Proceedings Paper


An Empirical Investigation of the Social and Musical Processes Involved in Children's Collaborative Compositions.

Raymond A.R. MacDonald, Department of Psychology, Glasgow Caledonian University, Cowcaddens Road, Glasgow G4 0BA, UK.
Telephone: + 44 (0) 141 331 3971
Fax: + 44 (0) 141 331 3636

Dorothy Miell, Department of Psychology, The Open University, Walton Hall, Milton Keynes MK7 6AA
Telephone: + 44 (0)1908 654 546

Laura Mitchell, Department of Psychology, University of Strathclyde, George Street, Glasgow GI 1QE, UK.



This paper reports two studies which have investigated the impact of social factors have upon children's musical creativity. We have been concerned to explore these social factors since making music is so essentially a social process, particularly in the collaborative settings of UK classroom music lessons, with the interaction between children both affecting and being affected by the evolving music. In exploring the processes involved we have drawn on the literature from social and developmental psychology, particularly the growing literature on collaborative learning, which emphasizes the importance of quality communication between children and the importance of agreeing shared goals and ways of working together (Kruger, 1993; Rogoff, 1990). However, most of the studies conducted on collaborative learning are of children's maths and science work, and there are few empirical studies of children's collaborations on more open-ended tasks such as music composition or creative writing (see Johnston, Crook & Stevenson, 1995 for an exception). A crucial aspect of studying collaborative music making is that music affords a channel of communication other than verbal interaction (Morgan, 1999). In addition, there is a reciprocal interaction between the ongoing musical and verbal communication between children in this context (MacDonald, Miell & Morgan, in press)

A key variable investigated in the studies reported here is the effect of an existing relationship between the children on the way in which they communicate and work together. We expected that, given the need for quality communication and for establishing a 'shared social reality' (Rogoff, 1990) in order to achieve successful collaboration, working with a friend would be particularly helpful for children. This might be expected to be particularly the case in creative, open-ended tasks where the children have to not only work together on the task itself (eg composing a piece of music), but also have to define the goals of their work and negotiate with each other without there being a 'right answer' to guide them, as well as stimulate and build on each other's creative input. Such interactional work is, we hypothesised, more likely to be achieved successfully when a child is working with someone they have an established friendship with, where they have experience of working, talking and playing together successfully.

Study One

In this study, 10-11 year old children were asked to compose a piece of music entirely of their own and in a style of their choosing to reflect the theme of 'the rain forest'. The children all began their involvement with the project by attending a workshop with one of the researchers during which they experimented with different instruments, rhythms, dynamics etc and discussed ways in which compositions can be developed and different effects achieved. The experimental sessions involved the children working on their compositions in same sex pairs and they were given 15 minutes to complete the task, using a full range of instruments typically available to them in school music lessons (tuned and untuned percussion and keyboards). Half the children worked on the task with one of their best friends while the other half of the children worked on the task with a child from a different class who they would have known by sight but who was not a friend.


We were interested in both the nature and quality of the interactive process as well as in the quality of the musical end product, and with this in mind we videotaped all the composition sessions and also recorded onto an audiocassette each pair's final performance of their composition. All the verbal utterances and musical motifs from the videotapes were transcribed and the talk was then coded in accordance with a system introduced by Berkowitz et al. (1980) and developed by Kruger (1992). This coding system divides utterances into 'transactive' and 'non-transactive' types. Transactive communication is defined as communication which builds upon and extends ideas that have already been voiced (either by the self or the partner) and the presence of transactive communication has been shown to be a key factor in good quality collaboration. We adapted this verbal coding system to allow us to also code the music played by the children as either transactive or non-transactive and to track the occurrence and elaboration of each musical motif throughout the composition session (MacDonald, Miell & Morgan, in press). The final compositions were rated for quality by a teacher from another school who worked from the audiotape of each composition and was unaware of the hypothesis of the experiment, the experimental conditions and all details of individual pairs. She rated the compositions using a set of marking scales developed by Hargreaves, Galton & Robinson (1996).

The results of this study highlighted the impact that social factors such as friendship have upon both the process and outcomes of children's collaborative compositional work. Looking first at the outcome measure, the teacher rated the compositions produced by friends as of significantly higher quality than the compositions of children who had been working in non-friendship pairs. Having established this difference in the overall quality of the music produced, we then turned to the measures of the processes involved in the talk and music of the interaction to see if there were also differences there which related to the outcome scores. We found that both the musical and verbal communication styles of the friendships pairs were qualitatively different from those of the non-friends. The friends both spoke and played more music in total than the non-friends, but also had a different pattern of interacting within these overall differences in amount. The friendship pairs used proportionally more transactive communication in both the verbal and the musical domains than the non-friends. This meant that the friends were building on, extending and elaborating on each other's ideas, expressed in both the talk and music, and developing their compositions by this gradual process of offering and refining suggestions. This style of interaction was found to be significantly positively related to the teacher's higher score for these pairs, suggesting that the presence of more transactive communication was what led to the higher quality compositions from the pairs of friends. In contrast, non-friends were more likely to spend their time in the session experimenting with the instruments for themselves and did not offer up or develop ideas together in the same way. The smaller amount of talk which they produced was characterised by information giving and simple, unelaborated agreements and disagreements with each other. Sometimes the music seemed to be played to cover their embarrassment and the lack of talk between them.

Thus it appears from this study that social factors such as friendship are key variables that influence the nature of children's interactions - in both the verbal and musical domains. The musical coding scheme which we developed allowed us to track interactive processes expressed musically as they occurred in the composition sessions and holds great promise for future studies of other groups and pairs collaborating to compose and improvise.


Study Two

A second study was designed to extend the first. Two key issues were identified as important for further investigation. The first was to explore the extent to which the friendship effect found in the first study might generalise to other settings. In particular, given the finding by Azmitia & Montgomery (1993) that working with a friend mainly helped children when they tackled difficult tasks, we wanted to vary the difficulty of the task. One way of changing the difficulty was to change the level of structure and guidance given to the children, so that children would find they had fewer choices and decisions to make for themselves. The participants in this case were limited to using only a keyboard and to starting their composition with a middle C (instructions derived from a study by Kratus (1989) looking at the composition process in 7 year old children). In order to see whether the friendship effect was also found in younger children, this second study also involved 8 year old as well as 11 year old children. As in the first study, children worked in same sex pairs of friends or non-friends, with one child in each pair being more experienced musically than the other. Again as in the first study, the interactions between the pairs and the outcome of the collaborations were examined. Dialogue between the pairs of friends and their musical interaction were analysed using measures of transactive and non-transactive communication. The musical processes used by the pairs were also examined. A school music teacher finally graded each composition.

Results highlighted that older children and those working with a friend took part in more transactive communication - both in their dialogue and in the music that they played. At 11 years old it appeared that there were no differences between the friends and non-friends in either the amount of transactive communication or the scores received for the compositions, whereas at 8 years old the differences between friends and non-friends were more apparent. Compositions by younger children paired with a friend were given a higher score by the teacher and used more transactive verbal and musical communication. In the analysis of the musical processes used, it also became clear that 8 year olds paired with a friend were able to organise their time to include sufficient quantities of development and rehearsal of their piece in a manner similar to that of the older children. The 8 year olds paired with a non-friend, however, were found to spend most time in individual exploration of sounds or silence.


The 8-year-old children paired with a non-friend took part in considerably less transactive dialogue than the older children and than those of the same age paired with a friend. At 11 years of age, little difference could be seen between the discussion style of the friends and non-friends, both taking part in high amounts of transactive and useful non-transactive dialogue such as making proposals. In this age group, scores given to the compositions were similar for the two groups, suggesting that the ways of working together on a structured music task were of a similar nature by this age level whether one is working with a friend or acquaintance, although we had observed differences in the previous study where the task was unstructured. At 8 years of age, however, the relationship with the partner appears to make more impact on the type of discussion that took place. At this age level, those paired with friends took part in more other-oriented transactive communication - in terms of their statements, questions and responses. The group scoring the lowest final composition score, the 8-year-old non-friends, were found to use the least amount of dialogue overall, in both transactive and non-transactive categories. Their counterparts of the same age paired with a friend, meanwhile, were discussing the task in a style much closer to that of the 11-year-old participants.

In order to find out whether the musical interaction between the children matched the differences found in the verbal interaction, an analysis of the amounts of transactive and non-transactive musical communication was then carried out. In comparing the two age groups, it became clear that the younger children had spent significantly more time playing 'music for self' - experimenting individually without any apparent attempt to communicate with the partner. The older children had also taken part in a significantly greater amount of musical repetition, rehearsing their composition more thoroughly. The variable of friendship also revealed expected differences such as the non-friends playing more 'music for self', and the friends using a greater number of other-oriented transactive musical motifs. Although these effects for friendship and age separately were found, it was again the interaction between these two variables that clarified the different ways of working in the pairings.

As found in the verbal interaction, little differences in musical communication could be seen between the friends and non-friends aged 11. In the 8-year-old group, however, the amount of transactive musical communication was found to be more than double that of the non-friends. At the same time, the musical activity in the friends of 8 years consisted of greater music for self, and less musical repetition. Whereas the older children and the younger children paired with friends spent time working on one another's suggestions, therefore, this group devoted most time to experimenting with sounds individually, and without aiming to rehearse and close on one musical product.

The musical techniques used by each child were analysed using the categories devised by Kratus (1989). In his study he examined the minute-by-minute development of the children's music as it fell into 4 categories: repetition, development, exploration and silence. The same analysis was conducted on the music played by the various pairs in the present study. The older children structured their time to include greater repetition, which gave a greater opportunity for rehearsing their composition. The 8-year-olds, meanwhile, were found to have longer periods of silence and of musical exploration - playing for themselves rather than their partner. The younger children spent more time in complete silence with no musical or verbal interaction going on, and when actually playing were more likely to be experimenting with new musical ideas for themselves. It was apparent that in the younger age group it was the non-friends who took part in a significantly greater amount of exploration. This group was also found to spend the least amount of time on repetition.

The analysis of the processes used at different times during the 10-minute period furthermore revealed the planning of time by children in each age group and type of friendship pair. The pattern of playing over time for the 11-year-old friends and non-friends could be seen to be generally similar, with repetition as the predominant technique, increasing over time in order to allow for sufficient rehearsal before the end. Exploration during this time decreased gradually, and development increased around the middle of the session yet decreased later in order to make way for further repetition as the end product was finalised.

The 8-year-old children working with a friend appeared to use their time in a manner similar to the older children, with only a slightly greater amount of silence distinguishing their use of the processes. As exploration and on-task talk decreased from their high levels in the first few minutes, repetition gradually increased to become the predominant process. The 8-year-old non-friends, however, as might be expected from the previous analyses, could be seen to plan their time more poorly, with large amounts of silence and little repetition or development.

The findings from the current analysis using Kratus's categories are therefore in line with those of his 1989 study, which had concluded that young children (7 year olds in his study, 8 year olds here) spent most time devoted to musical exploration compared to the equal division of time between exploration and repetition observable in our 11 year old participants. Again, in ratings given to the second performance of each composition, the current findings also backed those of Kratus, with older children able to replicate their composition more accurately, probably due to the increased attention given during the composition time to repeating the piece. The most noticeable difference, however, to Kratus's suggestion of the link between age and compositional ability was in the 8-year-old children paired with a friend. These pairs were found to use their time in a manner much closer to that of the older children, developing and repeating their piece in a way which led them to replicate their piece in a similar way to the 11 year olds. The 8-year-old non-friendship pairs meanwhile were found to use their time in an unstructured, unplanned manner that appeared to lead to their poor ability to replicate their composition accurately.

In comparing the current findings to those of Kratus, the theory of 'product' or 'process' orientation appears to be of particular significance. Kratus concludes that the 7-year-old children's compositional process consisted of trying one musical idea after another, their orientation appears more towards the process of making sounds rather than the actual production of one replicable melody. As musical exploration took precedence, therefore, no closure on the musical product was able to occur. Older children, however, undertook a process similar to that of adult composers, structuring their time to include sufficient amounts of development and rehearsal. Kratus therefore defines this as product oriented - aiming towards the refinement of one musical product.

It is clear from the current study that the process of working with a friend allowed the 8 years old children to be product oriented, a feature generally not observed when children of this age work individually on a composition. Whereas the 8 year olds working with a non-friend partner spent their time exploring new sounds, it appears that those working in a friendship pair came to realise quickly that rehearsal would be a necessary activity in order to complete the task successfully. The 8-year-old friends therefore took part in an amount of repetition similar to that of the older children.

Unlike the observation of children working individually at composition in Kratus's study, the work of Azmitia and Montgomery (1993) aimed at similar goals to the current study, examining why collaboration between friends might lead to increased success. Focusing on scientific reasoning skills, their findings suggested that the greater mutuality and involvement between friends led to greater support between the pair. This proposal is therefore backed by the current findings of greater amounts of transactive discussion between friends, particularly of the other-oriented variety where the children makes elaborations, responses to and revision of a partner's ideas rather than their own. A higher level of self-oriented transacts also between friends, however, relates to the suggestion by Newcomb and Brady (1982) that friends are more aware of the need to justify and explain one's ideas to a partner. It appears, therefore, that those children working with a friend felt more at ease being engaged with each other in the task, taking time to discuss aspects of their work more thoroughly.

One aspect of the collaborative work of friends suggested by Azmitia and Montgomery (1993) was that of greater checking and evaluating of solutions, a feature which in fact became apparent when carrying out the study. Watching the friends at work together even at this point revealed that a noticeable part of their discussion revolved around requesting to each other that they practice again, going over fine points quickly in order to do so. The findings of the current study particularly relating to the 8-year-old children were therefore found to support this theory. Whereas 8 year olds working with a non-friend took part in little rehearsal of their composition, those working with a friend used repetition within their 10 minutes at a level close to that of the 11 year olds. Relating to the suggestion of Nelson and Aboud (1985) that criticism of the partner's work allows friend to give a superior performance, it was also confirmed that friends do check and make suggestions on the partner's work to a greater extent, taking part in greater verbal and musical transactive other-oriented suggestions.

A further proposal from the work of both Azmitia and Montgomery (1993) and from Newcomb and Brady (1982) is that differences in performance between friends and non-friends are more likely to emerge in challenging tasks rather than straightforward ones. This can be seen as one explanation for the advanced interaction and superior performance of the younger friends, in that due to the younger children having had less musical, and indeed less collaborative problem solving experience in general, the task obviously presents a more challenging assignment for them than for the older children. For the older children, we would suggest that no differences were found between friends and non-friends in this study since the greater degree of structuring to the task made it easier than the more open ended task set in Study 1, where differences between 11 year old friends and non-friends were found. In Study 1, the children had to make a number of choices and decisions for themselves and it was suggested that the friends were more successful since they had a more developed style for working together and making such decisions collaboratively. With this more structured task, the advantage of being with a friend was less apparent and the effect was not observed (as indeed is often the case in studies of structured maths and science tasks with 11 year olds).

Hartup (1996) suggested that the important feature of friends' collaborative work was the ability to establish 'joint productive activity' and this is backed up by several features of the children's interactions in the current study. It became clear when conducting the study that the 8 year olds working with a non-friend partner required much greater amounts of prompting in order to become involved in the task. Even after this, it appeared that the non-friends at this age were unable to move from one stage of the process to the next with ease, choosing to maintain a high level of experimentation throughout the 10 minutes, or indeed waiting in silence for periods of time until one partner could make a suggestion. Non-friends, then, had to struggle to try and establish a way of working together before any productive activity could take place (particularly as here when no structure was provided in the task instructions) - a feature which comes naturally to friends with a history of such interactions.

Despite the findings of this and previous studies of the effective verbal interaction style of collaborating friends, the fear of off-task talk between friends often appears to stop teachers from pairing children in this way. Previous work by Miell and MacDonald (in press) and Hartup (1996), however, found that contrary to expectations, friends spent less time in off-task talk than non-friends. Although it was observable when carrying out the current study that friends appeared often more tempted to play a tune they knew to entertain their friend, the overall lesser amount of general exploration by the friends suggests that this did not affect the productiveness of the collaborations. Although, therefore, more off-task talk was found between the friends, and in particular those of 11 years old, this was compensated for by the much greater amount of on-task talk by the friends. It appears in the case of the 8-year-olds that more off-task talk in the friends did not mean that they were communicating less effectively, merely that the non-friends generally failed to communicate with one another at all.

A clear picture of the effect of friendship in these two groups has emerged, revealing that being in a friendship pair does allow 8 year olds to engage in transactive discussion, musical interaction and effective use of musical processes in a way that their non-friend same age counterparts cannot, and that older children show the same pattern when working on a less structured task. These results contrast with suggestions by Harrison and Pound (1996) that setting composition tasks to younger children may curb enthusiasm and imagination, it appears rather that collaborating with a friend allowed 8 year olds to use their friendship as a resource in encouraging increased motivational, organisational and imaginative ability.



The two studies reported here were designed to investigate the social processes involved in children's creative collaborations. The studies focused upon the process and outcomes of both the musical and verbal interactions. The first study highlighted that when children aged 11 work with someone they know well they produce proportionally more transactive communication. That is, communication that developed upon ideas previously proposed. This result was evident in both in the verbal and musical domains In addition, the compositions produced by pairs of friends are rated as being of a higher quality than the compositions produced by children working with some they did not know. In study two it was found that children aged eight produced similar findings to the first study when they were working on a more structured musical tasks. No differences of this nature were found between friendship and non friendship pairs of children aged eleven in this study. It is suggested that differences in the nature of the musical tasks employed help explain the differences in the findings of the two studies.


Azmitia, M. & Montgomery, R. (1993) 'Friendship, transactive dialogues, and the development of scientific reasoning.' Social Development, 2 (3), 202-221

Berkowitz, M.W., Gibbs, J.C. & Broughton, J. (1980) 'The relation of moral judgement disparity to developmental effects of peer dialogue'. Merrill-Palmer Quarterly, 26, 341-357

Gottman, J. (1983) 'How children become friends'. Monographs of the Society for Research in Child Development, 48(3).

Hargreaves, D.J., Galton, M.J. & Robinson, S. (1996) 'Teachers' assessments of primary children's classwork in the creative arts.' Educational Research, 38 (2), 199-211

Harrison, C and Pound, L (1996) Talking Music: Empowering children as musical communicators. British Journal of Music Education, 13 (3)

Hartup, W.W. (1996) 'The company they keep: friendships and their developmental significance' Child Development, 67, 1-13

Johnson, P.G., Crook, C.K. & Stevenson, R.J. (1995) 'Childs play: Creative writing in playful environments' in H.C. Foot, C.J. Howe, A. Anderson, A.K. Tolmie & D.A. Warden (Eds) Group and Interactive Learning. Boston: Computational Mechanics Publications.

Kratus, J.K. (1989) A time analysis of the compositional processes used by children aged 7 to 11. Journal of Research in Music Education, 37, 1, 5-20

Kruger, A.C. (1992) 'The effect of peer- and adult-child transactive discussions on moral reasoning'. Merrill-Palmer Quarterly, 38, 191-211

Kruger, A.C. (1993) 'Peer collaboration: conflict, co-operation or both?' Social Development, 2 (3), 165-182

MacDonald, R.A.R., Miell, D. & Morgan, L. (in press, 2000) Social processes and creative collaboration in children. European Journal of the Psychology of Education.

Miell, D. & MacDonald, R.A.R. (in press, 2000). Children's creative collaborations: The importance of friendship when working together on a musical composition. Social Development

Morgan, L. (1999) 'Children's Collaborative Music Composition: Communication through Music'. Unpublished dissertation, University of Leicester, UK.

Nelson, J. & Aboud, S. (1985) 'The resolution of social conflict between friends' Child Development, 56, 1009-1017

Newcomb, A. F, and Brady, J.E (1982) Mutuality in boys' friendship relations. Child Development, 53, 392-395.

Rogoff, B. (1990) Apprenticeship in thinking: Cognitive development in social context. Oxford University Press: Oxford.


 Back to index