+ All Categories
Home > Documents > MIlroy 1992

MIlroy 1992

Date post: 03-Jun-2018
Category:
Upload: linas-eriksonas
View: 228 times
Download: 0 times
Share this document with a friend

of 27

Transcript
  • 8/12/2019 MIlroy 1992

    1/27

    Social Network and Social Class: Toward an Integrated Sociolinguistic ModelAuthor(s): Lesley Milroy and James MilroyReviewed work(s):Source: Language in Society, Vol. 21, No. 1 (Mar., 1992), pp. 1-26Published by: Cambridge University PressStable URL: http://www.jstor.org/stable/4168309.

    Accessed: 13/12/2012 03:05

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp

    .JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    .

    Cambridge University Pressis collaborating with JSTOR to digitize, preserve and extend access toLanguage

    in Society.

    http://www.jstor.org

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/action/showPublisher?publisherCode=cuphttp://www.jstor.org/stable/4168309?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4168309?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=cup
  • 8/12/2019 MIlroy 1992

    2/27

    Language in Society 21, I-26. Printed in the United States of America

    Social network and social class:Toward an integratedsociolinguisticmodel'LESLEY MILROY AND JAMES MILROY

    Departmentsof Speech (L.M.) and EnglishLanguage(J.M.)Universityof Newcastleupon TyneQueen VictoriaRoad at St. Thomas'StreetNewcastle upon TyneNEi 7RU, UnitedKingdomABSTRACT

    In sociolinguistics, pproacheshat use the variablesof socioeconomicclass and social networkhaveoften been thoughtto be irreconcilable.In this article,we explorethe connectionbetweenthese variablesandsuggest the outlines of a modelthat can integrate hem in a coherentway. This dependson linkinga consensus-basedmicrolevelof networkwith a conflict-basedmacrolevelof social class.We suggest nterpreta-tions of certain ociolinguisticindings,citingdetailed vidence romre-search n Northern relandandPhiladelphia,whichemphasizehe needfor acknowledgingheimportance f looseknitnetwork iesin facilitat-ing linguistic nnovations.We then propose that the link betweennet-work and classcan be madevia the notion of weak network ies usingthe process-basedmodel of the macrolevel suggested by ThomasH0jrup's heoryof life-modes. Sociolinguistics,ociology, quantitativesocial dialectology,anthropologicalinguistics)

    One of the mostimportant ontributions f Labov'squantitativeparadigmhas beento allowus to examine ystematicallynd accountablyhe relation-ship between anguagevariationand speakervariables uch as sex, ethnic-ity, socialnetwork,and- mostimportantly erhaps social class.Languagevariation n largeandlinguistically eterogeneous ities as wellas in smallercommunitieshas been revealednot as chaoticbut as socially regular,andLabov andothershave shownhowinvestigatinghissociallypatterned ari-ation can illuminatemechanisms f linguisticchange.In this article,we fo-cus on the variablesof socialclassand socialnetwork,both of which haveappearedn someformin a largenumber ociolinguistictudiesof variationandchange.Ourprincipalnterest ies not in thecomplexsociological ssuesassociatedwithclass andnetwork, omeof whichwe discusshere,butin un-derstandinghe role of classandnetwork n patternsof linguisticvariationand mechanismsof linguisticchange.? 1992 Cambridge University Press 0047-4045/92 $5.00 + .00

    1

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    3/27

    LESLEY MILROY AND JAMES MILROY

    Socialclass s fundamentally conceptdesignedo elucidatearge-scaleo-cial, political,andeconomic tructuresndprocesses,whereas ocialnetworkrelates o thecommunityand interpersonalevel of socialorganization.Be-ginningwith Bott in 1958 (revised n 197I), a numberof Britishanthropol-ogists developednetwork-analytic rocedures ecause heyweredissatisfiedwith whatthey sawas anoverreliance n highlyabstract ocial, political,andeconomicframeworksn accounting or formsof behaviorof individuals.Personalsocial networkswere generally een as contextualizedwithinthisbroader ramework,whichwasbracketedoff to allow attention o be con-centratedon developingess abstractmodesof analysiscapableof account-ing for the variablebehaviorof individualsmoreimmediately.However, tis important o remember hat such bracketing ff is wholly methodologi-cal anddoes not reflectan ontologicalreality;no one claims that personalsocial network tructures independent f the broader ocial frameworkhatconstantlyconstrains ndividualbehavior.Whileacknowledginghesecon-straints,a fundamental ostulateof networkanalysts s that individuals re-ate personalcommunitieshat provide hemwitha meaningful rameworkfor solvingthe problemsof theirday-to-dayexistence Mitchell1986:74).Our own workin Belfasthas concentrated hieflyon detailedempiricalanalysisof linguisticand social variationat this interpersonal ndcommu-nity level, andin this articlewe want to proposea means of integrating e-searchat this levelwithresearchhat relatesanguage ariation o socialclass.Likethe BritishsociologistAnthony Giddens,who insisted hat thestudyof day to day life is integral o the reproduction f institutionalpractices(I984:282), we prefer o view the so-calledmicro-andmacrolevels f anal-ysis, to whichnetworkand classrespectivelymaybethought o correspond,as embodyingcomplementaryather han conflictingperspectives.A differentquestion, however, s the adequacyof the conceptualizationof class that is currentn muchcontemporaryociolinguistics,everal chol-ars havingremarked hat the social theory implicitlyadopted by sociolin-guistsis in need of explicitformulationand critique. Sociolinguists aveoften borrowed ocialconcepts n an ad hoc and unreflecting ashion,notusuallyconsideringriticallyhe implicit heoretical rameworkshat are im-portedwholesalealongwith such convenientconstructsas three-,four- ornine-sector scalings of socioeconomic status (Woolard I985:738).WhatWoolards criticizing ere s theprocedurewherebya particular o-cial class modelis importedas an initial ad hoc means of organizingdata,not becauseof its theoretical uitability,but forthepurelypragmatic easonthat it has beenwidelyusedin sociologicalsurveysand so is readilyopera-tionalizable.Thus,althoughmany impressively onsistentpatternsof vari-ation have emergedfrom urbansociolinguisticwork, an adequatesocialframeworkwithinwhichto interpretheirresults s stilllacking.Inattempt-ingnowto developsuch a framework,t seemsbest to startwiththe richso-

    2

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    4/27

    SOCIAL NETWORK AND SOCIAL CLASS

    ciolinguistic vidence hat has beengathered ver the last25 years, only thenlookingfor a socialtheory hat canaccount or it coherently.Working n thisorderwillallowprincipled ecisions o be madeaboutthe kindof frameworkrequired.We do not claim yet to havefound the ideal social class model; in this ar-ticle, we do no more than try to integrateexistingfindingsand suggestthekindof model that seems to be required.A numberof sociolinguistshaveremarkedhat the conception f socialclassunderlyingLabov'swork n NewYork City and Philadelphia s not particularly ppropriate Rickford 1986;Sankoff, Cedergren,Kemp,Thibault,& Vincent 989). His key sociolinguis-tic notion of speech communityemphasizessharednorms of evaluationthroughout he community,wherespeakersaresaid to agreeon the evalua-tion of thesevery inguisticnorms hatsymbolizehedivisionsbetween hem.Thissociolinguisticmodel seemsto reflecta consensusview of societyof thetype associatedwiththe sociologistTalcottParsons,wherebyhecommunityis envisagedas fundamentally ohesiveand self-regulating.Yet, the vitalityandpersistence f nonstandard ernacularommunitiesuncoveredby manyresearchers includingLabov) aremore readily nterpretable s evidenceofconflict and sharpdivisions n society than as evidenceof consensus.Althoughwe certainlyneed to assumesomekindof consensus o accountfor data such as the cross-classagreementon the phonolexical rules forraising and tensingof (a) in Philadelphia Labov I98I), scholars such asRickford I986), workingon Guyanese reole,have concluded hat conflictmodels of social class havebeenundulyneglectedby sociolinguists. ndeed,support for a conflictmodel of societyis providedby Labov's own recentworkin Philadelphia,where he found progressive egregationand linguis-tic differentiation etweenblackand whitenetworks Labov& HarrisI986).Furthermore,a conflict model is essential if we are to account for thephenomenonof linguistic hange,withwhichsomekind of social conflictisgenerallyassociated. Labov himself has acknowledged hat a thorough-goingstructural-functionalpproacho language ould be appliedonlyif lin-guistic systemsdid not undergo nternalchangeand development LabovI986:283).Although acknowledging hat the question here is one of the relativeweightgiven to conflict and consensusperspectives, ather han an absoluteoppositionbetween he two (cf. GiddensI989:705), we suggest hat a socialclass model based on conflict, division, and inequalitybest accountsformanyof the patternsof languagevariationuncoveredby the detailedworkof sociolinguists, enerally n phonological r morphological ariables.TheMarxistnotionof the linguisticmarkethas been usedin urbansociolinguis-tics (see Sankoffet al., I989, for a recentexample), he generalcontentionbeingthat languagerepresents form of socialand culturalcapitalthat isconvertiblento economiccapital.Dittmar, Schlobinski,and Wachs I988)

    3

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    5/27

    LESLEY MILROY AND JAMES MILROY

    provideda particularly sefulexpositionof the linguisticmarketconcept nrelation o theiranalysisof Berlinvernacular.However,WoolardI985) sug-gested hat standard/vernacularppositionemergingrom so muchresearchneeds to be discussed n termsof alternative inguisticmarkets.Thisis con-trary o Bourdieu'sI977, I984) viewof a singledominant inguisticmarketwhere the rule of the legitimate anguage s merelysuspended, ts domina-tion temporarily bsent,when the vernaculars used.Ourownwork as well as thatof otherssupportsWoolard's nalysis.Justas there s strong nstitutionalpressureo use varietiesapproximatingo thestandard n formalsituations, ffectivesanctionsare n force n nonstandarddomainsalso. Forexample, n Belfast,New YorkCity,and(no doubt)else-whereyoungmenareridiculedby theirpeers f theyuse middle-class orms.Woolardsuggested hat much recentsociolinguisticworkthat has concen-tratedon competing ocialvaluesusingcontrastivetatus/solidarityoncepts(or something imilar)offersa particularly romisingbridgebetween ocio-linguisticandsocial theory(see Brown&LevinsonI987, for a discussionofsuch work).A framework hat emphasizescompetingsocial valuesratherthan consensus offers a plausibleinterpretation f the mass of variablelinguistic ndsocialdetail rom nner-cityBelfastreportedn MilroyandMil-roy (1978), L. Milroy(I987a),J. Milroy I98I), andelsewhere.Thephono-logicalstructure f Belfastvernacular an be coherentlydescribed nly if itis analyzedas an internally onsistent but systematically ariable)vernac-ular, rather han an unsuccessfulapproximation o middle-classBelfastorstandardEnglishvarieties for a discussionsee J. Milroy1992, Ch. 3). Weinterpreted lose-knitsocial networksas mechanisms nablingspeakers omaintain uchvernacularodes, which hemselves onstitutean actively on-structed,symbolicoppositionto dominant, egitimized odes.An analysisntheseterms akesus partof theway,but it does notaccountfor widersocialstructures, nd so it needsto be supplemented y an appro-priatesocialclass model.The success,persistence,andprecise orm of thesymbolicoppositionenactedby small-scalenetworkswilldependnot uponcommunity-internalinguisticor interactionalactors,butuponthe relationof the resistinggroupto the nationaleconomyand to likegroups n othercitiesorstates seeGal I988). The levelof integrationf any givengroup ntothe widersocietyis likelyto be inverselyrelatedto the extentto whichitmaintainsa distinctivevernacular.This is whythe outcome n termsof lan-guagesurvivalor shiftin Belfastmaybe different rom thatin Parisor Co-penhagen; n Cataloniadifferent from Gascony.It will be constrainedbyvariations n political,economic, and social structureshat are specifictothese different ocalities.Furthermore,losestudyof networksand the lan-guagepattemsassociatedwith themcangiveus some idea of themechanismsthat give rise to correlationsbetween anguageand class.

    4

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    6/27

    SOCIAL NETWORK AND SOCIAL CLASS

    So far,we havetried o outline ome generalprerequisitesoran integratedandsociallycoherentsociolinguisticheory,constructedo takeaccountofwell-establishedinguistically etailed indingsof urbanvernacularesearch.In the following ections,we summarizehe chief principles nderlying net-work analysisof languagevariation, ookingfirstat close-knitcommunitiesand then at moreloose-knittypes of networkstructureof a kind generallyassociatedwith mobile individuals.Weargue that the structureand socialfunction of whatmight be describedas both strong nd weak networktypesneedsto beconsideredn order o integrate networkmodelwith a so-ciolinguisticallyplausibleand sociallyadequatemodelof class.

    NETWORK STRUCTURE IN CLOSE-KNIT COMMUNITIESA social networkmay be seen as a boundlessweb of ties that reachesoutthrougha wholesociety, linkingpeopleto one another, howeverremotely.But for practicalreasonsthe analyststudiessocialnetworksas anchoredto individuals,and interesthasmost often focusedon relatively trongfirst-ordernetwork ies - that is, thosepersonswith whomego directlyandreg-ularlyinteracts. This principleof anchorageeffectivelylimits the field ofstudy,generally o somethingbetween30 and 5o individuals,although t isassumed hat second-orderies to whomego is linked hroughothersare alsoinfluential(see Milroy i987a).Two types of personalnetworkcharacteristicsregenerallydistinguishedby anthropologists: tructural,whichpertains o the shapeand patternofthe network, and interactional, which pertains to the content of the ties.Both structural ndinteractional haracteristicsreimportantn constrain-ing socialaction. Investigatorsrom severaldisciplineswho havedevelopedformalmethodsof analyzingheproperties f networkshavetended o con-centrateon structural ropertiesuchas density,whereas ocial nvestigatorswho want to accountfor the observablebehaviorof individuals end to giveequalweightto interactional eaturessuch as the multiplexity,history,du-rability,frequency,andintensityof ties (see, e.g., Cochran,Larner,Riley,Gunnarsson, & Henderson I99o; Surra I988). Some importantstructuralandinteractional eaturesare conveniently istedby Mitchell(I986).Ouranalysisof the relationshipbetween anguagevariationandpersonalnetwork tructuren threeBelfast nner-city ommunities ttemptedo dem-onstratethat a close-knit,territoriallybased networkfunctions as a con-servative orce, resistingpressures or changeoriginating rom outsidethenetwork.By close-knitwe meanrelativelydenseand multiplex,these twoconceptsbeingof criticalmportancen a comparative nalysisof socialnet-works.In a maximallydenseandmultiplexnetwork,everyonewouldknow

    5

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    7/27

    LESLEY MILROY AND JAMES MILROY

    everyoneelse (density),and the actorswould know one another n a rangeof capacities multiplexity).Close-knitnetworks,whichvary n the extent owhichthey approximate o an idealizedmaximallydenseand multiplexnet-work, have the capacity o maintainandeven enforce ocal conventionsandnorms- including inguisticnorms- andcan providea means of opposingdominant nstitutional aluesand standardizedinguistic orms.Theircapac-ity to do this, however,seems o be dependent n theirterritorial estrictionto specificneighborhoods, he day-to-daybehaviorof individuals eing essconstrained y geographically ispersed etworks.Networkanalysis hus of-fers a basis for understandinghe community-levelmechanisms hat under-lie processesof languagemaintenance, nd the persistence ver centuriesofstigmatized inguistic orms and low-statusvernacularsn the face of pow-erfulnationalpoliciesof diffusingand imposing tandardanguagess indeedremarkable.Apart from its theoretical alue, a networkapproachhas been found use-ful in providinga suitablemethodologyfor studyingethnicor other sub-groups n the population n situationswherea socialclassmodel(particularlyone that focuses on consensus) s less practical.Quite apartfromany theo-reticalproblems,an initialapproachn termsof class s difficult f subgroupsare distributed nequallywithrespect o class.A networkapproachs morefeasiblewithgroupswho areeconomicallymarginal, rpowerless, r residentin homogeneousand territoriallywell-definedneighborhoods.Moreover,astrong senseof ethnicityor of local identityoften createsand maintains o-calizedcultural ndlinguistic ormsand valuesystems hat arepresented ndperceivedas sharplyopposedto the mainstream aluesof outsiders.Ap-proaching uch communities nitially n networkrather han class termscanallow the researchero get a gripon the relationbetween inguisticvariabil-ity and social structure.Examplesof sociolinguistic pplications f networkanalysisare: Schmidt(I985: Australianaboriginaladolescents),Bortoni-Ricardo (I985: rural immigrants to a Brazilian city), Gal (1979: bilingualpeasantworkers),Lippi-GreenI989: anAlpinerural ommunitynAustria),V. Edwards I986: Britishblack adolescents n the Midlands),and W. Ed-wards i99o: blackDetroitspeakers).Labovand hiscolleaguesn theirPhil-adelphianeighborhoodtudiesalso usedthe network onceptatthe fieldworkstage (Labov & Harris I986). So there is little disagreementon the practicalusefulnessof a network-basedmethodology.It has sometimesbeensuggested hat close-knit ypesof communitynet-worktendto beruraland that theyare nowadaysmarginalo urban ife. Thisis suggestedby, for example, he largesociological iterature n the stran-ger, themarginalndividualwho is often seenas typicalof themoderncitydweller. Harman I988) reviewedand evaluatedmuch of this work. In thesamevein, Wirth,an influentialmemberof the Chicagoschool of urbanso-

    6

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    8/27

    SOCIAL NETWORK AND SOCIAL CLASS

    ciologists,proposed haturbanconditionsgive riseto impersonality nd so-cialdistance Wirth1938). Thisfits in withthe argumentswe present hortlyon the roleof weakties in urbancommunities,but it doesnot tell the wholestoryabout urban ife. Forexample,recentcomparative esearchn EuropeandtheUnited Stateshas suggested hat although he personalnetworksofsociallyandgeographicallymobilepersonsare moregeographicallyispersedandless kin-based hanthe traditional ype of close-knitnetwork,they arealso larger,more supportiveandmore affectivelysatisfying Cochranet al.i99o). Furthermore, s Fischer I982) alsoemphasized,highlyeducatedandmobile individualsaremore able to be selective n their choiceof contactsthan the individualembedded n the localizedsolidarynetwork,whichcanbe oppressiveas well as supportive.Classicexamplesof suchlocalizedsoli-darynetworksare the ItalianAmerican urban illagers escribedby Gans(I962) or the close-knitYorkshireminingcommunitiesdescribedby Dennis,Henriques,and Slaughter1957).Thesetraditional lose-knit oftenindigenous)urbancommunities re lesssalient n AmericanandBritishcities thanthey once were,but theyare ap-parentlybeing replacedby similartypes of communitycreatedby newerimmigrants.More importantlyperhaps, as Giddens (I989) pointed out,neighborhoodsnvolvingclose kinshipandpersonal ies seemto be actuallycreatedby city life, andFischer I984) suggested hat whereassmalltownsdo not permitculturaldiversity, itiesdo. Forexample, hose who form partof urban ethnic communitiesgravitate o form ties with, and often to livewith, others from a similar inguisticor ethnic background.These ethnicgroupsseemto usethe close-knitnetworkas a meansof protecting heir n-terestswhile the communitydevelopsthe resources o integratemorefullyinto urban ife. For example,differences n the networkstructureof mem-bersof the Chinesecommunityn NewcastleuponTynecorrelatebothwithdifferentpatternsof languagechoiceand with different evels of integrationintonon-Chinese omainsof urban ife (Milroy&Li I99I). Bortoni-Ricardo(I985) made a similarpointwithregard o ruralmigrantso Brasilia.There-fore, thetypeof close-knit ommunitymosteasilyconceptualizedn networktermsis likelyto be a productof moderncity life rather han a residueofan earlier ype of social organization.Suchgroupsareimportant n providinga focal point for stigmatizedur-ban vernaculars nd othernonlegitimizedinguisticnorms,and so need tobe accountedor in anysociolinguisticheory.That s whysome form of net-workanalysis hat examines he relationship etween he individualandtheprimarygroupis so important.But the observable ndicatorsof networkstrength,a measureof integrationnto a close-knitgroup,willvaryin kindwithcommunity rganization. orexample,membershipn a religiousgroupmightbe irrelevantn a contemporaryorthernEnglishcoal-mining ommu-

    7

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    9/27

    LESLEY MILROY AND JAMES MILROY

    nity, but highlyrelevant n an EnglishMidlandsblackcommunity EdwardsI986).The Philadelphianeighborhood tudiesarerelevant o this point. Labovand Harris I986:21) suggested hat althoughsocialnetwork s useful as amethodologicalool, it has little explanatory alue n itself and mustbe sup-plementedat the interpretative tage by what theycall the socialhistoryof speakers: the kindsof socialexperience hey have had in dealingwithmembersof othergroups,theway theyhave usedlanguagen their ife. Atmany points in theirarticle hey appeared o represent ocial networkandsocial historyas alternativeand possiblycontradictorymodes of analysis,with socialhistorybeinga morepowerful xplanatoryategoryhannetwork.But there s in factno contradiction ere. The widersocial nteractions out-sideof theingroup)hattheydescribe s socialhistoryarethemselves escrib-able and intepretablen terms of a networkmodel. In that Labov and hiscolleaguesareexamininghe natureof the tie betweengroupandindividual,theyarecarryingout no more and no less than a networkanalysis,employ-ing indicatorsof network tructurehat relate o durationand contentof tie(cf. SurraI988; see also Cross I990, for an illuminating ccountof the re-lationshipbetweennetworkstructure ndethnicity).Still, they rejectsocialnetworkat the interpretative haseof theirwork anduseexplanationshatfall back on a rathervaguenotion of prestige, which s relatedultimatelyto a primarilyonsensual onceptof socialclass.However,muchof thePhil-adelphia data, as reported by Ash and Myhill (I986), is also open to an ex-plicitlynetwork-basednterpretation.Thisinterpretation as the advantageof avoidingpriormodelingof thebehaviorof individualsn termsof socialclass, power,or dominance,and we commentfurtheron this later. In gen-eral, it seemspreferableo carryout an initialanalysisof small-scale om-munitiesn termsof a notionsuchas network,which s designedorthis levelof socialorganization,priorto workingout an appropriatemodel of classthat relatesto the macrolevelof analysis.An important indingof the Philadelphia tudies(Labov& HarrisI986)is that the speechof blackand white ethnicgroupsis diverging n certainways,and it is thisdivergencehat theinvestigatorshoseto emphasize.Theyexpressedhisfinding n termsof participationn linguistic hanges:Certainchanges n progressn thewhitecommunityare not in progressn the blackcommunity.However, t can also be expressedn termsof the model we aredevelopinghere, whichwouldfocus on theconflictbetween he two groups:Thetwo communities o not agreeon normsof usage,andso in thisrespectwecan say that thesociolinguisticituation s one of conflictingnormsratherthanconsensus.It is alsolikelythat the linksbetweenblackandwhitecom-munitiesareon thewholerelativelyweak,in the sensethat cross-ethnic et-workstend not to be dense, multiplex,or territorially ased(Cross 99o). It

    8

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    10/27

    SOCIAL NETWORK AND SOCIAL CLASS

    is this notion of weakties to which we now turn, before examining ts ap-plication to the Philadelphia ommunities.STRONG AND WEAK NETWORK TIESOne importantcorollaryto the link between anguagemaintenanceand aclose-knitnetworkstructure s that outside nnovationand influencewill beassociatedwiththeweakening f sucha structure.This accounts or our find-ing in inner-cityBelfast hat speakerswhoseties to the localizednetworkareweakestare thosewho approximate eastclosely to vernacularnorms. Suchspeakersare mostexposedto external,often standardizing, ressures Mil-roy & MilroyI985). There s, however,a generalmethodologicalproblemassociatedwithnetworkanalysis.Although t canbe readilyoperationalizedto study speakerswhose networksare of a relatively lose-knit ype, it can-not so readilyhandlesociallyand geographicallymobilespeakerswhose per-sonal network ies are not predominantly ense or multiplex.So we cannoteasily demonstratehe effects of weak tiesby the quantitativemethods hatareused to demonstratehe effectsof strong ies, as in inner-cityBelfast, forexample.Fortunately, owever,a largeamountof linguistic vidences avail-able thatenablesus to follow the sociolinguistic mplicationsof the line ofreasoningdevelopedby Granovetter 1973, I982) in his examinationof thesocial function of weak network inks. Granovetterarguedthat althoughstrong ies (of the sortassociatedwithclose-knitnetworks)acilitate ocal co-hesion, they lead to overallsocial fragmentation.This seems to be the kindof interclass ndintercommunityragmentationhat we describedn Belfastandthat Labov pointedto in Philadelphiawithrespect o blackand whitegroups. However,Granovetter rgued hat it is the (often numerous)weakties betweenrelativelyclose-knitgroupsthroughwhich innovationand in-fluence flow and that lead to an overallsocial cohesion capableof balanc-ingthe fragmentationnd conflictassociatedwithstrong ies. It is importanthereto keepin mindGranovetter'snsistenceon theparadox hat weak tiescan be described s strong n thatthey provideinksto a community eyondthe immediate ocialcircle,informationabout educationand employmentopportunities . . . and access to diverse ideas and perspectives (Cochran1990:289). Persons who contractmainlystrong, localized, and often kin-based iesare deniedparallel ccess o theseresources, nd,as we havenoted,strongties of this kind can be norm-enforcing nd oppressive.FollowingGranovetter'sloselyarguedarticle,we haveproposednot onlythatgroups inked nternallymainlyby relativelyweaktiesaresusceptibleoinnovation,but also that innovationsbetweengroupsare generally rans-mittedby meansof weakrather hanstrongnetwork ies (e.g., throughcas-ual acquaintances ather han kin, close friends,or workmates).Weakties

    9

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    11/27

    LESLEY MILROY AND JAMES MILROY

    are, of course,likely to be much morenumerous han strongties. This ar-gumentruns somewhatcounter o the generalassumption hat diffusion oflinguisticchange s encouragedby relativelyopen channelsof communica-tion and discouragedby boundariesor weaknessesn linesof communica-tion. However,as we argued n somedetail(Milroy& MilroyI985), therearemanywell-knownpatternsof change hat are difficult to explain n thisapparently ommon-senseway. Some of these involve arge-scale ndlong-term changes overconsiderabledistances,as discussedby Trudgill 1983,I986). Examplesare the spreadof uvular[r]across nationalboundaries oaffect manynorthernEuropean ities,the spreadof certainLondonfeaturesto Norwich,and theappearancef similardevelopmentsn unrelated r dis-tantlyrelatedanguages e.g., preaspirationf voiceless tops n Icelandic ndScottishGaelic).It is veryhardto see how the relevantpopulations n suchcases couldbe linkedby strongties. Otherexamplesare at a more detailedcommunityevel,suchas the socialconfiguration f the spreadof [a]back-ing fromprotestantEast Belfast nto the Clonard a WestBelfast Catholiccommunity.Thisspread,which we look at in a littledetail,needs to be de-scribedwithina widerhistorical,social, and linguisticcontext.Westudied hevariables a) and (e)very ntensivelybothin the inner-citycommunitiesof Clonard,Hammer,and Ballymacarrettnd in the slightlyhigher statuscommunitiesof Andersonstown nd Braniel.We also estab-lisheda broadsocial class distributionby meansof a doorstepsurveycar-ried out on randomly ampledhouseholds n Belfast(J. MilroyI99I, 1992;L. Milroy 987b:82).The vowel a/ (asinman,grass)showsvariation crossa widephoneticcontinuumbetween ong, backed,roundedrealizations ndshorter ront and front-raised ealizations.Thevowel/e/ (asin went, ques-tion) variesbetween ong, midrealizations ndshort,lowerrealizationsseeJ. MilroyI98I, for an analysisof the phonologicalcomplexities).Realiza-tions of bothvariablesarestronglyaffectedby sex, networkstructure,andsocialclassof thespeaker.Theextensivequantitative nalysis eportedn de-tail elsewhere mostrelevantlyor thisargument: . Milroy1992; L. MilroyI987a;Milroy& MilroyI985) shows thatraised, engthened ariantsof /e/are associated rincipallywithwomenandmiddle-classpeakers, ndbackedvariantsof /a/ with menandworking-class peakers.It is clearfromthehistorical nddialectological ata presented y Patter-son (i86o), Staples I898), Williams1903), andGregg 1972) that /a/ back-ing and /e/ raisingare both relatively ecentphenomenan Belfast but arecharacteristic f modernScots andoriginatentheUlster-Scots-speakingi-alect area of Down and Antrim(as distinctfromthe Mid-andWest-Ulsternon-Scotshinterland).As FigureI shows, East Belfast adjoins the Ulster-Scots regionof NorthDown, whereasWestBelfast pointssouthwestdownthe LaganValley,the speechof which is Mid-Ulster,with less Scots influ-ence.Furthermore,mmigrationo WestBelfast s recentand is largely rom

    10

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    12/27

    SOCIAL NETWORK AND SOCIAL CLASS

    RathlinIsland

    mBallymena -\_Larne

    Lurgan * LBELFAST

    FIGURE I: Mapshowing he UlsterScotsarea(shaded)n relation o EastandWest Belfast.

    a Mid-andWest-Ulster interland.Present-dayociolinguisticvidence ug-geststhat the incomingvariantsof (e)and(a)arediffusing romeastto westof the city;scoresfor /a/ backingarehigher or EastBelfastworking-classmenthan for any othergroupstudied,whereasEast Belfastworking-classwomenusethelow, conservativeariants f (e)lessthananyother nner-citygroup.Thehigher tatusAndersonstownndBraniel peakers xhibita sim-ilarpatternof sex differentiationbut, as one mightpredictfromthe infor-mationpresented o far, use the incomingvariantsof (e) morefrequently,and the incomingvariantsof (a) less frequently, han inner-city peakers.

    11

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    13/27

    LESLEY MILROY AND JAMES MILROY

    TABLE . Contrasting atternsof distribution f two vowelsinvolved n change,according o social classand sex ofspeaker,relative requencyof innovatoryvariants,and level of correlationwithnetwork trengthHigh correlationChange led by with network strength

    (a) Males (working-class variant) Females(e) Females (middle-class variant) Males

    In summaryand this generaldistributions confirmedby a doorstep ur-vey), raisedvariantsof (e) are in the inner-cityassociatedparticularlywithwomen (and,we might add,with carefulspeechstyles). Theyare also asso-ciatedgenerallywithslightlymoreprestigiousouter-city peech.Incomingvariantsof (a) show an almostperfectlyconversepatternof social distribu-tion, beingassociatedwithmale,vernacularnner-city peech.Taking hisevidence ogetherwith the historicalandgeographical ataoutlinedearlier,we notethatalthough ncomingvariants f bothvowelsappear o have orig-inated n the samehinterland cotsdialect,eachhas assumeda diametricallyopposedsocial valuein its relativelynew urbansetting.Therelationship etween peaker hoiceof variantandindividual etworkstructureaddsa furthercomplexity o this pattern,andit is the overall re-lationshipamongsocial class, sex, andnetwork tructures f speaker hat isof particular elevancehere. Extensive tatistical nalysisof the relationshipbetweenlanguagevariation and social network has shown that whereaschoiceof variantcorrelateswithnetworkstructureamong some inner-citysubgroups, hesesociolinguisticpatternsare quite different or eachvowel.Although(a) is generally ensitive o network tructure, hoice of variant smore closely correlatedwith network tructure or women han formen;thisis despitethe fact that women(like middle-class peakersgenerally)use in-comingbackedvariantsmuch essfrequently thanmen.The converse s trueof (e);whereasmen useincoming aisedvariantsmuch essthanwomen,thecorrelationbetweenchoice of variantand networkstructure s higher formen. We arguedon the basis of these data that (e) functionsparticularlyclearlyfor men and (a) for women as a networkmarkerand noted that ineachcase it is the group for whom the vowel has less significance as a net-work marker that seems to be leading the linguistic change. Thecomplexre-lationshipamongclass and sexof speaker,network tructure, nd languageuse is summarizedn Table i, and the data upon which this discussion sbased arereported n Milroyand Milroy(I985), L. Milroy(I987a),and J.Milroy (1992).

    12

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    14/27

    SOCIAL NETWORK AND SOCIAL CLASS

    We are now in a position to relate hese patterns o the generalargumentoutlined earlier,namelythat a close-knitnetworkfunctions as a conserva-tive force, resistingpressures or change originating rom outside the net-work. Those whose ties are weakestapproximateeast closely to vernacularnorms and are most exposedto externalpressures or change. The analysispresented eresuggests hat the vernacular peakers ssociatedmost stronglywith the innovation re n eachcase those forwhomthe vowel functions eastprominentlyas a networkmarker.It is as if a strong relationshipbetweenthe network tructure f a given groupandchoice of phoneticrealization fa particular oweldisqualifieshatgroupfrom fulfilling he role of innova-tors with respect o that vowel. Conversely,he weakeningof the language/networkrelationshipwith respect o a groupof speakersmay be a necessarypreconditionof that group fulfilling the role of linguistic nnovators.Theseobservations rovide urther vidence o support he contention hata weakeningof network inks is implicatedn social processesof linguisticchange. Furthermore, omeinnovationsseemto have crossedthe sectarianboundary n working-classBelfastto producean intercommunity onsensuson normsamong he generation f speakerswho weremostrigidly egregatedfrom each other. The problemof explaininghow a linguistic hangesuch as(a) backingcouldpossiblydiffuseundersuchconditionsdissolves f we ac-ceptGranovetter's rinciple hat it is the multipleweak ties of casualinter-action (examplefor these speakersmightbe ties contracted n shops andsocial securityoffices) throughwhich innovation is routinelytransmittedrather han strong neighborhood ies of close association.The Philadelphia ata presented y Labovand Harris I986) and Ash andMyhill(I986) appearalso to be amenable o an interpretationn terms ofweak ties, even though he researchers ave preferredo account or the pat-terns revealedby theiranalysis n terms of prestigeand dominance.Labovand Harris I986:20-2I) mentioned he prestigeof the localized innovatorand spokeof the dominantdialectas opposedto the dominated.Thiscon-trastof dominancewasusedby Ash andMyhill n interpretingheirfindingswith respectto four groupsof speakers:a core white group, a core blackgroup, andtwo marginalgroups- a groupof blackswho have considerablecontact with-whiteshenceforthWBs)anda groupof whiteswho have con-siderablecontactwith blacks(BWs).It is thesemarginalgroupsthat inter-est us here.Ash andMyhill nterpretedhe pattern n Figure2 as evidence hat WBsconvergetoward whitemorphosyntactic ormsmoremarkedly han BWsconverge owardblacknorms,and this is explained s the resultof the dom-inanceof thewhitedialect.However,we do not need to invoke a macrono-tion such as dominance n interpretingdata at this community evel. Twopatternsare particularlynoticeablein the languageof these two contactgroups(white-orientedlacksand black-orientedwhites).First,on morpho-

    13

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    15/27

    LESLEY MILROY AND JAMES MILROY

    100

    ~80 -

    2 40

    20

    MA4 aaen 003

    3rdsg copula poss ain't/-s -s didn't?Blacks who have ittlecontat with whites ndex of contact8)* > > > | | | Whites whohavecaWiderableontactwithblacs............ .........Whitehohave litle contactwith blaks

    FIGURE 2: Average percentageuse of black English vernacular-markedmor-phosyntactic variants by four groups of Philadelphia speakers(adapted from Ash & Myhill I986:39).

    syntactic variation their scores average about the same, and on copula de-letion and ain't for didn't, the whites actually outperform the blacks onblack variants. Second, whereas the core black group uses these featuresquite variably (presumably also using the white variants), the core whitegroup does not use the black variants at all. So in their convergence pattern,the BWs and the WBs have different starting points. In ethnic group terms,the white group starts much further back on the black English dimensionthan the black group on the white dimension. And it follows from Figure 2that the core black vernacularincorporatesa resource not availableto main-stream white speakers - the capacity to alternate between black and whitemorphosyntactic variantsaccording to occasion of use. To this extent, thesespeakers resemble the inner-cityBelfast speakers, who also have at their dis-posal alternatingforms that carry different symbolic functions according to14

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    16/27

    SOCIAL NETWORK AND SOCIAL CLASS

    occasionof use.Typically, ne of thesealternants as in-group unctionsandbelongsto the vernacular,whereas he other hasout-group unctionsand ismore standardlike.We have discussed hemin variousplaces, mostrecentlyin J. Milroy i99i, 1992).Fromthis perspective,he convergence f WBstowardwhitenorms s notso remarkable as these white norms are alreadyavailable o themwithinan existingpatternof coreblackvernacular ariation.The reasonwhy thewhite-oriented lacks use the white norms moreoften than otherblacks isaccessible hrougha theoryof weakties, as it is clear fromthe authors'de-scriptionsof thesespeakers hat their contactswithwhites are of a classicweak-tie ype. They are describedas con men, hustlers,and politicalactiv-ists, andit is hardlyplausible hatcon men (for example)could successfullypracticeon personswith whomthey had contracted trong(denseand multi-plex)ties.The degree o which hesespeakers se thewhitenorms s increasedby therangeand numberof situations n whichthey have weak-tiecontactsoutsidetheircorecommunity,andfor themthe adoptionof more white us-ageis functional n their weak-tiecontacts.Thesuggestionby Ash and My-hill (I986:41) that prestige is the explanationfor this shift toward whitenormsseems o be quitea weak explanation,whichsimplybegsthe questionof what is meantby prestige n sucha context(for a discussionof prestige-based arguments, see J. Milroy I987).The convergence f black-orientedwhite speakers o blacknorms is in asense moreremarkable, s the corewhite dialect does not possessthe newvariants copuladeletion, etc.) that they adopt (to a certainextent)in car-ryingout what must presumablybe an act of linguisticaccommodation.These outsidevariantshaveto be acquired,and so someaffirmatoryeffortis involved.Althoughthe researchers o not giveprecise nformationas tothe strengthof thesespeakers'participationn blackculture,the model oflinguisticdiffusion and changewe have outlined n this section wouldpre-dict that theirties withbothcommunities relikely o be relativelyweak. Onthe basisof the informationprovidedby Ash andMyhill,we assume hat itis thisgroup,and not theWBgroup,who most resemble heperipheralhar-acterswhoRogersand Shoemaker197I) arguedaretypicalof the innovat-ingindividual.Cross's i99o) comparison f the friendship etworks f blackand white familiesoffers further nterestingperspectives n interethnic it-uations such as this.It seems therefore hatdespitethe fact that Labovand his colleaguesex-plicitly rejectedsocial networkas an interpretative oncept, some of theirfindingsare opento interpretationsasedon a networkanalysis hat distin-guishesbetween he social functionsof strongand weakties. Furthermore,it also seems that this typeof interpretation egs fewerquestions han onethat appealsto assumptionsabout the widersocialstructure,as implied nconcepts such as prestigeand dominance,and that we can best lay the

    15

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    17/27

    LESLEY MILROY AND JAMES MILROY

    groundworkor an integratedheoryby following hrough he implicationsof the weak-tiemodel that we have outlined n this section.THEORETICAL IMPLICATIONS OF A WEAK-TIES MODELSpeakerswhose ties to a localizednetworkareweakest,who approximateleastclosely to the norms of theirlocal community,and who are most ex-posedto externalpressures re frequently oundin the middle-class r up-per-working-classreasof cities.However,n the previous ectionwe alludedto thepracticaldifficulties n carrying ut empiricalnvestigations f loose-knit network tructures,whichcharacterizeesidents f Andersonstown ndBraniel.Othershave encountered he sameproblems,for example,in theprosperousBerlinsuburbof Zehlendorf LabrieI988; see also L. MilroyI987b: 98). But as manypeople(particularly ity dwellers,as Wirth sug-gested)contractweakties, we need to take suchties into account n our de-scriptionof sociolinguistic tructure.And despitethe empiricaldifficultiesin handlingweakties, an extensionof networkanalysis hatfocuseson theirpropertiesprovidesa crucial inkwith moreabstract ocialtheoriesof class.It is clear hatclass-specific etwork tructures renot arbitrarilyonstitutedbut emerge romlarge-scale ocialandeconomicprocesses hat themselvesgiverise to (forexample) he social andresidentialmobilityassociatedwithloose-knitnetworks.Therelationshipetween he variables f classand networkhavebeencon-sidered n some depth by Fischer I982) in SanFranciscoand by Cochranet al. (I990) in Germany,Sweden,Wales,andthe United States.Investiga-tors havegenerally mphasizedhe effects of educationandaffluence n af-fordingaccessto a sociallyandgeographicallywiderrangeof contactsandin enhancing heability o maintain hosecontacts.Generallypeaking,mid-dle-classnetworks consistingargelyof weakties in Granovetter'sense)arelarger, less kin- and territory-orientednd perceivedas moresupportive.Mewett I982) examined herelationship etweenclass and network rom adifferentperspective, rguinghatclass differencesn smallcommunities e-ginto emergeover time as theproportion f multiplexrelationships eclines.Observationsuch as thissuggesta route for constructing two-levelsocio-linguisticheory, inking mall-scaletructuresuchas networks,n which n-dividualsareembeddedand act purposivelyn theirdaily lives, withlargerscaleandmoreabstract ocialstructuresclasses)hat determineelationshipsof powerat the institutional evel.Fromthe point of view of the sociolinguist, t is smallerscale close-knitnetworks hatrenewand maintainocalsystemsof normsandvalueswithinwhichdiscourseprocessesof the kindanalyzedby GumperzI982) are un-derstoodandenacted.And it is network tructureshatlinktheinteractionallevelwith the political andeconomic,wherediverse ocal responsesof lin-

    16

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    18/27

    SOCIAL NETWORK AND SOCIAL CLASS

    guistic groups are constructed to material and cultural domination (GalI988). We need such a dual level of analysis if we are to understand the fre-quently negative self-evaluations of speakers of urban vernaculars, who nev-ertheless continue to use them in their daily lives. In this section, we use theweak-tie concept to link systematically a network and a class-based analysis.We have argued that weak ties between groups regularly provide bridgesthrough which information and influence flow and are more likely thanstrong ties, which are by definition concentrated within groups, to fulfill thisfunction. Thus, whereas strong ties give rise to a local cohesion of the kinddescribed in inner-city Belfast, they lead, as we have noted, to overall frag-mentation. Indeed, it is this potential for explaining both patterns - local sta-bility and cohesion versus overall fragmentation and conflict - that allowsus to relate a network analysis to a model of social structure at the macro-level. This is an important point, as some of the comments made in recentyears about network models in sociolinguistics by, for example, Labov (I986)and Guy (I988) assume that their application is limited to strong ties inclose-knit communities; and indeed they have been used chiefly in such com-munities (but see Bortoni-Ricardo I985). Guy's remarkthat network is a mic-rosociological concept, whereas class is macroscopic, seems reasonable if welimit network analysis to close-knit networks. But an analysis that takes intoaccount the function of weak ties allows us not only to link the two levelsin a principled way, but to develop a cleareridea of which type of social classmodel is appropriate. The analysis so far suggests an urban community thatconsists of clusters of individuals connected internally by differing propor-tions of weak and strong ties, which in turn are connected to other clustersby predominantly weak ties. Middle-class groups will tend to be internallyconnected with a higher proportion of weak ties than working-class groups.This conclusion is entirely consistent with Labov's finding that innovat-ing groups are located centrally in the class structure, characterized by himas upper-workingr lower-middle lass(Kroch1978; Labov I980:254). For,in British and American society at least, close-knit, territorially based, kin-oriented networks are located most clearly in the lowest classes, but upper-class networks are in some respects structurallysimilar, being relatively dense,small, close-knit and kin-oriented. Consider Mills's (1956) description of theAmerican power elite, and the close ties among British upper-class speakersacquired at a limited number of private schools and universities and subse-quently maintained for life. The majority of social and geographically mo-bile speakers fall between these two points. Thus, if we extend a networkanalysis to include an examination of loose-knit network types, which aresusceptible to outside (frequently standardizing) influences, it is evident thatnetwork-basedand class-basedanalyses are not contradictoryas is sometimessuggested; rather, they complement each other. Moreover, a network anal-ysis can give us an idea of the interpersonalmechanismsgiving rise to the ob-

    17

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    19/27

    LESLEY MILROY AND JAMES MILROY

    servable anguage/classcorrelations hat are such a prominent eatureofresearch n the quantitativeparadigm.AN INTEGRATED MODEL?At this point, we have a pictureof variousethnic andclassgroupsas bothinternally tructured nd connected o each otherwithvaryingproportionsandnumbers f strongand weak ies.Forexample, thnicsub-groupsn Brit-ainsuchas the blackspeakers tudiedby V. EdwardsI986) have a predom-inantlystrong-tie nternalstructurebut seem to be linkedby relatively ewweakties to whiteworking-class roups. These whitegroups n turnmighthave a similar nternalnetworkstructurebuthavemore weak tie links withotherwhiteworking-classroups.Verticalinksto middle-classroupsmightbe fewer thisseemed o be the case n Belfast)andmoreovero be frequentlyinstitutional o suchpersonsas doctors,lawyers, eachers,welfareperson-nel, and thelike.Middle-class roups or theirpart- professional,neighbor-hood, and friendshipgroups- are characterized y a higherproportionofweakties internally hanworking-classroups;hence heproblemsof study-ingthemsystematicallyn network erms n Zehlendorf ndin outer-cityBel-fast. But howeverwe interpret he conceptof classand howeverwe modelthese localizednetworks,Granovetter'sonceptof the weak tie can be usedto linkclose-knitcommunityevelgroupings o moreabstract nstitutionalstructures.Such an analysisattributeshe behaviorof speakerso theconstrainingf-fectsof thenetworkorto the diminution f thoseeffectsthatenables hele-gitimizedanguage o permeatenetworks, ather han by anydirecteffect ofprestigeas definedby the perceivedattributesof speakerswho are seentobelong o differentstatusgroups.Socialclass is not conceivedof hereasa gradedseriesof pigeonholeswithinwhich ndividualsmaybe placed.Fol-lowingthe analysisof the DanishMarxistanthropologistThomasH0jrup(I983), a view of socialclassmoreconsistentwithnetworkanalysis onceivesof it as a large-scaleandultimatelyeconomicallydrivenprocess that splitspopulationsnto subgroups.Thegroupssharing ertain ocial andeconomiccharacteristics ndlifestyles hat emerge romthis splitmay looselybe de-scribedas classes,but as we shall see H0jrup offered a moreexplicitlymo-tivateddescriptionn termsof life-mode.Theattraction f thisanalysis romour point of viewis thatdifferent ypesof networkstructure merge romthe conditionsassociatedwith the life-modesof thesesubgroups,andlocaland individualocialbehaviors seen as mediatedhrough hesesmaller calestructures ather handirectlyrelated o class.Whereasnetworkanalysisof the kindwe haveoutlinedso far can delin-eatevariouseconomic,political,andsubculturalroupingsn society, t can-not say anythingabout the varyingpotentialsof suchnetworks o exercise

    18

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    20/27

    SOCIAL NETWORK AND SOCIAL CLASS

    the economicand politicalpowerthat is the sourceof conflict and inequal-ity in society. In linguisticerms, his means hat powerfulnetworkshave thecapacity to impose theirlinguisticand culturalnormson others, whereaspowerlessones do not but can merelyuse the resourcesof the networktomaintainand at best renewtheir own linguisticand culturalnorms. There-fore, to supplement network analysis we need a social theory such asH0jrup's, which can explicitly ink a networkanalysisof subgroupswithinsociety o ananalysisof socialstructure t the political,nstitutional, nd eco-nomic levels.H0jrup'sanalysis s basedon ethnographicwork in Denmarkand exten-sive analysisof social and economic structure n other westernEuropeancountries.Although t begsas manyquestionsas it offerssolutions see, e.g.,Pedersen 99I, who questionedts applicability o women), t is particularlysuggestiven helping o constructa model of sociolinguistictructure hatin-tegrates he variablesof social class and social network.Withspecific ref-erence to western Europe, he proposed a division of populations intosubgroups hat aredescribedn termsof three ife-modes.These life-modesare seen as necessaryandinevitableconstituentsof the social structureas awhole. His conceptionof this largersocial structure s Marxist,and the ini-tialanalysiss in termsof modesof production ndconsumption.Thus,cru-cially, these subgroupsare not seen as sociallyor culturallyarbitrary ut asthe effectof fundamentalocietalstructureswhichsplitthepopulation ntofundamentally ifferent ife-modes H0jrupI983:47). Class is thus seen asa dynamicprocess hatgivesriseto these ife-modes.H0jrup'sanalysiss par-ticularlyhelpfulin suggestinga further ntegrationof the conceptsof net-work and class, becausethe different types of networkstructure hat wedistinguishedn the previoussection can be seento a considerable xtent asspringing rom differences n the life-modesof differentindividuals.Al-thoughthe argumentationupportinghis analysis s lengthyand complex,H0jrupuseda limitednumberof straightforwardoncepts o distinguishhethree life-modes. Life-mode i is the life-mode of the self-employed,Life-modes2 and3 of two different ypesof wageearner.Of critical mportanceis the ideologicalorientationof the threegroupsto work, leisure,and fam-ily. Wefocus a briefdescriptionof eachof them on evidentpointsof con-tact with our networkanalysis.Life-mode iThis is the life-mode of the self-employed,of which a close-knitfamily-centerednetworkwithlittle distinctionbetweenworkand leisure activitiesand a strong solidarity deologyis particularly haracteristic.The family-ownedbusinessmightbe in agriculture r fishing,a cornershop, or a res-taurant.In this life-mode,socialrelationshipsn the formof familyties orcooperativerelationsamong colleaguesbind the producers nto a cohesive

    19

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    21/27

    LESLEY MILROY AND JAMES MILROY

    productionunit. Theprimary oncern s to keeptheproduction olling,andall the family and other affiliatedproducersare involved n this. The pur-pose of the enterprises to be able to remain elf-employed,a means hatisits ownend. The conceptof free ime haslittlemeaning n thislife-mode,because he producers not putto workbutputshimselfor herself o workto gainindependence. hus, theconceptsof leisure nd work have a to-tallydifferentmeaning rom thatwhichtheyassumefor wageearners,andit is clear whya close-knit ype of networkstructureand a solidarityethicwillbe associatedwiththis life-mode,which tself followsfromthe typeofeconomicactivity n which the producersengage.H0jrupdid not see thiskind of life-modeas a relicof anearlierperiod cf. thepervasivenessf close-knitnetworks n moderncitiesnotedearlier)butas highlyefficientand com-petitive, given its flexibility of operation and the commitment of theproducers.Life-mode2Wageearnersare differentfrom Life-modeI commodityproducersn thattheyareincorporatedn a longandcomplexprocessof productionhattheydo not ownor control. Life-mode is thatof the ordinarywageearner, hepurposeof whoseworkaccording o H0jrup,is to providean income thatenablesa meaningful ife during he worker's ree time. The familydiffersfromLife-mode families n beingseparate romthewageearner'swork ac-tivities and is the frameworkwithinwhichnonproductive eisureactivitytakesplace. The Life-mode2 worker acksthe commitment o workchar-acteristicof Life-mode , beingpreparedo sell his or her labortherebybe-comingmobileandsevering xistingclose-knitnetwork ieswhere here s anadequate nducement o do so. If wagesare low however,the wageearnerhas to demandenoughto survive.Hence, the solidarity hat arises amongworkerswhoearn ittle- a solidarity eflectedat the institutionalevelin theestablishmentf tradeunions.At a neighborhoodevel,thissolidaritys em-bodied in the close-knitnetworksof the traditionalworking-class ocietyof the kindinvestigatedn Belfast.FollowingH0jrup'sanalysis,we surmisethat the solidarityethicwouldcollapseand network ies become weaker feconomic and political conditions allowedworkers o feel securein theirfutureprospects, f they earnedenoughto becomemobile, to buy betterhouses and cars, to take holidaysabroad,and so on. Theredo in fact ap-pearto be differencesof this kind in behaviorbetweendifferentgroupsofwageearner,as we notedin our analysisof the outer-cityversus nner-cityareas n Belfast.Moreover,as nationstatesvary withrespect o the wagesandconditions ffered o workers, lose-knit etworkswillbe associatedwithLife-mode2 workers n some countriesmorethan in others. Cochranet al.(I990) discussed n considerabledetail in severalchaptersthe effect of a

    20

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    22/27

    SOCIAL NETWORK AND SOCIAL CLASS

    whole rangeof political,economic,social,and cultural actorson socialnet-work structure.Lockwood's I989) classic nvestigation f class consciousness nd imagesof classstructuren Britain its in broadlywith such an analysis,particularlyin its distinctionbetween he outlooks described espectively s proletariantraditionalism ndprivatizedworker seealso Giddens 989:224). Privatizedworkers,exemplifiedby the Luton car workers tudiedby Goldthorpe t al.(I968-69), live apart from traditional working-classareas in the suburbs andsee workas a way of achieving satisfactoryifestyle orthemselves ndtheirfamilies.They apparentlyeject he traditionalworking-classolidarity thic,but Lockwoodstressed hat a certain evel of grievance endsto recreatehisethic as does the us/them, insider/outsider magerycharacteristic f close-knitcommunities nd of thetraditional roletariandeology.The persistenceandrenewalof this imagery and its associatednetworkstructures) eemtospringfairly directlyfrom changesin economic and power structures nsociety.Life-mode3Whereas he Life-mode wageearnerperformshe routine asksof the workforce at a givendailyor hourlyrate,the Life-mode3 wage earner s a higherprofessionalor managerial mployeewitha high levelof skill. This skill isitself a saleablecommodity, and the wageearner s paidto arrange,moni-tor, and controlthe productionprocess.Typically, heconceptof work andleisure and the role of the family are in sharpcontrast to those of Life-mode 2. This is because he Life-mode3 goal is to riseup through he hier-archy,obtainingcontrolthroughmanagerial ndprofessional olesso as toexerciseprogressivelymorepowerandultimatelyo escapefrom the controlof othersso as to controlresources ndexercisepoweron one's own account.Thisprocessdemandsan immersionof the individual n work, a competi-tive attitude o colleagues,and a blurringof the boundariesbetweenworkand leisure.The familyand its wayof life fulfillsa supportiverole in rela-tion to the career.Worktherefore s life to a high degree,and the conceptof freedom s not one of free time but is associatedwith the work situationand the careerperspective.Just as different ypesof network tructuremerge romthe economic on-ditionsassociatedwithLife-modes and2, so a certain ypeof personalnet-work structures likelyto followfrom Life-mode3. Thesewageearnerswillbe sociallyandgeographicallymobileas they pursue heircareers,formingmanyloose ties, particularly f a professionalkind,throughwhich innova-tions and influencemaybe transmitted.However, theywill also formrela-tively close-knit clusters and coalitions within their personal networksthroughwhichthey maycontrolconsiderable esources.This seemsto fit in

    21

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    23/27

    LESLEY MILROY AND JAMES MILROY

    Maintenance f non-legitimized Dominancef legitimizedlinguistic ode linguisticode

    Mainly trong Community-basedies Mainly eak

    A\A/ INon-community-Ix basedstrongiesUfe-modes 1 2 3 (coaltions,powerelites etc.)

    Macro evelsocial, political ndI economic structure lFIGURE 3: Macro-and microlevelsof sociolinguistic tructure.

    withour general haracterizationf the differing oleof loose-knitand close-knitnetwork ies. Theprimarilyoose-knitnetworkof the Life-mode3 in-dividualensures hat the dominant inguisticmarket as embodied n someform of legitimizedor standard anguage holds swaywithouthindrancefrom(inWoolard'serms)alternative ernacularmarkets.Figure3 is a sche-maticrepresentationf the relationof socialnetwork tructureo thesethreelife-modes.It is importanto emphasizehat theconceptof life-mode, ike that of net-work,is a structural ne, in thatthe ideologicaland culturalcharacteristicsof a particular ife-mode are determinedby its contrastto the other life-modesin the social formation.Theinterrelationshipsmongthe three ife-modes and theculturalpractices ssociatedwitheachone willtherefore akedifferent orms n, for example,Denmark, reland,England,andGermany.In eachof thesecountries, he three fundamentalmodes of production hatthe life-modes eflect willappearn differentvariants ndin differentcom-binations of oppositionand independence H0jrup I983:47). One conse-quenceof thischainof dependence unningrompoliticalandsocioeconomicstructureshrough ife-modes o networkstructure ndultimately o socio-linguistic tructuresee Figure3) is, as we havealready uggested,hatclose-knitnetworkswill be associatedwithLife-mode individualsn some nationstatesmorethanin others.Thisseems o bethe case f, forexample,wecom-22

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    24/27

    SOCIAL NETWORK AND SOCIAL CLASS

    pareBelfastwithCopenhagen. n Copenhagen, hesewageearnersare ap-parentlymore mobile and prosperousand less inclinedto live and worktogether n close-knitgroupsof the kinddescribedn Belfast(Gregersen&Pedersen 99i). Thisin turnwillgive riseto sociolinguistic atterns hat de-pendon varying ocal contingencies ndhenceto urbanvernaculars aryingin theirdegreeof focusingand vitality.CONCLUSIONThe purposeof thisarticlehasbeen to work towardan integratedmodelofsociolinguistic tructure hat linksin an explicitwaythe socialvariablesofsocioeconomicclassandsocialnetwork.Althoughthesevariablesare oftenpresentedas unrelatedor evencontradictory,we havetriedto demonstratethat, although heyare at differentordersof generality, t is usefulto pro-pose an interpretationof sociolinguisticspace that conceivesof them asinterrelated.Theyare, of course,related n realityalso. Particular onfigu-rationsof network tructure o not emergeaccidentallyorno particularea-son - the formtheytake is dependent n thelarge-scaleocial,political,andeconomicstructureshat sociolinguists enerally ccess n termsof socioeco-nomicclass. Thus, an attempt o integrateclassandnetworkas interpreta-tive categories s well motivated.However,the modelof socialclassto which we have appealedhere s notthe stratificational onsensus-basedmodelthat has beengenerally avoredby Western ociolinguists.WhereasLabov'sview of speechcommunityhasemphasized harednormsthroughout he communityand is thus related oa consensusmodelof socialclass, we havepreferredo emphasize he con-flicts and inequalitiesn societythat aresymbolizedby opposinglinguisticnorms.Thisanalysis mphasizeshebasisof personal ocialnetworksn con-sensus, whereasclass differences nvolve not consensusbut conflict. Theweak-tiemodelof Granovetteruggests meansof linkingnetworkandclass,as strong-tie ituationspredictagreementon norms,whereasweak-tiesitu-ations favorchangeand hence conflictof norms.The analysisof higher evel socialstructurehatwe havefoundmost use-ful heretreats socialand culturaldivisionsas emergingultimately romtheeconomic nequalities roducedbysocial class.H0jrup I983) assumed ucha frameworkn his process-basedmodel of life-mode,whichwe haveusedin conjunctionwithnetworkanalysis o developan integratedmodelof so-ciolinguistic tructure.By emphasizingn particularhe importanceof dis-tinguishingbetweenrelatively trongand relativelyweak networkties, wehavesuggestedhow theseeconomicallydeterminedife-modesgiverisenotonly to the socialand culturaldifferencesdescribedby H0jrup,but to dif-ferentkindsof networkstructure.Thiswill furtherenableus to specifythe

    23

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    25/27

    LESLEY MILROY AND JAMES MILROY

    conditions n which helinguistic ormsof the groupsare ikely o be focusedor diffuse, and the conditionsin which they are open to, or resistant o,change.NOTE

    I. Versions f thisarticleweregiven at the Sociolinguistics ymposium t RoehamptonnApril 99o andat theInternationalolloquium DesLangues t desVilles, rganized yCLAD,at L'Universite heikAnta Diop, Dakar,Senegal, n December990. Our hanks o colleagueswhocommented ndgaveadviceat thesemeetings.Particularhanks o Padraig Riagain,SallyBoyd,and DellHymes or theirusefulcomments; o Inge-LisePedersen,University f Copen-hagen, who drewourattention o H0jrup'swork on life-modes;andto MarinaBeale for as-sistancewiththe backgrounditeraturen sociology.

    REFERENCESAsh, S., & Myhill, J. (I986). Linguistic correlates of inter-ethnic contact. In Sankoff (I986).33-44.Bortoni-Ricardo,. M. (I985). Theurbanisationf ruraldialect peakers: sociolinguistictudyin Brazil. Cambridge:CambridgeUniversityPress.Bott, E. (197I). Familyand socialnetwork 2nd ed.). London:Tavistock.Bourdieu, P. (I977). The economicsof linguistic exchanges. Social Science InformationI6(6):645-68.(1984). Capital et marche linguistique. Linguistische Berichte 90:3-24.Brown, P., & Levinson,S. (1987).Politeness.Cambridge:CambridgeUniversityPress.Cochran,M. (i99o). Environmentalactorsconstraining etwork evelopment.nM. Cochranet al. (I990). 277-96.Cochran,M., Larner,M., Riley, D., Gunnarsson,L., &Henderson,C. R. (eds.)(i99o). Ex-tending amilies: Thesocial networks f parentsand theirchildren.Cambridge: ambridgeUniversityPress.Cohen,A. (ed.) (I982). Belonging.Manchester:ManchesterUniversityPress.Cross,W.E. (I99o). Raceandethnicity:The effects on social networks. nM. Cochran t al.

    (i0o). 67-87.Dennis, N., Henriques,F. M., & Slaughter,C. (1957). Coal is our life. London:Eyre andSpottiswoode.Dittmar, N., & Schlobinski, P. (I988). Thesociolinguistics f urbanvernaculars. erlin: deGruyter.Dittmar,N., Schlobinski,P., &Wachs,I. (I988). Thesocialsignificance f the Berlinurbanvernacular. n N. Dittmar& P. Schlobinski (I988). 19-43.Edwards,V. (i986). Language n a blackcommunity.Clevedon: Multilingual Matters.Edwards,W. (i99o). Socialnetwork heoryand language ariationn Detroit.Paperpresentedat the EighthSociolinguistic ymposium,Roehampton,London.Fischer, C. (i982). To dwell among friends: Personal networks in town and city. Chicago: Uni-versity of Chicago Press.(I984). The urban experience (2nd ed.). New York: Harcourt Brace Jovanovich.Gal,S. (I979). Language hift:socialdeterminantsf linguistichangenbilingualAustria.NewYork: Academic.(I988). The political economy of code choice. In Heller (I988). 245-63.Gans,H. J. (I962). Theurbanvillagers:Groupand classin thelife of Italian-Americans2nded.). New York: Free Press.Giddens, A. (I984). Theconstitutionof society.Cambridge: Polity.(I989). Sociology.Cambridge: Polity.

    24

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    26/27

    SOCIAL NETWORK AND SOCIAL CLASS

    Goldthorpe, . H. etal. (I968-69). Theaffluentworker n the classstructure.Cambridge: am-bridgeUniversityPress.Granovetter, M. (I973). The strength of weak ties. American Journal of Sociology 78:1360-80.(I982). The strengthof weak ties: A network heoryrevisited. n P. V. Marsden& N.Lin (eds.), Social structure nd networkanalysis.London:Sage. I05-30.Gregersen, ., & Pedersen, . L. (i99i). TheCopenhagentudy n urban ociolinguistics.Univer-sitetsjubilaeetsDanske Samfund,SerieA). Copenhagen:C. A. Reitzel.Gregg,R. J. (1972). The Scotch-Irish ialectboundariesn Ulster.In M. F. Wakelin ed.),Pat-terns n the olk speech of theBritish sles. London:Athlone Press. 109-39.Gumperz, J. (I982). Discourse strategies. Cambridge: Cambridge University Press.Guy, G. R. (I988). Language and social class. In Newmeyer (I988). 37-63.Harman,L. D. (I988). The modern tranger:On languageand membership.Berlin:Moutonde Gruyter.

    Heller, M. (I988). Code-switching.Berlin:Moutonde Gruyter.H0jrup,T. (I983). The conceptof life-mode:A form-specifyingmode of analysisapplied ocontemporarywesternEurope.EthnologiaScandinavica-50.Kroch, A. S. (I978). Toward a theory of social dialect variation. Language in Society 7:17-36.Labov, W. (1972). Sociolinguistic atterns.Philadelphia:Universityof Pennsylvania ress.(ed.) (I980). Locating anguage n timeand space.New York:Academic.(I98I). Resolving he neogrammarianontroversy.Language 7:267-309.(I986). Language tructure nd social structure. n S. Lindenberg t al. (eds.), Ap-proachesto social theory.New York:RussellSage.Labov,W., &Harris,W. (i986). De factosegregationf blackand whitevemaculars.nSankoff

    (I986). 1-24.Labrie,N. (I988). Commentson Berlinurbanvernacular tudies.In Dittmar& Schlobinski(I988). 191-206.

    Lippi-Green, . L. (I989). Socialnetwork ntegration nd anguage hange n progressn a ruralalpinevillage.Language n Society I8:2I3-34.Lockwood,D. (I989). Theblack-coated orker:A study n class-consciousness.xford:Claren-don. (Original work published i966)Mewett,P. (I982). Associational ategories nd the sociallocationof relationshipsn a Lewiscroftingcommunity. n Cohen I982). IOI-30.Mills, C. W. (I956). Thepower elite. Oxford:OxfordUniversityPress.Milroy,J. (I98I). Regionalaccentsof English:Belfast.Belfast:Blackstaff.(I987). The conceptof prestigen sociolinguistic rgumentation.YorkPapers n Lin-guistics I3:2I5-26.(I99I). The interpretationf social constraints n variation n BelfastEnglish.In J.Cheshireed.),Englisharound he world:Sociolinguisticerspectives. ambridge: ambridgeUniversityPress.75-85.(I992). Linguisticvariationandchange.Oxford: Blackwell.Milroy, J., &Milroy,L. (I978). Belfast:Changeand Variationn an urbanvernacular. n P.Trudgill ed.), Sociolinguistic atterns n BritishEnglish.London:Arnold. 19-36.(I985). Linguistic hange,socialnetwork,andspeakernnovation.Journalof Linguis-tics 21:339-84.Milroy, L. (I987a). Language and social networks (2nd ed.). Oxford: Blackwell.(I987b).Observing ndanalysingnatural anguage.Oxford:Blackwell.Milroy,L., & Li,W.(I99I). A socialnetwork erspectivencode-switchingnd anguagehoice:The exampleof the Tyneside ommunity. nPapers or theSymposium n Code-switchingin BilingualStudies:Theory, ignificance ndperspectives. Vol. i). Strasbourg:EuropeanScienceFoundation.Mitchell,J. C. (i986). Networkprocedures. n D. Frick ed.), Thequalityof urban ife. Ber-lin: de Gruyter. 3-92.Newmeyer,F. (ed.). (i988). Linguistics:TheCambridgeurvey.Vol.4. Cambridge: ambridgeUniversityPress.Patterson,D. (i860). Provincialisms f Belfast.Belfast:MayneBoyd.Pedersen, .-L. (I99I). Sociolinguisticlassificationna gender erspective.Unpublishedmanu-script.Universityof Copenhagen, nstitut or danskdialektologi.

    25

    This content downloaded on Thu, 13 Dec 2012 03:05:34 AMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 8/12/2019 MIlroy 1992

    27/27

    LESLEY MILROY AND JAMES MILROY

    Rickford,J. (I986). The need for new approaches o social classanalysis n linguistics.Lan-guage and Communication 6(3):215-2 1.Rogers,E. M., & Shoemaker, . F. (i97I). Communicationf innovations2nded.). New York:AcademicPress.Sankoff, D. (I986). Diversityanddiachrony.Amsterdam: ohn Benjamins.Sankoff,D., Cedergren,H., Kemp,W., Thibault,P., & Vincent,D. (I989). MontrealFrench:Language, lass,and deology. n W.Fasold& D. Schiffrin eds.),Language hangeand vari-ation. Amsterdam: ohn Benjamins. 07-I8.Schmidt,A. (I985). Youngpeople'sDjirbal. Cambridge:CambridgeUniversityPress.Staples,J.H. (I898). Notes on UlsterEnglishdialect. Transactionsf thePhilologicalSociety357-87.Surra,C. A. (I988). Theinfluenceof the interactive etworkon developing elationships. nR. M. Milardo ed.), Familyandsocial networks.Newbury,CA: Sage. 48-82.

    Trudgill,P. (I983). On dialect. Oxford:Blackwell.(I986). Dialects in contact. Oxford:Blackwell.Williams,R. A. (I903). Remarks n Northern rishpronunciation f English.ModernEnglishQuarterly 6:129-35.Wirth,L. (1938). Urbanism s a wayof life. AmericanJournalof Sociology 4(I): 1-24.Woolard,K. (I985). Language ariation nd culturalhegemony:Towardan integration f lin-guisticandsociolinguistic heory.AmericanEthnologist12:738-48.

    26


Recommended