jackendoff, r., & lerdahl, f. (1979). discovery procedures vs. rules of musical grammar in a...

Upload: goni56509

Post on 02-Apr-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    1/9

    Discovery Procedures vs. Rules of Musical Grammar in a Generative Music TheoryAuthor(s): Ray Jackendoff and Fred LerdahlSource: Perspectives of New Music, Vol. 18, No. 1/2 (Autumn, 1979 - Summer, 1980), pp. 503-510Published by: Perspectives of New MusicStable URL: http://www.jstor.org/stable/833000 .

    Accessed: 01/05/2013 13:49

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    .JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

    of scholarship. For more information about JSTOR, please contact [email protected].

    .

    Perspectives of New Music is collaborating with JSTOR to digitize, preserve and extend access to Perspectives

    of New Music.

    http://www.jstor.org

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/action/showPublisher?publisherCode=pnmhttp://www.jstor.org/stable/833000?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/833000?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=pnm
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    2/9

  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    3/9

    504tothephonetic tring n treatingt as an utterance fhisnativelanguage. In particular, hefamiliar reestructuresof transformationalrammar onstituteepresentationsfthesyntactic evel ofanalysis.Since language users are indeed able to assign linguisticanalysestophonetic tringsotherwise e couldnot under-stand each other), he inguistmust nquireas to whatgeneralprinciples he anguage useremploys n makingthisassignment. he linguist oucheshishypothesesbouttheseprinciplesn terms f a formal rammar, systemofrulesdescribing heclass ofpossibleanalyses and theway analyses are pairedwithphonetic trings.The formalgrammar, herefore,s meant as a hypothesis bout themental rammar,.e. the anguageuser'sunconscious now-ledgeof thesystematic rinciplesnvolved n assigningstructure o "stimulusobjects."The distinction etweenformalgrammar nd mentalgrammars crucial.All languageusershave a mentalgram-mar,which heyuse unconsciouslynd to which heyhavelittle r no introspectiveccess. Thus a linguist rying odevelop formal rammarmustuse indirect vidence, uchas theregularitiesnutterances fthe anguage, heresultsofpsychological xperiments,nd so forth,o form ypo-theses i.e. statements fformalgrammar) bout theprin-ciplesthe anguage useremploys.The use ofthe evidenceis not so differentrom hat n other cientific nterprises:usingregularitiesfobservablephenomena oform ypoth-eses about unobservablephenomena.This is, for xample,themethodologyfparticlephysicsand astronomy. heonlything pecial about thecase of inguistics s that theunobservable henomena rewithinus rather han in par-ticle accelerators rdistantgalaxies.It shouldalreadybe clearfrom his discussionthat theclass ofanalyticstatements boutthe stimulus bject, .e.the structuralescriptionfa sentence,s notbyanymeansidentical o thegrammar,s Keiler eemsto assume. Thegrammars rather systematic escriptionf thepairingsbetween timulus bjects nd analytic tatementsboutthem.Letus nowsee howthis distinction ears on the ssueofdiscovery rocedures, hichKeilerdiscusses t length.Wewill then urn o the mportant uestion fwhatthishas to dowithmusictheory.2. Discovery roceduresn linguistics

    As Keilerobservesp.181), henotion fdiscovery roce-dureappeared n the 40s and '50s in American inguistics,

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    4/9

    505and figuredmostprominentlyn thework f such inguistsas ZelligHarris.5The idea was thatone should be able tobeginwith an arbitrary orpusofunanalyzedsentencesfrom n arbitraryanguage,and be able to construct for-malgrammar,.e. a setofrules ssigning tructuralescrip-tionsto the sentences.Harris claimed to have developeda discovery rocedure or hisprocess-a mechanicaltech-niquefor orrectlyeriving grammar rom corpusof"stimulusbjects."Oneofthefirst reaksbetween homskyand his predecessorsHarriswas in facthis teacher)washis claimthat such a discovery rocedure as notfeasible.6Keiler's ejectionfdiscovery roceduress,we assume,based on thisargument fChomsky's, houghhe providesno citation.However, eiler'suse ofthe term discovery rocedure"is quitedifferent,ecause he confuses rammarwith hestructural escriptionshegrammar ssigns. Keiler n factuses "discovery rocedure"n the sense that inguistsusetheterm ruleofgrammar,".e. as partofmechanicalpro-cedure hatassigns structureso sentencesnota grammarto a corpus).The existence f"discovery rocedures"nKeiler's sense is uncontested ygenerative rammarians:in fact thenotionofgrammatical ule s one of themostcentralfeatures fgenerative inguistics.Thus Keilerhastreated ne oftheessentialaspectsofgenerative rammaras though tis something enerative rammarians eject.To makethedistinctionetween grammar nd a discov-eryprocedure learer, hinkofastronomy.A theory orgrammar) fplanetarymotionmustmechanicallypredictthepositionof theplanets;any theory hat does not do sois inadequate. By contrast, ne couldhardlyexpect ohavea mechanical rocedureor iscoveringheories fplanetarymotion-this is thedomainwhere cientificnspirationsis necessary.3. Whatdoesmusical analysis analyze?

    Turningnow tomusicalissues,we can statethegoalsofa generativemusictheoryn terms ltogether aralleltothose for inguistics. f we takethe"stimulusobject"tobe the musical surfaceofthepiece,thetheorymustassignto each musical surface musical analysis-a descriptionofthestructurefthepiecein terms fparameters frhythm, itch,motivic reatment,tructural rticulation,and so forth.n a mentalistic heory fmusic, heseana-lytic tatements re taken torepresent artofthestructurethat he istener imselfnconsciouslypprehendsr nfersfrom presentedmusical surface.

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    5/9

    506We contend hat n factmostgoodwork n musical ana-lysis s at least covertlymentalistic,n thesense just de-scribed, ince the value of a musical analysis is generallytakento be definedn terms fhow wellit explainstheanalyst's intuitions bout musicalcoherence. hus,for nextreme xample,an analysis that nvariablygroupedtogethervery hreemeasures of theEroica Symphonywouldbe rejected, n thegrounds hat tdoes notcorre-spondto any intuition boutthestructure f thepiece. notherwords, ne's choice ofacceptableanalyses for piecereflectsn a very trong ense thewayone apprehendsthepiece.Let us strengthenurcontentionhat musical analysesnecessarily eflecthecharacter fmusicalperception.Keilerpp.194-5)riticizes armour or seriouslyonfus[ing]theproblem fmusicperceptionwiththeproblem fclari-fying heformal omplexityfthe stimulus bjectthat sperceived." he question s: what are theformal ropertiesofthe"stimulusobject"? n fact, musical surface on-tains nothingbuta sequenceofpitch imultaneitiesandsilences)ofvarying ttack,duration,imbre,nd dynamics.Analytic erms uch as "downbeat," elaboration," pro-

    longation," structural ominant," nd "cadence" aresimply otapplicable o thestimulus bjectperse: they recharacteristicsttributed o it as partoftheprocessofmusicalperception. hus,in ourview,a merecharacteri-zationofthe stimulus bjectwould be virtually evoid ofmusical interest; raditionalmusical analysis is in factconcernedwithhowwe as listenershear patterns fpre-sented sounds.This view s consistentwithcontemporaryviewson perceptionn general; t is explicitly tated na number f the sources Keiler cites note26,p. 194).Fora simpleparallel,observe hat colorcertainly ppears tobe a characteristicf"stimulus bjects" n thevisual field;but n fact, he stimulus s a pattern felectromagneticradiationofvaryingwavelengths nd intensities.A theoryofcolor udgments herefores inherentlyoncernedwiththenature fvisualperceptionnd not ustwith he natureof"stimulusobjects." n short,Keiler himself s seriouslymistaken n trying o isolate theproblems fmusicalper-ceptionfrom heproblems fmusical analysis,sincemostofthe standard oncepts fmusicalanalysisareperceptualorcognitive ather han acoustic n nature.If,then,we acceptthat theanalyticstatements nemakes about a pieceare takenas formaldescriptions fone's intuitions,t is legitimate o ask what thesourceofthese ntuitionss. We see no alternative ut that the is-tenerhas a setofunconsciousgeneralprinciples, asedon some combination f nnateand environmentalnflu-

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    6/9

    507ences,whichhe employs n pairingmusical surfaceswithhis intuitions boutthem-including,ofcourse, ntuitionsofambiguity rvaguenessin thestructurefthepiece.Followingthe inguistic erminology, e can call the is-tener's principles a mental musical grammar. The theoristcan formulate ypotheses bouttheseprinciplesn termsof a formalmusical grammar which attemptsto pair musi-cal surfaceswithmusicalanalyses.The formalgrammarthereforeresupposes n independentntuitive ourceofjustification or heanalyses, ust as traditionalmusictheory oes.As in the case ofa linguistic rammar, rin-ciplesof formalmusicalgrammar re tobe evaluated nterms fhow wellthey ccountfor bservablephenomena,in this case theregularities f isteners' ntuitions boutvariouspassages ofmusic.4. Keiler's"discovery rocedures" s rulesofmusicalgrammar

    As suggested bove,whenKeiler uses theterm discov-eryprocedures"n speakingofmusictheory, e is using tin the sense "rule of formalmusicalgrammar," nd not nthe sense"procedureordiscoveringules offormalmusicalgrammar" the inguistic ense ofdiscovery rocedure).This is clearfrom is initial discussion p.181),wherehesays, "In Schenkerian tudies, he ssue [ofdiscovery ro-cedures] s relateddirectlyo the statusofreductive ro-cedures,.e. rules thatwill allowtheanalysttogo unequiv-ocallyand explicitly rom hemusical surface omoreunderlyingevelsofrepresentationsnd ultimatelyo theUrsatz." fone couldspecify rinciples freduction,heywould constitute ulesof formalgrammar,.e. principlesforpairingmusical surfaceswithanalyses.Keilerthen statestworequirementsn "discovery ro-cedures" i.e.formalmusicalgrammars) hat we findunex-ceptionable. he first s thatthey ndeedbe formal,hat s,sufficientlyxplicit hatany analyst usingthem orrectlywill derive he same analysisfor givenpiece.Second, heymustbe empiricallydequate, n that theanalyses theyassigntopiecesmustbehighly alued on intuitiverounds.But thenhe says (p. 182),"The circularitynherent n thenotionofdiscovery roceduresformal rammar] ies inthe factthat thequestionof their nternal ogicorconsist-encydependsnecessarily n thekinds ofanalytic udg-ments ntended n the first lace." However, here s nocircularityere.A formalmusicalgrammars not ntendedtodiscover ntuitions. ather, t is meant todescribe ntui-tionssystematically.t is motivatedn terms fsomeal-readyunderstood lass of ntuitions; ne test of ts ade-

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    7/9

    508quacy is its ability oextrapolate ts principles opredictone's intuitions boutpiecesnotpreviously nalyzed.Thisis theway any empirical heorys arrived t and tested.Keilercontinues hisparagraphbysaying"The mostcontroversialspectofthis ssue ofdiscovery roceduresderives, fcourse,from hoseanalyticresultsforwhichnoreal setofdiscovery roceduress demonstrable. he latterhas proven obe thecase in generativegrammar..." f weread thiscorrectlyn its context,Keiler s questioning heusefulness fsearchingfordiscovery roceduresn caseswhere heyhave provendifficulto find. t is notclear,however,n whichsensehe intends"discovery rocedure"here. fhe means it in the inguistic ense,he is certainlycorrect: homsky's rgument gainst discovery roceduresis convincing.f, however, e means the term n thesense"ruleofgrammar," s he does almosteveryplace lse,hehas seriouslymisunderstood heenterprise.fmusicalanalyses truly epresentomething bout listeners' ntui-tions, he istenermusthimself rriveunconsciously ttheanalysis;therefore emust ssumethat there s a men-tal grammarunderlyinghe ntuitions. he wholepointofa generative heorys to findout whatthe mentalgram-maris. Confronted ith a class ofanalytical nsightsforwhich tis difficultofind n adequate formalgrammar,one mustnotreject he notionofgrammaroutright,sKeiler on thisreading)appears to. This would be in effectto claim that istenershave no systematic asis for heintuitions heseanalytic nsightsexpress-a conclusionthat shouldcast doubton theanalyses themselves.Rather,thecorrect tance to adopt n such a situation s that ofall scientificnquiry: heprinciples re simplyhard tofind, nd theproblems heyraise shouldbe an impetus ofurtheresearch.Shortly hereafterp.183),Keiler criticizesNarmourforconstructingrulesof nference," hichKeilerequateswith"automaticdiscovery rocedures,"nd hencefindsinherently bjectionable. n fact,Narmour'srules consti-tute n attempt o statea fragmentfa formalmusicalgrammar-not procedureor iscoveringulesofgrammar.In that sensewe cannot faultNarmour'srules.On theotherhand,thegroundson whichtheycan be faulted rethose ofempirical dequacy-how welltheyaccountforan interestinglass of ntuitions boutpiecesofmusic-and here Keiler'scriticisms rewell taken.

    Abstracting way from he ssues we have beendiscus-sing,Keiler'scriticism f Narmour's"grammar"hereseemsto amountto an objection oderivingmusical ana-lysesin a linear fashionfromocal to global levels. Ifthis

    This content downloaded from 193.136.173.41 on Wed, 1 May 2013 13:49:30 PMAll use subject to JSTOR Terms and Conditions

    http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp
  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    8/9

  • 7/27/2019 Jackendoff, R., & Lerdahl, F. (1979). Discovery Procedures vs. Rules of Musical Grammar in a Generative Music Th

    9/9

    510111-171; ackendoffnd Lerdahl, GenerativeMusicTheoryand itsRelevance toPsychology," orthcoming;A DeepParallel betweenMusic and Language," Indiana Univer-sityLinguistics lub, 1980;Lerdahl and Jackendoff, Gen-erativeTheory fTonalMusic, n press MIT Press).5. ZelligHarris,Methods n Structural inguistics,Uni-versity fChicago Press,1951.6. Chomsky, yntactic tructuresop. cit.),Chapter6.7. Allan Keiler, The SyntaxofProlongation ," In TheoryOnly3.5 (August1977),3-27; Bernstein'sThe UnansweredQuestion nd theProblem fMusicalCompetence," usicalQuarterly 4.2 (1978),195-222.8. Our work n print odate ("Towarda FormalTheoryofTonal Music") does not n fact stateanyformal ules ofmusicalgrammar ither, s it was explicitly rogrammaticin nature.However, he otherworkscited n note4 abovedo staterules fgrammarn somedetail. t is onthestrengthofthese rules that we claim ourtheory ohave an empiri-cal status.

    Thi t t d l d d f 193 136 173 41 W d 1 M 2013 13 49 30 PM

    http://www.jstor.org/page/info/about/policies/terms.jsp