yanglingupennedu June 2010 Abstract We explore the implications of Zipfs law for the understanding of linguistic productivity Focusing on language acquisition we show that the item usage based approach has not been supported by adequate statistical e ID: 3789 Download Pdf
Yang Department of Linguistics and Psychology Yale University 370 Temple Street 302 New Haven CT 06511 USA Recent demonstrations of statistical learning in infants have reinvigorated the innateness versus learning debate in language acquisition This
EVA MARÍA GIL CENTENO , DESIRÉE MORENO POZO. PROFESOR COORDINADOR: JOAQUÍN RIVERO RODRÍGUEZ. George Kingsley Zipf (1902-1950), lingüista y filólogo estadounidense, aplicó técnicas de análisis estadístico al estudio de diferentes lenguas. .
Spoon River Anthology. Book of poems about a town called Spoon River. All poems have the a person’s name as a title. The book bears out the interactions among the people. They are all dead. Published in 1915.
Okon Department of Linguistics and Communication studies University of Calabar Nigeria ABSTRACT Bilinguals often switch between their two langu ages in the middle of a conversation Spolsky 1998 says code switches can take place between or even withi
guse Abstract This paper deals with the problem of recognizing and extracting acronym de64257nition pairs in Swedish medical texts This project applies a rulebased method to solve the acronym recognition task and compares and evaluates the results of
Act III. Act III, Scene I. Martha: a woman full of desperation and self-pity.. We see her starved for affection.. Martha’s ‘worship’ of her father has turned into admitting he cries all the time. .
scientific study. of language. . The . word ‘language’ here . means language . in general, not a particular language.. According to Robins (1985), linguistics. is concerned with human language as a universal and recognizable .
century England. Monarchy Problems. James I (VI of Scotland), 1603-25. Gunpowder Plot 1605 – restore Catholic monarch. Guy Fawkes Day – Nov. 5. Golden Age of Elizabethan literature. King James Version of Bible – “thou .
In thi s paper I examine both referential and quantificational accounts of complex demonstratives in order to show that neither side satisfactorily accounts for the all the data I then outline an alternative analysis on which although lexically univ
SulangText003 - v1 ‘Testimony’: Two texts from the Wotu language of South Sulawesi, Indonesia written and translated into Indonesian by Stefanus Syuaib with introduction, interlinearization, ann
Published bymyesha-ticknor
yanglingupennedu June 2010 Abstract We explore the implications of Zipfs law for the understanding of linguistic productivity Focusing on language acquisition we show that the item usage based approach has not been supported by adequate statistical e
Download Pdf - The PPT/PDF document "Whos Afraid of George Kingsley Zipf Char..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
AbstractWeexploretheimplicationsofZipf'slawfortheunderstandingoflinguisticproductivity.Focusingonlanguageacquisition,weshowthattheitem/usagebasedapproachhasnotbeensupportedbyadequatestatisticalevidence.Bycontrast,thequantitativepropertiesofaproductivegrammarcanbepreciselyformulated,andareconsistentwithevenveryyoungchildren'slanguage.Moreover,drawingfromre-searchincomputationallinguistics,thestatisticalpropertiesofnaturallanguagestronglysuggestthatthetheoryofgrammarbecomposedofgeneralprincipleswithoverarchingrangeofapplicationsratherthanacollectionofitemandconstructionspecicexpressions.2 syntacticcompetenceiscomprisedtotallyofverb-specicconstructionswithopennominalslots,ratherthanabstractandproductivesyntacticrulesunderwhichpresumablyabroaderrangeofcombinationsisexpected.Limitedmorphologicalinection.AccordingtoastudyofchildItalian(Pizutto&Caselli1994),47%ofallverbsusedby3youngchildren(1;6to3;0)wereusedin1person-numberagreementform,andanadditional40%wereusedwith2or3forms,wheresixformsarepossible(3person2number).Only13%ofallverbsappearedin4ormoreforms.Again,thelowlevelofusagediversityistakentoshowthelimitednessofgeneralizationcharacteristicofitem-basedlearning.Unbalanceddeterminerusage.CitingPine&Lieven(1997)andothersimilarstudies,itisfoundthatwhenchildrenbegantousethedeterminersaandthewithnouns,therewasalmostnoover-lapinthesetsofnounsusedwiththetwodeterminers,suggestingthatthechildrenatthisagedidnothaveanykindofabstractcategoryofDeterminersthatincludedbothoftheselexicalitems.Thisndingisheldtocontradicttheearlieststudy(Valian1986)whichmaintainsthatchilddeter-mineruseisproductiveandaccuratelikeadultsbytheageof2;0.Sofaraswecantell,however,theseevidenceinsupportforitem-basedlearninghasbeenpresented,andaccepted,onthebasisofintuitiveinspectionsratherthanformalempiricaltests.Forinstance,amongthenumerousexamplesfromchildlanguage,nostatisticaltestwasgiveninthemajortreat-ment(Tomasello1992)wheretheVerbIslandHypothesisandrelatedideasaboutitem-basedlearningareputforward.Specically,notesthasbeengiventoshowthattheobservationsabovearestatisticallyinconsistentwiththeexpectationofafullyproductivegrammar,thepositionthatitem-basedlearningopposes.Nor,forthatmatter,aretheseobservationsshowntobeconsistentwithitem-basedlearning,which,asweshallsee,hasnotbeenclearlyenougharticulatedtofacilitatequantitativeevaluation.Inthispaper,weprovidestatisticalanalysistollthesegaps.Wedemonstratethatchildren'slanguageuseactuallyshowstheoppositeoftheitem-basedview;theproductivityofchildren'sgrammarisinfactcon-rmed.Morebroadly,weaimtodirectresearcherstocertainstatisticalpropertiesofnaturallanguageandthechallengestheyposeforthetheoryoflanguageandlanguagelearning.Ourpointofdepartureisanamethathasbeen,andwillcontinueto,tormenteverystudentoflanguage:GeorgeKingsleyZipf.2ZipanPresence2.1ZipanWordsUndertheso-calledZipf'slaw(Zipf1949),theempiricaldistributionsofwordsfollowacuriouspattern:relativelyfewwordsareusedfrequentlyveryfrequentlywhilemostwordsoccurrarely,withmanyoccurringonlyonceinevenlargesamplesoftexts.Moreprecisely,thefrequencyofawordtendstobeapproximatelyinverselyproportionaltoitsrankinfrequency.LetfbethefrequencyofthewordwiththerankofrinasetofNwords,then:f=C rwhereCissomeconstant(1)IntheBrowncorpus(Kucera&Francis1967),forinstance,thewordwithrank1isthe,whichhasthefrequencyofabout70,000,andthewordwithrank2isof,withthefrequencyofabout36,000:almostexactlyasZipf'slawentails(i.e.,700001360002).TheZipancharacterizationofwordfrequency4 2.2ZipanCombinatoricsThelongtailofZipf'slaw,whichisoccupiedbylowfrequencywords,becomesevenmorepronouncedwhenweconsidercombinatoriallinguisticunits.Take,forinstance,n-grams,thesimplestlinguisticcombinationthatconsistsofnconsecutivewordsinatext.2Sincetherearealotmorebigramsandtrigramsthanwords,thereareconsequentlyalotmorelowfrequencybigramsandtrigramsinalinguisticsample,asFigure2illustratesfromtheBrowncorpus(forrelatedstudies,seeTeahan1997,Haetal.2002): 40 50 60 70 80 90 100 200 100 50 40 30 20 10 5 4 3 2 1 Cumulative%oftypesFrequencywords bigrams trigrams Figure2.Thevastmajorityofn-gramsarerareevents.Thex-axisdenotesthefrequencyofthegram,andthey-axisdenotesthecumulative%ofthegramthatappearatthatfrequencyorlower.Forinstance,thereareabout43%ofwordsthatoccuronlyonce,about58%ofwordsthatoccur1-2times,68%ofwordsthatoccur1-3times,etc.The%ofunitsthatoccurmultipletimesdecreasesrapidly,especiallyforbigramsandtrigrams:approximately91%ofdistincttrigramtypesintheBrowncorpusoccuronlyonce,and96%occuronceortwice.Therangeoflinguisticformsissovastthatnosampleislargeenoughtocaptureallofitsvarietiesevenwhenwemakeacertainnumberofabstractions.Figure3plotstherankandfrequencydistributionsofsyntacticrulesofmodernEnglishfromthePennTreebank(Marcusetal.1993).Sincethecorpushasbeenmanuallyannotatedwithsyntacticstructures,itisstraightforwardtoextractrulesandtallytheirfrequencies.3ThemostfrequentruleisPP!PNP,followedbyS!NPVP:again,theZipf-likepatterncanbeseenbythecloseapproximationbyastraightlineonthelog-logscale. 2Forexample,giventhesentencethecatchasesthemouse,thebigrams(n=2)arethecatchasesthemousearethecat,catchases,chasesthe,andthemouse,andthetrigrams(n=3)arethecatchases,catchasesthe,chasesthemouse.Whenn=1,wearejustdealingwithwords.3CertainruleshavebeencollapsedtogetherastheTreebankfrequentlyannotatesrulesinvolvingdistinctfunctionalheadsasseparaterules.6 SupposealinguisticsamplecontainsSdeterminer-nounpairs,whichconsistofDandNuniquedeterminersandnouns.(InthepresentcaseD=2foraandthe.)ThefullproductivityoftheDPrule,bydenition,meansthatthetwocategoriescombineindependently.Twoobservations,oneobviousandtheothernovel,canbemadeaboutthedistributionsofthetwocategoriesandtheircombinations.First,nouns(andopenclasswordsingeneral)willfollowzipf'slaw.Forinstance,thesingularnounsthatappearintheformofDP!DNintheBrowncorpusshowalog-logslopeof-0.97.IntheCHILDES(MacWhinney2000)speechtranscriptsofsixchildren(seesection3.2fordetails),theaveragevalueoflog-logslopeis-0.98.Thismeansthatinalinguisticsample,relativelyfewnounsoccuroftenbutmanywilloccuronlyoncewhichofcoursecannotoverlapwithmorethanonedeterminers.Second,whilethecombinationofDandNissyntacticallyinterchangeable,N'stendtofavoroneofthetwodeterminers,aconsequenceofpragmaticsandindeednon-linguisticfactors.Forinstance,wesaythebathroommoreoftenthanabathroombutabathmoreoftenthanthebath,eventhoughallfourDPsareperfectlygrammatical.Thereasonforsuchasymmetriesisnotamatteroflinguisticinterest:thebathroomismorefrequentthanabathroomonlybecausebodilyfunctionsareamoreconstantthemeoflifethanrealestatematters.Wecanplacethesecombinatorialasymmetriesinaquantitativecontext.Asnotedearlier,about75%ofdistinctnounsintheBrowncorpusoccurwithexclusivelytheorabutnotboth.Eventheremaining25%whichdooccurwithtendtohavefavorites:onlyafurther25%(i.e.12.5%ofallnouns)areusedwithaandtheequallyfrequently,andtheremaining75%areunbalanced.Overall,fornounsthatappearwithbothdeterminersasleastonce(i.e.25%ofallnouns),thefrequencyratiobetweenthemoreoverthelessfavoreddetermineris2.86:1.(Ofcourse,somenounsfavorthewhileothersfavora,asthebathroomandbathexamplesaboveillustrate.)Thesegeneralpatternsholdforchildandadultspeechdataaswell.Inthesixchildren'stranscripts(section3.2),theaveragepercentageofbalancednounsamongthosethatappearwithboththeandais22.8%,andthemorefavoredvs.lessfavoreddeterminerhasanaveragefrequencyratioof2.54:1.Eventhoughtheseratiosdeviatefromtheperfect2:1ratiounderthestrictversionofZipf'slawthemorefavoredisevenmoredominantoverthelesstheyclearlypointouttheconsiderableasymmetryincategorycombinationusage.Asaresult,evenwhenanounappearsseveraltimesinasample,thereisstillasignicantchancethatithasbeenpairedwithasingledeterminerinallinstances.Together,Zipandistributionsofatomiclinguisticunits(words;Figure1)andtheircombinations(n-gramsFigure2,phrasesFigure3)ensurethatthedeterminer-nounoverlapmustberelativelylowunlessthesamplesizeSisverylarge.Insection4,weexamine,anddiscoversimilarpatterns,fortheusagepatternsofverbalsyntaxandmorphology.Forthemoment,wedevelopaprecisemathematicaltreatmentandcontrastitwiththeitem-basedlearningapproachinthecontextoflanguageacquisition.3QuantifyingProductivity3.1TheoreticalanalysisConsiderasample(N,D,S),whichconsistsofNuniquenouns,Duniquedeterminers,andSdeterminer-nounpairs.HereD=2fortheandathoughweconsiderthegeneralcasehere.Thenounsthathaveappearedwithmorethanone(i.e.two)determinerswillhaveanoverlapvalueof1;otherwise,theyhavetheoverlapvalueof0.Theoverlapvaluefortheentiresamplewillbethenumberof1'sdividedbyN.Ouranalysiscalculatestheexpectedvalueoftheoverlapvalueforthesample(N,D,S)underthe8 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 80 90 100 ExpectedOverlapRank Figure4.Expectedoverlapvaluesfornounsorderedbyrank,forN=100nounsinasamplesizeofS=200withD=2determiners.WordfrequenciesareassumedtofollowtheZipandistribution.Ascanbeseen,fewofnounshavehighprobabilitiesofoccurringwithbothdeterminers,butmostare(far)belowchance.Theaverageoverlapis21.1%.UnderZipandistributionofcategoriesandtheirproductivecombinations,lowoverlapvaluesareamathematicalnecessity.Asweshallsee,thetheoreticalformulationherenearlyperfectlymatchthedistributionalpatternsinchildlanguage,towhichweturnpresently.3.2DeterminersandproductivityMethods.Tostudythedeterminersysteminchildlanguage,weconsiderthedatafromsixchildrenAdam,Eve,Sarah,Naomi,Nina,andPeter.ThesearetheallandonlychildrenintheCHILDESdatabase(MacWhinney2000)withsubstantiallongitudinaldatathatstartsattheverybeginningofsyntacticde-velopment(i.e,oneortwowordstage)sothattheitem-basedstage,ifexists,couldbeobserved.Forcomparison,wealsoconsidertheoverlapmeasureoftheBrowncorpus(Kucera&Francis1967),forwhichproductivityisnotindoubt.Werstremovedtheextraneousannotationsfromthechildtextandthenappliedanopensourceimplementationofarule-basedpart-of-speechtagger(Brill1995):5wordsarenowassociatedwiththeirpart-of-speech(e.g.,preposition,singularnoun,pasttenseverb,etc.).ForlanguagessuchasEnglish,whichhasrelativelysalientcuesforpart-of-speech(e.g.,rigidwordorder,lowdegreeofmorphologicalsyncretism),suchtaggerscanachievehighaccuracyatover97%.Thisalreadylowerrorratecausesevenlessconcernforourstudy,sincethedeterminersaandthearenotambiguousandarealwayscor-rectlytagged,whichreliablycontributestothetaggingofthewordsthatfollowthem.TheBrownCorpusisavailablewithmanuallyassignedpart-of-speechtagssonocomputationaltaggingisnecessary.Withtaggeddatasets,weextractedadjacentdeterminer-nounpairsforwhichDiseitheraorthe,andNhasbeentaggedasasingularnoun.Wordsthataremarkedasunknown,largelyunintelligible 5Availableathttp://gposttl.sourceforge.net/.10 15 20 25 30 35 40 45 50 15 20 25 30 35 40 45 50 EmpiricalPredictedidentity r=1.06 Figure5.Thesolidlinerepresentsthelinearregressiontoftheexpectedvs.empiricalvaluesofoverlapinTable1column5and6(r=1.08,adjustedR2=0.9716).Thedottedlineindicatesaperfectt(i.e.,theidentityfunctiony=x).Therefore,wecouldthatthedeterminerusagedatafromchildlanguageisconsistentwiththeproductiveruleDP!DN.Theempiricalstudiesalsorevealconsiderableindividualvariationintheoverlapvalues,anditisinstructivetounderstandwhy.AstheBrowncorpusresultshows(Table1lastrow),samplesizeS,thenumberofnounsN,orthelanguageuser'sagealoneisnotpredictiveoftheoverlapvalue.Thevariationcanberoughlyanalyzedasfollows;seeValianetal.(2009)forarelatedproposal.GivenNuniquenounsinasampleofS,greateroverlapvaluecanbeobtainedifmorenounsoccurmorethanonce.Thatis,wordswhoseprobabilitiesaregreaterthan1=Scanincreasetheoverlapvalue.Zipf'slaw(2)allowsustoexpressthiscutofflineintermswithranks,astheprobabilityofthenounnrwithrankrhastheprobabilityof1=(rHN).ThederivationbelowusesthefactthattheNthHarmonicNumberPNi=11=icanbeapproximatedbylnN.S1 rHN=1r=S HNS lnN(5)Thatis,onlynounswhoseranksarelowerthanS=(lnN)canbeexpectedtobenon-zerooverlaps.ThetotaloverlapisthusamonotonicallyincreasingfunctionofS=(NlnN)which,giventheslowgrowthoflnN,isapproximatelyS=N,atermthatmustbepositivelycorrelatedwithoverlapmeasures.Thisresultisconrmedinstrongestterms:S=Nisanearperfectpredictorfortheempiricalvaluesofoverlap(lasttwocolumnsofTable1):r=0.986,p0.00001.12 ciesfromthatchild'sinput(localmemorylearner)andthedeterminer-nounpairsalongwiththeirfre-quenciesintheentire1.1millionutterancesofadultspeech(globalmemorylearner).ForeachchildwithasamplesizeofS(seeTable1,column2),andforeachvariantofthememorymodel,weusetheMonteCarlosimulationtorandomlydrawSpairsfromthetwosetsofdatathatcorrespondtothelocalandglobalmemorylearningmodels.Theprobabilitywithwhichapairisdrawnisproportionaltoitsfre-quencyinthetwosetsofdata.Thus,amorefrequently-usedpairsintheinputwillhaveahigherchanceofbeingdrawn,whichreectsfrequencyeffectsinlearningsoftenemphasizedintheitem/usage-basedapproach(e.g.,Tomasello2001,2003,Matthewsetal.2005,Bybee&Hopper2001,amongothers).Eachsample,then,consistsofalistofdeterminer-nounpairswithvaryingoccurrencecounts.Wecalculatethevalueofoverlapfromthislist,thatis,thepercentageofnounsthatappearwithbothaandtheoverthetotalnumberofnouns.Theresultsareaveragedover1000draws.TheseresultsaregiveninTable2. Child SampleSize(S) Overlap(globalmemory) Overlap(localmemory) Overlap(empirical) Eve 831 16.0 17.8 21.6 Naomi 884 16.6 18.9 19.8 Sarah 2453 24.5 27.0 29.2 Peter 2873 25.6 28.8 40.4 Adam 3729 27.5 28.5 32.3 Nina 4542 28.6 41.1 46.7 First100 600 13.7 17.2 21.8 First300 1800 22.1 25.6 29.1 First500 3000 25.9 30.2 34.2 Table2.Thecomparisonofdeterminer-nounoverlapbetweentwovariantsofitem-basedlearningandempiricalresults.Bothsetsofoverlapvaluesfromthetwovariantsofitem-basedlearning(column3and4)differsig-nicantlyfromtheempiricalmeasures(column5):p0.005forbothpairedt-testandpairedWilcoxontest.Thissuggeststhatchildren'suseofdeterminersdoesnotfollowthepredictionsoftheitem-basedlearningapproach;itcertainlydoesnotseemtobetheresultofthechildretrievingjointlystoreddeterminer-nounpairsfromtheinputinafrequencysensitivefashion.Naturally,ourevaluationhereistentativesincethepropertestcanbecarriedoutonlywhenthetheoreticalpredictionsofitem-basedlearningaremadeclear.Andthatisexactlythepoint:theadvocatesofitem-basedlearningnotonlyrejectedtheal-ternativehypothesiswithoutadequatestatisticaltests,butalsoacceptedthefavoredhypothesiswithoutadequatestatisticaltests.4AnItemizedLookatVerbsTheformalanalysisinsection3canbegeneralizedtochildverbsyntaxandmorphology,whichareamongthemainsupportingcasesforitem-basedlearning.Unfortunately,theacquisitiondatainsup-portoftheVerbIslandHypothesis(Tomasello1992)andtheitem-basednatureofearlymorphology(Pizutto&Caselli1994)citedinsection1hasnotbeenmadeavailableinthepublicdomain.ButtheZipanrealityisinherent:thecombinatoricsofverbsandtheirmorphologicalandsyntacticassociatesaresimilarlylopsidedinusagedistributionasiswiththedeterminers.Wenowturntoexaminethestatisticaldistributionsofverbalmorphologyandsyntax.14 4.2AllverbsareislandsWenowstudythedistributionalpropertiesofverbalsyntaxthathavebeenattributedtotheVerbIslandHypothesis.Wefocusonconstructionsthatinvolveatransitiveverbanditsnominalobjects,includingpronounsandnounphrases.FollowingthedenitionofsentenceframeinTomasello'soriginalVerbIslandstudy(1992,p242),eachuniquelexicalitemintheobjectpositioncountsasauniqueconstructionfortheverb.Figure6showstheconstructionfrequenciesofthetop15transitiveverbsin1.1millionchilddi-rectedutterances.Processingmethodsareasdescribedinsection3.2excepthereweextractadjacentverb-nominalpairsinpart-of-speechtaggedtexts.Foreachverb,wecountitstop10mostfrequentcon-structions,whicharedenedastheverbfollowedauniquelexicalitemintheobjectposition(e.g.,askhimandaskJohnaredifferentconstructions.)Foreachofthe10ranks,wetalliedtheconstructionfrequenciesforall15verbs.8 4 4.5 5 5.5 6 6.5 7 7.5 8 0 0.5 1 1.5 2 2.5 log(freq)log(rank) Figure6.Rankandfrequencyofverb-objectconstructionsbasedon1.1millionchild-directedutterances.TheverbconstructionfrequencythusalsofollowaZipf-likepattern:evenforlargecorpora,averbap-pearsinfewconstructionsfrequentlyandinmostconstructionsinfrequentlyifatall.TheobservationofVerbIslands,thatverbstendtocombinewithoneorfewelementsoutofalargerange,isinfactcharac-teristicofafullyproductiveverbalsyntaxsystem.Asfarasweknow,thequantitativepredictionsoftheVerbIslandHypothesishavenotbeenspelledoutbutwemayestimatethenecessaryamountoflanguagesamplethatwouldmasktheseislandeffects.Theappealtounevennessofverbalconstructionfrequenciesseemstoreecttheexpectationthatunderfullproductivity,mostverbsoughttoappearwithmostofthepossiblerangeofarguments.Substitutingnounsanddeterminersforverbsandnominals,theformalanalysiscouldbecarriedoutfortheverbalsyntacticsystem.Insteadofcalculatingtheexpectednumbersofdeterminersthatanounappearswith,onewouldcalculatetheexpectednumberofobjectsaverbappearswith. 8Theseverbsare:put,tell,see,want,let,give,take,show,got,ask,makeeat,like,bringandhear.Thefrequencytalliesofthetop10mostfrequentconstructionsare1904,838,501,301,252,189,137,109,88,and75.16 (Jelinek1998);then-gramandruledistributionsdiscussedinsection2.2makethesepointsveryclearly.Forthelinguist,theZipannatureoflanguageraisesimportantquestionsforthedevelopmentoflinguistictheories.First,Zipf'slawhintsattheinherentlimitationsinapproachesthatstressthestor-ageofconstruction-specicrulesorprocesses(e.g.,Goldberg2003,Culicover&Jackendoff2005).Forinstance,thecentraltenetsofConstructionGrammarviewsconstructionsasstoredpairingsofformandfunction,includingmorphemes,words,idioms,partiallylexicallylledandfullygenerallinguis-ticpatternsandthetotalityofourknowledgeoflanguageiscapturedbyanetworkofconstructions(Goldberg2003,p219).YettheZipandistributionoflinguisticcombinations,asillustratedinFigure3fortheWallStreetJournalandFigure4forchilddirectedspeech,ensurethatmostpairingsofformandfunctionsimplywillneverbeheard,nevermindstored,andthosethatdoappearmaydosowithsufcientlylowfrequencysuchthatnoreliablestorageanduseispossible.Second,andmoregenerally,Zipf'slawchallengestheconventionalwisdomincurrentsyntacticthe-orizingthatmakesuseofahighlydetailedlexicalcomponent;therehavesuggestionsthatallmattersoflanguagevariationareinthelexiconwhichinanycaseneedstobeacquiredforindividuallanguages.Yettheeffectivenessoflexicalizationingrammarhasnotbeenfullyinvestigatedinlargescalestudies.However,usefulinferencescanbedrawnfromtheresearchonstatisticalinductionofgrammarincom-putationallinguistics(Charniak1993,Collins2003).Thesetaskstypicallytakealargesetofgrammaticalrules(e.g.,probabilisticcontextfreegrammar)andndappropriateparametervalues(e.g.,expansionprobabilitiesinaprobabilisticcontextfreegrammar)onthebasisofanannotatedtrainingdatasuchastheTreebankwheresentenceshavebeenmanuallyparsedintophrasestructures.Theperformanceofthetrainedgrammarisevaluatedbymeasuringparsingaccuracyonanewsetofunanalyzedsentences,therebyobtainingsomemeasureofgeneralizationpowerofthegrammar.Obviously,inducingagrammaronacomputerishardlythesamethingasconstructingatheoryofgrammarbythelinguist.Nevertheless,statisticalgrammarinductioncanbeviewedasatoolthatexploreswhattypeofgrammaticalinformationisinprincipleavailableinandattainablefromthedata,whichinturncanguidethelinguistinmakingtheoreticaldecisions.Contemporaryworkonstatisticalgrammarinductionmakesuseofwiderangeofpotentiallyusefullinguisticinformationinthegrammarformalism.Forinstance,anphrasedrinkwatermayberepresentedinmultipleforms:(a)VP!VNP(b)VP!VdrinkNP(c)VP!VdrinkNPwater(a)isthemostgeneraltypeofcontextfreegrammarrule,whereasboth(b)and(c)includeadditionallexicalinformation:(b)providesalexicallyspecicexpansionruleconcerningtheheadverbdrink,andthebilexicalrulein(c)encodestheitem-specicpairingofdrinkandwater,whichcorrespondstothenotionofsentenceframeinTomasello'sVerbIslandhypothesis(1992;seesection4.2).Byincludingorexcludingtherulesofthetypesaboveinthegrammaticalformalism,andevaluat-ingparsingaccuracyofthegrammarthustrained,wecanobtainsomequantitativemeasureofhowmucheachtypeofrules,fromgeneraltospecic,contributestothegrammar'sabilitytogeneralizetonoveldata.Bikel(2004)providesthemostcomprehensivestudyofthisnature.Bilexicalrules(c),similartothenotionofsentenceframesandconstructions,turnouttoprovidevirtuallynogainoversimplermodelsthatonlyuserulesofthetype(a)and(b).Furthermore,lexicalizedrules(b)offeronlymodestimprovementovergeneralcategoricalrules(a)alone,withwhichalmostallofthegrammar'sgeneral-izationpowerlies.ThesendingsarenotsurprisinggiventheZipannatureoflinguisticproductivity:18 Chang,F.,Lieven,E.,&Tomasello,M.(2006).Usingchildutterancestoevaluatesyntaxacquisitional-gorithms.InProceedingsofthe28thAnnualConferenceoftheCognitiveScienceSociety.Vancouver,CanadaChomsky,N.(1958).ReviewofLangagedesmachinesetlangagehumainbyParVitoldBelevitch.Lan-guage,34(1),99-105.Chomsky,N.(1965).Aspectsofthetheoryofsyntax.Cambridge,MA:MITPress.Chomsky,N.(1975).Reectionsonlanguage.NewYork:Pantheon.Chomsky,N.(1981).Lecturesongovernmentandbinding.Dordrectht:Foris.Crain,S.(1991).Languageacquisitionintheabsenceofexperience.BehavioralandBrainSciences.14,597-650.Culicover,P.&Jackendoff,R.(2005).Simplersyntax.NewYork:OxfordUniversityPress.Freudenthal,D.,Pine,J.M.,Aguado-Orea,J.&Gobet,F.(2007).ModellingthedevelopmentalpatterningofnitenessmarkinginEnglish,Dutch,GermanandSpanishusingMOSAIC.CognitiveScience,31,311-341.Freudenthal,D.,Pine,J.M.&Gobet,F.(2009).SimulatingthereferentialpropertiesofDutch,GermanandEnglishrootinnitives.LanguageLearningandDevelopment,5,1-29.Gabaix,X.(1999).Zipf'sLawforCities:AnExplanation.TheQuarterlyJournalofEconomics.114,739-767.Goldberg,E.(2003).Constructions.TrendsinCognitiveScience,7,219224.Ha,LeQuan,Sicilia-Garcia,E.I.,Ming,Ji.&Smith,F.J.(2002).ExtensionofZipf'slawtowordsandphrases.Proceedingsofthe19thInternationalConferenceonComputationalLinguistics.315-320.Hay,J.&Baayen,H.(2005).Shiftingparadigms:gradientstructureinmorphology.TrendsinCognitiveSciences,9,342-348.Jelinek,F.(1998).Statisticalmethodsforspeechrecognition.Cambridge,MA:MITPress.Kucera,H&Francis,N.(1967).Computationalanalysisofpresent-dayEnglish.Providence,RI:BrownUniversityPress.Legate,J.A.&Yang,C.(2002).Empiricalreassessmentsofpovertystimulusarguments.LinguisticReview,19,151-162.Li,W.(1992).RandomtextsexhibitZipf'slaw-likewordfrequencydistribution.IEEETransactionsonInformationTheory,38(6),1842-1845.MacWhinney,B.(2000).TheCHILDESProject.LawrenceErlbaum.Mandelbrot,B.(1954).Structureformelledestextesetcommunication:Deuxétudes.Words,10,127.Matthews,D.,Lieven,E.,Theakston,A.&Tomasello,M.(2005).TheroleoffrequencyintheacquisitionofEnglishwordorder.CognitiveDevelopment,20,121-136.McNeill,D.(1963).Thecreationoflanguagebychildren.InLyons,J.&Wales,Roger.(Eds.)Psycholin-guisticPapers.Edinburgh:EdinburghUniversityPress.99-132.Miller,G.A.(1957).Someeffectsofintermittentsilence.TheAmericanJournalofPsychology,70,2,311-314.20 Zipf,G.K.(1949).Humanbehaviorandtheprincipleofleasteffort:Anintroductiontohumanecology.Addison-Wesley.22
© 2021 docslides.com Inc.
All rights reserved.