The Cosmopolitan Vernacular

Author(s): Sheldon Pollock

Source: The Journal of Asian Studies, Vol. 57, No. 1 (Feb., 1998), pp. 6-37
Published by: Association for Asian Studies
Stable URL: http://www.jstor.org/stable/2659022
Accessed: 01/12/2010 02:22

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=afas.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

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@jstor.org.

Association for Asian Studies is collaborating with JSTOR to digitize, preserve and extend access to The
Journal of Asian Studies.

http://www.jstor.org
 The CosmopolitanVernacular
SHELDON POLLOCK

THROUGHOUT SOUTHERN ASIA AT DIFFERENT TIMES startingaround1000,butin

mostplacesby 1500, writersturnedto theuse oflocal languagesforliterary expression
in preference to thetranslocallanguagethathad dominatedliterary expressionforthe
previousthousandyears.This developmentconstitutes at thelevelofculturethesingle
mostsignificant transformation in theregionbetweenthecreationofonecosmopolitan
orderat the beginningof the firstmillenniumand anotherand fardifferent one-
throughcolonialismand globalization-at the end of the second.
The vernacularization of southernAsia is not onlythe most importantcultural
changein the late medievalworld-or perhapswe should say, in the earlymodern
worldthatit helps to inaugurate-but also the least studied.We have no coherent
accountof the matterforany region,let alone a connectedhistoryforsouthernAsia
or forthe largerEurasiaworldwherea developmentverysimilarin culturalform(if
not in social or politicalcontent)appearsto have occurred.We have no well-argued
theoreticalunderstandingof many of the basic problemsat issue. And, what is
especiallydisabling,we lack any reliableaccountof the politicaltransformations in
southernAsia to whichtheseculturalchangesare undoubtedlyif obscurelyrelated,
or a theoryofpowerand culturebeforemodernity thatwould allow us to makesense
of thisrelation.
What I aim to do in the space availablehereis tryto sketchout, first,a fewof
the larger conceptual issues that impinge on an analysis of cosmopolitanand
vernacularin literaryculture,and the narrowerquestions that pertain to their
historicization.The very idea of vernacularization depends upon understanding
somethingof theworldagainstwhichit definesitself,and thisI providewitha brief
accountofthehistoricalformation and ideationalcharacterofwhatI call theSanskrit
cosmopolis.Fortheformer I look at theriseand spreadofSanskritinscriptions,which
serveas a synecdochefora rangeof literary-cultural (and political-cultural)
practices;
forthe latter,I consideras paradigmaticthe space of culturalcirculationas this
structures the literaryand literary-criticalimagination.All this is preparatoryto an
analysisof one case of the formationof vernacularliteraryculture,that of early

SheldonPollockis theGeorgeV. Bobrinskoy ProfessorofSanskritand IndicStudiesat

theUniversity ofChicago.
I wishto thankT. V. Venkatachala
Sastry myguidein Old Kannada.Benedict
(Mysore),
Anderson (Ithaca)offered
helpful whenan earlier
criticism versionofthepaperwaspresented
at the1995 meeting oftheAssociationforAsianStudies.Thanksalsoto Chicagocolleagues
ArjunAppadurai, CarolBreckenridge,
DipeshChakrabarty,andStevenCollinsfortheirsug-
and Homi Bhabha,to whoseongoingworkon the "vernacular
gestions, cosmopolitan" in
postcolonialismthepresentpapermaybe viewedas something ofa precolonial
complement.
TheJournal 57, no. 1 (February
ofAsianStudies 1998):6-37.
? 1998 bytheAssociationforAsianStudies,Inc.

6
 THE COSMOPOLITAN VERNACULAR 7

Kannada. Here the localizationof the globalizing literary-cultural practicesand

representationsof Sanskrit constitutes a model instance of cosmopolitan
vernacularism.At thesametimeI hopeto show,throughone narrowbutsymptomatic
example (the historyof the literary-criticaldiscourseon the "Way" of literature,
mdrga),not onlyhow thevernacularreconfigures the cosmopolitan,but how the two
produceeach otherin thecourseoftheirinteraction. I end witha briefaccountofthe
failureof existing historicalexplanations(such as they are) to account for the
vernacularturn,and flagsomeofthe challengesforfutureinquiry,mostcruciallythe
relationshipof literarycultureto political culturein the non-Westand the very
problematicofpremodernglobalization.

HypothesizingVernacularization

The possibilityof conceptualizingand historicizingthe cosmopolitan/vernacular

transformationrequiresa workinghypothesiswith a numberof componentsthat,
althoughtheymay appearto attemptto settlethroughdefinitionwhat can onlybe
determinedempirically, can all be demonstrated These concerncultural
historically.
choice,therelativityof"vernacular," theliterary,
thehistoricalsignificance
ofwriting,
the meaningof beginnings,and the sociotextualcommunity.I addressthesebriefly
in order.

Cultural Choice
A language-for-literatureis chosenfromamongalternatives, notnaturallygiven.
Human linguisticdiversitymay be a fatality,in BenedictAnderson'smelancholy
formulation, but thereis nothingfated,unselfconscious, or haphazardabout literary-
languagediversity;it is willed. Vernacularliterarylanguagesthusdo not "emerge"
like buds or butterflies,theyare made. Not manyscholarsacknowledgethis factor
do muchwithit. One ofthefewwas Bakhtin,who saw moreclearlythananyonethat
"the activelyliterarylinguisticconsciousnessat all timesand everywhere (thatis, in
all epochsof literaturehistorically available to us) comes upon 'languages'and not
language.Consciousnessfindsitselfinevitablyfacingthe necessityof havingtochoose
a language"(1981, 295). Yet so faras I can see whatneitherBakhtinnoranyoneelse
has spelledout in detailedhistoricaltermsforspecificlanguagesin theeveryday sense
(by "language"Bakhtinusuallymeantsocioideologicalregisters)is what is at stake
in this choice,what else in the social and political world is being chosenwhen a
language-for-literatureis chosen.Forit is one thingto recognizethatliterary-language
diversityis willed, and anotherthing altogetherto specifythe historicalreasons
informing thiswill.

"Vernacular"/"'Cosmopolitan"
To definevernacularoveragainstcosmopolitanappearsto submergea numberof
Althoughnot all cosmopolitanlanguagesmayinitiallybe vernaculars-
relativities.
here the historyof Sanskritwhen Sanskritliterature(kdvya)is inventedat the
beginningof the commonera differssharplyfromthat of, say, Latin in the third
centuryB.C. whenLatinliteratureis abruptlyinvented-manyvernaculars themselves
do becomecosmopolitanfortheirregionalworlds.This is trueforBraj, whichwas
rendered rootlessly cosmopolitan by the elimination-conscious elimination,
accordingto some scholars-of local dialectaldifference
in the fifteenth
to sixteenth
8 SHELDON POLLOCK

centuries.'Kannada,too,thoughoftenthoughtofas a regionalliterary code,has long

been transregionalforwritersin yetsmallerzonessuchas Tulu Nadu or theKonkan.
But theserelativities
look less worrisomefromwithinthesubjectiveuniversesofthe
agents involved. Vernacularintellectualsdefine a literaryculture in conscious
oppositionto somethinglarger;theychoose to writein a language that does not
travel-and thattheyknowdoes not travel-as easilyas the well-traveled language
of the cosmopolitanorder.The new geoculturalspace theyimagine,whichI discuss
in whatfollows,fullytestifiesto this.That this"local" in turntypicallycomesto be
constructed as dominantand dominatingforsmallerculturalspacesis a further step
in the cosmopolitan-vernacular transformationand unthinkablewithoutit.

The Literary
However much contemporary thoughtwants to ignore,resist,blur, or trash
definitionsof "literature,"the historicalsocietiesstudiedheremade an unequivocal
distinction,practicallyand oftenby explicittheorization, betweena realmoftextual
productionthatis documentary and anotherthatis somethingelse-call it expressive,
interpretative, "workly"(dcasWerkhafte, Heidegger 1960), literary,or whatever.
Contemporary scholarshipis certainlyrightto questiontheselocal distinctions,and
to look fortheexpressiveorworklyin thedocumentary and constative,
and thereverse
(LaCapra 1983, 23-71). But that is a second-orderenterpriseand subsequentto
gaininghistorical-anthropological knowledgeofwhatpoetsin middle-period southern
Asia thoughttheyweredoing and whenand why.The distinctionbetweenrestricted
and elaboratedcodes,betweenthedocumentary and the literary,
was oftenproduced
and reproducedpreciselyby meansof languagechoice,as the historyof inscriptions
clearlyshows. Facts of social or culturalpower seem to have impingedupon this
choice, suggesting that restrictionand elaborationare potentialitiespermitted
developmentin theone case and deniedit in theother.When thisdenialis challenged
in the vernacularizationprocess,moreover,the challengetypicallytakesthe formof
domesticatingthe literaryapparatus (themes, genres, metrics,lexicon) of the
superposedculturalformation thatset the rulesof the literarygame.

Writing
The literaryin southernAsia comes increasinglyin the middle period to be
distinguishednot just fromthe documentary but fromthe oral,and to be evermore
intimatelylinked to writing,with respectto the authorityconferredby it, the
textualityassociatedwith it, and the historyproducedthroughit. The authorization
to writeis not,like the abilityto speak,a naturalentitlement.It is typicallyrelated
to social and politicalprivileges,which markliteraturein the restrictedsense as a
differentmode of cultural productionand communicationfrom so-called oral
literature.2Grantedthatliterateliteraturein SouthAsia retainsmanytext-immanent

'Such processeshave been noticedonlyby linguists,who discussthe matterin reference

to "koines"and typicallyignoremostofwhatinterestsculturaltheory.Cf.,e. g., Segal 1993.
For Braj, cf.Snell 1991, 30-32, and, moregenerally,Masica 1991, 54.
2Accordingto well-knownlegends,Tukaram,like Eknanthbeforehim, was forcedby
outragedbrahmansto "throwhis poems into the river."When he defendshis use ofMarathi,
he is thus clearlydefendingthe rightto write,not just to compose(cf. Pollock 1995, 121-
22).
 THE COSMOPOLITAN VERNACULAR 9

featuresoforality(whatthe late scholarofOld FrenchPaul Zumthorcalledvocalite'),3

and that the principalmode of consumptionwas auditory,still, writingaffected
literarycommunication in profoundways.These await systematic analysis,but there
is no doubtthatto writeliterarilyalwaysmeantrendering languagebothlearnedand
learned,to endow it with new normsand constraints.Historicallyspeaking,what
countsin the historyof vernacularliteraryculture,what makeshistorynot onlyfor
us (byprovidinghistoricalobjects)but fortheprimaryagentsthemselves (bymarking
a breakin the continuumof history)is literization,the committingof literatureto
writing.

Beginnings
When therefore throughan act of culturalchoicethe vernacularis deployedfor
theliteraryand theliterary attainsinscription, begins-that is, at particular
literature
timespeople begin to inscribetexts,or,whatcomesto thesame thingas a historical
issue,begin to considertextsinscribedin local languagesworthpreserving.In this
sensethe historyof vernacularliterarycultureis not coextensivewith the historyof
vernacularlanguage. Such literarybeginningsin South Asia are the object of
ethnohistoricalrepresentation and,despitethemanylogicaland ideologicaldifficulties
thatbesettheveryidea of beginnings,are oftensusceptibleto historicalanalysis(cf.
Pollock 1995). I am especiallyinterested
in vernacularinaugurations,
thoughofcourse
the choiceto be vernacularhas a continuinghistory.

Community
The last, and least disputable of my contentions-though also the least
historicized-is the mutuallyconstitutive relationshipof literatureand community:
literatureaddresses,sometimescalls intobeing,particularsociotextualcommunities.
These definethemselvesin significant if variablewayson the basis of the literature
theyshare,and theycreatenewliteratures in serviceofnewself-definitions.To choose
a language forliterature,then-to committo writingexpressivetextsas defined
accordingto dominant-culture models-is at the same timeto choosea community,
thoughits precisemeaningand the natureof the identitythat literatureconstructs
forit need to be investigated,
and not imagined,fortheworldbeforemodernity.
Absentthiskindofconceptualframework, it is hardevento perceivethechoices
to be vernacular-or cosmopolitan-let alone recovertheir historiesand social
meanings.
The choice to be vernacularin South Asia at the beginningof the second
millenniumwas made againstthebackgroundofSanskritand deeplyconditionedby
the literarycultureof Sanskrit.Without understandingthe historyof the literary
world Sanskritcreatedand the work it did there,it is difficultto understandits
supersession,what vernacularliterarylanguageswere called upon to do, when,and
why.I hope to suggestsomethingof the characterof thiscultureby lookingfirstin
a perhapsunexpectedquarter:thehistoryoftheSanskritinscriptional discourse.There
are three things I concentrateon here: the historyof the transregionalcultural
formation ofSanskrit,how it came to be and whatit consistedof;theroleofSanskrit

3Cf.Zumthor1987. RelevanthereforSanskritand earlyKannada textsare the literary-

linguisticphenomena(gunas,see below) or the modesofrecitation(pdthaorpathiti)described
by literaryscholarssuch as Rajasekharain the tenthcentury(KdvyamTmrmsd7), and Bhoja in
th eleventh(?rhgdrapraprakd?(a7, pp. 379 ff.).
10 SHELDON POLLOCK

as the vehicleof political expression;and, relatedto this,Sanskrit'shighlymarked

statusas theliterarylanguageoveragainstlocal languages.This real-world formation
providesthebackgroundforthebriefaccountofgeoculturalrepresentations to which
I thenturn.

Historicizingthe SanskritCosmopolis

As momentousas the vernaculartransformation at the beginningof the second

millenniumwas the creation,aroundthe beginningof the first,of the cosmopolitan
orderto whichit was theresponse.4 Two new,relateddevelopments werefundamental
to this order:the use of Sanskritin inscriptionsand the inventionof "literature."
Sanskritinscriptions,typicallyissuedfromroyalcourts,arecrucialbothas expressions
of the political,and forthe wider trendstheyrevealin literary-language use and
normsof literariness,whichthe historyof Sanskritliteratureconfirms.
For its first400 years,inscriptionalculturein South Asia is almostexclusively
non-Sanskrit(the languages used were instead the Middle-Indic dialects called
Prakrit),but thissituationchangeddramatically at thebeginningofthecommonera
when we firstbegin to findexpressivetexts eulogizing royalelites composed in
Sanskritand inscribedon rock-faces, pillars,monuments, a formthat
or copper-plates,
will later receivethe genre nameprasati(praise-poem).The most famousof these
texts,producedforor by the Indo-Scythian(Saka) overlordRudradaman(ca. A.D.
150), has been knownto scholarsformore than a century,and nothinghas been
discoveredsince to alterthe impressionthatit marksa profoundcultural-historical
break.Never beforehad Sanskritspokenas it does in Rudradaman'stext,out in the
open,in writtenform,in reference to a historicalking,and in aestheticizedlanguage.
And yetalmostimmediately thereafter,and forthenextthousandyears,it is thevoice
ofSanskritpoetrythatwould be heardin politiesfromthemountainsofPeshawarto
Prambanamon theplains of centralJava.
It is about this same time that what comes to be called kdvya("[written]
literature")in the emerging scholarlydiscourse of rhetoric(ala ika-rarscastra) is
crystallized,whenthegreatgenressuch as mahdkdvya (courtlyepic) and ndtaka(epic
drama)come into existencealong with the formaltechniques,such as the systemof
figuresof sound and senseand the complexquantitative-syllabic metrics,thatwere
to defineSanskritliteratureand have such resonancethroughoutAsia. Literary-
culturalmemory,as thismaybe discernedin literary criticismor in thekaviprafsamsds
(praisesof poets) that conventionally introduceSanskritliterarytexts,has no reach
beyondthesebeginningsin the earlycenturiesof the commonera,and it is difficult
forhistoricalscholarshipto showthatkdvyaas it will henceforth be practicedis much
earlierthanthis.Sanskritinscriptions such as Rudradaman'sshouldnot therefore be
viewed,as theyusuallyare, as the latestdate forthe existenceof literarySanskrit
(kdvya),but as theearliest.And the two together, kdvyaandpras'ati, areevidence,not
ofa renaissance(or "resurgence," or "revival")ofSanskritcultureafter
"re-assertion,"
a Mauryanhiatus,but of its inaugurationas a new culturalformation.Previousto
thisSanskritcultureappearsto have been restricted to thedomainofliturgyand the
knowledgesrequiredforits analysis;it can hardlybe said to haveexistedin anything
like the formit was soon to acquire.

4Thisand the followingsectiondrawon the detaileddiscussionin Pollock 1996.

 THE COSMOPOLITAN VERNACULAR 11

Whetheror not I overdrawthisdiscontinuity betweena highlyrestricted social

sphere of Sanskrit(liturgicaland scholastic)and a new political use of Sanskrit
accompaniedby whollynew formsof writtenliterature,the subsequenthistoryof
Sanskritin inscriptionaldiscourseis the historyof an unprecedentedand vast
diffusion.Once it came to be used forinscriptionalliteraturein NorthIndia in the
second to thirdcenturies,Sanskritwas adopted elsewherewith astonishingspeed.
Prakritdisappearedfromthe epigraphicalrecordthroughoutIndia in the space of a
century,neverto be revivedforinscriptionsthereafter, and retainedonlya residual
statusin the literary-culturalorder.
A cruciallyimportantdimensionto theuse ofSanskritin epigraphsand the rise
of kdvyais the divisionof linguisticlabor in inscriptional discourse,and, relatedly,
the literarysilenceof the vernacularsthroughoutthe cosmopolitanformation. Once
Sanskrithad becomethe languageforthe public literaryexpressionof politicalwill
throughoutmuch of southernAsia, it remainedthe only language used forthat
purpose.The vernacular was notprohibitedfromspeakingin theinscriptional domain,
but thepermissionwas restricted. A typicalinscriptioncommenceswitha genealogy
and praise-poemof the overlordwho issuesthe document,followedby thedetailsof
the transaction the inscriptionis meantto record(the boundariesof thegiftedland,
the conditionsof a templeendowment,and the like). When used at all vernacular
language is restrictedto the second or businessportionof the grant,and thus to
counting,measuring,and aboveall localizing.The literary function-whereby power
constructed foritselfits origins,grandeur,beauty,perdurance, and whichcanperhaps
thereforebe characterized as thefunction ofinterpreting theworldand supplementing
reality-was the workexclusivelyof Sanskritpoetry.The verycontrastgeneratedby
this divisionof labor,a relationof superpositionof unrelatedlanguagesthatI have
termedhyperglossia, servesto enhance the aestheticismin which one may locate
Sanskrit'ssupremeattractions.
Related to the empiricallyobservabledivision of labor in inscriptionsis the
discourseon literary languagein thealacgkdra tradition.Fromtheseventhcenturyon
it became a commonplaceof this traditionthatkadvya was somethingthatcould be
composedonly in a highlyrestricted set of languages.Chiefof thesewas of course
Sanskrit;farbehindboth in theoryand in actual literary productionwereMaharastri
Prakritand Apabhrams'a, two languagesthatundertheinfluence ofSanskrithad been
turnedinto cosmopolitanidioms,and which therefore could be and were used for
literarycompositionanywherein theSanskritcosmopolis.5Kdvyawas notsomething
made in thevernacular;thusa rangeof regionallanguagesfromKannada to Marathi
to Oriyawereliterarily silent.
As theturnto Sanskritis takingplace in theIndiansubcontinent forthecreation
of inscriptions at once political,literary,and publiclydisplayed,preciselythe same
phenomenonmakesits appearancein whatare nowthecountriesofBurma,Thailand,
Cambodia,Laos, Vietnam,Malaysia,and Indonesia,and with a simultaneity thatis
again striking.The firstSanskritpublic poems appear in Khmercountry,Champa,
Java,and Kalimantanall at roughlythesametime,theearlyfifth centuryat thelatest,
a
or notmuchmorethan coupleor threegenerations aftertheirwidespreadappearance

5The restrictionon literarylanguagesbeginswith BhamahaKdvydlankdra 1.16, 34-36.

Only near the end of the cosmopolitanepoch do Sanskritwritersadmit the possibilityof
producinggrdmya mahdkdvya, courtlyepics in the "vulgar"language(cf. the twelfth-century
Kdvydnus'asana8.6, p. 449). The linguistically"unlocalized"qualityof Apabhramsia is noted
by Shackle 1993, 266; cf.also Hardy 1994, 5.
12 SHELDON POLLOCK

in India itself.And theywill continueto be producedin some places forcenturies:

the last dated Sanskritinscriptionin Cambodia is around 1295, a little beforethe
abandonmentofAngkor.
Khmercountry,in fact,fromroughly600-1300 providesa good exampleofthe
politicsof literaryculturenoted above. Here the world of public poetryremained
resolutelya worldof Sanskrit.Inscriptionsin Khmer,to be sure,are producedfrom
virtuallythe same date as inscriptionsin Sanskrit;in fact,nearlyhalfof the extant
inscriptions are solelyin Khmer,whileone-thirdare in Sanskritalone,and a quarter
utilize both languages. But one invariablefeatureof them all is the linguistic
hyperglossiawe find in India: Sanskrit,and never Khmer, makes expressive
statements; Khmer(and rarelySanskrit)makesconstativestatements. When thefame
of the king is celebratedor his lineage or victoriesin battleproclaimed,the writer
employsSanskrit;whentheslavesdonatedto a templeare enumerated, thecatalogue
is givenin Khmer.Moreover,thetwolanguageshad a veryunequalrelationship with
each other.WhereasSanskritis, linguistically, uninfluenced by Khmer-indeed, it
retains an astonishinggrammaticaland orthographicregularityto the end of
Angkor-Khmer is massivelyinvadedbySanskritfromtheearliestperiod.Foralmost
a thousandyears-as the relationshipbetween political inscriptionand literary
literizationmentionedabove would lead us to expect-literatepoetryin Cambodia
is Sanskritpoetry,neverOld Khmer;literateliterary productionin Khmerdoes not,
in fact,seemto existbeforethefifteenth century, or morethana centuryafterAngkor
is abandonedand the last representative of the Sanskritcosmopolisin mainland
SoutheastAsia disappears(Khing 1990, 24-59). The characterof Khmerlanguage
usage in textsthatarepreservedto us and the laterhistoricaldevelopmentofKhmer
literaturetogethersuggestthatthelattercould notcome intoexistence,as a literized
entityforexpressivepurposes,untilSanskritliteraryculturewaned.
The spreadof politicalSanskrithappensnot onlywithextraordinary speed over
vastspace,but in a way thatseemsquite withoutparallelin worldhistory.First,no
organizedpoliticalpowersuchas theRomanimperiumwas involved.No colonization
ofSouthIndia or SoutheastAsia can be shownto haveoccurred;therewereno military
conquests,and no demographically meaningfulmigrations.Nor were any ties of
politicalsubservience, ofmaterialdependencyorexploitationeverestablished.Second,
Sanskritwas not diffusedby anysingle,unified,scripture-based religionimpelledby
religiousrevolutionor new revelation,but by small numbersof literatiwho carried
with them the verydisparate,uncanonizedtextsof a wide varietyof competing
religiousordersas well as textsof Sanskritliteraturehaving no religiouscontent
whatever.Third,Sanskritneverfunctioned as a linklanguagelike othertransregional
codes such as Greek,Latin,Arabic,Persian,Chinese.In fact,nothingindicatesthat
in thisperiodSanskritwas an everyday mediumof communication anywhere, not in
South let alone SoutheastAsia, or even functionedas a chancerylanguage for
bureaucratic or administrative
purposes.
What is createdin the periodthatcoversroughlythe millenniumbetween200
or 300 and 1300 (whenAngkoris abandoned)is a globalizedculturalformation that
seemsanomalousin antiquity.It is characterized by a largelyhomogeneouspolitical
language of poetryin Sanskritalong with a rangeof comparablecultural-political
practices (temple building, city planning, even geographical nomenclature);
throughoutit-to extendOliver Wolters'wordsas theydeserveto be, to the whole
of this cosmopolitanworld-elites in differentrealms shared "a broadlybased
communalityof outlook" and could perceive"ubiquitous signs" of a common,a
Sanskrit,culture(Wolters1982, 43). But it is producedand sustainedby noneofthe
 THE COSMOPOLITAN VERNACULAR 13

forcesthat operatein the othertranslocalformations of antiquity;it is periphery

withoutcenter,community withoutunity.One maywell wonderwhatthisglobalized
culturemeantifnoneof the familiarmaterial,governmental, or religiousconditions
of coherencepertainedto it. What culturalwork,forinstance,was performed by the
ubiquitousSanskritliterary textsinscribedand displayedby rulingelites?Sincethey
emergedfromthe verycentersof authoritythroughoutthis world,it is naturalto
factorthe political into any explanation,but it seems to be the political with an
obscure,unfamiliarlogic to it.
Even as we trygrasp this logic, the predicamentof theorizingthe premodern
fromwithina conceptualapparatusbequeathedby modernity loomsbeforeus. There
has largelyprevaileda singleparadigmforunderstanding the social foundationsof
Sanskrit cosmopolitan culture, namely, legitimation theory and its logic of
instrumental reason:Elites in commandof new formsof social powerdeployedthe
mystifying symbolsand codes of Sanskritsomehow to secure consent.But this
functionalistexplanationis not onlyanachronistic, but reallyis a mereassumption,
and an intellectuallymechanical,culturallyhomogenizing,and theoretically naive
assumptionat that.6
Ifwe contemplatetheSanskritecumeneat its height,fromthemiddleto thelast
fewcenturiesof the millennium,it appearsto consistof a limitednumberof large-
scale agrarianpolities (and their smaller-scaleimitators),"military-fiscal" states
gatheringtributefromlarge multiethnicpopulations,and definingtheirpolitical
aspirationsas universalist.Althoughnotoriously to definein concreteterms,
difficult
"empires"-the nameusuallygivento theworldsof the Guptas,forexample,or the
Gurjara-Pratiharas,or Angkor-seem to sharecertainsystemicculturalfeatures. One
mayevenpostulatean empire-system or empire-modelofpremodernity, a fieldas it
wereof the reproduction of empiresand of the deploymentof the empireform-in
thislike the systemof nation-states of modernity, wherethe structure of the system
itselfproducesa numberofculturaleffects (Balibarand Wallerstein1991, 91)-with
its own distinctiveculturalrepertory.
In thissystemimitationof an imperialformseemsto be successivelyrecreated,
not onlyin South and SoutheastAsia but elsewhere,both horizontally acrossspace,
perhapsthrougha processsimilarto whatarchaeologists call ''peerpolityinteraction,"
and verticallyin time throughhistoricalimagination.One could plot such a form,
on both axes, among a range of embodiments:Achaeminid(and Sassanian,and
Ghaznavid), Hellenic (and Byzantine),Roman (and Carolingian,and Ottonian),
Kushan (and Gupta, and perhapsAngkor)(see also Duverger1980, 21). In manyof
thesecases,qualifyingas empire,whetherimperialgovernancewas actuallyexercised
or not,seemsto have requireda languageof cosmopolitancharacterand transethnic
attraction,transcending or arrestingany ethnoidentity the rulingelites themselves
mightpossess. It had to be a language capable of makingthe translocalclaims-
howeverimaginarythesewere-that definedthepoliticalimaginationof thisworld.
Moreover,it had to be a languagewhosepowerderived,not fromsacralassociations
but fromaestheticcapacities,its abilityto make realitymore real-more complex
and morebeautiful-as evincedby its literaryidiom and style,and a literary history
embodyingsuccessfulexemplars of such linguisticalchemy. In the "Roma renovata"
of Carolingianand Ottonian Europe this language was Latin, which, though in
constantneedofrehabilitation, was retainedand reinforced as a crucialcomponentin

6Thenotioncontinuesto shapeworkon stateformation and culturein Southand Southeast

Asia, cf.e.g., Kulke 1993, and contrastPollock 1996, 236ff.
14 SHELDON POLLOCK

the politicaland cultural-politicalunderstanding of polity.In West Asia fromA.D.

1000 on, it was New Persian,whose firstgreatliteraryproduction,the Shahnama,
soughtto link the new politicalformations with an imaginedIranianimperialpast,
and along with otherbrilliantworksof literaryculturemade it the language that
rulingelites fromSistan to Delhi adopted perforceif theywere to participatein
"imperial"culturalpolitics,regardlessof what theymay have spoken in private.
Similar in its cultural-politicallogic to Latin and Persian,as in its temporaland
geographicspread,was Sanskrit.
More than just qualifyingthe polity for imperial status, however,Sanskrit
mediateda set ofcomplexaestheticand moralvaluesofimperialculture,whileat the
same time providinga code forthe expressionof key symbolicgoods-the most
importantamongthesebeingfame-in a wayno otherlanguagewas apparently able
(orpermitted)to do. The sourceofsuchcapabilitiesis to be locatedin thesophisticated
and immenselyinfluential Sanskritdisciplinesofgrammar,rhetoric, and metrics.
Imperiallanguage typicallypresupposedthe dignityand stabilityconferred by
standardizinggrammar.Only in a language constrainedby such a grammarand
thereforeescaping the danger of degenerationcould fame and distinctionfind
enduring expression. But there is more to grammaticalitythan such quasi
functionalism in the Sanskrittradition,somethingdeeper rooted.If the orderof
Sanskritpoetrywas tied to the orderof Sanskritgrammar,that orderwas itselfa
model or prototypeof the moral,social,and politicalorder.A just (sddhu)king was
one who himselfused and promotedthe use of correctlanguage (sddhusabda). Not
only was Sanskrittherefore the appropriatevehicleforthe expressionof royalwill,
but Sanskritlearningbecamea componentofkingliness.This is demonstrated bythe
numerousoverlordswho-from our Rudradamanin south Gujarat in A.D. 150 to
Siiryavarman II on Tonle Sap a thousandyears later-celebrated their Sanskrit
learning,especiallygrammaticallearning,in public poetry,and soughtto confirm
thislearningby patronizingthe productionof almosteveryimportantgrammatical
workknownto US.7
That the traditionof Sanskritrhetoricand metricswas centralto this whole
processis evidencedby the inscriptional poetryitself.But the textsoftheseformsof
knowledgealso circulatedas somethinglike globalized culturalcommodities,and
wereeventuallyto providea generalframework withinwhicha numberofvernacular
poetriescould themselvesbe theorized.Thus, forexample,the late seventh-century
rhetoricaltreatiseofDandin,the"MirrorofLiterature" [KAI), wasstudied
(Kdvyddars'a
and adaptedduringtheperiod900-1300 fromSri Lanka to Tamil countryto Tibet.
One could write an equally peripatetic account of metrical texts, such as
Kedarabhatta's"JewelMine of SanskritMeters"(Vrttaratndkara, ca. 1000). By way
of its twelfth-century Pali translationVuttodaya, it played a definingrole in the
creationof Thai poetryat the Ayutthayacourtin the seventeenth century(Terwiel
1996). It is instancessuch as thesethathelp us gauge the extraordinary importance
thattheinstruments ofSanskritculturalvirtuosity and their
possessedforintellectuals
mastersthroughout Asia.
As a resultof all this,Sanskritliteraturein general(kdvya)and politicalpoetry
in particularpossessa uniformity
(pras'asti) thatgivesa clearstylisticcoherenceto the
cosmopolitanculturalform.For withoutdenyingsome local coloring(thoughfor

7SeePollock 1996, 240 forreferences. HartmutScharfewas the firstto perceivea pattern

of royalpatronage(1977, 187), but it is fardenserthanhe knowsand his examplesare easily
multiplied.
 THE COSMOPOLITAN VERNACULAR 15

Angkor,forexample,thishas been exaggerated, cf.Wolters1982, 91), to participate

in the cosmopolitanordermeantpreciselyto occlude local difference. The Sanskrit
poethere-this is theinsistentimplicationoftheform,style,idiom,and evencontent
ofthousandsofinscriptional as well as morestrictlyliterarytexts-participatedboth
by theoreticaltrainingand literary practicein a transregionalculturalspheresimilar
to thatofhis Latin(and,I wouldguess,Chinese)peersat theotherendsoftheancient
world.8It is thisthatmakesit oftenvirtuallyimpossibleto localizeor date a workof
Sanskritliterature-which,by my argument,is exactlywhat constitutedone of its
greatestattractions.
Thereis no doubt fargreatercomplexityto theinteractions ofpowerand culture
in the SanskritcosmopolisthanI can capturein my briefaccount,or perhapseven
know. Yet it is arguable that imperial-cultural associationsand aestheticstyle,
especiallyas theseshapedpoliticalvocabularyand culture,had at leastas muchto do
with the makingof the cosmopolitandimensionof thisworldand its attractions as
persuasion,let alone misrecognition or mystification.Sanskritgave voice to imperial
politicsnotas an actual,materialforcebutas an aestheticpractice,and itwasespecially
thispoetryofpoliticsthatgave presenceto the Sanskritcosmopolis.
At the ideationallevel, the Sanskritcosmopolisfoundexpressionabove all in
certainrepresentations of the space of culturalcirculation.Two of theseneed to be
introduced,given theirrole in the theoryand practiceof literaryvernacularization:
the epic space ofpoliticalaction,about whichI will be verybrief,and the spacesof
literarystyle,whichneed some detail to make understandable.

Political Space in Cosmopolitan Vision

It is an insistentconcernofa wide varietyofkdvyaandprasastitextsto projecta
meaningfulsupralocalspace of political-cultural reference.The tenth-century poet
Rajasekhara,forexample,court-poetto the kingsof Tripura,was repeatinga long-
standingcommonplacewhendescribinghis patronsas universalrulers"in theentire
regionfromwherethe Gafngaemptiesinto the easternsea to wherethe Narmada
emptiesintothewestern,fromtheTamraparn.in thesouthto themilk-oceanin the
north"(ViddhalabhanJika4.21). So are the Kalachurikings themselveswhen they
repeatthis in theirepigraphs.The source,or at least most articulateforerunner,of
thisvisionis in theitihdsaor "epic" Mahdbhdrata, whereplottingthespace ofa large
world, a zone within which its political action was held to be operativeand
meaningful,is a central project of the narrative(a pure example, thus, of a
"chronotope,"and with the chronotope'spoliticsof space moreclearlyvisible than
Bakhtinhimselfunderstood,1981, 84-258). This unmappedmapping,in a different
but notunintelligible a numberoftheimportant
worldofhistoricalspace,constitutes
narrativejuncturesin the text,frombeginningto end. I describeseveralto give a
senseof thepractice.
On hiswanderings
duringhisself-exile a pathfrom
Arjunacharts Indraprasthanorth
andintotheeastern
toGafigadvara Himalayas,
southeasttoNaimisa,easttoKausikT,

8Jstressliterarypractice;variousSanskritswerein use outsidethe domainof kdvya.But

a widevarietyofPrakritsdivergentinphonology,
whereastraditionalscholarshipdifferentiated
morphology,and lexicon,no such distinctions(with the exceptionofdrsa or archaic,Vedic)
wereperceivedforSanskritin thepost-Paninianperiod(cf.,e.g.,SarasvatkanthAbharandlahkdra
2.5ff.).The comparableworld of earlyLatinityis well describedby A. H. M. Jones 1964,
1008.
16 SHELDON POLLOCK

southeastto Gaya, and further to Vanga, southdown the Kalifiga,overto Gokarna

on the west coast,northto Prabhasaand Dvaraka,northeastto Puskaraand thence
back to Indraprastha(MBh. 1.200-10). Beforehis consecration as emperorYudisthira
sendsout his brothersto conquerthe fourdirections:Arjunaproceedsto the north
(Anarta,Kashmir,and Bactria);Bhima to the east (Videha,Magadha,Anga,Vanga,
Tamralipi); Sahadeva to the south (Tripura, Potana, the lands of the Pandyas,
Dravidas, Coladrakeralas,Andhras; Nakula to the west (Marubhtumi,Malava,
Paficanada,as faras the land of the Pahlavas)(MBh. 2.23-29). Afterthe war,when
the Pandavas performthe Horse Sacrificeto affirmand confirmtheir universal
dominion,the wanderingsof the horse plot a map that runs fromTrigartato
Pragyotisa,Maniptura,Magadha, Vafiga, Cedi, Kdsl-,Kosala, Dravida, Andhra,
Gokarna,Prabhasa,Dvaraka, Paficanada,and Gandhara (MBh. 15.73-85). Lastly,
when theyrenouncetheiroverlordshipand begin their"Great SettingForth,"the
Pandavastravelfirstto the Lauhityariverin the east, "by way of the northern[i.e.,
northeasternlcoastoftheocean to thesouthwestquarter,"thento Dvarakaand from
thereto Himavan,Valukarfnava (the great"Ocean ofSand") and MountMeru(MBh.
17), thusperforming thelastcircumambulation oftheworld-the sortdescribedand
chartedrepeatedlybefore-forthe controlofwhichtheirfamilyhad been destroyed,
and ofwhichtheyfittingly take leave as theyprepareto die.

Thus, from the opening chapters of the principal narrative,and at its key
points-the royal consecration beforethe war, the reaffirmation of dominion afterthe
war, the ritual death-march at the end of the story-the epic insists continually on
concretely placing the action. It is the very fact of the existence of this spatial
imagination in the Mahdbhdrata that interests me, not its precision (indeed, it is
marked by uncertainty, confusion, and at times bizarre exoticism). There is a
conceivable geosphere, the narrativesuggests, where the epic's medium, the culture
of Sanskrit,and its message, a kind of political power, have application.
The spatial imagination that is found in the Sanskrit epics achieves sharper and
more concrete focus in the courtly literaturethat arises in the early centuries of the
common era, as in the "conquest of the quarters" motif appearing in courtly epics.
The most influentialexample, one studied as far as Khmer country,is that found in
Kalidasa's masterpiece, the "Dynasty of Raghu" (Raghuvamsfa 4). Here, the reality
effects,as it were, of the judicious choice of detail are quite apparent. The clearer
image of the spatial domain both of power and, implicitly,of the poetrythat fillsthis
domain and gives voice to power no doubt has something to do with the fact that
Kalidasa borrowed from the Allahabad Pillar inscription of the Gupta king,
Samudragupta (r. A.D. 335-76). It is not that there is something less literary,more
documentaryabout the inscriptionthan the poem (this would be so even if its author,
one Harisena, did not actually name it a kdvya,as he does) that somehow serves, as
model, to render the account of Kalidasa more historical or more "true." Rather, the
point of juxtaposing inscription and text in their historical relatedness is simply to
remind ourselves that the literarygeography of power in Sanskrit culture sometimes
achieved a kind of symmetrywith the living aspirations of historical agents.
However this macrospace may be defined(and note that it did not always embrace
the full cosmopolitan space as mapped by inscriptional and other cultural practices),
and whatever may be the precise nature of the imperial dominion and formof culture
it was imaginatively thought to comprise, it marks a wide range of epic and postepic
texts. And it is against this macrospace that a range of vernacular spaces of culture
and power were to be defined.
 THE COSMOPOLITAN VERNACULAR 17

The Space of SanskritLiteraryStyle

The Rajas'ekharawho wroteoftheuniversalsovereignty

oftheTripurakingsalso
wrotean allegoricalaccountoftheoriginofliterature,
thestoryofthe"PrimalBeing
of Poetry,"or "PoetryMan," Kavyapurusa:

Brahmacreateda sonfortheGoddessofSpeech,hismouthconsisting ofSanskrit,

hisarmofPrakrit, hisgroinofApabhramsa, hisfeetofPaisaca,hischestofmixed
language.Sahityavidya(PoeticsWoman)wascreatedto be hiscompanion, andwas
toldto followKavyapurusa whereverhe shouldgo. Theywentfirst to theeast,and
as Sahityavidyatriedto enticehim Kavyapurusa spoketo her in versesfullof
compounds, andstrings
alliteration, ofetymologically complexwords, whichbecame
knownas thegaudaPath(rLti). Nexthewentnorthto thecountry ofPanicala,
where
hespokeinverses withpartialcompounds, andmetaphorical
alliteration, expressions,
whichbecameknownas thepdocdlaPath.Eventually theyreachedthesouthwhere
he spokein verseswithmoderate alliteration,
no compounds, and simplewords,
whichbecameknownas thevaidarbha Path.
(Kdvyam2mcmsd 4)

Rajasekhara'sallegoryofliterature,brieflysummarizedhere,picksup several
themes already noted, including the geoculturalspace present to the Sanskrit
imaginationand the restrictions on the possible codes in which the literarycan be
composed.I cite thispassage,however,to introducethequestionofthetransregional
geographyofliterary style.Therewas a prehistory to Rajasekhara'saccountofmdrga/
rTti-the "Way" or "Path" of literaryculture-a somewhatconfusedand tangled
historyin its firstmanifestation, but reasonablystraightforwardin its development
by the tenthcentury.
Mdrga(thedominantand foundational term)carriestwoprincipalmeanings.The
firstis that of a "way" othershave gone before,and thus connotesa "custom"or
"tradition"of writing.Like the Greek odos("way"), mdrgaalso comes to imply
somethingof a method or a "followingof a way" (meth-odos) in the creationof
As a termin the Sanskritliterary-critical
literature.9 vocabularyit has a momentof
primacyin the seventhto tenth centuries-the Kashmiri theoreticianVamana
announcingin theearlyninthcenturythat"the Path is to literature as thesoul is [to
the bodyl"-and thoughit was eventuallyto cede thisposition,it remainsa crucial
termin the theorization of both cosmopolitanand vernacularformsofwriting.And
althoughthismayseemto be a narrowissueofphilologicalinquirygivenitsformalist
focus-for the Way concernsthe languagestuffof literature-wedo well to bearin
mindhow seriouslysuchquestionsweretakenby intellectualsacrossthegreaterpart
of southernAsia forcenturies.
As we see fromthe accountof Rajasekhara,the Way of Sanskritliteratureis
conceptualizedas plural and regional:thereis an "eastern"way (gauda, loosely,of
Bengal), a "southern"way (vaidarbha,of Vidarbha),a northernway (pdigcdla, of
Paficla, the northGangeticplain), latera westernway (Idt'y-a, of Lata or southern
Gujarat), and still later others.What differentiates these nominallyregionalized
proceduresof literatureare certainqualities of language use (guncas) at the level of
phonology(e.g., phonemic texture),syntax(e.g., degree of nominalization),and

9Forthe firstconnotation,cf.,e.g., Manu 4.178; forthe second,e.g. SRK 1729, 1733;

Vakpatiraja(ca. A.D. 730), Gaidavaho 84-85.
18 SHELDON POLLOCK

lexicon(e.g., the relativeprevalenceof primary[,rfdhijor derivative[yoga]Iwords).

Dandin in the late seventhcenturydefinesvaidarbhaas "endowed with all the
qualities,"whereasgaudcais characterized by theirinversionor absence(viparyaya).10
The formerthus shows a minimaldegreeof compoundingand of complexlexical
the lattera maximaldegreeof both.
derivatives,
FromthebeginningtheontologyoftheWaysofwritingis implicitlyorexplicitly
queried,and the generalunderstanding is thatwriterscould freelyadopt the one or
theother.ForVamana"theregionalappellationsmeanonlythatthesestylesarefound
in [thepoetsof]thoseparticularregions;theregionsthemselves contribute nothing."
One could and shouldchosethevaidarbhastyle(Kdvydlakdrascstra 1.2.6-10; 14-18).
Althoughhis remarks(like muchofhispresentation) aremorethana littleconfused-
fortheyexplainnothingabout whyregionalstylesshouldbe foundamongthepoets
in given regions-there is no ambiguitythat forhim regionwas not destiny,as it
was not,a fewcenturieslater,forthe criticKuntaka:
Ifdifferentiation
ofstyleweretrulybasedon thatofregion, theformer wouldbe as
numberlessas thelatter.Justbecausewriting a certain
exhibits ritidoesnotmeanit
as a regional
canbe classified custom,likecross-cousin
marriage. . . Furthermore,
it
cannotbe said to be a "natural"propertyin thesamewaythatcertainbeautiful
sounds,timbre,etc.,arenaturalto thesingingofa southerner.
1.24)
(Vakroktijivita
For mostofSanskrithistorywritersvoluntarily could adoptone styleor another.
The eleventh-century poet Bilhana,forexample,anotherKashmiri,tells of himself
thathe writesin vaidarbha("a rainofnectarfroma clearsky. .. guarantorofliterary
beauty-vaidarbhais grantedto only the finestpoets," Vikramdigkadevacarita vs. 9).
And, in fact, the freedomto choose from among regional styles grew into a
requirement as thedoctrineoftheWays was linkedevermorecloselyto thediscourse
on literaryemotions(,rasa):As theaffectivestateto be generatedin a sceneorpassage
varied,so would the Way. Thus forthe ninth-century writerRudrata,thevaidarbha
and pdAcd1aPaths are appropriatefor the moods of "love," "pity," "fear,"and
"wonder";the Ways themselveshe classifiesas anubhdvaor the verbalreactionsof a
characterin differentemotionalsituations(RudrataKdvydl/aikdra 15.20).
On the discursiveplane what the categoryof the Ways most insistently
communicatesis in factthe verycosmopolitanism of Sanskritliterature."Regional"
differencesarepartoftherepertory ofa global Sanskrit,thesignpreciselyofSanskrit's
transregionality:Theywerelocal coloringsthatwereproducedtranslocally, and thus
werean indexof Sanskrit'spervasionof all local space. Eventually,as we will see, it
is preciselythis implicitsense of the Way of Sanskritliteratureas a cosmopolitan
(ratherthan trulyregional)culturalformthat would be made explicit by a new
dichotomycentralto vernacularpoetriesthatarosein the late medievalperiod:Over
against mdrgaor the global Way of well-traveledSanskritculture came to be
constructed thedesior Place, thatwhichdoes not travelat all.
The Sanskritcosmopolis,createdin South and SoutheastAsia in a moreor less
simultaneoushistoricalprocess,possessedmarkedculturalsimilarities,such as the
productionof a code forpoliticalexpressionand of a literature whereadherenceto a

10Theevaluativejudgmentimplicithere,and the verydistinction,appear to have been

resistedas earlyas Bhamaha (Kdvydlankdra 1.31ff.),though the eleventh-century Kannada
writerNagavarmantakesBhamahato mean not thatthe north-south distinctionis meaning-
less,but thatthebeliefthattheone is superiorto theotheris mistaken(Kdvydvalokanam,s7tra
522).
 THE COSMOPOLITAN VERNACULAR 19

sophisticatedbodyofnormative discourseson grammar,rhetoric, and metricsensured

a uniformcharacterthroughoutthe cosmopolitanformation. The monopolizationof
literaryproductionin transregionalcodes was matched at the level of literary
representationby theprojectionofa supralocalframeofpolitical-culturalreferencein
and at the levelof literarytheoryby a doctrineofmodes
epic and postepicnarrative,
of writingwhose regionalityconnotesabove all Sanskrit'stranscendence of region.
These are among the keycomponentsof literaryculturethatwill be engagedin the
vernacularization
process.

ProducingtheVernacular

Few local literaryculturesof premodernity anywhereshow quite the same self-

consciousnessand permitus to followtheirdevelopmentwiththe same precisionas
we can achievein the case of Kannada,a languagefoundin what is now the Indian
stateofKarnataka.I wantbriefly to sketchthehistoryofKannadain theinscriptional
record,beforegoing on to consider in more detail the intense and long-term
negotiationbetweencosmopolitanand vernacularin Kannada literary production.
The statusof Kannada in the domain of the publiclydisplayedinscribedtexts
offersa textbookcase of the tendenciesdescribedabove. The earliestknowndynasty
of northwestern Karnataka-the locus of what was to become the prestigeliterary
dialect-the Kadambas (fourthcenturyon), neverused Kannada forpublic records.
The Gafigas,the oldestattesteddynastyin southwestern Karnataka(fourthto ninth
centuries), did not use Kannada for the documentary portionof copper-plategrants
until the time of Avinita in the sixth century.We are able to followthe literary-
culturalpolitics of Karnatakakingdomsmore closely,however,with the Badami
Calukyas,and especiallywiththeirsuccessors,the Rastrakuitas. What we findamong
the latter,whenwe look at the matterstatistically, is a slow but stunningdeclinein
the productionof Sanskritpublic poetrycommencingin the earlyninthcentury.
When thedynastyfirstbeginsissuinginscriptions startingaroundA.D. 750, Sanskrit
is used in morethan80 percentof the extantrecords;by its end 200 yearslater,less
than5 percentare in Sanskrit(Gopal 1994, 429-65).
Besidestheclearevidenceofshifting languagepreference, all theearlyinscriptions
in Kannada among the Badami Calukyas and Rastrakuitasremain resolutely
documentary. The firstexpressiveor "workly"inscriptions in Kannada fromwithin
the royalcourtcome to be producedonlyabout the timeof the reignof KrishnaIII
(939, EI 19, 289), or nearlyhalfa millenniumafterinscribedKannada firstappears
(Halmidi ca. 450).
It is notmanygenerations beforeKrishnaIII thatevidencefortextualizedliterary
productionin thelanguageis firstavailable,duringthe reignof the Rastrakuita king
NrpatunigaAmoghavarsa(ca. 814-80). In terms of literaryculture, this was a
remarkableperiod and in a
place manyrespects, site of what appearsto be literary
experimentation acrosslanguages.It was then,forexample,thatJainasturndecisively
to Sanskritfortheproductionoftheirgreatpoetichistories(as in theAdipurdna [A.D.
8371 of Jinasena II, the spiritual preceptor of Nrpatuniga, or Asaga's
Vardhamdnapurdna [8531, the first independent biography of MahavTra),and
undertooktheirfirstgrammaticalanalysisof Sanskritin perhapsfivecenturiesin the
20 SHELDON POLLOCK

?abddnufsdasanaof ?Skatayana.1IHere, too, a littlelateran importantnew currentin

Apabhramrsa, as we have seen the thirdcosmopolitanliterarylanguagealong with
Sanskritand Prakrit,findsexpressionin the workof Puspadanta(fl. 950), who was
probablythe firstto write a Jaina universalhistoryin the language.12 But the
historicallycrucialinnovationin literarycultureconcernsKannada.
No doubt attemptsto produceliterarytextsin Kannada precededthe periodof
Nrpatuiiga.In theterritorial imaginationofKannadaliterary culturethroughout the
medieval period, the "heartlandof Kannada" ("the very zone (nddu-e)between
Kisuvolal [Pattadakall, the renowned city of Kopana [Koppall, Puligere
[Lakshmeshvarl, and Omkunda[Okkundain the BelgaumDistrict]. . . is wherethe
veryessence[tirullof Kannada [is found]"[KRM 1.381),in otherwords,the royally
sanctionedprestigedialect,is placed not in northeastKarnatakawhereGovinda II
and his son Nrpatuniga built theircapital,but 250 km to the southwest,in the core
regionof thepredecessordynastyof the Calukyas."3Yet even if thiswerebecauseof
the presenceof a new Kannada literaturein Badami and Aihole,thiswould take us
back onlya fewgenerations-which,in fact,is about as faras the literary-historical
memoryof Kannada poets themselvesreaches,as this is embeddedin introductory
(the earliestauthorsmentionedare Asaga and Gunavarmaof the early
kaviprars'amsras
ninthcentury).The firstextanttextin Kannada describeshow difficult a task it is
forthe authorto identifyliterarymodelsforthe prescriptive projectbeforehim: he
is forcedto "huntforscraps"of Kannada literaturelike a mendicant:

Both Sanskritand Prakritare availableaccordingto one's wish (bagedante) for

composingliterature withrefinement (samari),forto be surethereare already
availablebothliterary
modelsandrules(laksya,laksana)in greatabundance foreach
ofthetwo.But thediscourse I presenthere[requiresi beggingscraps(irikoregozdvu)
to makeit intelligible.
[sc.,ofKannadaliteraturel It is thusdifficult
foranyoneto
do in thecaseofKannadathewaytheancientteachers [ofSanskritandPrakritdid).
(KRM 1.41-42)

Kannada literature(in thesenseI have beenusingthetermthroughout) was

a recentinvention,of perhapsthe eighthcentury,and it is preciselythe factof its
noveltyin thefaceofSanskritthatpromptedthewriterofthistextto puzzle through,
in a mostdetailedand subtleway,thecomplexdialecticbetweenthelocal and global
in medieval literaryculture. This singular work in the history of literary
vernacularizationis theKavirdjamdrga (ca. 875), "The Way of the King of Poets,"a
textto place besideDante's De vulgarieloquentia(1307)-or, rather,beforeit; it may
in factbe the firstworkin worldcultureto constitutea vernacularpoeticsin direct
confrontation with a cosmopolitanlanguage.14 There are considerablecultural-

"1HestyleshimselfAbhinavas'arvavaram in recognitionoftheearliermodel(Sarvavarma's
Kdtantra),and namesthe autocommentary on his grammarAmoghavrtti afterhis patron(men-
tionedin 4.3.208). The Jainaturnto Sanskritforkdvya-and JinasenaII clearlyregardshis
Adipurdnaas such-needs study,especiallythe earlyworksof Ravisena(678) and JinasenaI
(783). For a generalaccount,see Dundas 1996.
12Literaryproductionin Prakrithas been thoughtoddlyabsent(cf.alreadyAltekar1960a,
412), but as notedabove it had becomea residualor even archaicculturalfeature,as inscrip-
tionaldiscoursefromthe mid-fourth centuryon demonstrates.
"See KRM 1.37; Pa'mpaVAV 14.45. Cf. ChidanandaMurti 1978, 256.
14The Tamil Tolkdppiyam is no doubt earlier(its dating is much disputed;forone sober
assessmentsee Swamy1975), but thedichotomyoperativethereis notcosmopolitan/local but
standard/nonstandard, centamil[kotuntamill(Zvelebil 1992, 134-36).
 THE COSMOPOLITAN VERNACULAR 21

historicalparallelsbetweentheseworks,but also somesignaldifferences. At themicro

level,unlikethe Eloquentia, theKRM aims to producenot a unifiedlanguageforthe
polityfromamongcompetingdialects,but a languagequalifiedforliterature. At the
macrolevel,theKRM has a lesstransparent relationship
thanDante's workto political
theoryand practice,but its social locationand authorshipare clearand important.It
was writtenat thecourtofNrpatuniga and underhis guidance:the"Way oftheKing
of Poets" is the Way of Nrpatungahimself.15
Despite theimportance ofKRM forthecultural-politicalhistoryofmiddle-period
India, thereexistsno criticalanalysisor evendescriptiveaccountof theworkin any
language otherthan Kannada. Even Kannada-languagescholarshiphas not always
appreciatedits largerhistoricalsignificance.While Kannada in generalis unjustly
ignored everywherein South Asian research,Old Kannada (Halagannada), the
languageofthisand all literatureoftheregionbeforetheVTras'aiva culturalrevolution
at the end of the twelfthcentury,is understudiedeven in Karnataka-in largepart
becauseit is hardlyaccessiblewithoutknowledgeof Sanskrit.This paradoxicalfact,
like the text'srelationshipto the traditionof Sanskritpoetics,especiallyDandin's
"Mirrorof Literature," are two importantindicatorsof whatvernacularintellectuals
writingin Kannada weretryingto do. We have seen thatthe circulationof textson
Sanskritpoeticswas botha factorand a signofthecreationoftheSanskritcosmopolis
in Asia, and at thesametimeprovideda framework withinwhichlocal poetriescould
be conceptualized(in Siam,Sri Lanka,Tibet,and so on). The sameprocesstookplace
in the subcontinentitself,firstand nowheremore profoundlythan in Kannada
country.

MakingtheGlobal Local: theKavirajamarga

and theWays ofLiterature

The KRM fullyrecapitulatesthe structureof Dandin's "Mirror"and in some

importantwaysevenfunctionsas our oldestcommentary on the text.It firstdefines
literature, thatmarit (dosas)and makeit beautiful(gugnas)
describeslinguisticfeatures
(chap. 1), and thencataloguesfiguresofsound(chap. 2) and sense(chap. 3). In addition
to similarityin structure, perhapstwo hundredof the illustrativeversesare closely
adaptedfromSanskritantecedents.But theworkis not a translation of theSanskrit,
as oftenassumed.Not onlydoes "translation" as usuallyunderstoodmakeno cultural
senseforthisworldwhereliteracyin Kannada presupposedliteracyin Sanskrit,but
the workhas a quite different agenda fromits Sanskritmodel. What we are being
offeredin the KRM is an experimentin the localizationof a universalistic Sanskrit
poetics and an analysisof Kannada literaryidentity.Conversely,however,it has
somethingof interestto revealabout the creationof thispoetics,and about the real
dynamicsof local-globalexchange.I wantto illustratebothfeaturesby an analysisof
somethingthat has long confusedstudentsof the KRM: its appropriationof the
Sanskritdiscourseon the Way of literature.
The KRM firstintroducesthe categorymdrgain its broaderconnotation, literary
method,somethingcoded in the veryname of the work,Kaviradj'amdrga, "The Way
oftheKing ofPoetry.""Way" becomesa coveringtermfor"good literature," as such
(contrasting with "corrupt"poetry,dusya,2.7-8, so Jinasena,Adipurdna1.31; 208-

'5KRM 1.44, 147, etc. The actual redactorof the workwas a poet namedSrTvijaya.
22 SHELDON POLLOCK

9); "literatureof the Way" is the supremeuse of language,in all its formaland
aestheticcomplexity:

The manwhounderstands languagecan communicate

withothers,
disclosinghis
thoughts Wiserthanhe is themanwho can communicate
as he intended. large
meaninginbrief andwiserstillthemanwhoknowshowtomakehiswords
compass,
unitewithmeter.Morelearnedthanall is themanwhocanproduceworksofthe
greatWay(mahddhvakrtigal).
(KRM 1.15-16)

This is a perfectly
intelligibleusage. What has beenfoundpuzzlingis theKRM's
next move of adopting the notion of the regional Ways-whereby Sanskrit
demonstratedits pervasionof all literaryspace-for a differentiationof Kannada
poetryitself.
It is impossible fully
tocomprehend theprocedures oftheWayandreacha conclusion
aboutthemultiplicity oftheiroptions.Havingconsidered therulesonwordsofthe
earlier I willsaya littlewithrespect
sastras, toKannadaso thatthematteringeneral
maybe clear. . . Poetsarisein a worldwithoutbeginning in
andthusareinfinite
number, theirindividualized expressionsareofinfinite
kinds,andso theWayexists
in infinite variety... But to the best of my abilityI will discussbriefly the
distinction-their differences perceived bytheold [Sanskritjwriters
whoconsidered
thematter-between thetwoexcellent Ways,thenorthern inthe
andthesouthern,
manner I understand it. .. Ofthesetwothesouthern Wayhastenvarieties,
according
tothe[tenlanguagefeatures, Wayhasvarieties
gunasj. . .The northern differentiated
bythepresence oftheinverse ofthesefeatures.
(KRM2.46,49-51, 54-55)

This is followedby exhaustiveinventoryand illustrationof all the language

qualities taken over fromthe Sanskrittradition,which the author concludes is
foundational to Kannadapoetics:"Whateverthewordsemployedin a poem theywill
enhancethevirtuesofKannadaifmade subjectto thedifferent usagesassociatedwith
the Ways describedabove" (2.101). The KRM, in short,appearsto have completely
graftedthe discoursethat makes Sanskritcosmopolitan-the universalrepertory of
styles-onto the local worldof Kannada.
Modern Kannada scholarshave foundthis entireinquiry(of which thereis a
reprisein thesecondimportant medievaltexton Kannadapoetics,theKdvydvalokanam
of Nagarvarmaca. 1040, at the courtofJayasimhaII of the KalyaniCalukyasto be
not onlyirrelevant to actual Kannada poetry,but incoherent.No advancewhatever
has been made overR. Narasimhachar's impatientdismissalof the whole question:
"Northern"and "southern"in Kannada poeticsrefermerelyto the "schoolsor styles
in Sanskrit,"we are told, forthereis no evidencethatanythingcomparableexisted
in Kannada (1934, 121-22). Such a judgmentofcourseexplainsnothingofwhatthe
KRM intendsbyusingthediscourseon theWay foritsanalysisofKannadaliterature,
yet theredoes seem to be everyreasonto interpretit as alien and even meaningless
to a local literaryculture.Designed to reaffirm the real transregionality
of Sanskrit
literaturepreciselyby identifying quasi-regionalvarietiesthe madrgasappear to be
incongruously if not ludicrouslypasted onto a real regionalworldof Kannada. The
categorycapturesnothingwhateverin the local characterof the literatureand fits
onlyto the degreethisliteraturemimicsSanskrit.
The KRM is a textemergingfromthe verycenterof one of the mostpowerful
political formationsin middle-period India (cf. Inden 1990, 228ff.), and this fact, if
 THE COSMOPOLITAN VERNACULAR 23

no generalprincipleofhermeneutic charity,
shouldinviteus to ponderseriouslywhat
it meansbyusingthetalkofcosmopolitanSanskritto represent a vernacular-language
poetics. Metadiscursivelyone might argue that, faced with exclusion fromthe
transregionalityofSanskritand refusingto be caughtin thebracketsofthelocal,the
KRM seeksto remapthecosmopolitanWay ontothelocal worldofKarnataka.There
must therefore be a northernand a southernstyleof Kannada poetryitself-the
KannadaNadu mustbe shownto embracea northand a south,to constitute a regional
world-whetheror not such a divisioncorresponds to any reallyexistingpoetries.16
If Kannada is to participatein the worldof the literary(kdvya),a worlddefinedby
Sanskrit,it mustshowits characteristic features.In a word,the local mustevinceits
translocalcapacities.
An accountof thissortmaycapturesomethingof the cultural-political impulse
at workin theKRM, and otherevidenceI look at below seemsto corroborate it. But
there is anotherand more significant,if somewhatmore complicated,rationale
underpinning it. We beginto graspthiswhenwe considerhow theKRM differs from
and supplementsits Sanskritmodels. First,it renamesthe Ways as "north"and
"south"(the categoriesgauda and vaidarbhabeingofcourseimpossibleforKannada),
and therebymoderatesthe narrowlyspatial implicationsof the taxonomy.17More
importantis the distinction-whichfromthe vantagepoint of standardSanskrit
poeticsseemsodd enoughto constitutea categoryerror-that the KRM introduces
in distinguishing theWays accordingto thetwomaindivisionsofSanskritrhetorical
practice,indirectand direct("natural")expression(vakrokti and svabhdvokti):

TwoWaysaccordingly cameintoprominence, andwiththemtwodifferentforms of

expression,the indirect(vakra)and the direct(svabhdva). is an
Directexpression
invariable ofthesouthern
characteristic Way.Indirect ofmanyvarieties,
expression,
is foundin thecelebratednorthernWay.
(2.52-53)

For the Sanskrittradition,as we have seen, the Ways are differentiatedby the
presenceor "inversion"or absenceof certainlanguagefeatures(gugnas)at the level of
phonology,syntax,and lexicon. Yet here anotherdichotomyis introducedthat,
though largelyunspoken in that tradition,finallyhelps make the whole thing
intelligible:Southernpoetryis devoid of tropes and thus makes prominentthe
languageof literaryexpressionitself,whereasnorthern poetryreliesmoreon figures
of speech(the "manyvarieties"referred to above). Althoughthereappearsto be a
faintawarenessofthisfundamental distinctionearlierthantheKRM, we findit clearly

16The differentiation,it should be noted,reflectsno dialect divisionbetweennorthand

south in Old Kannada. Ganga poets in the south and Calukya poets in the northused a
homogenizedliteraryidiom, producingand reproducedby the philologicalworkdiscussed
below. (The Kannada Nighantu,s.v. uttaramdrga, therefore is mistakento gloss "uttarakan-
nada.")
'7"North"and "south" are used preferentiallyby Dandin's tenth-century commentator,
Ratnas'rTjnina(so, occasionally,by Dandin himself,KA 1.60, 80, 83). Ratna composedhis
commentary somewherein the Rastrakutaworld,his patronbeing one Sarvabhyunnatar-as-
named ?rTmattunfganaradhipa.
traktutatilaka And it appearsthat the two otherextantcom-
mentatorson Dandin workedin the Karnatakaregion(if the one, Vadijafighalais the Vadi-
ghafighalaBhatta mentionedin a tenth-century Gafiga grant [Annual Reportof theMysore
Archaeological
Dept.,19211 as niravadyasdhityavidycvydkhydnanipuna (1. 168); and iftheotheris
theTarunavacaspatiwho workedat thetwelfth-century Hoysala court).Evidentlyit was a text
thatspoke to southernintellectualswithspecial forcefulness.
24 SHELDON POLLOCK

articulatedonly in a somewhatlaterwork,the ?rggdraPrrakdfa of King Bhoja (first

quarteroftheeleventhcentury):"Therearethreesourcesofbeautyin poetry:Indirect
expression(vakrokti),direct expression(svabhdvokti), and expressionof emotion
(rasokti).Indirectexpressionis whenprominenceis givento figuresofspeech,simile
and thelike; directexpression, whenit is givento languagefeatures(gu?nas)" (678).18
How deviantfromtheSanskrittraditionthiscorrelation-ofguncas and thusvaidarbha
stylewith svabhdvokti, and vakroktiwithgauda-is thoughtto be appearsfromthe
wordsofBhoja's editor,who foundit altogetherunintelligible(Raghavan1963, 136-
37; it is in factunknownto Indologicalscholarship).In thelightofKRM it becomes
clear.
The logic ofargumentbothin KRM and oftheexamplesit adduces19producesa
geographyofKannadastylesthat,strippedto itsessentials,comprisesa realdichotomy
of practicesforvernacularwriters:(a) "southern"Kannada literatureis thatwhich
focalizeslanguage itself(literatureas "speech-directedspeech"), and accordingly
employsfiguratively unadorneddescription(the primarymeaningof svabhdvokti),
whereas"northern"Kannada literaturefocalizesrhetoric(vakrokti);(b) among the
mostdistinctive linguisticfeatures
listedamongthegu?nas is degreeofnominalization:
"southern"Kannada literatureis uncompounded;"northern"Kannada poetryis the
reverse;20 (c) "southern"Kannada literature is markedby theprevalenceoflocal (desi)
words (the analog of primarylexemes); northernpoetry by the prevalenceof
unmodifiedSanskritloans (samasamskrta [tatsamain othertraditions],the analog of
derivativelexemes).
The northern and southerntypesof Kannada literaturethusprefigure whatwere
eventuallyto be namedliteratureof the Way and literatureof the Place, magrga and
desi.Far fromanalyzingKannada againstan irrelevant set of categories,the KRM is
identifying thetwomodesofwritingthatconstitutethefundamental identitychoices
forKannada, and in factforall South Asian regionalliteratures.But thereis an
additionaland tellingironyin the dialecticof cosmopolitanand vernacular:For the
sourceof this organizingtaxonomyof Sanskritpoetrywould appear to lie not in
anythingto do with the natureof Sanskritpoetryas such,but ratherin underlying
inclinations ofsouthernpoets-such as Kannadapoetslike thoseat Nrpatufiga's court
or Tamil-bornpoets like Dandin himself-to writeSanskritin conformity with the
sensibilitiesof thesouthernlanguagesthatare finallymade visibleby theproduction
ofpoetryand poetictheoryin thevernacular.21 In theprocessoffullvernacularization

18Bhamaha regardsgauda as alankdravad,and vaidarbhaas avakrokti as well as prasanna,

komala,etc., i.e., endowedwith gunas,but he neverelaborates(Kdvydlankara 1.34-35), nor
does Dandin despite his explicitdichotomy(KA 2.360). Vamana illustratesvaidarbhawith
?dkuntala2.6, perfectsvabhdvokti, and gauda with Mahdviracarita 1.54, perfectvakrokti, but
otherwisegives no hintthathe understoodthe principlesin play.
19Thus KRM vss 2.60 and 62 can be distinguishedon the basis ofsvabhdvokti (southern)
and vakrokti(northern, the le5saku[-Jvalayam as can thetwohalvesof2.110
and otherfigures),
(the firstwithouttrope,the second with metaphorcompound).By contrast,in 2.109, the
operativedistinctionis theplay ofgugasin the firsthalfoftheverseindicatingsouthernstyle,
as opposedto the northern style,whichshowsnothingcomparable.
20Thestatusof ojas was ambiguousalreadyto Dandin, who while listingit as a quality
ofvaidarbhastylemakesit clearthatit is a peculiarfeatureofnorthern poetry(1.80), ofwhich
southerners make only moderate(andkulam)use (83). Vamana eliminatedit as a quality of
"pure" southernstyle(1.2.19), whereasforBhoja vaidarbhais "whollyuncompounded"and
gauda "compoundedto the fullestextentpossible"(?P 580).
2'ForRatna svabhdvokti is "expressionnatural"to southernpoets: "The vaidarbhaWay-
whichconsistsof beauty-factors relatingto wordsthemselves[as words),i.e., the tengunas-
is naturalto southernpoets (ddksindtydndm whereasthe 'eastern'courseofpoetry
svdbhdvikah),
 THE COSMOPOLITAN VERNACULAR 25

that was engaged in ninth-and tenth-century Karnataka,the stylesthat southern

writershad alreadytheorizedforSanskritwere naturallyretheorized as components
ofKannada,ofwhichnoncompounding, initialalliterationfprdsal,
directdescription,
and the like are realcomponents,as anypassageofOld Kannada poetrywill testify.
The largerprincipleto extractfromthis apparentlynarrowcase concernsthe
mutuallyconstitutive interaction
of the local and the global: As the cosmopolitanis
constitutedthroughculturalflowsfromthe vernacular,so the vernacularconstructs
itselfby appropriationfromthe cosmopolitan-a processthat sometimes,as here,
amountsto unwittingreappropriation.22

Philologization
and theProductionof
Difference

The KRM has othercultural-political aims, whichvariouslynuancethe project

of creatinga cosmopolitanidiom while at the same time identifyingKannada
difference.Kannada could not achieve its new rank unless it possessed both the
epistemologicalstatusof Sanskritand the dignityof its philologicalapparatus(i.e.,
laksanagranthasor rule-settingtexts).The KRM achievesthe formerby the veryfact
of engagingin a discourseon Kannada at all, and the latterby the explicitanalysis
ofliterary-languagenormswithwhichthegreaterpartoftheworkis concerned.The
textitselfis moreovera performance of its argument,forit constitutesKannada as a
languageof sciencein theact ofestablishingKannada as a languageof literature (by
contrast,theEloquentiacan onlymake its scholarlyargumentforthevolgare in
illustre
Latin).
The precociouslyearly philologizationwe find in the KRM will continue
uninterruptedly foranotherfourcenturies.Dictionariesarefoundfromtheend ofthe
tenthcentury.A numberof these,like the first,that of the poet Ranna (ca. 990,
fragmentarily preserved)are Kannada-Sanskrit, and glossingas theyoftendo simple
KannadawordswithSanskritequivalentsareaimedless at enhancingcommunication
thanachievinglanguageparity(cf.Nagaraj 1996, 223ff.).Fromthe same periodwe
find the firstin a long series of sophisticatedanalysisof Kannada metrics,the
Chandombudhi or "Sea of Meters" of Nagavarman I. Along with an elaborate
domesticationof the complexquantitative-syllabic metricof Sanskrit,thisprovides
an accountoftheten "native"meters,karntdtavisayabhdsdjdti, "indigenous[meters]of
the languageof the Kannada world" (5. 1). The grammaticaltraditionbeginswith
the Karndtakabhdsdbhisana or "Ornamentof the Kannada Language,"composedin
Sanskrits&trasby NagavarmamII (at the Kalyani courtin northeastern Karnataka
around1040), and culminatesin one of themostimportantgrammarsofprecolonial
India, the ?abdamanidarpanaof Kesiraja (at the Hoysala court, 1260). This
extraordinarywork,whichlike theKRM remainsvirtuallyunreadoutsideofKannada-
languagescholarship, would have to occupya centralplace in anyseriousaccountof
theprocessesofvernacular languageunification and standardization
beforemodernity.

takesnoteof semanticfiguresof speechand grandiloquence"(ad 1. 50). Accordingly, he sees

the differentWays as "inborn,""native," "specific"(tajja, sahaja, nija) to the poets of the
particularregionsjust like theirregionallanguage(on 1.40, p. 28).
22Comparethe intertextual linkagesthatshow the fifteenth-century Telugu poet Potana
to be reappropriatingand localizingin his campi7
Bhdgavatamu a Sanskritcourtlypurdna,the
tenth-century Bhdgavatam, which itselfappropriated(as Potana was probablyunaware)the
songsof the Tamil Alvars(seventh-ninth centuries).Cf. Shulman1993.
26 SHELDON POLLOCK

Sufficeit to say herethatin the ?MD, too, fromthe firstverseto the last, Kannada
difference is theorizedwithina Sanskritculturalepisteme;it is constructedas an
object of study fromthe perspectiveof a Sanskritthat definedwhat language,
especiallyliterarylanguage,is supposedto be.23
Everyfeatureoftheliterary in Kannada,foritsfirsthalf-millennium oflife,seems
to be markedby thekindsofnegotiationsofdifference and calculationsofvernacular-
cosmopolitanpredominancethatwe findin theKRM. This textdefinesvirtuallythe
whole rangeof literarythemesthatwill be meditatedoverforthe nextfouror five
centuries,everythingfromthe large questions of genre (KRM 1.33ff.)and the
construction, ifprematurely,ofa canonofKannadaproseand versepoetryjuxtaposed
to and complementing thatofSanskrit(KRM 1.28-32), to thestructure ofcompounds
and the microanalysis of whichSanskritand Kannada mayand maynot be joined in
compound(e.g.,KRM 1.51ff.).Such negotiations arenotjusttheoretical,either.They
informthe literaryproceduresof the poets themselvesover a whole rangeof texts
whose very titles-beginning with the earliest, the Karndta Kumdrasambhava
(attributedto Asaga, A.D. 853)-bespeak the localizationof the Sanskritglobal,and
suggestthata big partofwhatearlyKannadaliterature is aboutis theverypossibility
of makingliteraturein Kannada.

VernacularPoliticalSpace

No textmakesall thismoreexplicitthanthefirst literary workextantin Kannada,

Pampa's Vikramarjunavijaya (VAV, ca. 950). Pampa was the courtpoet of Arikesari
II, a Calukyaoverlordin whatis now westernAndhra(Vemulavada)who held actual
powerin the last decadesof Rastrakiuta rule.The Vikramdrjunavijaya, conceivedofas
the first"complete"vernacularversionof the SanskritMahdbhdrata, was solicitedby
the courtlyliteratiand paid forby the king himself:"The learnedfeltthatno great
poet in thepast had properly[re-]composedtheComplete Bhdrata-an unprecedented
thing-without damagingthe bodyof the tale and retainingits magnitude... and
that this was somethingonly Pampa could do. And so theygatheredtogetherand
besought[me]; I [therefore] undertaketo composethis work . . . Arikesarihimself
sent a messengerand gave [me] much wealth to have his fameestablishedin the
world,and in this fashionhad [me] composea historicalnarrative[itihdsakathal."24
The negotiationof culturaldifference mentionedabove is undoubtedlyone of the
work'smain preoccupations, and is signaledat its verycommencement: "A workof
literaturebecomesbeautifulif its imaginationis new . . . if it entersinto the poetry
of Place (desiyolpuguvudu),and havingdone so, penetratesintothepoetryoftheWay
(mdrgadol ta/vudu)"(VAV 1.8). But Pampa has additionalpurposesin mind,which
come into clear reliefonlyonce we recallsomethingabout the model he soughtto
overcome.
As my briefremarksabove tried to suggest,one of the things the Sanskrit
Mahdbhdrata is about is the productionor organizationof space and of a political

23Thelast verse in fact framesnine points of Kannada difference ("the uniquenessof

Kannada," aridu ... kannadan)over against Sanskrit,in termsof phonology,sandhi,com-
pounding,prosody,etc. (?abdamanidarpanza 342).
24VAV 1.11; 14.51. In fact,Peruntevanar's Tamil adaptation,the Pdrat-
(fragmentary?)
venpa,is about a centuryearlier(at the courtof NandivarmanIII Pallava, r. ca. 830-52).
 THE COSMOPOLITAN VERNACULAR 27

visionthatencompassesthisspace.As we saw, theheroes'travelsin theirexile,their

conquestof the quarterspriorto thedeclarationofuniversalsovereignty, thelevying
of troopsforwar when thatsovereignty is challenged,the wanderingsof the ritual
horsewhose compassmarksthe extentof theirreacquisitionof imperialstatusand
whose ritualslaughtermarksits confirmation, and the finalfunerealcircuitbefore
theirdeaths-when theyrenouncethe world of political power in despairat the
slaughtertheyengagedin to win it-reinforcethe imageofa vastyetbounded,ifso
hazilyboundedculture-sphere ofpoliticalreference,extendingfromNepal to Assam
(or theplacesnowso called) to thesouthernpeninsula,and thenceto Sind,Qandahar,
Kashmir.It is thisepic space,and thepoliticsthatfillit, thatPampa seeksto redefine
in his vernacularized version.
Pampa oftenrefersto his work as the samasta-bhdrata,wheresamastahas two
importantmeanings:the authorattemptsto reproduce,as noted,the "whole" of the
mainstoryoftheSanskritpoem.But also he wantshis epic to be seenas a "composite"
narrative.That is, it explicitlyidentifiesthe poem's patron,his family,overlord,
enemies,and his regionwiththeheroes,allies,antagonists, and worldoftheSanskrit
epic. To be sure,the poet is not a simple allegorist,and his touchis light.But his
directionsto readersareclearenough(he is explicitabouttheidentifications in 1.51),
and thestoryofCalukyapoliticalfortunes, as Arikesariassumesthemantleofprimary
vassal (sdmanta)amid the frayingstructureof Rastrakiuta power,is pushed through
the veil of the myth-epicat criticalpointsin the narrative.A good exampleof the
double narrativeis providedin the verycenterof the poem. When the sons of
Dhrtarastra in anticipationofbattlebeginto describethegreatdeedsoftheirenemy,
theepic heroArjuna-the hero'spridein fighting withgreatgod Siva and acquiring
magicweapons,thevalorhe showedin defeatingdemons,thegrandeurofhis sharing
the throneof Indra,king of gods-at this verypoint,where"Indra" king of gods
could just as well stand forIndra III Rastrakutta, Arikesari'smaternaluncle (and
"gods" could mean "kings"),thediscourseglidesseamlesslyintoa descriptionofthe
poet's royalpatron:
The majestyof this Sea of Virtues. . . who held his ground,shieldingand saving
KingVijayaditya, ForeheadOrnament oftheCalukyafamily,whenGovindaraja[IV
Rastrakuttalragedagainsthim;... whoattackedand conquered againthevassals
whocamein battalions on theorderofthesupreme Emperor Gojjega[Govindarajal
to King Baddega [ = Amoghavarsa
... and restoredimperialpower[sakalasdmrdjya-J
1111-whohadcometo himtrusting in him...
(VAV 9.51 +)

Arikesaridefeatsthe usurpingGovindarajaand restoresto power the rightful

ruler,but in doingso constituteshimselfas paramountoverlordin theDeccan in the
middleof the tenthcentury.
It is not thedetailsofthehistoricalcase thatdrawattention,but rathertheform
of culturalcommunicationPampa has inventedto presentthem.He has refashioned
in thevernaculara Sanskritepic discourseon thepoliticaland therebyrevisionedthe
transregional political order for anotherand very different kind of world. And,
accordingly, exactlylike the poem's politicaldiscourse,its geographicalimagination
is adjustedto theprimarynarrative project.The "CityoftheElephant,"Hastinapura,
which is home to the Bharataclan in the Sanskritepic, becomesVemulavada,the
Calukyancapital.The grand"circumambulation ofthequarters"ofthesubcontinent
thatrepeatedly organizestheactionoftheepic becomesa circuitofthecentralDeccan.
Even the list of riversfromwhich the watersare collectedforthe hero'scoronation
28 SHELDON POLLOCK

ritualat the end of theworkincludesa streamin the Kannada heartlandofBanavasi

(VAV 1.51ff.with Narasimhachar'snote ad loc.; 4.26ff; 14.31). In a word, what
Pampa has done is shrinkthe continentof the Bharatas(bhdratavarsa) to a Kannada
regionalworld,narrowthe visionofpoliticalpowerto the space in whichit actually
worked,and endowthiswithcomprehensible pointsofreference, narrativesense,and
literarystatus.It is now thekannadadandduof theKRM-"Between the Kaveriand
Godavaririversis thatregionin Kannada (nddaddkannadadol)f= the countryalso
called Karnataka],a well-knownpeople/region (janapada),an illustriousoutstanding
realm [viyayalwithinthe circle of the earth"(KRM 1.36)-that becomes the all-
importantpolitical and aestheticframework. And it is forthe moral and political
instruction ofthiscommunitythatPampa has writtenhis Bhdrata:"Having properly
[re-]composedthe celebratedworkof Vyasamuni... an expansivepoem of Place, is
it anywonderthat[Pampa,j the Sea of PoeticVirtues,has becomethe teacherofthe
Nadu?" (14.62).
It is by such an arrayof textsand practices-the KRM's assertingat once the
regionalityand supraregionality of Kannada,and its literaryvalue, by retrofittinga
Sanskrittaxonomy;Pampa's localizationof an epic space and politicalvisionto the
worldof Kannadanadu,and the rangeof otherculturalpracticesI have examined-
thatthe formof culturalcommunicationI want to call the cosmopolitanvernacular
comesto be produced.But if theKRM, thePampaBhdrata,and othertextscan give
us a vividsenseof thediscursiveand literarystrategiesby whichsuch a high-culture
vernacularis produced,how can we make sense of the time and the place of this
transformation? Why is it thatvernacularintellectualsstartingin theninthto tenth
centuries, fromwithinthecentersofpowerofdominantpolities(Rastrakuita, Calukya,
Hoysala), turnto Kannada forliteraryand politicalcommunication? What is their
interestthenand therein constituting theirlanguageas a newepistemological object,
an object of normativediscourse,a vehicle forcourtlyexpression?What is their
interestin renouncingwhatwas notonlypotentiallybut actuallythetranslocal, near-
global audienceof Sanskritand, forthe firsttime,speakinglocally?

Explaining Vernacularization

Similarprocessesto what we have foundin the creationof a Kannada literary

culturemay be observedall over the Sanskritecumenefromthe beginningof the
secondmillennium,fromAssam,Andhra,and Orissato SriLankaandJava,and from
Kerala, Maharashtra,and Gujarat to Tibet. Vernacularwriterstransformed the
inscriptionalrecord,so that the expressionof political will would henceforth take
place in thevernacular;thishappensmostspectacularly in Tamil underthe imperial
Co1as,but can also be seen,nascently,in Marathi,Oriya,Telugu. Theyappropriated
a Sanskritaestheticand a rangeof its literarymodels into theirlanguagesforboth
politicaland imaginativeexpression;Dandin, forexample,is reworkedin Sinhalain
the tenthcentury,Tamil in the twelfth,Tibetan in the thirteenth. They developed
new notionsofgeoculturalframeworks fortheirliterarynarrative the
representations,
same as those in which theirtextswould circulate.It was typicallyby way of a
localizationof the Sanskritepics-often with the double-narrative that we findin
Pampa-that all thesegoals were simultaneously achievedin a primalmomentof
Witnessin thisconnectionsuchJavanesetextstheRdmdyana
vernacularization. ofthe
tenthcentury(representative of a genrewheredouble-narrative is fundamental, cf.
 THE COSMOPOLITAN VERNACULAR 29

Robson 1983) and the prose Mahdbhdrata;Nannaya's Telugu version of the

Mahdbhdrataat the court of the Vefigi Calukyas in the mid-eleventh;Madhava
Kandali's Assamese Rdmayana,composed at the request of the Barahi king
Mahamanikyain the mid-fourteenth; Visnudasa'sBraj Mahdbhdrata (Pdndavacarita)
and Ramayanakathd writtenat the courtof the GwaliorTomarsin the mid-fifteenth
century;theOriyaversionsof theepicsand BhdgavatafromtheGajapati courtin the
laterfifteenth
century.This vasttransformation in thewaypeopleimaginedand wrote
theirnew regionalworldspresentsa complexofproblemsforhistoricalanalysisand
culturaltheory.We are nowherenearto unravelingany of theseforany partof the
newly vernacularized world, let alone constructing a unified theory of
vernacularization.But I thinkwe can identifysome conceptualdead ends and some
otheravenuesworthfollowing,and formulate a fewlargerprinciplesthatSouthAsia
vernacularizationsuggests.
In tworecentessaysa leadingpoliticaland culturaltheoristofSouthAsia,Sudipta
Kaviraj, considerssome centralissues of writingand being on the eve of British
colonialism(Kaviraj 1992a, 1992b). His reflectionsare invaluablefortheirinsistence
on thehistoricityand therefore ofrepresentations
variability ofcommunity, ethnicity,
identity,and theirterritoriallocalizations;even more so fortheirrecognizingand
chartingthe long-termtrend to "incommunication"in South Asia, that is, the
processesbywhichthemultilingualcapacitiesand enthusiasms ofspeakersandwriters
wereerodedby the monolingualization effected by modernity. But at the same time
a numberof receivedviewsabout the vernacularization of thisworldare reproduced
thathavegone uncontestedtoo long. Like everyotherscholarwho has writtenon the
issue, Kaviraj ties the "gradual separationof [the] emergingliteratures[of the
vernacularlanguages]fromthehighSanskrittradition"to "religiousdevelopments,"
indeed,religiousdevelopments hostileto thattradition,againstwhichthevernacular
literaturesmake an "undeclaredrevolution.""The origin of vernacularlanguages
appearsto be intimatelylinked to an internalconceptualrebellionwithinclassical
BrahminicalHinduism."25
In fact, there is precious little evidence to support these generalizations,
universallyaccepted though they are. There is of course no denyingthat some
relationshipmay be foundbetweenlanguagechoice and religiouspracticein South
Asian history;the resistanceto redactingthe Buddha's words in Sanskritand the
preferenceofJainasforeasternPrakritfortheirscriptures are familiarinstancesfrom
an earlyperiod.But by the beginningof the secondmillenniumthis relationshipis
much etiolated. Sanskrithad long ceased to be a brahmanicalpreserve,just as
brahmanshad long takento expressingthemselvesliterarily in languagesotherthan
Sanskrit,such as Apabhrams'aor indeed Kannada. The religiousdeterminantin
languagechoice in generalhas been vastlyoverdrawnforpremodernSouthAsia; in
the particularcase of the so-called rebellion in religious consciousnesstermed
devotionalism(bhakti),nothingsuggestsit can be isolatedas a significant let alone
primarydynamicin the historyof South Asian vernacularization. Some northern
Indianvernaculars came firstto be employedforwrittenliterature altogetheroutside
the brahmanicaltradition:Hindui in the west,forexample,by Mas'ud Sa'd Salman
in Lahoreca. 1100, Avadhiin theeast byMaulana DaiudinJaunpurat theend ofthe
fourteenthcentury.And many vernacularinaugurationsshow no concernwith
religiousdevotionalismwhatever.EarlyBraj crystallizedas a literaryidiom in the

25Whatis meantis the "origin"ofvernacularliteratures,

not languages,a commonslip-
page promptingmy remarksabove on writingand the beginningof literature.
30 SHELDON POLLOCK

writingsof Visnudas underthe patronageof the Tomars in Gwalior,and as Stuart

McGregorhas carefullydemonstrated, his vernacularepics have nothingto do with
bhakti(McGregorn.d.). The same holds true forthe westernend of the Sanskrit
culturalecumene,wherethe earliesttextsin Gujarati,of the fourteenth to fifteenth
centuries,include Bhalan's courtlyKddambariand the anonymouseroticphdgu,
Vasantavildsa, and forthe eastern,thepolitical-allegorical
kakawinsofJavanese.
In the case of Kannada,beliefin the religiousimperativeof vernacularization is
altogetherunchallengedin the scholarlyliterature.Here, however,the putative
impetusis notdevotionalismbut whatone scholara generationago describedas Jaina
loyaltyto "thepreceptofthefounderoftheirfaiththatthevernacular shouldbe used
forpreachingto themasses"(Altekar1960b, 314). Why it tookmorethana thousand
yearsfor this loyaltyto manifestitselfin literaryproductionin the language of
Karnataka,whereJainashad lived sinceperhaps300 B.C., is a mystery. Mysterious,
too, is the factthat,at the verytime and place when Kannada literaryproduction
finallydoes make history,the greatestofJainareligiouspoets-those whoseloyalty
shouldbe beyonddoubt-Jinasenaand Gunabhadra(ca. 850-900), choseSanskrit for
the spiritualpoetryof theirMahdpurdna, as manydid also forlaukikaor this-worldly
moralliterature, such as Pampa's contemporary at the Vemulavadacourt,theJaina
abbot Somadevasturi (authorof Ya?(astilakacamp7,
A.D. 959).
If a numberof the earlierKannada poets wereJainas,some weredecidedlynot.
It is no anomalythat when a brahmanministerof religiousaffairs(dharmakdryesu
niyukta) underVikramaditya VI ofthewesternCalukyas(end oftheeleventhcentury)
giftedland to a Mimaamsa college(aprdbhdkarasya vydkhyana?/d -the mostorthodox
ofall orthodoxies-thelongprasastihe composedwas equallydividedbetweenverses
in Kannada and Sanskrit(El 15, pp. 348ff.).As forJainaauthors,some werealmost
clearlyJainabrahmans(a categorypeculiarto the Digambaralay communityof the
Deccan), includingPampa (cf. VAV 14.49) and NagavarmanII (Kdvydvalokanam vs.
960). And muchoftheirworkhas littleor nothingto do withJainismas such.Some
mayhavecomposedtheologicalhistories,but theyalso composed,at leastforthefirst
three centuriesof literaryhistory,non-Jainaprose-versecourtlyepics (campz7s),
typicallyfornon-Jainapatrons(Ranna wrotehis Gaddyuddhaca. 1000, fora Saiva
prince,cf. 1.21). Pampa's Vikramarjunavijaya-which he calls a laukika poem in
contrastto hisjindgamaor theologicaltext,theAdiPurdna (VAV 14.60), and is, as we
saw, a workdeterminedin its everyimportantfeatureby politicalvision-not to
speakoftheKRM and suchhigh-culture vernacularizations as KarndtaKddambar7 (ca.
1030), providesevidenceenoughofan audienceand a literary cultureformedbyvalues
to which religiousidentitywas subordinate.The one value that the KRM itself
celebratesin describingthe literarycourtis culturalvirtuosity:

Anyonewhobetakeshimself to thegreatNrpatufiga to becomea memberofhis

circle(sabhd)mustbe committed
literary to thediscriminating understandingofall
this-worldly matters,as well as Uainal scriptural,
and eminent vaidikaquestions
He mustbe adorned
(laukikasdmdyikdruvaidikavis'esa). withdistinguished utterances,
analysis,and artsrelatingto the knowledge of literature(sdhita);he musthave
exceptional andhighlyskilledconduct,
insight, andbe totallyclear-thinking, fully
analyzing eachandeverydefinition andexample[ofliteraturel.
(3.219-20)

Not onlywas Kannadavernacularizationnotdrivenbyreligiousimperatives,

it was not in anymeaningfulsensepopular.Popularcommunication can hardlyhave
been servedby a literatureso thoroughly
presupposingSanskrittrainingin lexicon,
 THE COSMOPOLITAN VERNACULAR 31

syntax,metric,rhetoric;some textsexplicitlystatetheywerecommissionedby and

intendedfora learnedaudience,as we saw in the case of Pampa.
The dominantexplanations, therefore, derivedultimatelyfroma disciplinarybias
towardreligiousstudiesthatcan oftendeformthinkingaboutprecolonialIndia,26are
oflittlehelp in understanding theprimarymomentsofvernacularization thatmarked
much of South Asia in the earlysecond millennium.What is abundantlyclear,
however,in virtuallyeverycase we can historically capture-and again, Kannada is
paradigmatichere-is the roleof thecourtin thevernacularturn.It is cosmopolitan
elites-men like Pampa fullyin commandof Sanskritand enjoyingrankand status,
paid by the king for his work and rejoicingin his power and grandeur(VAV
13.49ff.)-writingcourtlypoetryfor theirpeers, who firstturnedKannada (and
Telugu, Malayalam,Braj, Assamese) into an instrumentforliteraryand political
expressivity, and who forthe next half-millennium will continueto produce the
literaryand philologicaltextsin thelanguage.What we needto understand, however,
is whatthiscourtlyliterature meantforthe self-understanding ofpolity,and whyit
came intoexistencewhenit did.
The common-sense of contemporary social theorysuggeststhatwe should seek
some instrumental fitbetweenvernacularpoetryand polity.The grammaticaland
literary-normative will-to-unification of the language,we may be led to assumeby
such theory,is intimatelyconnectedwith the politicalwill-to-unification, since the
poweroverlanguageis poweroverthe usersof thatlanguage-or moresimplyput,
grammarians and politicianssharethesamedelusions(Bourdieu1991, 43-65; Fabian
1986, 8). This axiom invitesus to look forsomethingnew politicallyhappeningin
the world of the Rastraku-tasand western Calukyas within which Kannada
vernacularization is occurring.One may,it is true,discerna different kindofpolitical
paradigmarisingin SouthAsia at theend ofthefirstmillennium.The old aspiration
of transregional and trans-"ethnic" rule,the "imperialpolity"thathad markedthe
subcontinent forthe previousthousandyears,had begun to give way to something
different,somethingperhapsto be called vernacularpolity.27Enduringdominance
was no longerto be soughtoutsidethe extendedcore area,whichforits partcame
increasingly to coincidewitha languageor culturearea-something thatthepolity,
by its cultural-politicalpractices,helped to create-vague though both areas
undoubtedlywerein conceptionand on theground.
When in late middle-periodIndia, one mightbe prone to suppose,kingdoms
beganto replacetheearliersupraregional empires,ordreamsofsupraregional empires,
withtherealityofregionalgovernance;whenkingsfromNrpatuniga in ninth-century
Kannadanaduto Airlanggain eleventh-century Java became less the cakravartins
of
cosmic imperia and more the overlordsof really existing regional polities, the
cosmopolitanexpressivity of Sanskritceded beforea vernacularthat could definea
regionalpoliticalspace thatactuallyworkedas such.And thuschoiceoflanguagefor
the making of literature-the inscriptionof new kinds of literarytexts in the
vernacular-whereby local cultureis authorizedand made availablefordiffusion and
permanence,could be takento constituteat the level of cultureand communication
a new sense of the permanenceand diffusionof the polityas a formof community

26Thishas broughtus to the point whereeven the most carefulstudentsof the subject
are proneto contrast"Sanskritization"as a processof "religiousculture"with,say,Islamici-
zation,whichis said to inhabitthe domainof the secular(Wagoner 1996, 872).
27Kulkeunderstandstheselaterkingdomsas "imperialpolities,"without,however,spec-
ifyingwhatdistinguishesthemfromthe earlierimperialformations (1995, 242-62).
32 SHELDON POLLOCK

self-understanding and solidarity. And the specific character of this newly

vernacularizing literature,as a cosmopolitanvernacular,suggeststhat it aimed to
usurpthepositionofthesuperposedliterary formation and to recreatetheconditions
of imperialcultureat the level of the region.
The troublewith thisapproach,I earliersuggested,is thatit restsupon a set of
beliefsabout the relationof cultureand power (whetheras instrumentalreason,
legitimation, or ideology)thathavebeenformedin theage ofcapitalin orderto make
sense of it (cf. Lefort1986, 181-236; Scott 1990, 70-107). These encouragea
conceptualstylethattypicallyreduceslanguageto powerand precludeseven asking
whatmaybe different about theirinteractionin thepast. It is no easymatter,to state
the difficulty moregenerally,to theorizea premodernworldwithoutdeployingthe
theoreticalpresuppositions-theonly ones we have-forged by modernity;to read
the precolonialfroma locationin the postcolonial,to displace let alone replacethe
notionof the nation formand the theoryof cultureit generates.It thus remains
unclearto me what warrantssuch presuppositionsin understandinga different-
potentiallyradicallydifferent-worldof the nonmodernnon-West.As I suggested
earlierto be the case in the Sanskritcosmopolis,one might instead theorizethe
presenceofsome altogetherdifferent culturallogic,wheretheaesthetic,forexample,
was centrallyin play, or some peculiarnew self-fashioning throughthe vernacular
distinctionof personsand places. Only moreempiricalwork,however,informedby
a stubbornconceptualautonomy,is going to be able to testsuch hypotheses.28
Developinga historically and culturallysensitiveaccountof the relationshipof
vernacularpoetryand polity beforewesternmodernityis, however,only part of a
biggercomplexofquestions,whichin lieu ofa prematurehistoricalconclusionabout
the cosmopolitanvernacularas such I want to tryto characterize, with respectboth
to its historicaland theoreticalchallenges.
This largercomplexis the problematicof premodernglobalization.What used
to be called "Indianization"is one ofthevarietyofhistorically important waysin the
past (othercrudebut still necessarycategoriesincludeHellenization,Romanization,
Sinicization)of being translocal,of participatingin social and culturalnetworksin
addition to materialnetworksthat transcendedthe immediatecommunity,and
againstwhicha wide rangeof vernacularculturesdefinedthemselves.Now, despite
the justifiablefascinationof the academywith the new globalization,the historical
studyof the culturaland social dynamicsof premodernglobalizationprocesses-
withoutwhichthenewnessofthepresentcase can onlybe imaginedand notknown-
has yetto begin in earnestforanypartof theworld.Considerfora momentonlythe
scholarshipon the Romanizationof the westernempire, a process of no little
consequence,I think,in the creationand construction of "Westerncivilization."In
1990 a leadinghistorianof the Roman empirecould say,"Thereseemsto havebeen
no scholarlyattentionpaid to anythingbut the symptoms"of Romanization;"Even
in so richlyinformeda workas . . . thereare onlytwo or threelines devotedto the
motivesforculturalchange;and I recallnothingmorethanthatin all my reading"
(MacMullen 1990, 60).
On the rareoccasionswhentheglobal and local are analyzedas waysofbeing in
interaction,both are typicallythought of as pregiven,sharplydefinedcultural
formations, theformer as theexogenous,greattradition,thelatteras theindigenous,

28 Theorizing vernacularpolitycomparatively
in SouthAsia and Europe,and thecurrently
dominantaccountsof vernacularization and nationalismin Europe (Gellner,Anderson),are
further addressedin Pollock 1998.
 THE COSMOPOLITAN VERNACULAR 33

littletradition-the clichesof the introductory area-studiescourse.They have taken

on the characterof stable thingsthat interact(or thingsthat "clash," in the more
cartoon-like versionofcivilizations)ratherthanbeingseenas a congeriesofconstantly
changingrepertories ofpractices;and iftheychangetheyare thoughtto do so notby
humanchoicefromamongsuch practicesbut as thingsin naturechange.
In an importantrecentessay on globalizationand localization in the early
nineteenth-century Pacific(East Asia, Polynesia,PacificNorthwest),MarshallSahlins
has argued that the world systemis not "a physicsof proportionate relationships
betweeneconomic'impacts'and cultural'reactions.'Rather,thespecificeffects ofthe
global-material forcesdependon thevariouswaystheyare mediatedin local cultural
schemes.""Indigenouspeoples,"thatis to say,variously"integratetheirexperience
of the world systemin . . . theirown systemof the world" (1988, 4-5). This is a
welcomeand necessarycorrective to the commonimage of the local as inertwax for
thedevelopmental imprintoftheglobal. It is, forexample,just suchlocal mediations
in the premodernglobalization process of "Indianization" that have interested
studentsof early Southeast Asia for several decades (Wolters 1982 remains a
stimulatingexample).
But implicitin Sahlins'saccountis a conceptionof local culturalschemesand a
systemof the world of indigenouspeoples as thingspermanently given. Manchu
emperorsin the eighteenthcenturythusare said to sharethe same systemas Ch'in
Shih Huang-tiin the thirdcenturyB.C. (Sahlins 1988, 22). But we knowsuch local
systemsconstantlychanged, sometimesradically.Certain componentsof literary
culture,forexample,werelongcentralto theChinesesystemoftheworld:The ability
to composeRecent Stylepoetrywas requiredto pass the civil serviceexamination
fromthe Sung periodonward.We now know that definingfeaturesof this poetry
were inventedin the T'ang by the importationof Sanskritliterarytheory,such as
Dandin's "Mirror"(Mair and Mei 1991), one of the importantculturalpreciosities
thatcirculatedin an Asian systemofpremodernglobalization.
Dehistoricizationand the ideology of indigenismthat depends upon it (the
indigenous being nothing but the conceptual consequence of a deficiencyof
historicization),whichusuallygovernthestudyoflocal mediationsofglobal cultural
forms, are even more prominentin the studyof theformsthemselves.Discussionsof
the impactof South Asian culturalflowson SoutheastAsia rarelyacknowledgethe
factthatno preternaturally unifiedIndian cultureexistedto produceIndianization;
what existedwas only a set of recentlydevelopedculturalcodes and acts, some of
whicharosealmostsimultaneously (perhapseven"convergently," Kulke 1990) outside
the subcontinent,and which only gradually coalesced into something like a
cosmopolitanunity.In fact,much of India itselfwas being Indianizedat the very
same periodas Java or Khmercountry-and in a hardlydifferent way-and it was
Indianvernacular themselves
intellectuals, Indianized,who drove the processforward.
Moreover,fromthe local perspective,we need to see thatwhen Sanskritcomes
to, say, earlyJava, it is not as a medium forthe articulationof realitiesthat are
"properlyJavanese"(Lombard1990, 13-14), as if realitywere constitutedpriorto
ratherthanlargelyby language,and Javaneseness somepreexistent thingoverwhich
Sanskritis laid ratherthana continuousprocessofbecomingin whichSanskritis one
element.The role of the Sanskritcosmopolitanin southeast,or southern,Asia was
less to bring"ancientand persistingindigenousbeliefsinto sharperfocus"(Wolters
1982, 9) in some "native" culturethat itselfretaineda "distinctiveness both as a
wholeand in itsparts"(Reid 1990, 1) thanto participatein theverycreationofthese
cultures,and to be itselfchangedin theprocess.
34 SHELDON POLLOCK

What needsto supplementSahlins'simportant critique,then,and futureresearch

on premodernglobalization,of whichthe cosmopolitanvernacularis an instance,is
appreciationofthefactthat"indigenous"culturesareproducedin thecourseoflong-
termtranslocalinteractions by theverysameprocessesthatproducetheglobal itself.
The local/globaldualism,therefore, needs to be historicizedout of existence,both
because nothingis globally self-identicaland because the local is always "newly
different while each becomesthe otherin constantlynew ways (Pred
differences,"
1995). But notonlyforthesereasons.If thedualismcontributes to people's"political
disarming"by producinga falseunderstanding of the largerforcesat workin their
lives, it may also contributeto theirarmingthemselves-to recreatesome "local"
thatneverexistedin the firstplace.

List of References

Kannada Texts:
KarndtakaKavirdjamdrgam. 1973. Edited by A. VenkataRao and H. SeshaAiyangar.
2d ed. Kannada Series.Madras:UniversityofMadras(Kannada Series).(KRM)
Kavirdjamdrgam.1983. Edited by K. Krishnamoorthy. Bangalore:IBH; ed. M. V.
Seetha Ramiah. Bangalore:KarnatakaSangha, 1968, reprint1994; ed. M. V.
SeethaRamiah,Mysore:D.V.K. Murthy,1975, reviseded.
Kdvydvalokanam.1967. Edited by R. Narasimhachar.3d ed. Mysore:Universityof
Mysore.
Pampa Bhdratam emba Vikramdrjunavijayam. [19263 1990. Edited by
Venkatanaranappa.Mysore: Universityof Mysore Press. Reprint Bangalore:
Directorateof Kannada and Culture.(VAV)
?abdamanidarpanaofKedirdja.1994. Edited by T. V. VenkatachalaSastry.Bangalore:
Directorateof Kannada and Culture.

SanskritTexts:
AdipurdnaofJinasena.1993. Edited by PannalalJain.4th ed. New Delhi: Bharatiya
Jnanapith.
El. Epigraphia
Indica
GaddavahoofVakpatiraja.Edited by N. G. Suru.Ahmedabad:PrakritText Society.
Kdvyd/a0kdra of Bhamaha.. 1981. Edited by Batuk Nath Sarma and Baldeva
Upadhyaya.2d. ed. Varansi:ChaukhambhaSanskritSansthan.
Kdvyalaksanam t=Kdvyddars'al. 1957. Edited byAnantalalThakurand UpendraJha.
Darbhanga:MithilaInstitute.(KA)
KdvyalAka-ra of Rudrata. 1928. Edited by Pt. Durgaprasadand Wasudev Laksman
ShastriPanshikar.Bombay:NirnayaSagar Press.
Kdvyda1Akdrasitraand VrttiofVamana. 1908. EditedbyRatnaGopal Bhatta.Benares:
Vidya Vilas Press.
Kdvyamimdrmsd of Rajas'ekhara.1934. Edited by C. D. Dalal et al. 3d ed. Baroda:
OrientalInstitute.
Kdvydnusasanam of Hemacandra. 1964. Edited by Rasiklal C. Parikh and V. M.
Kulkarni.2d ed. Bombay:Sri MahaviraJainaVidyalaya.
?dkadtyanavydkaranam 1971. EditedbySambhunatha
svopajndmoghavrttisamalankrtam.
Tripathi.Delhi: BharatiyaJnanapithaPrakasana.
Sarasvat7kanthdbharandlankdra. 1979. Edited by BiswanathBhattacharya. Varanasi:
VaranasiHindu University.
 THE COSMOPOLITAN VERNACULAR 35

Forthcoming.Edited by V. Raghavan. Harvard Oriental Series.

?r.gdraprakas'a.
Cambridge:HarvardUniversityPress.(SP)
ofVidyakara.1957. Edited by D. D. Kosambiand V. V. Gokhale.
Subhdsitaratnakosa
HarvardOrientalSeries42. Cambridge:HarvardUniversityPress.
of Kuntaka. 1977. Edited by K. Krishnamoorthy.
Vakroktij7vita Dharwar:Karnatak
UniversityPress.
Vikramdfkadevacarita of Bilhana. 1964. Edited by VishwanathShastriBharadwaj.
Varanasi:SamskritSahityaResearchCommitteeoftheBanarasHindu University.

SecondarySources
ALTEKAR, ANANT SADASHIV. 1960a. Rdshtrakfitas
and TheirTimes.Poona: Oriental
Book Agency.
In The EarlyHistoryoftheDeccan,edited by
1960b. "The Rashtrakiutas."
G. Yazdani. London:OxfordUniversityPress.
BAKHTIN, M. M. 1981. TheDialogicImagination:FourEssays.Austin:Universityof
Texas Press.
BALIBAR, ETIENNE, and IMMANUEL WALLERSTEIN. 1991. Classes,Nations,States.
London:Verso.
BOURDIEU, PIERRE. 1991. Languageand Symbolic
Power.Cambridge: Harvard
UniversityPress.
CHIDANANDA MURTI, M. 1978. "Kannada Language and Literature during the
Chalukyasof Badami (c. 540-750 A.D.)." In TheChalukyasofBadami,editedby
M. S. Nagaraja Rao. Bangalore:MythicSociety.
DUNDAS, PAUL. 1996. "Jain Attitudestowardsthe SanskritLanguage." In The
editedbyJanE. M. Houben. Leiden:Brill.
and StatusofSanskrit,
Ideology
DUVERGER, MAURICE. 1980. Le Concept d'Empire. Paris:PUF.
FABIAN, JOHANNES. 1986. Languageand ColonialPower.Cambridge: Cambridge
UniversityPress.
GOPAL, B. R., ed. 1994. TheRashtrakutas ofMalkhed.Bangalore:MythicSociety.
HARDY, FRIEDHELM. 1994. "CreativeCorruption: SomeCommentson Apabhramsra
Literature, Yogindu." In Studiesin SouthAsian Devotional
Particularly Literature:
ResearchPapers,1988-91, edited by Alan W. Entwistleand FranSoiseMallison.
New Delhi: Manohar;Paris: Ecole FranSaised'Extreme-Orient.
HEIDGGER, MARTIN. 1960. Der Ursprung desKunstwerkes.Stuttgart:Reclam.
INDEN, RONALD. 1990. Imagining India. Oxford: Blackwell.
JONES,A. H. M. 1964. TheLaterRomanEmpire.Oxford:ClarendonPress.
KAVIRAJ, SUDIPTA. 1992a. "Writing, Speaking, Being: Language and the Historical
Formationof Identitiesin India." In NationalstaatundSprachkonflikt in Sild-und
edited
Slidostasien, byDagmarHellmann-Rajanayagam and Dietmar Rothermund.
Stuttgart:Steiner.
. 1992b. "The ImaginaryInstitutionof India." In SubalternStudiesVII:
Writingson SouthAsian Historyand Society,edited by Partha Chatterjeeand
GyanendraPandey.Delhi: OxfordUniversityPress.
KULKE, HERMANN. 1990. "IndianColonies,Indianization orCulturalConvergence:
Reflections on the ChangingImage of India's Role in South-EastAsia." Semaian
3: 8-32.
. 1993. Kingsand Cults:StateFormation in India and Southeast
and Legitimation
Asia. New Delhi: Manohar.
, ed. 1995. TheStatein India: 1000-1700. Delhi: OxfordUniversityPress.
36 SHELDON POLLOCK

LACAPRA, DOMINICK. 1983. Rethinking

Intellectual
History:
Texts,Contexts,
Language.
Ithaca,N.Y.: CornellUniversityPress.
LEFORT, CLAUDE. 1986. The PoliticalFormsofModernSociety.
Edited by JohnB.
Thompson.Cambridge:MIT Press.
LOMBARD, DENYS. 1990. Le Carrefour globale.III: L'heritage
javanais,essaid'histoire
desroyaumes
concentriques.
Paris: Editionsde l'1Ecoledes hautesetudesen sciences
sociales.
MAIR, VICTOR H., and Tsu-LIN MEI. 1991. "The SanskritOrigins of Recent Style
Prosody."HarvardJournalofAsiaticStudies51(2): 375-470.
MASICA,COLIN. 1991. TheIndo-AryanLanguages.Cambridge:CambridgeUniversity
Press.
MACMULLEN, RAMSAY. 1990. Changesin theRoman Empire:Essays in theOrdinary.
Princeton,N.J.: PrincetonUniversityPress.
McGREGOR, R. S. 1984. HindiLiterature
fromitsBeginnings
totheNineteenth
Century.
Vol. 8, fasc.6. A History ofIndianLiterature.Wiesbaden:Harrassowitz.
n.d. "Visnudas(AD 15c) and his Ramayana-katha."
NAGARAJ, D. R. 1996. SdhityaKathana. (In Kannada.) Heggodu: Akshara
Prakashana.
NARASIMHACHARYA, R. 1934. History ofKannadaLanguage.Mysore:University of
Mysore.
. 1971. Panpabhdrataddpike. (In Kannada.) Mysore:Universityof Mysore
Press.
POLLOCK, SHELDON. 1995. "LiteraryHistory,Indian History,World History."
SocialScientist
23 (10-12): 112-42.
1996. "The Sanskrit Cosmopolis, 300-1300: Transculturation,
Vernacularization, and the Question of Ideology." In The Ideology and Statusof
Sanskrit,editedbyJanE. M. Houben. Leiden:Brill.
. forthcoming. "India in the VernacularMillennium:LiteraryCultureand
Polity 1000-1500." In Collective and PoliticalOrder,edited by Shmuel
Identities
Eisenstadtet al. Special Issue ofDaedalus.
PRED, ALLAN. 1995. "Out of Bounds and Undisciplined:Social Inquiryand the
CurrentMomentof Danger." SocialResearch 62(4): 1065-91.
RAGHAVAN,V. 1963. Bhoja's?rngdraPrakasa.2d ed. Madras:Punarvasu.
REID, ANTHONY. 1990. "An Age of Commerce' in Southeast Asian History."
ModernAsianStudies24(1):1-30.
SAHLINS, MARSHALL. 1988. "Cosmologiesof Capitalism:The Trans-Pacific Sector
oftheBritishAcademy
of the World System."Proceedings 74: 1-51.
SCHARFE, HARTMUT. 1977. Grammatical Vol. 5, fasc.2. A Historyof
Literature.
IndianLiterature.
Wiesbaden:Harrassowitz.
SCOTT, JAMES.1990. Domination and theArtsofResistance:
HiddenTranscripts.New
Haven, Conn.: Yale UniversityPress.
SEGAL,JEFFREY,ed. 1993. InternationalJournaloftheSociologyofLanguage.Vol. 99.
Special numberon koines.
SHACKLE, CHRISTOPHER. 1993. "Early Vernacular Poetry in the Indus Valley: Its
Contextsand its Character."In Islamand IndianRegions,edited by Anna Libera
Dallapiccolaand StephanieZingel-AveLallemant.Vol. 1. Stuttgart:
FranzSteiner
Verlag.
SHULMAN, DAVID. 1993. "Remaking a Purana. The Rescue of Gajendra in Potana's
Telugu Mahdbhdgavatamu." In PurancaPerennis: and Transformation
Reciprocity in
HinduandJainTexts,editedbyWendyDoniger.Albany:StateUniversity ofNew
York Press.
 THE COSMOPOLITAN VERNACULAR 37

A Braj BhdsdReader.London:
SNELL, RUPERT. 1991. TheHindi ClassicalTradition:
Universityof LondonSchoolofOrientaland AfricanStudies.
SWAMY, B. G. L. 1975. "The Date of the Tolkappiyam:A Retrospect."Annalsof
Oriental (Madras),SilverJubileeVolume: 292-317.
Research
History
TARLING, NICHOLAS, ed. 1992. TheCambridge Asia. Vol. 1: From
ofSoutheast
EarlyTimestoc. 1800. Cambridge:CambridgeUniversityPress.
TERWIEL, B. J. 1996. "The Introductionof Indian Prosodyamong the Thais." In
TheIdeologyand StatusofSanskrit, editedbyJanE. M. Houben. Leiden:Brill.
WAGONER, PHILIP. 1996. " 'Sultan among Hindu Kings': Dress, Titles, and the

Islamicizationof Hindu Cultureat Vijayanagara." JournalofAsianStudies55(4):

851-80.
WALLERSTEIN, IMMANUEL. 1991. Geopolitics EssaysontheChanging
and Geoculture:
World-System.Cambridge:CambridgeUniversityPress.
WOLTERS, 0. W. 1982. History, Culture,and Regionin SoutheastAsian Perspectives.
Singapore:Instituteof SoutheastAsian Studies.
ZUMTHOR, PAUL. 1987. La lettre etla voix:dela "litt6rature" Paris:Seuile.
m6die`vale.
ZVELEBIL, KAMIL. 1992. Companion StudiestotheHistoryofTamilLiterature.
Leiden:
Brill.