Distribution Type Public Restricted to

CLARITY:CROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTS,IST-2000-25310D2-2THEEFFECTIVENESSOFSTYLISTICANALYSISFOROTHERLANGUAGESTHANENGLISHPAGE1/1UPM©2002CLARITYCONSORTIUMISTR&DPROJECTSHARED-COSTRTDPROJECTHUMANLANGUAGETECHNOLOGIESPROJECTOFFICER:YVESPATERNOSTERclarityCrosslanguageinformationretrievalandorganizationoftextandaudiodocumentsTheeffectivenessofstylisticanalysisforotherlanguagesthanEnglishD2-2JUNE1,2002,SICS/WP2JUSSIKARLGRENCLARITY:CROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTS,IST-2000-25310D2-2THEEFFECTIVENESSOFSTYLISTICANALYSISFOROTHERLANGUAGESTHANENGLISHPAGE2/2UPM©2002CLARITYCONSORTIUMISTProjectNumberIST-2000-25310AcronymClarityFulltitleCROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTSEUProjectofficerYvesPaternosterDeliverableNumberD2-2NameTheeffectivenessofstylisticanalysisforotherlanguagesthanEnglishTaskNumberTNameWorkPackageNumberWP2NameInterfacev1.0DateofdeliveryContractual2002/06/01Actual2002/06/01CodenamecodenameVersion1.0draft
finalNaturePrototype
ReportSpecification
Tool
Other:DistributionTypePublicRestricted
to:Authors(Partner)[e-mail][phone]Fax[phone]Abstract(fordissemination)TheCLARITYprojectaimstoexploretheeffectivenessofstylisticanalysisinanimplementedinteractiveinformationretrievalsystem.Asreportedinearlierliterature,stylisticanalysisseemstoworkwellinpost-hocexperimentalsettings.Ithasnotbeentestedinrealusertests,however.TheCLARITYprojectaimstoimplementstylisticanalysistoolsandtesttheminongoinginteractiveusertests.Inaddition,mostexperimentsperformedonuseofstylisticsininformationretrievalhavebeenexclusivelyonEnglish-languagematerial.TheCLARITYprojectaimstoextendthestylisticanalysisfunctionalitytolanguagesotherthanEnglish,notablythelanguagesemployedintheCLARITYproject:Finnish,Swedish,Latvian,andLithuanian.ThisreportdescribesthecurrentstatusofthereimplementationofknownalgorithmsforthestylisticanalysisofEnglishandtheestablishmentofthepossibilitesforthetransferofEnglishlanguagealgorithmstoSwedish.Finnish,Latvian,andLithuanianarealsopartoftheproject,andeffectivenessandusefulnessofthetoolsforFinnishandfortheBalticlanguagescanbeextrapolatedatthispoint.KeywordsStylisticanalysis,multiplelanguagesCLARITY:CROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTS,IST-2000-25310D2-2THEEFFECTIVENESSOFSTYLISTICANALYSISFOROTHERLANGUAGESTHANENGLISHPAGE3/3UPM©2002CLARITYCONSORTIUMCONTENTSCONTENTS..............................................................................................................31SUMMARY...........................................................................................................42WHYSTYLISTICANALYSISINTEXTRETRIEVAL....................................53EARLYEXPERIMENTS.....................................................................................64GENRE,TOPIC,STYLE,ANDSOURCEININFORMATIONRETRIEVAL75CLARITYANDSTYLISTICCATEGORIES.....................................................96LINGUISTICFEATURESEMPLOYED............................................................107ISTHETEXTSUBJECTIVEANDARGUMENTATIVE?...............................118ISTHETEXTABSTRACTORCONCRETE?..................................................139ISTHETEXTANINTERVIEWORNOT?.......................................................1410LANGUAGESPECIFICITYOFLINGUISTICFEATURES....................1511FUTUREUSEREXPERIMENTS................................................................1612REFERENCES..............................................................................................17CLARITY:CROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTS,IST-2000-25310D2-2THEEFFECTIVENESSOFSTYLISTICANALYSISFOROTHERLANGUAGESTHANENGLISHPAGE4/4UPM©2002CLARITYCONSORTIUM1SUMMARYTheCLARITYprojectaimstoexploretheeffectivenessofstylisticanalysisinanimplementedinteractiveinformationretrievalsystem.Asreportedinearlierliterature,stylisticanalysisseemstoworkwellinpost-hocexperimentalsettings.Ithasnotbeentestedinrealusertests,however.TheCLARITYprojectaimstoimplementstylisticanalysistoolsandtesttheminongoinginteractiveusertests.Inaddition,mostexperimentsperformedonuseofstylisticsininformationretrievalhavebeenexclusivelyonEnglish-languagematerial.TheCLARITYprojectaimstoextendthestylisticanalysisfunctionalitytolanguagesotherthanEnglish,notablythelanguagesemployedintheCLARITYproject:Finnish,Swedish,Latvian,andLithuanian.ThisreportdescribesthecurrentstatusofthereimplementationofknownalgorithmsforthestylisticanalysisofEnglishandtheestablishmentofthepossibilitesforthetransferofEnglishlanguagealgorithmstoSwedish.Finnish,Latvian,andLithuanianarealsopartoftheproject,andeffectivenessandusefulnessofthetoolsforFinnishandfortheBalticlanguagescanbeextrapolatedatthispoint.CLARITY:CROSSLANGUAGEINFORMATIONRETRIEVALANDORGANISATIONOFTEXTANDAUDIODOCUMENTS,IST-2000-25310D2-2THEEFFECTIVENESSOFSTYLISTICANALYSISFOROTHERLANGUAGESTHANENGLISHPAGE5/5UPM©2002CLARITYCONSORTIUM