[PDF] Morphology

78790_7Katamba_ch1_3.pdf

Morphology

FrancisKatamba

1Introduction

1.1THEEMERGENCEOFMORPHOLOGY

Althoughstudentsoflanguagehavealwaysbeenawareoftheimportance ofwords,morphology, thestudyoftheinternalstructureofwordsdidnot emergeasadistinctsub-branch oflinguisticsuntil thenineteenthcentury. Earlyinthenineteenthcentury,morphologyplayedapivotalroleinthe reconstruction ofIndo-European.In1816,FranzBopppublishedthe results ofastudysupportingtheclaim,originallymadebySirWilliam

Jonesin1786,

thatSanskrit,Latin,PersianandtheGermaniclanguages weredescendedfromacommonancestor.Bopp'sevidencewasbasedona comparisonofthegrammaticalendingsofwordsintheselanguages.

Between1819

and1837,Bopp'scontemporaryJacobGrimmpublished hisclassicwork,

DeutscheGrarnrnatik.Bymakingathoroughanalytical

comparison ofsoundsystemsandword-formationpatterns,Grimm showed theevolutionofthegrammarofGermaniclanguagesandthe relationshipsofGermanictootherIndo-Europeanlanguages. Later,undertheinfluenceoftheDarwiniantheoryofevolution,the philologistMaxMullercontended,inhisOxfordlectures of1899,thatthe study oftheevolutionofwordswouldilluminatetheevolutionoflanguage justasinbiologymorphology,thestudy oftheformsoforganisms,had thrownlightontheevolutionofspecies.Hisspecificclaimwasthatthe study ofthe400-500basicrootsoftheIndo-Europeanancestorofmanyof thelanguagesofEurope andAsiawasthekeytounderstandingtheorigin ofhumanlanguage (d.Mliller,1899;citedbyMatthews,1974). Suchevolutionarypretensionswereabandonedveryearlyoninthe history ofmorphology.Inthis.centurymorphologyhasbeenregardedasan essentiallysynchronicdiscipline, thatistosay,adisciplinefocusing onthe studyofword-structureatonestageinthelifeofalanguageratherthanon theevolution ofwords.But,inspiteoftheunanimousagreementamong linguists onthispoint,morphologyhashadachequeredcareerin twentieth-centurylinguistics,asweshallsee.

1.2MORPHOLOGY

INAMERICANSTRUCTURAL

LINGUISTICS

AdherentstoAmericanstructurallinguistics,oneofthedominantschools oflinguisticsinthefirstpartofthiscentury,typicallyviewedlinguisticsnot somuchasa'theory' ofthenatureoflanguagebutratherasabodyof 3

4IntroductionTheConcept

ofChomskyanGenerativeGrammar 5 Thebulkofthisbook,however,presentsmorphologicaltheorywithinthe linguisticmodel ofgenerativegrammarinitiatedbyChomsky.Beforewe beginconsideringhowthistheoryworks,Iwillsketchthebackground assumptions madebygenerative grammarianssothatwecanplacethe theoryofmorphologyinthewidertheoreticalcontextofgenerative linguistics. Thecentralobjectiveofgenerativelinguisticsistounderstandthenature oflinguisticknowledgeandhowitisacquiredbyinfants.Inthelightofthis objective,afundamentalquestionthatatheory ofword-structuremust addressis, 'whatkindsofinformationmustspeakershaveaboutthewords oftheirlanguageinordertousetheminutterances?'Attemptstoanswer thisquestionhaveledto thedevelopmentofsub-theoriesofthelexicon (i.e.dictionary)and ofmorphology..., AccordingtoChomsky(1980,1981,1986),thecentralgoalofImgmstlc theoryis todeterminewhatitispeopleknowiftheyknowaparticular language.Chomskyobserves thatknowingalanguageisnotsimplya matterofbeingabletomanipulatealonglistofsentencesthathavebeen memorised.Rather,knowingalanguageinvolveshavingtheabilityto produceandunderstandavast(andindeedunlimited)numberofutter ances ofthatlanguagethatonemayneverhaveheardorproducedbefore. In otherwords,creativity(alsocalledproductivityoropen-endedness)is anaspectoflinguisticknowledgethatisofparamountimportance.

Linguisticcreativity

isforthemostpartrule-governed.Forin.stance, speakers ofEnglishknowthatitispossibletoindicatethatthereISm?re thanoneentityreferredtobyanounandthatthestandardwayofdomg thisis toadd-sattheendofanoun.Giventhenounbook,whichweall have encounteredbefore,weknowthatifthereismorethanoneofthese objectswerefer tothemasLikewise,giventhenonsenseword smiltsasinthesentenceThesmiltsstinkwhichIhavejustmadeup,you know smiltswouldrefertomorethanoneofthesesmellythings.Speakers

1.3THECONCEPTOFCHOMSKYANGENERATIVE

GRAMMAR

theAmericanstructuralistsshowedthatwordsareanalysableintermsof morphemes.Thesearethesmallestunitsofmeaningandgrammatical function.Previously,word-structure hadbeentreatedtogetherwith sentence-structure undergrammar.Thestructuralistsintroducedmor phologyasa separatesub-branchoflinguistics.Itspurposewas'thestudy ofmorphemesandtheirarrangementsinformingwords'(Nida,1949:1). Thecontributionofthestructuralistsinformsmuchofthediscussioninthe firstpartofthisbook.dealswithmeaning dealswithsoundsystemsdealswithsentence-structure dealswithword-structureSemanticlevel: r

Syntacticlevel:

r

Morphologicallevel:

r

Phonology(orphonemics):

Thelevelswereassumedtobeorderedinahierarchy,withphonologyat thebottomandsemanticsatthetop.Thetaskoftheanalystproducinga description ofalanguagewasseenasoneofworkingout,inseparate stages,first thepronunciation,thentheword-structure,thenthesentence structureandfinally themeaningofutterances.Itwasconsideredtheoreti callyreprehensibleto makeuseofinformationfromahigherlevel,e.g. syntax,whenanalysingalowerlevelsuchasphonology.Thiswasthe doctrine ofseparationoflevels. Intheearlydays,especiallybetween1920and1945,Americanstructura listsgrappledwiththe problemofhowsoundsareusedtodistinguish meaninginlanguage.

Theydevelopedandrefinedthetheoryofthe

phoneme (cf.Sapir,1925;Swadesh,1934;Twaddell,1935;Harris,1944). Astimewenton,thefocusgraduallyshiftedtomorphology.When structuralismwasinitsprime,especiallybetween1940and1960,thestudy ofmorphologyoccupiedcentrestage.Manymajorstructuralistsinvesti gatedissuesinthetheory ofword-structure(cf.Bloomfield,1933;Harris,

1942, 1946,1951;

Hockett,1952,1954,1958).Nida'scoursebookentitled

Morphology,whichwaspublishedin1949,codifiedstructuralisttheoryand practice.Itintroducedgenerationsoflinguiststothedescriptiveanalysisof words. Thestructuralists'methodologicalinsistenceontheseparationoflevels whichwenotedabovewasamistake,asweshallseebelowinsections(1.3.2) and(1.3.3).Butdespitethisflaw,therewasmuchthatwascommendable inthestructuralist approachtomorphology.Oneofthestructuralists'main contributionswas therecognitionofthefactthatwordsmay haveintricate internalstructures.Whereastraditionallylinguisticanalysishad treated

thewordasthebasicunitofgrammaticaltheoryandlexicography,descriptiveandanalyticalprocedures.Ideally,linguisticanalysiswas

expectedto proceedbyfocusingselectivelyononedimensionoflanguage structureatatime beforetacklingthenextone.Eachdimensionwas formally referredtoasalinguisticlevel.Thevariouslevelsareshownin [1.1]. [1.1] 6 IntroductionTheConceptofChomskyanGenerativeGrammar7 ofEnglishhavetacitknowledgeoftherulewhichsays'add-sfor plural' andtheycanuseittoproducethepluralformofvirtuallyanynoun.Ihave emphasisedthenotion ofrule,takingtheexistenceofrulesforgranted.

Iwillnowexplainwhyagenerative

grammarisasystemofexplicitrules whichmayapplyrecursively togenerateanindefinitenumberofsentences whichcan beaslongasonewantsthemtobe.Recursivenesshasthe consequencethat,inprinciple, thereisnoupperlimittothelengthof sentences.Agrammaticalconstituentlikeanounphrase(NP)orapre positionalphrase(PP)cancontainanindefinitenumber offurtherconstitu ents ofthatcategoryasinthesentenceJohnsawthepictureofthebabyon thetableintheattic.Therecursioncanbeseenclearlyinthetreediagram representing thatsentencein[1.2].Asseen,NPscancontainNPsandPPs whichinturncontainNPswhichcancontainNPsandPPs: [1.2] s NPVP N VNP

JohnsawNPPP

~

DETN PNP

I I I thepictureofNPPP I I

DETN PNP

I I I thebabyonNPPP I I

DETN PNP

I I thetableDETN I I intheattic

Notes:S -sentence;N -noun,

NP-nounphrase;V -verb,VP-verb

phrase;P -preposition,

PP-prepositionalphrase;DET-determiner.

Oneofourconcernswillbetodeterminewhethermorphologyshouldbe recognisedasaseparatelinguisticlevel(ormodule)thatisindependentof syntaxandphonology(see[1.1]aboveand[1.3]below).Domorphological ruleshavecertainpropertieswhichthey donotsharewithrulesinother partsofthegrammar?Arerecursiverulesofthekindfoundinsyntax neededinmorphology?Thisbookwilladdresstheseissuesindepth.Here IwillonlyattempttogiveyouaflavourofoneoftheissuesthatIwillbe exploring. There aremorphologicalprocesseswhicharesimilartosyntacticpro cesses. Forinstance,certainadjectiveswhichdescribeperiodsinhistory, suchas industrial,canhavetheprefixpost-beforethemasinpost industrial. And,giventheadjectivepost-industrial,wecanplaceanother post-beforeittoyieldpost-past-industrial.Clearly,theword-formation processwewitnesshereisrecursive.

Wehavetheruleattachingpost-toa

wordreapplyingtoitsownoutput.Thisraises aninterestingquestion:if morphologicalrules thatbuildwordsaresimilartosyntacticrulesthat buildsentences,whatreasonisthereforassumingthatmorphologyis essentiallydifferentfromsyntax?

Beforewegoanyfurtherwe

needtoclarifythetermsgrammarandruleof grammar.Thesetermsareusedbylinguistsinfourdistinctsenses.FIrstly,III -generativelinguistics'grammar' latedknowledgeofrulesandpDnJ::iples-Qf-th@idangu_ag<:-thaLpeople-havein tl1elrheads.Thistacitknowledgeenablesthemtodistinguishbetweenwell foITi:Ieoandlll-formedwordsandutterancesintheirlanguage.Forexample, manyEnglishspeakersmay notbeabletoexplaininanarticulatemanner whyitis'correct'tosayagrainbut'incorrect'tosayaoat.Neverthelesstheir knowledge ofEnglishgrammaticalstructureenablesthemtodeterminethat theformer iscorrectandthelatterisnot. Secondly,whereasintraditionalapproaches'grammar'onlyincludes morphology andsyntax,ingenerativelinguisticsthetermgrammaris employedinamuchwidersense.Itcoversnotonlymorphologyandsyntax butalsosemantics,thelexiconandphonology.Hence,therearerulesof grammarin everylinguisticmodule.Phonologicalrules,morpholOgical rn:tes,syntacticil:iIeSallosemaIJ.ticrulesareallregardedasrulesof grammar.

Thirdly,grammarandrules

ofgrammarmayrefertoabookcontainingaf\ ofrules_t~!i~E~IJ:i_lldthe}) behavIOuroflangu~~e.Theserules simplydescriberegularpatternsobserved ill1l1e-liiiguisticdata.

Lastly,somegrammars

arebookscontainingprescriptivestatements. Such gralllllJarscontainrulesthatprescribecertainkindsofusage.Outside linguisticsthisview ofgralllllJarisstillprevalent.Thereasonforthisis clear. Ineverydayliferulesarenormallymechanismsforregulatingbehav iour- thebehaviourofpupilsIna:scnool;-membetsofa.club,inmatesofa prison,etc. Inmanytraditionalpedagogicalgrammarsrulesservethesame purpose.They arestatementslike'Asentencesmustnotendwitha preposition.'Theyprescribewhat the'officiallyorsociallyapproved'usage is-intheopinionofthegrammarian. Inmuchofmodernlinguistics,however,ruleshaveadifferentfunction. 8

IntroductionTheConcept

ofChomskyanGenerativeGrammar9 (iI / ) Theyarenotprescriptionsofbehaviourwhichthegrammarianimposeson speakers,butrathertheyarestatementsofprinciplesresponsibleforthe observedregularitiesin thespeechorwritingorusers'-ofa-nartlculir I

~----------------~-------_._------_._---~--------------.--£'--------

1neclia:acterisationofregularitiesinobservedpatternsofusage

ISwhattheAmencanstructuralistsregardedastheprimaryobjectiveof linguisticinvestigations,Theirgrammaticalrulesweredescriptivestate mentslike 'ThearticleprecedesthenounintheEnglishnounphrase.'This statementreflectsthefactthatthebook,asinIreadthebook,isallowed *b~okthe,asin*1readbooktheisdisallowed.(Anasteriskmdlcatesadisallowedform.)

Chomskyhasshifted

thefocusoflinguistictheoryfromthestudyof observedbehaviourtotheinvestigationoftheknowledgethatunderlies thatgenerativelinguisticsrulesareintendedtogobeyond forpatternsinthedatatoacharacterisationofspeakers'

ImgUlstlcknowledge.

Theprimaryobjectiveofgenerativegrammaristo

modelaspeaker'slinguisticknowledge.

Chomskycharacteriseslinguisticknowledgeusing

theconceptsofcom petenceandperformance.Competence isaperson'simpliciLknQwiedge-of theproductionandullcler:stan.dingofan mdefimtelylarge nllmberofnewutterancespossiblewhileperformanceis theactualuseoflanguageinreal.sittiations:-Chomskyproposesthat :athertlian-p'erformance,istheprimaryobjectoflinguistic mqUiry. PutSimply,knowledgeofalanguageentailsmasteryofanelabor. atesystemofrulesthatenablesapersontoencodeanddecodealimitless numberofutterancesinthatlanguage.Onesub-setofthisrulesystemis therulesofword-formationwhichthisbookintroducesyouto.Insection (4.1.3) ofChapters4andsection(12.3.3)ofChapter12itwillbeshown thatspeakersofalanguagedonotjustcommittomemoryallthewords theyknow. Theircompetenceincludestheabilitytomanipulaterulesin ordertocreatenewwordsandtounscramblethemeaningsofnovelor unfamiliarwordswhichtheyencounter. Ifknowingalanguageessentiallyinvolvesmasteringasystemofrules, howdohumansaccomplishthistask?Chomskycontends thatthelinguistic ~fhumansThegeneralcharacteroflinguisticknowledge determmedbythenatureofthemindwhichisendowedwithaspecia hsed faculty.Thisfacultyisdeterminedintumbythebiologyof thebram.Thehumanchildisbornwithablue-printoflanguagewhichis calledUniversalGrammar. AccordingtoChomsky,UniversalGrammaristhefacultyofthemind whichdetermines thenatureoflanguageacquisitionintheinfantandof linguisticcompetence.Thepropertiesthatliebehindthecompetenceof speakersofvariouslanguagesaregovernedbyrestrictedandunified elementaryprinciplesrooted inUniversalGrammar.Thisexplainsthe

strikingunderlyingsimilaritybetweenlanguagesintheiressentialstruc-turalproperties.Admittedly,languagesdifferfromeach

other,butthe structuraldifferencesbetweenthemoccurwithinthefairlynarrowrange sanctionedbyUniversal

Grammar.Asweshallsee(especiallyin

Chapters3,8,9

and12)withregardtoword-fonnation,verysimilar word-buildingprinciplesrecurinlanguageafterlanguage.

Thelanguage

faculty ofthemindisessentiallythesameinallhumans.Hence.lan guagescanonlydifferfromeach otherwithinthe~redeterml~ed bytheneurologyandphysiologyofthehumanbram,whichdeterrmne the natureofUniversalGrammar.AndUniversalGrammarintum detenninesthekindsofgrammarsofparticularlanguagesthatcanbe acquiredbyinfants. Thedifferencesbetweenthegrammarsacquiredbyindividualspeakers of,say,EnglishandArabiccan beattributedtoexperience.An.indi vidual'sexperienceserves tospecifyaparticulargrammarforthepartu:ular languagewhich thatindividualisexposedto-withintherangepennltted byUniversalGrammar..' HowisUniversalGrammarstructured?Itismodularmstructure:It consists ofvarioussub-systemsofprinciples.Manyofitsprinciplesconsist ofparameterswhicharefixedbyexperienceonthebasisofevi dence ofthekindavailabletothechild.ChomskycomparesUmversal Grammartoanintricateelectricalsystemthatisallwiredup,butnot switchedon.Thesystemcontainsafinitesetofswitches,eachoneofwhich hasarestricted numberofpositions.Exposuretoaspecificlanguage experienceis requiredtotumontheseswitchesandgivethemtheappro priatesetting. Thebasicideaofparametersismeanttocapturethefactthatmanyrules areinterdependent.

Ifonechoiceismade,itmayeitherprecludesome

otherchoicesorsetinmotionotherrelatedchoices.Thismakesthetaskof languageacquisitionsimplerthanitwouldbeifeachrulehadtobeworked outindependentlyofallotherrules.Theparametricapproachassumesthat theinfantacquiringalanguagemakesverycleverguessesorhypotheses abouttherulesofthegrammarbeingacquiredonthebasisofrulesalready acquiredafterexperience ofaparticularlanguage. Foraconcreteexampleofaparameter,wewillconsidertheRight-hand

HeadlLeft-hand

HeadRulewhichwillbediscussedinChapter12.This

parameterisconcernedwiththepositionoftheheadofagrammatical constituent.Somelanguages,likeEnglish,normallyplacethe headonthe right,i.e.itisthelastelementofaconstituent.Forexample,inthenoun phrasethesebigbookstheright-handmostword,thenounbooks,ISthe head.Itmustcomelast.(Alternativeslike*booksbigtheseand*these booksbig areforbidden.) Asarule,theheadistheonlyobligatoryelementofaconstituentlikean NP. Booksisawell-formedNPbutneitherthesenorbigisapermissibleNP onitsown.Furthermore,intermsofmeaning,theheadbooksisthekey 10

IntroductionTheConcept

ofChomskyanGenerativeGrammar11 wordinthisNP.Thefunctionoftheseandbigismerelytospecifyfurther theparticularbooksreferredto.

Likewise,

atwordlevel,inacompoundlikefarmhouse,thehead,house, isthelast elementanditisthepivotalelementfromasemanticpointof view.(Afarmhouseisakindofhouse.)However,insomelanguages,such as Japanese,thereverseisthecase.Theheadofagrammaticalconstituent isnormallyontheleft.Onceaninfanthasworkedoutthepositionofthe headforoneconstructionthiscanbegeneralisedwithaconsiderable degree ofsuccesstootherconstructions.

Universal

Grammarconsistsofanumberofmoduleswhichareinter

related.This isshownin[1.3](whichyoushouldcomparewith[1.1]above): [1.3](i)LexiconandMorphology (ii)Syntax (iii)Phonetic

Form(PF)(whichdealswithrepresentationof

utterancesinspeech) (iv)Logical

Form(LF)(whichdealswithmeaning)

Asseen,UniversalGrammarincludesthelexiconandmorphology module.Knowledge ofword-structureisacentralaspectoflinguistic competence.Acasecan bemadeforrecognisingmorphologyasaseparate moduleofUniversalGrammar.Yetatthesametime,morphology(and thelexicon) arelikeabridgethatlinkstheothermodulesofthegrammar. Itis.thereforenecessarytoexaminemorphologynotinisolation,butin relatIOn totheothermodules.Morphologyinteractswithbothphonology andsyntaxaswellassemantics.So,itcanonly bestudiedbyconsidering thephonological,syntactic andsemanticdimensionsofwords.made importantcontributionstomorphologyduringthistime,asweshall see. Partofthereasonforthewidespreadneglectofmorphologyduringthe earlyyears ofgenerativegrammarwasthebeliefthatword-formation could beadequatelycoveredifitwaspartitionedbetweenphonologyand syntax. Itwasarguedthatnoseparatemorphologicallevelorcomponent was neededinthegrammar.Wayswerefoundofdescribingthestructure ofwordsinamodeloflanguagethathadaphonologicalcomponent,a syntactic componentandasemanticcomponentbutnomorphological component.Thoseaspects ofword-structurethatrelatetophonology(e.g. thealternation betweensane[sem]andsanity[sremtI]wouldbedealt withusingdevicesfoundin thephonologicalcomponent.Andthose aspects ofword-structurethatareaffectedbysyntaxwouldbedealtwithin thesyntacticcomponent. Thejobofthesyntacticcomponentofthegrammarwasthoughtofas beingtogenerate(i.e. tospecifyorenumerateexplicitly)allthewell-formed sentences ofalanguage,withoutgeneratinganyill-formedones.

Significantly,generating

allthesentencesofalanguagewasseenasmeaning generatingall thepermissiblesequencesofmorphemes(notwords),and showingwhich morphemegroupingsformedsyntacticconstituentslike nounphrases andverbphrases(alsoseep.13inthischapter).Aspecialised morphologicalcomponentandaproperlyarticulatedlexiconwere notpart ofthepicture.Thus,Lees(1960),thefirstmajordescriptivestudyproduced byagenerativelinguist,usedsyntacticrulestocreatederivedwordslikethe nounappointmentfromtheverbappoint.Asseenin[1.4a],Leesderivedthe sentencecontainingthenounappointmentfromasourcesentencewiththe verbappoint.Likewise,hederivedtheabstractnounpriesthoodfroma sourcesentencewith thenounpriest,asindicatedin[lAb]. [1.4]a.ThecommitteeappointsJohn.

Thecommittee'sappointmentofJohn.

(Sourcesentence:Lees,1960:67) WewillnotexaminetheparticularsofthesyntacticruleswhichLeesuses. OurconcernisthatLeessawthistypeofword-formationastakingplacein thesyntaxandbelievedthathecoulddispensewithmorphology.Wewill revisitthisissuein

Chapter12.

Letusnowturnourattentiontoquestionsofphonologicalrealisation.

Readjustment

rules(whichweremorphologicalrulesindisguise)playeda keyroleinthisarea.They operatedonthefinaloutputofthesyntactic component,makingwhatevermodificationswerenecessaryin orderto

1.3.1ThePlaceofMorphologyinEarlyGenerativeGrammar

Todaytheplaceofmorphologyingenerativegrammarissecure.Butthisis arecentdevelopment.Afterbeinginthelimelightwhenstructuralism peakedinthe1950s,morphologywasatfirsteclipsedwhengenerative grammarcame onthescene.Generativegrammariansinitiallyrejectedthe validityofaseparatemorphologicalmodule. Fromthepointofviewofadvancingourunderstandingofword structure,thisstancewasunfortunate.Sincegenerative grammarhasbeen thedominantschooloflinguisticsinthesecondhalfofthiscentury,it meantthatthestudyofword-structurewasintheshadowsformorethana decade.Morphologydidnotre-emergefromoblivionuntil themid-1970s.

Fortunately,theeclipsewas

nottotal.Afewisolated(forthemostpart non-generative)scholarssuchasRobins(1959)andMatthews(1972,1974)b.

Johnisapriest.

John'spriesthood.(Sourcesentence:Lees,1960:110)

12IntroductionTheConcept

ofChomskyanGenerativeGrammar 13 enablephonologicalrulestoapplytotherepresentationobtainedafterall syntacticruleshadapplied.

Unfortunately,

thereseemstohavebeennoconstraintonthepowerof readjustmentrules.Forinstance,inSPE(TheSoundPatternofEnglish) whichappearedin1968andwasthepivotalworkinthedevelopmentof generativephonologicaltheory,ChomskyandHalleproposed(onp.11) that thesyntaxshouldgenerateboththeregularpasttenseformmended [vf."mend}"past},.andtheirregularpasttenseformsang[v["sing}"past}". Thesebracketedstrings,whichweretheoutputofthesyntacticcomponent, wouldformtheinputtothereadjustmentrules.Next,thereadjustmentrules wouldremoveall thebracketsassociatedwiththepasttense.Inthecaseof mend,ageneralreadjustmentrulewouldreplacepastbyd,whileinthecase ofsingaspecialreadjustmentrulewoulddeletetheitempast,togetherwith theassociatedbracketlabels,giving[vsing}".Thesamereadjustmentrule wouldalsoattachthediacritic mark*tothevowelIIIindicatingthat eventuallyaphonologicalrulewouldchangeitintolre/.Thereadjustment ruleswouldgivetheforms [vf,.mend}vd}"and["s*ng}".Theserepresen tations-andall othersuchrepresentationsyieldedbyreadjustmentrules werereferredtoasphonologicalrepresentations.Finally,phonological representationswould beconvertedintothephoneticrepresentations [mendldJand [srelJJbyrulesinthephonologymodule.

Withthebenefit

ofhindsight,wecanseethatreadjustmentruleswerea mistake.

Theywereruleswithunbridledpower.Theycouldmakewhat

evermodificationswere deemednecessarytoenablephonologicalrulesto applytostrings ofmorphemesproducedbythesyntax.Itisveryundesir abletohaveabatch ofrulesthatempoweruslinguiststodowhateverwe like,wheneverwelike,solongaswecomeupwith theanswerwelike.A theorybecomesvacuousifithasrules thatcaninsertallmannerof elements,removeallmannerofelementsandmakeallmannerofelements exchangeplaceswheneverwechooseto,with noprinciplesrestrictingour freedom.Effectively,thismeansthatwearegivencarteblanchetostartoff withanyarbitraryinput,applytherules,andcomeupwith the'correct' answer. readjustmentruleswereabadideabecausetheyare eVIdence ofalackofinterestinwordsquawordsandinmorphologyasa level.Usingrulesofthesyntaxtospecifypermissiblesequences ofmorphemes,regardlessofwhethertheyoccurredinwordsorsentences, an?usingreadjustmentrulestotumstringsgeneratedbythesyntaxinto stnngsthatthephonologycouldprocessandassignapronunciationtowas merelyskirting roundtheproblem.Wordsareacentraldimensionof language.Theyhavecertainuniquepropertiesthattheydonotsharewith otherelementsoflinguisticstructurelikesentencesandspeechsounds.A theoryoflanguagemustincludeaproperlydevelopedmodelofword formation thatenablesthelinguisttodescribewordsontheirownterms-withoutoverlooking thewaysinwhichword-formationrulesinteract rulesinothermodules.Astimewentby,thisbecamecleartogeneratIve linguistswho,inincreasingnumbers,began toexploremoresatisfactory ways ofdealingwithword-structure.

1.3.2TheMorphology-PhonologyInteraction

Asregards

theinteractionwithphonology,theselectionoftheform manifestsagivenmorphememaybeinfluencedbythesoundsthatrealise neighbouringmorphemes.

TaketheindefinitearticleinEnglish.Ithastwo

manifestations.

Itisabeforeawordthatbeginswithaconsonant(e.g.,a

pear)andanbeforeawordthatbeginswitha(e.~.,anWe cannotdescribethephonologicalshapeoftheIndefimteartIcleWIthout referring tothesoundatthebeginningofthewordthatfollowsit.

1.3.3.TheMorphology-SyntaxInteraction

Asregardstheinteractionwithsyntax,

theformawordrna!beaffected by thesyntacticconstructioninwhichthewordISused.ForInstance,the verbwalkhasanumberofformsincludingwalk,walksandwalked.The selectionofaparticularformofthisverbonagivenisdependent on thesyntacticconstructioninwhichitappears.Thus,Inthepresent tense thechoicebetweentheformswalksandwalkdependsonwhether theoftheverbisthirdpersonsingular(inwhichcasewalks.is selectedasishe/shelitwalks)ornot(inwhichcasewalkisselectedasIII Ilyoulweitheywalk).Inthepasttense,walkisrealisedas. Chomsky(1957:39)dealswithallthesefactsasuncontroverslalsyntactIC phenomena,usingthephrasestructurerulebelow: [1.5]Sinthecont6!':tNPsing_

C0inthecontextNPpl-

past Notes:(i)standsfor'expand'or'rewriteas'.(ii).Cstandsfor -----variousverbalsuffixesthatmayberealisedas-s(asIIIwalks),0(I.e. zero)asin walkand-edasinwalked. Chomsky'sanalysisdoesnotseparatephrasestructurerules\e.? ~NP+VP;VPVerb+NP)whichenumeratepermISSIbleCOm?I nationsofwordsinphrasesandsentencesfromrulesofword-structurehke theonein[1.5]thatgiveswalksfromwalk.Alltheserulesarebanded togetherbecausetheyareconcernedwithenumeratingpermissiblecombi nations ofmorphemes(seeabove).

Note,however,

thatthistreatmentofsyntacticallymotivatedalternation intheform ofwordsiscontroversial.Wehavemerelyairedtheproblem forthepresent.WewillpostponedetaileddiscussionuntilChapter10. Turningtosemantics,theconnectionbetweenmorphologyandthe lexiconontheonehandwithmeaningontheotherisobvioussinceamajor roleofthelexiconordictionaryistolistthemeaningsofwords.Thisis becausenormally therelationshipbetweenawordanditsmeaningis arbitrary.Thereisnoreasonwhyawordhastheparticularmeaningthatit has. Forinstance,youjusthavetomemorisethefactthatthewordfaille referstoakindofhead-dresswornintheseventeenthcentury.Thereisno way thatyoucoulddiscoverthisfactfromthesoundsorthestructureofthe word.

Wewillcomebacktothistopicinsection(12.3.2).

Itislessimmediatelyobviousthat,inadditiontoindicatingthemeaning ofwordsandmorphemes,thelexiconmustalsostoreotherkindsof informationrelevanttotheapplicationofsyntacticandphonologicalrules.

Syntaxneedstohaveaccess

tomorphosyntacticproperties(i.e.properties thatarepartlymorphologicalandpartlysyntactic)suchaswhetheranoun iscountablelikespadesoruncountablelikeequipment.Thisaffectsits behaviourinphrases andsentences.Wemaysaythisspadeorthesespades butwecanonlysaythisequipment(not*theseequipments). Furthermore,somephonologicalrulesapplytowordsdifferentlydepen ding ontheirmorphosyntacticproperties.Forexample,somephonological rulesaresensitiveto thedifferencebetweennounsandverbs.Thus,inthe wordpermit,themainstress(shownherebyunderlining)fallsonthefirst syllableif thewordfunctionsasanoun(permit[noun])'Butifitfunctionsas averb (permitrverb])'mainstressfallsonthesecondsyllable.Obviously,for phonologicalrulesthatassignstresstoapplycorrectly,access tosuch morphosyntacticinformation isessential.Thisinformationmustformpart oftheentryofthewordinthelexicon. Thestudyofmorphology,therefore,cannotbeself-contained.The structuralistdoctrineoftherigidseparationoflinguisticlevelssketchedin (1.2)isuntenable.True,therearesomeissues thataretheinternal concerns ofmorphology.Butmanymorphologicalproblemsinvolvethe interactionbetweenmorphologyandothermodulesofthegrammar.For thisreason,muchofthespaceinthechaptersthatfollowisdevotedtothe interactionbetweenthelexiconandmorphologywiththemothermodules.

Thebookisorganisedasfollows:. . .

PartI(Chapters1-4)introducesbasicconceptsandtraditionalnotIOns which arefundamentaltoallmorphologicaldiscussions. PartIT(Chapters5-9)explorestherelationshipbetweenmorphology, phonology andthelexiconincurrentgenerative.. Partm(Chapters10-12)dealswiththerelatIOnshipbetweenmor- phology andsyntaxincurrentgenerativetheory. Overtheyears,therehavebeenseveralmorphologicaltheoriesthat havebeenproposedbylinguists.Onewayofyoutomor phologywould betopresentahistoricalandcomparatIvesurvey.Icould haveexaminedvarioustheoriesin turn,andperhapscomparedthem.Or, alternatively,Icouldhavebeenpolemicalandproselytising.Icouldha".e triedtopersuadeyouthatmypreferredtheoryisthebesttheory.ThatIS notwhatIshalldointhisbook.Instead,Ipresentyou,sympathetically butatthesametimecritically, with onetheoreticallycoherentapproachtomorphology,namely theoryofmorphologyincurrentmainstreamgenerativeThis decision issensiblenotonlybecausethisisthedominantmodelmthefield today, butalsobecauseIthinkitoffersthepromisingsolutionstothe perennialproblemsinmorphological Evensothebookisinevitablyselective.Ihavenotattemptedto represent:veryshadeofopinionwithinthegenerativeschool.RatherI havefocused onideasandpracticesthatseemtometoformpartofthe emerging'canon'inmainstreamgenerativemorphology.Obviously,to someextentthisisamatterofsubjectivejudgement.Insomecasesmy judgementmay notbethesameasthatofsomeother.linguists.. Ofcourse,morphologicaltheoryincurrentmamstreamgenerative grammardoes notenjoyamonopolyofinsight.Thedebtowedto.other approacheswillbeevident,especiallyintheearlychaptersandmthe bibliography..A majorfeatureofthebookisthatyouwillbeaskedtobeanactive investigator, notapassivereader.Ihavetoengage activelyandpracticallyindoingmorphologyrathermmerelylearnmg aboutitshistory andwatchingfromthestallshowItISdone.Asyouread eachchapter,youareaskedtopauseatplacesandanswerin-text andexercisesbeforeproceeding(thequestionsandexercisesareSignalled bylinesacross thepage).Eachchapter(afterthisone)ends:u rther exercisesdealingwithpointsraisedinthebodyofthetext.ThismSlstence ongettingyoutoanalysedataisduetomyco~vic~ionthat~e~t initiationforanyonewhowishestobecomeaImgUlstIStodolIngUIstic analysisrightfrom thestartratherthantoreadit. Inthetextnew morphologicaltermsappearmboldtypeandtheyare14

IntroductionOrganisation

oftheBook

1.4ORGANISATIONOFTHEBOOK

15 explainedwhentheyarefirstintroduced.(Theymayalsobeinboldtype theyappearsubsequentlyinacontextwheretheyneedtobehigh lIghted.) Keytermsfromotherbranchesoflinguisticsareexplainedina ~ttheend.ForanyotherlinguistictermsthatareunfamiIi ~lc~lOnaryoflinguistics,suchasDavidCrystal'sAFirstof

LIngUlstzcsandPhonetics(1980),shouldbeconsulted.Y

16

Introduction

2IntroductiontoWord-Structure

2.1WHATISAWORD?

Theassumptionthatlanguagescontainwordsistakenforgrantedbymost people.Evenilliteratespeakersknowthattherearewordsintheirlan guage. True,sometimestherearedifferencesofopinionastowhatunits are tobetreatedaswords.Forinstance,Englishspeakersmightnotagree whetherallrightisonewordortwoandasaresultdisputesmayariseasto whetheralrightisthecorrectwayofwritingallright.But,byandlarge, peoplecaneasilyrecogniseawordoftheirlanguagewhentheyseeorhear one.Andnormallytheirjudgementsastowhatisorisnotaworddo coincide.Englishspeakersagree,forexample,thattheformsplodyinthe sentenceThesplodycatsatonthematisnotanEnglishword-butallthe otherforms

2.1.1TheLexeme

However,closerexaminationofthenatureofthe'word'revealsasome whatmorecomplexpicturethanIhavepaintedabove.Whatwemeanby 'word'isnotalwaysclear.Asweshallseeinthenextfewparagraphs, difficultiesinclarifyingthenatureofthewordarelargelyduetothefact theterm'word'isusedinavarietyofsenseswhichusuallyarenotclearly distinguished.

Intakingtheexistenceofwordsforgranted,wetendto

overlookthecomplexityofwhatitiswearetakingforgranted. II Whatwouldyoudoifyouwerereadingabookandyouencounteredthe 'word'pockledforthefirsttimeinthiscontext? [2.1]

Hewenttothepubforapintandthenpockledoff.

I Youwouldprobablylookupthatunfamiliarwordinadictionary,not underpockled,butunderpock/e.Thisisbecauseyouknowthatpockledis notgoingtobelistedinthedictionary.Youalsoknow,thoughnobodyhas toldyou,thatthewordspocklingandpockleswillalsoexist.Furthermore, youknowthatpocklingandpockle,pocklesandpockledareallinasense different manifestationsofthe'same'abstractvocabularyitem. Weshallrefertothe'word'inthis senseofabstractvocabularyitem 17 Whichonesofthewordsin[2.2]belowbelongtothesamelexeme? [2.2]seecatchestallerboycatchingseessleepswomancatchsawtallestsleepingboyssleepseentall jumpedcaughtseeing jump womensleptJumpsjumpingusingthe termlexeme.Theformspockling,pockle,pocklesandpockled aredifferentrealisations(orrepresentationsormanifestations)ofthe lexemePOCKLE(lexemeswillbewrittenincapitalletters).Theyallshare acoremeaningalthoughthey arespelledandpronounceddifferently.

Lexemes

arethevocabularyitemsthatarelistedinthedictionary(cf.Di

SciulloandWilliams,1987).

19Morphemes:TheSmallestUnitsofMeaning

2.1.3TheGrammaticalWord

The'word'canalsobeseenasarepresentationofalexemethatis associatedwithcertainmorpho-syntacticproperties(i.e.partlymorpho logicalandpartlysyntacticproperties)suchasnoun,adjective,verb,tense, gender,number,etc.

Weshallusethetermgrammaticalwordtoreferto

the'word'inthissense. [2.3]a.UsuallyIcutthebreadonthetable. b.YesterdayIcutthebreadinthesink.

Thesameword-form

cut,belongingtotheverballexemeCUT,can representtwodifferentgrammaticalwords.

In[2.3a],cutrepresentsthe

grammaticalword cut[verb,present,non3rdperson],i.e.thepresenttense,non thirdpersonform oftheverbCUT.Butin[2.3b]itrepresentsthegramma- ticalword cut[verb,past]whichrealisesthepasttenseofCUT.. Besidesthetwogrammaticalwordsrealisedbytheword-form cutWhICh wehavementionedabove,thereisathirdonewhichyoucanobservein Janehasacutonherfinger.Thisgrammaticalwordiscut[noun.singular]'It belongstoaseparatelexemeCUT,thenoun.Obviously,CUT,thenoun, isrelatedinmeaningtoCUT,theverb.However,CUT,thenoun,isa separatelexemefrom

CUT,theverb,becauseitbelongstoadifferent

word-class(seesection3.5below). Thenatureofthegrammaticalwordisimportantinthediscussionofthe relationshipbetweenwordsandsentencesandtheboundarybetween morphologyandsyntax.Showwhy cutshouldberegardedasrepresentingtwodistinctgrammatical wordsin thefollowing: II thelexeme SEE SLEEP CATCH thelexeme JUMP TALL BOY WOMAN arerealisationsof arerealisationsof

IntroductiontoWord-Structure

18

Weshouldallagreethat:

Thephysicalword-forms

see,sees,seeing,saw,seen sleeps,sleeping,slept catch,catches,catching,caught

Thephysicalword-forms

jump,jumps,jumped,jumping tall,taller,tallest boy,boys woman,women

2.1.2Word-form

Aswehavejustseenabove,sometimes,whenweusetheterm'word'itis nottheabstractvocabularyitemwithaCommoncoreofmeaning:the lexeme, thatwewanttoreferto.Rather,wemayusetheterm'word'to refertoaparticularphysicalrealisation ofthatlexemeinspeechorwriting, i.e.aparticularword-form.Thus,wecanrefer tosee,sees,seeing,sawand seenasfivedifferentwords.Inthissense,threedifferentoccurrencesof anyoneoftheseword-formswouldcountasthreewords.Wecanalsosay thattheword-formseehasthreelettersandtheword-formseeinghassix. And,ifwewerecountingthenumberofwordsinapassage,wewould gladlycount see,sees,seeing,sawandseenasfivedifferentword-forms (belongingtothesameIexeme).

2.2MORPHEMES:THESMALLESTUNITSOF

MEANING

Morphologyisthestudyofwordstructure.Theclaimthatwordshave structuremightcomeasasurprisebecausenormallyspeakersthink of wordsasindivisibleunitsofmeaning.Thisisprobablyduetothefactthat manywordsaremorphologicallysimple.Forexample,the,fierce,desk, eat,boot, at,fee,mosquito,etc.,cannotbesegmented(i.e.dividedup)into smallerunits thatarethemselvesmeaningful.Itisimpossibletosaywhat the -quitopartofmosquitoorthe-ercepartoffiercemeans.

20IntroductiontoWord-StructureMorphemes:TheSmallestUnits

ofMeaning 21
ButverymanyEnglishwordsaremorphologicallycomplex.Theycanbe brokendownintosmallerunitsthataremeaningful.Thisistrueofwords like desk-sandboot-s,forinstance,wheredeskreferstoonepieceof furnitureandbootreferstooneitemoffootwear,whileinbothcasesthe-s servesthegrammaticalfunctionofindicatingplurality. Thetermmorphemeisusedtorefertothesmallest,indivisibleunitsof semanticcontentorgrammaticalfunctionwhichwordsaremadeupof.By definition,a morphemecannotbedecomposedintosmallerunitswhichare eithermeaningfulbythemselvesormarkagrammaticalfunctionlike singular orpluralnumberinthenoun.Ifwedividedupthewordfee[fi:] (whichcontains justonemorpheme)into,say,[f]and[i:],itwouldbe impossibletosaywhateachofthesounds[f]and[i:]meansbyitselfsince soundsinthemselves donothavemeaning. Howdoweknowwhentorecogniseasinglesoundoragroupofsounds asrepresentinga morpheme?Whetheraparticularsoundorstringof soundsistoberegardedasamanifestationofamorphemedependsonthe wordinwhichitappears.So,whileun-representsanegativemorpheme andhasameaningthatcanroughlybeglossedas'not'inwordssuchasun just and1m-tidy,ithasnoclaimtomorphemestatuswhenitoccursinuncle orinunder,sinceintheselatterwordsitdoesnothaveanyidentifiable grammatical orsemanticvalue,because-eteand-derontheirowndonot meananything. (Morphemeswillbeseparatedwithahypheninthe examples.)

Legoprovidesausefulanalogy.

Morphemescanbecomparedtopieces

oflegothatcanbeusedagainandagainasbuildingblockstoformdifferent words. Recurrentpartsofwordsthathavethesamemeaningareisolated andrecognisedasmanifestationsofthesamemorpheme.Thus,thenega tive morphemeun-occursinanindefinitelylargenumberofwords,besides thoselisted above.Wefinditinunwell,unsafe,unclean,unhappy,unfit, uneven, etc. However,recurrenceinalargenumberofwordsisnotanessential propertyofmorphemes.Sometimesamorphememayberestrictedto relativelyfewwords.Thisistrueofthemorpheme-dom,meaning'con dition,state,dignity',which isfoundinwordslikemartyrdom,kingdom, chiefdom, etc.(Myglosses,hereandelsewhereinthebook,arebasedon definitionsintheOxfordEnglishDictionary.) Ithasbeenarguedthat,inanextremecase,amorphememayoccurina singleword.

Lightner(1975:633)hasclaimedthatthemorpheme-ric

meaning'diocese'isonlyfoundinthewordbishopric.Butthisclaimis disputedbyBauer(1983:93)whosuggestsinsteadthatperhaps-ricisnota distinct morphemeandthatbishopricshouldbelistedinthedictionaryas anunanalysableword.Wewillleavethiscontroversyatthatandinstead seehowmorphemesareidentifiedinlessproblematiccases.lUsttwo otherwordswhichcontaineachmorphemerepresentedbelow: [2.4]a.-erasin play-er,call-er -ness as in kind-ness,good-ness -etteasinkitchen-ette,cigar-ette b.ex-asin ex-wife,ex-minister pre- asinpre-war,pre-school mis- asinmis-kick,mis-judge a. Writedownthemeaningofeachmorphemeyouidentify.(Ifyouarein doubt,consultagoodetymologicaldictionary.) b. Whatisthesyntacticcategory(noun,adjective,verb,etc.)ofthefor: whichthismorphemeattachestoandwhatisthecategoryoft e resultingword?

Iexpectyouranswer

toconfirmthat,ineachexamplei,n[2.4],ele~ents recognisedasbelongingtoagivenmorphemecontnbuteIdentifiable meaning tothewordofwhichtheyareapart.form-erISattached verbstoderivenounswiththegeneralmeamngsomeonewhodoes (whereXindicateswhateveractiontheverb..When-nessIS addedtoanadjective,itproducesanounmeamng.thestateor condition(e.g.,ofbeingkind)'.Theofthemorphe.m~ -ettetoanounderivesanewnounwhIchhasthemeamngsmallermSIze (e.g.,akitchenetteisasmallkitchenandcigaretteissmallerthana. Finallythemorphemesex-andpre-denvenounsfromnounswhdem1S- ,'f' derivesverbsfromverbs.Wecanglossthemorphemeex-asormer,pre- as'before'andmis-as'badly'.Sofarwehavedescribedwordswith justoneortwomorphemes.Infact, it ispossibletocombineseveraltogeth~rtoformmore.com plexwords.Thiscan beseeninlongwordshkeand nationwhichcontainthemorphemesun-faith-ful-nessand respectively.Butonwhatgroundsdodivide.upthesew.ordsm fashion?InthefollowingsectionswewillexammethebaSISonwhIch morphemesareidentified.

2.2.1AnalysingWords

Uptonow,wehaveusedthecriterionofmeaningtoidentifymOfJ:'hemes. InmanycasesformsthatsharethesamemeaningmaybesafelyassIgnedto the samemorpheme.Wherethemeaningofamorphemehasbeen whatobscure,youhavebeenencouragedtoconsultagoodetymological dictionary.Unfortunately,inpractice,appealingtomeaningslistedin etymologicaldictionarieshasitsproblems.

Historicallypterwasborrowedfrom

Greek,whereitmeant'featheror

wing'.Theformbibl-alsocamefromGreekwhereitmeant'papyrus scroll, book'.Doyouthinkpter-andbibl-shouldberecognisedasmor: phemesinmodernEnglish? I

23Morphemes:TheSmallestUnitsofMeaning

recognisingarecurrentword-buildingunitasamorphemealthoughwe cannotassignitaconsistentmeaning.

Thisis

trueof-ferinwordslikepre-fer,in-fer,de-fer,con-fer,trans-fer andre-fer. Anetymologicaldictionarywilltellusthat-fercomesfromthe

Latinwordmeaning'bear,bring,send'.However,wewould

behard pressed toidentifyacoifsis1encmeaninglike'bring'attributableto-fer-in everyinstanceabove.

Forthisreasonsomelinguists,suchasAronoff

(1976:8-10),haveargued thatitisthewordinitsentiretyratherthanthe morphemeperse thatmustbemeaningful.Whereasallwordsmustbe meaningfulwhentheyoccur ontheirown,morphemesneednotbe.Some morphemes,likeex-'former'asinex-wife andpre-'before'asinpre-war, haveatransparent,unambiguousmeaningwhileotherslike-jerdonot.

Theirinterpretationvariesdependingonthe

othermorphemesthatoccur togetherwith theminaword. Inviewoftheaboveremarks,whilesemanticconsiderationsmustplaya roleintheidentificationofmorphemes,giventhepitfallsofapurely semanticapproach,linguiststendtogiveahigherprioritytomoreformal factors.

2.2.2Morphemes,

MorphsandAllomorphs

Atonetime,establishingmechanicalproceduresfortheidentificationof morphemeswasconsideredarealisticgoalbystructurallinguists (d. Harris,1951).Butitdidnottakelongbeforemostlinguistsrealisedthatit wasimpossible todevelopasetofdiscoveryproceduresthatwouldlead automatically toacorrectmorphologicalanalysis.Noscientificdiscipline purports toequipitspractitionerswithinfallibleproceduresforarrivingat correcttheories.Creativegeniusis neededtoenablethescientisttomake thatleapintounchartedwaters thatresultsinascientificdiscovery.Whatis trueofscienceingeneralisalsotrue oflinguistics(d.Chomsky,1957:

49-60).Writingagrammar

ofalanguageentailsconstructingatheoryof howthatlanguageworksbymakinggeneralisationsaboutitsstructurethat gobeyondthe datathatareobserved.

Nevertheless,although

therearenoeffectivemechanicalproceduresfor discoveringthegrammaticalstructureofalanguageingeneral or,inour case,thestructureofitswords,thereexistreasonablyreliableandwidely acceptedtechniquesthathave beenevolvedbylinguistsworkingonmor phology.Thesetechniquesareoutlinedinthissection. The-llfainprincipleusedintheanalysisofwordsistheprincipleof contrast."liNecontrastformsthatdiffer(i)inphonologicalshapeduetothe '-SOlilldSU.sedand(ii)inmeaning,broadlydefinedtocoverbothlexical meaningandgrammaticalfunction.Thus,thephonologicaldifference between fbJI/and/g3:11correlateswithasemanticdifference.Thediffer enceinmeaningbetweenthetwosentencesThegirlplays andTheboy diptera bibliophilepteropus bibliographyIntroductiontoWord-Structure 22

Considerthefollowingwords:

[2.5]helicopter bible Ido notknowwhatyoudecided.ButIthinkitisquestionablewhether pter- isamorphemeofmodernEnglish.Ahelicopterisakindofnon-fixed aircraftwhichmostspeakersofEnglishknowabout;pteropusare tr.oplcalbatswith w~ngspopularlyknownas'flyingfoxes'and dlptera areflieS(whichfewofuswhoarenotentomologists knowabout).ObvIOusly,pter-doesoccursinmodernEnglishwords that havethemeaning'pertainingtowings'.Whatisdoubtfuliswhetherthis ispartofthetacitknowledgeofspeakersofEnglishwhoarenotversed m etyn:ology.Mostpeopleprobablygo.throughlifewithoutseeinga sem.an.tlcconnectionbetween'wings'and'helicopters'. . SI:nIlarly,aswehave.alreadynoted,thewordsbible,bibliographyand biblIOphilehavetodowithbooks.Probably manyEnglishspeakerscansee the bookinbibliographyandbibliophile.Butitisunlikelythat anyonelackmgaprofoundknowledgeofEnglishetymology(andaclassi cal.education) isawarethatthewordbibleisnotjustthenameofa scnpturebookandthatitcontainsamorphemewhichisfoundinanumber ofotherwordspertainingtobooks.

Clearly,

weneedtodistinguishbetweenetymologicalinformation whoserelevance isessentiallyhistorical,andsynchronicinformationthat ?artspeakers'competence.Ourprimarytaskasmorphologistsisto mvestIgatespeakers'tacitknowledgeoftherulesoftheirlanguagerather thantoperformhistoricalreconstruction.Weshalldiscussthisfurtherin

4.ThepointIammakingisthatover-relianceonmeaningin

morphemesputsusinaquandaryincaseswhereetymological meanmgsareshroudedinthemists ofhistoryandlosetheirsynchronicrelevance. co~mondefi~itionofthemorphemeasthe'minimalmeaningful umtImplIestheclaimthateverymorphemehasareadilyidentifiable meaning. Butthisisproblematic.Therearecaseswherewecanjustify 25
'wewillseeabook' 'wewillbuyalittlebook' 'theysawbooks' 'wewillseelittlebooks' 'theywillseea book' 'wewillseebooks' 'theysoldlittlebooks' 'theywillbuybooks' 'they boughtalittlebook' 'wesellbooks'kitabo katabo bitabo butabo kitabo bitabo butabo bitabo katabo bitabo [2.9]Itisrealisedas: a.

IIdliftheverbendsinIdlorItI

e.g./mend/-/mendrdlIpemt/-/pemtrd! 'mend''mended''paint''painted' b.IdlafteraverbendinginanyvoicedsoundexceptId! e.g.Ikli:nl-Ikli:ndlIwerl-Iwerdl 'clean''cleaned''weigh''weighed' c. ItIafteraverbendinginanyvoicelessconsonantotherthanItI e.g.Ipa:k!-Ipa:ktlImrsl-Imrstl 'park''parked''miss''missed'In[2.8],eachdifferentmorphrepresentsaseparatemorpheme.Butthis is notalwaysthecase.Sometimesdifferentmorphsmayrepresentthesame morpheme. Forinstance,thepasttenseofregularverbsinEnglishwhichis spelled-edisrealisedinspeechby/rd!,Id!orIt!.Thephonologicalproperties ofthelastsegment oftheverbtowhichitisattacheddeterminethechoice:

Morphemes:TheSmallestUnitsofMeaning

Hints:a.Thewordmeaning'book'appearsinallthesentencesbutin someit issingularandinothersplural. b. 'Book'sometimesreferstoanormalsizebook,andinother casestoalittlebook. c.Wehavethreedifferentverbs. d.

Theverbsareindifferenttenses.

e.

Theverbshavedifferentsubjects.

[2.8]-tabo'book',tu-'we',ki-'singular'(normalsize)nounprefix -laba 'see',ba-'they',bi-'plural'(normalsize)nounprefix -gula 'buy',-li-'future',ka-'singular'(smallsize)nounprefix -tunda'sell',-a-'past',bu-'plural'(smallsize)nounprefix

Theanswerto[2.7]isgivenin[2.8].

[2.7]tulilaba tuligula baalaba tulilaba balilaba tulilaba baatunda baligula baagula tutunda

IntroductiontoWord-Structure

[2.6a]and[2.6c] [2.6e] and[2.6f] [2.6d]and[2.6h] inall theexamples inalltheexamples parkisfoundinalltheexamples,sometimeswithan-ed suffix,sometimeswithan-ssuffixandsometimesonits own suffixedto parkin[2.6b,e,h] suffixedtoparkin[2.6d,f] 24
[2.6]a.Iparkedthecar.e.She parkedthecar.b.

Weparkedthecar.

f.Sheparksthecar. c.Iparkthecar.a

Weparkthecar.

o· d.Heparksthecar.h.

Heparkedthecar.Studythe

datain[2.6]andidentifythemorphs:

It!'-ed'

lsi'-s' DEFINITION:Themorphemeisthesmallestdifferenceintheshapeofa wordthatcorrelateswith thesmallestdifferenceinwordorsentence meaning oringrammaticalstructure. playsisattributabletothedifferenceinlexicalmeaningbetweenand Ig3:1I.Likewise,thedifferenceingrammaticalfunctionbetweenplay-s (presenttense)andplay-ed(pasttense)isresponsibleforthedifferencein meaning betweenThegirlplaysandThegirlplayed.

Themorphsare:

110rphItecursin

larl'1'

IIi:1'she'

Ihi:1'he'

1f5'd1'the'

Ika:!'car'

Ipa:rk!'park'

I1 Forournextexample,weshallperformananalysissimilartotheonewe have justdoneforEnglishondatafromalessfamiliarlanguage.Nowstudy the datain[2.7]whicharetakenfromLugandaandlistallthemorphs. (Although Lugandaisatonelanguage,toneisomittedforsimplicity'ssake asit isnotrelevanthere.)

Theanalysisofin~omorphemesbeginswith

--morIlhs.AmorphISaphysIcalformrepresentingsomemorphemeina language. Itisarecurrentdistinctivesound(phoneme)orsequenceof sounds(phonemes). NowcomparetheLugandaformsin[2.10]withthosein[2.7]above.27

Morphemes:TheSmallestUllitsofMeallillg

[2.12]a.impossible[rmpDsrbl] impatient [rmperfnt] immovable [rmuvgbl] b.intolerable[rntDl:lrgbl] indecent[rndi:sgnt] letusnowexaminesomeEnglishwords,focusingonthepronunciationof theunderlinedpartofeachword,whichrepresentsthenegativemorpheme ill-.Thismorphemecanroughlybeglossedas'not': Wecansaythat(i)lId!,IdlandIt!areEnglishmorphsand(ii)wecan groupallthesethreemorphs togetherasallomo:phsofthe.pasttense morpheme.Likewise,in

Lugandawecansaythat(I)tu-,tw-,-h-and-tabo

aremorphsandfurthermore(ii)tu-andtw-areallomorphsofthesame morphemesince theyrepresentthesamesuperordinateconcept,themor- pheme'first personplural'.. ThecentraltechniqueusedintheidentificationofmorphemesISbased on thenotionofdistribution,i.e.thetotalsetofcontextsinwhicha

Weclassifyasetofa~allomorphs

ofthesamemorphemeiftheyareindistributiOn.Morphs aresaidto beincomplementarydistributionif(i)theyrepresentthesame meaning orservethesamegrammaticalfunctionand(ii)theyarene:er foundinidenticalcontexts.So,thethreemorphsI-rd!,I-d!andI-tlwhich represent theEnglishregularpasttensemorphemeareincomplemen.tary distribution. Eachmorphisrestrictedtooccurringinthecontextsspecified in[2.9].Hence,theyareallomorphsofthesamemorpheme.~ame analysisappliesalsotoLugandatu-andtw-.Bothmorphsmeanweand theyareincomplementarydistribution.

Tu-occursbeforeconsonantsand

tw-beforevowels.Theyarethereforeallomorphsofthefirstpersonplural morpheme.Morphemesrealisedbyaninvariantform(e.g., futureand book)aresaidtohaveasingleallomorph(ct.Matthews,1974:83). b.Luganda e.g.morphememorphememorpheme 'firstpersallplural''future''book' II allomorphallomorphallomorphaIlomorph III I morphmorphmorphmorph

Itu-IItw-I/-li-II-tabol

morph ItI 'wesawabook' 'weboughtbooks' 'wesoldabook' morph Idl morpheme 'pasttense' kitabo bitabo kitabo

IlltroductiolltoWord-Structure

/Id/ morph 26
[2.10]twaalaba twaagula twaatunda Thefirstpersonpluralisrepresentedbytheformtu-in[2.7]andbytw-in [2.10].

Whatdeterminestheselectionoftuvstw-?

I Observethathereagainthedifferenceinformisnotassociatedwitha differenceinmeaning.

Themorphstu-andtw-bothrepresentthefirst

personpluralindifferentcontexts.Tu-isusedifthenextmorphemeis realisedbyaformbeginningwithaconsonantandtw-isselectedifthenext morphemeisrealisedbyaformthatbeginswithavowel. Ifdifferentmorphsrepresentthesamemorpheme,theyaregrouped togetherandtheyarecalledallomorphsofthatmorpheme.So,tu-andtw areallomorphsofthe'firstpersonplural'morpheme.(Forsimplicity's sake,for ourpresentpurposes,weareregarding'firstpersonplural'asa singleunanalysableconcept.)

Onthesamegrounds,lId!,IdlandIt!are

groupedtogetherasallomorphsofthepasttensemorphemeinEnglish. Therelationshipbetweenmorphemes,allomorphsandmorphscanbe representedusingadiagraminthefollowingway: [2.UJa.

English

e.g. a.Identifytheallomorphsofthisnegativemorpheme. b.Writeastatementaccountingforthedistributionofeachallomorph. Ihope thatyouhaveisolatedthefollowingallomorphsofthemorpheme in-:im-[rm-],in-[rn-]andin-[rg-]. Theselectionoftheallomorphthatisusedinaparticularcontextisnot random. In[2.12]thenasalconsonantinthevariousallomorphsofthe morphemein-ispronouncedinavarietyofways,dependingonthenature ofthesoundthatimmediatelyfollows.Topredicttheallomorphthatis selectedineachcase,arulelike[2.13]isrequired: [2.13]a.select[rm]beforealabialconsonant(e.g.,p,b,f,m)asin [ rm]possible,[rm]patient,[rm]movable. b.select [rg]beforethevelarconsonants[k](herespeltwith'c') and[g]asin[rg]compliance,[rg]gratitude. c.select[rn]elsewhere,i.e. beforeanalveolarconsonantlike[t, d,s,z,n],asin[rn]tolerable,[rn]tangible and[rn]decentor beforeavowelasin[rn]active,[rn]elegance. 29

Morphemes:TheSmallestUnitsofMeaning

. \'\ 1!!Ii (\ L-- [2.14] a."iljaIa\'hunger'iljaIa 'nails' mweezf'sweeper' mweezf 'moon' biJsa 'naked'biJsa 'dung' biJggya 'newness'buggya 'envy' aleeta 's/hebrings'aleeta 'onewhobrings' !\1LH '" \i,\.. b.asflka 's/hefries'asfika 'onewhofries' as6ma 's/hereads'asbma 'onewhoreads' ag6ba 's/hechases'agbba 'onewhochases'

In[2.14a],tonaldifferencesareused

todistinguishlexicalitems.Theword formsareidenticalinallrespectsexcepttone.

In[2.14b],ontheother

hand,toneisusedtosignalgrammaticaldistinctions.LHLHcorresponds to LHHFinthefirsttwoverbswhileinthelasttwo,LHHcorrespondsto LLF. Ineachcase,thefirstpatternrepresentsathirdpersonmainclause presenttense fonnoftheverbandthesecondpatternrepresentsthe relativeclauseform.Note:

Interpretthetonediacriticsasfollows:

•=Hightone(H),.=Lowtone(L),andA=FaIlingtone(F)consonantsand thevelarconsonant[g]in[rg]occursbeforevelar consonants. Ineachcasethetwoconsonantsendupsharingthesameplace ofarticulation.Thisexamplealsoillustrates anotherpoint,namelythatspellingisavery poorguide topronunciationinEnglish(andmanyotherlanguages). Where thepointatissuewouldotherwisebeobscuredbythestandar? orthography,phoneticorphonemictranscriptionwillbeusedasappropn ateinthisbook.(Seep.14forkey.) Inthelightofthisdiscussion,letusreturntotheearlierexampleofthe allomorphsof theEnglishregularpasttensemorphemein[2.9].Clearly, thedistribution ofallomorphsisphonologicallyconditioned:I-rdlis chosenafterthealveolarstopsIt!andIdl(withhibeinginsertedto separate thealveolarstopofthesuffixfromthefinalalveolarstopofthe verbtowhichitisattached);voicedI-dlischosenaftervoicedsegments other thanIdlandvoicelessI-t!ischosenaftervoicelessconsonantsother thanIt!.Sofar,alltheexamples ofmorphsthatwehaveseenhaveinvolvedonly vowels andconsonants.But,as[2.14]shows,morphemesmayalsobe signalledbytone,i.e.contrastiveuseofrelativepitch(cf.Hyman,1975;

Katamba,1989;Pike,1948):[

rntrend3rbl] [rnrektrv] [rnelrg;ms] [rgbmpli:t] [rgbmpretrbl] [rggretrtjud]

IntroductiontoWord-Structure

intangible inactive inelegance c.incomplete incompatible ingratitude Thethreeallomorphs[1m],[rn]and[rg]ofthemorphemein-arein complementarydistribution.Thismeansthatselecting oneprecludes selectingtheothers.Notwo ofthemcanoccurinidenticalenvironments.

Thisexampleillustrateswhat

isaverycommonstateofaffairs.Ifa morphenehasseveralallomorphs,thechoiceofallomorphusedinagiven contextisnormallyphonologicallyconditioned.Thismeans thattheallo morphselectedtorepresentthe morphemeinaparticularcontextisone whosephonologicalpropertiesaresimilartothoseofsoundsfoundina neighbouringallomorph ofsomeothermorpheme. Thephonologicalresemblancebetweenthenasalintheprefixandthe firstconsonantrepresentingthemorphemebeforewhichitisplacedisdue toassimilation.

Thepronunciationofthenasalintheprefixisadjustedto

matchtheplaceofarticulationofthefirstconsonantrepresentingthenext morpheme.Thus,in[2.12]thelabialconsonant[m]occursin[rm]beforea labialconsonant, thealveolarconsonant[n]in[rn]occursbeforealveolar28

2.2.3GrammaticalConditioning,LexicalConditioningandSuppletion

[2.15]

PresenttensePasttensea.walk

IWJ:k/

Walked

IWJ:kt!

kiss IkISI kiss-ed

IkISt!

grasp

Igra:spl

grasp-ed

Igra:sptl

b.weep

Iwi:pl

wep-t

Iwept!

sweep

Iswi:pl

swep-t

Iswept!

c.shake

IfeIk/

shook Ifuk/ take Ituk/ took Ituk/ In[2.15bJ,thechoiceofallomorphisgrammaticallyconditioned.The presenceofthepasttensemorphemedeterminesthechoiceoftheIwepl andIsweplallomorphsinverbsthatbelongtothisgroup.Fortheverbsin [12.15c]thepasttensedictates thechoiceoftheallomorphstookandshook oftheverbstakeandshakerespectively. In othercases,thechoiceoftheallomorphmaybelexicallyconditioned, i.e.use ofaparticularallomorphmaybeobligatoryifacertainwordis present.WecanseethisintherealisationofpluralinEnglish.

Normallytheplural

morphemeisrealisedbyaphonologicallycon ditionedallomorphwhosedistribution isstatedin[2.16]: 31
moths

Imo8s1

Examples:cupsleekscarts

lkApsl/li:ksl!ka:tsl

Morphemes:TheSmallestUnitsofMeaning

2.2.4UnderlyingRepresentations

J_I ThepairgoodandbetterisnotuniqueinEnglish.Findoneotherexample ofsuppletion. 1 Iexpectyoutohavefailedtofindaplausibleexplanation.Thereare<:ases wherefornoapparentreasontheregularrulein[2.16]inexplicablyfailsto apply.Thepluralofoxisnot*oxesbutoxen,althoughwordsthatrhyme with oxtaketheexpectedIrzJpluralallomorph(cf;-/f])ks!zJ-toxes.. /bDkslzJboxes).Thechoiceoftheallomorph-enisIt isdependentonthepresenceofthespecificnounox. Finally,thereexistafewmorphemeswhoseallomorphsshownophone ticsimilarity.Aclassicexample ofthisisprovidedbytheformsgoodlbetter whichbothrepresentthelexemeGOODdespitethefactthattheydonot haveevenasinglesoundincommon.Whereallomorphsofamorpheme arephonetically unrelatedwespeakofsuppletion. OtherexamplesofsuppletioninEnglishincludebad-worse(not*bad der);go -went(not*goed) c.selectallomorphl-zJelsewhere(i.e.ifthenounendsina voicednonstridentsegment;thisincludesallvowels andthe consonants/bd g d m nIJI r wj/).

Examples:

bardsmugsroomskeysshoes /ba:dzJImAgzJlru:mzJIki:zJIfu:zJ ,----------------------1 'Canyouexplainwhytherulein[2.16]failstoaccountfortherealisationof thepluralmorphemeinthewordoxen? 1

IntroductiontoWord-Structure

30
Wehaveseeninthelastsectionthatthedistributionofallomorphsis usuallysubject tophonologicalconditioning.However,sometimesphono logicalfactorsplaynoroleintheselectionofallomorphs.Instead,the choice ofallomorphmaybegrammaticallyconditioned,i.e.itmaybe dependentonthepresenceofaparticulargrammaticalelement.Aspecial allomorphmay berequiredinagivengrammaticalcontextalthoughthere might notbeanygoodphonologicalreasonforitsselection.Forexample, in[2.15a],which istypical,inEnglishthepresenceofthepasttense morphemeinthemajorityofcaseshasnoeffectontheselectionofthe allomorphthatrepresentstheverbitself.But,as[2.15b]and[12.5c]show, incertainverbs thepresenceofthepasttensemorphemerequiresthe selectionofaspecialallomorphoftheverb: [2.16]a.selectallomorphI-Iz/ifanounendsinanalveolaroralveo palatalsibilant(i.e.aconsonantwithasharp,hissingsound suchas

Iszf3tfd3/).

Examples:assesmazesfishesbadgesbeaches

h:eSIz/ImerzrzJIfrfrzJ/bred3IzJ/bi:tflzJ b.selectallomorph

I-slifanounendsinanon-stridentvoiceless

consonant(i.e. anyoneofthesoundsIpt k f8/). Abovewehavedistinguishedbetween,ontheonehand,regular,rule governedphonologicalalternation(asituation wherethechoicebetween alternativeallomorphs isregulatedinquitepredictablewaysbythephono logicalproperties ofthedifferentmorphsthatoccurnea:each(see section(2.2.2»andcases ofsuppletionwherethereISphonologrcally arbitraryalternationin therealisationofamorpheme(seesection(2.2.3». Thisisstandardingenerativephonology(cf.ChomskyandHalle,1968;

KenstowiczandKisseberth,1979;

Anderson,1974:51-61).

Merelylistingallomorphsdoes

notallowustodistinguishbetween :r he vitalpointtonoteisthatthethreepartsoftherulein[2.17]arenot Ill?ependentofeachother.Bymakingthreeseparatestatementswehave missedaAsuperiorsolutionwouldbetorestate'[2.17]as [2.18].reVIsedinwhichwepositasingleunderlying representatIOnfrom

WhIChthethreeallomorphsarederived,capturesthe

th~tthealternationintherealisationoftheseallomorphsisduetoasInglefactor,namelyassimilation.eccentric

alternationslikegood-bett(-er)andregularalternationslikethat shownbythenegativeprefixin-orbytheregular-spluralsuffix.Thelatter generala?dwillnormallyapplytoanyformwiththerelevantphonolo gical.propertIeS,itisspecificallyexempted.Thustheregularplural rule In[2.aboveIStoattachI-sl,I-vorl-IVtovirtuallyanynoun tha.tendsIn..appropnatesound.Bycontrast,aruleofsuppletionor COn?ItlOnIngonlyappliesifaformisexpresslymarkedasbeing sub!ecttoIt.Thus,example,ofallEnglishadjectives,onlygoodis sub!ect thesuppletlverulethatgivesbett-erinthecomparative;andonly oxISsubjectt?lexicallyconditionedrulethatsuffixes-entoyieldthe plural?xen.agrammaticallyconditionedrulewillonlybetrig geredIftheappropnategrammaticalconditioningfactorispresent.For example,theallomorphslep-ofthemorphemesleeponlyCO-occurswith thepasttense(orpastparticiple)morpheme.Itcannotbeselectedtoco withthepresentsleep+[past]yieldssleptIslepU(notI*sli:pt/) whIle sleep+[present]givessleepIsli:pl(notI*slep/). ~nngoutdistin~tionbetweenregularphonologicalalternation, whI.chISphoneticallymotIvated,andotherkindsofmorphologicalalter nationlackaphoneticbasis,linguistspositasingleunderlyingrep ?rbaseformfromwhichthevariousallomorphs(oralternants alternatIvephoneticrealisations)ofamorphemearederivedbyapply

Ingon.e

?rm.orephonologicalrules.Thestageswhichaformgoesthrough whenItISbemgconvertedfromanunderlyingrepresentationtoaphoneticrepresentationconstituteaderivation. . Foraexample,letusconsideragaintherepresentationofthe lll-morphemeIn[2.13],whichisrepeatedbelowas[2.17]forconvenience:

33Morphemes:TheSmallestUnitsofMeaning

aftersibilantse.g.IW:1:flwash-IW:1:frzJwashes aftervoicedsegmentsotherthansibilantse.g.lri:dlread lri:dvreads aftervoicelessconsonantsotherthansibilantse.g.Id3Ampl jump-Id3Ampsljumps I-sl I-IV I-v [2.20]I-IVaftersibilantse.g./lIVLiz-IrZIVLiz's

Thesameruleappliestogenitives:

[2.18]Thenasalrealisingthemorphemein-Imlmustappearinthe phoneticrepresentationasanasalconsonantthatsharestheplace ofarticulationoftheinitialconsonantoftheformtowhichitis attached. [2.19] Buthowcanwebecertainthatthebaseformis/rnlratherthanIIrnIor

1r1J!?Wehaveseenthatthenasalassimilatestotheplaceofarticulationof

theconsonantthatfollowsit.Thefactthatwhenavowelfollowswestill find[molappearingasin[m<):dlbl] inaudible,and[m-evltgbl]inevitable isveryrevealing.Fromaphoneticpointofview,vowelsdonothave definiteplaces ofarticulation,onlyconsonantsdo.So,aconsonantcannot assimilatetotheplaceofarticulationofavowel.Theoccurrenceof[m-] beforevowelsis notduetoplaceassimilation.Besides,thealveolarnasalis foundregardless ofwhetherthevowelthatfollowsismadeinthefrontof themouthlike[e],orinthebacklike[:1:].So,theinfluenceofthevowel cannot beresponsibleforthechoiceof[m-].

Asimplesolutionis

toassumethat[m-]isthedefaultform,i.e.theform selectedunlessthereareexplicitinstructionstodootherwise.Ifweposit this formastheunderlyingrepresentation,wedonotneedtochangeit beforevowels orbeforealveolarconsonants.Weonlyneedtochangeit beforenon-alveolarconsonants.

If,howeverweposited[Im-]or[IIJ-]as

theunderlying representation,wewouldneedrulestomodifythemwhen theyappearednotonlybeforenon-labialandnon-velarconsonantsre spectively butalsobeforevowels.Iftwoanalysescanbothaccountprop erlyforthefacts,theanalysisthatprovidesasimplersolutionispreferred. Obviously,inthiscaseitistheanalysisin[2.18](with1m-Iasthebase form)thatwins. Phonologicallyconditionedmorphologicalalternations tendtobevery general. Oftenallomorphsrepresentingdifferentmorphemeswilldisplay thesamephonologicalalternationsiftheyoccurinsimilarphonological environments. Thus,forexample,thevoiceassimilationprocessdisplayed by the-spluralsuffixin[2.16]isnotuniquetothatmorpheme.The-sthird personsingularpresenttensesuffixinverbsshowsexactlythesame alternations,asyoucansee:

IntroductiontoWord-Structure

[1m]alabialconsonant(e.g.p,b,f,m) as

In[Im]possIble,[Im]patient,[Im]movable.

b.select[IIJ]beforethevelarconsonants[k](herespeltwith'c') and[g]asin[IIJ]compliance,[IIJ]compatible,[IIJ]gratitude. select[rn]elsewhere,i.e. beforeanalveolarconsonantlike[t, d,s,z,nl, asin[m]tolerable,[rn]tangibleand[rn]decent orbeforeavowelasin[rn]active,[rn]elegance. a. c. 32
[2.17]

2.3THENATUREOFMORPHEMES

Thestatementin[2.21]showsthatthealternationinquestionisnotthe ?ropertyofanyonemorphemebutratherageneralphonolo 'p~ocessmthelanguage.Thetermsmorphophonemics(inAmerican hngUlstIcS)andmorphophonology (Europeanlinguistics)areusedtorefer to ofthiskindthataccountfortherealisationofphonologically condItIOnedallomorphs ofmorphemes.TherulefortherealisationIrn-Iin [2.18] isanotherexampleofamorphophonemicrule. [2.21]a.TheunderlyingrepresentationofanysibilantsuffixisIz/. b.Itisrealisedas: (i) frz/afteralveolarandalveo-palatalsibilants(e.g.Iszf3tf d31) (ii)Iz/aftervoicedsegmentsotherthansibilants(e.g.vowels andvoicedconsonantslike

Ibm dvi)

(iii)lsiaftervoicelessconsonantsotherthansibilantsIpt k f61

35TheNatureofMorphemes

sofa Is;mfalandballoonIbalu:nlcontaintwosyllableseachwhilecamera /k1£rnaralandhooliganIhu:lrganlcontainthreesyllableseach.(Ihave separatedsyllableswithaspace).

Butallthesewordshaveonlyone

morphemeeach.

Ontheotherhand,thewordbooksIbukslhasone

syllable, buttwomorphemes.Theyarethemorphemebookandthe final -slsiwhichrepresentsthepluralmorphemealthoughItISnota syllableinitsownright..

Whenwedivideawordintomorphemes,wefocus

onstnngsofsound thataremeaningfulregardlessofwhethertheyconstitutesyllables atthe phonologicallevel.Aquestion thatlurksinthebackgroundconcernsthe precisenature oftherelationshipbetweenstringsofsoundsandthe meanings thattheyrepresent.Thisisthequestiontowhichwenowtum.

ThediscussionthatfollowsdrawsonMatthews(1974).

Atfirst,itmightseemreasonabletoassumethattherelationshipbe tweenmorphemesandstrings ofphonemes,whichareidentifiedas morphs,isone ofcomposition.Inthatcase,wecouldsaythatthemor pheme bookIbuk/ismadeupofthephonemesfbi,lu/and/k/.

Aswewillseeinamoment,anapproachwhichassumes

thatmorphemes aremade upofphonemesleadstoatheoreticalcul-de-sac.Itispreferable toviewmorphemesasbeing representedorrealisedormanifestedby morphs. Itisunsoundtoassumethatmorphemesareactuallycomp.osedof (sequencesof)phonemesbecausethiswould ~hatthe:neanmg.ofa morpheme isafunctionofitsphonemiccompOSItIOn.ItISnot,smce phonemesinthemselvescannothavemeani