Ethnic and Cultural Diversity by Country

Ethnic and Cultural Diversity by Country Author(s): James D. Fearon Source: Journal of Economic Growth, Vol. 8, No. 2 (Jun., 2003), pp. 195-222 Published by: Springer Stable URL: . Accessed: 24/10/2013 17:24 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . . JSTOR is a 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 support@. .

Springer is collaborating with JSTOR to digitize, preserve and extend access to Journal of Economic Growth.



This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

JbJL

JournaolfEconomicGrowth8,, 195-222,2003

wV ? 2003KluwerAcademicPublisherMs. anufacturiendTheNetherlands.

Ethnicand CulturalDiversitybyCountry*

JAMES D. FEARON

DepartmenotfPoliticaSl cience,StanforUdniversitSyt,anforCdA94305-6044U, SA eFtohrtnhigecireromuppaisrcricoaeslvscaoluunattiroines*se,.*vEetrhanaliccgtrivoerueps,e"haorwcphervoegrirs, aamsilnsipecpoenrcyoomnicceasnptdA.pfotlietaridcdaslrceiessnicnecrgoenquceirpdetautaaalonnd practicaolbstaclesI,presenat listof822ethnicgroupsin 160countrietshatmadeupatleast1 percenotfthe countrpyopulatioinntheearly1990s.I comparae measuroefethnifcractionalizatbioansedonthislistwiththe mostcommonluysedmeasureI. also construcatn indexofculturaflractionalizatitohnatusesthestructural distancbeetweenlanguageass a proxyfortheculturadlistancbeetweengroupsina country. Keywordse:thnicheterogeneictuyl,turadliversitfyr,actionalizatieocno,nomicgrowth JEL classification05: ,D7

1. Introduction Does ethnicdiversitlyowera country'esconomicgrowthrateor levelofpublicgood provisiona,s EasterlyandLevine(1997) andAlesinaet al. (1997) claim?Aremore ethnicalldyividedstatesmorecivilwarprone?Less (or more)likelyto experience democratitcransitionosr stabledemocracyM? orelikelyto havehighlyfractionalized partysystemisftheyaredemocraticW? hatfactordsifferentieatthe nigc rouptshatrebelor havesecessionismt ovemenftsromthosethatdonot?Whatdifferentiattheesethnigcroups whosememberms obilizeinthepoliticaslpherefromthosethatremain"latent?"1

Empiricaelffortsoansweranyofthesequestionrsequirtehatwecollectdataonethnic groupsin differencotuntriesa,ndbeforeanysuchdatais collectedw, e needa listor sampleofethnicgroupsforsomesetofcountries.

Thispaperdiscussesconceptuaalndpracticapl roblemisnvolvedin constructinag cross-nationliasltofethnicgroups(Sections2 and3) andthenpresenttsheresultosfan efforttocarryoutthetask(Section4s -8). Restrictinatgtentiotnogrouptshathadatleast1 percenotfcountrpyopulationinthe1990s,I identif8y22 ethnicand"ethnoreligious" groupisn160countrieHs.ypothetilcyal,myobjectiviestoincludethosegrouptshatwould be listedmostoftenifrandomlcyhosenindividualisnthecountriynquestionwereasked "whatarethemainethnic(orracialorascriptiveg)roupsinthiscountry?I" lackthe resourcetsocarryoutsucha surveyandhavenotdoneso.ForlackofsomethinbgetterI, relyonthesecondarsyourcesandtheexistinlgistsdiscussedbelow.Thelistofferehdere

Presenteadtthe2002AnnuaMl eetingoftheAmericaPnoliticaSlcienceAssociationA,ugust29-Septembe1r, BostonS. omeoftheresearcfhorthispaperwassupportebdygrantfsromtheNationaSlcienceFoundatio(nSES9876477andSES-9876530)andtheCarnegieCorporatioFno. rexcellenrtesearchassistanceI wishtothank ChristinMa aimoneA, lexandeRrosas,AtsukoSuga,andNikolayMarinovI. alsothankDavidLaitinformany helpfuclonversatioanbsouttheprojectw, hichis linkedtoourlargerjointprojectoncivilconflict.

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

196

JAMESD. FEARON

shouldbeviewedas a continuawl orkinprogrestso,beimproveadsmorecountry-specific expertiseo,ractualsurveydata,is broughttobearcasebycase.

Section5 presentdsescriptivsetatisticSs.ection6 proposesa simplewayto usethe data to represenhtow ethnicstructuredsifferacrosscountriesa,s opposedto an aggregatemeasureof ethnicdiversityS.ection7 comparesthe standarddiversity measureconstructefdrommydatawiththecommonluysedmeasurebasedontheAtlas NarodovMira,publishedbySovietethnographeinrs1964.In Section8, 1 usethedata to construcatn indexofculturaflractionalizatitohnatusesa measureofthestructural relationshibpetweenlanguagesto take intoaccountthe culturaldistancebetween groupsin a countryF.orinstanceB, elarusand Cyprushavesomewhastimilarethnic structurebsu,tthegroupsin Belarusareculturallmy uchcloserthanthosein Cyprus. Usingthe structurdalistancebetweenlanguagesas a proxyforextentof cultural differencet,he culturalfractionalizatiomneasureattemptsto take such cultural proximitiyntoaccount.

1.1. RelationtoExistingWork Asnotedt,hebestknownandmostwidelyusedsimilarefforwtascarriedoutbya teamof Sovietethnographeinrtsheearly1960s,andpublisheadsAtlasNarodovMira.Theirlistof "ethnolinguisticg"roupsand populationfigureshas been employedby several generationofspoliticaslcientistsso,ciologistasn, d,morerecentlyec,onomisttsoproduce cross-nationeasltimateosf"ethnicfractionalizationM."o2stempiricasltudiecsoncerning theimplicationosfethnicdiversitysu, chas EasterlyandLevine'sworkon economic growthh,aveemployedthismeasureT. he Sovietteammainlyusedlanguagetodefine groupsb,utsometimeisncludedgroupsthatseemtobe distinguishbedysomenotionof racerathetrhanlanguagea,ndquiteoftenusednationaolrigin(e.g.,Anglo-Canadianarse listedintheUnitedStates).

MorerecentlyT,edR.Gurrandhiscollaboratohrsavedevelopeda listof"minoritieast risk" in 115 countriesa,long witha remarkablaerrayof variablescodinggroup characteristiscist,uationasn, dexperience(Gs urr1, 996).Thisdatasetallowsforthefirst timelarge-Nresearchon thecorrelateosf groupoppressionp,rotesta,nd rebellion. Unfortunatetlhye,selectioncriterifaorthesampl-e thegroupms ustbejudged"at risk" inoneormoreoffourway-s rendeprroblemateifcfortsousethedatatodrawinferences aboutthesephenomenTa.hedifficulitsythesameas thatoftryintgolearntheeffecotf SATscoresonacademicperformanbcyelookingonlyatelitecollegesI.fweconsideornly oppressedor disadvantagegdroups,we are truncatinvgariationon theindependent variablea,ndthusmakingithardertodetecta relationshbipetween(say)discrimination andrebellionT.hissampleselectionprobleminminoritieastrisk(MAR) is oneofthe motivationfsorthepresensttudy.

Alesinaet al. (2002) attemptto distinguisbhetweenethnicl,inguistica,ndreligious groupisna sampleofabout190countrieasn, dthenusetheirliststoconstrumcteasureosf ethnicl,inguistica,ndreligioufsractionalizatiAonlt.houghitis notclearhow"ethnic" and "linguistic"groupsare distinguishe(ads theythemselveasllow),thedescriptive statisticfsortheir"ethnic"measurelook broadlysimilarto thoseforthemeasure

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

ETHNICSTRUCTUREAND CULTURALDIVERSITYAROUNDTHE WORLD

197

constructhedereR. oeder(2002)hasmadeavailablea seriesoffractionalizatmioneasures for1961and1985basedalmostentireloyn Sovietethnographsiocurcesh; ismeasures appeartobeclosertothoseoftheAtlasNarodovMirathanminet, houghI havenotseen thegrouplistthatunderliehsisestimates.

2. CodingEthnicGroups

"Primordialistasr"esaidtobelievethatethnigc roupasreeitherfixedb,iologicallgyiven entitieos,r,iftheyaresocialconventiontsh,attheyaredeeplyrootedc,learlydrawna,nd historicalrlyigidconventions. toAcondyeo' 'newtehitnhipcgrrimouoprsd"iinalmliesaatnnyidnigfssfheoruelncdboteuqnutircikelsyId.t3irsaabpuidsleybdoefctohmemebscylueanrdtehrattoankeing mustmakeall mannerofborderline-arbitdraerciysionsa,ndthatin manycases there simplyis no singlerightanswerto thequestion"Whataretheethnicgroupsin this country?C"onstructivoisrtinstrumentaalirsgtumentasboutthecontingenftu, zzya,nd situationcahl aracteorfethnicitsyeemamplysupported.

Take,forexamplet,heUnitedStatesW. hatareitsethni(corracial)groupsL?etusmake thingms ucheasierbyrestrictinatgtentiotnogroupswithat least1 percenotfcountry population.I4f we consultofficiaclensuscategoriesw, e get three"races"- white, AfricaAnmericana,ndAsian- andanadditionaglroupH, ispanicw, hichthegovernment emphaticaldlyeclaresis "nota race."Is thistherighltistfortheUnitedStates?Whynot disaggregatHe ispanicinto PuertoRican AmericansC, uban AmericansM, exican Americansa,nd so on, or likewiseforAsian?5WhynotdistinguisbhetweenArab AmericanIsr,ishAmericanIst,alianAmericanGs,ermanAmericanasn, dso on?Andwhy shouldwe use thecurrenctensuscategoriesw, henearliercensusesformulatetdhe categorieqsuitedifferent(Nlyobles,2000)?

Lookingfarthearfieldd, oesSomaliahavea singleethnicgroup(Somalis),orseveral, correspondintogthemajorclans?Ifthelattera,t whatlevelin theextremeldyetailed hierarchicsayl stemofclanandsubclando we locatethe"ethnicgroups"forourlist? WhataboutcastesinIndia?WhataboutBerberisnseveraNl orthAfricacnountriews,here a largemajoritoyfthepopulatiocnould,iftheywishedc,laimBerberdescentb,utattitudes varyonwhethetrocharacteriozeneselfas "Arab"or"Berber?"WhataboutmanyLatin Americacnountriews,herethelinesbetween"indigenousa"nd"mestizo,"andbetween "mestizo"and"white,"areoftenvaguetothepointofbeingimperceptibolresituationdependentW. hataboutSudan,whereone mightcode oneselfas a (black,notArab) Southernienronecontextb,utas a Dinkainanother?

Anexplicidt efinitioonf"ethnicgroup"couldhelpwithsomeofthesequestionsb,ut itis importantto see thatit couldnotplausiblysolveall of themF. orexamplea,ny definitionfethnicgroupthatsaidunambiguoustlhyat"Hispanic"is anethnicgroupin theUnitedStatesbut"CubanAmericani"snotisprimfa acieimplausiblTe.henaturoef theconceptof"ethnicgroup"is suchthattherecanbe multiplwe aystospecifytheset ofethnicgroupsin a countryal,l ofwhichincludemore-or-leesqsuallyvalid"ethnic groups."

Thisobservatiohnas an importanimt plicatiofnorsocialscienceresearchthatuses

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

198

JAMESD. FEARON

measuresof ethnicdiversityto explainoutcomessuch as economicgrowthor political violence.If thereare multipleplausiblewaysof listinga country's"ethnicgroups,"we mustbe carefulthatwe do not,in effectc,hoose thecodingthatbestsupportsourtheory, afterthe fact. Somalia was viewed by the Soviet ethnographerisn 1960 as highly homogenous,a nationof ethnicSomalis sharingreligion,language,and customs.This was a perfectlyplausiblecodingthenand itremainsso today.Since thecivil warofthe 1990s,however,analystsseekingto explainpoorprioreconomicgrowthor thewar itself wouldbe drawnto arguethatSomalia is highlyethnicallyfractionalizedalongclan lines,

and thus a good example of the propositionthatethnic heterogeneitycauses poor economicperformancaend civil strifeD. esignatingSomalia as highlyfractionalizedis not implausibleeither,whetherfor 1960 or 1990. Or considerBotswana,a case often

used to supportthe argumentthat"Africa's growthtragedy" is explainedby ethnic heterogeneityW. ithits largeTswana ethnicgroup,Botswanacan be plausiblycoded as homogeneousby Africanstandards,and its economy has performedvery well. Yet Botswana's ethnic structureis fundamentalysimilar to Somalia's- the Tswana are dividedintoeightsubtribesthatare socially and politicallyconsequential.If forsome reasonBotswana's economyhad done poorlyover the last 30 years,and if it had seen significantinternalfightingalong triballines, it would have been viewed ex post as confirmatioonfthe"regularity"thatethnicdiversitymakesforlow growthanda greater riskof civil conflict!

So whatcan be done? Many of theproblematiccases notedabove have a common origin:Whereto locatethe"ethnicgroup" whenthereare twogroups,and groupB is a subsetof groupA? One approachis to avoid a decision,insteadincorporatintgheseset/ subsetrelationsin thestructuroefthedata.Thatis, we mightcode multiple"levels" of ethnicgroups.In theUnitedStates,forexample,thecensuscategorieswouldformlevel 1, a disaggregationbycountryoforiginlevel 2 (Mexican-AmericanV, ietnameseA- merican, etc.), and so on. FollowingScarrittand Mozaffar(1999), I partiallybuild such structure intothegrouplistsforsubSaharanAfrica,wherethisissue is particularlycommonand difficult.

Butformanypurposes,suchas producinga cross-nationaml easureofethnicdiversity, we willwanta singlelistofgroupsfora countryI.t is notevidentthatthe"levels" would correspondacross countries,makingit sensibleto compute"level 1 fractionalization," "level 2 fractionalization,e"tc. Moreover,setsand subsetsare nottheonlyproblemwe encounter.Should Mexico be dividedbetween"indigenous" and "mestizo/white,"or should"white" be brokenout?Or ifwe are listinghyphenatedAmericansfortheUnited States,do we include,say,"GermanAmericans,"evenifthisis at besta vaguecategory ratherthana groupinthesenseofa setofpeople whorecognizeandfeelmotivatedtoact on thebasis of thismembership?

Implicitin the idea of an ethnicgroupis the idea thatmembersand non-members recognizethedistinctionandanticipatethatsignificanatctionsareorcouldbe conditioned on it. So it is naturaland perhapsnecessarythatthe "rightlist" of ethnicgroupsfora countrydependon whatpeople inthecountryidentifyas themostsociallyrelevantethnic groupingsI. adoptthisapproachforthelistdiscussedbelow,inprincipleifnotliterallyin practice.Ideally,thestandardfor"the rightlist" thatI am seekingwouldbe definedbya procedurelike thefollowing:

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

ETHNICSTRUCTUREAND CULTURALDIVERSITYAROUNDTHE WORLD

199

1. Randomlysamplea largenumberof people in thecountry.

2. Ask each ofthemto listthemajoror mainethnicgroupsin thecountry.

3. Show themor read a listof manypossible formulationosf theethnicgroupsin the countrya,nd ask themto say of whichtheyconsiderthemselvesmembers.

4. Repeat(3), askingthemto say of whichgroupson thelistmostotherpeople in the countrywouldconsiderthemto be members.

5. Askthemtotrytorankthegroupstheyidentifieidn(3) accordingtohowstronglythey identifywiththe group (e.g., which is "most importantto you," or some such language).

Sucha surveycouldbe usefulformanyinterestinpgurposesbesidesthatofconstructing a list of ethnicgroupsby country(forwhichI would expect to analyze responsesto question2). It could be used to assess the degreeof social consensuson whatare the country's"ethnicgroups,"whichmightnotbe particularlhyighinmanycases. Iftakenat multiplepointsintime,itcouldbe usedtostudythepoliticaloreconomicdeterminantosf "situationalethnicity,"factorsthatlead people to see this-or-thaetlementof their "identityrepertoire"as moreor less importanatt differenttimes(Laitin, 1998; Posner, forthcomingR).ankingsby importancein question5 could allow a more subtle and nuancedmappingof levels of ethnicidentityand possibilitiesforreformulationand coalitions.The differencebsetweenanswersto questions3 and 4 could allow an inquiry intogaps betweensubjectiveunderstandingasnd objectiveassessmentsofethnicity(e.g., manywhiteAmericansmightidentify"Asian" as a race,buthowoftenwould "Asians" self-identiftyhisway?).

Withoutsurveydata of thissort,we are forcedto reviewexistinglistsand secondary sourcestoapplythisstandardas bestwe can. The mainsourcesemployedaredicussedin Section4.

Beforeproceeding,I stresstwo pointsthatfollowfromtheobservationthatwhatthe ethnicgroupsina countryaredependson whatthepeopleinthecountrythinktheyareata given time.First,it cannotbe assumed,withoutargumentt,hatethnicdistinctionsare whollyexogenous to otherpolitical,economic, and social variables of interest.For example,poor economicperformancceould exacerbatedistributionasltrugglesc,ausing people to see and act along lines of ethnicdivision that were formerlyconsidered unimportantB.y contrast,robusteconomic growthmightlead to the downplayingof ethnicdivisionsand a greateremphasison nationalidentityI.f Botswana seems more ethnicallyhomogeneousthanSomalia does at thispoint,it maybe thatthisis in parta resultratherthana cause ofeconomicgrowthL. ikewise,manyexamples,suchas Somalia, showthatpoliticalviolencecan lead orforcepeopletoidentifymorestronglyalongethnic linesthatformerlywereless salient(Kaufmann,1996; Fearonand Laitin,2000b). This maybe an argumenftorusinga listofethnicgroupsconstructeidn 1960,suchas theAtlas NarodovMira, to studysubsequenteconomicgrowthor politicalconflict.6

Second, we cannotuse the list to ask empiricallywhysome possible ethnicgroups

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

200

JAMESD. FEARON

becomeactualethnicgroupsata giventimeo, rwhyethnicas opposedtootherpolitical cleavagesdevelopW. emighwt anttoknowwhypossibleethnigc roupsuchas GermanAmericanorScots-Irisdho nothavethesamesocialandpoliticaslaliencethatwhiteand blackdo intheUnitedStatesatpresento,rwhyKenyannationaellectoracloalitionasre structurbedydivisionasmongKikuyuL,uo,KalenjinsK, amba,etc.,rathetrhanbetween menandwomeno,rrichandpoor.7Obviouslyif,a criteriofnorinclusioninthelististhat peopleinthecountrsyeethecategoriynquestionas anethnigcroupt,henwedonothavea sampleof all hypotheticalplyossibleethnic(or other)groupsN. onethelessa, listof "actual" orexistingethnicgroupswouldbe a prerequisitfeorsucha studyT. hetrick wouldbeconstructitnhgelistof"potential"ethni(corotherg) roupsS.inceitisnotclear thatthepopulatioonf"all possibleethnicgroupsina countryi"s well-defineedv,enin theorys,omesortofcase-contraopl proachwouldbe necessary.8

3. Ethnicity

I arguedinSection2 thatnoplausibledefinitioonf"ethnicgroup"willbyitselfimplya uniquelistof groupsfora countryS.till,a definitiownouldbe usefulto boundthe phenomenowne aretryintgocapturea,ndtoaddressquestionlsikethefollowingA.re Protestantans d Catholicsin NortherInrelando, r BosnianSerbsand Muslimst, o be includediftheonlysignificancutlturadlifferenicsereligion?

Standardefinitioonfs"ethnicgroup"intermosfa sharedbeliefofcommonancestry and/osrharedculturafleatureasreproblemat(iFcearonandLaitin,2000a).It is almost alwayspossibletogiveexampleosfgrouptshatfithedefinitiotnakenliterallybu, tthatare notintuitivel"yethnic,"or ofgroupsthatdo notfitthedefinitiobnutthatareoften describedas "ethnic."

FearonandLaitin(2000a)attempttodealwiththisproblembyexaminintgheimplicit rulesthatpeople(oratleastEnglishspeakersu) setodecidewhichgroupasre"ethnic"in everydatyalkT. heyarguethatincommonspeecha groupmaybedesignateads "ethnic" ifthegroupislargetrhana familyandmembershiinpthegroupisreckonepdrimariblyya descentrule.Thesearethecorecriteria,lthougthheconceptmaybe furthrerestrictetdo ruleoutcasessuchcastesornobilityan, dgrouptshatare"legislated"intoexistenc(ei.e., haveno"naturalizehdistorya"s a group)I.tis worthnotingthatsharedculturafleatures eseveemrytdoaptylaalykFn. oornexecamesspalre',yr'Joelewisn"awrheoefttheendaregscrroiubpecadasnabneetdhenscigcrriboeudapdse"septihtelnacick"iinng a commonlanguageu, niversallsyharedcustomso,r evencommonreligiouspractice (sincenon-believerasre typicallyincludedin thegroup,and it is contestedwhether conversiocnan makeone ethnicallJyewish)S.omaliclansare frequentrlyeferretdo as "ethnic"formationesv,enthoughtheirmemberdsonotseetheclansas culturalldyistinct inanysignificanretspect. coTmhmeroenrseullitgosifotnwhielolbredcionnarsliyadnegreu"daegteahnnailcy"-sinsaalsmoheleylwp,ehxepnmlaienmwbheerngshrioinputphseghraroiunpiags reckonepdrimarilbyydescentrathetrhanbypublicconfessioonffaithI.n Bosnia,thugs distinguishebdetweenSerbsandMuslimson thebasisoflocal knowledgaendrecords concernindgescentn,otbytestsofreligioufsaithorpracticeIn. theUnitedStateso,necan

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

ETHNICSTRUCTUREAND CULTURALDIVERSITYAROUNDTHE WORLD

20 1

make oneselfProtestanotr Catholicby adoptingtheappropriatereligiouspracticesand beliefs,somethingthatis hardor impossibleto do in NorthernIreland.

Anotherapproach to definitio-n in several ways more useful for the purpose of constructinaglistbycountrie-s is toemploytheidea of"radial categories"advancedby linguistsand cognitivescientists(Lakoff,1987; see Collier and Mahon, 1993, for a discussionwithrespectto political science). In practice,people may understandthe meaningofa conceptX byreferencteoprototypicaclases. Less prototypicaclases maynot shareall thefeaturesofa prototypea,ndyetstillbe validlyclassed as Xs, at leastin some circumstances.

Forexample,theprototypicaelthnicgrouphas thefollowingfeatures:

1. Membershipinthegroupis reckonedprimarilybydescentbybothmembersandnonmembers.

2. Members are conscious of group membershipand view it as normativelyand psychologicallyimportantto them.

3. Membersshare some distinguishingculturalfeatures,such as commonlanguage, religion,and customs.

4. These culturalfeaturesareheldtobe valuablebya largemajorityofmembersofthe group.

5. The grouphas a homeland,or at least "remembers"one.

6. The grouphas a sharedand collectivelyrepresentedhistoryas a group.Furthert,his historyis notwhollymanufacturedb,uthas some basis in fact.

7. The groupis potentially"standalone" ina conceptualsense- thatis,itis nota caste or caste-likegroup(e.g., Europeannobilityor commoners).

The term"radial" comesfromtheobservationthatbytakingawayoneormoreofthese featureso,ne maygettypesof "ethnicgroups" thatare notprototypicablutnonetheless are oftenseen as ethnicgroups.Forexample:

? Take away 2, 4, and 6, (and possibly othersexcept 1), and you get an ethnic category ratherthan an ethnic group. The extent or degree to which these conditionsapplymightbe said to determinethe"groupness" of a group(Brubaker, 2002).

? Take away 5 (and possiblyothersexcept1) andyougetsomenomadicethnicgroups, suchas theRoma.

? Take away 7 and you get castes in SouthAsia, or noble/commonedristinctionsin Europe.

This content downloaded from 171.65.249.4 on Thu, 24 Oct 2013 17:24:57 PM All use subject to JSTOR Terms and Conditions

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download