Journal of Ethnic and Migration Studies Vol.27, No. 1: 61± 79 January 2001

Theracial self-identi® cation of SouthAsians in the

Ann Morning

Abstract Theracial identity of South Asians haslong been asubject ofcontroversyin theUnited States. Theirinchoate racial status translates into avarietyof racial descriptorsbeing chosen by and for South Asians.This paper uses 1990census datato examinethe socio-economic anddemographic correlates ofthe racial self-identi® cation choicesmade by household heads of Indianorigin, both foreign- and US-born. The results ofmultinomial logit analysis show that respondentswho are more acculturated to theUnited States aremore likelyto describethemselves as `Black’ or` White’ thanare thosewith less familiarity with society.However, higher socio-economic levels areassociated with a greaterlikelihood of self-identi® cation asSouth Asian on thecensus race question. Finally,comparison with a sampleof Asian Indianchildren reveals the latter’s greatertendency to be identi®ed with a race otherthan South Asian,due both to theirmore extensivemixed ancestry andtheir larger share of US-born respondents.

KEYWORDS: RACIAL SELF-IDENTIFICATION; SOUTH ASIANS; INDIANS; UNITED STATES; LOGITANALYSIS; CENSUS

Introduction in generalseem to havedif® culty accepting complexity andambiguity. I’ m either Indianor American, white orblack ¼ Ican’t bebothand neither. (KavithaMediratta)

Thegreat thing about being Indian is thateverybody thinks you’re one of them (Amit K. Misra).1 The racialidentity ofSouth Asianshas long been asubject ofsomecontroversy in the United States. 2 In the earlyyears of the twentiethcentury, when whiteness (orAfrican ancestry) was a prerequisite fornaturalisation, American courts vacillatedon the questionof whether AsianIndians were white ornot. In contrastto and Armenians, who were deemed white forthe purposes ofcitizenship acquisition, and Japanese, Chinese, andFilipino applicantswho were not,the verdicton the racialclassi® cation of Indianschanged fromcase to case(Haney LoÂpez 1995).American uncertainty over South Asianracial identity hasalso been mirroredin the CensusBureau’s frequent changesin its classi®cation of this group. Overthe courseof the lastcentury, respondentsof South Asianorigin have been classi®ed variouslyas `Hindu’, `White’,`Other’, and` Asian’(Lee 1993). South Asiannewcomers are not alone, however, in confrontingan American raciallandscape that at ® rstseems to have no clear place forthem. Not only does the diversityof the United States’contemporary immigrant pool ensure asteady in¯ux ofpeople whodo not® teasilyinto the traditionalblack/ white dichotomy

ISSN1369-183X print/ ISSN1469-9451 online/ 01/010061-19 Ó 2001Taylor & FrancisLtd DOI: 10.1080/13691830020024867 CarfaxPublishing 62 A. Morning

(Bashiand McDaniel 1997;Mazumdar 1989), but in the pastas well, immigrants tested,stretched and molded the nation’sconceptions of racial categories. As Ignatiev(1995) has shown, Irish immigrants were notconsidered white until well aftertheir arrivalin the United States,and this was true ofotherEuropean groupsas well (Jacobson1998; Sacks 1994). Similarly, Chinese, Japanese,Korean, andFilipino Americanswere notalways considered to constitute a pan-ethnic Asianrace (Espiritu 1992;Takaki 1989). Butunlike the Irishwho have already become white,or the Chinese and Japanese whoare now Asian, the racialclassi® cation of South Asiansin the United Statesis still in ¯ux.Although they nowseem ®rmlyensconced in the census` Asian’category, this is a fairly recent development andone thatcame aboutonly afterconsiderable debate (Espiritu 1992).Moreover, several writers havedescribed anuneasy alliancebetween South Asiansand East Asians under the pan-ethnic `Asian’ rubric (see the contributionsin Shankarand Srikanth 1998a).Finally, otherAmericans seem unsure asto their racialstatus. F. James Davis(1991: 162) ® ndsevidence thatsome blacks consider Indians to be blackas well, andRosemary Marangoly George (1997)reports a widespreadconcern amongIndian Americansin Californiaover being takenfor Mexican or black. More broadly,Nazli Kibria (1996;1998) maintains that South Asiansare seen as `ambiguous non-whites’ in the United States. Given their inchoateracial status, South AsianAmericans may offer unusual insightinto the processof racialformation. De® ned by Omiand Winant (1994: 55)as `the socio-historicalprocess by which racialcategories are created, inhabited,transformed, and destroyed’, racial formation is both macro-level processand the culminationof myriadindividual encounters. It is constituted by `racialprojects large and small’ (Omi andWinant 1994: 61). This paper seeks to explore individual-level adoptionof some racial labels and resistance to others. The diversityof opinionsregarding their `appropriate’ racialclassi® cation leadsSouth Asiansto choose an array of labelson the censusform, even when censusinstructions explicitly directthem to one checkbox category.I use 1990 censusdata to examine the socio-economicand demographic correlates of the racialself-identi® cationchoices made by respondentsof AsianIndian origin,US- andforeign-born, whomake up the majorityof the South Asianpopulation of the United States. In the next section,I review the literatureregarding racialand ethnic identity in general andSouth Asianracial self-conception in particular.I then develop a seriesof hypotheses tobe tested,using dataand methods described in the third partof the paper. The lasttwo sections are devoted to analysis of the dataand discussionof the results.

SouthAsians and racial formation Theoriesof racial formation and of ethnicity Many branchesof race and ethnicity theorylend themselvesto the exploration ofhow South AsianAmericans come to view themselvesin racialterms, but centralto this investigation is Michael Omiand Howard Winant’s racial forma- tiontheory. In the ®rstedition of RacialFormation intheUnited States (1986:61± 2), they present their paradigmin the followingterms:

Themeaning of race is de® nedand contested throughoutsociety, in bothcollective and Theracial self-identi® cation ofSouth Asians 63

personal practice. Inthe process, racial categories themselves areformed, transformed, destroyedand reformed. We use the term racial formation to refer to the process bywhich social, economic andpolitical forces determine the content andimportance of racial categories, andby which theyare in turn shapedby racial meanings. Thuswe canthink ofthe South Asianexperience in the United Statesas acase studyof the processby whichracial categories are formed and transformed. More speci® cally,it provides an example ofwhatOmi and Winant call racialisa- tion:`the extensionof racial meaning toa previously unclassi®ed relationship, socialpractice or group’ (1986:64). Much ofthe literatureon racial formation privileges the broadsocial and politicalforces, historical and contemporary, which shape the development and spreadof racial schema (e.g. Davis1991; Jacobson 1998; Marx 1998). Students of racialformation writ large, however, often ignore the individual-level actions andencounters that shape racialisationon the ground.Instead, theories of ethnicity ±asopposed to race ± havedelved deeper intothe murky waters ofpersonal identity. Joane Nagel (1998:237) describes a prevailing paradigmof ethnicity that,in itsemphasis on volition, situational contingency, and social interactionwithin communities, places the individual atthe heartof the ethnic identi® cationprocess: According to this constructionist view, the origin,content, andform of ethnicity re¯ect the creative choices of individuals andgroups as they de® ne themselves andothers in ethnic ways¼ Thelocation andmeaning of particular ethnic boundariesare continuously negoti- ated,revised, andrevitalized, bothby ethnic groupmembers themselves aswell asby outside observers. Thissociological perspective alsodraws attention to the interplay of intern- ally-chosenand externally-imposed de® nition,or as Nagel putsit, ` what you think yourethnicity is,versus what they think yourethnicity is’ (1998:240). In doing so,it draws attention to what is perhaps the mostcrucial distinction between processesof racialand of ethnic formation,namely the varyingdegrees ofexternal versus internal de® nitionthat they incorporate. 3 Banton(1983: 10) contendsthat ` Membership in anethnic group isusually voluntary;membership in aracialgroup isnot.’ Moreover, there isa powerful element ofhierarchy embodied in the notionof race that is not inherent toethnicity: while `ethnic’ social relations arenot necessarily hierarchical, exploitative andcon¯ ictual, `race relations’ certainly appearto be.Although ethnic boundariesinvolve relations of power, andsocial categorizationis inherent to the internal± external dialectic of ethnic identi® cation, hierarchial difference is not de® nitive of ethnic relations ¼`Race’,however, unlike ethnicity, seems to bemuch more amatter of social categorizationthan of group identi® cation (althoughstill amatter of both).(Jenkins 1997: 74± 5) Although the spectrumof internalversus external de® nitionencompasses both the processesof ethnic andracial identi® cation,they fallat decidedly different pointson thisscale. Thus the `ethnic options’and ` twilightof ethnicity’ respect- ively revealed by Waters(1990) and Alba (1985)to characterise contemporary European Americansappear much lessplausible forthose whose lives are constrainedby race.

South Asians inracial terms Boththe volitionalaspect associated with ethnicity andthe externally imposed natureof race appear to be forcefully atwork in the caseof South Asians,who 64 A. Morning havenot yet cometo be closelyidenti® ed witha raciallabel inthe United States. Asdiscussedin the introductionto this article, the federal governmentcurrently classi®es most South Asiansas belonging tothe `Asian’ race,which atface value constitutesa powerful externalmechanism of categorisation(Nagel 1986). 4 Upon closerexamination this situation reveals a curiousconcordance of externaland internalde® nition.In the 1970s,Indian Americanassociations successfully petitioned forinclusion in stateand federal Asianracial categories, with an eye tobene® ting fromaf® rmative action measures (Espiritu 1992).Thus the group wasable toimpose its choice on the state± whichotherwise would have continued itspractice of designating South Asiansas white ±ratherthan the otherway round. Yetdespite the adoptionof this mutually agreed-upon of®cial classi® cation scheme,neither South Asiansnor other Americans seem toperceive themas raciallyAsian. Participation in pan-ethnic Asiancoalitions is frequently por- trayedas a politicalstrategy (Espiritu 1992;Visweswaran 1997) that has had littleimpact on South Asians’actual racial self-perception. Accordingto Kibria (1998:75), `South AsianAmericans have a sense of profound racialdifference fromother ,’ and this sense ofdifference isreciprocated by otherAsian Americans (Kibria 1996;Shankar and Srikanth 1998b).Instead, South Asians® nd themselvestreated at different timesas blacks or as whites (Davis 1991;Dworkin and Dworkin 1988; Gwaltney 1993; Mazumdar 1989). 5 Perhaps morecommon than their assignmentto one de® ned racialgroup is South Asians’treatment as ` ambiguousnon-whites’ (Kibria 1996:79), a term which neatlyre¯ ects the powerthat both external opinion andinternal self- identi® cationhave had in situatingthem in the Americanracial landscape. For if the `non-white’ label hasbeen assuredby USsocietyat large, the `ambiguous’ positioningappears to a considerableextent to have been the productof concertedeffort onthe partof South AsianAmericans. Kibria (1996:81) describestheir `ideologicaldisengagement fromthe USracialorder’ andGeorge (1997:31) observes `a certain reluctance to acknowledge aracialidentity for oneself andfor the communityat large.What is refused by nearly allupper and middle classSouth Asiansis not so much aspeci® cracialidentity but the very ideaof being raced.’Intentional obfuscation can also be readin the arrayof labelsthat Indian Americanshave offered when pressed fora racialidenti® er. Termssuch as `Aryan’,`Dravidian’,and` Indo-Aryan’ (Fisher 1980;George 1997; Kibria 1996)are more than simply notionswith some currency in the subconti- nent; they arealso obviously out of place andmeaningless in the American context.For all intents and purposes, claiming `Dravidian’ statusis like claiming noracial status in the United States.Such racialmarginalisation (Kibria 1996; Rajagopal1997) may be preferred toracialisation,however, if the racialgroup to whichone risksbeing appended isa low-statusone. Indeed, manysee South Asians’`refusal ofracialidentity’ (George 1997:32) as areactionto their non-white status.Rajagopal (1997: 52), for example, claims that:`Denied full acceptanceinto white society,they assumea deliberately blurred perception oftheir relationshipto the majoritysociety.’ On the other hand,George (1997)strongly refutes thisconnection, contending: `Upper-class andupper-caste Indiansin the USdonot want to pass for white orto escape blacknessas much asthey wishto move unconsciously and unobstructed throughthe public sphere, asthey doin .’ Integrally tied tothe controversialcontention that South Asianswish to be Theracial self-identi® cation ofSouth Asians 65 identi® ed aswhite isthe less-contestedassertion that they tryto distance themselvesfrom blacks and . Rajagopal (1997: 52) links the two explicitly when he writes,` Even asthey insiston being calledª brownº, the plea ofIndian immigrants not tobe calledblack is what is most audible. They aspire towardwhiteness ¼ ’Many researchershave noted the existence ofconsiderable anti-blackand anti- prejudice amongSouth Asians(Dworkin and Dworkin1988; Mazumdar 1989; Prashad 2000; Singh 1996),often accompanied by afear ofbeing mistakenfor black or Hispanic that Mazumdar (1989: 51) has characterisedas ` analmost paranoid response to even being thoughtof as black’.6 Although thisfear betraysbigotry, it is not necessarily unfounded. As one Indian Americaninterviewee put it,`This societywill place themin the samestatus as blacks in the blink ofan eye’ (Rajagopal1997: 52). And South Asiansare not the only onesto hold this view; one ofGwaltney’sAfrican-Amer- icanrespondents said of South Asians,` Alotof them don’ twantto admit their colorbecause they areafraid that these whitefolksover here wouldgive thema hardtime. Now they areright about that’ (cited in Davis1991: 162). As Singh (1996)points out, immigrants from South Asiawould not be the ®rstto avoid underdog groupsin orderto climb the Americansocial ladder. 7 The discussionabove points to the conclusionthat despite of® cial classi®cation as Asian, the debate overthe racialdesignation of South Asiansis farfrom over, both within and outside the community,and the meaningful pointsof reference arewhite andblack, not Asian. This outcome supports the claimby Bashiand McDaniel (1997),Ong (1996)and Song (1998)that all immigrantsto the United States,regardless of countryof origin,are absorbed intoa systemof racial strati® cation that is structured around only one axis:the black-white dichotomy.Ong (1996:751) elaborates: Imaintainthat the white± blackpolarities emergingout of the history of European-Ameri- canimperialism continue to shapeattitudes andencode discourses directed atimmigrants fromthe rest ofthe world thatare associated with racial andcultural inferiority. This dynamicof racial othering emerges in arangeof mechanisms that variously subject non-white immigrantsto whitening orblackening processes thatindicate the degree of their closeness to ordistance fromideal white standards. Thisview suggeststhat of® cial classi® cation of South Asiansas part of the Asian racialcategory represents littlemore than an agreement ofconvenience ora de tente in the struggle between South Asiansand society at large over a more fundamentalissue: where they ®talongthe black±white spectrum.Here the state isnot the realimposer of identity;the externalculprit isinstead the centuries of black±white polarisationthat have left USsocietyunable toconceive of its constituentsin anyother way. Regardless of the federal government’s classi®cation directives, it is local, transnational and situational norms instead ± the `everyday experience’ thatconstitutes the microlevel of racialformation (Omi andWinant 1994: 58) ± thatare more closely bound up in howSouth Asianssee themselvesand are seen by othersin racialterms. And the contextual variety¯ ourishes;you canbe Indian Americanand `white’ onyour driver’slicense, andyou canbe Indian Americanand `one ofus’ in ablack Brooklyncommunity. 8

Individual-leveldeterminants ofracial self-identi® cation: hypotheses Several studiesby Alejandro Portesand colleagues have explored the relation- 66 A. Morning ship between the negotiationof internaland external identities on one hand,and individual socio-economicand demographic characteristics on the other. Although these investigationsstudy ethnic self-identi® cation,their explicit theoreticalincorporation of the pressuresof externalde® nitionmakes them relevantto the studyof racial identity aswell. In a1984article entitled `The riseof ethnicity: determinants of ethnic percep- tionsamong Cuban exiles in ’,Portes et al. examine exiles’ `ethnic awareness’,whichthey de® ne as` the perception by members ofa minorityof the socialdistance separating them from the dominantgroup andthe existence ofdiscriminationbased on racial or culturaldifferences’ (Portes et al. 1984: 384). Thusthese immigrants’sense ofethnicity islinked totheir reception by the host society.More speci® cally,Portes et al. ®nd supportfor the `ethnic competition’ hypothesisthat increased exposure toand knowledge ofthe hostsociety heightens perceptions ofsocial distance and . Therefore education, English pro®ciency, andoccupation ± tothe extent they bring one outof the andinto the mainstreamlabour market ± arepositively associated withethnic awareness. Individuals’ interactionwith the largersociety is also at the heartof Portes andRumbaut’ s(1996)notion of `linear’ versus`reactive’ ethnicity.In thisview, immigrants’self-de® nitionis largely colouredby the culturalprescriptions of their homeland,while their children’sismuch moreattuned to the hostcountry context.More speci® cally,the secondgeneration’ s`reactiveethnicity’ isfostered in largepart by the injustices visitedupon the ®rst.As a result,the ethnic identity ofthe secondgeneration is distinct from the `linear’ identity maintained by their immigrantparents. Individual demographicand socio-economic characteristics are not important solelybecause they governexposure tothe hostsociety, however. Portes and McLeod’s(1996)study of second-generation schoolchildren demonstrates thatthese attributesare also associated with differing capacitiesto resist ethnic labelling imposedby the mainstream.Here Nagel’s (1998)dialectical process of ethnic formation,shaped by the interplay of internaland external de® nition, becomesa `contest’in whichimmigrants and their children attemptto fend off the `symbolicviolence’ ofimposed ethnic identities(Portes and MacLeod 1996:524, 528). In thisstruggle, some groups and individuals are better equipped toresist the externalimposition of labelsthan others. Portes and MacLeod® nd thatuse ofthe `Hispanic’pan-ethnic label, nowwidely applied in the United States,is more readily embraced by someLatin American-origin groupsthan others, with Nicaraguans being the mostreceptive andCubans the least.Within nationalgroupings, children whoare less acculturated (as measuredby their length of residence in the United Statesand their command ofEnglish) andless well off (re¯ected in family incomeand father’ seducation) aremore likely toadopt the Hispaniclabel. 9 These resultssupport the hypoth- esisthat groups and individuals with greater resources can resist outsiders’ labels. Taken together,these empiricalinquiries intoethnicity associatefour broad setsof factors with racial self-identi® cation:socio-economic status (e.g. edu- cation,income, occupation); acculturation (length oftime in the US, English usage,nativity); national origin; and demographic characteristics such as age andsex. Moreover, these variablesare featured in whatcan be termedas three theories: Theracial self-identi® cation ofSouth Asians 67

· ethnic awareness :higher socio-economicstatus and greater acculturation are associatedwith increased perception ofsocial distance from the dominant group; · contest ofethnicities :higher socio-economicstatus, greater acculturation, older age,and being maleall enable individualsand groups to resist the use of mainstream,or externally-imposed, identity labels;national origin also has an impacton self-designation; · generationaldivide :the ®rst,immigrant generation is less likely touse main- streamidentity labelsthan are second-generation individuals.

The translationof these ideasinto hypotheses concerning South Asians’racial self-description isnot straightforward. In particular,they posethe questionof whether there iscurrently amainstream-imposedracial label in the United Statesthat is at odds with one (ormore) internally preferred andde® ned by South Asiansthemselves. If we areto speak ofexternally imposedlabels, we mightconsider the categoriesof ` Black’ and`White’ justsuch terms. Regardless ofthe ostensibledifferences ofdesirability between them,together they consti- tutethe axisof mainstream American understanding of race,and the framework thatSouth Asianswho `refuse tobe raced’would eschew. Therefore Iwill treat South Asianlabels ± recordedon the UScensusrace question as national designations± asthe preferred optionsfor racial self-identi® cation.However, it shouldbe kept in mindthat a signi® cantportion of the literatureon this topic suggeststhat racial designation as white wouldbe preferred even todesignation asSouth Asian. The theoriesat issue can now be operationalisedin the followinghypotheses regarding South AsianAmericans’ racial self-de® nition:

· ethnic awareness :higher socio-economicstatus and greater acculturation are associatedwith a decreasedtendency ofSouth Asiansto self-identify as `White’ andan increased tendency toidentify as`Black’; · contest ofethnicities :higher socio-economicstatus, greater acculturation, older age andbeing maleare all associated with a decreasedtendency toself-ident- ify asboth` Black’ and`White,’ andan increasedtendency toidentify asSouth Asian;national origin also has an impact on South Asians’racial self-desig- nation; · generationaldivide :the ®rst,immigrant generation of South Asiansis less likely touse the mainstreamracial labels of `Black’ and`White’ thanare second-gen- erationindividuals.

Dataand methods Data In orderto assess the hypotheses above,this inquiry utilisesa 1per cent IntegratedPublic Use MicrodataSeries (IPUMS) sampleof the 1990US Census (Ruggles andSobek 1997).For the purpose of investigatingthe racialself- descriptionof South Asianrespondents, this data set offers the advantagesof being nationallyrepresentative, being largeenough toinclude representativesof the fairly smallSouth Asianpopulation (which constitutesroughly 0.3per cent ofthe totalpopulation), and including ancestryand birthplace informationthat permitsthe identi® cationof South Asian-originrespondents. 68 A. Morning

Although the foregoing discussionof racialidentity hasapplied tothe South Asiancommunity as a whole in the United States,the principal analysisbelow focusesexclusively onrespondentsof AsianIndian origin (who made up 80per centof the AmericanSouth Asianpopulation in 1990).The scopeis limited in thisway because only Indians® nd acheckbox (i.e. the `AsianIndian’ category) targetedfor them on the censusform; others must write in their responseson the racequestion (e.g. `Pakistani’).Asa result,Indians face aqualitativelydifferent processof self-identi® cationwhen responding tothe UScensusthan do those of otherSouth Asianbackgrounds. However, the racialreporting ofnon-Indian respondentswill alsobe presented. Respondentsare identi® ed asbeing ofAsianIndian originif they indicatedon the censuseither Indian ancestryor an Indian birthplace.However, this group- ing doesnot include respondentsborn in Indiaif they describe themselvesas havingbeen bornabroad to American parents or if they indicatetheir ancestry tobe entirely fromanother region. An importantexception tothe latterdecision rule, however,is the inclusionof individuals who reported their birthplace as `India’ anddescribed their ancestryas `American Indian’, apparentlyunaware thatthe CensusBureau usesthis term to refer toNative Americans. The total sizeof the Indian-origin subsamplecomes to 6,722 respondents. The reporting andcoding of ancestry responses present particularproblems forthis study. In additionto the problem ofSouth Asiansdescribing their ancestryas `American Indian’ ,itis also quite possible thatCensus Bureau coders madeerrors in assigningwritten-in ancestryresponses of ` Indian’ incorrectlyto either the `AmericanIndian’ or`Asian Indian’ categories.As a result,the delimitationof the Indian-origin communitymay mistakenly include some nativeAmericans and exclude someSouth Asians.I addressthe latterproblem by including asAsian Indian thoserespondents born in Indiawhose ancestry hasbeen recordedas `AmericanIndian’. ButI donot attempt to reverse the ®rst problem ±i.e.American Indians mistakenly designated as Asian Indian ±both because there isno clear guideline forexclusion and because the problem is likely tobe minor.Only 19of the 73Indian-origin respondentswho marked their raceas ` AmericanIndian’ were bornin the United States,so the likelihood issmall that racial self-reporting as`American Indian’ isbeing driven by miscodednative Americans. 10

Methodology The mainvariable of interest in the analysesto follow is the raciallabel chosen by respondents,i.e. the boxchosen from the 1990census’ listof seven main racialgroupings: `White’, `Blackor Negro’, `Indian (Amer.)’, `Eskimo’,`’ , `Asianor Paci® c Islander’,and `Other race’(see Figure 1).The Asiancategory offered aseriesof subcategories, of which `Asian Indian’ wasthe only South Asianoption, and it provided a®ll-in blank sothat one mightwrite in responses suchas `Pakistani’.Topreserve thisinformation, I use asthe dependent variable arecodedmeasure of racial response with the followingfour categories:White, Black,South Asian(grouping here the `AsianIndian’ checkbox aswell as written-in Bangladeshi,Pakistani and Sri Lankanresponses), and Other (contain- ing the remainder). The independent variablesto be consideredare age, sex, education, an occupationalprestige index (Duncan’sSocio-EconomicIndex), family income, Theracial self-identi® cation ofSouth Asians 69

Figure 1. 1990US Census race question language spoken athome, status, and nativity. 11 Ialsoinclude an indicatorof non-South Asianancestry; since respondentswere allowedto listup totwoancestry types, some recorded both a South Asianand a non-South Asian ancestry(e.g. French, Brazilian,Japanese). The analysiswill proceed in twostages. First I drawa subsamplecontaining only the Indian-origin headsof household(n 5 2,090),and use multinomial logisticregression to analyse the relationshipbetween their racialself-descrip- tionand the aforementionedindependent variables.Because household heads arethe onesmost likely to® ll outthe censusform, it is appropriate to consider their responsesan expression of self-identi® cation,but the sameassumption cannotbe extended toother household members, especially children. Yetbe- causethe racialreporting of children, especially the US-born, mayfollow a patterndifferent fromthat of their parents,I present descriptivestatistics on the racialdescription of Indian-origin children aswell asother children in house- holdsheaded by Indian-origin respondents.Finally, variationsin racial self-description by nationalorigin will be brie¯y reportedusing anexpanded sampleof South Asianhousehold heads.

Analyses Theracial self-designation of Indian-origin heads of household Onthe 1990US census,90 per centof Indian-origin headsof household designatedtheir raceas South Asian(see Table 1).This is not surprising given the presence ofan `Asian Indian’ checkbox onthe census,but itrepresents a strikingchange fromthe censusof 1970, which was the lastto lack such an optionor directions expressly forSouth Asianrespondents. In thatyear, nearly three-quartersof the South Asian-originpopulation was designated as ` White,’ due bothto self-classi®cation as white andto the CensusBureau’ sreassignment ofAsianIndians from the `Other’ racecategory to the white one (Espiritu 1992). The structuringof the censusrace question and its tabulation by the Census Bureau clearlyhave a tremendousimpact on the racialclassi® cation of South 70 A. Morning

Table 1. Raceand socio-economic characteristics: descriptivestatistics

Asian Indianheads of household All US Socio-economic heads of characteristics White BlackSouth Asian 1 Other household

Meanage (years) 41.7 41.4 40.3 39.7 48.4 Male (%) 71.1 58.7 90.3 74.0 67.5 Meaneducation 2 11.2 10.0 13.7 11.5 10.4 Meanfamily income ($) 39,24230,605 57,038 44,903 35,710 Socio-economic index (SEI) 3 50.6 31.2 61.0 44.7 43.0 Foreign-born(%) 45.6 58.7 97.8 89.0 9.5 UScitizen (%) 75.6 67.4 43.5 41.1 95.9 Englishmonolingual (%) 73.3 89.1 15.8 20.5 87.9 Non-SouthAsian ancestry(%) 42.2 30.4 1.1 8.2 84.7 N 5 2,090 90 46 1,881 73 917,641

Notes: 1 `SouthAsian’ responses include boththe `Asian Indian’checkbox andwritten-in responses such as`Pakistani’and `Sri Lankan’. 2 Educationallevel codes are:10 5 HighSchool graduate/GED’;11 5 `Somecollege, nodegree’; 12 5 `Assoc. degree (occupational)’;13 5 `Assoc. degree (academic)’;14 5 `Bachelor’s degree’. 3 SEI 5 Duncan’sSocio-Economic Index,scaled from1± 96. The 3.6% of the sample thatreceived anSEI score of 0±meaningthey are not employed ±is excluded here. Source: IPUMS1990 1% unweighted sample.

Asians.However, these factorsdo not tell the whole storyof how South Asians chooseracial identi® ers.As Table 1shows,10 per centof the 1990Indian-origin headsof household identi® ed themselvesas white, black, or otherwise (includ- ing AmericanIndian), despite censusinstructions that directed them to the `Asian Indian’ category.And aswe see later,this ® gure risesto 16per centamong their children. Accordingly,I nowturn to the examinationof how the selectionof raciallabel varieswith individual-level characteristics.

Cross-tabulationsof race andsocio-economic characteristics AsTable 1shows,Asian Indian householderswho choose different raciallabels forthemselves vary considerably in their socio-economicand demographic attributes.Those who describe themselvesas `Black’ showthe lowestlevels of education,family incomeand occupational prestige (Duncan’sSEI), despite havingthe second-highest averageage. This may be due in partto the relatively high shareof female householders:women make up over40 per centof the `Black’ respondentsin thissample. This is also one ofthe twomost acculturated groups,having the highest percentage ofEnglish-only speakers. 12 `White’ re- spondents,however, show the highest ratesof US birth andcitizenship. In contrast,the group whichchose a South Asianlabel onthe racequestion hasthe highest income,occupational and educational levels. Italso shows a limited degree of acculturation:the respondentswho use aSouth Asianracial descriptorhave the highest percentage of foreign-born, the lowestshare of monolingualEnglish speakers,and the second-lowestrate of UScitizenship. Boththe `South Asian’race and `Other’ racerespondents are less acculturated thanthose who have adopted the `Black’ and`White’ labelsthat constitute the primaryaxis of American racial strati® cation. Finally, ancestryalso appears to be associatedwith the raciallabels selected by Theracial self-identi® cation ofSouth Asians 71

Table 2. Raceand socio-economic characteristics:multinomial logit results (relativerisk ratios)

Exponentiated results

White Black Other Independent variables RRRS.E. RRR S.E. RRR S.E.

Age 1.0200.011 1.015 0.013 0.995 0.011 Male 0.9130.311 0.510 0.194 0.481* 0.144 Education 0.874*0.036 0.788* 0.039 0.848* 0.029 Familyincome 1.0000.000 1.000 0.000 1.000 0.000 Foreignbirth 0.086*0.030 0.258* 0.113 0.316* 0.149 Englishonly 4.979*1.488 25.447* 12.927 0.893 0.303 Non-SouthAsian ancestry7.624* 2.987 4.386* 2.073 3.454* 1.844 Pseudo R-square 0.275 N 5 2,090

Note:* Signi®cant at the 95%level. Source: IPUMS1990 1% unweighted sample,Asian Indianheads of household,base category: SouthAsian race. the Indian householdheads. As the lastrow in Table 1shows,individuals who gavea racialresponse other than South Asianwere much morelikely toreport sometype ofnon-South Asianorigin; this was especially true forthose describ- ing themselvesas black or white.

Multinomiallogit results forAsian Indianheads of household Totest the hypotheses outlined earlier,I turnto the multinomiallogit results shownin Table 2.The independent variablesin thisregression include: the demographicvariables of age andsex; the socio-economicvariables of education andfamily income;the acculturation-relatedvariables of nativity and English monolingualism;and ® nally,the variablere¯ ecting mixed ancestry(i.e. report- ing ofsome ancestry other than South Asian).Duncan’s Socio-EconomicIndex (SEI) isnot included due toitshigh correlationwith education and income, and the citizenshipvariable is excluded because ofits strong correlation with nativity. In general, the demographicvariables of age andsex have little effect onthe householders’ choiceof racial category. The socio-economicvariables appear somewhatmore relevant: although family incomeshows no signi® cantrelation- ship withracial identity, education consistently demonstrates a signi® cant associationwith the outcomevariable. More speci® cally,moving up one level in the educationalattainment schema (see Table 1notesfor the educationalcodes) decreasesa respondent’s probability of selecting `White’ insteadof aSouth Asianterm by over10 per cent.The sameincrease in educationallevel is associatedwith an even steeper drop± over20 per cent± in the probability of selecting the `Black’ label insteadof a South Asianone. The variablesrelated to acculturationalso demonstrate signi® cantand sizeable effects onrespondents’ choiceof racialcategory. Foreign-born respondentshave a90per centlesser probability thanthe native-bornof choosing the `White’ label ratherthan a South Asianone, and a nearly 75per centsmaller likelihood of 72 A. Morning

Table 3. Nationalorigin and race

Race selected on1990Census (%)

South Asian

`Asian Indian’Nationality Nationalorigin NWhite Black box write-in Other

Indian 2,0904.3 2.2 88.8 1.2 3.5 Pakistani 299 6.7 0.3 25.8 65.9 1.3 Bangladeshi 53 1.9 0.0 43.4 50.9 3.8 Sri Lankan 38 7.9 0.0 26.3 65.8 0.0 Total 2,4804.6 1.9 79.3 11.1 3.2

Source: IPUMS1990 1% unweighted sample. selecting `Black’ insteadof aSouth Asianterm. Similarly, speaking English only increasesdramatically one’s probability ofself-identifying as`White’ (by nearly 5times)or `Black’ (by afactorof 25) as opposed to South Asian. Finally, the variablere¯ ecting mixed ancestryalso produces striking results. Respondentswho report some ancestry other than (or in additionto) South Asianare over seven timesmore likely thanthose who do not to describe themselvesas ` White’ ratherthan South Asian,and over four timesmore likely tocall themselves `Black’ thanSouth Asian. In sum,education, acculturation and mixed ancestryall seem torelate stronglyto Asian Indian householdheads’ choiceof raciallabel onthe census± even in acontextwhere the framingof the questionpoints them to one category abovethe rest.However, this group constitutesonly aportionof the larger South Asiancommunity of all ages and family statusto which we mightwish togeneralise these ®ndings.In the nextsection, simple descriptivestatistics are presented toshed somelight onthe racialidenti® cationof otherSouth Asian headsof household, and ® nally onAsian Indian children, the next generation.

Takingnational origin and generation into account The hypothesisthat national origin affects choice of racial label cannotbe thoroughlyexplored here, bothbecause the sampleof non-Indian South Asians isa very smallone, and because ±asnotedearlier ±the racequestion on the 1990 censusis structured differently forIndians than it is for other South Asians. However,Table 3presentsthe distributionof racialresponses for the four largestSouth Asiangroups, from which at least two points of interestemerge. First,there appearto be distinctdifferences in the patternsof race reporting acrossnational origin, although the smallnumbers ofsomeof the groupsentail largecon® dence intervalsaround these estimates(not shown). Indians display the highest likelihood ofselecting the `Black’ category,while Sri Lankans followedby Pakistanisare most likely todescribe themselvesas white. Second, the extent towhich non-Indian groupsused the `AsianIndian’ checkbox to describe themselvesis considerable, and it also varies by nationalorigin. Only half ofthe respondentsof Bangladeshi origin identify themselvesas such; roughly the otherhalf simply checks` AsianIndian’. Atleast two explanations maybe given forthis `borrowing’ ofthe Indian checkbox.One maybe that Theracial self-identi® cation ofSouth Asians 73

Table 4. Generation, race,nativity and ancestry

Nativity andancestry

US-born Foreign-born

Mixed Nomixed Mixed Nomixed Generation Race ancestryancestry ancestry ancestry All

Householdheads South Asian 7.8 56.1 60.7 93.7 90.0 Other 92.2 43.9 39.3 6.3 10.0 100.0 100.0 100.0 100.0 100.0 N 5 51 66 28 1,945 2,090

Children South Asian 21.2 92.5 41.4 91.3 83.7 Other 78.8 7.5 58.6 8.7 16.3 100.0 100.0 100.0 100.0 100.0 N 5 255 1,333 29 713 2,330

Notes:1. The sample of children includes all children of Asian Indian-originhousehold heads,as well asall other children identi® ed asbeingof Indian origin, regardless of the ancestryor birthplace of their household head. 2.Herethe `Other’ categoryof race includes all responses other thana SouthAsian one(e.g. White, Black,etc.). 3.`Mixed ancestry’re¯ ects reports of some ancestryother thana SouthAsian one. Source: IPUMS1990 1% unweighted sample,Asian Indianhouseholders andchildren. respondentsgravitate to the printed categorythat most closely resembles their self-de® nition,even if itis not a perfect ®t.In thissense, the `AsianIndian’ categoryon the UScensusbecomes a pan-South Asianumbrella forall respon- dentswith roots in the subcontinent.The second,and likely simultaneous, explanationis that South Asianrespondents who report their ancestryor birthplace outsideIndia but indicate`Asian Indian’ raceare a testimonyto the shifting of geopoliticalborders in South Asiaover the courseof the twentieth century. Itis likely thatmany respondents in thissample of heads of household were bornin whatwas once India but isnow or . Letus now examine generation.As explained above,the racialreporting of householdmembers otherthan the head,and especially minorchildren, ismuch lesslikely tore¯ ect self-identi® cationchoices. Nevertheless, the racialdescrip- tionof youthmay still provide someinsight into the waysthat future changes in the nativitybalance of the South Asianpopulation ± immigrantversus native-born± mayaffect the group’s racialpro® le in the United Statesas a whole. Table 4comparesthe racialreporting ofAsian Indian householdersto that of asampleof children 18and younger thatincludes the children ofthese householders,as well asother children whoare identi® ed asbeing ofAsian Indian origin(by birthplace orancestry), regardless of the originof their heads ofhousehold. 13 The lastcolumn of Table 4indicatesthat the children are60 per centmore likely thanthe householdersto be raciallyidenti® ed assomething otherthan South Asian. Cross-tabulatingboth samples by nativity(US-born versusforeign-born) and ancestry(mixed ancestryversus no mixed ancestryreported) offers someinsight intothese differing patternsof racialidenti® cation.Among the householders,the 74 A. Morning vastmajority (93 per cent) ismade up offoreign-born respondentswho do not reporthaving any ancestry other than South Asian.This group, in turn,is unlikely touse anyracial descriptor other than a South Asianone: only 6per centdeviate from this pattern. 14 In contrastto the adults,the sampleof children ismuch morediverse: only 30 per centare foreign-born respondentswho do not report any non-South Asian ancestry.Instead, the largestcomponent of thissample is made up of US-born children withoutmixed ancestry,but they makeup only alittleover half ofthe sample.However, neither ofthese groupsdemonstrates patterns of racial reporting thatare substantially different fromthose of the headsof household. Instead,the greatestcontributor to the differences between householders’ and children’s racialreporting overallappears to be the higher degree ofmixed ancestryamong the children. Respondentswho list some ancestry other than South Asianmake up lessthan 4 per centof the householder sample,but they contributeover 12 per cent± three timesas many ± of the sampleof children. Amongthese mixed-ancestrychildren, only 23per centare identi® ed asSouth Asianon the racequestion; instead, 54 per centare reported as `White’ and almost10 per centas `Black’ (detailsnot shown). This result is broadly consistent withAllen andTurner’ s(1996)® nding thatover 80 per centof the children of AsianIndian/ white marriageswere listedas `White’ onthe 1980US census± a higher rateof white identi® cationthan for any other Asian/ white combination.

Concludingdiscussion Together,the multinomiallogistic regression on the AsianIndian householder sample,coupled withthe descriptiveresults for all South Asianhousehold heads andfor Asian Indian children, yield mixed resultsfor the three hypotheses outlined earlier.The `generationaldivide’ hypothesisreceives the mostsupport, asimmigrants appear more likely thanthe US-born tobe raciallyidenti® ed as South Asianin boththe samplesof Indian householdersand children. However, in the sampleof children, the theorised` linear’ versus`reactive’ ethnicity divide doesnot seem tobe atwork so much asdifferences in the degree ofmixed ancestry.The absence ofa`linear’/`reactive’contrast among the children maybe due tothe factthat among them, nativity does not distinguish immigrants from their children, but ratherit marks ` child immigrants’from ` immigrants’children’ ±thatis, the `one-and-a-half’ generationfrom the second. Some evidence appearsfor differences in racialself-identi® cationby national origin,but the limitationsof the data± namely,small sample sizes for the non-Indian groups± preclude de® nitive conclusions. Perhaps the moststriking pattern to emerge fromthe analysisof Indianheads of householdis the consistentlyparallel directions in which the probabilitiesof reporting as`White’ or`Black’ areaffected by the independent variables.In other words,the datarefute the implicitassumption behind the `ethnic awareness’ hypothesisthat the probability ofadopting the blackracial label isinversely relatedto the probability ofadopting the white one.Instead, independent variablesthat increase the likelihood of selecting `White’ alsoincrease the likelihood ofselecting `Black’. The regressionresults vindicate the assumptionembodied in the `contestof ethnicities’ hypothesisthat `Black’ and`White’ constitutebinary componentsof asingle systemof raciallabelling, one towhich the use ofSouth Asianlabels Theracial self-identi® cation ofSouth Asians 75 standsas an alternative. Still, the supportfor this hypothesis is mixed. While socio-economicstatus (when measuredby educationalattainment, but not income) ispositively correlated with the probability of identifying asSouth Asian,greater acculturation (as measured by USbirthand English monolingual- ism)increases the likelihood of using mainstream,non-South Asianlabels. Nor dothe age andsex variables provide supportfor the `contest’idea. Although none ofthe three hypotheses receives unconditionalsupport in this analysis,several ® ndings meritfurther explorationin connectionwith likely future developments in the AmericanSouth Asiancommunity. First,the South Asianpopulation of the United Statesis growing more diverse in termsof national origin (Morning 2001).A growingshare of Pakistanis, for example, mightraise the overallrate of self-identi® cationas white in the South Asiancommunity, as well asdiscourage any trend toward using the `Asian Indian’ censuscategory as a pan-ethnic South Asianidenti® er.In contrast,a growingBangladeshi presence mightproduce the oppositeeffect. And either development mightaffect whichracial categories are listed on the censusform itself in the future. In additionto becoming morediverse in termsof nationality,immigrant in¯ows from South Asiancountries have also become moresocio-economically andoccupationally heterogeneous since the early1970s. Although the multino- mialanalysis conducted here found noevidence of associationbetween family incomeand racial label, itdid reveala considerablein¯ uence by educational attainment.Among the Indian householderswith a Bachelor’sdegree orhigher (69per centof the sample),over 95 per centdescribed themselvesas South Asian,but amongthe less-educatedhouseholders, less than 80 per centdid. Thusshifts in the educationalattainment of the South Asiancommunity might modify itsracial reporting patterns.Greater occupational and income diversity couldhave a similareffect, asa largershare of working-class members might expose the South Asiancommunity to greater discrimination. Misir (1996: 57) writes:

Working-classor state school-educated second generationIndian Americans donot see a naturalalliance orunity with other Asian American groups.They do know that when they arecalled `Dothead’ the slur is meantfor them, and others who look like them.These young IndianAmericans aremore inclined thantheir elite peers to socialize exclusively with each other attheir universities’ largeHindu youth, South Asian students orIndian cultural groups.

Thusthe classevolution of the South AsianAmerican population warrants future inquiry. The `generationaldivide’ notionthat second and later generations have signi® cantlydifferent attitudestoward racial self-de® nitioncompared to the ®rst generationhas important implications for the South Asiancommunity in the United States.As a relativelynew, post-1965immigrant ¯ ow,this group cur- rently haslittle generational diversity; in 1990it was largely a®rst-generation community.The extent towhichthis remains the casewill depend largely onthe future volumeof immigration (and emigration) and on South AsianAmerican fertility rates.Assuming, however, that the ®rstgeneration will cometo rep- resentan ever-declining shareof the totalSouth Asiancommunity, we would expect the use ofthe racialterms ` White’ and/or` Black’ toincrease, all else equal. When allAsian Indians from the 1990census sample are considered, 76 A. Morning regardlessof age orhousehold status, and the children of allIndian household headsare included aswell regardlessof their reportedancestry and birthplace, 83per centof this sample of 7,758 describe themselvesas South Asian.Among the US-born segment ofthis sample, however, only 65per centuse aSouth Asianterm. Instead, 25 per centof the secondgeneration is identi® ed as`White’, and5 per centas `Black’ . Bothmixed ancestryand some type of`reactive’ethnicity appearto contribute todifferences in racialreporting between ®rst-generationimmigrants and their second-generationoffspring. However,it is an unusual kind ofreactiveethnicity thatresults in the secondgeneration being morelikely toracially identify itself withthe dominantwhite group. In Portesand Rumbaut’ soriginalmodel (1996), the ®rstgeneration faces considerable discrimination, which mobilises the se- condgeneration to de® ne itself asdistinct from the dominantgroup. Butthe highly af¯uent Indian Americancommunity of the latetwentieth century, concentratedin professionaloccupations yet smallin numbers andgeographi- callydispersed, has not faced the samevirulent prejudice asothercontemporary immigrantgroups. As a result,the secondgeneration’ sreactionto the American raciallandscape appears to be fuelled lessby the ®rstgeneration’ sexperience of discriminationand more by itspurposeful disengagement fromtraditional Americanracial schema. Where George (1997)® ndsthe refusal tobe raced especially adamantamong the ®rstgeneration, Mazumdar (1989: 53) hypothe- sisesthat `US-born children maynot be able toescape colorprejudice aseasily asimmigrant parents’. Kibria (1996:81) sums up the generationalidentity difference in the followingmanner:

`Native’ conceptions of racial identity arelikely to havefar less meaningfor second-gener- ationSouth Asian Americans thanthey do forthe immigrantgeneration. Particularly given their exposure andsocialization into apost-civil rights political environment, onethat has beencharacterized by a heightened consciousness of race,second-generation may® ndit more necessarythan their parentsto directly confront the dynamics of U.S.racial thinking.

Finally, if the South Asianpopulation in the United Statescontinues to growin bothabsolute and relative terms, attitudes about racial group member- ship arelikely tochange aswell. Kibria (1996,1998) suggests at least two ways in whichthis might take shape. First, in largernumbers South Asiansmay be increasingly perceived asan economic threat, either towhites or to racial minoritygroups. In thisinstance, the sizeof the populationinteracts with its classcomposition to determine whichgroups are most likely tosee South Asians ascompetitors. This competition and possible con¯ict could be instrumentalin redrawingracial boundaries. Second, Kibria predictsthat a largerSouth Asiancommunity will make`greater effortsto gain a politicalvoice in the United States,’and that as partof thisprocess, it will increasingly turnto `racial self-de® nitionand positioning’ (Kibria 1998:73). The ®ring ofracial and ethnic entitiesin the furnace ofpolitics has been observedin countlessinstances bothin the United Statesand abroad (Bell 1975;Horowitz 1985; Portes and Rumbaut1996). The commonalityof thisprocess, coupled withthe United States’long historyof de® ning allits citizens along racial lines, makeit unlikely thatany South Asian` refusal tobe raced’will survivelong pastthe second generation. Theracial self-identi® cation ofSouth Asians 77

Acknowledgements Iwouldlike tothank Dr AlejandroPortes, Ms Jyoti Thottam and an anonymous reviewer fortheir thoughtful commentson previous drafts. This investigation wassupported by the USNationalInstitutes of Health, National Research Service AwardT32HD01763 from the NationalInstitute for Child Healthand HumanDevelopment.

Notes 1Interview, 17June 2000, DC. 2SouthAsia is commonlyunderstood to include Bangladesh,, India, , Pakistan and ;, Burma and the arealso included bysome writers. Iuse the term `SouthAsian’ to refer to individuals with ancestralorigins in this region,but who donot necessarily identify themselves aspart of aSouthAsian collectivity. As NazliKibria (1996:77) notes,the term is `highlyproblematic, maskingdeeply salient divisions of nationality,culture, religion andlanguage’ . 3But see Omi andWinant (1994) and Bashi and McDaniel (1997)for more comprehensive discussion of the differences between race andethnicity. 4Afghansare the exception: the CensusBureau classi® es them as`White’. 5Ablackformer sailor told Gwaltney (1993:90) of his travels in Madras,India and Alexandria, Egypt:`¼ theygot some stone blackcitizens in bothof them towns, not to mention the places in between `em!Hell, no,they ain’ white, an’if youcould see `em yousho’ wouldn’ tbeaskin’ noquestion like that.’ 6Ironically, this reaction maybind South Asians closer to the blackcommunity, by usingthe latter asa constantpoint of reference (Rajagopal1997: 52). 7Singh(1996: 101) aptly cites GeorgeS. Schuyler’s1929essay, `Our Greatest Gift to America’, which refers to the boonthat came in the formof `the sense ofsuperiority over blackswhich new Europeanimmigrants were ableto maintainin adjustingto the painful realities oftheir American existence’. See alsoPrashad (2000). 8Thanksto Jyoti Thottamfor these anecdotes. 9Portes andMacLeod (1996) also ® ndthat girls aremore likely thanboys to accept the pan-ethnic label,as are younger children. 10Miscoded native American Indianancestry might also be in¯ating the counts of those reporting aswhite orblack, by mistakenly including Europeanand who intended to report native .However, since halfof the household headswho report aswhite orblack are foreign-born, it seems unlikely thatmiscoded Europeanand African Americans could approachanything near a majorityof these categories. 11Nativity ±i.e. whether oneis foreign-bornor US-born ± alsotends to re¯ect immigrant generation.Although normallybeing born in the USis insuf® cient to indicate whether oneis second-,third- ora later generation,the relative recency of immigration in¯ows fromSouth makeit afairlysafe assumption thatthe bulkof US-bornrespondents of SouthAsian origin are second-generation.The extremely youngage structure of this native population supports this assumption:its meanage is 12years, and the 75 th percentile amere 15years old. 12The respondents who describe themselves as`American Indian’± not shown separatelyhere but ratherincluded in the `Other’ category± alsohave relatively low socio-economic levels, but they show amuch lesser degree of acculturation. This choice of label probablyre¯ ects forthe most partan unfamiliarity bothwith of®cial US documentation andwith the nationalpractice of referring to indigenous peoples asAmerican Indian. 13Note thatthis procedure includes all children of Indianhousehold headsbut does not automati- cally include the children of Asian Indianswho arenot household heads.Unless theyare otherwise identi® ed asIndianthrough their birthplace orancestry data, children of Asian Indians who aremarried to non-Indianhousehold headsare excluded. 14It should benoted here thatalthough US-born respondents makeup less than6 per cent of the householders’ sample,this groupappears to havea strikingly highlikelihood of usingracial labels other thanSouth Asian. This probably stems in partfrom some erroneous inclusion of individuals of native American Indianancestry, due to the complications of census coding discussed earlier in the paper. 78 A. Morning

References Alba,R.D. (1985) : Into theTwilight of Ethnicity .Englewood Cliffs, NJ:Prentice- Hall. Allen, J.P.and Turner, E. (1996)`Racial andHispanic Identity Preferences forChildren of Mixed Heritage’.Paper presented atthe annualmeeting of the Association of American Geographersin Charlotte,, April 1996. Banton,M. (1983) Racialand Ethnic Competition .Cambridge:Cambridge University Press. Bashi,V. andMcDaniel, A.(1997)`A theoryof immigration andracial strati® cation’, Journal of Black Studies,27(5):668± 82. Bell, D.(1975)`Ethnicity andsocial change’,in Glazer,N. andMoynihan, D. (eds) Ethnicity:Theory andExperience .Cambridge: Press, 141±74. Davis, F.J.(1991) Whois Black? One Nation’s De® nition .University Park,PA: PennsylvaniaState University Press. Dworkin, A.G.and Dworkin, R.J.(1988) `Interethnic of acculturating Asian Indiansin the United States’, PluralSocieties ,18(1):61± 70. Espiritu, Y.L.(1992) Asian American :Bridging Institutions and Identities. : University Press. Fisher,M.P. (1980) TheIndians of City:A Studyof Immigrants from India . New : Heritage. George,R.M. (1997) ª `Fromexpatriate aristocrat to immigrantnobodyº : SouthAsian racial strategies in the SouthernCalifornia context’, Diaspora,6(1):31± 60. Gwaltney,J.L. (1993) Drylongso:A Self-Portrait of BlackAmerica. New York:The New Press. HaneyLo Âpez,I.F. (1995) `White bylaw’, in Delgado,R. (ed.) CriticalRace Theory:The Cutting Edge . Philadelphia: Temple University Press, 542±50. Horowitz,D.L. (1985) EthnicGroups inCon¯ ict .Berkeley:University of CaliforniaPress. Ignatiev,N. (1995) How theIrish Became White .New York:Routledge. Jacobson,M.F. (1998) Whitenessof aDifferent Color: EuropeanImmigrants andthe Alchemy of Race . Cambridge,MA: Harvard University Press. Jenkins,R. (1997) RethinkingEthnicity: Arguments and Explorations. London:Sage Publications. Kibria,N. (1996)`Not Asian, black or white? Re¯ections onSouth Asian American racial identity’, Amerasia Journal ,22(2):77± 86. Kibria,N. (1998)`The racial gap:South Asian American racial identity andthe ’, in Shankar,L.D. and Srikanth, R. (eds) APart, Yet Apart: SouthAsians inAsian America , Philadelphia: Temple University Press, 69±78. Lee,S.M. (1993) `Racial classi® cations in the U.S.census: 1890±1990’, Ethnicand Racial Studies , 16(1): 75± 94. Marx,A.W. (1998) MakingRace andNation: AComparison of SouthAfrica, the United States, andBrazil . Cambridge:Cambridge University Press. Mazumdar,S. (1989)`Racist responses to racism:the Aryanmyth and South Asians in the United States’, SouthAsia Bulletin ,9(1):47± 55. Mediratta,K. (1999)`How doyou say your name?’, in Min,P.G. and Kim, R.(eds) Strugglefor Ethnic Identity:Narratives byAsian American Professionals .Walnut Creek,CA: Altamira Press, 77±86. Misir, D.N.(1996) `The of NavrozeMody: race, violence, andthe searchfor order’, Amerasia Journal,22(2):55± 76. Morning,A. (2001)`’, in Encyclopediaof American Immigration .New York:M.E. Sharpe. Nagel,J. (1986)`The political construction of ethnicity’, in Olzak,S. andNagel, J. (eds) Competitive EthnicRelation s.Orlando, FL: Academic Press, 93±112. Nagel,J. (1998)`Constructing ethnicity: creating andrecreating ethnic identity andculture’, in Hughey,M.W. (ed.) New Tribalisms: TheResurgence of Race andEthnicity .New York:New York University Press, 237±72. Omi, M.andWinant, H. (1986) RacialFormation inthe United States .New York:Routledge &Kegan Paul Inc. Omi, M.andWinant, H. (1994) RacialFormation inthe United States: From the1960s to the1990s . New York:Routledge, 2ndedition. Ong,A. (1996)`Cultural citizenship assubject-making: immigrants negotiate racial andcultural boundariesin the United States’, CurrentAnthropology ,37(5):737± 62. Portes, A.andMacLeod, D. (1996)`What shall Icall myself? Hispanic identity formationin the second generation’, Ethnicand Racial Studies ,19(3):523± 47. Portes, A.,Parker, R.N. and Cobas, J.A. (1984) `The rise ofethnicity: determinants of ethnic perceptions amongCuban exiles in Miami’, American SociologicalReview ,49(3):383± 97. Theracial self-identi® cation ofSouth Asians 79

Portes, A. andRumbaut, R.G. (eds) (1996) Immigrant America: APortrait .Berkeley:University of CaliforniaPress, 2ndedition. Prashad,V. (2000) TheKarma of Brown Folk .: University of MinnesotaPress. Rajagopal,A. (1997)`Transnational networks andHindu nationalism’, Bulletinof ConcernedAsian Scholars,29(3):45± 58. Ruggles,S. andSobek, M. (1997)`Integrated Public Use MicrodataSeries: Version 2.0’.Minneapolis: University of ,Historical CensusProjects, URL:http:/ /www.ipums.umn.edu/ Sacks,K. (1994)`How did become white folks?’,in Gregory,S. andSanjek, R. (eds) Race. New Brunswick, NJ:Rutgers University Press, 78±102. Shankar,L.D. and Srikanth, R. (eds) (1998a) APart, Yet Apart: SouthAsians inAsian America . Philadelphia: Temple University Press. Shankar,L.D. and Srikanth, R. (1998b)`Introduction: closing the gap?South Asians challenge ’, in Shankar,L.D. and Srikanth, R. (eds) APart, Yet Apart: SouthAsians inAsian America.Philadelphia: Temple University Press, 1±22. Singh,A. (1996)`African Americans andthe new immigrants’,in Bahri,D. andVasudeva, M. (eds) Between theLines: South Asians andPostcoloniality .Philadelphia: Temple University Press, 93±110. Song,M. (1998)`Pahkar Singh’s argument with Asian America: color andthe structure ofrace formation’,in Shankar,L.D. and Srikanth, R. (eds) APart, Yet Apart:South Asians inAsian America . Philadelphia: Temple University Press, 79±102. Takaki,R. (1989) StrangersFrom aDifferent Shore:A History of Asian Americans .New York:Penguin. Visweswaran, K.(1997)`Diaspora bydesign: ¯ exible citizenship andSouth Asians in U.S.racial formations’, Diaspora,6(1):5± 29. Waters,M. (1990) EthnicOptions: Choosing Identities in America .Berkeley:University of Press.

Author details AnnMorning is completing herPhD in Sociology atPrinceton University. Shemay be contacted at the following address: Of® ce of Population Research 229Wallace Hall Princeton, NJ08544 USA E-mail:[email protected]