foundations-of-generative-syntax

AppliedPsycholinguistics16:2231BookReviewsElman,J.L.(1990).Findingstructureintime.CognitiveScience,14,179-211.(1993).Learninganddevelopmentinneuralnetworks:Theimportanceofstartingsmall.InC.Umilta&M.Moscovitch(Eds.),AttentionandperformanceXV:Consciousandnonconsciousinformationprocessing.Hillsdale,NJ:Erlbaum.Fodor,F.A.,&Pylyshyn,Z.W.(1988).Connectionismandcognitivearchitecture:Acriticalanalysis.Cognition,28,3-71.Lachtcr,J.,&Bever,T.G.(1988).Therelationbetweenlinguisticstructureandassociativetheoriesoflanguagelearning—Aconstructivecritiqueofsomeconnectionistlearningmodels.Cognition,28.195-247.Levine,D.S.(1991).Introductiontoneuralandcognitivemodeling.Hillsdale,NJ:Erlbaum.McClelland,J.L.,&Elman,J.L.(1986).Interactiveprocessesinspeechperception:TheTRACEmode!.InJ.L.McClelland&D.E.Rumelhart(Eds.),Paralleldistributedprocessing(Vol.2).Cambridge,MA:MITPress.McClelland,J.L.,&Kawamoto,A.H.(1986).Mechanismsofsentenceprocessing:Assigningrolestoconstituents.InJ.L.McClelland&D.E.Rumelhart(Eds.),Paralleldistrib-utedprocessing(Vol.2).Cambridge,MA:MITPress.Mozer,M.C.(1991).Theperceptionofmultipleobjects:Aconnectionistapproach.Cam-bridge,MA:MITPress.Pfeifer,R.(Ed.).(1989).Connectionisminperspective.NewYork:Elsevier.Pinker,S.,&Mehler,J.(Eds.).(1988).Connectionsandsymbols.Cambridge,MA:MITPress.Quinlan,P.T.(1991).Connectionismandpsychology:Apsychologicalperspectiveonnewconnectionistresearch.Chicago:UniversityofChicagoPress.Rumelhart,D.E.,&McClelland,J.L.(1986).OnlearningthepasttensesofEnglishverbs.InJ.L.McClelland&D.E.Rumelhart(Eds.),Paralleldistributedprocessing(Vol.2).Cambridge,MA:MITPress.St.John.(1992).Thestorygestalt:Amodelofknowledge-intensiveprocessesintextcompre-hension.CognitiveScience,16.271-306.Schneider,W.(1987).Connectionism:Isitaparadigmshift?BehaviorResearchMethods,Instruments,andComputers,19,73-83.CatherineL.HarrisBostonUniversityFoundationsofgenerativesyntax.RobertFreidin.Cambridge:MITPress,1992.Pp368.ThisintroductiontoGovernmentBindingTheory(a.k.a.PrinciplesandParametersTheory)isintendedforadvancedundergraduateandgraduatestudents"witheithersomebackgroundinlinguisticanalysisorsomefamil-iaritywithformalanalysisinotherdisciplines"(p.4).Itsprimarygoalis"tobridgethegapbetweenintroductorylecturesandthefrontiersofre-searchinsyntax"(p.1).Unlikemostothertextbookswiththesamepur-pose(see,e.g.,thenowclassicvanRiemsdijk&Williams,1986),Founda-tionsinGenerativeSyntax(henceforthFGS)doesnotchooseahistoricalapproach,butinsteadtriesto"providearationalreconstructionofthecurrenttheoryofgrammarintermsofthebasicconceptsonwhichitisbuilt"(p.1).ItthereforecomesasadisappointmentthatFGSignoressomeofthemostimportanttheoreticaladvancesmadesinceChomsky(1981).Nomentionismadeofderivationalandrepresentationaleconomy,acon-ceptthat,inthewakeofChomsky(1989),hascometoplayacentralroleofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2232BookReviewsinthetheoreticdebate.EvenmoreproblematicisthefactthatFGS-unlikeotherrecentintroductorytextbookssuchasHaegemann(1991),Cowper(1992),orNapoli(1993)-doesnotreflecttheveryinfluentialattemptofChomsky(1986)tounifyvarioussyntacticlocalityconstraints.Asaconsequence,thebookleavesthestudentlessthanperfectlyequippedtotacklemuchoftheprimarysyntaxliteratureofthe1990s.Moreover,theargumentationinthisbookissometimesflawed,anditsoverallstructurebecomesconfusingtowardstheend.(Iwillreturntotheseproblemsindetail.)WhileIcannotrecommendFGSasthebasisforthebeginner'sstudyofcontemporarysyntax,thebookcontainsnuggets,suchasaverydetailedanalysisoftheEnglishauxiliarysystem,whicheverylinguist(in-cluding,butnotlimitedto,first-yeargraduatestudents)willfindvaluable.'Thefirstchapter("TheStudyofSyntax")introducesbasicanalyticaldevicessuchasphrases,phrasestructuretrees,andphrasestructurerules.Italsolaysoutsomeofthepsychologicalunderpinningsofthetheory,inparticular,theassumptionofaninnateUniversalGrammarastheanswertotheproblemofhowchildrenacquirelanguagedespitethepovertyofthestimulus.Chapter2("IntroductiontoFormalGrammar")developsaphrasestructuretheory,definessomeofthestructuralrelationsapparentinphrasestructures(e.g.,c-commandandgovernment),andillustratestheirrolewithrespecttoCaseassignmentinEnglishnounphrases,whichisgivenatransformationalanalysis.Thenextchapter("Recursion,Movement,andBounding")discusseswh-movementandextrapositionandtworestrictionsontheseprocesses,subjacencyandtheEmptyCategoryPrinciple(ECP).TheanalysisofclausestructureisrefinedandacomparisonofboundingnodesinEnglishandItalianservesasafirstillustrationofcrosslinguisticvariation.Chapter4("TheEnglishVerbalAuxiliarySystem")givesacom-prehensiveaccountofthebehaviorofEnglishmodalsandaspectuals,whichisbasedonwell-knownruleslikeAffixHopping,Subject-VerbIn-versionandDo-Support.Thefifthchapter("Infinitivals,Case,andGov-ernment")returnstoCaseTheoryandshowshowsubjectsofinfinitivalsententialcomplementsareassignedCaseviaExceptionalCaseMarkingbythematrixverborNP-movementtothesubjectpositionofthehigherclause.ThediscussionisextendedbeyondthescopeofEnglishbyalookatlexicalCaseinRussianandIcelandic.Chapter6("TheAnalysisofEmptyCategories")analyzesthedistributionofPRO,wh-trace,andNP-traceasdeterminedbypreviouslyintroducedmodulesofthetheorythatwereintroducedearlierandaddstotheseThetaTheoryandProperBinding.Theseventhchapter("BindingTheory")investigatesthestructuralrelationbetweenanaphorsandpronouns,theirrespectiveantecedents.Thelastchapter("ExtensionsofBindingTheory")extendsthisinvestigationtoin-dependentlyreferringexpressionsandemptycategories.Eachchaptercon-tainsexercisesandbibliographicalcomments.Thebookhasendnotes(anunusualand,giventhatsomeofthesenotesareessentialfortheunder-standingofthetext,confusingchoiceforatextbook),abibliographyandanindex.IwillnowdiscusssomeofwhatIperceivetobeshortcomingsofFGS.ofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2233BookReviewsChapter2dealswithphrasestructureand,hence,cruciallywiththead-junct-argumentdistinction.Butthestructuraldefinitionofthesetermsin(1),adaptedfrompage40,misidentifiessubjectsasadjuncts2whiletheconceptualexplanationthat"thedirectobjectphysicsin(36a)[i.e.,(2)below]isacomplementofstudies,whereasatMITistakentobeanadjunctonthegroundsthattheactofstudyinginvolvesBlancheandphysics,butnotMIT'(p.41)wronglyimpliesthatsubjectsare,infact,complements.1.[X"1"adjuncts...[X*complementsX]]2.Blanche[VPstudiesphysicsatMIT]]Notethatfewstudentswillbeconvincedbytheclaimthattheactofstudy-inginvolveswhatisstudied,butnotwhereitisstudied.Sincenounequivo-calsyntactictestsforadjunct-versusargument-hoodaregiven,thesecon-ceptsremainunclear.SubjacencyisacentraltopicofChapter3,butnotincurrentresearch-comparethefactthatlastyear'svolumesofLinguisticInquiryandNaturalLanguageandLinguisticTheory,thetwoleadingGBjournals,eachmen-tionSubjacencyonlyonceinpassing.Ineffect,subjacencyhasnotsurvivedChomsky(1986).Itisthussurprisingthatthe"variousattemptstointegratetheformulationsofSubjacencyandtheECPintermsofthenotionofgovernment"areonlybrieflymentionedinanappendix,wheretheyarediscountedas"extremelytechnical(andatpresentspeculative)innature"(p.133).ThediscussionofSubjacencyinFCSdoesraiseaninterestingpsycholinguisticpoint,whichthepresentationunfortunatelymuddles.Frei-dinpointsoutthat"achildacquiringEnglishwillnotbeexposedtotherelevantsentencesthatdeterminethatIPratherthanCPisthecorrectparameter[forboundingnodes]forEnglish;inotherwords,thechildwillnothearsentenceslikethosein(88)"(p.119),[i.e.,(3)below],andIPmustthereforeconstitutethedefaultsettingoftheboundingnodeparam-eter.3a.*[CPwho,[IPdidJohnsay[CPwhatj[IPe;sawes]]]]b.*[CPwhOj[IPdid[NPEliot'slackofinterestinejsurpriseVirginia]]CrucialfortheconclusionthattheEnglishparametersettingcannotbelearnedisnottheabsencefromthechild'sinputof(3a,b)(i.e.,ungram-maticalsentenceswhichshowthatIPisaboundingnodeinEnglish;suchevidenceisneveravailable),butrathertheabsenceofgrammaticalsenten-ceswhichshowthatCPisnotaboundingnodefromtheinputoftheEnglish-learningchild(suchevidenceisoftenavailable;cf.[NPtuofratello[CP[PPacui]i[IPmidomando[CP[NPchestorie]j[IPabbianoraccon-tatotjtj]]]]]"yourbrothertowhomIwonderwhichstories(they)told"-which,asmentionedbyFreidin,showsthatIPisnotaboundingnodeinItalian).Wh-movementacrosstwoCP-nodesalsocrossestwoIP-nodes(cf.[4a])andwh-movementacrossaCP-nodeandanNP-nodealsocrossesone,IP-node(cf.[4b]),resultingbothtimesinaSubjacencyviolation.ofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2234BookReviews4a.*[CPwhatj[Cdo[IPyouwonder[CPwhOj[IPMarytoldtj[CPt,'[IPJohnlikest,]]]]]]]b.*[CPwhati[C'do[IPyouwonder[CP[NPwhoseinterestint,]j[IPtjsurprisedMary]]]]]Giventhegrammaticallyof[CPwhOj[Cdo[IPyouthink[CPV[IPtjatethecake]]]]],thechildknowsthatIPandCPcannotbothbeboundingnodes.SincepositiveevidenceagainstCPasaboundingnodeisunavailable(cf.[4a,b]),thecorrectchoice(IP)mustbeprovidedbyUGasthedefaultsetting.Thispointisworthmaking,butitcannotbemadeonthebasisof(3)alone.TwomainproblemsstickoutfromtheotherwisethoroughtreatmentofEnglishauxiliariesinChapter4.First,theaccountisentirelybasedonlanguage-andstructure-specificrules.Thisreflectsalong-standingtradi-tionofworkonthistopic,butisfarremovedfrom"thecurrenttheoryofgrammar"andits"thebasicconcepts"withwhichFGSwantstofamiliarizethestudent.Thus,thefactthatamodal,perfectivehave,andprogressivebemustbeimmediatelyfollowedbyaninfinitival,pastparticiple,andprogressiveparticiple,respectively,isexplainedbytransformingtheD-structurein(5a)viaAffixHoppingintotheS-structurein(6a),aprocesswhereeachaffixslotisfirstvacated(excepttherightmost)andthenfilled(excepttheleftmost)vialocalloweringofanabstractaffix.5a.[IP[INFLTNS[MODAL+0]][VP[HAVE+EN][VP[BE+ING][VP[VERB+e]]]]]b.[IP[INFLe[MODAL+TNS]][VP[HAVE+0][VP[BE+EN][VP[VERB+ING]]]]]ItisunclearwhyFGSdoesnotconsideranalternativetothiscomplicatedandoutdatedanalysis.InChapter3welearnedthatwantandwonderselectcomplementswhoseheadsaremarked[—finite]and[+wh],respectively.Wecan,therefore,capturetheco-occurrencerestrictionsonauxiliariesandverbswithoutrecoursetoAffixHopping(or,forthatmatter,syntacticaffixes)simplybyassumingthatmodalsandaspectualauxiliariesselectcomplementswhoseheadsaremarked[+infinitival],[+past,—finite]and[+progressive],respectively.Notethat,whiletheAffixHoppingap-proachhastowriteanadjacencyrequirementintotherule,nospeciallocalityconditionisneededundertheselectionapproachsinceselectioninvolvesgovernment.Second,FreidingivesthefollowingrulesofDo-SupportandNegativeInsertion,whichinteractinnegativedeclarativestoobligatorilysupplyanauxiliarydo.6a.NegativeInsertion3X-[+aux]-SD:12SC:12VP3not+3Y44ofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2235BookReviewsb.Do-SupportX-[+aux]-YSD:123SC:1do/23JustwhatmotivatesDo-Supportremainsunclear.Freidin'sstatementthatdo"appearstohavethesolefunctionofbearingthetensemarkingofthesentence"(p.171)echoesawidelyheldbelief,butisincompatiblewiththetheoryofFGS,whereAffixHoppingmustbeabletoprecedeNegativeInsertionandtheungrammatical"*JohnnotwatchedtheO.J.Simpsonchase"can,therefore,havethederivationin(7),whichisperfectasfarasthebindingofthetenseaffixisconcerned.7a.[IPJohn[I'[INFL[+past][+aux]][VP[WATCH+e]theO.J.Simpsonchase]])b.[IPJohn[V[INFLe[+aux]][VP[WATCH+[+past]]theO.J.Simpsonchase]]]c.[IPJohn[I'[INFLe[+aux]][VPnot[VP[WATCH+[+past]]theO.J.Simpsonchase]]]]Nothinginthissystempreventstheoptionalgenerationofanempty[+aux]intheINFLofanaffirmativedeclarativesentencecontainingnoovertauxiliaryatD-structurewhichwouldresultintheill-formed(ifnon-emphatic)"*JohndidwatchtheO.J.Simpsonchaseforhours."ThisimportantproblemisnotevenmentionedinFGS,aseriousoversightthatcouldhavebeenavoidedifthetheoryofderivationaleconomydevelopedinChomsky(1989)hadbeenincludedintheexposition.ThediscussionofExceptionalCaseMarkinginChapter5containsthefollowingsentence:"Recallthatunderourreviseddefinitionofm-command(withrespecttogovernment)infinitivalIPdoesnotconstituteamaximalphrasalprojectionbecauseitisnotprojectedfromalexicalhead.Therefore,infinitivalIPwillnotbeabarriertogovernmentlikeVP,NP,AP,PP,CP,andfiniteIP"(p.195).Theindexguidesthestudenttothedefinitionofgovernment.("Alexicalheadagovernsacategory/3ifam-commands/3and/3m-commandsa"[p.65])andasimilarcomment("Recallthatitfollowsfromthisdefinitionthatamaximalprojectionofalexicalheadisabarriertogovernment"[p.124]).Needlesstosay,itdoesnotfollowfromthisdefinitionthatmaximalprojectionsofnonlexicalheadsarenotbarrierstogovernment,norcantheyberecalled,sincetheyhavenotbeenmentionedbefore.Thesentencecitedatthebeginningofthisparagraphactuallyreferstoapassagethatisnotlistedintheindex:"Wecanavoidthisproblem[Howcanthecomplementizercasemarkthesub-ject,althoughthelatterdoesnotm-commandtheformer?]ifonlymaximalphrasalprojectionsoflexicalcountfordeterminingm-commandrelations.Assumingthattheinfinitivaltoisagrammaticalformativeratherthanalexicalhead,infinitivalIPdoesnotcountasamaximalphrasalprojectionforthispurpose.Hence,...CandNPwouldm-commandeachother"(p.ofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2236BookReviews187).Themoodofthelastsentenceleavesitopenwhethertheproposalhasbeenadoptedornot.NotefurthermorethattheconclusionthatfiniteIPisabarrierpresupposesthatfiniteINFLisalexicalhead,anassumptionthatisconfirmedinafootnote("WewouldwanttosaythatfiniteIPissome-howprojectedfromaheadwithlexicalfeatures"[p.331]),butrejectedinthetext("Inflisthe(nonlexicalinthisinstance)headgoverningthesubjectNP"[p.185];inthefinitesentence,'HeannoysSusan').Confusionofthistypeiswidespreadthroughoutthebook.Chapter6notonlyintroducestheProperBindingCondition("*[NPe],where[NPe]isatracethatisfreeinitsbindingdomain"[p.239])beforeBindingTheoryhasbeendiscussed,butalsoconsistentlymisidentifiesthebindingdomainofNP-trace:"BydefiningthebindingdomainforNP-traceasthec-commanddomainoffiniteInfl,weareabletodistinguishbetweentheill-formed(55a.i-ii)[i.e.,(8a-b)]andthewell-formed(55b.i)"(p.237)[i.e.,(8c)]and"ThebindingdomainofatraceresultingfromNP-movementisthec-commanddomainofeitherac-commandingAGRorac-commandingsyntacticsubject"(p.239)[cf.(8d)].Thedefinitionofc-commandadoptedinFGS("Acategoryac-commandsacategory0ifandonlyifthecategorythatimmediatelydominatesaalsodominates/3,andneitheranor/3dominatestheother"[p.44])determinesI''asthec-commanddomainoffiniteINFLorAGRandhencethebindingdomainofe,in(8c).Sincee,remainsunboundinI''(whichexcludesJohn{),thesentenceisincorrectlyruledoutasaviolationoftheProperBindingCondi-tionalongside(8a,b,d).Thismistakeisfinallycorrected20pageslater,inthenextchapter(cf.p.258),wherem-command("Acategoryam-commandsa[sic]category(8ifandonlyifthefirstmaximalprojectiondominatingadominates/3,andneitheranorj3dominatestheother"[p.64])insteadofc-commandisusedtodelimitthebindingdomainofovertanaphors,preciselyinordertoruleinsentenceslike(8c),whichmakesitevenmoremysteriouswhysuchabasicmistakewasmadeinthefirstplace.8a.*Johniwasreported[CP[IPejhadrecommendedMary]]b.*Maryiwasreported[CP[IPJohnhadrecommendedej]c.[IP1Johnj[I''[I1was]reported[IP2e,tohaverecommendedMary]]]d.*Mary,wasreported[IPJohntohaverecommendede(]Notethat(8a,b,d)violatetheindependentlymotivatedCaseUniquenessPrinciple:"AlexicalNPmayhaveonlyoneCasemarking"(p.60).TheythereforedonotconstituteevidencefortheProperBindingCondition.Infact,allexamplesadducedinsupportoftheProperBindingConditionexceptthelast("*MaryisunusualforJohntobeproud")clearlyviolateoneormoreindependentlymotivatedprinciplesofgrammarinadditiontotheProperBindingCondition.4Thisissometimesnotedinthetextorafootnote,butsometimesitisnot,asinthecaseof(8a,b,d).TheevidencefortheProperBindingConditionis,atthispointindeed"notoverwhelm-ing"(p.245),andthestudentwhohasnotyetbeenintroducedtoBindingofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2237BookReviewsTheory,ingeneral,andPrincipleA,inparticular,shouldquestionwhysomuchmachineryisintroducedtodosolittlework.5TheclassiccasesofBindingTheory(i.e.,thebindingofovertreflexives,reciprocals,andpronouns)arediscussedonlyinChapter7(i.e.,toolate).Again,thebindingdomainforPrincipleA("AnanaphormustbeboundindomainD"[p.255])isinitiallywronglydefinedtoexcludethesubject("DomainD...isthe...c-commanddomainofsubject....Let'c-commanddomainofsubject'bedefinedasthesetofcategoriesthatarec-commandedbythesubject.ThisincludesallcategoriesdominatedbyIP,minusthesubjectitself,giventhatthec-commandrelationholdsonlybe-tweencategoriesinalinearprecedencerelation"[pp.255-56]),amovethatpredictstheperfect,"MaryandJohnlikeeachother,"tobeungrammatical.ThediscussionofPrincipleAshouldobviouslyhavecomebeforethedis-cussionoftheProperBindingCondition;thatthetwoaresimilarisnotedhereonlyinafootnote.FreidinpointsoutthatinRussian,areflexivecanbeboundacrossaninterveningsubjectifbotharecontainedinanembeddedinfinitivalclause(cf.[9a])oranounphrase(cf.[9b]),butnotifbotharecontainedinanembeddedfiniteclause(cf.[9c]).9a.onnerazreSaetmne[CPPROproizvodit'opytynadhe-NOMnotpermitme-DATto-performexperimentsonsoboj]self-INST"hefdoesnotallowmej[CPPROjtoperformexperimentsonhimself/myselfj]"b.jaCital[NPegostat'juosebe]I-NOMreadhisarticle-NOMaboutself-LOC"Ijread[NPhis;articleaboutmyself/himself;]c.Vanjaznaet[CPCtoVolodjaljubitsebja]Vanja-NOMknowsthatVolodja-NOMlovesself-ACC"VanjajknowsthatVolodjajloveshimselfcy,"Freidingoesontosaythat"onewaytoaccountforthereflexiveparadigmwouldbetomodifythedomainstatementforPrincipleA.Themodifica-tionmustbebasedonadistinctionbetweenthesyntacticsubjectsoffiniteandnonfiniteclauses"(p.267)andconcludesthatRussianreflexives"mustbeboundinthedomainofaboundSUBJECT"(p.268),whereasubjectisboundifitiscoindexedwithafiniteINFL.YetamuchsimpleraccountisreadilyavailableaccordingtowhichthesyntacticsubjectdoesnotplayanyrolewhatsoeverindelimitingthebindingdomainofRussianreflexives,andthelatterisinsteaddefinedasthem-commanddomainofac-commandingfiniteINFL(oranINFLcontainingAGR):thatis,thematrixclausein[9a,b]buttheembeddedclausein[9c].NoteinparticularthateveninFreidin'saccount,"itisnecessarytoincludeAGRintheformulationofthedomainstatement(hence'SUBJECT'andnot'subject')toaccountfortheabsenceofnominativereflexivepronounsinRussian"(p.268),be-causetheRussianequivalentof"*Vanjajknows[CPthatselfjlovesVolod-ofuse,availableathttp:/www.cambridge.org/core/terms.http://dx.doi.org/10.1017/S0142716400007128Downloadedfromhttp:/www.cambridge.org/core.BeijingNormalUniversity,on11Sep2016at15:11:31,subjecttotheCambridgeCoretermsAppliedPsycholinguistics16:2238BookReviewsj]"ispresumablyungrammatical.Freidin'sformulationproperlycontainsthealternativefor