and Kumyks speak Turkic languages, and the Ossetians speak an Indo-European language, (Indo-Iranian branch.)

Remarkably, even though these Caucasian groups speak languages from four different language families–one of which may have originated in far-off Mongolia–they are genetically quite similar to each other.

from Haak et al.

The Iranians have a small but noticeable chunk of bright green, which shows up in tiny quantities in some of the other populations in this group. The bright green is highly characteristic of India, where it is found in large quantities.

Iran speaks an Indo-European language, of the Indo-Iranian branch. (Given present politics, it is a bit of a wonder that the Aryan Nation and its ilk are actually named after the Muslim nation of Iran, but there you go, that’s history for you.) So I suspect that Iran got its language due to a small group of Indians conquering the place, imposing their language, and marrying into the local population, but this isn’t really supposed to be a post on the history of Indo-European.

What about Turkey’s neighbors to the south? How much do Turks resemble them? Here are some folks in the local vicinity (Syria and Iraq border Turkey to the south, but Iraq doesn’t seem to have made it into this dataset):


The most noticeable thing here are the big chunks of purple, which reach their maximum in the Bedouins. However, I suspect the purple is (in some manner) related to the dark blue which it replaces; if you glance up at the dataset used for the image at the top of the blog, you’ll note that it shows the same basic ancestral DNA groups for the Middle Easterners as Europeans (albeit in different proportions.) The technical differences between these two data sets aren’t worth getting into; suffice to say that I think the Haak dataset is just showing us a finer grained level of detail, which is why I am primarily leaning on it.

At any rate, the purple is distinctive. The Turks (and Iranians) have some purple, but not a lot; the Caucasians very little. The Middle Easterners also have a bit of pink (and a touch of blue) which hail from Africa. These colors, interestingly, appear not to have made it into the Turkish samples at all.

So while the Turks are similar to the Syrians and other neighbors to the south, I hold that they are genetically more similar to their neighbors in Iran and the Caucuses.

DNA from various Asian peoples

But what about the red and yellow bits? Those come from central Asia. Russia has similar levels of red, which is found all over Siberia and northern Eurasia, including the Sami; Yellow is common across far east Asia, including China, Japan, and Mongolia. Most of the countries that Americans mean when they say “Asian” have a mix of red and yellow.

Since the first written Turkic we have comes from the middle of Mongolia, it is sensible that folks in Turkey, today, might have DNA that appears to have come from the region. However, they don’t have a lot of this DNA, suggesting that the overall number of migrants or conquerors, (Turkic or Mongolian or of some other Asian origin,) was relatively low compared to the rest of the population. Today’s Turks, therefore, are probably descended primarily from the ancient Anatolian population that was there before the Turks, Mongols, Indo-Iranians, or other folks showed up.

Geographically, Turkey is located on a plateau and markedly greener than its neighbors to the south. That alone may account for differences between the Turkish people and their southern, more desert-dwelling neighbors.

What about the other Turkic peoples?

There are a lot of them:

The term Turkic represents a broad ethno-linguistic group of peoples including existing societies such as the Turkish people, Azerbaijanis, Chuvashes, Kazakhs, Tatars, Kyrgyz, Turkmens, Uyghurs, Uzbeks, Bashkirs, Qashqai, Gagauz, Altai, Khakas, Tuvans, Yakuts, Crimean Karaites, Krymchaks, Karakalpaks, Karachays, Balkars, Nogais and as well as past civilizations such as Yenisei Kirghiz, Dingling, Tiele, Chuban, Pannonian Avars, Göktürks, Bulgars, Kumans, Kipchaks, Turgeshes, Khazars, Seljuk Turks, Ottoman Turks, Mamluks, Timurids, Khiljis, and possibly Huns, Xiongnu, Wusun, Tauri and the Tuoba.

And we don’t have time to run through all of them. We will mention those who are included in Haak’s dataset, though:

TurkishDNAfromHaakNogai balkar Chuvash Kumyk Kyrgyz Turkmen Altaian yakut

(Chuvash? Are you sure?)

These guys have a lot in common–most of them have, at least broadly, similar varieties of DNA–but not enough to be considered a single ethnic group. Like most groups, they tend to be more closely related to their neighbors than to folks far off, and the Turkic peoples are pretty scattered. The especially odd thing about them, though, is that none of these–at least, none of the folks in Haak’s dataset–look like the Mongols, despite the Turkic languages having probably originated somewhere near Mongolia. (And the Mongolian-like DNA they do have might be more easily explained by Mongolian expansions than by Turkic ones.)

Wikipedia comes to a similar conclusion:

The physical characteristics of populations of speakers of Turkic language stretch across a range as wide as the land they inhabit. The Turkic peoples in Europe look European – with the exception of some Crimean Tatars and Turkics in the Caucasus (Kumyks, Nogays, etc.) who look European+Northeast Asian, while Turkics in the Middle East resemble the peoples of the Middle East, those in Central Asia mostly look mixed but have mostly northeast Asian features. Turkics in northeast Asia resemble populations in that region. In trying to answer such questions as what “race” were the Proto-Turkic speakers, neither anthropometric nor genetic studies have been of much assistance to date. What few DNA analyses have been done arrive at the problem as an answer: affinity to primarily western populations in the west, eastern in the east, and a mixture on a gradient from east to west or vice versa in between.[2] These biological circumstances suggest that racial evolution over the region is earlier than can be considered in the time of the distribution of languages; i.e., the languages may have evolved among populations that were already mixed.

The extremes of the Eurasian continent–Europe, India, SE Asia–have wide zones with a fair amount of genetic homogeneity (even where there are multiple ancestral groups.) In between these zones, however, we get a mixing zone, where different groups come together and new ethnicities are born. All of the Turkic groups here have, to greater or lesser degrees, the tri-color pattern typical of Europe (orange, teal, dark blue) and the di-color pattern typical of SE Asia (red and yellow,) though this is greatly attenuated at the extremes of Turkey and the Yakut. Some groups also have the green characteristic of Indo-Iranians, probably due to bordering those zones.

The Turkic language groups may therefore represent a kind of genetic mixing zone between the large, homogenous zones to their east, west, and south. How long have the steppes (and the mountains to their south) been mixing zones? We don’t know. But the idea that the Turkic peoples were ethnically mixed and heterogenous long before they began speaking Turkic languages at all seems reasonable.

But if Turks aren’t particularly Turkic, why do they speak a Turkic language at all?

Surprisingly, the Turks didn’t even exert military dominance over Turkey until about the 1,000. Prior to this, Anatolia, as we may call the pre-Turkic area–was ruled by the Byzantines, eastern successors to the Roman Empire. The local population was Greek-speaking Christians.

The origins of the Turkic peoples are shrouded in mystery, mostly because of the lack of good written records. There is much speculation, for example, about whether or not the Huns were Turkic, but unless someone can come up with a Hunnic dictionary, we’ll probably never truly know.

The first confirmably Turkic group we know of was the aptly-named Goturks, who lived in parts of China and Mongolia, beginning around the 500s. They apparently controlled a rather large region:


We know of the Goturks because they left behind written records of themselves (beginning in the early 700s,) the Orkhon inscriptions. Interestingly, these Old Turkic inscriptions are written in an alphabet derived from Aramaic (which is, in turn, derived from Phoenician):


What were a bunch of nomadic herders doing making a bunch of monuments inscribed with a derivative form of the Aramaic alphabet up in the middle of Mongolia in the 700s? For that matter, why weren’t they using something derived from Chinese, who lived much nearer?

My best guess is that the alphabet arrived with some eastern variant of Christianity, spread by Christian missionaries through the Persian empire and beyond. (Remember, Iran wasn’t conquered by the Muslims until 651; before that, Christianity had a much larger foothold in the East.) This is not to say that the Goturks were Christians in the way that we typically practice it today, (shamanism focused on the sky god Tengri, whom they shared with the Mongols, appears to have been the dominant religion,) but that they may have had contact with Christian missionaries or religious texts.

At any rate, it looks like the Turkic peoples get on too well with the Chinese, and probably weren’t too keen on the Mongols, (no one was too keen on the Mongols,) which may have inspired them to start migrating. (Or perhaps they were always migrating. They were nomads, after all.)  Either way, by the 800s, a Turkic-speaking people called the Seljuqs had pitched their yurts north of the Caspian sea.

From there they migrated southward, encountering Muslims in Iran, (where they picked up Islam,) and eventually reaching Turkey around the year 1,000. (These migrations probably should not be thought of as single, organized movements of people, but of many migrations, mostly of tribes simply wandering in search of pastures for their animals, conquering neighbors, fleeing conquerors, and generally being a complicated, disorganized bunch of humans.)

At any rate, the Seljuk Empire, founded in 1037, absorbed the crumbling Persian Empire, and invaded the Byzantine Empire in 1068. By 1092, it stretched from the Bosphorus, down through Palestine, across Iran, around Oman, through several -stans, and up to the far western end of China:


This all helped inspire the Crusades, launched in 1096 to help the Byzantines repel the Seljuks, but that is a story for another day. The Mongols showed up around 1243, but by the 1400s, the Turks were in charge again. In 1453, the Ottomans took Constantinople–now Istanbul (which is really just a slight corruption of the Greek for “to the city,” “εἰς τὴν πόλιν”)–ending the last vestige of the once vast Roman Empire.

An observer described the looting:

Nothing will ever equal the horror of this harrowing and terrible spectacle. People frightened by the shouting ran out of their houses and were cut down by the sword before they knew what was happening. And some were massacred in their houses where they tried to hide, and some in churches where they sought refuge. …

Old men of venerable appearance were dragged by their white hair and piteously beaten. Priests were led into captivity in batches, as well as reverend virgins, hermits and recluses who were dedicated to God alone and lived only for Him to whom they sacrificed themselves, who were dragged from their cells and others from the churches in which they had sought refuge, in spite of their weeping and sobs and their emaciated cheeks, to be made objects of scorn before being struck down. Tender children were brutally snatched from their mothers’ breasts and girls were pitilessly given up to strange and horrible unions, and a thousand other terrible things happened. …

Temples were desecrated, ransacked and pillaged … sacred objects were scornfully flung aside, the holy icons and the holy vessels were desecrated. Ornaments were burned, broken in pieces or simply thrown into the streets. Saints’ shrines were brutally violated in order to get out the remains which were then thrown to the wind.

The Wikipedia estimates that 4,000 were killed and 30,000 deported or sold into slavery. 4,000 sounds like a low estimate to me, given the nature of warfare, not to mention reports like Barbaro’s:

Barbaro described blood flowing in the city “like rainwater in the gutters after a sudden storm”, and bodies of the Turks and Christians floating in the sea “like melons along a canal”.[50]

As I have mentioned before, I strongly recommend not getting conquered.

The Ottoman Empire continued to expand, reaching its greatest extent in 1683:


The few small Turkic-speaking communities in Europe today probably owe their genesis to the Ottoman empire, though some might have arrived on their own, via more northerly routes.

And as for the guys in Siberia? They probably just decided to try walking north instead of south.

Genetic Map of Europe

source: Big Think: Genetic map of Europe
source: Big Think: Genetic map of Europe

Isn’t it beautiful?

Some thoughts:

Sardinia and the Basques: low Indo-European genetic component.

I’ve been saying for a while the Ashkenazim should be considered a European group. Genetically, they’re half-Italian, and they’ve been living in Europe for about 2,000 years.

Tuscany has, I think, a very high Neanderthal %? I’m not sure why different parts of Italy would be so far apart, but I can see why Italy might have problems with staying unified or forming national-scale institutions.

Finns and Hungarians are not divergent enough from other Europeans to consider their unique language situation indicative of major genetic differences.

World’s most famous Chuvash:

Vladimir Lenin

Chuvash are a “Turkic” people of Russia. (Turks are also a “Turkic” people, though ironically, they actually don’t have a lot of Turkic DNA.)

Hrm, I should do a post on the Turkic Peoples.