Old Turkic is the language of the runiform inscriptions of the eighth-century steppe and of the manuscripts that follow them. It is the oldest Turkic that can be read, and every later Turkic language is described by what it kept from this stage and what it gave up. Vilhelm Thomsen, who deciphered the script, concluded that the language of the inscriptions is a pure Turkic and belongs to a stage older than any Turkic language then known.[1]

The language is known from its own monuments, not from foreign report. The Kül Tegin inscription gives some fifty lines of continuous Turkic prose, dated by the text itself to 731 and 732, and the Tonyukuk inscription gives a second, older voice beside it.[2] That is a rare thing. Most languages of the early medieval world are recovered from what their neighbours wrote down about them. This one is recovered from what its speakers carved for themselves, at length, in their own script, about their own affairs.

The texts are also unusually candid. A khan addresses his people in the second person and tells them they ruined themselves once and could do it again.[2] A counsellor writes in the first person and quotes the advice he gave and the orders he disobeyed.[3] For the historian that is a political document. For the linguist it is something scarcer still: connected prose with argument in it, from a period that elsewhere leaves mostly names and titles.[4]

The name and what it covers

Old Turkic is a term of scholarship, not of the sources. The inscriptions name a people, türk bodun, and never a language.[2]

What the term covers is a matter of convention. The conventions differ. At its narrowest it is the language of the Orkhon monuments alone, for which the narrower label Orkhon Turkic is also used.[5] Kök Turkic, Runic Turkic and Old Oghuz have also been used for it.[6] At its widest it takes in the Yenisei epitaphs, the Talas and Altai inscriptions, the Irk Bitig, and the Buddhist, Manichaean and Christian manuscripts of Turfan and Dunhuang. In that wider sense it runs from the eighth century to the Mongol invasion of the thirteenth. It takes in the Karakhanid material and the dictionary of Mahmud al-Kashgari of the eleventh on the way.[7]

Usage varies on both edges. Some scholars keep the name for the language of the Orkhon stones alone and set even the runiform inscriptions of the Uyghur steppe empire apart. Others place the Karakhanid texts in Middle Turkic.[8] The view that has become standard among specialists takes the whole corpus as one language in several dialects, to be treated as alike unless a difference is shown.[8] The language was then not confined to the east. It was presumably spoken, with minor variation, in West Turkestan, where Karakhanid was spoken, and as far south as Bactria.[9] At the late end the Uyghur manuscripts run on past the eleventh century to the Mongol conquests of the thirteenth. Those written under Mongol rule can be difficult to tell from the earlier ones.[10]

The span is worth holding in view, because it is long. Between the Orkhon stones and the Mongol conquest lie some five centuries, a change of religion, a change of script, a move from the open steppe to the oasis towns, and a shift from carving to copying. A language does not stand still across that, and the label covers the whole of it.

The corpus

Four bodies of text preserve the language, and they are unequal in every way: in length, in date, in subject, and in what they can be asked.

The Orkhon inscriptions (Kül Tegin, Bilge Khan, Tonyukuk, and the Ongi and Küli Čor stones beside them) are the largest continuous Turkic prose of the khaganates. They are official and commemorative, spoken in the first person by a khan or a counsellor. They were meant to be read aloud to an assembly standing in front of the stone.[2]

The Yenisei epitaphs are shorter, far more numerous and much harder to date, and they speak in the voice of the dead man: he takes leave of his wife, his sons, his herd, his lord.[11] They give the ordinary vocabulary of a life that the great inscriptions have no room for: the words for a household, a marriage, a debt, a horse.

The Irk Bitig, the book of omens, is the one long runiform text on paper rather than stone.[12] That matters for more than its contents. A text on paper is a text that can be copied and carried. Its existence shows that the script was not reserved for monuments.

The manuscripts of the Uyghur period and after, which are Buddhist and Manichaean translations out of Sogdian, Chinese and Tocharian written in several scripts, are by word count the largest body of Old Turkic by a wide margin. They belong to a settled, monastic world rather than to the steppe of the khaganates. Their language has moved accordingly, with more abstraction, more borrowing, longer periods, and a vocabulary built to carry doctrine that the stones never had to carry.

Grammars of the language count three groups, not four. The first is the runiform inscriptions of the seventh to the tenth century. An overview of 2004 gave some two hundred of them. They were found mostly in Mongolia and the upper Yenisei basin, with a few in the Altai, along the Irtysh, in Kazakhstan and Kyrgyzstan and in the north-east of Chinese Turkestan. Most are epitaphs, some are graffiti on rocks beside main roads, and a few stand on objects such as coins, mirrors and bowls.[13] The second is the Uyghur manuscripts of eastern Xinjiang and Gansu from the ninth century on, in Uyghur, Manichaean, runiform, Brahmi, Sogdian, Syriac and Tibetan script. About three quarters of them are Buddhist; the Manichaean texts make up less than a tenth, though most of these are early, and the Christian texts are the fewest.[14] The third is the Karakhanid works of the eleventh century, the Kutadgu Bilig and the dictionary. The only direct manuscript evidence of the period is a set of land-sale documents from Yarkand in Uyghur script, since the surviving manuscripts of the two works are later copies.[15]

Later counts are made region by region and differ in date and in what they include. In Mongolia, 182 inscriptions at 101 sites had been recorded by 2022, counting even the smallest rock texts.[16] For the upper Yenisei basin a figure of about 250 is quoted, which includes inscribed objects.[17] In the Altai, a count of 2021 gives more than a hundred inscriptions, cut on rock faces.[18] Road building has damaged many of them, and much of the material cannot be read in full as a result.[19] A catalogue of 2012 and an online edition with new squeezes and comparative readings record the texts.[19] In Kyrgyzstan, more than thirty inscriptions were known about 2000, nearly all of them cut in stone.[20]

The four are therefore not four samples of one thing. They are four registers from four situations. A statement about “Old Turkic” that does not say which of them it rests on is worth less than it looks.

Turkic before the inscriptions

Turkic words are on record for much longer before the first inscription, but always in someone else’s script and someone else’s sound system.[4] Chinese histories transcribe Göktürk names and titles with characters chosen for their sound. Sogdian, Greek and later Arabic and Persian writers do the same in their own alphabets. The Bugut inscription, raised in the First Khaganate, is in Sogdian, not Turkic. It shows a Göktürk court using a foreign chancery language for its own monument.[21]

These transcriptions preserve the shape of a word only poorly. A Chinese character had to be picked from a syllabary that has no ö, no ü, and no final consonant clusters of the Turkic kind. In addition, the pronunciation of that character in the seventh century must itself be reconstructed before the Turkic behind it can be guessed at. This is why the equations between a Chinese form and a Turkic name are so often argued and so rarely settled. The name Bumın behind 土門 is a reconstruction, not a transcription.

Some Turkic is on record earlier still. From about the beginning of the Christian era neighbouring peoples wrote down words that are Turkic by their shape. One is the place name suvlag, ‘watering place’, in early Chinese sources, whose suffix is Turkic.[22] Phrases, as against single words, appear first in documents in Bactrian from Afghanistan, the earliest dated to the first half of the seventh century. One of 640 has tapaglıg, ‘revered’. The language of such phrases is the same as that of the later Old Turkic sources. They belong to the time of western Göktürk rule over Bactria, after the Hephthalites had fallen.[9] The biography of the pilgrim Xuanzang, written about the early seventh century, has him ask the new shad of the western Göktürks for envoys and for 鄔落, wuluo.[23] This is taken to be the oldest recorded form of the Turkic word ulaq, “relay horse”, about a century before the first runiform inscriptions that can be dated.[24] The same biography calls the officer who led Xuanzang to the yabgu’s camp 答摩支. This may be the title tamaci, a kind of escort officer, rather than a name. The title is known later among the Mongols and the Khitan.[25] Of the first khaganate’s own court the only text so far known is the Sogdian stone of Bugut. By one assessment, no Turkic etymology is available for any of the titles it names.[22]

The same Chinese chapter that describes the sixth-century Göktürks says in one clause that their written characters resemble those of the Hu, the Sogdians.[26] That clause may report the runiform script, a Sogdian chancery hand of the kind the Bugut stone uses, or something else again. Under the script, Turkic Atlas takes up the question.

The inscriptions end the guessing. From 716 onward the language is written in a script made for it, with letters that distinguish the sounds it distinguishes. A second dating of the inscriptions runs from about 680 to 740.[27]

Sounds

The vowel system has eight vowels, four back and four front. They are paired: a against e, ı against i, o against ö, u against ü. Counted after the opposition of length had been given up, there are nine, because long ä did not merge with its short partner but survived as a vowel of its own.[28] Whether the language also had long vowels as a separate class is argued. Some take a set of spellings in the inscriptions and in the later material as evidence for primary long vowels inherited from an earlier stage; others treat them as orthographic and holds that length was not distinctive.

Harmony is the rule that makes the rest of the grammar possible. Within a word, all vowels are front or all are back. A suffix has no fixed vowel of its own. It has a slot, and the stem fills the slot. The accusative appears as -ıg on a back stem and as -ig or -üg on a front one. The inscriptions show the two side by side: bodunug, the people, against isig küçüg, toil and strength.[2] Rounding harmony works the same way in the second layer: a rounded stem vowel pulls a rounded vowel into the suffix.

The effect is that a Turkic word of five syllables is heard as one unit rather than as a stem with suffixes attached. It is also the first thing by which a word is recognised as Turkic at all, before anything is known about its meaning. A word whose vowels do not agree is a word that came from somewhere else, or one in which something has gone wrong.

The consonants have two series where a European ear hears one. The k and g of kagan and yagız are made further back in the mouth than the k and g of kök and beg. The runiform script writes them with different letters, and it does the same for b, d, l, n, r, s, t and y.[2] A script with two whole series of consonants, sorted by the vowels they stand next to, fits this language exactly and fits no other. Under the script, Turkic Atlas follows that where it leads.

The word

Old Turkic builds words by adding suffixes to an unchanging stem, one after another, each with one job. Nothing is prefixed, nothing is fused, and the word can be read from left to right as a sequence of decisions: this root, in this derivation, in this number, belonging to this person, in this case.

The case system is usually counted at six, but the count depends on how two pairs are handled. The genitive is -nIŋ. The accusative is -(X)g on a bare stem and -In after a possessive suffix. The dative and the directive share -kA. The locative and the ablative share -DA. The instrumental is -In.[2] Whether the dative-directive and the locative-ablative are one case each or two is a question about analysis rather than about the texts. The texts use one ending for both jobs.

Two of these are worth a second look. The locative-ablative doing duty for both place and source is a genuine economy. Later Turkic splits it. The instrumental in -In goes the other way. It is alive in the inscriptions and gone from the modern languages, which use a postposition instead. A case that a language loses is a better dating tool than one it keeps.

The phrase türk bodunıŋ ilin törüsin, the realm and the law of the Göktürk people, shows three of the endings working together in four words: genitive on the possessor, possessive on each of the two possessed nouns, accusative after it.[2]

Two nouns can also stand side by side with no ending between them. A name and its title do so, as in Bumın kagan, Khan Bumın. A place name and the feature it names do the same, as in Kögmen yış, the forested mountain Kögmen.[29] In both, the proper name is subordinate to the common noun.[29] Whether kaŋım kagan, “my father the khan”, belongs to the same type is argued. It may be an apposition, in which each noun keeps its own weight.[29]

Numerals count towards the next ten. Bir yegirmi, “one towards twenty”, is eleven, yeti otuz is 27 and bir kırk is 31. The inscriptions can also put such sums with artukı, “in addition”: otuz artukı bir is 31.[30] The same way of counting lived on among the Ganzhou Uyghurs.[31]

Derivation is where the language is richest, and where the inscriptions do their political work.

The suffix -çI makes the man who does the thing: yogçı sıgıtçı, the mourner and the wailer, the two words the Kül Tegin inscription uses for the peoples who came to the founders’ funeral.[2] The suffix -lXg makes the one who has it and -sXz the one who lacks it: başlıg, having a head, against biligsiz, without understanding.[2]

The pair that carries the argument of the inscription is -sIrA-, which makes a verb out of a lack. Elsire- is to lose the realm. Kagansıra- is to be without a khan.[2] The two words are not descriptions of a political situation. They are the political situation, compressed into a verb stem, and the text can then conjugate it.

The same economy runs the other way. Bilig is practical political wisdom, the quality a ruler is supposed to have. It is the root of the name Bilge. Its opposite in the text is not ignorance but añıg bilig, evil knowledge, which is what China is said to conceive once a people has settled close.[2] The charge is not that the neighbour is stupid. It is that the neighbour has the same faculty and uses it the other way.

The verb

Causative, passive, reflexive and reciprocal are suffixes, and they stack. Yükün- is to bow. Yükün-dür- is to make bow. The inscription’s başlıgıg yükündürmis, tizligig sökürmis means that they made those who had heads bow, and those who had knees kneel. It contains two causatives in one breath. It is the fixed formula for breaking an independent people.[2]

One verb is worth following across two monuments. Kılın- is the reflexive-passive of kıl-, to make, and so means to be made, to come into being. The Kül Tegin inscription opens the world with it: when the blue sky above and the brown earth below were made, between the two the sons of men were made, kişi oglı kılınmıs.[32] The Tonyukuk inscription opens a life with the same verb in the first person: ben özüm tabgaç iliŋe kılıntım, I myself was made in the realm of China.[33] The word that makes mankind makes Tonyukuk. He uses it of his own birth in a foreign country.

Tense is less elaborate than in the modern languages. The marking of how the speaker knows is stronger.

The past in -DI tells what the speaker witnessed. The narrative past in -mIš tells what is known at a remove: reported, inherited, or reconstructed from its results. The distinction is not stylistic. The Kül Tegin inscription uses it structurally. The account of the ancestors runs in -mIš, and when the speaker turns to what he did himself, the verb turns with him.[2] A reader who follows the endings can see where hearsay stops and testimony begins, which is more than most medieval texts allow.

Beside these stands the particle erinç, which marks an inference the speaker is confident of but did not see: bilge kagan ermis, alp kagan ermis… anı üçün ilig ança tutmıs erinç, they were wise khans, they were brave khans, and that, surely, is why they held the realm as they did.[2] The inscription is reasoning about its own past and saying so.

The future participle in -sIk gives one of the flattest lines in the corpus. Türk bodun, ölsikiŋ is not a threat and not a curse. It is a statement of what is coming.[2] The same form makes an idiom in Tonyukuk that no translation improves: tün udısıkım kelmedi, küntüz olursıkım kelmedi, which means my sleeping at night did not come, my sitting by day did not come.[3] The Kül Tegin inscription uses the same two verbs in the negative as a set phrase.[34]

The sentence

What a European language does with subordinate clauses, Old Turkic does with converbs and participles. The converbs in -(X)p and -(y)X hang one action on the next: sü sülepen, having led out an army; teŋri töpüsinte tutup yügerü kötürmis, taking them by the crown of the head, he raised them up.[2]

Because the machinery is suffixal and the verb comes last, a sentence can accumulate circumstance for several lines and still resolve on its final word. The Kül Tegin inscription opens with exactly that. It gives the making of the sky, the making of the earth, the making of mankind and the seating of the founders over them, all in one period that does not break.[2] The claim being made is that the Göktürk state is the third thing that exists, after the sky and the earth. The grammar expresses this claim as much as the words do.

Speech is quoted, not reported. The words stand as they were said and the particle tep or teyin, saying, follows them.[2] This is why the inscriptions can put a sentence in the mouth of a people and in the mouth of Heaven without any change of construction. The people’s sentence is illig bodun ertim, ilim amtı kanı: I was a people that had a realm, where is my realm now.[2]

Tonyukuk argues, and to argue he needs a conditional. The suffix -sAr supplies it. The inscription uses it for a proverb that is one of the few pieces of gnomic wisdom the corpus preserves: yuyka erkli topulgalı uçuz ermis, yinçge erkli üzgüli uçuz. yuyka kalın bolsar topulguluk alp ermis, yinçge yogun bolsar üzgülük alp ermis. In translation, what is thin is easy to pierce and what is slender is easy to snap; but if the thin grows thick it is hard to pierce, and if the slender grows stout it is hard to snap.[3]

Three different verbal nouns do the work in those two sentences: the supine in -GAlI for the purpose, the verbal noun in -GUlXk for the thing to be done, and the conditional in -sAr for the case assumed. The proverb is about a small people that must grow before it can survive, and it is made entirely out of grammar.

A second passage of the Tonyukuk text, on lean and fat bulls, has the form of a proverb too, though how it should be explained is disputed.[34] For the first proverb, an earlier explanation gives the sense “strength is born of unity”.[34]

Paired words and repeated formulas

The prose of the inscriptions has a shape. Two devices build it.

The first is the pair: two near-synonyms joined into one expression that means more than either. Yogçı sıgıtçı is mourner and wailer. Arkış tirkiş is caravan and convoy. Açsık tosık is hunger and fullness. İsig küçüg is toil and strength.[2] The second element is rarely dispensable and the pair is usually fixed. İsig küçüg is the standard expression for the labour and the military service a subject owes. It survives into modern Turkish as iş güç. Other fixed pairs are il törü, realm and law, and yir sub, earth and water.[35] Pairs of opposites work the same way: içre taşra, inside and outside, ırak yağuk, far and near, and ini eçi, younger and elder brother.[36]

The second device is the parallel clause used as a refrain. Başlıgıg yükündürmis, tizligig sökürmis returns through both royal inscriptions. So does the address türk bodun with a verb in the second person after it.[2] The effect on a listener standing at the stone is the effect of a speech rather than of a document. That is what the texts were for. Parallel clauses can also match in length. One count finds 13 and 8 syllables in the two halves of ilgerü Šantuŋ yazıka tegi süledim, taluyka kiçig tegmedim, and 14 and 8 in birigerü Tokuz Ersinke tegi süledim, Tüpütke kiçig tegmedim.[37]

At its sharpest the style puts two things beside each other and lets the arrangement make the argument. The Kül Tegin inscription says that Chinese words are sweet and Chinese silk is soft. With sweet words and soft silk, it says, they draw distant peoples close. Once a people has settled close, the evil knowledge is conceived there.[2] Nothing in that is a complaint about treachery in battle. It is an analysis of how generosity works as a weapon. It is set down a few lines after the same speaker admits that his people are taking the gifts.

The vocabulary

The core is Turkic and needs no explanation from outside: kinship, herd animals, the parts of the body, the numbers, the words for sky, earth, water, road and year.

One count puts the vocabulary of the Orkhon inscriptions at about nine hundred individual words.[38] A study of the Altai texts finds 26 roots that occur there only in derived form, six of them nouns and twenty verbs.[39] Some of these roots are attested in other inscriptions or in manuscripts, and others in no text at all.[40] One derived verb, yıgla-, “to weep”, occurs in the whole corpus of inscriptions and manuscripts only in an Altai text from Kızıl-Kabak.[41]

The political vocabulary is Turkic too, and it is the part that matters most for the history of the khaganates. İl is the realm, the political order that exists while a khan rules and stops existing when he does not.[2] Törü is law in the widest sense, the customary order, the way things are properly done.[2] Kut is the fortune granted from heaven that makes rule legitimate.[2] Bodun is the people as a political body, and kagan, šad, yabgu and beg are the ranks of rule.[2] To lose the il is defeat. To lose the törü is to stop being a people at all.[2]

Loanwords are few in the inscriptions and many in the manuscripts. The steppe borrowed from Chinese for goods and offices, from Sogdian for trade, writing and religion, and later, through Sogdian and Tocharian, from Sanskrit for the whole vocabulary of Buddhism.[42] One example from Sogdian is kend, “town”, which comes from the Sogdian kant.[43] On one account, kul means a dependant or servant in Old Turkic and Uyghur, and kün a female slave. Karabaş has the same sense, and Uyghur texts add küngüz for a slave woman.[44] That the earliest texts borrow so lightly is a fact about their subject rather than about the language. An inscription about war and rule has less occasion for foreign words than a translated sutra. The translators of the Manichaean texts also coined Turkic words member for member on Sogdian models. Tınlıg, “living being”, from tın, “breath”, is probably built like Sogdian wātδār, “holder of breath”. It occurs in Manichaean and Buddhist texts but not in the inscriptions.[42]

One statement about the lexicon has to be set against another. The Zhou Shu, describing the Göktürks of the sixth century, says that they do not know the reckoning of years and mark time only by the greening of the grass.[26] The inscriptions of the eighth date themselves by the twelve-year animal cycle, in the Year of the Sheep and the Year of the Monkey.[2] A hundred and fifty years lie between the two, which is room enough for the cycle to have been adopted in the interval. The contradiction is not necessarily an error on either side.

Dialects and stages

The three groups form a coherent set of dialects, differing most in vocabulary, and the differences inside each group are probably no greater than those between them.[45] The runiform stones of Mongolia already differ among themselves. The Tonyukuk stone has ben, ‘I’, as an independent pronoun and men inside the verb phrase. The Kül Tegin and Bilge Khan stones have men everywhere. The Šine-Usu stone of the Uyghur khaganate has ben alone again, and the Tariat stone has it even after a verb.[46] The Kül Tegin and Bilge Khan stones, and the Šine-Usu stone with them, write the vowel e out as i more often than the Tonyukuk, Ongi and Küli Čor stones and the Uyghur stones of Tariat and Tes, which nearly always leave it unwritten.[46] The future ends in -dAçI on the Orkhon stones and in -gAy in the rest of the corpus, though -gAy occurs in some Yenisei inscriptions.[47]

Within the Uyghur manuscripts, scholars distinguish a pre-classical, a classical and a late stage. Most of the Manichaean texts and early Buddhist ones such as the Sängim manuscript of the Maitrisimit are pre-classical, and the Tantric texts are late.[48] Whether a difference between two texts is one of dialect or of stage is argued. The old palatal nasal ñ is a standard test. The runiform inscriptions keep it. The Argu dialect described by Kashgari merges it with n. Most Uyghur texts show y, while some show both. This suggests that copyists with a pronunciation of their own changed what stood in their model.[49] One account counts at least two dialects of Old Turkic, that of the Orkhon inscriptions and Uyghur. It counts at least three within Uyghur, according to what became of the old palatal nasal: it stays ñ in the first, becomes n in the second and y in the third. Most Uyghur manuscripts are in the y dialect.[50] The manuscripts in Brahmi script were once held to be a dialect of their own. They appear rather to stand outside the written norm and to show features of the spoken language.[51]

One view holds that the written language was wholly distinct from the speech of its scribes. The variants that stand side by side in the Kutadgu Bilig, chosen for the sake of the verse, speak against it. They are real forms that existed together in speech.[52]

The line between d and y

One consonant draws the line that marks off the corpus from what lies beside and after it. Old Turkic, in the inscriptions and in the Uyghur manuscripts alike, keeps d where Oghuz, Kipchak and later texts have y: adak, foot, against the later ayak.[53][27] Uyghur texts that show the change, such as kaygu for kadgu, sorrow, are not counted in the corpus, since the change is a dialect trait no less than a sign of lateness. Mahmud al-Kashgari already records that the Oghuz and the Kipchak say koy- where others say kod-.[53] An Uyghur inscription of 1334 has the change together with Persian and Mongolian loanwords.[54] Where a text stands on that line is one of the ways it is dated and placed.

The division does more work than a single sound seems able to bear. It is one isogloss. It is used as a proxy for a whole chronology because the corpus offers few others that are as easy to see.

Whether the Yenisei epitaphs are the same language as the Orkhon monuments, a neighbouring dialect, or the ancestor of the Kirghiz material of later centuries, is argued and not settled. The texts are short, formulaic and mostly undated, which is exactly the material on which such a question cannot be closed.

Place among the Turkic languages

Old Turkic belongs to the Common Turkic side of the family, against the Oghuric side whose one survivor is Chuvash. The split between the two turns on a pair of sound correspondences, r against z and l against š. It is older than any written Turkic. It is reconstructed, not recorded. Within Common Turkic the language of the inscriptions stands close to the ancestor of the Siberian group and not far from what the Karluk languages later became.

It is not, however, the ancestor of all modern Turkic in a simple line. It is one attested stage of one variety. The modern branches descend from varieties that were spoken beside it and were never written down. The temptation to treat the Orkhon language as the common ancestor is strong precisely because it is the only eighth-century Turkic anyone can read. That is an argument about the evidence rather than about the family. The survey of the Turkic languages sets out the branches and their order.

Particular forms show how the language stands to its relatives. Modern Turkish keeps ben, ‘I’, where the Kül Tegin and Bilge Khan stones have only men, the assimilated later form. The verb buyur-, ‘to command’, which Old Turkic replaced with yarlıga-, lives on in practically all Turkic languages outside Siberia. It is the source of the Old Turkic title buyruk.[55] A further argument, drawn from the hortative in -(A)lIm and the future in -dAçI, holds that the Orkhon language stands especially close to Oghuz, the south-western branch.[45]

Limits of the corpus

The inscriptions are the speech of a court in a formal register, cut in stone for an audience standing in front of it.[2] From the khaganate period there is no private letter in Turkic, no contract, no account book, no verse that is certainly verse, and no woman speaks in any of the great inscriptions.[2]

What the language sounded like in a tent, in a market or between two herdsmen is not recorded and cannot be recovered from what is. The Yenisei epitaphs come closest. They are formulaic in their own way. Everything said about the ordinary register of eighth-century Turkic is therefore an argument from the shape of the language rather than from a text that uses it.

The corpus is the runiform inscriptions, a book of omens and a library of translations. It is enough to describe the grammar with confidence and not enough to hear the language spoken.

References

  1. ^ Thomsen 1993, p. 15
  2. ^ a b c d e f g h i j k l m n o p q r s t u v w x y z aa ab ac ad ae Kül Tegin inscription
  3. ^ a b c Tonyukuk inscription
  4. ^ a b Erdal 2004, p. 2,4
  5. ^ Tekin 1968, p. 7–8
  6. ^ Tekin 2003, p. 17–20
  7. ^ Erdal 2004, p. 5
  8. ^ a b Erdal 2004, p. 21
  9. ^ a b Erdal 2004, p. 4
  10. ^ Erdal 2004, p. 5, 7
  11. ^ Tekin 2003, p. 230–232
  12. ^ Erdal 2004, p. 7,40
  13. ^ Erdal 2004, p. 6–7
  14. ^ Erdal 2004, p. 7
  15. ^ Erdal 2004, p. 8
  16. ^ Badam 2024, p. 25
  17. ^ Erhan 2011, p. 5
  18. ^ Karaman Dezember 2021, p. 34
  19. ^ a b Karaman Dezember 2021, p. 35
  20. ^ Çetin 2000, p. 69
  21. ^ Bugut inscription
  22. ^ a b Erdal 2004, p. 2
  23. ^ Life of Xuanzang, juan 2
  24. ^ Pelliot 1929, p. 220
  25. ^ Pelliot 1929, p. 220–221
  26. ^ a b Zhou Shu (Book of Zhou), juan 50
  27. ^ a b Tekin 1968, p. 7
  28. ^ Erdal 2004, p. 45
  29. ^ a b c Adams 1979, p. 34
  30. ^ Tekin 2003, p. 128–129
  31. ^ Barthold 2004, p. 46
  32. ^ Kül Tegin inscription; Tekin 1968, p. 344
  33. ^ Tonyukuk inscription; Tekin 1968, p. 344
  34. ^ a b c Aksan 1990, p. 7
  35. ^ Aksan 1990, p. 9
  36. ^ Aksan 1990, p. 8–9
  37. ^ Aksan 1990, p. 11
  38. ^ Aksan 1990, p. 6
  39. ^ Karaman Dezember 2021, p. 46
  40. ^ Karaman Dezember 2021, p. 46–47
  41. ^ Karaman Dezember 2021, p. 45
  42. ^ a b Özertural 2013, p. 94–95
  43. ^ Ögel 2000, p. 215–217
  44. ^ Parlatır 1983, p. 809–810
  45. ^ a b Erdal 2004, p. 11
  46. ^ a b Erdal 2004, p. 12
  47. ^ Erdal 2004, p. 10, 14
  48. ^ Erdal 2004, p. 7–8
  49. ^ Erdal 2004, p. 12–13, 19
  50. ^ Tekin 2003, p. 18, 26
  51. ^ Erdal 2004, p. 17
  52. ^ Erdal 2004, p. 35
  53. ^ a b Erdal 2004, p. 19–20
  54. ^ Erdal 2004, p. 20
  55. ^ Erdal 2004, p. 10

Sources

Literature

  • Peter B. Golden, An Introduction to the History of the Turkic Peoples (1992)
  • Wilhelm Radloff, Die alttürkischen Inschriften der Mongolei. Zweite Folge: Die Inschrift des Tonjukuk (1899)
  • E. Denison Ross (after Vilhelm Thomsen), The Orkhon Inscriptions; The Tonyukuk Inscription. Being a Translation of Professor Vilhelm Thomsen's Final Danish Rendering (1930)
  • Talat Tekin, Irk Bitig: The Book of Omens (1993)
  • Vilhelm Thomsen, Inscriptions de l'Orkhon déchiffrées (1896)
  • Emre-E. Yavuz, The Gokturk Empire: 2nd Edition (2025)
  • Marcel Erdal, A Grammar of Old Turkic (2004)
  • Zekine Özertural, Mitteliranische Einflüsse auf die manichäisch-türkische Wortbildung (2013)
  • Talât Tekin, A Grammar of Orkhon Turkic (1968)
  • Talât Tekin, Orhon Türkçesi Grameri (2003)
  • V. V. Barthold, Orta Asya Türk Tarihi Dersleri (Orta Asya Türk Tarihi Hakkında Dersler) (2004)
  • Bahaeddin Ögel, Türk Kültür Tarihine Giriş I: Türklerde köy ve şehir hayatı (2000)
  • İsmail Parlatır, Türk sosyal hayatında kölelik (1983)
  • Talât Tekin; Mehmet Ölmez, Türk Dilleri: Giriş (2003)
  • Paul Pelliot, Neuf notes sur des questions d'Asie Centrale (1929)
  • Vilhelm Thomsen, Çözülmüş Orhon Yazıtları (mit: Orhon ve Yenisey Yazıtlarının Çözümü: İlk Bildiri) (1993)
  • B. S. Adams, A Feature of the İzafet in the Orhon Inscriptions (1979)
  • Doğan Aksan, Göktürk Yazıtlarında Söz Sanatları: Güçlü Anlatım Yolları (1990)
  • Ahmet Karaman, Dağlık Altay Yazıtlarında İleri Ögeler (Dezember 2021)
  • Azzaya Badam, Moğolistan’da 2018–2021 Yılları Arasında Bulunan Runik Yazıtlar (2024)
  • Aydın, Erhan, Yenisey Yazıtlarında Geçen Unvanlar ve Unvan Niteleyicileri (2011)
  • Çumagulov, Çetin, Göktürk Harfli Yazıtların Kırgızistan’daki Araştırılması, Muhafazası ve Bugünkü Durumu (2000)

Related

What links here9

Languages
Chinese transcriptions of Turkic · Oghur Turkic · Oghuz and Kipchak speech in Kashgari · Old Uyghur · The Khalaj language · The Turkic languages
Sources
Irk Bitig · Kül Tegin inscription · Yenisei inscriptions

Cite this article

Yavuz, Emre-E. “Old Turkic.” Turkic Atlas, revised 8 November 2026. https://turkicatlas.org/languages/old-turkic/

Last revised: 8 November 2026