The story of Turkic
01
A Family Nobody Had to Discover
Most language families in this atlas had to be proved. Turkic mostly did not. Its languages are close enough that their speakers, and the scholars who wrote about them, long treated them as varieties of one tongue. In the 1070s Mahmud al-Kashgari, writing in Baghdad for Arabic-speaking readers, described the speech of the Oghuz, the Kipchaks and other Turkic peoples as dialects of a single 'language of the Turks', and recorded where they differed in sounds and vocabulary.
European comparative work began in the 18th century. In 1730 Philip Johan von Strahlenberg, a Swedish officer who had spent years as a prisoner of war in Siberia, published word tables that set the 'Tatar' languages beside Finnic, Mongolic and others. His tables lie behind the later, now abandoned, 'Ural-Altaic' hypothesis.
The 19th century supplied the evidence. Russia's expansion into Siberia and Central Asia brought scholars into contact with dozens of Turkic peoples, and Wilhelm Radloff (Vasily Radlov) collected their oral literature and compiled a four-volume comparative dictionary of the 'Turkic dialects' (1888–1911). Vilhelm Thomsen's decipherment of the Orkhon inscriptions in 1893 then gave the family a securely dated 8th-century stage against which the modern languages could be measured.
The one real question was Chuvash. Its r and l, where the other languages have z and š, and its heavily altered vowels made some scholars doubt it was Turkic at all. Once those differences were shown to be regular, Chuvash was accepted as the descendant of a branch that had left the rest very early. In the 20th century, classifications such as Aleksandr Samoilovich's (1922) sorted the remaining languages by a handful of diagnostic sound changes, and the branches in this atlas descend from that tradition.
02
From the Eastern Steppe to the Mediterranean
Almost everyone places the Turkic homeland in the east, somewhere between the Altai mountains, southern Siberia and Mongolia. The reconstructed vocabulary fits herders who kept horses, cattle and sheep, and it also includes words for forest trees such as the birch, which is one reason many historians look to the zone where the steppe meets the Siberian forest. Very early loanwords point the same way: exchanges with Mongolic, and borrowings from Chinese and from Iranian languages of Central Asia, such as Old Turkic känt 'town', from Sogdian, which survives in the name Tashkent.
The family is young. A Bayesian analysis of basic vocabulary by Alexander Savelyev and Martine Robbeets (2020) put Proto-Turkic at about 2,100 years ago, with a wide margin of uncertainty, which agrees with older estimates that place the split between Oghur and Common Turkic around the turn of the era. Whether the Xiongnu confederation of the 3rd century BCE, or the Huns who reached Europe in the 4th century CE, spoke Turkic is unknown; the few recorded words of both have been read as Turkic, Yeniseian and other things.
The spread itself is well documented. From the 6th century onward Turkic-speaking elites, first the Türk khaganate and later the Seljuks, the Kipchaks and the armies of the Mongol empire, carried the language west and south. Genetics shows how this worked. A study by Bayazit Yunusbayev and colleagues (2015) found that most Turkic-speaking populations carry some ancestry traceable to southern Siberia and Mongolia, but usually only a minority of it: Turkish speakers in Anatolia, for example, are genetically close to their non-Turkic neighbours. For the most part the language spread by conquest and language shift, not by the replacement of one population by another.
03
Harmony, Suffixes and Verb-Final Sentences
A reader who learns one Turkic language finds much of the grammar of the others already familiar. The most noticeable feature is vowel harmony. Suffixes change their vowels to match the vowels of the word they attach to, front with front and back with back, and in most languages rounded with rounded. Turkish ev 'house' makes ev-ler-imiz-den 'from our houses', while at 'horse' makes at-lar-ımız-dan 'from our horses'.
That example also shows the second feature: words are built by adding suffixes one after another, each with a single job (plural, then 'our', then 'from'). There are no prefixes, no grammatical gender and usually no definite article. Sentences normally end with the verb, and modifiers come before what they modify, including whole clauses: Turkish okuduğum kitap 'the book that I read' is literally 'my-reading book'. Where English has prepositions, Turkic has postpositions and case endings.
Several features are less common among the world's languages. Most Turkic languages have no verb 'to have'; possession is expressed as existence, as in Turkish kitabım var 'I have a book', literally 'my book exists'. Many mark in the verb whether the speaker saw an event: Turkish geldi 'came' implies first-hand knowledge, while gelmiş means 'apparently came' or 'came, I hear'.
The family has also been written in an unusual number of scripts: the Old Turkic runes, the Sogdian-derived Uyghur alphabet, Arabic, Cyrillic and Latin, and, among Christian and Jewish communities, Greek, Armenian and Hebrew letters. Ottoman-era Turkish-speaking Orthodox Christians wrote Turkish in Greek script, a tradition known as Karamanlidika.
04
What Is Still Argued Over
The largest dispute concerns Turkic's relatives. The Altaic hypothesis joins Turkic with Mongolic and Tungusic, and its wider 'Transeurasian' form adds Koreanic and Japonic. Its main modern defence is the etymological dictionary of Sergei Starostin, Anna Dybo and Oleg Mudrak (2003); Martine Robbeets and colleagues argued in 2021 that the family spread with early millet farmers in northeast China. Many specialists, including Alexander Vovin, Stefan Georg and Juha Janhunen, reject a genetic relationship. They argue that the shared words, especially between Turkic and Mongolic, are loans from long periods of contact, and that the structural similarities are typical of the region rather than inherited.
The Oghur question is linked to this. If Oghur r and l are older than Common Turkic z and š, some of the best Turkic–Mongolic matches may be early Oghur loans into Mongolic rather than evidence of common descent.
Inside the family, the Oghur/Common Turkic split is accepted by nearly everyone, but the lower branches are less settled. Khalaj, spoken by a few tens of thousands of people in central Iran, keeps an initial h- lost elsewhere (hadaq 'foot', Turkish ayak), and Gerhard Doerfer argued that it forms a branch of its own. The Siberian languages (Tuvan, Khakas, Altai, Sakha) may not form a single group. Kyrgyz and Salar sit awkwardly between branches, and classifications based on sound changes, on shared vocabulary and on history do not always agree.
Dating also remains open. Estimates for Proto-Turkic range from the later first millennium BCE to the turn of the era, and they depend partly on whether steppe peoples such as the Xiongnu are counted as Turkic speakers.