The practice architectures of middle leading in early childhood education

Background Across the globe, public sector leadership is generally organised in a hierarchical and linear way with decisions distributed to an appointed leader at each level in the organisation; this structure is particularly evident in the Swedish school system. Principals at each school are responsible to a director of education, and teachers are directly responsible to their principal. In this climate, a new kind of leadership is evolving with the specific purpose of facilitating collegial professional learning and the development of quality student learning in schools and preschools. Leaders in these roles generally have a recognised and often appointed role in supporting the professional development of their colleagues, but seldom this role is taken to be a more formal paid position in the linear or hierarchical formation of organisation. In Sweden, these leading roles are recognised as utvecklingsledare, handledare, förstelärare (development teacher, facilitator, first teacher) or curriculum coordinators. Moreover, these leaders are teachers whose work is mostly that of a classroom teacher. Their leading role has often been generated Abstract

from past personal engagement in professional development (Bennett et al. 2007;Fluckiger et al. 2015).
Elsewhere, we have described these teachers as middle leaders and have identified them in joint studies conducted in Sweden and Australia (Grootenboer et al. 2015b). This paper will be another contribution to the field where we more deeply focus on the practice architectures of middle leading work and how they enable and constrain activities at a particular site. This term middle leader reflects how these teachers sit in the middle positionally (i.e. between the principal and the staff ), philosophically (i.e. as a leader among peers) and practically (i.e. in the practices of leading). Other theorists such as Hargreaves and Ainscome (2015) use the term middle leader; however, their descriptions generally concern the leading work undertaken at a regional level in large-scale reforms related to organising and building networks. Counter to their renderings, our study revealed that although the middle leaders may also take up such work, it is in specific, localised sites whereby their leading work is predominantly being practised among colleagues. The more general term teacher leaders (Muijs and Harris 2003) has also been used to describe teachers' leading roles; a term connected to concepts like distributed leadership, communities of practice and professional learning communities. These are discussed in the next section.
In this article, we want to emphasise the role of these middle leaders and the way they were responsible for enacting the systematic quality work 1 among their peers related to the implementation of the revised curriculum beginning in 2011. The purpose of the article is to examine the intricacies of the leading practices of 15 middle leaders in one city district in Sweden, using a practice theory framework. Specifically, we aim to understand how these middle leaders were supported in their organisations, and how their leading practices facilitated the improvement of the teachers' professional work in their units in line with curriculum. To explore and understand these issues we used the theory of practice architectures (Kemmis et al. 2014), as it draws attention to the particular arrangements (made visible in language, work and power) and the cultural-discursive, material-economic and social-political conditions that enable and constrain the practices of these middle leaders in their systematic quality work. The article begins with a short review of leadership research literature followed by a brief description of the theory of practice architectures. The description of the site followed by methods of data collection will then be presented. Finally, the results will be analysed and discussed by using the theory of practice architectures as a lens to examine what enable and constrain middle leading practices.

Leading in education
The educational leadership literature is vast; but it mostly focuses on the work of principals and positional heads. That said, there is a substantial body of literature concerned with teacher leadership (e.g. Muijs and Harris 2007). However, much of this work is dominated by a direct focus on its effect on student learning. We instead shift the focus to the practices of leading (of teachers who lead peers) and the impact on adult learning. In other words, our concern is fundamentally on site-based staff professional development which sees teachers as learners, and how this development work is facilitated by particular nuanced practices of leading from the middle (Bennett et al. 2007;Grootenboer et al. 2015b).
There are a range of models and theories of educational leadership; relevant in this article is the kind of leading connected to what has been described as distributed leadership (Spillane 2006). We suggest this connection simply because middle leaders often are the ones to whom leadership is distributed. In a review of distributed leadership, Liljenberg (2015) found that most studies reported positive results of distributed leadership. However, some studies identified challenges when introducing distributed leadership at an organisational level (e.g. Leithwood et al. 2007). Harris et al. (2014) summarised these challenges as being related to time, culture, professional reluctance and the fear of "getting it wrong". Another way of generating leadership among teachers is to establish teacher teams considered to be influential in the evolution of a genuine working culture based on collaboration and capacity building. Most often this culture is steered by school positional heads (e.g. the principal) with the teachers being collectively responsible for the implementation of the curriculum (Hargreaves 1994). Hargreaves referred to this as a form of "contrived collegiality" (pp. 135). An example of contrived collegiality is when a facilitator comes from the district office and teachers are involved in a centrally mandated programme to develop new ways of teaching in a school; it is contrived in the sense that teachers are expected to work collegiality 'into' their own school in more topdown mode. However, this is not always the case; our ongoing research on middle leading shows otherwise (Edwards-Groves and Rönnerman 2013; Grootenboer et al. 2015b;Rönnerman et al. 2015).
Rather than being related to "contrived collegiality", teacher leading can be generated through action research collaborations built from the ground up. It is based on, and develops through, genuine shared meaning-making and capacity building (Kemmis et al. 2014;Salo and Rönnerman 2013;Rönneman 2012) which facilitates a multidimensional culture of relational trust (Edwards-Groves et al. 2016). Relational trust between participants and facilitators is commonly argued as being a central condition for generating professional learning communities and making possible the kind of leading required for professional change and action (Bryk and Schneider 2003;Cranston 2011) in preschool settings (Thornton and Cherrington 2014). Building on the work of Bryk and Schneider, Cranston connects the success of learning communities with the notion of trust, which is articulated as 'being relational in its orientation and developed around group norms of safety, risk-taking, and change orientation ' (2011, pp. 59). Grootenboer et al. (2015b) go further to explain that relational trust is apparent in five realms: interpersonal, intersubjective, interactional, intellectual and pragmatic. These realms form critical conditions for site-based change of the nature described in this article. Although the middle leaders in this study were asked to lead their colleagues it was their primary position in the school as a teacher (e.g. 4 h was allocated for planning and leading the dialogues) that prefigured the more sustainable building of collective knowledge in teams. Their leading was not in the traditional mould of a sole up-front crusader.
Significant for the study reported in this article is Liljenberg's (2015) conclusion that only a few studies in the Swedish context have focused on leadership in teacher teams. Instead many studies centre on co-production and concepts such as "professional learning communities" (Stoll et al. 2006) have been arbitrarily applied to groups of teachers working together. However, Liljenberg did find that in these Swedish studies, collective learning that takes place through shared values and processes of negotiation of meaning were central. In this article, we wish to contribute to these understandings by focusing on the leading practices that were central to these processes and practices in preschools, where teachers as a matter of routine work in teams.
Taking these things into account, we specifically focus on middle leading practices. In our understanding, there is a significant difference between concepts like distributed leadership (which we earlier have shown as being generated, i.e. as a process of external and internal generative practices) and middle leading (Edwards-Groves and Rönnerman 2013). Although both descriptions of leading are localised and site-based, broad accounts of distributed leadership (by for example Spillane 2006) are characterised by the tendency for the 'leading' to be initiated by, and connected to, the principal, whereas middle leading is specifically teacher-driven, site-responsive team-based professional learning and development in the school (Rönnerman et al. 2015).
Middle leaders are important to the sustainability of practices in schools because, unlike principals, they have the capacity to make a significant difference to classroom practice. Lingard et al. (2003) noted that the "principal's effects on student outcomes were small and indirect" (pp. 51), but, "teachers have the greatest impact upon student learning of all 'educational variables' . The effect of principals' practices on student learning are, in contrast, heavily mediated and limited" (pp. 148). Middle leaders are positioned in and near the classroom and so they can impact pedagogy and student learning more substantively since they are in the classroom themselves (Bennett et al. 2007;Fluckiger et al. 2015). The classroom is the critical site in education, and it is where the decisions of principals and leaders have to be enacted (Grootenboer and Marshman 2016). Proximity is a key factor. We argue therefore that middle leaders are central in the development of substantive and sustainable quality education because they exercise leading practices close to and in classrooms.
We noted earlier that "middle leaders are those who have an acknowledged position of leadership in their school, but also have a significant teaching role (e.g. development leaders, senior teacher, Head of Mathematics Department)" (Grootenboer et al. 2015b, pp. 278), and that their leading is practised 'below' the positional head (e.g. principal) and simultaneously 'above' and 'among' the teaching staff (Grootenboer et al. 2015a). The notion of acknowledgement is crucial for teachers engaged in development work, as for some middle leaders, their work is not always a paid position in the hierarchical structure of the school; but it is work acknowledged by their teacher peers. Therefore, the concept of middle leading has significance in three ways: 1. Positionally-middle leading is structurally and relationally practised 'between' the school senior management and the teaching staff. They are not in a peculiar space of their own, but rather than are practising members of both groups.

2.
Philosophically-middle leading is practised from the centre or alongside colleagues. In this sense, middle leaders are not the 'heroic crusader' leading from the front, but rather alongside and in collaboration with their colleagues. 3. In practice-middle leading is understood and developed as a practice. To this end, the focus is on the sayings, doings and relatings of leading rather than the characteristics and qualities of middle leadership. (Grootenboer, Edwards-Groves and Rönnerman 2014, pp. 18).
In the educational leadership literature much is written about schools, but there is less about preschools (early childhood education) 2 (e.g. Heikka et al. 2013, pp. 36). Hujala (2004) found from focus group interviews that leadership was connected to the position of a leader. However, Haikki (2014a, b) studied leadership in Early Childhood Education (ECE) in her thesis. She found that distributed leadership is an emerging field in ECE but still the trust between the macro and micro level faced difficulties in sharing responsibilities according to the interviews. This is interesting as will be shown in this article that trust is something highly valued between macro and micro levels. In one article Heikka et al. (2013) did a literature review on distributed leadership in ECE and have mapped out what characterise distributed leadership in different studies in the Anglo Saxon context (pp. 32-33). The findings are rather related to the concept per se and theoretical underpinnings than to the leading taking place as is the focus in this article. We find this literature important and this study may contribute to the filed by conceptualising a specific practice-based framework of leading where the teachers themselves spend most of their time in the classroom, and also have a significant role leading the professional learning among peers. As we found that at the time, the work of the middle leader was crucial for sedimenting the kind of quality work in preschools required by newly revised curriculum.

Theoretical framework
To provide a theoretical framework for the study and the subsequent analysis reported in this article, we take a practice perspective of leading. In particular, this means that we have not understood or interrogated leading as a set of personal leadership qualities and/ or characteristics, even though this propensity has seemed to prevail in educational leadership literature. Rather, we foreground leading practices or, as described by Wilkinson and Kemmis (2015), leading as practice. This approach centres on middle leading practices as opposed to middle leadership or middle leaders 3 per se. Aligning with Schatzki (2002), this idea highlights the situatedness and sociality of middle leading practices as these are enacted and experienced in sites. Of course, there are many theoretical understandings of practice (Nicolini 2013), but here we are drawing on the work of Kemmis and Grootenboer (2008) and Kemmis et al. (2014) and their "theory of practice architectures". The point of differentiation of this theory concerns its ontological perspective that stresses the primacy of actual practices (and practice architectures) as they are experienced in sites.

Theory of practice architectures
The term 'practice' is ubiquitous in education. It is used widely, and often uncritically, to generally describe any particular activity (like reading or teaching or even schooling). However, recently there has been a turn towards practice theories to better understand and critique the unfolding of education in particular sites and settings. According to Kemmis and Grootenboer (2008), practices are constituted in more than just the dimension of activity, but in three interdependent, interconnected and simultaneously occurring realms. These three realms always occur simultaneously as sayings, doings and relatings that are influenced by practice architectures (or the specific conditions that shape the doing of an activity). The theory of practice architectures attends to the ways practices are formed socially involving characteristic: 1. Forms of understanding (or language and discourses made present in the saying of things); 2. Modes of action (or activities made present in the doing of things) and 3. Ways in which people relate to one another and the world (or relationships made present in the relatings between people).
These three dimensions 'hang together' in a distinctive project (Kemmis et al. 2014). Furthermore, practices are distributed among practitioners; orchestrated, so that individual's involvements in the practice cohere around a common purpose (or shared project); diverse or differentiated, so that different individuals contribute in distinctive ways to the overall project of the practice and to be accomplished they are ecologically dependent on other practices.
However, practices do not develop or unfold in a vacuum-they are enabled and constrained by practice architectures that prefigure, but do not predetermine, practices in a particular site (Kemmis and Grootenboer 2008). Practice architectures are the preconditions that prefigure practices; these make practices possible and hold them in place. These are specifically the: 1. Cultural-discursive arrangements, in the medium of language, that make the sayings of the practice possible; together with, 2. Material-economic arrangements, in the medium of physical-space time, that make the doings of the practice possible; together with, 3. Social-political arrangements, in the medium of power, solidarity and agency, that make the relatings of the practice possible (Kemmis et al. 2014).
The theory of practice architectures is useful for examining the site-based local conditions that influence what happens when people (like the teachers and middle leaders and principals in our study) come together to work with one another, and how particular sayings, doings and relatings influence these happenings.

The context and organisation of middle leaders in a Swedish city district
Since 1996 preschools have been part of the Swedish educational system. Preschools are considered as an initial educational stage of schooling and its practices are governed by a state legislated curriculum known as Lpfö98 (in Swedish). The state dictates that the education practices in the Swedish school system (from preschools to high school) are controlled by the aims and guidelines in the three curricula (Lpfö98, Lgr11, Lpy11) with the expressed aim "that the three curricula should link into each other and take a common view of knowledge, development and learning" (Lpfö98,pp. 3). This means that preschool principals have been forced to become managers, a difference from earlier days when he/she was part of a team (Nihlfors et al. 2014). There are distinct and differing aims between preschool and school. For preschool, there are no knowledge requirements to achieve, but rather the aims are related to students' development and learning through everyday knowledge and activities. The curriculum for early childhood education sets out the fundamental values for the preschool, and the tasks, goals and guidelines for the activities that are to be carried out.
The Swedish School Inspectorate, established in 2008, emphasised leadership work with teachers in the development of systematic quality work (The Swedish Schools Inspectorate 2012) this feature is also recognised in the curriculum. It states that the principal has responsibility for: … carrying out systematic work on quality together with preschool teachers, child minders and other staff as well as providing the child's guardian with opportunities to participate in work on quality. (Lpfö98 2010, pp. 16).
The dissemination of the revised 2010 curriculum required new district initiatives. As a response, the city concerned in this study organised a seminar series to address different aspects of the revised curriculum. The initiative involved university researchers conducting seminars connected to the new curriculum goals. The first seminars were about action research, documentation and the revised curriculum itself. Others were related to specific goals focused on young students' reading, writing, mathematics, gender and sustainable development. Essentially, this process relied on middle leading practices.
The city district involved in this study nominated ten teachers (from different preschool sites) to participate in the seminars. These teachers, plus another five volunteers, became middle leaders tasked with facilitating the development of all personnel in the city district pre schools. This process involved setting up a new organisation consisting of six distinctive groups, each with its own purposes and tasks, as depicted in Fig. 1 (noting the arrows representing the both ways communication flow between groups).
Middle leaders facilitated Learning Dialogues among all preschool staff and acted as "bridge and broker" (Wenger 1998) between the district decision-making body, the principal and school-based implementation. Middle leaders were each responsible for leading the dialogues around quality development in their particular preschool site. In addition to their situated preschool role, middle leaders facilitated monthly meetings in the Middle Leader Group, took part in the Learning Dialogues Group (in Swedish: Lärande samtal) and organised the report back to the district education office via the Development Group, where they participated twice a semester. Two middle leaders were also selected to be overall district coordinators. As shown in Table 1, almost 360 teachers were involved across 13 preschool sites (larger sites had more than one middle leader).
The organisational structure points to the kind of brokerage the two coordinating middle leaders were responsible for (Wenger 1998). They also acted as messengers between the different groups by being responsible for the meetings documentation, including assembling the notes from the Learning Dialogues which were sent to the Development Group for discussions and new decisions.

The study
Given the apparent gap in the literature related to middle leadership in the early childhood sector, this study focused on developing empirically based insights into the practices of a specified group of middle leaders in a particular context. As will be clear from the preceding theoretical discussion of practice, middle leading is understood as site-based, and so here the data collection is localised, and the findings will provide a touchstone for others involved in leading in preschools to reflect on their policies and practices. The findings were generated through analysis that was both emerging (i.e. themes that were identified through an open reading of the data) and related to the key foci (i.e. from the theoretical frame and the questions), and this allowed for new insights, and also added specific understanding related to the identified 'gap' .
The study was conducted in a district of a major city in Sweden (the processes and structures described above), and the participants included 15 middle leaders (including the two coordinators). Data were collected at different occasions during 2014 and  2015 with a focus on the leading practices of these 15 middle leaders. The overarching research question of the study was: How are the leading practices of middle leaders enabled and constrained as they lead professional learning in their early childhood sites? In 2015, a survey was conducted with five broad questions and several follow-up subquestions. The broad questions were: 1. What factors influenced you to take on a middle leading role? 2. How have you organised meetings with your colleagues? 3. How do you relate to colleagues, to other middle leaders and to the principal and vice versa? 4. How should an ideal situation look like for you as a middle leader? 5. Do you have other thoughts about what enables and constrains you as a middle leader?
The qualitative surveys were sent by email as an attachment to the coordinators who distributed them to all middle leaders. Time was provided for middle leaders to individually respond to the survey during one of their monthly meetings. Completed surveys, written in Swedish, were sent directly back to the Swedish researcher for transcription by the Swedish researcher, following Kvale's (1997) process for condensing answers. First, all answers were condensed to a sentence summarising the key points; second, all condensed sentences were further condensed to create a short summary of responses. This process was completed for each question. Third, a ten-page description was constructed with one of the coordinators for member checking. This description was later expanded with details exemplifying the middle leaders' views in a full report (Rönnerman and Wilhelmsson 2016). Survey responses and citations used to illustrate the teachers' voices were translated from Swedish into English for the purposes of conducting further analysis by the international research team and for publication; we note that sometimes a direct translation was not possible, so for comprehensibility for an international audience some words have been changed. Aligned with ethical considerations, informed consent was given by all participants, and for anonymity, participants are referred to with T. followed by a letter (A-N).
Data from observations and recorded conversations of Learning Dialogues from an evaluation meeting among middle leaders were also collected during 2014 (for more detail see Rönnerman et al. 2015). Additionally, interviews with one principal and the district development leader were conducted. Other complementary data include relevant documents such as the revised curriculum (Lpfö98 2010), the specific document of systematic quality work prepared in the city district by the district development leader in consultation with the middle leaders (called the SQW-booklet), minutes from the meetings where the coordinators met middle leaders and reports from the middle leaders meetings. Each of these data was examined in relation to the theoretical dimensions delineated by the theory of practice architectures; that is, to understand the enabling and constraining conditions (or practice architectures) of the work of middle leaders. According to the theory of practice architectures, the intent for analysis is to examine the cultural-discursive arrangements (or discourses and language), the material-economic arrangements (or activities, resources and materials) and the social-political arrangements (or power relationships, agency and solidarity).
Analysis involved modifying Braun and Clarke's process for thematic analysis (2006, pp. 16-23) whereby data were examined by: • Familiarisation with and immersion in the data.
• Generation of transcripts.
• Formulating initial codes in relation to the three dimensions of practice architectures as a priori. • Searching for and reviewing themes that connect to the cultural-discursive, material-economic and social-political arrangements. • Defining, naming and elaborating and cross-checking themes across the range of data.
This process allowed us to understand the conditions that enabled and constrained the middle leading practices through the middle leaders' expressions, their described activities and how they related to one another and to artefacts in their preschool site. While we appreciate that there are a number of localised conditions that prefigured leading practices, here we focus on the arrangements that formed the practice architectures across the preschool sites.

Results
In this section, we present the data vis-à-vis the conditions and arrangements that enabled and constrained the leading practices of the participants as they led professional learning in their early childhood sites. To do this, we utilise the three distinctive realms of theory of practice architectures for analysing the cultural-discursive arrangements, the material-economic arrangements and the social-political arrangements. It is important to note that the theory has served an analytical function, since in real life practice, these three sets of arrangements are interrelated and enable and constrain the conduct of the other simultaneously.

Middle leading practices were enabled and constrained by cultural-discursive arrangements related to language and discourse
In the realms of the cultural-discursive, the new language and the shift in expected and required discourses that appeared in the revised curriculum for preschools were most significant.

Developing a shared language
It was evident in notes taken by middle leaders that certain concepts, language and discourses were introduced and used in their activities. Specifically, the language used in Learning Dialogues was heavily influenced by the language of the curriculum and was reflected in the meeting agendas. The language was connected to the government policy about preschools where, for example, it stated that "all education should be based on scientific ground and proven experiences" (SFS 2010). The meanings of these directives were discussed and problematized in the Learning Dialogues and in the final evaluations.
For example, in the last section of the summary in the column "to be developed" one middle leader comment indicated:

We lack clarity in expectations concerning the connections between science and proven experiences. We need to critically scrutinize ourselves and what we read (T. A)
This quote illustrates that although time is allocated for discussion about all concepts there is still uncertainty in how to use the concepts and the meaning of them; they can use the language but their comprehensibility of its meaning was restricted. This formed a constraining dimension of the work they were doing. In contrast, the middle leaders mentioned "that newly qualified teachers had a better understanding of these concepts" and among this group there was a belief that working with the local SQW-booklet "will strengthen their scientific grounding". Middle leaders recognised that for newly qualified teachers, the new policy was an enabling feature because of the recency of their training they were more attuned to the practice demands of the new policy.

Engaging with professional readings
In the protocols developed by the Middle Leaders Group, coordinators suggested that readings be used as preparation for the Learning Dialogues. An example of the types of readings included books written about how to conduct systematic quality work by other preschool teachers and/or by researchers. Usually chapters or specific pages relating to the particular purpose of the upcoming meeting were selected. Other input sources suggested included to log into the National Agency for Education webpage and take part of the uploaded film about the curriculum, by listening to lectures from researchers uploaded on YouTube and other YouTube clips (e.g. Break womb at https://www.youtube.com/watch?v=4xio4mOXu3Y). 'Readings' were connected to the specific area that upcoming Learning Dialogues addressed. Often the reading or film sequence was connected to a number of reflective questions that would frame the ensuing discussion in the Learning Dialogues. This literature shaped the participant discourses, forming an enabling feature of the development work therefore creating particular cultural-discursive arrangements that influenced the kind of language and discourses used in the Learning Dialogues. Teacher language was transformed by the curriculum reform; this 'travelled' from the local SQW-booklet scaffold by readings into the discussions in the Learning Dialogues and the teacher evaluations when documenting their everyday work.

Middle leading practices were enabled and constrained by material-economic arrangements related to time and space
The middle leaders had time from their teaching responsibilities allocated for their facilitation. A substitute teacher replaced the middle leader. Each meeting took place on Mondays between 2 and 4 pm.

Time
It was compulsory for all teachers to join the Learning Dialogues. Most middle leaders found the allocated time (4 h/week) to be adequate for the required tasks, although some requested more time for their own professional development (reading for professional knowledge development and to better develop their leading practices).

Resources are given, but I would like more time for my own education and to strengthen my role as a middle leader (T. F)
Indeed, the issue of a lack of time was prominent, and many found that they had to use the weekends to prepare.
Most weeks the Learning Dialogues were well attended, but sometimes some teachers were unable to participate due to a lack of substitutes, illness or an unwillingness to leave the children because they had been too many substitutes in the unit that day.

There has not been enough time given to the teachers to read what is said at each meeting. It is not always substitutes present or the teacher does not want to leave the children with a substitute teacher. (T. J)
These circumstances constrained the implementation of the planned agenda prepared by the middle leaders and made some of them frustrated and disappointed. This is not an uncommon issue in most educational initiatives, as the time for meeting and discussing their work is a matter that concerns teachers. Among the quotes, the same pattern is evident by these teachers. Even if they had allocated time to meet they were still asking for more time to be able to prepare in a better way and by that become more professional. The most concerned issue related to time seems to be when teachers did not turn up at a meeting, reasons all related to material-economic arrangements, conditions of lacking substitutes or illnesses among other teachers in the team which made it impossible to leave the class.

Physical space
Both the Middle Leader Group and teacher Learning Dialogues took place off-site (e.g. at City Hall or nearby school, respectively), where participants would not be interrupted. They were seated in a circle with the middle leader and the principal as part of the dialogue group; other materials included desks, white boards, books of readings, equipment to view YouTube clips, work notes brought by teachers or middle leaders. These arrangements (off-site and the circular design) enabled each participant to feel more able to contribute to the conversation more fully. While these details may seem somewhat mundane, it is important to note that these physical arrangements (i.e. the time and room) enabled the middle leaders to undertake their educational development practices.

Group sizes and set-ups
Teachers participating in the Learning Dialogues were drawn from across each preschool site. Depending on how many units in the preschool, the group size varied. Note larger preschools required two middle leaders. One teacher from each unit attended the meetings in turn, enabling all to attend over a four-week period. From observations it was evident that the principal would take notes on a laptop, whilst the middle leader initiated discussions often provoked by questions; teachers were enabled to share experiences, student work samples, developing knowledge, issues and questions. The middle leader was alert to group dynamics, as well as individual and collective learnings, writing keywords/concepts on the whiteboard. During the Learning Dialogues, teachers were organised into smaller groups to allow greater opportunities for everyone to speak.
To facilitate these discussions, middle leaders emphasised the importance of having all voices heard and managed the process to enable all to speak in turn around the table.
Our collegial learning activities started with a shared reading, video or a dilemma which was carried through to discussions in smaller groups. As a leader of the conversations I distributed the key word and inserted new angles and comments. All participants were asked to raise inquiries to one another to go deeper in the discussions and not remain in conversations about sharing tips and tricks. Sometimes we divided into smaller groups. Each session always ended with a summary of the meeting and a report. (T. C) At monthly meetings, middle leaders met to share and critique experiences and reflections from facilitating Learning Dialogues. Evaluation meetings were held at the end of the semester. During these meetings, observations showed how participants organised themselves into smaller groups to summarise and collectively condense meeting notes. Each middle leader took on self-initiated responsibilities to complete small group tasks. Collated summaries were reported back to the whole group forming a more comprehensive overall summary that was collected by the coordinating middle leaders and sent it to the development group.

The self-allocation and distribution of activities and tasks along with different grouping configurations in the physical space of the meetings provided enabling material-economic arrangements that facilitated robust and equitable discussions between teachers, principals and middle leaders. Additionally, middle leaders had space and time for leading the Learning Dialogues in their preschool sites and attending the Middle Leader's Group. To be able to meet from other preschools mean we get to share experiences and knowledge. Colleagues express how they have got new insights. I have a possibility to design my learning dialogues in a variety of ways, which is wanted from my colleagues. (T. B)
As acknowledged by participants in surveys, as is often the case in reform programmes, lack of real time for practising and addressing the in-class reform activities emerged a constraint on the quality and depth of the discussions. On the other hand, data show how they appreciated the time they got to be able to share and deepen knowledge.

Middle leading practices were enabled and constrained by social-political arrangements related to power and solidarity
The organisation in the city district into six specific groups involved different practices that shaped, and were shaped by, one another to achieve better learning opportunities for all children in the district. It appeared that the leading practices that emerged were ecologically arranged with other leadership practices in the preschools with the middle leaders acting as the brokers between principals and teachers, preschools and city districts. In that way the relational practices were prefigured by the reciprocity between practices among and across groups (see Table 2).

Collaborative work with staff, principals, other middle leaders and the district director
Coordinating middle leaders were participants in all groups except the Leading Group (see Fig. 1). Power was visible in their experiences of practice, as all participants had roles at specific 'levels' in the organisation. For example, power was specifically present when the principals made it compulsory for teachers to attend the Learning Dialogues-a decision made in the Leading Group that in turn required compliance by all principals and personnel. Against such directives, however, some principals exercised their autonomy to do otherwise; using the allocated time for other activities. Power and solidarity emerged through the practices and experiences in the different groups as the Development Group made decisions and shaped the other practices regarding the focus of their meetings. In other instances there was also a flow-back from the Middle Leading Group, where collated summaries were sent to the Development Group for their consideration in future planning. This flow-back showed the reciprocity and agency evident in the different relationships. In one way, the middle leader was an operative in the power structure of the district's policy agenda, and consequently the ways they related to others were constrained by this.

Confidence of the principal
Most middle leaders took on the role upon the suggestion and encouragement of their principal. For example, two middle leaders commented

My principal motivated why she thought I would be good as a middle leader. (T. G)
My principal gave me the opportunity to work as a middle leader. She thought I had the qualities for this role. (T. I) Table 2 How the practice architectures are prefiguring practices around systematic quality development work

Example of sayings in the semantic room Examples of cultural-discursive arrangements
The language used referred to the curriculum; personnel and middle leaders mentioned specific goals from the curriculum. Reports from preschool units followed the SQW-template. Critical aspects were mentioned concerning understanding of concepts The revised national curriculum and Education Act has specific goals that are highlighted. The local policies and SQW-booklet prescribed a specific format for reporting the work from each preschool unit. Specific readings were used about quality. Lectures were given about quality. A course for middle leaders was given. The middle leaders planned the Learning Dialogues

Example of activities in the physical space Example of economic-material arrangements
Sharing experiences in Learning Dialogues and in Learning Group. Reporting on documentations in the site by using iPad and the SQW-template. Participants were divided into small groups in discussions. Summarising Learning Dialogues and reporting back to the Development Group. Tensions and disruption in groups when teachers did not attend Time for leading and participating were given to all personnel. Middle leaders were also given time to plan and report back. Middle leaders organised space for the meeting. Computer and iPads were given to each preschool unit and used for documentation. The SQW-booklet stated how the quality work in the site should be documented

Example of relations in the social space Example of social-political arrangements
In discussion, relationships are made through the teachers' own documentation. Relating to one another in sharing experiences on certain issues. Teachers related to middle leaders and vice versa. Trust between the middle leaders and the teachers, and the principal Certain aspects of quality were focused each month and mentioned in the plan by the coordinating middle leaders. New readings were given locally as lectures or as books. The coordinating middle leaders suggested to the development group. Compulsory to participate in Learning Dialogues. Principals involved but not leading Learning Dialogues These sentiments were echoed by many of the other middle leaders, and so principal support and encouragement appeared to be an important aspect of the social-political arrangements that enabled the middle leaders to enact leading with confidence. We connect this feature to trust between different partners in the site.

Trust between middle leaders, principals and teaching colleagues
Closely related to the previous point, the data revealed that trust was an enabling and evolving feature in their work practices. Trust in the work of middle leaders was demonstrated by the principal, by teaching colleagues and in each other as middle leaders. In recent research, the importance of different dimensions of relational trust is highlighted (Edwards-Groves et al. 2016), and for the participants in our study, trust in and by middle leaders was also perceived critical. Trust was central in their ability to lead Learning Dialogues in ways that sufficiently institute the reform measures required by the district. In particular, it was found that the relational positioning of their practicesin the middle-became particularly poignant and critical if change processes were to be implemented and sustainable.
Of course, the nature and characteristics of the trust involved was complex and multifaceted, that is, participants (teachers, principals and middle leaders) had to have a strong sense of interpersonal trust (that they could relate personally with the others in the group), intersubjective trust (that they could construct shared meanings with others), interactional trust (that they could interact with agency and confidence), intellectual trust (that the processes and content were delivered with a high degree of scholarship and academic integrity) and pragmatic trust (that they could achieve the goals set). Relational trust encountered in these "five dimensions of trust for in educational leading practice" (Edwards-Groves et al. 2016) emerged as a significant dimension of the practice architectures that enabled the facilitating practices or the middle leaders (or conversely constrained them when trust in the range of realms was not present). Middle leaders described the role of being in the middle "as a role in its own right"; to them, it became a leading role within which they felt secure, and one that entailed strong and durable collaborations with their principals and all the personnel who trusted them and entrusted their professional development too. One of them even expressed it as being a model for future development work (T. C). Having trusting relationships with colleagues is clearly considered to be a double role, as expressed by teacher E.
When it is Monday afternoon and we meet in Learning Dialogues I take on my leading role. I get many questions from my colleagues and they are positive and like the Learning Dialogues because we discuss quality matters. When I am among the children I work as one of the teachers at the preschool. (T. E) It was evident that middle leaders attempted to practice democratic ways of working; accounted for in these excerpts when it is suggested individuals "do not have more knowledge than others", rather their "task is to lead the dialogues". In this way they were relationally positioning themselves alongside their teaching colleagues. The role as a middle leader was considered challenging, but at the same time inspiring and motivating as they were challenged by their peers in the dialogues:

I have a role that challenges and questions in a stimulating way (T. B)
However, across the dataset there were also examples of where particular socialpolitical arrangements constrained their middle leading practices. Some middle leaders expressed how their colleagues saw them more or less as a teacher, which was unexpected.

Most of my colleagues know I don't have the answers on the topics we discuss. But still, there are some who tell me they have not done the readings. Every start I have told them you are not reading for my sake but for the collegial learning in this group (T. J)
Also, another issue mentioned by a few middle leaders concerned continuing their facilitation with the full team: I wish I could follow up our discussions in the Learning Dialogues together with the whole team to be able to discuss the development in the practice. Changes could be done all the way and not end with discussions. (T. C) This seemed to be an issue as they worked towards coherence and consistency across all the teacher teams in their site. Finally, a critical dimension of the social-political arrangements that enabled and constrained the middle leaders' practices was the involvement and relationship with their principal.

Principals as participants, not leaders
A way for principals to become more involved in teacher learning through the Learning Dialogues was to participate in them, which significantly all did. However, principal participation was generally not as a fully active discussant but rather as listener and note-taker. This means they were present but not necessarily participating. From the principal´s point of view, this was a good way of getting to know my teaching personnel well and to appreciate the topics that were discussed in the Learning Dialogues (Principal). In a principal interview it was mentioned that this provided her with a good way of getting to know problems that could be picked up in the annual individual conversations she had with all staff.
According to the middle leaders, this kind of peripheral involvement by the principal is not uncommon (see e.g. Heikka 2014). However, one noted that some teachers found it uncomfortable when the principal was present; it was felt that when the principal was absent for any length of time, discussions had a different tone and depth. From this, the involvement of the principal served as both an enablement and a constraint.
Their peripheral presence and participation in the Learning Dialogues created influential social-political arrangements that both enabled and constrained the social space as power shifted in different ways among the participants. Data showed that the arrangements, in general, enabled solidarity for the task of being a middle leader. But there were signs of power and restrained agency when principals were not genuine interactants in the discussions or that they allocated the time for other tasks where middle leaders practices were constrained because they did not get enough time for their work.

Practices prefigure and are prefigured by other practices
The initial analyses outlined above show how one practice can prefigure another practice through particular practice architectures. Sometimes positively, sometimes negatively, the practice architectures made middle leading and site-based learning practices coherent and held in place making it more or less sustainable, or in other circumstances the practice architectures created disruptions, uncomfortable or unsustainable practices. Furthermore, analyses illustrated how the distinctive cultural-discursive, material-economic and social-political arrangements enabled and constrained the practices of middle leading, and made particular sayings, doings and relatings possible. To understand the leading practices required for the development of systematic quality work in the city district, we need to further understand how the preschools are enmeshed with the arrangements that make them possible and hold them in place (Kemmis et al. 2014). In Table 2 we provide an example how this occurred, based on the analyses presented earlier. Table 2, and earlier analyses, shows how the cultural-discursive arrangements were distinctive and visible in shaping the middle leaders' practices. The national curriculum for early childhood education (Lpfö98 2010) reached all the way into the preschool units, filtered by local policies and a locally developed and a written book (SQW-booklet), with advice and guidelines about how to conduct the systematic quality work in each local site. This was specifically done by material-economic arrangements that enabled all teachers to take part in dialogues led by middle leaders. In the unfolding of practices amongst all those involved, the coordinating middle leaders played a central role as brokers between the practices in the organisation related to quality work. This was important as the social-political arrangements enabled the middle leaders to realise the projects of their practices in social-political space in a medium of solidarity and agency.
Examining the middle leaders' practices-that is their sayings, doings and relatings revealed in language, work and power-we can see how these were shaped by the different practice architectures in their particular sites (the Development Group, the Middle Leading Group, the Learning Dialogues). Their role as brokers helped the organisation to realise the goals of quality work as they prefigured and configured the practices in the organisations of systematic quality work. For example, practices enacted in the Learning Dialogues were prefigured by the middle leader's coordination and plan for certain themes to follow in the discussions. Through their discussion summaries from the Learning Dialogues to the Development Group and by that, in a sense, prefigured the discussions and decisions to be made for future directions for the city district.
The last, but most important, link in the organisation that focused on children's learning and development is the teacher team and the practical teaching work that happens in each unit. Teacher comments' in one of the Learning Dialogues (in 2014) revealed a tension for the practices of the learning dialogues and previous practice, e.g. I wish we could go back to action research. This teacher had earlier taken part when the quality work was located much more at the site and based on action research. According to her, the organisation as a whole has neglected recognition and contact with this most important development practice using where the everyday work with the students happen. On the one hand, practice architectures around the broader 'top-down' practices in the organisation in the city district constrained the development of quality work at the site. On a contradictory note, most arrangements were considered by the teachers as enabling ways to participate in discussions and share their experiences, although it seemed unable to change the teaching activities in the unit as put by one middle leader. One explanation could be that these teachers are not in the same Learning Dialogues at the same time, rather three different sessions led by the same middle leader. Another might be that the teacher team itself needed more time for development and change. Middle leaders recognised this weakness in the organisation with many suggested continuing with the teacher teams after the formal Learning Dialogues to promote more possibilities for sustainable practice development in the different units.

Concluding comments
In this article, we examined the leading practices of middle leaders in a city district in Sweden. We found the organisation of the systematic quality work through middle leading going across the more familiar hierarchical organisation of the public sector, and identified the middle leaders and the coordinating middle leaders as the hub in the work of systematic quality work in early childhood education. The coordinating middle leaders were present and played a crucial role as brokers (Wenger 1998) in all the different groups in the organisation. These coordinators played a central role in the sustainable development of the implementation of the new curriculum through systematic quality work. This case illustrates the important role of middle leaders in a contemporary organisation steered in a predominantly hierarchical way. Although the middle leaders were initially asked by the principals to become middle leaders, their practices developed from being among the group of middle leaders. But at the same time, each recognised that they were confident in the role and were supported by their colleagues, their principal and other middle leaders, opposite to what Heikka (2014) found in her study where a lack of trust was found between macro and micro levels in directors having difficulties in sharing responsibility for leadership in the organisation. This issue can further be related to the discourse of professional learning communities (Stoll et al. 2006) and collegial learning as a way to strengthen knowledge among the teachers which in turn benefit students' learning (Timperley 2012). But also the development of relational trust among the teachers is something to consider for further quality in the work with students.
To understand their practices, the theory of practice architectures was used for illuminating what enabled and constrained the leading practices of middle leaders-i.e. the arrangements that figured and prefigured the practices. These analyses highlighted how different cultural-discursive, material-economic and social-political arrangements influenced what particular practices were possible in the quest to improve the quality at the local site. A main conclusion is that middle leading practices were influenced by practice architectures that distinctly and distinctively shaped the language and discourses, the activities and physical set-ups and the social relationships required for creating (or not) development and sustainable change in Swedish preschools. In other words, the teachers developed a common language to be able to have deep understandings of their work and in a sense to create meaning-making. In this sense, the relations to one another and to the principal were crucial to be able to create these meetings over time and in the whole city district which in turn would have effects on the students learning and the quality in early childhood education.