Psychology of Music - SAGE Publications · Psychology of Music Volume 31 2003 Editor: Susan Hallam....

sempre :Society for Education, Music and Psychology Research

SAGE Publications ISSN 0305-7356

Psychology of MusicVolume 31 2003

Editor: Susan Hallam

Psychology of MusicVolume Number January

E D I T O R I A L 5

A RT I C L E S

Jeanne BambergerThe development of intuitive musical understanding:a natural experiment 7

John McCormick and Gary McPhersonThe role of self-efficacy in a musical performance examination:an exploratory structural equation analysis 37

Luan Ford and Jane W. DavidsonAn investigation of members’ roles in wind quintets 53

Elizabeth S. NawrotThe perception of emotional expression in music: evidence frominfants, children and adults 75

Stephanie WilsonThe effect of music on perceived atmosphere and purchaseintentions in a restaurant 93

B O O K R E V I E W S

Robert Rowe, Machine MusicianshipReviewed by Jonathan Impett 113

Tim Miles and John Westcombe (eds), Music and Dyslexia:Opening New DoorsReviewed by Katie Overy 116

The development of intuitivemusical understanding: a naturalexperiment

7A R T I C L E

Psychology of Music

Psychology of MusicCopyright ©

Society for Education,Music and Psychology

Researchvol (): ‒

[- ():; ‒; ]

J E A N N E B A M B E R G E RM A S S A C H U S E T T S I N S T I T U T E O F T E C H N O L O G Y

A B S T R A C T Tracing the compositional process of two musically untrained college students, this close case study demonstrates their ability to producearchetypal tonal melodies, even when working initially within the constraints oftonally and metrically ambiguous melodic materials. The two students wererepresentative of a sample of about 75 who participated in a new approach to music fundamentals supported by a novel, interactive computer music environment. Students’ logs, including their composition sketches, decision-making, analysis of progressive modifications and completed compositions, serveas evidence and data for analysis. It is argued that, when students work at theirown pace with immediate sound feedback, can modify given materials and haveaccess to multiple representations at differing levels of detail, they are able tomake explicit their intuitive criteria for compositional decision-making, as well asproposing an intuitive model of a ‘sensible tune’.

K E Y W O R D S : archetypes, experimental methodology, learning, music theory, musicalintuitions, perception, structural functions

Introduction

This article reports on a close case study of two musically untrained collegestudents as they go about the task of composing melodies within the constraintsof certain given materials. The two students,1 who are the subjects in thisnatural experiment in music cognition and the development of musicalunderstanding, are representative of some 75 students who have participatedin a new approach to music fundamentals, supported by a novel interactivecomputer music environment. The students’ logs trace their compositionsketches, decision-making and analysis of progressive modifications. In thisarticle, these logs, together with their completed compositions, serve as thedata for analysis.

sempre :

The data show that the students, taken as typical musically untrainedadults, are able to produce coherent tonal melodies, even when given tonallyand metrically ambiguous melodic materials with which to work. With theopportunity to work at their own pace – with immediate sound feedback frominterim sketches, together with access to multiple representations at differinglevels of detail – the students are also able to develop, to some extent, explicitcriteria for their decision-making as they design-in-action.

I argue, in particular, that the students’ decision-making processes and theresulting compositions embody the schemes Rosner and Meyer (1982) havecalled ‘archetypes’:

[Archetypes] establish fundamental frameworks in terms of which culturallycompetent audiences . . . perceive, comprehend, and respond to works of art . . .Archetypes may play a significant role in shaping aesthetic experience andfostering cultural continuity in the absence of any conscious conceptualizationabout their existence, nature, or kinds. Rather, they may be and usually areinternalized as habits of perception and cognition operating within a set ofcultural constraints. (p. 318)

While Rosner and Meyer demonstrate the validity of these claims by askingsubjects to listen for instances of archetypal structures, in the cases discussedhere, students actually generate these archetypes. Even when given unfamiliar(modal) materials with which to work, the musically untrained studentsshape the materials to conform to the archetypal features and relations weassociate with the commonplaces of familiar folk and popular songs.

Analysis of the work of the two students complements but also raisesquestions concerning this growing body of research in cognition andperception. The vast majority of these experiments in the field have onlyinvolved perception, i.e. subjects are asked just to listen to carefully con-trolled, often very brief, stimuli and to make judgments along some pre-determined rating scale. Paradigmatic of these earlier studies and probablymost distant from the present naturalistic study are the so-called ‘probe-tone’experiments where subjects listen to musical stimuli intended to establish atonal context, and are then asked to make ranked judgments with regard tothe ‘fittedness’ of selected ‘probe tones’ (Leman, 2000: 481). While therehave been a number of variations on this approach, the work of Krumhansland her collaborators is paradigmatic of the model (Krumhansl and Kessler,1982; Clarke and Krumhansl, 1990; Krumhansl, 1990).

The work of Deliège et al. (1996) on ‘real-time listening’ and particularlythe so-called ‘puzzle’ experiment (pp. 141ff) most closely resembles the studyunder discussion here. However, there are significant differences in results.The authors of the ‘puzzle’ experiment report that the data ‘seeminglydemonstrate that non-musician subjects possessed little capacity to producecoherent tonal structures’ (p. 143). Moreover, ‘The results appear to indicatethat sensitivity to tonal–harmonic structure and function derives largelyfrom formal musical training’ (p. 155).

8 Psychology of Music 31(1)

A question to be addressed, then, is this: in spite of the similarities betweenthe two experiments, how can we account for the differences in results? Thereare a number of factors involved including the difference in materials. Non-musician subjects in the ‘puzzle’ experiment were asked to listen to segmentsfrom a Schubert dance written for piano, which they had not previouslyheard. The task was to

. . . recreate the most coherent piece possible within a given time using the ‘kit’of [prepared] segments. Subjects built a piece simply by moving these icons soas to arrange them in a linear order . . . Subjects could listen to the segmentsand to their constructed ‘piece’ as often as they wished. (p. 141)

The musically untrained students in our experiment were also givensegments represented by icons on a computer screen and asked to make acoherent piece by arranging the icons in a linear order. However, the seg-ments in the students’ situation were taken from Ambrosian chant (in contrastto a full-textured piano piece) which they had not heard previously, but thetask was simply to make a coherent melody, i.e. with no harmonic accompani-ment. Further, the boundaries of segments chosen from the Schubert piece insome cases (e.g. segments 1 and 2) seemed (to this listener) to interruptmelodic grouping in favor of harmonic boundaries. Given a single melodicline, the segments in the students’ experiment consisted of 5–8 melody notesspecifically chosen to be consistent with structural melodic boundaries.

Equally important were differences in the working conditions in the twosituations. In the students’ experiment:

● the task was open-ended; ● participants worked at individual work stations in a computer music lab

and in their own time; ● there were no time-constraints.

Further, working in the given computer environment, participants wereencouraged to:

● make small changes in the pitch or duration of the segments if they felt itnecessary in building coherence;

● listen critically and frequently to the results of their ongoing experiments;● actively reflect on their strategies by keeping a running log of decisions

and results;● make use of multiple kinds and levels of representations that were made

available.

The focus for the students, themselves, and also for the teacher/researcher,was on the evolution not the evaluation of the students’ work. I argue that allof these components together provide a greater potential for deeply inter-rogating the mental strategies guiding working perception – the ‘knowledge-in-action’ of musically untrained subjects (Schön, 1983: 59).

The contrasting views of experimental research illuminate the tension

Bamberger: Intuitive musical understanding 9

between, on the one hand, researchers who strive in their experimentaldesign and methodology for objectivity, including controlled environments(stimuli and choice of subjects), as well as consistent, statistical measures asunits of analysis; and, on the other hand, researchers whose experimentalenvironments are designed to be exploratory and to enhance the potential forclose naturalistic observation and probing analysis of generative behavior.(See Auhagen and Vos, 2000, for a view of this tension with respect to‘tonality induction’, in particular.)

In the Introduction to Vygotsky’s Mind in Society (1978), the editors(Michael Cole et al.) describe this contrast more generally:

. . . the purpose of an experiment as conventionally presented is to determinethe conditions controlling behavior. Quantification of responses provides thebasis for comparison across experiments and for drawing inferences aboutcause-and-effect relationships.

For Vygotsky, the object of experimentation is quite different . . . Vygotskybelieved that . . . to serve as an effective means . . . the experiment must providemaximum opportunity for the subject to engage in a variety of activities thatcan be observed, not just rigidly controlled. (pp. 11–12)

But rather than pitting one approach against the other, it is moreinteresting to assume that knowledge gained in each situation is useful andthen to think about the possible meanings of ‘rigor’ and ‘relevance’ in bothtypes of experimental design. In the light of such reflections, differences inthe nature of evidence and of results that accrue can be more productivelyand practically understood.

The task, the environment and the materials

The two students, whose work is followed in detail in this article, are typical ofthose undergraduates at Massachusetts Institute of Technology (MIT) whoelect to take the elementary music fundamentals course to satisfy a portion ofthe humanities/arts requirement for graduation. In classes of 12 to 15students, the majority are majoring in a science or engineering subject, mosthave had no formal music background, while one or two play someinstrument a little (usually self-taught guitar). The classes, which meet for 3 hours per week for about 12 weeks each semester, typically include a few1st-year, mostly 2nd- and 3rd-year, and a few 4th-year students. Over thepast 3 years, the students’ work has been facilitated by the text, DevelopingMusical Intuitions, and its accompanying computer environment, Impromptu(Bamberger, 2000).

The composition project, which is the primary focus here, is usuallyassigned in the 3rd week of the semester. A previous introductory projectinvolves students in simply reconstructing given tunes (some familiar, someunfamiliar) using as their ‘units of work’ melodic segments we call‘tuneblocks’. Figure 1 shows an abbreviated version of Impromptu’s computer


screen for reconstructing the tune, ‘Did You Ever See a Lassie’, one of some20 tunes included in the tuneblocks catalog with which students can work.New sets of tuneblocks are easily made.

Each of the patterned icons in the TUNEBLOCKS area, when clicked, playsone of three brief and structurally salient motives (‘tuneblocks’) needed toreconstruct the tune. The patterns on the tuneblocks icons are simply neutraldesigns with no reference to the melodic shapes. The intention is to focus thestudents’ attention on listening rather than looking. The tuneblock labeledLASS in the TUNEBLOCKS area of the screen plays the complete tune,‘Lassie’.

For those unfamiliar with the tune, ‘Lassie’, the score and the tuneblocksare shown in Figure 2.

To reconstruct the tune, students listen as often as they like to the wholetune and to the tuneblocks, individually. Then, dragging blocks into thePLAYROOM area, they experiment with arranging them and listening backto the results as they search for the order of occurrence that plays the com-plete tune. Pressing the space bar causes the synthesizer to play the blocks currently in the PLAYROOM while the GRAPHICS Window at thebottom of the screen shows a ‘pitch contour’ representation for these blocks –an easily accessible rough sketch of melodic shape. The PLAYROOM area in


F I G U R E 1 ‘Lassie’ in the tuneblocks window.

TUNEBLOCKS

LASS

PLAYROOM

CATALOG

GRAPHICS

F I G U R E 2 ‘Lassie’ in conventional music notation and tuneblocks.

Figure 1 shows the arrangement ofblocks that plays the first part of thetune, ‘Lassie’.

Nearly all students (including child-ren as young as 6 years of age) are ableto complete this task. This seems strongevidence that these structurally mean-ingful elements (tuneblocks) are intuitive

units of perception. The process of reconstructing a tune is one of‘constructive analysis’. That is, the students make and hear the structure of atune gradually emerging. When reconstruction is complete, the result ismappable onto a conventional schematic, e.g. a b a c, shown in Figure 3.

To help students reflect on their work, an ‘Explorations’ section in the textpoints out certain common organizing principles found in the reconstructedtunes. These include antecedent–consequent phrase relations (as in the

beginning of ‘Lassie’), rep-etition, return, sequenceand structural hierarchies.Hierarchies are representedin several ways, including‘structural trees’ such asthe one shown in Figure 4.Students were reminded ofthese analyses as they wenton to compose their owntunes in the next project.

Composing original tunes with tuneblocks

In this second project, the one with which this article is primarily concerned,students work within the same computer environment but now using a set ofunfamiliar tuneblocks as the material constraints with which to composetheir own original melodies. A critical part of both projects is asking studentsto reflect on and keep a log of the process as an integral part of the process itself.Thus, there are two rounds of investigation here: in the first round, thestudents’ reflections constitute research into their own intuitive under-standing; the second round involves meta-investigations into the results ofthe students’ personal research.

The students’ working process and their papers follow instructions givenin the text. Therefore, to provide the reader with relevant background, anabbreviated version of these instructions follows:


F I G U R E 3 Static, conventional schematic.

F I G U R E 4 Structural tree.

Instructions

Using a set of unfamiliar tuneblocks, make a tune of your own that makes senseand that you like. There is no given tune to match; there is no right answer.Consider the following questions as you listen to and experiment with theblocks:

● What are the specific features and relations that differentiate one block fromanother?

● What are the musical features that seem to generate the possible structuralfunction of each block (beginning, ending, middle, etc.)?

● Which blocks seem to go well together and why? Why do you dislike aparticular sequence of blocks? What did you do to fix it, and why is the newsequence better?

Make a description of the structure of your completed tune including thefunctional relations among the blocks, e.g. antecedent–consequent relations,repetition, return, etc. Also describe how you group tuneblocks to form biggerblocks (phrases and sections). Be sure to keep a log of your progress and try toaccount for the decisions you make along the way. (Bamberger, 2000: 26–7)

The sets of blocks with which the students composed tunes early in thiscomposition project were in the familiar tonal style of common folk songs.The set with which the students are concerned here were modal blocksactually taken from an Ambrosian chant. Students were not told that blockswere taken from an existing melody.

These materials and a subsequent atonal set were specifically chosenbecause it was expected that the students would hear them as ‘strange’. Withthis in mind, students were given the following questions as a basis forreflecting on their response to this material.

● In what ways are the features (e.g. rhythm, pitch relations) of these blocksdifferent from the others you worked with?● What can these differences tell you about the kinds of relations that you areused to and that you have come to take for granted as generating coherence inthe tunes you find ‘ordinary’? (Bamberger, 2000: 29)

As the papers will show, it was critically important that students wereencouraged to make small changes in the given blocks. They did so by usingthe Edit Window to open up the blocks and look at their ‘contents’ (see Figure5). Specifically, students were told:

If you find that a block just doesn’t work for you, you can experiment withchanging some of its pitches and/or durations to make it work better. But if youdo make changes, keep track of the changes in your log. In your paper, try to saywhat you didn’t like about the original block and how your changes improvedit.

Opening the Edit Window for Block 1, notice the two lists of numbers, onelabeled P, for pitches; the other labeled D, for durations. A good way to explorethe meaning of these numbers is to listen to what happens if you change them.(Bamberger, 2000: 30–1)


The analysis that follows focuses particularly on the features that emergeduring the evolution of the students’ work. The analysis is intended as anexample of ‘thick description’ in exploring phenomenologically dense andprovocative data. Thomas Kuhn, in his book, The Essential Tension (1977),comments on this direction of research in an essay titled, ‘The Function ofMeasurement in Modern Physical Science’:

. . . much qualitative research, both empirical and theoretical, is normallyprerequisite to fruitful quantification of a given research field. In the absence ofsuch prior work, the methodological directive, ‘go ye forth and measure,’ maywell prove only an invitation to waste time. (p. 213)

Following Kuhn, the mode of empirical qualitative research pursued here isintended to raise questions that are relevant to research in music cognitionand also to re-thinking curriculum, particularly in the elementary musicfundamentals classroom (see also, Bamberger, 1996).

1. The student papers

The first paper is by a student I call Linz. Linz was a 4th-year student with noformal music background, majoring in biology. The second paper is by a 1st-year student I call Keven, a computer science major. Keven played drums inthe school band, knew how to read drum notation, but ‘not notes’, i.e. he hadno experience of playing melodies or reading pitch notation.

Figure 6 shows the Ambrosian blocks, along with the graphic (pitchcontour) and notated versions of each.

The representations students actually saw were the icons for theAmbrosian tuneblocks along with pitch contour graphics for icons that theyhad placed in the PLAYROOM. So that students could refer to the blocks intheir papers, the blocks were given number names, 1–5, according to theorder in which they appear in the TUNEBLOCKS area. Staff notation has beenadded for the reader’s convenience.

The Impromptu screen for working with the Ambrosian blocks is shown inFigure 7.

The students’ papers are presented here with only minimal edits so that


F I G U R E 5 Edit Window for Ambrosian, Block 1.

readers can follow the evolution of the students’ reflections as their musicalintuitions gradually emerge. The students’ papers and the evolving tuneshave been divided into a series of developing ‘sketches’, using as boundary


F I G U R E 6 The Ambrosian blocks.

F I G U R E 7 Ambrosian TUNEBLOCKS screen: Block 1 in PLAYROOM.

Blocks Pitch contour Staff notation

Tuneblocks - AMBROSIANSpeed

PLAYROOM

CATALOG

GRAPHICS

TUNEBLOCKS

criteria moments when a student shifts direction or focus of attention, orwhen a particular problem becomes an extended source of development.Analytic comments are inserted between successive sketches.

LINZ’S PAPER

First impressions

Listening to each of the blocks once and then going through each of them asecond time, I noticed how Blocks 1 and 5 began with the same three notes –perhaps they could make a combination together. I also noticed that, of the fiveblocks, only Block 3 seemed to make a suitable ending. Therefore, I tentativelycalled Block 3 my ending block. Block 5 seemed to make a good beginning.

I noticed how all the blocks shared the same tempo – I mean, the durationbetween each of the notes was equal. This actually made the song seem verymonotonous and boring.

This feeling of monotony was strengthened after looking at the pitch contourwhere I saw that none of the blocks seemed to have any large jumps down or up.This gave the feeling that the tune sort of hovered around one note and theconstant stepwise movement left my ears wanting some excitement and actuallyneeding to hear jumps to widely spaced-apart pitches.

CommentsRight from the outset, Linz associates specific features of the blocks withcertain feelings – for example, equal durations with ‘monotonous’ and stepwisemovement with ‘wanting some excitement’. These initial associations guideLinz to emergent design criteria: variation in rhythm and in pitch contour aregoing to be necessary for a tune that she likes and that will make sense.

Linz tentatively assigns an ending function only to Block 3. From this wecan assume that she is hearing pitch C, with which Block 3 ends, as the moststable pitch, i.e. a quasi ‘tonic’. However, Linz later develops additionalcriteria necessary for generating a convincing ending for her melody (seeSketches 7–9).

Sketches 1 and 2: beginnings

I decided to start with the combination of Block 5 going to Block 1.

I liked the sound of Block 5 as a beginning because to my ears, the sense ofstarting something is best portrayed with a block which seems to go in differentdirections – up and down.


F I G U R E 8 Block 5 going to Block 1 (Sketch 1).

Sketch 1

However, I also noticed that Block 1 sounded like it wanted to go somewhere butwas stopped abruptly halfway there – a sort of question that needed an answer.To utilize this potential call and answer format, I placed Block 1 before Block 5. Irepeated Block 1 because the repetition seemed to give it more of a sense of ahalf-finished idea.

CommentsLinz shows an unusual ability to shift her focus among modes of attentionand this has a reciprocal effect: a perceived potential structural function,‘beginning’, leads to noticing particular features, ‘wide range’. In turn,specific kinds of features suggest potential structural functions: for instance,of Block 1 she says, ‘repetition . . . gives a sense of a half-finished idea’. Ascriteria for a ‘sensible tune’ begin to emerge, Linz critiques her initialdecisions: the ‘question’ features of Block 1 win over the ‘up and down’features of Block 5 for an effective beginning.

Sketch 3: first modifications

At this point I wanted to break up the monotony of the tempo, so I decided tomodify Block 1 so that the fifth note was held for the same amount of time as thefirst four notes combined.

Immediately, this changed the character of the piece and placed an emphasis onthe first note and fifth note of the block.The long holding of the note also added


F I G U R E 9 Question–answer (Sketch 2).

Sketch 2

F I G U R E 10 Editing Block 1.

to the anticipation I had of hearing something else. Now I felt there should besomething that answered the call of the repeated blocks. Block 5 was a very goodstart because it began the same as Block 1 but instead of stopping halfwaythrough, it continued forward and seemed to finally get somewhere.

CommentsRecall that Linz has no experience with music notation or with the specificmeaning of Impromptu numbers, yet she chooses to lengthen the last note ofBlock 1 by an amount proportional to the other notes in the block, i.e. by ‘thesame amount of time as the first four notes combined’. While her purposewas simply to break up ‘the monotony of the tempo’, the proportionalextension of the last note of Block 1 results in actually transforming themetric structure. An unfamiliar 5-beat metre becomes a familiar duple metrewhich she succinctly describes as: ‘an emphasis on the first note and fifthnote of the block’. Not surprisingly, the transformation to a familiar metre,‘immediately changed the character of the piece’.

Adding Block 5, attentive also to motion and more global structuralfunctions, Linz hears that the extended and elaborated Block 5 ‘continuedforward and seemed to finally get somewhere’.

Sketches 4 and 5: looking ahead

Next, I tried to find the continuation of the answer. I didn’t like the way Block 3sounded because it felt too much like the ending of the piece and I didn’t feel thatmy song could finish there because there had been no development yet.


F I G U R E 11 ‘An emphasis on the first and fifth note . . . Block 5 seemed to finally getsomewhere’ (Sketch 3).

F I G U R E 12 Duple metre.

Sketch 3

I decided to keep Block 2 after Block 5. Block 5 seemed to naturally divide intogroups of four notes with the strong beat being on the 1st and 5th notes, so tokeep with this trend, I modified Block 2 so that the final note would be the sameduration as the first four notes combined.

I was starting to get an idea of how I wanted the form of my piece to be. Block 1ais introduced as the start of something that we haven’t figured out yet. It getsrepeated again but it doesn’t really get any further. Finally, with the addition ofBlock 5, we get the movement of the piece into an actual idea. However, wethrow in a second block which doesn’t quite finish off the idea (Block 2a). If weplay Block 5 again, we can see that we have an antecedent consequent phrase thatneeds to be completed.

CommentsAlert now to the question of when a piece sounds finished, Linz reasons thatit is too soon to end because ‘there had been no development yet’. Pausing toreflect on the large design of her song, Linz shifts from narrative mode, whereshe represents her emerging song as if it were an unfolding story plot orperhaps a logical argument (‘something that we haven’t figured out yet . . .


Sketch 4

Sketch 5

F I G U R E 13 ‘There had been no development yet’.

F I G U R E 14 Modified Block 2 (Sketch 4).

F I G U R E 15 ‘An antecedent–consequent phrase that needs to be completed’ (Sketch 5).

Block 1a Block 1a Block 5 Block 2a

(To be completed)

the movement of the piece into an actual idea’) to the logic of musicalfunctions: ‘we have an antecedent–consequent phrase that needs to becompleted’.2

Sketch 6: a generative problem

The final thing to do was complete the antecedent consequent phrase using thefinal two blocks (Blocks 3 and 4). I still heard Block 3 as the only block I could useas an ending so I placed it at the end and put Block 4 before it. I didn’t like havingthe note that is shared between Blocks 4 and 3 being repeated because it was likea stop in the motion of the piece.

CommentsMoving from detail to larger design, Linz identifies a problem: the repeatednotes form a ‘stop in the motion of the piece’. Her effort to solve that problembecomes the generative force driving the whole series of modifications thatfollow.

Sketches 7, 8 and final tune: evolving solutions

I switched Blocks 2 and 4 (again modifying Block 4) but still had the problem ofthat same note being played three times.

I tried repeating Block 2 so that it would have more motion preceding therepeated note but this made the song seem boring. Next, I deleted the fifth noteof Block 2 (making a new Block, 2aa), and repeated it so that there would be a


F I G U R E 16 ‘A stop in the motion of the piece’ (Sketch 6).

Sketch 6

Sketch 7

F I G U R E 17 ‘Still the problem of the same note played three times’ (Sketch 7).


Block 5 Block 5 Block 4

constant upward and then downward stepwise progression of notes without anyrepetition in the middle.

I didn’t like how the notes (5 2aa 2aa 3) were played with exactly the sameduration. In order to keep the music going forward . . . I kept the repetition butchanged the block so that the first two notes get played twice and ‘twice’ as quick(Block 2b).

With 2aa 2b, I had the sense that just as I was about to get bored with therhythm, there was a sudden quickening of the tempo that pushes the songforward to the end. In addition, I extended the last note of Block 3 so that itwould make a more convincing ending.

Final tune

The single letter ‘a’ denotes where I changed the rhythm so that the duration ofthe last note was longer.


Sketch 8

F I G U R E 18 Repeated Block 2aa: ‘upward and downward progression . . . without repetition’(Sketch 8).

F I G U R E 19 ‘First two notes get played twice and “twice” as quick’ (Sketch 9).

F I G U R E 20 ‘Pushes song forward to the end’.


Contents of Block 2aa “...played twice as quick”

Analysis

The structural hierarchy: there is a brief introduction followed by the antecedentphrase and a consequent phrase. The consequent phrase is longer than theantecedent phrase and made up of more blocks.

Final commentsBetween Sketch 6 and the Final Song, Linz goes through a series oftransforming modifications, all of them directed towards solving the problemshe identified on listening to Sketch 6, thus keeping the motion of the piecegoing forward to the end.

It is notable that Linz has not deviated from her initial identification ofBlock 3 as the only one with which her song could end. The thrust of this


F I G U R E 21 Final song.

F I G U R E 22 Linz’s tree diagram.

final series of modifications is her search for a way to make that ending aconvincing one – a convincing outcome of what she has already made.

As Schoenberg (1975) comments:

Even in the relatively simple forms, those most nearly related to thefundamental tones . . . tonality does not appear automatically, of itself, butrequires the application of a number of artistic means to achieve its endunequivocally and convincingly. (p. 274)

Linz’s problem-solving is particularly characterized by her quick shiftsfrom one mode and one kind of representation to another, resulting in agradually evolving and expanding ‘repertoire of possibles’ with respect tocreating coherence. It is in this process that Linz makes tangible and explicither emergent intuitive criteria for ‘a tune that makes sense’.

Table 1 summarizes the series of modifications in Sketches 6–8. The tableincludes the identified problem, the actions Linz takes, and the purpose ofthese actions towards a solution of the problem she has set.

The emergent features of Linz’s final tune include the following basiccharacteristics of tonal melodies:

● clearly articulated and (mostly) balanced phrases;● consistent (duple) metre;● resolution to a tonic cadence;● antecedent/consequent phrases.


TA B L E 1 Linz’s final design solutions

Problem Action Purpose of action

Sketch 6Repeated notes: ‘like a stop Switch Blocks 2 and 4 To reduce 4 repetitionsin the motion of the piece’ to 3

Sketch 7Still 3 repeated notes Repeat Block 2 To make more motion

before repeated notesToo boring Delete 5th note of Block 2 To make constant

(new Block 2aa) and repeat upward and downwardstepwise progressionwith no repetition

Sketch 8Each note is played with ‘First two notes played twice To keep the musicsame duration and “twice” as quick’ (new moving forward.

block (2b). Repeat it.

Sketch 9Unexciting Block 2aa followed by To ‘push song forward to

Block 2b the end’.

Beyond these basics, her work also includes attention to larger scalerelationships:

● detail as means towards larger design;● goal-directed motion;● motivic development;● rhythmic contrast.

Linz’s effective use of multiple means of generating coherence andmultiple modes of representation calls to mind Marvin Minsky’s (1986)comments on the importance of multiple representations:

A thing with just one meaning has scarcely any meaning at all. That’s why it’salmost always wrong to seek the ‘real meaning’ of anything.

Rich meaning-networks, however, give you many different ways to go: if youcan’t solve a problem one way, you can try another. True, too manyindiscriminate connections will turn your mind to mush. But well-connectedmeaning structures let you turn ideas around in your mind, to consideralternatives and envision things from many perspectives until you find one thatworks. And that’s what we mean by thinking! (p. 64)

KEVEN’S PAPER

First impressions

1(a) The process

OK, so when these blocks were described as weird, you weren’t joking.Ambrosian was a lot harder to make sense of, and I was thankful that I couldmodify the blocks.

One of the first things I noticed was that making balanced sections is going to bedifficult. I also realized I hadn’t found any sections that sounded like a goodending.The note that I felt should be the tonic was not found at the end of any ofthe blocks.

Block 4 felt a bit like an ending, so I decided to work with it to find out why. Idiscovered that the 3rd note was actually what I thought the tonic should be, butinstead of coming back to it, it stayed up a step, which sounded horrible. Based onthis, I modified Block 4 to make Block 6, which returned to the tonic. Here’s myBlock 6:

P: E D C D C

D: 4 4 4 4 4


F I G U R E 23 Block 6 returns to the tonic.

I also observed that there was no rhythmic variation whatsoever. I suspected thatthis would make separation of sections difficult.

CommentsUnlike Linz, Keven uses ‘measure’, ‘group’ and ‘section’ interchangeably torefer to a phrase, but never actually uses the term ‘phrase’. Further, Kevenmakes ‘balance’ a priority right from the outset. Similar to antecedent–consequent, ‘balance’ was defined and exemplified in the text and in therecorded compositions in the previous project.

Even before beginning to compose, Keven notices that the given materialsare going to be problematic. For example, he comments, first, that to make‘balanced sections’ is going to be a problem since the given blocks differ innumber of beats; second, he fails to find a ‘section’ that sounds ‘like a goodending’; and third, the need for ‘separation of sections’ is a problem becausethere is ‘no rhythmic variation’.

The issue of finding a good ending block presents an interesting musicalpuzzle. Notice that Keven does hear a possible tonic (C) in the middle of Block4. To use this found tonic, Keven modifies Block 4 to return to the C, thuscreating an ending block that he finds satisfactory. But why, then, does hereject Block 3 since it, too, ends with the designated tonic, C?

This seems a clear example of the influence of situation or context.Consider, for example, the difference in situation generated, even on such asmall scale, by Blocks 3 and 4:

Within Block 3, the C is preceded by a falling perfect 4th (D4–A3), directlyapproached by a rising minor third (A3–C4), and with no leading tone. Byconvention, this context only weakly generates C as a tonic. In contrast, the Cin the middle of Block 4, which Keven does hear as a tonic, is approachedstepwise from above, E–D–C, an archetypal tonic-generating gesture – i.e. byconvention heard and labeled as scale degrees 3–2–1.

Further, Keven comments that Block 4 ‘instead of coming back to [the C] . . . stayed up a step, which sounded horrible’. Keven’s strong response could beaccounted for as his intuitive hearing of an unfulfilled implicative relationship:

An implicative relationship is one in which an event – be it a motive, a phrase, andso on – is patterned in such a way that reasonable inferences can be made bothabout its connections with the preceding events and about how the event itselfmight be continued and perhaps reach closure and stability. (Meyer, 1973: 110)


F I G U R E 24 Situation and function.

Block 3 Block 4

Following this view, Keven’s modification of Block 4, the return to C,satisfies the implication of the previous gesture – E–D–C – and also satisfieshis tacit criterion for an ending block. Moreover, Keven’s hearings of Blocks 3and 4, which initially seemed inconsistent with one another, now become notonly reasonable but also evidence for his strategic know-how – what Schön(1983) has called ‘knowing-in-action’.

Sketches 1–3

I decided to start with Blocks 5 and 2 because there was similarity between them.Each had an ‘arched’ section that went up two notes then down two notes. Since5 had its arch at the end and 2 was only the arch, I put 5 first.This puts the archescloser together and made the sequence more obvious.

However, this arrangement felt very unresolved, so I put my newly created Block6 ending after the 5, and it felt like a nice closing. However, I still wanted to usethe sequence I first created, so I put them together: 5–2 5–6.

This sounded ok at first, but the second time I listened to it, I realized I didn’t likethe way the two parts ran together. As I had anticipated, there was no separationbetween the antecedent phrase and the consequent phrase. I fixed this by makingBlock 7, which was just Block 2 modified so that the last note was a half noteinstead of a quarter note.

To keep things balanced, I modified Block 6 so it also had a half note at the end.


F I G U R E 25 ‘Each had an arched form’ (Sketch 1).

Sketch 1

Sketch 2

F I G U R E 26 ‘A nice closing’ (Sketch 2).

F I G U R E 27 Block 7, ‘half note instead of quarter note’.

Comments on Sketches 1–3Both Linz and Keven use ‘similarity’ as a basis for coupling blocks at thebeginning of their tunes. However, their similarity criteria are significantlydifferent: Linz is hearing a similarity (‘Blocks 1 and 5 began with the samethree notes’), while Keven is most likely seeing a visual similarity (‘Each hadan “arched” section’). Indeed, the functional relationships among the pitches ofthe ‘arches’ in Blocks 5 and 2, including intervals, accents and impliedharmonic functions, suggest that Keven’s attention to visual appearanceleads him to spuriously label the ‘arched shapes’ a ‘sequence’. (For an incisivediscussion of the prevalent mismatch between visually seen transformationsversus musically heard transformations among beginning music students, seeNarmour, 2000: 376–83.)

In Sketch 3, Keven encounters and resolves an anticipated designconstraint – a ‘separation’ problem. Extending the last note of Block 2, hesolves the ‘separation’ problem, and extending the last note of Block 6 hesatisfies the priority he has put on ‘balance’, as well.

At this point I wondered whether there would be a simple way to modify theother blocks so that the balancing would be easier. The first half of my piece (partA) had 14 beats per group. I would like to try maintaining the 14 beat groupingthroughout. I would likely have to leave Block 5 out of part B because itdominated part A.That left me with only 5 beat blocks. I’d also have to have anextended block like 6 or 7 to gain separation. That left me with 8 beats to fill. Ithought about trying to stretch one of the blocks, but decided to just chop 2notes off of one of the blocks and combine it with a 5-beat block, instead.

Playing around, I discovered that Block 2 could make a good beginning, too, so Iused it with this different function. Block 1 sounded good after it, except for thelast bit that sounded too much like the end of Block 2 again. Since I was lookingfor a block to cut anyway, I created Block 8 by cutting the last 2 notes out ofBlock 1.


Sketch 3

Sketch 4

F I G U R E 28 ‘To keep things balanced’ (Sketch 3).

Antecedent Consequent

Block 5 Block 7 Block 5 Block 6

F I G U R E 29 Similar visual form, but a sequence?

Next I needed a way to end this 14-beat ‘measure’. I hadBlocks 2 and 8, so I was looking for a 6-beat block to fill itout. I didn’t like 6 so I went with 7 again.

Comments on Sketch 4Still focused on ‘balancing’, and once again looking ahead, Keven proposesthree design constraints specific to the current situation before going on:

1. maintain the 14-beat grouping;2. leave out block 5 – it dominated Part A;3. have an extended block to gain separation.

Having made a plan, Keven feels free to begin ‘playing around’ – i.e. hereturns to experimenting, listening and working by ear. Block 2 can serve asboth a beginning as well as an ending; Block 1 conveniently fits his plan for aneeded 3-beat block – a perfect candidate to ‘chop 2 notes off of ’; adding abeat to Block 7 works to meet his primary constraint – he has another 14-beat phrase.

Evolving solutions and the final tune

I was then looking for a way to end the piece. Repeating the previous measurewith antecedent–consequent sounded like it could work. It was a little odd having2 right after 7 since they are essentially the same measure, but since they wereserving different functions, it was OK.


F I G U R E 30 Block 8.

F I G U R E 31 ‘To end this 14-beat measure’ (Sketch 4).

Sketch 4

F I G U R E 32 ‘Antecedent–consequent’?

Antecedent Consequent

Block 5 Block 7 Block 5 Block 6

I still had Block 3 yet, which I didn’t really like the sound of at all. While it didend on the tonic, I didn’t like the way it repeated it twice. It sounded like itwould be better if the 4th note was up a little so it could come down to the tonic.However, when I tried it, it didn’t sound as good as it did when I sang it tomyself. I discovered that I had subconsciously raised the 3rd note as well, andwith that modification, it finally started to sound like something! Moving thesecond note up as well made it a little bit better, too. In addition to moving the 3middle notes up one step, I also made the last note twice as long, as I had withthe other 6-beat blocks. Thus, my final Block 9 was:

P: D E B D C

D: 4 4 4 4 8

1(b) Making an accounting


F I G U R E 33 Block 3 transforms into Block 9.

F I G U R E 34 Final tune.

A A·

A B

F I G U R E 35 Keven’s tree chart.

Much of my motivation was derived from concern for balanced sections. Irecognized initially that I was going to have to modify the length of some blocks inorder to make a coherent piece. I would also need to modify the length of somenotes in order to break the monotony of straight quarter notes. Looking over thestructure of the piece, familiar patterns are visible. AA’ and BB’ both formantecedent–consequent pairs.There is a lot of repetition of the motif in Block 2/7which helps tie the whole piece together.There is a sequence with Block 7 andthe end of Block 5.

Final comments After proposing another ‘antecedent–consequent’, Keven leaves behind hisanticipatory calculating and takes off to explore in search of a block tofunction as an ending. But there is a surprising change here. Recall that whileKeven initially heard a tonic (C) in the middle of Block 4 and modified theblock so as to end it on that tonic, he did not hear the C with which Block 3ends as also a tonic. Now, after working with the blocks, listening to them innew situations, but apparently without noticing the change, Block 3 becomesa possible ending block, as it was for Linz from the beginning. However, whileKeven is satisfied that Block 3 ends with the tonic, that is not sufficient initself to make an acceptable ending. And once more, as with Linz, it isrepetition that is a problem: ‘I didn’t like the way it repeated it [the tonic]twice’.

The need for closure triggers a whole series of modifications, but withquite different strategies and quite different results as compared with Linz. Insearch of a satisfying close, Keven entirely abandons his pre-planning lists ofconstraints. Improvising, singing to himself, ‘subconsciously’ experimenting,he tests and reflects on the results. Through this process, one-by-one he‘pushes’ all but the first and last pitches of Block 3 up one step, with each‘push’ a response to the newly created implications for continuation of theprevious change.


F I G U R E 36 ‘Moves the 4th note up to come down to the tonic’.

F I G U R E 37 ‘Finally started to sound like something!’.

Of particular significance is Keven’s response to the modification thatbrings in the missing leading tone: ‘it finally started to sound like something!’(Figure 37). This response is evidence that Keven clearly recognizes he hasstumbled upon something useful to his quest. The whole process seems aremarkable example of a series of intuitively guided choices that result in thetransformation of a modal motive that Keven heard initially as not evenending with a tonic, into a typical tonal cadential figure. Indeed, Keven hascreated a motive that is close to one of Rosner and Meyer’s (1982) archetypes– a ‘changing note melody’.

A changing note melody is one in which the main structural tones of thepattern consist of the tonic (1), the seventh or leading tone of the scale (7), thesecond degree of the scale (2), and then the tonic again. (Rosner and Meyer,1982: 325)

To complete this series of improvised modifications, Keven returns to hisby now familiar turn – assuring balance: making ‘the last note twice as long’,the 5-beat block becomes a 6-beat block, and he has 14 beats in all. Thesemodifications could be seen as similar in goal to the single modification Kevenmade to Block 4, but in reverse. That is, in modifying Block 4, Keven waslooking for an appropriate continuation for what was already implied, i.e. areturn to the implied tonic. Now the problem Keven solves is to build upimplication so that the final closure is a satisfying one.

Reflecting back on his process and his final tune, Keven searches for themeans he has found useful to the task of making a ‘coherent piece’. Keven’sexpressed criteria, together with his improvised modifications in achievingthem, make clear that his knowledge-in-action includes an intuitive feel formaking the pitch relations that convincingly create the particular coherencehe is seeking.

While Keven’s strategies and procedures differ from those of Linz,primarily in the degree to which Keven tends to plan ahead in making hisdesign constraints, their final tunes share many of the basic features thatcharacterize tonal melodies. The following are the primary emergent featuresof Keven’s tune:

● balanced phrases;● clearly articulated phrase boundaries;● resolution to a tonic cadence;● hierarchical structure (motive, phrase, section);


F I G U R E 38 ‘Second note up made it a little bit better, too’.

● clearly defined structural functions including:● antecedent–consequent phrase relationships● realization of implications for continuation● development (motivic repetition and variation)● motion towards closure and an archtypical tonal cadence.

As Schön (1983) observes:When we go about the spontaneous, intuitive performance of the actions ofeveryday life, we show ourselves to be knowledgeable in a special way. Often wecannot say what it is that we know. When we try to describe it we find ourselvesat a loss, or we produce descriptions that are obviously inappropriate. Ourknowing is ordinarily tacit, implicit in our patterns of action and in our feel forthe stuff with which we are dealing. It seems right to say that our knowledge isin our actions. (pp. 49–50)

3. Conclusions

SUMMARY OF IMPLICIT AND EXPLICIT CRITERIA FOR A ‘SENSIBLE TUNE’Results of the two close case studies indicate that, as anticipated in theIntroduction, both students were able to shape tonally and metricallyambiguous melodic materials so as to produce coherently structured tonalmelodies. Further, the students were able to develop, to some extent, explicitcriteria for their decision-making. However, despite the following similaritiesin features and relations embodied by the two students’ melodies, theirstrategies and their priorities clearly differ.

Balanced phrasesKeven gives precedence to balanced phrases, making it an explicit constraintright from the outset. Linz does not explicitly state ‘balance’ as a desiredfeature, but she implicitly does so by successfully making each of her innerphrases the same (8 beats) in total time.

Articulation of phrase boundariesAgain, Keven is explicit about the articulation of phrase boundaries when henotices that lack of rhythmic variation will make the ‘separation of sections’difficult. Linz is not as explicit but, also bothered by the unvaried rhythm, sheproportionally lengthens the last notes of all the 5-beat blocks and in doingso also clearly articulates their boundaries.

MetreAs for metre, Linz’s proportional lengthening of phrases results inunambiguously generating duple metre. Keven does not explicitly speak ofaccents or of metric considerations, but being insistent on balanced phrases,he does, in this sense, make a 14-beat metric. The melody may also be heardin 2/4, in which case there are 4 phrases, all 7 measures long and grouped


(irregularly) as 4+3 bars (14 beats). I find the latter proposal less satisfactoryespecially because, given the other features, the 7-bar phrases feel somewhat‘tipsy’.

TONALITY AND STRUCTURAL FUNCTIONS

With respect to tonality, both students leave no doubt that they are able tohear, to make, and appropriately to use a tonal center along with otherstructural functions in relation to it. However, once again the students differwith respect to how they go about satisfying these criteria and to the specificfeatures that they accept as meeting their demands.

Most powerful with regard to context, function and higher-level melodicgrouping structure is the quest by both students in the ending phase for asense of progressive movement towards a stable goal. For Linz, her aim to defineand solve this problem is explicit. Through a series of cumulating, primarilyrhythmic modifications, the block chosen for her ending (Block 3), despite itsweakly defined tonic, successfully functions to achieve a stable resolution.Keven, in contrast, does not initially hear Block 3 ending on the tonic at all.However, once he does, he focuses on his dissatisfaction with the pitchrelations within the block. Incrementally changing one pitch at a time, herecognizes the power of the leading tone when he hears it and, by using it,creates an archetypal tonal cadence.

It is important to emphasize that the characteristics I attribute to thestudents’ tunes are interpretations, made after-the-fact and after-the-acts, andonly then couched in music-theoretic terms. The students’ own criteria wereemergent, evolving primarily as actions or reactions in the process ofdesigning, improvising and building their melodies.

It is interesting in retrospect to compare evidence from the students’ work,particularly in the last series of modifications, with the formal ‘probe-tone’experiments of Krumhansl and others. In contrast to the predesigned butoften rather musically impoverished context-creating stimuli with whichthese formal experiments begin, students in the informal compositionsituation demonstrate their perception of tonality as a structural functionwithin self-generated contexts. In particular, the perception of tonality isembedded in efforts to satisfy situated structural implications – a feel for thetension of moving forward towards the stability of arrival.

Hasty’s (1997) remarks in relation to the importance of situation as afunction of musical process effectively captures this sense of evolution andemergence. He says ‘a piece of music or any of its parts . . . while it is goingon, is open, indeterminate, and in the process of becoming a piece of music ora part of that piece (p. 3).

Evidence from analysis in this natural experiment also helps to account forthe differences in results as compared with those of subjects in the Deliège etal. (1996) experiment who ‘seemingly demonstrate that nonmusician subjectspossessed little capacity to produce coherent tonal structures’ (p. 144ff). As


argued earlier, the evidence now makes it clear that if musically untrainedstudents are given time, an environment that encourages reflection and theopportunity to evolve critiera as they ‘play with’ given material, they areindeed able to produce coherent melodic (at least) tonal structures.

EDUCATIONAL IMPLICATIONS

If a general pedagogical approach emerges from this study, it rests on thefinding that the basic characteristics of tonal structure are already part ofmusically untrained students’ intuitive knowledge-in-action. Thus, acurriculum for elementary music fundamentals classes should recognize, buildon and help students develop these intuitions in at least the following ways:

● first, give students ‘units of work’ that are consistent with their intuitive‘units of perception’ – aggregated, structurally meaningful entities such asmotives, figures and phrases;

● second, provide a working environment such that materials are easilymanipulated at mutiple levels of structure – for instance, at the aggregatemotive level, and also easily modified at the more detailed level of theirpitch and duration ‘contents’;

● third, encourage compositional, action-based projects that necessarilydirect students’ attention to context and within contexts to structuralfunctions;

● fourth, give students easy access to a variety of representations thatinclude: multiple sensory modalities, multiple graphics and multiplelevels of musical structure;

● fifth, encourage students to invoke strategies that will help make theirintuitive knowledge explicit, e.g. listening critically, designing, improvising/experimenting and reflecting on decision-making criteria, along withtrying to account for results.

The advocated approach is noticeably different from that assumed in moreconventional music fundamentals texts. These differences are well describedby Granados (2001) in the distinction he makes between ‘problem space’ and‘design space’ in relation to educational strategies more generally (Granados,2001: 504–5).

‘Problem space’ (as Granados uses it) best characterizes exercises at thebeginning of traditional music fundamentals classes where there is a ‘well-defined problem’ and an unambiguous solution. ‘Design space’ describes aprocess of defining and re-defining problems as an inherent part of ongoingwork. Instead of being given a priori names for elements, and specificstrategies for finding problem solutions, students progressively notice newelements as these emerge with each new modification (for more on thisapproach, see Bamberger, 1991/5).

Later, when students are introduced to conventional notations andtheoretical units of analysis, these traditional basics serve as a source of


answers to questions that students have put to themselves in their previousreflective conversations back and forth with their materials. Thefundamentals thus become a necessary framework within which studentsmore fully describe and account for their own initially tacit and intuitiveperceptions of musical coherence. Going forward from here, the foundationhas now been laid for students to learn to hear and appreciate more complex,less immediately accessible compositions as their abilities for inquiry andacquisitiveness grow and deepen. Rather than giving up their intuitions,students are learning how to understand them better and then build on them.

N O T E S

1. The two students featured in this article were chosen because their papers weremore complete and more clearly written than some others, not because thecontent was particularly exceptional.

2. Linz has taken the term, ‘antecedent–consequent phrase’, from the previous tunebuilding project where it was defined and examples given (Bamberger, 2000: 25).Linz recognizes a potential instance of the type and with it the possibility ofactually making one.

R E F E R E N C E S

Auhagen, W. and Vos, P.G. (2000) ‘Experimental Methods in Tonality InductionResearch: A Review’, Music Perception 17: 417–34.

Bamberger, J. (1991/5) The Mind behind the Musical Ear: How Children Develop MusicalIntelligence. Cambridge, MA: Harvard University Press.

Bamberger, J. (1996) ‘Turning Music Theory on Its Ear: Do We Hear What We See: DoWe See What We Say?’, International Journal of Computers and Mathematics Learning1(1): 48–74.

Bamberger, J. (2000) Developing Musical Intuitions: A Project-Based Introduction toMaking and Understanding Music. New York: Oxford University Press.

Clarke, E.F. and Krumhansl, C.L. (1990) ‘Perceiving Musical Time’, Music Perception 7:213–51.

Deliège, I., Melen, M., Stammers, D. and Cross, I. (1996) ‘Musical Schemata in Real-Time Listening’, Music Perception 14: 117–60.

Granados, R. (2001) ‘Constructing Intersubjectivity in Representational DesignActivities’, Journal of Mathematical Behavior 19: 503–30.

Hasty, C.F. (1997) Meter as Rhythm. New York: Oxford University Press.Krumhansl, C. (1990) Cognitive Foundations of Musical Pitch. New York: Oxford

University Press.Krumhansl, C. and Kessler, E. (1982) ‘Tracing the Dynamic Changes in Perceived

Tonal Organization in a Spatial Representation of Musical Keys’, PsychologicalReview 89: 334–68.

Kuhn, T.S. (1977) ‘The Function of Measurement in Modern Physical Science’, in TheEssential Tension. Chicago: University of Chicago Press.

Leman, M. (2000) ‘An Auditory Model of the Role of Short-Term Memory in Probe-Tone Ratings’, Music Perception 17: 481–509.

Meyer, L.B. (1973) Explaining Music. Berkeley: University of California Press.


Minsky, M. (1986) The Society of Mind. New York: Simon and Schuster. Narmour, E. (2000) ‘Music Expectation by Cognitive Rule-Mapping’, Music Perception

17: 329–98Rosner, B.S. and Meyer, L.B. (1982) ‘Melodic Processes and the Perception of Music’,

in D. Deutch (ed.) The Psychology of Music. New York: Academic Press. Schoenberg, A. (1975) Style and Idea. New York: St Martins Press.Schön, D.A. (1983) The Reflective Practitioner: How Professionals Think in Action. New

York: Basic Books.Vygotsky, L.S. (1978) Mind in Society: The Development of Higher Psychological

Processes, ed. Michael Cole, Vera John Steiner, Sylvia Scribner and EllenSouberman. Cambridge, MA: Harvard University Press.

J E A N N E B A M B E RG E R is Professor of Music at the Massachusetts Institute ofTechnology where she teaches music theory and musical development. Her researchis interdisciplinary, focusing on perceptual change through the observation andanalysis of children and adults in moments of spontaneous learning situations. Shewas a student of Artur Schnabel and Roger Sessions, and performed extensively in theUS and Europe as piano soloist and in chamber music ensembles. She attendedColumbia University and the University of California at Berkeley, receiving degrees inphilosophy and music theory. Professor Bamberger’s most recent books include TheMind behind the Musical Ear (Harvard University Press, 1995) and Developing MusicalIntuitions: A Project-Based Introduction to Making and Understanding Music (OxfordUniversity Press, 2000). Forthcoming publications include: ‘Changing MusicalPerception through Reflective Conversation’, in R. Horowitz (ed.) Talking Texts:Knowing the World through the Evolution of Talk about Text (International ReadingAssociation) and ‘Restructuring Conceptual Intuitions through Invented Notations:From Path-Making to Map-Making’, in S. Strauss (ed.) The Development of NotationalRepresentations (Oxford University Press).Address: Music and Theater Arts, Massachusetts Institute of Technology, Room 4-246, 77 Massachusetts Avenue, Cambridge, MA 01239-4307, USA. [email: [email protected]]


The role of self-efficacy in a musical performance examination: an exploratorystructural equation analysis

37A R T I C L E

Psychology of Music



Researchvol (): ‒

[- ():; ‒; ]

J O H N M C C O R M I C K a n d G A RY M C P H E R S O NU N I V E R S I T Y O F N E W S O U T H WA L E S

A B S T R A C T The study reported here investigated cognitive mediational processesin the context of a music performance examination. The prime purpose was tofocus on an aspect of musical learning – graded music examinations – that hashitherto received little research attention. The sample consisted of 332 instrumentalists who were completing Trinity College, London, graded, externally assessed performance examinations. Analysis of survey data was carried out using structural equation modelling. The analysis suggested that, inmotivational terms, and consistent with research carried out in other academiccontexts, self-efficacy was the best predictor of actual performance. The authorsconclude that whilst practice plays a vital part in the development of a musician’s capacity to perform well, it should not be considered in isolation frommotivational and related variables.

K E Y W O R D S : examinations, musical performance, practice, self-efficacy, strategy use

Introduction

Recent years have seen a considerable growth in the number of studiesaddressing the cognitive mediational processes which impact on positivemotivation in academic achievement contexts (Murphy and Alexander,2000). However, relatively little of this research has been validated in thedomain of music, especially musical performance (McPherson andMcCormick, 1999). This is remarkable on at least two counts: first, the funda-mental role played by achievement motivation, particularly in educationaldomains (Covington, 2000; Eliot and Church, 1997; Maehr and Meyer,1997); and second, the acknowledged high demands placed on musical per-formers who seek to achieve at a high level (Ericsson et al., 1993; Lehmannand Ericsson, 1997; Sloboda and Davidson, 1996).

sempre :

Given the strength of findings in academic research, two componentsinvolved in regulating learning seem especially important for music perform-ance. First, the cognitive strategies musicians employ to monitor and controltheir learning have been shown in academic subjects such as mathematicsand science to result in higher levels of cognitive engagement, which in turnlead to higher levels of achievement (Pintrich and De Groot, 1990). Resultsfrom a music investigation by McPherson and McCormick (1999) suggestthat these findings might also apply in music learning. In their study, pianostudents (aged between 9 and 18) who did greater amounts of practice weremore likely to rehearse music in their minds plus make critical ongoing judge-ments concerning the success or otherwise of their efforts. Students whoreported more time spent practising were more capable of organizing theirpractice in ways that provided for efficient learning; for example, practisingthe pieces that needed most work and isolating difficult sections of a piecethat needed further refinement. This result suggests that student musicianswho are more cognitively engaged while practising not only tend to do morepractice, but may also be more efficient with their learning. This is also con-sistent with other researchers (for example, Hallam, 1998; O’Neill, 1997;Williamon and Valentine, 2000) who have concluded that

. . . while the length of time learning to play an instrument and an estimate oftime spent practising are important predictors of learning outcome, they arenot the only factors in accounting for achievement and that a range of differentfactors may come into play when different aspects of achievement are consi-dered. (Hallam, 1998: 125)

In addition, the second author has identified a hierarchy of cognitivestrategies that students employ when performing music in various ways, andreported that these help predict the skills of performing rehearsed repertoire,sight-reading, playing from memory, playing by ear and improvising(McPherson, 1993, 1996, 1997). His results parallel findings of academicsubjects, and have implications for music pedagogy, in that better studentmusicians are likely to possess a sophisticated repertoire of strategies thatthey use when performing and practising. In contrast, weaker instrumental-ists are likely to display a naive understanding of how to transfer informationobtained from musical notation or aurally into an appropriate musicalresponse (McPherson, 1993, 1996, 1997). This is further illustrated inHallam’s (2001a, 2001b) studies with young instrumentalists. She con-cludes that learners’ use of effective strategies when practising and perform-ing depends on how successfully they have acquired the range of auralschemas that allow them to monitor and control their performance. In herview, efficiency of a strategy increases with expertise and depends on a mix ofmetacognitive and domain knowledge upon which students draw to monitorand evaluate their playing.

Extending these findings, Williamon and Valentine (2000) have reportedthat the relationship between the amount of practice and practice efficacy


may be more subtle than previously thought, and have argued that both thequality and quantity of a musician’s practice need to be examined wheninvestigating the determinants of musical skills. This is especially importantgiven Hallam’s (1998) conclusion that even with purposeful practice, stu-dents will vary on a variety of dimensions, including how successfully theyare able to concentrate while practising, adopt appropriate improvementstrategies, and understand the nature of what they are learning. These subtlecognitive processes impact on both the quality and quantity of a musician’spractice. However, more work is needed to determine more precisely how theyoccur, and under what conditions they may be taught.

A second important element concerned with regulating one’s learningwhile practising a musical instrument includes self-regulation, employedwhen students decide to manage their own learning by blocking out distrac-tions or making a conscious effort to practise. In academic subjects such asmathematics and science, self-regulatory processes are thought to stimulatecognitive engagement and help students perform better (Pintrich and DeGroot, 1990; Zimmerman, 1995).

Research in academic areas suggests, however, that ‘knowledge of cogni-tive and metacognitive strategies is usually not enough to promote studentachievement; students also must be motivated to use the strategies as well asregulate their cognition and effort’ (Pintrich and De Groot, 1990: 33).Practising merely for recreation does not guarantee increased skill, so with-out a real desire to learn, no student can expect to improve, let alone main-tain whatever level she or he has already attained (Lawler, 1977). Successfulmusicians of all types are characterized by a desire to improve their perform-ance and increase their skills, and there are many parallels in the rigour andfocus of self-taught rock and classically trained student musicians. Successfulmusicians in both groups not only invest large amounts of time practisingbut possess a deep desire to master their craft (Ericsson et al., 1993; Shehan-Campbell, 1995; Walser, 1993). Consequently, our theoretical frameworkincludes motivational components such as intrinsic value, defined as thelearner’s ‘beliefs about the importance and interest of the task’ (Pintrich andDe Groot, 1990: 34) and anxiety, which deals with the level of anxiety a stu-dent experiences when performing in public or for one of the graded externalperformance examinations which are typical in music education.

The concept of self-efficacy, which originated in the work of Bandura(1977), is also of critical importance, because it is particularly salient in specific performance activities (Bandura, 1997; Zimmerman, 2000). Self-efficacy is defined as ‘the conviction that one can successfully execute thebehavior required to produce the outcomes’ (Bandura, 1977: 79). In this sense,self-efficacy may seem closely tied to theories of self-concept and self-compe-tence in that it does include personal judgements of ability (Pintrich andSchunk, 1996). Where it differs, however, is that self-efficacy also includes being able to organize and execute the actions or skills necessary to demonstrate

McCormick and McPherson: Self-efficacy in a musical performance examination 39

competent performance. For example, self-efficacy for music performance notonly implies a self-recognition of being a good instrumentalist, but alsoexplicit judgements about the skills necessary to perform in front of others,such as in a music examination or concert. Another important distinction isthat self-efficacy judgements are made in relation to a specific type of per-formance (Pintrich and Schunk, 1996; Stipek, 1998). A trumpeter mightlower his or her efficacy judgements for playing a high note in a particularpiece because of a sore embouchure; a pianist may display lower self-efficacywhen faced with the challenge of learning a difficult piece in what she or hemight feel is too short a timeframe.

In studies of academic achievement, perceptions of personal competence‘act as determinants of behaviour by influencing the choices that individualsmake, the effort they expend, the perseverance they exert in the face of diffi-culties, and the thought patterns and emotional reactions they experience’(Pajares, 1996b: 325; see also Bong and Clark, 1999; Hackett, 1995).According to Pintrich and Schunk (1996), the optimum level for self-efficacyis to have slightly higher perceptions of efficacy than is justified by one’s actual real ability. Indeed, students with high self-efficacy in a particulardomain are more likely to choose more challenging tasks, exert more effort,persist longer and be less likely to experience debilitating anxiety (Bandura,1986; Pajares, 1996a; Zimmerman, 2000). In contrast, even accurate self-perceptions can result in ‘lower optimism and lower levels of self-efficacy’sprimary functions – effort, persistence, and perseverance’ (Pajares, 1996b:340).

Academic research shows that students tend to avoid tasks and situationsfor which they feel inadequate, and concentrate on tasks and activities withwhich they feel they can cope (Pintrich and Schunk, 1996). Although nospecific research has been found in music that focuses on self-efficacy (asdefined by Bandura and cited earlier), Hallam’s (1998) review of evidencedoes suggest that students who perceive themselves as musically inadequatetend to turn to other sporting and leisure activities and are less likely to con-tinue learning their instrument.

Given the context and how they are organized, externally graded perform-ance examinations such as those offered by Trinity College, London and theAssociated Board of the Royal Schools of Music offer a unique frameworkfrom which to examine young musicians’ self-efficacy perceptions. This isparticularly important given that these perceptions are likely to fluctuate,depending on the person’s physical condition and mood, plus external factorssuch as the nature of the task and the social milieu (Pintrich and Schunk,1996).

The study reported here investigated each of these cognitive mediationalprocesses in the context of a music performance examination, using structur-al equation modelling to explore the relationships among these aspects ofmotivation, music practice and performance. Our purpose was to focus


attention on an aspect of musical learning in which many hundreds of thou-sands of children worldwide participate each year, but which has receivedalmost no research attention; namely, graded music examinations.

Method

The sample consisted of 332 instrumentalists, between the ages of 9 and 18years (mean12.81; SD = 2.32), who were completing Trinity College, Londongraded, externally assessed performance examinations at 15 different region-al and metropolitan centres across three Australian states. Students who par-ticipated were learning to play either piano, or a string, brass or woodwindinstrument.

The Trinity College, London examinations, involve requirements that aresimilar to other examination systems such as the Australian MusicExaminations Board and the Associated Board of the Royal Schools of Music.Candidates are required to perform prepared pieces with piano accompani-ment, technical exercises, and études from a graded syllabus in front of atrained, professional examiner who provides a mark according to whetherthe performance was unsatisfactory (below 65), a pass (65–74), a merit(75–84) or a distinction (85 or above). Students undertaking these types ofexaminations can progress through the various grades from initial, grades 1to 8, and diplomas (Associate, Licentiate and Fellowship).

Although estimates of examiner reliability were not available for theTrinity College external performance examination, they were still consideredappropriate, based on the efficient manner in which these examinations areadministered and controlled, the rigorous system of accrediting examiners,and the fact that these types of awards are generally acknowledged by theprofession in the countries in which they are offered as an important indica-tion of a musician’s overall ability to perform music (see further, McPhersonet al., 1997).

In the months preceding the study, teachers who were preparing studentswere sent copies of a letter and asked to distribute them to all students (andtheir parents) who were undertaking all levels of the Trinity College examina-tions. The letter explained the aims of the study and provided examples of thetypes of questions to be asked. It also invited candidates to participate in thestudy by arriving at the examination centre earlier than required in order tocomplete a questionnaire immediately before entering their examination.Importantly, this procedure allowed for efficacy beliefs and actions to bemeasured in close temporal proximity to the students’ examination, based onevidence that the ‘closer in time, the better the test of causation’ (Bandura,1997: 67). Like other researchers (e.g. Bandura, 1997; Bong and Clark,1999) we felt that the most reliable indication of self-efficacy would comefrom a measure administered immediately before candidates entered theirexamination.


MEASURES

Approximately 65 percent of subjects who received the invitation completeda self-report questionnaire that included 16 items on self-regulatory learningcomponents (Cognitive Strategy Use, Self-Regulation) and motivational com-ponents (Intrinsic Value, Anxiety, Self-Efficacy) of instrumental learning.Subjects were instructed to respond to items using a 7-point Likert-type scale(1 = ‘not at all true of me’ to 7 = ‘very true of me’). Items were adapted froma research self-report questionnaire that has proved effective in studies ofacademic learning (Pintrich and De Groot, 1990) and earlier work by theresearchers (McPherson and McCormick, 1999, 2000).

The Cognitive Strategy Use items focused on rehearsing strategies (e.g. ‘If Ican’t play a piece I always stop to think about how it should go’), elaborationstrategies (e.g. ‘I’m always thinking about pieces I’m learning by singingthem through in my mind’ and ‘When I’m practising I often stop playing andthink about how the music should go’) and organizational strategies (e.g.‘The first thing I do when I practise is ask myself “What do I need to practisemost today?’”, ‘When I practise I always say to myself “I made a mistake, Imust try this section again”’ and ‘When I learn a new piece, I spend most ofmy time practising the most difficult sections’).

The Self-Regulation items were concerned with effort management (e.g. ‘IfI can’t play a piece, I leave it until the next lesson’, ‘I sometimes forget to domy practice’, and ‘I don’t like to learn hard pieces’). The Intrinsic Value itemswere concerned with the students’ interest in learning their instrument (e.g.‘Doing well on my instrument is important to me’ and ‘Playing my instru-ment is my favourite activity’). The Anxiety items were concerned with per-formance anxiety on the test each candidate was about to undertake (e.g. ‘Ihave an uneasy, upset feeling when I perform in front of people’ and ‘I’mscared I might freeze up when the examiner asks my scales’).

Self-Efficacy was assessed using the item ‘I have fully mastered the require-ments for today’s examination’. This item reflects how the candidate feltabout his or her capacity and skills to perform well in the examination. Toextend the measure of Self-Efficacy, candidates were also asked ‘How good amusician do you think you are, in comparison with other students of yourown age? Would you say that you are poor, below average, average, aboveaverage, or outstanding in comparison with other students of your sameage?’ and ‘What result do you think you will get for your exam today?’ For theformer question they indicated their response by ticking one of seven cate-gories: unsatisfactory (below 65), low pass (65–9), high pass (70–4), lowmerit (75–9), high merit (80–4), low distinction (86–90), or high distinction(91–100). These types of measures are consistent with self-efficacy measuresused in academic research, in which students are asked to rate themselves ona scale according to how confident they feel about their ability to complete atest or performance task (Pintrich and Schunk, 1996).

The questionnaire included a further 11 items which sought to obtain


information on how much practice had been undertaken during the monthleading up to the examination, and the frequency with which subjects prac-tised various activities on their instruments. Subjects were asked how manytimes a week they practised and for how long. An estimate of their PracticeTime was computed by multiplying the number of practice sessions per weekby the number of minutes averaged for each session. Subjects also used a 5-point scale (never, rarely, sometimes, often, every day) to indicate the frequen-cy with which they practised different types of activities during their homepractice, such as Informal Aspects of Practice (i.e. playing by ear for ownenjoyment, improvising own music), Formal Aspects of Practice (i.e. using awarm-up routine, practising scales/arpeggios, plus studies and études, andsight-reading music), and Repertoire (playing new unlearned pieces, playingolder familiar pieces).

At the end of each day of performance examinations, the secretary overseeing the Trinity College examinations collated each of the completedquestionnaires and then wrote on the top right-hand corner each student’s per-formance examination result. The grade level was represented by the grade(preliminary or grade 1 through 8) of the examination each student was sitting.

Analysis and discussion

Structural equation modelling was carried out using LISREL 8.3 (Joreskogand Sorbom, 1996) and the data were normalized using the procedureincluded in LISREL 8.3. This methodology was chosen for two principal reasons. First, structural models specify causal relationships among latentvariables and, although the data are cross-sectional in nature, the studentscompleted their questionnaires prior to the performance examination.Second, unlike multiple regression, LISREL estimates measurement errors.

Although the weighted least squares method, using the asymptotic covari-ance matrix estimated from the polychoric correlation matrix, is generallypreferred with ordinal data, this was not possible, given the modest samplesize. Consequently, maximum likelihood was employed with a covariancematrix as recommended by Joreskog and Sorbom (1996). Preliminary analy-sis consisted of confirmatory factor analyses. Anxiety (2 items), CognitiveStrategy Use (5 items), Self-Regulation (4 items), Self-Efficacy (3 items),Formal Practice (frequency: 4 items) and Informal Practice (frequency: 2items) were found to have adequate fit (GFI > .90) either as single factorswhen there were sufficient degrees of freedom, or in pairs. Fit statistics ofother conceptualized factors, Intrinsic Value and Repertoire, were unsatisfac-tory; consequently, the related items were not included in further analysis.Three variables, Practice Time, Result and Grade Level, were represented by single measures. Following Joreskog and Sorbom (1989), error variancewas set for these three variables, somewhat arbitrarily, at .15, which was con-sidered better than assuming no error term at all.


ANALYSIS 1A tentative initial model with Self-Efficacy as an endogenous variable wasposited. The first step in composing the initial model was to consider whichvariables could be conceptualized as directly related to self-efficacy and theexamination result, and which as intervening. It should be emphasized thatthis did not involve any notion of causality, given the cross-sectional natureof the data and the exploratory nature of the analysis. However, it can beargued that this model reflects temporal relationships, because the measureof self-efficacy was taken immediately prior to the examination, and the othervariables largely reflected past behaviours and practices before the students’formal examination.

The key feature of the initial model is that Anxiety, Practice Time, FormalPractice, Informal Practice and Grade Level do not have direct paths to theexamination result, but are hypothesized to have indirect effects (throughSelf-Efficacy, Grade Level and Cognitive Strategy Use) to Informal Practice,Practice Time and Formal Practice. The hypothesized relationships werebased on logical argument. For example, it was considered that time spentpractising could be predicted by Anxiety, Grade Level and Self-Regulation, butCognitive Strategy Use was more likely to be related to the form of practice,either formal or informal. So whilst the initial model is only one of many pos-sible models, its specification is based on substantive grounds (Hair et al.,1992; Joreskog and Sorbom, 1993).

Overall, the model was found to be approaching a reasonable fit of thedata (χ2 = 475.16, df = 211, p < .01; χ2/df = 2.25; standardized RMSR =.07; RMSEA = .06; GFI = .89; AGFI = .85) and is shown in Figure 1 withstandardized path coefficients. However, as the purpose of the analysis wasessentially ‘model generating’, the parameter estimates, the residuals and themodification indices were examined (Joreskog and Sorbom, 1993). Pathsfrom (1) Anxiety to Self-Efficacy and Practice Time, (2) Grade Level to FormalPractice and achievement, and (3) Practice Time to Performance were foundto be non-significant. It should be noted that there were no significanthypothesized paths from Anxiety to any other variable. All non-significantpaths were eliminated and there was a concomitant small improvement inthe fit (for example, GFI = .90). Perusal of modification indices for latent vari-able paths suggested a single additional direct path from Cognitive StrategyUse to Self-Efficacy. Logically, Cognitive Strategy Use could be predicted to beassociated with Self-Efficacy. Analysis was repeated with this path, resultingin a marginal improvement in the fit (χ2 = 337.46, df = 159, p < .01; χ2/df =2.17; standardized RMSR = .06; RMSEA = .06; GFI = .91; AGFI = .88). Thefinal model is shown in Figure 2.

A key feature of the final endogenous model is that Anxiety is no longerincluded. The sole direct path to Performance is from Self-Efficacy with a rela-tively high standardized coefficient of .68. Interestingly, two variables, GradeLevel and Cognitive Strategy Use, are negative predictors of Self-Efficacy. For


Grade Level, this relationship may reflect the increasingly demanding exami-nation requirements as students move from lower to higher grades, and perhaps, appreciate more the gap between their actual and desired gradelevel. This result is consistent with a Wigfield et al. (1997) study that showeda decline, over time, in children’s ability-related beliefs in a number ofsubjects, including instrumental music.

The paths from Grade Level through Practice Time, through Self-Efficacyto Performance, with positive coefficients, are also worth noting. This illus-trates the value of examining grade and amount of practice together since, asgrades increase, respective examinations are longer and more demandingand more practice is therefore required.

The negative relationship of Cognitive Strategy Use with Self-Efficacy maybe explained by the items used in the questionnaire which infer focusing onerrors and correcting them, e.g. ‘When I practise I always say to myself “I


F I G U R E 1 Self-efficacy endogenous standardized initial model.

F I G U R E 2 Self-efficacy endogenous standardized final model.

made a mistake, I must try this section again”’. However, in this model, thatnegative relationship is somewhat offset by indirect effects through formalpractice and informal practice. Each variable, except for Self-Regulation, has a direct path to Self-Efficacy. Hence, the most striking feature of thismodel is that Self-Efficacy is a strong mediator between other variables andPerformance.

ANALYSIS 2A competing model was posited on the basis of references in the literature toself-efficacy affecting choice of activities, effort, persistence and self-regula-tion (Zimmerman, 2000), all of which may be conceptualized as related to musical practice. Of course, this model does not reflect that the studentscompleted the questionnaires immediately prior to their performance examinations. However, although Bandura (1997) argued that temporalproximity of measurements was an important factor in establishing accuraterelationships between efficacy beliefs and actions, it may not be a necessaryrequirement (Bong and Clark, 1999). This proposed model with Self-Efficacyexogenous is shown in Figure 3. One Anxiety item had negative error covari-ance and the model could not be fitted to the data. The Anxiety variable andrelated paths were eliminated and the procedure repeated. Paths from GradeLevel to Cognitive Strategy Use and from Self-Regulation, Formal Practice,Informal Practice and Practice Time to Performance were non-significant,and eliminated, and the analysis repeated. Fit statistics for this model were: χ2 = 395.24, df = 160, p < .01; χ2/df = 2.47; standardized RMSR = .08;RMSEA = .07; GFI = .89; AGFI = .86. Again, as this was a model-generatingprocess, modification indices were inspected and a path added from Cognitive


F I G U R E 3 Self-efficacy exogenous initial model.

Strategy Use to Self-Regulation. There was a small improvement in model fit(e.g. GFI = .90), but the path from Cognitive Strategy Use to Performance wasnon-significant and was subsequently eliminated. Inspection of modificationindices suggested adding a path from Cognitive Strategy Use to InformalPractice. This resulted in yet another non-significant path from Self-Efficacyto Cognitive Strategy Use. This path was eliminated and the analysis repeated.The final model with Self-Efficacy as an exogenous variable is shown inFigure 4. It should be noted that the fit statistics are very close to those for theSelf-Efficacy endogenous model (χ2 = 333.05, df = 158, p < .01; χ2/df =2.11; standardized RMSR = .06; RMSEA = .06; GFI = .91; AGFI = .88).

The key features of the final self-efficacy exogenous model are not unex-pected, given the endogenous model. There is a strong direct effect from Self-Efficacy to Performance, with no direct effects from other variables on Performance. As was highlighted earlier, this is only one of a number of possiblemodels. Although the fit statistics of the two final models are very close, itwould seem logical to consider the self-efficacy endogenous model more appro-priate for these data. Notwithstanding, the literature has conceptualized self-efficacy both as a mediating variable between other cognitive variablesand performance, and as a variable which directly affects other cognitive andbehavioural variables (Zimmerman, 2000). Of course, direction of causalitycannot be established by these data. However, the competing models suggestthe desirability of these phenomena being studied with longitudinal data.

Conclusion

This study suggests that the relationship between self-efficacy and perform-ance quality appears to be consistent with other academic contexts (Bong


F I G U R E 4 Self-efficacy exogenous final model.

and Clark, 1999; Pajares, 1996b; Pintrich and De Groot, 1990; Zimmerman,2000). The study reported an exploration of relationships among variablesrelated to motivation, musical practice and performance. The principal resultis the strong association between self-efficacy and actual performance andthe former’s clear superiority as a predictor of actual performance in a gradedexternal music examination.

It is likely that the failure to fit the anxiety construct within both the mod-els is a reflection of the suitability of the items rather than the actual role ofanxiety in the performance examination for these young musicians. Theintention was to develop items specific to the performance examination con-text. In retrospect, the use of a well-established instrument, such asSpielberger et al.’s (1983) State–Trait Anxiety Inventory or Nagel et al.’s (1989)music-specific measure, would have been a better choice. Moreover, state andtrait forms of anxiety could be predicted to play quite different roles in terms ofpractice and performance. The inclusion of state and trait anxiety wouldrequire different initial models to those employed in the analysis reported here.

Whilst performance examinations have much in common with otherforms of academic examinations, they differ from many in an important way.In a highly charged music performance examination, the performer only hasone opportunity to perform at his or her best. There is no time to return andrevise an earlier decision, or to make more time available for one sub-task byquickly dispatching another. Moreover, performance is arguably the mostimportant image-forming component of an individual’s identity as a musi-cian. Consequently, it is perhaps not surprising that students’ perceptions ofself-efficacy should play a major role in how they perform. However, we stilldo not understand properly the mechanisms whereby students come tobelieve in their own abilities to perform well. Notwithstanding, this studyreinforces the view that whilst practice plays a vital part in the developmentof a musician’s capacity to perform well, it should not be considered in isola-tion from motivational and related variables (Hallam, 1998; O’Neill, 1997;Williamon and Valentine, 2000).

Further research with longitudinal data and refined instruments is desir-able to unravel the mediating relationships of self-efficacy with aspects ofinstrumental practice.

A C K N O W L E D G E M E N T S

The authors wish to acknowledge valuable advice from the editor and two anony-mous reviewers.

R E F E R E N C E S

Bandura, A. (1977) Social Learning Theory. Englewood Cliffs, NJ: Prentice-Hall.Bandura, A. (1986) Social Foundations of Thought and Action: A Social Cognitive Theory.

Englewood Cliffs, NJ: Prentice-Hall.


Bandura, A. (1997) Self-Efficacy: The Exercise of Control. New York: Freeman.Bong, M. and Clark, R.E. (1999) ‘Comparison between Self-Concept and Self-Efficacy

in Academic Motivation Research’, Educational Psychologist 34(3): 139–53.Covington, M.V. (2000) ‘Goal Theory, Motivation, and School Achievement: An

Integrative Review’, Annual Review of Psychology 51: 171–200.Eliot, A.J. and Church, M.A. (1997) ‘A Hierarchical Model of Approach and

Avoidance Achievement Motivation’, Journal of Personality and Social Psychology72(1): 218–32.

Ericsson, K.A., Krampe, R.T. and Tesch-Römer, C. (1993) ‘The Role of DeliberatePractice in the Acquisition of Expert Performance’, Psychological Review 100:363–406.

Hackett, G. (1995) ‘Self-Efficacy in Career Choice and Development’, in A. Bandura(ed.) Self-Efficacy in Changing Societies, pp. 232–58. New York: CambridgeUniversity Press.

Hair, J.F, Jr, Anderson, R.E., Tatham, R.L. and Black, W.C. (1992) Multivariate DataAnalysis with Readings, 3rd edn. New York: Macmillan.

Hallam, S. (1998) ‘The Predictors of Achievement and Dropout in InstrumentalTuition’, Psychology of Music 26(2): 116–32.

Hallam, S. (2001a) ‘The Development of Expertise in Young Musicians’ Strategy Use,Knowledge Acquisition and Individual Diversity’, Music Education Research 3(1):7–23.

Hallam, S. (2001b) ‘The Development of Metacognition in Musicians: Implicationsfor Education’, British Journal of Music Education 18(1): 27–39.

Joreskog, K.G. and Sorbom, D. (1989) LISREL 7: A Guide to the Program andApplications. Chicago: SPSS.

Joreskog, K. and Sorbom, D. (1993) LISREL 8: Structural Equation Modeling with theSIMPLIS Command Language. Chicago: SSI.

Joreskog, K. and Sorbom, D. (1996) LISREL 8: User’s Reference Guide. Chicago: SSI.Lawler, J.D. (1977) The Learning and Performance of Physical Skills. New York: Prentice-

Hall.Lehmann, A.C. and Ericsson, K.A. (1997) ‘Research on Expert Performance and

Deliberate Practice: Implications for the Education of Amateur Musicians andMusic Students’, Psychomusicology 16: 40–58.

Maehr, M.L. and Meyer, H.A. (1997) ‘Understanding Motivation and Schooling:Where We’ve Been, Where We Are, and Where We Need to Go’, EducationalPsychology Review 26: 371–409.

McPherson, G.E. (1993) ‘Factors and Abilities influencing the Development of Visual,Aural and Creative Performance Skills in Music and their EducationalImplications’, Dissertation Abstracts International, 54/04-A, 1277, University ofSydney (University Microfilms No. 9317278).

McPherson, G.E. (1996) ‘Five Aspects of Musical Performance and their Correlates’,Council for Research in Music Education 127: 115–21.

McPherson, G.E. (1997) ‘Cognitive Strategies and Skills Acquisition in MusicalPerformance’, Bulletin of the Council for Research in Music Education 133: 64–71.

McPherson, G.E., Bailey, M. and Sinclair, K. (1997) ‘Path Analysis of a Model toDescribe the Relationship among Five Types of Musical Performance’, Journal ofResearch in Music Education 45(1): 103–29.

McPherson, G.E. and McCormick, J. (1999) ‘Motivational and Self-RegulatedLearning Components of Musical Practice’, Bulletin of the Council for Research inMusic Education 141: 98–102.


McPherson, G.E. and McCormick, J. (2000) ‘The Contribution of Motivational Factorsto Instrumental Performance in a Music Examination’, Research Studies in MusicEducation 15: 31–9.

Murphy, P.K. and Alexander, P.A. (2000) ‘A Motivated Exploration of MotivationTerminology’, Contemporary Educational Psychology 25(1): 3–53.

Nagel, J.J., Himle, D.P. and Papsdorf, J.D. (1989) ‘Cognitive-Behavioral Treatment ofMusical Performance Anxiety’, Psychology of Music 17: 12–21.

O’Neill, S.A. (1997) ‘The Role of Practice in Children’s Early Musical PerformanceAchievement’, in H. Jorgensen and A.C. Lehman (eds) Does Practice Make Perfect?Current Theory and Research on Instrumental Music Practice. Oslo: Norgesmusikkhogskole.

Pajares, F. (1996a) ‘Self-Efficacy Beliefs in Academic Settings’, Review of EducationalResearch 66: 543–78.

Pajares, F. (1996b) ‘Self-Efficacy Beliefs and Mathematical Problem-Solving of GiftedStudents’, Contemporary Educational Psychology 21: 325–44.

Pintrich, P.R. and De Groot, E.V. (1990) ‘Motivational and Self-Regulated LearningComponents of Classroom Academic Performance’, Journal of EducationalPsychology 82(1): 33–40.

Pintrich, P.R. and Schunk, D.H. (1996) Motivation in Education: Theory, Research andApplications. Englewood Cliffs, NJ: Prentice-Hall.

Shehan-Campbell, P. (1995) ‘Of Garage Bands and Song-Getting: The MusicalDevelopment of Young Rock Musicians’, Research Studies in Music Education 4:12–20.

Sloboda, J. and Davidson, J. (1996) ‘The Young Performing Musician’, in I. Deliègeand J. Sloboda (eds) Musical Beginnings: Origins and Development of MusicalCompetence, pp. 171–90. New York: Oxford University Press.

Spielberger, C.D., Gorsuch, R.L., Lushene, R., Vagg, P.R. and Jacobs, G.A. (1983)STAIC: State-Trait Anxiety Inventory for Adults. Sampler Set, Manual Test, ScoringKey. Palo Alto, CA: Consulting Psychologists Press.

Stipek, D. (1998) Motivation to Learn, 3rd edn. Needham Heights, MA: Allyn and Bacon.Walser, R. (1993) Running with the Devil: Power, Gender and Madness in Heavy Metal.

Hanover, NY: Wesleyan University Press.Wigfield, A., Eccles, J.S., Yoon, K.S., Harold, R.D., Arbreton, A., Freedman-Doan, K.

and Blumenfeld, P.C. (1997) ‘Changes in Children’s Competence Beliefs andSubjective Task Values across the Elementary School Years: A Three Year Study’,Journal of Educational Psychology 89: 451–69.

Williamon, A. and Valentine, E. (2000) ‘Quantity and Quality of Musical Practice asPredictors of Performance Quality’, British Journal of Psychology 91: 353–76.

Zimmerman, B.J. (1995) ‘Self-Efficacy and Educational Development’, in A. Bandura(ed.) Self-Efficacy in Changing Societies, pp. 202–31. Cambridge: CambridgeUniversity Press.

Zimmerman, B.J. (2000) ‘Self-Efficacy: An Essential Motive to Learn’, ContemporaryEducational Psychology 25: 82–91.

J O H N M c C O R M I C K is a senior lecturer in educational administration in the School of Education at the University of New South Wales in Sydney, Australia. His broadarea of research interest is motivation in various settings. In addition to collaborativework with Gary McPherson, he has carried out research in the areas of stress, anxiety,leadership and organizational decision-making.


Address: School of Education, The University of New South Wales, Sydney, Australia2052. [email: [email protected]]

G A RY M c P H E R S O N completed his doctorate at the University of Sydney and is cur-rently an associate professor of music education at the University of New South Walesin Sydney, Australia. He has served on a variety of editorial boards for journals inmusic education and music psychology, including Psychology of Music, and is current-ly the editor of Research Studies in Music Education. His published research addressesvisual, aural and creative aspects of musical performance in young developing musicians, as well as self-regulatory and motivational issues that influence musicaldevelopment.Address: School of Music and Music Education, The University of New South Wales,Sydney, Australia 2052. [email: [email protected]]


An investigation of members’roles in wind quintets

53A R T I C L E

Psychology of Music



Researchvol (): ‒

[- ():; ‒; ]

L UA N F O R D a n d J A N E W. DAV I D S O NU N I V E R S I T Y O F S H E F F I E L D

A B S T R A C T This article presents new findings regarding group processes inchamber ensembles. Following work by Young and Colman, Butterworth, andMurnighan and Conlon on the string quartet, a questionnaire was distributed towind quintets throughout the UK. A total of 55 respondents, representing 20ensembles, participated in the study. Quantitative and qualitative data revealedinformation about group formation, personnel recruitment and the use ofdeputies, leadership, seating positions, verbal and non-verbal communication,conflict, rehearsing and performing, audience effects, social aspects and administration. The current study focuses on the role of the horn player in thequintet and explores players’ attitudes and perceptions of their individualcareers, quintet repertoire, concert promoters and string quartets as an opposingmedium. Results are discussed with reference to the existing quartet studies andto social psychological theory of group dynamics.

K E Y W O R D S : chamber music, group dynamics, small group behaviour, social interaction, wind ensembles

Introduction

RESEARCH PRECEDENTS

Applied psychological and sociological research into group dynamics, fordecades, has provided invaluable insights into teams in working environ-ments, investigating ways in which managers and group leaders can helptheir organizations to function most effectively (Young and Colman, 1979;see Allport, 1924; Bales and Bargatta, 1955; and Zajonc, 1965); Shaw,1971; Baumeister, 1984; Forsyth, 1990; Zander, 1994). Such research hasdirect relevance in the field of music and a number of studies have been car-ried out relating to the orchestral working environment and interpersonal

sempre :

dynamics involved in group function, including Faulkner (1973), Small(1987), Frederickson and Rooney (1993), Atik (1995), Allmendinger et al.(1996), and Levine and Levine (1996). The chamber music field has beenlargely neglected with only a handful of studies to date (Young and Colman,1979; Butterworth, 1990; Murnighan and Conlon, 1991; Davidson andGood, 2002), all of which focus on one medium: the string quartet. (Alverno[1987] also studied string quartets but in the context of the performing artsmedicine field rather than that of group dynamics. Therefore the authorshave not considered this research to be of direct relevance to the currentstudy.)

Young and Colman (1979) were the first to suggest that psychologicalresearch into group dynamics could have relevance for chamber musicensembles. However, their article simply summarizes some of the research upto that time and discusses the theoretical issues of leadership within ensem-ble conflict, and audience effects within a string quartet. While revolutionaryfor its time, the authors are aware of the speculative nature of their writingand point out the need for empirical research.

Butterworth (1990) was the first psychologist to undertake a case study ofa chamber group. Her work takes the form of 15 hours of observation as wellas questionnaires and interviews with each of the members of the DetroitString Quartet, a part-time ensemble drawn from a full-time professionalorchestra, the Detroit Symphony Orchestra. Butterworth provides a com-mentary on the rehearsals observed, followed by an analysis of leadershiproles, member interactions, group aspirations, personalities and conflict. Thestudy demonstrates that this quartet embodied many of the group processesdiscussed by Young and Colman (1979), although one of the drawbacks ofButterworth’s case study approach is that it only offers a one-sided view.

Blum (1986) and Rounds (1999) attempted to explore the group processesof quartets by examining biographical texts. Whilst some interesting ques-tions were raised, the second-hand nature of the biographical reportingmakes both these studies anecdotal. The authors are highly aware of the diffi-culties involved in trying to validate the biographical reporting.

Murnighan and Conlon (1991) undertook a detailed and empirically validlarge-scale study of 20 professional British string quartets, using semi-structured interviews (supported by archival material) and some observa-tions. With such a large sample they were able to provide a much more balanced view of how quartets work, focusing on the relationship betweeninternal dynamics and group success.

Davidson and Good’s (2002) study focused on a string quartet comprisingstudents who had been working together for just six months. Through videorecordings and semi-structured interviews, the researchers examined thesocial and musical co-ordination between the four players as they preparedfor their first recital. The ‘newness’ of the ensemble produced rather differentdata from the professional quartets studied by Butterworth, and Murnighan


and Conlon, but it provided new insights into the ways in which chambermusicians develop an awareness of their colleagues’ styles of playing, musi-cal gestures and communicative behaviours.

The work of Davidson and Good (2002), and Murnighan and Conlon isthe most recent and far-reaching to date in the field of group dynamics ofchamber ensembles, and along with Young and Colman, and Butterworth,they form the foundations for further investigation. The first author of thiscurrent investigation has professional experience of wind quintets and, inorder to pursue the small group processes of chamber ensembles in moredetail, the comparable medium of the wind quintet became the focus of thestudy.

WIND QUINTETS

The first wind quintets (flute, oboe, clarinet, horn, bassoon) were writtenaround 1800 and many composers wrote for the ensemble during theClassical period. It was a popular professional and amateur form of musicmaking. Although the popularity of chamber ensembles, in general – andstring quartets, in particular – waned in the second half of the 19th century,wind quintets continued to be written, often by leading performers whose in-depth knowledge of the instruments meant their works were very appealingfor players and audiences alike (Stoneham et al., 1997), and the 20th centu-ry saw a great surge of interest in writing for wind instruments (Secrist-Schmedes, 1996). From a musicological perspective, it appears that the contrasting colours and timbres available in a wind quintet are appealing tomany composers. From a psychological perspective, the contrasting timbres,types of instrument and playing techniques make the wind quintet a poten-tially fascinating source of investigation into how individuals behave in orderto function as a group. Indeed, perhaps these contrasts will make the inter-personal dynamics of the ensemble very different to those of a string quartetwhere the timbres and techniques of the instruments are much closer.Additionally, working with five as opposed to four people in the wind ensemble, the dyadic and triadic allegiances make the potential for conflict quite different to those facing the string quartet members. Thesequestions along with others covered by the string quartet researchers werepursued in the current project in a questionnaire study.

Methodology

RESEARCH INSTRUMENTS

The need to question as many quintets and address as many issues as possiblein the absence of any previous research meant that an observation and inter-view study would be impractical. Given time and financial constraints, andthe fact that wind quintets are geographically spread far and wide, it was feltthat a postal questionnaire would be the most effective method of gathering

Ford and Davidson: Members’ roles in wind quintets 55

data (in line with recommendations by Oppenheim, 1966). Subsequently, a21-page one-sided questionnaire was prepared consisting of 50 questionsusing 5-point Likert scales and open-ended semi-structured questions.1 Theaim of the questionnaire was to enquire about quintets in general, whilstinvestigating individual players’ points of view (additional comments wereencouraged wherever respondents felt it appropriate to add them). The questionnaire emerged out of studying the previous quartet literature (as discussed earlier), and the first author’s professional experience of wind quintets. The questions related to the following themes: details of formation,personnel changes, deputies, gender, age, careers of individual players, con-cert fees, frequency of rehearsals/concerts, rehearsal techniques, viewsabout over-rehearsal, performance nerves, verbal and non-verbal communi-cation, leadership, seating plans, disagreements between players, socialaspects both in and outside rehearsals/performances, administration, pro-gramming policies, individual practice, players’ attitudes about string quartets as an opposing medium, players’ concepts about wind quintet reper-toire, concert promoters’ prejudices, positive and negative aspects of playingin a wind quintet, priority of the quintet in their lives and how long theyenvisage their group staying together.

The questionnaire was then piloted, working with the first author’s ownprofessional wind quintet. The remaining four players of this quintet wereasked to complete the questionnaire independently of one another and, at the same time, to proof-read the text, check that the questions could not bemisinterpreted, amend errors and make suggestions for improvement. Thecompleted questionnaires were then returned to the first author for furtherrefinement.

PARTICIPANTS

The first author felt it necessary to find the largest number of professionalwind quintets possible by searching through national musical directories, thenames of which cannot be disclosed for reasons of anonymity. Following thesearch, 30 quintets from all over the UK were contacted by telephone andinvited to participate in the study. All but three (who had disbanded) wereinterested in being involved. Questionnaires were subsequently sent out to allthe members of 27 quintets throughout the UK (130 players in total). Two ofthese players returned two questionnaires each because they were membersof two different quintets, and two groups were incomplete at the time of thestudy, one group having just four members, the other only three. In total, 55questionnaires were returned representing 20 quintets (a response rate ofover 60% of the 30 quintets originally contacted) generating quantitativeand qualitative data. Eight of these quintets were represented by just one member, two quintets by two members, three quintets by three membersand one quintet by four members; six groups returned a full set of five ques-tionnaires. Of the total sample, 17 of the groups were professional, three


semi-professional. Three quintets consisted entirely of players in full-timeorchestral positions; four quintets were split between players in full-timeorchestral positions and freelance players, while the remaining 13 quintetswere combinations of freelance musicians, instrumental teachers, artsadministrators, music retailers, students and non-musicians.

Results and discussion

In this article, some general results are explored by discussing both the quan-titative and the qualitative data, and a case study of horn players is madeowing to the large numbers of changes in horn personnel discovered. A sys-tematic content analysis was carried out (in line with Miles and Huberman,1994) due to the large amount of qualitative data collected, and direct quota-tions from quintet members are used to illustrate the emergent themes. Inorder to facilitate discussion in the following exploration of our data, theemergent themes are compared with the findings of related psychologicalstudies and the previous string quartet work.

GENERAL FINDINGS

(i) Gender compositionApplying a frequency analysis to the data collected, it was discovered that noone male–female combination of players was different from any other (χ2 =6.4, df = 5, n.s.). Table 1 shows the distribution in terms of percentages,revealing that only 15 percent (three quintets) were single-sex groups. Theseresults contrast strongly with the string quartet field where single-sex groupstend to be in the majority (Murnighan and Conlon, 1991).

The distribution of men and women playing the five instruments of thewind quintet differed from that expected by chance (χ2 = 72, df = 1, p <.001). Figure 1 shows the percentages of men and women playing eachinstrument. From these data it is evident that flute players are mainly femaleand horn players are predominantly male. This finding corresponds to workby Abeles and Porter (1978), Delzell and Leppla (1992) and O’Neill (1997)


TA B L E 1 Gender composition in wind quintets

% of quintetsMales Females (N = 20)

1 4 152 3 353 2 154 1 205 0 50 5 10

who, in studying preferences for musical instruments, found that maleswould more typically choose brass instruments than woodwind, and that theflute was largely popular with females.

(ii) How did the group form?Applying a frequency analysis to the data collected, a significant result wasobtained (χ2 = 7.29, df = 2, p < .05). Table 2 shows the percentage ofresponses given to possible ways of forming. From this, it is evident that the majority of the wind quintets in the study formed whilst studying atmusic college, or else forged links with colleagues later on in their careers. Inthe qualitative data, players explained this categorization more fully byexpressing the desire to join forces with friends or acquaintances or to workalongside people they liked and respected. For example, one bassoonist wanted to

work with friends who have honest exacting standards.

One flute player expressed the wish to

improve my ensemble playing by working with other excellent musicians.

Changes over time From the entire data set, only two quintets have survivedwithout any changes of personnel since their formation. In fact, 77 percent


F I G U R E 1 Gender/instrument distribution in wind quintets (N = 20).

0

10

20

30

40

50

60

70

80

Flute Oboe Clarinet Horn Bassoon

males

%

females

of the quintets represented reported personnel changes. Of these quintets,respondents revealed various reasons for players leaving groups: a change ofcareer, moving away to study, family problems, lack of commitment, person-ality clashes, geographic constraints or sub-standard playing. When the fre-quency of instrument changes was examined according to instrument, a sig-nificant effect was found (χ2 = 15.85, df = 4, p < .01). The percentages ofchanges according to instrument are shown in Table 3. From this, it is obvious that there are far more changes in horn player than any of the otherinstruments. This startling difference has led us to make a case study ofthe horn players and the data collected about them. Their results will be discussed when examining the qualitative data collected from the question-naire.

All the ensembles studied reported that each time a member left and a newone joined, the group progress was affected – whether this was to the good orthe detriment of the group. This finding is consistent with the general psy-chological research on group dynamics (see Forsyth, 1990). Using a deputyor substitute player has a similar effect. Yet, all wind quintets use substitutes,which is generally considered necessary in the event of an individual playerin the ensemble wishing to accept alternative more financially rewardingwork, or the ensemble having to cancel a lucrative concert because of theunavailability of a player. Here is one indicative comment from a very busyfreelance horn player:

All ensembles except string quartets and piano trios use deputies. One can onlyget away with never using deputies if the quintet is full time (I can’t think of


TA B L E 2 Group formation in wind quintets

% of quintetsPlace of formation (N = 20)

At music college 50With colleagues 45As an off-shoot from another ensemble 5Other 0

TA B L E 3 Personnel changes in wind quintets

Instrument % of changes

Flute 9.5Oboe 9.5Clarinet 21Horn 43Bassoon 17

any full-time quintets in the UK) or if the fee is so good that one can turn downlarge blocks of work that clash.

This attitude is in stark contrast to that of many string quartets who givepriority to their ensemble, only earning their ‘filthy lucre’ with extra free-lance work (Murnighan and Conlon, 1991: 168) if and when time allows.String quartets, being much more of a full-time ensemble, use deputies veryrarely if at all:

This wasn’t a job in which, with one person sick, you simply call the musicians’union for a substitute. Even if an experienced violinist with a compatible musi-cal temperament could be found – no small task – it would take months for thenew group to jell. (Steinhardt, 1998: 277)

Other commitments All wind quintets in the study, though professional,classed themselves as part time. They rehearsed on average once a month, orfailing that, immediately prior to concerts. Unless members lived close to oneanother and/or the group gave a lot of concerts, it was not practical torehearse more frequently than this due to players’ other work commitments.A frequency analysis of the types of jobs undertaken by the quintet playersshowed a statistical difference (χ2 = 192.4, df = 6, p < .0001). Table 4 showsthe distribution – in percentages – of the primary sources of income for thequintet members in our study. From this, it appears that the significantmajority of quintet members follow freelance careers and that the most com-mon source of work is instrumental teaching.

(iii) Seating positionsWhen rehearsing and performing, wind quintets – like string quartets – haveto make decisions about how they should sit, and who they must sit oppositeor next to. The quintets do not have standard positions, whereas string quar-tets tend to choose between two basic positions (see Figure 2).

The variety of instruments and the additional player in a wind quintet cre-ate a greater number of choices. Nevertheless, it is possible that wind quin-tets may have derived some of their ideas from string quartet seating, as one


TA B L E 4 Primary sources of income for wind quintet players

% of quintet playersOccupation (N = 100)

Full-time orchestral position 22Freelance/instrumental teaching 60Solo 1Quintet 4Music admin/retail 8Student 2Non-musical 3

quintet member commented:

. . . flute is almost always in the same position, the rest varies a lot betweengroups – bassoon may be at front or back depending on acoustics. The maindecisions are normally flute and clarinet opposite each other? Bassoon on out-side to strengthen bass line? Direction of horn bell? Horn and bassoon together?

The flute invariably sits in the corresponding seat to that of the first violinin the string quartet and many quintets like to have the bass (i.e. the bassoon)coming from the middle, as is sometimes the case with cellists in quartets. It isalso important that the horn bell faces away from the audience so that thesound does not project straight out at them. The clarinet, often sharing theleading voice, is sometimes placed opposite the flute, although in somegroups the oboe assumes this position and the clarinet sits next to the flute.The string quartet idea of keeping the inner parts on the inside is not strictlyadhered to in wind quintets, probably because the part writing is such thatthe ‘inner parts’ are shared around the group rather than being confined tocertain instruments. The flute, for example, can provide beautiful soaringmelodies one minute, and supporting mid-range harmony the next.

A frequency analysis of the various permutations of seating plan wasundertaken, and a significant effect was found (χ2 = 25, df = 5, p < .001).Figure 3 shows the six different seating plans our quintet members made.Seating plan (1) is the most favoured amongst the quintets in the currentstudy and reinforces the ideas discussed earlier.

Of these quintets, half experimented with their seating plan before settling


F I G U R E 2 String quartet seating plans: V1 = 1st violin, V2 = 2nd violin, Vla = viola, C = cello.

V2 Vla V2 CV1 C V1 Vla

F I G U R E 3 Wind quintet seating plans: F = flute, O = oboe, H = horn, B = bassoon, C = clarinet.

(1) 55% H (2) 25% BO B O H

F C F C

(3) 5% H (4) 5% BC B H C

F O F O

(5) 5% C (6) 5% FB H C O

O F B H

for their layout in an attempt to produce ‘the best homogeneous sound’, where-as 18 percent of quintets reported that that they ‘just followed tradition’,although it is unclear exactly where or when such a tradition originated.

Leadership Traditionally, as outlined by Young and Colman (1979), stringquartets have always been led by the first violin. In Murnighan and Conlon’s(1991) study too, the terms ‘leader’ and ‘first violinist’ were used almostinterchangeably by interviewees. In the wind quintets interviewed here, how-ever, the leadership role is not so clearly defined. When asked whether theyconsidered their quintet to have a leader (as is the case for the string quartet),significantly 63 percent said ‘no’, 32.7 percent said ‘yes’, and 3.7 percent didnot respond. Indeed, from the additional comments it was clear that, in mostcases, players took equal responsibility for leading the ensemble, as one clari-nettist from a long-established group pointed out:

Each instrument has its own role and identity and we work as a team.

and as a horn player stated:

All members have strengths in different areas, so we are democratic in decisionmaking and respect each others’ strengths.

The term ‘leader’ can mean different things for different people. Sometimesquintet leadership is perceived as an administrative role only, as illustrated byone clarinettist:

The organization is done by me, but musical aspects are shared responsibilitiesand interpretative ideas come from the ensemble; works well.

These players see their colleague as leader of their ensemble because:

. . . she does basically all the admin [and therefore] runs it!

while this group, who share the administrative duties, do not feel they have aleader:

We do equal amounts of work for the quintet.

Returning to musical aspects such as expression and interpretation,respondents felt it appropriate to share responsibility around the group. Astwo flautists from different quintets pointed out, the concept of leading themusic was something which passed around the group according to the com-position of the piece and the musical style:

It varies depending on the music being played. We are very aware of each otherand find that different players lead at different times.

The team is led by the player with the most important tune.

A frequency analysis revealed that the distribution of responses for eachinstrument differed to that expected by chance (χ2 = 21.99, df = 5, p < .001).Table 5 shows the percentages of responses, revealing that the majority con-sidered the flautist to be leader.


The flautist was in this position only in as far as she or he is responsible forbringing other players in and off at the beginnings and ends of sections ofmusic. Players felt that the responsibility was then passed on to whoever hadthe leading voice during the course of a piece. Respondents also regardedtheir colleagues as leaders if they tended to have clear ideas about the direc-tion of rehearsals, had strong personalities, or if they took on the responsibil-ity for administrative duties.

Psychologists have found that in small groups, people sitting on the ends ofan inverted ‘V’ shape tend to emerge as leaders more frequently than thosesitting further in (see Bass and Klubeck, 1952). This study significantly rein-forces this theory, as in the majority of quintets, the leader (if there is one) sitsat one of the wings. In this study, only two other identified leaders (an oboistand a bassoonist) sit further in. Figure 4 shows which instruments are con-sidered to be leader and how many of those sit at one of the wings.

In cases where more than one member of the same quintet responded tothe questionnaire, there were differences of opinion between these membersas to whether they have a leader and who that leader is. This suggests that, likethe successful quartets in Murnighan and Conlon’s (1991) study, wind quin-tets advocate democracy. It seems, however, that they still have their fairshare of conflict. According to Young and Colman (1979), conflict is likely in


TA B L E 5 Who do you consider to be leader?

% of responses(N = 100)

Flute 55.6Oboe 16.7Clarinet 16.7Horn 0Bassoon 5.5Unanswered 5.5

F I G U R E 4 Seating positions of leaders.

3X

2 X X 41 X X 5

Seating position 1 2 3 4 5

Flute 10Oboe 1 1Clarinet 3Bassoon 1

a quartet which can easily divide equally into two. In a quintet, the playersform sub-groups, causing a majority–minority split of 3/2 or 4/1. All groupssaid that the majority view was the accepted one. This observation leads tothe next topic of discussion.

(iv) Group functionZander (1994) states that people are more likely to achieve group goals ifthey speak freely with one another and they are more likely to do this if theyhave spent a lot of time together. This idea is reinforced by one clarinettistwhose quintet is long established, successful and busy:

We all get on really well together but feel free to take anyone or anything to task– resulting in some pretty awful rows sometimes!

Conversely, some groups said they do not disagree at all. Given the humantemperament, it seems unlikely that there would be no disagreements, but ifthey haven’t been playing together for very long and they rehearse less thanonce a month, they may not feel comfortable in taking other players ‘to task’(see Argyle, 1994: 9; Zander, 1994: 69).

According to Murnighan and Conlon (1991), string quartets play muchmore than they talk when rehearsing. The wind quintets in this study show avariety of approaches, as this horn player remarks:

With some groups there is so much empathy that it all happens in the playing.Some need to talk. Both approaches are equally valid if the end result is good.

One clarinettist feels that:

Too much stopping to talk tends not to give good results for us, but often ‘looks’and body language say more than words!

Indeed, eye contact was the most popular form of non-verbal communica-tion, although the questionnaire did not successfully measure its frequencyor effectiveness. Expressive body movement was discussed less than might beexpected, although it is possible that players are not consciously aware oftheir movements whilst playing, therefore rendering the results inaccurate(see Davidson, 1997, for more details on expressive body movement in per-formance). In addition, unlike string players, wind instrumentalists cannotsee how their colleagues are producing their sound. Articulation, tone anddynamics are produced inside the body and the instrument and so wind playersmay need to use discussion to clarify technical matters. The ratio of discus-sion to actual playing was not successfully measured by the questionnaire.

The very different characteristics of the wind instruments lead the playersto comment that they need to be equally skilled as soloists and accompanists,switching from leading voice to supporting harmony. Most reported beinghappy to perform this juxtaposition of roles, although the questionnairerevealed some striking findings regarding horn players. While many enjoyed thechallenges of wind quintet playing, some found the experience unrewarding,


uncomfortable, difficult or dull. It has already been shown that quintets havehad significantly more changes of horn player than any other instrument.None of the horn players in this study take on leadership roles, nor do theyadmit to having any responsibility for administrative duties. For this reasonwe now examine our data more closely to explain why this may be so.

(v) Horn playersOf the quintet members questioned, one third reported bad experiences theyhad encountered with horn-playing colleagues, claiming that their previoushorn players had perhaps been more committed to building up orchestralcontacts than to the future of the quintet and eventually this had an adverseeffect on the group. For example, one flute player commented:

The original player was extremely busy doing freelance playing which not onlycut down our rehearsals but affected [their] embouchure!

The horn player’s lack of sensitivity often frustrated colleagues who obviouslydid take their group seriously and wanted it to work, as one oboist commented:

The original horn player didn’t really take wind quintets seriously and didn’twant to rehearse, also didn’t really listen as a chamber music player.

Another group suffered from a lack of continuity in the format of theirquintet, owing to a series of horn players who were either unable or unwill-ing to commit themselves, as one bassoonist reported:

We had four changes of horn player! Two career changes, one long-termdeputy, one sacking (poor attitude and attendance).

If taken at face value, these comments paint a rather bleak picture of hornplayers, suggesting that they lack the ability to commit themselves, display lit-tle talent and are incapable of taking chamber music seriously. However, thiswas only true in one third of the quintets, which shows that there are plentyof good, committed players in quintets.

Horn players’ own comments may help to provide a clearer and fairer ideaof what it is like to be the only brass player in a wind quintet and might beginto explain why so few of them stick at it for very long.

Chamber music is completely different from orchestral playing.

This comment from one horn player and teacher seems rather surprisingbecause orchestral playing and chamber music playing do, of course, havecommon elements, both requiring high levels of discipline, concentrationand listening skills. However, Reynolds (1997) seems to agree with this state-ment when he describes some of the difficulties of wind quintet playing froma horn player’s point of view. He believes that woodwind players are able toreproduce the music as written with relative ease, while

The orchestral horn player has to make conscious and deliberate adjustmentsin volume, articulation, and release in order to play the traditional music written for wind quintet. (p. 185)


Consequently, as one very experienced horn player (having played in severalwind quintets) remarks:

For the horn, wind quintet is much harder than any other type of playing,whether solo, chamber or orchestral.

Reynolds (a leading US horn player and teacher) goes on to explain whythis might be the case. Because woodwind players ‘generally have an attitudethat is less metronomically bound than brass players’ (p. 185), horn playersmay feel uncomfortable about the timing of breaths and nuances. They areless able to move around their instruments with the ‘astounding rapidity’ offlutes and clarinets whose instruments are highly mechanized. Horn playersmust be very sensitive to the range of colours offered by each of the wood-wind instruments. The flute, for example, often dominates the quintet on thetop line, but when providing an inner voice in a similar range to that of thehorn it tends to project considerably less well, so careful balancing is needed.The clarinet poses similar problems for the horn player because its ‘softestnotes are barely audible’, although the two can blend well together when thevolume is at mid-range and when parts are fairly close together. The oboe isso different from the horn that ‘it is better to find ways to use the wonderfuldifferences in quality between these two instruments rather than to distortone or the other, or both, in an attempt to blend’. Conversely, bassoonistsshare a similar range and somewhat similar tone quality to that of the horn,and can generally balance well, except in particularly loud passages.

Do the demands of wind quintet playing ever cause horn players to feelinadequate alongside four woodwind players? Possibly as, according to Kemp(1996), horn players tend to be more susceptible to anxiety and are generallymore introverted than their trumpet and trombone-playing colleagues.Additionally, Baumeister (1984) found that performance suffered under pres-sure if participants felt self-conscious or anxious about their abilities. Heargues that self-consciousness increases performers’ awareness of stress,pain or fatigue, causing them to hold back and not necessarily give of theirbest. This may go part of the way to explain why, as some horn playersbecome tired during a particularly tough concert programme, they tend to‘split’ notes. One player certainly feels uncomfortable at times and does notlike

. . . feeling responsible if I don’t play as well as hoped on a performance

while another hates

. . . being the one who can’t always play it right.

In contrast, one rather dissatisfied player seems to have overlooked all of thechallenges that wind quintets offer:

Sometimes the part isn’t very interesting!

These negative feelings can be related directly to Murnighan and Conlon’s


(1991) study on two counts. First, just as second violinists seldom have theopportunity to express themselves freely in a solo line but are, nevertheless,required to be as equally skilled and flexible as the first violin in order to sup-port his or her sound, intonation, dynamic and so on, according to Reynolds(1997):

Horn players must develop a willingness to trade the glamour of the leadingvoice for the nobility of playing an inner voice superbly. This can be difficult foran orchestral player for whom ‘projection’ is very important. (p. 186)

Second, it is worth pointing out that in a full-size symphony orchestra,horn players usually play in a section of four, all of whom immediatelyunderstand the technical difficulties of the instrument. In a wind quintet, thehorn player is suddenly the only brass player, so it seems that they need to feelvalued by their colleagues in the same way that second violinists requirerecognition and respect from other quartet members. One rather contentedhorn player confirms this idea when he describes what is for him the mostpositive aspect of playing in a wind quintet:

... nice to be accepted and respected by the other members and by musiciansoutside the group.

In fact, just as the quintets in this study have had more changes of horn play-er than any other instrument, Murnighan and Conlon (1991) reported that‘everyone felt that a second violinist was the most likely member to leave aquartet’ (p. 175).

On a more positive note, Reynolds (1997) sees wind quintet playing as apositive learning experience for horn players, and as a challenge to the devel-opment of technical and music skill:

Wind quintet playing can have a refining effect on horn players, particularly inthe areas of style, balance, attacks, dynamics and rhythmic nuances.Woodwind players tend to use dynamic and rhythmic nuances more freely thanbrass players. This is probably due to their having a richer fund of solo litera-ture. Brass players prize power and precision; woodwind players prize line andelegance. These two attitudes can merge in a wind quintet if horn players arewilling to expand their musicianship to include these freedoms. (p. 186)

It seems that many horn players in the study are willing to ‘expand theirmusicianship’ and are thoroughly positive about the whole experience ofplaying in a wind quintet. This full-time orchestral player actually prefersplaying in a wind quintet:

My priorities are first solo work, second quintet, third orchestra.

This perhaps relates to Butterworth’s (1990) finding that musicians like toplay chamber music because they can influence how the music should go, aluxury not possible in an orchestra where the conductor has the final say inmusical decisions. The fact that this player seems to be primarily pursuing asolo career further suggests a desire to be more musically independent.


On the other side of the coin, players feel that wind quintet playing helpsthem in their orchestral work or other aspects of their careers. They seem tobe glad of

. . . the chance to explore new repertoire and to develop the chamber musicskills necessary in today’s musical environment

and the fact that wind quintet playing is

. . . stretching (technically and musically)

offers

. . . more challenging music than orchestral parts and is good for tuning and lis-tening practice

and provides

. . . good discipline for individual standards.

Just as, in string quartets, the role of the second violinist has been high-lighted as being of particular interest (Murnighan and Conlon, 1991), thenumber of personnel changes, lack of leadership roles, reluctance to under-take administrative tasks, and the fact that some horn players appear lessconfident about their abilities than their woodwind colleagues suggest thathorn players need to feel valued by other members of their group becausethey have a very difficult role, being the only brass player.

(vi) Allied sociological issuesWind quintets seem to face barriers that string quartets simply do not comeacross. Not only have wind quintets been ignored by researchers, they areoften sadly neglected by concert promoters too, judging by the overallresponse to the following statement from the questionnaire: ‘Concert promot-ers prefer to book string quartets rather than wind quintets.’

Many of the wind quintet players mentioned low morale. At least half ofrespondents apologized for or complained about the wind repertoire andcompared it with that for string quartet whose ‘repertoire is much moreextensive and well known’.

Some players feel that the

lack of respect from music societies as a whole

and

getting overlooked by people who favour string quartets

are unfair and possibly the most negative aspects of playing in a wind quintet.Other players tried to understand the situation from a financial point of

view. While it is not necessarily in the concert promoter’s interest to make aprofit on ticket sales, it is important to break even and not run at a loss; other-wise, a music society may have to fold. Several respondents seemed to thinkthat having ‘one less person to pay’ could be yet another incentive to booking


a string quartet as opposed to a wind quintet, although wind quintets tend tobe paid less than string quartets anyway. A fee of over £250 per person isconsidered ‘very rare’ for a wind player in a quintet.

To follow up this point, personal communications with a London concerthall confirmed that this is indeed the case, even with an internationallyrenowned wind quintet attracting only the same fee as a middle-rankingstring quartet. Additionally, one of the world’s leading string quartets attract-ed three times the fee per player of a nine-piece wind ensemble of similarstanding.

Fees for wind quintet concerts start from less than £50 per player. Perhapswind players feel they have to keep their fees low so as not to deter concertpromoters from booking them. Wind quintets in general do not have the rep-utation of string quartets: only 40 percent of wind quintets in this study havemade commercial recordings and, of these groups, only two have made morethan five recordings. Players claim this is due to lack of public recognitionand the fact that the wind quintet repertoire is less well known than that ofstring quartets, as one flute player suggests:

String quartets are a known factor and can generally programme a well-knownpiece alongside a lesser-known work. Much of the wind quintet repertoire isonly known to wind players.

If audiences are used to hearing string quartets and continue coming tohear string quartets why book an alternative? One reason might be to attractmore children and young people to concerts. Indeed, personal communica-tion with a concert promoter revealed that a wind quintet recital had attract-ed a younger audience than that which usually attended their various stringquartet concerts because more children learn to play the flute and clarinetthan any other instrument. So there are evidently some powerful social andcultural factors affecting employment of wind quintets.

Blending Young and Colman (1979) pointed out that the string quartet is‘made up of four instruments of the same family, similar in timbre andmethod of sound production’ (p. 12). If, as they suggest, some composershave found difficulty writing for the string quartet because of its ‘lack ofcolour contrasts’ (p. 12), it seems ironic that the wind quintet with its diver-sity of instruments does not attract more attention. While for some players,

It’s a fairly unusual and eccentric mix of sounds and can be quite a powerfulexperience!

others understand that the ‘lack of a really homogeneous sound’ may be off-putting for listeners.

Comparing string quartets and wind quintets, one very experienced free-lance horn player said:

They [strings] have to rehearse much harder and in much greater depth.


The suggestion that string quartets have to work harder is interesting. Is thisbecause the repertoire is ‘better’ as has been suggested by several respond-ents, and therefore requires more in-depth analysis? Various biographicalaccounts of string quartets (Blum, 1986; Steinhardt, 1988; Rounds, 1999)describe how players rehearse together in enormous detail, even down to theproduction, attack or articulation of individual notes. Some quartetsrehearse every day even though they are performing the same work up to 40times in a season.

In comparison, it is ironic that 90.6 percent of wind quintet membersthink that a piece of music can be over-rehearsed even though they rehearsetogether less than once a month! Many wind quintet players do not feel suffi-ciently committed to the repertoire for whatever reason, and therefore maynot be inclined to spend hours and hours studying it. Because they have somany other commitments they do not have the luxury of such in-depthrehearsal. They have to earn a living and wind quintets do not generate agreat deal of income – ‘rehearsals without pay’ and ‘hard work for littlemoney’.

Some respondents feel that comparing the two mediums is difficultbecause string quartets tend to work full time and wind quintets do not.However, as one flute player points out:

If we were booked and playing as much as string quartets this would be irrelevant.

Finally, respondents were asked how much of a priority their wind quintetwas in their life at the moment and how long they envisaged their group stay-ing together. Applying a frequency analysis to the data collected revealed that


TA B L E 6 Attitudes and future aspirations

% of responses(N = 100)

How long do you envisage your quintet staying together?Always 1.8For a long time 54.5For a short while 5.5Not long at all 5.5Don’t know 30.9NA 1.8

How much of a priority is your quintet in your life at the moment?1 not at all 9.12 30.93 27.34 205 very much so 12.7

plenty of players clearly value the benefits of belonging to a wind quintet (χ2 = 76.9, df = 5, p < .001). Table 6 shows the percentage responses reflect-ing the attitudes and future aspirations of the quintet members. More thanhalf were positive about their group’s future, hoping that it would keep goingfor a long time. For the majority of players, however, their quintet was not ahigh priority when considered alongside other commitments (χ2 = 9.44, df =4, p < .1).

Thus it seems the wind quintet does not attract the social and culturalinterest of the string quartet, so important in enabling individuals to earn aliving.

Conclusions

This study has provided a large amount of data from a sample of ensemblesspread over a wide geographical area. It shows that wind quintets share cer-tain features in common with string quartets: both ensembles require goodinterpersonal dynamics in order to function effectively. In addition, however,the wind quintet has its own distinctive set of issues: working as a team andsharing the responsibilities of leadership; time management when rehearsaltime is limited; adapting to the change in group dynamic introduced by theuse of a deputy has a major effect; experimenting with seating arrangementin order to achieve the best overall blend of sound; and promoting the ensem-ble and its repertoire in order to obtain work and earn respect can be prob-lematic. Additionally, the horn player in particular needs to be either carefullytreated or trained to integrate more easily into the wind ensemble.

Obviously these are vital psychological, sociological and cultural issuesrequiring much more study. Future research, especially in the form of obser-vation, could provide more data, especially on rehearsal and concert behav-iour; semi-structured interviews could also help clarify any ambiguities thatmay have arisen through misinterpretation of our postal questionnaire.

Another research possibility might be to extend the existing questionnairestudy to quintets abroad. Any similarities between British and foreign groupswould help substantiate the findings of the present study whilst insights intotheir differences would be fascinating. Another different angle could be toundertake a more sociologically based interview study of concert promoters,investigating their likes, dislikes and preferences for booking certain types ofensembles and repertoire, in order to investigate if the thoughts and feelingsof the respondents of this study are justified.

There is no doubt that the wind quintet is a little-known medium worthyof greater recognition. By introducing it to the field of music psychology wehave aimed to reawaken an interest in the social dynamics of chamberensembles whilst making some headway towards raising the profile of thewind quintet.


N O T E

1. Copies of the questionnaire can be obtained from the first author via the email orpostal addresses cited at the end of this article.


The authors would like to thank all the quintet players who filled in and returnedquestionnaires and, in particular, the Chameleon Ensemble for their valued input intothe pilot study.

R E F E R E N C E S

Abeles, H.F. and Porter, S.Y. (1978) ‘The Sex-Stereotyping of Musical Instruments’,Journal of Research in Music Education 26: 65–75.

Allmendinger, J., Hackmann, J.R. and Lehman, E.V. (1996) ‘American Musics: Lifeand Work in Symphony Orchestras’, The Musical Quarterly 80(2): 194–219.

Allport, F.H. (1924) Social Psychology. Boston, MA: Houghton Mifflin.Alverno, L. (1987) ‘Music Student Quartets and Psychotherapy’, Medical Problems of

Performing Artists 2: 75ff.Argyle, M. (1994) The Psychology of Interpersonal Behaviour, 5th edn. London: Penguin.Atik, Y. (1995) ‘People Factors in the Performing Arts’, Performing Arts Medicine News

3(3): 7–12.Bales, R.F. and Bargatta, E.F. (1955) ‘Size of Group as a Factor in the Interaction

Profile’, in A.P. Hare et al. (eds) Small Groups in Social Interaction, pp. 396–413.New York: Knopf.

Bass, B.M. and Klubeck, S. (1952) ‘Effects of Seating Arrangement on LeaderlessGroup Discussions’, Journal of Abnormal and Social Psychology 47: 724–7.

Baumeister, R.F. (1984) ‘Choking under Pressure: Self-Consciousness and ParadoxicalEffects of Incentives on Skillful Performance’, Journal of Personality and SocialPsychology 46: 610–20.

Blum, D. (1986) The Art of Quartet Playing: The Guarneri Quartet in Conversation withDavid Blum. Ithaca, NY: Cornell University Press.

Butterworth, T. (1990) ‘Detroit String Quartet’, in J.R. Hackman (ed.) Groups thatWork (and Those that Don’t), pp. 207–24. San Francisco: Jossey-Bass.

Davidson, J.W. (1997) ‘The Social in Musical Performance’, in D.J. Hargreaves andA.C. North (eds) The Social Psychology of Music, pp. 209–28. Oxford: OxfordUniversity Press.

Davidson, J.W. and Good, J.M.M. (1997) ‘Social Processes in Western Art ChamberMusic: Some Aspects of the Social Psychology of the String Quartet’, in A.Gabrielsson (ed.) Proceedings of the Third Triennial ESCOM Conference, pp. 329–32.Uppsala: Uppsala University Press.

Davidson, J.W. and Good, J.M.M. (2002) ‘Social and Musical Co-ordination betweenMembers of a String Quartet: An Exploratory Study’, Psychology of Music 30(2):186–201.

Delzell, J.K. and Leppla, D.A. (1992) ‘Gender Associations of Musical Instruments andPreferences of Fourth Grade Students for Selected Instruments’, Journal of Researchin Music Education 40: 93–103.

Faulkner, R.R. (1973) ‘Orchestra Interaction: Some Features of Communication andAuthority in an Artistic Organisation’, Sociological Quarterly 14: 147–57.


Forsyth, D.R. (1990) Group Dynamics, 2nd edn. Pacific Grove, CA: Brooks/ColePublishing.

Frederickson, J. and Rooney, J.F. (1993) A Sociological Study of the Freelance ClassicalMusician – In the Pits. New York: Edwin Mellen Press.

Kemp, A.E. (1996) The Musical Temperament: Psychology and Personality of Musicians.Oxford: Oxford University Press.

Levine, S. and Levine, R. (1996) ‘Why They’re Not Smiling: Stress and Discontent inthe Orchestral Workplace’, Harmony (Forum of the Symphony OrchestraInstitute), Vol. 2, April [www.soi.org].

Miles, M.B. and Huberman, M.A. (1994) Qualitative Data Analysis, 2nd edn. ThousandOaks, CA: Sage.

Murnighan, J.K. and Conlon, D.E. (1991) ‘The Dynamics of Intense Work Groups: AStudy of British String Quartets’, Administrative Science Quarterly 36: 165–86.

O’Neill, S.A. (1997) ‘Gender and Music’, in D.J. Hargreaves and A.C. North (eds) TheSocial Psychology of Music, pp. 46–66. Oxford: Oxford University Press.

Oppenheim, A.N. (1966) Questionnaire Design and Attitude Measurement. London:Heinemann.

Reynolds, V. (1997) The Horn Handbook. Portland, OR: Amadeus Press.Rounds, D. (1999) The Four and the One: In Praise of String Quartets. Fort Bragg, CA:

Lost Coast Press.Secrist-Schmedes, B. (1996) Wind Chamber Music: Winds with Piano and Woodwind

Quintets: An Annotated Guide. Lanham, MD: The Scarecrow Press.Shaw, M. (1971) Group Dynamics: The Psychology of Small Group Behavior. New York:

McGraw-Hill.Small, C. (1987) ‘Performance as Ritual: Sketch for an Enquiry into the True Nature

of a Symphony Concert’, in A.L. White (ed.) Lost in Music: Culture, Style and theMusical Event. London: Routledge & Kegan Paul.

Steinhardt, A. (1998) Indivisible by Four: A String Quartet in Pursuit of Harmony. NewYork: Farrar, Straus and Giroux.

Stoneham, M., Gillaspie, J.E. and Clark, D.L. (1997) Wind Ensemble Sourcebook andBiographical Guide. Westport, CT: Greenwood Press.

Young, V.M. and Colman, A.M. (1979) ‘Some Psychological Processes in StringQuartets, Psychology of Music 7: 12–16.

Zajonc, R.B. (1965) ‘Social Facilitation’, Science 149: 269–74.Zander, A. (1994) Making Groups Effective, 2nd edn. San Francisco: Jossey Bass.

L UA N F O R D trained as a performer at Birmingham Conservatoire, and later read for aMasters Degree in Psychology for Musicians at the University of Sheffield. She hasbeen clarinettist with the Chameleon Ensemble since 1993; this has led to a particularinterest in the social interaction of small chamber ensembles. In addition she enjoys acombination of instrumental and classroom teaching, music animateur work andorchestral playing. She is Co-ordinator of Woodwind and Brass at BirminghamConservatoire Junior School, and in 2001 she was awarded an honorary degree fromBirmingham Conservatoire ‘in recognition of distinguished work within the field ofmusic’.Address: Department of Music, University of Sheffield, Sheffield S10 2TN, UK. [email: [email protected]]


JA N E DAV I D S O N is Reader in Music at the University of Sheffield. She has publishedwork ranging from musical development to performance. She teaches MA Psychologyfor Musicians and MA Psychology of Music, runs the MA Music Theatre Studies andcurrently supervises 15 PhD students. She also runs undergraduate modules onDevelopment of Musical Ability, Psychological Approaches to Performance and MusicTherapy. In addition to this, she is a busy performer and director, working in Operaand Music Theatre. She was Editor of Psychology of Music from 1997–2001.Address: as Luan Ford. [email: [email protected]]


The perception of emotionalexpression in music: evidence frominfants, children and adults

75A R T I C L E

Psychology of Music



Researchvol (): ‒

[- ():; ‒; ]

E L I Z A B E T H S . N AW R O TM I N N E S O TA S TAT E U N I V E R S I T Y M O O R H E A D

A B S T R A C T Two studies investigated the development of the perception ofemotion in music. In Study 1, preschool children and adults matched nine piecesof music to five photographed facial expressions (happy, sad, anger, fear and neutral). While children did not agree with the adult majority interpretation formost pieces, their pattern of responding to the music, both with photographchoices and spontaneous verbal labels, was similar to the adults. Importantmethodological differences between this and previous research could explain theinconsistencies. Study 2 used happy and sad music along with a dynamic visualdisplay in an intermodal matching experiment with 5- to 9-month-old infants.Infants preferred the affectively concordant happy display but did not look longerto the affectively concordant sad display as predicted. Taken together, theseresults begin to explore how emotional perception from music may be due toinnate perceptual predispositions together with learned associations that developin childhood.

K E Y W O R D S : development, emotion, infant, intermodal, music, perception

Infants, children and adults are sensitive to emotional information specifiedfrom a wide range of sources, including facial and vocal expressions andmusic. There is a growing body of research examining the development ofaffective interpretations in music (Cunningham and Sterling, 1988; Dolginand Adelson, 1990; Gregory et al., 1996; Kastner and Crowder, 1990;Kratus, 1993; Robazza et al., 1994; Terwogt and Van Grinsven, 1991). Muchof this work compares the responses of younger to older listeners in order to identify consistencies in the interpretation of affect. However, the results so far are equivocal. For example, Dolgin and Adelson (1990) found that 4-year-olds correctly identified emotions, including happy, sad, anger andfear, from sung melodies. Furthermore, Kastner and Crowder (1990) found

sempre :

06Nawrot 2/12/02 10:06 am Page 75

that 3-year-olds matched positive (happy/interested) and negative (sad/angry)faces to melodies played either in the major or minor chord consistent withadults. However, Gregory et al. (1996) reported that 4-year-olds’ responses tohappy and fearful music were not above chance and others have found thatchildren confuse emotions of fear and anger in music (Robazza et al., 1994;Terwogt and Van Grinsven, 1991).

Research has pointed out that confusion of fear and anger from musicmight be the result of a mixed message. In the case of music with a compos-er’s intent to express anger, an appropriate reaction might be fear. In thiscase, a particular response could represent the composer’s feeling (anger) orthe listener’s reaction (fear). Children’s egocentrism may predispose them tointerpret the music from only one perspective, namely their reaction ratherthan the composer’s intention, and this may lead to an apparent confusion(Terwogt and Van Grinsven, 1991).

The use of inconsistent methodologies together with children’s low levelsof responding under some task conditions may help explain other inconsist-encies in the literature. For example, some research has judged children’sinterpretations in comparison to adults’ presumed correct answers (e.g.Cunningham and Sterling, 1988) which becomes difficult when children do not respond at high levels. Responding above chance levels is influenced by the number of possible choices, and research designs vary from a two-alternative forced-choice task (e.g. Kratus, 1993) to a four-alternative forced-choice task (e.g. Dolgin and Adelson, 1990), with some including neutralaffect categories, and others not. Kratus (1993) found greater consistency inchildren’s responses compared to adults using a simpler, two alternativeforced-choice task, whereas Dolgin and Adelson (1990) found significant dif-ferences when more choices were presented, and when children’s responseswere analyzed in terms of correctness. Moreover, some studies elicited visualresponses (pointing to schematic or photographed facial expressions), andothers used verbal response sheets (Cunningham, 1988; Robazza et al.,1994).

In addition to the research on children’s perception of emotion frommusic, research has investigated the origins of emotional perception withinfant listeners. The ability of infants to detect emotional expressions wouldbe especially remarkable as it suggests a biological readiness for the discrimi-nation of socially relevant signals. There is a rich literature demonstratinginfants’ perception of facial expressions of emotion (e.g. Nelson, 1985) andvocal expressions of emotion (Walker, 1982; Walker-Andrews and Grolnick,1983; Walker-Andrews and Lennon, 1991).

There is reason to believe that infants would be sensitive to emotionalinformation specified in music, too. For example, studies of infants’ lookingand listening preferences show that they perceive many of the formal proper-ties of music similarly to adults, including information about pitch, melody,tempo and musical phrase structure (Schellenberg and Trehub, 1996;


06Nawrot 2/12/02 10:06 am Page 76

Trehub et al., 1987, 1999; Trehub and Schellenberg, 1995; Trehub andTrainor, 1993). Trainor et al. (2000) concluded that the musical quality ofinfant-directed speech makes it a particularly effective means of emotionalcommunication. Fernald (1989) agreed that when it comes to communicat-ing emotion through infant-directed speech, the melody is the message.According to Rock et al. (1999) music ‘may be a more powerful medium thanspeech for affective communication with infants’ (p. 527). But whether or notinfants interpret a discrete emotional message from music is not clear.

In a recent study, Rock et al. (1999) found that adults could reliably dis-cern which type of music (lullaby or play song) was played to an infant basedon their attention. Lullabies could be considered as infant-directed music, inkeeping with the idea that caregivers adjust their communication style toaccommodate the listener (similar to infant-directed speech [IDS]). Infantsalso appear to prefer the pleasing sounds of consonant music. Trainor andHeinmiller (1998) reported that 4-month-olds listened longer to a set of con-sonant interval notes than dissonant interval notes and that this preferencewas maintained when the dissonant intervals were placed into a naturalisticmusical context (a Mozart minuet). Furthermore, Zentner and Kagan (1998)found that 6-month-olds not only preferred to listen to a consonant over adissonant version of a melody, but showed other meaningful behaviors aswell, including more vocalizing and less fretting during the consonantmelodies. Infants’ reactions to different types of music imply that they candetect and act on the emotional message.

Gentile (1998) used an ecological approach to investigate the perception ofemotion from music, stating that the musical structure itself may specify anemotional content. If this is the case, then adult listeners, as well as musicallynaive and very young listeners, should be consistent in their judgments ofemotion from music. Adult listeners judged the emotional content of 28musical excerpts using a combination of forced-choice and Likert scales.Children as young as 3 years showed agreement with the adults’ judgmentsfor happy, sad and angry excerpts. Gentile went on to test whether 8-month-old infants could detect an emotional message from music previously rated bythe children and adults. Infants were first exposed to a 3-minute familiariza-tion phase where they saw six dynamically posed happy expressions while listening to six different happy musical excerpts. Following the familiarizationtrials, one of three types of change conditions was presented: the musicchanged (e.g. from happy to sad); the facial expression display changed (e.g.from happy to sad); both the music and the display changed (e.g. bothchanged from happy to sad). Analysis of the infants’ looking times found evi-dence for discrimination in all three change conditions. For example, whenthe music changed to sad, infants looked less to the happy display. Likewise,their looking time decreased when the faces changed to sad. When both thedisplay and music changed, infants looked less during the concordant sad dis-plays than the concordant happy displays. Gentile’s (1998) study is among

Nawrot: Perception of emotional expression in music 77

06Nawrot 2/12/02 10:06 am Page 77

the first to test whether infants could detect a discrete emotional messagefrom musical structure based on findings with children and adult listeners.

The present series of studies were similarly designed to investigate the per-ception of emotion from music across these three age groups. Study 1 com-pared the interpretation of music by pre-school-aged children and adults.Identical procedures were used for children and adults, including photo-graphed facial expressions and a neutral emotion category. Analyses com-pared the pattern of choices made by each listener, rather than a judgment ofthe correctness of the children’s answers, and verbal labels were included forconsideration.

In Study 2, 5–9-month-olds participated in an intermodal matchingexperiment designed to measure their responses to affectively concordant anddiscordant visual and musical stimuli. The intermodal matching paradigmwas chosen because research has shown that infants are capable of extract-ing invariant affective information from intermodal displays (Soken and Pick,1999; Spelke and Owsley, 1979; Walker, 1982; Walker et al., 1983; Walker-Andrews and Lennon, 1991). While previous research used facial and vocalemotional expressions, the present study was intended to extend that result toinclude affective information contained in music. If infants in the presentstudy understood the emotional message contained in the music, they shouldlook longer to affectively concordant displays versus discordant displays.Specifically, it was predicted that infants would look longer to a happy facialexpression when it was accompanied by happy music and likewise they wouldlook longer to a sad facial expression when it was accompanied by sad music.

Study 1

METHOD

ParticipantsChildren attending two local day-care facilities were invited to participatethrough letters sent home. Parental consent was received for 24 children,ranging in age from 3–5-year-olds. Their mean age was 3.9 years (35 to 68months) and 14 were boys. In addition, 20 adults ranging in age from 18 to37 years participated (mean age 23.3 years; 11 women). Adult participantswere college students solicited through the department’s research subjectpool.

StimuliFacial expressions Stimuli included five photographs of a female model fromthe shoulders up, posing naturalistic facial expressions including happy, sad,anger, fear and neutral. These five photographs were selected from a group of36 expressions by two undergraduates who rated these as the best examplesof each expression. Photographs were glossy finish, 9 × 13 inch color printson cardboard backing.


06Nawrot 2/12/02 10:06 am Page 78

Musical selections Nine musical pieces were selected to express the emotionshappy, sad, anger and fear by an experienced judge (a professor of music).Two undergraduates then selected the best 20-second passage from eachpiece to illustrate a discrete emotion. Pieces 1 and 2 both came from Mozart’sViolin Concerto No. 5 (Allegro aperto); piece 3 came from Beethoven’s SymphonyNo. 6 (Pastoral selection from ‘Awakening of happy feelings on arriving in thecountry’); piece 4 was Bach from Gould’s Goldberg Variations (Aria 16, open-ing movement); piece 5 was from Vask’s Message (selection from Lauda); piece6 was Barber’s Adagio for Strings from Platoon; and pieces 7, 8 and 9 all camefrom Prokofiev’s Alexander Nevsky. In the following discussion, the pieces aregrouped by happy (pieces 1–3), sad or neutral (pieces 4–6) and fear or anger(pieces 7–9). All nine pieces came from digital recordings.

ProcedureAll participants were individually tested. The five photographs were laid outon a table in front of the child. To ensure that each child could correctly iden-tify the emotion in the faces, a pre-test was administered where the child wasasked to point to the face that looked happy, sad, afraid and angry. All of thechildren correctly identified each facial expression. Following the pre-test,children were told that they would listen to different pieces of music andwould have to ‘point to the face that goes with the music’. The nine 20-secondselections were played in a different random order for each participant. Inaddition to their choices of the facial expressions, children were also asked togive a verbal label to each piece of music. The procedure was identical foradult participants with the exception of administration of the pre-test.

RESULTS AND DISCUSSION

Table 1 shows how the majority of adults (50% or greater) matched themusical pieces to the photographs. These choices can be taken as the pre-sumed correct interpretation of each piece. The corresponding percentage ofchildren who agreed with that correct interpretation is also given. As piece 5(from Vask’s Message) did not elicit a majority interpretation by either adultsor children, it was not analyzed further. The children’s choices were morehighly variable than the adults, with only one piece of music eliciting amajority (over 50%) response that matched the adults. On average, 75 percentof the children also provided a verbal label to each piece, and these wereconsistent with their photo choices for seven of the eight pieces. Althoughonly a small percentage of children agreed with the adults’ majority choice ineach case, the performance levels of these 4-year-olds were similar to previ-ous reports (e.g. Dolgin and Adelson, 1990). However, the children’s low levelof agreement with a single, adult choice does not preclude the possibility thatthey were distributing their choices in a way that was consistent with theadults. Therefore, the data were further analyzed to compare the pattern ofchoices made by both children and adults.


06Nawrot 2/12/02 10:06 am Page 79

For each musical piece, a chi-square statistic (α = .05) was computed to determine whether the children’s and adults’ choices across all five photographed facial expressions were associated. Overall, the children’s andadults’ pattern of choices did not differ for five of the eight pieces. Figures 1 to3 show the results for the group of happy (pieces 1–3), sad/neutral (pieces 4,6) and anger/fear (pieces 7–9) pieces, respectively.

Children were no different than adults for some music that adults inter-preted as happy and neutral/sad. For example, for piece 1, children andadults most often selected happy, χ2(4) = 2.0, p > .05, and for piece 4, sad andneutral choices were not distributed differently, χ2(4) = 5.26, p > .05. Therewas similar consistency for the fear/anger music. The pattern of respondingto pieces 7 and 8 was not different between children and adults, χ2(4) = 3.79,p > .05 and χ2(4) = 7.4, p > .05, respectively. The similarity of the pattern ismore striking considering neither group responded to either piece with amajority interpretation. The children’s verbal labels of these pieces were also nodifferent from either their photo choices, χ2(3) = 3.53, p > .05 or the adults’choices, χ2(3) = 4.91, p > .05 (for piece 7). For piece 7, both children and adultsmost often selected fear, followed by anger. For piece 8, both children andadults equally selected fear and anger. Children’s verbal responses included‘scarey’ followed by ‘mad’. Other responses to piece 8 included ‘monsters’ and‘haunted house’. In these cases, there was remarkable similarity between chil-dren and adults in the interpretation of the music. In agreement with previousreports (Gregory et al., 1996), these 4-year-olds did not respond above chanceto some pieces, perhaps implying that they confused angry and fearful pieces(Robazza et al., 1994; Terwogt and Van Grinsven, 1991). However, by ana-lyzing the pattern of their responses rather than the correctness of the majori-ty choice, the consistency between children and adults becomes apparent.

For three of the eight pieces of music, children’s and adults’ patterns ofchoices were different. For both pieces 2 and 3, most adults selected happy,but the children’s choices were not significantly different from an equal distri-bution across the five facial expressions for both piece 2 (χ2(4) = 2.01, p >.05) and piece 3 (χ2(4) = 5.17, p > .05). Likewise, most adults found piece 6


TA B L E 1 Percentage of children agreeing with adult majority choices for eight musical piecesa

Piece 1 Piece 2 Piece 3 Piece 4 Piece 6 Piece 7 Piece 8 Piece 9Age group happy happy happy neutral sad fear fear/anger anger

Adults 70 95 90 55 60 50 45b 50Children 61 26 38 17 22 35 29b 26

aPiece 5 had no majority interpretation for either age group and so is not included here.bThe majority interpretation of piece 8 was tied equally between fear and anger forboth adults (45% each emotion) and children (29% each emotion).

06Nawrot 2/12/02 10:06 am Page 80


F I G U R E 1 Percentage of adults (filled bars) and children (open bars) selecting each of thefive photographed facial expressions that matched the ‘happy’ music: piece 1 Mozart’s ViolinConcerto No. 5 (upper chart); piece 2 Mozart’s Violin Concerto No. 5 (middle chart); andpiece 3 Beethoven’s Symphony No. 6 (lower chart).

% o

f p

arti

cip

ants

ch

oo

sin

g e

ach

fac

e

happy sad anger fear neutralPiece 1: Mozart’s Violin Concerto No. 5


happy sad anger fear neutralPiece 3: Beethoven’s Symphony No. 6

06Nawrot 2/12/02 10:06 am Page 81

sad, while children again chose each photograph equally often (χ2(4) = 9.39,p = .05). However, in all three cases where the children failed to choose a sin-gle photograph to match the music, their verbal labels were no different fromthe adults’ results. Figure 4 illustrates this by comparing the distribution ofboth photo and verbal responses to piece 2. In the top chart, adults and chil-dren’s choice of photographs are different, with the children showing noclear consensus. The bottom chart shows the children’s verbal labels of thatsame piece of music. Now the adults’ and children’s pattern of responding isno different (χ2(4) = 2.19, p > .05) with 73 percent of children calling thispiece ‘happy’. In this case, the children’s verbal labels were a better match to


F I G U R E 2 Percentage of adults (filled bars) and children (open bars) selecting each of thefive photographed facial expressions that matched the ‘sad/neutral’ music: piece 4 Bach fromGould’s Goldberg Variations (upper chart), piece 6 Barber’s Adagio for Strings fromPlatoon (lower chart).

% o

f p

arti

cip

ants

ch

oo

sin

g e

ach

fac

e

happy sad fear anger neutralPiece 4: Bach from Gould’s Goldberg Variations

happy sad fear anger neutralPiece 6: Barber’s Adagio for Strings from Platoon

06Nawrot 2/12/02 10:06 am Page 82


F I G U R E 3 Percentage of adults (filled bars) and children (open bars) selecting each of thefive photographed facial expressions that matched the ‘anger/fear’ music: pieces 7–9 (upper,middle, lower charts), all taken from Prokofiev’s Alexander Nevsky.

% o

f p

arti

cip

ants

ch

oo

sin

g e

ach

fac

e

happy sad anger fear neutralPiece 7: Prokofiev’s Alexander Nevsky



06Nawrot 2/12/02 10:06 am Page 83

the adults’ interpretation; this is also the piece where the children’s verballabels were most different from their photo choices (χ2(4) = 6.56, p = .07).

An important methodological difference between this and most previousresearch was the use of a neutral facial expression choice. Gentile (1998)found that adults did not use the ‘no-emotion’ label on a forced-choice verbalresponse sheet, but he did not use this category for the children’s responses.In the present study, both children and adults chose the neutral face aboutequally often overall, although both groups chose this face most often in com-bination with the sad music (see Figure 2). The choice of a neutral face mightindicate that some pieces of music are not as transparent at communicatingan emotion (i.e. listeners might use this as a default choice when they areunsure). Or they might be using this neutral face to represent a less intenseemotion, especially when it comes to sad music. For example, children only


F I G U R E 4 Percentage of adults (filled bars) and children (open bars) matching a facialexpression to the happy music for piece 2. The top chart compares adults’ photograph choicesto children’s photograph choices, whereas the bottom chart shows children’s verbal labels.

% o

f p

arti

cip

ants

ch

oo

sin

g e

ach

fac

e

happy sad anger fear neutral


06Nawrot 2/12/02 10:06 am Page 84

generated neutral verbal labels, namely ‘fine’ and ‘not sad’, for piece 6, whichadults clearly felt was sad.

Despite these differences, both children and adults consistently judged theemotional quality for most of the musical selections. Study 2 was designed toaddress the question of emotional interpretation of music with infant listen-ers. An intermodal matching paradigm was used to test whether infantscould detect the common affect from paired visual and musical stimuli. Inprevious studies using this technique, infants looked longer to an affectivelymatching visual display with accompanying vocal expression of emotion(Soken and Pick, 1999), suggesting that they can extract an invariant emo-tional message in this type of situation.

Study 2

METHOD

ParticipantsInfants were recruited from birth announcements in the local paper. Parentsresponded to invitations and brought their infants to the lab where informedconsent was received. A sample of 26 healthy, full-term 5–9-month-oldinfants participated (15 girls; mean age of 31.8 weeks). Six additional infantswere dropped from the study due to fussy behavior.

StimuliFacial expressions The same female model who posed the static facial expres-sions for the photographs in Study 1 was videotaped for the happy and sadexpressions in this study. The model was videotaped from the shoulders upunder normal lighting while telling either a happy or sad story for two min-utes. The model practiced each story and viewed the recording for feedbackbefore making a final version for each emotion. This type of naturalistic affec-tive display was created similarly to other research with infants (Gentile,1998; Haviland and Lelwica, 1987; Soken and Pick, 1999).

Music Happy music piece 1 (from Mozart’s Violin Concerto No. 5) and sadmusic piece 4 (Bach from Gould’s Goldberg Variations), which reflected thechildren’s interpretations of happy and sad in Study 1, were used for Study 2.Each of the two 20-second pieces were digitally edited to loop seamlessly andcontinuously for approximately 2 minutes, the duration of each trial for theinfants.

ProcedureInfants were seated on their parent’s lap inside a 120-cm high three-sidedenclosure designed to reduce distraction and hide the recording equipment.Infants were approximately 45 cm from the display, which included two 30-cm color monitors side by side and two centrally located speakers beneath


06Nawrot 2/12/02 10:06 am Page 85

the monitors. Behind the display and occluded by the back wall of the enclo-sure, a video camera was positioned 138 cm above the monitors. A 2-cmaperture in the enclosure was used to view the infant and record lookingbehaviors for later scoring.

The infants were presented with four 2-minute trials consisting of affec-tively concordant and discordant music/facial expression pairs. In each trial,the infant listened to a single piece of music (either happy or sad) while view-ing the pair of dynamic facial expressions. The lateral position of thehappy/sad displays was counterbalanced across the four trials for eachinfant. The order of the first piece of music played was counterbalancedacross infants. However, for each infant the music was blocked so that thesame piece was played for the first two trials (repeated measures) followed bythe different piece for the last two trials. During the trials, the parents woreheadphones and listened to an unrelated piece of music to minimize the pos-sibility that their behavior might influence their infants.

An observer naive to the lateral position of the facial expression and themusical condition scored each infant’s looking behavior from videotape. Theobserver measured the total amount of looking time to each facial expressionfor the two music conditions across each of the repeated trials. Fixations offscreen were not included in the total looking time.

RESULTS AND DISCUSSION

Table 2 shows the mean looking time of the 26 infants to each of the fourconditions in Study 2. Because there was no effect of trial (t(25) = 1.52, p=.14), the looking times were averaged across the two trials in all analyses.Concordant displays were those where the music and facial expressionmatched (e.g. looking time to the happy expression when listening to thehappy music), while discordant displays were those where the music andfacial expression did not match (e.g. looking time to the happy expressionwhen listening to the sad music). Overall, the infants looked longer to thehappy displays than the sad displays in both the affectively concordant (M =71.73) and discordant conditions (M = 62.08).

The looking time data were further analyzed following Walker (1982) andSoken and Pick (1999). The proportion of total looking time (PTLT) was cal-culated for each of the affectively concordant and discordant pairs (see Table2). In each case, this was the amount of looking time to a particular expres-sion (e.g. happy expression during happy music) divided by the total lookingtime in that condition (e.g. happy expression + sad expression during happymusic). The PTLT for the affectively concordant and discordant conditionswere then compared using a matched sample t test (significance levels weretwo-tailed with α = .05). That is, the PTLT for the happy expression/happymusic (happy affectively concordant) and sad music (happy affectively discordant) was compared. Likewise, the PTLT to the sad expression/sadmusic (affectively concordant) and happy music (affectively discordant) was


06Nawrot 2/12/02 10:06 am Page 86

compared. Because this second comparison is identical to the first (the valuesare one minus the other), only one t test is reported. The infants’ proportionof looking time to the affectively concordant displays (happy M = .55, sad M = .46) was not significantly different from that to the affectively discordantdisplays (happy M = .45, sad M = .54; t(25) =.25, p > .05). However, infantslooked proportionally longer to the happy expression (M = .55) compared tothe sad expression (M = .45; t(25) = 3.18, p < .01).

Although the infants showed the predicted looking preferences only for theaffectively happy display (and not the affectively sad display), this preferenceis evidence for discrimination. Gentile (1998) reported decreased lookingtimes to happy displays following a change to a sad piece of music, conclud-ing that this was evidence for discrimination of the emotional content of themusic. The overall lower looking times and greater variability in the sadmusic condition compared to the happy music condition (see Table 2) suggestthat the infants might have been more distracted and possibly upset by thesad music.

These results support and extend Soken and Pick’s (1999) finding that 7-month-olds discriminated between happy and sad facial and emotionalexpressions. As in this study, the infants in their study did not show differ-ences in looking time to the affectively concordant and discordant displays,but they did show a significant preference for the happy expression. Alongwith Gentile’s (1998) report, this is one of the first studies to extend the inves-tigation of emotional interpretation in music from children to infants.

General discussion

Taken together, the results from Study 1 and 2 begin to address importantissues in the development of affective interpretation in music. In agreementwith previous research, Study 1 found that children as young as 4 years


TA B L E 2 Mean total looking time (TLT) and proportion total looking time (PTLT) to thehappy and sad facial expressions in concordant and discordant music conditions in Study 2(SDs in brackets)

Affective display type TLT PTLT

Happy expressionAffectively concordant music 71.73 (22.40) .55a

Affectively discordant music 62.08 (25.29) .45

Sad expressionAffectively concordant music 54.15 (24.00) .46Affectively discordant music 58.35 (21.38) .54

a Infants looked proportionally longer at the happy expression, t(25) = 3.18, p < .01.

06Nawrot 2/12/02 10:06 am Page 87

could interpret discrete emotions by matching music to facial expressions(Dolgin and Adelson, 1990; Gentile, 1998; Terwogt and Van Grinsven,1991). Children’s responses to all eight musical pieces were similar to adultswhen both forced-choice photograph and child-generated verbal labels wereanalyzed. However, unlike some previous reports (e.g. Robazza et al., 1994),children’s judgments of anger and fear in music were similar to those ofadults. Important differences in methodology could explain this, as could pre-vious findings of children’s apparent inconsistencies. For example, children’slow levels of consensus in this and previous reports might be the result ofboth task demands and analysis techniques. The use of a five-alternativeforced-choice task, while it has obvious ecological value, increases inconsis-tencies in children’s responding (Dolgin and Adelson, 1990; Kratus, 1993).This, along with a comparison to adults’ correct interpretation, may increasethe apparent difference between children and adults. Under these conditions,it may be beneficial to analyze the pattern of responding. When the adults’and children’s distribution of choices were compared, their results were moreconsistent. When inconsistencies did arise, the children’s verbal labelsappeared to be a better match to the adults’ responses.

As Gentile (1998) points out, inconsistencies may also be the result of thedifferent communicative value of the particular piece of music. Some piecesmay be more emotionally transparent than others, which could be due to anynumber of reasons ranging from the composition (composer’s intent to com-municate emotion), to the performance (performance media or quality), tothe listener (task demands, listener’s musical expertise or age). In Study 1,adults varied from 45–95 percent agreement in their responses to the musi-cal pieces, demonstrating the importance of careful choice of music. Theinfluence of the listener, too, is apparent. In the case of anger, the listenermay be responding to the expression of anger in the musical composition, orto his or her own reaction of fear. In order to clarify these alternatives, it isimportant to understand how listeners processed the task, especially withregard to potential age differences. The use of a multiple-alternative forcedchoice may complicate the task more for children who tend to respond froman egocentric perspective.

The results of Study 2 support and extend previous research on infants’perception of affect (Soken and Pick, 1998). Results indicated that infants didshow discrimination in the intermodal display; however, their overall prefer-ence for the happy expression warrants a cautious conclusion regarding theirunderstanding of the emotional content of the music specifically. Their look-ing preference during the sad condition, along with their overall decrease inlooking time during the sad music, does indicate a discrimination, but it wasnot in the predicted direction in this intermodal matching task. Soken andPick (1999) explain their findings that infants preferred the happy expres-sions using the familiarity hypothesis which states that increased exposure toa particular expression promotes learning and discrimination. This may also


06Nawrot 2/12/02 10:06 am Page 88

explain why the infants in Gentile’s (1998) study were able to demonstratediscrimination during the affect change conditions as they were all first famil-iarized to 3 minutes of happy faces and happy music.

In the present study, as in Gentile’s (1998) research, infants looked lessduring the sad music and looked less at the sad displays. However, for anintermodal matching task, it is reduced looking time to discordant pairs, notjust lower overall looking time, that is necessary to demonstrate discrimina-tion. It is likely that the infants’ lower overall looking during the sad musicmight have prevented them from attending to the intermodal display, thusmasking their discrimination abilities during these trials. Alternatively, theinfants’ preference for the happy expression may simply have reflected howthe dynamics of the model’s movements captured their attention. A compari-son of happy versus angry emotions would help to rule out this possibilityand could also help identify the component of the music (e.g. tempo) that isassociated with the emotion.

Previous studies of infants’ perception of music have focused on decon-structing the music into basic structural elements to determine what, if any,musical component gives rise to a predictable reaction (e.g. looking prefer-ence). If such an element is found early in life, it might suggest an innatemechanism for the detection of affect. One candidate for such a basic dimen-sion is consonance (Crowder et al., 1991; Trainor and Heinmiller, 1998;Zentner and Kagan, 1998). But the preference could be based on sensoryconsonance resulting from a processing bias in the auditory system, or psy-chological consonance resulting from a learned preference for tonally pleas-ing notes. Even 6-month-olds could have learned such a preference fromaccumulated experience with music through caregivers’ lullabies. In eithercase, it is important that additional studies investigate infants’ ability todetect discrete emotional messages from music.

Future research could also address whether infants can categorize emo-tion across different pieces of music in the way that they categorize the emo-tional tone of facial expressions (e.g. Kestenbaum and Nelson, 1990).Infants’ categorical perception of speech sounds, color, and even some com-ponents of music such as octave equivalence (Demany and Armand, 1984)suggest a systematic representation of the perceptual world. This mightreflect a predisposition inherent in the perceptual systems for vision and audi-tion (Lynch et al., 1992), which could arise from the inherent stimulus struc-ture. The ecological approach would predict that musically naive listeners,including young infants, should be able to detect emotional invariants(Gentile, 1998), as this ability is sure to have evolutionary value as a meansof emotional communication (Trainor and Heinmiller, 1998). Indeed, thesimilarities in music processing between adults and infants (Trehub andSchellenberg, 1995) may also suggest that infants could interpret emotion inmusic in a predictable way.

Lynch et al. (1992) concluded that the perception of music by infants


06Nawrot 2/12/02 10:06 am Page 89

younger than one year is unaffected by musical acculturation. Indeed, infantlisteners often outperform adults (even musically trained adults) on some dis-crimination tasks, such as detecting mistunings in unfamiliar and unequalstep scales (Trehub et al., 1999). This suggests that development involvesboth tuning of the perceptual mechanisms as well as some loss of sensitivity,as in the case for the perception of speech sounds (Eimas, 1985). If the effectsof culture and experience accumulate on music perception during childhood,it is even more important to recognize that the emotional quality of musicmay not be the same for an infant as an adult.


This research was supported by a Faculty Improvement Grant from Minnesota StateUniversity Moorhead. I would like to thank Bob Schieffer, Amanda Lindseth and DrLuke Howard for help with stimulus generation, Dr Douglas Gentile for information onhis research and Dr Mark Nawrot. I would also like to thank the student researchteam including Christopher Geisler, Kerri Stenseth, Shelley Cervantes, TashaMcGregor, Meredith Haugen and Sonia Balliet for help with data collection. The com-ments of two anonymous reviewers are gratefully appreciated. Portions of thisresearch were presented at the American Psychological Society meetings inWashington DC, May 1998 and Denver, CO, May 1999 and appeared in the BrownUniversity Child and Adolescent Behavior Letter, November 1999.

R E F E R E N C E S

Crowder, R.G., Reznick, S.J. and Rosenkrantz, S.L. (1991) ‘Perception of the Major/Minor Distinction: V. Preferences among Infants’, Bulletin of the PsychonomicSociety 29(3): 187–8.

Cunningham, J.G. and Sterling, R.S. (1988) ‘Developmental Change in theUnderstanding of Affective Meaning in Music’, Motivation and Emotion 12(4):399–413.

Demany, L. and Armand, F. (1984) ‘The Perceptual Reality of Tone Chroma in EarlyInfancy’, Journal of the Acoustical Society of America 76: 57–66.

Dolgin, K.G. and Adelson, E.H. (1990) ‘Age Changes in the Ability to Interpret Affectin Sung and Instrumentally Presented Melodies’, Psychology of Music 18: 87–98.

Eimas, P.D. (1985) ‘The Perception of Speech in Early Infancy’, Scientific American204: 66–72.

Fernald, A. (1989) ‘Intonation and Communicative Intent in Mothers’ Speech toInfants: Is the Melody the Message?’, Child Development 60: 1497–1510.

Gentile, D. (1998) ‘An Ecological Approach to the Development of Perception ofEmotion in Music’, Dissertation Abstracts International 59 (5-B): 2454, University ofMinnesota.

Gregory, A.H., Worrall, L. and Sarge, A. (1996) ‘The Development of EmotionalResponses to Music in Young Children’, Motivation and Emotion 20(4): 341–9.

Haviland, J. and Lelwica, M. (1987) ‘The Induced Affect Response: 10-week-oldInfants’ Responses to Three Emotional Expressions’, Developmental Psychology 23:97–104.

Kastner, M.P. and Crowder, R.G. (1990) ‘Perception of the Major/Minor Distinction:IV. Emotional Connotation in Young Children’, Music Perception 8(2): 189–201.


06Nawrot 2/12/02 10:06 am Page 90

Kestenbaum, R. and Nelson, C.A. (1990) ‘The Recognition and Categorization ofUpright and Inverted Emotional Expressions by Seven-Month-Old Infants’, InfantBehavior and Development 13: 497–511.

Kratus, J. (1993) ‘A Developmental Study of Children’s Interpretation of Emotion inMusic’, Psychology of Music 21: 3–19.

Lynch, M.P., Eilers, R.E. and Bornstein, M.H. (1992) ‘Speech, Vision, and MusicPerception: Windows on the Ontogeny of Mind’, Psychology of Music 20: 3–14.

Nelson, C.A. (1985) ‘The Perception and Recognition of Facial Expressions inInfancy’, in T. Field and N. Fox (eds) Social Perception in Infancy, pp. 101–25.Norwood, NJ: Ablex.

Robazza, C., Macaluso, C. and D’Urso, V. (1994) ‘Emotional Reactions to Music byGender, Age, and Expertise’, Perceptual and Motor Skills 79: 939–44.

Rock, A.M.L., Trainor, L.J. and Addison, T.L. (1999) ‘Distinctive Messages in Infant-Directed Lullabies and Play Songs’, Developmental Psychology 35(2): 527–34.

Schellenberg, E.G. and Trehub, S.E. (1996) ‘Natural Music Intervals: Evidence fromInfant Listeners’, Psychological Science 7(5): 272–7.

Soken, N.H. and Pick, A.D. (1999) ‘Infants’ Perception of Dynamic AffectiveExpressions: Do Infants Distinguish Specific Expressions?’, Child Development70(6): 1275–82.

Spelke, E.S. and Owsley, C.J. (1979) ‘Intermodal Exploration and Knowledge inInfancy’, Infant Behavior and Development 2: 13–27.

Terwogt, M.M. and Van Grinsven, F. (1991) ‘Musical Expression of Mood States’,Psychology of Music 19: 99–109.

Trainor, L.J. and Heinmiller, B.M. (1998) ‘The Development of Evaluative Responsesto Music: Infants Prefer to Listen to Consonance over Dissonance’, Infant Behaviorand Development 21(1): 77–88.

Trainor, L.J., Austin, C.M. and Desjardins, R.N. (2000) ‘Is Infant-Directed SpeechProsody a Result of the Vocal Expression of Emotion?’, Psychological Science 11(3):188–95.

Trehub, S.E. and Schellenberg, E.G. (1995) ‘Music: Its Relevance to Infants’, Annals ofChild Development 11: 1–24.

Trehub, S.E. and Trainor, L.J. (1993) ‘Listening Strategies in Infancy: The Roots ofMusic and Language Development’, in S. McAdams and E. Bigand (eds) Thinking inSound: The Cognitive Psychology of Human Audition, pp. 278–327. Oxford: OxfordUniversity Press.

Trehub, S.E., Schellenberg, E.G. and Kamenetsky, S.B. (1999) Infants’ and Adults’Perception of Scale Structure’, Journal of Experimental Psychology 25(4): 965.

Trehub, S.E., Thorpe, L.A. and Morrongiello, B.A. (1987) ‘Organizational Processes inInfants’ Perception of Auditory Patterns’, Child Development 58: 741–9.

Walker, A.S. (1982) ‘Intermodal Perception of Expressive Behaviors by HumanInfants’, Journal of Experimental Child Psychology 33: 514–35.

Walker-Andrews, A.S. and Grolnick, W. (1983) ‘Discrimination of Vocal Expressionsby Young Infants’, Infant Behavior and Development 6: 491–8.

Walker-Andrews, A.S. and Lennon, E. (1991) ‘Infants’ Discrimination of VocalExpressions: Contributions of Auditory and Visual Information’, Infant Behaviorand Development 14: 131–42.

Zentner, M.R. and Kagan, J. (1998) ‘Infants’ Perception of Consonance andDissonance in Music’, Infant Behavior & Development 21(3): 483–92.


06Nawrot 2/12/02 10:06 am Page 91

E L I Z A B E T H S . N AW RO T is currently an Associate Professor of Psychology atMinnesota State University in Moorhead; she received her undergraduate degree fromCarnegie Mellon University in 1988, followed by a PhD in Psychology from theUniversity of California, Berkeley in 1992. She then completed two years of post-doctoral work in neuropsychology at the University of Iowa Hospitals and Clinics. Herresearch background is the study of developmental psychology and visual psy-chophysics. She has published several papers in the area of visual perception, includ-ing studies of infant color vision and adaptation, and the effects of brain damage onvisual function. Address: Department of Psychology, Minnesota State University, 1104 7th StreetSouth, Moorhead, MN 56563, USA. [email: [email protected]]


06Nawrot 2/12/02 10:06 am Page 92

The effect of music on perceivedatmosphere and purchase intentions in a restaurant

93A R T I C L E

Psychology of Music



Researchvol (): ‒

[- ():; ‒; ]

S T E P H A N I E W I L S O NU N I V E R S I T Y O F N E W S O U T H WA L E S

A B S T R A C T Extending research by North and Hargreaves (1998), this studyinvestigated the effect of music on perceived atmosphere and purchase intentionsin a restaurant. Four musical styles (jazz, popular, easy listening and classical)and no music were played in a restaurant over two consecutive weeks. Resultsindicated that different types of music had different effects on perceived atmos-phere and the amount patrons were prepared to spend. Classical, jazz and popu-lar music were associated with patrons being prepared to spend the most on theirmain meal. This value was found to be significantly lower in the absence of musicand when easy listening was played. There was some evidence that the type ofmusic also had an effect on the amount of money patrons actually spent in therestaurant. Overall, the study contributes to the development of a model thatseeks to account for the relationship between music and consumer behaviour.

K E Y W O R D S : consumer behaviour, listening, music psychology, perceived atmosphere, restaurant

Introduction

Despite the amount of money spent on music-related resources in the com-mercial sector, research which investigates relationships between music andconsumer behaviour is relatively sparse. Studies in social psychology havedemonstrated a range of interactions between music and the social context inwhich it is heard (Fried and Berkowitz, 1979; Stratton and Zalanowski,1984; Zillmann and Bhatia, 1989; Bleich et al., 1991; Zullow, 1991; Stackand Gunlach, 1992; Standley, 1995). Few studies, however, have consideredthe influence of music in everyday commercial environments such as restau-rants and other retail outlets. Research in this area has the potential to pro-vide commercial practitioners with guidelines regarding how to enhance

sempre :

business and increase sales through the use of music, in addition to making acontribution to an important area of social psychology.

THE EFFECT OF MUSIC ON PERCEIVED ATMOSPHERE

In 1996, North and Hargreaves studied the effect of music on atmosphere ina university cafeteria (North and Hargreaves, 1996). Results suggested thatcustomers’ ‘liking’ of the cafeteria increased with their ‘liking’ for the musicplayed within it. In a follow-up study, North and Hargreaves (1998) investi-gated whether affective responses to music of different styles could influencethe perceived characteristics of a commercial environment. Results suggestedthat different musical conditions influenced subjects’ perceptions about thecafeteria. Furthermore, the nature of subjects’ perceptions of the cafeteriashowed a relationship to their perception of the music.

North and Hargreaves (1998) stated that the implications of their studyneeded to be validated by future research. In response, the present studyexplores the effect of music on perceived atmosphere and purchase intentionsin a restaurant. While the methodology employed in the North andHargreaves study is maintained to a large extent, the present research designhas been expanded to incorporate a broader range of musical conditions.Further, as suggested by North and Hargreaves, the present study utilizes anon-student subject sample.

THE EFFECT OF MUSIC ON PRODUCT CHOICE AND PURCHASE INTENTIONS

A number of studies suggest that music has the potential to influence prod-uct choice (Alpert and Alpert, 1990; Areni and Kim, 1993). The idea thatmusic conveys and triggers relevant information that may prime consumers’beliefs about a product has been described as musical ‘fit’. Areni and Kim(1993) applied the idea of musical ‘fit’ to shopping behaviour by playing classi-cal music and Top-40 music in a wine store. The results showed that classicalmusic led to consumers purchasing more expensive wine. These results sup-port MacInnis and Park’s (1991) notion that persuasion is enhanced whenthe music is appropriate for the context in which it is played, and Yalch andSpangenberg’s (1990) suggestion that classical music is associated with theperception of higher-priced store items.

The effect of musical tempo on the speed of consumer behaviour and theamount of money spent has also been investigated. For example, Milliman(1982) found that slower music was associated with a slower shopping pace,and increased gross sales. In a further study, Milliman (1986) tested the effectof fast and slow music in an ‘upscale’ restaurant. The results showed thatdiners ate more quickly when fast music was playing (see also Robally et al.,1985). On the nights when slow music was playing, customers spent signifi-cantly more time in the restaurant and more money on alcoholic beverages.

Bruner (1990) suggests that the style of music played is likely to have a more significant effect on customers’ perceptions and choices. This is


supported by Yalch and Spangenberg (1990), who investigated the influenceof different styles of music on customers’ estimates of the amount of timethey spent shopping. The study suggested that consumers who are exposed to‘non-typical’ stimuli in the environment (for example, younger customersencountering easy listening music) tend to overestimate time durations.

Several studies of in-store music have drawn on an environmental psy-chology model proposed by Mehrabian and Russel (1974). This model statesthat people respond to environments according to two primary factors: pleas-ure and arousal. In a more pleasurable environment consumers are morelikely to demonstrate ‘approach behaviours’ towards it. In turn, an environ-ment with arousal-evoking qualities is likely to enhance the effects of pleas-ure. Dube et al. (1995) demonstrated the effect of pleasure and arousalinduced by music on consumers’ desire to engage in buyer–seller interactionsin a bank. Similarly, North and Hargreaves (1996) found that ‘liked’ musicwas positively related to patrons’ willingness to return to a dining area, andtheir willingness to interact with others.

The literature reviewed above provides evidence that music can influencefactors such as the speed of consumer activity, product choice, customers’desire to affiliate, and their estimation of time. One area that has been givencomparatively little attention in the literature is the extent to which musicmight actually influence the amount of money patrons are prepared tospend. Introductory evidence has been provided by North and Hargreaves(1998), who found that the type of music played in a cafeteria influenced theamount of money patrons were prepared to spend. In addition, they foundthat sales figures for the cafeteria were significantly higher when classicaland popular music were played.

AWARENESS AND APPROPRIATENESS

It has been suggested that people are likely to spend more time and money ina restaurant or retail environment if the music being played is consideredappropriate (Radocy and Boyle, 1997). Therefore, patrons in the presentstudy were asked to indicate whether they thought the music being played inthe restaurant was appropriate. Responses are considered in terms ofpatrons’ perception of atmosphere and the amount of money they are willingto spend. Patrons were also asked to rate their level of awareness of the music being played. Based on the characteristics of stimulative and sedativemusic described by Radocy and Boyle (1997), it was predicted that subjects’awareness of the music would be greater in the popular, jazz and control conditions.

Aim

To extend the literature on perceived atmosphere and purchase intentions inthe commercial environment, the present study has four aims:

Wilson: The effect of music in a restaurant 95

1. To investigate the extent to which the type of music being played influ-ences patrons’ perception of a restaurant environment;

2. To examine the influence of different musical styles on the amount ofmoney patrons are prepared to spend, and the amount of money theyactually spend;

3. To investigate interactive effects between music, perceived atmosphereand a range of other variables such as the number of people dining, thenumber of times patrons have been to the restaurant before, the amountof alcohol consumed and patrons’ perception of the quality of food andservice; and

4. To examine whether the appropriateness of the music influences patrons’perception of the restaurant and the amount of money they spend.

Method

PARTICIPANTS

All patrons present in the restaurant during the testing period were eligiblefor the study. The sample comprised 300 subjects, that is, a total of 50 sub-jects for each condition over the 12-day testing period. Testing was carriedout between 7.30 pm and 11.30 pm from Monday to Saturday over the twoweeks. Of the 300 subjects, 45.4 percent were male and 54.6 percent werefemale. The distribution of participants according to gender and age is pre-sented in Figure 1. As shown, the majority of subjects were aged between 20and 39 years.

Patrons were approached at their tables at the end of their meal and asked to


F I G U R E 1 Distribution of participants by age and gender.

%

0

5

1 0

1 5

2 0

2 5

3 0

<10 yrs 10–19 yrs 20–29 yrs 30–39 yrs 40–49 yrs 50–59 yrs 60–69 yrs 70 yrs

AgeMale Female

complete a questionnaire about the restaurant. The first questionnaire of theevening was not administered until the music had been playing for at least 30minutes to ensure that participants had sufficient exposure before responding.

DESIGN

The experiment was conducted at Out Of Africa, a popular Sydney restaurant.The choice of restaurant was governed by the following criteria: (a) seatingcapacity >100; (b) high-quality stereo system and speakers; (c) close com-petitors with other restaurants in the area; and (d) a diverse range of clien-tele. In week one the presentation of conditions occurred in the followingorder: jazz, no music, classical, easy listening, control and popular. This orderwas changed in the second week (classical, jazz, popular, control, no music,easy listening) to reduce the effect associated with day of the week.

The experiment was based on a time series design with the aim of examin-ing the intervention of a series of conditions. As shown, a control group wasincorporated by administering questionnaires under ‘normal conditions’ ontwo nights during the testing period (where restaurant owners would playthe music they would normally play). The control condition comprised a mix-ture of world music (played on CD), and live music (a 3-piece African band).While the live music component may be seen as reducing the homogeneity ofthe control as a single condition, its inclusion was viewed as providing a morebalanced and accurate representation of what was normally played (andexperienced) in the restaurant.

MATERIALS

With the exception of the ‘no-music’ condition and the non-interventioncontrol, each condition employed several hours of music that was previouslyrecorded onto a series of CDs (see Appendix). The four musical styles present-ed included classical, popular, easy listening and jazz. Following North andHargreaves (1998), the pieces selected for the study were typical examples ofthe style.

On each night during the study, the music was played on a high-qualityCD player through four speakers that were suspended in each corner of therestaurant. The volume of the music was held constant, and at a level whereit was clearly audible while still allowing patrons to talk over it comfortably.

QUESTIONNAIRES

Two separate questionnaires were designed for the study. The first, completedby restaurant management prior to testing, was designed to collect informa-tion about the type of music typically played, and details of the consistency ofother factors in the restaurant (for example, lighting, decor, service, menu)which might influence the results of the study. The second questionnaire,described below, was administered during the testing period to restaurantpatrons.


PATRON QUESTIONNAIRE

The first section of the questionnaire asked patrons to indicate the number oftimes they had been to the restaurant previously, their reason for dining, thenumber of people they were dining with, and whether they thought theywould return to the restaurant. This information was obtained to allow amore detailed investigation of the interaction between music and consumerbehaviour. For example, we might expect these extraneous factors to enhanceor reduce the impact of music on patrons’ perception of the restaurant envi-ronment, and the amount they are prepared to spend. These early questionsalso acted as ‘distracters’ – the intention being that subjects would be lessinclined to think that the survey was about the music being played. Followingthis, subjects were asked to rate the characteristics of the restaurant accord-ing to a list of 20 adjectives. For the purpose of comparison, the adjectivesused in the study were identical to those used by North and Hargreaves(1998). Patrons made their responses on an 11-point scale (0 = ‘the restaurant definitely does not possess this characteristic’ to 10 = ‘the restau-rant definitely does possess this characteristic’). Subjects were then asked to rate the quality of food and service in the restaurant, the extent to whichthey were aware of the music being played (0 = ‘not at all aware’ to 5 =‘extremely aware’), and whether they thought the music being played wasappropriate.

The second section of the questionnaire asked patrons to respond to themusic being played. This section was answered by all subjects with the excep-tion of those in the ‘no music’ group. Subjects were asked to rate the musicaccording to the same set of adjectives used previously so that the relation-ship between music and perceived atmosphere could be examined directly.Once again, patrons responded on an 11-point scale (0 = ‘the music definitelydoes not possess this characteristic’ to 10 = ‘the music definitely does possessthis characteristic’). As a means of examining purchase intention, the finalsection of the questionnaire asked patrons to indicate the maximum amountof money they were prepared to spend on their main meal.

Results

PERCEIVED CHARACTERISTICS OF THE RESTAURANT

Differences between the four musical styles, no music and the control condi-tion on participants’ ratings of atmosphere were investigated using a MANO-VA. The overall difference between musical styles was found to be significant,F(294, 5) = 12.55, p < .0001.

Table 1 summarizes differences between the conditions for subjects’ rat-ings of atmosphere. The results of Tukey HSD post-hoc tests are also present-ed, indicating several significant differences between conditions.

The relationship between responses to the restaurant and responses to themusic for each adjective was investigated by a series of correlations. The


result for each adjective was found to be significant at p = .001, with r valuesranging between .23 and .52.

A factor analysis was conducted to further examine subjects’ responses tothe restaurant according to the 20 adjectives. Varimax rotation of the princi-pal components solution generated five factors with eigenvalues greater than1. In total, these factors accounted for 61.3 percent of the variance inpatrons’ responses. Table 2 presents the details of factor loadings greaterthan ± .30.

These loadings led to Factor 1 being interpreted as upbeat; Factor 2 aspeaceful/passive; Factor 3 as tacky; Factor 4 as invigorating/stimulating; andFactor 5 as upmarket/sophisticated. Differences between the musical condi-tions on each factor were explored using factor scores. The results of one-wayANOVAs and Tukey HSD post-hoc tests are reported in Table 3.

These results suggest that different musical styles produced differences in the general perceived characteristics of the restaurant. For example, no music was associated with the restaurant being perceived as the leastupbeat, classical music with the restaurant being perceived as the most


TA B L E 1 MANOVA and Tukey tests concerning the effects of music on the perceived characteristics ofthe restaurant

M

No EasyCharacteristic music listening Classical Jazz Popular Control F p

Upmarket 4.77abcde 6.44c 6.36b 7.02a 6.34e 6.68d 6.98 < .0001Sophisticated 4.33abcde 6.17c 6.13b 6.71a 6.29e 6.61d 7.35 < .0001Happy 6.93abcd 8.12b 7.78e 8.22a 8.20d 8.87ce 5.74 < .0001Restful 5.77 5.25 6.56a 6.32 4.66a 5.92 3.11 < .05Fresh 6.63 7.47 7.71 7.56 7.00 7.68 1.82 .11Exciting 6.37a 7.49 6.78 7.20 7.17 7.71a 3.09 < .05Rebellious 4.27 3.90 4.87 4.68 3.97 4.71 1.50 .19Cerebral 4.13 4.66 4.69 3.80 3.54 a 4.77 a 2.26 < .05Feminine 2.97b 3.78d 4.49abc 3.02a 2.14cde 3.55e 5.39 < .0001Fun 6.43a 7.58 6.60b 7.24 7.40 8.01ab 4.34 < .05Downmarket 2.37 2.29 1.80 1.93 1.94 2.07 0.45 .81Youthful 5.73 6.69 6.40 5.76 6.66 6.63 1.68 .14Peaceful 4.77 3.47ac 6.18cde 5.44ab 3.37be 4.00d 8.89 < .0001Spiritual 3.13ad 3.03beg 5.49def 5.17abc 3.29cf 4.54g 8.33 < .0001Tacky 2.43 a 2.24 1.40 1.17 a 1.51 1.28 2.34 < .05Masculine 4.07 3.58a 4.87c 5.34ab 3.14bcd 4.61d 5.23 < .0001Invigorating 4.80a 4.81b 5.31 5.63 4.80c 6.28abc 3.82 < .05Aggressive 2.37 3.15a 2.04 2.44 2.29 1.86a 1.89 .09Fashionable 6.20 6.47 6.67 6.90 7.03 6.68 0.65 .66Sensual 4.57b 4.66ac 5.24cd 6.00a 6.62abd 6.62bce 16.64 < .0001

Note. Means marked by similar subscripts differ at p = .05 for each characteristic.

upmarket/sophisticated, and popular music with the restaurant being per-ceived as the most upbeat. Jazz music was associated with the restaurant beingperceived as the least peaceful/passive and the most invigorating/stimulating,and easy listening with the restaurant being perceived as the most tacky.


TA B L E 2 Factor analysis of responses to the restaurant environment

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5Characteristic loading loading loading loading loading

Upmarket 0.77Sophisticated 0.71Happy 0.83Restful 0.71Fresh 0.62Exciting 0.79Rebellious 0.65Cerebral 0.38 0.47Feminine 0.68Fun 0.90Downmarket –0.60Youthful 0.72Peaceful 0.79Spiritual 0.42 0.45Tacky 0.74Masculine 0.37 0.55Invigorating 0.79Aggressive 0.83Fashionable 0.34 0.35Sensual 0.62Eigenvalue 5.61 2.46 1.81 1.29 1.07% of variance 28.1 12.3 9.0 6.5 5.4

TA B L E 3 One-way ANOVAs and Tukey HSD tests to investigate differences between factor scores andconditions

M

No EasyFactor music listening Classical Jazz Popular Control F p

1. Upbeat –0.50abc 0.16ad –0.43def –0.26g 0.31beg 0.18cf 5.94 < .00012. Peaceful/passive –0.14 0.06 0.21 –0.37 –0.26 0.22 2.49 .033. Tacky 0.26 0.36a –0.98 –0.17 –0.16 –0.24a 3.01 < .054. Invigorating/

stimulating –0.14a –0.42bd 0.06 0.59abc –0.44ce 0.25de 6.95 < .00015. Upmarket/

sophisticated 0.17 –0.22c 0.63cde 0.36ab –0.23be –0.28ad 7.82 < .0001

Note. Means marked by similar subscripts differ at p = .05.

PURCHASE INTENTIONS

Based on an ANOVA, the effect of music on the maximum amount of moneypatrons were prepared to spend on their main meal was also significant (p =.001). The no music condition produced the least maximum price subjectswere willing to pay (Aus$17.12). Tukey HSD tests showed that this amountwas significantly lower (p = .05) than every other condition: easy listeningAus$19.67, classical Aus$20.20, control Aus$20.63, popular Aus$21.01,and jazz Aus$21.82. Easy listening produced the next lowest amount subjects were willing to pay, and this was significantly different from the jazzcondition.

It could be expected that patrons who had been to the restaurant morethan once might be familiar with restaurant prices, and that this may haveinfluenced the maximum price they were prepared to spend. However, no correlation was found (p = .05). Similarly, while it might be expected thatsubjects with a higher income would be prepared to spend more money ontheir main meal, this relationship was not found to be significant (p = .05).

SALES FIGURES

Restaurant sales figures were obtained during the testing period, and for thesame days two weeks before and after testing. The variation in sales figures inthe weeks before and after the testing period made it difficult to determine theinfluence of the intervening musical conditions on the amount of moneyspent. For example, a chi-square goodness-of-fit test revealed significant dif-ferences in sales over the six weeks. However, on four of these days, the high-est and lowest figures occurred outside the testing period. This suggests thatthe number of people dining in the restaurant was largely responsible for differences in sales figures. On the remaining days, sales were lower whenclassical music was played, and higher when popular music was played thanon the same days before and after testing. As information regarding the fre-quency of patrons was not available to the researcher, it is impossible to drawconclusions regarding the relationship between the type of music played andthe amount of money patrons actually spent. However, there are a number ofother factors which suggest a relationship between sales figures and the typeof music played.

Due to the fact that all subjects completed the questionnaire while waitingfor their bill, we can assume that no more purchases were made after thistime. The final item on the questionnaire asked subjects to indicate what timeit was. This information provided some indication of the amount of time (andperhaps money) subjects spent in the restaurant, or at least which conditionsmay have contributed to patrons remaining in the restaurant later in theevening.

Table 4 shows that classical music was associated with relatively few people remaining in the restaurant after 11 pm, and a greater number ofsubjects leaving the restaurant earlier in the evening, between 8 pm and


10 pm. An ANOVA showed that the time subjects filled out the questionnairewhen classical music was played differed significantly from all other condi-tions (p = .05). This implies that the type of music being played had an effecton how long patrons remained in the restaurant.

SITUATIONAL INFLUENCES ON THE PERCEPTION OF ATMOSPHERE

In addition to responding to the characteristics of the restaurant and themusic being played, subjects were asked to provide details such as (a) thenumber of people they were dining with; (b) the reason they were dining out; (c) the amount of alcohol they had consumed; (d) the number of timesthey had previously dined at the restaurant; and (e) the quality of food andservice.

Patrons were asked to indicate the number of people they were dining with(including themselves) on a 5-point scale (1 = one to 5 = five or more people).Based on an ANOVA, the relationship between the number of people diningand patrons’ responses to the characteristics of the restaurant overall was notsignificant. However, the number of people dining did yield a significantresult for the adjectives ‘cerebral’ and ‘aggressive’ (p = .05). In both the jazzand control conditions, mean responses for these adjectives increased incre-mentally with the number of people dining.

It was considered that patrons dining with friends or family might responddifferently to the environment than patrons attending a business meeting or function. However, based on an ANOVA, subjects’ reason for dining out did not produce significant differences in their responses to the restaurant (p = .05).

The figures reported in Table 5 suggest that the more upbeat styles ofmusic were associated with a greater number of people consuming three ormore drinks. Only a relatively small proportion of subjects (11.1%) consumedthree or more drinks on the nights when classical music was played. AnANOVA showed that the number of drinks consumed differed significantlybetween classical music and every other condition except no music (p = .05).

A MANOVA was used to investigate the interaction between the number of


TA B L E 4 Frequency distribution of subjects (%) according to the time the questionnaire wascompleted

Condition 8–9 pm 9–10 pm 10–11 pm After 11 pm

Jazz 2.44 21.95 70.73 4.88No music 3.33 35.00 50.00 11.67Classical 6.67 37.78 53.33 2.22Easy listening 0.00 15.25 74.58 10.17Control 5.63 16.90 64.65 12.82Popular 0.00 2.86 87.14 10.00

drinks consumed, subjects’ perception of the atmosphere, and musical condi-tion. The amount of alcohol consumed did not significantly affect patrons’responses to atmosphere overall. However, an interactive effect was foundbetween the number of drinks consumed, the type of music played, and sub-jects’ responses to ‘invigorating’ and ‘masculine’ (p = .001). Specifically, thejazz condition produced significant correlations (p = .001) between the num-ber of drinks consumed and the following adjectives: invigorating, masculine,sophisticated, rebellious and cerebral. In each case, mean ratings increasedwith the amount of alcohol consumed.

Patrons’ responses to the characteristics of the restaurant were also con-sidered in terms of the number of times they had been to the restaurantbefore. While there was no significant correlation overall, the number oftimes subjects had been to the restaurant was found to correlate with theirperception of the restaurant as ‘fun’. This relationship was found to be signifi-cant in the jazz and control conditions only (p = .05).

Patrons’ responses to the quality of service indicated that 92.2 percent ofsubjects rated the service in the restaurant as ‘slightly above average’ or high-er. A MANOVA was used to investigate the interaction between musical con-dition, patrons’ responses to service, and their responses to atmosphere.Overall, a significant effect (p = .001) was found between patrons’ responsesto service and their responses to the restaurant. More specifically, significantresults were found for the following adjectives: upmarket, sophisticated,happy, fresh, exciting, fun and fashionable. The jazz condition yielded a signif-icant correlation between patrons’ responses to service and their responses to‘happy’, ‘fun’ and ‘fresh’, while popular music generated a significant corre-lation between patrons’ responses to service and their responses to ‘exciting’.That is, the higher the rating for service, the higher the rating responses forthese adjectives. Classical, easy listening, no music and the control conditiondid not yield significant correlations between patrons’ responses to serviceand the characteristics of the restaurant.

It was considered that the time of night subjects filled out the question-naire may have influenced their ratings of the restaurant. Based on anANOVA, a significant effect was found between the time subjects filled out thequestionnaire and their responses to ‘invigorating’ (p = .001). Inferential statistics showed that rating responses for ‘invigorating’ were lower when


TA B L E 5 Number of alcoholic beverages consumed (%)

EasyNo. of drinks Jazz No music Classical listening Control Popular

None 14.6 10.0 24.4 16.9 15.5 17.11–2 41.5 63.3 64.4 40.7 42.3 45.73 or more 43.9 26.7 11.1 42.4 42.3 37.1

responses were made later in the evening for the easy listening and no musicconditions only.

Two additional factors were explored in relation to patrons’ responses to the restaurant: patrons’ awareness of the music, and the degree to which they thought the music being played was appropriate. Figure 2 reportsthe distribution of responses according to patrons’ awareness of the musicbeing played. Patrons were less aware of the music in the classical condition,and more aware of the music in the jazz, popular and control conditions.Based on an ANOVA, the relationship between patrons’ awareness of themusic and their responses to atmosphere was not found to be significant (p = .05).

In terms of appropriateness, the control condition generated the most pos-itive response (94.4% considered the music appropriate), followed by jazz(87.8%) and popular music (77.1%). Generally, the more upbeat styles ofmusic were considered to be more appropriate than classical (46.7%) andeasy listening (62.7%). Classical music was considered the least appropriatestyle for the restaurant and, based on an ANOVA, was significantly differentfrom every other condition (p = .01). Easy listening was found to be signifi-cantly different from jazz and the control condition, and the number of


%

0

5

10

15

20

25

30

35

40

45

50

Not at all aware Slightly aware Moderately

aware

Very aware Totally aware

Level of awareness

Jazz Classical Easy listening Control Popular

F I G U R E 2 Distribution of participants according to awareness of the music.

people who thought the music was appropriate was significantly higher inthe control condition than the popular condition (p = .01).

Subjects’ responses to the appropriateness of the music were found to cor-relate significantly (p = .05) with their responses to the restaurant for theadjectives ‘downmarket’ and ‘tacky’. That is, patrons perceived the restaurantto be more downmarket and more tacky when they did not consider themusic to be appropriate. Inferential statistics also suggested that subjects whohad been to the restaurant before considered classical and easy listeningmusic to be less appropriate than those who were dining at the restaurant forthe first time; however, an ANOVA did not yield a significant difference (p = .05). While 53.3 percent of patrons in the classical group considered themusic inappropriate, a t-test showed that the amount of money they wereprepared to spend did not differ significantly from the 46.7 percent who con-sidered classical music appropriate.

Based on an ANOVA, the age of subjects was not found to significantlyinfluence their responses to the appropriateness of the music, the amount ofmoney they were prepared to spend, or their responses to atmosphere ormusic (p = .05). Similarly, a t-test did not reveal significant differencesbetween males and females according to these factors. An ANOVA was alsoused to investigate whether there was a relationship between subjects’income and their responses to the appropriateness of the musical style beingplayed. No differences were found between income groups (p = .05).

Discussion

PERCEPTION OF THE RESTAURANT

The results reported in Table 1 indicate that different styles of music, and theabsence of music, influenced patrons’ perceptions of the restaurant environ-ment. A positive relationship was also found between patrons’ perceptions ofthe restaurant and their perception of the music. These findings are consis-tent with North and Hargreaves (1998). A factor analysis of responses to therestaurant provided evidence that different styles of music (and no music) ledto differences in the general perceived characteristics of the restaurant, afinding also consistent with the North and Hargreaves (1998) study.

PURCHASING (INTENTIONS AND ACTUAL SALES)Patrons’ responses to the maximum amount of money they were prepared tospend on their main meal yielded several differences between conditions. Themost extreme differences occurred between no music and each of the fivemusical conditions. In accord with North and Hargreaves (1998), theseresults provide evidence that different musical styles, and no music, have thepotential to influence patrons’ purchase intentions.

Due to the fact that the number of people dining in the restaurant fluctu-ated considerably on the same day before, during and after the testing period,


it is difficult to assess the influence of music on actual sales. However, classical music was associated with relatively few people remaining in therestaurant after 11 pm, and a greater number of people leaving the restau-rant earlier in the evening. This may be related to the fact that 53.3 percent ofsubjects considered classical music to be inappropriate. While this did notappear to affect the amount of money patrons in this group indicated theywere prepared to spend, the perceived inappropriateness of the music mayhave influenced the amount of time and perhaps money they actually spent(supporting Radocy and Boyle, 1997). Classical music was also associatedwith fewer drinks being consumed, and consequently less money spent onalcoholic drinks.

INTERACTIONS

The results of this study suggest that several other factors were influencingthe relationship between music and consumer perceptions. For example,results showed that the more people dining at a table, the more the restaurantwas perceived as ‘cerebral’ and ‘aggressive’. This relationship was found to besignificant in both the jazz and control conditions. Results also suggested thatan interactive effect occurred between the style of music, subjects’ perceptionof the atmosphere, and the amount of alcohol consumed. For example, rat-ings for several adjectives increased with the amount of alcohol consumedwhen jazz was playing.

A significant interactive effect was also found between the style of musicplayed, patrons’ responses to the quality of service, and responses to atmos-phere. Lending support to Dube et al. (1995) and North and Hargreaves(1996), these results suggest that music may be positively related to patrons’willingness to interact. Findings also suggested an interaction betweenmusic, perceived atmosphere and time of night. That is, the restaurantatmosphere was perceived as less invigorating later in the evening for theeasy listening and no music conditions.

AWARENESS AND APPROPRIATENESS OF THE MUSIC

Results showed that subjects’ awareness of the music differed across condi-tions. Overall, subjects were more aware of the music when upbeat styleswere played. In contrast, classical and easy listening conditions were associ-ated with more people indicating that they were ‘not at all aware’ of themusic being played.

Findings suggest that subjects considered the more upbeat (or stimulative)styles of music to be more appropriate for the restaurant. This may be attrib-uted to the concentration of subjects aged between 20 and 39. For example,the number of patrons indicating that classical and easy listening music wereappropriate increased according to age. Conversely, the number of peoplewho considered popular, jazz and the control condition appropriate decreasedwith age (although not significantly). Subjects who did not consider the


music to be appropriate perceived the restaurant as more downmarket andmore tacky.

The music that was considered more appropriate for the restaurant can bedescribed as having a high level of ‘fit’. In this environment, the degree of fitbetween the music and the restaurant appears to have influenced several fac-tors which are directly related to enhancing business and increasing sales.For example, the musical styles with a higher degree of ‘fit’ were associatedwith more alcohol being consumed, higher purchase estimates, more positiveresponses to atmosphere, and more patrons remaining in the restaurant laterin the evening. While there was no significant difference between classicalmusic and the more upbeat styles in terms of the amount patrons were pre-pared to spend, the lack of fit between classical music and the restaurant mayhave caused patrons to spend less time in the restaurant, and consequentlyless money on food and drinks.

IMPLICATIONS

The fact that patrons were prepared to spend more when popular, jazz andclassical music were played suggests that spending might be increased bymusic that creates the perception of an upbeat or upmarket environment.This finding corresponds with North and Hargreaves (1998) who found thatclassical and popular music had a more positive effect on purchase intentionsthan easy listening and no music. Similarly, Areni and Kim (1993) found thatpeople were prepared to spend more in a wine store when classical music wasplaying. In the present context, classical music was not considered to beappropriate by a large number of patrons. This suggests that there may besome discrepancy between the amount of money patrons indicated theywould be prepared to spend, and the amount of money actually spent.Overall, however, the findings reported here are consistent with a model ofthe effects of music on purchasing which states that ‘the nature of people’sresponses to music activate contextually relevant knowledge or behavior inother domains’ (North and Hargreaves, 1998: 2267).

The present findings support Radocy and Boyle’s (1997) suggestion thatpeople might be inclined to spend more time and money in a restaurant orstore when the music being played is considered appropriate. Similarly, find-ings support MacInnis and Parks’ (1991) notion that persuasion is enhancedwhen the music is appropriate for the context in which it is played.

Findings suggest that responses to the characteristics of the restaurantwere positively influenced by factors such as the number of people dining at atable, the amount of alcohol consumed, the quality of service, and the number of times patrons had been to the restaurant before. These positiverelationships were only found to exist when the more upbeat styles of musicwere played. Again, these findings support the notion that people’s responsesto music may activate contextually relevant behavior in other domains.

There are several practical applications of the results reported in this


study. Firstly, results suggest that music can be used by restaurant and store owners to create a specific atmosphere which will distinguish the envi-ronment from competitors. Findings also suggest that stores which playupbeat or upmarket music may be able to charge higher prices. Overall, theabsence of music had the most negative effect on atmosphere and theamount of money patrons were prepared to spend.

This research has demonstrated that music can influence the perceivedcharacteristics of the environment in which it is played. In addition, it has provided evidence that different types of music can produce specificatmospheres such as upmarket and upbeat. Importantly, the study demon-strated that music can influence the amount of money patrons are preparedto spend, and perhaps the amount of money they actually spend. Overall, it is clearly evident that music has the potential to influence commercialprocesses.


I am grateful to APRA (Australasian Performing Right Association) for providingfinancial support for this study and especially to Dean Ormston for his contribution tothis project. Thanks are extended to Andrew Becker from SMA (Satellite MusicAustralia) for assisting with the musical stimuli used in the study. I would also like tothank the owners of Out Of Africa, Omar Madji and Hassan M’Souli, for their partici-pation in this study, and Meredith Wilson, Associate Professor Gary McPherson andDr Emery Schubert for their valuable comments on this paper.

R E F E R E N C E S

Alpert, J.I. and Alpert, M.I. (1990) ‘Music influences on mood and purchase intention’, Psychology and Marketing 7: 109–33.

Areni, C.S. and Kim, D. (1993) ‘The Influence of Background Music on ShoppingBehavior: Classical versus Top-40 Music in a Wine Store’, Advances in ConsumerResearch 20: 336–40.

Bleich, S., Zillmann, D. and Weaver, J. (1991) ‘Enjoyment and Consumption ofDefiant Rock Music as a Function of Adolescent Rebelliousness’, Journal ofBroadcasting and Electronic Media 35: 351–66.

Bruner, G.C. (1990) ‘Music, Mood and Marketing’, Journal of Marketing 54: 94–104.Dube, L., Chebat, J.-C. and Morin, S. (1995) ‘The Effects of Background Music on

Consumers’ Desire to Affiliate in Buyer–Seller Interactions’, Psychology andMarketing 12: 305–19.

Fried, R. and Berkowitz, L. (1979) ‘Music Hath Charms . . . and Can InfluenceHelpfulness’, Journal of Applied Psychology 9: 199–208.

MacInnis, D.J. and Park, C. (1991) ‘The Differential Role of Characteristics of Musicon High- and Low-Involvement Consumers’ Processing of Ads’, Journal ofConsumer Research 18: 161–73.

Mehrabian, A. and Russel, J. (1974) An Approach to Environmental Psychology.Cambridge, MA: MIT Press.

Milliman, R.E. (1982) ‘Using Background Music to Affect the Behavior ofSupermarket Shoppers’, Journal of Marketing 46(3): 86–91.


Milliman, R.E. (1986) ‘The Influence of Background Music on the Behavior ofRestaurant Patrons’, Journal of Consumer Research 13: 286–9.

North, A.C. and Hargreaves, D.J. (1996) ‘The Effects of Music on Responses to aDining Area’, Journal of Environmental Psychology 16: 55–64.

North, A.C. and Hargreaves, D.J. (1998) ‘The Effect of Music on Atmosphere andPurchase Intentions in a Cafeteria’, Journal of Applied Psychology 28(4): 2254–73.

Radocy, R.E. and Boyle, J.D. (1997) Psychological Foundations of Musical Behavior, 3rdedn. Springfield, IL: Charles C. Thomas.

Robally, T.C., McGreevy, C., Rongo, R.R., Schwantes, M.L., Steger, P.J., Wininger, M.A.and Gardner, E.B. (1985) ‘The Effect of Music on Eating Behavior’, Bulletin of thePsychonomic Society 23: 221–2.

Stack, S. and Gunlach, J. (1992) ‘The Effect of Country Music on Suicide’, SocialForces 71: 211–18.

Standley, J. (1995) ‘Music as a Therapeutic Intervention in Medical and DentalTreatment: Research and Clinical Applications’, in T. Wigram, B. Saperstone andR. West (eds) The Art and Science of Music Therapy. Langhorne, PA: HarwoodAcademic/Gordon and Breach.

Stratton, V.N. and Zalanowski, A. (1984) ‘The Effect of Background Music on VerbalInteraction of Groups’, Journal of Music Therapy 21: 16–26.

Yalch, R. and Spangenberg, E. (1990) ‘Effects of Store Music on Shopping Behavior’,Journal of Services Marketing 4: 31–9.

Zillmann, D. and Bhatia, A. (1989) ‘Effects of Associating with Musical Genres onHeterosexual Attraction’, Communication Research 16: 263–88.

Zullow, H.M. (1991) ‘Pessimistic Rumination in Popular Songs and News MagazinesPredict Economic Recession via Decreased Consumer Optimism and Spending’,Journal of Economic Psychology 12: 501–26.

Appendix: musical stimuli

JAZZ

A Night in Tunisia: Clifford Brown A Stanley Steamer: Earl Hines A Taste of Honey: Charlie Bird Be Yourself: Kenny Burrell Better Get It in Your Soul: Charles Mingus Black Coffee: Earl Hines Blues for ZW: Leroy Jones Deodato: Bangles and Beads For All We Know: Dave Brubeck Gone with the Wind: Dave Brubeck Indiana (Back Home Again in Indiana): Milt Hinton Love for Sale: Miles Davis Midnight at the Oasis: Hubert Laws Miles: Miles Davis My Funny Valentine: Chet Baker and Gerry Mulligan Rumble in the Jungle: Max Roach Something Else: Miles Davis


Take Five: Dave BrubeckThat Beautiful Sadness: Mark Isham The New Message: Art Blakey and the Jazz Messengers This Can’t Be Love: Ellis Marsalis What Now My Love: Lou Donaldson Yesterday’s Dreams: Freddie Hubbard

POPULAR

All in Your Hands: Lamb Alone: Ben Harper Appletree: Erika Badu At the River: Groove Armada Blow Up the Pokies: The Whitlams Buses and Trains: Bachelor Girl Crash and Burn: Savage Garden Don’t Call Me Baby: Madison Avenue Even When I’m Sleeping: Leonardo’s Bride Everybody Here Wants You: Jeff Buckley Freshmint: Regurgitator Friendly Pressure: Jhelisa Glockenpop: Spiderbait Half the Man: Jamiroquai I Think I’m in Love with You: Jessica Simpson I Try: Macy Gray It Ain’t Over ‘Til It’s Over: Lenny Kravitz Karmacoma: Massive Attack Keep Me Lifted: Spearhead Lucky Star: Alex Lloyd Nothing Much Happens: Ben Lee One More Time: Groove TerminatorRevenge on the Number: Portishead Shine: Vanessa Amorosi Spinning Around: Kylie Minogue Still a Friend of Mine: Incognito Sunshine on a Rainy Day: Christine Anu Thank You (For Loving Me at My Worst): The Whitlams Tropicalia: BeckTry Whistling This: Neil FinnWeir: Killing Heidi Why Does My Heart Feel So Bad?: Moby

CLASSICAL

Bach: Air on a G String Bach: Brandenberg Concerto No. 1 in E, RV 269 (II: Largo)


Beethoven: Moonlight SonataBeethoven: Symphony No. 4 in B flat major, Op. 60 (II: Adagio) Debussy: Prelude to the Afternoon of a FaunElgar: Serenade for Strings, Op. 20 (II. Largetto) Grieg: Morning (from Peer Gynt) Handel: Water Music: Suite (II) Hummel: Piano Concerto in A minor, Op. 85 (II: Larghetto)Mahler: Symphony No. 5 in C sharp minor (IV: Adagietto)Mendelssohn: Violin Concerto No. 2 in E minor, Op. 64 (Andante)Mozart: Piano Concerto No. 19 KV 459 (Allegretto)Mozart: Piano Concerto No. 20 KV 466 (Romance) Mozart: Piano Concerto No. 21 in C major, K467 (II: Andante) Rachmaninov: Concerto for Piano and Orchestra No. 2 in C minor, Op. 18 (II:Adagio Sostenuto) Schubert: Symphony No. 5 in B flat Major, D485 (II: Andante con moto)Sibelius: Andante FestivoTchaikovsky: Piano Concerto No. 1 Vivaldi: The Four Seasons (Spring)

EASY LISTENING

Adeline: Richard Clayderman All At Once: Fairfield All I Have to Do is Dream: John Fox All My Life: Nick Ingman Always On My Mind: Pan Pipes Annie’s Song: Johnny Pearson Are You Free: Images Around Every Corner: Grant Geissman Catalina: Fernando Jonas Cinema Paradiso: Roger Woodward Drivetime: Tommy Emmanuel Endless Love: Richard Tucker Everything I Do I Do It For You: Guitar Moods Forever Love: Gary Barlow Greensleeves: London Symphony Orchestra Lara’s Theme: Richard Clayderman Memory: James Galway Never on Sunday: The 101 Strings Orchestra Once Again: Earl Klugh Sadness: Roger Woodward Saving All My Love For You: Paul Mauriat Stay Another Day: Pierre Belmonde Stranger on the Shore: Kenny G The Greatest Love of All: Starsound Orchestra


The Summer Knows: Glenn Long and His Orchestra Thinking of You: Oscar Lopez Tonight I Celebrate My Love For You: Hill/Wiltstchinski Guitar Your Song: Samantha Blue

S T E P H A N I E W I L S O N obtained her PhD at the School of Music and Music Education atthe University of New South Wales in 1999. Her doctoral thesis investigated patternperception and temporality in the music of Steve Reich. Since this time, she has con-ducted research in the area of music psychology for the Australasian PerformingRight Association and has been involved in both music teaching and performance.Currently, she is working as Project Development Co-ordinator for the Office of thePro-Vice-Chancellor (Education) at the University of New South Wales. This roleinvolves supporting a range of learning and teaching initiatives and projects acrossthe University. Address: Learning and Teaching@UNSW, Office of the Pro-Vice-Chancellor(Education), University of New South Wales, Sydney, Australia 2052. [email:[email protected]]


Book reviews

113

Psychology of Music



Researchvol (): ‒

[- ():; ‒; ]

ROBERT ROWE, Machine Musicianship. Cambridge, MA: MIT Press, 2001.416 pp. + CD-ROM. ISBN 0-262-182206-8 £32.95

The straightforward clarity of Rowe’s writing belies the enormity of the proj-ect represented by Machine Musicianship. He provides a very readable surveyof the wide, heterogeneous and often technical space in which research inmusic cognition and computational music creation intersect. The book effec-tively falls into two parts: theories and models addressing various aspects ofmusic cognition are discussed in terms of their utility in the machine imple-mentation of musical competence; and applications in different modes ofcomputer music are trawled for functionality which replicates or engageswith human musical skills. His aim is to improve the functionality ofmachine music processing in its various modes by better embodying humanmusicianship. Throughout, Rowe offers a full account of his own implemen-tation of the most relevant models in the context of his Machine MusicianshipLibrary, a common environment for evaluation and experimentation.Explanatory code in C++ is incorporated in the text; full code, applicationsand sound examples are on a CD ROM. These provide well-rounded support,but the narrative also succeeds in making its points for the reader who prefersto leave to one side the technical detail; indeed, in order to consider the widerimplications of the book, this might be the best first approach.

That Rowe’s selection of models for consideration reads like a summary ofwork in music cognition over the last two decades is indicative of the extent towhich the computational paradigm has suffused the field. One wonderswhether a model (a metaphor – let’s be honest) which cannot be expressed incomputational terms is thinkable or allowable. Given the conservative way inwhich the role of computers tends to be imagined, however, this may alsoprove to be a vision-limiting factor. Models of chord identification, key induc-tion, metric analysis, segmentation and pattern identification are surveyedcomprehensively, although Narmour’s implication/realization model is per-haps under-represented, given its naturally computational formulation. Theexplanations of the various computational paradigms involved – from ‘goodold-fashioned AI’ to neural networks and intelligent agents – are concise and

sempre :

unusually lucid. The treatment of pattern processing is judicious in terms ofthe musical literature, but this is itself conditioned by issues of representa-tion, usually of events. There are large bodies of work in the areas of data-mining and nonlinear time-series analysis which might have much to offer inthis respect, but which are perhaps less intuitive in terms of conventionalmusical terminology.

New issues emerge from Rowe’s exploration of the practical implementa-tions of the most likely candidates, particularly the speed with which quiteplausible theories become divorced from biological reality (I don’t think Imaintain values for 24 candidate keys when listening to classical music), andthe difficulty of locating ‘fundamental’ behaviours for different parameters.Pitch recognition may be hard-wired, categorization probably less so, but itseems unlikely on a physiological level that, as Rowe suggests, ‘as key induc-tion is to chord identification, metrical analysis is to beat tracking’. Rowe isconcerned with certain properties of the models he considers: their cognitiveverisimilitude, the degree to which they afford algorithmic implementation,musical utility and real time performance. This last property is a function ofseveral factors: computational cost, representational transparency, thedegree of background knowledge necessary and the capacity to deal with thesequential ordering of real time data. The formal distinction between modelswhich can work sequentially, and in real time, and those which rely on theout-of-time analysis of complete sets of data is an area which is still open forinvestigation. As computing power becomes less of an issue, this becomes amatter of very practical concern. What emerges from Rowe’s work is that, inpractice, background or a priori knowledge and cognitive performance intime are inextricably convolved.

The field of the use of models of musical skills in creative applications isnaturally less orderly. There are chapters on compositional techniques, algo-rithmic expression and music cognition, interactive improvisation, interactivemultimedia and installations, of which the last, by virtue of the vast range ofpossible modes of activity, is the least comprehensive. They constitute anaccurate survey of current modes of practice, although in its pragmatismsuch an approach risks impeding the evolution of new practice. Many strate-gies are illustrated with examples in the Max graphical programming envi-ronment – the lingua franca of interactive music. It is particularly interestingto see instances from the works of Manoury, Campion and Teitelbaum, and thestyle-learning techniques of Dannenberg. Much of this work is published else-where but is less than easily accessible, and in digest form the common issuesemerge from piece-specific detail. George Lewis – one of the most significantfigures in terms of both music and thought – is given rather short shrift in thesection on improvisation, perhaps because his techniques are less open toexternal scrutiny. Rowe touches on the combinatorial complexity of workingwith multiple strands of interaction; in fact his own Cypher program – layersof listener, player and critic agents – affords intuitive expansion in this way.


Behind the vast body of work presented lie two issues which the individualprojects often manage to avoid: the precise explanatory level on which a par-ticular model holds true (culturally situated cognitive metaphor, biologicaldescription, machine-implementable algorithm); and the nature of the rela-tionship between a tool for creativity and its human user. In the latter case,whilst interactive music systems must certainly engage with human cogni-tive mechanisms, it is likely that the design of a system which functions as aprosthesis to human creativity must be fundamentally different to that of acybernetic musician. The musical outcome is likewise likely to be different(and here we return to the issue of tonality as a cognitive primitive). Many ofthe theories and models discussed are responses to computational or cogni-tive paradigms – often simultaneously both super-biologically metaphoricaland sub-biologically technical – rather than thoughts about how musicworks in a space bounded by human biology, culture and technology. Most ofthe techniques considered are predicated on some sort of pre-embodiedknowledge, which for reasons of encapsulation, learning or falsification rep-resent the abstraction of a musical skill rather then its embodiment. Timeand context swiftly exert their authority as the prime conditioning agents ofany real-time implementation (or musical activity, for that matter), but in thereal-world dynamics of human theory construction are too soon relegated tonuisance value. What would be left without random numbers, caricatures ofmainstream jazz and disembodied exam-passing Western music theory? Toover-generalize: while the functional modality of much of the work proves tobe too vague in the face of real-world application, the knowledge it requires istoo specific.

The areas of music cognition and interactive production inevitably overlap– inevitably, that is, from the point of view of creation. The post-hoc culturalgeneralization of theories of music cognition that have any cultural, contex-tual or historical specificity may be unavoidable – useful, even, in an experi-mental context – but serves to underplay the real-world, real-time aspects ofboth the creation and impact of music. Unsurprisingly, then, there are dis-continuities of discourse and of function in redirecting such theories to theproduction of music. The relationship between the two areas may appearasymmetrical: whilst research in psychology can exist in modes other thanthe computational, computer-based interactive music must crucially engagewith the terms of its own engagement – the fundamental premise of Rowe’sbook. One implicit lesson might be, however, that ‘abstract’ cognitiveresearch might usefully situate itself more fully in the particulars of a specificmusical activity.

The case for the relationship between the two halves of the book could bemade, if anything, more strongly. In both cases, recent work in other areas ofAI (from Herbert Simon on artificiality to Rodney Brooks’ radical rethinkingof robotics) has suggested that turbo-charging algorithms, tinkering withthem to make them cleverer, or adding a level of randomness or ‘humanizing’

Book reviews 115

is not a step along the path between the two. Rather it is the situating of suchbehaviour – the environment in which they find themselves; the complexityand even incommensurability resulting from their number; their uniquetimeline by which their current state reflects their history – which generatesrichness of detail. Rowe’s presentation and examination of this body of workobliges the reader to address two questions which both general theoreticaland specific technical studies easily sidestep:

– can we extrapolate from individual cognitive mechanisms to the adaptivecomplexity of musicianship of any interest?

– are the principles derived from investigating and imitating individualaspects of generic musics transferable to the activity of generating inter-esting music?

There are other books which survey recent work in music cognition, wellknown to readers of this journal. Rowe’s earlier Interactive Music Systemspresents his personal approach to the issues, and Tod Winkler’s InteractiveMusic Composition provides recipes for common techniques in the Max lan-guage. Machine Musicianship is the first publication to deal coherently withthe mutual implications of these areas of activity and, as such, should be avital part of the grounding of anyone setting out in either direction. There isa crucial paradox at its heart: the musicianship discussed remains that of keyidentification, metre recognition and score following; the set of skills requiredto fully appreciate Rowe’s book is broader, more contemporary, technological,adaptive and dynamic, and embodies a wider cultural perspective. The latteris perhaps the set of skills which creative systems will have to address.

Jonathan ImpettUniversity of East Anglia

TIM MILES and JOHN WESTCOMBE (eds), Music and Dyslexia: Opening NewDoors. London: Whurr, 2001. 200 pp. ISBN 186-156205-5 £16.50

There is a growing awareness that dyslexic children can experience particu-lar difficulties when learning to play a musical instrument. In 1985, theMusic Committee of the British Dyslexia Association produced a leaflet out-lining some common problem areas, and in recent years a number of teach-ers and researchers have sought to analyse and define the difficulties moreclearly.

The present collection of essays, edited by Tim Miles and John Westcombe,offers an interesting and valuable addition to the small amount of literatureon this topic. The book includes summary chapters by some of the key figuresin the area (including Tim Miles, John Westcombe, Sheila Oglethorpe andMargaret Hubicki), along with a collection of personal accounts by dyslexic


musicians of all ages and degrees of professionalism. The broad range ofexperience held by these contributors leads to a variety of insights into themany ways in which dyslexia can affect music-learning; insights which willbe welcomed by music teachers, parents, dyslexic musicians and researchersalike.

A key message of the book is that dyslexic children and adults can and doenjoy great musical success, despite the potential difficulties. Importantly,that elusive quality ‘musicality’ does not seem to be negatively affected bydyslexia (on the contrary, some teachers have noted a particular musical sen-sitivity amongst their dyslexic pupils). The problem is rather with the tools ofmusic – the physical co-ordination, the written notation, the complex timingskills and the learning and memory work involved. All of these skills canbecome very well developed if the teaching is supportive and carefully struc-tured, and the student’s motivation is high. The current volume contains agreat deal of practical advice for teachers, along with tips for students andprofessionals, making it a perfect shelf-partner for Sheila Oglethorpe’sInstrumental Music for Dyslexics (2002). The real value of the collection, how-ever, lies in the detailed personal descriptions of the actual experience oflearning music as a dyslexic, such as Jacob Wiltshire’s description of the con-fusion and frustration encountered in his music lessons (ch. 11), Siw Wood’sexplanation of her difficulties with music notation (ch. 12), or CarolineOldfield’s account of her problems when faced with a new piece of flutemusic (ch. 5). On the more positive side, Nigel Clarke provides inspirationwith the story of his success as a composer (ch. 7), and Sheila Oglethorpeoutlines the remarkable achievements of hard-working young dyslexic cho-risters (ch. 17).

A particular strength of the volume is the care with which the editorsguide the reader through the book. From the brief biography of each con-tributing author and the introductory chapter on dyslexia to the final sum-mary chapter, the suggestions for further reading, and the checklists for rec-ognizing dyslexia, it is clear that the editors have carefully considered theneeds of their audience. The presentation and organization of the book arealso extremely clear and user-friendly; ideal for dipping into.

One unfortunate consequence of gathering together a number of essayson the same topic is the inevitable repetition of information and ideas. In thiscase, the editors have chosen to include all the repetition, suggesting that itserves to emphasize the similarities of experiences between different musi-cians and teachers. An advantage of this approach is the resulting richness ofthe personal and contextual information, which a more concise editing stylemight have lost. Such detailed material provides an invaluable resource forthose interested in the specific nature of dyslexics’ musical difficulties, andleads the way for more qualitative research to be conducted in this area.

In this regard, it would be useful to analyse which particular difficultiesoccur most frequently amongst dyslexics, which occur less often, and which

Book reviews 117

occur similarly amongst non-dyslexics. There is also a possibility that dyslexics’interpretation of their difficulties with music may be influenced by what theyhave read about dyslexia and music, resulting in a slight ‘contamination’ oftheir reports. Addressing such challenging issues will be important for futureresearch – the current volume provides an extremely useful starting point.

R E F E R E N C E

Oglethorpe, Sheila (2002) Instrumental Music for Dyslexics. London: Whurr.

Katie OveryUniversity of Sheffield


EUROPEAN SOCIETY FOR THECOGNITIVE SCIENCES OF MUSIC (ESCOM)

5TH TRIENNIAL ESCOM CONFERENCE8–13 September 2003

University of Music and Drama, Hanover, Germany

Extension of deadline for submissions

We are delighted about the great interest shown in the forthcoming 5thESCOM Conference.The title of ESCOM5 is ‘Experience Music in Science– Science in Music’ and the focus of the Conference is on interdisciplinarypresentations, discussion and dissemination of new research relating tomusic perception and cognition.

A large number of interesting submissions have already been received forpapers and symposia. However, many applicants have asked for an extension of the deadline for their submissions.We are pleased to be ableto comply with these wishes and have now extended the deadline for submissions to:

1 February 2003

As we are running to a tight schedule, no submissions will be acceptedbeyond this new deadline. Please take advantage of this opportunity toactively participate in the 5th ESCOM Conference.

Details regarding the submissions are available on the internet athttp://www.escom5.de

Reinhard Kopiez (Conference Chair)Andreas Lehmann (Conference Co-Chair)Irving Wolther (Conference manager)

sempre:Society for Education, Musicand Psychology Research

Conference announcement and call for papers

Teaching and learning in musicIn honour of Arnold Bentley

Centre for International Research in Music Education,Faculty of Education, University of Surrey Roehampton

Saturday 12 April 2003

Keynote speaker: Professor Gary McPherson, University of New South Wales, Australia

‘Motivational and self-regulatory learning components of musical development’

Gary McPherson will present results from a large Australian Research Council project onover 700 young musicians between the ages of 10 and 20. The study examines the role ofself-regulation and motivation in musical practice and achievement, as measured bystandard performance examinations, and considers the implications for teaching andlearning in music.

Gary is Course Coordinator for Music Education at the University of New South Wales.He is former National President of the Australian Society for Music Education, Treasurerof the International Society of Music Education, and Editor of Research Studies in MusicEducation. His work is widely known internationally: his recent publications include TheScience and Psychology of Music Performance: Creative Strategies for Teaching andLearning (with Richard Parncutt, Oxford University Press, 2002).

Please send abstracts (200 words) to Caroline Freeland at the address below. Submissionson any aspect of learning and teaching in music are welcomed. Also, submissions frompostgraduate students and researchers early in their career are particularly welcome.

For further information please contact:Caroline Freeland, Centre for International Research in Music Education, Faculty of

Education, University of Surrey Roehampton, Southlands College, Roehampton Lane,London SW15 5SL [email: [email protected]]orRaymond MacDonald, SEMPRE Conference Secretary, Department of Psychology,

Glasgow Caledonian University, Cowcaddens Road, Glasgow G4 0BA [email:[email protected]]

He is former National President of the Australian Society for Music Education, Treasurerof the International Society of Music Education, and Editor of Research Studies in Music

5E D I T O R I A L

Psychology of Music



Researchvol ():

[- ():; ; ]

It is a great pleasure to be writing in the first issue of Psychology of Music to bepublished by SAGE Publications. We hope that it will be a long and fruitfulpartnership which will enable the journal to meet the needs of readers in the21st century. The journal will be published quarterly in January, April, Julyand October and, in addition to being available in hard copy, each issue can now be accessed online in institutional libraries subscribing to the jour-nal. Readers are also invited to subscribe to the journal’s free ContentsAlerting Service. More information can be found on SAGE’s website:www.sagepub.co.uk

This issue reflects the remit of the journal in providing a forum for articleswhich address issues at the intersection of psychology, music and music education. The articles, all empirical, draw on a wide range of methodologicalapproaches to consider issues as diverse as the development of intuitive musi-cal understanding to the effects of music on behaviour in real-life settings.The book reviews similarly reflect the unique focus on psychology, music andmusic education. Articles which consider theoretical perspectives or providereviews of the literature in particular fields will be especially welcome forfuture issues.

Forthcoming issues will include papers drawn from the conference‘Investigating Music Performance’ held at the Royal College of Music, London,from 12–13 April 2002, and a mapping exercise of UK music education and related psychological research accompanied by commentaries from colleagues around the world, giving brief accounts of similar research in theirown geographical area. A themed issue is also planned focusing on the topicof composition. If you have any suggestions for other themed issues, I wouldbe pleased to hear from you.

Susan HallamInstitute of Education, University of London

[email: [email protected]]

sempre :

Psychology of Music - SAGE Publications · Psychology of Music Volume 31 2003 Editor: Susan Hallam....

Documents

Transcript of Psychology of Music - SAGE Publications · Psychology of Music Volume 31 2003 Editor: Susan Hallam....