认知语言学与语言类型学的关系.docx

认知语言学与语言类型学的关系.docx

《认知语言学与语言类型学的关系.docx》由会员分享，可在线阅读，更多相关《认知语言学与语言类型学的关系.docx（18页珍藏版）》请在冰豆网上搜索。

认知语言学与语言类型学的关系

CognitiveLinguisticsandlinguistictypology

JohanvanderAuweraandJanNuyts

UniversityofAntwerp

0.Introduction

ThischapterlooksintotherelationsbetweenCognitiveLinguisticsandlinguistic

typology.Thefirsthalfofthechapteroffersa‘neutral'characterizationofthefieldof

linguistictypology.Linguistictypologyisdefinedasacross-linguistic,descriptiveaswellasexplanatoryenterprisedevotedtotheunityanddiversityoflanguagewithrespecttolinguisticformortherelationbetweenlinguisticformandmeaningorfunction.ThesecondhalfisdevotedtoanexplorationoftherelationsbetweenlinguistictypologyandCognitiveLinguistics.Itisarguedthatthetwostrandsareeminentlycompatible,thatthereisworkthatillustratesthis,butalsothatmostcognitivelinguistsandtypologistsneverthelessworkindifferentspheres.Inafirstsectionwediscussthedifficultyofapplyingtypology'ssamplingmethodinCognitiveLinguistics.Inasecondone,wefocusonthetypologists'eoprireimntationongrammarandtheirhesitationtorelatetheirstrictly

speakinglinguisticgeneralizationstowidercognitiveconcerns.

1.Whatislinguistictypology?

Theterm‘linguistictypology'israthergeneral.Itcouldbetakentomeannomore

thantheinvestigationoflinguistictypes.Linguistictypesappearwhenthelinguisthasclassifiedlinguistic

entitiesinvirtueofasimilarity.Inthissense,anylinguisticdisciplinecountsastypology.Inmorphology,forinstance,prefixesandsuffixescanbesaidtobeentitiesofthesametype,called‘affixes';andaffixesandrootsorstemsarealsoentitiesofthesame

type,called‘morphemes'.Insociolinguistics,mostAustralianlanguagesandmostnative

Americanlanguagesareofthesametype:

theyareallthreatenedlanguages.Orin

historicallinguisticsonecansaythatNorwegianandDanisharelanguagesoftheGermanictype.Inreality,however,theterm‘linguistictypology'isusedinanarrowe

way.Although,inpartasaresultofthegeneralityoftheliteralmeaningjustdescribed,therearevariouscontroversiesastoitsexactnature,thedefinitionin（1capturesatleastitsmostcentralconcerns.

（1Linguistictypologyisacross-linguistic（adescription（bandexplanation（coftheunity

anddiversityoflanguages（dwithrespecttolinguisticform（eortherelationbetweenlinguisticformandmeaning/function（f.

Intheabovedefinition,sixfeaturesaresingledout.Wewilldiscusstheminsomedetail.

Sayingthatlinguistictypologyshouldbecross-linguistic—feature（a—meansthatobservationsshouldbebasedonawidevarietyoflanguages.Inprinciple,onecannotdotypologyonthebasisofonelanguage,notevenifthelanguageisaconglomerateofdivergentdialects.Also,instudiesofonlyahandfuloflanguagesonedoesnotusuallyspeakabout‘typology',butabout‘contrastivelinguistics'.Thelanguagesselectedsfurthermoreconstituteasample.Thesizeofthesample（whichcanvaryconsiderab—lycf.the22languagesofXrakovskijed.2001onimperativestothe272ofSiewierska1999onverbalagreementisgearedtowardsbeingrepresentativeofthevariationinthetotalityoftheworld'slanguage.Ofcourse,represen-tnaetisvseisnotsolelyamatterofsamplesize.

Typologistsnowhaveincreasinglybettermethodsto

controlforgeneticorarealbias—i.e.thedangeroftakingtoomanylanguagesof（respectivelythesamefamilyorthesameare—aandevenfortypologicalbias—i.e.thedangeroftakingtoomanylanguagesofwhichitisalreadyknownthattheyaretypologicallysimilar（seeDryer1999;Rijkhoff&Bakker1998;Croft2003:

19-28.

Astofeature（b,typologistsfirstofallneedtodescribethefacts.Thisislessobviousthanitmaysound,however.Descriptionsarebasedonanalyticconcepts,whichareunavoidablyinspiredbytheories.Hencenodescriptioncanbefullytheoryindependent.Thisisamatterofdegree,however.Inextremecases,descriptionscanvarytremendously,tothepointevenofbeingincomprehensibletoanybutlinguistsofthesametheoreticalpersuasion.Sincetypologicaldescriptionsshouldbeusefultolinguistsofdiversetheoreticalorientations,however,itisessentialtoreducetheirtheory-dependenceasmuchaspossible.Aversionofthisaimforneutralitycoupledtoanaversiontothecurrentproliferationoflinguistictheorieshasbeencalled“basiclinguistictheory

Dixon（1997:

128-135.

Beforeweturntofeature（c,concerningexplanation,letusclarifywhatitisthatshouldbedescribed.Feature（dstatesthattypologistsarelookingfortheunityanddiversityoflanguages.Typologistsdescribehowlanguagesdiffer,butatthesametimealsohowtheyaresimilarorevenidentical,relativetooneormoreparameters.Featurescharacterizingalllanguagesarecalled‘universals'.Therearewhatmaybecalled‘absoluteuniversals',whichapplytoalllanguages,asillustratedin（2,andthe-rearenon

absoluteor‘statistical'universals,whichholdtrueofmostlanguages,asillustratedin（3.

（2a.Alllanguageshavenounsandverbs.（Whaley1997:

59

b.Alllanguageshavestops.（Maddieson1984:

39

（3MostlanguageshaveeitheranSOVoranSVObasicwordorder.（Tomlin1986:

22Theuniversalsin（2and（3makeaclaimaboutapropertythatdoesnotdependonanyotherpropertyoflanguage,i.e.theyarenot‘tchoenpdriteiofenrareldte'rm—orn—ot

‘implicational'.Buttherearealsoimplicationaluniversals,anditisthesethathavebeenmostprominentinthelastfewdecades.Theytoo,canbeabsoluteorstatistical.（4givesexamplesofabsoluteimplicationaluniversals.

（4a.IfalanguagehasadominantVSOwordorder,itwillhaveprepositions.（Greenberg

1963:

78

b.IfalanguagehasNPinternalagreement,thentheagreementfeaturesmayincludecase,butnotperson.（Lehmann1988:

57

Particularlyinterestingaboutanimplicationaluniversalisthatitdoesnotonlytellusaboutunitybutalsoaboutdiversity.（4a,forinstance,impliesthreesubsetsofpossiblelanguages:

（5a.dominantVSOorderandprepositions

b.nodominantVSOorderandprepositions

c.nodominantVSOorderandnoprepositions

Inlogicalterms,thiskindofuniversalisamaterialimplication.Therearethreesituationsthatmakeittrue:

antecedenttrueandconsequenttrue;antecedentfalseandconsequenttrue;antecedentfalseandconsequentfalse.Hencepostulatingthiskindofuniversalgoeshandinhandwithaclassificationoflanguages.Animplicationaluniversaldoesruleoutonesituation,of

course,viz.thatofatrueantecedentandafalseconsequent.Thus,（4arulesoutthecombinationin（5d.

（5d.dominantVSOorderandnoprepositions

Actually,typologistsnowbelievethatlanguagesoftype（5ddoexist,afterall（see

Song2001:

46.Thismeansthattheuniversalin（4aisstatisticalonlyand,infact,themoretypicaluniversalhasnowbecomethestatisticalone（Dryer1998.Ofcourse,thisobservationinnowaydiminishesthevalueoftheuniversal.Onthecontrary,typologistsmustnowexplainboththeverystrongtendencytoruleout（5d,aswellasthefactthatsomelanguagescanneverthelesswithstandthistendency.

Thistakesustofeature（cofthedefinitionin（1,viz.explanation.Dotypologistsalsoattempttoexplaintheregularitiestheyobserve?

Theydo,butinsomecornersoflinguisticstheirexplanationsaretakentobeofnegligibleorinsufficientquality.Thereasonisthatexplanationrequiresatheory,andnotalltheoriesarecompatible.Asstatedbefore,mosttypologicaldescriptionsaimtoberelativelytheory-neutralandtooffer‘descriptive'or‘empirical'observations,o-f（4th.eThkeinsdeinge（n2eralizationscanthenserveasinputforvarioustheories.Inasimpleworld,then,thetypologistscouldbedeliverersofdata,anditisuptotheoreticianstoexplainthese.Butintheactualworld,thedivisionoflaborisnotthatsimple.Inmoderntypology,mosttypologistsattempttoexplainthedatathemselves,andthispartoftheworkisnottheory-neutralatall.Intermsofthecurrentsharpdivisioninlinguisticsbetweenformalistandfunctionalistparadigms,typologiststendtobefunctionalists.1Asaconsequence,thenon-typologicaltheoreticianofthefunctionalistbrandwillusuallynotonlyappreciatethedatafromthetypologist,butalsohis/hertheoreticalconsiderations.Buttheformalistnon-

typologicaltheoreticianwillusuallyatbestbegratefulforthedatabutfeelfreetoneglectthetypologist'stheory.

Whatcana‘typologicalexplanation'be,then?

Letusfirstdiscusstwofeatures

whichitshouldnothave,atleastnotaccordingtomanytypologists:

itcannotrelyon‘geneticinhe-rtiance',anditcannotbe‘areal'.Bothelementsrequiresomeelaboration.

First,sayingthattypologicalexplanationcannotrelyon‘geneticinheritance

thatasimilaritybetweenlanguagescannotbeaccountedforbysimplyreferringtothe

hypothesisthattheyinheriteditfromacommonancestorlanguage.（Notethatthisonlyconcernsgeneticinheritanceperse,andnotgenetic/diachronicexplanationingene—ralseebelow.Forexample,partofthereasonwhybothmodernDanishandmodernDutchhavetwotypesofpreterite—withadentalsuffixorwithastemvowelchange—isthattheparentlanguagehadthemtoo.Or,mostTibeto-Burmanlanguagesareverbfinalandpostpositional,andtheymayhaveinheritedthisfromProto-Sino-Tibetan（DeLancey1987:

806.Butofcourse,theseobservationsassuchcannotbethewholestory,forlanguagesdoalsoeasilydiscardpartoftheirinheritance,viz.throughlanguagechange.Theessentialquestionis:

whydolanguages（ancestorsandinheritorshavesuchfeatures,andwhydidtheyordidtheynotkeepthemindiachronicchange?

isCroft

Wearetouchinghe