JournalofExperimentalPsychology: ©2009AmericanPsychologicalAssociation HumanPerceptionandPerformance 0096-1523/09/$12.00 DOI:10.1037/a0013482 2009,Vol.35,No.1,264–280 The Role of Accent Salience and Joint Accent Structure in Meter Perception Robert J. Ellis and Mari R. Jones TheOhioStateUniversity Previousresearchindicatesthattemporalaccents(TAs;accentsduetotimechanges)playastrongrole inmeterperception,butevidencefavoringaroleformelodicaccents(MAs;accentsduetopitchchanges) ismixed.TheauthorsclaimthatthismixedsupportforMAsistheresultofafailuretocontrolforaccent salienceandaddressedthishypothesisinExperiment1.Listenersratedthemetricalclarityof13-tone melodiesinwhichthemagnitudeandpatternofMAsandTAswerevaried.Resultsshowedthatmetrical clarityincreasedwithbothMAandTAmagnitude.InExperiment2,listenerswereaskedtoratemetrical clarityinmelodieswithcombinationsofMAandTApatternstoallowtheauthorstoascertainwhether these two accent types combined additively or interactively in meter perception. With respect to the additiveorinteractivedebate,thefindingshighlightedtheimportanceof(a)accentsalience,(b)scoring methods,and(c)conceptualversusstatisticalinterpretationsofdata.Implicationsfordynamicattending andneuropsychologicalinvestigationsarediscussed. Keywords:accentsalience,meterperception When a listener taps to a musical event, taps are typically in An important contributor to meter is the relative timing of synchronywithaperceived“beat,”reflectingaperceivedtemporal accents. An accent refers to a tone in a sequence that stands out regularityinthemusicalsurface(cf.Snyder&Krumhansl,2001; from other tones along some auditory dimension (e.g., pitch, Temperley,2001).Howthesebeatsareorganizedintimeisoneof intensity, timbre, duration). More concretely, an accent is a “de- the functions of meter. Meter is fundamental to music; it is what viation from a norm that is contextually established by serial makesawaltzawaltzandamarchamarch.Itisalsosignificant constraints” (Jones, 1987, p. 624); thus, an accent acquires its psychologically, in that the perception of meter depends on a statusfromsurroundingtones(Cooper&Meyer,1960).Accented listener’s ability to recover temporal regularity from a sound tonesoftencorrespondtostrongerbeats(Cooper&Meyer,1960; patterncontainingtemporalirregularities.Furthermore,becauseof Handel,1989;Huron&Royal,1996;Jones,1987,1993;Lerdahl its periodic nature, an established metrical framework affords a &Jackendoff,1983). basisforalistenertogenerateanticipationsabouttimingsoffuture Two common types of accents are melodic accents (MAs; events, thereby guiding attending in a dynamic fashion (Jones, accents due to pitch relationships) and temporal accents (TAs; 1976, 1990, 2001, 2004; London, 2001). Picking up on temporal accentsduetotimerelationships).TheprecisenatureofMAand regularity in a musical event facilitates listener performance in a TAcontributionstomusicperceptioncontinuestobeanimportant number of tasks, including (a) judgments of event duration (e.g., research topic. One line of research has sought to determine Boltz,1991,1998;Jones&Boltz,1989),(b)judgmentsofmelodic whetherMAsandTAs,takentogether,makeadditive(e.g.,Palmer phrase completeness (e.g., Boltz, 1989a), (c) discrimination of & Krumhansl 1987a, 1987b) or interactive (e.g., Jones, 1987, pitchchanges(e.g.,Jones,Boltz,&Kidd,1982;Jones,Johnston& 1993; Jones & Boltz, 1989) contributions to the listening experi- Puente,2006;Jones,Moynihan,MacKenzie,&Puente,2002),and ence. As Schellenberg, Krysciak, and Campbell (2000) have (d) discrimination of time changes (e.g., Large & Jones, 1999). pointed out, the answer to this question has implications for Even very young listeners are sensitive to differences between low-level representations of pitch and duration as well as higher metrical frameworks (Hannon & Johnson, 2005; Hannon & level representations of phrase grouping, expectancies, and emo- Trehub,2005,2006). tion,makingitacriticalissuetothefieldofmusicpsychology. Alternatively,otherresearchsuggeststhatMAscontributeneg- RobertJ.EllisandMariRiessJones,DepartmentofPsychology,The ligibly to music perception (e.g., Hannon, Snyder, Eerola, & OhioStateUniversity. Krumhansl, 2004; Huron & Royal, 1996; Snyder & Krumhansl, ThisresearchwascompletedaspartofRobertJ.Ellis’smaster’sthesis 2001). However, such findings create a logical problem: If MAs and was generously supported by the Caroline B. Monahan Fund for failtoreliablycontributetometerperception,thendiscussionsof Experimental Research Support in the Music Cognition/Perception Area whether MAs and TAs combine additively or interactively in within the Department of Psychology. We gratefully acknowledge com- meterperceptionbecomemoot.Thus,beforeaddressingtheissue mentsfromDevinMcAuleyandMollyHenryandextensivefeedbackfrom of additivity versus interactivity (the focus of Experiment 2), we PeterPfordresher. firstexaminethemixedevidenceregardingtheimpactofMAsin CorrespondenceconcerningthisarticleshouldbeaddressedtoRobertJ. Ellis, 175 Psychology, The Ohio State University, 1835 Neil Ave., meter perception. In reviewing this literature, we note that one Columbus,OH43201.E-mail:[email protected] overlooked reason for mixed outcomes involves the stimuli in- 264 METERPERCEPTION 265 volved.StimuliwithTAsthataremorepronouncedthanMAsmay tokenterminologyisusefulhereindistinguishingbetweengeneral bias listeners toward TA information. Only when the salience of and particular; cf. Esposito, 1998; Levelt, 1989). For example, a MAsiscomparabletothatofTAscantheirrelativecontributions salient pitch change might occur if three successive tones, each (withinthesamemelody)beassessed.Asananalogy,supposethat ascending by 1 semitone (ST), were followed by a pitch that astudyreportsthatsubjectsexhibitapreferenceforeatingapples ascended by 5 ST. A salient time change might occur if three from one bowl over oranges from another, but neglects to report successiveshortIOIswerefollowedbyalongIOI. thatthebowloforangeswasoutofsubjects’reach.Thisconclu- TheMAtyperefersbroadlytoanysalientlocalserialchangein sionismeaningless:Onlyifthebowlsofapplesandorangeswere pitch relationships and has three common tokens (cf. Huron & equally accessible would the finding that subjects preferred one Royal, 1996; Jones, 1987, 1993). First, a pitch-contour accent overtheotherbeinteresting. dependsonatemporalorderingofpitches;itresultsfromalocal Issues of additivity versus interactivity in perception were ad- change in the direction of pitch trajectory (e.g., ascending to dressedinclassicworkbyGarner(1970,1974),whoproposedthat descending), with accentuation on the inflection point (cf. Tho- relationships between two dimensions exist along a continuum massen,1982).Second,apitch-leapaccentfallsonthesecondof rangingfromseparable(i.e.,additive)tointegral(i.e.,interactive). twotonesformingapitchinterval,whenthatintervalislargerthan However, Garner also noted that when two constituent stimulus preceding pitch intervals in a series (cf. Tekman, 1997, 1998; dimensionsarenotmatchedforsalience,themoresalientdimen- Thomassen,1982).Third,atonalaccentarisesfromaserialshift sioncanaffecttheperceptionofthelesssalientdimension,thereby in“stability”withinatonalcontext,thatis,fromtheleadingtone obscuring these relationships (Garner & Felfoldy, 1970; see also tothetonic(keynote;cf.Bharucha,1984;Dawe,Platt,&Racine, Melara & Marks, 1993, 1994; Tekman, 1997, 1998, 2001). Our 1993;Smith&Cuddy,1989). investigation of MA and TA salience fits within this broader The TA type refers to any salient local serial change in time context. relationshipsandhastwocommontokens.Arhythmicaccent(or Inwhatfollows,weoutlinetwoopposinghypothesesaboutthe pause accent; e.g., Jones & Pfordresher, 1997) results from a respective roles of MAs and TAs in meter perception. The joint change within a serial pattern of IOIs (e.g., three 200-ms IOIs accentstructurehypothesis,whichpositsrolesforbothMAsand followedbya600-msIOI),anditslocationdependsontheserial TAs,iscontrastedwithatemporalaccentbiashypothesis,which context(Jones,1987;Jones&Pfordresher,1997;Narmour,1996; holds that meter perception depends primarily on TAs. Both hy- Povel & Essens, 1985; Povel & Okkerman, 1981). A duration potheses have received support, but because previous investiga- accentoccursonatonethathasalongerduration(i.e.,fromtone tions have failed to systematically calibrate the salience of MAs onset to offset) than neighboring tones in a sequence (Woodrow, and TAs in the stimuli employed, their conclusions remain tenu- 1951;cf.Castellano,Bharucha,&Krumhansl,1984). ous.Thisleadsustoformulateathird,anaccentsaliencehypoth- esis,whichmakesexplicittherelationshipbetweenserialchange The Role of MAs in Meter Perception: Two andaccentsalience. Contrasting Views Meter, Accent Types, and Accent Tokens Dynamic Attending and Joint Accent Structure Meter involves a succession of strong and weak beats, equally One view that favors a role for both MAs and TAs in meter spacedintimeandorganizedintometricalframeworks(Lerdahl& perception comes from dynamic attending theory (Jones, 1976, Jackendoff, 1983; Justus & Bharucha, 2002). Two of the most 2004; Jones & Boltz, 1989; Large & Jones, 1999; McAuley & commonmetricalframeworksintheWesternclassictraditionare Jones, 2003). This approach assumes that recurrent time spans duple and triple meter (cf. Huron, 2006, p. 195). Duple meter withinreal-worldeventscansynchronizeattendingviathemech- refers to a pattern of alternating strong (S) and weak (w) beats anismofentrainment.Entrainmentisthephysicalprocesswhereby (SwSwSw ... ), frequently heard in marches. Triple meter, con- internalattendingperiodicitiesbecomeattunedtosalientrecurrent sistingofastrongbeatfollowedbytwoweakbeats(SwwSww...), stimulustimespans.Resultingattentionalsynchroniesarepossible is associated with waltzes. Meter implies temporal invariance in atmultipletimescales,andtheyarefacilitatedwhentimespansat two levels: shorter (lower order) time spans marked by the inter- differenttimescalesarehierarchicallynested.Inauditorypatterns, onset intervals (IOIs) between successive S and w elements; and suchtimespansaremarkedbyaccentsthatarisefromsalientserial longer (higher order) time spans, or metrical periods, marked by changesinpitch(MAs)and/ortiming(TAs).WhenbothMAsand successiveSelements(cf.Benjamin,1984;Lerdahl&Jackendoff, TAs are present in a single pattern, they contribute to the emer- 1983;Yeston,1976). gence of a common higher order time structure, or joint accent Accentsareusedtomarkstrongbeatsandhenceareimportant structure(JAS;e.g.,Boltz&Jones,1986;Jones,1987,1993). tometer(Cooper&Meyer,1960;Handel,1989;Huron&Royal, Atabasiclevel,aJASreferstotherelativetimingsofdifferent 1996;Jones,1987;1993;Lerdahl&Jackendoff,1983).Intheory, accent types within a melody. These properties are illustrated in accents of different types correspond to defining acoustic dimen- Figures 1 and 2. In Figure 1, local serial changes in a single, sions(e.g.,pitch,time).Ourapproachgoesfurtherthanothersin isochronous, melodic line instantiate both pitch-leap accents (an itsexplicitacknowledgmentoftheroleoftimeindefiningaccents MA token; large open circles) and duration accents (a TA token; ofalltypes.Thatis,accenttypeistakentomeanafamilyofsalient large solid circles). In this example, the accent periods of both local changes over time along a common dimension (e.g., pitch, MAsandTAsareconsistentlydoublethelowerordertimespans loudness,duration).Inauditorysequences,serialvariationsalong (IOIs).Thisrelationshipresultsininvariantaccentperiodswithin each dimension (type) express token accents of that type (type/ amusicalevent.Accordingtodynamicattendingtheory,invariant 266 ELLISANDJONES long short ttoonnee ttoonnee period= 2 ×IOI (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) MA aacccceenntt ttyyppee TA (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) period= 2 ×IOI 11 22 3333 44 5555 6666 77 serial position Figure1. Anexampleofthejointaccentstructure(JAS)foranisochronousmelody.Melodicaccents(MAs) temporalaccents(TAs),andunaccentedtonesareindicatedbythelargeopencircles,largesolidcircles,and smallcircles,respectively.IOI(cid:1)inter-onsetintervals. accent timing should promote more effective entrainment than 1991; Boltz & Jones, 1986; Deutsch, 1980; Drake, Dowling, & variableaccenttiming(Large&Jones,1999). Palmer,1991),(d)detectdevianttonesmoreaccurately(Dowling, Figure2illustratesfourJASsthatemergefromcombinationsof Lung,&Herrbold,1987;Monahan,Kendall,&Carterette,1987), duple (D (cid:1) 2 (cid:2) IOI) and triple (T (cid:1) 3 (cid:2) IOI) MA and TA (e)identifyconstituentaccentpatternsmoreaccurately(Keller& patterns.1Throughoutthisarticle,weusethephraseaccentpattern Burnham, 2005), and (f) synchronize to constituent tones more (eitherMApatternorTApattern)torefertoatemporalsuccession precisely (Jones & Pfordresher, 1997; Pfordresher, 2003). In ad- ofaccentsandthephrasemetertorefertolisteners’percepts.We dition,thetemporalstructureofasequencecanhighlightthepitch useashorthandnotationtodescribeJASs,witheachaccentperiod ortonalstructureofamelody(Jonesetal.,1982;Laden,1994). (D, T) subscripted by its accent type (MA, TA); for example, In the dynamic attending approach, specific predictions are D T denotes a melody with a duple MA pattern and a triple madeaboutmeter.Indiscordant-JASmelodies,temporalcomplex- MA TA TApattern.Complexityreferstotheoverallirregularityofaccent itiesresultinlessefficiententrainment,reducingalistener’sability timing;JAScomplexitydependsuponconstituentaccentpatterns totracksuccessivetonesinrealtime(Jones&Pfordresher,1997; (i.e., both MA and TA periods). An index of JAS complexity is Pfordresher,2003)andreducingmetricalclarity.Akeyprediction given by the quotient (or ratio; e.g., Jones, 1987) of the larger isthatconcordantJASswillelicitfasterandstrongerperceptions accentperioddividedbythesmalleraccentperiod.InbothFigures of metrical clarity than discordant JASs, due to facilitation from 1and2A,forexample,theMAandtheTApatternsarebothduple regularly timed accents in concordant JASs and/or interference (DMADTA),andtheaccentperiodquotientis1((cid:1)2/2),indicating fromconflictingaccentperiodsindiscordantJASs.Thisprediction low temporal complexity (high regularity). In Figure 2B assumes that MAs and TAs have roughly equal salience. We (TMATTA),bothaccentpatternsaretriple,andthequotientisagain restate this as a JAS hypothesis: When multiple accent types or 1 ((cid:1) 3/3), again indicating low complexity. In general, small tokensarepresentinamelody,meterperceptionisdeterminedby integer quotients (1, 2, 3, 4) indicate low temporal complexity. JASproperties.Thus,aconcordantJASwillelicittheperceptofa ThesehighlyregularJASpatternsaretermedconcordant(Jones& metermoreclearlyandimmediatelythanadiscordantJAS. Pfordresher,1997).Bycontrast,morecomplexJASsemergewhen dupleandtripleaccentpatternsarebothpresentinasinglemelody, asinFigures2C(T D )and2D(D T ).Inthesepatterns, 1Inthesepatterns,allinitialphasedifferencesbetweenaccentpatterns MA TA MA TA accent timing is less regular, as indexed by a noninteger accent were0;thatis,bothdupleandtripleaccentpatternsbeginwithanaccent periodquotient,1.5((cid:1)3/2).TheseJASsaretermeddiscordant. onTone1.Accentphasinghasbeenconsideredelsewhere(Boltz&Jones, Dynamic attending theory implies that lower JAS complexity 1986; Jones, 1987, 1993; Pfordresher, 2003). Phase shifting two accent periods(thatareotherwiseconcordant)leadstoslightlyworseperformance facilitates listener comprehension, due to more efficient entrain- (i.e.,inmelodyreproductionortappingsynchronization)thannonshifted ment.Thisviewissupportedbyempiricalevidence.Whenlisten- concordantmelodies,butbetterperformancethanmelodieswithdiscordant ingtomelodieswithaconcordantJAS(comparedwithlisteningto JASs(whetherphase-shiftedornot),dependingonthephaseofashift.In discordant-JAS melodies), listeners (a) judge the melodies’ end- thepresentstudy,settingtheinitialphaseto0wasnecessarytopreserve ings as more complete (Boltz, 1989a, 1989b), (b) reproduce the globalaspectsofmelodies(i.e.,twoascendingsix-tonecells,withaccents melodies’durationsmoreaccurately(Boltz,1998;Jones,Boltz,& onTones1,7,and13)despitevariationinconstituentaccentpatternsover Klein,1993),(c)recallmelodicstructuremoreaccurately(Boltz, thesession;seetheMethodssectionofExperiment1formoredetails. METERPERCEPTION 267 Accent serial position Accent JAS Period period type 1 2 3 4 5 6 7 8 9 10 11 12 13 quotient label D MA A MA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 2 1.0 D D TA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 2 MA TA D TA T MA MA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 3 B 1.0 T T TTAA (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) 33 MA TA T TA T MA C MMAA (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) (cid:127)(cid:127) 33 1.5 T D TA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 2 MA TA D TA D MA D MA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 2 1.5 D T TA (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) (cid:127) 3 MA TA T TA Figure 2. A schematic illustration of four joint accent structures (JASs) formed by combinations of accent periods (duple, triple) marked by melodic accents (MA) and temporal accents (TA). All tone onsets are isochronous.ConcordantJASsappearinPanelsAandB,anddiscordantJASsinPanelsCandD. We note that the dynamic attending approach is not the only AdditionalevidencefavoringtheTAbiashypothesiscomesfrom theoryinwhichMAsarepositedtoplayaroleinperception(cf. ananalysisofmusicalscores.Randomlysamplingalargecorpusof Dixon & Cambouropoulos, 2000; Temperley & Bartlette, 2002; folktunes,HuronandRoyal(1996)correlatedtemporallocationsof ToiviainenandSnyder,2003).However,dynamicattendingtheory MA and TA tokens with their respective metrical positions (as in differsfromotherapproachesinitsincorporationof(a)accentsas scoredtimesignatures).Onlyrhythmicaccentsyieldedpositivecor- serialchanges,(b)JASpatterns,and(c)entrainmentactivities. relations between accent size and scored metrical position (their Experiment 1). Furthermore, even in melodies containing no TAs (their Experiment 2), neither pitch-leap nor pitch-contour accents Temporal Accent Bias significantlycorrelatedwiththestrongmetricalpositions.Huronand Royal(1996)concludedthatthis“callsintoquestion”theclaimthat Although the JAS hypothesis assumes that both MA and TA pitch-leapsfunctionasaccents(p.501). accent patterns can contribute equally to meter, this assumption Finally, a number of theoretical approaches also assume (ex- hasbeenquestioned.Inavarietyoftasks,MAshavebeenfound plicitlyortacitly)thatperceptsofmeterarebiasedtowardtempo- tocontributerelativelylittleinformationcomparedwithTAs.Inan ral accent patterns (e.g., Desain & Honing, 1989; Drake & Ber- earlystudy,Woodrow(1911)foundthatMAswereinconsistentin trand,2001;Johnson-Laird,1991;Longuet-Higgins&Lee,1982; theirabilitytoinducetonegroupings;somelistenersheardthemas Parncutt, 1994; Povel & Essens, 1985; for a review, see Clarke, beginning a group, others as ending a group. Monahan and Cart- 1999).Togetherwiththeempiricalfindings,theseaccountscanbe erette(1985)examinedratingsofsimilarityforpairsofmelodies summarizedinaTAbiashypothesis:Metricalperceptsarebiased that varied in both MA and TA properties and found that most towardthepatternoftemporal(ratherthanmelodic)accents.When subjectsbasedtheirratingsonTA,notMA,properties.Snyderand both MA and TA patterns are present, listeners are biased to use Krumhansl (2001) asked subjects to “tap to the beat” of two TApatternswhenperceivingmeter. versions of ragtime piano music (rhythms only vs. pitch plus rhythms);theyfoundthattheadditionofpitchinformationdidnot facilitate beat finding. Hannon et al. (2004) found that listeners’ Accent Salience: A Neglected Concept ratings of meter in folk songs relied on both MA (including pitch-leap, contour) and TA (rhythmic accents, duration accents) TheJAShypothesisassumesthatbothMAsandTAsaresalient, factors,butMAsfailedtoinfluenceratingssignificantlyinmelo- whereastheTAbiashypothesisassumesthatTAsareselectively diesthatcontainedbothTAsandMAs.TheyconcludedthatTAs salient for meter perception. The former approach considers MA havea“primaryrole”inpredictingmeter(p.968).Itisnotewor- salience necessary; the latter approach treats MA salience as thy, however, that MA factors were more predictive in melodies irrelevant.Noempiricalinvestigationshavesystematicallyexam- thatcontainedtemporallyregularMApatterns. inedtherelativesalienceofMAsandTAs.Insomeinvestigations, 268 ELLISANDJONES accentlocationsaredefinedaprioriinexperimenter-createdstim- within a global serial context (details in the Methods section of uli. For example, MA locations might be defined as contour Experiment 1). All other things equal, accent salience should be inflectionsortonesofatonictriad,andTAsasthetonesfollowing systematically affected by the magnitude of these local serial apause(i.e.,amissingbeat)inanotherwiseisochronoussequence changes. (Boltz, 1991; Boltz & Jones, 1986; Jones & Pfordresher, 1997; Jones(1987,1993)alsoarguedthattheaccentuationpotentialof Pfordresher,2003).BetterperformancewithconcordantJASsthan anytonedependsuponthenumberofaccentsthatcoincideonthat discordant JASs provides only indirect support for MA salience, tone. Here we propose that accent salience monotonically in- because the degree to which MA factors contribute to the JAS is creases with the number of local serial changes on a given tone. leftunexplored.Towit,MonahanandCarterette(1985)hypothe- Forexample,ifatoneisbothlengthenedindurationandacontour sizethatTAsaliencemighthavebeengreaterthanMAsaliencein inflection, then it should have greater salience than a tone with their melodies, potentially explaining listeners’ reliance on TA only one of these two accent tokens. Because little evidence informationinjudgingsimilarity. supportssuchclaims,wesoughttoevaluatetheminExperiment2. A disparity in salience between MAs and TAs may be more Anaccentsaliencehypothesisaddressesbothoftheseissues.It pronouncedinresearchthatexamineslistenerresponsestooriginal holdsthatthesalienceofatoneasanaccentincreaseswith(a)the compositions (Boltz, 1989a, 1989b, 1989c, 1991; Hannon et al., magnitudeoflocalserialchangesassociatedwiththattoneand(b) 2004;Palmer&Krumhansl,1987a,1987b;Snyder&Krumhansl, thenumberofserialchanges(indifferentdimensions)associated 2001) or in analyses of metrical information within excerpts of withthattone.Thus,a5-STpitchleapshouldhavegreatersalience “real” music (Dixon & Cambouropoulos, 2000; Huron & Royal, thana2-STpitchleap,anda5-STincombinationwitha100-ms 1996;Temperley&Bartlette,2002;Toiviainen&Snyder,2003). tone duration (vs. a 60-ms tone duration) should have greater Although dealing with naturalistic stimuli increases ecological saliencethaneithera5-STpitchleapora100-msdurationalone. validity, it denies the experimenter control of important indepen- dent variables such as accent magnitude. Failure to control for Plan of Experiments differential MA versus TA salience sets the stage for misleading We addressed three main questions in this research. First, in conclusions.Forinstance,ifaselectedexcerpthappenstocontain Experiment1,weasked,“Insimplemelodies,whatmagnitudeof TAs that are more salient than MAs, then a listener would likely serialchange(e.g.,inapitchleap,oratonelengthening)makesan ignorethelatteranduseonlyTAstoaidmeterperception.Onthe accent salient?” In this experiment, we tested the accent salience surface,suchafindingwouldappeartofavortheTAbiashypoth- hypothesispredictionthataccentsalienceincreaseswiththelocal esis. But because the salience of constituent MAs and TAs (in magnitudeofindividualserialchanges.Wemanipulatedthemag- isolation) is typically unknown, little can be claimed about their nitude of pitch-leap accents (MAs) and duration accents (TAs) relativecontributionswhenpresenttogether.OnlywhenMAsand across different melodies that contained minimal global variation TAsareexperimentallyequatedforsaliencecantheirrelationship in serial structure (details in Experiment 1, Methods). We chose bevalidlyassessed. theseparticularaccenttokensbecausetheirmagnitudesareeasyto If accent salience depends upon serial change along a dimen- quantify(i.e.,asthenumberofsemitonesinthepitchleaporthe sion, then a resolution of conflicting evidence about the role of duration of the tone in milliseconds) and hence to “titrate.” Lis- MAs (vs. TAs) in meter perception (highlighted in contrasting tenersheard13-tonemelodieswithpatternsofpitch-leapordura- predictions of the JAS and TA bias hypotheses) may turn on tionaccentsandratedperceivedmetricalclarity. calibrating accent salience. We argue that these conflicting find- Second, in Experiment 2, we asked, “Do concordant patterns, ings reflect a research orientation to accent structure in which which contain two different coinciding accent types (MA plus differencesinthesalienceofvariousaccenttypesareoverlooked TA), induce meter more clearly than patterns with having only a or left uncalibrated. To address this, we offer an operational single accent type?” This experiment evaluated the other predic- definition of accent salience involving two properties: (a) the tion of the accent salience hypothesis: that metrical clarity in- magnitude of a change along a given acoustic dimension and (b) creases with the number of different (coinciding) accents. We thenumberofserialchangesassociatedwiththattone. tested this by having MAs and TAs occur either together or In linking accent salience to the magnitude of a local serial separatelyindifferentmelodies. changealongsomedimension,weshouldnoteanimportantqual- Third,inExperiment2,wealsoevaluatedpredictionsoftheJAS ificationthatinvolvesglobalcontext.Alllocalchangesnecessarily hypothesisversusthoseoftheTAbiashypothesisusingmelodies happenwithinalargerprevailing(global)serialcontext.Thus,the thatcontainedbothMAsandTAs.Thesehypothesesofferdiffer- salience of any local serial change is modulated by the structural entanswerstothequestion,“Domelodyandrhythminteractsuch variability within a surrounding context. For instance, greater that concordant JASs facilitate quicker and clearer metrical per- globalvariabilityinrhythmicpatternslowersthesalienceoflocal ceptsthandiscordantJASs?”TheJAShypothesisansweris“Yes”: time changes (Jones & Yee, 1997; Large & Jones, 1999, Experi- a concordant JAS should facilitate meter perception, whereas a ments1and3).Similarly,amusicaleventcontainingawidepitch discordantJASshouldhinderit.TheTAbiashypothesisansweris rangecanattenuatethesalienceofembeddedMAs,therebylead- “No”: salient MAs in JASs will not interfere with meter percep- ingtoconclusionsconsistentwiththeTAbiashypothesis,apoint tion,whichdependssolelyuponthepatternofTAs. weconsiderintheGeneralDiscussion.Becauseneitherlocalnor globalaspectsofaccentsaliencehavebeencontrolledinprevious Experiment 1: MA and TA Salience research,wemanipulatedthemagnitudeoflocalserialchangesof two accent tokens (pitch-leap accents and duration accents) by InExperiment1,wesystematicallymanipulatedthemagnitude holding at feasible minima the variability in pitch and timing of local serial changes along two different dimensions (pitch and METERPERCEPTION 269 time) to assess the accent salience hypothesis. In different melo- randomly assigned in equal numbers to two presentation orders. dies,thetwoaccenttypes(MA,TA)wereembedded,respectively, Subjects had an average of 4.6 years of formal musical training as pitch-leap and duration tokens at serial locations specified by (SD(cid:1)3.3;range(cid:1)0–12);allreportednormalhearing. eitherdupleortriplemetricalframeworks.Ourgoalwastoeval- Apparatus. Stimuli were programmed on a PC-compatible uate the accent salience hypothesis prediction that a listener’s 200 MHz Pentium computer (Intel, Santa Clara, CA) running abilitytodifferentiateduplefromtriplemeterimprovesasaccent Version 6.0 of the MIDILAB program (Todd, Boltz, & Jones, magnitude (i.e., local serial change) increases. We controlled the 1989). The computer interfaced with a Roland MPU-401 MIDI variability of both the melodic and the temporal global serial processing unit (Roland Corp., Hamamatsu, Japan), which con- contextthatsurroundedlocalserialchangesbycreatingmelodies trolled a Yamaha TX81Z FM tone generator (Yamaha Corp., based on simple, ascending pitch trajectories that unfolded with Hamamatsu, Japan) set to the “sine wave” voice. Stimuli were uniformIOIs. amplified by a Kenwood KA-5700 amplifier (Kenwood USA, ArelatedgoalwastouncoverMAandTApatternsofcompa- Long Beach, CA) and delivered individually to subjects through rable salience levels in preparation for subsequent evaluations of AKGK-270headphones(AKGAcoustics,Vienna,Austria). the JAS and temporal bias hypotheses in Experiment 2. To cali- Stimuliandconditions. Thirty-twounique,13-tonesequences brateaccentsalienceratings,webuiltuponthemagnitudematch- (“melodies”)werecreatedwitheitherpitch-leapaccents(anMA) ing paradigm of J. C. Stevens & Marks (1980; cf. Marks & ordurationaccents(aTA).Inallmelodies,IOIsbetweensucces- Gescheider,2002).Unlikethecross-modalitymatchingparadigm sivetoneswerealways500ms.Allmelodiessharedthesamebasic (J. C. Stevens & Marks, 1965; S. S. Stevens, 1959), wherein pitch contour structure: two identical, ascending 6-tone “cells” subjects adjusted the level of one dimension (e.g., brightness) to (Tones1–6,Tones7–12)followedbya13thtone(arepeatoftones matchthelevelofanother(e.g.,loudness),themagnitudematch- 1 and 7). Table 1 shows accent serial locations in duple and triple ing paradigm required subjects to rate brightness and loudness accentpatterns.AccentsintheduplepatternoccurredatTones1,3, separately, but on the same numerical scale. Stimulus values of 5,7,9,11,and13;inthetriplepattern,theyoccurredatTones1,4, lights and tones receiving the same ratings became x and y coor- 7, 10, and 13. These constraints preserved global pattern structure dinatesonthecross-modalfunction.Inthepresentstudy,melodies over the experiment; each 6-tone cell could either have three with MA patterns or TA patterns were both rated on a common groupsof2tonesortwogroupsof3tones. 6-pointscaleofmetricalclarity.MagnitudesofMAsandTAsthat Our choice of pitch-leap and duration accents was also moti- producedsimilarratingsfromsubjectswereconsideredcompara- vatedbyglobalpatternstructure.Whilecontouraccentshavebeen blysalient. usedinpreviousinvestigationsofmeter(e.g.,Jones&Pfordresher, 1997; Pfordresher, 2003), using them as the MA here would Method introducedramaticdifferencesinmelodiccontourshapebetween Subjects. Twenty-eight Ohio State University (OSU) under- dupleandtripleaccentpatterns.Usingpitchleapaccentsallowed graduatesinpsychology(whoparticipatedforcoursecredit)were us to hold the locations of contour inflections (i.e., pitch peaks) Table1 ASchematicIllustrationoftheIsochronousMelodiesUsedinExperiments1and2 Serialposition Accentpattern/magnitude 1 2 3 4 5 6 7 8 9 10 11 12 13 Temporalaccents1 Duple (cid:1) ● (cid:1) ● (cid:1) ● (cid:1) ● (cid:1) ● (cid:1) ● (cid:1) Triple (cid:1) ● ● (cid:1) ● ● (cid:1) ● ● (cid:1) ● ● (cid:1) Melodicaccents2 Duple 2ST F5 F#5 G#5 A5 B5 C6 F5 F#5 G#5 A5 B5 C6 F5 3ST D#5 E5 G5 G#5 B5 C6 D#5 E5 G5 G#5 B5 C6 D#5 4ST C#5 D5 F#5 G5 B5 C6 C#5 D5 F#5 G5 B5 C6 C#5 5ST B4 C5 F5 F#5 B5 C6 B4 C5 F5 F#5 B5 C6 B4 Triple 2ST F#5 G5 G#5 A#5 B5 C6 F#5 G5 G#5 A#5 B5 C6 F#5 3ST F5 F#5 G5 A#5 B5 C6 F5 F#5 G5 A#5 B5 C6 F5 4ST E5 F5 F#5 A#5 B5 C6 E5 F5 F#5 A#5 B5 C6 E5 5ST D#5 E5 F5 A#5 B5 C6 D#5 E5 F5 A#5 B5 C6 D#5 Note. Accentpatternsaredupleandtriple.Magnitudeismeasuredsemitones(ST). 1Temporalaccents(largecircles)hadaconstantmagnitudewithinapattern:80-ms,100-ms,120-ms,or140-mstonedurations.Unaccentedtones(small circles)hadadurationof60ms. 2Melodicaccentsarehighlightedinbold.OnlythosemelodieswithapitchpeakofC6areshowninthistable;patternswithapitchpeakofA5were constructedsimilarly. 270 ELLISANDJONES constantthroughouttheexperiment.Inasimilarvein,usingrhyth- Subjectswererandomlyassignedtooneoftwopseudorandom mic accents to create duple versus triple accent patterns would presentationorders.Withineachorder,nothreesuccessivemelo- haveintroducedvariabilityintheIOIstructureofmelodiesacross dieshadthesamepitchpeak,accenttype,oraccentmagnitude.At the experiment. Duration accents preserve isochrony, helping to theendoftheexperiment,subjectscompletedaquestionnaireon minimizeglobalpatternvariability. theirmusicalbackground(training,musicalpreferences),taskper- Duration accents assumed one of four magnitudes over the ceptions,andstrategies. course of the experiment: 80 ms, 100 ms, 120 ms, or 140 ms. Data reduction. Collapsed over the pitch peak variable, each Tones without accents had a duration of 60 ms. Interstimulus melodywasheardfourtimesoverthecourseoftheexperiment.To intervals(i.e.,thetimebetweentheoffsetofonetonetotheonset quantify a subject’s perception of each melody, we mapped the ofthenext)wereadjustedtopreservesequenceisochrony. sevenMIDILABbuttonsontotheintegersfrom–3to3,fromleft Pitch-leapaccentsalsoassumedoneoffourmagnitudes:2ST, toright.Asignedclarityscore(Cscore)wasthencalculatedasthe 3 ST, 4 ST, or 5 ST; tones without pitch-leap accents were 1 ST meanofthefourresponsestothatmelody.ThesignofaCscore aboveprecedingtones.Boththenumberandmagnitudeofapitch identifies the meter choice: negative for duple and positive for leapincreasedthedistancebetweenthelowestandhighestpitches triple.Thenumberitselfreflectstheclarityoftheidentifiedmeter, ofamelody,whichrangedfrom5STto13ST.Accordingly,the with1correspondingtoslightlyclearand3correspondingtovery startingpitchofamelodywasadjustedsothatthehighestpitchof clear. A C score near 0 indicates that a subject either (a) consis- that melody (the pitch peak) was either A5 or C6. Pitch peaks tently pressed the neutral/can’t decide button or (b) pressed a combination of both groups of two and groups of three buttons, locations were constant over all melodies (Tones 6 and 12), and suggesting overall uncertainty about the meter. Each of the 28 werenotpitch-leapaccents. subjects produced 16 C scores (2 accent types (cid:2) 2 accent pat- Tones 1, 7, and 13 did not have the same pitch-leap accent terns (cid:2) 4 accent magnitudes), resulting in a total of 448 data magnitudes as other accent locations. Tones 7 and 13 were low points. pitches, preceded by a pitch that could be up to 13 ST higher, makingthedirectionoftheleapdifferentfromotheraccents.Tone 1couldonlyformanintervalwithTone2,precludingitfrombeing Results a pitch-leap accent. However, because Tones 1, 7, and 13 were In all experiments, we report the eta-square values, calculated common to both duple and triple frameworks, the presence of a fromsumsofsquares(SS)tablesasSS /SS ,whichindicates (potentially larger) pitch accent at these locations could not be effect total theproportionofvarianceuniquelyexplainedbyaneffect(Keppel usedtodifferentiatethem. &Wickens,2004). Design. Theprimarydesignwas2(cid:2)2(cid:2)4,withaccenttype Signed clarity scores. Signed C scores for MA melodies and (MA, TA), accent pattern (duple, triple), and accent magnitude TA melodies were analyzed separately, each with a 2 (accent (four values) as within-subjects factors. All melodies were pre- pattern) (cid:2) 4 (accent magnitude) repeated-measures analysis of sented twice, in one of two presentation orders. Trial-to-trial variance (ANOVA). We used care in performing ANOVAs that constraints held that no three consecutive melodies had the same includedaccenttypeasafactor,becausesuchadesignimplies,a accenttype,pattern,magnitude,orpitchpeak. priori,certainequivalencesbetweenMAandTAmagnitudes(i.e., Procedure. Subjects listened to recorded instructions while assigning 2-ST pitch-leap accents and 80-ms duration accents as observing a task diagram. They were asked to attend to the the first level in a unified accent magnitude variable). A main grouping pattern of strong and weak tones within each melody. effectforaccenttypewouldthusbemeaningless. Subjectshadupto3.5sfollowingsequencecessationtopressone Figure 3 presents C score means as a function of accent type ofsevenhorizontallyorderedbuttonsonaMIDILABresponsebox (MA, TA), accent pattern (duple, triple), and accent magnitude toindicategroupingclarity.Fromlefttoright,buttonswerelabeled (fourlevels).Asignificantmaineffectappearedforaccentpattern asfollows:verycleargroupsoftwo,moderatelycleargroupsoftwo, inbothMAmelodies,F(1,27)(cid:1)48.05,(cid:3)2(cid:1).31,p(cid:4).001,and slight groups of two, neutral/can’t decide, slight groups of three, TA melodies, F(1, 27) (cid:1) 125.41, (cid:3)2 (cid:1) .51, p (cid:4) .001. Melodies moderately clear groups of three, and very clear groups of three. with a duple accent pattern received negative scores (indicating Verbal,ratherthannumerical,labelswereusedtominimizeinter- that subjects heard tones grouped by twos) and melodies with a subjectvariabilityinscaleuse(cf.Borg,1982;Marks&Geschei- triple accent pattern received positive scores (tones grouped by der, 2002). Subjects were told that the task was subjective and threes). Accent pattern also interacted with accent magnitude in weregivennoinstructionsregardingresponsespeedotherthanto bothANOVAs.Scorestoduplemelodiesbecamemorenegative, withholdaresponseuntileachmelodyfinished. andscorestotriplemelodiesmorepositive,asTAaccentmagni- Subjectsreceivedsixpracticetrialswithfeedback.Tohelporient tudeincreased,F(3,60)(cid:1)224.69,(cid:3)2(cid:1).26,p(cid:4).001.Scoresto listeners,weusedlargeraccentmagnitudesinthepracticetrialsthan triplemelodiesbecamemorepositiveasMAmagnitudeincreased, in the experimental trials (pitch-leap accent (cid:1) 7 ST; duration ac- but scores to duple melodies were not modulated by MA magni- cent (cid:1) 180 ms); that is, melodies were either strongly duple or tude,F(3,57)(cid:1)17.63,(cid:3)2(cid:1).10,p(cid:4).001.Ingeneral,differences stronglytriple.Responsesweremadeontheseven-buttonbox.Fol- betweenduplemeterandtriplemeterincreasedwithaccentsize. lowingaresponse,anLEDscreenontheresponseboxdisplayeda If a given accent magnitude has the potential to differentiate single digit indicating the response that “most people might have duplefromtriplemeter,thenCscoresformelodieswithdupleand given”: 2 for groups of two and 3 for groups of three. Following triplepatternsshouldbestatisticallydifferent.Weperformedfour practice,subjectsheardeachofthe32melodiestwiceoverthecourse planned comparisons from within the Accent Pattern (cid:2) Accent oftheexperiment,withoutfeedback. Magnitudeinteraction(oneforeachlevelofaccentmagnitude)in METERPERCEPTION 271 Figure3. Experiment1:Signedclarityscore(Cscore)meansasafunctionofaccenttype,accentpattern,andaccent magnitude.Errorsbarsshowstandarderrors.MA(cid:1)melodicaccent;TA(cid:1)temporalaccent;ST(cid:1)semitone. bothANOVAs.Thesecomparisonsrevealedthatthethreelargest accountformorethan3%ofthevariance;|C|scoresincreasedwith MAmagnitudes(3ST,4ST,and5ST)andthethreelargestTA accentmagnituderegardlessofwhetherthoseaccentsappearedin magnitudes(100ms,120ms,and140ms)allsignificantlydiffer- dupleortripleaccentpatterns. entiatedduplefromtriplemeter(ps(cid:4).01). One goal of Experiment 1 was to identify comparably salient Absolute value of C scores. We also analyzed the data in a accentmagnitudes.InspectionofFigure4suggeststhatthe4-and differentmanner,onthebasisofthefollowinglogic.TheCscore 5-ST pitch-leap accent patterns (left panel) and the 100-ms dura- scale treats duple and triple meter as polar opposites on a quasi- tionaccentpattern(rightpanel)elicitedsimilarratingsofmetrical continuous variable. Such a scale, however, may not accurately clarity (|C| score means (cid:5)1.25) in both duple and triple accent reflect listeners’ decision process. That is, rather than making a patterns.Weconfirmedthisbyperforminga2(accenttype)(cid:2)2 single decision (implied by the C scale), listeners may implicitly (accent pattern) (cid:2) 2 (accent magnitude) ANOVA and a Tukey’s partitionthistaskintotwosubtasks,thefirstemphasizingcatego- honestlysignificantdifference(HSD)testonthethree-wayinter- rization(“Isthegroupingpatterndupleortriple?”)andthesecond action. (It was necessary to include accent type as a factor, since emphasizing perceptual clarity (“How clear is the grouping?”). Strongmaineffectsforaccentpattern(explaining31%and51%of the variance of ratings to MA and TA patterns, respectively) 3 3 indicate that subjects were adept at the categorization portion of Accent Pattern the decision. To isolate the metrical clarity portion, we took the DDuuppllee absolutevalueofasignedclarityscore(|C|).Thisreflectsthefact Triple that melodies rated as “C (cid:1) –3” and “C (cid:1) 3” reflect the same y) 2 2 degree of metrical clarity (very strong groups) instantiated by earit orCl different accent patterns. Thus, performance on the |C| scale iso- Scal l4a-tpeosintthescacllea)r,itfyromwitihtscwahteicghorylis(tdeunpelres,tprieprlcee)i,vwedithaamhiegtheerr(socnorea | |CMetricr 1 1 ( indicatinggreaterclarity. Two 2 (accent pattern) (cid:2) 4 (accent magnitude) ANOVAs were conductedon|C|scoresforMAmelodiesandTAmelodies.Figure4 0 0 presentsthe combined data as a function of these two factors for 1 2 3 4 5 6 60 80 100 120 140 160 each accent type. A main effect for accent magnitude, with a MA Magmnaitude (ST) TA Magntiatu de (ms) stronglinear(lin)trendcomponent,emergedinboththeMA(left panel) and TA (right panel) melodies, F (1, 27) (cid:1) 36.92, (cid:3)2 (cid:1) lin Figure 4. Experiment 1: Absolute value of a signed clarity score (|C| .58, p (cid:4) .001, and Flin(1, 27) (cid:1) 103.55, (cid:3)2 (cid:1) .79, p (cid:4) .001, score) means as a function of accent type, accent pattern, and accent respectively. Although an Accent Magnitude (cid:2) Accent Pattern magnitude.Errorsbarsshowstandarderrors.MA(cid:1)melodicaccent;TA(cid:1) interaction was significant in both ANOVAs, in neither did it temporalaccent. 272 ELLISANDJONES we were interested in comparisons of all data points.) The |C| Table2 scoresforthe4-and5-STMAandthe100-msTAdidnotdiffer NineJointAccentStructuresFormedbyMelodicAccentand ineitheraccentpattern(ps(cid:6).8). TemporalAccentPatterns We also conducted an experiment in which accent type was a between-subjects factor (versus as a within-subjects factor, re- TApattern portedhere).Thegeneralpatternoffindingswasquitesimilar. MApattern Duple Neutral Triple Discussion Triple TMADTA TMANTA TMATTA Neutral N D N N N T MA TA MA TA MA TA Duple D D D N D T OurmaingoalinExperiment1wastoassessmetricalsalience MA TA MA TA MA TA as a function of accent type and magnitude. We found that mag- Note. Jointaccentstructure(JAS)patternsarelabeledinshorthandwith nitudeincrementsofserialpitchandtimechangesincreasedper- accent period (D (cid:1) duple; N (cid:1) neutral; T (cid:1) triple) subscripted by the ceived differences between duple and triple meter (C score anal- accenttypeinstantiatingit(MA(cid:1)melodicaccent;TA(cid:1)temporalaccent). ysis) and enhanced metrical clarity (|C| score analysis). These ConcordantJASsaremarkedinbold;discordantJASsareitalicized. findingssupporttheaccentsaliencehypothesis. On the whole, it could be argued that TA patterns conveyed cordantmelodieswillelicitstronger(andfaster)ratingsofmetrical moremetricalinformationthanMApatterns,asevidencedbythe clarity than discordant melodies, due to differences in temporal larger effect sizes for accent pattern (in C scores) and accent complexityofthesejointaccentpatterns.InaCscoreanalysis,this magnitude(in|C|scores)inTAmelodiesthanMAmelodies.Such shouldresultinmoreextremeratingsforconcordantmelodies(i.e., findingswould,onthesurface,appearconsistentwiththeTAbias closertoeitherendpointofthescale)andmoreneutralratingsfor hypothesis.Wecautionagainstsuchaninterpretationofthis“main discordant melodies (i.e., closer toward C (cid:1) 0). In a |C| score effect”ofaccenttype,however,becauseMAandTAmagnitudes analysis, this should result in higher |C| values for concordant were chosen a priori; thus, the two magnitudes assigned to the melodiesandlower|C|valuesfordiscordantmelodies.Similarly, samelevelinanANOVA(e.g.,140ms,5ST)cannotbedirectly response times (RTs) should be faster for concordant melodies compared.Instead,itismoreappropriatetoconcludethatmetrical than discordant melodies. The TA bias hypothesis, by contrast, clarity was higher for some TA magnitudes than for some MA doesnotpredictthatconcordantJASswillfacilitatemeterpercep- magnitudes,andviceversa. tion; instead, C scores, |C| scores, and RTs should be similar for Finally,arelatedgoalinExperiment1wastouncoverMAand anytwomelodieswiththesameTApatternregardlessoftheMA TA accents of comparable salience levels. On the basis of the pattern(e.g.,D D andT D ). MA TA MA TA analyses, we selected the 5-ST pitch-leap accent and the 100-ms Second, the accent salience hypothesis predicts differences in durationaccentascomparableintheirabilitytoevokemeter. performancebetweenconcordantJASsandsimpleJASsthathave the same TA pattern (e.g., D D vs. N D ). According to MA TA MA TA Experiment 2: MAs and TAs Combined in Melodies theaccentsaliencehypothesis,metricalclarityandresponsespeed should increase with the number of co-occurring accent tokens. Experiment 2 was designed to compare predictions of the JAS Thus, the presence of simultaneous MAs and TAs in concordant andtemporalbiashypotheses.Thesehypothesesfeaturedifferent melodies(e.g.,D D )shouldimproveperformancerelativeto MA TA rolesforMAsinmeter.Toassessthem,wecreatedmelodiesthat melodieswithTAsonly(e.g.,N D ).TheTAbiashypothesis MA TA containedbothMApatternsandTApatternsinordertodetermine does not predict this difference in performance: the MAs in con- whetherMAshadanyimpactonperformance,andifso,whether cordant melodies are irrelevant and should not facilitate that impact reflected an additive or interactive relationship with performance. TAs.WedevelopedasetofnineJASsbyfactoriallycrossingthree MA patterns with three TA patterns. That is, in addition to the Method duple (D) and triple (T) accent patterns used in Experiment 1, a “neutral”(N)accentpatternwascreated,withaccentlocationson Subjects. Twenty-five OSU undergraduates in psychology Tones1,7,and13(i.e.,theaccentlocationscommontobothduple participated for course credit. They averaged 3.6 years of formal and triple accent patterns). This factorial approach is outlined in musical training (SD (cid:1) 2.6; range (cid:1) 0–9) and reported normal Table 2 in the shorthand notation introduced earlier (metrical hearing.Theywererandomlyassignedtooneoffourpresentation periodsubscriptedbyaccenttype).InTable2,thetwoconcordant orders(ns(cid:1)6,6,6,and7). JASs(D D andT T ;cf.Figures2Aand2B)aremarked Apparatus. TheapparatuswasidenticaltothatusedinExper- MA TA MA TA inbold,andthetwodiscordantJASs(T D andD T ;cf. iment1. MA TA MA TA Figures 2C and 2D) are italicized. Table 2 also illustrates four Stimuliandconditions. Twenty-sevenmelodieswerecreated. “simple JASs”; these patterns contain either a duple or a triple All retained the same invariant IOI of 500 ms and pitch contour accentpatterninonedimensionandaneutralaccentpatterninthe used in Experiment 1 melodies. However, in Experiment 2, each other (D N , T N , N D , N T ). Finally, a “neu- melodyhadbothanembeddedMApatternandanembeddedTA MA TA MA TA MA TA MA TA tral JAS” (N N ) contained neutral accent patterns in both pattern (neutral, duple, or triple). The nine possible JASs are MA TA dimensions. shown in Table 2. MA patterns were marked by 5-ST pitch-leap The TA bias hypothesis leads to predictions about meter that accents and TA patterns by 100-ms duration accents. Because a contrast with those of both the JAS hypothesis and the accent possibleconfoundingofpeakpitchwithaccentmagnitudeisnota salience hypothesis. First, the JAS hypothesis predicts that con- factor in Experiment 2 melodies (due to a single pitch-leap mag- METERPERCEPTION 273 nitude),allmelodieswereanchoredbytheirstartingpitch(rather F(2,38)(cid:1)39.38,(cid:3)2(cid:1).17,p(cid:4).001;andtheirinteraction,F(4, thanbypitchpeakasinExperiment1).Equallyoften,thestarting 76)(cid:1)2.76,(cid:3)2(cid:4).01,p(cid:4).05.TheMAfactoraccountedforthree pitchwasC4,D#4,orF#4. timesmorevariancethantheTAfactor,suggestingthatwhenboth Design. The primary design was 3 (cid:2) 3 repeated-measures, MA and TA patterns are present in the same melody, the TA with factors MA pattern (neutral, duple, triple), and TA pattern patternisnotnecessarilythestrongermetricalmarker.Compared (neutral,duple,triple). with the main effects, the interaction explained a negligible Procedure. TheprocedurewasidenticaltothatusedinExper- amountofvariance. iment 1, with the following exceptions: (a) For each of the four In Figure 5, it is clear that concordant JASs (T T and MA TA presentationorders,thetrial-to-trialconstraintswerethatnothree D D )receivedthemostextremeCscores(1.87and–1.96, MA TA consecutivetrialscouldpossessthesamestartingpitch,thesame respectively),whereasdiscordantJASs(T D andD T ) MA pattern, or the same TA pattern; and (b) All melodies were MA TA MA TA received less extreme scores (0.19 and –0.94, respectively). heardthreetimesoverthecourseoftheexperiment. ThisevidencesupportstheJAShypothesisbutnottheTAbias Datareductionandanalysis. Datafrom5subjectswhofailed hypothesis. (on the basis of their C scores) to follow instructions were ex- Absolute value of C scores. The |C| scores isolate the clarity cludedfromfurtheranalysis.2Remainingdatawereanalyzedasin portion of a listener’s decision from the categorization portion. previous experiments. RTs were recorded from the onset of the According to predictions of the JAS hypothesis, concordant JAS final tone of each melody; all melodies were exactly the same melodiesshouldelicittheperceptofmeter(DorT)moreclearly length.BecauseRTdistributionsarepositivelyskewed,afewlong (i.e.,larger|C|values)thandiscordantJASmelodies.Inaddition, response times can dramatically alter RT means (Ratcliff, 1993). the accent salience hypothesis predicts that concordant JAS mel- To correct for this, we excluded RTs that were more than 2.5 odiesshouldelicitclearermetricalperceptsthancomparablesim- standarddeviationsawayfromeachsubject’sgrandmeanRT.On pleJASmelodies. average,7%ofindividualRTswereexcludedforeachsubject. First,wesubmitted|C|scorestothesame3(cid:2)3ANOVAused earlier.Onceagain,MApatternexplainedmorevariancethanTA Results pattern, F(2, 38) (cid:1) 16.12, (cid:3)2 (cid:1) .15, p (cid:4) .001; F(2, 38) (cid:1) 7.21, SignedCscores. WeanalyzedsignedCscoresusinga3(cid:2)3 (cid:3)2 (cid:1) .05, p (cid:4) .005, respectively. Compared with Experiment 1, ANOVA,withfactorsMAPattern(neutral,duple,triple)andTA however,theamountofexplainedvariancebythetwomaineffects Pattern(neutral,duple,triple).Figure5presentsCscoremeansas was markedly less. Instead, the MA Pattern (cid:2) TA Pattern inter- afunctionofthesetwofactors.Significanteffectswerepresentfor action, F(4, 76) (cid:1) 19.73, (cid:3)2 (cid:1).16, p (cid:4) .001, accounted for MA pattern, F(2, 38) (cid:1) 71.26, (cid:3)2 (cid:1) .51, p (cid:4) .001; TA pattern, substantially more variance in the |C| score ANOVA than the C scoreANOVA(16%vs.(cid:4)1%). Next,wedirectlytestedapredictionoftheJAShypothesisthat 3 contrastswiththatoftheTAbiashypothesisbyfocusinguponthe subsetofconcordantanddiscordantJASs.Withinthissubset,the MMAAPPattttern ps of three”uoups of three”o 21 TDNreuipupltleeral TMANTA TMATTA JchlaoAonydnpSMicocohtAahryndepPdasoinasttttehtdmeeJroAspniesSos(cid:2)rspanrltoTehatdAa.cinccTPethfsnaoattgrttserditersrnaiu,stcceiinftorutrtecrdherlaeaasnrciipttstiryoJepAndr(.eiiS.csCeste.,eo,ndnwtl,asdhiritsgeetpsreeehrenaont|sCudwle|tdhnivtecehaymlTtuheoAeerfsg)JbmeAfioaaeSss-r gro“gr T D N T hypothesis, the 2 (cid:2) 2 design revealed a main effect for neither “ MA TA MA TA e factor,Fs(cid:4)1,(cid:3)2s(cid:4).01,butinsteadapronouncedMAPattern(cid:2) C ScorCS ScoreS neutral”nneutral” 0 NMANTA (TTAhisPastatmerne 2in(cid:2)tera2ctAioNnO, FV(A1,o1n9)C(cid:1)sco6r0e.s5,8b,y(cid:3)2co(cid:1)ntr.a2s9t,,rpes(cid:4)ulte.0d0i1n. ““ N D onlyaslightinteraction,with(cid:3)2(cid:1).01.)Asshownintheleftpanel MA TA o”o”-1 ps of tws of tw DMANTA DMATTA the2JBAeScaaucsceoudnatta,cfarroemwtahsistaekxepnertoimeelinmtitneastteedsusbevjeecrtaslwmheoteirncporerrdeiccttliyonusseodf ououpu--22 the MIDILAB response box, under the following logic. Experiment 1 “grgro DMADTA establishedthatdupleandtripleaccentpatternsconsistentlyelicitedneg- “ ative and positive C scores, respectively (or, at the very least, scores to -3 triple-meter patterns that were more positive than scores to duple-meter Duple Neutral Triple patterns). We reasoned that a subjects’ TMATTA scores should be more positivethanD D scores,thusT T (cid:7)D D (cid:6)0.Becausetwo DuratiToAn APcactetenrtn Pattern MA TA MA TA MA TA adjacentresponsebuttonsare1.0Cscoreunitsapart,thefinalinclusion criteria was restated as “include if T T (cid:7) D D (cid:1) 1.0.” Five MA TA MA TA Figure 5. Experiment 2: Signed clarity score (C score) means as a subjectsfailedtomeetthiscriterion.Intheoriginalsample(n(cid:1)25),the functionofmelodicaccent(MA)patternandtemporalaccent(TA)pattern meanvalueofT T (cid:7)D D was2.73(SD(cid:1)2.03).Forthereduced MA TA MA TA fortheninejointaccentstructures.Errorsbarsshowstandarderrors.T(cid:1) sample(n(cid:1)20),themeanvalueofT T –D D was3.82(SD(cid:1) MA TA MA TA triple;N(cid:1)neutral;D(cid:1)duple. 1.04).
Description: