Prisons and primary schools

The multifaceted nature of music education takes us on unexpected journeys. After a career in woodwind and singing teaching within Primary and Secondary schools, including ensemble and choir direction, as well as some Secondary class teaching, a journey into music education in the Criminal Justice System was one that I did not expect. However, the opportunity arose to participate in a prison music project and I was struck by how similar the learning processes within a completely different context to my normal teaching environment were. Two common elements in these learning processes are the relationship between the learner and the tutor and the evolving identity, or transformation, of the learner.

The relationship between learner and tutor has been highlighted as a key element in the learning processes of older adults. In identifying the characteristics of a ‘good’ music facilitator, Hallam et al. (Reference HALLAM, CREECH, MCQUEEN and VARVARIGOU2011) found that learners believed ‘good’ music facilitators should respond to the learners and motivate them through supportive feedback. When discussing their learning experience throughout a week-long Javanese Gamelan prison project, similar characteristics were reported by participants (Henley et al., Reference HENLEY, CAULFIELD, WILSON and WILKINSON2012). Moreover, the relationship between facilitator and learner within a criminal justice context is considered a vital component in developing social capital (personal, professional and peer relationships and social interaction) and individual agency (confidence, motivation, ownership and self-determination) attributed to the process of moving away from a criminal identity known as desistance from crime (McNeill et al., Reference MCNEILL, ANDERSON, COLVIN, OVERY, SPARKS and TETT2011).

The process of desistance from crime involves a transformation of identity, and this has been a key theme in recent research in criminal justice. Within the Good Vibrations Project, Caulfield et al. (Reference CAULFIELD, WILSON and WILKINSON2010) found that participants had an increased sense of self-esteem and were able to identify themselves as members of the Javanese Gamelan within the prison. This contributed to a rise in confidence and a transformative effect leading to offenders taking on new responsibilities within their prison communities. In turn, this contributed to the development of a more positive identity that they hoped would take them through the transition back into the community. Transformation of identity has also been seen as key in wider offender education and preparation for transition into the community. Christopher (Reference CHRISTOPHER2013) explored the process of preparation for work given within prisons. She developed a model that can be applied to any employment training, which will help ex-offenders move into employment. The model comprises Identity, self-Esteem, Competence and Resilience (IECR), identity development being the first step in the transformative process; identity needs to be developed before self-esteem, self-esteem is vital in developing competence, and competence leads to developing resilience.

As well as being a key concept in recent research in criminal justice, identity development and transformation has also been discussed in music education. Freer (Reference FREER2009, Reference FREER2010) investigated issues relating to gender in choral singing using the concept of Possible Selves (Markus & Ruvolo, Reference MARKUS, RUVOLO and Pervin1989). He suggests that within a choral singing context, learners look for positive role-models that help to develop a positive possible self. This positive possible self is crucial in the development of a positive identity of a male choral singer (Freer, Reference FREER2010). Hoffman (Reference HOFFMAN2012) looked at the relationship between engagement, exclusion and identity construction in ensemble learning within a diverse socio-economic environment. She investigates the relationship between socio-economic status and academic success in the USA and how the structures of the ensemble contribute to the construction of identities, calling us to move away from hierarchical ensemble structures so as to foster emerging positive identities. Taylor's (Reference TAYLOR2011) study of older adults' musical learning experiences documents how identity is constructed at a mature age. Through analysing the life histories of her participants she was able to understand the complex nature of mature musical identity and the way adults use their youthful musical identity in learning later in life. These three studies all provide examples of the centrality of identity within transformative processes in different learning contexts; without the construction of a specific identity, change could not have occurred. Furthermore, the facilitator or tutor plays a considerable role in inspiring this change.

With common elements found in both music and prison education, it seems reasonable to draw on research carried out in these educational contexts to explore how the learning processes contributing to identity and transformation can firstly be viewed and secondly understood. However, as the contexts are so different, a conceptual framework that allows interactions between the individual and the social context to be viewed and understood, but is not specific to either context is needed. Cultural Historical Activity Theory (CHAT) provides such a framework.

CHAT and music education

CHAT is a way of looking at any activity involving a subject, or learner, and a desired object, or goal. At the core of CHAT is Vygotsky's (Reference VYGOTSKY, Van der Veer and Valsiner1930/1994) concept of mediation. Wertsch (Reference WERTSCH, Daniels, Cole and Wertsch2007, p. 178) explains:

In other words, people use tools, such as symbols and language, to carry out activity; the tools carry some kind of meaning that can be interpreted by the individual in the social context. Therefore, the tools act as mediators between the individual and the social and physical world. Mediation can occur explicitly in the form of tools that are intentionally introduced into an activity, but it can also occur implicitly. During the process of teaching, a teacher may introduce a way of assimilating information in order to enable the student to organise the information to gain meaning. For example, a music teacher may teach a student a rhyme so as to remember where notes lie on the stave. Once this is memorised, the student can then organise this information and realise it by interpreting the written notes into sounds.

The rhyme is an example of a tool of explicit mediation. Over time, the student internalises this tool until the process of reading music becomes what Hutchins (Reference HUTCHINS1995, p. 310) calls an ‘automized skill’. This skill has been mediated by a rhyme, however with continued interaction between the student and the notation, the rhyme has become obsolete yet the process still exists. Moreover, the notation then becomes itself an explicit tool of mediation between a physical activity and a sound.

The example Wertsch (Reference WERTSCH, Daniels, Cole and Wertsch2007) gives of implicit mediation is thinking aloud. When thinking and speaking at the same time, a person is exploring their thought processes through the tool of spoken language. In musical terms, an example of this is when a student is trying to learn to play a tune that they know, say for example Happy Birthday, by working out the notes aloud on an instrument against a tune in their head. They are exploring the musical sounds in order to find the ones that they want to fit in with the tune, the cognitive tools mediating the activity as it progresses.

The crucial factor of mediation is that individual cognitive factors and environmental factors (including other people and their cognitive factors) are not separated, but are integral to each other. Furthermore, as environmental factors are culturally and historically developed, the individual cannot be separated from the cultural historical context of the activity. Cole (Reference COLE1996, p. 103) tells us that

The example Cole gives to demonstrate this is that of tying a knot in a handkerchief as an external way to control memory. This highlights the point that tools are culture-specific in that if they were in a different context, they would not have the same meaning; without the culturally embedded knowledge of what the knot in the handkerchief means, the knot cannot mediate the act of recall. Moreover, a handkerchief is a cultural object and the act of tying a knot in a handkerchief to aid memory recall is a custom that is passed down through generations. If handkerchiefs are no longer used by a younger generation, perhaps replaced by paper tissues, the custom of tying a knot in a handkerchief might disappear. The disappearance of the custom would be the result of interactions between people and their environment. Furthermore, if we attempted to analyse why the custom has disappeared, we would find it difficult to separate out the younger generation from their environment.

Over the past 25 years, Engeström has developed CHAT and it continues to evolve as a theoretical concept (Engeström, Reference ENGESTRÖM1987, Reference ENGESTRÖM1990, Reference ENGESTRÖM, Chaiklin and Lave1993, Reference ENGESTRÖM, Engestom, Miettinen and Punamaki1999, Reference ENGESTRÖM2001, Reference ENGESTRÖM2005, Reference DANIELS and WARMINGTON2007, Reference ENGESTRÖM2008). The triangular model of the activity system put forward by Engeström (Reference ENGESTRÖM, Engestom, Miettinen and Punamaki1999) is shown in Figure 1. It is a framework for viewing mediation and practical activity within the context of the social environment. In other words, it is a way of ‘examining the relationships that hold between the individual and her environment’ (Barrett, Reference BARRETT, Miell, Macdonald and Hargreaves2005, p. 264).

Fig. 1. Engeström's activity system (Engeström, Reference ENGESTRÖM, Engestom, Miettinen and Punamaki1999, p. 31).

Engeström (Reference ENGESTRÖM, Chaiklin and Lave1993, p. 67) describes the activity system as incorporating

Therefore the functional aspects of an activity can be viewed in relation to the social aspects of an activity. The triangular model represents interactions within an activity between the different activity constituents. The subject, the object and the mediating artefacts (also known as tools) interact with the community of the activity, the rules of that community/activity and the division of labour within the community/activity to produce an outcome. In terms of musical learning, the learner is the subject, the object is the particular musical learning objective, and the tools are both physical instruments and cognitive processes (such as aural learning, note reading, non-verbal communication, etc.); a person learns to play the flute through manipulating their instrument, developing technique and engaging in musical expression and cognition. The person may be learning within a group, so the learning group is the community, the way the learning group is organised and managed (whether it is learner-led, teacher-led, whether there is peer learning etc.) constitute the rules, and the division of labour is the amount the person is contributing to the music; when playing with others, do they lead, do they follow, do they play everything, do they miss notes out and let others provide fluency? All of these things contribute to musical learning, and all are inseparable. For example, if a person is learning to play electric guitar, the type of community (a rock band) might predict the rules of the group (learner-led), and the rules of the group might predict the cognitive tools used (a mixture of guitar tablature and aural skills). The way the labour is divided in the group (guitar solos) contributes to musical development through increased expertise of guitar playing. Mapping these interactions using the activity system enables the analysis of how activity is taking place without separating individual cognitive processes from the socio-cultural and historical context, and provides a lens through which to view musical activity that recognizes music-making as a cultural practice.

In addressing the multifaceted nature of music education, Welch (Reference WELCH2007) discusses the use of CHAT as a methodological approach that allows the complex processes within musical learning to be viewed in a holistic way, so that the investigator can ‘combine both macro and micro perspectives’ (p. 27). In applying the activity system to a case study of female choristers within a previously all-male environment, Welch was able to begin to understand how female choristers are inducted into an all-male environment and the resulting cultural transformation of both the participants and the socio-cultural environment. Moreover, using the activity system framework to view the cultural environment of the cathedral chorister, Welch (Reference WELCH and Barrett2011) is able to find explanations for the maintenance of the choral tradition within cathedrals. With this case study, Welch highlights the potential for using CHAT to view both activity and transformation as a result of an activity.

Burnard and Younker (Reference BURNARD and YOUNKER2008, p. 63) applied Engeström's model of CHAT to the micro processes of musical learning so as to ‘identify and describe how cultural tools are used and characterise interaction in composing and arranging.’ Focusing on the collaborative interactions involved in group improvising, they used a qualitative methodology to observe children's responses to composing and arranging tasks. In analysing video observations and field notes, they applied CHAT as a framework to view the different types of interactions and collaborations taking place within the tasks. They found that within these tasks, there were in fact different activity systems at play; that composing and arranging involved different interactions and therefore different activity systems. They concluded that the triangular model of activity as viewed in the activity system may not be the best framework for viewing multiple activities.

An interesting example of how the activity system has been adapted to view a specific complex musical environment can be seen in Burrows' (Reference BURROWS2004) exploration of the interactions between musicians within an improvisation. Burrows applied CHAT to free improvisation within a performance context in order to answer questions relating to the nature of improvisation and how a group of musicians improvise together. For Burrows, the triangular model posed difficulties in the way the complex interactions within improvisation are identified and labelled. He found when he applied the model to free improvisation he was not able to understand improvisation through purely viewing the constituents within the activity. However by taking the constituents out of the activity system, and placing them within a model that linked group cognition and individual cognition, he was able to see ‘the contribution and interchange of musical ideas between individuals’ and understand ‘the aural tapestry which is a kind of nexus for distributing cognition’ (p.8). Burrows is able to manipulate the activity system so as to draw out the elements of activity and then find a way of linking together the activity of multiple subjects in order to understand the transaction between musicians within free improvisation.

Although Burnard and Younker (Reference BURNARD and YOUNKER2008, p. 62) say that ‘in music education, however, Activity Theory research remains relatively under-represented’, there is potential for using CHAT to view transformation within music education, and also potential to manipulate the constituents of activity and create a flexible activity system that can be applied to multiple contexts and activities.

Moving from two dimensions to three dimensions

As a doctoral student looking for ways to untangle the complexities of learning to play an instrument within an ensemble, discovering CHAT was a key moment. Whilst grappling with analysing data that related to both individual cognition and the social environment, trying to separate the two in order to understand what was taking place within the learning proved difficult. The different theoretical concepts being used related to individual learning within group activity (see Table 1). All indicated that context was important, but I had no theoretical underpinning of how these concepts joined together or supported each other.

Table 1 Main theoretical concepts brought together by the activity system.

Table 2 Attributes relating to the development of desistance from crime (adapted from McNeill et al., Reference MCNEILL, ANDERSON, COLVIN, OVERY, SPARKS and TETT2011).

Whilst exploring CHAT I recognized Russell's (Reference RUSSELL, Daniels and Edwards2004) description of context as a ‘web or network of sociocultural interactions and meanings that are integral to the learning.’ To this, Russell added that CHAT enables us to look ‘beyond the individual learner, the interface and the ‘material’ to understand the social and material relations that affect complex human learning' (p. 310). It helps us move beyond the what and into the how. I then set about applying CHAT to my data to see if it enabled me to look deeper into the learning process and understand these social and material relations.

The initial coding was regrouped into categories headed by the constituents of the activity system: subject; object; mediating artefacts; rules; community and division of labour. In doing so, the individual theoretical concepts that had been used to explain different aspects of the learning process could be brought together (see Table 1). Moreover, by bringing them together through the activity system the relationships between the concepts could be explored, providing an understanding of how they interacted and complemented each other.

A detailed snapshot of the learning processes and social interactions that come into play during a learning activity embedded in active music-making could be drawn up. However, the research findings also alluded to the transformative power of learning through active music-making, not only in terms of musical development but also in the development of identities, and this appeared to be related to the re-mediation of tools. Furthermore, identity seemed to be a key factor in the continuation of learning (Henley, Reference HENLEY2009). Russell (Reference RUSSELL, Daniels and Edwards2004, p. 311) acknowledges that mediation is an evolving process, suggesting that

Yet the activity system as it stood was not able to show me how the tools were re-mediated and therefore how progression and transformation occurred.

Daniels and Warmington's (Reference DANIELS and WARMINGTON2007) critique of CHAT highlights how contradictions within the activity system make it an unstable framework for analysis, with reference to three important aspects: contradiction and labour-power (i.e. the potential of the group being greater than the potential of the individual), subject positioning and identity within activities, and emotional experiencing in personal transformation. All three of these aspects also became problematic in using the two-dimensional system in my research; not only in that contradictions appeared, but also the activity itself changed at different points in the learning process. Sometimes these changes in activity were context-driven – moving from rehearsal to performance, but more often they were learner-driven. Deeper analysis suggested that these learner-driven changes in activity were initiated by the learner in order to cope with external events or internal perceptions of failure (Henley, in press).

Nardi (Reference NARDI1997, p. 75) recognises that the constituents of activity do change:

In the case of my doctoral research, not only did the constituents of an activity realign themselves as a result of changing conditions, they switched places with another constituent. So, whereas the object of activity might be to learn to play the clarinet, and the mediating artefacts may be an amalgam of different cognitive and environmental factors, the person carrying out the activity would change the object and mediating artefacts as it suited them. When the person was not feeling secure in their musicianship, musical identity or learning, they changed the object to something extra-musical, for example relieving stress, and music changed to the position of mediating artefact so as to account for why they were not ‘very good’ at that point in time: Playing the clarinet relieves my stress, I do not have to be ‘good’ at it.

On deeper analysis of changes in activities I realised that there was a common element: identity. Each time the activity changed, it was initiated by a change of identity. Furthermore, learners deliberately changed their identities in order to change the activity: I am not a clarinettist, I am a person who is stressed and this activity is therefore my stress relief. By doing this, learners were able to motivate themselves to continue learning rather than give up. This suggested that identity is the key that allows interaction between activities, and transformative process to occur as a result. After exploring a number of different models incorporating identity as a constituent, I devised a three-dimensional activity system that allows for different activities to be placed parallel to each other, and identity trajectories that allowed the switch between activities to be viewed. This activity system is shown in Figure 2.

Fig. 2. Henley's three-dimensional activity system (Henley, Reference HENLEY2009, p. 210).

This then provides a conceptual framework that acknowledges the fluidity of activity and enables the interactions between the individual, the socio-cultural context and multiple activities to be analysed and understood.

Three research projects; different methodologies, same analytical tool

Since its development, the expanded model of CHAT has been used in three subsequent research projects. Although located in very different social contexts, each project centred on collaborative music-making and investigated some aspect of identity development.

Project 1 – Music Partnership Project

The first project was an evaluation of a nine-month Music Partnership Project (MPP) funded by the Department for Education (DfE) that was carried out in Birmingham, UK (Hoskyns & Henley, Reference HOSKYNS and HENLEY2010). The objective of the project was to bring together a network of music education providers within the city who had begun working together collaboratively to provide music enhancement experiences for children in Primary and Secondary schools. The aims of the project were:

• • To reach as many children and young people as possible across the city.

• • To bring together different music educators through collaborative activity.

• • To bring together Primary and Secondary schools in order to improve transition.

• • To engage children and young people in creative music-making.

• • To provide continuing professional development (CPD), particularly for Primary teachers who feel under-confident in teaching music.

• • To develop partnerships between schools and music practitioners that can be sustained.

The project was complex and involved the creation of five clusters each comprising one arts organisation, one Secondary school and three Primary schools. Within the clusters, performing musicians from the arts organisations and instrumental teachers from the local music service worked with children both in individual schools and together as clusters to form creative ensembles. The clusters shared their work both within their cluster and across the clusters, and regular sharing events to facilitate this were organised as a key element of the MPP.

The MPP involved a commitment to working together and to develop shared understandings of creative music-making ensembles. Each organisation involved, including the schools, had different attitudes toward creative music-making and ensembles, but the intention was to share some of these in order to be better able to work collaboratively and cooperatively with a range of teachers, musicians and young people. Of particular interest within this was the way the Primary generalist teachers worked with Secondary specialist teachers, instrumental teachers and performing musicians to develop their understanding of creative music-making and enhance their own pedagogy. In addition to the development of shared understandings through collaborative working, CPD events were organised throughout the project. These included a Practitioners' Ensemble that brought together the different practitioners musically and who also performed at project sharing days.

The research team was engaged as external partners to evaluate the impacts of the project and how far the aims had been met. The evaluation brief was to take an objective stance and to look critically at the project in terms of how it was achieving its aims. The research used a mixed-methods approach and comprised three strands. The first strand involved an investigation of the impacts of the project on the musical development of the children and young people. This strand explored the ways in which the project enabled the participants to access music-making, and the impacts of working in a creative way on their perceptions of themselves as musicians. The second strand involved an investigation of the ways in which the parties collaborated. This strand explored the impacts of collaborations between different practitioners as well as collaborations between children, young people and practitioners. The final strand involved an investigation of the impacts in terms of the professional development of the practitioners. This involved an exploration of the transformations of the practitioners as teachers, and looked particularly at whether generalist Primary teachers were empowered to continue to teach music as a result of the project. Impacts were interpreted as being on both a macro and micro level. On a macro level, data were required to explore how many participants had accessed music-making as a result of the project, what their perceptions of the quality of this music-making was, how visible the project had been within the schools, and how visible the project was to the wider community. On a micro level, data were required to explore individual impacts in terms of the ways in which the project had enriched the musical lives of participants (both adults and children) and also how far the project had changed participants' perceptions of themselves as learners and teachers. This being the case, a mixed methodology was adopted that allowed the researchers to collect both quantitative and qualitative data, and triangulate these in order to construct an understanding of the impacts of the project, what led to these impacts and to see how far they related to the project's aims.

Data were collected throughout the nine-month project using four layers of data collection. Observation data was collected from 50% of participating schools (one Secondary and one Primary in each cluster), from all project sharing days, one CPD day, and one Practitioners' Ensemble. This was collected via video observation where parental permission had been granted and by audio recording and field notes where video permission was not granted. Interview data were collected via semi-structured interviews with Secondary, Primary and instrumental teachers and Cluster Leaders from each cluster, short semi-structured interviews with children both at sharing days and in school, and via dialogue with practitioners during observations. Qualitative and quantitative questionnaire data were collected via two online questionnaires; one for practitioners and one for students. Finally, statistical data were provided by clusters on participation rates across the project.

Video observations and interviews were transcribed and initially coded using a constant comparative process that allowed themes to emerge (Boeije, Reference BOEIJE2002). The thematic analysis was repeated with questionnaire data and themes were compared, categorised and analysed using the expanded model of CHAT as a framework. This enabled connections between the themes that emerged from the data and the processes within the music-making activities as seen by CHAT to be made, uncovering the micro impacts of the MPP. Statistical data were analysed so that an understanding of the macro impacts of the project could be constructed. These were then cross-referenced to see how the micro and macro impacts interacted. The results gave us a detailed understanding of the impacts of the project as a whole, the impacts of the project on each cluster and the impacts of the project on individuals.

Project 2 – Good Vibrations Javanese gamelan with male and female offenders

The second project investigated the short- and medium-term impacts of a Good Vibrations Javanese gamelan project within prisons (Caulfield et al., Reference CAULFIELD, WILSON and WILKINSON2010). The research involved four simultaneous research strands:

• • Strand one: male offenders who had participated in a Good Vibrations project previously in order to investigate the medium-term and potential long-term impact of the project.

• • Strand two: male offenders who had recently completed a Good Vibrations project in order to investigate the short-term impact on male offenders.

• • Strand three: male ex-offenders who had previously completed a Good Vibrations project in order to investigate the potential medium- and long-term impact as offenders move out into the community.

• • Strand four: female offenders who had recently completed a Good Vibrations project in order to investigate the short-term impact on female offenders.

The research built on previous research findings (Wilson et al., Reference WILSON, CAULFIELD and ATHERTON2008) and adopted a similar methodology. The field researchers were all experienced in researching within prison environments and working with vulnerable populations. Data were collected via interviews and document analysis. In strands two and four, participants were interviewed daily and an emotional scale was used so as to compare behavioural and emotional change during the project. Participants were then interviewed at the end of the project in depth to investigate behavioural and emotional change after completion of the project. Prison staff were interviewed in strands one, two and four and disciplinary and other relevant records were analysed (Henley et al., Reference HENLEY, CAULFIELD, WILSON and WILKINSON2012).

A structured approach to data analysis was taken using a method derived from Braun and Clarke's (Reference BRAUN and CLARKE2006) six-point process. Detailed line-by-line coding was carried out using themes initially drawn out of previous research literature, and new codes were used as they emerged out of the data. Codes were then compared and cross-referenced in order to identify themes, and a list of key themes was drawn up. These themes were then checked to ensure that enough evidence could be found to support their identification as key themes. The themes that emerged during this analysis included positive change, anger management and stress reduction, coping and emotional and psychological impacts amongst others (Caulfield et al., Reference CAULFIELD, WILSON and WILKINSON2010).

Once these findings had been thoroughly analysed and written up, a further analysis of the data was conducted. As I was not part of the fieldwork team, the secondary analysis was carried out with no personal relationship with the participants. This enabled a certain level of objectivity in analysing the data. The themes suggested that there were multiple outcomes and impacts, and therefore a framework was needed that drew together these outcomes and enabled the elements of the project that contributed to these outcomes to be viewed. The expanded model of CHAT was applied and through the secondary analysis we were able to see how social music-making acted as a catalyst for the positive change found in the primary analysis (Henley et al., Reference HENLEY, CAULFIELD, WILSON and WILKINSON2012).

Project 3 – Good Vibrations Javanese gamelan with young offenders

The findings from Project 2 produced an exciting possibility that the processes of social music-making could in fact be shared processes with behavioural and emotional change in offenders. So, I wanted to investigate these learning processes in more depth to understand the relationship between musical learning and behavioural and emotional change.

The aim was to explore the learning processes in depth and gain a detailed understanding of the workings of a Good Vibrations project. Therefore, a qualitative methodology that comprised participant observation as the main data collection method was used. In addition to participant observation, follow-up semi-structured interviews six weeks after the project had finished were carried out. Both the project facilitator and the prison officer in charge of the project were also interviewed and access to Good Vibrations' own evaluation data, including pre- and post-project questionnaires and a post-project focus group, was given.

Labaree (Reference LABAREE2002, p. 103) suggests there are four interlinked values of participant observation: ‘the value of shared experiences; the value of greater access; the value of cultural interpretation; and the value of deeper understanding and clarity of thought for the researcher.’ However, he warns that these values are not without their challenges. Maintaining objectivity and accuracy, and not allowing familiarity and comfort within a situation to hide the mundane are two such challenges and the participant observer must ensure that their position inside the researched environment does not engulf their responsibility as a researcher to question that environment. To minimise this, a ‘cooling off’ period took place before the observation data were analysed, and observation data was triangulated and cross-referenced with the other data sets so as to fully interrogate the data.

Previous research and the literature suggested that the interactions between tutor and learner are a significant factor within the transformative learning process. Therefore participant observation was carried out in the role of tutor. Undertaking the role of support facilitator allowed the relationships between project participants and facilitator to be understood more fully through experiencing the interactions first-hand, but also through observing the interactions between the participants and the main project facilitator; it allowed me to exploit the multiple positions of insider-outsider within participant observation in order to collect data through different lenses (Labaree, Reference LABAREE2002).

A participant observation schedule was drawn up using CHAT as a framework. Field notes were collected and collated using categories from the model of CHAT and interview questions were designed to collect data relating to the different constituents of the model. The data provided by Good Vibrations – pre- and post-questionnaire and focus group data – were coded using codes derived from the model of CHAT. The themes that emerged through using CHAT in this way were closely related to those of previous research (Caulfield et al., Reference CAULFIELD, WILSON and WILKINSON2010), and two overarching categories were identified relating to individual processes and group processes. On referring back to literature surrounding desistance theory, what had emerged as a result of collecting and collating the data using the expanded model of CHAT were findings closely linked to attributes relating to the development of desistance from crime (McNeill et al., Reference MCNEILL, ANDERSON, COLVIN, OVERY, SPARKS and TETT2011). The individual processes related to Individual Agency and the group process related to Social Interactions. Moreover, I was able to align the codes and categories from CHAT with these attributes and analyse the data using a framework derived from desistance theory.

Questions for the follow-up interviews were prepared based on the initial analyses of the project data sets, including the responses of the participants to the focus group and post-questionnaire. These follow-up interviews provided detailed information of the participants' reflections six weeks after the project and their aspirations as a result of the project. Moreover, on coding the data using the codes and categories aligned with desistance theory, I was able to see the contribution that the musical learning process had on the development of the attributes relating to desistance from crime and suggest that the development of individual agency and social interactions is a shared process with the development individual musical understanding and ensemble skill (Henley, Reference HENLEY2015).

The three-dimensional activity system in action

The findings related to the outcomes of these three projects are inspiring (Hoskyns & Henley, Reference HOSKYNS and HENLEY2010; Henley et al., Reference HENLEY, CAULFIELD, WILSON and WILKINSON2012; Henley, Reference HENLEY2015). However, the concern with this discussion relates to the research process. In all three projects, the three-dimensional activity system was applied but used in slightly different ways (see Figure 3). Project 1 (Music Partnership Project) used CHAT as the primary analytical tool. Project 2 (Good Vibrations Male and Female Offenders) used CHAT for a secondary analysis to understand the processes that led to primary findings. Project 3 (Good Vibrations Young Offenders) used CHAT in the design of the data collection, the collection and collation of the data, and in initial analysis. This then enabled an analysis to take place using a context specific framework.

Fig. 3. The use of CHAT in each project.

In Project 1 (Music Partnership Project), using CHAT as the primary analytical tool enabled the interactions between the Primary generalist teachers, their own musical development and the socio-cultural environment to be viewed. After participating in the MPP a generalist Primary teacher commented:

By seeing how the MPP functioned in terms of the development of individual confidence through interaction between the teachers and the mediating tools, and how this progressively developed as the division of labour changed and the teacher took on more responsibility, we were able to begin to understand this transformation. Similarly, it enabled us to begin to understand why some teachers responded in a negative way, ‘I have learned nothing, and gained no experience or knowledge as a result.’ Data in this case showed that the teacher did not interact with the community, the rules of the activity, and was not able to share the division of labour within the project in the same way as others who reported some kind of change.

Project 2 (Good Vibrations Male and Female Offenders) used the three-dimensional activity system in a similar way, but was used as a secondary analytical tool rather than for the primary analysis. The activity system was applied to understand the findings of the research after they had been analysed and reported for evaluation purposes. These findings included a change in behaviours during the project that related to individual feelings and emotions:

Also, change in behaviours that impacted on how participants were viewed by others:

Through the application of CHAT we were able to see the processes that led to these changes in behaviour. Moreover, using a three-dimensional model allowed us to understand how a participant was able to engage in the musical learning in a multi-layered way.

Creating ownership, finding a voice, working together and, importantly, feeling normal appeared to contribute to the development of positive change identified by the primary analysis (Caulfield et al., Reference CAULFIELD, WILSON and WILKINSON2010). By understanding the different layers of activity within the project, we could begin to understand how and why this was the case. The prime musical activity centred on creative, collaborative music-making through improvisation and composition. This led to the participants developing ownership of the project in terms of their own creativity. However, in different layers of activity, such as group discussion, planning a performance, and formative assessment, participants found that they were able to find a voice, work together, feel normal and address particular personal challenges of working within a group such as this. This too contributed to a sense of ownership in terms of owning the decision-making process and the ownership of the community. In fact, there appeared to be a myriad of different layers of activity that each individual participant engaged with during the musical learning process within the project.

This was also apparent in Project 3 (Good Vibrations Young Offenders). CHAT was used in the design and collection of data as well as the initial analytical tool, but it led to the application of a framework that enabled a way of understanding how the processes within musical learning contributed to context-specific transformation. By applying the three-dimensional model of CHAT it was clear that there were common project outcomes but also that different participants were gaining very different things from the project. As with previous research (Caulfield et al., Reference CAULFIELD, WILSON and WILKINSON2010), participants reported that they were able to sleep as a result of the project.

This was explained by the participants as a direct result of the music.

The activity system enables us to see how this may have occurred through a process of listening and internalising the music, but also how the tools of the prime musical activity, creative music-making, were re-mediated in another activity producing an entirely different outcome. It was clear that there were many different activities relating to listening, communication, working together, ownership, developing relationships and agency, and in each of these activities tools were re-mediated to produce the desired outcome. When the links between these different activities and the components of desistance theory were discovered, I was able to take this further and understand the significance of this in terms of the particular context of the participants.

Moreover, four further individual case studies compiled as a result of this research demonstrate a further four different transformations and their link to desistance from crime can be seen.

1. 1. It was important for one participant to complete the week even though he became very frustrated on a number of occasions and constantly reported that he did not like the music: ‘If I don't stick to it now, it makes me think I can't stick it when I go out and get a job.’

2. 2. The project enabled another participant to engage with a sense of himself: ‘It was an incredible break from such a horror. Being trapped in here with no way of moving on. It lifts you up and gives you perspective.’

3. 3. A third participant overcame his personal feelings for others in the group for the benefit of the musical outcome: ‘There were people on that course that I couldn't stand. They knew I didn't like them. I avoided putting myself into a situation that could get me into trouble. I'd not do that on other courses.’

4. 4. A fourth participant demonstrated how he overcame his nervousness of working with others through music: ‘I wasn't really comfortable working in groups before.’

If the activity of the project had been viewed through one singular activity system, it would have been a more difficult task to tease out these different individual outcomes, and the re-mediation of tools may have been missed. The point is that through a dynamic model of CHAT the dynamic nature of musical learning and its relationship with a particular context can be understood.

Moving the model forward

The commonality of these three research projects is that transformation appeared to have occurred as a result of the interaction between the different constituents of the three-dimensional activity system and that identity is a key component in this. Furthermore, whereas the doctoral research that led to the development of this activity system model found that a change of identity inspired a change of activity, these research projects suggest that a change of activity can inspire a change of identity. Therefore there is potential for the expanded model to show a two-way transaction between identity and activity.

The benefits of using CHAT as an analytical tool are clear. Not only is it useful for understanding how particular activities lead to musical transformation and change, but it is also useful for leading us to new understandings of the relationship between musical learning and personal and social transformations. However, this is not to say that it does not have its challenges. The reason that the initial analysis using the three-dimensional model in Project 3 (Good Vibrations Young Offenders) led to a different theoretical framework is because it showed nothing new about the musical learning process itself. Whilst this might suggest that the musical learning process in very different contexts is the same, the value is seeing how the socio-cultural context and the learner interact. Perhaps the real value then of CHAT is its capacity to be used as a tool to explore the inter-relationship between musical, social and emotional behaviours in different socio-cultural contexts, and how these lead to transformation.

In practical terms, the potential for CHAT to be applied at a praxial level when designing learning is significant. If research has demonstrated the way that learners interact with the specific elements of CHAT, and these interactions lead to transformation, then perhaps there is a value in using CHAT in the planning and organisation of musical learning. Moreover, if more researchers and educators use and develop this three-dimensional activity system, our understanding of CHAT will evolve. Not only will this give both the music education research and teaching community a tried and tested tool for understanding the complexities of developing identities and developing musicianship, but it will also give the wider education community a lens for viewing learning processes in diverse contexts, and a potential framework for pedagogical development.