搜索
JournalofExperimentalPsychology:Learning,Memory,andCognition,1999,Vol.25,1347-1365.PragmaticConstraintontheInterpretationofComplexNounPhrasesinSpanishandEnglishRobertThornton,MaryellenC.MacDonald,andMarielaGilUniversityofSouthernCaliforniaFourexperimentsexaminedtheroleofapragmaticconstraint,themodifiabilityofnounphrases(NPs),inthemodificationofcomplexNPs.Experiment1demonstratedthatNPsthathadreceivedrelativelyspecificpriormodificationwerelesslikelytotakeadditionalmodificationthanNPswithlessspecificmodification.ThiseffectobtainedinbothSpanishandEnglishusingtwooff-linetasks.Experiments2and3demonstratedon-linemodifiabilityeffectsforbothlanguagesusingaself-pacedreadingtask.TheresultsfurthersuggestedthatalthoughSpanishandEnglishspeakersmayhaveopposingmodificationpreferences,modifiabilityconstrainedtheirinterpretationsinthesamedirec-tion.TheresultsofExperiment4suggestedthatdiscrepanciesbetweentheoff-lineresultsfromExperiment1andtheon-lineresultsfromExperiment3maybeduetotaskdifferences.Implicationsarediscussedinrelationtocurrentmodelsofsentenceprocessing.Althoughcross-linguisticinvestigationhasplayedanimportantroleinsometheoriesoflanguagecomprehen-sion(e.g.,Bates&MacWhinney,1979),workinsen-tenceprocessingandsyntacticambiguityresolutionhasuntilrecentlyfocusedprimarilyonEnglishconstructions.Thisnarrowfocusisunfortunate,asalternativetheoreticalapproachestosentenceprocessingcanmakeverydiffer-entclaimsregardingtheuniversalityofprocessingmech-anisms.Garden-paththeory,forexample,holdsthatspeak-ersofalllanguageswillinitiallyresolvesyntacticambigui-tiesthroughverygeneral,universalparsingprinciples(Fra-zier,1987).Morerecently,anumberoftheorieshavesug-gestedthatambiguityresolutionprocesseswillbeverysen-sitivetospecificpropertiesofindividuallanguages,thoughthesetheoriesdifferintheirclaimsofexactlywhatthosesensitivitiesmightbe(Cuetos&Mitchell,1988;Gibson,Pearlmutter,Canseco-Gonzalez,&Hickok,1996;Gilboy,Sopena,Clifton,&Frazier,1995;MacDonald,Pearlmutter,&Seidenberg,1994;Mitchell&Cuetos,1991;Trueswell&Tanenhaus,1994).Inthisarticle,weinvestigatehowSpan-RobertThornton,DepartmentofPsychologyandNeuro-scienceProgram,UniversityofSouthernCalifornia;MaryellenC.MacDonald,DepartmentsofPsychologyandLinguisticsandNeu-roscienceProgram,UniversityofSouthernCalifornia;MarielaGil,DepartmentsofPsychologyandLinguistics,UniversityofSouthernCalifornia.ThisresearchwassupportedbyNSFGrantSBR-9511270tothesecondauthor.WewouldliketothankJoeAllen,JudithKroll,DonMitchell,andtwoanonymousreviewersforusefulcommentsanddiscussion.AddresscorrespondenceandreprintrequeststoRobertThornton,HedcoNeurosciencesBuilding,Uni-versityofSouthernCalifornia,LosAngeles,CA90089-2520.E-mail:thornton@gizmo.usc.edu.ishandEnglishspeakersusepragmaticinformationtocom-prehendanambiguitythatispresentinbothlanguages.Thesimilaritiesanddifferencesinambiguityresolutionacrossthesetwolanguagesoffersomeimportantinsightsintothe-oriesofsentenceprocessing.Cross-LinguisticDataThemostinformativecross-linguisticinvestigationsofambiguityresolutiontodatehavebeenthosethatcomparetwoormorelanguagesthatshareaparticularsyntacticam-biguity,ratherthancomparisonsofdifferentambiguitiesindifferentlanguages.OnesyntacticconstructionthathasbeenparticularlyusefulinthisregardisanounmodificationambiguitywiththestructureNP1prepositionNP2modifier,inwhichthefinalmodifyingphrasemaymodifyeitherthemoredistantnounphrase,NP1,orthenearer(oftentermedlocal)nounphrase,NP2.ASpanishexampleisin(1):(1)Lahijadelcoronelquetuvoelaccidente“Thedaughterofthecolonelwhohadtheaccident”Thisambiguityispresentinalargenumberoflan-guages,includingEnglish,Dutch,Spanish,French,Ital-ian,andGerman(Brysbaert&Mitchell,1996;Cuetos&Mitchell,1988;DeVincenzi&Job,1995;Gibsonetal.,1996;Gilboyetal.,1995;Hemforth,Konieczny,&Scheep-ers,inpress;Zagar,Pynte,&Rativeau,1997).Comprehen-dersofmostoftheselanguagesappeartoexhibitanoverallpreferencetointerpretambiguousmodifiersasmodifyingthedistantsite(NP1),whereasEnglishspeakerstendtoin-terpretmodifiersasattachingtothelocalsite(seeMitchell&Brysbaert,1998,forareview).Therehavebeenanumberofproposalsattemptingtoaccountforthesedata.Forexam-ple,MitchellandCuetos(1991)offeredthelinguistictun-inghypothesis,anexposure-basedparsingmechanismfor2THORNTON,MACDONALD,ANDGILwhichinitialparsingdecisionsaresetineachlanguageonthebasisofthefrequencydistributionoftheoccurrenceofdifferentstructures.Althoughtuningcantheoreticallyoc-curusingavarietyofinformation(Mitchell,Cuetos,Corley,&Brysbaert,1995),workdonewithinthisframeworkhastypicallyassumedthattuningoccursatastructuralgrain,inwhichthefrequenciesofsyntacticstructures,independentoflexicalordiscoursecontent,shapeambiguityresolutionpreferences.Incontrast,Gibsonandcolleagues(Gibsonetal.,1996;Gibson,Pearlmutter,&Torrens,inpress)havepro-posedthatcross-linguisticvariationinthesecasesiscausedbydifferentialweightingoftwouniversalparsingprinci-ples:recencypreference,whichfavorslocalattachment(cf.lateclosure,Frazier&Rayner,1982,andrightas-sociation,Kimball,1973),andpredicateproximity,whichfavorsdistantattachment.