12garrett - home | linguistics
TRANSCRIPT
139
ConvergenceintheFormationofIndo‑EuropeanSubgroups
Chapter12
ConvergenceintheFormationofIndo‑EuropeanSubgroups:PhylogenyandChronology
atedialectsintheprehistoriccontinuum,hascreatedthehistoricalmirageofabranchyIEfamilywithitsmanydistinctivesubgroups.
If thismodel is right, it alsohas ramificationsfortheproblemofIEchronology:When(andwhere)wasProto‑IEspoken,andwhatprocesses led to itsspreadacrossawideEurasianterritoryby1000JK?Iwillsuggestinsection3thataconvergencemodelofIEphylogenyaddstothedossierofevidenceagainstthe early chronology proposed by Renfrew (1987),andinfavourofthetraditionalview.Ishouldaddasacaveatthatwhatfollowsisinpartspeculativeandprogrammatic; further linguistic work is certainlyneeded.OnereasontheconvergenceprocessesIde‑scribehaveeludeddiscoveryisthattheirnatureistoerase the evidence for earlierdialect continua.OurevidencefortheIElanguagesmostlybeginswiththeresultsoftheprocessesIdescribehere;onlyinrarecases,likethatofLinearBandMycenaeanGreek,dotheaccidentsofarchaeologicaldiscoveryofferaclearwindowontheformationofanIEbranch.
2.Phylogeny:theformationofGreek
Work by Alice Kober, Michael Ventris, and JohnChadwick led fi[y years ago to the discovery thattheLinearBwritingsystem,usedonCreteandtheGreekmainlandinthesecondmillenniumJK,wasasystemofwritingGreek.ItisnowwellestablishedthatthedialectoftheLinearBtexts,Mycenaean,thoughdocumentedoverfourcenturiespriortothefirstsig‑nificanta]estationofotherGreekdialects,mustbetreatedasaGreekdialectandnotasProto‑GreekoraseparateIEdialect.ThisisbecauseMycenaeansharesinnovationswithindividualGreekdialects,suchastheassibilationof*‑ti>‑si(asinehensi‘theyare’),sharedwithArcado‑Cyprian,EastAeolic,andA]ic‑Ionic(vsWestGreekentí).Basedonsharedinnovationpa]erns,the scholarly consensus is thatMycenaean ismost
AndrewGarre]
1.Introduction
InthischapterIaddresstwointerrelatedproblems.ThefirstistheproblemofIndo‑European(IE)phy‑logeny:HowistheearlyfiliationoftheIElanguagefamilybestmodelled,andifourmodelsaretree‑likewhat should the trees look like? Iwill suggest thatconventionalmodelsofIEphylogenyarewrong.TheirbasicpresuppositionisthatIEhasasetoftenormorefamiliarsubgroups—Anatolian,Indo‑Iranian,Greek,etc.—whichcaninturnperhapsbeorganizedintohigher‑order subgroups suchas ‘Italo‑Celtic’or thenon‑AnatoliansubgroupIwillcallNuclearIE(NIE).The la]er is now widely accepted (Melchert 1998;Ringeetal.2002;Jasanoff2003),butmosthigher‑ordersubgrouping proposals are controversial, becausethe shared innovations said to justify them are farless robust than thosedefining thewell‑establishedsubgroups.Thereis,inshort,anessentialdifferenceinlinguisticprofilebetweenthefamiliarIEsubgroupsand proposed higher‑order subgroups. No modelwhosesolemechanismoffiliationissimplebranchingseemswellsuitedtocapturethisbasicdifference.
Iwillsuggestanalternativemodel:thefamiliarbranches arose not by the differentiation of earlierhigher‑ordersubgroups—from‘Italo‑Celtic’toItalicandCeltic,andsoon—butbyconvergenceamongneighbouringdialects in a continuum.Dialect con‑tinua are typical in shallow‑time‑depth languagefamilies;initsearlyhistory,Iwillsuggest,therewerealso IE continua fromwhich the familiar branchesemergedbymutualassimilationasadjacentdialectscametooccupyanddefinenewlinguisticandsocio‑cultural areas (Celtic,Germanic, etc.). The adjacentdialects fromwhichnewgroups emergedmaynothaveformedsubgroupswithinanearliercontinuum;dialects may even have shared innovations withneighbours that eventually fell intoother linguisticgroups.Convergence,togetherwithlossofintermedi‑
140
Chapter12
closely affiliated with Arcado‑Cyprian (MorpurgoDavies1992).
Itisalsowellestablishedthattherearelinguisticchangesfoundinallfirst‑millenniumGreekdialects,including Arcado‑Cyprian, that are not found inMycenaean. Before the decipherment of Linear BsuchchangeswereassumedtobeProto‑Greek,butnowitisclearthattheyreflectarealdiffusionacrosstheGreek‑speakingarea.Themasculine‑neuteractiveperfect participle presents a typical case. All first‑millenniumdialectsreflectasuffix*‑wot‑,asinHo‑mericarērót‑‘fashioned’<*arār‑wot‑,butthisisaGreekdevelopment;thecorrespondingNIEsuffixwas*‑wos‑.YetMycenaeanhasformslikeneuterpluralarārwoha<*arār‑wos‑andnonewith‑wot‑.AnapparentProto‑GreekinnovationisunreconstructiblefortheancestorofallGreekdialects.Howgeneralisthispa]ern,anddoesitaffectouroverallviewofProto‑Greek?InthiscontextMorpurgoDavies(1988,102n4)writesthat‘itwouldbeausefulexercisetocollectallthefeatureswhichwewouldhavea]ributedtoCommonGreekbeforethedeciphermentofLinearB’.
2.1.TheevidenceforProto‑GreekThisisnottheplacetopresentindetailtheresultsofthe exerciseMorpurgoDavies advocates, but I cansummarizeitsfindings.Ihaveexaminedfeaturesat‑tributedtoProto‑GreekbyMeillet(1913),wellbeforethedeciphermentofLinearB, excluding those thatarenotuniquetoGreek.Itturnsoutthatli]leremainsofMeillet’sProto‑Greek;excludingpost‑Mycenaeaninnovations, fewuniquechangesdistinguishGreekphonologicallyormorphologicallyfromNIE.1
IninflectionalmorphologyitiswellknownthattheGreekverbsystemisquitearchaic,butthefirst‑millenniumsystemofnouninflectionhasundergonesignificantchange.Meillet(1913)stressedthelossofthespatial(ablative,instrumental,andlocative)cases,which he called ‘one of the traits that characterizeCommon Greek’ (1913 [1975], 46); these categoriessurvived innoGreekdialectknownin1913.WhiletheMycenaean case system is still controversial inpart,Hajnal(1995)arguesthattheinstrumentalandlocativecasesbothsurvivedandthatinamajorin‑flectional class, animate athematic consonant‑stemnouns,theonlycase‑markingchangefromPNIEtoMycenaeanwasadative‑locativepluralsyncretism.Thenewending‑si(vsearlierloc.pl.*‑su)showstheonlyclearnominalform‑changethatisbothuniquetoGreekandpan‑Greek,butitisatrivialadaptationbasedonloc.sg.‑iandinstr.pl.‑phiwithfinali.ThelossoftheablativehadbeguninIEinasmuchasitsformswereparasiticonthegenitiveinthesingularandonthedativeintheplural;sincetheGreekgeni‑
tiveexpressesablativefunctions,thelossoftheabla‑tive can be viewed as an extension of the singularsyncretismintotheplural.TheinflectionalsystemoftheProto‑GreeknounthusdifferedonlymarginallyfromthatofitsPNIEancestor.
Inphonology,thediscussionisusefullydividedintothreeareas:segmentinventory,syllablestructure,andwordstructure.Intheareaofsegmentinventory,thequestioniswhatthesoundsofPIEwereandhowthey have changed. Since the Greek vowel systemis famously conservative, this amounts to examin‑ingtheconsonantsandsyllabicsonorants.Tobeginwiththela]er,itiswellknownthatPIE*l•,*r 8,*m8,and*n 8mainlydidnotsurviveinIElanguages;theirlossisamajorcauseofthecollapseoftheinheritedmor‑phologicalablautsystem.InGreek,reflexesof*l•and*r 8showaandovocalismvaryingacrossdialects;nopan‑Greek development can be reconstructed. Thenasals*m8and*n8becomeainfirst‑millenniumdialectsbutinsteado[enshowoinMycenaeanwhenprecededby labial consonants,as in*spermn >spermo ‘seed’.TheIEsyllabicsonorantswouldthereforestillhavebeendistinctphonologicalcategoriesintheancestorofMycenaeanandotherGreekdialects.
Among other segment types,Mycenaean alsoretainsthelabiovelarstopskw,gw,kwh,aswellasyandwinmostpositions;indeed,thechangeofy>hbeforesonorantsisrecentandongoinginMycenaean.Associ‑atedwiththegenerallossofyispalatalizationofmanyconsonanttypesinCyclusters;theFirstPalatalizationaffecting *t(h)y occurred beforeMycenaean, but theSecondPalatalizationaffectingabroaderrangeofCyclusterswasarguablyatleaststillongoing.
Insegmental terms, then,anyProto‑Greekan‑cestraltoMycenaeanandthefirst‑millenniumdialectsmusthavehadtherelativelyarchaicsegmenttypeskw,gw,kwh,y,w,l•,r8,m8,andn8.InfacttheonlyIEsegmenttypesmissing inProto‑Greekwouldhavebeen thelaryngeals*h1,*h2,and*h3,segmentslostinallNIElanguages and probably already at least partly inPNIE.Onapurelysegmentallevel,themostsignifi‑cantchanges tohaveprecededMycenaeanseemtohavebeentheFirstPalatalizationandtheconditionedchangeof*yand*stoh.ThesegmentalchangesthatdistinguishPNIEandProto‑Greeklooklesssubstan‑tial than those differentiating English, French, orGermandialects(notallofwhicharealwaysmutuallyintelligible,ofcourse).
Under the syllable‑structure rubric can begrouped various changes simplifying original CsCandobstruent‑sonorant clusters infirst‑millenniumdialects.AsshownbyMycenaeanformslikeaiksmā‘spear’,hehraphmenā‘sewn’,anddleukos‘sweetwine’,theseinnovationscannotbereconstructedforProto‑
141
ConvergenceintheFormationofIndo‑EuropeanSubgroups
Greekevenwheretheyaffectalllaterdialects.Steriade(1993)hasshownthatsuchclusterchangesreflectabasicchangeinsyllablestructure:MycenaeanretainedanIEsyllablestructurecanonallowingmanymoreon‑settypesthantherelativelyimpoverishedsetofclus‑ters(suchasstop+liquid)oflaterGreekdialects.
Finally,undertheword‑structurerubricIcon‑siderasetofchangesnoto[enseenasrelated.Withcharacteristic insight,Meillet (1913)wrote that ‘theendof thewordisdistinct;withoutpresentinganyconstant particularity it was felt in a preciseman‑ner’(1913[1975],26).HemeantbythisthatseveralGreek changes conspired to demarcateword ends:theaccentualDreimorengesetz,thelossoffinalstops,andthemergerofthenasals.Iwouldextendthisap‑proach,andsuggest that it isanorganizingfeatureofanumberofGreekinnovationsthattheyservetodemarcate prosodicwords both at the le[ and therightedge.
At the le[word‑edge, twoGreekchangescanbeseenasby‑productsofthedevelopmentofaspira‑tionasaninitial‑syllableprosody.Oneisaspirationmetathesis,bywhichanhinasecond‑syllableonsetsometimesmigrated to the beginning of theword,asin*euhō>héuō‘singe’(Lejeune1982,95–6,137–8).Thisisclearlypost‑Mycenaean,since‘wheel’isspelled<a‑mo>inLinearBandmustbeinterpretedasarhmo,not†harmo,whichwouldbespelled<a2‑mo>;aspira‑tionmetathesisisseeninlaterhárma.
Thesecondle[‑edgechangeisGrassmann’sLaw,bywhichaninitialaspiratedstopisdeaspiratedwhenanaspiratefollowsintheword,asingen.sg.*thrikhos>trikhós‘hair’.TheLinearBscriptdoesnotdistinguishaspiration,butGrassmann’sLawmustpostdateMyc‑enaeanbecauseitmustpostdatethepost‑Mycenaean*phm > mm change (compare hehraphmenā ‘sewn’above). This in turn is shown by later forms liketethramménos‘havingbeennourished’withoutGrass‑mann’sLaw,fromtheroot*threph‑(tréphō‘nourish’).IfGrassmann’sLawprecededMycenaeanwewouldexpect †tetraphménos > †tetramménos, like Homericepépithmen ‘we had been persuaded’ from the root*pheith‑(péithō‘persuade’).
Underlying both aspiration metathesis andGrassmann’s Law is a single pa]ern: aspiration isa temporally extended phonetic feature stretchingacross the entire first syllable.A dissimilatory lossofaspirationofthistypeoccurswhenaspirationas‑sociatedwith thefirst stop is reinterpreted (due toits extended duration) as a coarticulatory effect ofthesecondstop,whileametathesisas seen inhéuōorhármaariseswhenthephonologicalsourceofex‑tended‑duration aspiration is phonetically obscure(Blevins&Garre]1998;2004).
At the right word‑edge, a set of changes oc‑curredthatcanberelatednotjustphonologicallyasdemarcativebutphoneticallyviatheinverseofinitialaspiration:finallaryngealization.Thesechangesareashi[ofthepositionoftheaccent,whichoriginallycouldoccupyanysyllableofthewordbutinGreekisrestrictedtothelastthreesyllables;thelossofallfinalstops;andthemergerofword‑final*mand*nasn.ThedefectsofLinearB,whichwritesneitheraccentnor coda consonants,make it hard to tell whetherthese changes had occurred inMycenaean. But anindirectsuggestioncanbemadethatthefinalnasalmergermaynothavetakenplace,sincethetransferofhistoricalm‑stemnounsintotheclassofn‑stemshadnothappened,asshownbythedativesingularhemei ‘one’.2That transfer is in turn a consequenceof themerger ofword‑final *m and *n asn. If it isplausible that the three right‑edge sound changesare interrelated, it is also plausible that none hadtakenplaceinMycenaean.NotethatthetraditioninMycenaeanstudies is to interpretMycenaeansoastobeassimilaraspossibletofirst‑millenniumGreek,evenwhereLinearBgivesusnoevidence.Theop‑positestrategymaybeasappropriateeveniftosomeextentspeculative.
The relationship among the three right‑edgechangescanbeunderstoodas follows. It is likely inGreek,asinmanylanguages,thatfinalstoplosshadanintermediatestagewithglo]alizedstops([ ],etc.).Theretentionofstopsbeforeword‑finals(asinthríks‘hair’)supportsthisview,sincestopsbeforeswouldarticulatorilyhaveaspreadglo]is,preventingglo]ali‑zation.IsuggestthattheGreekaccentshi[regularizedfallingpitchattherightedgeoftheword;previouslytherewasnocorrelationbetweentherightword‑edgeandpostaccentualfallingpitch.Thisfacilitatedword‑finalstopglo]alization,perhapsasareinterpretationoftheambientlaryngealizationo[enassociatedwithfallingpitch.Mergerofword‑finalstopsas[/]>∅andneutralizationofword‑finalnasalplacecontrastsmaythenhavebeenthesamechange: lossofdistinctionscuedbyfinalVCtransitions.AsIhavesuggested,itisreasonabletospeculatethatthesechangesalloccurredinthecenturiesa[erMycenaean.
Insum,especiallyifweallowthatatleastafewpost‑Proto‑GreekchangesmustalreadyhaveaffectedMycenaeanbeforeitsa]estation(itisa[erallaGreekdialect),detailedanalysisreducesthedossierofde‑monstrableanduniquelyProto‑Greekinnovationsinphonologyandinflectionalmorphologytonearlyzero.Proto‑GreekretainedthebasicNIEnounsystem,verbsystem,segmentinventory,syllablestructure,andar‑guablyphonologicalwordstructure.Inalltheseareasoflinguisticstructure,GreekwasnotyetGreekearly
142
Chapter12
inthesecondmillennium.Butifso,ithardlymakessensetoreconstructProto‑Greekassuch:acoherentIEdialect,spokenbysomeIEspeechcommunity,ances‑traltoallthelaterGreekdialects.ItisjustaslikelythatGreekwasformedbythecoalescenceofdialectsthatoriginallyformedpartofacontinuumwithotherNIEdialects,includingsomethatwentontoparticipateintheformationofotherIEbranches.WiththisinminditispossibletoseeexternallinksforsomeGreekdialectpa]erns.Forexample,thefirst‑personpluralendings‑mesand‑menaredistributedsuchthat‑mesoccursinWestGreek,acrosstheAdriaticfromItalic(withsinLatin‑mus),while‑menoccurselsewhere,acrosstheAegeanfromAnatolian(withninHi]ite‑wen).Theisoglossseparatingprepositionalvariantsprotí(asinHomer)andpotí(WestGreek)likewisecorrespondstotheIndo‑IranianisoglossseparatingSanskritprátiandAvestanpaiti.
IfProto‑GreekdidnotexistassuchandMyce‑naeanphonologyandinflectionareminimally‘Greek’,whatmakesMycenaeanGreek?Chadwick,seeingtheessenceoftheproblem,haswri]enthat‘theremusthavebeenatimewhentheancestrallanguagecouldnotfairlybedescribedasGreek’,addingthatthebestevidence thatMycenaean isGreek ‘comes fromthevocabulary,whichcontainsnumerouswordswhichare ... specific to Greek’ (1998, 27). In short, Greekin the secondmillenniumalreadyhadadistinctivederivational,lexical,andonomasticprofile.ItmightnotoverstatethecasetosaythatMycenaeanwasalateNIEdialectwithGreekvocabulary;adistinctivelyGreekphonologicalandinflectionalprofilewaslargelyadevelopmentofpost‑Mycenaeanhistory.
2.2.SystemscollapseandlinguisticinnovationThe finding that numerous linguistic innovationsspreadacrosstheGreekdialectareainthecenturiesa[erMycenaeanmakes sensebothhistorically andsociolinguistically.Twopointsarekey.First,archaeo‑logical evidencepoints tomassivepopulation shi[and economic change during the Greek DarkAgec.1200–800JK.Morris(2000,195–6)writesof‘giganticupheavalsallacrosstheeastMediterraneanaround1200’,includingthedestructionoftheMycenaeanpal‑aces,migration,famine,disease,‘economicdisaster’,andmassivedepopulation.Thearchaeologicaldata,accordingtoDickinson(1994,87),‘surelyreflectcon‑siderablesocialchanges’.Thelinguisticeffectsofthesechangeshavebeennotedbefore;forexample,duringtheDarkAge,nearly‘thewholeoftheterminologyconnectedwiththesystemsofland‑tenureseemstohavedisintegrated’(MorpurgoDavies1979,98).
Second, toward the end of theDarkAge andsubsequentlythereisawealthofevidenceforemerg‑
ingsystemsofinteractionthatlinkedtheGreekworldeconomically,socially,andpolitically.InthiscontextSnodgrass(1980)mentionsarablefarming,metallurgy,colonization, panhellenic sanctuaries, ship‑buildingandnavigation,polisrivalries(inarchitecture,athlet‑ics,etc.),andwritingandliteracy,thoughitmustbesaidthatsomeatleastofthesesystemsemergedonlylaterintherelevantperiod.
Inshort,prototypicalexamplesoftwopa]ernsareseenbetweentheMycenaeanperiodandthere‑emergenceofGreekwritinginthefirstmillennium:a systems collapse (Tainter 1988, 10–11) and theemergenceofanewsystemalsobasedonpeer‑polityinteraction(Renfrew&Cherry1986).Asalientfeatureofthenewsystemisthewell‑knownsenseofGreekethnicidentity,whichbydefiningtheboundariesofaGreekdialectareamusthavefavouredthediffusionofinnovationsacrossthatareaandnofarther.
These historical phenomena are importantsociolinguisticallybecausetheyletusfitGreekintoa broader picture of language change. Linguistsstudyingsocialstructurehavefoundthattightsocialnetworksare
animportantmechanismoflanguagemaintenance,inthatspeakersareabletoformacohesivegroupca‑pableofresistingpressure,linguisticandsocial,fromoutsidethegroup...Oneimportantcorollarytothelinkbetweenlanguagemaintenanceandaclose‑knitterritorially‑basednetworkstructureisthatlinguisticchangewillbeassociatedwithabreak‑upofsuchastructure(L.Milroy1987,182–90).
Bycontrast,loosesocialnetworks,thosewithmanytiesoutsidetheirnetworks,‘arelikelytobegenerallymoresusceptibletoinnovation’(J.Milroy1992,181).
Ethnographicsociolinguistsmainlystudylocalsocialcontexts,butextrapolatingtoabroaderscaleand longuedurée fewhistorical se]ings couldmoreaptlybecalled‘thebreak‑upofaclose‑knitterritori‑ally‑based network structure’ than the GreekDarkAge.Complex systems collapse should yield rapidlinguisticchange;citingthewell‑knowncaseof theAlgonquian languages Arapaho and Gros Ventre,Bakker(2000,586)writesthat‘[i]nsituationsofgreatsocialupheavalandchangesonecanwitnessphono‑logical changewhich takes placemuch faster thanotherwise’.ForGreek,theperiodbetweentheendofLinearBdocumentationandthere‑emergenceofwrit‑inginthefirstmillenniumshouldhavebeenaperiodofrelativelyrapidlinguisticinnovation.Thischange,Isubmit,wastheformationofGreekasweknowit.
2.3.TheoriginofIndo‑EuropeanphylogenyDoesthemodelpresentedaboveapplyonlytoGreek,orcanitbegeneralized?Inanearlierarticle(Garre]
143
ConvergenceintheFormationofIndo‑EuropeanSubgroups
1999)IsuggestedthatGreekmaybetypicalofIEsub‑groups,andthatthereasonweseethepa]ernclearlyinGreekisthatwehaveMycenaean.3FornootherIEbranchdowehavecomparabledata—anItalicdialectof 1000 JK, or an Indo‑Iranian variety documentedearlyinthesecondmillennium.ButthecoherenceofotherIEbranchescanbedoubtedtoo.ThequestionofItalicunityhasbeendebatedbylinguistsforatleast75years.EvenforIndo‑Iranian,notalong‑standingproblemlikeItalic,theNuristanilanguagesshowthatthexyz{soundchangepostdatedProto‑Indo‑Iranian,andthepa]erningofearlyloansintoUralichassug‑gestedthatIndo‑Iranianwasalreadydialectallydif‑ferentiatedc.2000JK(Carpelanetal.2001).
IftheformationofGreekwasalocaleventfacili‑tatedbylocalinteractionpa]ernsandethnicidentity,itisalsorelevantthatIEbrancheslikeIndo‑Iranian,Slavic,Celtic, and even the poorly a]estedVeneticshowevidenceofacollectivesenseofethnicidentity.Insuchcases,asNichols(1998,240)putsit,‘acomplexnativetheoryofethnicityandastrongsenseofethnicidentitycanbereconstructed,andboththetheoryandtheidentitywerebasedonlanguage’.
IhavearguedthataMycenaeansystemscollapseprecipitated a period of rapid innovation inGreekdialectsandthecreationofacharacteristicGreekpho‑nologicalandmorphologicalprofile,butthecollapsewasnomereparochialeventoftheeasternMediter‑ranean.AccordingtoCunliffe(1997,41),
[t]heimpactoftheAegeansystems‑collapseontheEuropean hinterland was considerable. Existingexchangesystemsbrokedownorweretransformed.Some communities, once part of European‑widenetworks,foundthemselvesisolatedandnewcon‑figurationsemerged.
Itisthuspossiblethatthedynamicsbehindtheemer‑genceofCeltic,Italic,andotherIEbranchesofEuroperefractthesamehistoryasthosebehindtheemergenceofGreek.InAsia,thoughtherecanhardlybedirectevidence,wemayimaginesimilarprocessesatplayintheformationofIndo‑Iraniana[erthecollapseoftheBactria‑MargianaArchaeologicalComplexc.1750JK(Parpola2002,91–2).
IfthisframeworkisappropriateforIEbranchesgenerally,we cannot regard IE ‘subgroups’ as sub‑groupsinaclassicalsense.Rather,thelossor‘pruning’of intermediatedialects, togetherwithconvergenceinsituamongthedialectsthatweretobecomeGreek,Italic,Celtic,andsoon,haveintandemcreatedtheappearanceofatreewithdiscretebranches.ButthetruehistoricalfiliationoftheIEfamilyisunknown,anditmaybeunknowable.
Iconcludesection2bynotingapa]erninneedofanexplanation.Earlyinthesecondmillennium,Ihave
suggested,IEbranchessuchasGreekhadacquiredmuchoftheirlexicalandderivationalprofile,whiletheir grammatical apparatus continued to have itsbasicNIEcharacter.Speakinginthebroadestterms,earlyIElanguagespreadwasthusatwo‑phaseproc‑ess.Inthefirstphase,localIEdialectsacquiredtheirspecificlexical,derivational,andonomasticfeatures;in the secondphase, late in the secondmillenniuminsomecases,changesthatgavedialectareas theircharacteristic phonology and morphology sweptacrossthoseareas.Whatsociolinguisticallyplausiblescenariocouldgiverisetosucheffects?
3.Chronology:thedispersalofIndo‑European
PhylogeneticreconstructionsmayalsocontributetothedebatebetweenthetwochronologicalframeworkspositedfortheinitialIEdispersal.InwhatIwillcallthe first‑agriculturalists framework (Renfrew 1987),PIEwasspokenaround7000JKandIEspreadwiththediffusionofagriculturefromAnatoliaintoEuropeintheseventhmillennium.OnthisviewthemodernIElanguageshavedivergedforabout9000years.InwhatIwillcallthesecondary‑productsframework,thetimedepthofIEissomethreemillenniashallower:PIEwasspokenandIElanguagedispersalbeganinthefourthmillennium.Thischronologicalframeworkistraditional;generalpresentationsfromthispointofviewincludethatofMallory(1989).ThenameIusealludestothesecondaryproductscomplex.UnderthisrubricSherra](1981;1983;1997)hasidentifiedseveralemergentusesofdomesticatedanimals—ploughing,carting,wool, anddairy— thatarose inEurope inthelatefourthandearlythirdmillennia;hereferstoa‘revolution’that ‘markedthebirthofthekindsofsocietycharacteristicofmodernEurasia’(1981[1997,161]).Newpropertytransmissionsystems,land‑usepractices,andsocialnetworkpa]ernsaresaidtobeaspectsofthetransition.
3.1.ImplicationsofconvergenceInsofaras the formationof IEbrancheswasa localprocess,andtheircharacteristicinnovationstookplacelaterthanusuallysupposed,theirphonologicalandmorphological structuresmust have been closer inthecenturiesaround2000JKthanhasbeenthought.Table 12.1 shows reflexesoffivePNIEnumerals inthreeintermediateprotolanguagesandrepresentativemodern descendants: Greek; Spanish; and Waigali(Nuristani,Indo‑Iranian:Turner1962–6).
Thesimilarityoftheintermediateprotolanguagesisobvious,andclearlyalsofewerchangesoccurreden route to each intermediate protolanguage thansubsequently.ModernGreekisthemostphonologi‑
144
Chapter12
callyconservativelanguageinthesampleofHeggarty(2000),andevenforGreekTable12.1showsonlyfoursoundchangesenroutetotheintermediateprotolan‑guage(h1>e,kè>k,s>h,irregularnnin‘nine’)butatleastelevenhistoricallydistinctlatersoundchanges:syllabicnasals>a;kw>tbeforee;lossesofy,w,andh;ee>ē;ē>ī;ea>ja;lossofvowellength;stops>fricativesbeforestops;andashi[frompitchtostressaccent(notshown).Intheotherlanguagesthelaterchangesareplainlynumerous,alsoincludingpitch‑to‑stressshi[s,whileProto‑ItalicandProto‑Indo‑Iranianeachshowonlyfourchanges.4TheGreekreconstructionsfollow§2.1,anditisworthaddingthattheIndo‑IranianandItalicformsmaybetooinnovatorypreciselybecausewedonothavetheequivalentofMycenaeanGreekprovingthepresenceofareallydiffusedchanges.Ifanything, theextentofphonologicalchanges inthemodernlanguagesisunderstated.
The time depth from the intermediate pro‑tolanguages to theirmoderndescendants is on theorder of 4000 years (Proto‑Italicmay be somewhatyounger),andduringthisperiodsignificantlymorephonologicalchangehas takenplace thanoccurredenroutefromPNIE.5NotethatallthreeintermediateprotolanguagesretainthebasicPNIEsystemofnomi‑nalcasesandinflection,andintheverbalsystemthethree‑wayPNIEaspectcontrastamongpresent,aorist,andperfect(Meiser2003).Noneofthissurvivesinthemodernlanguages.
The first‑agriculturalists model posits a spanof3000–4000yearsbetweenPNIEand2000JK.Thismeans assuming two typologically incomparableperiods,eachthreeorfourmillennialong:aperiodmarkedby lessphonological or inflectional changethan isobserved inanydocumented language, fol‑lowedbyaperiodwhenallIElanguagesweretrans‑formedbyaccumulatingwavesofphonologicalandmorphological change. That is, themodel requirestheunscientificassumptionthatlinguisticchangein
the period for which we have nodirect evidence was radically dif‑ferent from change we can studydirectly.6
There is nothing new in aconclusionthat linguisticevidencefavours the secondary‑productschronology over than the first‑ag‑riculturalists chronology (Nichols1998,254–5;Darden2001),thoughIhopenewlightisshedontheques‑tion if IE subgroups are productsof secondary convergence. Othertypesofrelevantlinguisticevidenceinclude especially the evidence of
linguisticpalaeontology,amethodwithwell‑knownpitfalls whose results in this case have been chal‑lenged;Iwillconsiderthisissueinsection3.2.Mostimportantly, asRenfrew (1987) has remindedus, itbehoovesaproponentofanyviewofIEdispersaltosituatethatviewinaplausiblemodelofancientsocialdynamics.The centralquestionshavealwaysbeen:WhatcausedthespreadofIndo‑European,andwhydiditspreadoveritsbroadEurasianterritory?Iwillsketchanapproachtothesequestionsinsection3.3.
3.2.LinguisticpalaeontologyInessence,theargumentfromlinguisticpalaeontol‑ogy is that IE is reconstructedwithwords for sec‑ondaryproducts(plough,wool,yoke)andwheeledtransport (axle, nave, thill, wagon, wheel); sincethese technologiesdidnotarisebefore4000JK, theIEdispersalcannotbeassociatedwiththediffusionof agriculture severalmillennia earlier. In the first‑agriculturalistsframework,PIEandevenPNIEdatefrombefore5000JK,neitherlanguagecouldhavehadsecondary‑productsorwheeled‑transportterms,andtheentireterminologicalensemblemustbealinguis‑ticmirageifitseemsreconstructibletoPIEorPNIE.7Howthenarethedataexplained?Eventheadvocatesoflinguisticpalaeontologyrecognizethatthemethodhasgeneralpitfalls,butthespecificdatamustbescru‑tinizedcritically.8
One alternative account is independent for‑mation: apparent cognates do not reflect commoninheritance from a single ancestral prototype, butwere separately formed in several languages. Forexample,perhapstheapparentPIE*h2wr 8gis ‘wheel’(basedon*h2werg‑‘turnaround’)reflectsindepend‑entformationinHi]iteandTocharian.ButsuchanaccountishardlypossibleforPNIE*kwekwlos‘wheel’(in Germanic, Greek, Indo‑Iranian, Tocharian, alsoborrowed early into Uralic): though derived from*kwelh1‑‘turn’,areduplicatedC1e‑C1C2‑o‑nounisso
Table12.1.FivenumeralsinPNIEandthreeNIEbranches.
‘three’ ‘five’ ‘seven’ ‘eight’ ‘nine’
PNIE *treyes *peŋkwe *septm8 *okètō *h1newn8
Proto‑Greek *treyes *peŋkWe *heptm8 *oktō *ennewn8
ModernGreek tris pente e[a oxto eñja
Proto‑Indo‑Iranian *trayas *pañča *sapta *aćtā *nawa
Waigali tre pũč sot os. t. nũ
Proto‑Italic *trēs *kWeŋkWe *septm8 *oktō *newn8
Spanish tres siŋko sjete očo nweve
145
ConvergenceintheFormationofIndo‑EuropeanSubgroups
unusualmorphologically that parallel independentformationisexcluded.Likewise,forPIE*yug@om‘yoke’(*yeug@‑‘harness,join,yoke’),withreflexesinAnatolianandalmostallotherbranches,whileneuterthematicnounsare common in some IE languages, theyarequiterareinAnatolianandithaslongbeenknownthatthecategorywasnotveryproductiveinPIE.Itisunlikelythat*yugomwasindependentlycreatedinAnatolianandotherbranches,allthemoresobecausetheformismorphologicallyinvariant;haditbeencre‑atedindependentlywemightexpectotherformationsinsomecases.Otherwordsposeadifferentproblem:PIE*h2erh3‑ ‘plough’(everywherebutAlbanianandIndo‑Iranian)isarootandPNIE*akès‑‘axle’(inBaltic,Celtic,Germanic,Greek,Indo‑Iranian,Italic,Slavic)isnottransparentlybasedonaroot;thereforeneithercanbe the result of any re‑formation. In almost allcases,theformsofsecondary‑productsandwheeled‑transport terms must be reconstructed for PIE orPNIE.
Fortheirmeanings,secondarysemanticshi[isapossiblealternativeaccount.ThusRenfrew(2001,46)suggeststhatPIE*h2/3wl•h1neh2 ‘wool’ (inAnatolian,Baltic,Celtic,Germanic,Greek, Indo‑Iranian, Italic,Slavic) might originally have referred to ‘the fiberfromthesheep’,perhapsusedforrugs,clothing,orfelt,shi[ingitsmeaninglaterassheepwerebredforwool.Indeed,formorphologicalreasons,itisclearthatthewordreferredoriginallytofiberthatwasusedinsomeway:*h2/3wh1neh2isderivedviathesuffix*‑neh2fromtheroot*h2/3welh1‑‘topluckorpullhair’(Latinvellere).Thewordcannotsimplyhavereferredtohair;itmusthavearisenatatimewhensheephadcoatswhosehairswereplucked.Butthereisnoreasontobelievethatsuchapracticeexistedbeforethebreedingofwoollysheep:theearliestusesofthesheep’scoatwereinfelting,forwhichthereisnoevidencebeforethe thirdmillennium, and spinning into thread forweaving,knownforwoolfromthefourthmillenniumintheNearEastandlaterinEurope.9Inshort,woolmakesnomorphologicalsenseonthefirst‑agricultur‑alistsIEchronology.
Argumentsinvolvingsecondarysemanticshi[canbeenvisionedforotherterms,likethosefor‘nave’and‘thill’,whichrefergenerallytopolesinsomelan‑guages;theshi[toavehicularcontextcouldbeinde‑pendent(Specht1947,100–102).Butinothercasesitishardtoseehowsecondarysemanticshi[canexplainthedata.Perhaps*yugom‘yoke’originallyreferredtoanother joining result, butwhat? andwhywas theoriginalmeaning lost throughout IE? In thecaseofwordsfor‘wheel’,whatdidtheyoriginallydesignateifnotwheels?Asemanticshi[fromconcrete‘wheel’toabstract ‘circle,cycle’ isplausiblebutthereverse
shi[(abstract>concrete)isunusualatbest(Sweetser1990;Traugo]&Dasher2001).
It remains to comment on the possibility ofvocabularydiffusionorborrowinga[ertherelevanttechnologyarose.Borrowingordinarilybetraysitselfphonologically; if ‘yoke’ had been borrowed fromIndo‑Iranian into other IE branches, itwould havetakentheform*yugama[ertheIndo‑Iranianchangeofotoa.Onecouldsuggestthatvocabularydiffusiontookplacebeforethephonologicalchangesofindivid‑ualbranches,butgiventheoverallrequirementsofthefirst‑agriculturalistschronologythiswouldmeanthatIEhistoryfirsthadaperiodofseveralthousandyearswithjustvocabularychangeanddiffusion,andthatlinguisticchangeasitoccursinclearlydocumentedlanguageswouldonlyhavebegunwiththesecond‑ary‑productscomplex;asnotedinsection3.1,thisideaisunrealistic.Apossiblealternativeaccountinvoking‘etymologicalnativization’(Hock1991,392–3)wouldfounderonthedivergentprofilesofthatprocessandtheIEdata.
Itisworthaddinginconclusionthatlinguisticreconstruction yields not just isolatedwords but aterminologicalensembleinacoherentsemanticfield.Theexplanatorystrategyforcedbythefirst‑agricul‑turalistsframeworkmissesthebigpicturebyinvokinganunsystematicjumbleofadhocalternatives.
3.3.ThedispersalofIndo‑EuropeanRenfrew(1987)originallyarticulatedthefirst‑agricul‑turalists frameworkusing themechanismofdemicdiffusion,theslowmovementofpeoplesovermanycenturies as one farmer a[er anothermoves a fewmilestoclearnewfarmland.Thismeansoflanguagespreadisclearlydocumentedinmanycases(Bellwood&Renfrew2002;Diamond&Bellwood2003).ForIEitselfRenfrew(2000)haslargelyabandonedtheidea,allowinginresponsetocriticsthatdemicdiffusionwasnotthemechanismofIElanguagespreadinwesternandnorthernEuropeandinsteadinvokinglanguageshi[(atermthatlabelsaphenomenonwithoutoffer‑ingasociolinguisticmechanismormodeltoexplainit).ForIndo‑Iranian,aconnectionbetweenthespreadoffarmingandlanguagedispersalinanareacompa‑rableinsizetoEuropehasbeenabandonedaltogether(Renfrew2000,423–4).
Asrecentlyas2001Renfrewhaswri]enthat‘theonlyprocessoreventofsufficientlygeneralsignificanceforthewholeofEuropetoaccountfortheIndo‑Europe‑anizationofalmostanentirecontinentwasthecomingoffarming’(Renfrew2001,37).Suchastatementhidestwomisleadingassumptions.First,Sherra]andotherscontendthatthesecondaryproducts‘revolution’wasawatershedinEuropeansocialhistory;thedifference
146
Chapter12
betweenachronologybasedonfarmingandonebasedonsecondaryproductsisjustwhatisatstake.Second,inthiscontextcallingEurope ‘anentirecontinent’ isdistractinglyEurocentric.Europeisalso ‘asmallpe‑ninsulaoftheEurasianlandmass’(Richards2003,142),and the IE spread is abroadEurasianphenomenonthatshouldbeseenassuch.AninterpretationthatsetsasidehalftheIEarea,offersanexplanatorymodeloflanguagespread(demicdiffusion)onlyforpartofEu‑rope,andmainlydismisseslinguisticevidencecannotberegardedasasatisfactoryaccountofwhatisa[eralla linguisticprocess:thedispersalofIElanguagesacrossEuropeandAsia.
ItakeitthatPIEwasspokenc.3500JK,perhapssomewhat earlier, in a part ofwhatMallory (1989,239)callsthe‘circum‑Ponticinteractionsphere’;thePIEareacouldnothavebeenlargerthanthatofeco‑logicallycomparablelanguages,forexamplethesizeof Spain (Anthony 1995). It is traditional to situatePIEinthePontic‑Caspiansteppe,thoughawesternPonticPIEmaysuitdialectgeographybe]er(Sherra]&Sherra]1988proposeacircum‑PonticPIEnotlongbefore4000JK).ItisimportanttobearinmindthatPIEmayhavehadlinguisticallyrelatedneighbours;wecannotknowhowtheensemblewouldappearinthearchaeologicalrecord.
TheoldestIEsplit,betweenAnatolianandNIE,mayhavebegunasasmall‑scalecollapseofthePIEspeechcommunitywithasocioculturalreorientationofitsnorthernandsouthernhalves.Perhapsthesouth‑ernhalfofthecommunitywasdrawnintotheinterac‑tionalsphereofthelatefourth‑millenniumAegean,orthenorthernhalfwasdrawnintotheBalkans.Inanycase,PonticIEspeakerscametobeorientedtowardstheBalkansandthesteppe,whileotherswereorientedsocioculturallytowardstheAegeanandAnatolia.
In the Pontic area NIE began to differentiate,withTocharianitseasternmostdialectalongtheBlackSeaandthefirstknownIElanguagetomakeitswaytoCentralAsia.ThesecondwasIndo‑Iranian,whosespreadonthesteppeandc.2000JKtoBactria‑Mar‑gianaiswidelyaccepted(Mallory1998;2002;Parpola1988;1998;2002;Renfrew2000,423–4).TocharianhadperhapsseparatedfromtheNIEareabyc.3000JK,withIndo‑Iranianspreadingeastwardonthesteppeduringthethirdmillennium.
TheEuropean expansion, even if it representsonlypartofIEdispersal,isacrucialproblem.Renfrew(1999; 2000; 2001) suggests thatNIE dialects had alongepisodeofmutualconvergenceinwhathecalls‘OldEurope’,followingGimbutas(1973)—aBalkanandEast‑centralEuropean interactionalsphere thatflourished in the fi[h and fourth millennia beforefragmenting. In his view this period was marked
bydiffusionacrossmostNIEdialects;atitsend‘thestrongculturalinteractionsmarkingthe“OldEurope”episodediscussedbyGimbutascametoanend,andthevarioussub‑regionstendedtogotheirownsepa‑rateways’(Renfrew2001,42).
Renfrew(1979)hasstronglyemphasizedthecon‑sequencesofsystemscollapse.Theseincludelinguisticdiffusionasinpost‑MycenaeanGreek(section2.1),orlinguisticreplacementasdiscussedbyRenfrew(1987,133–7)forseveralstatecollapses.Butnotallcomplexsocietiessubjecttosystemscollapsearestates;somearenetworksoftheOldEuropetype.Inhisanalysisofthecollapseofcomplexsocieties,includingsocie‑tiesofseveralorganizationaltypes,ageneralpa]ernemphasizedbyTainter (1988, 191) is that ‘[i]neachcase,peoplesontheperiphery...rosetoprominencea[ertheoldersocietyhadcollapsed’.
InthescenarioIhavesketchedNIEdialectswerespokenontheperipheryofOldEurope,andIsuggest,inwhatIhopeisnotanunholyallianceofthedoc‑trinesofGimbutasandRenfrew,thatitwasthecollapseoftheOldEuropeinteractionalsystemthatfacilitatedtheinitialspreadintoEurope.LikeGimbutas,IseeIEdispersalasrelatedtowhatMallory(1989,238)sum‑marizesasmid‑fourthmillennium‘culturalchaos’and‘somethingofaBalkan ‘darkage’.ButIagreewithRenfrewthatitisnotnecessaryordesirabletoimagineinvasions bywarrior Indo‑Europeans; systems col‑lapsenaturallyledtorapiddispersalofthespeechofitsperiphery.10InacomplexsystemIEspeakersmustalreadyhaveinteractedwithmorecentralparticipantsinroleswecannotknow(perhapssomewerespecialistwainwrights,weavers,orherders).ThepointisthatanIEspreadintotheBalkansandEast‑centralEurope,in the late fourth and early thirdmillennia,wouldbeanaturalaspectofthecollapseofOldEurope.Asarguedinsection2.3,thelateremergenceofEuropeanIElanguagesthatweredistinctivelyCeltic,Italic,andsoonmayhavefollowedtheAegeansystemscollapseofthelatesecondmillennium.
ViewingtheIEdispersalsbroadly,itispossibletodiscernthreemajorpa]erns.OneisthesteppespreadthatledtothedispersalofTocharianandIndo‑Iranian.Asecondpa]ernischaracteristicoftheIEspreadintoEuropeandlinguisticchangesthattookplacethere:dispersalwasassociatedwithsystemscollapse(OldEurope, the late‑second‑millennium Aegean) andthesocialreorganizationsofthesecondaryproductscomplex.TheIndo‑IranianspreadintoIranandSouthAsia a[er the collapse of the Bactria‑MargianaAr‑chaeologicalComplexcanalsoperhapsbeassimilatedtothispa]ern.
Thethirdpa]ernisnotwidelynotedbutseemsquiterobust:anorth–southspread into the interac‑
147
ConvergenceintheFormationofIndo‑EuropeanSubgroups
tionalspheresoftheurbanizedzonethatrunsfromtheAegean throughAnatolia and theNearEast toBactria‑Margiana. This significant Eurasian pa]ernhasatleastfourinstantiations:• theinitialsplitofPIE,insofarasitwasassociated
with a reorientationofProto‑Anatolian towardsthe Aegean and Anatolia, with the subsequenteastwardspreadofHi]ite;
• thespreadofGreekdialectsintoGreeceandtheMinoansocioculturalworld;
• theIndo‑IranianspreadintotheoasiscitadelsofBactria‑Margiana;
• thespreadoftheMitanniIndo‑IraniandialectintoSyria.
Ineachcasetheresultingsocioculturalprofileshowssignificant continuity with indigenous patterns,respectivelyHa]ic (in the case ofHi]ite),Minoan,Bactria‑Margiana,andHurrian.
Mallory’s analysis of the Indo‑Iranian spreadmay be broadly applicable here. Mallory & Mair(2000,267)commentasfollowsontheinteractionofAndronovo‑culturesteppeIndo‑Iraniansandtheur‑banoasisdwellersofBactria‑Margianac.2000JK:
[T]heAndronovanswouldhavecomeintocontactwiththeoasis‑dwellers,adopteditemsoftheirmate‑rialculture,someoftheirreligiousbeliefsandcul‑turalpractices(suchasthefirecultandconsumptionofthehallucinogenic*sauma),butnotthelanguageof the oasis‑dwellers. Rather, the language of thesteppe‑dwellerswouldhaveoperatedasthelinguafranca of exchange between regions, thenperhapswithinthese]lementsthemselvesuntilsomevarietyof Indo‑Iranianhadbecome themain languageofWestCentralAsia....
Mallory(2002,39)writesfurther:Indo‑IraniantribesfromthesteppelandsenteredintothepoliticalsphereoftheBMAC[Bactria‑MargianaArchaeological Complex] and absorbed from it asuiteofreligiousinstitutionsandtheirnamesaswellas theconceptofasuperordinate tierwithin theirsocialorganisation.This tier ...providedasystemofcoordinationbetweenthedifferentelementsbothwithintheBMACandthemobileunitsoutside. ItlinkedoasisdwellersandsteppenomadsinCentralAsiaand...itcouldalsobringtogetherpeopleprac‑tisingdifferentse]lementandeconomicstrategiesonthenorthernsteppe.
ForGreece,Palaima(1995,127)describesaprocesswherebytheestablishedHelladic/Aegeanand Indo‑European features of mainland cultureweretransformedandmadepartoftheLateHelladicpalatialculturethroughastrong,selectiveadaptationofdiverseelementsofMinoanmaterialcultureandMinoansocial,politicalandreligiousideology.
Across Eurasia generally, I suggest, IE language
spreadmaybeinterpretedpartlyasaresultofsuchinteractions between a northern periphery and asouthernurbanzone.
To speculate further, the same pa]ern of in‑teractionmaywellliebehindthetwo‑stageprocessidentifiedinsection2.ThefirststageofIElanguagespreadischaracterizedbyadistinctivelexical,deri‑vational,andonomasticprofile; thiscorrespondstourbanizationandtheuseofindigenoussocioculturaltraditionsbyspeakersof IE languages. InAnatolia,Greece,andBactria‑Margianarespectively,comparethe‘dominantroleofHa]icelementsinOldHi]itereligionandcultandideologyofkingship’(Melchert2003,17),includingHa]icloanwordslikehalmaššuiX‑‘throne’;theelitesemanticprofileof‘Minoan’loansinGreek(Renfrew1998),includingthevocabularyofkingship(Mycenaeanwanaks>ánaks,perhapsgwasileus>basiléus);andthedossierofborrowedIndo‑Iraniansocialandreligiousvocabulary,includingimportanttermslike*indra‑‘Indra’and*dāsa‑‘(hostile)people’(Lubotsky2001).11Insuchcircumstances,weexpectsignificantlexicalchangeaswellaschangesinmoresocioculturally embedded aspects of morphology,such as onomastics and ways of deriving occupa‑tionalterms,ethnicadjectives,andthelike.WhatisresponsibleforthisfirststageofIEdispersalisthusthesocioculturalcontinuityweseeinAnatolia,Greece,andBactria‑MargianaasIElanguagesarrive.
The second stage with its phonological andinflectional transformations corresponds, on thisview,totheemergenceoflocalethnicidentitiesandnetworks.Insomecasesthismayhavebeenalong,gradualprocess;inothersasystemscollapsemayhavefacilitatedrapidinnovation,asinGreek,Indo‑Iranian(ifaBactria‑Margianacollapsec.1750JKplayedaroleintheemergenceofdistinctiveIndo‑Iranianphonol‑ogyandmorphology),andperhapssomeEuropeanIElanguages.
4.Conclusion
Ihavemadetwomainargumentsinthischapter.Insection 2, based on a new analysis of Mycenaean,I argued that the apparent features of Proto‑GreekmainlydiffusedthroughoutGreeceduringanda[ertheMycenaean period. It follows that Proto‑Greek—orifthisdidnotexist,IEspeechofc.2000JKthatwastobecomeGreek—waslinguisticallyclosertoIEthanhasbeensupposed.IsuggestmoregenerallythatweshouldcontemplatemodelsofIEphylogenythatassignagreaterroleintheformationofIEbranchestoconvergenceinsitu.
Insection3,Iexploredthechronologicalconse‑quencesofthisviewofIEphylogeny.Ifthelinguistic
148
Chapter12
changesinvariousIEbranchestookplacerelativelylate in their histories, then it is unlikely that PIEwasspokenc.7000JKasinthefirst‑agriculturalistsframework.SpeculativelybutIhopeconstructively,IbrieflysketchedascenarioforIEdispersalthatfitsthelinguisticfactsandmayperhapsanswerwhatRenfrewrightlyasksofIndo‑Europeanists,thatanyaccountbesituatedinaplausiblemodeloflinguisticchangeandsocialdynamics.
Acknowledgements
ThankstoJamesClackson,PeterForster,andColinRenfrewforinvitingmetoparticipateintheCambridgesymposiumandtootherparticipantsforusefuldiscussion.Forvaluablecommentsonawri]endra[of this chapter,whichhavesavedmefrommanyerrors,IamgratefultoJulie]eBlevins,Peter Forster, Jay Jasanoff, Leslie Kurke, Nino Luraghi,CraigMelchert,AnnaMorpurgo Davies, Colin Renfrew,andMichaelWeiss(fewifanyofwhomagreewithallmyconclusions).
Notes
1. ForinformationaboutMycenaeanGreekreadersmayconsultthehandbookofBartone&k(2003)andthelexiconofAuraJorro(1993),bothwithfullreferencestootherliterature.AchangeintheverbsystemthatshouldbenotedbecauseitisseeninMycenaeanisthedevelop‑mentofthematic3sg.*‑eti>‑ei(which,Iwouldargue,isindirectlyrelatedtotheFirstPalatalizationmentionedbelow).
2. Onthisview<ko‑to‑na‑no‑no>(PYEa922)cannotbeinterpreted as haplography for <ko‑to‑na‑na‑no‑no>withacc.sg.ktoino #n(Morpurgo1963s.v.),andtheepi‑graphicallyuncertainformatPYEq146.11cannotbeinterpretedas <i.‑qo.‑na‑to‑mo>withgen.pl.hikwkwo #n‘horses’(Chadwick1979,25).AsfarasIknow,unam‑biguousMycenaeanformswhereafinalnasaliswri]eninsandhihavenotyetbeenfound.
3. SeeZimmer (2002) fordiscussionof relatedperspec‑tives.
4. TheIndo‑Iranianchangesarelaryngealloss,palatalstopaffrication,theLawofPalatals,andthemergerofnon‑highvowelsandsyllabicnasalsasa.TheItalicchangesarelaryngealloss,yloss,ee>e #,andthep>kwchangein‘five’.ForProto‑ItalicIfollowMeiser(2003,30–31)exceptthatItakeew>owasasecondarydevelopmentin light of early Latin forms likeneuna. In any case,amongtheItalicchangesonlylaryngeallossissecure:ylossispreciselyachangeformerlyreconstructedforProto‑GreekbeforethedeciphermentofLinearB,ee>e #contractionisdependentonyloss,andthep>kwchangeisonlyweaklyreconstructible.
5. NotethattheReader(referee)ofthisvolumeobjectedthatifwehadadditionalinformationfromotherwiseunknownlanguagesdescendedfromPNIE(andsuchlanguages certainly existed), our changed PNIE re‑constructionsmight amplify the changes en route to
thea]esteddaughterlanguages;therelativeclosenessofPNIEandtheintermediateproto‑languagesmightrepresentamirage.Inprinciple,ofcourse,thisisafairpoint.However,ontheonehand,somethingofthisgen‑eralsortdidindeedhappenwiththediscoveryofAna‑tolian,andbynowamajoreffecthasbeenthewidelyacceptedredrawingoftheIEfamilytree;ontheotherhand,itisjustastruethatnewlydiscoveredlanguagesordialects (likeMycenaeanGreek) can change one’sreconstructions of the intermediate proto‑languagesand therewouldbeno a priori reason for thesenewdiscoveries tohavea systematiceffecton theoveralldifferencesunderdiscussion.
6. Tobesure,inawidelypublicizedstudy,Gray&Atkin‑son(2003)havesuggestedthatcomputationalphylo‑geneticanalysismaysupportthefirst‑agriculturalistschronology.Butevense]ingasidemethodologicalques‑tions,anddoubtsaboutlinguisticreconstructionfromlexicaldata, thespecificresultsofGray&Atkinson’sresearchare likely tobe inerrorasa resultofabiasin theunderlyingdata (Dyen et al. 1997).Modern IEbranchesshowexampleswhere,inaparticularsemanticslot,theirknowncommonancestor(Latin,Sanskrit,etc.)hasonewordwhichhasbeenreplacedbyadifferentwordinallormostdescendantlanguages.ThusLatinignis‘fire’hasbeenreplacedbyreflexesofLatinfocus‘hearth’ throughout Romance, and archaic Sanskrithanti‘kills’hasbeenreplacedbyreflexesofayoungerSanskrit formma #rayati throughout Indo‑Aryan. ThisprocessespeciallytargetswordsofIEantiquity,whicharemoreo[enirregularandthereforepronetoreplace‑ment.Thispa]erncreatestheillusionofaslowerrateofchangeintheinternalhistoriesofmodernbranches:it seems in retrospect (say) that Latin focus replacedanIEwordfor‘fire’intheprehistoryofLatinandnotlaterinRomance.BecausetheoverallratesofchangepositedinGray&Atkinson’smodelarebasedonap‑parentrateofchangeinmodernbrancheswithknownhistories,theoverallratesofchangeassumedwillbetooslow.OvertheevolutionfromPIEtoProto‑Italic,Proto‑Indo‑Iranian,etc.,thetimedepthcalculatedwillthusbetoolong.Icannotassessthepreciseeffectsofthisbias,buttospeculate,if5percentofthedataislikefocusa‘true’averagelexicalretentionrateof80percentover1000yearswillinsteadlooklike85percent;thisisequivalentto23percentretentionover9000years,whilean80percentrateisequivalenttoasimilarfigure(26percent)over6000years.Preciselythis3000‑yeardifference distinguishes the first‑agriculturalists andsecondary‑productsmodels.
7. Itisnottrue,asallegedbyRenfrew(2000,432–4;2001,45),thatthemorphologyofsuchvocabularyshowsthatitispost‑PIE,noristhissuggestedbySpecht(1947)orLehmann(1993),whomhecites.Rather,theargumentisthatbecauseathematicnounsareolderthanthematicnounswithin the prehistory of PIE, wheeled‑transporttermswererelativelynewinPIE.
8. ForanexcellentreviewoftheevidenceseenowDarden(2001).ThesensiblyskepticalassessmentofClackson(2000)treatsmainlytheweakestevidenceinthedossier,
149
ConvergenceintheFormationofIndo‑EuropeanSubgroups
andinacrucialcaseheoffersaninconsistentanalysis,rightlynotingthat‘thill’and‘yoke’terms‘donotneedthereconstructionofachariot,butcouldalsoapplytoaplough’(445),butthensuggestingthat‘claimslinkingIndo‑Europeanto...the“secondaryproductsrevolu‑tion”...canalsobechallengedinmuchthesameway’(447). If theapparentsecondary‑productsvocabularyisillusorythenitshouldnotbeusedtoexplainothervocabulary.
9. SeeBarber(1991,24–5,221)andSherra](1983[1997,203]).According toBarber (1991,24n8), ‘thekempiertypeofwildsheep’have‘virtuallyunspinnable’coats.Becausewoolwaspluckedortornbeforeitwasshorn(Barber1991,21),*h2/3welh2‑givestherightsensewhereaverbroot ‘cut’wouldnot.OutsideLatin theverbalformsofthisrootarehardertojudge;forrelevantma‑terialseenowRix(2001,674–9),andseeDarden(2001,196–204)onthearchaeologicalevidence.
10. ForareviewofarchaeologicaldataseeWhi]le(1996),whonotes‘extensiveandprofoundchangesthroughoutsouth‑eastEurope’c.4000–3500JKandsuggeststhatIElanguages‘mayhavespreadaYerthesechangeswereunderway,notastheirprimarycause’(126–7).
11. It is importanttoemphasizethatHa]ic linguistic in‑fluenceonHi]itehasbeenoverstatedinthepastandthatthecaseisstrongerforLuvianlinguisticinfluence(Melchert2003).ForGreek,Renfrew’s(1998)otherwiselucid treatment ismarredby a failure todistinguishtwo processes of contact‑induced language change,borrowingviamaintenanceandinterferenceviashi[(Thomason&Kaufman1988;Thomason2001);thetwoleavedistinctlinguistic‘footprints’,andtheGreekdatashowborrowing.
References
Anthony,D.,1995.Horse,wagon,andchariot:Indo‑Europe‑anlanguagesandarchaeology.Antiquity69,554–65.
Aura Jorro,F., 1993.DiccionarioMicénico, 2vols.Madrid:ConsejoSuperiordeInvestigacionesCientíficas.
Bakker, P., 2000. Rapid language change: Creolization,intertwining, convergence, in Renfrew et al. (eds.),585–620.
Barber,E.J.W.,1991.PrehistoricTextiles.Princeton(NJ):Prin‑cetonUniversityPress.
Bartoněk, A., 2003. Handbuch des mykenischen Griechisch.Heidelberg:C.Winter.
Bellwood,P.&C.Renfrew(ed.),2002.ExaminingtheFarm‑ing/LanguageDispersalHypothesis.(McDonaldInstituteMonographs.) Cambridge: McDonald Institute forArchaeologicalResearch.
Blevins,J.&A.Garre],1998.Theoriginsofconsonant‑vowelmetathesis.Language74,508–56.
Blevins,J.&A.Garre],2004.Theevolutionofmetathesis,inPhoneticallybasedphonology,ed.B.Hayes,R.Kirchner& D. Steriade. Cambridge: Cambridge UniversityPress,117–56.
Carpelan,C.,A.Parpola&P.Koskikallio(eds.),2001.EarlyContacts betweenUralic and Indo‑European:Linguisticand Archaeological Considerations. (Mémoires de la
SociétéFinno‑Ougrienne242.)Helsinki:Suomalais‑UgrilainenSeura.
Chadwick, J., 1979.TheuseofMycenaeandocuments ashistoricalevidence, inColloquiumMycenaeum:ActesduSixièmeColloque International sur lesTextesMycé‑niens et Egéens tenu à Chaumont sur Neuchâtel du 7au13septembre1975,eds.E.Risch&H.Mühlestein.Genève:Droz,21–32.
Chadwick, J., 1998.TheGreeknessofMycenaean.Aevum72,25–8.
Clackson,J.,2000.TimedepthinIndo‑European,inRenfrewetal.2000,441–54.
Cunliffe,B., 1997.TheAncientCelts.Oxford:OxfordUni‑versityPress.
Darden,B.,2001.OnthequestionoftheAnatolianoriginofIndo‑Hi]ite,inDrews(ed.),184–228.
Diamond, J.&P. Bellwood, 2003. Farmers and their lan‑guages:thefirstexpansions.Science300,597–603.
Dickinson, O., 1994. The Aegean Bronze Age. Cambridge:CambridgeUniversityPress.
Drews,R. (ed.), 2001.GreaterAnatolia and the Indo‑HiXiteLanguageFamily:PapersPresentedataColloquiumhostedbytheUniversityofRichmond,March18–19,2000.Wash‑ington(DC):JournalofIndo‑EuropeanStudies.
Dyen,I.,J.B.Kruskal&P.Black,1997.ComparativeIndo‑Europeandatabase.Downloadedon3July2004fromhttp://www.ntu.edu.au/education/langs/ielex/IE‑DATA1.
Garre],A.,1999.AnewmodelofIndo‑Europeansubgroup‑inganddispersal,inProceedingsoftheTwenty‑FiYhAn‑nualMeetingoftheBerkeleyLinguisticsSociety,February12–15,1999,eds.S.S.Chang,L.Liaw&J.Ruppenhofer.Berkeley(CA):BerkeleyLinguisticsSociety,146–56.
Gimbutas,M.,1973.OldEuropec.7000–3500JK:theearli‑est European civilization before the infiltration oftheIndo‑Europeanpeoples.JournalofIndo‑EuropeanStudies1,1–21.
Gray,R.D.&Q.D.Atkinson,2003.Language‑treedivergencetimessupporttheAnatoliantheoryofIndo‑Europeanorigins.Nature426,435–9.
Hajnal,I.,1995.StudienzummykenischenKasussystem.Berlin:WalterdeGruyter.
Heggarty,P.,2000.Quantifyingchangeovertimeinphonet‑ics,inRenfrewetal.(eds.),531–62.
Hock,H.H.,1991.PrinciplesofHistoricalLinguistics.2ndedi‑tion.Berlin:MoutondeGruyter.
Jasanoff,J.H.,2003.HiXiteandtheIndo‑EuropeanVerb.Oxford:OxfordUniversityPress.
Lehmann,W.P.,1993.TheoreticalBasesofIndo‑EuropeanLin‑guistics.London:Routledge.
Lejeune,M.,1982.PhonétiquehistoriqueduMycénienetduGrecancien.2ndedition.Paris:Klincksieck.
Lubotsky,A.,2001.TheIndo‑Iraniansubstratum,inCarpe‑lanetal.(eds.),301–17.
Mair, V.H. (ed.), 1998.The BronzeAge and Early IronAgePeoplesofEasternCentralAsia,2vols.(JournalofIndo‑EuropeanStudiesMonograph26.)Washington(DC):JournalofIndo‑EuropeanStudies.
Mallory,J.P.,1989.InSearchoftheIndo‑Europeans:Language,ArchaeologyandMyth.London:Thames&Hudson.
150
Chapter12
Mallory,J.P.,1998.AEuropeanperspectiveonIndo‑Euro‑peansinAsia,inMair(ed.),175–201.
Mallory,J.P.,2002.ArchaeologicalmodelsandAsianIndo‑Europeans,inSims‑Williams(ed.),19–42.
Mallory,J.P.&V.H.Mair,2000.TheTarimMummies:AncientChina and theMystery of theEarliestPeoples from theWest.London:Thames&Hudson.
Meillet,A., 1913 [1975].Aperçu d’une histoire de la languegrecque.8thedition.Paris:Klincksieck.
Meiser,G.,2003.Venividivici:DieVorgeschichtedeslateinischenPerfektsystems.München:C.H.Beck.
Melchert,H.C., 1998. The dialectal position ofAnatolianwithin Indo‑European, inProceedings of theTwenty‑FourthAnnualMeetingoftheBerkeleyLinguisticsSociety,February14–16,1998:SpecialSessiononIndo‑EuropeanSubgroupingandInternalRelations,eds.B.K.Bergen,M.C.Plauché&A.C.Bailey.Berkeley(CA):BerkeleyLinguisticsSociety,24–31.
Melchert,H.C.,2003.Prehistory, inTheLuwians,ed.H.C.Melchert.Leiden:Brill,8–26.
Milroy,J.,1992.LinguisticVariationandChange:OntheHistori‑calSociolinguisticsofEnglish.Oxford:Blackwell.
Milroy,L.,1987.LanguageandSocialNetworks.2ndedition.Oxford:Blackwell.
Morpurgo,A, 1963.Mycenaeae Graecitatis Lexicon. Rome:Ed.dell’Ateneo.
MorpurgoDavies,A.,1979.Terminologyofpowerandter‑minologyofworkinGreekandLinearB,inColloquiummycenaeum: Actes du sixième Colloque internationalsurlestextesmycéniensetégéenstenuàChaumontsurNeuchâteldu7au13septembre1975,eds.E.Risch&H.Mühlestein.Geneva:Droz,87–108.
MorpurgoDavies,A.,1988.MycenaeanandGreeklanguage,inLinearB:a1984Survey,eds.A.MorpurgoDavies&Y.Duhoux.Louvain‑la‑Neuve:Peeters,75–125.
MorpurgoDavies,A.,1992.Mycenaean,Arcadian,Cyprianand some questions of method in dialectology, inMykenaïka,ed.J.‑P.Olivier.(BulletindeCorrespondanceHellénique,Supplément25.)Paris:Boccard,415–32.
Morris,I.,2000.ArchaeologyasCulturalHistory:WordsandThingsinIronAgeGreece.Malden(MA):Blackwell.
Nichols,J.,1998.TheEurasianspreadzoneandtheIndo‑Europeandispersal, inArchaeology and Language II:CorrelatingArchaeologicalandLinguisticHypotheses,eds.R.Blench&M.Spriggs.London:Routledge,220–66.
Palaima,T.G.,1995.Thenatureof theMycenaeanwanax:non‑Indo‑European origins and priestly functions,inTheRole of theRuler in thePrehistoricAegean, ed.P.Rehak.(Aegaeum,11.)Liège:UniversitédeLiège,119–39.
Parpola,A.,1988.ThecomingoftheAryanstoIndiaandtheculturalandethnicidentityoftheDāsas.StudiaOrientalia64,195–302.
Parpola,A.,1998.Aryanlanguages,archaeologicalcultures,and Sinkiang: Where did Proto‑Iranian come intobeing,andhowdiditspread?,inMair(ed.),vol.1,114–47.
Parpola,A.,2002.FromthedialectsofOldIndo‑AryantoProto‑Indo‑AryanandProto‑Iranian,inSims‑Williams(ed.),43–102.
Renfrew,C.,1979.Systemscollapseassocialtransformation,inTransformations,MathematicalApproachestoCultureChange, eds. C. Renfrew& K.L. Cooke. New York(NY):AcademicPress,275–94.
Renfrew,C., 1987.Archaeology andLanguage: thePuzzle ofIndo‑EuropeanOrigins.London:Cape.
Renfrew,C.,1998.WordofMinos:TheMinoancontributiontoMycenaean Greek and the linguistic geographyof theBronzeAgeAegean.CambridgeArchaeologicalJournal8(2),239–64.
Renfrew, C., 1999. Time depth, convergence theory andinnovationinProto‑Indo‑European:‘OldEurope’asaPIElinguisticarea.JournalofIndo‑EuropeanStudies27,257–93.
Renfrew,C.,2000.10,000or5000yearsago?—Questionsoftimedepth,inRenfrewetal.(eds.),413–39.
Renfrew,C.,2001.TheAnatolianoriginsofProto‑Indo‑Eu‑ropeanandtheautochthonyoftheHi]ites,inDrews(ed.),36–63.
Renfrew,C.&J.F.Cherry(eds.),1986.PeerPolityInteractionand Socio‑political Change. Cambridge: CambridgeUniversityPress.
Renfrew,C.,A.McMahon&L.Trask(eds.),2000.TimeDepthinHistoricalLinguistics,2vols.(PapersinthePrehis‑toryofLanguages.)Cambridge:McDonaldInstituteforArchaeologicalResearch.
Richards,M.,2003.TheNeolithicinvasionofEurope.AnnualReviewofAnthropology32,135–62.
Ringe,D.,T.Warnow&A.Taylor,2002.Indo‑Europeanandcomputationalcladistics.TransactionsofthePhilologicalSociety100,59–129.
Rix,H.(ed.),2001.LIV:LexikonderindogermanischenVerben:DieWurzelnundihrePrimärstammbildungen.2ndedi‑tion.Wiesbaden:Reichert.
Sherra],A.,1981.Ploughandpastoralism:aspectsof thesecondaryproductsrevolution,inPaXernsofthePast:Studies in Honour of David Clarke, eds. I. Hodder,G. Isaac & N. Hammond. Cambridge: CambridgeUniversityPress,261–305(reprintedinSherra]1997,158–98).
Sherra],A.,1983.ThesecondaryexploitationofanimalsintheOldWorld.WorldArchaeology15,90–104(revisedinSherra]1997,199–228).
Sherra],A.,1997.EconomyandSocietyinPrehistoricEurope:ChangingPerspectives.Princeton(NJ):PrincetonUni‑versityPress.
Sherra],A.&E.S.Sherra],1988.ThearchaeologyofIndo‑European:analternativeview.Antiquity 62,584–95(reprintedinSherra]1997,471–85).
Sims‑Williams, N. (ed.), 2002. Indo‑Iranian Languages andPeoples. (Proceedings of the BritishAcademy 116.)Oxford:OxfordUniversityPress.
Snodgrass,A., 1980.ArchaicGreece: theAgeofExperiment.London:J.M.DentandSons.
Specht,F.,1947.DerUrsprungderindogermanischenDeklina‑tion.Gö]ingen:Vandenhoeck&Ruprecht.
Steriade,D.,1993.GreekProsodiesandtheNatureofSyllabifica‑tionProcesses.NewYork(NY):Garland.
Sweetser,E.1990.FromEtymologytoPragmatics:MetaphoricalandCulturalAspectsofSemanticStructure.Cambridge:
151
ConvergenceintheFormationofIndo‑EuropeanSubgroups
CambridgeUniversityPress.Tainter, J.A., 1988.TheCollapse ofComplexSocieties.Cam‑
bridge:CambridgeUniversityPress.Thomason,S.G.,2001.LanguageContact.Washington(DC):
GeorgetownUniversityPress.Thomason, S.G. & T. Kaufman, 1988. Language Contact,
Creolization and Genetic Linguistics. Berkeley (CA):UniversityofCaliforniaPress.
Traugo],E.C.&R.B.Dasher,2001.Regularity inSemanticChange.Cambridge:CambridgeUniversityPress.
Turner,R.L., 1962–6.AComparativeDictionaryof the Indo‑AryanLanguages.London:OxfordUniversityPress.
Whi]le,A.,1996.EuropeintheNeolithic:theCreationofNewWorlds.Cambridge:CambridgeUniversityPress.
Zimmer,S.,2002.TheproblemofIndo‑Europeanglo]ogen‑esis.GeneralLinguistics39,25–55.