Cohesion in English--Halliday & Hasan

Lingua45(1978)333-354漏North-HollandPublishingCompanyI~VIEWARTICLEONCLASSIFYINGANAPHORICRELATIONSAreviewofM.A.K.HallidayandR.Hasan,CohesioninEnglish*RodneyHUDDLESTONUniv.ofQueensland,St.Lucia,AustraliaO.introdecfioaHallidayandHasan'sCohesioninEnglish(1976)dealswithvarioustypesofnon-structuralrelationsthatlinkonepartofatextwithanother.Theyseeatextasasemanticunitratherthanastructuralone;itdiffersfromunitslikesentence,clauseorphrasenotinsizebutinkind.Atexthas'texture'ratherthanstructure,thiste~t.u~ebeingwhatdifferentiatesitfromarandomcollectionofsentences.Thetexttte;.saproductnotsimplyofthecohesiverelationsinthetext"alsc,~nvolvedarecertainaspectsofsentencestructure,namelythe'theme"and'information'systemsdescribedinsomedetailinHatLiday1967,andalsothestructureJofdiscourse,thatis,thestructureassociatedwithparticulargenressuchasnarrative,sonnet,operatinginstructionsorwhatever.ThebulkofteebookisdevotedtoaquitedetaileddescriptionofthelinguisticresourcesinEnglishthatcanestablishcohesivetiesbetweensentences.Thesearerelationsofdependencebetweenelementsindifferer~tsentencessuchthatthedependenttermcannotbeeffectivelyinterpretedexceptbyrecoursetotheotherterm(p.4).Fivetypesofcohesiverelationaredistinguished:reference,substitution,ellipsis,conjunctionandlexicalcohesion.Referentialcohesioncoverspersonalpronouns,demonstratives(includingthedefinitearticle),andcomparatives-ofidentity(likesame),similarity,difference,quantityorquality.Themainsubstitutesdiscussedareone(s),soandnot,thesame,do,doso.Thedescriptionofellipsisisorganizedintermsofellipsiswithinthenominalgroup,theverbalgroup*Longman,London,1976.334Reviewarticleandwithinoroftheclause.1.*Conjunctiveexpressionsincludesuchitemsasalso,furthermore,but,however,therefore,otherwise,then,nextandmanyo~bers;theyareclassified,attheprimarydegreeofdelicacy,asadditive,adversative,causalandtemporal.Finally,lexicalcohesion,whichisdiscussedverymuchmorebrieflythantheothers,involveseitherreitera-tionofthesameorsemanticallysimilarlexicalitems,or'collocation',theuseoflexicalitemsthatregularlyco-occur,suchaslaughandjoke,etc.Inadditiontothefivechaptersdealingseparatelywiththesedifferenttypesofcohesion,therearetwogeneralchaptersdiscussingtheconceptsoftext,texture,cohesionandthedistinctionbetweenthefivetypes,andaconcluding,chapterwhichexplainsandexemplifieshowtheearliermaterialcanbeusedintextualanalysis.Thefairlypracticalpedagogicalaimofthislastchapterprobablyaccountsforthefactthatthebookcontainsaratherlargeamountozrepetition.Thebookoffersmuchtointerestthoseworkingonthelinguisticpropertiesoftexts,onthetheoryofanaphcraa.,ldonEnglishgrammar.';r9Itwillalsoappealtothosein~erestedinsystemicgammar9forthedescriptionispresentedintermsofthatmodelandthebookaddscon-siderablytothereadilyavailablematerialgivingHaUiday'ssystemicanalysisofEnglish.2IndeedIthinkitcontainsagooddealmoreofthisthanisjustifiedbythecentralconcernsofthebook.Forexample,indiscussingthetypeofellipsisfoundin(1).say,(1)Hemusthavedestroyedthem.-Hemayhave,Isuppose.HaUidayandHasan(henceforthH-H)taketheanaphoricrelationtoholdbetweenmayhaveandmusthavedestroyedthem,whichinvolvestheminalengthydescriptionoftherespectsinwhichthe'presupposing'and'presupposed'verbalgroupsmaydiffer-andthisinturninvolvesafairlydetailedaccountofHaUiday'sanalysisoftheverbalgroup,ananalysiswhichdiffersmarkedly(e.g.byrecognizinga36-termtensesystem)fromthatadoptedbynon-systemicgrammarians.Mostofthiscouldbedispensedwithifonetooktheanaphorin(1)tobezero,withdestroyedthemastheantecedent.Moreover,whateverone'sgeneralevaluationofHalliday'sanalysisoftheverbalgroup,itseemstomethatthefactsconcerningellipsislendnosupporttoit-ratherthecontrary.Consider,forinstance,suchexamplesas(2i)Whatishegoingtodowithallthatparaphernalia?-Catchfish.*Fortextoffootnotesseepp.352-354.Reviewarticle335(2ii)Tomisgoingtojointheexpedition.Lizis(goingto),aswell.(3i)Whatishetryingtodowithallthatparaphernalia?-Catchfish.(3ii)Tomistryingtojointheexpedition.Lizis(tryingto),aswell.H-Hanalysebegoingtoasanauxiliary(markingfuturetense),whereastryisalexicalverb.Thismeansthattheirdescriptionof(2)doesnotgeneralizeto(3),asitshould,andindeedtheiraccountdoesnotprovidefor(3i)atall.Further,althoughtheydevote30pagestoverbalellipsis,theydonotdiscusssuchmattersasthedistinctionbetweenusedandoughtexemplifiedin(4i)Sheoughttoplaytennis.Tomought(to)aswell.(4ii)Sheusedtoplaytennis.Tomusedtoaswell.wherethetoisobligatoryinthesecondclauseof(4ii)butoptionalint'Ai~_~ine~th~t~h~rai~nntthar~aliTatlnnnfany~v~tarni~fa.atnraitk"~,/.....J"Jreadilyescapesattentionwithinthesystemicframework.Iwouldhope,however,thatitsuseofthesystemicmodelwillnotdeterthoseunsympathetictothismodelfromusingthebook.ForitcontainsagreatdealofexcellentdescriptivematerialontheanaphoricdevicesinEnglish,significantlymorethanonefindsinthelargereferencegrammars.ItiswithoutquestionaninvaluablesourceofinformationandinsightconcerninganareaofEnglishgrammarthatisofconsiderablerelevancetocurrentissuesinlinguistictheory.Intheremainderofthisreview,Iwanttotakeupthoseaspectsofthebookathatbearonthegeneraltheoryofanaphora.1.EndophoraandexopheraOnadimensionthatcutsacrosstheonealongwhichreference,substitu-tionandellipsisaredistinguished,H-Hclassify'phoric'relationsaccordingtothefollowingschema(p.33):(5)i.[phora]4_!!iv.anaphoraIii.endophoraILv.cataphora1iii.exophora336ReviewarticleAnendophoricrelationisonebetweentwoelementsofthetext,whereasanexophoricrelationholdsbetweenatextualelementandsomethingoutsidethetext,inthe'situation'.Inanaphora,thephoricelementfollo,~stheitemonwhichitisdependentforitsinterpretation,incata-phoraitprecedesit.Thethreeterminalcategoriesofthetaxonomyareexemplifiedin(6)Johndidn'tcome.-Washeill?(7)Iwouldneverhavebelievedit.They'veacceptedthewholescheme.(8)I~aveaheadache.In(6),heisanaphorictoJohn,in(7)itiscataphorictothewholeofthefollowingsentence,in(8)Iisexophorictothespeaker.H-Hthususetheterm'anaphora'inanarrower(butmoretradaitional)sensethanitnormallyhasintheextensivegenerativeliteratureonthesubject.Forreasonswhichwillemergepresently,Icannotacceptthetaxonomyin(5),andthusdonotwishtoadopttheterminologythatgoesw'thit.Ishallaccordinglyuse'Anaphora',withacapitalA,forH-H'scategoryii,distinguishingcategoriesivandvasForwardsAnaphoraandBackwardsAnaphorarespectively)ThetwotermsinanAnaphoricrelationIshallspeakofastheAnaphorandtheAntecedent6_notethat,since'Anaphora'isappliedtocategoryii,bothAnaphorandAntecedentarewithinthetext.WhenfollowingH-H'sterminology,Ishalluselowercaseinitials.Ishalladoptthesameconventiontodistinguishbetween'Refer'andH-H's'refer',andsimilarlyfortheirderivatives.Thus'Reference'isusedasinthephilosophyoflanguageandlinguisticwritingstakingoverthatusage(cf.Lyons1977:Ch.7):itisdistinguishedfromdenotation,sothatwemightsaythatthenoundogdenotestheclassofdogswhereasinanappropriateutteranceofThatdoglookshungrytheNPthatdogisusedtoRefertoacertainentity,topickouttheentityaboutwhichsomethingisbeingasserted;'reference',bycontrast,isH-H'stermforoneofthethreemajortypesof'phoric'relations.7Nowwhatseemstometobetheoreticallyunsoundinthetaxonomyof(5)isthatittreatscategoriesiiandiiiasdifferentsub-casesofasinglemoregeneralphenomenon.H-Harequiteexplicitaboutthis-thattherelationbetweenanAnaphoranditsAntecedentisthesameasthatbetweenanexphoricelementandthecorrespondingelementinthesitua-tion:'therelationofreferencemaybeendophoricaswellasexophoric-[...]areferenceitemcanrefertoanelementintherex,aswellastoanReviewarticle337elementinthesituation'(p.306).Buttheserelationsaresurelyquitedifferent.Considerexamples(6)and(8)above.Therelationwhichholdsinanutteranceof(8)betweenIandthespeakerislikethatwhichholdsinanutteranceof(6)betweenheandthepersonJohn,notbetweenheandthelinguisticexpressionJohn"thequestionisaboutthehealthofaperson,notofacertainnounphrase.In(theintendedinterpretationofanutteranceof)(6),heisAnaphorictoJohnbutReferstoJohn.ItisbyvirtueofitsAnaphoricrelationtoJohnthatthelistenerisabletoidentifytheReferentasJohn,buttheAntecedentitselfisnottheReferent.In(8)noAntecedentisrequiredbecausethespeakernecessarilyfiguresinanysituationinwhichspeechoccurs.Johnofcourseisnotanecessarypartofanyspeechsituation,buttheutteranceofthefirstsentenceof(6)createsacontextofsituationinwhichhedoesfigure,thuspermittingReferencetohimbymeansofhe.EarlierH-Hequate'referencetoanotheritemwithinthetext'andreference'toathingasidentifiedinthesurroundingtext'[3_9].Again,however,theseareverydifferentthings,ascanbeseenfromanexamplelike(9)EvenAcidEddie'smothercallshimthat.IntheiwendedinterpretationthatReferstoAcidEddie,'anotheritemwithinthetext',whereashimReferstoAcidEddie,'athingidentifiedinthesurroundir~gtext'.ForH-H,bothhimandthatin(9)willbeAnaphorictoAcidEddie,butinfactitisonlyhimthatisAnaphorichere.AcidEddieisnottheAntecedentofthatbutitsReferent:therelationshipisnotAnaphorabutwhatLyons(1977)calls'textualdeixis'.H-H'sobservationthat'beingpresentinthetextis,asitwere,aspecialcaseofbeingpresentinthesituation'isthustrue,butnotinthesenseinwhichtheymeantit"itholdsfortextualdeixis,butnotforAnaphora.(Sentence(9)istakenfromGrinder(1971),whoalsotreatsAcidEddieastheAntecedentofbothhimandthat.HeisthereforeledtoproposethattworeferentialindicesshouldbeassignedtoAcidEddie,oneidenticaltothatonhim,theothertotheindexonthat.Thisproposal,however,issurelymisconceived,fortheroleofreferentialindicesinthestandardtheoryofgenerativegrammarwastomarkcoreference,whereasitisclearthattherelationbetweenthatandAcidEddieisoneReference,notco-Reference.)NowthefollowingobjectionmightbemadetotheabovecriticismsofH-H.AlthoughtheyfrequentlyuseexpressionslikethoseIhavequoted,338Reviewarticlewhereareferenceitemissaidtorefertoanotheriteminthetext,theyexplain'referential'Anaphorabysayingthatwhatispresupposed(i.e.theAntecedent)isnotawordorword-sequencebutratherameaning(p.145).Itisthereforewrongtoimputetothemaviewwhichleadstotheabsurdityofananalysiswherethesecondsentenceof(6)isequivalentto,'WasthelinguisticexpressionJohnill?'.However,thisinterpretationdoesnothingtosave(5).'WasthemeaningofJohnill?'isnolessabsurdanequivalence,unlesswetakethemeaningofJohntobeJohn.Theobjectionstosucha'Referential'theoryofmeaningaretoowellknowntoneedrehearsing,andIamobviouslynotsuggestingthatH-Hsubscribetoit.Butwhatevertheymeanbysayingthatwhatispresupposedisameaning,itisclearthatwecannotmaintainof(6)both(a)thatthereferentofheisthepersonJohn,and(b)thatitsreferentis'withinthetext'.TheAntece-dentofhecannotbeJohn,forJohnisnon-linguistic;itsAntecedentcan....I.,kor..t.....,else.....L.._^^,,:....:_.-................J_,鈥�,.,,u~v~,~,~,,,,,vi~,ala,~~wttnutnngutnu~:structureunoerlylngJohn.Ineithercase,however,theAntecedentisdistinctfromtheReferent,andtherelationofhetoitsAntecedentisofaquitedifferentkindfromthatofhetoitsReferent.Theheof(6)andtheIof(8)haveitincommon,ofcourse,thatbothareReferential.NotallAnaphorsareReferential,however,sothatwecannotregardAnaphoraingeneralasaspecialcaseofReference.WhatiscrucialtoAnaphoraisthatthereisarelationbetweenAnaphorandAnte-cedent.ButthereisnothingequivalenttothisrelationinH-H's'exophora',andhencenomoregeneralconceptsubsumingcategoriesiiandiiiin(5).ItistruethattherearemanyexpressionsthatcanbeusedeitherAnaphori-鈥�1"11lcallyornon-Anaphonca~,),butitdoesnotfollowfromthisthatthereisanysubstancetotheclaimthatinbothcasestheyarebeingused'phorically'.82.DelimitationoftheAnaphorInsomecasestherearenon-trivialproblemsindecidingjustwhatisthedependentterminanAnaphoricrelation.Someoftheseinvolvethe6istinctionbetweensubstitutionandellipsis,whichIshallbetakingupinsection4;hereIwanttoconsidertwocasesfromtheareaof'referential'cohesionwhereIdisagreewithH-H'sanalysisinthisregard.Thefirstinvolvesthedefinitearticleanddemonstrativedeterminers.InanexamplelikeReviewarticle339(10)MrsSmithhasarrived.-I'mafraidIhaven'ttimetoseethewoman.H-Htaketheasthe'referenceitem',theAnaphor.Iwouldargue,rather,thatitiswholeNPthewomanthatishereAnaphorictoMrsSmith.Itistherelationshipofthewoman,notthe,toMrsSmithwhichparallelsthatofthepronominalAnaphortoitsAntecedent:in(6)theAnaphoricrelation-shipofhetoJohnpermitstheidentificationofJohnastheReferentofhe,andin(10)theAnaphoricrelationshipofthewomantoMrsSmithpermitstheidentificationofMrsSmithastheReferentofthewoman.AndindeedthisanalysisfitsinbetterwithH-H'sgeneralaccountofcohesioncitedearlier"itisnotthebutthewomanthat'cannotbeeffectivelydecodedexceptbyrecoursetoMrsSmith";andalsowiththewayinwhichtheydistinguishcomparativereferencefromothertypes,includingtheoneinvolvedhere:comparison'isbasednotonidentityofreferencebutonnon-identity'(p.313)-therelationofidentityofreferencein(10)mustclearlyholdbetweenMrsSmithandthewoman.(Moreover,inatheorywhich,aswehaveseen,equatestherelationofAnaphortoAntecedentwiththatofReferringexpressiontosituationalReferent,itissurelyveryunsatisfactorytospeakofMrsSmith,oritsmeaning,asthereferentofthe.)TheissueofwhethertheAnaphorin(10)istheorthewomanisnotamerelyterminologicalmatter,foritbearsontheanalysisofsuchexpres-sionsas(11)Thehouseonthecornerisforsa|e.whereH-H(andmanyothers,e.g.Quirketal.1972:155)treatthefirsttheascataphorictothemodifieronthecorner.IfwetaketheAnaphorin(10)tobethewomanratherthanthe,thenclearlytherewillbenoAna-phoraatallin(11)-assumingacontextinwhichtherehasbeennopreviousmentionofthehouseinquestion.Andthisseemstomeasitshouldbe.Considerthesetofexamples(11)-(14):(12)Thehouseisforsale.(13)Iftheytarehereontime,bringyourparents1aswell.(14)They'rehere.spokenincontextswheretherehasbeennopreviousmentionofthehouseorpeopleinquestion-thus(14)mightbeutteredasenehearspeople340Reviewarticlearrivingonthedoorstep.(13)isaparadigmexampleofbackwardsAnaphora.Butthedifferencebetween(14)and(13)isquiteunlikethatbetween(12)and(11).Infactthereisnodifferenceatallbetween(12)and(11)asfarasthedefinitearticleisconcerned.Ineachcaseitintroducesadefinitedescription,onewhich(assumingtheutteranceachievesthespeaker'scommunicativeintentions)containsenoughinformationabouttheReferentfortheaddresseetobeabletoidentifyitinthecontextofdiscourse:whetherthedescriptiveinformationisgivenintheformofasinglenounorofanounplusmodifier(s)makesnodifference,foritisthefullnounphrasesthehouseandthehouseonthecornerthatRefertothesituationalentity,nottheorhouseorwhatever.OnthecornerisnomoretheAntecedentofthein(11)thanisMrsSmithin(10).掳SimilarconsiderationsapplytothesecondcaseIwanttodealwithhere,thatofcomparativereference,exemplifiedin(15)Maxisintelligent.ButTomiswittier.(16)TomiswittierthanMax.ForH-H,wittierisanaphorictoMaxin(15)andcat~phorictoMaxin(I6).Obviouslythereissomekindofcohesivetiebetweenthetwosentencesof(15),andthussomeresemblancebetween(15)andtheparadigmexampleofanaphoragivenin(6)above.ButIdonotbelievethatitexplainsthisresemblanceinanywaytosaythatwittierreferstoMaxashereferstoJohn.Identifyingwittierasthereferenceitem,astheAnaphor,isagaininconsistentwithH-H'sgeneralaccountofreference.Asnotedabove,theysaythat"comparisondiffersfromotherformsofreferenceinthatitisbasednotonidentityofreferencebutonnon-identity:thereferenceitemisinterpreted,notbybeingidentifiedwithwhatitpresup-poses,butbybeingcomparedwithit"(p.313).ButclearlythecomparisonJn(15)isnotbetweenwittierandMax(orbetweentheirmeanings),butbetweenTomandMax,orratherbetweenhowwittyTomisandhowwittyMaxis.Itwillnotdotodistinguish(15)from(6)bysayingthatwhereasin(6)heandJohnareidenticalinreference,in(15)wittierandMaxarenon-identicalinreference;wittierissimplynotaReferringexpression.WittierisnotitselfanAnaphor;ifthatnotionistobeappliedin(15),theAnaphorwillbewithinthestandardofcomparisonthatmustbep~'esentatsomeabstractlevelintherepresentationofthesecondsentence-theconstructionisonethatcallsforamoreabstractconceptionofAnaphorathanthatwhereAnaphorandAntecedentareactuallyidentifiableintheReviewarticle.341surfacestructure.Asforexample(16),ifitisagreedthatwittierisnotanAnaphor,thenitwillagainbeclearthatthereisnobackwardsAnaphora(cataphora)atallhere,justasthereisnonein(11).1掳3.ReferenceandsubstitutionH-Hdistinguishfourtypesof'phoric'relations"reference,substitution,ellipsisandlexicalcohesion.Theterm'substitution'~susedattwolevelsofgenerality,onewhereitisopposedtoellipsisandanotherwhereellipsisisaspecialcaseofit-thecasewherethesubstituteiszero.Itisinthelatter,moregeneralsensethatIshallbeconcernedwithitinthissection;insection41shalltakeupthedistinctionbetweenellipsisandthenarrowertypeofsubstitution,andinsection51shallreturnbrieflytothefinaltypeofphoricrelation,iexicalcohesion.Thedifferencesbetweenreferenceandsubstitution,accordingtoH-H'saccount,aresetoutinthefollowingtable(slightlyadaptedfromtheonetheygiveonp.145):(17)referencesubstitution(a)Levelofabstraction(b)Whatispresupposed?(c)Isclassp~eserved?(d)Isreplacementpossible?(e)Primarysourceofpresuppositionsemanticmeaningsnotnecessarilynotnecessarilylexicogrammaticalitems(wordsetc.)yesyessituationtextPoints(a)-(d)areclose.!yrelated.AreferentialAnaphorisAnaphorictoameaningratherthantoanitem(orform),andbecausetherelationshipholdsatthissemanticlevelthereisnorequirementthattheAntecedentshouldbeofthesamesyntacticcategory(orhavethesamesyntacticfunction)astheAnaphor,andhencenorequirementthattheAnaphorshouldbereplaceablebyacopyoftheAntecedent.Point(e)isthatwithreference,exophoraismorebasicthanendophora,andviceversaforsub-sfitution;althoughIwouldnotdenytheinterestofthequestionofwhetherAnaphorieusesoftheitemsinquestionaremoreorlessbasicthannon-Anaphoricones,IthinkthatH-H'sdiscussionistoobriefandimpressionis-tictocontributetoanexplicationofthedistinctionbetweenthetwotypes342ReviewarticleofAnal'hera.Itisclearthat(a)-(d)areintendedtoprovidethemainbasisfor*hea~sti~ction,anditisjustthesepointsthatIshallcons{derinthisanalysis.H-Hillustrate(c)and(d)forreferencewithexamplesthatincludethefollowing(pp.54--55):(18)John'shouseisbeautiful.Hehaditbuiltlastyear.(19)Johnhasmovedtoanewhouse.Hiswifemustbedelighted.In(18)theAnaphorheisapronounfunctioningasheadofanominalgroupwhiletheAntecedentisapossessivedeterminerfunctioningas'deictic',whereasin(19)hisisapossessivedeterminerfunctioningasdeicticwhiletheAntecedentisanounfunctioningashead;inr~eithercasecanwereplacetheAnaphor(he,his)bytheAntecedent(John's,John).Theseexamplesareunconvincing:theyde~endonapurely'surface'identificationofAnaphorandAntecedent.ifwesaythattheAntecedentin(18)isJohn,notJohn's,andtheAnaphorin(19)isnothisbutthiswiththepossessivecomponentabstractedaway(i.e.thelexemeHEasopposedtoaparticularinflectionalformofthelexeme),replacementispossible.Moreover,theinflectionaldifferencesbetweenAntecedentandAnaphorexhibitedbytheaboveexamplesarisealsowiththesubstitutivetypeofAnaphora:(20)Youmusttel!her.-I've~lreaaydoneso.wherewhatcouldbereplacedb3acopyoftheAntecedenttellherisnotdoneso,butthiswiththeen-inflectionabstractedaway.itIwouldcharacterizethedifferencewithrespecttoreplaceabilityinsomewhatdifferentterms.Considersuchexamplesas(21)Johnwasdrunkagainlastnight.Thatsortofbehaviourwilldohimnogood.(22)ThebossisinItalyatpresent.ShewillbebacknextTue;day.In(21),ifwereplacetheAnaphorthatsortofbehaviourbyaninflectionallyadjustedformofitsAntecedent,i.e.byJohn'sbeingdrunkagainlastnight,wechangethem~ning,becausetheAnaphorencodeselementsofmeaningnotexpressedintheAntecedent.Andthesameholdsfor(22):replacingshebythebosschangesthemeaningbecausethesemanticReviewarticle343element'Female'islost.TheseexamplesshowthatthereisnofundamentaldifferenceinkindbetweenpersonalpronounAnaphorsand(non-pro-nominal)definiteNPAnaphors:bothcanencedeinformationthatisnotexpressedintheirAntecedent.12H-H'ssubstitutiveAnaphorsdonotintroduceadditionalcomponentsofmeaninginthisway.ButitisnotpossibletoacceptH-H'sviewthatwhatispresupposedbysubstitutiveAnaphorsisthe'wording'(asopposedtothe'meaning')oftheAntecedent,th