<<

THE EARLIEST TURKISH LOAN IN MONGOLIAN Author(s): GERARD CLAUSON Source: Central Asiatic Journal, Vol. 4, No. 3 (1959), pp. 174-187 Published by: Harrassowitz Verlag Stable URL: http://www.jstor.org/stable/41926439 . Accessed: 15/10/2014 13:15

Your use of JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

Harrassowitz Verlag is collaborating with JSTOR to digitize, preserve and extend access to Central Asiatic Journal.

http://www.jstor.org

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIEST TURKISH LOAN WORDS IN MONGOLIAN

by Sir GERARD CLAUSON London

In an articlein theCentral Asiatic Journal , II, 3, entitled"The Case against theAltaic Theory", statedthe reasons for which I do notbelieve that the Turkishand Mongolianlanguages are geneticlyrelated, and in an in Studia Altaica, the Festschriftfor Prof. Poppe publishedin the Ural- AltaischeBibliothek (1957), entitled"The TurkishY and relatedsounds", I producedevidence which shows that the Mongols or theirancestors at three differentperiods borrowedwords from three differentTurkish dialects (or languages1),one of whichwas older than any whichnow survive. In the presentarticle, a summaryof whichwas read at the 24thInter- national Congress of Orientalistsin Munisch in September,1957, I proposeto carrythe subjecta stagefurther and to discuss,first how it is possibleto determinein whichof the two languagesa commonto bothis nativeand in whichit is a loan word,secondly what light is thrown on the problemby a studyof the historyand pre-historyof the two peoples, and finallyat what periods and fromwhat dialectsthe earlier Turkish loan words in Mongolian were borrowed. The second part involvesinter alia an examinationof thearchaeology of an area in which littlesystematic work has yetbeen done and of whichno comprehensive account has yetbeen published. This part of the articleis thereforeno more than a preliminaryand incompletestudy of a verycomplicated subjecton whichwe may confidentlylook to the futureto throwmore light. It is commonground that a greatmany words, including even verbs, 1 Thequestion when two later forms of a singlelanguage cease to "dialects"and become"" in theirown right has beendiscussed at greatlength without reachinga satisfactoryconclusion for dealing with doubtful cases. In thepresent article,for the sake of clarityI haveused "" only for "Mongolian" and "Turkish"and "dialect"for any form of thoselanguages, although some of these "dialects"are generally, and quite rightly, recognised as languages in their own right.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 175 occurboth in one or moreTurkish dialects and in one or moreMongolian dialects,either in exactlythe same formor in formsso similarthat their identityis certain. If it is acceptedthat Turkish and Mongolianare des- cendedfrom a common"Altaic" ancestor,this is easilyexplained; but if thisis not accepted,and personallyI do not accept it, thenit mustbe shown that the words concernedare native to one language and loan words in the other,and methodsmust be worked out of determining whichlanguage is thelender and whichthe borrower. It seemsto me that thereare at leastsix methods,one or moreof whichcan be appliedin any individualcase. One method,and veryoften the simplest, is to examinethe word in both languagesand ascertainwhether it is the traditional,perhaps the only, word fora particularconcept in one languageand a moreor less super- fluousduplicate of some other word in the other. When one people dominatesanother, as for examplethe Mongols dominatedthe Tuvans rightdown to the earlypart of the presentcentury, the language of the latterpeople becomes completelysaturated with the vocabularyof the former,words being borrowedright and lefteven when perfectlygood nativewords exist for the same concept. The same thingmay happen when two peoples speakingdifferent languages are in intimatecontact withone another,even withoutany politicalrelationship of superiority and subordination.Let me give a specificexample. There are in 8th centuryTurkish at least two wordsmeaning "to fear,be afraid",kork - and iirk-, as wellas otherwords with related meanings, like eymen- "to be nervous,shy". All thesewords are stillin currentuse in manyTurkish dialects. Neverthelessat thepresent time there is currentin all partsof the Turkish-speakingworld except Azerbaijan and Turkeya word çoçi- "to be afraid",which is the normalword forthis concept in Mongolian. It musthave been borrowedat a comparativelyrecent date, since there is no authorityfor its existencein Turkishprior to the 19thcentury (this does not mean thatit was borrowedquite as late as this,there are manygaps in the record). Converselyeven in theearliest surviving Mongolian dialect, that of the Secret Historycompiled in the middle of the 13th century,there are several pairs of words, one peculiar to Mongolian and seeminglythe normal word for the concept concernedand the other found also in Turkish and there traceable back to the 8th century;for example for "year" hon and also jil (in Turkish cil),2 for "half" diili and also 2 For MongolianI use the standard system of used, for example, by Prof.Poppe, for Turkish the official alphabet of theTurkish Republic with a few

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 176 SIR GERARDCLAUSON jarim (carini) and for "camping ground" nutuyand also ayil (agii). Thus if one of the two languagesuses a word,let us call it "word A", for a particularconcept and the other language uses another word, "word B", forit and also wordA, itis safeto assumethat word A is a loan word in the second language. Secondly,if a word commonto both languagescan be identifiedas a derivedword, for example a deverbalnoun derivedfrom a verb, or a denominalverb formedfrom a noun, and the word fromwhich it is derivedis peculiar to one of the two languages,then it can safelybe assumedthat the derived word is a loan wordin the otherlanguage. Let me give a specificexample. Thereis a Turkishword verbyar- "to split, divide", in some dialects pronouncedcar-; fromit are derivedyarim (carini),a Noun of SingleAction, meaning "a singleaction of dividing, i.. half", and yargu(carģu), a Noun of Actionmeaning "dividing" in a metaphoricalsense, presumably between right and wrong,hence "a legal process or decision". Both the basic verb and its derivativesoccur in Turkishfrom the earliest period. In Mongolianthere is no traceof a verb yar- (Jar-)"to divide", but Jarimand faryuboth occur with the same meaningas in Turkish,and bothmust be loan words,Jarim both for this reasonand forthe reasons already stated (under the first method). The thirdmethod rests on a studyof the way in whichderived words are formed.Derived words are formedin thesame wayin bothlanguages by attachingsuffixes to the basic formsof verbsor nouns. Such may be simple,as forexample the Turkishsuffixes -im and -ģu (see the precedingparagraph), or ,as for example a nominal attachedto a verbalsuffix, used to forma verbalnoun froma denominal verb. Generallyspeaking the suffixes used in the two languagesare quite different;where they are identicalthe explanation seems to be eithersheer coincidence,or the borrowingof the actual suffixitself. In the case of compoundsuffixes the fact that borrowing has takenplace can sometimes be provedby thefact that the constituentparts of the suffixdo not both occurin theborrowing language.8 A good exampleof a purelyMongolian suffixis -langj-leng,used to formNouns of Action or State fromverbs, additionalletters as explainedin the preamble to"The Turkish andrelated sounds". Onecorrection tothe usedin that paper must be made. A studyof the bP*ags-paalphabet has convinced me that the velar sound which occurs in the Secret Historyshould be transliteratedy and not % (). Theborrowing ofconstituent parts of words as wellas wholewords is not unusual in otherlanguages. For example, in English philo-, "fond of", borrowed from Greek, is usedto make compound words like "philoprogenitive", ofwhich the second element is notGreek.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 177 e.g.fobalang "suffering" from joba - "to suffer".A good manywords with thissuffix occur in some modernTurkish dialects, which have been under strongMongolian influence. They are all loan words. Converselythere is a Turkishdesiderative verbal suffix -sa-/-se- attached both to verbsand nouns to formverbs meaning"to wish to", e.g. sev- "to love", hence sevse-"to wishto love" ; suv "water",hence suvsa - "to wishfor water, be thirsty".This verbalsuffix is unknownin Mongolian. From it is formed thenominal suffix -sag/seg "fond of", e.g. in Turkisherseg "(a woman) fondof men,nymphomaniac". A numberof such wordswere borrowed byMongolian from Turkish ; laterthe suffix was foundto be so convenient to expressa conceptwhich could not otherwisebe expressedas simplyin Mongolian that it was attachedto purelyMongolian basic words,e.g. erne"woman", hence emeseg"a ladies' man"; miqa(n) "meat", hence miqasay"fond of meat". Althoughthese words cannot be describedas Turkishloan words,since they do not occurin Turkish,they are certainly not pureMongolian since they contain a foreignelement. The fourthmethod rests on a studyof thephonetic structure of a word commonto the two languages. Turkishbasic wordsare usuallymono- syllablesor dissyllables,and, ifthe latter, usually (not invariably)end in a consonant. A Turkishword of morethan two syllablesis almostinvari- ably a derivedword of whichsyllables after the second are suffixesof one kindor another. In Mongolian,dissyllables ending in vowelsand words of morethan two syllablesare not uncommon,though no doubtmany of the latterare derivedwords. Words like cerge"a row,line" and itavun "a partridge"look like loan wordsin Turkishand in factare Mongolian loan words,borrowed in the 13thor 14thcentury; but structureby itself, unsupportedby otherconsiderations, is not conclusive. Fifthly,chronology by itself may prove almost conclusively that a word commonto bothlanguages is nativeto one and a loan wordin theother. There is a gap of about fivecenturies between the earliestsubstantial Turkishtexts, the so-called"runic" or "Orkhon" inscriptionsof thefirst half of the 8th century,and the earliestsubstantial Mongolian text,the SecretHistory, supposedly compiled in themiddle of the 13th century, and there are many other Turkish texts,some of them very voluminous, representingthe Turkishlanguage as it existedbetween these two dates. From these textsa vocabularycontaining thousands of words can be compiled,representing the Turkishlanguage as it existedin the period beforeits speakerscame intoclose contactwith the Mongols as theresult of theconquests of ChinggisKhan. In thewhole of thislarge vocabulary thereare no wordswhich could be identifiedas Mongolianloan wordsby

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 178 SIR GERARDCLAUSON employingthe methodsenumerated above. From this two conclusions follow. Firstif a wordcommon to the two languagesoccurs in pre-13th centuryTurkish, it is certainlynative Turkish, and in Mongolian a loan word. Secondly if such a word is not traceable in pre-13th century Turkishthere is a strongpresumption that it is nativeMongolian and in Turkisha loan word,but thispresumption, unless supported by the use of one of the precedingmethods, does not amountto absolutecertainty, sincewe cannotbe surethat the pre- 13th century Turkish texts which have survivedcontain every single Turkish word whichthen existed. Sixthlyand finally,a wordcommon to thetwo languagesmay be prov- isionallyclassified as nativeto one languageand a loan wordin theother because of theeconomic or culturalbackground which it implies. This takes me to my second topic, the lightwhich is thrownon the subjectby a studyof the historyand pre-historyof the two peoples. In both cases theirhistory can be tracedback muchfurther than the names by whichthey are now known. The name "Turk", of whichthe earliestknown formseems to have been Tiirkii,was the name of a tribewhich did not become knownto historyuntil the second half of the 6th centuryA.D., but peoples with othernames, who certainlyor almostcertainly spoke a formof Turkish, can be tracedback to a much earlierperiod. All thesepeoples seem to have beenspecificly nomadic steppe-dwellers, ranging over the great plains or steppeswhich are bounded on the northby the southernedge of the forestbelt running across Eurasia fromthe Baltic to thePacific and on the south by the highmountains on the northernedge of Tibet,and extend withpartial interruptions from the mountainsof Shansi,which roughly formedthe boundary of ancientChina, and theirnorthern extension, the Khinganrange, in the east,to thecomplex net-work of mountainsin the west formedby the Pamirs,the Tannu-ola mountainsand the various rangeswhich lies betweenthem. Of thesepeoples mentionmust first be made of the Huns. I imagine that all Turcologistsaccept ProfessorPritsak's contentions in his recent book Die BulgarischenFürstenliste und die Sprache der Protobulgaren (Wiesbaden, 1955), that the Hsiung-nuof the Chinese historieswere, broadlyspeaking, the ancestors of theEuropean Huns, and theEuropean Huns theancestors of the Protobulgars. This does notof course necessari- ly implythat the Hsiung-nu,the Huns and the Protobulgarswere pure and homogeneous racial groups and that no additional Turkish or foreignelements were incorporated in thesepeoples in thecourse of their long wanderings,or even that they necessarilyretained their original

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 179 language,4subject of course to inevitablewear and tear; butthere is a very strongpresumption amounting almost to certaintythat modern Chuvash is a directdescendant of ancientBulgar; Bulgar,in its turn,a laterform of the European Hunnish; and that, in its turn,a later formof the languageof the Hsiung-nu.If so, thatlanguage must have been an early formof Turkish, and thisseems to be confirmedby the scanty and obscure remnantsof it (words, titles and phrases) preservedin the Chinese records. These are so distortedthat it is hard to make anythingof them, but they clearly contain some words which were later Turkish,like teijri"heaven" and the title of the supremeruler, shan-yü, which is almostcertainly yavgu in an earlierform, probably davģu. And so it is a reasonable hypothesisthat the Hsiung-nuwere the ancestorsof theTurks and theirlanguage an earlierform of Turkish;and as a resultwe can use theinformation about theHsiung-nu in theChinese records as a guide to the economic and cultural backgroundof the ancestorsof the 6thcentury Turks as farback as thoserecords will take us. The most importantof theseis Chapter 110 of the Shih Chi, which was compiledat about the end of the 2nd centuryB.C. no doubt partly frommuch earlier material. The earlierparts of the history,which pur- portto go back beyondthe middleof the 2nd millennium,are obviously mythicaland thereis no solid historyin the subsequentnarrative until about themiddle of the3rd century B.C., whenthe Hsiung-nu became an organizedtribe with a rulercalled a shan-yü. But ifthe record is shorton concretehistorical facts, of whichthere may well have been none worth mentioning,it paints an admirablyclear pictureof the kind of people theseancestors of the Turkswere at thisdate and probablya good deal earlier. Theycan be described,in ethnologicalterms, as nomadicsteppe- dwellers,who had passed throughthe first,or "savage", stage of existence,that of food-gatheringand huntingwild animals,and reached the second, or "barbaric", stage of stock-owningand stock-breeding, perhaps with a littleprimitive agriculture as a side line and no doubt importantsurvivals of food-gatheringand hunting.Their habitatin this periodwas thegreat plains defined above and primarilythe eastern part of them; archaeologicalresearch in this area has not advanced sufficiently to makeit possibleto correlatethe facts of history and thoseof archaeolo- gy and thus to show wherethe cradle of this race was, but thereis no evidenceat all to show thatthey came fromanywhere else. Obviouslythere was some humanpopulation in thegreat plains from a 4 It is ofcourse well known that the Danubian Bulgārs lost their own language and adopteda Slavonicone, but the Kama Bulgārs did not.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 180 SIR GERARDCLAUSON

veryearly period,but its characterand extentcannot be known at all clearlyuntil further work has been carriedout. There is, however,one archaeologicalfact which takes us farback intothe 2nd millenniumB.C., perhaps even into the 3rd. The Chinese sockettedbronze celt can be linkedby a seriesof intermediate types with a prototypeoriginating either at somepoint in southernEurope, if the movementwas all eastwards,or moreprobably at some pointrather further East, ifthe movement was in both directions;the distributionmap shows thatit reachedChina by a route along the steppesprobably not veryfar south of the edge of the forestbelt. Othercultural influences, for example the technique of making paintedpottery, seem to have reachedChina bythe same route. Obvious- lythese influences must have movedgradually through a seriesof peoples, and it seemsnot unreasonableto suppose that the last link in the chain beforeChina was a tribeof remoteancestors of the Turks,unless we are to assume the existenceof a people who later completelydisappeared. For a periodwhich may be verytentatively dated to about the turnof the 2nd and 1st millenniawe have rathermore concreteinformation. In his fascinatingbook DrevnyayaIstoriya YuzhnoySibiri (i.e. "The Ancient History of SouthernSiberia"), 2nd edition (, 1951), S. B. Kisilev traces in detail the successivearchaeological periods in SouthernSiberia, that is in thiscontext an area in the forestbelt not far northof the north-westcorner of the greatplains, and points out that down to the end of the Andronovoperiod the human skeletalremains indicatethat the area was inhabitedby "Europeoids",but that in thenext, or Karasuk,period, dated roughlyfrom the 12thto the8th centuries B.C., thereare tracesof immigrationsof "Mongoloids" into thisarea fromthe south-east,and the appearanceof artefactsshowing clearly the influence of Chineseart and culture. "Mongoloid" is, of course, an anthropolo- gical termof artand includesmany races besidesMongols (in factit very seldommeans Mongols proper). It is veryunlikely that representatives of thegreatest Mongoloid race of all, theChinese themselves, reached so far westas earlyas this; remoteancestors of the Turks may well have beenthe middle-menthrough whom these artefacts reached this area. Thereis an interestinglittle piece of linguisticevidence which seems to pointto the presenceof ancestorsof the Turksfar to the westat a very earlyperiod. One, and onlyone, Indo-Europeanpeople (it would prob- ably be saferto say a people talkingan Indo-Europeanlanguage) crossed thegreat mountain barrier at an earlydate and establishedthemselves in the regionof the greatplains, the people commonly,but probablyincor- rectly,known as "Tokharians". In thesecond half of thefirst millennium

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 181

A.D. theywere establishedin the Kucha-Turfanarea and fragmentsof Buddhistscriptures in two dialectsof theirlanguage, "Tokharian-A" or "Agnean" and "Tokharian-B" or "Kuchaean", have been found at Turfanand Kucha respectively.As I pointedout in "The Case againstthe AltaicTheory" it was probablyfrom the Tokharian-B word okso thatthe Turks got theirword for "ox", ökiiz, a word whichthey subsequently passed on to the Mongols in theaberrant form ( h)iiker. If thisis so, then it was probablyfrom the Tokharians that they got not onlythe name, but also the animal itself,and veryprobably the firstrudiments of stock- breedingand animalhusbandry. Indeed further research might show that otherTurkish words with a similarcontext had the same origin. There are no Chineserecords of comparableantiquity relating to the ancestorsof theMongols. The name "Mongol" appearsfairly late in the historyof Mongolian-speakingpeoples, perhaps not much before the 12th centuryA.D., and, as in the case of the Turks,their ancestors must be soughtunder an alias. At one timeor another,and forwholly inadequate reasons,various early peoples have been describedas Mongolian-speak- ers; these include the Hsien-peiof the 3rd centuryA.D., who almost certainlyspoke a formof Turkish,the Hephthalitesof the 5th century A.D., who clearlyspoke an Indo-European(Iranian) language,5and the Avarsof the 6th century, who wereprobably Turkish-speakers. But there do seem to be solid groundsfor supposing that the Kitañ (whose name became Kitayin the 8thor 9thcentury) spoke a formof Mongolian,and the Chineserecords trace their history back to the beginningof the 5th centuryA.D. when theywere in Manchuria,in or not far out of the forestbelt. The generalimpression conveyed by the veryunclear references in the Chineserecords is thatthe earliest ancestors of the Mongols werea forest people centredsomewhere north of theedge of theforest belt between the longitudeof Lake Baikal and the Pacific,and thatneither the Kitañ nor the Mongols became knownto the restof the worlduntil they emerged fromtheir forests and adopted a nomadicform of life. Much thesame is trueof thepeople knowncollectively as Tungus. The researchesof A. P. Okhladnikovand otherSoviet archaeologists seem to have provedthe existence of a continuousseries of ancientcultures on the westernshores of Lake Baikal and thencenorthwards down the Angara valley,of whichthe earliest goes back perhapsto the3rd millennium B.C.

6 See R. Ghirshman,Les Chionites-Hephthalites(Mémoires la délégationarchéo- logiquefrançaise Afghanistan, Tome Xlil) (Cairo,1948).

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 182 SIR GERARDCLAUSON and thelatest merges into the culture of theTungus speakers who stilllive in thatarea.6 If thissummary appreciation of the evidence of history and archaeology is moreor less correct,then during that formative period in humanhistory when, according to the generallyaccepted theory,languages as we understandthe termtoday originated and weredeveloped, the cradlesof theTurkish-, Mongolian- and Tungus-speakingpeoples lay farapart, the firstin thegreat plains west of China,the second in theforest belt between Lake Baikal and thePacific and thethird in theforests west and northof that lake. And so we mightexpect that the primevalTurkish language would be wellprovided with terms used by steppe-dwellers,including the namesof animals native to, or introducedfrom elsewhere into, the steppes, but poorly providedwith names for forestanimals and termsfor the techniquesfor hunting them, while the Mongolian and Tunguslanguages shouldbe poorlyprovided with the former, but rich in thelatter, category of words. Obviouslythis state of affairswill not be fullyreflected in stagesof the languages as late as 8th centuryTurkish and 13thcentury Mongolian. The edge of the forestbelt is not an unclimbablefence, separating two differentworlds as theshore separates land and sea; thereare intermedi- ate areas of wooded steppe,and even withinthe forestthere are fairly large areas in the rivervalleys in whichsuch typicallysteppe forms of economiclife as animal husbandryare not onlypossible, but even enjoy exceptionalamenities. By the 8th centurythe Turks were alreadyfully aware of theseamenities and of the advantagesto the steppe-dwellerof takingrefuge from his enemiesin the mountainforests. Indeed in the introductoryparagraph to hisbrother's memorial tablet7 Bilge Kagan said thathe reckonedthat the ötüken mountainforest was thesafest place for his people to settledown. Converselythe Mongols, and theKitañ before them,had emergedfrom the forestsinto the steppes. But these move- mentsmust all have takenplace long afterthe languages had takenshape, and theirbasic vocabularies,when stripped of loan words,should reflect theconditions of theirprimeval environment. Names of animals are particularlysignificant in mattersof this kind, and the twelve-yearanimal cycle, which was used byboth peoples from a • SeeNarody Sibiri ("The Peoples of Siberia"), edited by Levin and Potapov (Mos- cow,1956), p. 704. ' Orkhoninscription I, South side, 1.8; see H. N. Orkun,Eski Türk Yazitlari, I (Istanbul,1936), p. 26,and various other editions, e.g. A. vonGabain, Alttürkische Grammatik,p. 248.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 183 veryearly period, gives a convenientlist of such names. In thefollowing listthe animalshave been arrangednot in the orderin whichthey occur in the cycle,but in a logical orderwhich will facilitatetheir study; the formsof thenames are theearliest available, that is forTurkish 8th or 9th centuryand forMongolian 13thcentury forms. Turkish Mongolian Tiger bars bars Monkey biçin bičin Dragon lu lu Mouse stçgart quluqana Snake yilan moyay Pig laģzin yaqay Dog it noqay Ox siģir,ud Qi)iiker Sheep koñ qoni Fowl takiģu takiya Hare taviçgan tďulay Horse at morin The firstthree animals are exoticboth to the Siberianforests and to the steppes; for these animals both languages use the same foreignloan words; for "tiger" an Indo-European (?specificlyIranian) word, for "monkey"an Indo-Europeanword, the preciseorigin of whichis uncer- tain (but cf. Persianbüzina ), and for"dragon" the Chineseword lungin the abbreviatedpronunciation which was currentin north-westChina in the 8th century.There can be no reasonable doubt that these words reachedMongolian throughTurkish. The nextfour animals, mouse, snake,pig and dog, are nativeboth to thesteppe and theforests ; for these four the two languages use completely differentwords. The next fouranimals, ox, sheep, fowl and hare are typicallysteppe animals,the first three being among the earliestanimals domesticated by man in the "barbaric" stage. For these Mongolian uses Turkishloan words,three of themshowing phonetic peculiarities which I shall discuss later. It is odd thatfor "ox" Mongolianuses a Turkishword which is not actuallyused in theTurkish version of thecycle. The horsetoo is,prima facie , a typicallysteppe animal, and it is curious thathere the two languagesuse completelydifferent words, in contrastto theIndo-European languages which in theirearlier stages all use thesame wordin various forms,which are oftenused to illustratethe Indo-Euro- pean phoneticlaws. The conclusionseems to be that the Turks and Mongols metthe horseindependently and beforethey met one another. The main conclusionto be derivedfrom this list is thatthe two Ian-

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 184 SIR GERARDCLAUSON guagesevolved in isolationfrom one another(animals 4-7 & 12),but that the Mongols learnttheir "steppe" terminologyfrom the Turksand very likelyalso the rudimentsof animal husbandry(animals 8-1 1). Another word pointingin the same directionis thatfor "calf", Turkishbuzaģu , Mongolianburďu. I now turnto mythird topic, the periods at whichand thedialects from whichthe earlierTurkish loan wordsin Mongolianwere borrowed. The latest period whichI propose to discuss in this connectionis the 13th century;the processof interchangebetween the two languageshas been more or less continuoussince then,but does not presentany particular pointsof interest;it may well have been moreor less continuous,at any rate so far as Mongolian borrowingsfrom Turkish are concerned,for some long periodbefore that date, but a carefulstudy of thewords which in standardTurkish begin withy-9 to whichI devotedmy attentionin "The Turkish Y and related sounds", shows that there were three distinctphases, one in the 13thcentury and two earlier. The lastest,or 13thcentury, phase began whenthe firstattempts were made to convertthe Mongols to Buddhismand Buddhistscriptures in Turkishwere translated into Mongolian. This involvedintroducing into Mongoliana numberof Turkishwords for which Mongolian equivalents could not be found. The Turkishdialect concerned was Uygur,which is in the main line of traditionof standardTurkish, and loan wordslike yilvi"magic" and yirtinçii"universe" which began withy- in Uygurwere borrowedin that form. In this,as in earlierphases, certainphonetic changeswere made in some loan words in orderto bringthem into con- formitywith Mongolian phonetics.In particularfinal (-Š) in Turkish wordswas alteredto -s , as in ulu§"country", which became ulus; and since finalunvoiced and affricatesdid not existin Mongolian a vowel was added at the end of wordswith these finals (see "TurkishY" p. 37, footnote). The intermediatestage ended whenthe third phase began,early in the 13thcentury, and presumablybegan when the Mongols properfirst came into contactwith Turkish-speaking tribes. Well beforethe 13thcentury, certainlyas early as the middle of the 11th and perhaps earlier,the pronunciationin some Turkishdialects had begun to divergefrom the standard. In particularinitial y- had become c-, forexample yil "year" had becomecil. One suchdialect, presumably the ancestor of themodern SouthSiberian dialects ("languages") Xakas, MountainAltai, Tuvan and the rest,seems to have been spoken by a north-easternTurkish tribe in contactwith the Mongols priorto the expansionwhich occurred under

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 185

ChinggizKhan, and extensiveborrowings from it seem to have taken place. In the languageof the SecretHistory, which is antecedentto the thirdphase and containsno Turkishloan wordsbeginning with y -8 about 6% of the words beginningwith / - appear to be Turkishloan words beginningwith y- in standardTurkish, e.g. jil (icilcyil ), Jarim(carim < yarim) and jaryu (carģu y, thefourth the >r and the fifth§>L Words were takenfrom this dialect into Mon- golian more or less unchanged,apart fromthe regularphonetic adjust- mentsmentioned above, exceptthat the voiced dental spirant d, whichdid not existin Mongolian,became plosived. The followingtable gives typicalexamples of Turkishloan-words in Mongolian, which show one or more of these peculiarities,and also parallelexamples in theother group of Turkishdialects referred to above.

MongolianProto-Turkish 8th century Protobulgaretc. Turkish 1. ¿/retaineddayi(n ) < daģi"enemy" > yagi dilan"snake" > yilan < dilom(Protob.) 2. ñ-retained nodurya< ñodruk"fist" > yodruk ñaz"summer" > yaz < nyár(Turkish loanword in Magyar)

8 Onlyone word beginning with y-, yarn "wound", occurs also in Turkish, and as that wordis notfound in Turkishprior to the13th century, it can reasonably,on the principlelaid downin "method5" above,be regardedas a Mongolianloan word.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions 186 SIR GERARDCLAUSON

MongolianProto-Turkish Protobulgaretc.

3. -g->-y- takiya < takigu"fowl" ayil < aģīl"camping ground" dayi(n) seeabove 4. z>r ( h)iiker< ökiiz"ox" (< Tokharian B okso) burďu < buzaģu"calf" > paru(Chuvash) tokuz/tokkuz"nine" > taxarjtaxxar (ditto nyàr(see above) 5. £>/ tďulay < taviçgan"hare" ta$"stone" > t'ul'çul(ditto)

It is surelymore than a merecoincidence that none of thesetraits, except the third,which is of minorimportance and not uncommonin Turkish dialectology,have anyparallel in anyother Turkish dialects except Proto- bulgar(including Protobulgar loan wordsin Magyar)and itsdescendants mediaevalBulgar and modernChuvash, and that in these dialectsthey are all foundtogether. It is at firstsight disconcerting to findthat the only close parallelto the hypotheticalfar-north-eastern language from which these borrowings of theearliest period were made is to be foundin themost westerly of all the Turkishdialects, but we need not reallybe surprisedat findingan early languagesurviving in isolatedpockets far from one another. Exactlythe same thinghas happenedin the case of Basque and the only otherlan- guagesgeneticly related to it,Abkhaz and someother Caucasian language. Moreoverthere is anothercurious link betweenthe far westand the fareast. In the Chuvashlanguage çu- (properlyťu- witha palatalizedt-) regularlyrepresents standard Turkish ta- (as in ta§>çul in the table above); standardTurkish medial -ģ- is frequentlyomitted; and in the moderndialect standard becomes$ as regularlyas in the earlierdialect standard$ became /. In short,although other explanations are theoretic- ally possible, the most reasonable explanationof the modern name Chuvash(T'uvaç) is thatit is a laterform of Tavgaç, the name of a very famousancient Turkish tribe. Admittedlythere is no recordin historyof a partof theTavgaç having accompaniedthe Hsiung-nuin theirmovement to the west,but equally thereis no recordin historyof the Bulgārshaving ever been in the east, fromwhich we know theymust have come. We know thatthe Bulgārs, who came to thewest in a subordinateposition, ultimately imposed their nameon thehorde of which they were part, and thereis nothingimpossible in thetheory that a sectionof Tavgaç also formedpart of thehorde, and

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions THE EARLIESTTURKISH LOAN WORDS IN MONGOLIAN 187 ultimatelyimposed theirname on it, afterthe Bulgārs had lost their importanceas one of the ultimateconsequences of the Mongol invasion of southernRussia. The Tavgaç (in Chinesetranscription To-pa) firstappeared in history in the 4th centuryA.D. and very rapidly became important. They seizedlarge parts of northernChina and governedit fora centuryand a half,the ruling clan rapidlybecoming completely Sinicized and assuming thedynastic title of "NorthernWei" (rulingfrom A.D. 386 to 535). It was preciselyto theNorthern Wei courtthat the Kitañ began to send annual missionswith gifts (that is tribute)in A.D. 440. It seems reasonableto suppose thatthis was the firstsubstantial contact between a Mongolian- speakingand a Turkish-speakingpeople and thatit was fromthe Tavgaç thatthe Kitañ startedborrowing words for concepts for which they had no words of theirown, as primitivepeoples always have done when they have come into contactwith more advanced peoples. Indeed it seems reasonablealso to suppose thatit was duringthis period that the Kitafi, recentlyemerged from their Siberian forests but alreadyfamiliar with the horse,for whichthey had a name of theirown ( morin), firstlearnt the rudimentsof animal husbandryand acquired fromthe Tavgaç not only oxen,calves, sheep and domesticfowls, but also names forthem, just as centuriesearlier the ancestors of theTurks had acquiredoxen and a name forthem from the Tokharians. My suggestiontherefore is thatthe earliest phase of Mongolianborrow- ingsfrom Turkish started in about the fifthcentury A.D. and thatthese earliestloan words wereborrowed by the Kitañ fromthe Tavgaç. It is not oftenthat a date practicallyimposed by considerationsof a purely phoneticnature agrees so neatlywith the recordedfacts of history.My onlyregret, and it is a verysincere one, is thatI do not at presentsee any means of provingpositively that this theory is correct.

This content downloaded from 141.213.236.110 on Wed, 15 Oct 2014 13:15:31 PM All use subject to JSTOR Terms and Conditions