Music Education Research, Vol. 3, No.

2, 2001

Conference Keynote
Musical development Theories Revisited

KEITH SWANWICK, Institute of Education, University of London, 20, Bedford

Way, London WC1H 0AL,UK (Email: k.swanwick@ioe.ac.uk)

ABSTRACT Work on musical development continues in a variety of ways. Associated

with this is a range of theories and assumptions, though these are not always explicit.
In this paper some of the major contributions are reviewed. These include theories of
separate developmental channels for different musical activities, theories of musical
concepts and symbol systems, and the generic theory of Swanwick and Tillman, a theory
of musical understanding. Relationships between these theories are suggested along with
implications for teaching and learning.

Introduction
Much of the early research in music education was concerned with the development of
the discrimination of isolated pitch and rhythm differences and the ability to distinguish
between other sub-musical sound materials, such as recognition of the number of notes
in a chord or judgements of intensity and timbre. This tendency to separate and measure
different musical functions still continues to this day. For example, Brophy maintains
that because music consists of melody, rhythm, harmony, timbre, and form, the
development of understanding of each of these elements is acquired at different rates for
individual students, (Brophy, 2000). Hargreaves (1986) asserts that the development of
rhythmic skills are among the rst to emerge; and earlier, Bentley argued that the ability
to discriminate rhythm develops before an ability to discriminate pitch and to perceive
chords (Bentley, 1966). The problem is that such data are gleaned from dramatically
varying instruments of measure. For instance, a test of rhythm pattern discrimination,
such as Bentley gives, cannot be compared with his test based on counting the number
of notes in chords. These measures are calibrated differently. It is no more logical to
compare the results of such tests than it would be to compare tyre pressures on a car with
the oil level. There is no connective theory. There is no base-line for comparing say,
melodic with rhythmic development unless we have a model of musical development
ISSN 1461-3808 print; ISSN 1469-9893 online/01/020227-16 2000 Taylor & Francis Ltd
DOI: 10.1080/14613800120089278
228 K. Swanwick

and understanding, a meta-theory that links speci c measures of individual items: for
example again, perhaps the car needs a service!
Similar problems exist with empirical studies of early musical development. Follow-
ing Moog, other researchers (Davidson, 1983; Dowling, 1982, 1984), observed the
development of singing, recorded the spontaneous musical utterances of young children
and noted increasing control over melodic contours, interval size and tonality through the
early years. Several researchers have noted the intermingling of spontaneous singing and
conventional songs, especially around the age of 3 and after (Hargreaves, 1986,
pp. 7174). Such work has produced interesting observations and is essentially observa-
tional, which is a step away from testing. It still, though, lacks a theoretical frame within
which to locate speci c and limited developments. I wish therefore in this paper to focus
on major musical development theories and on their implications for understanding
musical development and music education.

Theories of Development and Musical Development

From the literature it is possible to identify four speci c criteria for evaluating
developmental theories in general and for music in particular (Hargreaves & Zimmer-
man, 1992). These are that:
Any theories and associated evidence should comprehensively re ect the nature of
musical behaviour.
Theories and underlying assumptions should be valid across a range of musical
activities or modalities: composing (including improvisation), performing and audi-
ence-listening.
Evidence should be systematically and reliably produced to support or challenge
theoretical assertions.
Developmental theories should take into account both the natural developmental
inclinations of individuals and the cultural environment in which their development is
realised.
Stating this last criterion opens up a crucial issue. There is an old debate about the
relative emphasis on what used to be called nature or nurture. Traces of the
developmental theory of Rousseau with its implication of maturational unfolding and
ripeness can be found in Gesell, Freud, Erikson and Piaget. Alternatively, from the
tabula rasa of Locke through the behaviourism of Pavlov, Watson and Skinner, to the
social-learning theories of Bandura and Vygotsky, researchers have emphasised the
environmental and social shaping of development. Of course, no major maturational
theorist has denied the crucial developmental role of environment and no environmental-
ist discounts the predisposition of a person or other organism to respond to physical or
cultural conditions. For example, Piaget never suggested that children could learn in a
cultural vacuum. And although he was not particularly concerned to investigate what
educational strategies or social conditions might affect development, he saw no problem
with the idea that each child might have his or her own optimal developmental rate.
Vygotsky is often seen as challenging the Piagetian view of development as an
unfolding of an organism without reference to cultural and physical environment
(Hargreaves & Zimmerman, 1992) But there are also strong similarities between Piaget
and Vygotsky. The latter saw development as a spiral and thought that children passed
through the same point at each new revolution while advancing to a higher level
(Vygotsky, 1978, p. 56). Vygotsky, like Piaget, viewed development as a complex
 Musical development theories revisited 229

dialectical process embodying qualitative transformations. What distinguishes Vygot-

skys work from Piagets is his concern with what he calls the zone of proximal
development, the space between the level of independent problem solving and the level
of potential achievement with adult guidance or in interaction with able peers, i.e. with
the social process. This theory is not a denial of maturation but recognition of the
possible effects of educational transactions.
Developmental stage theorists, such as Piaget, Erikson and Kohlberg, took the view
that a theory of cognitive stages should meet ve quite stringent criteria (Crain, 1992,
p. 264):
Stages imply qualitatively differentiated patterns of behaviour.
Stage descriptions deal with general issues rather than speci c achievements. (This is
very similar to the second criterion above referring to general validity across different
activities or modalities.)
Stages unfold in an invariant sequence.
Stages are cultural universals. (This may be thought highly problematic. We might,
however, expect a good theory to have explanatory power beyond a single cultural
group, though widespread might be a better word than universal.)
Stages are hierarchic, in that early structures are integrated into later ones.
Although some theories do not meet all of these criteria or resist the very idea of stages,
we do expect to nd patterns, for example of linguistic or sensory-motor development
in growing children. Such expectations do not violate respect for individual or cultural
differences, for the concept of a stage is not so xed as to leave no possibility for unique
interpretation, cultural difference or variation. Nor are stages rigidly linked with ages,
though there may be strong connections.
I wish to probe how far music development theories may meet not only the general
criteria but also how they match these speci c developmental stage criteria. If a theory
of cognition claims to address qualitative difference, how much more so should a theory
of an art such as music? And where a theory does not deal with general issues, but is
developed around speci c and limited musical skills, to what extent is it then possible
to nd explicit theoretical connections with other modalities? Can such a theory be
applied beyond the isolated skill? The idea of an invariant sequence suggests that we
would expect to nd some predictive power in any theory of musical development. We
ought to know what to look for, surely as important an ability for music educators as for
the general eld of child development.
In summary then, theories of musical development should:
have musical validity;
have relevance across different musical activities;
take account of both maturation and cultural setting;
identify qualitative, sequential and hierarchical changes;
have widespread cultural application;
be supported by reliable data.

Sera ne Rocks the Boat

Mary Louise Sera ne offers a direct challenge to traditional psychological models and
her approach is concerned with underlying cognitive processes. In her book, Music as
Cognition, she counters reductionist work in the psychology-of-music and musical
230 K. Swanwick

development (Sera ne, 1988, pp. 67). Sera ne was unimpressed by the earlier fascina-
tion for codifying music research into measurable parameters, such as time, timbre,
pitch and loudness. For Sera ne, isolated pitches and scales and chords are not musical
entities but analytical devices, the product of thinking about music rather than thinking
in music. This is a helpful distinction I make myself and there is a conceptual
relationship between Sera nes coherent musical units, which are heard moving in
time, and my expressive gestures or character(Swanwick, 1979, pp. 810), where
sound materials are transformed into musical entities, and tones are heard as tunes
(Swanwick, 1999).
Strongly critical of atomistic attempts to measure musical development, Sera ne puts
forward what she claims to be an over-arching theory of core cognitive processes. A
fundamental question informing her work is what is the nature and source of musical
thought? (Sera ne, 1980, p. 1). She posits a generic set of cognitive processes underly-
ing all musical production and musical perception (though for perception we should
perhaps substitute musical conception). Sera ne thus attempted to present a meta-psy-
chological model that stood outside of speci c and different musical activities or
modalities. The main characteristic of this universal cognitive activity is awareness of
movement in time. Tones are not heard in isolation nor as pairs of stimuli to be identi ed
or discriminated, but are sensory experiences from which the listener constructs musical
properties.
This theory meets at least some of our criteria for music developmental theories. First,
it claims to comprehensively re ect the essential nature of musical activity. Sera ne
established a generic meta-theory overlaying different speci c musical activities, in this
way meeting the criterion of validity across different modalities. However, her empirical
work has only been in the audience-listener mode: the children involved are basically
doing tests. The criterion of systematically acquired and reliable evidence is by no
means clearly met. It also should be noted that the relationship between Sera nes theory
and her observational method is not particularly clear. Several tests were given to
children including pitch discrimination, Piagetian number conservation, and several tasks
requiring correct answers to questions involving recognition of relationships between
small musical items. For example, the motivic chaining task asked whether or not a
longer phrase is made up of previously heard smaller phrases. It also seems clear that
the in uence of cultural (including an educational) environment is not a variable for
consideration. Musical development from her standpoint is predominantly ontogenetic in
emphasis.
Whether Sera nes theory is able to map musical development across a range of
musical and cultural settings is problematic, though this might be said of many
candidates for a theory of musical development. For example, we might notice her
reference to pieces of music rather than performances of music. There is here an
implication of certain Western traditions involving the presentation of notated composi-
tions. This may be a limiting aspect of her theory, in so far as many types of musical
production relying solely on aural transmission could be excluded. There is no evidence
to support the stage theory criterion of observability in different cultural settings.
Sera ne did not deal systematically with age-related changes. She implies the
existence of qualitatively differentiated developmental layers but there is little systemat-
ically gathered evidence for this. Replication and extension of Sera nes empirical
observations would be required in order to give suf cient con dence that the criterion
of methodological reliability was convincingly met.
 Musical development theories revisited 231

Symbol System Theories and Musical Concepts

The work of Howard Gardner and others in Project Zero has made a considerable impact
upon arts educators. Gardners theory of multiple intelligences has given teachers
working in these areas a sense that their activity is important and that the arts are
intelligent activities. His theory focused on the concept of symbol systems, which he
de ned as follows: symbolisation requires appreciation of an object and the capacity to
link the object known to a picture, label, or other kind of element that denotes it
(Gardner, 1973, p. 90). For Gardner, music and the other arts are distinctive modes of
symbolic communication, indeed his concept of multiple intelligences is really an
elaboration of the idea of multiple symbolic forms, an older formulation familiar to
readers of Langer and others. Symbol use is thus seen as central to childrens artistic
development.
As members of the Project Zero team, Davidson and Scripp also focused on childrens
use of symbols; however, they extended their argument to conclude that the development
of children could be studied through a visual symbolic medium. They asked children
between 5 and 7 years of age to write down a familiar song so that other children would
recognise it. From these data Davidson and Scripp identi ed a range of notations, from
simple marking to approximations of western staff notation. These data were to some
extent age related. Davidson and Scripp also compared children and adults with or
without musical training. They concluded that the use of conventional notation is an
indicator of higher levels of music development, surely a problematic assertion in view
of the aural/oral basis of much of the worlds music. They are also quite prescriptive:
when literacy skills are not developed alongside technical instrumental
training, the result is a fragmented knowledge of music and ultimate disinte-
gration of skills. When literacy skills fail to develop, students compensate for
the lack of integration in their training by substituting what they know about
music for what they hear. (Davidson & Scripp, 1989, p. 77.)
According to Davidson and Scripp, the interaction of motor and literacy skills enables:
the student to link performance, concept and percept. Re ective thinking
appears as an important dimension of musical development that arises from the
more enactive stages where skills are rst manifest, and are later linked to the
symbolic literacy skills of the musical culture. The most effective levels of a
music education must encompass this perspective. (Ibid., p. 80.)
In their paper Surveying the Coordinates of Cognitive Skills in Music, Davidson and
Scripp linked musical production directly with the making of musical scores (Davidson
& Scripp, 1992, p. 396). They then annexed the concept of symbols to focus on visual
notations. In this respect, they departed from Gardner, who argued that the sounding
forms of music de ne musical intelligence. Gardner identi ed as savants those who have
marked musical performance and aural abilities independent of notational ability. While
the savants that Gardner described do not use music notations, they did engage in a
symbolic form at a fairly high level. It is therefore questionable whether the study of
visual representations of music can, by itself, comprehensively inform us about chil-
drens musical development. In terms of our other criteria, it is also problematic as to
whether this procedure would hold for non-western cultures, or indeed for western
people with musical backgrounds where the emphasis is strongly aural/oral, such as pop
and rock music. Thus, these studies appear to be strongly biased towards musical
232 K. Swanwick

traditions involving staff notation, arguing that the closer a students notations resemble
staff conventions, the more musically developed the student may be. As we have seen
with Gardners savants, this is not necessarily so.
I want to argue that musical development neither depends on nor is best observed by
translating musical images into notations or words. Such procedures shift the methodo-
logical focus from musical development to notational or linguistic developmentvery
different domains. Thus any observed development might be explained within the
domain of notational representation rather than music.
Despite these shortcomings, a number of music development researchers have placed
signi cant weight on the analysis of notations. For example, Barrett asserted that the
notations children use to symbolise their experience of the world is representative of
their thinking about the world (Barrett, 1997, p. 71). But music itself is an activity that
is in some way representative of our experience of the world. Music is a primary
symbolic system. Notations, verbal descriptions or graphic representations are secondary
systems, offering a translation from one representational domain to another. In this
process some loss of information is inevitable.
In terms of our criteria for a strong developmental theory, this kind of work does not
appear to comprehensively re ect the nature of musical activities and behaviour. It does
not appear valid across a range of musical activities or modalities, nor does it address
the relationship between the natural developmental inclinations of individuals and their
cultural environment. When evaluated by the stringent criteria of stage theory, the
development of notations is not necessarily qualitatively differentiated, rather, it is
quantitative, more of the same at a greater level of detail. It is also questionable as to
whether the developmental descriptions of Davidson and Scripp effectively address
general issues, rather than speci c achievements. Finally, there appear to be insuf cient
data to determine whether there is an invariant developmental sequence or whether these
stages can be generalised to non-western cultures.

Bambergers Theory of Developmental Cumulation

Jeanne Bamberger also studied musical development, focusing on the relationship
between childrens notations and musical performance, speci cally playing Montessori
bells. From her observations, she concluded that childrens musical development is
multiple and cumulative.
I argue that the changing mental organizing structures that guide hearings,
constructions, and descriptions at various ages and stages of musical develop-
ment do not constitute a unidirectional progression in which earlier mental
structures are replaced by later ones. Rather, foci of attention among relevant
aspects of musical structure shift but also cumulatively build on one another.
I conclude that the goal of musical development is to have access to multiple
dimensions of musical structure, to be able to choose selectively among them,
to change focus at will (Bamberger, 1991, pp. 34).
Bamberger found that the use of notations helped to shape their users internalised,
active organizing constraints (Ibid., p. 15). She argued that musical symbols (notations)
help childrens musical understanding in a cognitive process, which involves several
levels, which she referred to as multiple hearings. According to Bamberger, as students
increase their capacity to make multiple hearings, their musical skills become more
developed (Ibid., p. 124).
 Musical development theories revisited 233

Bambergers systematic methodology explored the relationship between visual appar-

atus and childrens musical development, but it did not involve children experiencing or
learning music in other ways. As part of a laboratory procedure, her observations were
based on normative values, correct answers in the performance of simple melodies.
Bambergers theories and associated evidence do not comprehensively re ect the nature
of musical activities and behaviour and are not valid across a range of musical activities
or modalities such as composing (including improvising), performing and audience-lis-
tening. Furthermore, Bamberger did not consider the interaction between the natural
developmental inclinations of individuals and of the cultural environment in which
childrens musical development is realised.
A close reading of Bambergers theory of developmental cumulation suggests that
there may be stages with qualitatively differentiated patterns of behaviour, although
Bamberger stressed the interchangeability of levels rather than an invariant sequence.
While Bamberger did not produce evidence for cultural universality, she has suggested
in her work that there may be hierarchical stages where early structures are integrated
into later ones.

Musical Concepts
As we have seen, much of the thinking of the Cambridge (Mass.) researchers closely
identi ed music with notations of one kind or another. There is also a related assumption
about the cognitive nature of musical development and especially the idea of musical
concepts which is shared by many who approach music from the cognitive psycholog-
ical angle. Under the heading concepts and schemata in music, Hargreaves and
Zimmerman discussed the in uence of Piaget on the psychology of music. From multiple
encounters with music we develop musical concepts which enable us to make compari-
sons and discriminations, to organise sounds, to generalise, and, nally, to apply the
emerging concepts to new musical situations (Hargreaves & Galton, 1992, p. 385).
Hargreaves and Zimmerman distinguished concepts, and higher levels of generalisation,
from mental images. Bruner also made this distinction in terms of iconic and symbolic
modes of representation, as did Piaget through his notion of internalised mental actions
and formal operations. Hargreaves and Zimmerman, however, asserted that conceptual
labelling, categorising and organising musical perceptions provides the key for later
study and enjoyment of the complexities of music. Musical development thus proceeds
from sensory-motor schemata, though representation in musical images to the ability to
handle an increasing number of concepts (Ibid., p. 386).
Is musical development really an increase of labelling and categorising? I doubt it.
There are many conceptually inarticulate ne musicians. Likewise, tests of conser-
vation appear to be a somewhat narrow approach to musical development and, like
analysis of notations, may constrict our concept of musical understanding. Studying
notations and test results may be help to some extent but they do not in themselves lie
within the sonorous, symbolic mode of music. They are not musical behaviours.
Observing actual musical production is likely to be more informative.

The Swanwick/Tillman Developmental Model

Developmental theories of music education which depend on analysis of secondary
symbol systems (notations) or on separated observations of melodic, rhythmic, harmonic
and other behaviours, or rely on testing only aural perception may have attractive
234 K. Swanwick

scienti c possibilities in terms of control and reliability. However, the issue of musical
validity is a real one for such methodologies. In an attempt to address this dif culty,
Swanwick and Tillmans 1986 study of musical development is based on a generic
theory of musical experience and on observations of actual music making under fairly
open conditions. The work has been widely cited and extensively reviewed and further
research by myself and others have since addressed questions both of validity and
reliability.

The Theoretical Basis

In A Basis for Music Education, Swanwick set forth categories of hierarchical educa-
tional objectives for music (Swanwick, 1979). These include skill acquisition (later to
become materials), recognising and producing expressive gesture (later to be short-
handed as expression), identifying and displaying the operation of norms and devia-
tions (later designated as form) and aesthetic response (ultimately to be labelled
value). This approach was further developed and subsequently linked to Piagetian
concepts in The Arts in Education: Dreaming or Wide Awake? In that transcript of an
earlier professorial lecture, I argued that mastery, imitation and imaginative play are
essential psychological elements in all artistic engagement (Swanwick, 1983, 1988). In
speci cally musical terms these are identi ed with perceiving and controlling sound
materials, projecting and locating expressive character, and awareness of interrelation-
ships between expressive gestures, i.e. dynamic structure. This is the theoretical basis of
the later detailed developmental theory and it was an attempt to synthesise and
psychologise major strands of debate in aesthetics. Technique, expression, form and
value are thus not seen as competing but as complementary elements of music
experience. The most developed form of the theory is that music is a multi-layered
human experience, where layers not only interact vertically but also laterally, as minds
assimilate and accommodate to musical processes.
Swanwick thus attempts to sketch a generic epistemology for music, that is to say, to
answer Sera nes question: What is the nature and source of musical thought? It is this
issue which received attention rather than the creation of a developmental model. There
was initially little engagement with Sera nes other question of how children acquire
musical capabilities, though educational implications of the epistemology are explored in
some detail. The link between the epistemological issue and the subsequent developmen-
tal model is made explicit, particularly in the article by Swanwick and Tillman (1986)
and later in Tillmans PhD thesis (1987). Further empirical and theoretical work has been
discussed by Swanwick (1991, 1994) and by Swanwick and Franca (1999).

The London Study

The data collected by Tillman were crucial for the articulation of the eventual develop-
mental theory. Children, mostly aged 311, from several ethnic and cultural groups in
London schools produced un-notated compositions. These were part of normal classroom
activities and the pieces were tape-recorded, an activity which continued over 4 years,
nine times each year, yielding 745 compositions from 48 children. There was thus a
cross-section of music from children of different ages and in some cases a longitudinal
spectrum of compositions from individual children. The compositions ranged from brief
spontaneous utterances to more sustained and rehearsed musical inventions. Collecting
compositions in this way was thought to be both ecologically sensitive and having
 Musical development theories revisited 235

greater musical validity than giving written tests. These data were grouped into clusters
having similar properties and these tted with Swanwicks original conception of four
levels or layers of musical thinking: materials, expression, form and value (Swanwick &
Tillman, 1986).
Further evaluation of the compositions led to seeing that on each of four levels there
was a transformation from assimilatory, personal response to music (the left side) to
accommodatory social sharing (the right side). This dialectical relationship was more
fully developed in Musical Knowledge, where under the generic headings of intuition
and analysis is developed the conceptual basis of the left and right side of the spiral
model (Swanwick, 1994, p. 176). These are associated respectively with assimilation and
accommodation, with musical encounter and musical instruction, the aesthetic and the
artistic, personal interpretation and cultural transmission. Koopman rightly saw the left
side as a subjective pole, concerned with self-directedness while the right side
represents a dialogical relation with the external world (Koopman, 1995, p. 60).
The developmental spiral thus consists of eight modes of musical functioning, two
on each layer or level. The terminology for these modes drew on earlier observations by
Bunting of children composing in the classroom (Bunting, 1977). Each of these modes
was seen as a qualitative shift. An initial phase of exploring sounds transforms into
manipulative control. With this technical ability musical expression becomes possible, at
rst spontaneous but later more conventional with vernacular commonplaces of phrase
and sequence. These conventions are later assimilated into musical form, initially as
speculation and then in speci c styles or idioms. Beyond this lies the possibility of
symbolic value for the individual and systematic musical commitment. The concept of
value involves more than enjoyment and denotes the phenomenon of people becoming
consciously aware of the importance of music, a meta-cognitive process. The essence of
these developmental elements has been captured in short criterion descriptions.
Although social contexts were not explicitly exempli ed, they are implicit in the
right-hand dimension: in the concept of manipulative control that makes it possible for
people to make music together; in the idea of the vernacular: shared music processes;
and in the idiomatic with its implications of social conventions of style and genre. Even
the systematic, with its connotations of creating new musical processes and forms, can
be seen as social, in that existing conventions may be challenged. For these reasons the
original spiral diagram had an arrow from left to right captioned towards social sharing.
Further theoretical and empirical work has been reported in subsequent publications
(Swanwick, 1988, 1991, 1994). Consequently it can be argued that the theoretical basis
of the criteria for composing are good indicators of more general musical development
(Swanwick, 1994, p. 85).

Validity
There are of course issues of validity concerning this model and the extent to which it
may be said to give an adequate picture of the elements of musical development. One
of our criteria is that any evaluation of such a theory has to take into account the range
of musical phenomena it attempts to explain. This theory can hardly be attacked for
limitations of this type. The explicit integration of expressive, formal and value
dimensions is musically more comprehensive than many previous attempts to model
musical development in terms of limited functions, for example measuring discrimi-
nation between sound materials (higher and lower, or longer and shorter) or attempting
236 K. Swanwick

to study conservation of melodic or rhythmic elements. Nor is development assessed

through secondary functions, such as the ability to handle graphic or notational
representation, or by verbal description. Swanwick argues that the four layers with their
eight modes between them capture the richness of musical understanding (1994). This
has yet to be supplanted by an alternative theory of equivalent descriptive and analytic
power and educational relevance.
A second validity question concerns the relationship between the layers and age. Here
there have been misunderstandings. For example, both Davies and Marsh pointed out
that a child might work in several developmental layers at the same time (Davies, 1992;
Marsh, 1995). This is certainly true. Swanwick drew attention to the spread of the
compositions of individuals at any age and also to the cumulative nature of the
developmental sequence. We do not merely pass through one of these modes but carry
them forward with us into the next (Swanwick,1988, pp. 6364). Popular convention
asserts (wrongly) that Piaget thought each stage somehow separate from the others. For
example, Gardner asserted that for Piaget the child does not even have access to his
earlier forms of understanding. Once he is out of a stage, it is as though the prior stage
had never happened (Gardner, 1993, pp. 2627). This is a curious interpretation of
Piaget. For example, when writing of the development of children through what he calls
the successive structures (sensory-motor, symbolic, preconceptual, intuitive and rational)
Piaget told us plainly that it is essential to understand how each of these behaviours is
continued in the one that follows, the direction being from a lower to a higher
equilibrium. It is for this reason that in our view a static analysis of discontinuous,
strati ed levels is unacceptable (Piaget, 1951, p. 291).
Nor should we assume that the child in Piagets theory somehow stands outside of
cultural in uence. As Sera ne has pointed out, Piaget was an interactionist, emphasising
the reciprocal effects of both the external milieu and the internal cognitive structures
(Sera ne, 1980, p. 3). Similarly, the Swanwick/Tillman model supposed that musical
development is structural, embodying a coherent system, subject to transformations and
tending towards equilibrium between assimilation and accommodation, between personal
motivation and musical cultural conventions.
Although Swanwick and Tillman proposed an invariant sequence of development, in
that certain structures of musical thought precede others, there was no suggestion of a
rigid age timetable. In general though, very young children do not usually aspire to
idiomatic authenticity but enter the world of music with some excitement over sound
materials and enjoy control of them before engaging with vernacular conventions. By the
age of 14 years or so, making music in idiomatic ways becomes a strong imperative for
many young people. For Swanwick and Tillman the process of musical development was
not a once-in-a-lifetime, linear affair. The broken ends of their helix indicate that the
layers are recursive: for example, when people encounter new music, or a performer
begins to work on a new piece, or when a composer engages with a new computer music
program, in these situations of challenge they are likely to nd themselves once again
at the start of the process.

Reliability
Questions have been raised as to whether the initial assessment or coding of the
compositions was suf ciently objective (Hargreaves & Zimmerman, 1992). There are
also issues concerning the sample of children: for instance, could these ndings be
repeated in another cultural setting? Further, there was a relative dearth of data from
 Musical development theories revisited 237

older children to support the valuing layer of the spiral, which is not adduced from
compositional data. Koopman also drew attention to this conceptual discrepancy between
the rst three layers and the fourth.
Partly to meet such legitimate concerns, a replication of the original study was
conducted in a different cultural setting, that of the Greek part of Cyprus (Swanwick,
1991). Over 600 recordings of childrens compositions were collected and from these 28
were selected at random with a single sorting rule, that there should be seven items from
each of four age groups: 4/5, 7/8, 10/11 and 14/15. Seven primary and secondary music
teachers were asked to independently assign each of the compositions to one of the
criterion statements on a best t basis. The relationship between the actual ages of the
children and the placing of compositions by the spiral criteria was statistically
signi cant. There was a clear ascending relationship between age and the order of the
criteria and high levels of inter-judge agreement (Swanwick, 1994, p. 111).

The Musical Modalities of Musical Production, Performance and Perception

It is not correct to say that the Swanwick and Tillman model is based on childrens
compositions and cannot explain activities in other modalities (Hargreaves and Zimmer-
man, 1992, p. 388). The speci c data of this study is indeed that of childrens
compositions but as we indicated earlier, major elements of the theory derive from
earlier work by Swanwick on the nature of musical knowledge, on a generic epistemol-
ogy. There is also some empirical evidence on this issue of the effectiveness of the
theory to deal with what Hargreaves and Zimmerman call musical production and
musical perception. Similar criteria to those for composition have been employed in the
assessment of musical performance and high levels of agreement have been reported
between judges on the hierarchical order of these performance statements (Swanwick,
1994, pp. 108110).
The structure and theoretical concepts of the developmental spiral have also been used
by Hentschke in both England and Brazil to examine childrens perception of music as
audience-listeners (Hentschke, 1993). Younger children tended to comment on sound
materials and expressive character while reference to musical form appeared mostly
among the children around and after the age of ten, hardly ever before. It seemed that
the development of audience-listening follows the same sequence as composing and that
the layers of musical understanding might indeed provide a reasonably robust genetic
epistemology for music. Franca conducted a study using a version of the criteria in
Portugal for the area of audience-listening and found signi cant judge accordance in
sorting the randomised sets of statements into a hierarchy which matches the predicted
order (Silva, 1998; Swanwick & Franca, 1999).
This study and its associated theory raise an interesting and important issue related to
modalities. If it is possible to have a meta-theory of musical development, which applies
to production, performance and perception, might an individual show developmental
differences between the modalities? As Koopman (1995) said, the concept of musical
understanding overarches composing, interpreting, listening and judging.
It is important to distinguish between activities and understanding, between behav-
iour and cognition. We may characterise cognition or understanding as one of the
residues of activity. It is what we take away with us when the activity is over and
bring back on subsequent occasions. The same is true of skills. It is not illogical to
say that a tennis player is exceptionally good, even at a time when she is not actually
playing. In the same way we may have a high level of understanding
238 K. Swanwick

even if on speci c occasions we are unable to demonstrate it. For example, if we

converse for a time with a fairly normal child of 4 or 5 years we are likely to
become aware of a wide vocabulary and a con dent use of most grammatical forms.
Conversely the written language of the same child is likely to reveal much less linguistic
ability and would not be a good indicator of the childs language (rather than literacy)
ability.
The particular activity, the modality of articulation, can to a greater or lesser extent
reveal or conceal levels of understanding. For instance, a gifted jazz improviser who may
be asked to perform at sight from a dif cult notated score by Debussy may be unable
to demonstrate a high level of musical development. Similarly, a uent and sensitive
performer of notated western classical pieces may be quite inarticulate if asked to
improvise. Only if we have at least an implicit theory over-arching diverse activities is
it possible to make this kind of comparison.
Franca argued that the theoretical basis of the spiral offers a valid general
theory of musical understanding (Silva, 1998; Swanwick & Franca, 1999). She worked
with 20 Brazilian children between 11 and 13 years of age. Each child made tape
recordings of three memorised piano performances, recorded three of their own
compositions, and discussed and made written notes on three recorded pieces of
music, all of which were heard three times. Judges (experienced teacher-
musicians), using criterion statements based on the eight spiral modes, assessed these
products in the three modalities of performing, composing and audience-
listening. There were high levels of inter-judge reliability. An interesting nding is
that, while most of the childrens work displayed matching levels of musical understand-
ing for composing and audience-listening, their performances were usually less
developed. The education relevance of this kind of study may be signi cant, partic-
ularly in terms of the range of activities suggested in, or mandated by curriculum
guidelines.
In evaluating this work according to our criteria, it seems that the theory and
associated evidence is valid in that it comprehensively re ects the nature of music across
a range of activities. The various empirical replications and extensions of the project
strengthen the claims of reliability. The model is premised on qualitatively differentiated
patterns of behaviour and early structures are integrated into later ones. There is no
evidence to contradict the claim that the musical layers (or stages) unfold in an invariant
sequence and there is some evidence that the developmental sequence is culturally
widespread.
One outstanding issue is the relationship of the spontaneous natural develop-
mental of individuals and the cultural environment in which this development
is realised. The left and right sides of the spiral raise but do not develop this
issue and work is needed along Vygotskian lines, exploring musical production
and perception in terms of the zone of proximal development, looking particularly
at the in uence of language and peer-modelling. As mentioned earlier, Vygotsky
saw development as a spiral, passing through the same point at each new
revolution while advancing to a higher level (Vygotsky, 1978, p. 56). It should
also be noted that the original spiral diagram had an arrow pointing from left
to right labelled towards social sharing. The manipulative, vernacular, idiomatic and
systematic modes are essentially social and accommodatory. The theory does not
therefore ignore or rule out the social dimension, although much work remains to be
done in this area.
 Musical development theories revisited 239

Relationships amongst the Theories

There are a number of possible connections amongst some of the theories we have
reviewed. For example, the left and right of the spiral may be compared to Bambergers
intuitive and formal types of musical knowledge (Bamberger, 1978) and with what
Hargreaves calls empiricisma childs idiosyncratic construction of experience and
rationalismformal, adult ways of thinking (Hargreaves, 1986, p. 78). Similar distinc-
tions are also embodied in Bambergers gural and metric features of childrens
notations, though she resisted equating the representation conventions of metric notation
with musical understanding and warned that, where school music concentrates on
reading and writing standard notation, there may be a loss of intuitive musical
sensitivity. As Hargreaves pointed out, phrase shapes and various subtleties of articula-
tion are left out of metric notations but may be captured more effectively in gural
shapes (Hargreaves, 1986, p. 99). Bamberger believed that formal music education
linked to metric notation may contribute to a loss of musical sensitivity. Swanwick also
warned against losing opportunities to experience the left-hand elements of the spiral
model in the formal transactions of classrooms (Swanwick, 1994).

Implications for Teaching and Learning

From this review of developmental theories a number of implications can be drawn for
teaching and learning. There seem to be two main areas in which we see the implications
of a comprehensive developmental theory for musical transactions in classrooms and
studios.
The rst implication is in terms of broad curriculum planning in schools and colleges.
Curriculum activities may be focused toward speci c aspects of musical development at
different broadly identi ed stages, working with the grain of development. In the very
early years of schooling and at pre-school level, sensory exploration and the encourage-
ment of manipulative control would be the main aim. In the rst years of schooling this
could be taken further forward and expressive elements of music would come more
sharply into focus. Kratus noted that there are few studies of childrens emotional
response to music. His study of the responses of children aged 612 years suggested that
his sample interpreted the emotional character of a piece of music (Glenn Gould playing
excerpts from Bachs Goldberg Variations) with a high degree of consistency (Kratus,
1993). His conclusions included the view that, since 6-year-olds were able to interpret
emotion in music to the same extent as those children who were 12, formal music
education does not seem to have an effect. Nor should teachers feel they need to teach
students the associations of happy/sad, excited/calm though these might be starting
points for analysing musical elements.
While this may be true, there surely is work for educators to do assisting in the process
of re ning perception of feeling qualities and in facilitating students expressive
production. This may involve movement, dance, drama and visual images, all of which
might help promote, stimulate and intensify awareness of expressiveness. By the age of
10 or so we would also be looking to further the production and recognition of musical
speculation, an awareness of the uniquely contextual nature of the contrasts and
repetitions essential to musical form.
Young people in high schools are often seeking to enter a grown-up world of music.
The resources of formal education are not always organised to match the musical
240 K. Swanwick

development of young people. These include appropriate instrumental and performance

settings and equipment, along with the expertise of a range of teachers, so that there can
be some basis of choice for the student who is becoming idiomatically aware or who has
high music aptitude with commensurate achievement. Drawing on musical agencies
outside the school may be one powerful strategy in providing elements of idiomatic
authenticity.
The second way in which an understanding of musical development might inform
music education concerns individual development. Although schools are organised into
classes, people develop as individuals. It is surely helpful for a teacher to be able to
relate to developmental change with awareness of subsequent possibilities. For example,
that engagement in the vernacular may evolve into the more imaginatively speculative.
This may enable the teachers to ask a stimulating question, make a suggestion or choose
material that may have more meaning and developmental consequences for the individ-
ual. For example, when beginning to learn a new instrument or to perform a new piece
we may observe and assist student development. It may be necessary from time to time
to move to a lower layer in order to move to other layers more musically and securely.
For instance, it may be a bene t to take an instrumentalist back to the enjoyment of the
exploratory sensory mode.
The recursive implications here are suggestive for organising teaching sequences or
musical projects. In one sense progression in music may be thought of as linear,
quantitative or horizontal. We may see education in music as proceeding from relatively
simple to more complex and technically advanced material. Skill and content sequences
can be used to guide the horizontal sequence of the curriculum with long- and
short-range objectives, providing a sequential scheme for introducing and extending
awareness and control of musical materials and expressive features. But progression may
also be seen as layered, qualitative or vertical. For example, a curriculum sequence may
be viewed as recurring spirals along the horizontal where the level of musical material
changes over time, perhaps from highlow or loudsoft distinctions, through pentatonic
melodies to other tonal series or harmonic progressions. The vertical dimension
informs the critical judgement of teachers and students, and raises the question of
quality. How many layers are involved in musical production or in audience response.
Students would thus be provided with a foundation for developing musical understanding
in an organised and coherent way without losing the potential richness of musical
experience.

Conclusion
Very broadly, the main implications for music education to be drawn from this review
are as follows:

Research has been able to identify patterns of musical development, some of these in
speci c areas of musical perception and production and having generic theoretical
implications.
These theories have broad musical validity, are relevant across different activities, take
account of both nature and nurture, identify qualitative, sequential and hierarchical
changes, have widespread cultural application and are supported by systematically
gathered data.
Musical development may be generically construed as consisting of cumulative layers.
 Musical development theories revisited 241

Understanding musical development is suggestive for curriculum design and for

organising instructional activities.
Once students have developed beyond early childhood, they may move freely between
all or any of the layers, provided that the activity is rich enough in musical
possibilities.
Developmental sequences are recursive and may be reactivated in new musical
contexts.
Integration of different musical modalities may provide opportunities for the develop-
ment of musical cognition and response.
As well as providing musical models and structures, it is important to leave room for
assimilatory activity, where the student absorbs and decides for him or herself.

While there is still considerable information we do not have about how individuals
develop musically, we do know quite a lot about musical development and how to
promote it. There is strong evidence from several sources for a broad developmental
sequence during early childhood and that this may be reactivated subsequently with new
musical experiences. We also know that education may positively in uence this.
However, it is not at all clear that formal education (schooling) is necessarily the
optimum environment. Much work remains to be done on musical learning outside of
instructional programs as part of lifelong enculturation and on the effect of different
levels of instructional framing, especially less-directive and more loosely sequenced or
informal teaching and learning in the early stages of ones development. The future of
music education may not depend so much on schools as we know them but on such
things as opportunities in local communities and the global communities of the web.
Musical development is likely to become increasingly problematised and pluralised and
we may expect existing models and theories to be further challenged.

Acknowledgement
The author is grateful to have discussed the material in this paper with Maria Runjola
of the University of Buffalo, with whom he has recently collaborated on a larger study
of this topic.

REFERENCES
BAMBERGER, J. (1978) Intuitive and formal musical knowing: parables of cognitive dissonance, in: S.S.
MADEJA (Ed.) The Arts, Cognition and Basic Skills (St Louis, Missouri, Cemrel).
BAMBERGER, J. (1991) The Mind Behind the Musical Ear (Cambridge, Massachusetts, Harvard University
Press).
BARRETT, M. (1997) Invented notations: a view of young children s musical thinking, Research Studies
in Music Education, 8, pp. 214.
BENTLEY, A. (1966) Measures of Musical Abilities (London, Harrap).
BROPHY, T.S. (2000) Assessing the Developing Child Musician (Chicago, Illinois, GIA Publications Inc.).
BUNTING , R. (1977) The Common Language of Music: Music in the Secondary School Curriculum (York,
York University).
CRAIN, W. (1992) Theories of Development (New Jersey, Prentice Hall).
DAVIDSON, L. (1983) Tonal structures of children s early songs, Bulletin of the British Psychological
Society, 36, pp. 119120.
DAVIDSON, L. & SCRIPP, L. (1989) Education and development in music from a cognitive perspective. In
D.J. HARGREAVES (Ed.) Children and the Arts (Milton Keynes, Open University Press).
DAVIDSON, L. & SCRIPP, L. (1992) Surveying the coordinates of cognitive skills in music, in: R. COLWELL
(Ed.) Handbook of Research on Music Teaching and Learning (New York, Schirmer Books).
242 K. Swanwick

DAVIES , C. (1992) Listen to my song: a study of songs invented by children aged 5 to 7 years, British
Journal of Music Education, 9, pp. 1948.
DOWLING , W.J. (1982) Melodic information processing and its development, in: D. DEUTSCH (Ed.) The
Psychology of Music (New York, Academic Press).
DOWLING , W.J. (1984) Development of musical schemata in children s spontaneous singing, in: A.J.
CROZIER (Ed.) Cognitive Processes in the Perception of Art (Amsterdam, Elsevier).
GARDNER , H. (1973) The Arts and Human Development (New York, Wiley).
GARDNER , H. (1993) The Unschooled Mind (First published 1991) (London, Fontana).
HARGREAVES, D.J. (1986) The Developmental Psychology of Music (Avon, Cambridge University Press).
HARGREAVES, D.J. & GALTON, M. (1992) Aesthetic learning: psychological theory and educational
practice, in: B. REIMER & R.A. SMITH (Eds) Yearbook of the Arts in Education (Chicago, NSSE).
HARGREAVES, D.J. & ZIMMERMAN, M. (1992) Developmental theories of music learning, in: R. COLWELL
(Ed.) Handbook of Research on Music Teaching and Learning (New York, Macmillan).
HENTSCHKE, L. (1993) Musical development: testing a model in the audience-listening setting, Unpub-
lished PhD (University of London, Institute of Education).
KOOPMAN, C. (1995) Stage theories of musical development, Journal of Aesthetic Education, 29,
pp. 4966.
KRATUS, J. (1993) A developmental study of children s interpretation of emotion in music, Psychology
of Music, 21, pp. 319.
MARSH, K. (1995) Childrens singing games: composition in the playground? Research Studies in Music
Education, 4.
PIAGET, J. (1951) Play, Dreams and Imitation in Childhood (Norton Library, 1962 edn) (New York,
Norton and Co).
SERAFINE, M.L. (1980) Piagetian research in music, Council for Research in Music Education Bulletin,
62, pp. 121.
SERAFINE, M.L. (1988) Music as Cognition: the Development of Thought in Sound (New York, Columbia
University Press).
SILVA , M.C.C.F. (1998) Composing, performing and audience-listening as symmetrical indicators of
musical understanding, Unpublished PhD (University of London, Institute of Education).
SWANWICK , K. (1979) A Basis for Music Education (London, Routledge).
SWANWICK , K. (1983) The Arts in Education: Dreaming or Wide Awake? (London, University of London
Institute of Education).
SWANWICK , K. (1988) Music, Mind and Education (London, Routledge).
SWANWICK , K. (1991) Further research on the developmental spiral, Psychology of Music, 19 pp. 2232.
SWANWICK , K. (1994) Musical Knowledge: Intuition, Analysis and Music Education (London and New
York, Routledge).
SWANWICK , K. (1999) Teaching Music Musically (London and New York, Routledge).
SWANWICK , K. & FRANCA, C.C. (1999) Composing, performing and audience-listening as indicators of
musical understanding, British Journal of Music Education, 16, pp. 317.
SWANWICK , K. & TILLMAN, J. (1986) The sequence of musical development: a study of children s
composition, British Journal of Music Education 3, pp. 305339.
TILLMAN, J. (1987) Towards a model of the development of musical creativity: a study of the
compositions of children aged 311, Institute of Education (London, University of London).
VYGOTSKY , L.S. (1978) Mind in Society: the Development of Higher Psychological Processes (Cam-
bridge, Massachusetts, Harvard University Press).