搜索
1TypologyandUniversalsWilliamCroft,UniversityofNewMexico1.Introduction:thetypologicalandgenerativeapproachestolanguageuniversalsTypologyrepresentsanapproachtothestudyoflinguisticstructurethatdiffersincertainimportantrespectsfromthegenerativeandthefunctionalistapproaches(Wasow,thisvolume;VanValin,thisvolume),althoughitiscloserinspirittothelatter.Themostimportantdifferencebetweentypologyandtheseotherapproachestolinguisticstructureisthatthetypologicalapproachisfundamentallycrosslinguisticinnature.Aformalistlinguistcan,andmostformalistlinguistsdo,analyzeasinglelanguageinthesearchforuniversalsoflanguagestructure.Thereareformalistanalysesofmanydifferentindividuallanguages,butrelativelyfewcrosslinguisticformaliststudies(anotableexceptionistheworkofMarkBaker,e.g.Baker2003;butseeCroft2008).Afunctionalistlinguistcan,andoftendoes,analyzeasinglelanguageinthesearchforuniversalsoftherelationshipoflanguagestructuretolanguagefunction.Somefunctionalisttheoriesaremoreextensivelysupportedbycrosslinguisticdata,notablyFunctionalGrammar(Dik1997)andRoleandReferenceGrammar(VanValinandLaPolla1997).Thesetheoriesrecognizethevariationingrammaticalstructuresacrosslanguages,buttheirframeworksgenerallyfocusonwhatisbelievedtobecommontoalllanguages(see§5).Typology,ontheotherhand,isfundamentallycomparative.Agenuinelytypologicalanalysisofagrammaticalconstruction,oraphonologicalpattern,orotheraspectsoflanguage,examinesthevariationacrossalargenumberoflanguages.Inthisrespect,typologyresemblescomparativehistoricallinguistics.Thegoalsoftypologyandcomparativehistoricallinguisticsareverydifferent,however,althoughtheresultsofeachareessentialtotheother(see§2).Comparativehistoricallinguisticsseeksgenetic(familytree)relationshipsamonglanguages,inordertodiscoverthehistoryofthelanguagesandtheirspeechcommunities(Joseph,thisvolume).Typologyexaminesabroadsampleoflanguagesinordertodiscoveruniversalsoflanguagestructureandproposeexplanationsofthoselanguageuniversals.Forthisreason,typologyislinkedtolanguageuniversals.Forsomelinguists,typologysimplymeansthedescriptionofvariation,thatis,howlanguagesdifferintheirstructure.Forexample,asimpledescriptivetypologyofthewordorderofnumeralandnounwoulddividethelanguagesoftheworldintothreebroadtypes:•Thoseinwhichnumeralsnormallyprecedethenountheymodify,asinEnglish(Indo-European)twowomen;•Thoseinwhichnumeralsnormallyfollowthenountheymodify,asinMa’di(Nilo-Saharan,Uganda-Sudan)àgO@su@‘menfour’;•Thoseinwhichnumeralsmayeitherprecedeorfollowthenountheymodify,asinWardaman(Australian,Australia)guyaminyimulurruwuyayi‘twoold.women’ormarlugalege‘old.manone’2However,typologyinthelinguisticsenseismorethanaclassificationofhowlanguagesdifferintheirstructure.Adescriptivetypologyleadstogeneralizationsthatconstrainhowmuchlanguagescanvary;thosegeneralizationsarelanguageuniversals.Severalexamplesoflanguageuniversalswillbegiveninthischapter.Thesearchforlanguageuniversalsissharedbytypologyandgenerativegrammar.However,thelanguageuniversalsderivedfromtypologicalresearcharequitedifferentfromthosederivedingenerativegrammar,althoughthegenerativeandtypologicalapproachesaroseataroundthesametime.Althoughthebeliefinlanguageuniversalshasconsiderablemoderncurrency,itisbynomeansanecessaryfactoruniversally-heldopinion,andinfacttheoppositeviewwaswidelyheldinAmericanlinguisticsuntilaround1960.Toaconsiderabledegree,thedifferencebetweenthegenerativeandtypologicalapproachestolanguageuniversalscanbetracedtothedifferenttraditionstowhichtheirrespectivefounders,NoamChomskyandJosephGreenbergresponded.Thegenerativeapproachrepresentsareactionagainstbehavioristicpsychology,whilethetypologicalapproachrepresentsareactionagainstanthropologicalrelativism.Thebehavioristviewoflanguage,inparticularlanguagelearning,isanti-universalistinthatitpositsnoinnate,universalinternalmentalabilitiesorschemas.Inthebehavioristview,linguisticcompetenceisacquiredthroughlearningofstimulus-responsepatterns.Incontrast,thegenerativeapproachpositstheexistenceofinnateinternallinguisticabilitiesandconstraintsthatplayamajorroleintheacquisitionoflanguage.Itistheseconstraintsthatrepresentlanguageuniversalsinthisapproach.TheargumentusedbyChomsky(e.g.,Chomsky1976)fortheexistenceofinnateuniversallinguisticcompetencereferstothe‘povertyofthestimulus’.Itisarguedthatthechildhasanextremelylimitedinputstimulus,thatis,theutterancesthatitisexposedtofromthemotherandothercaregivers.Thisstimulusisincapableofpermittingthechildtoconstructthegrammaroftheadult’slanguageinaclassicbehavioristmodel;therefore,thechildmustbringinnateuniversalsofgrammaticalcompetencetobearonlanguageacquisition.Hencetheprimaryfocusonuniversalsinthegenerativetraditionhasbeenontheirinnatecharacter.Theanthropologicalrelativistviewoflanguageisthatthelanguagesoftheworldcanvaryarbitrarily:‘languagescoulddifferfromeachotherwithoutlimitandinunpredictableways’,inafamousquotationfromthelinguistMartinJoos(Joos1957:96).ThisviewoflanguagewasparticularlystrongamonganthropologicallinguistsstudyingNorthAmericanIndianlanguages,whichindeeddifferradicallyinmanywa