Available via license: CC BY 4.0
Content may be subject to copyright.
Languages2020,5,10;doi:10.3390/languages5010010www.mdpi.com/journal/languages
Article
LinguisticDiversity,Multilingualism,andCognitive
Skills:AStudyofDisadvantagedChildreninIndia
IanthiMariaTsimpli
1,
*,MargreetVogelzang
1
,AnushaBalasubramanian
1
,
TheodorosMarinis
2,3
,SuvarnaAlladi
4
,AbhignaReddy
4
andMinatiPanda
5
1
TheoreticalandAppliedLinguisticsSection,FacultyofModernandMedievalLanguages&Linguistics;
UniversityofCambridge,CambridgeCB39DA,UK;mv498@cam.ac.uk(M.V.);ab2668@cam.ac.uk(A.B.)
2
DepartmentofLinguistics;UniversityofKonstanz,78457Konstanz,Germany;t.marinis@uni‐konstanz.de
3
SchoolofPsychology&ClinicalLanguageSciences;UniversityofReading,Reading,RG66AL,UK;
t.marinis@reading.ac.uk
4
DepartmentofNeurology,NationalInstituteofMentalhealthandNeuroSciences(NIMHANS),
Bengaluru,Karnataka560029,India;alladisuvarna@hotmail.com(S.A.);meenu.6692@gmail.com(A.R)
5
SchoolofSocialSciences,JawaharlalNehruUniversity,NewDelhi,Delhi110067,India;
minatipanda@mail.jnu.ac.in
*Correspondence:imt20@cam.ac.uk
Received:16February2020;Accepted:11March2020;Published:16March2020
Abstract:MultilingualismandlinguisticdiversityarethenorminIndia.Althoughstudieshave
shownarelationbetweenbilingualismandcognitivegains,linguisticdiversityhassofarbeen
ignoredasapotentialfactoraffectingcognitiveskills.Thisstudyaimstofillthisgapbyexamining
howcognitiveskills—asmeasuredbythen‐backandRaven’sColoredProgressiveMatricestasks—
areaffectedbymultilingualismand/orsociolinguisticdiversityinalargecohortof
socioeconomicallydisadvantagedprimaryschoolchildrenintwourbansitesofIndia:Delhiand
Hyderabad.Wepresentaquestionnaireestimatingsociolinguisticdiversityandshowthatthis
measureassessesadistinctconstruct,ascomparedtoachild’smultilingualism.Childrenwere
classifiedasgrowingupmonolinguallyorbilingually,dependingonwhethertheygrewupwith
oneormorelanguagesinthehome.Regardingcognitiveperformance,bilingualswerefoundto
outperformmonolingualsonthen‐backtask,aswellasontheRaven’stask.Inaddition,asocially
andlinguisticallydiverseenvironmentseemstoenhancecognitiveperformanceforchildrenwho
arenotmultilingualthemselves.Finally,severalcontextualfactorssuchascitywerefoundto
influencecognitiveperformance.Overall,thisshowsthatcognitivetasksaresubjecttocontextual
effectsandthatbilingualismandlinguisticdiversitycanenhancecognitiveperformanceofchildren
indisadvantagedcontexts.
Keywords:bilingualism;linguisticdiversity;cognition;workingmemory;fluidintelligence;
disadvantagedcontexts;socio‐economicstatus
1.Introduction
Languageandcognitivedevelopmentinteractinnumerouswaysfrombirthonwards.In
monolinguallearners,earlycognitiveskillsarenecessaryforvocabularydevelopment(e.g.,Clark
2004),whilelanguageabilitiesaresimilarlyimportantforthedevelopmentofcognitiveskillssuchas
theoryofmind(DeVilliers2007)andverbalreasoning(GentnerandLoewenstein2002).In
bilingualism,muchrecentresearchhasfocusedontheimpactofduallanguagedevelopmentanduse
oncognition,revealingconflictingornullresultsacrossstudies(EngeldeAbreuetal.2012;Sanchez‐
Azanzaetal.2017;Paapetal.2017).However,researchonmultilingualismandcognitiveeffectsis
sparse,possiblyduetotheassumptionthatmultilingualismiscumulativebilingualism,inthesense
Languages2020,5,102of21
thatmorelanguagesentailmoreswitchingcostsand,byextension,havesimilareffectsoncognition
inthemultilingualspeaker.Sincebilingualismandmultilingualismareformsoflanguage
experiences,itisessentialtoevaluatehowtheseexperiencesaffectanindividual’slinguisticand
cognitiveprofile.Inresearchonchildbilingualism,therehasbeenconsiderableprogressinhowto
measurelanguageexperience,understoodasquantityandqualityofinputandoutputineachofthe
child’slanguages(Byers‐Heinleinetal.2019;Bedoreetal.2012;Unsworth2013).Inthesemeasures,
quantifyingandqualifyinglanguagepracticesathomeandtheschool,dependingonthechild’sage,
areofprimaryimportance.Assumingmultilingualismtobeanextensionofbilingualism,wewould
expectthatthesamemeasures(addingoneortwolanguagestothequestionsincluded),should
sufficetoprofilethemultilingualspeakerandunderstandpossibleeffectsontheirlinguisticand
cognitiveskills.Despitethesoundnessofthisassumption,someconsiderationshavebeenleft
unexplored.First,linguisticdiversityatthesocietallevelmayormaynottranslateasmultilingualism
attheindividuallevel.Forinstance,achildmaybesurroundedbyspeakersofdifferentlanguages,
butmaynotbeproficientenoughtousemostoranyoftheselanguagesthemself.‘Knowledge’inthis
case,ismostlyawarenessofdiversity,whiletheactuallinguistic,receptive,orproductiveskillsofthe
individualmayvaryalongacontinuumfromzerotofullability.Non‐westernmultilingualsocieties
suchasIndiapresentacaseinpoint:ThelanguageexperienceofanIndianchildcannotignore
contextuallinguisticdiversityasinstantiatedinthecommunity,theschool,andthehome
environmenttowhichthechildisexposed.Linguisticdiversitycanplayasanimportantroleas
languageinputforthechildgrowingupinamultilingualsociety—andcrucially,shapesthechild’s
understandingoftheroleoflanguage(s)inthesocio‐culturalcontext,theirsensitivityto
sociolinguisticvariation,andtheprecociousdevelopmentofmetalinguisticskills(Ochsand
Schieffelin2008;Mohanty2019).Althoughstudieshaveshownthatproficiencyineachofthe
bilingual’slanguagesiscrucialforcognitivegains,especiallyinthedomainofmetalanguage
(Pattanaik2004;PattanaikandMohanty1984;Bialystok2013),contextuallinguisticdiversityhasso
farbeenignoredasapotentialfactoraffectingcognitiveskillsintheindividual.Thisstudyaimsto
fillthisgapbyexamininghowcognitiveskillsareaffectedbymultilingualismand/orsociolinguistic
diversityinalargecohortofsocioeconomicallydisadvantagedprimaryschoolchildrenintwourban
sitesofIndia,Delhi,andHyderabad.
1.1.TheIndianContext
MultilingualisminIndiaisthenorm,asislinguisticdiversity,withclosetoonethousand
indigenouslanguagesbelongingtofourmajorlanguagefamilies.Ofthese,theIndo‐Aryanand
Dravidianfamiliesincludethelanguagesspokenby95%ofthepopulation(1971,Censusdata
reportedinVasantaetal.2010).AccordingtothePeople’sLinguisticSurveyofIndia
(http://www.peopleslinguisticsurvey.org/)launchedbyDevyin2010,thereare780different
languagesinIndia,manyofwhichareendangered1.Mostrecently,Ethnologuesuggeststhenumber
ofdifferentlanguagesislower,andstandsat462(SimonsandFennig2018).Thislinguisticdiversity
hasbeenamajorconsiderationbystateandcentralgovernmentsinrelationtoeducationandnational
policies,aswellasinrelationtotheroleofEnglishasalanguageofpower.Englishisstrongly
supportedbystatepolicies,especiallyincaseswhereconflictsbetweendifferentlanguagegroupsare
resolvedthroughacommonlyacceptedchoiceofEnglishasthemediumofinstruction(Mohantyet
al.2009;Mohanty2019).ParentsandteachersacknowledgeandsupporttheroleofEnglish,with
parentalpreferencesbeingparticularlystrongforEnglish‐mediuminstructioncurricula.This
1 InIndiathereisnoonenationallanguage.TheIndianconstitutioninits8thschedulerecognizes22languages
asscheduledlanguages.TheyaremostlyalsousedasmediaofinstructionasrelevanttodifferentIndian
states.ThelanguagesareAssamese,Bangla,Bodo,Dogri,Gujarati,Hindi,Kashmiri,Kannada,Konkani,
Maithili,Malayalam,Manipuri,Marathi,Nepali,Oriya,Punjabi,Tamil,Telugu,Sanskrit,Santali,Sindhi,and
Urdu.Twooftheseareclassicallanguages,SanskritandTamil,asrecognizedbytheIndianconstitution.
HindiandEnglishfunctionaslinklanguages,withthecentralgovernmentrecognizingHindiastheofficial
languageandEnglishastheprovisionalsub‐language(Devy2018).
Languages2020,5,103of21
multilingualdiversitypresentsbothopportunitiesandchallengestotheeducationsystem,giventhe
hugepopulationandfrequentlylimitedresources.
InurbanareassuchasinNewDelhiandHyderabadwherethisstudywasconducted,the
percentageofminorityspeakersisoverallsmallerthaninruralareassuchasBihar,wherethenumber
ofminorityspeakersisover50%,speaking48differentmother‐tongues(1971censusdata,Bihar).
However,urbanslums—where17%ofurbancitizensinIndialive—includealargenumberof
internalmigrantswhomayspeakothermother‐tongues,orvarietiesoftheregionallanguage.Urban
slumsaresettlementswithinadequateaccesstosafewater,sanitationandinfrastructure,poor
structuralqualityofhousing,overcrowding,andinsecureresidentialstatus.Schoolattendancerates
forchildrenlivinginDelhislumsisaround54%,comparedtoattendanceratiosinDelhischools
overall,whichwas90%in2004(Tsujita2009).Asreportedinthesamestudy,around73%ofslum
childrenattendingStdIinDelhischoolsareover‐aged,forreasonsrelatedtoschoolcapacityand
administrativeissuesincreasingstructuralinequalitiesandaffectingeducationquality.Over‐aged
childrenbringchallengesandbenefitstotheclassroomthough(AlcottandRose2017).
Akeyfactoraffectinglearningoutcomesissocialclass.Morespecifically,socio‐economicstatus
(SES)isoneofthecrucialfactorsinfluencingchildren’scognitiveperformance(BritoandNoble2014;
Ghoshetal.2015;Burneo‐Garcésetal.2019),andchildrenoflowSESaremorelikelytounderperform
thanchildrenfrommoreaffluentmiddle‐classfamilies.AlthoughmiddleclasschildreninIndiaare
morelikelytoattendprivateschools,aspointedoutbytheASER2016report(Pratham,2017),the
differencesbetweenchildren’sperformanceingovernmentschoolsandprivateschoolscannot
necessarilybeattributedtodifferencesintheeffectivenessofthesetwoschooltypes,becausethey
mayalsobeduetotheinterplayofawiderangeofothervariables,includingSES(AlcottandRose
2017).ThereareseveralindicatorsofSES,suchasparentaleducation,income,occupation,facilities
athomeandintheneighborhood,sanitation,aswellasnutritionalandpsychologicalwellbeing
(Hackmanetal.2010;BritoandNoble2014;Royetal.2018).Childrenfromlow‐SESbackgrounds
oftensufferfromlackofnutrition,poorhousingfacilities,andlackofparentalcareandcognitive
stimulationathome(PetrouandKupek2010;Walkeretal.2011).This,inturn,isassociatedwith
poorlearningoutcomesinchildrenintermsofschoolskills,language(seePaceetal.2017fora
review),andcognitivedevelopment(AlcottandRose2017;CunhaandHeckman2008;Fernaldetal.
2012;HackmanandFarah2009;Kellyetal.2011;Nobleetal.2005;PaxsonandSchady2007).For
example,ithasbeenfoundthatchildren(aged8–11)fromhigh/middleSESperformedbetterthan
low‐SESchildrenonlanguage,memory,andexecutivefunctiontasks—namely,workingmemory,
cognitiveflexibility,andinhibition(Aran‐FilippettiandRichauddeMinzi2012;Arán‐Filippetti2013).
SimilarresultswereobtainedbyastudyinEcuador,where7to11‐year‐oldmiddle‐SESchildren
outperformedlow‐SESchildrenonmemory,language,andexecutivefunctiontasks(Burneo‐Garcés
etal.2019).Sincepoorperformanceofchildrenfromlow‐SESbackgroundshasbeenattributedto
lowparentaleducationandlackofparentalstimulation(Hamadanietal.2014;PetrouandKupek
2010;Walkeretal.2011),therelationshipbetweenSESandcognitivedevelopmentmaybeespecially
relevantforchildrenfromdevelopingcountries(UrsacheandNoble2016).
Inthewesterncontext,itiswell‐establishedthatthereisadisparityinchildren’slearning
outcomesduetodifferencesinSES(CarneiroandHeckman2002,2003;Nobleetal.2005;Jerrimand
Vignoles2013).Thisisalsoevidentinmiddle‐incomedevelopingcountriessuchasIndiaand
Pakistan.Aslametal.(2019)examinedtheinfluenceofSESonlearningoutcomesin1683children
aged8–12yearsstudyingingovernmentschoolsinPakistan.Theyusedanindexofhouseholdassets
asameasureofSES,whichwascomprisedofquestionsrelatingtowhethertheirhousehadatable,
chair,radio,stove,mobilephone,colortelevision,bicycle,motorbike,car,fridge,orwaterfacilityin
thehouse,aswellastheconstructionmaterialusedtobuildtheirhouse.TheresultsindicatethatSES
wasasignificantpredictorforimprovedperformanceinliteracy,butnotnumeracyinanacademic
year.IntheIndiancontext,morethan50%ofchildreninGrade5areunabletoreadaGrade2text
(Pratham2014),andthelearningoutcomesinschoolsarelaggingby2grades(Bhattacharjeaetal.
2011).ThereisadisparityinaccesstoprimaryschooleducationinIndia,withchildrenfrommiddle‐
andhigh‐incomehomesoptingforprivateschools,andchildrenfromlow‐incomefamiliesoptingfor
Languages2020,5,104of21
governmentschools(Agrawal2014;AlcottandRose2015;Pratham2017).Therearedifferences
betweengovernmentandprivateschoolswhichcontributetothedisparityinlearning,butitis
importanttoconsiderotherfactors,suchasSES,whichmayalsonegativelyinfluencelearning
outcomes.ChildrenoflowSESusuallytendtoenrollingovernmentschoolswhichhaveverylittle
funding;thesechildrenalsodonotreceiveenoughparentalsupportathome.Theirexposuretoprint
isalmostabsent—orrestrictedonlytomarketplaces(Tsimplietal.2019).Therefore,thereisanurgent
needtounderstandthefactorsaffectinglearningoutcomesinchildrencomingfromdisadvantaged
backgrounds.
Oneoftheaspectsthatmaypositivelyinfluencethelearningoutcomesoflow‐SESchildrenin
Indiaismultilingualism,asIndiaprovidesastrongmultilingualsocietalcontext.Numerousstudies
haveexaminedthepotentialadvantageofbi‐ormultilingualsovermonolingualswhenitcomesto
cognitiveperformance(seeBialystok2009,foranoverview).Executivefunctioning,inparticular,
seemstobeenhancedinbilinguals(e.g.,Bialystoketal.2004;Bialystok2009;Costaetal.2008;Soveri
etal.2011a),althougheffectsarenotalwaysreplicated(Colzatoetal.2008;Costaetal.2009;Paap
andGreenberg2013).Ithasbeensuggestedthattheseinconsistenciesinresearchoutcomemay
dependonexternalfactors,suchasthepopulationunderinvestigation(Hansenetal.2016).
Therefore,itisunknownifmultilingualchildreninIndiafromlow‐SESbackgroundsalsoexperience
thecognitiveadvantagesassociatedwithmultilingualisminothercontexts.Inaddition,itis
unknownwhetherthelinguisticallydiverseenvironmentinwhichchildreninIndiagrowup
providesanycognitiveadvantagesindependentofthemultilingualismofachild.
1.2.MultilingualismandCognition
ToinvestigatethepotentialcognitiveadvantagesofmultilingualchildreninIndia,thechildren’s
cognitiveskillsinourstudywereassessedwithboththeRaven’sColouredProgressiveMatricestask
andthen‐backtask.Raven’sColouredProgressiveMatrices(Ravenetal.2008)isafrequentlyused
measureofgeneral,orfluidintelligence.Wewerespecificallyinterestedinwhetherlinguistic
diversityandmultilingualismwererelatedtoachild’sfluidintelligence.Thetaskisfrequentlyused
withchildren(e.g.,Belacchietal.2010;Cottonetal.2005;Torregrossaetal.2019;Weichboldand
Herka2003;WeyandtandWillis1994)fromasearlyasage4(Raven2012).Althoughsimilartests
havetraditionallybeenconsideredasculture‐independent(Cattell1940;CattellandCattell1963),a
cross‐culturalmeta‐analysisofRaven’sProgressiveMatrices(threetypes,Advanced,Colored,and
StandardProgressiveMatrices,Raven1938,1956)showedthatperformanceofthetaskcanbe
influencedby“country‐levelindicatorsofeducationalpermeation”,aswellasnumberofyearsof
education(Brouwersetal.2009,p.330).Thus,itseemsthatfluidintelligence—oratleastperformance
ontheRaven’stask—issubjecttocontextualeffects.Nevertheless,therelationbetween
multilingualismandperformanceontheRaven’staskremainsunclear.Jarvisetal.(1995)foundno
relationshipbetweendegreeofbilingualismandRaven’sscoresinadults.ResearchfromBialystok’s
labreportsnodifferencebetweenmonolinguals’andbilinguals’Raven’sscoresinchildren(Bialystok
andShapero2005)oryoungerorolderadults(Bialystoketal.2004).Incontrast,Diaz(1985)founda
positivecorrelationbetweendegreeofbilingualismandRaven’sscoresinlow‐proficiency
kindergarten‐agebilingualchildren(butnotinhigh‐proficiencychildren).Otherstudieshavealso
foundevidenceofbilingualchildrenoutperformingmonolingualsontheRaven’stask(Mohantyand
Das1987;PealandLambert1962).
InadditiontotheRaven’stask,then‐backtaskwasalsoadministered.Then‐backtask(Kirchner
1958;Mackworth1959;MooreandRoss1963)isacommonmeasureofworkingmemorythathas
frequentlybeenusedwithschool‐agedchildren(e.g.,Ciesielskietal.2006;López‐Vicenteetal.2016;
Scudderetal.2014;Vuontelaetal.2003),childrenwithdevelopmentaldisorders(Epsteinetal.2011;
López‐Vicenteetal.2016),andchildrenwithtraumaticbraininjury(Levinetal.2002).Asthetask
involvesupdatingoftask‐relevantinformationthroughexecutivefunctioning(Cohenetal.1997;
Miyakeetal.2000),itmaybeexpectedtobeinfluencedbybilingualism.Littlehasbeenreported
aboutpotentialculturalorcontextualeffectsonthen‐back,butneitherfrequencyoflanguage
switchesineverydaylife,norageofacquisitioninthesecondlanguage(L2)(Soverietal.2011b),
Languages2020,5,105of21
norbilingualismingeneral(Moradzadehetal.2015;YowandLi2015)seemtoinfluencen‐back
performanceinadults,althougheffectsofbilingualismhavebeenfoundinhigh‐conflictversionsof
thetasks(Teubner‐Rhodesetal.2016).Incontrast,positiveeffectsofbilingualismonn‐back
performancehavebeenfoundinchildren,butonlyforyoungerchildrenaroundage7and8inL2‐
immersionschools(Hansenetal.2016).
Generally,itisquitewell‐documentedthatSEShasanegativeinfluenceonlearningoutcomes
acrosstheglobe.Itisalsogenerallyassumedthatbilingualismhasapositiveinfluenceoncertain
aspectsofcognition,specificallyinthenon‐verbaldomain.However,thereisadearthofstudiesin
theIndiancontextthatcapturetheeffectsofsociolinguisticdiversityandmultilingualismon
cognitiveperformanceinchildrenfromlow‐SESbackgrounds.Therefore,thecurrentstudyfocuses
onexaminingtheinfluenceoflinguisticdiversity,multilingualism,andSESonalargesampleof
primaryschoolchildrenfromgovernmentschoolsinIndia.Theparticipatingchildreninourstudy
arefromlow‐SEShomes,livinginslumornon‐slumareas,andarealldisadvantagedanddeprived
withrespecttohousingfacilities,lackofparentaleducation/stimulationathome,waterfacilities,lack
ofadditionalliteracysupport,etc.,andarestudyingingovernmentschools.Althoughallthechildren
arefromunderprivilegedcontexts,thereisstillsomeamountofvariabilityintheirSES,aswellas
variabilityintheirmultilingualismandlinguisticdiversityintheirdailycontext.Tocapturethis,we
developedquestionnaires,eachcontainingaseriesofquestionsaddressingdifferentaspectsof
multilingualism,linguisticdiversity,andSES.Usingtheoutcomesofthesequestionnairesasproxies,
weinvestigatetherelationshipbetweenlinguisticdiversity,multilingualism,andcognitiveskillsin
disadvantagedchildreninIndia.Specifically,weformulatedthefollowingresearchquestions:
1. Howcanwequantifythedistinctionbetweenmultilingualismintheindividualandthe
linguisticallyandculturallydiverseenvironmentofchildreninIndianprimarygovernment
schools?
2. Istherearelationshipbetweenbilingualismand/orsociolinguisticdiversityandcognitiveskills?
3. Towhatextentdosocio‐economicstatus,mediumofinstruction,andotherdemographic
variablescorrelatewithcognitiveskills?
Ourhypothesiswithregardtoquestion2wasthatbilingualismandlinguisticdiversitywill
influencecognitiveperformance.Althoughacertaindegreeofcorrelationbetweenthetwoconstructs
isexpected,weassumethatinamultilingual,highlylinguisticallydiversecountrysuchasIndia,the
constructsmaybedistinguishedandhence,contributeindependentlytochildren’scognitiveprofiles.
RegardingSES,followingseveralstudiesshowingalinkbetweenSESandcognitivedevelopment,
weexpectdifferencesamongparticipantsthatarerelatedtovariationinSES,despitetheoverall
disadvantagedbackgroundofourcohort.Finally,againfollowingstudiesindicatingthatfluid
intelligencecanbeaffectedbycontextualfactors,wehypothesizethatdifferencesinmediumof
instructionacrossschoolsinthetwourbansitesmaybefound.
2.MaterialsandMethods
2.1.Participants
694childrenfromtheIndiancitiesofDelhiandHyderabadparticipatedinthestudyduring
schoolhours.AllchildrenattendedYear4(Standard4)inGovernmentprimaryschools.Theschools
hadeithertheregionallanguage(HindiorTelugu)orEnglishasanofficialmediumofinstruction,
andwereineitherslumornon‐slumareasofthecity.Formalwrittenconsentwasobtainedfromthe
principalorheadteacherofall28participatingschools,aswellasthechildrenparticipatinginthe
study.ThestudywasconductedinaccordancewiththeDeclarationofHelsinki,theESRC’s
FrameworkforResearchEthics(ESRC2010),andtheguidelinesoftheIndianCouncilforMedical
Research(ICMR2006).TheprotocolwasapprovedbytheEthicsCommitteeoftheUniversityof
Cambridge(RG83665),theEthicsCommitteesofJawaharlalNehruUniversity,andtheNational
InstituteofMentalHealthandNeurosciences.
Languages2020,5,106of21
Childrenwereclassifiedasgrowingupmonolinguallyorbilingually,dependingonwhether
theyweregrowingupwithoneormorelanguagesinthehome.Childrenwithanyreadingdifficulty
basedonreportsfromtheteacherswereexcludedfromthestudy.Sevenchildrenwhoperformed
belowchancelevel(scoredbelowzero;seeScoringsection)onthen‐backtaskwereexcludedbecause
theydidnotdisplayanunderstandingofthetask.Ademographicoverviewoftheremaining
childrenispresentedinTable1.
Table1.Overviewofthedemographicinformation.
LocationNumber GenderAge(SD)AgeRangeMediumof
InstructionSchoolSite
Delhi371178f,
193m8.75(0.65)8–1275Hindi,
296English
173slum,
198non‐slum
Hyderabad316176f,
140m9.56(1.20)8–15199Telugu,
117English
170slum,
146non‐slum
Total687354f,
333m9.12(1.03)8–15
274regional
language,
413English
343slum,
344non‐slum
2.2.Methods
Thechildrencompletedanumberofexperimentalcognitivetasksandquestionnaires.Thetasks
wereadministeredbyresearchassistantsfromIndiawhowereproficientinboththeregional
language(Hindi/Telugu)andEnglish.ThecognitivetasksconsistedoftheRaven’sColoured
ProgressiveMatrices(henceforthRaven’stask;Ravenetal.2008)asameasureofgeneralintelligence,
anda2‐backtask(avariantofthen‐backtask;Kirchner1958)asameasureofworkingmemory.The
questionnairesconsistedofquestionsaboutlinguisticdiversity,bilingualism,andSES.
TheRaven’stask(Ravenetal.2008)isanon‐verbaltaskthatconsistsof36multiple‐choice
exercisesinwhichtheparticipantisaskedtoidentifyoneofsixpossiblepiecesthatmatcheswitha
givenpattern.Thetaskisofincreasingdifficulty.Itwaspresentedtochildrenaspicturesonalaptop,
andchildrenwereaskedtoidentifythecorrectpiecebyeitherpointingtoitorreportingitsnumber.
TheRaven’staskiswidelyusedasameasureofgeneral/fluidintelligence.Thefactthatitisnon‐
verbalmakesitanappropriatetaskforoursamplepopulationof(oftenmultilingual)children;a
standardizedversionhasbeendevelopedfortheIndianpopulation(Raven’sEducationalCPM/CVS
(India);Raven2012).
The2‐backtask(Kirchner1958)isataskthatrequiresparticipantstomonitorasequenceof
characters(weuseddigitsratherthanletters),anddeterminewhetherthepresenteddigitmatches
thedigitthatwaspresented2digitsback.Thetaskwaspresentedtochildrenonalaptopusing
PsychoPy(Peirce2007).Ifthecurrentlypresenteddigitwasindeedidenticaltothedigitthatwas
presented2digitsback,thechildshouldhaverespondedwithakeypress(key‘J’).Ifthepresented
digitdidnotmatchthedigitthatwaspresented2digitsback,thechildshouldnothavepresseda
key.Digitswerepresentedonthescreenfor500mseachandwereseparatedbyablankslidethat
wasdisplayedfor2500ms.Thetaskconsistedof60digitsintotal,ofwhich20matchedthedigitthat
waspresented2digitsback(hits),whichshouldhavebeenrespondedtowithakeypress,and40did
notmatchthedigitthatwaspresented2digitsback(falsealarms).The2‐backtaskiswidelyaccepted
asameasureofworkingmemoryrequiredtostoreeachdigittemporarily,theupdatingofworking
memorywhenremovingdigitsthatarenolongerrelevant,andtheinhibitionofirrelevantdigits(e.g.,
MorrisandJones1990;Miyakeetal.2000).
ThechildquestionnaireswereadaptedfrompreviousstudieswithbilingualsbyKaltsaetal.
(2019)andRothouandTsimpli(2017).Regardingbi/multilingualism,childrenwereaskedwhich
languageswerespokenathomebytheirparents,siblings,orotherrelativesandbestfriends—andin
whichlanguagestheyusetorespond.Importantly,thesequestionsrelatetothenumberofdifferent
languagesusedbythechild,andtothechild,inthehome/privateenvironment,andnotintheschool
environment.Inaddition,socialandsociolinguisticdiversitywasinvestigated.Thispartofthe
Languages2020,5,107of21
questionnaireaskedthechildforthenumberofdifferentindividualswithwhomtheyhadregular
interactionswithinthecontextofschool,family,orthecommunity(e.g.,themarketplace).Foreach
oftheseinteractions,thechildrenwereaskedtoreportthegenderandageofthepersonwithwhom
theyinteractedandthelanguage(s)thechildknewthispersonspoke.Althoughsomeoverlapwith
themeasureofmonolingualismorbilingualismathomewasexpected,thesociolinguisticdiversity
questionsprovidedinformationadditionaltothelanguage(s)thechildusedathome.Notethatthe
differentlanguagesthechildreportedinthesociolinguisticdiversityquestionnaireneedednotbe
languagesthatthechildspoke,butlanguagesthattheindividualsinteractingwiththechildhad,of
whichthechildwasaware.
Thesocio‐economicstatusofthechildrenwasinvestigatedthroughachildquestionnaire
speciallydevelopedforthispopulationofIndianchildrenthathadanexclusivesectiononhome
activitiesandparents’work.Itincludedquestionsabouttheirparents’income,numberofroomsin
thechild’shome,numberofgadgetsathome(gadgetsbeingaTVandphone,butalsoafridge,cooler,
fanetc.),whethertheyhaveadrinkingwatertapintheirhomeorwaterhadtobecollectedfrom
outsidethehouse,whethertheydidchoresathome,andwhethertheyspentanytimedoing
homework.Weobtainedinformationaboutparent’soccupationinthesamequestionnaire,butthere
waslowvariability(around4outof5workingmotherswereworkingasmaidsandfatherswere
workingaslaborers).
2.3.Scoring
TheRaven’staskwasscoredbasedontherawscores,i.e.,thenumberofcorrectanswersthata
childgaveforthe36questionsintotal.AlthoughthestandardscoresweredevelopedfromtheIndian
population(Raven2012),theyarebasedonchildrenfromupper‐middleclassfamiliesthatattend
English‐medium(oftenprivate)schools,thusmaynotberepresentativeofoursampleoflow‐SES
children.Therefore,weperformedallanalysesrelatingtotheRaven’staskwiththerawscores,with
lowerscoresreflectingpoorerperformance.
Forthe2‐backtask,acompositescore(A’orA‐prime,firstintroducedbyPollackandNorman
1964)wascalculatedbasedonparticipants’hitratesandfalse‐alarmrates(accordingtotheAscores
inZhangandMueller2005).Thisisacommonwaytomeasurenon‐parametricsensitivityin
detectiontaskssuchasthe2‐backtask.AhigherA’scoreindicatesthataparticipantwasbetterable
toperformthetaskingeneral,withfewermissesandwrongkeypresses;anA’scoreabovezero
meansthattheparticipantwasabletodiscriminatehitsfromfalsealarmsandthusabletoperform
thetaskproperly(abovechance).
Regardingbilingualism,childrenwerecategorizedasbi/multilingualiftheyhadmorethanone
languageintheirhomeenvironment.Noadditionaldistinctionsweremadebasedonthenumberof
additionallanguages.Importantly,thismeasuredoesnotprovideinformationaboutthefluencyofa
childinthesecondlanguageortheamountofexposuretoit.Itdoespresuppose,however,thatthe
childcouldspeakandunderstandthesehomelanguages.
Sociolinguisticdiversitywascalculatedasacompositescorebasedonthechildquestionnaire
section.Specifically,thethreesectionsthatthequestionnaireincludedwithregardtotheage,gender
andlanguageprofileoftheindividualswithwhomthechildinteractedregularlyinschool,within
thefamily,andinthecommunity(e.g.,themarketplace)wereeachwasgradedasfollows:
1pointiftheconversationalpartnerwasofthesamegendervs.2pointsifthepartnerwasofa
differentgender;
1pointiftheconversationalpartnerwasofthesameage,2pointsiftheconversationalpartner
wasofadifferentage,and3pointsiftherewereconversationalpartnersofboththesameanda
differentage;
1pointiftheinteractionswereinthesamelanguageasthechild’shomelanguagevs.2pointsif
theinteractionswereinadifferentlanguage.
Iftherewasnoconversationalpartnermentionedinacertainsection,childrenwouldnotreceive
anypoints.Thechildrencouldthusreceiveaminimumof0pointsandamaximumof7pointsfor
Languages2020,5,108of21
eachsection,makingatotalmaximumscoreof21points,withahigherscorereflectingmore
sociolinguisticdiversity.
Thechildren’ssocio‐economicstatusscorewascalculatedasacompositescorefromthedifferent
questionsonthequestionnaire.Specifically,thechildrenreceived:
1pointforincomewhenonlyoneparentworkedvs.2pointswhenbothparentsworked;
1pointwhentheirhousehad1roomvs.2pointswhentheirhousehadtwoormorerooms;
1pointwhentheyhad1gadgetathome,2pointswhentheyhad2gadgetsathome,and3points
whentheyhad3ormoregadgetsathome;
1pointwhentheyhadtapwaterinthehousevs.0pointswhentheydidnot;
1pointwhentheydidnochoresathomevs.0pointswhentheydid;
1pointwhentheydidhomeworkaftergoingbackhomevs.0pointwhentheydidnot;
Thisaddeduptoatotalscorebetween3and10points,reflectingarangeofsocio‐economic
statuses,withahigherscorereflectingahigherstatus.Recallhowever,thatallchildreninourstudy
areoflowSES.
2.4.Analysis
TheinternalconsistencyofourquestionnairesforsociolinguisticdiversityandSESwas
calculatedasCronbach’sαusingthe‘psych’packageinR(Revelle2019).Theinfluenceofthese
measuresonperformanceonthecognitivetaskswasexaminedwithlinearmodelsinR.Sinceweare
interestedininfluenceofsociolinguisticdiversity,bilingualism,andSESoncognitiveskills,thetwo
cognitivetasks(Raven’sand2‐back)weretakenasdependentvariablesintheanalyses,andall
additionalvariables(bilingualisminthehome,sociolinguisticdiversityscore,SESscore,age,gender,
city,mediumofinstruction,andschoolsite)weretakenasindependentvariables.Allvariableswith
multiplelevels(asopposedtonumericalvariables)werecodedusingeffectcoding.Modelswere
constructedtestingtheinfluenceofspecificindependentvariablesonthecognitiveskills,basedon
theresearchquestions.Themodelscontainingdemographicvariablesweresimplifiedintermsof
removingindependentvariableswhenmodelcomparisonsrevealedthatthesevariablesdidnot
significantlyimprovethemodelfit.Forpost‐hocanalyses,thedatasetwassplitintomonolingualsat
homeandbilingualsathome,aswellasintochildrenfromDelhiandchildrenfromHyderabad.The
relativeeffectsizesofsignificantmaineffectswerecalculatedusingCohen’s(‘effsize’packageinR;
Torchiano2019).Inaddition,therelationshipbetweenthedifferentdependentvariables(i.e.,the
possiblecorrelationbetweenRaven’sand2‐back)andbetweenthedifferentindependentvariables
(e.g.,thepossiblecorrelationbetweenSESscoreandsociolinguisticdiversityscore)intheformof
correlationmatriceswerecomputedwiththesamesoftware.
3.Results
Theresultsarediscussedbyresearchquestionbelow.
3.1.HowCanWeQuantifytheDistinctionbetweenMultilingualismintheIndividualandtheLinguistically
andCulturallyDiverseEnvironmentofChildreninGovernmentPrimarySchoolsinIndia?
ThescoresofthechildrenonthedifferenttasksandquestionnairesarelistedinTable2.The
children’sscoresonbilingualism,onsociolinguisticdiversity,andonSESarediscussedinmoredetail
inthissection.Approximately41%ofthechildreninthestudywerebilingualormultilingual,based
onthepresenceofmorethanonelanguageintheirhomeenvironment.Thedistributionofthe
children’sscoresonsociolinguisticdiversityandSESarepresentedinFigure1.Thesegraphsshow
thatforbothmeasures,thechildrenshowadistributionoverthehigherrangesofthescores,with
mostchildrenhavingasociolinguisticdiversityscorebetween9and19,andmostchildrenhavingan
SESscorebetween6and10.Thereliabilityofthesociolinguisticdiversityquestionnaire,reflectedin
Cronbach’sα,is0.68,whichisconsideredanacceptablelevelofreliabilityandinternalconsistency
betweenthedifferentquestionsthatmakeupthetotalscore(Streiner2003;Ursachietal.2015).The
samecalculationofreliabilityfortheSESquestionsrenderedaCronbach’sαof0.32,indicatinglow
Languages2020,5,109of21
internalconsistency.Theresultsalsoshowthatthiscannotberesolvedbyremovingonespecific
question,andthusindicatethatthequestionsmaybetappingintodifferentaspectsofSES.The
childrenfromthetwodifferentcities(Delhivs.Hyderabad)showsimilarmeanscoresandasimilar
distributionofscoresofsociolinguisticdiversity,butshowdifferentdistributionsofSES,with
HyderabadscoringhigherthanDelhi(seeTable2andAppendixA).
Table2.Overviewoftheaveragescoresonthecognitivetasksandthequestionnaires(SES=socio‐
economicstatus;SD=standarddeviation).
Location Raven’sRaw
Score(SD)
n‐Back
A‐Prime
(SD)
Sociolinguistic
Diversity
max.21(SD)
SES
max.10(Sd)
Delhimonolinguals
(n=228,61%)20.1(5.9)0.69(0.14)14.9(1.1)6.8(1.0)
bilinguals
(n=143,39%)21.1(5.6)0.70(0.15)15.4(1.4)7.1(1.2)
totalDelhi20.5(5.8)0.69(0.14)15.1(1.3)6.9(1.0)
Hyderabadmonolinguals
(n=175,55%)15.7(5.1)0.65(0.17)13.7(3.0)8.2(1.0)
bilinguals
(n=141,45%)17.3(6.2)0.69(0.15)15.8(2.6)8.4(0.9)
totalHyderabad16.4(5.7)0.67(0.17)14.7(3.0)8.3(1.0)
Total
average
monolinguals
(n=403,59%)18.2(6.0)0.67(0.15)14.4(2.2)7.4(1.2)
bilinguals
(n=284,41%)19.2(6.2)0.70(0.15)15.6(2.1)7.8(1.2)
Total18.6(6.1)0.68(0.15)14.9(2.3)7.5(1.2)
Figure1.Distributionofthescoresonsociolinguisticdiversity(leftgraph)andsocio‐economicstatus
(SES,rightgraph),basedonthequestionnaires.
Whenexaminingtheoverlapbetweenthemeasuresobtainedfromourquestionnairescompared
tothemoretraditional,binarymeasuresofbilingualisminthehome(yes/no)andschoolsite
(slum/non‐slum)inFigure2,itcanbeseenthatthesebinarymeasureshavedifficultiescapturingthe
diversitydisplayedbythesampleofchildren.Morespecifically,althoughonemightexpectchildren
frombilingualhomestohavehighersociolinguisticdiversityscoresthanthosefrommonolingual
homes,therewasconsiderableoverlapinsociolinguisticdiversityscoresbetweenthetwogroups,as
shownbytheareashadedinpurple.Likewise,onemightexpectchildrenfromschoolsinslumareas
tohavelowerSESscoresthanchildrenfromschoolsinnon‐slumareas,butagaintherewasavery
largeoverlapbetweenthetwogroups.Thus,themeasuresofsociolinguisticdiversityandSES
seemedtoassessdifferentsocial,cultural,and/orlinguisticaspectsthanthesebinary
categorizations—reflectingthelinguisticallyandculturallydiverseenvironmentofthesechildren.
Languages2020,5,1010of21
Figure2.Distributionofthescoresofsociolinguisticdiversityforbilingualsandmonolinguals(left
graph)andsocio‐economicstatus(SES)forchildrenfromschoolsinslumandnon‐slumareas(right
graph).Purpleareasreflecttheoverlapbetweenthetwogroupsonthesociolinguisticdiversityor
SESscores.
Inaddition,weexaminedwhetherthescoresonthedifferentcognitivetaskscorrelatewitheach
other,andwhetherthescoresonthequestionnairescorrelatewitheachother.Theresultsare
presentedinFigure3.WefoundasignificantnegativecorrelationbetweenRaven’scoreandSES
score(r=−0.17,p<0.001),indicatingthatchildrenwithhigherSEShadlowerRaven’scores,reflecting
moreinteractionsindifferentlanguages.Anadditionalpositivecorrelationwasfoundbetween
children’sscoresonthe2‐backtaskandtheirscoresontheRaven’stask(r=0.27,p<0.001),indicating
thatperformanceonthetwocognitivetaskswasrelated.
Figure3.Correlationmatrixpresentingthecorrelationsbetweenthecognitivescoresandthesocio‐
economicandsociolinguisticdiversityscoresinvestigatedinthispaper(SES=socio‐economicstatus).
Correlationswithablueorpinkbackgroundarestatisticallysignificantata0.01level.
3.2.IsThereaRelationshipbetweenBilingualismand/orSociolinguisticDiversityandCognitiveSkills?
Thesecondresearchquestionfocusesonthepotentialrelationshipsbetweenbilingualism,
linguisticdiversity,andcognitiveskills.Tothisend,separateanalyseswereperformedinvestigating
theinfluenceofbilingualismandsociolinguisticdiversityonthe2‐backtaskandonRaven’stask..
Theresultsshowthatchildrenfrombilingualhomesperformbetteronthe2‐backtask,withamean
scoreof0.67forchildrenfrommonolingualhomes,comparedtoameanscoreof0.70forchildren
frombilingualhomes(ß=0.03;t=2.2;p<0.05,seeFigure4,leftgraph).Thesizeofthiseffectislarge
sociolinguistic diversity score
number of children
0 5 10 15 20
0 50 100 150 200
bilinguals
monolinguals
Languages2020,5,1011of21
accordingtoaCohen’sdof1.97.Nosignificantinfluenceofsociolinguisticdiversitywasfoundfor
the2‐backtask(ß=‐0.00;t=‐1.5;p=0.13).
TheresultsfortheRaven’staskshowapositiveeffectofbilingualismaswell,withanaverage
scoreof18.2forchildrenfrommonolingualhomes,comparedto19.2forchildrenfrombilingual
homes(ß=15.5;t=4.7;p<0.001,seeFigure4,rightgraph).Thesizeofthiseffectwaslarge,basedon
aCohen’sdof3.8.Nomaineffectofsociolinguisticdiversitywasfound(ß=0.05;t=0.5;p=0.61),but
aninteractionwasobservedbetweenbilingualismandsociolinguisticdiversity(ß=−0.96;t=−4.4;p
<0.001).ThisinteractionindicatesthattheeffectsofsociolinguisticdiversityscoresontheRaven’s
taskweredifferentforchildrenfrombilingualcomparedtothosefrommonolingualhomes(see
Figure5).Investigatingthisinteractionfurther,wesplitupthedatasetintoagroupofchildrenfrom
bilingualhomesandagroupofchildrenfrommonolingualhomesandinvestigatedtheeffectsof
sociolinguisticdiversityinthesegroupsofchildrenseparatelywithlinearmodels.Theresultsshow
thatforchildrenfrommonolingualhomes,therewasasignificantpositivecorrelationbetween
sociolinguisticdiversityscoreandRaven’sscore(ß=0.42;t=3.2;p<0.01),whereasforchildrenfrom
bilingualhomesanegativecorrelationwasfound(ß=‐0.53;t=−3.1;p<0.01).
Figure4.Overviewofthepositivecorrelationsbetweenbilingualismandcognitivescores;2‐back
scoreintheleftgraphandRaven’sscoreintherightgraph.
Figure5.Visualrepresentationoftheinteractionbetweensociolinguisticdiversityandbilingualism
ontheRaven’sscore.
Languages2020,5,1012of21
3.3.ToWhatExtentdoSocio‐EconomicStatus,MediumofInstruction,andotherDemographicVariables
CorrelatewithCognitiveSkills?
ThefinalresearchquestioninvestigatesthepossibleinfluenceofSES,gender,age,city,medium
ofinstruction,andschoolsiteonthescoresonthecognitivetasks.Whenexaminingthe2‐backscores,
theonlyfactorthatimprovedthemodelfit—andwasthusretainedasanindependentvariable—was
mediumofinstruction.Theresultsshowthattheaverageperformanceonthe2‐backtaskdiffered
betweenthethreemediumsofinstruction,withchildreninHindi‐mediumschools(meanscore0.71)
outperformingchildreninTelugu‐mediumschools(meanscore0.65;ß=0.05;t=2.6;p<0.01;large
effectsize;Cohen’sd=2.0).Neitherofthesegroupsperformedsignificantlydifferentlyfromchildren
inEnglish‐mediumschools(meanscore0.69).Notethattheseweremedium‐of‐instructioneffects
ratherthancityeffects,ascitydidnotprovetobeasignificantpredictor(withmeanscoresof0.69in
Delhiand0.67inHyderabad).Thisisinlinewiththedistributionofscoresfromthe2‐backtaskfor
childrenfromDelhiandHyderabad,whichareverysimilar(seeFigure6,leftgraph).NeitherSES
scorenorschoolsite(slumornon‐slum)weresignificantpredictorsofperformanceonthe2‐back
task.
FortheRaven’sscore,onlycitycontributedtothebestmodel.Thislargeeffectofcityindicates
thatchildrenfromDelhiperformedbetterontheRaven’staskthanchildrenfromHyderabad(ß=
4.22;t=9.6;p<0.001;Cohen’sd=4.5)
2
.ThiscanbeseeninthemeanscorespresentedinTable2as
wellasinFigure6(rightgraph),whichshowsmorelowscoresforchildrenformHyderabadthanfor
childrenfromDelhi.NotethatthecorrelationbetweenSESandRaven’sscore,aspresentedinFigure
3,didnotreachsignificanceinthecurrentanalysesduetotheoverlapinexplanatorypowerbetween
cityandSES;childrenfromDelhihadlowerSESscoresandperformedbetterontheRaven’stask
thanchildrenfromHyderabad.ToinvestigatethepossibleeffectofSESfurther,wethereforesplitup
thedatasetintochildrenfromDelhiandchildrenfromHyderabadandinvestigatedtheeffectsofSES
inthesegroupsofchildrenseparatelywithlinearmodels.Theresultsshowthatwithinthesecities,
nosignificanteffectsofSESonRaven’sscorewerefound(Delhi:ß=0.44;t=1.5;p=0.12,Hyderabad
ß=−0.29;t=−0.9;p=0.39).
Figure6.Distributionofscoresonthe2‐backtask(leftgraph)andontheRaven’stask(rightgraph)forchildren
fromDelhiandHyderabad.Purpleareasreflecttheoverlapbetweenthechildrenfromthetwocitiesonthe2‐
backandRaven’sscores.
2
Basedonthisfinding,were‐examinedwhetherthefindingsonbilingualismandsociolinguistic
diversityinfluencingRaven’sscoresstillheldwhencitywastakenintoaccountasanindependent
variableinthemodel.Theresultsshowthattheeffectsofbilingualismandtheinteractionbetween
bilingualismandsociolinguisticdiversitywereunaffectedbytheadditionofcityasafactor.
Languages2020,5,1013of21
4.Discussion
Thestudy’sgoalwastoevaluatetheroleofmultilingualismandlinguisticdiversityinthe
cognitiveskillsofprimaryschoolchildrenfrompoorsocioeconomicbackgroundinurbanIndia.To
thisend,weexploredvariabilityindemographicssuchasage,genderandSES,aswellas
sociolinguisticdiversityinthedailycontextsofthechild’slife,i.e.,school,homeandcommunity,as
possiblefactorsaffectingthechild’sperformanceontasksassessingfluidintelligenceandupdating.
Bi‐ormultilingualismathomewasconsideredaseparatefactorthataddresseswhetherachilduses
onlyonelanguageathome,ormorethanone.Incontrast,sociolinguisticdiversityaimedtoassess
thechild’sfamily,school,andsocialencounterswithindividualswhomayhavethesameordifferent
ageorgenderasthechild,andwhomayspeakoneormorelanguagesofwhichthechildisaware,
butdoesnotnecessarilyknoworuse.Thus,althoughanoverlapbetweenthechild’smultilingualism
andsociolinguisticdiversityisexpected,asmultiplehomelanguageswillalsoincreasethediversity
index,thetwoconstructsarenotidentical.
Ourfirstresearchquestionfocusedonthedistinctionbetweenmultilingualismintheindividual
childandsociolinguisticdiversityinthechild’sdailyenvironment.Bothfactorswereevaluatedusing
aquestionnaire.Ourdataconfirmedourhypothesis,namely,thatthemeasureofsociolinguistic
diversityassessesadistinctconstructfromthechild’smultilingualism.Inparticular,sociolinguistic
diversityshowedalargeoverlapbetweenmonolingualsandbilinguals,confirmingthatthesetwo
measuresreflectdifferentaspectsofachild’slinguisticprofile.Furthermore,theoverlapconfirms
ourexpectationthatchildreninIndia,whethermonolingualorbilingual,experienceconsiderable
socialandlinguisticdiversityintheirdailylives.Similarly,thefindingsshowthatourproxymeasure
ofSESreflectssomethingdifferentfromtheslum/non‐slumdistinctionthatisbasedonthelocation
oftheirschools.Specifically,theoverlapinthedistributionofSESscoresacrosschildrenfromslum
andnon‐slumareasindicatesthatwhateverthedistinctionencodes,itisnotcomprehensively
capturedbythemeasureofSESalone.Furthermore,slumsdonotnecessarilysharethesame
characteristicsacrossthetwocities,withconditionsinurbanslumsinHyderabadbeingconsiderably
morechallengingthanthoseinDelhi(GlaeserandSims2015;GovtofNationalCapitalTerritoryof
Delhi2015).Specifically,thenumberofurbanpoorinHyderabadhasincreasedby264%since2001,
suggestingthatonethirdofthepopulationlivesinslumsinconditionsconsiderablyworsethanin
otherurbansites,asprovisionsarenotadequate.ThismakesthecomparisonbetweenDelhiand
Hyderabadslums—andtherespectivepovertylevels—hardtodraw.Althoughwewillattemptthis
infutureresearch,theoverlapinSESscoresinslum/non‐slumschoolsremainsvalid.Theimportance
ofthesefactorsisillustratedbytheirdifferentialeffectsoncognitiveperformance,asdiscussedbelow.
Thesecondresearchquestionwasconcernedwiththerelationshipbetweenbilingualism,
sociolinguisticdiversity,andcognitiveskills.Itwasfoundthatbilingualsoutperformmonolinguals
onboththen‐backtaskandtheRaven’stask(inlinewithDiaz1985;Hansenetal.2016;Mohantyand
Das1987;PealandLambert1962,butcontraBialystokandShapero2005;Jarvisetal.1995;
Moradzadehetal.2015;YowandLi2015),reflectingenhanced(updatingof)workingmemoryand
fluidintelligence,respectively.Inaddition,itwasfoundthatthemonolinguals’scoresontheRaven’s
taskwerepositivelyinfluencedbylinguisticdiversityintheirenvironment.Thus,alinguistically
diverseenvironmentseemstoenhancecognitiveperformanceforchildrenwhoarenotmultilingual
themselves.Nevertheless,nomaineffectofsociolinguisticdiversitywasfound,indicatingthat
bilingualismandsociolinguisticdiversityarenotadditiveeffects;theybothleadtobetter
performanceoncognitivetasks,butabilinguallivinginahighlylinguisticallydiverseenvironment
doesnotoutperformotherbilinguals.Anunexpectedfindingregardingtheeffectsofsociolinguistic
diversityconcernsthenegativeimpacttheyseemtohaveontheRaven’sscoresofbilingualchildren.
Althoughweareunabletoexplainthisfindingatthispoint,somepossiblereasonsmaybe
considered.First,thenumberofbilingualchildrenfromeachcityissimilar,althoughbilingualsare
proportionatelymoreinHyderabad.LookingmorecarefullyatthebilingualprofiledataoftheDelhi
andHyderabadchildren,weobservethatallofthe143bilingualchildreninDelhihaveHindiasone
oftheirhomelanguages.Incontrast,ofthe141bilingualchildreninHyderabad,81,i.e.,57%,donot
haveTeluguasoneoftheirhomelanguages.Thishasimplicationsforthedifficultiesfacedbythe
Languages2020,5,1014of21
HyderabadbilingualswhoattendeitherTelugu‐mediumorEnglish‐mediumschools,asinbothcases
thechildrenarepresentedwithschoollanguagesthatareunfamiliartothemfromhome.Incontrast,
inDelhi,theoverwhelmingdominanceofHindithere—andinmanyNorthIndianstatesfromwhere
manyofthebilingualchildrenhavemigrated—impliesthatlinguisticfamiliaritywiththeschool
languageisinplace.ItisworthnotingatthispointthattheEnglish‐mediumschoolsinDelhiprovide
veryloworalinputinEnglish,andmostoftheclassisdeliveredinHindi(Lightfootetal.,submitted).
Asaresult,HindidominanceisalsoattestedinofficiallyEnglish‐mediumschools,enablingtheDelhi
bilingualgrouptobenefitfromamatchbetweentheschoolandthehomelanguage.Thisdifference
betweentheDelhiandtheHyderabadbilingualgroupsmaycontributetothenegativeimpactof
increasedsociolinguisticdiversityonchildren’sRaven’sscores,asthesociolinguisticdiversityfaced
bythebilingualsinHyderabadmaybeunderstoodasachallengeimposedbythelackoffamiliarity
withlanguagesoutsidehome.Anotherrelatedreasonmaypossiblystemfromthedifferencesamong
thepopulationsoftheurbanpoorinthetwocities.Asmentionedabove,Hyderabadischallenged
byarecentincreaseinthenumberofslum‐dwellersandanoverallhigherpercentageofurbanpoor
comparedtoDelhi.Itissuggestedthatthisincreaseisduetoaninfluxofruralmigrantswhoare
culturallyandlinguisticallydiverse(GlaeserandSims2015).Itisthuspossiblethatthecombination
ofmigrationandpovertyfurthercontributetothenegativeimpactofsociolinguisticdiversityonthe
bilingualgroup’sperformance.
Regardingthepositiveeffectsofbilingualismonfluidintelligence,ourresultsarecompatible
withliteraturesuggestingcontextualeffectsonRaven’sperformance(e.g.,Brouwersetal.2009).In
ourstudy,thecontextualeffectsonRaven’sstemnotonlyfrombilingualism,butalsofrom
sociolinguisticdiversityinthemonolingualgroup.Recallthatourparticipantsareallfrom
disadvantagedbackgrounds,albeitwithsomevariationintheirSESscores.Furthermore,
sociolinguisticdiversityandbilingualismarecorrelated.Thiscorrelationisintuitivelyplausibleon
thegroundsthatmultilingualparticipantsinevitablyexperiencehigherlinguisticdiversity.Variation
isthenexpectedtobefoundinthemonolingualparticipantsonly,andheretheeffectisinthe
directionofimprovedperformanceonRaven’s,withhighersociolinguisticdiversityinthe
monolingualchildren’senvironment.NeithersociolinguisticdiversitynorSESaffectedperformance
onworkingmemory,however.Bilingualismwastheonlysignificantpredictorfor2‐backscores.
Turningtothethirdresearchquestionofourstudy,thefindingsshowseveraleffectsof
demographicvariablesoncognitiveskills.NoeffectofSESonfluidintelligencewasfound,whichis
inconsistentwithpreviousliteratureontherelationshipbetweenSESandcognitiveperformance
(Burneo‐Garcésetal.2019;Paceetal.2017).However,ourdatawereobtainedfromasampleof
disadvantaged,low‐SESchildren,sodifferencesinsocioeconomicstatuswithinthiscohortmaybe
small.ThelowerSESofourparticipantsmaybeexemplifiedindifferentways.Specifically,someof
thechildrenwerefirst‐generationlearners,i.e.,theycomefromhouseholdswithverylowlevelsof
literacy,afactwhichmayalsobededucedfromparents’occupation(maid/laborer).Severalofour
researchassistantsfurtherconfirmedthedisadvantagedstatusofmanychildreninthiscohort.In
someoftheirreports,researchassistantsconfirmedthat“mid‐daymeals[arethe]reasonbehind
students’highattendanceinforenoonclasses;gradually[leadingto]poor[er]attendanceinthe
afternoonclasses.”Inaddition,“[thechildren’s]housesdonothaveproperspaceforsleepingand
evenforanyactivity”,and“workingonacomputerforcognitivetaskswassomethingverynewto
manystudents.”Thisissupportedbyfindingsfromtheliteraturethatprovisionofamid‐daymeal
improvesschoolparticipation,especiallyingirls,inruralIndia(DrèzeandKingdon2001).We
additionallyknowthatschoolattendanceinIndiaispositivelyrelatedtotheeducationlevelof
parentsandnegativelyrelatedtopovertyandhouseholdsize(Jayachandran2002).
Incontrast,aneffectofmediumofinstructiononthe2‐backtaskwasfound,withchildrenfrom
HindimediumschoolsoutperformingchildrenfromTelugumediumschools,andaneffectofcityon
theRaven’stask,withchildrenfromDelhioutperformingchildrenfromHyderabad.AsallHindi
mediumschoolswerelocatedinDelhi,andallTelugumediumschoolsinHyderabad,theseeffects
arelikelytoberelated.Weconsiderithighlyunlikelythatlanguageitselfwasthecauseofthese
effects,hencethefindingswarrantfurtherinspectionofthedifferencesbetweenchildrenfromDelhi
Languages2020,5,1015of21
andHyderabad.Asbilingualismwasestablishedasapredictorofbetterscoresoncognitivetasks,
wefurtherexaminedwhichtypesofschoolsbilingualandmonolingualchildrenattend(seeTable3).
Adifferencewasfoundbetweentheschoolsthatbilingualsattendinthetwocities,with
monolingualsinHyderabadattendingprimarilyTelugu‐mediumschools,andmonolingualsinDelhi
attendingbothHindi‐mediumandEnglish‐mediumschools.Wethereforepostulatethatthe
differencesinmediumofinstructionmayhavebeenduetoTelugu‐mediumschoolsbeingattended
mainlybymonolinguals,whoweregenerallyoutperformedbybilinguals.
Table3.OverviewofthepercentageofmonolingualandbilingualchildrenineitherEnglishor
regionallanguage‐mediumofinstructionschools(childreninEMI(EnglishMediumofInstruction),
HMI(HindiMediumofInstruction),andTMI(TeluguMediumofInstruction)schools).
LocationMonolingualsBilingualsSociolinguisticDiversity
Delhi46.1%EMI–53.9%HMI65.0%EMI–35.0%HMI15.0EMI–15.2HMI
Hyderabad