historical syntax and the generative paradigmlinguistica.sns.it/rdl/20.2/gianollo.pdf · on the...

HistoricalSyntaxandtheGenerativeParadigm

ReviewarticleofDavid lighTfooT 2006. How New Languages Emerge. Cambridge:CambridgeUniversityPress.Ian roBerTs 2007. Diachronic Syntax. Oxford: Oxford UniversityPress.

ChiaraGianollo

1. Introduction

The two monographs under review are recent contributions byeminentscholarsinhistoricallinguistics,bothwrittenwiththeintentof summarizing the main results and the new perspectives offeredby decades of formal studies in diachronic syntax. Treating the twobooks inparallel,however, calls foraproviso, as theyare explicitlyconceivedfortwoquitedifferentkindsofaudience.

Lighfootexploresthecontributionofgenerativehistoricalsyntaxto the study of change within human communities, by focusing onthedevelopmentofwhat theauthorbelieves tobe “amoresophisti-cated analysis of history and change” than what has been proposedbyevolutionaryanddevelopmentalbiologistsandpoliticalhistorians(Lightfoot 2006: viii). His intent is to make it accessible to scholarswithinterestsattheinterfacewithlinguistics(anthropologists,soci-ologists,psychologists,neuroscientists).

Ontheotherhand,Robertshasdesignedacomprehensivetext-bookintroducingthedisciplineofdiachronicsyntaxtostudentsinlin-guistics,whoapproachforthefirsttimetheformalstudyofsyntacticchangeand,possibly,generativesyntactictheoryitself.

Themodesofexpositionofthetwobooksvaryaccordingly,asdothe choice of arguments and the space allotted to many topics theyshare. Indiscussing the import of the twovolumesunder review, itisofcourseimportanttokeepinmindtheauthors’basicdifferenceofintent.However,suchadifferencecannotpreventonefromnoticingasignificant similaritywhichhopefullymakesaparallel treatmentofthe twomonographsworthpursuing:bothareessentiallymotivatedby the major growth witnessed by diachronic syntax over the pastfew decades. This growth, unexpected as it was at the beginning oftheSeventies,hastransformedthefieldintooneoftheliveliestareas

Rivista di Linguistica19.2(2007),pp.347-373 (ricevuto nell’aprile 2008)

Chiara Gianollo

348

of contemporary linguistic theory, which has obtained a place of itsown in many undergraduate and graduate curricula in linguistics.Inaddition,itisconstantlyinteractingwithmanyotherneighboringdisciplines,therebyofferingvaluableinsightsforresearchonculturalvariationand change.Anoverviewof theprogressing efforts of dia-chronically-mindedgenerative linguistscanbegrasped fromanum-berofvolumeswhichcollect theworkdiscussedduring thebiennialDIGS (Diachronic Generative Syntax) meetings (Battye & Roberts1995, Van Kemenade & Vincent 1997, Pintzuk, Tsoulas & Warner2000,Lightfoot2002,Battlori,Hernanz,Picallo&Roca2005,Jonasto appear,Crisma&Longobardito appear).

Apparently both Lightfoot and Roberts felt that, as in anyquickly-developing field, the moment had come to offer a synthesis,highlighting,ontheonehand,thecontributionofhistoricalsyntaxtolinguistictheoryitselfandtootherconnectedscientificareas,butalsopinpointingtheprioritiesandchallengesahead.

The exposition will proceed as follows: first I will summarizethebasic linesofagenerativeapproach todiachronicsyntax,whichrepresent the common background for the two studies (section 2).Then,aftergivingashortoverviewofeachvolume(section3),Iwillconcentrate on some issues which appear to be particularly salientinevaluatingthepastandfutureroleofhistoricalstudiesofsyntaxandwhichreceiveaninsightfulandofteninnovativetreatmentinthetwo books (section 4). Sometimes the two authors will be shown toholdquitedifferentpositions, and the comparisonwill target issueswhich combined synchronic and diachronic research may hopefullyenlighten.

2. Diachronic syntax in a generative perspective

The central tenet in both books, strongly in compliance withthe Chomskyan view about language, is that language change isbest understood as change in individuals, in the internal system ofgrammatical knowledge arising through first-language acquisition(internal, I-language). Such connection between diachronic changeandlanguageacquisitionwasfirstcastingenerativetermsbyDavidLightfoot’s (1979) influential monograph, Principles of diachronic syntax.Languagechangecametobeseenasan inherentpossibilityarising, given certain circumstances, during the process of acquisi-tionofgrammaticalstructures,currently interpretedasaprocessofparametersetting.Anewgenerationmayconvergeoveraparameter

Historical Syntax and the Generative Paradigm

349

valuedifferentfromthatofthepreviousgeneration,oncethetrigger-ingexperience,i.e.thecorpusofprimarydatachildrenareexposedto(theexternal,E-language),changessignificantlyfromonegenerationtotheother.

Lightfoot(1979)viewedlanguagechangeas“afunctionofchanceand necessity” (a formula which deliberately echoed Monod’s (1970)famousessayongenetictransmission).Chanceresidesinoscillationsinthetriggeringexperience,yieldedeitherbytheexistenceofearlierchanges in the grammar or by extra-grammatical factors (contact-induced borrowing, massive imperfect second-language acquisitionofthetargetgrammar,desireforexpressivityandconsequentvaria-tion in frequency of a given construction). Necessity is induced bya series of universal, biological characteristics of human language,namely “that the grammar should not allow excessive opacity, thatsurfacestringsshouldbeprocessedwithminimalperceptualdifficul-ty,andthatgenerationsshouldmaintainmutualcomprehensibility”(Lightfoot1979:396).Insum,languagechangeisinterpretedastrig-geredbylocalcauses,eitherinternalorexternaltothegrammaticalsystem, and carried out during the stage of language acquisition, aprocesswhichatthesametimeensuresconservativity(mutualintel-ligibility),eliminationofopacity,andobediencetoinnaterestrictionsontheformofUniversalGrammar.

For the first time, the understanding of language change cameto be strictly tied to the elaboration of restrictive theories of gram-mar, which could define the limits of variation and, thus, the pos-sible outcome of change given a certain triggering experience. In1979,Lightfootheldtheopinionthatthesecondaryroleofthestudyofdiachronicsyntax inhistorical linguisticresearchwas“a functionof inadequate theoriesofsynchronic syntaxon thepartofneogram-marians, American structuralists and transformational generativegrammarians alike” (Lightfoot 1979: vii). In this respect, he consid-ered work on word-order implicational universals originating fromGreenberg’s(1963)researchtobeextremelypromising.And, infact,the real takeoff of generative diachronic linguistics has been deter-minedbytheriseof thePrinciples&Parametersapproachtosyntac-tic variation, since its first formulation in Chomsky’s Pisa Lectures(Chomsky1981).

The Principles&Parameters framework offers to historical lin-guistics thepossibilityofoperatingwithanextremelypowerful toolintheinvestigationofgrammaticalchange,and,inparticular,intheexplanationofits‘bumpiness’,i.e.thefrequentoccurrenceofclustersof changes appearing simultaneously in a given language, which

Chiara Gianollo

350

wereconsideredalreadybyLightfoot(1979:402)inapre-parametriceraas“varioussurfacerealizationsofasinglechangeintheabstractgrammar”.

A parametric analysis of syntax, based on a deeply deductivetheory of language variation, helps capture the connection betweendifferentco-occurringsuperficialchangesatleastintwoways.First,it often allows embracing various apparently scattered grammaticalchanges under only one, abstract point of variation, thereby favor-ingexplanatoryapproachesoflanguagechangebasedonthestudyofacquisitionalmechanisms.Second, itoffersawaytohandleclustersofdistinctparametric changesbyproposinga theoryof thecomplexinterdependencies existing among parameter values, which mighttriggerachain-shifteffectinthelanguage.

The Principles&Parameters framework thus represents a theo-retical model which can account both for clusterings of synchronicpropertiesofgrammar(i.e.implicationaluniversals),andforcluster-ingsofdiachronicproperties.

Historical linguists working in a generative perspective areconvinced that all major kinds of syntactic change can be reframedin termsofparameter-resettingoperations,whichappear tobe “theprincipal explanatory mechanism in diachronic syntax” (Roberts2007:121).Atthesametime,theybelievethatabetterunderstand-ing of change in terms of parameter resetting has to be consideredfundamental in order to answer more general questions at the syn-chronic level of explanation, such as, among others, the nature ofparameters and parametric networks, the mechanisms of first-lan-guageacquisitionandthenatureofthetriggeringdata,theexistenceofdefaultvalues,andthelevelofgrammaticalvariabilityinlinguisticcommunities.

The interpretation of language change as a particular instanceofparametersetting,i.e.asanacquisition-drivenphenomenon,leadsgenerativehistoricallinguisticstorefuteanytheoryofchangewhichappeals toexplanations involvingmorethanonegeneration.Syntaxisseenasanessentiallyconservative,inertmodule,guaranteeingtheoverallconvergenceofthenewborngrammarswiththeparentalonesfoundinideally‘normal’situationsoflanguagetransmission(i.e.withnosubstantialmutationintheprimarydata).Syntacticchangedoesnot arise, unless it is locally caused; in Keenan’s original formula-tionoftheInertiaPrinciple,“Thingsstayastheyareunlessactedonbyanoutsideforceordecay”(Keenan1994:2).InLightfoot’sterms,“structuralchangeinI-languagesiscontingent,resultingfromchang-esinthegrammarsorintheuseofgrammarsofearliergenerations


351

thathavetheeffectofchangingtheavailabilityofgrammaticalcues”(Lightfoot2006:164).

Historycannottranscendspeakers;grammars,beingconstructedbyindividuals,cannotretain“racialmemories”(Lightfoot1979:391),and there is no space in the theory for explanatory notions such asdiachronic universals or long-term teleological changes. Lightfootentertainsquitea radical viewon thispoint,which is criticallydis-cussedbyRobertsinhisbook,aswewillseebelow(4.3).

Noprinciplesofhistoryand,thus,nopredictivetheoryofchangecan be formulated according to such premises. Lass, in his (1980)monograph, held the extreme view that language change would inprinciple not be subject to explanation, as it is not deterministic,henceunpredictable.AccordingtoLightfoot,grammaticalchangecanbeexplainedinsofarasitisconsideredtobeachangeofsomeprop-erties of the individual’s grammar from one generation to another.What cannot be generally subject to a principled explanation is theprimitivechangeinthelinguisticenvironment,intheE-language(asimilarpointismadebyLass1987,1997):“Grammatical,structuralchangesneedexplanation;butthereisnotheoryofwhytriggerexpe-riences should change, except insofar as they change as a result ofearlier structural changes” (Lightfoot 1999: 207). Once extra-gram-matical factors enter the picture, “Clio is free to play idly with herwaterclock”(ib.).

The radical disruption with a long historicist tradition broughtaboutbygenerativist thinking on language change ispatent; ithasentailed a drastic revision of many traditional treatments of long-standingissues inthestudyof languagechange,andhasprompted,during the past decades, a lively debate, which is conspicuouslyreflectedinthetwovolumesunderreviewhere.Afterashortgeneraloverviewof their respective content (3.1and3.2), Iwill concentrateontwomainaspectswhichappeartobeparticularlychallengingforthegenerativeapproachtothestudyofdiachronicsyntax:thesearchfor local causes in processes of change and the connected theory oflearnability(4.1and4.2),andthenecessityofdealingwithlong-termprocessesinthehistoryoflanguages(4.3).

3. Overview of the two volumes

3.1 Lightfoot (2006)DavidLightfoot isnotnew to the taskof offering linguistsand

the broader scientific community comprehensive monographs deve-

Chiara Gianollo

352

loping a wide-ranging integrated approach to the problem of lan-guagechange(cf.e.g.Lightfoot1982,1991,1999,and,withStephenAnderson, 2002). With his latest book, he is particularly concernedwiththe‘outsideworld’,intwodifferentsenses.

First, fromthepointofviewoftheintendedaudience,thebookisespeciallydevotedtonon-linguists,to“peoplewhohavethoughtalittle about language but who do not necessarily work on syntactictheory,whohavenoconcernwhethersyntax isminimalistorcogni-tive or unificationalist or systemic, but who might be interested intheimplicationsofthatworkforunderstandinghownewsystemscandevelop”(Lightfoot2006:viii).Forthisreason,technicalitiesarekepttoaminimumandvarioussectionssumupthehistoryofcoretheore-ticalproblemsinthestudyoflanguagehistory.

Secondly,fromamoresubstantialpointofview,Lightfoot’smaininterest shifts from the internal-language dimension to the mutualrelationshipbetweenchangeattheinternal-languagelevelandintheexternal language, two dimensions feeding each other in a dynamicinteractionwhichhasnotyetbeensatisfactorilystudied.

The book contains eight chapters, some of which represent theexpansionandtherevisionofpreviousworksbytheauthor,especial-lyLightfoot(1991)and(1999),inlightofnewresearch.

The first chapter introduces basic concepts underlying thecognitive approach to the study of language change, which is pur-sued in the rest of the volume; it represents a clear and syntheticoverviewoftheroleofhistorical linguisticswithinthebiolinguisticframework.Inparticular,thereaderispointedtothecoreproblemof generativist research on change and variation, namely the factthat, once it isacknowledged that there existsahuman “languageorgan”, which is genetically transmitted and invariant within thespecies, the origin of theactual variationamong the observed lan-guages of the world becomes a real paradox. In order to solve it,it has to be posited that the possibility of variation is biologicallybased,andthathumanlanguagecapacityisactuatedbyarangeof‘phenotipical shapes’; however, the observed variation cannot havea biological basis, rather it must be determined by environmentalfactors, which become crucial during the stage of language acqui-sition. The child, in building her mental grammar, her internalI-language, is guided by innate principles to develop grammaticalstructures which are recognized within the corpus of primary lin-guisticdata.ThesedatacomefromtheexternalE-language,whichisa functionof theuseof I-grammarsby individuals in thechild’senvironment. The emergence of new internal systems of linguistic


353

knowledge must be linked to a shift occurring in the external pri-marycorpus,andviceversa.Thecentralchaptersofthebookare,infact,devotedtothedefinitionofadistinction,bothinthecausesandin the actuation, between two different types of linguistic change:while E-language changes affect the primary corpus for languageacquisition, possibily preparing I-language changes, the latter areformalchangesinanewgeneration’smentalgrammar.Thus,E-lan-guage,whichdependsonthevariousspeakers’useoftheirinternallinguistic knowledge, is a costantly changing entity, inherently influx.Onthecontrary,I-languagesusuallyremainstableduringtheadultage:theyonlychangefromonegenerationtoanother.Itistheinterplaybetweeninternalandexternallanguageswhichgivesrisetonewgrammaticalsystems.

Thesecondchapterisdevotedtoashortsurveyoftraditionalapproaches to language change, starting with the ComparativeMethod in the Nineteenth Century and discussing the structureof Nineteenth-Century historical explanations. The third chapterrepresents an expansion of the first chapter with respect to thenotions of I-language and the poverty-of-stimulus argument insupportofauniversalinnatebasisforlinguisticknowledge.Issuesconcerning learnability theories and the nature of acquisitionalmechanisms–atopicIwillcomebacktoin4.1–areillustratedinthefourthchapter.ThefifthchapterisconcernedwithanaccountofsyntacticchangeattheI-languagedimension,andpresentsevi-dencefromthehistoryofEnglishmodalverbsandfromthestudyofverbmovementinconnectiontomorphology.Inthesixthchap-terthetriggeringmechanismofstructuralchangesisinvestigated,relating it tothechange intheuse,atthe levelofE-language,ofpre-existingconstructions;heretheempiricalmaterialisrepresen-tedbyastudyofthesplitgenitiveconstructionfromOldtoMiddleand Early Modern English, and of the shift in the verb-objectorder in the history of various Germanic languages. The seventhchapter deals with the accelerated process of new language crea-tiontakingplaceinthecaseofcreolesandsignedlanguages.Theconcludingchapter takesupagaintheclassicalconcernsofhisto-ricallinguisticsdiscussedinthesecondoneandsurveyssomenewperspectivesraisedbythecognitiveapproachtolanguagechange,withparticularattentiontotheissueofreconstruction.AccordingtoLightfoot,thefactorofchanceintrinsictodiachronicprocesses,which makes every predictive theory of change impossible, alsohas theeffect of severely restricting thepossibility of reconstruc-tingproto-grammars.

Chiara Gianollo

354

3.2 Roberts (2007)Ian Roberts, with his textbook, has tried to satisfy a profound

needofmany linguisticsdepartments, that of introducing thegene-rative study of syntactic variation and change to a public which issupposedtohavenopreviousknowledgeofformalsyntaxnorofhisto-ricallinguistics.Althoughtheauthorexplicitlywarnsthereaderthatthebookisnot intendedtobeanintroductiontosyntactictheoryoramanual forsyntacticanalysis,hesucceeds inmakinghistextbookself-sufficient.Thetaskisaccomplishedthankstotheintelligentchoi-ceandclearexpositionofa limitedbutsignificantnumberofpointsofsyntacticvariation,whichareinvestigatedinthehistoryofvariouslanguagesandtackledfromavarietyofperspectivesindifferentsec-tionsofthebook.

The accessibility of the text is substantially helped, especiallyfrom a didactic point of view, by the use of separate boxes for theillustration of more technical aspects of the theory, by the presenceofausefulglossary,andbythethoroughandwell-organizedindexofsubjects.Eachchapterisenrichedbyaconcludingsectioncontainingdetailedsuggestionsforfurtherreading,whereeachreferenceisbrie-flypresented in itsprincipalachievements, constituting in thiswayalsoavaluablesurveyofdifferentapproachestophenomenamentio-nedinthediscussionandaguidetorecentdevelopments.

ThefirstchapterintroducesthePrinciples&Parametersapproa-ch to syntactic variation in its Minimalist manifestation and thenfocusesonthegroupofsixparameterswhoseeffectsinvariousstagesof different languages will be used as the main empirical materialthroughoutthebook:theexistenceofnullcategoriesinsubjectposi-tion,themovementofthelexicalverbtoafunctionalpositionknownas V-to-T movement, the possible successive movement of the verbinstantiatingthephenomenonofVerbSecond,aspectsof thesyntaxofnegation(negativeconcord),themovementof interrogativeopera-tors(Wh-movement),andfinallythemuch-debatedhead-complementparameter.Parametersarepresentedfirstintheirsynchronicdimen-sion,byadoptinga trulycomparativeperspectivewhichcoversdatafromarichsampleoflanguages(it isnotfrequenttofindabookonformalsyntaxwithafour-pageIndexofLanguages);thenfollowsananalysisofwitnessedchangesrelativetothevalueofeachparameter,conductedwithathoroughdiscussionofreferencesonthesubject.

The second chapter is concernedwithdifferent types of syntac-tic change, such as reanalysis, grammaticalization, and change inargumentstructure,complementation,andwordorder.Allreceiveanexplanationintermsofparameterresetting.Thesectiononreanalysis


355

isparticularlyinteresting,asittakestheauthortoaddresstwocru-cialproblemsforanytheoryofchange,termedbyRobertsthe‘Regressproblem’andthe‘Chicken-and-Eggproblem’(Roberts2007:125-127).Thesearebothrelatedtotheabductivenatureofgrammaticalchange,andrefer respectively to theparadoxof languageacquisition (anewgenerationabducesadifferentgrammarfromwhathasgeneratedtheprimarycorpusithasbenexposedto)andtotheproblemconcerningthecausalrelationshipbetweentwocorrelatedchanges.

The role of first-languageacquisition in language change is thesubjectofthethirdchapter,whichintroducesaPrinciples&Parametersapproach to thedynamics of acquisitionanddiscusses thenature oftriggers,theroleofmorphology,thedefinitionofstructuralsimplicityanditsimportanceinguidingreanalyses,andthesignificance,inthisrespect,ofatheoryofmarkednessofparametervalues.Thischapterisanoteworthyoriginalcontribution,blendingtheexpositionofsomefirmlyestablishedpointsingenerativeacquisitionalresearchwiththeillustrationofnovelperspectives on theanalysis of syntactic changestemmingfromtheauthor’srecentresearch.Iwillcommentonsomeoftheseaspectsin4.2and4.3.

Thediscussionof thedynamicsof syntactic change, the ‘transi-tion problem’, in the fourth chapter is the occasion to introduce theissuecreatedbyanI-languageperspectiveon languagechangewithrespect to the apparent gradualness in the diffusion of innovation.Robertsreviewsthemechanismsoflexicaldiffusion,whichbringstheauthor to the concept of microparametric variation and change, theroleof formaloptionality incausingapparentgradualness, thecom-peting-grammarsapproach.Thenhepresentsamajordebateinhisto-ricalstudies, thatconcerningthenotionof ‘drift’,whichIdiscuss in4.3.Theconcludingsectionofthechapterisdedicatedtothetopicofsyntacticreconstruction.Thefifthandlastchapterdealswiththeroleofcontactincausingsyntacticchange:inthefirstpart,theeffectsofmassivesecond-languageacquisitionand,moreingeneral,of lingui-sticsubstrataareintroduced,whereasthecreationofcreolesandsignlanguagesarethesubjectofthesecondpart.

4 Challenges for generative diachronic syntax

4.1 Acquisition and the nature of the triggering experienceInthefourthchapterofhislatestbook,Lightfootincisivelysum-

marizeshisoriginalapproachtolearnability,whichhehasbeendeve-loping since the late Eighties (Lightfoot 1989). Lightfoot considers

Chiara Gianollo

356

learnabilitytheoriesbasedonevaluationmetricstobepsychologicallyimplausible,as theyrely tooheavilyontheprocessofgrammarcom-parisonover finite corporaofunanalyzedsentences, i.e. onE-langua-ges. Models such as Gibson & Wexler’s (1994) Triggering LearningAlgorithm,anerror-driven,serialdevice,presuppose, in fact, thechi-ld’saccess toa storeddata set, formedbyE-languageelements, andtheirbatchprocessing,yieldingseriousfeasibilityproblemscausedbythe“exponentialre-explosion”(Fodor2001:736)ofworkloadnecessaryto evaluate parametric systems of a plausible extension (Gibson &Wexler’ssimulationinvolvedjustthreeparameters).Amajorfeatureofnaturallanguages,alreadyforcefullypointedoutbyGibson&Wexlerthemselves(butseealsopreviousobservationsbyClark1989),isrepre-sentedbypervasiveambiguityintheinputsentencesconstitutingthetriggeringexperience,whichinprincipleenablesthechildtogeneratemore than one grammar for given sentence types. Once ambiguitygeneratesanerror insettingoneparameter,thiserrormaycause, inturn,successiveinputtoreceiveawronginterpretation,leadingtoanunrecoverablesituationgiventhesekindsofmodelsofacquisition.

LightfootarguesthatthisaporiacanbeavoidedonceoneadoptsaperspectiveonlearnabilitygenuinelycenteredonI-language,accor-ding to which the real triggering experience is not to be recognizedin sentences or word-strings, E-language elements, but rather inpieces of structure, pure I-language entities. The same strategy isfollowedbyFodorinproposingtheStructuralTriggersLearnermodel(Fodor1998and subsequentwork, especiallyFodor2001andFodor&Sakas2001):childrendonotlearnfromambiguousinput,becausetheyareguidedby innate structural triggers called ‘treelets’, piecesof structure adopted in the grammar only if successful in parsingstringscomingfromtheexternalinput.Eachinputsentencebecomesparametrically unambiguous once it receives a complete structuraldescription, thus each structurally represented sentence will repre-sent an unambiguous trigger once triggers are supposed to be pie-ces of structure. If a surface string may receive different structuraldescriptions, it will be disregarded during the process of parametersetting: the learner, within this model, is able to detect parametricambiguityandtobewareof it, learningonly fromunambiguousevi-dence (seeFodor1998:23-27 foradiscussionof casesofsubset lan-guages,wherethelearnermightbethoughttobecompelled,infact,tolearnfromambiguousevidence).

Similarly,Lightfoot (1997, 1999, 2006: 77-86)proposesamodelbasedonstructural triggers, called “cues”,piecesof structureprovi-dedbyUniversalGrammarthatguidethechildtothementalrepre-


357

sentationsofsentencescomingfromtheE-language.Suchsentenceswillbeconsideredtoexpressacueonly if theyunambiguouslyneedthatpieceof structure inorder tobeanalyzed.So, for instance, thechildlooksforacuelikeIV,i.e.inflectedverbsintheInflection/Tenseposition:theinstantiationofthiscueinasufficientnumberofE-lan-guageutterances(e.g.verb-initialinterrogativesentencesornegativesentencesinModernFrench)willimplyforthechildtheassumption,in her I-grammar, of a V-to-I (in current Minimalist framework, V-to-T) movement. Cues/treelets are “global” triggers in Gibson andWexler’s(1994:409)sense:theyareunambiguoustriggersforagivenparametervalueineverylanguage.

A further hypothesis couched in Lightfoot’s model of acquisi-tionconcernsthesyntacticspacewhichisrelevantfortheprocessofparametersetting:Lightfoot(1991)introducestheideathatchildrenlook for cues only in simple structures, unembedded domains, suchassimpleclausesandthetopofembeddedclauses(complementizersand subjects). The “degree-0” learnability hypothesis stems mainlyfromdiachronicobservations,clearlysummarizedinLightfoot(2006:123-136).HeretheauthordiscussesthechangefromOVtoVOorderin thehistory ofEnglish, connecting it toVerb-Secondproperties ofmainclauses,whichincreasethenumberofVOordersintheE-lan-guagecorpus,andalsotothereanalysisofmodalsandtotheriseoftheperiphrasticdoconstruction.Thedifferentrateofchangeinmainand embedded clauses, gradual in the former but later in time andmuchmorerapidinthelatter,isconsideredtobeevidenceofthefactthatchildren,inchangingtheirgrammarfromOVtoVO,onlyconsi-deredunembeddedevidence,whichhadgraduallyshiftedtoapreva-lenceofsuperficialVOordersduetoaccidentalvariationintheuseofE-languagebythelinguisticcommunity;assoonasthechangehappe-ned,thenewvaluewasgeneralizedtoembeddeddomainsaswell(foradiscussionofdifferentanalysesof thischange,andforreasons forskepticism about its suddenness in embedded domains, see Roberts2007:175-198andcitedreferences).

Within a cue-based approach, the notion itself of parametercomes to be cast into a different perspective: in fact, Lightfoot sta-tes that “there isnoneed foran independentnotionofparameters”(Lightfoot2006:78);children,inprocessingdatacomingfromthepri-marycorpustheyareexposedto,scantheenvironmentforcues:“cuesthatarerealizedonlyincertaingrammarsarethepointsofvariationbetweengrammars”(ib.).Roberts(2007:242-245)confrontsthisview,andraisessomecritiques,themostimportantofwhich,inmyopinion,concernstheunrestrictednatureofthecue-basedapproach:“ifthere

Chiara Gianollo

358

isno independentnotionofcues, thenwehavenowayofspecifyingthe class of possible parameters, and hence the range along whichlanguagesmaydiffer,synchronicallyordiachronically”(Roberts2007:244).Iftheformatofcuesisnotexplicitlydefinedandshowntoobeyrestrictions following from properties of Universal Grammar, a cue-basedapproachruns intotheriskofoverloadingthegeneticendow-ment,whichwouldcontainapotentiallyinfinitelistofpiecesoftreestructurestobematchedwiththementalrepresentationsofexternaldata during the process of language acquisition. Moreover, it wouldfail toofferaprincipledaccountforthe interactionsamongparame-tervalueswhichareapervasivecharacteristicofnaturallanguages.It has to be added, though, that the same methodological difficul-ties may arise within a parametric approach, if parameters are notsubjecttoacritiqueconcerningtheirpossibleformatandtheirhierar-chicalorganization(seeGianollo,Guardiano&Longobardito appearforsomeremarksandaproposalconcerningarestrictivetheoryoftheformofparameters).Thestudyof languagechangemayprovetobedecisiveinyieldingbetterformalizationsofthespaceofgrammaticalvariation,whosesuccess,aswehaveseen,isstrictlyconnectedtoaninvestigationoftheproblemofparametricexpressioninthePrimaryLinguisticData(henceforth,PLD).

Beyond the issue of the relevant format for triggers, historicalconcernschallengeclassical learnability theoriesalsowhen it comestoaccountforwhatistraditionallyreferredtoasthelogicalproblemof language change, also termed the “Regress problem” by Roberts(2007: 126): the paradox to be explained consists of the fact that,when language change happens, the trigger experience produced byagenerationwhichhasacquiredagivengrammarisnotsufficientforthenextgenerationtoconvergeonexactlythesamegrammar.Thus,anewgrammarhastobeascribedtoamutationinthecorpusgene-ratedbyspeakersofthe‘older’grammar,immunetothegrammaticalchange.Suchprimitivemutationisassumednottobegrammaticalinnature; itarises fromvariation in theactualuseof thepre-existingsystem,whichmaychangethenextgeneration’sprimaryexperiencetoathresholdlevelwhichtriggersanewgrammaticalsystem.

In order to cope with these facts in elaborating a learnabili-ty theory, the learning algorithm to be assumed cannot be strictlydeterministic, but rather “weaklydeterministic” (Roberts2007:231)andallowforparametricchangetotakeplace (this iswhyProbablyApproximately Correct Algorithms of language learning, such asthosediscussede.g.byClark&Roberts1993andNiyogi2006,havebeenproposed).


359

Given a weakly deterministic model, the factor assuring sub-stantial convergence with the parent grammar is a drive towardsconservativity inherent to the language acquisition device, formula-tedastheInertiaPrinciple(seediscussioninRoberts2007:231-232),whichblockschangesnotnecessarilytriggeredbylocalcausesintheprimarycorpus.Inthepresenceofanadequatetrigger(P-expressioninRoberts’terms,cueinLightfoot’s),thevalueattributedtoagivenparameterbyanewgenerationconvergesonthatofthepreviousone;otherwise,oncethetriggerforagivenvaluehasbecomelessrobust,eitherbecauseoftheinterventionofextra-syntacticfactorsorastheconsequence of an independent syntactic change, it paves the waytowardsgrammatical change.But then,how todefine the ‘strength’thresholdofatrigger,i.e.theextenttowhichthelearnercanbesaidto find unambiguous parametric expression in the PLD? Possibleanswers to this core question are presented at length, in constantconfrontationwithLightfoot’spositionsexpressedover theyears,byRoberts(2007),inhisdiscussionofgrammaticalizationand,especial-ly,reanalysis.Wewilladdressthisissueinthefollowingparagraph.

4.2 Economy of representation and diachrony Grammatical change is abductive in nature and, as such, open

to ‘error’:giventheprinciplesofUGandaprimarycorpus,thechildabducesagrammar,whichmayormaynot convergeon thesystemwhich has yielded the primary corpus itself. The reason for this, asdiscussed in the previous paragraph, must lie in the variable useof grammars within a linguistic community, which may reduce thestrength of some triggers and result in parameter change. Roberts(2007)arguesthatchangeinparametervaluesisgenerallyassociatedwith a process of reanalysis of a given syntactic pattern, a processwhichinvolvesMoveandAgreeoperationsaffectingfunctionalheads(whose feature set is assumed, since Borer 1984, to be the locus ofparametricvariation).Reanalysishappenswhenagivenstringsurfa-cinginthePLDisassignedbythelearnerastructuralrepresentationdifferentfromtheonewhichhasgeneratedit.Theconditionscausingreanalysisare investigatedbyRoberts (2007) inhis second chapter:Robertsconnectsreanalysistothenotionoftransparency introducedby Lightfoot (1979), which is however innovatively interpreted intermsofeconomyofrepresentation.

The “Transparency Principle” invoked by Lightfoot (1979: 98-115,121-141) inhis studyof the changing categorizationofEnglishmodalshadbeenconceivedasaUGprinciplepromptingthelearner’s“therapeutic”reaction,intermsofreanalysis,incaseofaccumulated

Chiara Gianollo

360

opacity in the primary data. The Transparency Principle “requiresderivations to be minimally complex and initial, underlying struc-tures to be ‘close’ to their respective surface structures” (Lightfoot1979:121).Thus,theclusteringofexceptionalfeaturesontheclassofEnglishmodalverbscausedthemtobeopaqueasmainverbsandledthelearnertotheirnewcategorizationasauxiliaries.

Roberts,atleastsinceRoberts(1993),hasattemptedtoreachaformalcharacterizationofthenotionofopacity,byrelatingittothatof simplicity, in the wake of Lightfoot’s original formulation of theTrasparencyPrinciple.HeconnectsopacitytoambiguityanddefinesthelatterbyreferringtoClark&Roberts’(1993)notionofparameterexpression and P-ambiguity. Strong P-ambiguity, whereby a stringmay express both values of a given parameter, is considered to belinkedtoreanalysis.StrongP-ambiguitymayariseasaconsequenceofchangesinothermodules(phonology,semantics)ormaybecausedbyothersyntacticchanges.

In developing the notion of “Diachronic Reanalysis”, Roberts(1993: 153-160) considers structural ambiguity arising from opacityto be addressed by the learner by appealing to a notion of structu-ral simplicity: in the presence of competing structural representa-tions for a given string, the learner would opt for the simplest one.Hedefinessimplicity intermsofnumberof linkswithinasyntacticchain, and proposes that the notion plays a role in acquisition bymeansofwhathecallsthe“LeastEffortStrategy”,accordingtowhich“[r]epresentations assigned to sentences of the input to acquisitionshould be such that they contain the set of the shortest possiblechains (consistentwith (a)principlesofgrammars, (b)otheraspectsofthetriggerexperience)”(Roberts1993:156).ReanalysesguidedbytheLeastEffortStrategymaycreatetheconditionsforasubsequentparametricchange,byreducing,andultimatelyremoving,structuralevidencefortheolderparametersetting.

Roberts(ib.)isveryresoluteinsettingtheLeastEffortStrategyapart from similar claims made by Chomsky (1989, reprinted withminorrevisionsasthesecondchapterofChomsky1995),who,accor-dingtoaresearchagendawhichwillbecome,initsmatureform,theMinimalist Program, considers principles of simplicity in derivationandrepresentationtobepartofUGingeneral,andnotjustanacqui-sitional strategy to deal with ambiguity. Nonetheless, the notionof Diachronic Reanalysis and the connected proposal concerningthe Least Effort Strategy are strongly criticized by Lightfoot (1999:216-220) as “an attempt to explain some changes entirely throughUG, independently of changes in trigger experiences” (Lightfoot


361

1999:216),and,assuch, suggestingadeterministicviewofhistory.AccordingtoLightfoot,DiachronicReanalysesoccurwhereagramma-ticalchangehasalreadytakenplaceand,thus,havenorealexplana-torypower:therealexplanationliesinshiftsintheinput,whicharetheonlycause forparametricchanges. Inhisview, theLeastEffortStrategy,andhisownTrasparencyPrinciple,representawrongturntowards the search for “endogenous” tendencies towards “optimiza-tion”(Lightfoot1999:218),foranexplanationofgrammaticalchangesmotivateduniquelybyinternalfactors,suchaseconomy.

Roberts(2007:132)partiallyanswersthiscritiquebydefendingtheusefulnessofanotionofDiachronicReanalysisinhighlightingtheroleofstructuralambiguityinparametricchange.Inhisdiscussionofgrammaticalchangesarisingthroughreanalysis,healsoshowsthathis approach does not entail abstracting away from actual changesinthetriggeringexperience:onthecontrary,inordertoattainarealexplanationof thechange, it isnecessarytosingleouta localcausewhich prevented reanalysis in a previous generation and motivatesitsappearanceinthenewone.Forinstance, inhisdiscussionofthedevelopment of the French question particle ti, he argues that thetriggertoreanalysiscomesfromapreviouschangeinthephonologi-calsystem,althoughheacknowledgesthatthisinterpretation,whilesolving the problem for syntax, shifts the burden of explanation tophonology (seeRoberts1993:220-224and2007:129-132).However,Roberts does not deny that higher-order cross-linguistic principlesmightbeactiveinguidingacquisitionand,therefore,change.

Roberts (2007) recasts the Least Effort Strategy in terms of a“simplicity preference”, according to which “reanalysis is motivatedbyageneralpreferenceon thepartof languageacquirers toassignthe simplest possible structural representations to the strings theyhear” (Roberts 2007: 131). He further elaborates on the notion ofsimplicitybyproposinga“simplicitymetric”,drawnfromrecentjointworkwithAnnaRoussouongrammaticalization(Roberts&Roussou2003). Roberts & Roussou (2003: 200 ff.), while noticing that, inprinciple,manydifferentapproachestosyntacticcomplexitymaybeproposed(basedonnumberofnodes,branchingnodes,traces,chainlinks -cf. Roberts 1993-, symbols or features), favor a feature-coun-ting approach, following the proposal made by Chomsky & Halle(1968) for treating complexity in phonological systems. A simplerrepresentation for a given string will be that containing fewer for-mal features. Roberts and Roussou further assimilate complexitytomarkedness; theirapproach isdiscussedbyRoberts (2007) inhisthird chapter, where he connects this proposal to the similar one

Chiara Gianollo

362

madebyChomsky&Halle(1968) forthedistinctive-featuresysteminphonology.Accordingly,anasymmetryisrecognizedinparametervalues:“theunmarkedvalueofaparameterdeterminesagrammarwhich generates simpler structures than those generated by themarkedvalue”(Roberts2007:254).

As featuresareresponsible formovement, that is,aProbecau-singmovementwillbemorecomplexthanonenotcausingit,itispos-sibletodevelopamarkednesshierarchyforgrammaticaloperations,alongthelinesofRoberts&Roussou(2003:210-213):Move>Agree>neither.

Marked parameter settings are associated with opaque, relati-vely complex, constructions, e.g. constructions involving movement;this explains, for instance, the mechanism of grammaticalization,which creates exponents of functional categories out of lexical ele-ments: given certain co-occurring circumstances, such elements arereanalyzed as items directly merged in the functional position (andnotcopiedtherebymeansofacostliermovementoperationfromthelexicallayer).

Also,morphologyisconsideredtobeassociatedwithmarkednessand,asaconsequence,withcomplexity.Roberts(2007:264)suggeststhat“[i]faformalfeatureofacategoryCisinflectionallyexpressed,thenCisassociatedwithamarkedparametervalue”;asaconsequen-ce,morphologicallossisrecognizedasadecisivefactorinparameterchange,yieldinganunmarkedsetting.

Thisformulation,however,shouldnotbetakenasaninstantia-tionoftheisomorphicviewofsyntaxandinflectionalmorphologycon-vincinglycriticizedbyAnderson(2002).Morphologymayactasacueformovement,andthusitslossmayperturbatetheprimarydataandweakentheparametricexpression.Thisdoesnotentail,however,asarguedbyLightfoot(2002;seealsotheothercontributorstothesamevolume), that there be a two-way relationship between morphologyandsyntax,wherebythereisnomovementwherethereispoormor-phology(forasimilarremark,seeLightfoot2006:106).

FollowingChomsky&Halle(1968),Roberts’markednesstheorydoes not posit the existence of a single unmarked value for a givenparameter; markedness has to be contextually determined, in lightof parametric interactions, and markedness reversals might takeplace. This helps avoid an impending paradox: once a preferencefor unmarked systems is built into the learning theory, one wouldsuppose grammatical change to be directional, only from marked tounmarkedvalues.Asthisisobviouslynotthecase,Robertsdevelopsasystemwherebychangesfromunmarkedtomarkedarepossible,once


363

what he calls a “markedness-induced harmony” (Roberts 2007: 275)isatwork.Accordingly,themarkednesshypothesiscomestoberefer-rednot justto individualparameters,buttosystemsofparameters:“themarkednessofaparticularparameterwilldependonthevaluesassumedbyotherparametersinagivensystem”(Roberts2007:273).Crucially,RobertsassumesthatsomethinglikeVennemann’s (1974)andHawkins’(1983)cross-categorialharmony,inducingapreferenceforlanguageswherethereisahomogeneousorderingofconstituentsacross phrasal categories, is active in grammatical systems: he sta-tesitintermsofa“preferenceforpotentialmovementtriggerstoacttogether”(Roberts2007:194),harmonically.Thus,forinstance,evenif a given head having an EPP feature – causing movement – mayrepresentacomplexfeatureforagrammar,oncethisheadwillcon-formtootherheadsinthelanguageinhavingsuchanEPPfeature,itsvaluewillnotbemarked,becauseapreferenceforharmonicorde-ringwilloverridethemarkednessofthemovementoperation.

We see here that Roberts recasts the typological notion ofcross-categorial harmony as a preference for simple, i.e. relativelyunmarked,grammars.Inhisdiscussionofthehead-complementpara-meter,Robertstakesintoconsiderationtypologicalapproachestotheproblem,tryingtorelateimplicationaluniversalsoftheGreenbergiantraditiontoclusteringeffectscreatedbysingle,deepparameters,ortothecomplexsystemofinteractionsexistingamongvaluesofdifferentparameters. In discussing the issue of cross-categorial harmony intheorderofheadsandtheircomplements,heappearstobeconvincedofthefactthatastrongtheoryofsyntaxshouldbeseriouslyconcer-nedwithfrequentlyobservedcasesofharmonicordering,despitetheobvious counterexamples. In particular, he proposes that harmoniceffectscannotbetheresultofaunitarysyntacticoperation,butratherthatoftheinteractionofdistinctgrammaticalfeatures,andthat“thepreferencefor‘harmonic’orderingmaythusderivefromanoverridingtendencyforindependentparameterstoconspiretoproduceacertaintypeofgrammar”,incompliancewith“ahigher-ordercross-linguisticprinciple” (Roberts 2007: 101-102). The concept of cross-categorialharmonyhasrecentlybeenrevivedinparametrictermsbyRobertsinworkwithTheresaBiberauer(Biberauer&Roberts2005;Biberauer&Robertsto appear),togetherwiththatofparametricconspiracy.

AccordingtoBiberauerandRoberts(2005),effectsofparametricharmonymayariseasasymptomofageneraldrivetowardssimpli-fication of structures which is active during language acquisition, a‘least-effort’ strategy forcing reanalysis once primary data becomeopaqueorambiguouswithrespecttoagivenparametervalue.There

Chiara Gianollo

364

isnointernaltendencytomutationwithinsyntaxitself,andthetrig-gerforthemechanismisconsideredtoresideintheexternallangua-ge,inthelinguisticcommunity;however,assoonasambiguityarisesin the input, deep overarching grammatical principles of simplicityandharmonywillaffectthelearner’sreanalysisoftheprimarydata.

Obviously, saying that a principle of simplicity guides learningand,asa consequence, syntactic change,doesnotamount to sayingthatchangemakesgrammarssimpler:besidesthefactthatreanaly-sis processes may cause complications in other areas of grammars(see the example taken from French grammar in Roberts 1993:177-186), the result of parametric change is just another grammarwithdifferentproperties,innowaysimplerthanthoseoftheformersystem.

However, the interaction of simplicity with the notion of har-monyinRoberts’modelofchangeintroducesintothetheoryadirec-tionalforceabletoconspiretowardsasyntactic‘type’overperiodsoftime transcending generations. Roberts’ model, therefore, entails arevision of ‘orthodox’ generative treatments of long-range historicalphenomena,whichareusuallyseenasepiphenomenal,non-directio-nalclusteringsofparametricproperties.Thecontroversialperspecti-vesinitiatedbysuchanapproachclearlyappearfromLightfoot’sandRoberts’ discussions of the notion of ‘drift’, to which we turn in thenextparagraph.

4.3 The notion of driftWhenSapirintroducedthenotionof‘drift’in1921,heconsidered

it,asappropriatelystressedbyLightfoot(2006:37;cf.also1979:386ff.,1999:208ff.),asanexplanandum.Sapirconsidereddriftinalan-guagetobethe“unconsciousselectiononthepartof itsspeakersofthoseindividualvariationsthatarecumulativeinsomespecialdirec-tions”(Sapir1921:155).DriftsforSapirrepresentedapeculiarfacetofvariation,i.e.thatpartofvariationwhichwasnotrandom,butper-ceivedbyhindsightasdirected.Nonetheless,phenomenaofdriftwereassumedtobeduetospecificlocalcauseswhichcouldbeisolatedbylinguistic analysis (see the analysis of four factors involved in the“driftawayfromwhom”inSapir’s(1921)seventhchapter).

However, Lightfoot claims that, since the first applications oftypologicalobservationsonword-orderharmony to thestudyof lan-guagehistory(e.g.Lehmann1974,Vennemann1975),‘drift’becameasortofexplanatoryforce,anexplanans,withinatheoryofchange roo-tedinadeterministicviewofhistoricaldevelopments,accountingforthem without looking for local causes. Basically following the histo-


365

ricist thinkingofNineteenth-Century linguistics, typologists startedto investigate the existence of diachronic universals, of directionalgrammaticalphenomenacapableofspanningcenturiesandtransfor-mingone‘pure’linguistictypeintoanother,followinguniversallypre-dictablehistoricalpathsandgivingrisetointermediate,transitionalstagesrepresentedby‘mixed’grammars.Aparticularlydebatedphe-nomenon in thisareaof studyhasbeen therelativeorderof consti-tuents(subject/verb/object,noun/adjective,auxiliary/verb,andsoon),i.e.whatinparametriclinguisticsisknownasthehead-complementparameter. Typological approaches to these facts assume a cross-categorialcoherenceintheorderofheadswithrespecttotheircom-plements,whichisperturbedduringtheprocessoflanguagechange;accordingly,a‘mixed’systemwillbedrivenbyaninternalforcealongapredictablesuccessionofstagestoattainanewequilibrium.

Such long-spanning change is incompatible with an approachwhichrecognizesthelocusforchangeintheprocessoftransmissionofindividualgrammars,asmostexplicitlystressedinLightfoot’s(1979:385-405) discussion of drift. Nonetheless, gradualness and disconti-nuityofchange,aswellasconvergenceeffectsareundeniablepheno-mena.AsRobertsemphasizes (Roberts2007:348; cf.also2001:91),giventhehugespaceofpossibleparametricvariationcreatedbyafewbinary parameters (thirty independent binary parameters generate230languages=1,073,741,824),thefactthatonecan,infact,observelinguistictypesordiachronictendenciesisastonishing.Thishintstothefactthatchangeislessrandomthatwhatmightbesupposedandcalls foranexplanationwhichgoesbeyond the individual.ToadoptLass’(1987)powerfulsimile,“speakersseemratherlikeTolstoy’s‘lit-tlemen’,caughtupingreathistoricalcurrentswhoseimportthey’reunaware of, but who nevertheless play their ‘ordained’ parts in thelargerdesign”(Lass1987:162).

Lightfoot (1979: 396, 402 ff.) readily acknowledged the existen-ce of drag-chains in syntactic change, and, while judging that theyhad been overstated by typological research and misunderstood intheirrole,heconsideredthem“ofenormous importance foratheoryof grammar” (Lightfoot 1979: 396). Furthermore, he was persuadedthat advance in syntactic theory could represent an important steptowards their understanding. For instance, he was confident in thesignificance of X-bar conventions for diachronic linguistics, as theyconstrainedpossible innovationsbyprovidinga restrictive theoryofphrasestructureand,inparticular,couldaccountforobservedcross-categorialgeneralizations intheorderofspecifiers,heads,andcom-plements(Lightfoot1979:402-403).

Chiara Gianollo

366

More recently, inhis1999book, comingback,whenevaluatingNiyogi & Berwick’s (1997) computational simulations of langua-ge change, to the issue of the existence of diachronic trajectories,Lightfootstatesthat“theexplanationforlong-termtendencies,iftheyemerge,willbeafunctionofthearchitectureofUGandthelearningprocedure and of the way in which populations of speakers behave.Inthiswaythehistoricaltendencieswillturnouttobeepiphenome-na, derived in an interesting fashion, not stipulated by brute force”(Lightfoot1999:225).

Inhis fourth chapter,Roberts (2007)discussesLightfoot’s criti-quetothetypologicalinterpretationofthenotionof‘drift’,embracinghis perspective; nonetheless, being convinced that “something likeSapir’s notion of drift is required on both empirical and conceptualgrounds” (Roberts 2007: 350), he lays the basis for a parametrical-ly oriented explanation of long-term trajectories by exploring thevalidityofthenotionof ‘parametricdrift’.Hesinglesoutthreemainproblems an explanation has to face, which can be summarized asfollows: the causation problem (what causes purported linguistic‘cycles’), the directionality problem (which natural direction a givendriftwouldhave),andtheincompatibilitywithaprincipleofInertiawhichissupposedtocauseconservativenessingrammaticalsystems.

Roberts proposes that drift, and, in general, implicationalsequences of syntactic changes, which include also grammaticaliza-tionphenomena,beinterpretedinparametrictermsasa“cascadeofchanges,akindof ‘dominoeffect’intheparametricsystem,wherebyaninitial,exogenouschangedestabilizesthesystemandcausesittotransitthroughaseriesofmarkedstatesuntiliteventuallyrestabili-zesasarelativelyunmarkedsystemagain”(Roberts2007:341-342).Roberts interprets thegradualdiffusionof changewithin thegram-maticalsystemsasobeyingdynamicssimilartothoseobservedinthephenomenonof lexicaldiffusion:as soundchangesgraduallydiffusethroughthelexiconbyinvolvingoneitematatime,inthesamewayparametricchangesintheformalfeaturesoffunctionalheadsmightgraduallydiffuse,headbyhead,inthesetoffunctionalcategories.

Summarizing recent joint work with Theresa Biberauer(Biberauer & Roberts 2005), he offers an exemplification of ‘para-metric drift’ by analyzing changes in the English verbal systemas a cascade of parametric resetting operations spanning from theFifteenth to theSeventeenth century, starting from the loss of verbmovementtotheCpositionandfromthereoninvolvingeachachan-ge in the feature compositionof elementsmerging in theTposition(Roberts2007:351-356).


367

Thetwonotionsofstabilityofasystemandmarkednessofstatesraise a series of crucial questions for an acquisition-based theory ofchange,andareat riskofbringingback into thepicture thesortoflong-term teleology refuted by generative historical research. First,if a system is unstable, i.e. at least dispreferred by some principlesof UG, how is it possibly acquired as such by new learners? Whyshoulditlastforlongperiods,withoutbeing‘normalized’inthespaceofa fewgenerations? It seems tome that theactuation problem, touse the classical terminology introduced by Weinreich, Labov, andHerzog’s(1968)seminalwork,isthereagain.

Moreover, when successive cohorts of acquirers “reanalyse dif-ferent aspects of the PLD which have been rendered marked by anearlier change” (Roberts 2007: 342), what drives the gradual choiceof suchaspects?Roberts’ ideawith respect to this lastpoint is that“eachparameterchangeskewsthePLDinsuchawaythatthenextis favoured”(Roberts2007:356).Causes,thus,arestill local,withinthe triggeringexperience;however, theymaynotbe inducedby for-cesexternaltosyntax,butratherbecouchedinthesyntacticsystemitself,andappearonebyone,eachastheconsequenceofapreviousparameter resetting operation. As for the causation problem, there-fore,theimmediatecauseforthedriftmechanismtobesetinmotionisconsideredtoresideintheoutsideworld,whereevenasmallvaria-tionintheprimarycorpusforacquisitionmaytriggeraprocessoflan-guagechange;this,inturn,wouldentailaseriesofinternally-causedshiftsinthePLDtriggeringfurtherparametricchangesandcausingthe‘cascade’effect.

There is nothing predetermined in the observed sequences ofchange,insofarastheyareneverinevitable:theyjustproceedaslongas no other factor intervenes. This is accounted for by revising theformulationoftheprincipleofInertia,accordingtowhichitdoesnotnecessarilyentail stasis,butpersistence inagivendirection,unlessexternalinterferenceontheprimarydataoccurs.

Asconcernsthedirectionalityproblem,asolutiontoitispropo-sedbyinvokingmarkednessconsiderations,whichplayanimportantrole in Roberts’ line of reasoning: “certain areas of the parameterspace attract grammatical systems, by being relatively unmarked”(Roberts 2007: 350). A drift would be directional as a result of itspreferencetowardsunmarkedareasofthegrammaticalspace,i.e.invirtue of amotivationbuiltwithinUG itself.Theprocess of changewould not be completed within the individual, but would continuebeingpulledtowardsa“basinofattraction”.

Chiara Gianollo

368

Ithastobenoticed,however,thatcontemporarylinguistictheoryseems to be still far from the result of elaborating a solid theory ofmarkedness,whichmayguidehistoricalexplanations.

Roberts is concerned with markedness at length in the thirdchapterofhisbook.Theconceptofmarkednessappearstobetightlyconnected,inhisaccount,tothatofstructuralsimplicity,aswehaveseenin4.2.

Lightfoot (2006) discusses markedness in connection to creolesstudies in his seventh chapter. It is well known that an importantaspectofBickerton’s(1984aandb)LanguageBioprogramHypothesisconsistedinthepromiseoffindingintheinvestigationofcreolegene-sisspecialevidenceconcerningunmarkedsettingsofUGparameters.More specifically, creoles would represent a collection of parametersettings coming in part from the superstrate language and, muchmore substantially, instantiating unmarked values innately prede-fined by UG, which would manifest themselves in response to anextremely impoverished triggering experience, such as that offeredby pidgins. Lightfoot criticizes this hypothesis, offers an alternati-ve explanation in terms of cue-based acquisition, and argues thatnothinginprincipleforcesonetothinkthat“markedsettingsrequi-reaccess tomoreextensiveexperience,andperhaps to fairly exoticdata,andthatthis isnotavailableto firstspeakersofacreole. […]Onecaneasily imagineamarkedsettingbeingtriggeredbyreadilyavailable data, even in the first forms of a creole” (Lightfoot 2006:144). Most interestingly for our discussion here, he maintains thattheonlybasisforpostulatingmarkednessvalueswithinUGisrepre-sented by arguments from the poverty of the stimulus, i.e. by thenecessityofrankingparametricvaluesinordertoavoidresortingtonegativeevidenceduringacquisition.Markedness,underthisanaly-sis,resultsfromtheapplicationoftheSubsetPrinciple,bywhichthelearnerchoosesthemostrestrictivegrammarwhichcangeneratealland only the structures found in the primary corpus: the smallestlanguage derives from the least marked parameter setting. Oncemarkedness is interpreted in this principled way, it can be shownthat‘radicalcreoles’,likeSaramaccan,dodisplaymarkedparametricvalues (Lightfoot 2006: 142-144). The determination of marked vs.unmarkedparametricvaluesonthebasisof theSubsetPrinciple iscriticized by Roberts (2007: 257-261), who, nevertheless, acknowle-dges the explanatory power of markedness considerations based ontheSubsetPrincipleincasesofgenuineformaloptionalityandinthediachronicprocessof restrictionof functionapplyingtogivengram-maticaloperations.


369

Despitetheauthors’attemptstolaythebasisforaserioustheoryof markedness, the problem appears to remain open, together withtheconnectedissuesrelativetoaprincipledaccountoflong-rangedia-chronicphenomena.A renewed interest in such topicsbygenerativesyntacticians is, however, welcome. Although the study of syntacticpersistenciesandcross-generationaltendencies in languagechange isanotoriouslytrickyfield,alsoforobjectivedifficultiesduetothenatureofourhistoricalrecordsformostlanguages,Ifeelthatonemustconclu-dewithLassthat“[i]fthelinguisttieshimselfdowntotheparochial,to the individualbrainand itsspatiotemporallyboundknowledge,heimposesonhimselfamethodologicalandconceptuallimitationthatnootherhistorianwouldlethimselfbeconstrainedby”(Lass1987:157).

5. Conclusion

Thevolumesunderreviewarebothsuccessfulattemptstobrid-gegaps,withLightfoot focusingon therelationshipwithdisciplinesattheinterfacewithlinguisticsinthestudyofhumanbehavior,andRobertsbeingconcernedwithatoo frequentdichotomybetweenfor-malandhistoricalstudiesof language,and,especially,ofsyntax.Inchoosing to concentrate indepthon justa fewaspectsofLightfoot’sandRoberts’argumentation,Ihaveruntheriskofoverlookingsomeotherimportantcontributionsoftheirdiscussiontothecurrentdeba-te. However, I hope to have given a clear idea of the importance ofoperatingat theedgeofdifferentbranchesof linguistics in order toforcefully address diachronic issues which might, in turn, contribu-te to amoregeneralunderstanding of thehuman language faculty.Tosay itwithLightfoot, “amodernhistorical linguistneeds tobeageneralist and to understand many different subfields – grammati-caltheory,variation,acquisition,theuseofgrammarsanddiscourseanalysis, parsing and speech comprehension, textual analysis, andtheexternalhistoryoflanguages”(Lightfoot2006:6).

The volumes by Lightfoot and Roberts highlight the particu-larly urgent necessity of a better connection between acquisitionaland historical studies. Roberts (2007: 225) regrets that “there issomething of a sociological divide between linguists working on L1acquisitionandthoseworkingondiachronicsyntax”.Thisisespecial-lyunfortunategiventheweightthatnotionssuchasthatofsimplicityandcross-categorialharmonyaregaininginhistoricalexplanations,aweight which calls for a better investigation of the validity of suchconstrualsbymeansofobservationalstudiesoflanguageacquisition.

Chiara Gianollo

370

Another point which clearly emerges from the two books con-cerns the significance of conclusions drawn from the observation ofnetworks of parameters, and not just scattered points of variation:approaches to theories of cascade resettings or parametric conspi-racies such as Roberts’ have to be elaborated over large parame-tric systems, displaying their full range of interactions, which maybe understated when studying only a few parameters at a time.Crucially, therefore, sound historical analyses of syntactic changeshouldwiden their scope to encompass full-fledgedsystemsofpara-meters. In this respect, the application of the Modularized GlobalParametrization strategy, first proposed by Longobardi (2003), todiachronic investigation seems to qualify as particularly promisinginorder tounderstandmechanismsofpropagationofchangewithinparametric networks. According to this method, an entire, coherentmoduleofsyntax–nounphrasestructure inLongobardi’swork– isanalyzed in a significant number of languages, and parameters areformulated in order to capture minimal contrasts existing betweenany two varieties and to single out interactions among parametricfeatures. In this way, it becomes easier to comprehend instances ofco-variationandco-evolution.

Inclosing, letmeobservea factwhichmightseemmarginal tothediscussionbut, indeed,directlyor indirectyenablesmanyof theproposals sketched above, namely the dramatic improvement in thefine-granedanalysisofdatayieldedbytheexistenceofsyntacticallycodedelectroniccorporafordifferentstagesoftheEnglishlanguage,theYork-Toronto-Helsinki Corpus of Old English, thePenn-Helsinki Parsed Corpus of Middle English, and the Penn-Helsinki Parsed Corpus of Early Modern English (respectively,YCOE,PPCME2andPPCEMEinthereferences).Insharpcontrast, themuchmorefrag-mentarypictureofthesyntactichistoryof languagesforwhichsuchtechnicaldevelopmentsareyettocomeisapparent.

Address of the Authors:

Chiara gianollo,UniversitàdiTrieste([email protected])

References

andersonStephenR.2002.SyntaxandMorphologyareDifferent:CommentsonJonas’“ResidualV-to-I”.InlighTfooTDavid(ed.).Syntactic Effects of Morphological Change.Oxford:OxfordUniversityPress.271-275.


371

anderson Stephen R. & David lighTfooT 2002. The Language Organ: Linguistics as Cognitive Psychology. Cambridge, UK: CambridgeUniversityPress.

BaTTlori Montserrat,MariaLluïsahernanZ,CarmepiCallo&FrancescroCa2005(eds.).Grammaticalization and Parametric Change.Oxford:OxfordUniversityPress.

BaTTye Adrian & Ian roBerTs 1995 (eds.). Clause Structure and Language Change.Oxford:OxfordUniversityPress.

BiBerauerTheresa&IanroBerTs2005.ChangingEPP-parametersinthehis-toryofEnglish:accountingforvariationandchange.English Language and Linguistics9.1.5-46.

BiBerauerTheresa&IanroBerTs to appear.Cascadingparameter changes:internally-driven change in Middle and Early Modern English. IneyThórsson, Thórhallur & Jan Terje faarlund (eds.). Grammatical Change and Linguistic Theory: the Rosendal Papers. Amsterdam:Benjamins.

BiCkerTon Derek 1984a. The language bioprogram hypothesis. Behavioral and Brain Sciences7.2.173-188.

BiCkerTon Derek 1984b. Creole is still king. Author’s response to commen-tary.Behavioral and Brain Sciences7.2.212-218.

Borer,Hagit1984.Parametric Syntax.Dordrecht:Foris.Chomsky Noam1981.Lectures on Government and Binding.Dordrecht:Foris.Chomsky Noam1989.Somenotesoneconomyofderivationandrepresenta-

tion.MITWPL10.43-74.Chomsky Noam1995.The Minimalist Program.Cambridge,MA:MITPress.Chomsky, Noam & Morris halle 1968. The Sound Pattern of English. New

York:HarperandRow.Clark,Robin1989.Ontherelationshipbetweenthe inputdataandparam-

etersetting.NELS 19.48-62.Clark, Robin & Ian roBerTs 1993. A Computational Model of Language

LearnabilityandLanguageChange.Linguistic Inquiry24.2.299-345.Crisma,Paola&GiuseppelongoBardito appear(eds.).Historical Syntax and

Linguistic Theory.Oxford:OxfordUniversityPress.fodor, JanetD.1998.Unambiguoustriggers.Linguistic Inquiry29.1.1-36.fodor, JanetD.2001.Settingsyntacticparameters.InBalTin,Mark&Chris

Collins(eds.).The Handbook of Contemporary Syntactic Theory.Oxford:Blackwell.730-767.

fodor, Janet D. & William G. sakas 2001. The structural triggers learner.In BerTolo, Stefano (ed.). Language Acquisition and Learnability,Cambridge,UK:CambridgeUniversityPress.172-233.

gianollo, Chiara, Cristina guardiano & Giuseppe longoBardi to appear.Three fundamental issues in parametric linguistics. In BiBerauer,Theresa(ed.).The Limits of Syntactic Variation.Amsterdam:Benjamins.

giBson Edward & Kenneth Wexler 1994. Triggers. Linguistic Inquiry 25.3.355-407.

greenBerg Joseph 1963. Some Universals of Grammar with ParticularReference to theOrderofMeaningfulElements. IngreenBerg, Joseph(ed.).Universals of Language.Cambridge,MA:MITPress.73-113.

Chiara Gianollo

372

haWkins,JohnA.1983.Word order universals.London:AcademicPress.jonas, Dianne to appear (ed.). Grammatical Change: Origins, Nature,

Outcomes.Oxford:OxfordUniversityPress.keenanEdward1994. Creatinganaphors.AnhistoricalstudyoftheEnglish

reflexivepronouns.LosAngeles:UCLA.Ms.lass Roger1980.On explaining language change.Cambridge,UK:Cambridge

UniversityPress.lass Roger 1987. Language, Speakers, History and Drift. In koopman,

Willem, Frederike van der leek, Olga fisCher, Roger eaTon (eds.).Explanation and Linguistic Change. Amsterdam: Benjamins. 151-176.

lass Roger1997.Historical linguistics and language change.Cambridge,UK:CambridgeUniversityPress.

lehmann WinfredP.1974.Proto-Indo-European Syntax.Austin:UniversityofTexasPress.

lighTfooT David 1979. Principles of Diachronic Syntax. Cambridge, UK:CambridgeUniversityPress.

lighTfooTDavid1982.The language lottery: Toward a biology of grammars.Cambridge,MA:MITPress.

lighTfooT David 1989. The child’s trigger experience: degree-0 learnability.Behavioral and Brain Sciences12.2.321-334.

lighTfooT David 1991. How to set parameters: arguments from language change.Cambridge,MA:MITPress.

lighTfooT David1997.Catastrophicchangeandlearningtheory.Lingua 100. 171-92.

lighTfooT David1999.The development of language: acquisition, change and evolution.Oxford:Blackwell.

lighTfooT David 2002 (ed.). Syntactic Effects of Morphological Change.Oxford:OxfordUniversityPress.

longoBardiGiuseppe2003.MethodsinParametricLinguisticsandCognitiveHistory.Linguistic Variation Yearbook3.101-138.

monod Jacques1970.Le Hasard et la Nécessité: essai sur la philosophie natu-relle de la biologie moderne.Paris:Seuil.

niyogi Partha 2006. The Computational Nature of Language Learning and Evolution.Cambridge,MA:MITPress.

niyogi Partha & Robert BerWiCk 1997. A Dynamical Systems Model forLanguageChange.Complex Systems11.161-204.

pinTZuk,Susan,GeorgeTsoulas&AnthonyWarner 2000 (eds.).Diachronic Syntax. Models and Mechanisms.Oxford:OxfordUniversityPress.

PpCeme.Penn-Helsinki Parsed Corpus of Early Modern English.Maintainedby Anthony kroCh & Beatrice sanTorini - University of Pennsylvania.http://www.ling.upenn.edu/hist-corpora.

ppCme2. Penn-Helsinki Parsed Corpus of Middle English. Maintained byAnthonykroCh&AnnTaylor-UniversityofPennsylvania.http://www.ling.upenn.edu/hist-corpora.

roBerTs Ian 1993. Verbs and Diachronic Syntax. A Comparative History of English and French.Dordrecht:Kluwer.


373

roBerTs Ian 2001. Language change and learnability. In BerTolo Stefano(ed.), Parametric Linguistics and Learnability. Cambridge, UK:CambridgeUniversityPress.81-125.

roBerTsIan&Annaroussou2003.Syntactic Change: a Minimalist Approach to Grammaticalization.Cambridge,UK:CambridgeUniversityPress.

sapir Edward 1921. Language. An introduction to the study of speech. NewYork:HarcourtBrace&Co.

van kemenade, Ans & Nigel vinCenT 1997 (eds.). Parameters of Morphosyntactic Change.Cambridge,UK:CambridgeUniversityPress.

vennemann Theo 1974. Topics, subjects and word order. From SXV to SVXvia TVX. In anderson John M. & Charles jones (eds.), Historical Linguistics.Proceedings of the First International Congress of Historical Linguistics.VolII.Amsterdam:North-Holland.339-376.

vennemann Theo1975.Anexplanationofdrift. InliCharlesN. (ed.).Word Order and Word Order Change,Austin:UniversityofTexasPress.269-305.

WeinreiCh,Uriel,WilliamlaBov,&MarvinherZog1968.Empirical founda-tionsforatheoryoflanguagechange.InlehmannWinfredP.&Yakovmalkiel(eds.).Directions for historical linguistics: A symposium.Austin:UniversityofTexasPress.95-188.

yCoe. York-Toronto-Helsinki Corpus of Old English. Maintained by AnnTaylor, Anthony Warner, Susan pinTZuk, Frank BeThs - University ofYork.

http://www-users.york.ac.uk/~lang22/YcoeHome1.htm.

historical syntax and the generative paradigmlinguistica.sns.it/rdl/20.2/gianollo.pdf · on the...

Documents