Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
DOMINIQUESPORTICHE
UCLA ECOLENORMALESUPRIEURE,PARIS INSTITUTJEANNICOD
BERKELEYLINGUISTICSOCIETY February,12th2012
1
Humans exhibit a uniquely fine-tuned behavior in the linguistic domain. Studying language to understand human individual cognition is perhaps the founding question of generative grammar. This uniquely fine-tuned behavior seems unique to our species. It is thus not unreasonable to study language with the perspective of uncovering features of human language cognition that sets it apart from other human cognitive properties, and features of language cognition that sets humans apart from other species. While it is important to stress that generative grammar has no a priori view about the answers to these questions, although it may have hypotheses about them, this quest about language and human singularity also contributes to defining the research project generative grammar has pursued for the past half century. It is tempting to assume that, pathological cases aside, such features would be shared by all humans, hence all languages. This is why one way to undertake this kind of study is to look first for linguistic universals, and any reasonable method is a priori fine. For example, exploring superficial cross linguistic similarities, e.g. perceptually salient properties such as word order, could yield categorical universals or universal tendencies. Even if it did it is not clear one way or the other for any superficial property - it would be hard on that basis to conclude much with respect to the questions above. Indeed, the language faculty, the capacity to develop language, whatever it may be, can be seen as a function that maps, for individuals, data a language learner is exposed onto his/her final system. Whatever properties this final system exhibits, they may in fact reflect properties of the input data that are accidental, shaped by historical events or other contingencies or properties that are necessary because of physical laws or other irrelevant impositions from the point of view of the questions at hand; a further argument would be needed to show their relevance to these very questions. Conversely, finding no such properties by sampling these salient properties is not telling either. Indeed, universal features of this mapping function may require, in order to express themselves, specific triggers in the learners available data which may be contingently missing. This has been well understood for some time. Another way, is to try to study deeper, hidden or highly intricately structured properties of natural languages, properties that are not perceptually salient, seem not to be grounded in perceptual regularities, seem not to be
convergent solutions of general algorithms (e.g. of a statistical sort), that is, properties that have a good chance of not being inferable by this mapping function from the learners data. Such properties are, a priori, better candidates to lead to the type of features alluded to above. Finding contingent (although perhaps biologically not so) such properties, whether they are shared by many or all languages (say like some common genetic trait) should be very telling. If such properties are found in a very small sample of languages, nay only a single one (like an extremely rare genetic condition), it should be very telling too (think of the role played by the study of abnormalities, whether of the brain or the rest of the body, in understanding the normal). This has also been well understood for some time. This is in part why generative linguists have tried in the last half century to initially go deep and narrow rather than wide and shallow (and in the long run deep and wide of course) sometimes being criticized of anglocentrism - to discover such hidden properties, to address questions of singularity. And research has uncovered several such systems (binding theory, polarity item licensing, the theory of presupposition, constraints on movement, etc...) which exhibit a high degree of internal complexity. In this article, I outline the properties of one such a complex system, called Binding Theory, based on English and French data. As it turns out, many of the properties I will be discussing are known to hold at least quite widely crosslinguistically, and perhaps do universally.
In a recent article from 2010 published in Cognitive Science Language Acquisition Meets Language Evolution, They sketch a range of alternative approaches that provide a *promising starting point* [my emphasis, DS] for understanding binding as arising from domain-general factors. They provide in fact suggestions which do not even begin to address very elementary properties of the English binding system. This is recognized even by staunch proponents of treatments in terms of pragmatic principles which are thought of as rational solutions to problems of communicative coordination such as Levinson (1991) who states (p. 108) that computation of pragmatic inferences of these sorts must be made over a level of linguistic representation which includes rich configurational, lexical and semantic information.
Theproblem: 1a.Johnlikeshimself 1b.*Johnlikeshim 2a.*Johnsmotherlikeshimself 2b.Johnsmotherlikeshim 3a.*JohnsaysthatMarylikeshimself 3b.JohnsaysthatMarylikeshim 4a. JohnsaysthatMarylikesSueandhimself 4b.JohnsaysthatMarylikesSueandhim
Twohardquestions: Question#1:Thereversepatternisunattestedand seemsruledoutinprinciple(selfanaphorizesthe pronounasveryoftennotedbut)why? [Explainingwhyintensification/focalizationleadsto anaphoricitycf.Charnavelsforthcoming dissertation] #2:Grantingthepattern,howshoulditbedescribed (andultimatelyexplained,thatisreducedtosimpler things)
THEPLAN: Remindyouofexistinganswerstoquestion#2andgive youasenseofwhytheseanswerswereadopted. ProvidenewdatamostlyfromFrench(butalsosome fromEnglish)supportingsomeofthemostinfluential answersandchallengingothers. CONCLUSION(forFrenchandEnglish):Three(atleast) ingredientsareneeded Adichotomybetweenanaphorsandexemptanaphors ConditionAoftheBindingtheorylaChomsky(1986) Aspecialsystemforsubjectorientedanaphors(reflexive voice)
8
ClassicalTheory:(frompronominalization/reflexivizationin thelate1960stotheBindingtheoriesofthemid1980s) ConditionA:areflexiveRmustbeboundinD(R) ConditionB:apronounPmustbenotboundinD(P) IfD(R)=D(P)complementarydistribution. Isthiscorrect?[Notoverall(butpossiblyoverasubsetof syntacticpositions)] Wheredoesitcomefromifandwhenitiscorrect? Naturalidea:Competitition onesentenceisillformedbecausetheotheriswellformed.
9
(Pure)PragmaticReasoning(e.g.Levinson1991)
(notarealsynchronictheory,ratherasketchofhowsuchsystemscouldhavearisen basedonpragmaticreasoningandgottengrammaticizedalthoughhowexactlyit wouldhavecometosettleonthecurrentsystemisunclear)
10
Predictions:
Thereshouldalwaysbeawinneranditshouldbeunique. Overrideshouldbepossibleifthereisameaningreason (whichcouldmakethenormalloserpossible: Implicaturesarecancellable)
Problems
Caseswithnowinner Caseswithnoloser(eventhoughthereisnodifference) Caseswithnooverridewhereitwouldmakesense(and noescapefromstructures:attheveryleastsensitivityto quiteabitofgrammaticalstructures)
11
Caseswithnowinner(somefurtherconstraints) 1) JeanetMarieconnaissentelle/JohnandMarylikeher 2) JeanetMarielaconnaissent/JohnandMarylikeher 3) JeanetMarieseconnaissent/JohnandMarylikeherself Caseswithnoloser(eventhoughthereadingwiththe pronounislessspecificthanthereadingwiththe reflexive)Difficultyforpragmaticapproach 4) Maryputitnexttoher/herself 5) Maryreadstoriesabouther/herself 6) MarysaidthatJohnhadinvitedeveryonebuther/butherself
12
Withsmallerconstituentsthanclauses: (nosimplefixlimitingimplicaturecomputationtopropositions) 9)With[BillkafraidofhiskmotherandJohntoo],weareintrouble 10) With[BillkafraidofhimselfkandJohntoo],weareintrouble 11) With[BillkafraidofhimkandJohntoo],weareintrouble 12) Both[BillkspicturesofhiskmotherandJohns]arenice 13) Both[BillkspicturesofhimselfkandJohns]arenice 14) Both[BillkspicturesofhimkandJohns]arenice Conclusion: Pureimplicaturebasedsystemsareunpromising(butattheveryleast theyshouldbesensitivetoquiteabitofsyntacticstructures)
14
PureGrammaticalcompetitition:
Theremaywellbenowinneratall Butthereshouldbenomorethanonewinner ProblemswithGrammaticalcompetition:
15) 16) 17) Caseswithnowinner(somefurtherconstraints) JeanetMarieconnaissentelle/JohnandMarylikeher JeanetMarielaconnaissent/JohnandMarylikeher JeanetMarieseconnaissent/JohnandMarylikeherself Caseswithnoloser(whetherornotthereisameaningdifference) Maryputitnexttoher/herself Maryreadstoriesabouther/herself MarysaidthatJohnhadinvitedeveryonebuther/herself
Earlyversionsofbinding(purecompetition)abandoned
15
2.Fundamentalfacts:
Caseswithnowinners:suggestssomeintrinsic(grammatical) constraints Caseswithnoloser:overalllackofcomplementarydistribution pronouns/reflexives
Twowaystofixtheclassicaltheory
16
Butweknowthisisgoingtobeinsufficientbecauseof contrastlikethese:
#1MarysaidthatJohnhadinvitedherself #2MarysaidthatJohnhadinvitedeveryonebutherself
SecondWay:PREDICATEBASEDTHEORIES
PollardandSag(1991)/ReinhartandReuland(1989,1993): (somedifferencesbetweenthem) Positionssubdivideintwodisjointsubsets InSUBSET#1(grammaticallydefined):complementarity betweenanaphorsandpronounsANDnoexemptanaphors InSUBSET#2(=complementofSUBSET#1):No complementaritybetweenanaphorsandpronounsAND exemptanaphors GENERALIDEA:Lackofcomplementarityarisesbecause exemptanaphorsdonotcompetewithpronouns
18
SUBSET#1:IfRorPisasyntacticargumentofsome predicateandhascoarguments:
PredicateBasedConditionA(PBCA): Rmusttakeahigher(e.g.ccommanding)coargumentasantecedent. ConditionB:apronounwiththesameantecedentisdisallowed Rulesin1ahencerulesout1b 1a.Johnlikeshimself 1b.*Johnlikeshim Rulesout2aand3a,hencerulesin2band3b 2a.*Johnsmotherlikeshimself 2b.Johnsmotherlikeshim 3a.*JohnsaysthatMarylikeshimself 3b.JohnsaysthatMarylikeshim
19
20
Regularanaphors(notinexemptpositions): Mustbeboundbyasyntacticcoargument NoConstraintsoncontentormeaning Anaphorsinexemptpositions: Noccommandrequirement AllowLongDistanceAntecedents Constraintsonthecontentofantecedents?Yes,but variableandillunderstood,e.g. o Humanantecedent,consciousantecedent o Antecedentbealogophoriccenter o Anaphorbereaddese o Antecedentcontrasted(?)
21
3.TwosetsofProblems Problemset#1:Thereareanaphorsinexemptpositionsbut subjecttostrictlocality:PBCAistooweak Charnavel:inanimatesonpropre(hisown)andluimme (himeven=himself) Problemset#2:Anaphorsnotinexemptpositionsbut ConditionAviolated(PBCAistoostrong) Charnavelonsonpropre(sapropredescriptiondePierre) ZribiHertzforEnglishandforFrench KaiserandRunnerexperimentalevidence ReinhartandReuland:FocusAnaphors
22
23
Charnavelsfindings:
TheFrenchexpression[sonpropreN]e.g.sonproprepre(hisownfather)is threewaysambiguous Plainpropre:sonpropreprecontrastswithsonprewhichisnothisown (e.g.nothisownfather,hisadoptedfather) PossessumPropre(mainstressonNoronpropre):inducesalternativeson theNwithscalarityeffects(=evenhisownfather) Possessorpropre(mainstressonpropreonly):inducesalternativesonthe possessor(hisfatherasopposedtosomeoneelses).
24
Examples:
Ccommandrequired 1) a.Ceproblmeinclutsa(propre)solutionetcelleduproblmeprcdent. 'Thisproblemincludesits(own)solutionandthatofthepreviousproblem.' b.Lesannexesdeceproblmeincluentsa(*propre)solutionetcelleduproblme prcdent. 'Theappendicesofthisproblemincludeits(*own)solutionandthatofthe previousproblem.'
Clausematenessrequired
2) a. Cette auberge fait de l'ombre son (propre) jardin et au jardin de la maison voisine. 'This inn gives shade to its (own) garden and to the garden of the neighboring house.' b.Cetteaubergebnficiedufaitqueson(*propre)jardinestplusspacieuxque celuidesaubergesvoisines. 'Thisinnbenefitsfromthefactthatits(*own)gardenismorespaciousthanthatof theneighboringinns.'
26
InterveningSubjectProhibited
3) a. Cette peinture possde ses (propres) composants et des composants plus communs. 'This paint includes includes its own components and more common components.' b.Cettepeinturearendulesouvriersallergiquesses(*propres)composantset ceuxd'unautretypedepeinturesimilaire. 'This paint made the workers allergic to its (*own) components and to those fo anothertypeofsimilarpaint.' 4) a.Cetteentreprisesuscitel'admirationdesemployspourson(*propre)patronet leurcolrecontrelespatronsconcurrents. 'This company arouses the admiration of the employees for its (*own) manager andtheirangeragainstthecompetingmanagers.' b.Cetteentreprisesuscitel'admirationdeson(propre)patronetlacolrecontre lespatronsconcurrents. 'Thiscompanyarousesadmirationforits(*own)managerandangeragainstthe competingmanagers.'
27
SIDENOTE: TheFrenchdatawascollectedbyIsabelleCharnavelusingquantitatively controlledmethods.Thequestionnaireusedinthisstudywasaccessibleona dedicatedwebsite.19Itconsistedof120targetsentences.Theyincludedtwo mainvariables: natureofthepossessive(son/sonpropre) animacyoftheantecedent(animate/inanimate), whichcreated4conditions: A:animateson B:inanimateson C:animatesonpropre D:inanimatesonpropre Moreover,thefourconditionswererepresentedin9differentsyntactic contexts. TheOK/*distinctionreportedherallcorrespondtostatisticallysignificant differences.
28
29
30
31
32
sonpropre
ZribiHertz(1989,1995)(forEnglishand)forFrench
Longdistancehimself.(cf.ZribiHertz,1989,Baker,1995) 11) Johnicouldn'tresistthehungerforrevengewhichfilledhimselfi Longdistanceluimme.(cf.ZribiHertz,1995) 12) L'avenirdeMarienedpendpasd'ellemme,maisdesesparents. 'Mary'sfuturedoesnotdependonherself,butonherparents.'
33
Runneretal.(2003,2006)experimentalevidence
13) JoesawKenspictureofhimself
ReinhartandReuland(1993):FocusAnaphorsareexempt
As focus, a free SELF anaphor can occur even in an argument position at SStructure. Though such examplesarehardertofindandaremoremarked
14) a.Thisletterwasaddressedonlytomyself b.Whyshouldthestatealwaystakeprecedenceovermyself? c. "Bismarck's impulsiveness has, as so often, rebounded against himself." (quotedinZribiHertz1989)
ParticularlyProblematicinthatinanimateluimme(orpossessorsonpropre)can (must?)befocusedbutstillarenotexempt
34
CONCLUSIONS
1.Necessarytodistinguishexemptanaphorsfromplainanaphors 2.Inspectionofthepositionofananaphordoesnotdetermine whetheritisexemptornot:thesamepositioncanyieldnonexemptor exemptanaphors 3.Inspectionofthecontentofananaphordoesnotdeterminewhether itisexemptornot:thesameanaphorinagivenpositioncanbeexempt ornot 4.Whenwecantell(inanimates),whatdetermineslocalityofbinding isnotbeingargumentofapredicate,itisanotionofsyntacticdomain.
A(nonexempt)anaphor mustbeaclausemateofitsantecedent Neednotbeacoargumentofitsantecedent CantbeseparatedfromitsantecedentbyanInterveningsubject
35
37
Frenchluimme:anaphor,exemptornot
1) Jean(pensequeMarie)parledeluimme Jean(thinksthatMarie)istalkingabouthimself 2) JeanmontreMarieellemme JeanisshowingMarietoherself 3) *Jeanexamineluimme (Unlessheavilystressed) Jeanisexamininghimself 4) Jeansestexamin JeanREFLisexamining 5) *JeanpensequeMarieexamineraluimme(unlessheavilystressed) JeanthinksthatMariewillexaminehimself IfReflexiveVoiceisallowed(onlyforsubjectorientedanaphors),itmustbeused unlessheavystressispresent(doesrecalloneaspectofRRstheory,butonlyfor subjectorientedanaphorsthatcancliticize)
38
Yes,butinEnglish??
39
TwoPuzzles:Ahn(2010,2011,forthcomingdissertation) Puzzle#1: 19) a. WhosawJohn? b. BILLsawJohn c. JOHNsawJohn d. JohnsawHIMSELF prosodicallyanomalous QuestionAnswerCongruence:AnappropriateanswertoaWHquestionmustbe (semanticallyandprosodically)focused. 20) a. WhodidMaryshowBillto? b. #MarySHOWEDBilltohimself c. #MaryshowedBILLtohimself d. MaryshowedBilltoHIMSELF
Nothingwrongwithstressingareflexive.
40
Puzzle#2
Whathappened? 1) JohncutBILL JeanacoupPIERRE NormallyDefaultSententialStressshouldfallonthedirectobject(themostdeeply embeddedelement) 2) #JohncutHIMSELF #PierreacoupluiMME 3) JohnCUThimself PierresestCOUP 4) MarydescribedBilltoHIMSELF MarieadcritPierreluiMME 5) #MarydescribedBILLtohimself #MarieadcritPIERREluimme
41
]]]
Puzzle#1:ReflexiveVoiceisnewinformationandthisiswhatisstressed Puzzle#2:ThereflexiveissyntacticallytoohighforDSS:thelowestelementisactually
42
43
THANKYOU
44
APPENDIX:
Moreexamplesof:Clausematinessrequired(variouscases)
1) a.Cetteaubergefaitdel'ombreson(propre)jardinetaujardindelamaisonvoisine. 'Thisinngivesshadetoits(own)gardenandtothegardenoftheneighboringhouse.' b. Cette auberge bnficie du fait que les touristes prfrent son (*propre) jardin ceux des auberges voisines. 'Thisinnbenefitsfromthefactthatthetouristspreferits(*own)gardentothatoftheneighboringinns.' 2) a. Cette montagne est moins rpute pour son (propre) sommet que pour le sommet voisin auquel elle donneaccs. 'Thismountainislessrenownedforits(own)summitthanfortheneighboringsummititgivesaccessto.' b.Cettemontagneattirebeaucoupdegensparcequeson(*propre)sommetestl'undessommetslesplus escarpsdupays. 'This mountain attracts many people because its (*own) summit is one of the steepest summits in the country.' 3) a.Ceproblmeinclutsa(propre)solutionetcelleduproblmeprcdent. 'Thisproblemincludesits(own)solutionandthatofthepreviousproblem.' b.Ceproblmeprsentepeudedifficultspourqueleslvespuissenttrouversa(*propre)solutionplus rapidementquecelledesproblmesprcdents. 'Thisproblempresentsfewdifficultiessothatthestudentscanfindits(*own)solutionmorequicklythan thatofthepreviousproblems.'
45
4) a.Cettedfaitesupplmentaireaentransespropresconsquences. 'Thisadditionaldefeatentaileditsownconsequences.' b.Cettedfaitesupplmentaireapoussleshabitantssupporterses(*propres)consquencesenplus decellesdeloccupation. 'This additional defeat led the inhabitants to endure its (*own) consequences on top of those of the occupation.'
MoreexamplesofInanimateluimmeinexemptpositions(adjunct,only
argumentofDP,PP)withalocalantecedentrequired.
1) a.Cetteatrophietuetout,hormisellemme. 'Thisatrophykillseverythingexceptitself.' b. *Cette atrophie est trs dangereuse parce que rien ne rsiste ce genre de problme hormis elle mme. '*Thisatrophyisverydangerousbecausenothingresiststothiskindofproblemexceptitself.' 2) a.Cesrsultatsontdelavaleureneuxmmes. 'Theseresultshavevalueinthemselves.' b. *Ces rsultats sont prometteurs mme si le chercheur ne reconnat pas encore de valeur en eux mmes. '*Theseresultsarepromisingeveniftheresearcherdoesnotrecognizevalueinthemselvesyet.'
46
MoreExamplesofnopartialbinding
3) c.[L'coleetlesmaisonsdesinstituteursontchacuneunjardin.]L'colefaitdel'ombre leurs(*propres)jardins,maispasaujardindelamairie. '[Theschoolandtheteachers'houseshaveeachagarden.]Theschoolgivesshadeto their(*own)garden,nottothegardenofthetownhall.'
47
Interpretivepropertiesoflogophoricsonpropre/luimme
Consciousvs.unconsciousantecedentofsonpropre(pronounsmaybeusedinboth cases). 6) a.Lesembaumeursdupharaonprennentsoindeson(propre)preetdupredeson pouse. 'ThePharaoh'sembalmerstakecareofhis(own)fatherandhiswife's.' b.Lesembaumeursdupharaonprennentsoindeson(*propre)corpsetdeceluide sonpouse. 'ThePharaoh'sembalmerstakecareofhis(own)bodyandhiswife's.' 7) a.Lepharaonctoiesouventlesembaumeursquiprendrontsoindeson(propre) corpsetducorpsdesonpouse. 'ThePharaohoftenmeetstheembalmersthatwilltakecareofhis(own)bodyand hiswife's.' b.Lepharaonctoyaitsouventlesembaumeursquiprsentprennentsoindeson (*propre)corpsetducorpsdesonpouse. 'ThePharaohoftenmettheembalmersthatarenowtakingcareofhis(*own)body andhiswife's.'
48
Discourseparticipationinattitudecontexts 8) a.PierreditqueMarieetLiseaimentbienses(propres)photos,maispascellesde sonpouse.(source) 'PetersaysthatMaryandLizlikehis(own)pictures,butnotthoseofhiswife.' b.MarieetLiseontditPierrequ'ellesaimaientbiensespropresphotos,maispas cellesdesafemme.(addressee) 'MaryandLiztoldPeterthattheylikehis(own)pictures,butnotthoseofhiswife.' c.MarieetLiseontditauxvoisinesdePierrequ'ellesaimaientbienses(*propres) photos,maispascellesdesafemme.(nondiscourseparticipant) 'MaryandLiztoldtheneighborsaboutPeterthattheylikehis(own)pictures,but notthoseofhiswife.' d.MarieetLiseontditdePierreauxvoisineslesquellesdesespropresphotoselles aimaient,maisn'ontpasparldecellesdesafemme. 'MaryandLiztoldtheneighborsaboutPeterwhichonesofhis(own)picturesthey like,butdidnottalkaboutthoseofhiswife.'
49
50
Evidenceforreconstruction 10) a. Marie a dit de Pierre aux voisines quelle photo de lui mmeLisevoulait. #C'taitenfaitunepeinturelaquelleLisepensait. 'MarytoldtheneighborsaboutPeterwhichpictureofhimself Liz wanted. #In fact, it was a painting that Liz was thinking about.' b. Pierre a dit aux voisines quelle photo de luimme Lise voulait.C'taitenfaitunepeinturelaquelleLisepensait. 'PetertoldtheneighborswhichpictureofhimselfLizwanted. Infact,itwasapaintingthatLizwasthinkingabout.'
51
b.whetheranelementissubjecttoitcannotbedeterminedonthebasisofthecontentofthiselementaloneinspectionoftheantecedentis necessary (soRRonchainsisnotgoingtowork:toorestrictive) (discussRRconditiononchains) RRdonotenforceccommand.RathertheyrequireofifbindingisimposedbyConditionA,theset(antecedent,reflexive)satisfyalocality conditioninwhichtheantecedenthastobehigher(whichlooksverymuchliketheoldconditionA) However:thisconditionwouldhavetoholdonlyofinanimatesonpropreorluimme(butthisspendsontheantecedent). 5.CanweshowthatanimatesaresubjecttoConditionA? Possibilyifwecontrolforviewpointetc..: Out? Why? Johnsmotherreadstoriesabouthimself
52