Children’s friendships in diverse primary schools: teachers and the processes of policy enactment

Abstract Drawing on data from a project exploring children's and adults’ friendships across social class and ethnic difference, this paper focuses on the enactment of national and institutional policy around children’s friendships as realized in three primary schools in diverse urban areas in London. Through a focus on the way in which social and emotional learning (SEL) and teachers’ understandings of children’s friendships seek to govern children’s friendship behaviours, we turn to Foucault’s work to explore how power shapes relations between policy frameworks and teachers’ practices, and between those who teach and those who are taught. We discuss the disciplinary potential of SEL and teachers’ ‘common sense’ understandings of children’s friendships, but conclude by noting possibilities for teachers to create spaces in which all children can safely explore the nature of friendships.

these sets of relationships with reference to policies on social and emotional learning (SEL), and teachers' 'common sense' understandings of children's friendships and how these should be. Whilst offering a critique of how some aspects of SEL play out in teachers' practices, we conclude by noting the importance of taking children's friendships seriously, the possibilities SEL could offer here, and the possibilities for agency for teachers concerned with the support of children's friendships.
First, however, we briefly introduce the three case study schools. The three schools are all within a six-mile radius of each other, and are located in London's inner suburbs, the ring of residential and small-scale commercial settlement that surrounds central London. The three schools -Leewood, Junction and Fernhill -are located in areas with super-diverse populations and experiencing various degrees of gentrification. The schools were chosen as sites of class and ethnic diversity, using reports from school inspectors, Ofsted, local knowledge and government data (e.g. the percentage who are awarded the Pupil Premium, an extra funding allowance based on the take up of Free School Meals (FSM), a proxy indicator of relative poverty). The ethnically diverse and variously gentrifying nature of the localities, the varying percentages of FSM, and the mixed ethnic intake, indicate the super-diverse nature of the pupil populations. 2 All three schools were rated as 'good' by Ofsted in 2013.
In each school, we focused on one Year 4 (8/9-year-olds) class. We spent one or two half days a week over 10-12 weeks in each classroom, spending time with the children, observing them in the classroom and playground, and then conducting interviews with children and teachers. Across the three schools, we spoke to 78 children about their friends, and they drew friendship 'maps' to illustrate their social networks. Forty-six of their parents were also interviewed, using interpreters where necessary, about their understanding and management of, and desires for, their child's friendships. We also spoke with 12 teachers/governors across the three schools, which included all the class teachers, one teaching assistant, and members of senior management and chairs of governors. It is this latter set of interviews which this paper particularly, but not exclusively, analyses. Teachers were asked about the friendship relationships in the case study class, and the school's reactive and proactive policies and practices around friendships. School policy was not the anticipated focus of the project, but as we conducted the fieldwork, this area assumed greater significance, hence our decision to focus on it here.

Policy enactment, school friendships and community cohesion
We are here adopting a policy sociology approach, to understand the 'workings out' of policy within institutions. The work of Foucault has been a particular influence on policy sociologists, (see Ball 2013;Ball, Maguire, and Braun 2012;Gale 2001;Scheurich 1994) and we reflect some of this influence here, whilst not at all claiming to be 'Foucauldians' . Policy shapes the possibilities for the ways in which individuals understand themselves and others, and we are interested in the way in which policy allows teachers to take up particular understandings of the children, their relationships, and their families in these three schools. We understand teachers to be 'both policy subjects and policy actors within the policy process' (Braun, Ball, and Maguire 2011, 581); hence the term 'enactment' , as developed by Ball, Maguire, and Braun (2012), which points to the role of teachers at the chalkface in interpreting and translating central government policy. Part of their interpretation is shaped by their common sense understandings of what makes sense for their school and the children it serves. These common sense understandings are assumed to be 'true' , and are also of course 'intimately bound up with power' (Burr 2004, 68). As such, adopting a 'common sense' view can limit the likelihood of discussion, questioning, reflection, as thought, priorities and action seem obvious, the taken-for-granted approach. A policy sociology approach also emphasizes the importance of context, both in terms of the national context set for schools, and also in terms of the institutional context: for example, the relationships between staff, the reputation of the school, and as noted above, the teachers' sense of the ethos and cultural climate of the school ('who we are and how we do things here'). Mostly we focus here on similarities between the schools, but as our analysis is on-going, we are also developing, elsewhere, our account of the differences between the three sites. The data analysis presents an example of 'policy enactment' , the 'workings out' of policy in schools, namely how national and school policy, and particular teacher reactions and responses on and to SEL, play out in the three schools. Some of the data displays a direct teacher reaction to policy -the dominance of the attainment agenda in schools for example -but much of the analysis does not illustrate direct responses to formal policy statements, rather the more diffuse way in which the 'enabling' policies around SEL allows space for differentiated teacher understandings and practices around friendship.
At least since the beginning of this century, there has been, and continues to be, a policy climate of hyperactivity in schools, resulting in a 'policy soup' on the ground where policies meet, collide, jostle for position (Braun, Ball, and Maguire 2011). Depending on the degree of mandate that accompanies various policy initiatives, different policies are treated differently in schools (Ball 1994;Braun, Ball, and Maguire 2011;Taylor et al. 1997). The national policy set relevant to this paper, and to the subject of friendships, is policies around SEL, the latter being part of the curriculum area, Personal, Social and Health Education (PSHE). Acronyms are confusingly prolific here, as teaching PSHE topics is frequently understood as a means for schools to respond to their duty to promote pupils' spiritual, moral, social and cultural development (SMSC). 3 As we illustrate below, these policies allow the case study schools considerable leeway in their enactment.
Policy acts to identify particular phenomena as problems -'the target of social regulations at any given moment' (Foucault 1983, lecture 6 p. 2, cited in Ball 2013, so the existence of policy in the area of SEL suggests that pupils' social relationships with peers and adults are positioned as potentially problematic and requiring state intervention. The nature of that 'social regulation' requires further examination. Ball (2013) notes that Foucault's concerns with both the disciplining of the individual body and the regulation of the population come together in schooling which 'is one point on the "crossroads where that power over and invested in, individual bodies and populations would converge in technologies of discipline and regularization" (Stoler 1995, 83)' (Ball 2013, 54). We will illustrate this idea with reference to the 'disciplining' of individual children's emotions through particular policy initiatives, whilst also noting the way in which SEL in general seeks to regulate and direct affective bonds with others, thereby attempting to 'govern' 'individuals and collectives by practices of correction, exclusion, normalization, disciplining, therapeutics and optimization' (Lemke 2010, 430). As Wetherall argues, we need to consider how within particular historically specific 'emotional regimes' (Reddy 2001), 'affective practices sediment in social formations' (Wetherall 2012, 103/104). The focus of our data analysis then is the government of the individual body and emotions of the child, as directed (haphazardly) by policy and understood by teachers. These attempts are, we suggest, focused on making the child emotionally and behaviourally responsible in terms of his/her relationships with others. However, we also argue that both policies on SEL and the teachers' understandings of children's friendships offer only a partial conceptualization of and attempt to 'govern' the children's friendships. As we illustrate in other project papers (e.g. Iqbal, Neal, and Vincent 2015), the affective and connective bonds of the children's friendship remain, to a great extent, a untameable space in which they exercise agency. We turn now to consider this claim in more detail.

Learning to be better: PSHE and SEL
As indicated above, the set of national policies directing school practices around pupil relationships focuses around PSHE, a non-statutory, but recommended curriculum area. PSHE includes SEL alongside a wide variety of other topics including drugs, safety and health education. Recently, the PSHE area has been understood as a potential vehicle through which to prevent the dissemination of 'extremist' religious and political views in schools, and to promote ' fundamental British values'; the latter a requirement placed upon schools in November 2014 (Department for Education 2014). PSHE is a baggy curricula 'hold-all' for topics that do not fit within the formal, assessed curriculum. For schools and teachers therefore, PSHE is, at the time of writing, a liminal area in a curriculum governed by the need -at primary stage -to ensure 'good' results in English and maths when children leave primary school. These subjects are mandatory, currently PSHE is not. Different aspects of PSHE have, for schools, become more or less visible over time, the promotion of British values referred to above (and which is itself a clear requirement placed upon schools) being the latest example of prominence. In terms of the PSHE dimension on relationships and SEL, these were given a particular distinction under the recent Labour Governments as part of the Every Child Matters suite of policies, and funding was given to develop and support schools in using SEAL (Social and Emotional Aspects of Learning) materials. These materials were explicitly designed to comprise three 'waves' of intervention -whole school, small group, individual -and practices in each 'wave' should all interact (DfES 2005). The funding for this programme was withdrawn by the recent coalition government, but the materials are still available for use. 4 The withdrawal of funding heightened the degree to which teachers could choose how and to what extent to focus on SEL. All the three schools still used the SEAL materials, in a piecemeal fashion, as their source for work on friendships. Now the focus on PSHE/SEL may return, with an increased emphasis on schools' role in fostering children's 'emotional literacy' (Ecclestone and Lewis 2014;George and Clay 2013). In May 2015, the Department for Education announced 14 projects, including rugby coaching and first aid, that will receive funding through the Department for Education's £3.5 million character grants scheme. 'Character education' is defined as education which helps more children develop a set of character traits, attributes and behaviours that underpin success in education and work, such as perseverance, resilience and grit, confidence and optimism, motivation, drive and ambition, neighbourliness and community spirit, tolerance and respect, honesty, integrity and dignity, conscientiousness, curiosity and focus (see https://www.gov.uk/government/news/character-education-apply-for-2015-grant-funding, last accessed December 2015).
This situation of varying emphases over time is indicative of the way in which the policy context in which teachers are trying to support children to develop positive relationships is 'multi-layered -a palimpest of previous policies and policy legacies' (we are grateful to Sarah Kerr, a former student at UCL IOE, for this observation) .
In a hyperactive policy climate (Braun, Maguire, and Ball 2011), there is always a new set of demands. As noted above, Ball, Maguire and Braun argue that policy enactment involves a creative process of translation and interpretation, but also a constrained process of 'unstable juggling between irreconcilable priorities, impossible workloads, satisficing moves and personal enthusiasms' (Ball, Maguire, and Braun 2012, 71). Here, we indicate three contexts that we argue affect policy enactment on SEL. The first is the 'vocational habitus' (Colley et al. 2003) of primary school teachers, the second the performative demands of primary school teaching, and the third, the different emphases in different institutions which result from the processes of 'juggling' referred to by Ball, Maguire and Braun. Vocational habitus refers to 'a powerful aspect of the vocational culture: the combination of idealized and realized dispositions' denoting 'the right person for the job' (Colley 2006, 25). Prioritizing children's emotional well-being is part of the 'professional context' within which teachers (of young children especially) operate and which directs their response to particular policies (Braun, Ball, and Maguire 2011). This concern and caring was evident in all three schools. All the teachers to whom we spoke appreciated the importance to the children of having positive friendships. Some of the senior leadership spent protracted periods of time resolving disputes as equably as possible, and our interviews with them revealed sensitive understandings of the children's friendship practices. Within this general framework of concern and caring, we illustrate in a later section, the 'work' done by teachers' 'common sense' readings of children's friendships, in terms of defining 'proper' friendships.
Our second point, the pressurized context in which primary school teachers currently work, is fundamental to understanding their practices. The 2013 Department for Education Workload Survey reported an average working week of 59 hours for primary school teachers. Additionally, proactive work on friendships is particularly vulnerable to the constraints on teacher time and effort imposed by a national climate that emphasizes attainment outcomes in core subjects (also Peterson et al. 2014). Despite genuine concern over the children's well being, in practice, with a crowded timetable, disputes -a sign of the unruly nature of friendship -were also seen as annoying distractions for the class teachers. They were very clearly operating under pressure in a system that ranks primary schools by the results of national tests taken by children in their last year of primary school (also George & Clay 2013). Although the case study classes were two years away from these tests, ensuring the children met targets was of utmost importance. The disciplinary framework provided by Ofsted judgements was clear in the research. Two out of the three schools had moved in 2013 from 'satisfactory' to 'good' in Ofsted terms, and were seeing the results in increased enrolments and pupil-based income, with Fernhill generating a waiting list for the first time in many years. Leewood which moved from 'outstanding' to 'good' became the focus of a considerable number of interventions and close monitoring by the local authority. Tanya at Junction School describes the performative pressure on teacher time, energy and practices which means that not disrupting the smooth flow of the classroom becomes paramount.
I think that what happens sometimes in a whole school is that this whole sort of academic targets, and them [test results at 11] being published, and everything being driven by that, sometimes that means for the class teacher that is the main priority and they sometimes lose sight of other things that are absolutely essential for a child.
[…] I think our senior management team do see the great [i.e. bigger] picture, but you know, we are sort of driven by external forces a little bit (Tanya, senior leader, Junction School).
Our third point identifies the role of specific institutional contexts which led to some differences in policy enactment. First, it is important to note that there were considerable areas of commonality amongst the case study schools. They all had a range of both reactive and proactive strategies with regard to supporting and developing friendships. The former included, circle time, 'buddy' schemes, alternative spaces for children not happy in the playground, and interventions tailored for particular children. 5 More proactive strategies included circle time, Anti-bullying Week (a national initiative), 'action boxes' (private messages for the teacher), drama, adult mentors in the playground organizing games and art activities, reiterations of school vision statements about 'how we behave to others on our school' and so on. Out of the three schools, Fernhill was the only one to have a named policy on PSHE. Recently developed (summer 2014), it set out the curriculum content of PSHE in the school, adhering closely to the programmes offered by the PSHE Association. The text was short and rather bland in tone, referencing the importance of respecting difference, and commenting that through the curriculum, pupils 'learn to understand and respect common humanity, diversity and differences so that they go on to form effective, fulfilling relationships that are an essential part of life and learning' . Such texts can play a role in demonstrating that a requirement for policy in this area has been met, without changing practice (Braun, Ball, and Maguire 2011), but Fernhill had a history of prioritizing SEL. The school had a long-established staff group, so the same senior member of staff (Holly) had been involved in the original trials of the SEAL materials a decade ago. Her interest had ensured the continuity of the SEAL focus over time, and the work had become a key part of how the school understood itself. In Junction, and Leewood, several staff changes in senior management, over the last 5-7 years, meant that it was harder to trace a continuous policy focus on this area. A lack of written policy however did not mean that there was no commitment to the area. Junction, like Fernhill, had been eligible in the past for funding for areas of deprivation (New Labour's Excellence in Cities policy) and used this to appoint staff to develop SEL work. After the funding finished, both schools acted to partially retain this staffing from their own budgets. Fernhill's original interest was provoked by teachers' understanding of the children in their traditionally working class, racially diverse intake as having pronounced needs.
[A decade ago] we were dealing with some very high end children that were probably misplaced in mainstream schools […] so we worked with the whole team of schools [trialing SEAL] around behavior improvement (Holly, deputy head Fernhill) Although gentrification and the increasing popularity of the school were understood to be bringing in a cohort of children who were less problematic in terms of behaviour -a point also made at Junction which was witnessing a similarly increasing middle class intake -Holly argues that the school's continued commitment to SEAL is necessary in order to allow disadvantaged children to access the curriculum, and to supplement parental deficit. I think our kids come from such a range of backgrounds [now], you know we have some very eloquent children, some very well supported and privileged children, but we have a huge number of children that don't have those advantages … so they come to school at a disadvantage, their vocabulary is limited. You know perhaps they don't have the social skills. The parents may not have the aspiration for them to do well and achieve […] I don't think we could deliver the curriculum unless we had a commitment to that [SEAL].
We develop below this notion of SEL as particularly important for non-middle-class children.

Social and emotional learning
Policies to promote SEL in schools have received criticism from some researchers who argue, broadly speaking, that it encourages a particular form of 'self-seeking' , what Foucault (1995) calls 'normalization' . Processes of classification, central to schooling, are reaching beyond measurable forms of academic ability and focusing on emotional stability and literacy. Watson et al. (2012) in their consideration of school conflict resolution strategies argue that children are expected to 'continually "get on" with all their peers' (2012, 133), a goal many adults would identify as unrealistic. Non-problematic relationships with others is often understood as a characteristic of a 'good' student, as the ability to negotiate successfully relationships with teachers and peers is seen as an indicator of emotional maturity and possession of an attitude and set of behaviours that are conducive to learning (Archer and Francis 2006), as well as a form of capital accumulation for the individual. Researchers in this area have argued that material commonly used in 'well-being' segments of PSHE teaches techniques which help children work on themselves, in order to better conform to school, by managing themselves more effectively. For example, Clack (2012, 499) identifies resilience, stoicism, optimism, altruism, emotional regulation and mindfulness as the psychological constructs deemed necessary for emotional well-being, very similar to those now proposed as 'character education' . We note here that friendship -a complex mixture of emotional and emotive connections -is often reduced in SEL practices to a series of attributes and learnt techniques. An example of this work on the disciplining of the self was given by a teaching assistant at Fernhill School (the focus of his post was responsibility for working with children who were perceived as having barriers to learning).
Some of the children I work with, they say, 'I've got no friends' and it really affects them. So I had to work with a particular child, explain to that child how we are going to make friends, 'What do you think we need to do, to make friends? What do you need to do?' And we went back to, he has to be friends with himself first, so him as a person and once we developed that, then he was able then to make friends. So how you behaved, what you showed other people, your facial expressions, your body language. When you are playing games are you fair? When you are doing activities are you able to let someone else have a choice? Can you also listen to people's points? So we worked a lot on that. Your behaviour in class, do you listen, because if you don't listen people see that you are getting in trouble, they might not want to be drawn to that. So we worked a lot on that and now the child […] is there playing with all kinds of different children (Darren) Darren presents the child with a series of technical steps, suggesting an incremental path to the development of a successful friendship. Work that embeds a previously isolated child within the peer networks of his class is clearly valuable. However, the focus on the individual and his/her (usually his) disciplining of their selves has been criticized for ignoring the wider context in which children live, and the structural inequalities which shape those contexts (Clack 2012, 502, Gillies 2011. Individual responsibility, the production of a particular kind of subject, as emotionally self-responsible, is key in this model of well-being, a classic neoliberal framing (Holloway and Pimlott-Wilson 2012). Again using a Foucauldian analysis, Lemke makes a similar argument, citing Cruikshank (1996) with regard to the 'self esteem approach' in the United States.
The "self esteem" approach considers a wide variety of social problems to have their source in a lack of self esteem on the part of the persons concerned […] It promises to solve social problems by heralding a revolution … against the (wrong) way of governing ourselves. In this way, the angle of possible political and social intervention changes. It is not social-structural factors which decide whether [social problems] can be solved, but instead individual-subjective categories. (Lemke 2001, 202) Clack agrees, noting: 'well being against this ideological and conceptual framework becomes something the individual controls through the correct management of their inner life' (2012, 502). The most obvious (but not the only) examples of this work across the three schools were directed at developing the interpersonal skills of three boys, one in each class, who exhibited the most challenging behaviour, and who all had learning and social support from teaching assistants (TAs). 6 Given research that notes the over-representation of black boys in categories suggesting behavioural and/or emotional disturbance (Rollock et al. 2015), it is important to point out the three boys, made subject to this 'field of visibility' (Foucault 1995, 202), were all black/mixed race. At least two of the boys had experienced very challenging circumstances outside school that were largely beyond the reach of school staff. Their actions are designed to try to address the anger and distress with which the boys presented in school. However, the point remains that the school response worked to reduce friendship to a series of techniques, helping the children to more effectively regulate their emotions (Clack 2012). Similarly, Gillies argues in her study of a unit for teenagers identified as having social and emotional issues, that 'empathy becomes a strategic technique designed to better manage social encounters ' (2011, 196). The school steps in where the family is seen to have failed in developing resilient, emotionally balanced children (Hoffman 2010). Hoffman cites Martineau 'who points out that in the context of efforts to teach resilience in inner city schools, resilience functions as an ideological code for social conformity' (Hoffman 2010, 391). Developing this theme, Ecclestone and Lewis (2014) argue that social and emotional well-being is, a 'new form of governance that shapes desirable citizens' (203). Children who are already able to appear as 'emotionally competent, literate and confident … gain new advantages' (2013) in the moral economy of schools. Hoffman agrees, 'now even emotional competencies can be subject to measurement and those who don't measure up can be found lacking' (2010,(391)(392). Helen, a pupil at Leewood (white British working class), provides an example, as she was positioned by her class teacher as someone who did not 'know' how to fit in. Describing an afternoon out in the woods, the teacher notes the children came together for a collective and spontaneous game, They [the class] were building this place [shelter] for the animals in the woods. They were all together, it was wonderful … And they were, 'You get the bigger bit [of wood]' 'Let's put it there' [mimics excited discussion] Except Helen. Helen put herself right out of it. 'They hit me' , you know it was some sort of reason. If you ever ask them to find a partner, she's … right out of it. She doesn't know how to (fades away). (Jessica, class teacher, Leewood, emphasis added) Note the use of the active tense here -Helen put herself right out of it -which situates the exclusion as self-exclusion. Some children are understood to be more emotionally competent and 'better at' interpersonal relationships than others, having friendships with 'deeper roots' as another teacher (Phillip) comments below. Disciplinary power orders individuals in space, what Foucault referred to as 'the art of distributions' , setting up new divisions between 'normal' and 'abnormal' . The identity of those defined as not-normal is important. Indeed, when we look at who in the case study schools was positioned as challenging in terms of their friendships, it was the three boys and, to a lesser extent, Helen mentioned above and another 10 children spread across the three schools. Twelve out of these fourteen children were black or mixed heritage children (a third of the total number of black and mixed heritage respondents), and ten were boys. As our data-set derives from three classes only, we are not able to make any claims about the role of race in this process, but given that qualitative research (e.g. Gillborn & Youdell 2000, Gazeley et al. 2013Gillborn et al. 2015) suggests that black and mixed race boys are more likely to be positioned as emotionally or behaviourally problematic, this is a concerning observation, albeit based on a very small sample size.
We are arguing that SEL interventions focus attention on children's interactions. Yet, this is, in many ways, an ordering, a bringing into the spotlight (Foucault's 'field of visibility' again) and a 'setting-down' in policy texts of the common sense understandings that teachers and other adults carry of children's friendships. In the following section, we consider how teachers read children's interactions. Foucault argued that 'knowledges are produced within power relations … in the sense that some groups or institutions have been able to speak knowledgeably about "others", subaltern groups who are concomitantly rendered silent' (Ball 2013, 15). Adults -parents, teachers and others -are able to speak with claims to authority about children. However, the data collected with the children lead us to argue that teachers do not define nor proscribe children's relationships for the children themselves (see Iqbal, Neal, and Vincent 2015). Although the understandings of teachers and other adults do impact upon the children, and contribute to their sense of their own abilities and competencies in relation to their friendships, the sphere of their affective relationships remains to varying degrees (depending on who the child is) a space of independence. Power relations can be found in all social interactions, but social interactions are not reducible to 'a binary structure with "dominators" on one side and "dominated" on the other, but rather a multiform production of relations of domination' (Foucault 1980, 142). We turn now to teachers' understandings of the children's friendships.

Teachers' readings of children's friendships
Within the teachers' general framework -their vocational habitus -of concern and caring, we wish to emphasize the work done by adults' 'common sense' understandings of friendship, and we identify here the way that such understandings deliver the idea that children's friendships were often light, flexible relationships of convenience. Some examples from the class teachers:

Mary Healy in her research into friendship comments,
The issue of friendship and how we bond with others ought to be an important concept for education, yet schools rarely take the forming, nurturing and nourishing of friendship beyond helping to deal with disputes between friends when they disrupt school life. The general attitude tends to be a 'bus theory' of friendship: do not worry if you miss one, another will be along in a minute. (Healy 2011, 442) However, our data illustrate that children experience their friendships keenly, and value loyalty highly. Their own fears were around the potential fragility of friendships, and there were frequent references to friends who had left the schools. In the extract below, two children in Crimson class at Leewood discuss their concern over losing friends.

Andy (white British, middle class):
Every time a new person arrives in class I get worried that they are going to be a best friend to somebody who is my best friend, one of my friends.
Helen: And they are going to be taking your best friend? That is what I always worry. 7 Teachers' rather more cavalier reactions may be understood as a defence against the children's negative emotions, their displays of anger, pain (also Watson et al. 2012). More prosaically, they could also be responding to the tedium and time pressure of sorting out recurrent issues with 'the same children' (Phillip, second class teacher, Leewood). On a more practical level, understanding the children's relationships as flexible allowed teachers, by directing bodies -into new seating plans for example -to use the children's affective relationships as a management tool to produce calm behaviour in the classroom. George (2007) finds the same minimizing response from teachers in her study of girls' friendships: [Teachers] appeared to trivialize the emotional impact [changes in group relationships] had on individual group members. It would appear that these teachers drawing on the cultural ideas of 'real' friendship (Allan 1989) discounted any ruptures to the girls' relationships, brushing aside their upset, assuming that the next day will see a restoration in the friendship. (2007,95) However, clearly teachers are not all the same and even within our relatively small sample; there were clear differences in how much weight was given to the children's friendships. Jacqui, for example, spoke of how in her seating arrangements, she tried to both respect children's existing relationships, and also direct pairings in ways that 'formed new bonds' . Tanya (with a pastoral care post) and Gary (a class-teacher) at Junction provide an example of the differences in approach between teachers. Tanya described the policy of placing of Action Boxes in each class which allowed the children to post a message to their teacher, alerting them to the child's concerns. She acknowledges that their use 'depends on the class teacher really' and we found that Gary's was rarely in use. Also at Junction, in an attempt to reverse the marginalization of PSHE in a crowded timetable, senior leaders had decided to remove an assembly, so that the time freed up could be used for circle time. However, Gary has other priorities for this time.
They had a huge amount of that last year, partly because the teacher they had… was in charge of PSHE and that sort of thing. I use [the time] in a sense to kind of get them just to reflect on what they have been doing over the past week maybe, in terms of how they approached things. But I use it more particularly for their learning behavior because it was so poor when they started in Y4, so poor […] their attitudes to what has been asked of them, what is expected of them, how they are going to approach those challenges. (Gary) Our second point concerning teachers' understandings is that buried in some of their accounts seemed to be a sense that there was a proper way to 'do' friendship, what Allan (2011) calls a 'friendship blueprint' , which generated relationships more whole, complete and profound than others. One teacher at Leewood (Phillip) notes that it is a close-knit group of five white middle-class girls in Crimson class, with their out of school networks, shared activities, sleepovers, etc. whom he sees as having friendships with 'deeper roots' . However, this understanding ignores the extent to which such outof-school activities (paid-for classes, summer camps, shared holidays abroad) depend on and derive from classed resources. As a result, a range of opportunities the girls have to be together out of school are simply not available to other children.
In contrast, children with complex learning needs had their ability to develop friendships routinely questioned by their class teachers. Peter, Lucas (Leewood School) and Jordan (Junction School) all had complex learning needs. They were understood by their class teachers as not being able to have 'proper' friendships with children without special needs. Peter and Lucas were close friends. The class teacher and the boys' parents had different views on this. The teacher felt the friendship should be celebrated, a positive view which seemed to be undercut with a sense that the boys could not hope, because of their disabilities, to maintain friendships with any of the other children.
What is terribly interesting about Lucas and Peter was, I had a bit of a fight with the parents last year because they didn't want Lucas and Peter to be friends with one another […] And you are kind of thinking, well how fantastic that they have -because they love one another […] but we have been kind of pushed to say -and I understand because I think they felt that the others weren't being friends with them but I don't think they ever will -I don't think they can. They can be really lovely to them and really friendly. But I don't think they can be friends.

INT:
Because of the learning difficulties the two children have? R: Yes, and yet they found one another. They have found a friendship which is really profound. I think that is so worth celebrating but [their parents] didn't want it. I did wonder why, on that level. (Jessica, Class teacher, Crimson class, Leewood School) Their parents, however, feeling that the other children were encouraged to be kind to their boys, rather than to seek them out as equals, challenged this view that the able-bodied/disabled divide was sufficiently fixed so as to make friendships impossible across this dimension of difference. In an example of Foucault's 'bio-politics' -contestations over 'the vital characteristics of human existence' (Rabinow and Rose 2006, 196) -Lucas' and Peter's mothers, wanting more mixing, insisted that their children were not always put together and that they spent more time in the classroom (not outside it with their TAs).
Jordan at Junction School was on the autistic spectrum. Jordan had recently transferred from another school, and has not as yet made any friendships in his new class according to his mother. At the time, we conducted the interviews with the children (although this later changed), he received the most segregated educational experience of all the children across the three schools, as he spent nearly all his time sitting in a corner of the classroom with his TA. His mother described his friendships with children out of school as including a close friendship with another child on the spectrum, and also with neuro-typical children. However, his teacher commented that the most that could be expected was for the other children to 'be kind' .
The children kind of make sure he is accommodated and they are very, very accepting and supportive of him […] If he wants to play games, they will play with him and they will involve him, and he kind of just does what he does […] They are being kind to him. I think it would be difficult to form friendships with Jordan because the communication is very difficult … I don't know how much there is a kind of conversation going on that you would characterize as a friendship, a strong friendship. (Gary, teacher, Burgundy class, Junction School) We argue here that the 'truth discourse' concerning friendship and children with complex needs as promulgated by the two class teachers at Leewood and Junction Schools laid a limited expectation on the other children of 'being kind' . This was understood by the teachers as a realistic common sense response to the realities of disability. It was then left to the children's parents to challenge these 'barriers to being' (Thomas 1999;cited in Connors and Stalker 2007) as best they could.

Concluding thoughts
We have been concerned to contribute here to the literature on policy enactment in schools, by focusing on teacher responses to a policy set which enables, but does not enforce. In their study of policy enactment, Ball, Maguire, and Braun (2012) have argued that performance (attainment) dominates as a focus of education policy; it is around performance where the action and activity, the sound and fury of both policy and practice lie. But, as they also note, performance is not the 'all' of schools, and we have focused here on a set of policies that are not directly related to performance, and are not experienced as enforcing or prescriptive by our case study schools. Ball et al. argue that these 'writerly' policies present teachers with opportunities for creativity, sense-making, and personal commitment (2012,94). Indeed, we can see this in our research in the sustained commitment of two senior leaders (Tanya at Junction School and Holly at Fernhill) over a period of years to emphasize the importance of children's friendships to their well-being. We also, however, note the easy assumptions made about children's capacities for friendship. Additionally, we want to draw attention to the inconsistencies of national policy, the flow and ebb of central government agendas which create different emphases and priorities. Altogether, our aim has been to illustrate the complex ways in which, through processes of policy enactment, power circulates between government, teachers and children.
To conclude our more detailed arguments in relation to policy on SEL: staff in the three schools understood children's emotional well-being as fundamental to enabling them to learn and flourish at school, and were often highly reflexive concerning the school's role in enabling well-being. However, they were working within contexts shaped by the demands of a crowded timetable and a performative policy agenda. As a result, Ball argues (2003) that practitioners are likely to incline towards a set of values and behaviour which allow them to better manage the demands placed upon them. They are required to 'organize themselves as a response to targets, indicators and evaluations. To set aside personal beliefs and commitments and live an existence of calculation' (Ball 2003, 215). However, we suggest this process is not a complete one and we make two points here. The first is that, as noted above, some of the teachers work hard to try to create and maintain a space in which the children can discuss emotional and social issues. Second, power does not simply operate from the top-down, instead power relations are 'a dense web that passes through apparatuses and institutions without being exactly localized in them' (Foucault 1998, 96). The formation of the children's friendships in school is affected by the web of power relations between one child, his/her peers, his/her teachers and parents. Particular children, identified as emotionally vulnerable, become the focus for teacher attention in order that the children work on themselves. However, we argue that the teachers' understandings do not define the children's subjectivities, and even the most 'vulnerable' retain some freedom and agency with regard to their friendships. Much of the intricacies of the peer group remain hidden from adults (including researchers). It is difficult for adults to 'see' as pre-teen children see, hence the frequent adult emphasis on the casual and fluid nature of many of the children's alliances which acts to rationalize and minimize what to the children seem like emotive and major happenings. Our interviews with the children lead us to argue that what the teachers' reactive and targeted strategies miss is the quieter exclusion and marginalization of children, girls especially (also George 2007), who conduct their relationships under the radar. Only some children make visible problems and tensions. Tayna (Junction School) talks of 'allegiances that you would not call friendships' , where there is an unbalanced play of power for instance by one child over another. Then there were children whose loneliness was not recognized. These children were generally not seen as problematic in terms of friendship because their difficulties and distress do not play out in confrontations.
We have rehearsed some themes within the critiques of SEL, but we are not suggesting that this work should disappear from schools. There is limited national policy to support teachers, PSHE has largely been marginalized. However, the role of the individual teacher is important here (Holloway and Pimlott-Wilson 2012). If teachers find the space in which to emphasize SEL as part of PSHE, they have a fair amount of autonomy over how they do so. A recent study of SEAL initiatives (Banerjee, Weare, and Farr 2014) recommends whole-school and universal (rather than targeted) approaches, and also highlights the variability in school practices. As we have noted, SEL has been criticized for its disciplinary focus on the individual. However, we finish by suggesting that SEL could develop beyond such a critique, if teachers were able to focus on the subject of friendship as an important topic for all children, and not solely as a site for the production of well-behaved, self-governing bodies. This would require that issues and topics for discussion would not be guided primarily by teachers' concern with behaviour management, but be identified by the children, in order to reach the concerns that they hold. As Watson et al. (2012) argue, fluid, open debate and discussion with children may have the potential to diminish alienation, and offer safe spaces for children -all children -to explore the nature of friendship, and friend-making and maintaining.

Notes
1. Vertovec argues that the multiplicity of origins of those who arrive in London and other major cities, their varying immigration status, gender, age, religion, patterns of spatial distribution and so on contribute to super diverse urban populations. How people negotiate with others in these dynamic and fluid circumstances is a growing area of research in geography and sociology. 2. At Leewood, the relevant FSM percentage is 19.7%, at Junction 32.8% and at Fernhill 38.7%. The three classrooms contained pupils from White British backgrounds (Britain, Ireland), White Other (Western and eastern Europe, Turkey), Black (Black Caribbean and Black African origin), South and East Asian origin, Arabic and mixed heritage backgrounds. Thirty-four out of the seventy-eight children who participated in the research were Black British (26) or mixed heritage (8); 19 children were White British, 16 White other (including 5 Turkish/ Kurdish). Six were of S Asian origin, 1 of E Asian origin, and 2 of Arab origin. 3. See for example Peterson et al. (2014), http://www.doingsmsc.org.uk/. 4. http://teachfind.com/national-strategies/seal-primary-school-assembly-0?current_search=primary%20SEAL (last accessed December 2015). 5. 'Restorative justice' is a process that resolves conflict by encouraging children to take responsibility for their actions and respect the views of others. See for example http://www.restorativejustice4schools.co.uk which emphasizes the use of the approach for behaviour management, another example of the affective nature of friendship being reduced to emotional and behaviour self-responsibility (see Watson et al. 2012 for a critique). 'Circle time' provides a space in which children can gather together to discuss personal/emotional issues, and can be used as a reactive or proactive strategy with regard to friendship. 6. The role of adult TAs in shaping friendships is worth noting as some children were accompanied by their TA into the playground. 7. Helen's understandings of her own friendships are different to those her class teacher (Jessica cited above) holds.
This illustrates the point that teachers do not frame the children's sense of their friendships.

Funding
This work was supported by the ESRC [grant number ES/K002384/1].

Notes on contributors
Carol Vincent is a professor of Sociology of Education at University College London, Institute of Education. Her research interests include parenting and family life, relationship between parents and educational institutions, and how these relationships are mediated by social class and ethnicity, and education policy.
Sarah Neal is a reader in Sociology at the University of Surrey. Her research interests lie in the following areas and their inter-relationships: ethnicity, race, multiculture; community, belonging and place, rurality and the countryside; social policy-making, inequalities and inclusion. A recently completed related research project is Living Multiculture: the new geographies of ethnicity and the changing formations of multiculture in England (ES/J007676/1).
Humera Iqbal is a researcher at University College London, Institute of Education. Her research interests include the diversity of Muslim communities in the UK, and parenting and family life in multicultural settings, in particular how ethnicity and race influence parenting practices.