Links

Posting Introduction

1.    In the first two chapters of Turkish and Mongolian Studies, Royal Asiatic Society Prize Publication Fund, vol. XX, London, 1962 (hereafter referred to as Studies), I gave a brief sketch of the history of the Turkish peoples and of their languages from the earliest period (the eighth century a.d.) at which they become directly known to us down to the medieval period, and attempted to identify and name the various dialects and languages which evolved during that period.

2.    The broad thesis of this sketch was that a unitary Turkish language, which was not genetically connected with any other language known to us, and specifically not connected genetically with the Mongolian and Tungus languages, took shape, almost certainly in the steppe country to the west and north of the Great Wall of China, at some date which we cannot now determine, but certainly long before the start of the Christian era; that this unitary language split into two main branches, ‘standard Turkish’ and 'l/r Turkish’, not later than, and perhaps before, the beginning of the Christian era; and that during the first millennium a.d. standard Turkish slowly broke up into two or three dialects, which soon became independent languages. No substantial early texts of any form of 1/r Turkish have survived, but there are in Chinese historical works a few words of Tavğaç, the language of the Turkish tribe called by the Chinese T’o-pa, who were the ruling element in the Chinese Northern, or Yiian, Wei dynasty (a.d. 386-535), and there are some Turkish loan-words in Mongolian and Hungarian which were almost certainly borrowed from an 1/r language, by the Mongols probably in the fifth or sixth centuries, and bv the Hungarians probably in the ninth. The slight remains of Proto-Bulgar are in a similar language and so too are a few funerary inscriptions of the Volga Bulgars of the thirteenth and fourteenth centuries. The material is, however, very scanty and I have made only a few references herein to words in 1/r Turkish, and those mainly where the only surviving 1/r language, Chuvash (Çuv.), throws light on the phonetic structure of individual standard Turkish words. These citations are from N. I. Ashmarin, Thesaurus linguae Tschuvaschorum, 17 vols., Kazan-Cheboksary, 1928-50, cited as Ash. followed by the volume and page.

3.    As I pointed out in Studies, p. 53, an essential basis for any scientific study of the comparative etymology of the Turkish languages is a list of all germine Turkish words which are known to have existed before the various medieval Turkish languages were subjected to a muss invasion of loan-words from the langurs of the foreign peoples (mainly Arabs, Iranians, and later Mongols) with > rTvmi the Turkish peoples then came into contact. The present work contains just such a list. I have endeavoured to include in it all the words known t oyna vc existed in the earliest known standard Turkish languages, that is Türkü,
vi
Uygur, including Uyğıır-A, Old Kırğız, Xakani, Oğuz, Kıpçak, and sonic minor dialects, prior to the Mongolian invasion at the beginning of the thirteenth century, excluding, except in a few special eases, words borrowed directly or indirectly from the Indian, Iranian, and Semitic languages. All references to ‘early Turkish' and ‘the early period’ relate to the Turkish languages as we know them in the eighth to twelfth centuries inclusive.

4.    One of the special characteristics of the Turkish languages is the manner in which derived words are constructed from the basic words of the language by a process of attaching suffixes to them. It is important, therefore, not merely to compile a bare list of these early words and their meanings, but also to indicate what kind of words they are (nouns, verbs, adverbs, etc.), verbs being distinguished from the rest by the attachment of a hyphen, e.g. at (noun) ‘a horse’, at- (verb) ‘to throw’, and in addition, in the case of derived words, to analyse their structure and indicate from what basic words they were derived, since this often explains their semantic evolution. This is all the more important since, even at the earliest date at which it becomes known to us, Turkish had already had a long history, in the course of which some basic words had already become obsolete, leaving the words derived from them, so to speak, ‘in the air’, with no obvious origin; on this point see, for example, Studies, pp. 140 ff. In giving these morphological explanations I have used the grammatical terminology customary among English scholars of Turkish, eschewing the new terminologies which are now taking shape in some academic circles.

5.    My second purpose has been to establish the original meanings of all words in the list and trace their subsequent semantic evolution by illustrating each word by a quotation, or series of quotations, of passages in early texts in which they occur. For this purpose I have examined all the surviving texts in the languages enumerated above to which I have been able to obtain access. These texts are enumerated in paras. 20-6 below.

6.    ’I'he only words which have been included in the list without at least one such quotation are a few which can first be traced in an actual Turkish text only in the medieval period, that is during the thirteenth to sixteenth centuries inclusive, but which demonstrably existed at an earlier period for such reasons as that they exist both in standard Turkish and in Chuvash (in a distinctive form peculiar to that language) so that they must already have existed before the split between standard and I/r Turkish, or that words derived from them are attested in the early period, or that they occur as loan-words in foreign languages in forms which indicate that they must have been borrowed in, or before, the early period. ’I’he texts which have survived from the early period are of course only a fraction of those which once existed and there are no doubt other words which existed in the early period but do not happen to occur in the texts which have survived. Some of these words occur in medieval texts or still exist in modern Turkish languages, but I have not attempted to identify and include them in the list, since the evidence that they were really early words is lacking.
vii

7.    A surprisingly large number of early words are hapax legomena (Hap. leg.), that is, occur only once, and have not so far been discovered elsewhere; they have, however, been included in the list because they seem to be genuine Turkish words and may later be discovered elsewhere; but the fact that they are Hap. leg. makes it impossible to determine whether they are correctly transcribed. Properly speaking the term implies one single occurrence, but it has been stretched to include verbs listed in Kaş. with an enumeration of the Perf., Aor., and Infin. forms, since experience shows that if a word is mis-spelt in the Perf. it may also be mis-spelt in the Aor. and Infin.

8.    When an early word has survived into the medieval period, I have also added a selection of quotations from the texts enumerated in paras. 27-52, and when it still survives in one or more modern languages I have usually recorded this fact also. There are two reasons for doing this: the first is that evidence of this kind is often required to fix the exact meaning and pronunciation of the word in the early period; the second is that, as I understand it, the main purpose of compiling a dictionary of this kind is not only to help students to read and understand the meaning of early texts, but also to provide a solid basis for the etymological dictionaries of medieval and modern Turkish languages which are now being compiled or will be compiled in future. I must, however, make it clear that my quotations from medieval texts and enumerations of modern languages in which a word survives do not profess to be as comprehensive as the quotations from early texts.

THE SYSTEM OF TRANSCRIPTION

9.    The problems of transcribing early Turkish texts are discussed at length in Studies, Chap. 3 and elsewhere, and it is unnecessary to cover the whole ground again. There is, of course, no means of discovering exactly how Turkish was pronounced between 900 and 1200 years ago; the alphabets employed are all to a greater or less degree unscientific; the most that can be achieved is a broad system of transcription which will give an approximate idea of the pronunciation of the words concerned in the languages enumerated in para. 3. The transcription alphabet which I suggested in Studies and which is used here can best be described as the official alphabet of the Turkish Republic with one minor modification (regarding the use of ğ and g) and a few extra letters to represent sounds which either do not exist in Republican Turkish (Rep. Turkish) or, if they exist, are not represented by a distinctive letter. This alphabet was selected on the basis that the phonetic structure of early Turkish was probably much the same as that of Rep. Turkish, apart from certain sounds for which special letters have been provided. These sounds can be tabulated as follows:

10.    Vowels. There are four back and five front vowels, all of which may be either short or long. The only additional letter is 6 for close e, a sound which
viii

certainly existed in early Turkish and still survives at any rate in some dialects of Rep. Turkish.

„ , (short aiou liack {,

(long a: 1: o: u:

.. (short e e i ö ii

1‘ront .    ......

(long e: c: 1: o: u:

These arc arranged, when circumstances demand it, in the following order: a, a:, e, e:, e, e:, i, i:, i, i:, o, o:, u, u:, ö, ö:, ü, Ü:; note that u, u: precede ö, Ö:. The round vowels represent a range of sounds which merge imperceptibly into one another in the various languages.

In most, but not all, modern languages the vowels o and ö occur only in first syllables of words, but there is good evidence, which I assembled in ‘Three Notes on Early Turkish’, Türk Dili Araştırmaları Yıllığı, 1966, pp. 1 fF., that they were also used in the second and later syllables of many words in early Turkish. This evidence is, however, too fragmentary and imperfect, and in some cases too inconsistent, to justify restoring such spellings at the beginning of all the entries concerned. I have therefore entered all these words in the conventional spelling with u/ü in the second and later syllables. Where there is good evidence in regard to individual words that these vowels were originally o/ö I have entered these spellings in brackets, e.g. törü: (törö:), but I must emphasize that these sounds occurred in far more words than are singled out for this treatment, for example, in all words with the Suffix -uk/-ük (-ok/-ök).

11. Consonants. The conclusion reached in Studies was that the sounds which existed in early Turkish can be tabulated as follows, v. representing voiced and u. unvoiced sounds:

     M.

Liquids (v.) 1 r (Aspirate (u.) h)

The post-palatal sounds occur only in association with front vowels and the velar sounds only in association with back vowels, so only one letter, k, is required to represent both unvoiced plosives. The unvoiced labial fricative f probably did not exist in early Turkish as an independent sound, but seems to
ix
have arisen as a secondary form of v in association with some unvoiced sounds. The voiced palatal sibilant probably did not exist as an independent sound in early Turkish, but is noted as a secondary sound in one or two words containing ç, e.g. çöj- probably for çöz-. It also occurred in a few Sogdian loan-words, e.g. a:ju:n, and a few other words, e.g. üjme:, which are probably, but not demonstrably, Sogdian loan-words. The labial semi-vowel w was not a native Turkish sound, but evolved as a secondary form of v in one or two languages like early Oğuz and also occurred in some loan-words. The voiced post-palatal g was normally a plosive, but may in certain positions have been a fricative; conversely the voiced velar ğ was normally a fricative but may in certain positions have been a plosive. The unvoiced velar fricative x probably did not exist as an independent sound in early Turkish, but seems to have arisen as a secondary form of k in association with some sounds, and also occurred in a few loanwords, some of them very old like xağan and xan. Kâşğarî says that the simple aspirate h was not a native Turkish sound but occurred in some ‘impure’ dialects and in one or two words like ühi: in which it seems to be a secondary form of g.

The consonants are arranged, when the circumstances demand it, in the following order: b, p, v, (f), (w); c, ç, j; d, d, t; ğ, (velar) k, X; g, (postpalatal) k, (h) ; 1; m ; n, r), n ; r ; s ; ş ; y ; z. Note that the letters are arranged in classes, and not in the normal conventional order, but that the first letters of the classes are arranged in the conventional order.

12. There are some particular difficulties in transcribing some early languages, of which the following are the most important:

(1)    In Türkü there was a tendency towards dissimilation, so that when a suffix beginning alternatively with a voiced or unvoiced consonant like the Perf. suffix -di:/-ti:/-di:/-ti: was attached to a verb ending in a consonant, the unvoiced form was attached to a voiced consonant and vice versa, e.g. erti:, tapdi:. In Xak. the position tended to be exactly the opposite, at any rate to the extent that suffixes with voiced initials were attached to verbs ending with voiced consonants, e.g. erdi:, although cases of suffixes with voiced initials being attached to verbs ending in unvoiced consonants are very common, at any rate in our manuscript of Kaş. Chronologically speaking Uyğ. lay squarely between Türkü and Xak. but nearly all the alphabets in which it is written are so ambiguous that it is quite uncertain whether it was a dissimilating language like the former or an assimilating language like the latter, or whether the change took place at some date between the eighth and eleventh centuries. I cannot therefore claim to have been in the least consistent in transcribing Uyğ. texts, usually following the scriptions in the published texts.

(2)    The Runic alphabet (see Studies, pp. 51 ff.) in which nearly all the Türkü texts have survived, the Manichaean Syriac alphabet, and the Brâhmî alphabet had different letters for b and p and for g and post-palatal k, the other early alphabets did not, and the Brâhmî spellings are very erratic; there is indeed some doubt regarding the phonetic value of some of the letters of this alphabet
ix
in its Central Asiatic form. It is therefore almost impossible to decide how some words should be transcribed. In some cases assistance can be got from the modern languages, but some of these have undergone quite considerable phonetic changes.

(3) Only one of the alphabets concerned, Brâhmî, had separate letters for o, u, ö and ü, and the spellings of the texts in this alphabet are often inconsistent. In this case, however, the correct form can often be established if the word was a loan-word in Mongolian, or survives in modern languages; but the spellings in modern languages are often inconsistent, particularly as between languages in the Oğuz group and the rest.

13,    The alphabet set out above has also been used for transcribing (or perhaps it would be more accurate to say transliterating, since the system adopted is purely mechanical) quotations from Arabic and Persian texts, with the addition of the following letters: t, d, t, h, ş, and z, hamsa'ayn and q for the unvoiced velar plosive, which must be distinguished from the unvoiced post-palatal plosive in these languages. In these languages, too, vowel length is indicated in the usual manner by a superscribed line, e.g. ā. The same letters are also used to transcribe Mongolian, but Sanskrit is transcribed in the conventional way.

THE ARRANGEMENT OF WORDS IN THE DICTIONARY

14.    As pointed out above, the Turkish texts quoted in this book are written in a variety of alphabets, all more or less ambiguous, and it is often impossible to determine the correct transcription of a number of words; moreover, some words were pronounced slightly differently in different languages. It would, therefore, not be sensible to arrange the words in the strict alphabetical order to which we are accustomed in the dictionaries of European languages, since this would involve a great many double or multiple entries and greatly add to the difficulty of finding individual words. This problem and the means of solving it are discussed in detail in Studies, Chap. 4. The arrangement which emerged from this discussion can be summarized as follows:

(i) The words are broken down into fourteen groups, containing respectively words beginning with the following sounds:

(2) Each of these fourteen groups is divided into sub-groups; in each case the first sub-group contains monosyllables ending in an open vowel; the order of the remainder is determined in the case of the first group (words beginning with vowels) by the first consonant, and, if there is one, the second consonant or failing that the final vowel, if any, and in the case of other groups by the second consonant, and the third consonant, if any, or failing that the final vowel, if any. For purposes of classification these are the only significant sounds; vowels other than initial and final vowels are disregarded except for purposes of arranging the order of the words in a sub-group.

15.    Each sub-group is identified by from one to three code letters representing the significant sounds enumerated above, the code letters being as follows:

A, or if one of the significant sounds is post-palatal g or k, E, representing vowels.
B representing labials, b, p, v, (f), (w) but not m.
C representing denti-palatal affricates c, ç, (j).
D representing dentals d, d, t but not n.
Ğ representing velars ğ, k, x but not ŋ.
G representing post-palatals g, k, (h) but not I).
L representing 1.
M representing m.
N representing n, ŋ, ñ.
R representing r.
S representing s.
Ş representing ş.
Y representing y.
Z representing z.    '

Each sub-group is divided into the following parts:

(1)    monosyllables, if any, coded as Mon.
(2)    monosyllabic verbs, if any, coded as Mon. V.
(3)    disyllables, coded as Dis.
(4)    disyllabic verbs, coded as Dis. V.
(5)    longer words, coded as Tris.
(6)    longer verbs, coded as Tris. V.

16.    In order to locate a word in the dictionary:

(1)    declensional and conjugational suffixes must be removed;
(2)    the part in which it appears in the dictionary must be determined by converting the significant sounds in it into the appropriate code letters, account being taken of the fact whether it is either a verb or not a verb, e.g.:

xii

            17.    Within each part of a sub-group the words are arranged in the following order, which is designed to make it as easy as possible to locate a word written in any of the various scripts:

(1)    in Mon. and Mon. V. in the alphabetical order of the vowel, and when two or more words have the same vowel (o and u, and ö and ü, respectively being regarded as identical since they are not distinguished in these scripts) in the order of the final consonants; thus in Mon. V. AD- the order is ad-, at-, et-, ı:d-, it-, u:d-, ut-, üd-, öt-, üt-.
(2)    in Dis. and Dis. V.,
(a)    all words ending in a closed syllable precede those ending in an open vowel, thus in Dis. V. ADL- ütül- precedes atla:-;
(b)    if two words contain, in addition to the letters represented in the heading, further consonants belonging to different classes, these determine their order; thus in Dis. V. SCD- sıçtur- precedes saçtaş-;
(c)    in other cases, subject to (a) above, the order is determined by the first vowel; thus under Dis. V. SCD- the order is saçıt-, sıçıt-, suçıt-, süçlt-, and under Dis. V. ADL- atla:-, edle:-, or, if the vowels are identical, by the order of the consonants following, thus under Dis. V. ADL- odul, udul-, otul-.
(3)    in Tris. and Tris. V. the same principles are observed, mutatis mutandis.

18.    Most words in the dictionary are preceded by code letters in capitals, only basic words of which the pronunciation is reasonably certain not being preceded by one. These letters have the following meaning:

(1)    C compound; used when a word is composed of two shorter words, e.g. C içkur ‘belt', composed of iç and kur. Such words are rare,
(2)    D Derived; placed before every word other than a compound which is not a basic word, e.g. D evdil-, the Passive form of evdi:-.
(3)    F Foreign: placed before words which are known to be loan-words, e.g. F a:ju:n borrowed from Sogdian. Words which are believed to be loan-words are preceded by ?F if definite proof is lacking.
(4)    PU Pronunciation uncertain; see (6).
(5)    S Secondary; placed before words which are secondary forms of other words in the list; such words are followed by a reference to the words of which they are the secondary forms.
(6)    VU Vocalization uncertain. This prefix is used when the vocalization of a word is uncertain but there is no doubt about its consonantal structure, e.g. VU eteç indicates that this is the most probable pronunciation of the word but that it might be ataç, atıç, atuç, etc. Words are preceded by PU when there is some doubt about the pronunciation of one or more consonants in them whether or not the vocalization is certain.
xiii
(7)    When a word is known to have existed at an earlier date but was already obsolete by the eighth century so that it is known only from its derivatives the basic word is entered with a preceding asterisk and followed by cross-references to some of the words derived from it, e.g. 1 *ad-, see adın, adjr-.
(8)    E Erroneous; unfortunately a few completely non-existent words have found their way into some medieval and modern dictionaries and word-lists owing to misreadings of ancient texts. A small selection of these is included in the list, preceded by E and followed by an explanation, in order to indicate that such words really did not exist and have not merely been overlooked, e.g. E amuç which arises from an erroneous scription of the Arabic word asahh in the manuscript of Kaş.

AUTHORITIES QUOTED, WITH SHORT TITLES

I . EARLY TEXTS

19.    All quotations in the paragraphs about individual words are preceded by the name of the language in which they were written and the date of composition expressed in centuries denoted by roman numerals; viii = eighth century a.d. and so on. When two or more quotations are taken from the same language they are grouped under a single head, possibly with different chronological indications.

A. Türkü

20.    Two periods are distinguished:

(1) viii. Several major inscriptions in the Runic alphabet fall under this heading; there is a brief note on the best editions of them in Studies, p. 68, but my own quotations are derived, wherever possible, from a personal examination of the photographs and squeezes of the inscriptions. As a result in one or two cases I have, I think, found better readings of some difficult passages than those in any of the editions; examples will be found under olğurt-, berüki, sın-. In addition to the major inscriptions listed separately below there are some shorter inscriptions, but as they are undated and may be later than viii they are included under the next heading. The major inscriptions are the following:

(a)    the funerary inscription of Tonukuk (this is the customary transcription of the name but it is VU) composed probably in the second, but possibly the third decade of viii; cited as Tfollowed by the number of the line;
(b)    the funerary inscription of Kül Tegin, composed in A.D. 732, cited as I followed by the side (E, W, N, S) of the stone and the line on the side;
(c)    the funerary inscription of Bilge: Xagan, composed in a.d. 735, cited as II similarly followed;
xiv
(d)    the funerary inscription of the Küli Çors at Ixe-Xu şotu in central Mongolia, roughly contemporary with the preceding, cited as Ix. followed by the line;
(e)    the relatively brief inscription at Ongin, cited as Origin followed by the line.
(2) viii ff. Under this heading are included texts which may have been composed in viii, but were possibly, in some eases probably, composed in IX, or perhaps in one or two cases even later. These fall under the following heads:

(a)    a few minor inscriptions from Mongolia, of which the most convenient editions are those in the second volume of H. N. Orkun, Eski Türk Yazıtları, 3 volumes and index, Istanbul, 1936-41, (cited as ETY followed by the volume in roman and the page in arabic figures). These are occasionally cited Under the names given in that volume followed by a reference to ETY 77;

(b)    those of the so-called Yeniseian (Yen.) inscriptions which were found in Khakassia and were probably erected by minor Western Türkü chieftains who escaped to Khakassia after the fall of the Türkü empire in the middle of viii, see Studies, p. 69. The most convenient edition is that in S. Ye. Malov, Yeniseiskaya pis’niennost' iyurkov, Moscow, 1Ç52; they are cited as Mal followed by the number of the inscription and the line, but my readings sometimes differ from Malov’s;

(c)    a few inscriptions found at Ayrtarri Oy near the town bf Talas ori the river of the same name and probably datable to ix or x (see Studies, pp. 72 and 256). The texts are almost illiterate and in a deplorable state, and I have quoted them only occasionally, citing them as Talas followed by the number öf the inscription and the line;

(d)    paper manuscripts in the Runic alphabet. The only substantial one is a book of divination called Irk Bitig, which I cite as Irk B followed by the number of the paragraph and not, as has usually been done, by the page, since the page numbers used do not correspond to those of the actual manuscript. The most convenient edition of the rest, which are rather miscellaneous in character, is in ETY II, but as Orkun’s system of references is rather chaotic 1 have substituted my own, followed by a reference to the page in ETY II. It is commonly believed that these documents can be dated to about ix;

(e)    Manichaean (Man.) texts in the Manichaean Syriac or Uygur alphabets and what Prof. A. von Gabain in her Alttiirkische Grammatik, Leipzig, 1941, p. 5 (cited as v. G. ATG) called ‘the N-dialect’, but which seems in fact, see Studies, p., 118, to be Türkü written iri a non-Runic alphabet. The longest and most important of these texts is the ‘Confession of Sins’ known as the Chuas-tuanift (more correctly Xwastwaneft). The best manuscript, almost complete, is in the British Museum, and there are other fragments, some of the beginning of the text missing in that manuscript, in Germany. The most convenient edition, with a facsimile of the B.M. manuscript, is still that by A. von Le Coq in JRAS 1911, pp. 279 ff., although the translation is not entirely accurate. I cite it as Chuas. followed by the line of the B.M. manuscript or, in the case of the first part, as Chuas. I followed by the line in von Le Coq’s numeration. The other
xv
Manichaean Türkü texts which I quote have all been published in the Abhand-lungen or Sitzungsberichte of the Academy of Sciences in Berlin (originally the Königliche preussische, later Preussische, and later still Dcutsche, Akademie der Wissenschaften). As these publications also contain Manichaean texts in Uyğur and Uyğur-A it is more convenient to list all the publications of Manichaean texts here to avoid additional references in subsequent paragraphs. The main bulk of them is in A. von Le Coq’s Tiirkische Manichaica aus Chotscho I (AKPAW, 1912); II (APAW, 1919); III (APAW, 1922), cited as MI, II, and III followed by the page and line, the last followed, by (i), (ii), etc. if the same line number appears more than once on the same page. Others were published in a series called Tiirkische Turfantexte (cited as TT), edited at one time or another by W. Bang, A. von Gabain, and G. R. Rachmati (later R. R. Arat), which also contains Uyğur Buddhist and Civil texts. Those parts which contain Manichaean texts (mostly not in Manichaean Türkü) are TT 77 (SPAW, 1929) cited by page and line, and TT 777 (SPAW, 1930) and TT IX (ADAW, 1958) both in Uyğur and cited by line. Other short Manichaean texts are in A. von Le Coq, Eirt manichâisch-uigurisches Fragment aus Idiqut-Schachri (SKPAW, 1908) citcd as Man.-uig, Frag, followed by page and line, and W. Bang and A. von Gabain, Eirt uigurisches Fragment iiber den manichauchen Windgott (Ungarische jahrbücher VIII, pp. 247 ff.), cited as Wind, followed by the line.

B. Uyğur and Uyğur-A (Uyğ. and Uyğ.-A)

21. The reasons for supposing that two separate, but closely related, Uyğ. dialects existed are stated in Studies, p. 42. There are obvious traces of Uyğ.-A in some Buddhist texts, e.g. those in TT VI and TM IV(althoiight it is stated in the introduction to the latter that the Uyğ.-A forms had been ‘corrected'in the transcription so that they are no longer apparent); but such texts do not seem to be entirely homogeneous, so that it is only in the, Manicheaean. texts that it seems possible more or less to isolate the two dialects. Three periods can be distinguished:

(1)    viii. The earliest and most substantial Uyğ. inscription in the Runic alphabet is that situated at Şine-usu in Outer Mongolia, which is the funerary monument of an Uyğur Xagan who ruled from a.d. 746 to 759. It is >cited as Şu. followed by the side and line on the side. ^;    .■ ^{1 :}

(2)    IX. The second such inscription, only fragments of the Uyğ. text of -which still survive, is the funerary memorial at Kara Balgasuh of an Uyğur Xagan who probably ruled from a.d. 808 to 821. It is cited as III followed by a letter and line and a reference to ETY II. The third is the short funerary inscription at Suci in Outer Mongolia, probably datable to the first quarter of ix, and cited as Suci followed by the line. ...........    •;

(3)    viii ff., that is, in this case, texts the earliest of which may go back as far as vui, although the actual manuscripts are probably later, while the latest, apart from the xviii manuscript of the Suvarnaprabhâsa Sutra, the composition_; of which was much earlier, is probably a Civil document dated firmly to the .second
xvii
lines i to 324, the second lines 1763 to 2160, the rest being unpublished); and the Turkish translation of the Suvarnaprabhâsasütra, published by V. V. Radlov and S. Ye. Malov, St. Petersburg, 1913 ff.; cited as Suv. followed by the page and line.

(d) Civil (Civ.) texts, that is not religious. These fall into four classes:

(i)    Medical texts, probably mere translations of texts in other languages. The main collection is in G. R. Rachmati, 7.ur Heilkunde der Uiguren I (SPAW, 1930) cited as HI followed by the line, and II (SPAW, 1932) cited as H II followed by the page and line. There are also one or two such texts in TT VII and VIII.

(ii)    Astronomic, astrological, magic, and omen texts, published in TT I (SPAW, 1929), cited by the line, TT VII and VIII and USp. Texts of the first two kinds are mainly translated from the Chinese, generally with an Indian background, and are mostly late in the period; the line between unorthodox Buddhist and purely magical texts is indefinite and I may have put some texts the wrong side of it.

(iii)    Commercial and legal documents and (iv) miscellaneous texts. The main collection of these is in USp,; there is one in TT VII, No. 42, a very late text apparently transcribed from an original in the Arabic alphabet, and there are one or two published elsewhere. When quoting the last I have indicated where they may be found.

A good many of the Uyğ. texts listed above have been completely or partially indexed; but no index is available for others, including Suv. which is much the longest, and I cannot guarantee that I have listed every word which occurs in these unindexed texts.

C. Old Kırğız (O. Kir.)

22.    It is commonly believed that those of the Yeniseian inscriptions which were found not in Khakassia but in Tuva were set up by Kırğız chieftains in that area and are in the Old Kırğız language. It has been suggested that they are the oldest monuments in the Runic alphabet and may date back as far as vi, but L. R. Kyzlasov has recently proved by archaeological methods (see Studies, p. 70) that most of them date from ix and x and that some may be even later. I have classified them as ix ff. The most convenient edition is that of Malov mentioned in para. 20 (2) (b), but A. M. Shcherbak has published revised texts of some of them, and some photographs and squeezes are available. I have therefore not always followed Malov’s readings.

D. Xakani (Xak.)

23.    Xak. is a language closely related both to Türkü and to Uyğ., but sufficiently distinct from both to be regarded as a separate language. It was certainly not directly descended from the latter, indeed it existed side by side with Uyğ. for two or three centuries, and was perhaps not quite directly descended from the fof-mçr, It is first identified in xi, and there are two Xak. xi texts of major
xvi
quarter of xiv. The actual date of composition of the remainder is quite uncertain; some of the legal and commercial documents are certainly not older than xiii and contain Mongolian, Arabic, and Persian loan-words, but I have taken them into account with suitable precautions, since it is quite impossible to divide these documents neatly into two categories, those anterior and those subsequent to a.d. 1200. These texts can be divided into the following classes:

(a)    Christian texts (Chr.). These are likely to be quite early in the period. Only two seem to have survived, the well-known apocryphon about the Magi published in F. W. K. Müller, Vigurica (AKPAW, 1908) pp. 5 ff., cited as U 1 followed by the page and line, and a fragment published in M III cited by page and line.

(b)    Manichaean (Man.) texts in Uyğ. and Uyğ.-A published in the various works listed in para. 20 (2) (e) above. All are fragmentary and vary greatly in length, though none are very long; in the case of the longer texts it is usually possible to determine whether they are in Man. Tiirkü, Uyğ. or Uyğ.-A; when this is impossible they are cited as Uyğ. The earliest of these texts probably go back to viii and the latest are unlikely to be much later than ix.

(c)    Buddhist (Bud.) texts. These have been puhlished in substantial quantities; the following is a list of those which I have quoted. There are four volumes of Uigurica, the first three edited by F. W. K. Müller, the last by A. von Gabain. U 1 has already been mentioned under («); U II (AKPAW, 1911), U III (APAW, 1922), and UIV(APAW, 1931) are cited by page and line. The following volumes of the TTseries mentioned in para. 20 (2) («) contain Buddhist texts: TT1V(SPAW, 1930); TT F(SPAW, 1931); TT F7(SPAW, 1934); TT VU (APAW, 1937); TT VIII (ADAW, 1954, the volume containing the texts in the Brâhmî alphabet), and TT X (ADAW, 1959). Of these VI and X are cited by the line, IV and V by page and line, and VII and VI11 by document and line. There is a useful and complete Analytischer Index zu den jünf ersten Stikhen der TT(SPAW, 1931) and VI contains a partial and Vll, VIII, and X a complete index of words. Another AKPAW volume containing Buddhist texts is F. W. K. Müller, Ztvei Pfahlinschriften aus dem Turfanjunden, 1915, cited as Pfahl. followed by the page and line. W. Radloif, Uigurische Sprachdenkmaler, Leningrad, 1928, cited as USp. followed by document number and line, contains inter alia a number of Buddhist texts. Other publications which I have quoted are the following: Paul Pelliot, La Version ouigoure de I’histoire des princes Kalyânamkara et Pâpamkara, T’oung Pao XV. 2, 1914, cited as PP followed by the page and line; W. Bang and G. R. Rachmati, Türlü Cehennemler üzerine Uygurca Parçalar, Türkiyat Mecmuasi IV, İstanbul, 1934, cited as TM IV followed by the line; Şinasi Tekin, Kuanşi İm Pusar, Erzurum, i960 (superseding RadlofF’s partial edition, St. Petersburg, 1911) cited as Kuan. followed by the line; W. Radloff, Tisastvustik, ein in tiirkischrr Sprachebearbeiletes buddhistisches Sutra, St. Petersburg, 1910, cited as 7Yj., followed by the manuscript page and line; A. von Gabain, Die uigurische Übersetzung der Biographie Hiien-tsangs (SPAW, 1935) and Brie/e der ıdgurischen Ilüen-tsang-Biographie (SPAW, 1938)1 both cited as Hiien-ts. followed by the line (the first containing
xviii
importance. The first is the earliest, and by far the most important, Turkish-Arabic dictionary, the Divinn Lıtğâti'1-Turk of Mahmud al-Kâşğari, dating from the third quarter of xi and cited as Kaş. The quotations in the form Kaş. followed by a roman number and an arabic number and sometimes a second arabic number are from the Turkish translation by B. Atalay, Divattü Lvgat-it-türk Tercümesi, 3 volumes and index, Ankara, 1940-3, since this edition is reasonably accessible, and the index is excellent and comprehensive. The principle which

I have followed is to cite the volume and page for main entries and the line as well for quotations from other parts of the book. I have, I hope, included every word which it contains. Atalay’s translation is not wholly satisfactory and I have in every case consulted the facsimile {the printed text contains many rriinor inaccuracies) and normally quote the original Arabic in my citations, so that the accuracy of the translation can be checked. In a number of cases I have altered Atalay’s readings and spellings. The second major text is the Kutadğtı: Bilig of Yûsuf Xaşş Hâcib of Balasagun, the earliest surviving purely literary text in any Turkish language, also dating from the third quarter of xi. My quotations are taken from R. R. Arat’s critical edition, Istanbul, 1947, cited as KB followed by the line in that edition. In cases of doubt I have also consulted the facsimiles of the three manuscripts; in some cases tny spellings are slightly different from Arat's. As his index has not yet been published, I cannot guarantee that 1 have listed every word which occurs in this text, but I have searched it fairly thoroughly. I have classified all the words from these two sources as xi since this date is certain. The Verse Preface of KB is not part of the original text and its language is obviously a little later; quotations from it have therefore been entered under the heading of xii (?) KB VP followed by the line. The Prpse Preface is later still and perhaps not really Xak. but a few quotations from it have been entered as xm(?) KBPP followed by the line.

E. Oğuz

24. Kişğarî lists a certain number of words in a language which he calls ‘the language of al-Gtizziya' and says that owing to long contact between the people who spoke it and Persian-speakers it had acquired a number of Persian loanwords, often in a rather corrupt form (see ören and turma:). It was no doubt the language spoken by the Oğuz tribes, who were then living in the neighbourhood of the Aral Sea and had to some extent lost contact with the other Turkish tribes. They were descendants of the Oğuz (Tokkuz Oğuz etc.) who had played quite an important part in history during vii and viii and probably even earlier. There is no reason to suppose that during that period they spoke a language markedly different from that of the Türkü; but although even by xi the Oğuz language had undergone some phonetic decay and accepted a number of loanwords, it retained, and its modern descendants still retain, some characteristics more primitive even than those of Türkü, e.g. initial d- in certain words (see para. 14 (1)). Only a few specifically Oğuz words are listed in Kaş. but they are the earliest surviving specimens of the language.
xix

F. Kıpçak {Kıp.)

25.    Kâşğarî also cites a number of words in a language which he calls ‘the language of Qifcâq’, a language which he describes as similar to Oğuz. Several words are mentioned as being common to Kip. and Oğuz but not known in Xak. In xi the Kıpçak were west of the Oğuz in southern Russia and Kaş. is the earliest authority for their language.

G. Minor eleventh-century dialects

26.    Kâşğarî describes a few words as belonging to other contemporary dialects with which he was acquainted; those mentioned most often are the dialects of Arğu:, Çigil, and Gancak (the last probably not really Turkish, see Studies, p. 132 and the remarks on kendük below). I have included these words with Kâşğari’s descriptions of them.    .

II. MEDIEVAL TEXTS

27.    The Uyğ. language lingered on for some centuries in Chinese Türkistan (Sinkiang); indeed the latest surviving Bud. Uyğ. manuscript, that of the Suvarnaprabhösasütra already mentioned, was copied in xviii, but the only accessible authority for Uyğ. in this latest state is the Chinese Uyğ. dictionary (Chin.-Uyğ. Dict.) prepared by a committee of Chinese scholars towards the end of xiv. Some words from this work are included in Radloff's Worterbuch and a complete list from another edition was recently published by Prof. L. Ligeti in Un Vocabulaire sino-ouigoure des Ming, le Kao-tcKang-kouan Yi-chou du Bureau des Traducteurs, Dissertationes Sodalium Instituti Asiae Interioris II, Budapest, 1966. I have included references to this work as Ligeti, followed by the page.

28.    The difficulty about classifying most of the remaining medieval texts, which were, with only one or two exceptions, written by or for Moslem Turks, is that in xi the only literary language in which texts have survived, and probably the only literary language then in use, was Xak., but that towards the end of that century or very soon afterwards literary languages emerged which were written in different areas where the spoken languages were rather different from one another, and so very soon began to pursue different courses. There are interesting discussions of the problem of classifying the texts concerned in Philologiae Turciçae Fundamenta I, Wiesbaden, 1959 (cited as Fundamenta) and more particularly in J. Eckmann’s article ‘Das Chwarezmtiirkische’ in that volume, and also in the introduction to A. K. Borovkov, Leksika sredneaziatskogo Tefsira XIII—XV w., Moscow, 1963 (cited as Tef.).    ;    . -

29.    What can perhaps be called the direct line of descent, or southern stream, evolved in the area called in the early medieval period Khorasan and the southern part of Mâ wara’u’l-nahr (Transoxiana). By xv the literary language used in that area was what is commonly, if inaccurately, called Çağatay (Çağ.), but at any
xx
rate three, and probably four, texts have survived which were written between xi and xv (both excluded) in languages intermediate between Xak. and Çağ. The spoken language in this area during this period seems to have been fairly homogeneous with the literary language, except that there was probably a greater concentration of Arabic and Persian, and perhaps from xiii onwards also Mongolian, loan-words in the literary than in the spoken language.

30.    Possibly the earliest of these intermediate texts which, since it contains some Mongolian words, cannot be earlier than xiii, is the didactic poem edited by R. R. Arat under the title Atebetii'l-hakayik, Istanbul, 1951, cited as At. followed by the lines.

31.    An even earlier date has been suggested for a few Turkish tafsirs, that is interlinear translations of the Koran (mostly bilingual, some trilingual with a Persian translation of the Arabic and a Turkish translation of the Persian) with commentaries and other additional matter. The difficulty in determining the date of these texts and the exact language in which they were composed lies in the fact that the surviving manuscripts of them were all written later, perhaps a good deal later, than the date at which they were composed and have been subjected to a good deal of modernizing by the copyists, later phonetic forms, and even actual words, being substituted for the author’s own phonetic forms and words. It is not at all clear whether all, or at any rate all the earlier, bilingual manuscripts go back to a single archetype or whether some are independent translations and compilations, but the latter is the more probable. Borovkov in Tef. has indexed the Turkish vocabulary of one such work, the manuscript of which was found in the town of Karshi. The name of the author, the date of composition, the name of the copyist, and the date of copying are all equally unknown, but the manuscript seems to be later than that of a similar tafnr found in Turkey dated a.d. 1333 (see Tef., p. 4) which is said to have a text close to that of the Karshi manuscript. An examination of the vocabulary shows that it is very heterogeneous. It includes a number of words of great antiquity for some of which, e.g. 2 ap, it is the latest authority. On the other hand, it also includes Arabic, Persian, and even some Mongolian, loan-words, which proves that in its present form it cannot be earlier than xm, and it even contains different forms of the same word, e.g. adrıl-, ayrıl-, ayrul-, which belong to different periods in the history of the language. It has been suggested that the.se are evidence that the language used, even if basically Xak., has been heavily influenced by Oğuz and Kip. and that it must therefore have been composed in some northern area and ought to be classified as Xwarazmian (see para. 37). There is not, however, any conclusive evidence that this is so, and there are other possible explanations of the non-Xak. elements in the text. In the circumstances it seems better to regard the language as intermediate between Xak. and Çağ,, to date Tef. as xm(?) and cite it immediately after At.

32.    Next in time, and in much the same language, comes the Qisasu’l-anbiyâ of Nâşiru’l-dîn al-Rabğûzî composed in a.d. 1310. In this case, too, all the surviving manuscripts are later than the work itself and have been subjected to a good deal of modernization. No index of the vocabulary has been compiled, but a number of words from a XIX printed edition are cited in RadJoff’s Worterbuch and some of these citations have been included here as xiv Rbğ., followed by a reference to Radloff.

33.    One Arabic-Turkish vocabulary, the Hilyatu -inşân wa halbatu l-lisân of Camâlu’l-din ibnu’l-Muhannâ (see Studies, p. 193) is probably of about the same date as Rbğ. It contains an extensive list of Arabic words with the Turkish equivalents. The autho^ in his preface says that they are in two languages ‘those of our country and of Türkistan’, and enumerates the phonetic differences between the two. The facts given suggest that the first is an early form of Azer-bayjani and the second a late form of Xak., but, except in a minimum number of cases, it is impossible to decide whether any given word belongs to one of these languages or both. Very probably the great bulk of them are early Azerbay-jani, but as a matter of convenience it is easier to cite this work as Xiv Muh. immediately after Rbğ. followed by the page and line of Melioranski’s edition (Mel.) and the page of Kilisli Rif'at’s edition (Rif.)', the two editions are not identical and it is likely that the latter, which is based on a single manuscript, contains a good many words added to the author’s text by later copyists.

34.    Finally there are many texts in Çağatay, the earliest perhaps late xiv, the great bulk xv or even a little later. The language still survives in a later form as modern Uzbek, and no attempt has ever been made to define the latest date at which Çağ. proper was still in use. Apart from a very extensive literature, three major Çağ.-Persian and Çağ.-Old Osmanli dictionaries have been published which, although they all profess to be primarily dictionaries of the language used by Mır 'Ali Şır Nawa'I, probably give a fairly exhaustive list of the Çağ. vocabulary, excluding the Arabic and Persian, but not the Mongolian, words which it includes. These are, in order of age:

(1)    Bada i'll’l-luğat written in Herat by Tali'-i Harawi during the reign of the Timurid Sultan Husayn(A.D. 1438-1506), a Çağ.-Persian dictionary of which the only(?) surviving manuscript, dated a.h. 1117 (a.d. 1705-6) was published in facsimile with a comprehensive index by A. K. Borovkov, Moscow, 1961.

(2)    The anonymous Çağ.-Old Osmanli dictionary commonly known (after the first entry in it) as the Abuşka, compiled during the first half of xvi and published by V. de Veliaminof-Zernof as Dictionnaire djagatai-turc, St. Petersburg, 1869.

(3)    The Sanglax, a Çağ.-Persian dictionary written by Muhammad Mahdi Xan and finished in A.H. 1172 or 1173 (roughly a.d. 1759). A facsimile of one of the best manuscripts was published with an introduction and comprehensive indices by myself as E. J. W. Gibb Memorial, New Series XX, London, i960.

Of these the Sanglax, although the latest, is much the most extensive and best, the author correcting a good many mistakes made in previous dictionaries, including the two mentioned above and others which have not survived, I have
xxii
cited the first occasionally as Bati, followed by the page, the second more frequently as Vel. followed by the page and the last, almost comprehensively, as San. followed by the page and line of the facsimile.

35.    Exaggerated importance has been attached by some scholars to Şayx Sulaymaıı Iiuxari, Luğat-i Çağatay via Turki 'Otmâni, Istanbul, a.h. 1298 (a.d. 1881); it contains very little original matter, apart from the mistakes which abound in it (see San., p. 31) but I have occasionally cited it as ŞS. followed by the page.

36.    There is in Bokhara an important manuscript of the Muqaddimatu'I-Adab of Mahmüdu’l-Zamaxşarî, a classified Arabic glossary compiled early in XII with an interlinear translation in Turkish (as well as in Persian and Mongolian). The manuscript is dated a.d. 1495 and there is no prima facie reason for supposing that the Turkish translation is much older. The language is therefore likely to be Çağ. and this is confirmed by an examination of the Turkish equivalents in N. Poppe’s index to the Mongolian translation in Mongol'skii slovar' Mukaddi-mat al-Adab, Moscow, 1938. Prof. Borovkov, when he died in 1963, had almost completed an index of the Turkish words in this manuscript, but it is uncertain whether it will now be completed and published. The Turkish entries in Prof. Poppe’s work are not in a convenient form for use in preparing a Turkish dictionary, and are not, according to Prof. Borovkov, wholly accurate. I have not, therefore, quoted this work except occasionally (as Zattı.).

37.    The history of the northern stream of literary languages is a great deal more complicated. There was undoubtedly in xn a second literary focus further north than the area described in para. 28, centred perhaps in the court of the Xwa-razmşâhs who first became important early in xn. The most convenient name for texts belonging to this stream is that recently adopted in such works as the Fundamenta, Xwarazmian (Xwar.), although the literary language which evolved there must also have been in use over a much wider area extending into southern Russia. The spoken languages in this area at this period were markedly different from Xak.; the Turks in Xwarazm and the Aral Sea area were Oğuz and those in southern Russia Kıpçak and already in xi Kâşğarî regarded Oğuz and Kip. as separate languages fairly close to one another but different from Xak. Thus; so far as literary works are concerned, it is safe to assume that Xak. was very quickly modified to bring it closer to the local languages.

38.    The oldest text from this area, if its colophon is to be relied on, was in fact composed not in Xwarazm but in the Crimea. It is the Qişşa-i Yûsuf written by a certain 'Alî, of whom nothing further is known, in A.H. 630 (a.d. 1233). As usual the manuscripts are a good deal later and have been more or less modernized. D. Brockelmann listed a number of words from this work in Alt’s Qissa-i Yusuf, der alteste Vorlâufer der osmanischen Literatür (AKPAW, 1917). It has long been recognized that the language is not Osmanli and I have cited the words quoted in
xxiii
this paper as Xwar. xiii 'Ali followed by the page, but it may well contain more Oğuz elements than most other Xwar. texts.

39.    Another text which can be tentatively identified as coming from the Aral Sea area is the enigmatic text sometimes called Oğuz Nâme (see Studies, pp. 48, etc.) written in the Mongolian Official Alphabet and edited by W. Bang and G. R. Rachmati, Die Leğende von Oghuz Qaghan (SPAW, 1932). It contains several Mongolian loan-words and so cannot be earlier than xm, but can hardly Be much later. Considering its subject-matter, the language is likely to be much more specifically Oğuz than most other contemporary texts, but the ambiguity of the alphabet used makes it impossible to determine whether it has such a specifically Oğuz trait as initial d-. It is cited as xm(?) Oğ., followed by the line in the edition mentioned.

40.    Next in order of time come five texts which are all more or less solidly dated. The oldest is the Mu'mu' l-murid written in Xwarazm in a.d. 1313 (see Eckmann, op. cit., p. 115), but I have not been in a position to cite any words from it.

41.    The next is the Xusraw u Şîrîn, written by a certain Qutb at the court of Tini Beg Xan of the Golden Horde in a.d. 1341-2. It has been admirably edited by A. Zajqczkowski, Najstarsza wersja turecka husrâv u Sirin, 3 vols., Warsaw, 1958-61, with a facsimile of the only manuscript, a transcription, and a full index excluding the Arabic and Persian loan-words. It is cited as xiv Qutb followed by the page in Zajqczkowski’s index.

42.    The third is the Muhabbat Nâma of Xwarazmi. Two manuscripts of this poem have survived, the second containing a number of verses which did not form part of the original text; see my paper on this poem in CAJ VII. 4, 1962. Unfortunately verses 437 and 440 which state that the poem was composed ‘on the banks of the Sir Darya in a.h. 754 (a.d. 1353)’, are among the interpolated verses but the date must be approximately correct, since it falls within the reign of Cani Beg Xan of the Golden Horde, during whose reign the poem was composed. It is cited from the recent editions (see Studies, p. 48) as xiv MN followed by the number of the verse.

43.    Next there is the Nafıcu'l-Farâdis written by Mahmud ibn 'Ali in Xwarazm, or perhaps Saray on the Volga, sometime before a.d. 1358, the date of the earliest manuscript, but not necessarily much before, since there is fairly good evidence that the author did not die until a.d. 1360. A reproduction of one of the best manuscripts was published in J. Eckmann, Nehcii’l-Feradis I, Ankara, 1956, but unfortunately his index is not yet published.

44.    Finally there is the free translation into Turkish of Sa'di’s Gülistan. Only öne manuscript of this work has survived, and a facsimile of it, with a preface by
xxiv
Prof. F. N. Uzluk, was published by the Türk Dil Kurumu in 1954 under the title Seyfi Serâyî Gülistan Tercümesi, The translator was a native of Sarai on the Volga, but made his translation at the Mamluk court in Egypt in a.d. i391.

45.    Quite recently there has appeared the first half of an admirable dictionary by E. Fazylov of Qutb, MN, the Nahcu’l-Farüdis, and the Gülistan, with extensive quotations and references, under the title Starouzbekskii yazik. Khorezmiiskie pamyatniki XIV veka, Tashkent, 1966. I have used this work extensively to provide references to the Nahcul-Faradis, under the title Nahc. followed by the page and line in Eckmann’s facsimile for those words contained in this part of the dictionary; for the rest of the vocabulary I have had to rely on a very inadequate list of words published by Kivâmettiıı in I'M IV. I have not thought it necessary to add references to the Gülistan (Gul.) except very occasionally, since its vocabulary is practically identical with that of the other works mentioned.

46.    Apart from the texts mentioned above a few other Xwar. texts have survived and one or two short ones have been published, at any rate in facsimile, but are

not cited herein.

47.    Seven vocabularies of medieval Kip. have been published, some of them containing some Türkmen (Tkm.) words. The most important, and the only one which is purely Kip., is the Codex Cumanicus, a handbook of the Koman (Korn.) language, a Kip. dialect, in Latin, compiled early in xiv (see Studies, p. 48 and Fundamenta). The text is not entirely homogeneous; the earlier part was compiled for Italian merchants, but it also contains later additions by German missionaries. An index to it was published in K. Grenbech, Komanisches IVörterbuch, Copenhagen, 1942, and is cited as xiv Kont. CCI (for the Italian) and CCG (for the German part); Gr. sometimes followed by the page.

48.    The remaining six are all in Arabic, one a list of Turkish words in alphabetical order with Arabic equivalents, the others Arabic handbooks with Turkish equivalents. All these vocabularies have a specified or inferable Egyptian Mamluk background, and it seems clear that basically they are handbooks of the languages spoken by Turkish slaves brought to Egypt from southern Russia, some of them spoke Kip. and others Tkm., an Oğuz dialect. For example, Hou. (see para. 49) seems from internal evidence to have been compiled from oral information collected in the Mamluk possessions in Syria. In some cases an Arabic word is translated by different Kip. and Tkm. words, in others individual words are described as Kip. or Tkm., but in the great majority of cases no such indication is given. The presumption in nearly all cases is that the word is Kip., but there are one or two passages in Id. (see para. 50) which seem to imply the contrary.

49.    Until almost today it was generally accepted that the oldest of these vocabularies was the anonymous Arabic-Turkish vocabulary in Leiden manuscript
xxvv
517, Warner, published in M. T. Houtsma, Ein türkisch-arabisches Glossar, I>eiden, 1894, cited herein as xiii Hou. followed by the page and line of the Arabic text. But as this book was actually going to press an article by Barbara Flemming (Hamburg) was published in Tier Islam, Band 44, Berlin, June 1968, pp. 226 ff., in which she announced that Houtsma, who purported to publish the exact text of this unique manuscript had converted the date in the colophon into the figures 643, whereas the facsimile published in her article shows quite clearly the words talâta v>a arba'in wa sab'umi'ya that is (Şa'bân) a.h. 743 equivalent to (January) A.D. 1343. Instead, therefore, of dating back to xiii and being the oldest of these vocabularies, it goes b^ek only to the middle of xiv and comes chronologically between the two vocabularies mentioned in the following paragraphs. This announcement unfortunately appeared too late for it to be possible to correct the numerous references to Hou. in the dictionary.

50.    There are two other xiv vocabularies. The Kitâbu'1-idrâk li-lisâni'1-atrâk, the only list of Turkish words with Arabic equivalents, was written in Egypt, probably in a.d. 1313, by Atiru’l-din Abü Hayyân Muhammad ibn Yusuf, originally a native of Granada. A critical edition was published by A. Caferoğlu, Abu Hayyân, Kitâb al-ldrdk li-lisdn al-Atrâk, Istanbul, 1931. It is cited as xiv Id. followed by the page in the Arabic text.

51.    The Kitâb Bulğatı l-muştöq fi luğati’l-turk wa'l-qifçâq was written by Camâlu’l-din Abü Muhammad 'Abdullah al-Turki probably in the middle of xiv and perhaps also in the Mamluk dominions in Syria. The only manuscript, which is not quite complete, was published with a comprehensive index by A. Zajqczkowski, Manuel arabe de la langue des Turks et des Kiptchaks, Warsaw, I, 1938; II (title in Polish), 1954. It is cited as xiv Bui. followed by the page and line of the printed text in I and the page of the manuscript in II (verbs).

52.    There are two XV vocabularies, both surviving in unique manuscripts and dating from early in the century, but it is uncertain which is the older. Al--tuhfatul-sdkiyafiluğati'l-turhiya, an Arabic-Turkish vocabulary in alphabetical order, mainly Kip. with a number of Tkm. words, was written almost certainly in Egypt and before a.d. 1426 (the date of a note on the first page). It was edited with a facsimile, translation, and index in B. Atalay, Ettuhfet-iiz-zekiyye fil--lûgat-it-türkiyye, Istanbul, 1945. It is cited as Ttih. followed by the page and line of the facsimile.

53.    Al-qawaninti'l-kulliya fi-d.abti'1-luğati'l-turhiya was compiled by an unknown author in Cairo by cross-questioning Turks living there. The book is undated, but a reference in it to Tamerlane makes it certain that it was written in xv, and probably early in the century. The text was published by Köprülüzade Mehmed Fuad (Prof. F. Köprülü), Istanbul, 1928. It is cited as Kav. followed by the page and line.
xxvi

54.    Quite recently Prof. Zajqczkowski published an account of a short Arabic-Kip. vocabulary entitled al-Durratu l-mudi’a fi luğati'l-turkiya discovered in Florence, with an extract from the text and an index to the extract, in Rocznik Orientalistyczny XXIX, Pt. i, pp. 39 ff. The work is anonymous and undated but clearly belongs to the first half of xv. It adds little to our knowledge of Kip. but I have quoted a few words from it as xv Dur. followed by the page.

55.    The Oğuz language which is conveniently, but not quite accurately, called ‘Old Osmanli’ (Osm.), since the earliest texts in it date from before the foundation of the Ottoman Empire, is recorded from mid xm, but the xiii material is scanty. The earliest is a few verses in the works of Mawlana Calalu’l-din al-Rumi (a.d. 1207/8-1273/4) and a few verses in those of his son commonly called Sultan Vcled. These verses might more properly be classified as Salcuk, but as they are the first stage of a continuous literary tradition it seems more convenient to classify these very early texts and other slightly later texts which are not strictly Osm., like the Kitâb Dede Korkut, which seems to be in a Tkm. language, probably xv., as Osm. so as to avoid an excessively complicated terminology. A good many early works of this kind have been published with partial or complete indices, but it did not seem to me necessary for the purposes of this dictionary to do more than cite under the heading xiv ff. Osm. (since the xiii material is so inconsiderable) those of the words contained in this dictionary which reappear in the Türk Dil Kurumu publication Tamklariyle Tarama Sözlüğü, 4 vols., Istanbul, 1943-5; Ankara, 1953-7 (TTS I-IV) or as ‘Rumi’ words in San. A new edition of TTS is now being published, but is not yet sufficiently far advanced to make citation from it very convenient.,

THE MODERN LANGUAGES

56.    It would not be possible to include in this book all the evidence which could be assembled regarding the survival in modern languages of the early words listed herein without a much greater delay in its publication and a much greater increase in its bulk than would be regarded as advantageous. In any event it is sufficiently obvious to all students of the Turkish languages that common words like at ‘a horse’ and al- ‘to take’ still survive almost everywhere to make a detailed proof of the point unnecessary. Nor is it very important, when a basic word is known to survive in some or all modern languages, to know whether its various derivatives survive also. On the other hand, for the reasons stated in para. 8, the subject cannot be entirely neglected, and it is important to include at any rate some references to modern languages. This immediately raises the question how modern languages should be classified and grouped. A good many systems of classification have already been suggested; there are two quite different ones in Fundamenta; but for present purposes it is hardly necessary to use anything more elaborate than a simple geographical system. Even this is open to some objections, since there are no clear-cut geographical lines between the modern descendants of the early languages, and some modern languages
xxvii
descended from different ancient languages have in the course of time developed characteristics which bring them closer to one another than to languages with which they are genetically related. After a good deal of reflection I finally decided upon a sixfold division: north-eastern (NE), south-eastern (SE), north-central (NC), south-central (SC), north-western (NW), and south-western (SW). Of these the NW languages are, genetically speaking, Kip. and the SW Oğuz; the SC and probably SE are, broadly speaking, descended from Xak. but by different lines of descent.

57.    In the NE group a^e included the languages spoken, but until recently not written, in eastern Siberia and adjacent areas. Yakut, however, which has been isolated from the rest so long that it has acquired very special characteristics of its own, is seldom cited, the authority used being E. K. Pekarsky, Slovar' yakutskogo yazyka, St. Petersburg, 1907-30, cited as Pek. The most important authority for the vocabulary of these languages is W. RadJoff, Versuch eines Worterbuches der tiirk-Dialecte (Opyt slovarya tyurkskikh narechii), 4 vols., St. Petersburg, 1888-1911, cited as R followed by the volume in roman, and the column in arabic numerals, individual languages being mentioned in the abbreviated forms used in that work with the minor alterations of spelling shown in the List of Abbreviations. I have also used the Khakassko-russkii slovar’, edited by N. A. Baskakov, Moscow, 1953, cited as KJiak. sometimes followed by the page, the Tuvinsko-russkii slovar', edited by A. A. Pal'mbakh, Moscow, 1955, cited as Tuv. sometimes followed by the page, and occasionally the modern Russian-Khakas, Russian-Tuvan, and Russian-Altay dictionaries.

58.    In the SE group are included the Turkish languages and dialects of Chinese Türkistan and adjacent areas, which have traditionally been called Eastern Türki and are now called Neo-Uyğur by Soviet and Chinese scholars. These fall into two main groups, the literary language written in Arabic script and the spoken dialects. I have called both Tiirki. For the first, citations are from R. B. Shaw, A Sketch of the Turki Language, Calcutta, 1878, cited as Shaw followed by the page, and Burhan Şehidi, Uyğurça-Xenuça-Rusça Luğat (Uigursko-kitaisko-russkii slovar’), Pekin, 1953, cited as BŞ followed by the page. A good many words in one dialect, Tarançı, spoken in southern Siberia are included in R., and are cited as Tar. followed by a reference to R. For the rest, mainly dialects spoken in southern Sinkiang, I have used G. Jarring, An Eastern Turki-English Dialect Dictionary, Lund, 1964, cited as Jarring followed by the page. One language in this area is in a class by itself, that of the Sarığ Yuğur in Kansu, the only language which can reasonably be regarded as directly descended from early Uyğ. It is recorded in S. Ye. Malov, Yazyk zheltykh uigurov, Alma Ata, 1957, cited occasionally as Sarığ Yuğ. followed by the page.

59.    In the NC group are included Kırğız (Kir.) and Kazax (Kzx.), called in R Kara Kırğız and Kırğız respectively. Historically Kir. belongs to the same family as the NE languages and Kzx. seems to be a Kip. dialect, but the peoples talking
xxviii
them have lived in close propinquity to one another for so long that they now have many common characteristics. The authorities which I have used principally arc, for Kır., K. K. Yudakhin’s Kirgizsko-russkii slovar', Moscow, 1965, cited as Yud. followed by the page, and for Kzx. Kh. Makhmudov and G. Musa-baev, Kazakhsko-russkii slovar', Alma Ata, 1954, and IÎ. N. Shnitnikov’s Kazakh-English Dictionary, The Hague 1966, cited as MM and Shnit. respectively followed by the page. I have also occasionally used the equivalent Russian-Kir. and Russian-Kzx. dictionaries, and only occasionally quoted R since the modern authorities seem to be rather fuller and more reliable.

60.    There are several modern SC spoken dialects, but the only literary language is Uzbek (Uzb.) for which I have used A. K. Borovkov, Uzbeksko-russkii slovar', Moscow, 1959, cited as Bor. followed by the page, and less often the Russian-Uzb. dictionary of 1954. The vocabulary in H. Vambery’s Ğagataische Sprachstudien, Leipzig, 1867, is very heterogeneous İn character. The bulk of the words in it are classical Çağ. taken, often with some errors, from various oriental authors; the remainder seem to be colloquial xix Uzb. words collected by the author himself, and occasionally cited as Vam. followed by the page.

61.    The NW group comprises a rather wide range of languages. Those quoted in R include Kazan Tatar (Kaz.) and three Karaim dialects, those of the Crimea, Lutsk, and Troki (Kar., Krm., L., and T.) but he also uses Krm. for the non-Karaim language spoken in the Crimea which is indistinguishable from Osm., so that the only safe indication that a word is Karaim is that it is given in the Hebrew alphabet. The words quoted from these languages are reproduced with the same abbreviations and the reference in R. For Kar. T. I have occasionally used T. Kowalski, Karaimische Texte im Dialekt von Troki, Cracow, 1929, cited as Kmc. followed by the page. For (Kazan) Tatar I have used the Tatarsko-russkii slovar', Moscow, 1966, cited as Kaz. or Tat., and for Karakalpak (Kk.) and Nogay (Nog.) I have used the Karakalpaksko-russkii slovar', Moscow, 1958 and the Nogaisko-rttsskii slovar', Moscow, 1963, both edited by N. A. Baskakov and cited by the page, and occasionally the Russian-Kk. and -Nog. dictionaries. For Kumyk (Kurn.) there is at present available only the Russko-kumykskii slovar', Moscow, i960, edited by Z. Z. Bammatov; Bashkir has diverged so far from the rest of the group that I have not taken it into account. No convenient authorities for minor languages like Karaçay and Balkar are at present available. As regards Chuvash (Çuv.) see para. 2.

62.    In the SW group are included only three languages, Azerbavjani (Az.), Osm.    • (including Rep. Turkish), and Tkm. For Az. I have used various authorities published in the Soviet Union, not citing them individually. There are many authorities for Osm., but those which I have generally used are Sir James Redhouse, A Turkish and English Lexicon, in the Constantinople, 1921, reprint, cited as Red. followed by the page; Şayx Sami, Qâmüs-i 1'urkt, 2 vols., Istanbul, A.h. 1318 (a.d. 1900-1) cited as Sami followed by the page; and occasionally the
xxix
a.h. 1306 (a.d. 1888) edition of the Lehce-i Osmaniya, cited as Leh. followed by the page. In noting the survival of words in modern languages I have not included those words which have recently been re-introduced into Rep. Turkish to replace Arabic and Persian loan-words. Some of these are genuine old words which had become obsolete in Osm., others are old loan-words like acun (for a:ju:n, a word borrowed from Sogdian) and çağ ‘period’ (a medieval loan-word from Mong.), but they have not had a continuous history in Osm.

63. As pointed out in paras. 2 and 12 it is often important in tracing the history of a Turkish word to kr»6w whether it became a loan-word in Mongolian and, if so, when, since this often throws light not only on the age of a word but also on its original pronunciation and meaning. The question of such loan-words is discussed at length in Studies, Chap. 11, and it is there suggested that these words were borrowed in three main periods, the first, probably v and vi, during which words were borrowed by a Mongolian-speaking people, possibly the Kitan, from an I/r Turkish-speaking tribe, possibly the Tavğaç; the second, probably between viii and xn; and the third during and after the reign of Chinggis. The earliest substantial remains of Mongolian are not, however, earlier than xm. In citing Mongolian words borrowed from Turkish I have normally used three authorities:

(1)    E. Haenisch, Worterbuch zu Manghol un Ni'uca Tobca’an, Leipzig, 1939, which contains a list of the words in the well-known Mong. xiii Secret History (SH), cited as Haenisch followed by the page.

(2)    The standard, though now somewhat antiquated, dictionary of Classical Mongolian, J. E. Kowalewski, Dictionnaire mongol-russe-français, Kazan, 1841 ff., cited as Kow. followed by the page.

(3)    A good dictionary of modern written Mongolian, The Mongol-English Practical Dictionary, compiled by M. Haltod and others and published by the Evangelican Alliance Mission (in the U.S.A.) 1949-53, cited as Haltod followed by the page.

OTHER MODERN AUTHORITIES

64. In addition to the modern authorities quoted above use has been made of the following other modern authorities:

(1)    A. Caferoğlu, Uygur Sözlüğü, 3 parts, Istanbul, 1934-8, a useful list of the words contained in the indices to the texts enumerated in paras. 20 (2) (e) and 21 above and one or two others, cited as Caf.

(2)    v. G., ATG, see para. 20 (2) (e).

(3)    G. Doerfer, Türkische und mongolische Elemente im Neupersischen; I Mongolische Elemente, Wiesbaden, 1963; II Tiirkische Elemente, alif bis tâ, 1965; III ditto jim bis kâf 1967, cited as Doerfer, followed by the volume and the number of the word.
xxx

(4)    C. Brockelmann, Osttiirkische Grammatik der islamischen Litteratwr-sprachen Mittelasiens, Leiden, 1954, cited as Brockelmann followed by the number of the paragraph.

(5)    A. M. Shcherbak, ‘Nazvaniya domashnikh i dikikh zhivotnykh v tyurk-skikh yazykakh’ (‘The names of domesticated and wild animals in the Turkish languages’), one of several articles in Istoricheskoe razvitie leksiki tyurkskikh yazykov, Moscow, 1961, cited as Shcherbak, followed by the page.

(6)    The reproductions of MS. No. S. J. Kr. 4638, Packet 8 in the manuscript collection of the Leningrad branch of the Institute of the Peoples of Asia, recently published in D. I. Tikhonov, Khozyaistvo i obshchestvennyi stroi uigurskogogosudarstva X—X1V vv., Leningrad, 1966, and again with transcription and translation by E. R. Tenishev in an article called ‘Khozyaistvennye zapisi na drevneuigurskom yazyke’ in Issledovartiya po grammatike i leksike tyurkskikh yazykov, Tashkent, 1966. The manuscript is a late Uyğ. family archive, and is cited as Fam. Arch, followed by the line.

THE ARRANGEMENT OF INDIVIDUAL ENTRIES

65.    In paragraphs 4, 5, and 8 some account has been given of the kind of information which is supplied about individual words. There is not a single word regarding which the full range of information is available, and it may therefore be useful to set out here what the full range is and how the various items are arranged. The word may be preceded by a code letter (see paragraph 18). If it is a hapax legomenon (see paragraph 7) it is immediately followed by ‘Hap. leg.’. From this point the order varies slightly. It is sometimes convenient to enter next the original meaning possibly with some remarks on later developments, but this is superfluous if the word is Hap. leg., since the quotation will supply it, and if the word is derived it is more logical to record the derivation before the meaning. After this it may be desirable to add some other remarks. If the fact that the word is a loan-word in Mongolian is relevant, this is mentioned next. Information regarding the survival of the word then follows. It cannot be given earlier since there are many instances of words surviving in modern languages, usually NE, not because they have existed continuously in those languages but because they have been re-borrowed from Mongolian, perhaps quite recently. Finally, cross-references may be given to other words etymologically related or other words with the same or a similar meaning. The quotations then follow in the following order: Türkü viii; viii ff.; viii ff. Man.; viii ff. Yen.: Uyğ. viii; ix; viii ff. Chr.; Man.-A; Man.; Bud.; Civ.; Xiv Chin.-Uyğ. Diet.: O. Kir. ix ff.: Xak. xi (including KB); xn(?) KBVP; xm(?) KBPP; At.; Tef.; xiv Rbğ., Muh.: Arğu, Çigll, Gancak xi: Çağ. xvff.: Oğuz xi: Xwar. xiii Ali; xiii(?) Oğ.; xiv Qutb, MN, Nahc.: Kıp. xi: Kom. xiv CCI, CCG; Gr.: Kip./Tkm. xiii Hou.; xiv Id., Bui.; xv Tuh., Kav., Dur.: Osm. xiv ff.

66.    In conclusion may I very humbly submit that this book contains a vast nnmKpr nf nnotations. translations, and references, and that it is hard to believe
xxxi
that in such a large flock there are no black sheep? May I plead, as the father of Turkish lexicography did nine centuries ago, that yazma:s atım yağmur yai)ilma:s bilge: yai)ku: ‘the only shot that never misses is the rain, the only scholar who never makes a mistake is the echo’ Kaş. Ill 379, 20, and present in advance my apologies for these and other shortcomings.

It would not be proper for me to end this preface without expressing my sincere gratitude to the British Academy for their generous contribution towards the cost of producing this volume.

August 1968
Gerard Clauson