Musicperformanceprovidesarichdomainfor studyof bothcognitiveandmotorskills. Empirical research in music performance is summarized, with particularemphasis on factors that contribute to the formation of conceptual interpreta-tions, retrieval from memory of musical structures, and transformation intoappropriate motor actions.For example, structural andemotional factors thatcontributeto performers’ conceptual interpretations are considered. Researchon the planning of musical sequences for production is reviewed, includinghierarchical andassociative retrieval influences,style-specifi c syntactic influ-ences,andconstraintsontherangeof planning.Thefinemotorcontrol evidencedin music performanceisdiscussed in termsof internal timekeepermodels,motorprograms, andkinematic models. The perceptual consequences of music per-formance arehighlighted, including the successful communication of interpre-tations, resolution of structural ambiguities, and concordance with listeners’expectations. Parallels with other domains support the conclusion that musicperformanceis notuniquein its underlying cognitivemechanisms.
Serial Order and Timing Issues........................................................................................... 117Methodological Issues......................................................................................................... 117
Music performanceprovidesa rich domainfor study of both cognitive andmotorskills. Performersdominatemanyaspectsof our musicalculturetoday.Concertattendanceand recordingsales,for example,often reflect listeners’preferencesfor performersand abilities to distinguish amongperformances.Althoughpublic consumption of musictendsto highlight performancediffer-ences,thereare also strong commonalities acrossperformancesthat reflectcognitive functions of grouping, unit identification, thematic abstraction,elaboration,and hierarchicalnesting.Thus, music performanceis basedonbothindividualistic aspectsthat differentiateperformersandnormativeaspectssharedby performers.Both the commonalities anddifferencesamongmusicperformancescanbemodeledtheoretically in termsof generalcognitive abili-ties.
Themajority of studiesfocuson theperformanceof musical compositionsfor which notation is available, thus providing unambiguous performancegoals.The focus has also beenon piano performance,in which pitch andtiming measurementsaresimplified. Commonformsof music performanceintheWesterntonaltradition includesight-reading(performingunfamiliarmusicfrom notation), performing well-learned(prepared)music from memory orfrom notation,improvising,andplayingby ear(performingmusic from auralpresentation).Correlationsamongtheseabilities tend to be high and to in-creasewith training (McPherson1995, Nuki 1984), althoughsomestudiesshow differencesin abilities acrossperformers.For instance,accompanistsperform betterthan soloists onsome sight-reading tasks (Lehmann & Ericsson1993).Although thereare few studies of long-termchangesin performanceability, diary and interview studiessuggestthat differencesin performancelevels acrossindividuals are largely a function of experienceand practice(Ericssonet al 1993,Slobodaet al1996).
Psychological studiesof music performanceaim to developtheoriesofperformancemechanisms(what cognitiveor motor constraintsinfluenceper-formance).A secondaim is to explainthe treatmentof structuralambiguities(in what contextsdo ambiguities arise,what kinds of choicesdo performersmake).A third aim is to understandrelationships betweenperformanceandperception(how are listenersinfluencedby performanceaspects).During aperformance,musical structuresandunits areretrievedfrom memoryaccord-ing to the performer’s conceptualinterpretation,and are then preparedforproductionand transformedinto appropriatemovements.The following sec-tionsof thereview—Interpretation, Planning,andMovement—focuson thesecomponentsof performance.Topicsthatarecoveredelsewhereincludestylis-tic performanceconventions,expertiseand skill development (Ericsson&Lehmann1996),sight-readingand improvising (Sloboda1985b),and social
116 PALMER
and evaluativeaspectsof performance(Gabrielsson1997). This chapterre-views onlythoseperceptualstudiesthataddress performance issues.
SerialOrder andTimingIssues
Speaking,typing, andperforming musicareamong themostcomplexformsofskilled serial action producedby humanbeings.Seminaltheoriesof motorcontrol (Bernstein1967,Lashley1951) often usemusic performanceas theultimateexampleof human motorskill. Basedoncapacitiessuchasthetrilli ngspeedof concertpianists(on the order of 16 notes/s),Lashley (1951) sug-gestedthat successiveelements ofthis kind of activity mustbe centrallylinked;a centrallycontrolledmechanismdeterminesmovements in a predeter-minedorder.This open-loop(motorprogram)theoryis basedon two typesofevidence:Thereis little time for feedbackto affect the planning of the nextmovement(Keele1968),andsomeskills canbe performedin the absenceofkinestheticfeedback (Keele &Summers1976,Lashley 1951).
The control of complex, temporally structuredbehaviorssuchas speechproductionor musicperformanceembodiestwo problems:the serialorderofsequenceelements,andtheir relativetiming. The serialorderproblemarisesfrom the fact thatchain-likeorganizationof behavioris inadequateto explaincertain serial order effects in sequenceperceptionand production.For in-stance,strongconstraintson the order of words within phrasesand of pho-nemeswithin words must be met for speechto be acceptable.Musical andlinguistic sequencesthat are well-formed in their serial order,however,areoften not understandableunlessadditional constraintshold on the relativetiming of the individualsequence elements. Music performed withoutaccuratetemporal control is considered defi cient because it lacks the property ofrhythm, in which the timing of elementsis influencedby the timing of other(adjacentandnonadjacent)elements (Vorberg& Hambuch1978).Thedomainof music performanceis ideal for developingmodelsof timing mechanismsbecauseit offers theoreticalconsensuson thenatureof the temporalrelation-shipsthatmustbepresentfor a sequenceto beconsideredaccurate.Questionsof serial order, relative timing, and how rhythm (temporalpatterning)con-strainsthe planningand productionof musicalsequencesareaddressedbelow.
Methodological Issues
Severalmethodological issues influence the interpretationof researchin musicperformance.First, the wealth of data from a single performance(roughly3000piecesof information in onesecondof digital audiosoundrecordedat alow sampling rate) resultsin problemsof separatingsignal from noise.CarlSeashore(1936, 1938), one of the first to conductpsychological studiesofmusicperformance,developeda pianocamerasystemto recordonly gestural
MUSIC PERFORMANCE 117
(movement-based)datafrom hammerandfoot-pedalmovements, greatly re-ducingtheamountof datanecessaryto captureessentialperformanceaspects.Currentcomputermusictechnologyreliesheavily on movement-basedinfor-mation and recordsonly event onsets,offsets,and their relative intensitiesfrom electronic-or computer-monitored musicalinstruments.
Despite the reductionof information, problemswith separatingthe sig-nal—performanceexpression—from randomnoisefluctuationsremain.Per-formanceexpressionrefersto the largeandsmall variationsin timing, inten-sity or dynamics,timbre,andpitch that form themicrostructureof a perform-ance and differentiate it from another performance of the same music.Musicianscanreplicatetheir expressivepatternsof timing anddynamicsfor agivenmusicalpiecewith high precision(Gabrielsson1987a,Henderson1936,Seashore1938,Shaffer& Todd 1987),andattempts to play without expres-sion significantly dampenthesepatternsbut do not removethem altogether(Bengtsson& Gabrielsson1983, Palmer1989, Seashore1938), which sug-gests that some variations areintentional. Expression isoftenanalyzedaccord-ing to thedeviationof performedeventsfrom their fixed or regularvaluesasnotatedin a musicalscore(Gabrielsson1987a,HG Seashore1936).However,performancecan be expressivewithout referenceto a score(as in musicalimprovisation). Expressioncan also be analyzedrelative to the performanceitself; for instance,expressionwithin a unit suchasa phraseis thepatternofdeviationsof its partswith respectto theunit itself (Desain& Honing1991).Consequently, measurements of performance expression sometimes differacross studies,which makescomparisonsdifficult.
A second methodological problem is determining which performancesshouldbe consideredrepresentative,given the largevariationsthat canoccuramongcompetentperformancesof the samemusic.Thereare few objectivecriteriafor performancesuccess;mostexperimentersopt for arecognizedlevelof performerexpertise.Largesamplesof famousperformersarehardto find,however,andexploratory(nonexperimental)methodsor casestudymethodsareoftenused.A similar representativenessproblemarisesin choiceof musi-cal stimuli. Becauseof complexityissues,experimentersoftenusesimplifiedor reducedmusical compositions. For thesereasons,the domain of musicperformancereliesheavilyon convergingevidencefrom bothsmallandlargesamplestudiesconductedwith different musicalstimuli.
INTERPRETATION
Music performanceis often viewedaspart of a systemof communication inwhich composerscodemusicalideasin notation, performersrecodefrom thenotationto acousticalsignal,andlistenersrecodefrom theacousticalsignalto
118 PALMER
ideas(Kendall & Carterette1990).Eachperformerhasintentionsto convey;the communicative content in music performanceincludesthe performers’conceptualinterpretation of themusicalcomposition. Westerntonalmusichasdeveloped anotation that representspitch andduration information fairlyexplicitly but intensity and tone quality only approximately.Other relation-ships,suchasgroupboundaries,metricallevelshigherthanthemeasure,andpatternsof motion, tension,andrelaxationareunspecifiedor only implicitlyspecifiedin notation.Thus,ambiguitiesin musicalnotationallow a performerconsiderablefreedomin decidinghow to interpretthe music’s content.Inter-pretationrefers toperformers’ individualisticmodelingof a piece accordingtotheirown ideasor musical intentions. Differencesin interpretation canaccountfor why thesamemusicalscoreis performeddifferently by differentperform-ers or why the sameperformermay perform a piecedifferently on separateoccasions.
As in other art forms, there is no single ideal interpretation for a givenmusical piece;every performanceinvolves some kind of interpretationoranalysis(Cone1968,Levy 1995,Meyer 1973).The field of music analysisoffers variousexplanationsfor the contentof a given composition. For in-stance,a piececanbe viewedasa hierarchyof part/wholerelationships,asalinear coursethat follows the harmonictension, or asa seriesof moodsthatresultin a unity of character (Sundin1984).However,musicanalysisdoes notindicatehow a performeractuallyproducesa desiredinterpretation(Dunsby1989). One goal of interpretationis to convey the meaningof the music.Definitions of musical meaningabound,but several theorists define it ashavingmajor componentsthat relateto structure, emotion,andphysicalmove-ment(Gabrielsson1982,Meyer1956),which contributeto performers’ inter-pretations.
One function of interpretation is to highlight particularstructuralcontent(Clarke 1987). Someexperimentalwork evaluatesthe effectsof individualperformers’ structuralinterpretations on performanceexpression.Nakamura(1987) comparedmusicians’ performancesof a baroquesonatawith theirnotatedinterpretations of musicaldynamics (patternsof intensity changes).Performers’ notatedintentions generallycorrespondedto changesin soundlevel. Listeners’ perceiveddynamicsmatchedperformers’ intendeddynamicsfairly well, evenwhen underlyingacousticchangeswere not identifiable.Palmer(1989) comparedpianists’ notatedinterpretationsof phrasestructureand melody with expressivetiming patterns.Onsetsof the melodic voiceprecededothervoice onsetsin notatedsimultaneities (termedmelody leads),and slowing in tempowas greatestat phraseboundaries.Expressivetimingpatternsdecreasedwhenpianistsattemptedto play without interpretation,andthesepatternsincreasedin exaggeratedinterpretations, similar to other find-
MUSIC PERFORMANCE 119
ingsof modulationsin expressivelevel (Kendall& Carterette1990,Seashore1938).Furtherstudiesindicatedthat the expressivetiming patternsincreasedfrom novicesto experts,increasedduringpracticeof anunfamiliarpiece,andchangedacrossdifferent interpretationsof the samepieceperformedby thesame pianist(Palmer 1988).
Interpretationsof structuralcontentaffect both the expressivemarkingofindividual eventsandthelikelihood thateventswill becorrectlyretrievedandproduced.Error analyses(basedon comparisonwith the notatedscore)ofpiano performanceswith different phrasestructureinterpretations indicatedthatpitchdeletionstendedto occurwithin phrasesandperseverationsatphraseboundaries,which suggeststhat interpretationsstrengthenphraseboundariesrelative to other locations(Palmer1992). Thesefindings were replicatedinlater experiments,which alsoindicatedthat melodicinterpretations increasedthelikelihoodthatmelodic eventswerecorrectlyretrievedandproducedrela-tive to nonmelodic events(Palmer & vande Sande 1993,1995).
Anotherfunctionof interpretationis to highlight particularemotionalcon-tentof the music.An extreme viewholdsthat the structure ofmusic is isomor-phic to thestructureof moodsor feelings;music shouldsoundtheway moodsfeel (Langer1953).Gabrielsson(1995)comparedperformers’ interpretationsof emotionalcontentwith their useof expression.Flute andviolin perform-ancesof the samemusic interpretedwith different emotionalcharactersindi-catedgeneralpatternsof changein expression.Performancesof happyandangry emotionswere played with faster tempo and larger dynamic range,whereas softandsad emotions were performed withslowertempoand smallerdynamicrange.Toneonsetswereabruptin theangryversionandmoregradualin thesadversion.Relatedpatternsof performanceexpressionwerefound inviolin performancesof a Beethoventhemewith tenderor aggressiveinterpre-tations (Askenfelt 1986). Later experiments replicatedthesepatterns,andmostof the emotioncategorieswereaccuratelyconveyedto listeners(Gabrielsson& Juslin 1996).The emotional contentof music has also beenexaminedrecently in terms of narrative,with emphasison dramaticcharacterization,thematiccontent,andconceptionsof large-scalestructures(Schmalfeldt1985,Shaffer1995).
Musicalexperienceenhancesbothperformers’ useof expressionto empha-sizeinterpretationsandlisteners’ ability to identify interpretationsandexpres-siveaspectsof performance(Geringer& Madsen1987,Johnson1996,Palmer1988,Sloboda1985a).Listenerswithout musical experiencedo pick up someinterpretiveaspects.Nonmusicianlistenerswere able todiscern general differ-encesamongmechanical(inexpressive),expressive,andexaggeratedlevelsofperformanceasaccuratelyasmusician listeners(Kendall & Carterette1990).Someevidencesuggeststhattypeof musicalexperiencematters:All musician
120 PALMER
listenerswereinfluencedby expressivetiming cueswhenaskedto choosetheintendedphrasestructurein pianoperformances,but only listenerswith pianotraining were influenced by expressive timing cues (melodyleads) whenchoosingamongmelodyinterpretations(Palmer 1988, 1996b). Althoughthesestudiesaddressthe sufficiency of expressivefeaturesto conveyperformers’interpretations,theydo not addresshow necessarytheyare(seesectionbelowon Perceptionof PerformanceExpression).
PLANNING
Planningandmemoryretrievalprocessesin musicperformancereflectmulti-dimensional relationshipsamong melodic, harmonic,anddiatonicelements. InWesterntonal music,individualpitches,chords,andkeysarepositedascon-ceptuallydistinct unitsof knowledge,thatreflectlevelsof melodic,harmonic,and diatonic structure,respectively.Somecompositional structures,suchashomophonic music, emphasizeacross-voice(chordal) associationsbetweenmelody and accompaniment,whereasothers,such as polyphonic structure,emphasizewithin-voice (single-note) associationsamongmultiple importantvoices.Analysesof pianoperformancesindicatedthat chorderrorsoccurredmore often in homophonic stylesand that single-noteerrorsoccurredmoreoften in polyphonic styles, which suggeststhat the relevantmusical unitschangeacrossdifferentmusical contexts(Palmer& vandeSande1993,1995).Knowledgeof diatonicandharmonicstructureinfluencesperformanceaswell.Mistakesweremore likely to originatefrom the key of the piecethan fromanotherkey andto beof thesamechordtypeaswhatwasintended(Palmer&vandeSande1993).Child singers’ pitch errorswerealsolikely to beharmoni-cally related to intendedevents(Moore 1994), and pianists’ errors duringsight-readingof piecesin which deliberatepitch alterationshadbeenplacedindicatedtacit knowledgeof likely melodicandharmonicrelationships (Slo-boda 1976).
Theoriesof skilled performanceoftenassumethatpeoplepreparecomplexsequencesfor production by partitioning them into shortersubsequences(cfvan Galen& Wing 1984).Phrasestructureis onefeaturethat influencesthepartitioning of musical sequences;evidencefrom performancetiming anderrors suggeststhat musical sequencesare partitioned during planning intophrasesegments(Palmer& vandeSande1995).Errorsthatreplacedintendedpitchesin piano performanceswere more likely to originatefrom the samephraseasthe intendedeventthanfrom differentphrases.Interactingelementsrarelycrossedphraseboundaries,similar to findings in speecherrors(Garcia-Albeaet al 1989,Garrett1980).Segmentationduringperformanceplanningisalsoinfluencedby relationshipsamongmusicalaccentstructures.Adult pian-ists’ and children’s abilitiesto reproducemelodieswereincreasinglydisrupted
MUSIC PERFORMANCE 121
the morethat melodic,metrical,andrhythmic groupingaccentswereshiftedoutof alignmentin theperformed tunes(Drake et al 1991).
Both structuralrelationsand the serial distancebetweensequenceeventsinfluencethe rangeover which performerscanplan, presumablybecauseoflimitationson memorycapacity.Supportingevidence isseen ineye-hand spantasks,in which pianistsreproducedbriefly presentedmusicalsequences.Themeaneye-handspanwas7–8eventsbeyondthelocationatwhich thenotationdisappeared,andit tendedto extendonly to phraseboundaries(Sloboda1974,1977).However,eye-handspanmeasuresmayreflecteffectsof bothmemorycapacityand anticipatoryeye movements.Rangeof planningin memorizedpianoperformances(with nonotation)wasaffectedby bothserialdistanceandstructuralrelationsamongsequenceelements(Palmer& vandeSande1995).Errorsandtimingmeasuresindicatedthattheplanning of currentelementswasaffectedby elementsthatspannedlargerserialdistancesin theabsenceratherthanin thepresenceof interveningphraseboundaries,similar to interactionsofdistanceandstructuralconstraints in languageproduction(Garcia-Albeaet al1989).Thesefindingssuggesttwo possible invariantsin theplanning of com-plex serialbehaviorsin manydomains:the co-occurrenceduring planning ofelementsthatsharestructuralfeatures,andconstraintsof structuralboundarieson serialdistancesover whichelementsareconcurrentlyplanned.
Syntaxof MusicalStructure
Theperformanceof musicis alsoconstrainedby style-specificsyntacticprop-ertiesthat transcend individualinterpretations. Manytheories of Westerntonalmusichavemeterandgroupingastheirprimarysyntacticelements(Cooper&Meyer 1960,Lerdahl& Jackendoff1983).Meter refersto periodic features:theregularalternationof strongandweakbeats.Positionsof metricalaccentsform hierarchicallevels,with differentperiodicitiesrepresentedat eachlevel.Meterprovidesa temporalframeworkin performancefor whento do what,assupportedby evidencethatonly those rhythmic patterns that canbeaccommo-datedto a metricalframeworkarecorrectlyreproduced(Povel1981,Povel&Essens1985),and the samedurationpatternis performedwith different ex-pressivetiming when placed in different metrical contexts(Clarke 1985).Groupingrefersto thesegmentation of a sequenceinto smallersubsequencesthatalso formhierarchical levels,based largelyon pitch relationships(Lerdahl& Jackendoff1983).Somemetricalandgroupinglevelsaremoresalientthanothers.Tactusrefersto the most salientperiodicity or metrical level, whichcorrespondsto the rate at which one might tap a foot to the music (Fraisse1982),andphrasesarethoughtto be themostsalientlevel of groupingstruc-ture.Eventsat the mostsalientlevelsarecommonly emphasizedin perform-ance(cf Repp1992b,Todd 1985)andmay be mostpreciselyor consistentlyproduced andperceived (see sectionbelow onTimekeeper Models).
122 PALMER
Probablythemostwidespreadstructuralcharacteristicof Westernmusicisits hierarchicalnature;both pitch and rhythm structuresarerepresentedin aseriesof levels,betweenwhich relationships of reductionor elaborationoper-ate(cf Clarke1988,Lerdahl& Jackendoff1983,Schenker1969).For instance,Schenker’s (1969)musictheoryviewsthemelodic andharmonicorganizationof a musicalpieceasa seriesof progressivelymorecomplexelaborationsof asimplefoundation,thebackground,from whichthesurfacelevelor foreground(thenote-to-noteaspectsof themusicalscore)is generated.Thesehierarchicallevelsnot only embodymusic-theoretic principlesbut alsohaveimplicationsfor perceptual andcognitiveprocesses,suchasthepredictionthatmoreimpor-tant eventsareprocessedat deeperlevelsandthusmemoryshouldbe facili-tated for thoseevents.
Improvisationtaskshavebeenusedto addresshierarchicalimplicationsformusic performance.Pianists’ improvisations on a musical themetendedtoretainfrom the themeonly structurallyimportanteventsfrom abstracthierar-chical levelsof reduction(Largeet al 1995).A neuralnetworkmodeltrainedto producereducedmemoryrepresentationsrepresentedstructurallyimportanteventsmoreefficiently thanothers,by accountingfor themusicalreductionintermsof a recursive auto-associativemechanism. The network’s weightingsofrelativeimportancecorrespondedwith boththemusicaleventsretainedacrossimprovisations andthepredictions ofstructural importancefrom areductionistmusictheory(Lerdahl & Jackendoff1983), which suggests that reductionmaybe a naturalconsequence of hierarchical encodingsof musicalstructure(Largeet al 1995).Schmuckler(1990)usedanimprovisationtaskto testperformers’expectancies for whicheventswould follow in open-endedmusical fragments.Performers’ improvised continuations reflected influencesof boththe contentsof the musical fragments andthe abstract tonal andmetricalhierarchiestypicalof Westernmusic(Krumhansl& Kessler1982,Lerdahl& Jackendoff1983).Otherstudiesindicateda correspondencebetweenthe eventsmostoften pro-ducedin improvisationsand listeners’ ratingsof how highly expectedthoseeventswere(Schmuckler1989).Thesefindingssuggestthatmusicperceptionandperformanceareboth influencedby thehierarchicalpropertiesof musicalstyles.
Structure-ExpressionRelationshipsMany findings haveestablisheda causalrelationshipbetweenmusicalstruc-tureandpatternsof performanceexpression(Clarke1988,Palmer1989,Slo-boda1983).Oneof themostwell-documentedrelationships is themarkingofgroupboundaries,especiallyphrases,with decreasesin tempoanddynamics(Henderson1936). Patternsof rubato (tempomodulations)often indicateahierarchyof phrases,with amountof slowing at a boundaryreflecting thedepth of embedding(Shaffer & Todd 1987; Todd 1985, 1989). The more
MUSIC PERFORMANCE 123
importantthemusical segment,basedon a hierarchicalanalysisof meterandgroupingprinciples(Lerdahl& Jackendoff1983),thegreaterthephrase-finallengthening.The greatestcorrespondencebetweenexpressivetiming and in-tensityin performanceis foundat anintermediatephraselevel (Palmer1996a,Todd1992),andperformers’ notatedandsoundedinterpretationstendto differmostat levelslower than thephrase (Palmer 1989,Repp1992b).
Metrical structurealsoinfluences performanceexpression. Metrical accents(eventsalignedwith strongbeatsasimpliedby notatedmetricalinformation)are often emphasizedby lengtheneddurationsand delayedonsetsin pianoperformance(Henderson1936) and in vocal performance(Palmer& Kelly1992).Pianistspresentedwith thesamemelodiesin differentnotatedmetricalcontextsplayedeventsalignedwith metricalaccentslouder,with longerdura-tions, and with more legato (smooth) articulation (Sloboda 1983,1985a).Listeners’ subsequent judgments of meter for the different performancesalignedwith performers’ metrical intentionsmost often for the mostexperi-enced pianists’ performances (whose expressive markings of meter wereclearer) (Sloboda 1983). When the different expressive cues were inde-pendently manipulated in computer-generated simulations, li steners mostoften chosethe intendedmeter primarily on the basisof articulation cues.Loudnesscuesalonecommunicatedmeteralso,but theywerenotpresentin allperformances(Sloboda1985a).In all, thesefindings suggestthat thereis noone setof necessary and sufficientexpressivecues todenote meter.
Oneof the first typesof musicalstructurefor which systematic patternsofperformanceexpressionwere documented is the durationpatternsthat formcharacteristicrhythms(Bengtsson& Gabrielsson1977). An exampleis theViennesewaltz(basedonarepeatingpatternof threeequal-durationbeatswitha metricalaccenton the first beat),typically performedwith a shortfirst beatand a long secondbeat (Askenfelt 1986, Bengtsson & Gabrielsson1977).Gabrielsson(1974) documented systematic deviations in the note durationsand amplitudes ofpianists’ and percussionists’ performances ofrepeatingrhythmic patternsto a metronomic tempo;the first noteof eachmeasurewaslouder, and notatedduration ratio relationships were increased. Listeners’ratingsof similarity amongtheseperformedrhythms(Gabrielsson1973a)andperformancesof polyphonic (multivoiced)rhythms(Gabrielsson1973b)sug-gestedthat the expressivetiming patternscanbe groupedaccordingto threefactors:structure,motion,andemotion. Structureincludedmeter,accentpat-tern, andsimplicity (of durationratios).Motion included rapidity(sound eventdensity),tempo,and forward movement. Emotion includedvitality, excited-ness,andplayfulness(Gabrielsson1982).Factoranalysesof the timing pro-files from pianoperformancesof a Mozartsonatareplicatedsomeof thesame
124 PALMER
structure-expressionrelationships found with the simpler rhythm patterns,inwhich othertypesof musicalstructurewere not present(Gabrielsson1987a).
Performanceexpressionalsoservesto differentiateamongsimultaneouslyoccurringvoicesin multivoiced music.Voicescanbe distinguishedby theirintensity or timing. Early analysesof Duo-art (player piano) rolls indicatedthat pianistsplayed tonescomprising the melodic voice soonerthan othertonesnotatedassimultaneous(Vernon1936).Recordingsof wind, string,andrecorderensemblesalsoindicatedasynchroniesamongthe voicesfor notatedsimultaneities,with a spreadof 30–50ms anda small relativelead(7 ms) oftheinstrument leadingtheensemble(Rasch1979).Theamountof spreadwaslarger for instruments whose rise(attack) timewas longer,which suggests thatmusiciansmayadjusttheasynchroniesto establish appropriatetiming of per-ceptualonsets.Measurementsof both acousticandelectronicpianoperform-ancesindicateda 20–50 ms lead of the melody over other voices (Palmer1989,1996b),longerthanthe20msneededfor listenersto determinetheorderof two isolatedtone onsets(Hirsh 1959). As interpretationsof the melodicvoice changedacrossperformances,the voice that precededother notatedsimultaneitieschangedaccordingly(Palmer1996b).Melody leadsmay serveto separatevoicesperceptually.Experiments with simpletonesequencesindi-catethat tonesthataretemporallyoffset tendto beperceivedasbelonging toseparate streams(Bregman &Pinker 1978).
Do performersusea syntaxor formal setof rulesto generateexpression?Accordingto theview thatmusicalstructureis relatedto performanceexpres-sion in termsof explicit generativeprinciples,systematic patternsof expres-sion result from transformations of the performer’s internalrepresentationofmusicalstructure(Clarke 1993, 1995). Threetypesof evidencesupporttheview thatstructuresystematically generatesexpression:theability to replicate
MUSIC PERFORMANCE 125
thesameexpressivetiming profile with very smallvariability acrossperform-ances(cf Henderson1936,Seashore1938),theability to changeaninterpreta-tion of a pieceand producedifferent expressionwith litt le practice(Palmer1989, 1996b), and the ability to perform unfamiliar music from notation (sight-read)with appropriateexpression(Palmer 1988,Shaffer1981,Sloboda1983).
Structure-expressionrelationshipshavebeenformalizedin computationalmodels that apply rules to input structural descriptions of musical scores(Sundberget al 1983a,b).In one model, three types of rules affect eventdurations,intensities,pitch tunings,andvibrato.Differentiationrulesenhancedifferencesamongcategories,groupingrulessegmentthe music,andensem-ble rulescoordinatemultiple voicesor parts(Sundberget al 1991).Anothercomputational model of performanceexpressionformalizesthe inner pulses(reflectingindividuality andviewpoint) of individual nineteenth-centurycom-posers(Clynes1986);pulsesdefinedatdifferentlevelsof musicalstructureareappliedsimilarly to all piecesby a given composerto generateperformanceexpression(Clynes 1977, 1983). Perceptualjudgmentsof model-generatedsimulations (Clynes1995,Repp 1989,Thompson1989,Thompsonetal 1989)andcomparisonswith live performanceexpression(Repp1990)providesomesupportfor thesemodels,but they indicatein generalthat piece-specificfac-tors contributeto performanceexpressionasmuchas the piece-transcendentfactors captured bythe models’ rules.
The view that musical structuregeneratesexpressionalso predicts thatperformersshouldfind it moredifficult to imitatea performancethatcontainsanarbitraryrelationshipbetweenexpressionandstructurethana conventionalone.In fact, pianistsmostaccuratelyimitateda performancethat containedaconventionalrelationshipbetweenphrasestructureandphrase-finallengthen-ing, but they could also reproducesynthesizedversionsthat containeddis-tortedstructure-expressionrelationships(Clarke1993,Clarke& Baker-Short1987).Reproduction accuracy worsenedwith increasinglydisruptedstructure-expressionrelationships, althoughaccuracyimprovedover repeatedattemptsevenfor themostdistortedtiming patterns.Listeners’ ratingsof thequality ofthe performancesdecreasedas the structure-expressionrelationshipbecamemoredisrupted(Clarke1993).Evidencethatperformerscanimitateexpressivetiming patternsthat have an arbitrary relationshipto the musical structuresuggeststhat performanceexpressionis not generatedsolely from structuralrelationships(Clarke 1993).
Perception of PerformanceExpressionWhat perceptualfunctionsdo expressiveaspectsof performanceserve?Per-formanceexpressioncan communicateparticularinterpretationsand resolvestructuralambiguities,assuggestedby the studiesreviewedabove.Perform-ance expression may also function to compensate for perceptual constraints
126 PALMER
of the auditory system. According to a bottom-up argument based on psy-choacoustic mechanisms,musiciansplay someeventslouder or longer be-causethey are heardas softer or shorterotherwise(Drake 1993). Listenersshoweddecreaseddetectionaccuracyfor experimentally lengthenedeventsplacedright beforea long durationin simplerhythmicpatterns(Drake1993),the samelocationsat which performerstendedto lengtheneventsin richermusical contexts(Drake& Palmer1993,Palmer1996a).Similar findings havebeennoted for intensity changes.Under instructions to play melodic toneswith equal intensity,pianistssystematically intensified thesecondtoneof eachgroupof four tones(Kurakataet al 1993),contraryto predictionsof metricalaccentuationon the first tone of eachgroup.Perceptualratingsof the samesequencesindicatedthat the tonesin original performancesaswell assimu-latedequal-intensityversionswerejudgedto haveequalintensities,comparedwith simulated versionsof randomizedor alteredintensities (Kurakataet al1993).Theseinitial findingssuggestthatperceptualsensitivity to temporalandintensitychangesis modulated by structuralaspectsof musicalsequences,andperformance expressionmay compensatefor thosemodulations.
The compensatorypsychoacoustic explanationof performanceexpressioncan be contrastedwith a top-downexplanationthat musicalstructureelicitsexpectationsvia listeners’ internalrepresentationof structure-expressionrules(Repp1992c).Listeners’ detectionof a single lengthenedeventin an other-wise temporally uniform (computer-generated) performance indicatedthatlengtheningwas more difficult to detectin placeswhereit was expectedtooccur (at endsof structuralunits, strong metrical positions, and points ofharmonictension) (Repp1992c).Furthermore,listeners’ detectionaccuracy(percentcorrectpereventlocation)for lengthenedeventswasinverselycorre-lated with a performer’s naturaluseof expressivelengtheningin the samemusical piece. Detectionaccuracyalso correlatedwith bottom-up acousticpropertiesof musical stimuli, including intensity and tone density charac-teristics inherentin the musical score.Thesefindings were takento reflectboth top-down and bottom-up influenceson the perceptionof performanceexpression(Repp1992c).Furtherexperimentsreplicatedthe detectionfind-ingsfor lengtheningsandextendedthedetectionparadigmto intensity changes(Repp 1995a). Although bottom-up and top-downexplanations cannotbecompletelyseparated,thefindingssuggest that the structure given in amusicalcomposition hasinherentrelationalpropertiesthat constrainboth perceptionandperformance,ratherthanperceptionsimply constraining performanceorvice versa (Repp1995a;see also Jones1987).
Psychological testsof music-theoreticmodelsof musicalexpectancyandtension-relaxationpoint to a similar explanationof the influenceof composi-tional structure in perception andperformance.Narmour’s (1990,1996)model
MUSIC PERFORMANCE 127
of melodic expectancypredictswhich eventsare most likely to occur in agiven musicalcontext.The more expectedeventsare thosethat matchtheirpreceding contextualimplications. Lerdahl’s (1996) model predicts patterns oftonal tension andrelaxationthat arisefrom harmonicrelationshipsacrosslargemusicalsections.Both musictheoriesarebasedon a combination of bottom-up (hard-wired)andtop-down(acquired)processesthataccountfor listeners’expectations.Perceptualexperimentssuggestthat listenerscanapprehendthemusic-theoreticpredictionsof melodicexpectancies(Cuddy& Lunney1995,Krumhansl1995)andtension-relaxation(Krumhansl1996)from just thecate-gorical score information presentedin computer-generated(expressionless)performances.Comparisonsof themusic-theoretic predictionswith pianoper-formanceindicatethat expressivecuesemphasizemelodic expectanciesandtension-relaxation(Palmer1996a).Unexpectedeventswereplayedlouderthanexpectedevents,andeventswith higher tensionwereperformedwith longerdurations.Thesefindings suggestthat performers’ and listeners’ interpreta-tions of certainstructuralrelationships areconstrainedin similar waysby themusicalcomposition.
MOVEMENT
After musicalstructuresandunits areretrievedfrom memoryaccordingto aperformer’s conceptualinterpretation, theymustbe transformedinto appropri-ate movements.Movement plays many roles in theoriesof music and itsperformance;for example,musicalrhythm is often definedrelative to bodymovement(Fraisse1982, Gabrielsson1982). Different views exist on thecausalrelationships betweenmusical rhythm andmovementin performance.For instance,movementcangeneraterhythmandtiming,or rhythmandtimingcangeneratemovement(Clarke 1997).Thesetwo viewsareconsideredbelow.
TimekeeperModels
Movementgeneratingtiming is themotorcontrolview: Structural information(suchasa sequence’s rhythm)maybetheinput to a motorsystem,which thenproduces somekind of temporally structuredbehavior,perhaps withthe useofinternal clocksor timekeepers.Internalclockswereproposedto accountforbehaviorssuchas the anticipation and coordinationof gesturesor acts,e.g.accompanyingmusicalsoundswith tapping.Accompaniment reflectsa syn-chronizationbetweenperceptionandproductionthat requirestheanticipationof upcomingevents.In music performance,motor systemsare thought toconstructthe information for upcomingmovementson the basisof internalclocks,which act as timekeepersby controlling the time scaleof movementtrajectories(Shaffer1981).A clock constructsbeatsat an abstractlevel thatprovidetemporalreferencepointsfor future movements.Theprimary role of
128 PALMER
an internal clock is to regulateand coordinatecomplex time seriessuchasthoseproduced between handsor betweenperformers.
Evidenceto supportclock modelscomesmainly from reproductiontasks,in whichsubjectshear and thenreproduce musicalrhythms by tapping. Peoplearemoreaccurateat reproducingmusicalrhythmswhoseinteronsetintervalsarebasedon 1:1or 2:1 ratios than on other ratios (Essens& Povel 1985, Povel1981). Both musiciansand nonmusicians reproduceduration patternsmostaccuratelywhenthedurationsarerelatedin integerratio relationships(Essens1986).Early modelsof the temporalcontrol of rhythmic sequencespositedasingle clock (Essens& Povel 1985, Povel & Essens1985), whereasotherscontrastedmultiple timekeepers(Vorberg& Hambuch1984;for a review,seeJones1990).Becausereproductiontaskscombineperceptualandmotorproc-esses,somemodelsof reproductiontiming attribute internal timekeepingtoperceptualencoding(Povel & Essens1985), whereasothersattribute it toproductionmechanisms (Vorberg &Hambuch 1978).
At what hierarchicallevel of musicaltime doesan internalclock operate?Most clock modelsexert their influenceat the level of the tactus,or mostsalientmetrical level in a musicalsequence(Essens& Povel1985,Parncutt1994).Evidencefrom sometaskssuggeststhat 600 ms may be the preferredpaceof the tactus:Peoplemostoftengeneratebeatpatternsaround600ms inspontaneousrhythmic tappingtasks(Fraisse1982),thetypical interstepinter-val foundin neutralwalking is 540ms(Fraisse1982,Nilsson& Thorstensson1989),andlistenersmostoftenusemotion termsto describe rhythmicpatternswhoseinterbeatintervalscenteraround650 ms (Sundberget al 1993).Mostinternal clock modelsapplied to music performanceproducetime periodsgreaterthanor lessthantheprimarytiming level by concatenatingor dividingbeatperiods,ratherthanby positing additional clocks (Clarke1997,Shaffer1982).
A furtherimplication of a motorsystempacedby aninternaltimekeeperorclock is that temporalvariancein performedeventdurationsmay be attribut-ableto the timekeeperor to the executingmotor system.Early modelsof thetiming mechanismsunderlying tappingbehaviorspartitioned the temporalvarianceinto lack of precisiondueto aninternaltimekeeperanddueto motorresponsedelays,basedon covarianceanalysesof the interresponseintervals(Wing & Kristofferson1973a,b).Extensionsof this modelweredevelopedtotest hierarchicalorganizationsof timekeepersoperatingat multiple metricallevelsor beatperiodsin single rhythms(Vorberg& Hambuch1978)and inpolyrhythms (Jagacinskiet al 1988).Covarianceanalysesalsoallow compari-son of whetherthe timing of eventdurationsis constructeddirectly or indi-rectly; performeddurationsat themetricallevel directly controlledby a time-
MUSIC PERFORMANCE 129
keepershould be less variable than the durationsof residualnestedeventswithin thatlevel.
Tests of hierarchicalclock modelsoperatingat various metrical levels,basedon covarianceanalyses,wereappliedto musicperformance.Compari-sonsof temporalvariancein skilled pianoperformancesindicatedthat time-keepingwasmostdirectly controlled(leastvariable)at intermediatemetricallevels of the subbeat(below the tactus),the beat,or the bar (Shaffer1980,Shafferet al 1985). Further testsof solo piano performancesindicatedthattiming wasdirectly controlledat the beatlevel (abovethe level of individualnotes),which allowedthe two handssometemporalindependencein coordi-natingnoteeventsbelowthebeatlevel thatdifferedin duration(Shaffer1984).In extensionsof covarianceanalyses,Shaffer(1981)concludedthat separatetimekeeperscontrolledthe timing of individual handsin pianoperformance.Duet performancesindicatedthat eachpianist’s timing hadhighestprecision(leastvariance)at thebar level,which suggestshow performersmight coordi-natein theabsenceof anexternalconductor(Shaffer1984).Although covari-anceanalysesrely on an assumption of constantglobal tempothat is rarelyseenin musicperformance,thesefindings suggestthat temporalprecisioninperformanceis influencedby thestructureof thesequence—inparticular,thesalience of thebeat levelor tactus.
Performancetiming canalsoexhibit stability at moreabstracthierarchicallevels, such as entire musical pieces.The durationsof string quartetsoverrepeatedperformancesby thesameperformerswere highlyconsistent(Clynes& Walker 1986). The standarddeviationof the total pieceduration(30–45min) was about 1%, smaller than that of individual movements within thepiece. If onemovementwas shortened,anothercompensatedin duration,which suggeststemporalcontrol at a level higher than the individual move-ments.A relatedtheory predictsthat the performancetemposof successivesectionsof music form simple integerratios,calledproportional tempos(Ep-stein 1995). The various periodicities that comprisea performancedisplayphasesynchrony,particularlyat structuralboundaries.Like Clynes& Walker(1986), Epstein proposedoscillator mechanisms thattrack periodicities oftempoin performanceandperceptionandspecifyrelationshipsamongsucces-sive movement durations and tempo changes in quantized steps. Similarmechanismshavebeenproposedin a modelof rhythmic attending,basedoninternalreferentperiods(preferredattentionalperiodicities)thatmaybesharedby performers andperceivers (Jones 1987).However,large-scale tempo meas-urementsmay reflect performers’ memoryfor tempo(Levitin & Cook 1996)as well as timekeeper stability, and findings based on live performances(Clynes& Walker 1986)arelimited by practicalconstraints suchasconcert
130 PALMER
hall rentalperiods.Nevertheless,thesetheoriesdosuggest thata largerangeofperiodicitiesinfluences thetiming of musicperformance.
Motor ProgramsAnother theory of temporalcontrol of performancestemsfrom motor pro-grammingviews. A motor programcontainsrepresentationsof an intendedactionandprocessesthat translatetheseinto a movementsequence(Keele&Summers 1976, Shaffer1981). The basicidea isthatasequenceof movementscanbe coordinatedin advanceof its execution.The goal of motor program-ming is to accountfor motor equivalenceacrosscontexts,the fact that thesamesequencecanbeperformedwith differentactionsandretainits fluency,expressivity, andadaptivity. One view accountsfor performers’ ability toproducethesamesequencein differentwayswith a singlegeneralizedschemathat takesparameters(Rosenbaumet al 1986, Schmidt 1975). Changesinglobal tempoacrossperformancesof the samemusicalpiecehavebeencon-ceptualizedin termsof a parameterchange.If timing of musicperformanceisrelationallyinvariantacrosstempochanges,thena changein tempoamountsto multiplying all eventdurationsby a constantvalue.Relationalinvariancewould support the existence of ageneralized motor program, in which avariablerateparameteraccountsfor performers’ ability to producethe samesequenceat different rates.Testsof relationalinvariancefor speech,typing,and walkinghave producedmixedresults(cf Gentner 1987).
Testsof relationalinvariancein musicperformancegenerallyindicatethatthe relativedurationsof noteeventstendto vary acrossperformancesof thesamemusic playedat different tempi by the sameperformer(Clarke 1982,Desain& Honing1994,MacKenzie& vanEerd1990,Repp1995b),althoughin somecasestherelativetiming patternsremainhighly similar (Repp1994).Onehypothesisfor the relativetiming changesacrosstempi is that structuralinterpretationdoes not remain constantacrossperformancetempo; for in-stance,thenumberof groupboundariesincreasedwith slowertempoin pianoperformancesof the samemusical piece (Clarke 1982). Lack of relationalinvariancesuggestsa failure of transferof learning;practicinga patternat adifferent ratethanthe intendedperformanceratemight becounterproductive.Thesefindings also warn againstdrawing structural conclusionsbasedonperformance data averaged or normalizedacross tempi.
Is the perception ofmusical structure invariant acrosstempo changes?Perceptualexperiments withperformedmonorhythms (Gabrielsson 1973a)and polyrhythms (Handel& Lawson 1983) suggestthat tempo changesdoaffect theperceptionof durationpatterns.If performersuseexpressivetimingto bringaboutadesiredstructural organizationfor aparticulartempo,differentperceptionsmight resultfor thesamerelativeexpressivetiming patternplayedat a different tempo.Repp(1995b)independently manipulatedtheamountof
MUSIC PERFORMANCE 131
expressivetiming (incrementedin termsof a powerfunction) andthe globaltempo(incrementedin termsof total pieceduration)of performances.Listen-ers gavehigher ratingsof aestheticquality to the reducedexpressionat fasttempoandto the augmentedexpressionat slow tempofor the samemusicalpieces, whichsuggeststhatlisteners preferredthe amount ofexpressive timingto changewith tempo(Repp1995b).Although theseperceptualfindings donot indicatethemechanismscontrolling performancetiming, theysuggestthata perceptualanalogueexistsfor the tempoeffectson expressivetiming docu-mentedin performance.
KinematicModels
The viewthat rhythmgeneratesmovementis reflectedin thenotion that musicperformanceandperceptionhavetheir origins in the kinematic anddynamiccharacteristicsof typical motoractions. Forexample,regularitiesobservedin asequence of footmovementsduringwalking orrunningaresimilar to regulari-tiesobservedin sequencesof beatsor notevalueswhenamusicalperformancechanges tempo.A rhythmicframework may betransmitted fromperformers tolistenersthroughsound(Shove& Repp1995),assuggestedby computationalmodelsof musicperformancein which the auditorysysteminteractsdirectlywith the motor system(Todd 1995). The kinematicsof movementallow acommonorigin for performanceandperceptualphenomena,basedon similarkinematicpropertiesapplyingacrossindividuals. Consequently, aestheticallysatisfyingperformancesshouldbe thosethat satisfykinematic constraintsofbiologicalmotion(Shove &Repp1995).
Kinematic modelswerefirst appliedto the largedecelerationsin perform-ancetempothatcommonly occurat theendsof pieces,calledthe final ritard.Pianists’ final ritards were modeledin two parts—avariable timing curvefollowed by a systematic,constantdecreasein tempo(called linear tempo)(Sundberg& Verrillo 1980).The “motor music” usedin the studies,whichcontainsa regular sequence of eventswith short durations, may createassocia-tionsfor listenerswith experiencesof physicalmotion(Kronman& Sundberg1987).Feldmanet al (1992) modeledboth ritards andpositive accelerationsthat occurredthroughoutperformances.Basedon modeling fits to the timingof a few ensembleperformances,cubic polynomial modelswere chosentominimizethejerk or jumpinessin connecting points oftempochanges(ritards)to theconstanttempothatprecededthem.Repp(1992b)modeledtheexpres-sive timing of a shortmelodicgesturein pianoperformancesof a Schumannpiece,finding a best-fittingquadraticpolynomial. Thethreeparametersrepre-sented apositiveconstant that corresponded tooveralltempo,anegativelinearcoefficient that correspondedto vertical and horizontaldisplacementof theparabola,anda positivequadraticcoefficient that correspondedto degreeof
132 PALMER
curvature.Synthesizedperformancesfor the samemelodic segmentbasedonalteredparametervalueswereplayedfor listeners,who preferredtiming pro-files thatfit the originalparabolic functions(Repp 1992a).
Although most modelsof motion in performanceaddresstiming, someapplyto dynamic(intensity) changes as well. Somemeasurementsof perform-ancesuggesta coupling betweenexpressivetiming anddynamicsin singing(Gjerdingen1988,Seashore1938)andpianoperformance(Gabrielsson1987a,Palmer1996a),in which tempoand intensity increaseanddecreasetogetherovera musicalsectionsuchasa phrase.Todd(1992)proposedanunderlyingkinetic energymodelfor performanceexpression, inwhich intensity ispropor-tional to the squareof musical velocity (numberof eventsper unit time).Contrasting the fit of different parabolicmodelsto intensity and timing pat-ternsin pianoperformances,Todd settledon a modelwith constantaccelera-tion (linear tempo).Like Sundberg& Verrillo, Todd (1992) proposedthatmusicalexpressioninducesa percept of self-motion in listeners.
Thenotion thatperformanceexpressionhasits originsin thekinematicanddynamicpropertiesof motor actionswasextendedin a generalframeworkofperceptionandperformance(Todd1995).A lineartempomodelequivalenttoKronman & Sundberg’s (1987) was fit to the expressivetiming of pianoperformancesegments,which wereidentifiedby changesin thesignof accel-eration.Todd (1995)proposedanauditorymodelof rhythmperformanceandperception,basedonatime-domainprocessthatcomputes temporalsegmenta-tion of onsets(low-passfilters) and afrequency-domain processthatcomputesa periodicity analysis(bandpassfilters). In addition,a sensory-motor feedbackfilter hastwo periodic components:the tactus(a filter centeredat 600 ms),modelingbeats,and body sway (a filter centeredat 5 sec),modelinglarge-scalebody movements. Performers’ body and limb movements can specifysomeaspectsof music performance,as evidencedin observers’ ratings ofperformances based on visual information only from point-light displays(Davidson1993,1994).Todd’s (1995)modelrequiresfurthertesting to elimi-natepotentialoverfitting of data,andits identification of line segmentscanbeproblematic.Themodel’s advantageis that it is a purelybottom-upsegmenta-tion method that requires noinput structural markers,as are requiredbyseveral of thekinematicmodelsdiscussedabove.
suggestthatcategoricaldistinctionsunderlietheperceptionof rhythmic struc-ture,andperformersuseexpressivetiming to separatedurationalcategoriesofnoteeventsevenmorewhenthe events’ absolute durationsareconvergingatfast tempi(Clarke1985).Thus,tempochangesin performancemayoperateina noncontinuous,stepwisefashionacrossabsolutedurationsto retaintheper-ceptionof intendedrhythmic categories(Desain& Honing 1992), which isanotherexplanationfor why relationalinvariancemay not hold acrosstempochanges.Although this explicationis not yet fully developed,it incorporatesperceptualconstraintsand sensitivity to musical structurein explaining thecontrolof movementin musicperformance.
CONCLUDING COMMENTSScientificstudyof musicperformancehaswitnessed tremendousgrowth inthepast tenyears, dueto both technologicaladvancesandtheoreticalinterest fromthe related fields of psychoacoustics, biomechanics,artificial intelligence,computermusic,musictheory,and music education.Performancestudiesnowdrawon conceptsfrom music theory,andstructuralparallelsfrom psycholin-guisticsareoftenfruitful. Distinctionsbetweenthepsychological mechanismsproposedfor music perceptionand performanceare becomingblurred. Forexample,listeners’ (and performers’) abilities to track the beatand recovercategoricalinformation in continuously varying performancesarenow activeissuesfor researchers inbothperceptionandperformance.Music performanceoffersa well-defineddomainin which to studybasicpsychological constructsunderlyingsequenceproduction,skill acquisition, individual differences,andemotional response,all of whichwill bethefocusof futureresearchdirections.Finally, interdisciplinary approachesto this domainaregrowing, in part be-causecurrent findings documentmusic performanceas a seeminglyuniquehumanability thatis notuniquein its underlying cognitivemechanisms.
ACKNOWLEDGMENTS
Preparationof this chapter was supportedin part by NIMH grant R29-MH45764.I gratefully acknowledgethe commentsof Eric Clarke,PeterDe-sain,CarolynDrake,HenkjanHoning,RichardJagacinski,Mari RiessJones,Bruno Repp,andJohnSloboda,andthe aid of PeterKnapp,RosaleeMeyer,BrentStansfield,and Timothy Walkerin preparing thismanuscript.
LiteratureCited
Askenfelt A. 1986. Measurement of bow mo-tion and bow force inviolin playing. J.Acoust.Soc. Am.80:1007–15
BengtssonI, Gabrielsson A. 1977. Rhythm re-
searchin Uppsala. In Music, Room, andAcoustics,17:19–56. Stockholm: R. Swed.Acad.Music
Bengtsson I, Gabrielsson A. 1983. Analysis
134 PALMER
and synthesis of musical rhythm. SeeSundberg1983, pp. 27–60
Bernstein N. 1967. The Co-ordination andRegulation of Movements. New York: Per-gamon
BregmanAS, PinkerS.1978. Auditory stream-ing andthe buildingof timbre.Can. J. Psy-chol. 32:19–31
ClarkeEF.1982. Timing in the performance ofErik Satie’s ‘Vexations’. Acta Psychol. 30:1–19
Clarke EF. 1985. Structure and expression inrhythmic performance. In Musical Struc-ture and Cognition, ed.P Howell, I Cross,R West,pp. 209–36. London: Academic
ClarkeEF. 1987. Levels of structure in the or-ganization of musical time. Contemp.Mu-sicRev.2:211–38
ClarkeEF. 1988. Generative principlesin mu-sic performance. In Generative Processesin Music: ThePsychology of Performance,Improvisation, and Composition, ed. JASloboda, pp. 1-26. New York: OxfordUniv. Press
Clarke EF. 1995. Expression in performance:generativity, perception and semiosis. SeeRink 1995, pp. 21–54
ClarkeEF.1997. Rhythm andtiming in music.See Deutsch1997. In press
ClarkeEF,Baker-Short C.1987. The imitationof perceived rubato: a preliminary study.Psychol. Music 15:58–75
Clynes M. 1977. Sentics: The Touch of Emo-tions.NewYork: Doubleday
ClynesM. 1983. Expressive microstructure inmusic,linkedto living qualities.SeeSund-berg1983, pp. 76–181
Clynes M. 1986. When time is music. InRhythm in Psychological, Linguistic andMusical Processes,ed. JR Evans, M Cly-nes,pp. 169–224. Springfield, IL: Thomas
Clynes M. 1995. Microstructural musical lin-guistics: composers’ pulsesare liked mostby the best musicians. Cognition 55:269–310
Clynes M, Walker J. 1986. Music as time’smeasure.MusicPercept.4:85–120
Cone ET. 1968. Musical Form and MusicalPerformance.New York: Norton
Cooper G, Meyer LB. 1960. The RhythmicStructure of Music. Chicago: Univ. Chi-cagoPress
Cuddy LL, Lunney CA. 1995. Expectanciesgeneratedby melodic intervals: perceptualjudgments of melodic continuity. Percept.Psychophys.57:451–62
Davidson JW. 1993. Visual perception of per-formance manner in the movements ofsolomusicians.Psychol. Music21:103–13
Davidson JW.1994. Which areasof a pianist’s
body convey information about expressiveintention to an audience? J. Hum. Mov.Stud. 26:279–301
Desain P, Honing H. 1991. Towardsa calculusfor expressive timing in music. Comput.MusicRes.3:43–94
DesainP, Honing H. 1992. Music, Mind, andMachine: Studies in Computer Music,Mu-sic Cognition, and Artificial Intelligence.Amsterdam:Thesis
DesainP, Honing H. 1994. Does expressivetiming in music performance scalepropor-tionally with tempo? Psychol. Res. 56:285–92
DeutschD, ed.1997. ThePsychologyof Music,Vol. 2. NewYork: Academic. In press
DrakeC. 1993. Perceptual andperformedac-cents in musicalsequences.Bull . Psychon.Soc. 31:107–10
DrakeC, Dowling WJ,PalmerC.1991. Accentstructures in the reproduction of simpletunesby childrenandadult pianists.MusicPercept. 8:315–34
DrakeC, PalmerC. 1993. Accentstructuresinmusic performance. Music Percept. 10:343–78
Dunsby J. 1989. Guesteditorial: performanceand analysis of music. Music Anal. 8:5–20
Epstein D. 1995. ShapingTime: Music, theBrain, and Performance. New Y ork:Macmillan
Ericsson KA, Krampe R, Tesch-Romer C.1993. The role of deliberatepractice in theacquisitionof expert performance.Psychol.Rev.100:363–406
EssensPJ,Povel D-J. 1985. Metrical andnon-metrical representations of temporal pat-terns.Percept. Psychophys.37:1–7
Feldman J, Epstein D, Richards W. 1992.Forcedynamicsof tempo changein music.MusicPercept.10:185–204
FraisseP. 1982. Rhythm and tempo. In ThePsychology of Music, ed. D Deutsch, pp.149–80. NewYork: Academic
GabrielssonA. 1973a. Similarity ratings anddimensionanalysesof auditory rhythmpat-terns. I. Scand. J. Psychol. 14:138–60
GabrielssonA. 1973b. Similarity ratings anddimensionanalysesof auditory rhythmpat-terns. II. Scand. J. Psychol. 14:161–76
GabrielssonA. 1974. Performanceof rhythmpatterns.Scand. J. Psychol. 15:63–72
GabrielssonA. 1982. Perception andperform-anceof musical rhythm. In Music, Mind,and Brain: The Neuropsychology of Music,
MUSIC PERFORMANCE 135
ed.M Clynes,pp. 159–69. New York: Ple-num
Gabrielsson A. 1987a. Onceagain: the themefrom Mozart’s piano Sonata in A Major(k.331). See Gabrielsson 1987b, pp. 81–104
Gabrielsson A, ed.1987b. Action and Percep-tion in Rhythm and Music. Stockholm: R.Swed.Acad.Music
Gabrielsson A. 1995. Expressive intention andperformance.In Music and the Mind Ma-chine, ed. R Steinberg, pp. 35–47. Berlin:Springer-Verlag
Gabrielsson A. 1997. Music performance.SeeDeutsch1997. In press
Gabrielsson A, Juslin PN. 1996. Emotional ex-pression in music performance: betweenthe performer’s intention andthe listener’sexperience.Psychol. Music24:68–91
Garcia-AlbeaJE, del Viso S, Igoa JM. 1989.Movementerrors and levels of processingin sentenceproduction. J. Psycholinguist.Res.18:145–61
Garrett MF. 1980. Levels of processingin sen-tenceproduction. In Language Production.Speech and Talk, ed. B Butterworth, 1:177–220. London: Academic
Gentner DR. 1987. Timing of skilled motorperformance:testsof theproportionaldura-tionmodel. Psychol. Rev.94:255–76
Geringer JM, Madsen CK. 1987. Program-matic research in music: perception andperformanceof intonation. In Applicationsof Research in Music Behavior, ed. CKMadsen, CA Prickett, pp. 244–53.Tus-caloosa:Univ. Ala. Press
GjerdingenRO.1988. Shape andmotion in themicrostructure of song. Music Percept.6:35–64
Handel S, LawsonGR. 1983. The contextualnature of rhythmic interpretation. Percept.Psychophys.34:103–20
HendersonMT. 1936. Rhythmic organizationin artistic pianoperformance.SeeCE Sea-shore 1936, pp. 281–305
Hirsh IJ. 1959. Auditory perception of tem-poral order. J. Acoust. Soc. Am. 31: 759–67
Jagacinski RJ, Marshburn E, Klapp ST, JonesMR. 1988. Tests of parallel versus inte-gratedstructure in polyrhythmic tapping. J.Mot. Behav. 20:416–42
Johnson C. 1996. An empirical investigation ofmusicians’ and nonmusicians’ assessmentof perceived rubato in musical perform-ance.J. Res.MusicEduc.44:84–96
Jones MR.1987. Perspectivesonmusical time.See Gabrielsson1987b, pp. 153–75
JonesMR. 1990. Musicaleventsandmodelsofmusicaltime. In Cognitive Models of Psy-chological Time, ed. RA Block, pp. 207–40. Hil lsdale,NJ:Erlbaum
KeeleSW. 1968. Movementcontrol in skilled
motor performance.Psychol. Bull . 70:387–403
KeeleSW,Summers JJ.1976. The structure ofmotor programs.In Motor Control: Issuesand Trends, ed.GE Stelmach,pp. 109–42.NewYork: Academic
Kendall RA, Carterette EC.1990. Thecommu-nication of musical expression. MusicPer-cept. 8:129–64
KronmanU, Sundberg J. 1987. Is the musicalritard an allusion to physical motion? SeeGabrielsson 1987b, pp. 57–68
Krumhansl CL. 1995. Music psychology andmusic theory: problemsandprospects.Mu-sicTheory Spect.17:53–80
KrumhanslCL. 1996. A perceptual analysisofMozart’s PianoSonata,K. 282: segmenta-tion, tension, and musical ideas. MusicPercept. 13:401–32
KrumhanslCL, KesslerEJ. 1982. Tracing thedynamic changesin perceivedtonalorgani-zation in a spatial representationof musicalkeys.Psychol. Rev.89:334–68
KurakataK, Kuwano S, NambaS. 1993. Fac-tors determining the impression of theequality of intensity in pianoperformances.J. Acoust.Soc. Jpn. E 14:441–47
Langer S. 1953. Feeling and Form: A Theoryof Art. New York: Scribner
Large EW, Palmer C, Pollack JB. 1995. Re-duced memory representations for music.Cogn. Sci. 19:53–96
Lashley K. 1951. The problem of serialorderin behavior. In Cerebral Mechanisms inBehavior: The Hixon Symposium, ed. LAJeffress,pp. 112–36. New York: Wiley
Lehmann AC, EricssonKA. 1993. Sight-read-ing abili ty of expert pianists in the contextof piano accompanying. Psychomusicology12:182–95
Lerdahl F. 1996. Calculating tonal tension.MusicPercept.13:319–64
Lerdahl F, Jackendoff R. 1983. A GenerativeTheory of Tonal Music. Cambridge, MA:MIT Press
Levitin DJ,Cook PR. 1996. Memory for musi-cal tempo: additional evidence that audi-tory memoryis absolute. Percept. Psycho-phys.58:928–36
Levy JM. 1995. Beginning-ending ambiguity:consequencesof performance choices.SeeRink 1995, pp. 150–69
MacKenzieCL, VanEerdDL. 1990. Rhythmicprecision in the performance of pianoscales: motor psychophysics and motorprogramming. In Attention and Perform-ance: Motor Representation and Control,ed. M Jeannerod, 13:375–408. Hil lsdale,N.J.:Erlbaum
McPhersonGE. 1995. The assessmentof mu-sical performance:development and vali-dation of five newmeasures.Psychol. Mu-sic23:142–61
Moore RS.1994. Effectsof age,sex,andme-lodic/harmonic patterns on vocal pitch-matching skills of talented8-11–yearolds.J. Res.Music Educ. 42:5–13
Nakamura T. 1987. The communication of dy-namics between musicians and listenersthrough musical performance. Percept.Psychophys.41:525–33
Narmour E. 1990. The Analysisand Cognitionof Basic Melodic Structures: The Implica-tion-Realization Model. Chicago: Univ.ChicagoPress
Narmour E. 1996. Analyzingform andmeasur-ing perceptual content in Mozart’s SonataK. 282: a new theory of parametric ana-logues.MusicPercept.13:265–318
Nilsson J,ThorstenssonA. 1989. Ground reac-tion forces at different speeds of humanwalking andrunning. Acta Physiol. Scand.136:217–27
Nuki M. 1984. Memorization of pianomusic.Psychologia 27:157–63
Palmer C. 1988. Timing in skilled piano per-formance.PhD thesis.Cornell Univ., Ith-aca, NY
Palmer C. 1989. Mapping musical thought tomusical performance. J. Exp. Psychol.:Hum.Percept.Perform. 15:331–46
PalmerC. 1992. Therole of interpretiveprefer-encesin music performance.In CognitiveBasesof Musical Communication, ed.MRJones,S Holleran, pp. 249–62. Washing-ton, DC: Am. Psychol. Assoc.
Palmer C. 1996a. Anatomy of a performance:sourcesof musical expression. Music Per-cept.13:433–54
Palmer C. 1996b. On the assignmentof struc-ture in music performance.MusicPercept.14:21–54
Palmer C, Kelly MH. 1992. Linguistic prosodyandmusical meterin song. J. Mem.Lang.31:525–42
Palmer C, vande Sande C. 1993. Units ofknowledgein music performance. J. Exp.Psychol.: Learn. Mem. Cogn. 19:457–70
Palmer C, van de Sande C. 1995. Range ofplanning in skilled music performance. J.Exp.Psychol.: Hum.Percept. Perform. 21:947–62
Parncutt R. 1994. A perceptual model of pulsesalience and metrical accent in musicalrhythms.MusicPercept. 11:409–64
Povel D-J, EssensP. 1985. Perception of tem-poral patterns.MusicPercept. 2:411–40
Rasch RA. 1979. Synchronization in per-
formedensemble music. Acustica43: 121–31
ReppBH. 1989. Expressive microstructure inmusic: apreliminaryperceptual assessmentof four composers’ “pulses.” Music Per-cept. 6:243–74
ReppBH. 1990. Patternsof expressivetimingin performancesof a Beethovenminuetbynineteen famous pianists. J. Acoust. Soc.Am.88:622–41
ReppBH. 1992a. A constraint on the expres-sive timing of a melodic gesture: evidencefrom performance andaesthetic judgment.MusicPercept.10:221-42
ReppBH. 1992b. Diversity and commonalityin musicperformance: an analysis of tim-i ng mi crostructure i n Schumann’ s“Traumerei.” J. Acoust. Soc. Am. 92:2546–68
ReppBH. 1992c. Probing the cognitive repre-sentation of musical time: structural con-straints on the perception of timing pertur-bations.Cognition44:241–81
ReppBH. 1994. Relational invarianceof ex-pressive microstructure across globaltempo changes in music performance: anexploratory study. Psychol. Res. 56:269–84
ReppBH. 1995a. Detectabil ity of duration andintensity increments in melody tones: apartial connection between music percep-tion and performance. Percept. Psycho-phys.57:1217–32
ReppBH. 1995b. Quantitativeeffectsof globaltempo on expressive timing in music per-formance:someperceptual evidence. Mu-sicPercept. 13:39–57
Rink J,ed.1995. ThePracticeof Performance:Studies in Musical Interpretation. Cam-bridge:CambridgeUniv. Press
Rosenbaum DA, Weber RJ, Hazelett WM,Hindorff V. 1986. The parameter remap-ping effect in human performance: evi-dencefrom tonguetwistersandfinger fum-blers.J. Mem. Lang. 25:710–25
Schenker H. 1969. FiveGraphic Music Analy-ses.NewYork: Dover
Schmalfeldt J.1985. Ontherelation of analysisto performance: Beethoven’s Bagatell esOp. 126, Nos. 2 and 5. J. Music Theory29:1–31
Schmidt RA. 1975. A schematheory of dis-cretemotor skill learning.Psychol. Rev.82:225–60
Schmuckler MA. 1989. Expectation in music:investigation of melodic and harmonicprocesses.MusicPercept.7:109–50
Schmuckler MA. 1990. The performance ofglobal expectations. Psychomusicology 9:122–47
Shaffer LH. 1980. Analyzing piano perform-ance:a study of concertpianists.In Tutori -als in Motor Behavior, ed.GE Telmach, JRequin, pp. 443-45. Amsterdam: North-Holland
Shaffer LH. 1981. Performancesof Chopin,Bach, andBartok: studies in motor pro-gramming. Cogn. Psychol. 13:326–76
Shaffer LH. 1982. Rhythm andtiming in skill .Psychol. Rev.89:109-22
Shaffer LH. 1984. Timing in soloandduet pi-ano performances. Q. J. Exp. Psychol.36A:577–95
Sloboda JA. 1974. The eye-handspan:an ap-proach to the study of sight reading. Psy-chol. Music 2:4–10
Sloboda JA. 1976. The effect of item positionon the likelihood of identifi cation by infer-encein prose reading and music reading.Can. J. Psychol. 30:228–37
SlobodaJA. 1977. Phraseunitsasdeterminantsof visualprocessingin music reading. Br.J. Psychol. 68:117–24
Sloboda JA. 1983. The communication of mu-sical metre in piano performance. Q. J.Exp.Psychol. A 35:377–96
Sloboda JA. 1985a. Expressive skill in two pi-anists: metrical communication in real andsimulated performances. Can. J. Psychol.39:273–93
Sloboda JA. 1985b. The Musical Mind: TheCognitive Psychology of Music. Oxford:Clarendon
Sloboda JA,DavidsonJW,HoweMJA, MooreDG. 1996. Therole of practicein thedevel-opment of performing musicians. Br. J.Psychol.
Sundberg J, ed. 1983. Studies of Music Per-formance.Stockholm: R. Swed.Acad.Mu-sic
Sundberg J, Askenfelt A, Fryden L. 1983a.Musical performance:a synthesis-by-ruleapproach.Comput. MusicJ. 7:37–43
SundbergJ, FribergA, FrydenL. 1991. Com-
mon secretsof musiciansandlisteners:ananalysis-by-synthesisstudy of musical per-formance.In Representing Musical Struc-ture, ed. P Howell, R West, I Cross, pp.161–97. London: Academic
SundbergJ, FribergA, FrydenL. 1993. Musicand locomotion—perception of toneswithenvelopes replicating force patterns ofwalking. Presented at Stockholm MusicAcoust.Conf., Stockholm
Sundberg J,Fryden L, Askenfelt A. 1983b.What tells you the player is musical? Ananalysis-by-synthesis study of music per-formance.See Sundberg1983, pp. 61–75
SundbergJ, Verrillo V. 1980. On the anatomyof the retard: a study of timing in music. J.Acoust.Soc. Am.68:772–79
Sundin N-G. 1984. Musical interpretation inperformance: music theory, musicology,and musical consciousness.J. Music Res.5:93–129
Thompson WF. 1989. Composer-specific as-pects of musical performance:an evalu-ation of Clyne’s theory of pulse for per-formancesof Mozart andBeethoven. Mu-sicPercept. 7:15–42
Todd NPM. 1985. A model of expressivetim-ing in tonalmusic.MusicPercept.3:33–58
Todd NPM. 1989. A computational model ofrubato. Contemp.MusicRev.3:69–88
Todd NPM. 1992. The dynamicsof dynamics:a model of musical expression. J. Acoust.Soc. Am.91:3540–50
Todd NPM. 1995. The kinematics of musicalexpression. J. Acoust. Soc. Am. 97:1940–49
vanGalenG, WingAM. 1984. Thesequencingof movements. In The Psychology of Hu-manMovement, ed. M Smyth, AM Wing,pp. 153-82. London: Academic
Vernon LN. 1936. Synchronization of chordsin artistic piano music. SeeCE Seashore1936, pp. 306–45
Vorberg D, Hambuch R. 1978. On the tempo-ral control of rhythmic performance.In At-tention and Performance, ed. J Requin, 7:535–55. Hil lsdale,NJ:Erlbaum
VorbergD, Hambuch R. 1984. Timing of two-handed rhythmic performance. Ann. NYAcad. Sci.423:390-406
Wing AM, Kristofferson AB. 1973a. Responsedelaysandthe timing of discretemotor re-sponses.Percept. Psychophys.14:5–12
Wing AM, KristoffersonAB. 1973b. The tim-ing of interresponse intervals. Percept.Psychophys.13:455–60
