association for psychological science

Infants' Perception of Phrase Structure in MusicAuthor(s): Carol L. Krumhansl and Peter W. JusczykSource: Psychological Science, Vol. 1, No. 1 (Jan., 1990), pp. 70-73Published by: Sage Publications, Inc. on behalf of the Association for Psychological ScienceStable URL: http://www.jstor.org/stable/40062394 .

Accessed: 23/04/2013 16:17

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

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Sage Publications, Inc. and Association for Psychological Science are collaborating with JSTOR to digitize,preserve and extend access to Psychological Science.

http://www.jstor.org

This content downloaded from 132.236.27.111 on Tue, 23 Apr 2013 16:17:53 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=sage

http://www.jstor.org/action/showPublisher?publisherCode=assocpsychsci

http://www.jstor.org/stable/40062394?origin=JSTOR-pdf

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


INFANTS' PERCEPTION OF

PHRASE STRUCTURE IN MUSIC

Carol L. KrumhansP and Peter W. Jusczykb "Cornell University and Institut de Recherche et Coordination Acoustique et Musique, Paris, and

b University of Oregon and Labor atoire de Sciences Cognitive s et Psycholinguistique and

Ecole des Hautes-Etudes en Sciences Sociales, Paris

PSYCHOLOGICAL SCIENCE

Research Report

Abstract - A visual preference procedure was used to examine 6- and ^-month- old infants* sensitivity to phrase structure in music. Sections of Mozart minuets were divided into segments that either did or did not correspond to the phrase structure of the music. Infants in both age groups listened significantly longer to the appropriately segmented versions. Their behavior accorded well with judgments of the same materials made by adults, suggesting that protracted musical experience may not be necessary to perceive phrase structure in music. Strong correlations were found between certain musical variables and the infants* preferences for the musical passages, pointing to acoustic proper- ties that may be important for defining musical phrases.

When William James (1890) described the infant's world as a "blooming, buzz- ing confusion," he was reflecting on how infants are able to organize the diverse array of continually changing impres- sions received through different sensory modalities into a coherent picture of the structure of the world. Since James' time, much has been learned about the perceptual capacities that allow the infant to pick out objects and events in the environment (e.g., Haith & Campos, 1983; Mehler & Fox, 1985). Neverthe- less, some fundamental questions remain to be answered. One such question is how infants segment complex acoustic signals so as to arrive at a description that identifies the important component

events and expresses the relations among them.

Perhaps the best known example of segmentation is the ability of a native speaker of a language to analyze utter- ances into a hierarchy of discrete units such as words, phrases, and clauses de- spite the fact that pauses between suc- cessive units are often absent from the acoustic signal (Jusczyk, 1986). Al- though this ability, essential for recover- ing the meaning of an utterance, is not fully understood, experimental evidence indicates that boundaries between important units may be cued in part by rhythm and intonational contour (Coo- per & Paccia-Cooper, 1980; Nakatani & Dukes, 1978). Not only are adult listeners sensitive to such cues (Collier & t'Hart, 1975; Lehiste, Olive, & Streeter, 1976; Scott, 1982; Scott & Cutler, 1984), but recent studies demonstrate that 6- month-old infants show clear listening preferences for correctly versus incor- rectly segmented speech (Hirsh-Pasek, Kemler-Nelson, Jusczyk, Wright- Cassidy, Druss, & Kennedy, 1987; Kem- ler-Nelson, Hirsh-Pasek, Jusczyk, & Wright-Cassidy, 1989). The latter results imply that the infant is already detecting some sort of organization of segments in- herent in the acoustic stream of speech.

Music shares with speech the charac- teristic of being hierarchically organized into units of varying temporal extents (Deutsch & Feroe, 1981; Lerdahl & Jackendoff, 1983). A short segment con- stitutes a unit of one level and then joins with other segments to form longer units at higher levels in the hierarchy. Percep- tually, the segmentation of music de- pends on a variety of factors, including contrasts of pitch range, dynamics, and timbre, lengthening of durations, changes of melodic contour, and metri- cal, tonal, and harmonic stress (Clark & Krumhansl, in press; Palmer & Krum-

hansl, 1987a, b; Simon & Sumner, 1968). Although perceptual segmentation of music by infants has not been investi- gated previously, recent studies have shown that infants in the first year of life are sensitive to changes of melodic contour and rhythm (Trehub, 1987). In addition, perceptual grouping by infants on the basis of changes of pitch range, timbre, and dynamics has been demon- strated (Thorpe & Threhub, 1989; Thorpe, Trehub, Morrongiello, & Bull, 1989). Thus, these factors may contrib- ute to segmentation of music at an early age.

The present study was undertaken to investigate the degree of which infants are sensitive to phrase structure in music and to identify the factors underlying this sensitivity. The experiment tested their preferences for alternative segmenta- tions of simple musical passages, which were the initial sections of Mozart minuets chosen for their relatively obvious and regular melodic, harmonic, and rhythmic structure. In some versions (called Natural), the musical passage was divided by short pauses into segments that corresponded to the phrases in the music. In other versions (called Unnatrual), the pauses were inserted in the middle of phrases. This was accom- plished by beginning the Unnatural version at some point other than the start of the piece while keeping unchanged the pattern of inserted pauses. Because the initial sections of the pieces are re- peated, the two versions contained the same musical material, presented in the same order as the original music. Figure 1 shows, for one of the minuets, the Nat- ural and Unnatural versions. The experiment employed the headturn preference procedure used previously in studies of speech perception by infants (Fernald, 1985; Hirsh-Pasek et. al., 1987; Kemler- Nelson et al., 1989).

Correspondence and reprint requests to: Carol L. Krumhansl, Dept. of Psychology, Uris Hall, Cornell University, Ithaca, NY 14853 or Peter W. Jusczyk, Dept. of Psychol- ogy, University of Oregon, Eugene, OR 97403.

70 Copyright © 1990 American Psychological Society VOL. 1, NO. 1, JANUARY 1990




Carol L. Krumhansl and Peter W. Jusczyk

Fig. 1. Top: Shows one Mozart minuet in which pauses are inserted at phrase boundaries; this is the Natural version. Bottom: Shows the Unnatural version of the same minuet; the pattern of pauses is identical but this version begins at the sixth measure of the original music.

METHODS

Subjects

The first group contained 24 infants of approximately 6 months of age (avg. age = 25.6 weeks); two additional infants were tested, but they cried and did not complete the session. The second group contained 24 infants of approximately 4V2 months of age (avg. age = 19.4 weeks); six additional infants did not complete the session - three because they failed to meet the requirement of looking for at least 3 sec to either side of the test booth (see procedure), two because they cried, and one because of equipment failure.

Apparatus and Stimulus Material

Sixteen Mozart minuets were se- lected from Trente pieces facile pour piano (Bruxelles: Schott Frfcres). Each contained an inital section of either 8 or 10 measures followed by a repeat sign. A given stimulus consisted of the initial section and its repetition, so that all stimuli were either 16 or 20 measures in length. All were notated in three-quarter

time and played at a tempo of 125 quarter-notes per minute by a Yamaha TX802 FM tone generator synthesizing a piano timbre under the control of a Macintosh Plus computer. Each stimulus consisted of either 4 or 6 phrases (of measure lengths 2, 4, or 6). Two versions of each stimulus were produced. In the Natural version, two-beat pauses (of duration .96 sec) were inserted at the end of each phrase as determined by the experimenters' intuitions. In its Unnatural counter- part, the pattern of pauses was identical. However, the segments began after 1,3, 5, 7, or 9 measures of the music (and ended at the same point after playing through the music twice); this had the consequence that the pauses did not co- incide with the phrase endings.1 The

Natural and Unnatural versions of each minuet were recorded on audio tape and later redigitized and stored on an LSI 1 1/ 73 computer for on-line presentation during the experiment.

Procedure

The infant was seated on its parent's lap in the center of a three-sided test booth. A red light and a loudspeaker were mounted at eye level on each side of the side panels and a green light was mounted on the center panel facing the infant. An observer hidden behind the center panel looked through a peephole and recorded the direction and duration of the infant's headturns by pressing but- tons on a response box linked to an LSI 1 1/73 computer in an adjacent room. The observer was not informed as to which loudspeakers the Natural and Unnatural samples were assigned. In addition, during the course of the experiment, both the observer and the infant's parent listened to recorded music over headphones to keep them unaware of the Nat- ural and Unnatural versions. The infant consistently heard the Natural versions of the minuets through the loudspeaker on one side of the booth and the Unnat- ural versions on the other side. Whether the Natural versions appeared on the left or right side was counterbalanced across subjects.

An 8-trial familiarization period served to acquaint the infants with the assigned positions of the Natural and Unnatural versions. Each trial began by blinking the green light to draw the infant's attention to center. When the infant was so oriented, the center light was extinguished and the red light above one of the loudspeakers began to flash. When the infant made a head-turn of at least 30° in the direction of the loudspeaker, the musical sample began to play and con-

1. As a check that the locations chosen for inserting pauses in the Natural samples coincided with adult listeners1 perception of phrase structure, a group of 14 adults was asked to judge which version of each minuet sounded "more natural." Small groups of 3 to 6 subjects were tested in a sound insulated room equipped with individual test cubicles; they listened to the stimuli over headphones. On a given trial, subjects listened to both versions of each minuet; on half the trials, the Natural version occurred first; on the other half, the Unnatural version occurred first. Overall, the Natural version was preferred on 79% of the trials (/(13) = 7.22, p < .0001), confirming the experimenters' intuitions concerning the phrase structure of the pieces. That performance was not higher might be attributable to the demands of having to remember the two relatively extended passages for comparison, and the fact that these listeners had little or no musical training.

VOL. 1, NO. 1, JANUARY 1990 71




Infants' Perception of Musical Phrases

tinued until completion or until the infant failed to maintain the 30° headturn for 2 sec. Minuets numbers 1, 3, 7, and 24 were used on the familiarization trials. The procedure was unchanged for the 12 test trials that followed. During this phase, half the infants heard the Natural and Unnatural version of 6 of the minuets (numbers 4, 5, 8, 14, 15, and 16), and the other half the infants heard the two versions of the other six minuets (numbers 9, 12, 13, 17, 18, and 23). Within the block of 12 test trials, no effort was made to present the alternative versions of a minuet successively. The ordering of the stimuli was random with the provision that for three of the minuets, the Natural version occurred before the Unnatural version, and vice versa.

RESULTS AND DISCUSSION

Preferences for the alternative versions of the minuets were indexed by measuring the amount of time that the infants oriented to the loudspeaker playing each type of sample. For the 6- month-old infants, the mean orientation times for the Natural and Unnatural version were 10.23 and 8.03 sec, respectively. Twenty-two of the 24 infants showed this pattern, and the difference is highly significant statistically (f(23) =

5.08, p < .0001). Moreover, for each of the 12 minuets, the mean orientation time was longer for the Natural version than for the Unnatural version (f(ll) =

6.09, p < .0001). A variety of analyses were under-

taken to determine the musical correlates of these looking time differences. The first analysis considered the pitch of the tones surrounding the pauses. (This, and the following analysis involving durations, considered the first tones of each stimulus as "after" a pause and the last tones as "before" a pause. In other words, the gap between trials was counted as a pause.) The top Figure 2 shows the average pitch of the last three tones before the pauses and the first two tones following the pauses on a scale of pitch height from low to high where C4 is middle C. Whenever two tones were sounded simultaneously, the higher tone was included in the analysis. As can be seen, there is a sharp drop in the average pitch height before the pauses in the Nat- ural version, but not the Unnatural ones.

Fig. 2. Top: Shows the average pitch of the last three tones before the pauses and the first two tones after the pauses for the Natural (solid line) and Unnatural (broken line) versions. Bottom: Shows the average pitch duration of the last three tones before the pauses and the first two after the pauses for the Natural (solid line) and Unnatural (broken line) versions.

The values for the individual stimuli were correlated with the orientation time data. Significant correlations were found for the second to last position (r(22) = - .422, p = .04) and the last position (K22) = - .490, p = .014) before the pauses.

A second analysis looked at the durations of tones surrounding the pauses. Because the durations in the melody tones showed greater variability, only these were considered. The average durations of the last three tones before the pauses and the first two after the pauses are shown at the bottom of Figure 2 on a scale of pitch duration where a quarter- tone is equal to .48 sec. A sharp increase in pitch duration before the pauses oc- curs for the Natural versions, but not for the Unnatural ones. When the values for the individual stimuli were correlated with the orientation times, significant correlations were observed for the second to last (r(22) = .42, p = .039) and last (r(22) = .535, p = .007) positions before the pauses.

Of the other musical variables considered, only one showed a significant rela-

tionship with the orientation times. This variable was based on the interval formed by the last pitch of the melody and its accompanying bass note. The proportion of intervals preceding pauses that were octaves was much higher for the Natural (.583) than for the Unnatural (.111) versions. Moreover, the octave measure correlated significantly with orientation times (r(22) = .472, p = .020). Thus, orientation times were correlated with three characteristics of the ends of the musical segments before the pauses: drops in pitch, lengthening of durations, and presence of octave intervals. The results did not depend on the beginning characteristics of the segments. In particular, there appeared to be no effect attributable to the fact that the Natural versions began at the start of the piece whereas the Unnatural versions did not. The two versions did not differ significantly in terms of the pitch or duration of the first two tones, nor did these variables correlate with the orientation time data. Nor can the results be attributed to the final ending, because the infants did not listen to the very end of the stimuli.

72 VOL. 1, NO. 1, JANUARY 1990




Carol L. Krumhansl and Peter W. Jusczyk

The results for the 4V2-month-old infants were similar to those for the 6- month-old infants. They oriented longer to the Natural (12.81 sec) than to the Un- natural (10.05 sec) versions of the minuets. Again, this difference was quite stable. Twenty of the 24 infants showed this

pattern, and the difference was highly significant statistically (f(23) = 3.341, p = .0028). Moreover, for 11 of the 12 minuets, the mean orientation time was

longer for the Natural version than the Unnatural version (f(ll) = 3.182, p =

.0087). The pattern of correlations between the orientation times and the musical variables was similar. The pitch height and pitch duration of the last position before the pause correlated significantly with the orientation times (r(22) = -.502,/? = .0125 and K22) = .513,/? = .0104, respectively). Although in the same direction, the correlation between the octave measure and orientation times was not significant (r(22) = .291, p =

.1671) for this group. Again, the orientation times were not correlated with characteristics of the first tones after the

pauses, or those at the very beginning of the stimulus, in particular.

This study demonstrates that infants

prefer to listen to musical passages that are interrupted at phrase boundaries

compared to ones interrupted in the middle of phrases. This result implies that

they perceive segmental organization in musical passages in the sense that some

places appear to afford better stopping points than others. In this respect, infants behave much like adult listeners (Clarke & Krumhansl, in press; Palmer & Krumhansl, 1987a, b), which suggests that the ability to perceptually segment passages into phrases may not require protracted musical experience. Instead, certain acoustic markers appear to in- duce appropriate segmentation of the musical stream. The correlations between musical variables and the infants' orientation times offer some suggestions as to the potential cues for segmentation, viz. a drop in pitch height at the ends of

phrases, a phrase-final increase in pitch duration in the melody, and a predomi- nance of octave simultaneities at the ends of phrases.

Our demonstration of sensitivity to segmental organization accords well with other observations that have been made of music perception by infants, particularly their sensitivity to melodic contour and rhythmic patterns (Trehub, 1987), grouping on the basis of pitch range (Thorpe & Trehub, 1989; Thorpe et al., 1989), and octave equivalence (Demany & Armand, 1984). The present study goes a step further in suggesting that such variables may provide the means for segmenting the musical stream. In addition, there are some obvious parallels between the present study and those demonstrating infants' sensitivity to acoustic correlates of segments in speech (Hirsh-Pasek et al., 1987; Kemler-Nelson et al., 1989). Hence some of the proposed acoustic cues for segmental organization in speech, such as falling intonational contour and syllable lengthening, seem com- parable to the pitch height and duration variables present in music. Indeed, when considered in conjunction with the work on speech segmentation, the present study offers some intriguing suggestions about general acoustic cues that the infant may use to segment the auditory world.

Acknowledgments - This research was supported by grants from N.I.C.H.D. (HD-15795) to PWJ and by N.S.F. (8701948) to CLK. The authors would like to thank Tracey Schomberg and Nan Koenig for their help in testing subjects, Eric Bylund for his technical assistance, and Cort Lippe for advice on musical is- sues. The ordering of the authors was determined arbitrarily; both authors partici- pated equally in the project.

REFERENCES

Clarke, E.F., & Krumhansl, C.L. (in press). Per- ceiving musical time. Music Perception.

Collier, R., & t'Hart, J. (1975). The role of intona- tion in speech perception. In A. Cohen & S.G. Nooteboom (Eds.), Structure and process in speech perception. Heidelberg: Springer.

Cooper, W.E., & Paccia-Cooper, J. (1980). Syntax and speech. Cambridge, MA: Harvard Univer- sity Press.

Demany, L., & Armand, F. (1984). The perceptual

reality of tone chroma in early infancy. Jour- nal of the Acoustical Society of America, 76, 57-66.

Deutsch, D., & Feroe, J. (1981). The internal repre- sentation of pitch sequences in tonal music. Psychological Review, 88, 503-522.

Fernald, A. (1985). Four-month-old infants prefer to listen to motherese. Infant Behavior and De- velopment, 8, 181-195.

Haith, M., & Campos, J.J. (Eds.). (1983). Carmi- chael's handbook of child psychology : Infancy and developmental psychology (Vol. 2). New York: Wiley.

Hirsh-Pasek, K., Kemler-Nelson, D.G., Jusczyk, P.W., Wright-Cassidy, K., Druss, B., & Kennedy, L. (1987). Clauses are perceptual units for young infants. Cognition, 26, 269-286.

James, W. (1890). Principles of psychology. New York: Holt.

Jusczyk, P.W. (1986). A review of speech perception research. In K. Boff, L. Kaufman, & J. Thomas (Eds.), Handbook of perception and human performance (Vol. 2). New York: Wiley.

Kemler-Nelson, D.G., Hirsh-Pasek, K., Jusczyk, P.W., & Wright-Cassidy, K. (1989). How the prosodic cues in motherese might assist language learning. Journal of Child Language, 16, 53-68.

Lehiste, I., Olive, J.P., & Streeter, L. (1976). The role of duration in disambiguating syntactically ambiguous sentences. Journal of the Acousti- cal Society of America, 60, 1199-1202.

Lerdahl, F., & Jackendoff, R. (1983). A generative theory of tonal music. Cambridge, MA: MIT Press.

Mehler, J., & Fox, R. (Eds.). (1985). Neonate cognition. Hillsdale, NJ: Lawrence Erlbaum.

Nakatani, L., & Dukes, K. (1978). Locus of segmental cues for word juncture. Journal of the Acoustical Society of America, 64, 1582-1592.

Palmer, C, & Krumhansl, C.L. (1987a). Indepen- dent temporal and pitch structures in perception of musical phrases. Journal of Experimen- tal Psychology: Human Perception and Per- formance, 13, 116-126.

Palmer, C, & Krumhansl, C.L. (1987b). Pitch and temporal contributions to musical phrase perception: Effects of harmony, performance tim- ing, and familiarity. Perception & Psychophys- ics, 41, 505-518.

Scott, D.R. (1982). Duration as a cue to the perception of a phrase boundary. Journal of the Acoustical Society of America, 71, 996-1007.

Scott, D.R., & Cutler, A. (1984). Segmental phonol- ogy and the perception of syntactic structures. Journal of Verbal Learning and Verbal Behav- ior, 23, 450-466.

Simon, H.A., & Sumner, R.K. (1968). Pattern in music. In B. Kleinmuntz (Ed.), Formal repre- sentation of human judgment. New York: Wiley.

Thorpe, L.A., & Trehub, S.E. (1989). Duration illu- sion and auditory grouping in infancy. Devel- opmental Psychology, 25, 122-127.

Thorpe, L.A., Trehub, S.E., Morrongiello, B.A., & Bull, D. (1989). Perceptual grouping by infants and preschool children. Developmental Psy- chology, 24, 484-491.

Trehub, S.E. (1987). Infants' perception of musical patterns. Perception & Psychophysics, 41, 635-641.

(Received 5/18/89; Accepted 7/18/89)

VOL. 1, NO. 1, JANUARY 1990 73



association for psychological science

Documents