Perceptin & Psychphysics /9, 48 (I), 9/-/ The inability f yung children t reprduce intensity differences in musical rhythms CLAIRE GERARD and CAROLYN DRAKE Labratire de Psychlgie Experimentale, CNRS, Paris, France A musical rhythm can be described in terms f bth its tempral and its dynamic structure. Hwever, althugh 6-year-ld children are able t perceive and reprduce simple tempralstructures, even 8-year-lds rarely reprduce intensity differences. Fur experiments n the perceptin and reprductin f musical rhythms by 5- t Syear-ld children demnstrate that even thugh dynamic structure is clearly dminated by its tempral supprt, intensity differences playa rle in reinfrcing the tempral structure. The inability f children t reprduce intensity differences appears t be due neither t an inability t cntrl the intensity f their tap respnses nr t the fact that they cannt perceive such changes in intensity. Rather, the results seem best interpreted in terms f the allcatin f attentinal resurces. With simple stimulus material (Experiments 1-3), the children fcused n tempral infrmatin, and nly when the prcessing f tempral infrmatin was mastered did they have "enugh attentin left" t direct it t intensity differences (Experiment 2). With mre cmplex rchestral music (Experiment 4), attentin was primarily allcated t the dynamic structure. Psychlgists cncerned with the perceptin f music have tended t cncentrate n either meldic r tempral aspects f the auditry sequence. The last few years, hwever, have seen the appearance f studies fthe relatin between the tw, with tw main explanatins emerging. As fr the first, Jnes (1976, 1987) has prpsed a mdel f perceptual interactin, in which perceived tempral structure affects pitch structure by guiding attentin t the pitch events that cincide with stressed tempral events. Sme empirical supprt fr this mdel has been prvided by Deutsch (1982) and Jnes, Bltz, and Kidd (1982). The secnd explanatin suggests, n the cntrary, that tempral and pitch infrmatin culd be perceived independently. Palmer and Krurnhansl (1987a, 1987b) have presented sme empirical evidence which indicates that the perceived dimensins are prcessed separately, with the listeners attending t ne stimulus dimensin at the expense f the ther. Studies invlving factrial analysis f discriminatin tasks (Gabriels sn, 1973; Mnahan & Carterette, 1985) als describe tw rthgnal factrs (rhythm and pitch), with the frmer being mre fundamental. The idea f independent prcessing has als been develped by Thmas and Weaver (1975), wh prpsed tw infrmatin prcessing systems, ne respnsible fr analyzing tempral infrmatin and an- The authrs are grateful t G. Vallee, F. Leprince, and C. Bass fr cllecting part f the data. They wish t thank the annymus reviewers fr their helpful criticisms, M. C. Blte fr her help in cntrlling the stimuli and her cmments n the paper, and W. Jay Dwling, L. Demany, and S. McAdams fr their suggestins regarding earlier drafts. Requests fr reprints may be sent t Claire Gerard, Labratire de Psychlgie Experimentale, Universite Rene Descartes, EPHE, EHESS, URA CNRS 316, 28 rue Serpente, 6 Paris, France. ther respnsible fr analyzing the characteristics f the sund, such as intensity r pitch. In this paper, we examine a similar questin cncerning the perceptin f musical rhythm. In the same way as the dependence r independence f pitch and duratin has been examined, we can ask what the relatins between tempral and intensity structures are. Are they als integrated int a single unit (interactin), r is each ne prcessed independently and integrated at a later stage? Althugh the relatins between the tw have been described in empirical studies, it is imprtant t g beynd simple descriptin and explain the underlying prcesses invlved (such as perceptin, attentin, and mtr cntrl). One f the advantages f develpmental studies is that, by studying the way prcesses prgressively emerge and develp, ne can glean an understanding f hw the same prcesses wrk in the adult. In this paper, we will therefre present a series f experiments in which we examined the perceptin and reprductin f musical rhythms by yung children, investigating the psychlgical prcesses invlved in the extractin fbth tempral and dynamic infrmatin. Rhythm is smetimes cnsidered simply t cnsist f variatins in the time dmain, with the duratin f ntes and the interval between the ntes influencing the resulting perceptin f the rhythms. Hwever, variatins in many ther dmains are als knwn t influence this perceptin, s that sme elements in the rhythm becme mre perceptually salient r accented than thers. Since Cper and Meyer (19), many authrs (e.g., Thmassen, 1982) have insisted n the fact that dynamic structure can be created by variatins in intensity, duratin, pitch, and timbre. Handel (1974) classified these varius surces f accentuatin int tw types f rganizatinal principles: 91 Cpyright 19 Psychnmic Sciety, Inc.
92 GERARD AND DRAKE (1) frequency and intensity patterns, which are patterns in time, and (2) duratin and interval patterns, which are patterns ftime. He suggested that these tw rganizatinal principles may be gverned by distinct rules that determine their perceptin. Mnahan and Carterette (1985) classified five majr surces f cues f accent: tempral patterning, pitch pattern shape, tnal system, dynamic (ludness) patterning, and timbral patterning. Hwever, much ambiguity exists in the literature cncerning wrds such as "accents," "accenting," and "accentuatin." Authrs have felt the need t distinguish between the physical dimensin f sunds (playing luder, lnger, r n a particular instrument) and their psychlgical dimensin (the perceptin f an event as being mre salient than thers). T cnfuse the.issue, it is knwn that althugh the psychlgical dimeti~in usually cincides with the physical dimensin, accents can be perceived in the absence f any physical variatins (subjective accents; see Fraisse, 1956; Pvel & Okkerman, 1981). Mnahan and Carterette (1985) defined "accenting" as a psychlgical event (p. 4), but Mnahan, Kendall, and Carterette (1987) referred t physical cues as "accenting" and t perceptual phenmena (whether based n physical variatins r nt) as "accent." Thmassen (1982) preferred the term "accentuatin" fr the first and "accent" fr the secnd; Fraisse (1956) wrte f an ppsitin between bjective and subjective accents, the secnd fwhich was nt based n physical variatins; Pvel and Okkerman (1981) have called interval-prduced "accent" the subjective increment in ludness prduced by tempral intervals between tw tnes. T avid any cnfusin in terminlgy, fr this series f experiments, we will use the wrd accent fr variatins in the physical dimensins f sund. In this paper, we examine ne particular surce f accentuatin: that f accents created by variatins in intensity. Unless therwise specified, all references t accents will hencefrth be directed tward dynamic accents resulting frm physical intensity differences. First, let us examine physical accents. Thmassen (1982) cnsidered that a 4-dB increase in sund level is adequate t make the luder sund be perceived and interpreted as an accent. When Gabrielssn (1974) asked subjects t prduce accents n a pian r drum, differences f 2-4 db and 5- db were bserved, respectively. Schmidt (cited in Fraisse, 1956), wh asked subjects t tap a regular beat, bserved spntaneus gruping with a regular cyclic variatin in intensity. The pertinence fdynamics in the perceptin f music has been demnstrated by Nakamura (1987), wh asked musicians t play a piece f music with a crescend, which was later crrectly interpreted by nnmusicians. As regards subjective accents, psychphysical studies have demnstrated that perceived ludness f tnes varies as a functin fnumerus parameters (such as frequency, duratin f tne, r time interval between tw tnes.) When tw tnes are presented in shrt successin, ludness masking ccurs. Frward masking appears nly when the interval between them is less than 1-0 msec, and backward masking appears when the interval is less than msec. Thus, due t the tempral limits given, masking nly appears with fast musical tempi. Pvel and Okkerman (1981) have als shwn that subjective accents are perceived in equitne sequences (frequency, spectral cmpsitin, intensity, and duratin f tnes being cnstant) in which nly the time interval between the tnes has been mdified. When repeated grups f tw tnes were presented, subjective accents were perceived n the first r secnd tne, depending n the duratin f the interval between the tw. The tendency t hear accents n the last tne f the grup nly ccurred if the betweengrup interval was cnsiderably lnger than the withingrup interval and if the latter did nt exceed a duratin f abut 2 msec. The strength f the subjective accent seemed high: One subject wh was asked t adjust the intensity f the first tne until the tw tnes sunded equally accented needed an increase fabut 4 db (Pvel & Okkerman, 1981, Experiment 4). In ur experiments, using slwer tempi, we culd assume that the phenmena f masking and f subjective accentuatin wuld nt have much effect n ur stimuli-thugh f curse, as will be seen later, they culd nt be ignred cmpletely. As we have seen, physical intensity differences cnstitute an aspect f rhythm that is perceivable and reprducible by musicians and nnmusicians alike. We were therefre surprised with the results f experiments n the perceptin and reprductin fmusical rhythms in yung children, because, althugh tempral structure was ften reprduced well, intensity differences were reprduced very rarely (Gerard, 1988a; Gerard & Auxiette, 1988; Gerard & Remnd, 1985). The fact that children (up t the age f 8 years) d nt spntaneusly reprduce systematic variatins in intensity culd be explained in tw ways: One hypthesis (Hypthesis 1) wuld be that their mtr cntrl is inadequate t gauge the strength f beat necessary fr the cntrl f the ludness f the taps prduced; the ther hypthesis (Hypthesis 2) wuld be that tempral and dynamic infrmatin are prcessed independently, s that attentin must be divided between the tw. Yung children wuld thus cncentrate n ne at the expense f the ther. Ifthis is the case, it becmes necessary t ask whether tempral r dynamic infrmatin receives pririty under varius cnditins, and whether r nt the situatin changes during develpment. We devted this study t the prblem f accentuatin in the perceptin and reprductin fmusical rhythms by yung children in rder t examine these tw hyptheses. In the first three experiments, we were cncerned with intensity accents; in the furth experiment, we cnsidered a wider range f phenmena. EXPERIMENT 1 One f ur aims in Experiment I was t examine the relative imprtance attributed by children t tempral and intensity differences as criteria fr judging rhythms (Hypthesis 2). By cnsidering the respnses f 5-, 6-,
CHILDREN AND INTENSITY DIFFERENCES IN RHYTHM 93 7-, and 8-year-ld children, ne might see hw these abilities evlve and whether r nt the same imprtance wuld always be attributed t the same infrmatin. Methd Subjects. A ttal f 1 pupils frm a large schl cmplex in Anthny (a suburb f Paris) tk part in the experiment. The yungest children came frm the last year f nursery schl. They were frm 5 years t 5 years and 9 mnths ld. The ther subjects frm the primary schl were divided int three age grups: first grade (6 years t 6 years, 9 mnths), secnd grade (7 years t 7 years, 9 mnths), andthird grade (8 years t 8 years, 9 mnths). Nne f the children had a hearing impairment, and "bad pupils" (as judged by the teacher) were excluded frm the sample. Stimuli. Twelve pairs f rhythmic patterns were cnstructed: 6 pairs with identical patterns (i) and 6 pairs with different patterns (d). Figure 1 presents the patterns, indicating the temp and, when necessary, the accented sunds. Within a pair, the first rhythm was repeated three times, and then, 2 sec later, the secnd rhythm was repeated three times. These rhythmic patterns were played by a musician wh was asked t tap the accents clearly and mre ludly than the ther ntes when and nly when it was indicated in the mdel (see Figure 1). A wd blck was used t btain a clear staccat sund. The pairs f patterns were recrded n a Sny TC D5M tape recrder. After recrding, the pairs were mixed and rerecrded t btain the three presentatin rders, which were designed t prevent the ccurrence f mre than three is r ds in successin and t assure that each pattern appeared at the beginning, middle, and end f the sessin. At a later date, the actual intervals, intensities, and duratins f the sunds used in the experiment were measured by a digital strage scillscpe (Guld as 42(0). The mean duratin f the wd-blck sunds was 237 msec (range, 0-2 msec), and the internset intervals crrespnded well with IDE N TIC A L PAT T ERN S IItrrp ("'5'C.) 1 j j j j /I j J j j J 0 2 ~ ~ ~ ~ /I ~ ~ ~ ~ J 0 3 JID n ~ II JID Jl J J 0 4 ~ 1'~. ~ II J ) J. J J 0 5!JJ[J~/I[J~[J J J 0 6 ~.J]JJ /I.r]Il~J J 0 - - I F FER E N T PATTERNS 1 J J r; J II J n J J J 0 2 J J.n J /I JJiJJ J 0 - - 3 JJ.rJJ /I J n ~ ~ J 0 - - 4 J J J J II ~1.n n r; J 0 5 j J J J /I J ~ ~ ~ J 0 6 J J J J /I J ~l J n.i I 0 Figure 1. Rhythmic patterns presented in Experiment 1. Table 1 Percentages f Crrect Respnses in the Discriminatin Task in Experiment 1 Age Grups Pair 5 years 6 years 7 years 8 years Identical Patterns i1 86.67 83.33 83.33 0 i2 86.67 86.67 76.67 96.67 i3 76.67 93.33 i4 43.33 46.67 i5 86.67 i6 76.67 86.67 Different Patterns dl 53.33 86.67 63.33 d2 63.33 66.67 d3 96.67 d4 23.33 36.67 36.67 23.33 d5 73.33 83.33 0 d6 93.33 Nte-i = identical pair, d = different pair. the mdels presented in Figure 1. The variatins due t the natural executin perfrmed by the musician were as fllws: (1) Internset intervals were ± 7 msec fr a quarter nte, ±4 msec fr an eighth nte, and ±3 msec fr a sixteenth nte; (2) in an unaccented sequence, the greatest difference between the intensity f the sftest and that f the ludest sund was 5.7 db SPL, whereas in an accented sequence, the intensity differences between unaccented and accented sunds were at least 14 db SPL (14-19.8 db). Thus, the manually created sunds had a natural quality, but they were als within the range f values required by the experimenters. Prcedure. The children were examined individually in a quiet rm in the schl. They sat abut 1 m frm the tape recrder and heard the stimuli ver a ludspeaker. The fllwing instructins were given t each child: "Yu are ging t hear sme music played by Paul and Cecile n their drum. First yu will hear Paul's music and then Cecile's music." The experimenter tapped 1.111.1. and then I... I... n the table in frnt f the child. "Yu see that the 2 pieces f music are nt the same. Paul and Cecile dn't play the same thing. Listen carefully nw." The experimenter tapped I... 111I and then I... 1111 n the table. "Nw Cecile and Paul play the same thing, dn't they?" (This was dne t make sure that the child understd the cncepts f "same" and "nt the same"). "Nw yu will hear ther pieces f music played by Paul andcecile. I want yu t tell me whether Paul's music is the same r nt the same as Cecile's music." The tape recrder was stpped after each pair, and the experimenter nted whether the child cnsidered the tw rhythms t be the same r nt. Results Table 1 presents the percentage f children wh answered the cmparisns crrectly. Identical patterns were frequently identified crrectly with the exceptin f Pair i4, which cntained the nly rhythm with a rati f 1:3 between shrt and lng intervals and seemed difficult t encde and recgnize as "the same." This phenmenn is well knwn with respect t bth adults (pvel, 1981; Pve1 & Essens, 1985) and children (Drake & Gerard, 1989). Fr the different pairs, the results were clsely related t the type f difference between the stimuli. When the temp f ischrnus sequences was mdified (Pair d4), the children did nt cnsider the tw rhythms t bediffer-
94 GERARD AND DRAKE ent. Inversely, when an ischrnus sequence was cmpared t a cmpsite rhythm (Pairs d5 and d6), perfrmances were better and imprved with age. In this paper, we will fcus mainly n the relative imprtance f tempral and intensive rganizatins, s Pairs dl, d2, and d3 will be cnsidered in detail: In Pair d l, the tw rhythms have a different tempral rganizatin and are nt accented; in Pair d2, the tw rhythms have the same tempral rganizatin, but ne is accented and nt the ther; in Pair d3, the tempral and intensive rganizatins are different. A chi-squared test was used t evaluate significant variatins frm chance level. Percentages between 36% and 64% d nt differ frm % at the 5% significance level (p <.05). First, Pair dl, which varied nly in tempral structure, was discriminated abve chance level nly by the 6- andsvyear-lds, s tempral structure is a dimensin alng which children were able, in sme cases, t discriminate the tw rhythms. Sevenyear-ld children seemed t have had a failure in perfrmance. These lw scres were nt related t ther behavir bserved during the experiment. This apparent failure must be recnsidered in the light f the fllwing bservatin: The 6-year-ld children were very attentive and interested in the experiment. The teacher in the classrm was very prud f this' 'excellent grup, this particular year. " S the accident f the curve shws a cmbined effect f gd results fr the yunger grup and pr perfrmances fr the lder grup. The nly cnclusin we can draw is that fr 6- and 7-year-lds, success in the task is nt well established and is thus unstable. Secnd, Pair d2 can nly be discriminated by the presence r absence f accents. Up t the age f 8, the children cnsidered the tw rhythms t be the same, fr the scres were belw chance level. Intensity differences did nt appear t be taken int accunt. Hwever, the picture was nt as simple as this. Pair d3 had the same tempral difference as dl, but ne f the rhythms als had accents. Fr this pair, discriminatin was abve chance level fr all the children-ntably better than fr the discriminatin f the same tempral structures withut accents (Pair d1). The presence f accents enhances the discriminatin f tempral structure. Discussin This experiment demnstrates that whereas 5-year-lds apparently d nt perceive a difference between tw simple rhythms n the basis f either tempral r dynamic infrmatin, 8-year-lds are able t use tempral infrmatin but neglect dynamic infrmatin. Hwever, accents enhance discriminatin based n tempral structure fr 5- and 7-year-lds. It culd be cncluded either that accents are nt systematically and cnsciusly perceived, r that accents are perceived by the children but are cnsidered unimprtant with regard t what makes rhythms similar r different. At least up t the age f 8, tempral infrmatin is dminant while dynamic infrmatin is neglected. In rder t verify this cnclusin (and t be sure that accents are perceived), a secnd experiment was perfrmed. EXPERIMENT 2 Hypthesis 2 states that children can nly cncentrate n ne dimensin f rhythm at a time. Iftempral structure is reduced t a minimum, enugh attentin shuld be released t prcess infrmatin abut intensity. It shuld then be pssible fr 6-, 7-, and 8-year-ld children t discriminate between tw sequences n the basis f their intensity differences. What is mre, if the prblem f sharing attentin is eliminated, the rle f mtr cntrl can be examined (Hypthesis I). Children were asked t discriminate between tw sequences n the basis f intensity differences and t reprduce them. Methd Subjects. Pupils in a small private primary schl in England tk part in Experiment 2. Three age grups (6, 7, and 8 years) f children were examined. Nne f the children suffered frm any hearing impairment. Stimuli. The sunds were synthezised n a BBC cmputer and recrded n tape. All the ntes were a C3 n the pian, and they all lasted msec. The tempral structure fthe rhythmic patterns used was always a regular ischrnus beat, but intensity varied; metric accents were intrduced. The five basic stimuli are presented in Table 2. The primary accents were played % luder than the nnaccented ntes and the secndary accents 25% luder (n the linear scale f the cmputer). All the differences in intensity were easily discriminated by the experimenters. Readings taken at a later date frm the scillscpe gave a difference f 14 db SPL between primary accents and unaccented ntes and 8 db SPL between secndary accents and unaccented ntes. Each stimulus cntained 12 ntes. In SI, all ntes were f the same intensity. In S2, I in every 2 is a primary accent (time signature = 2/4). In S3, I in every 3 is a primary accent (time signature = 3/4). In S4, I in every 2 ntes is accented, alternating between primary and secndary accents (time signature = 4/4), and in S5, I in every 3 ntes is accented, als alternating between primary and secndary accents (time signature = 6/8). Fifteen pairs f stimuli were derived by taking the five stimuli tw at a time, separated by a pause f 2 sec, thus making 5 identical (i) and different (d) pairs as shwn in Table 2. The pairs were recrded n a Sny TC-D5M tape recrder. Prcedure. The children were examined individually in a small rm in the schl. They heard a pair f stimuli ver a ludspeaker and were asked t say whether r nt the tw were the same. Next they heard the first stimulus f the pair again and were asked t Table 2 Rhythmic Patterns Presented as Stimulus Pairs fr Discriminatin and Reprductin in Experiment 2 Basic Stimuli Stimulus Stimulus Rhythmic Pattern Sl S2 S3 S4 S5 SI UUUUUUUUUUUU il dl d2 d3 d4 52 PUPUPUPUPUPU (2/4) i2 d5 d6 d7 53 PUUPUUPUUPUU (3/4) i3 d8 d9 S4 PUSUPUSUPUSU (4/4) i4 dlo S5 PUU5UUPUUSUU (6/8) i5 Nte-U = unaccented; P = primary accent; S withut a fllwing numeral = secndary accent; i = identical pairs; d = different pairs. Time signatures are indicated between parentheses fr S2 t S5.
CHILDREN AND INTENSITY DIFFERENCES IN RHYTHM 95 Table 3 Percentages f Crrect Respnses in the Discriminatin Task fr Identical Patterns in Experiment 2 Identical Patterns Age/Temp il i2 i3 i4 i5 M altl alt2 71 alt3 a2tl 0 0 0 a2t2 84 a2t3 0 a3tl 0 0 0 0 a3t2 89 a3t3 0 Mean 81 82 81 85 79 81 Nte-i = identical pair. al = 6 years; a2 = 7 years; a3 = 8 years. Fast temp, tl = 0 rnsec; mediumtemp, t2 = 0 msec; slw temp, t3 = 0 msec. reprduce it n a drum. Finally, they heard the secnd stimulus f the pair and reprduced that as well. Three tempi were used, crrespnding t three sessins n 3 cnsecutive days in a cunterbalanced rder, ne temp at each sessin: 0, 0, and 0 msec nset t nset. Within a sessin, the 15 pairs f stimuli were presented in a cunterbalanced rder. Fr each subject, each stimulus was heard and reprduced six times in all (e.g., 51 ccurred twice in ii and nce in each f d I, d2, d3, and d4), s that six trials were btained fr each child fr each stimulus. The whle sessin was recrded n a secnd tape recrder, and the reprductins were later analyzed by an analgic extractr f tempral and intensive characteristics (AETIC). Scring. The percentage f identical r different pairs crrectly identified in the discriminatin task was analyzed. Accents were detected in the reprductin task in the fllwing way: Fr a given subject, each tap was cded as a percentage f the ludest tap, and intensity grups were created. Three categries were defined fr each reprductin, depending n the range f intensities used. Taps in the ludest categry were cnsidered t be accents. Results Discriminatin. The percentage f crrect respnses fr identical pairs are presented in Table 3. The verall success rate was high (81%), and perfrmance imprved with age [6 years == 71 %, 7 years == 84 %, 8 years == 89%; F(2,27) == 3.849, p <.05]. The fast and slw tempi were easier than the intermediary ne [fast == 87%, medium == 73%, slw == 85%; F(2,54) == 4.2, P <.05]. N difference was seen between the pairs, and nne f the interactins were significant. Hwever, these results prvided n cncrete cnclusins, since the children might have been making crrect identificatins n the basis f the ttal number f ntes (12 ntes) r the ttal duratin f the pattern, rather than the variatins in intensity. Table 4 presents the results fr different pairs and shws that the verall success rate n the lo different pairs was high (77%). It imprved with age [6 years == 69%, 7 years == 78%, 8 years == 83%; F(2,27) == 3.34, p <.05], and the fast temp was easier than the ther tw [fast == 82%, medium == 75%, slw == 75%; F(2,54) == 3.864, p <.05]. The cmparisn invlved in each pair played an imprtant rle. Cmparisns f Stimulus 1 (withut intensity differences) with the ther stimuli were relatively easy (dl == 86%, d2 == 88%, d3 == %, and d4 == 92%) and were f equal difficulty [F(3,8l) == 0.757, n.s.]. Thus, children are indeed able t discriminate metric rganizatin frm a simple ischrnus beat. The easiest cmparisn was made between S2 and S3 (d5 == 95 %), where binary meter was cmpared with ternary meter. The ther cmparisns were mre difficult, due t the cnfusins between primary and secndary accents. The results fr d6 were very lw: S2 and S4, which differed nly by the presence f secndary accents, were hardly ever discriminated. In the same way, S3 was cnfused with 85 (d9 == %), whereas the discriminatin between 82 and S5 was easy (d7 == 93%), shwing that, in these cnditins, the children assimilated the 6/8 time signature with a ternary rather than a binary rganizatin, with the secndary accents being assimilated with primary accents. The same assimilatin culd explain the high scres fr d8 (85%) and dlo (83%). Table 4 Percentages f Crrect Respnses in the Discriminatin Task fr the Different Patterns in Experiment 2 Different Patterns di d2 d3 d4 d5 d6 d7 d8 d9 dio Age/Temp 5152 5153 SIS4 SIS5 S253 S2S4 S2S5 53S4 5355 S4S5 altl 0 0 alt2 alt3 a2tl 0 0 0 0 0 0 a2t2 a2t3 0 0 a3tl 0 0 0 0 0 0 a3t2 0 0 0 0 0 0 a3t3 0 0 0 0 0 0 Mean 86 88 92 95 39 93 85 83 Nte-d = different pair. 5 fllwed by a numeral = stimulus. al = 6 years; a2 = 7 years; a3 = 8 years. Fast temp, tl = 0 msec; medium temp, t2 = 0 msec; slw temp, t3 = 0 msec.
96 GERARD AND DRAKE Reprductin. A reprductin was cnsidered crrect if it cntained an intensity accent every 2, 3,4, r 6 ntes in accrdance with the primary accent f the stimulus. Table 5 presents the results. The percentage f crrect reprductins increased with age [6 years = 31%, 7 years = 43%, 8 years = %; F(2,27) = 5.442, p <.05]. The lwest perfrmances were bserved in the fast temp [fast = 35 %, medium = 44 %, slw = 45 %; F(2,54) = 11.179, p <.(01). The scres als depended n the stimulus [F(4,8) = 36.16, p <.001]. Sl (withut accents) and S2 and S3 (nly primary accents) were reprduced crrectly mre ften than were S4 and S5 (which als cntained secndary accents). The perfrmances did nt imprve ver the six trials [Trial 1 = 45%, Trial 2 = 42%, Trial 3 =:38%, Trial 4 = 42%, Trial 5 = 42%, Trial 6 = 42%~F(5,135) = 0.852, n.s.]. The interactin between age and stimuli [F(8,8) = 2.133, p <.05] is shwn in Figure 2. N imprvement with age was seen fr SI [F(l,27) = 0.962, n.s.], but an imprvement was seen fr all the ther stimuli. Discussin Under these simplified tempral cnditins f ischrnus sequences, frm the age f 6, children were able t discriminate between patterns n the basis f intensity differences. We can cnclude that this type f accent is perceived and integrated by yung children. When attentin des nt have t be fcused n tempral structure, children are free t cncentrate n dynamic aspects f structure. What is mre, under these ptimal cnditins, children frm the age f 7 are able t reprduce patterns f accents-their mtr cntrl is sufficient t dse the strength f each tap at will fr basic patterns f 2/4 and 3/4. Thus, Hypthesis 1, which pstulated inadequate mtr cntrl, is invalid, at least frm a certain stage f develpment, and fr simple intensive cnfiguratins. EXPERIMENT 3 Anther way t test the attentin-sharing hypthesis is t examine the effect f training n varius abilities (perceptin, attentin, and mtr cntrl). In a reprductin task, it shuld be pssible t draw children's attentin t a particular aspect f the rhythm-in this case, the presence f accents-and t see whether r nt tempral and/ r dynamic perfrmance imprves. It is knwn that 6 year-lds are able t reprduce tempral structures but nt dynamic nes, and children f this age were therefre chsen s that we might study whether r nt training can enhance perfrmance. Methd Subjects. The 23 children wh tk part in the experiment came frm the last year f a' Parisian nursery schl. Their mean age was 5 years, mnths (5 years t 6 years, 6 mnths). Nne f the children suffered frm any hearing impairment. "Bad pupils" (as judged by the teacher) were excluded frm the sample. Stimuli. Fur rhythmic patterns were cnstructed. They are presented with their time signatures and their tempi in Figure 3. Table 5 Percentages f Reprductins Cntaining Crrect Intensity Accents in Experiment 2 Age Grup/ Temp altl alt2 alt3 a2tl a2t2 a2t3 a3tl a3t2 a3t3 Mean altl alt2 ait3 a2tl a2t2 a2t3 a3tl a3t2 a3t3 Mean altl alt2 ait3 a2tl a2t2 a2t3 a3tl a3t2 a3t3 Mean altl alt2 alt3 a2tl a2t2 a2t3 a3tl a3t2 a3t3 Mean alii alt2 alt3 a2tl a2t2 a2t3 a3tl a3t2 a3t3 Mean 0 IO IO IO IO 2 3 Stimulus SI Stimulus S2 Stimulus S3 Stimulus S4 IO Stimulus S5 Trials 4 61 51 0 0 22 0 0 0 IO 0 14 5 6 0 Nte-al = 6 years; a2 = 7 years; a3 = 8 years, Fast temp, tl = 0 msec; medium temp, t2 = 0 msec; slw temp, t3 = 0 msec,
CHILDREN AND INTENSITY DIFFERENCES IN RHYTHM 97 % 6 7 Age... S3.. ' Figure 2. Evlutin with age f the percentage f crrect reprductins f intensity differences in Experiment 2. The patterns were played n a wd blck by a musician wh was asked t tap all the metric accents clearly and mre ludly than the ther ntes. This prcedure f hearing and reprducing wd-blck sunds makes stimuli and respnses as similar as pssible and apprximates "real" musical situatins in terms f type f sunds as well as type f mvement. The patterns were recrded n a Sny TC-D5M tape recrder. The actual intervals, intensities, and duratins f sunds prduced by the musician were measured by the scillscpe as in Experiment I. Once again, the tempral intervals crrespnded well with thse required (mean variatins were ± msec fr a quarter nte, ±5 msec fr an eighth nte, and ±3 msec fr a sixteenth nte). The smallest difference between accented and unaccented sunds in a given sequence was 12 db SPL (12-15.6 db). The mean duratin f the sunds was 228 msec. Prcedure. The children were examined individually in a quiet rm in the schl. They heard a first pattern and were asked t reprduce it n a wd blck. Their attentin was then drawn t the dynamic structure fthe rhythms by saying, "Listen carefully, sme beats in the 'music' are luder than thers. If yu hear a lud beat, hit luder, and if yu hear a sft beat, hit sfter t." Fllwing these instructins, the children were asked t hit the table three times "ludly" and three times "sftly," and then t alternate lud and sft beats fr a cycle fa dzen beats. Then a rhythm was presented three times, the child being asked t reprduce it after each presentatin. The presentatinrder was cunterbalanced. The whle sessin was recrded n a tape recrder. Scring. The tempral structure was scred by tw independent judges as fllws: 0 = a simple pulse r tempral structure different frm the mdel; I = part fthe tempral structure reprduced; 2 = crrect tempral structure. Intensity reprductins were studied by the AETIC and decded by cmputer, as described fr Experiment 2. Results Intensity reprductins. First, recall that in Experiment 2 n imprvement in the reprductin f accents was bserved ver trials. Similar results have been fund by,/ 8 Gerard and Rernnd (1985) with mre cmplex tempral structures. Once again, in Experiment 3, accents were either absent frm the reprductinsr anarchic (i.e., with n apparent crrespndence t the intensity differences in the mdels). The analyses issued frm AETIC decding were incnsistent. This was true fr the trials bth befre and after accents were explained. The warnings between Trials I and 2 resulted in greater general ludness f all the reprduced taps, but nt in precise metric accents lcalizatin. Out f the 23 children examined, nly 11 reprduced audible accents, and nly 6 (ut f the II) did s with an acceptable tempral structure; the ther 5 children reprduced ischrnic pulses r repeated patterns f three taps, whatever the stimulus. As a cnsequence, the gd scres fr tempral reprductins crrespnded mainly t unaccented reprductins. Analyses perfrmed n the 6 subjects wh had accented sme taps als shwed that the reprduced accents did nt cnfrm either t the metric accents f the stimulus r t any ther systematic chice. The ttal number f accents increased n the secnd trial (13 taps accented fr the 6 subjects x 4 patterns n the first trial, 28 taps accented fr the 6 subjects X 4 patterns n the secnd, 26 n the third and 22 n the furth), but the accents were equally distributed n each tap f the patterns. Therefre, because the number f bservatins was t lw and even the slightest trend tward systematic behavir was absent, the study had t be abandned. An explanatin is required. As described in the intrductin, in Pvel and Okkerman's (1981) research an "interval-prduced accent" ccurred with tempral patterns cntaining repeated grups f tw tnes (e.g., 11.11.). These patterns are hard t perfrm when the metric accent crrespnds t the shrt first tne (as bserved by the authrs, using tw examples, p. 571). In ur experiments, we never emplyed such tw-tne patterns, but tw patterns in the present experiment (amng the 22 different patterns presented in Experiments 1, 2, and 3) began with a series fshrt tnes (I. I.1... r 11111...). If the "interval-prduced accent" culd be generalized t grups f mre than tw tnes, the intrductin f metric accents culd have made ur sequences ambivalent, with the metric accent n the first tne f the grup and the interval-prduced accent n the last lng tne. Hwever, Pvel and Okkerman demnstrated that interval- patterns I4!mp IIJJ PJJJ(i) J 0 2 fjd1 DDDlal J 0 3 jjjj 1m lid J(il J 0.1 nj 1 )JW J 0 Figure 3. Rhythmic patterns presented in Experiment 3. Fr eacb pattern, the time signature is indicated in brackets and the internset interval between tw quarter ntes is given in millisecnds.
98 GERARD AND DRAKE / / / ~_I./ ~ /,.2 patterns <::r--;=~:... -/ 2 3 TRIALS Figure 4. Imprvement in the tempral structure f the reprductins ver trials in Experiment 3. The child's attentin is drawn t the accents between Trials 1 and 2. prduced accents disappear if the interval fllwing the shrt tne is lnger than 2 msec. With ur relatively slw tempi, we were prbably ut f the range f the phenmenn. Nevertheless, we are currently reprducing Pvel and Okkerman's experiment with lnger grups f tnes and with children t verify this pint. A final related pint must be mentined. Pvel and Essens (1985) asserted that the initial and final tnes f a cluster cnsisting f three r mre tnes were perceptually marked. In this case, metric accents wuld reinfrce this trend and nt cnflict with it, but the questin still remains: Why dn't yung children reprduce accents? Tempral structures. Figure 4 presents the imprvements f the scres ver trials. Fr tempral structure, the figur~ shws that fr three f the fur rhythms used, a greater unprvement was bserved between the first and secnd reprductins (when the child's attentin was drawn t the presence f accents) [F(3,198) = 17.542, p <.05] than between the ther trials, where very little imprvement was bserved [F(1,198) = 0.071, n.s.]. Once again, this cnfirms the ineffectiveness f simple repetitin withut changing the child's cgnitive strategy. Discussin Experiment 3 shws that it is pssible t direct a child's attentin tward the dynamic structure. Althugh accents were still nt directly reprduced, the children must have taken them int accunt, fr when attentin was drawn t them, the reprductin f the tempral structure imprved. This result cntradicts Hypthesis 2, since variatins in ne dimensin(intensity) influence the perceptin f variatins in the ther dimensin (time), demnstrating a certain degree f interactin, as in Experiment 1. We will cme back t this very interesting bservatin in the general discussin. EXPERIMENT 4 In the first three experiments, dynamic structure was limited t variatins in intensity, but as stated in the intrductin, ther events can create accents in "real" music. Lerdahl and Jackendff (1983a, 1983b) distinguish three types f accents: phenmenal accents (events n the musical surface that give emphasis r stress t a mment in the musical flw, such as sudden changes in dynamics r timbre, leaps t relatively high r lw ntes, r harmnic changes), metric accents (the peridic return f any beat that is relatively strng in its metrical cntext), and structural accents (the cnjunctin f meldic, harmnic, r metric events). In Experiment 4, we examined children's ability t synchrnize with the metric structure f Tchaikvsky's Sugar Plum Fairy. Once again, Hypthesis 2, regarding separate cgnitive channels, was tested, but this time tw pssibilities were cnsidered. Either phenmenal and structural accents enhance the metric structure f the music, making it mre easily perceivable and thus helping synchrnizatin, r they distract the children's attentin frm the tempral structure they are trying t fllw, thus disrupting synchrnizatin. In the latter case, attentin has t be shared between different surces f infrmatin. This time, "real" music will be examined. In Experiment 4, 5.5-year-ld children were selected, fr tw reasns. The first, as previusly seen, was tied t the critical perid f develpment (between ages 5 and 6), and the secnd was related t the fact that, at this age, children are knwn t be able t synchrnize easily with a regular beat. Dwling and Harwd (1986) and Mg (1976) cnsider that synchrnizatin is achieved at the age f abut 3 r 4 years. We wanted t see whether this behavir wuld bebserved in situatins mre cmplex than thse bserved by these authrs, wh had children accmpany their wn singing f familiar sngs. Methd Subjects. Thirty-three children with a mean age f 5.5 years (range, 5-6 years) were selected frm the upper classes f a nursery schl in Ruen. They were cnsidered by the teacher t have nrmal hearing and t be "gd pupils." Prcedure. During the week preceding the experiment, an afternn was spent getting the children used t the videtape recrder (Sny Camescpe VT 8 AF). The children were filmed in the schl library during grup games with the teacher. The camescpe was left in the rm fr a week. On a separate day, the children listened t the Sugar Plum Fairy Dance (played by the Lndn Symphny Orchestra under the directin f Antal Drati; a Philips stere recrding) twice in successin. The fllwing week, they were examined individually in the schl library and asked t accmpany the Sugar Plum Fairy music by clapping their hands. Each child was sitting n a chair in frnt f the camera, the experimenter asked if the child was ready; and when the child was, a TC-D5M tape recrder played the music and the camescpe recrded the sessin. Scring. First, the sund events f the music were cded with the cmputer by marking the beginning f the piece, the metric structure, the theme variatins, the beginnings and ends f the meldic
CHILDREN AND INTENSITY DIFFERENCES IN RHYTHM 99 phrases, and the expressive variatins in intensity tied t the interpretatin. All f these marked events were digitally cded 0 r 1 and lcated alng a tempral chain crrespnding t the unflding f the music. Frm this infrmatin, structural accents were derived, which were defined, in the cntext f this experiment, in terms f the simultaneus ccurrence f at least tw events. Next t this were pltted the perids f synchrnizatin and desynchrnizatin bserved by tw independent judges fr each child. The musical piece cntained: A metric intrductin (7 sec) given by the strings; Theme 1 (Tl), the appearance f tw meldic phrases ver the metric by the celesta (15.5 sec); Theme 2 (1'2), the repetitin and develpment f the same meldy withut the underlying rnetrics (13. 5 sec); Theme 3 (T3), the successin f seven shrt meldic phrases; Theme 4 (T4), an arpeggiated transitin (9 sec); Theme 5 (T5), the repetitin f Tl; Theme 6 (T6), the repetitin f T2. Results Table 6 presents the mean percentage f ttal desynchrnizatin time during each theme. Synchrnizatin with the intrductry "metric" was almst perfect fr all children. Half the children beat the quarter nte when the ther half beat the eighth nte, but in bth cases the beats were very regular and well synchrnized. Desynchrnizatin appeared when the metric disappeared at the end f T1 and during T2, even thugh the meldy was the same. Desynchrnizatins were als frequent in T3. This is surprising, since the main musical line and the shrt meldic phrases were very regular. Only T4 had a mre fluid rganizatin, s it is nt surprising that it was accmpanied by almst % desynchrnizatin time. It was als pssible t determine at what pint in the music the desynchrnizatins started by examining the sequence f cded musical events and the parallel sequence f desynchrnizatin. This analysis shwed that desynchrnizatins were particularly numerus when expressive variatins in intensity appeared; increasing intensities, far frm regularizing beats, had a destabilizing influence. Mrever, the beginnings and ends f mel- Table 6 Percent Desynchrnizatin Time During Each Theme in Experiment 4 %Desynch. Theme Time Difference Tl 7 T2 19.7 F(I,15) = 5.007, p <.05 T5 8.7 T6 22 T3 29.7 T4 39 Tl, T2, T5, T6 T3, T4 F(I,15) = 4.317, P <.05 Nte-T1 and T2 have the same meldic lines but metric accmpaniment disappears in T2. This difference is repeated in T5 and T6. T3 and T4 are cmpsed differently: small meldic phrases in T3, arpeggis in T4. These tw last themes d nt differ frm each ther but d differ frm T1 T2 T5 T6 tgether. n.s. F(I,15) = 13.262, P <.01 dies disturbed synchrnizatin t. Since structural accents were derived frm the cnjunctin f several imprtant events, it culd be predicted that they shuld disturb perfrmances as well. This was indeed bserved. Discussin Structural accents are the cnjunctin in time f several imprtant events in the music, s it culd be predicted that if they drew attentin t a particular pint in time, children wuld find it easier t synchrnize with the music. The ppsite effect was fund; when several events ccurred simultaneusly, the children lst track f the metric structure f the music. This culd be interpreted by the fact that mre prcessing time is required when mre infrmatin arrives simultaneusly and nging activity is thereby inhibited. In this real-life music situatin, expressive structure f a piece f music camuflages the metric structure. GENERAL DISCUSSION In develpmental studies f rhythm perceptin, researchers have cncentratedn the perceptin f tempral structure. It is significant that hardly anyne, even thse wh have carried ut cnsiderable studies f children, refers t the perceptinf accents (Dwling & Harwd, 1986; Hargreaves, 1986; Slbda, 1985; Zenatti, 1981). Can we therefre cnclude that children (and indeed the researchers we have just mentined) cnsider accents t be unimprtant? The starting pint fr this series f experiments was ur bservatin that 6-year-lds are able t reprduce simple tempral structures, but that even 8-year-lds d nt reprduce intensity differences. We have shwn that this inability cannt be explained by prblems f mtr cntrl, since under ptimal cnditins (Experiment 2) 7 year-lds can sufficiently cntrl the intensity ftheir taps at will. In a similar way, frm the age f 6, children are able t discriminate between tw sequences that vary in tempral structure, but even 8-year-lds are nt able t discriminate n the basis f accentuatin alne in cmplex tempral structures (Experiment 1). We exclude the pssibility that children are unable t hear the differences in intensity between the lud and sft taps, since, under ptimal cnditins (Experiment 2), they are able t discriminate n the basis f metric accents. In ther wrds, children are able t hear and reprduce accents, but they d nt d this spntaneusly befre the age f 8, and, even then, they d s inefficiently. We have als seen that even if accents are neither cnsidered imprtant by themselves nr reprduced, they d help t structure the tempral aspects f a rhythm (Experiments 1 and 3). S accents enhance the perceptual rganizatin f a rhythm. We therefre need t explain hw accents can play such a rle. The peridic ccurrence f accents is nt nly a cue fr segmentatin int grups (Cper & Meyer, 19; Deliege, 1987; Teplv, 1966), but als the basis f met-
0 GERARD AND DRAKE ric rganizatin (Cper & Meyer, 19). Cnsequently, perhaps by underlying the gruping r metric structure f rhythm, accents allw a higher level f perceptual hierarchy. It wuld therefre be interesting t examine frmal analyses f the hierarchical structure f music (Lerdahl & Jackendff, 1983a, 1983b; Lnguet-Higgins & Lee, 1984) in the light f develpmental studies. But these frmal cnsideratins d nt explain the psychlgical prcesses invlved, which will be discussed nw. In sme psychlgical studies, it has been suggested that meldy and rhythm are prcessed independently; likewise, we have suggested that accentuatin and tempral aspects f rhythm are prcessed independently. Althugh we have seen that yung children are able t cncentrate n the dynamic dimensin alne (Experiment 2), we have als seen that the presence f intensity accents influences the perceptin f the tempral structure f the rhythms (Experiments 1 and 3), demnstrating a certain degree f interactin. We can therefre reject the nw dated ntin f tw separate channels (ne fr intensity and ne fr time) with independent prcessing. In its place appears the ntin f independence in respnse rganizatin; it is as if the children were unable t cncentrate n bth aspects f rhythm at the same time. Our findings can be explained in relatin t studies f divided attentin and characteristics f attentinal resurces. Histrically, the first explanatins f attentinal selectivity (in the framewrk f "filter mdels") emphasized the richness and cmplexity f the infrmatin presented at ne time, alng with the risk f verlad (Bradbent, 1958). Filter mdels f attentin tend tday t have been replaced by "resurce capacity mdels" (Kahneman, 1973; Kahneman & Treisman, 1984). Ifwe cnsider the cncept f resurce in the sense used by Nrman and Bbrw (1975), resurces include "such things as prcessing effrt, varius frms f memry capacity, and cmmunicatin channels" (p. 45). The term resurce is mainly reserved fr the fuel metaphr, yet Hirst (1986), Hirst and Kalmar (1987), and Hirst, Spelke, Reaves, Caharack, and Neisser (19) cnsider (1) that cgnitive structures and skills are resurces; (2) that resurces must differ alng stimulus and respnse dimensins; and (3) that the number fdistinct resurces is nt fixed, but can change with subjects' practice and experience. These theretical cnsideratins fit well with ur results. Easy and adequate executin f ur tasks will depend n the cmplexity f the material, the subject's familiarity with this material, and the degree fmaturatin fthe system. As expertise (acculturatin, learning, maturatin) increases, less prcessing time is required. Thus, fr musicians with a high degree f expertise, prcessing ccurs in an autmatic fashin and requires little attentin. Reprductin f bth tempral and dynamic structure can be realized effectively. Hwever, fr yung children, wh have nt as yet develped the necessary expertise, each type f executin requires cnsiderable attentin, s they cncentrate n either ne r the ther f the aspects f the stimulus, nt n bth. We have seen that tempral structure seems t be mre fundamental than accent structure when accent structure reinfrces the metric ne; when tw types f infrmatin are in cmpetitin, pririty fr attentin is given t tempral aspects. Hwever, when dynamic structure is cnsidered within a larger framewrk, as in Experiment 4, we have als seen that dynamic infrmatin can upset tempral structure. The questin f hw this expertise might be acquired is cmplex. First, it culd be that maturatin is sufficient fr the cgnitive prcesses t becme functinal. Secnd, acculturatin (passive acquisitin f musical laws thrugh expsure t music) culd explain hw children gradually acquire the necessary rules. Finally, active learning culd be necessary fr the acquisitin fcertain abilities; withut training, these abilities might never be acquired. Sme experiments n the rle f musical cmpetence in rhythm reprductin have shwn that active learning imprves dynamic structuratin mre than tempral structuratin (Gerard, 1988b; Gerard, Drake, & Auxiette, 19). New experiments n the relative abilities f 6- and 8-year-ld children, adult nnmusicians, and adult musicians regarding tempral and dynamic structure shuld clarify this pint (Drake & Gerard, 19). Let us end n a musical nte. We have seen that tempral infrmatin rather than peridic accent is mre fundamental in children; the abilities invlved in perceptin seem t develp at an earlierage fr the frmer. A parallel can be drawn with the histry f western music. The rhythmic structure f religius chants was based n the dynamics f speech. A repetitive, regular intensity structure attesting t a metric rganizatin nly appeared later, when ppular dance music wn the status f "music" (Lachartre, 1985). Mrever, if we examine a musical scre tday, tempral infrmatin is clearly indicated (the length f each nte in relatin t the thers is explicit; the temp is stated), whereas dynamic structure is left much mre t the musician's intuitive knwledge f the accents invlved in bth metric and gruping structures. Perhaps everyne can prduce a tempral structure, but nly musicians are able t recnstitute it with the apprpriate dynamic structure that will make it a living, musical rhythm. REFERENCES BROADBENT, D. E. (1958). Perceptin and cmmunicatin. Lndn: Pergamn. COOPER, G., & MEYER, L. B. (19). The rhythmic structure fmusic. Chicag: University f Chicag Press. DEuEGE, I. (1987). Gruping cnditins in music listening: An apprach t Lerdahl and Jackendffs gruping preference rules. Music Perceptin, 4, 325-3. DEUTSCH, D. (1982). The internal representatin f infrmatin in the frm f hierarchies. Perceptin & Psychphysics, 31, 596-598. DWUNG, w. J., & HARWOOD, D. L. (\986). Music cgnitin. Orland, FL: Academic Press. DRAKE, c., & GERARD, C. (\989). A psychlgical pulse train: Hw yung children use this cgnitive framewrk t structure simple rhythms. Psychlgical Research, 51, 51. DRAKE, c., & GERARD, C. (19). The reprductin fsimple rhythms
CHILDREN AND INTENSITY DIFFERENCES IN RHYTHM 1 by children and adults (musicians and nnmusicians} is imprved by the intrductin f intensity accents. Manuscript in preparatin. FRAISSE, P. (1956). us structures rythmiques. Luvain: Publicatins Universitaires de Luvain. GABRlELSSON, A. (1973). Similarity ratings and dimensin analyses f auditry rhythm patterns: I. Scandinavian Jurnal fpsychlgy, 14, 138-1. GABRlELSSON, A. (1974). Perfrmance f rhythm patterns. Scandinavian Jurnal f Psychlgy, 15, 63-72. GERARD, C. (l988a). L'enfant et Ie rythme: Etudes experirnentales de la discriminatin et de la reprductin de rythmes musicaux. Vibratins, 6, -5. GERARD, C. (I988b). Perceptin du temps et du rythme: L'enfant dispse-t-il d 'une hrlge interne? Marsyas: Revue de l'lnstitut de Pedaggie Musicale et Chrgregraphique, 8, 75-85. GERARD, C., '" AUXIETTE, C. (1988). The rle played by meldic and verbal rganisatin in the reprductin f rhythmic grups by children. Music Perceptin, 6, 173-192. GERARD, C.; DRAKE, c., '" AUXIETTE,C. (19). The rle played by musical cmpetence in the reprductin frhythmic patterns by children. Unpublished manuscript. GERARD, C., '" REMOND, C. (1985). Evlutin genetique des reprductins rythmiques seln les cnditins de presentatin du mdele. Unpublished master's thesis, Labratire de Psychlgie Experimentale, Paris V. HANDEL, S. (1974). Perceiving meldic and rhythmic auditry patterns. Jurnal fexperimental Psychlgy, 3, 922-933. HARGREAVES, D. J. (1986). The develpmental psychlgy fmusic. Cambridge, England: Cambridge University Press. HIRST, W. (1986). Aspects f divided and selective attentin. In J. Ledux & W. Hirst (Eds.), Mind and brain (pp. 5-141). New Yrk: Cambridge University Press. HIRST, W., '" KALMAR, D. (1987). Characterizing attentinal resurces. Jurnal f Experimental Psychlgy: General, 116,68-81. HIRST, W., SPELKE, E., REAVES, c., CAHARACK, G., '" NEISSER, U. (19). Dividing attentin withut alternatin r autmaticity. Jurnal f Experimental Psychlgy: General, 9,98-117. JONES, M. R. (1976). Time, ur lst dimensin: Tward a new thery f perceptin, attentin and memry. Psychlgical Review, 83, 323-355. JONES, M. R. (1987). Dynamic pattern structure in music: Recent thery and research. Perceptin & Psychphysics, 41, 621-634. JONES, M. R., BLTZ, M., '" KIDD, G. (1982). Cntrlled attending as a functin f meldic and tempral cntext. Perceptin & Psychphysics, 32, 211-218. KAHNEMAN, D. (1973). Attentin and effrt. Englewd Cliffs, NJ: Prentice Hall. KAHNEMAN, D., '" TREISMAN, A. (1984). Changing views fattentin and autmaticity. In R. Parasuraman & D. R. Davies (Eds.), Varieties fattentin (pp. 29-6i). New Yrk: Academic Press. LACHARTRE, N. (1985). Mesures musicales. In Encyclpaedia Universalis (pp. 1225-1227). Paris: Encyclpedic Universalis. LERDAHL, F., '" JACKENDOFF, R. (l983a). A generative thery ftnal music. Cambridge, MA: MIT Press. LERDAHL, F., '" JACKENDOFF, R. (l983b). An verview f hierarchical structure in music. Music Perceptin, I, 229-252. LNGUET-HIGGINS, H. c., '" LEE, C. S. (1984). The rhythmic interpretatin f mnphasic music. Music Perceptin, I, 424-441. MOOG, H. (1976). The musical experience f the preschl child (c. Clarke, Trans.). Lndn: Schtt. MONAHAN, C. B., '" CARTERETTE, E. C. (1985). Pitch and duratin as determinants f musical space. Music Perceptin, 13, 1-32. MONAHAN, C. B., KENDALL, R. A., '" CARTERETTE, E. C. (1987). The effect f meldic and tempral cntur n recgnitin memry fr pitch change. Perceptin & Psychphysics, 41, 576-0. NAKAMURA, T. (1987). The cmmunicatin f dynamics between musicians and listeners thrugh musical perfrmance. Perceptin & Psychphysics, 41, 525-533. NORMAN, D. A., '" BBROW, D. J. (1975). On data-limited and resurcelimited prcesses. Cgnitive Psychlgy, 7, 44-64. PALMER, C., '" KRUMHANSL, C. L. (1987a). Independent tempral and pitch structures in determinatin f musical phrases. Jurnal fexperimental Psychlgy: Human Perceptin & Perfrmance, 13, 116-126. PALMER, C., '" KRUMHANSL, C. L. (1987b). Pitch and tempral cntributins t musical phrase perceptin: Effects f harmny, perfrmance timing, and familiarity. Perceptin & Psychphysics, 41. 5-518. POVEL, D. J. (1981). Internal representatin f simple tempral patterns. Jurnal fexperimental Psychlgy: Human Perceptin & Perfrmance, 7, 3-18. POVEL, D. J. '" EsSENS, P. (1985). Perceptin f tempral patterns. Music Perceptin, 2, 411-4. POVEL, D. J., '" OKKERMAN, H. (1981). Accents in equitne sequences. Perceptin & Psychphysics,, 565-572. SLOBODA, J. A. (1985). The musical mind: The cgnitive psychlgy f music. Oxfrd, England: Clarendn Press. TEPLOV, B. M. (1966). Psychlgie des aptitudes musicales. Paris: Presses Universitaires de France. THOMAS, E. A. C., '" WEAVER, W. B. (1975). Cgnitive prcessing and time perceptin. Perceptin & Psychphysics, 17, 363-367. THOMASSEN, J. M. (1982). Meldic accent: Experiments and a tentative mdel. Jurnal f the Acustical Sciety f America, 71, 1596-15. ZENATTI, A. (1981). L'enfani et sn envirnnement musical. Paris: Presses Universitaires de France. (Manuscript received March, 1989; revisin accepted fr publicatin January 31, 19.)