Discursive institutionalism: towards a framework for analysing the relation between policy and curriculum

Abstract Discourse approaches in education policy analysis have gained prominence in the last decade. However, though the literature on policy discourses is growing, different conceptions of the ‘discursive’ dimension and its potential for empirical analysis related to the field of curriculum policy have not yet been fully researched. To address this gap in education policy research, this article explores the framework of discursive institutionalism. Using background and foreground ideas and coordinative and communicative discourses on three analytically distinct levels, this article proposes and discusses a framework for empirically analysing, explaining and understanding education reforms on the transnational and local levels. The introduced conceptual framework represents an integration of discursive institutionalism (DI) and curriculum theory (CT) to provide a more multifaceted set of concepts to explore the lending and borrowing of transnational education policies and their application at both national and local levels. These concepts have been applied as analytical tools in a research study on the most recent curriculum reform in Sweden, and they may serve as an example of how different ideas, discourses and levels can be distinguished in research studies to maintain the complexity of education reforms.


Introduction
Education, which has previously been understood as a primarily national affair, is increasingly influenced by transnational policies. These international policy flows take different forms in different countries due to distinct historical, social and cultural traditions (Sellar and Lingard 2013). Beck (2006, 77) criticises the dominance within the social sciences of what he terms 'methodological nationalism' , calling for a replacement of the national perspective with a 'methodological cosmopolitanism' that 'contains the national project and at the same time extends it' . Sassen (2006) supports this critique by arguing that the structures and procedures constituting globalisation are taking place in what has historically been constructed as a national context. Global policies and regulations require an institutional base for policy-making and for the regulation of transnational agreements, and this base has been primarily situated in national institutional spaces either contained within or controlled by the state. In light of these dynamics, how can the interconnections among these global, national and local policy levels and the actual education systems be understood and analysed?
In this article, drawing on a larger four-year research project, we introduce an integrated framework developed from both curriculum theory (CT) and Vivien Schmidt's (2008Schmidt's ( , 2010Schmidt's ( , 2012aSchmidt's ( , 2015 'discursive institutionalism' (DI) to provide a more multifaceted set of concepts to explore the lending and borrowing of transnational education policies and their application at the national and local levels. The concepts have been applied as analytical tools in a research study of the most recent curriculum reform in Sweden, and they may serve as an example of how different ideas, discourses and levels can be distinguished in research studies to maintain the complexity inherent in the field of education policy and reform. Comparative education has consistently argued that transnational education policy indisputably affects, but does not determine, the framing of national policies (Rizvi and Lingard 2010;Steiner-Khamsi 2012;Savage and O'Connor 2015). We argue that a theory of discursive institutionalism (Schmidt 2008(Schmidt , 2010(Schmidt , 2012a(Schmidt , 2015 might contribute to a deeper understanding of what happens within the 'black boxes' of curriculum codes (Lundgren 1989) and the conceptions of education (Englund 2005) built on curriculum theory by introducing a more articulated notion of institutional change and drawing attention to the discursive nature of transnational policy transfer (Steiner-Khamsi 2012). The term 'curriculum' , as it is used in Swedish, is close to the English designation of the same term. Specifically, the concept refers to the basic goals and direction of education. Within curriculum theory, the term 'curriculum' denotes the overarching philosophy and conceptions underlying an actual curriculum document (Lundgren 1989).
We build our understanding of curriculum theory on a Nordic and Continental European conception of the same, which is linked to the idea of Didaktik (Gundem and Hopmann 1998;Westbury, Hopman, and Riquarts 2000). The Nordic tradition of curriculum research is also historically related to the new sociology of education (Young 1971) and to a focus on the subjects that comprise a curriculum. 1 An important aspect of the Nordic approach to curriculum theory is Ulf P. Lundgren's conceptualisation of 'curriculum codes' , which identifies the underlying curriculum principles inherent in school subjects from a reproduction theory perspective (Bernstein and Lundgren 1983). The basic levels of analysis in Nordic curriculum research (Lundgren 1989) are also represented in Anglo-Saxon curriculum research texts by, for example, Doyle (1992) and Deng and Luke (2008), as follows: the society (institutional) level, the actual curriculum (programmatic) level and the classroom level. Using these analytic levels, it is possible to critically explore the intended, enacted and achieved curriculum (Anderson-Levitt 2008) by using social theories to explain outcomes. While a Nordic version of CT can be characterised as a system theory perspective on the interdependence of relationships in education systems, there is a lack of institutional theories on the discourses and discourse recontextualisations that underpin a communicative understanding of policy transfer and enactment in the transnational, national and local arenas.
The purpose of this article is to contribute to an on-going conceptual discussion of how to understand and explain the formation and configuration of policy discourses on different institutional arenas. The key question that foregrounds the conceptual inquiry in this article is 'What concepts can form an analytical framework that considers the different arenas, discourses and social actors through which education policies are framed and performed?' To answer this research question, we draw on 'classical' CT and its different levels of analysis: classroom, curriculum and societal (Lundgren 1989;Englund 2005;Deng and Luke 2008). We also draw on DI (Schmidt 2008(Schmidt , 2010(Schmidt , 2012a(Schmidt , 2015 and its concepts of ideas and coordinative and communicative discourses. The three levels of analysis in CT correspond to the three general levels of ideas in DI: the different arenas for policy solutions, the programmatic ideas that underpin policy ideas and the level of 'public philosophies' , or the values and principles that form the basic assumptions of societal norms. Exploring theoretical concepts at the intersection of policy and curriculum contributes to policy research by offering empirical means to define and contextualise those specific policy programmes that have comprehensive effects; this exploration will also contribute to CT by going beyond its traditionally national theoretical framework to offer effective analytic tools for analysing transnational policy discourses.

A reciprocal relationship between transnational and national levels
A salient feature of both transnational and national education policy is what Grek (2009) calls 'governing by numbers' . Measurement and monitoring are preferred techniques within the normative ideal of neoliberalism (Ball 2015); they have dominated the Western world since the end of the 1980s (Schmidt 2009). As a consequence of this new global neoliberal policy paradigm, comparative research has emerged as a way to grasp the complex processes of policy framing at the international, national, regional and local levels. According to Simola et al. (2013), comparative research in education risks both simplifying and overestimating policy comparisons and transfers. Takayama (2009) offers another line of critique, arguing that we must move beyond an unreflective discussion of national differences to instead recognise how global discourses of neo-liberalism have been made possible through the re-articulation and recontextualisation of local historical contestations and politics. In the Nordic countries, Carlgren and Klette (2008) call for more nuanced and sensitive descriptions, distinguishing the differences embedded in the similarities in responses to shared transnational policy ideas and ideologies.
Drawing on Steiner-Khamsi's (2012) arguments for a 'culturalist model' -in which the transfer of education models is not taken for granted, but, rather, is considered to be rooted in the interactions of experts from different parts of the world and the possibility that the actual impacts of different education systems remain an empirical question -we extend the focus of the transfer of education models to include education discourse. Thus, we are interested in the specific features of education policy discourses and their shifts when transferred across arenas (including not only exogenous events and shifts, but also endogenous [internal] agency for explaining change). Second, we agree with Steiner-Khamsi (2012) that education transfer should be understood as a 'circular movement' instead of a linear idea, the latter of which implies that it is possible to determine a certain system of origin and a receiver system. The interactions of politicians and experts in the transnational arena contribute policies, knowledge and experiences from different national arenas to the formation of transnational policy trends; thus, a national education system can 'pick' pieces from different national and transnational policy ideologies, policy programmes and solutions. Our starting point for the exploration of a framework for analysing the transfer of education policy discourses is in line with Steiner-Khamsi's (2012) overview of recent research, which focuses on education policy as a discursive practice that links transnational policy movements to national education reforms in which the tangible imprints of discourses in practice cannot be taken for granted.

Ideas, arguments and discursive interactions in institutional contexts
New institutional research has theorised that it is possible to find a form of policy convergence between different school systems, a so-called 'world model of schooling' , because world-society models shape nation-state identities, structures and behaviours through global cultural and associational processes. One example of such uniformity is the choice by many countries to structure their mass schooling systems in the form a six-year primary level and three-year junior and senior secondary levels, following the UNESCO statistical system (Meyer et al. 1997). This perspective on rationalised modernity has been criticised because it tends to ignore power relations, choosing instead to highlight governments' tendencies towards both competition and coordination (Anderson-Levitt 2008). The so-called Stanford School world society theory, presented by John W. Meyer and his colleagues, suggests a sociological neo-institutionalist approach that emphasises the role of ideational factors in international politics (Buhari-Gulmez 2010). In the sense that a primacy of ideas implicates a constructivist approach, there are similarities between the Stanford School's theory and the 'fourth' new institutionalism, discursive institutionalism, 2 as promoted by Schmidt (2012a). Although sociological and discursive institutionalism both represent new institutionalist approaches to cultural sociology, they differ in that the Stanford School mainly studies the diffusion by scripts of a rationalist world culture that constitutes actors as dependent performers of global models. By contrast, discursive institutionalist researchers have highlighted the processes through which global ideas are integrated into local contexts and emphasised the key role of conscious actors in shaping motivational discourses in their studies of local-global interactions (Alasuutari 2015). Like Sassen (2006), researchers within the field of discursive institutionalism have directed their interest toward actual local practices through which global ideas are framed and translated into local contexts.

Institutions as a social reality
New institutionalism emerged in the mid-1980s as a response to a research field that, as we have argued, overemphasised the potential for agency within institutions that eschewed institutional constraints. New institutionalism research, by contrast, risks overemphasising institutional constraints by subordinating institutional actors. According to Schmidt (2008), DI tries to balance the risk of over-or underemphasising the role of institutions by stating that institutions are both given and contingent. They are given in the sense that agents act within a given institutional context, but contingent in relation to these agents' actions (i.e. what they think, write and speak): 'These institutions are therefore internal to the actors, serving both as structures that constrain actors and as constructs created and changed by those actors' (Schmidt 2008, 314).
Institutions are not 'real' in any material way, but they are socially real and have causal effects in a social reality. Schmidt (2012a) rejects critical realism on the basis that it is rooted in a scientific meaning of reality that begins with particles and scales up to people. Instead, Schmidt (2012a, 96) roots her point of departure in a philosophy of society and conversely proceeds 'from people to particles' . Thus, she places herself on the constructivist side, noting that most constructivists do not deny a material world. Following this line of reasoning, the core question is not whether reality exists, but, rather, what is real even if it is not material (Schmidt 2012a). Institutions exist not in the field of what Searle (1995) calls 'brute facts' (e.g. a mountain exists whether affected by human actions or not), but in the field of 'institutional facts' , which are formed by culture and society.
In a critique of constructivist/discursive institutionalism, Bell (2011) claims that discursive institutionalism relegates institutions to a vague and almost meaningless role, while overlooking the question of what constitutes an institution. In her response to Bell (2011), Schmidt (2012b claims that institutions can be explained as simultaneously constraining and enabling structures of meaning that are both external and internal to 'sentient agents' . Actors' background ideational abilities enable them to both create and maintain institutions using coordinative discourse and to change institutions using their foreground ideational abilities in communicative discourse. In accordance with Schmidt's (2012) explanation of institutions, this study uses discursive institutionalism to improve analyses of institutional changes and displacements in relation to analyses of different levels and arenas within curriculum theory, historically anchored in the sociology of knowledge (Lundgren 1989;Deng and Luke 2008). DI studies are characterised by (1) taking an interest in the content of ideas in terms of 'what is' and 'what ought to be' , (2) being concerned with the interactive processes of ideas and the ways in which ideas are exchanged and modified through discourse, (3) understanding institutional structures as both constraining and enabling and (4) providing insights into the dynamics of institutional change (Schmidt 2010). Thus, because of its focus on institutional ideas and discourses from a local perspective, DI is useful for tracing transnational ideas across contexts and arenas.

Discourses understood as the content and exchange of ideas on different levels
DI focuses on two basic forms of discourse: (1) coordinative discourse among policy actors in policy arenas and (2) communicative discourse between policy actors and the public. These two forms of discourse have two functions: formulating the content of ideas and developing the interactive processes through which ideas are conveyed. Central to both types of discourse are the actors involved in the policy process. Actors on different organisational levels establish discourse communities to interpret, shape, recontextualise and influence policy. Discourse communities constitute the basis for the shaping of different and conflicting discourse coalitions. Discourse communities also unite the social construction of discourse with those social practices that fall outside of and yet affect discourse (Schmidt 2012a).
Ideas can be understood as the basis for shaping institutional discourse at three different levels of generality. In this respect, DI exhibits a clear parallel with CT. In DI, the most basic level of ideas is a kind of social philosophy or social atmosphere: a 'public philosophy' or public worldview. This level of ideas, which underpins policies and programmes, is difficult to identify via specific criteria because these ideas are often left unarticulated, serving exclusively as background knowledge for 'public philosophies' or worldviews (Schmidt 2008, 306). The DI level reflecting the emergence of ideas that shape a society's basic conception of education and other institutional activities has its parallel in the first level of CT, the societal level, where ideological ideas on education are formed. Ideological-level studies within CT examine how knowledge, values and experiences are distinguished, selected and organised for education. These questions involve historical, cultural and socioeconomic aspects of society and form the principles from which a national curriculum can be created (Lundgren 1989;Deng and Luke 2008). According to Campbell (2004), background assumptions and public sentiments on the institutional level (ideological ideas) are cognitive and normative ideas that restrict the range of programmes and policies that are available to politicians and decision makers. Public philosophies contribute to collectively shared perceptions and opinions about what measures should be considered legitimate. In the framework suggested in this article, background ideas will be related to the institutional/societal level as conditions for change, since, as Schmidt (2008, 306) notes, 'the philosophical ideas generally sit in the background as underlying assumptions that are rarely contested…' . Foreground ideas, by contrast, are related to the programmatic level, where politicians and policymakers try to change institutions by communicating programmatic ideas to the public through reform proposals. For this reason, coordinative discourses are linked to the conditions shaped by common background ideas as a 'public philosophy' at the institutional/societal level, and communicative or 'persuasive' discourses are linked to the programmatic level, where politicians try to persuade the public of the benefits of their proposals. Like all frameworks for theoretical analysis, this framework represents a schematic simplification of a far more complex reality. Its purpose is to highlight core concepts and arenas by unravelling and focusing on theoretically valid and empirically significant distinctions.
At the second level of generality, the programmatic level ('programmatic beliefs' in DI [Schmidt 2008] and 'the intended curriculum' in CT [Lundgren 1989]), researchers seek patterns describing how ideas from public philosophy are formulated into more programmatic ideals, problems and desired norms in specific policy approaches or political programmes. Programmatic ideas comprise the character of the problems to be solved; the ideals, norms and values to be considered; the applicable instruments to be used; etc. A well-known example of a clear rupture of ideas at DI's programmatic level is the shift in economic politics that occurred when Prime Minister Margaret Thatcher led the British government during the 1980s (Schmidt 2008). In CT, the second level refers to the actual governing of a nation's school system, including its curricula, education acts and authoritative guidelines. At the programmatic level, the focus is on the governing, implementation and evaluation processes and on how discourses on curriculum content and assessment are shaped and contested in different arenas. Lastly, the more specific level of ideas in DI comprises 'the specific policies or 'policy solutions' proposed by policy makers' (Schmidt 2008;306). In DI studies, scholars who focus on this level try to answer the question of why some ideas succeed and evolve into policy while others do not (see also Kingdon 1995). An important approach within this field of institutional studies is that good ideas (i.e. ideas that appear relevant to the problems at hand) succeed while others fail. However, many ideas that could have been considered 'good' have failed, and vice versa. Thus, it seems a difficult task to develop these sorts of answers solely by examining policy discourse at this third and most specific level. This is an important argument for including all three levels in policy analysis. Within CT, this DI 'policy solution level' corresponds in our example to a self-governing municipal level, where, in the Swedish case, local politicians are responsible for the provision of schooling in accordance with state goals and regulations. Within CT, we call this municipal level 'the first local level' (i.e. a local policy arena) to distinguish it from the discursive/practical classroom level, or 'the second local level' . This second local arena does not really correspond to the third policy solution arena in DI; however, DI is helpful for analysing how discourses are communicated and formed within the school and classroom arenas.
In sum, we argue for a connection between DI and CT because, as Schmidt (2010, 15) argues, 'without discourse, understood as the exchange of ideas, it is very difficult to explain how ideas go from individual thought to collective action' . CT has long lacked a theoretical understanding of the exchange of ideas and the path of institutional change. What CT contributes to DI, instead, is a rich theoretical tradition of studies of reproduction, power and the types of knowledge that are most worthwhile (Young 1971;Bernstein 2000;Young 2008;Bernstein and Lundgren 1983).
We combine DI and CT by distinguishing among understandings of ideas and discourses at three different levels: the overarching societal philosophy in terms of public norms and assumptions; the programmatic level, where some ideas are selected for implementation in curricula and governing norms; and the local level, where policy solutions actually take shape and are implemented with consequences for actual institutional activities, like schooling. Whereas CT often considers the societal level as the first level and the 'classroom level' as the third level, Schmidt (2008) identifies the more specific 'policy solution level' as the first level and the general 'philosophical' level as the third level. Both DI and CT include programmatic levels in the middle of their analytical frameworks. In our combination of DI and CT, we have retained the CT tradition by identifying the societal level as the first level; however, it is important to emphasise that all analytical levels must be considered and that it is often advisable to begin an analysis at the third and most specific level.

Two different kinds of ideas and discourses
There are two different sites for ideas: in the background, as underlying assumptions, and in the foreground, as conscious perceptions. Ideas can also be cognitive or normative. Normative background ideas are more or less unspoken thoughts on values and identities that fit within the concept of public opinion or public philosophy (e.g. 'all children have the right to go to school'). Cognitive background ideas are often termed paradigms: They are basic thoughts with a common understanding so strong that they are taken as self-evident (e.g. 'children need to learn to read and write').
Foreground ideas are more visible than background ideas and, thus, are more regularly contested. Normative foreground ideas constitute frames of possible proposals in debates that enable politicians and decision makers to legitimise their programmes for institutional change on certain specified grounds (e.g. a common and compulsory school organisation for all pupils for nine years). Cognitive ideas are typically foregrounded in programmatic debates that discuss cognitive concepts and analyse ways to solve different types of problems, thereby promoting certain forms of decision-making and institutional change (e.g. the actual structure and content of a curriculum). All four types of ideas -cognitive and normative background ideas and cognitive and normative foreground ideas -have distinct effects on the process of institutional change (Campbell 2004).
In essence, one could say that the concept of background ideational abilities explains how institutions are created and how they persist. Furthermore, the concept of foreground discursive abilities explains why institutions change or remain unchanged (Schmidt 2010). The term 'discourse' refers to both the representation of ideas and the process of idea exchange. Agents of communicative discourse include policy actors, professional actors, parents, the media, social movements 'and even ordinary people through their everyday talk and argumentation … ' (Schmidt 2012a, 102). In the policy field, communicative discourse represents the processes through which political actors present and argue for their ideas for political legitimation.
When the DI concept of discourse, which developed within political science, is moved to the educational research arena at the intersection of policy and curriculum research, it becomes possible to analyse communications among different levels of institutions. In such an analysis, there is a need to distinguish between at least two types of communicative discourse. The term 'persuasive communicative discourse' refers to communication from the political arena and/or the national authority arena designed to 'interpret' a problem and to persuade the public regarding its solution. Persuasive communicative discourse denotes authoritative and asymmetrical (i.e. one-way) communication in which where critical voices are absent or do not make themselves heard. In the Swedish case, this kind of persuasive communicative discourse has been obvious in communications from the national authority arena (the programmatic arena) to the first and second local arenas. The other important aspect of communicative discourse, which is closely connected to Habermas (1994) belief that deliberative conversations among all those concerned with an issue should be referred to as a 'deliberative communicative discourse' , can be understood as critical deliberation within and outside the school as an institution reflecting the meanings of policy discourses and key concepts. Deliberative communicative discourse also contains persuasive elements; however, in this definition of discourse, the persuasive elements are open to more symmetrical and reciprocal contestations and counterarguments. In our study, such deliberative communicative discourse on curriculum reform is largely absent. When it is relevant to talk about deliberative or persuasive communicative discourses is, thus, a primarily empirical question.
With its concepts of coordinative and communicative discourse, DI contributes to an understanding of the complex and interactive processes of policy influence. It also helps to distinguish between discourses of first-order change, in which only the policy instruments are replaced; second-order change, in which objectives change; and third-order change, in which the core of the policy as a whole shifts (Schmidt and Radaelli 2004). However, DI has less to say about what happens when discourses are conveyed and/or moved across different arenas. Here, DI can be complemented with the concept of recontextualisation familiar to CT research. Bernstein's (2000) principle of recontextualisation furthers our understanding of how educational policy texts are shaped by multiple recontextualising agencies, including teachers' enactment of policies in schools. A transformation takes place when a discourse moves from its original site into a pedagogic arena, where different elements of the discourse's meaning are selectively appropriated, relocated and/or refocused. Although distinct elements of the discourse can be recognised in these arenas, the discourse is no longer the same discourse, since the potential for partially new meanings has been created via the transformation from one arena to another, which is partly embedded in other material contexts. Singh, Thomas, and Harris (2013) show how principles of recontextualisation can contribute to theorisations of power and control in the policy process, with a focus on the significance of mid-level policy actors. Similarly, Tsataroni, Ravanis, and Falaga (2003) employ the concept of recontextualisation in their study on the effects of changes in education policy on pre-school practice. Schriewer and Martinez (2004) develop the theory of 'externalisation' to describe the 'filtering' of the reception and description of an international environment in relation to changing situations internal to a given system. Steiner-Khamsi (2002) suggests extending this concept of externalisation in order to highlight three distinct phases in the process of policy borrowing: (1) externalisation, (2) recontextualisation and (3) internalisation. By focusing on the local level and on what has not been affected by transnational policies, it becomes possible to analyse local policies and priorities.
In Figure 1, we present an overview of the various terms used above and their relations in order to clarify the framework we use to analyse how a national school system can be institutionally maintained and changed through communicative actions in terms of discourses.

Arenas
Background ideas  Figure 1. an analytical framework for understanding educational reforms at different levels in terms of ideas, discourses and actors. notes: in the column to the left, the different levels of ideas from di (public philosophy/institutional level, programmatic level and policy solution level) have been combined with the levels of ct (societal level, curriculum level and the first local and the second local level). the analytical framework represents a schematic explanation to understand the relation between policy impact and education reforms.
To study the ways in which ideas affect both change and stability, it is important to (1) establish what is meant by ideas and distinguish between different forms of ideas and the spheres in which they are formed, (2) define what is meant by discourses and distinguish between different forms and functions of discourses, (3) establish which actors are crucial to how ideas are transferred and transformed through different discourse coalitions and include the different arenas and institutional levels relevant for understanding how various ideas affect education, (4) determine how ideas are related to and affect decisions and activities at various levels and time scales and (5) identify the conditions that seem relevant to how discourses both stabilise and change the institutional base for school education. Like the pluri-scalar governance model of education suggested by Dale (2005), which illustrates that education policy can no longer be seen as the exclusive preserve of individual nationstates, the present framework comprises different scales. However, Dale's (2005) arguments for a new functional, scalar division of education governance labour do not address the ways in which policy ideas would be communicated within this new scheme. In a more recent article, Dale (2010) explores the mechanisms through which globalisation affects national policy with a focus on delivering global mechanisms, rather than on discourses of transnational policies. Likewise, Phillips and Ochs (2013) have proposed explanatory and analytical devices demonstrating the extensive process of policy borrowing at different stages, rather than highlighting communicative dimensions for disseminating or resisting policy borrowing.
By relating core concepts from DI to CT, we argue that both policy researchers and curriculum theory researchers have access to strong theoretical concepts for investigating aspects of how, when and why certain policy ideas emerge, gain influence and are operationalised in actual education, while others remain superficial and still others remain stable over time. Programmatic initiatives and local activities need to be taken into account to explore the significance of policy ideas on all three levels of recontextualisation processes: the institutional, the programmatic and the local.

Travelling policies and the processes of recontextualisation
In this section, we draw attention to several implications of the theoretical framework in relation to the empirical case of curriculum change for compulsory schools in Sweden. The 2011 Swedish curriculum reform, Curriculum for the Compulsory School, Pre-School Class and the Leisure-Time Centre, or Lgr 11 (NAE 2011), is located at the nexus of transnational and national policy discourses. According to the Swedish Ministry of Education (Government Bill 2008/09:87), this most recent curriculum reform is part of a package of reforms introduced with the aim of strengthening national governance and reorienting schools towards a clear focus on knowledge, while emphasising continuity with former national curricula. The most important question regarding the overall construction of the curriculum is where its centre of gravity lies. An investigation into the knowledge focus of the curriculum requires a close and critical text analysis of curriculum type, construction and content.
Our analysis of the national curriculum arena shows that transnational influences and policy flows have increased considerably since the pervasive Swedish education reform of 1991, which represented a discourse of third-order change in which the core of the policy as a whole shifted. 3 Since this 1991 reform, two curriculum reforms have been undertaken in Sweden: Curriculum for the Compulsory School System, Pre-School Class, and the Leisure-Time Centre, or Lpo 94 (NAE (Swedish National Agency for Education) 1994), and the aforementioned Lgr 11 (NAE (Swedish National Agency for Education) 2011). The processes of recontextualisation and translation to national conditions, however, remain unrecognised. The first step in our analysis revealed that Lgr 11 was essentially an extension of the 1991 reform, with a general framework built on management by objectives and results. However, whereas the first compulsory school curriculum within the framework of the 1994 reform, Lpo 94, focused on competencies, its successor, Lgr 11, also included strong demands regarding measurable outcomes. The knowledge focus has, thus, shifted from a competence focus in the 1994 curriculum (Lpo 94) to a results focus (Nordin and Sundberg 2016;Wahlström 2016). The second step in our analysis revealed that central elements of international standards-based curriculum reforms had been appropriated into the core construction of the Swedish national curriculum of 2011 (Lgr 11), including standardised knowledge requirements and their alignment with grading criteria. In this regard, the Swedish curriculum is an example of a standards-based curriculum, in line with the international technical-instrumental curriculum discourse and combined with a neoconservative perspective on curriculums as a given and uncontested body of knowledge (Sundberg and Wahlström 2012). In the following, we analyse the process of recontextualisation in relation to the ideological/societal, programmatic and enacting levels in order to demonstrate how DI concepts can be used in empirical policy studies on curricular issues.

The institutional/societal level
In the examination of policy flows and policy content, there is a need to include both transnational and national policy arenas that affect discourses on education because today, more than ever, policies are 'boundless' in their discursive communication. At the same time that documents were being prepared for the 2011 Swedish curriculum reform, the normative background ideas in the transnational arena were dominated by a neoliberal agenda characterised by state deregulation and global competition (see Schmidt 2009). Though such policies seem similar, they are not identical when recontextualised in different national arenas, since different states take different positions on how and to what extent different sectors should be deregulated. Though different nations experienced different outcomes of their neoliberal agendas, the transnational policy arena was, overall, heavily influenced by the neoliberal doctrine (Gunter et al. 2016). The main actors in relation to the Swedish curriculum reform were the Organisation for Economic Co-operation and Development (OECD) and the European Union (EU), as well as private international companies, such as McKinsey & Company and Pearson.
The paradigmatic principle following from these background ideas is mainly that the nations of the OECD and the EU need to cooperate more closely to cope with global competition. A second cognitive conclusion is that future opportunities will centre on individuals (i.e. their knowledge, skills, abilities, etc.) and 'human capital' , such that individuals will be regarded as one of the primary factors of production. For the OECD, this line of thinking yielded a focus on comparative knowledge assessments in the form of the Programme for International Student Assessment (PISA). Such assessments would allow member countries to compare their school and learning systems to those of the most successful nations (Meyer and Benavot 2013). However, initial positive expectations in terms of Europe's ability to rapidly transform into a competitive and sustainable economy with a greater degree of social cohesion ended in 2005, when a theme of crisis emerged. Globalisation and rapid changes legitimised the crisis discourse, creating both a sense of urgency and a 'point of no return' (Nordin 2012). The image of crisis, in turn, legitimised the need for radical change through comprehensive reforms in the EU member states (Robertson 2008). These reforms were largely underpinned by extensive evaluations and comparisons, a technical paradigm of knowledge and numbers (Ozga 2008;Grek 2009).
In the Swedish arena, the shift in normative background ideas was epitomised in an official report on democracy and power published in 1990(Official Report 1990. Until then, the Swedish school system was considered to be an integral part of the Swedish welfare model, and it was based on the idea of education as a civil right in a welfare state. The official report launched an individual-centred democratic tradition as an alternative to the former society-centred democratic tradition. The starting point for this newer tradition was that of the sovereign individual who takes responsibility for his or her own destiny, a perspective that aligned with transnational neoliberal developments in other Western countries (Englund 1998;Schmidt 2009). Thus, through the displacement from a welfare state to a welfare society, the paradigm shifted from a state-governed common school system to a decentralised school system, which opened the door for private schools and a voucher system for free school choice. A simultaneous change occurred in the coordinative discourse concerning the public philosophy of how schooling should be governed, shifting the rhetoric from state control marked by detailed financial regulations to a decentralisation of the school system via management based on results and control (Wahlström 2009).

The programmatic level
The policy goal 'Lifelong Learning for All' was identified in 1996 as a programmatic framework (normative foreground idea) for the OECD, which the EU echoed in its proclamation that 1996 was the 'European year of lifelong learning' . The underlying assumption was that rapid societal changes related to the continuous development of information technology and the knowledge-based economy highlighted the need for people to be able to respond quickly to structural changes in their working lives. The international networks and contacts that made up the transnational policy arena formed coordinative normative discourse of common interests and similar worldviews (Lawn and Grek 2012). The challenge was not just that the demands on the workforce (in terms of skills, abilities and knowledge) were increasing, but that members of the workforce were also expected to change jobs and professions several times during their working lives. The cognitive policy solution to this problem became the 'key competencies' concept. In response to this solution, the OECD (2005) launched the programme 'The Definition and Selection of Competencies' (DeSeCo), and the EU (European Communities 2007) formulated a European reference framework for eight key competencies for lifelong learning. The period between the late 1990s and the early 2000s was one in which adapting to the discourse of lifelong learning was a necessary prerequisite for education reforms in Europe (Grek 2013). The aim of these programmes was to include learning competencies in national school curricula by leveraging member states' own lifelong learning strategies.
Though the EU programmes 'Education and Training 2010' (European Commission 2003), 'Europe 2020' (European Commission 2010) and 'Rethinking Education' (European Commission 2012), the cognitive foreground of programmatic discourse was shared with the member states through communicative discourse. Consequently, while coordinative discourse grew stronger in the transnational policy arena, the national arena witnessed the rise of communicative discourse. Against the backdrop of a strong coordinated transnational discourse, member states' communication began to focus more on implementing an agreedupon agenda. Thus, the coordinative discourse in the transnational arena contributed to a specific form of education policy denationalisation in the national arena (Robertson 2013;Wahlström 2016).
An analysis of the three most recent curriculum reforms in Sweden reveals that a displacement of the normative foreground ideas regarding the legitimisation of educational reforms occurred between 1970 and the early 2000s. Concerning the organisation of the reforms, the state moved from a tradition of official investigations and reports in the 1970s, in which the task of the investigation was to pose relatively open questions across broad representations of different parts of society, to an increasingly narrow approach in the 1990s and the beginning of the twenty-first century, in which task of the investigation in terms of policy solutions was specified beforehand and the investigation was carried out by an individual investigator. This shift represented a simultaneous change in the role of the state, which transformed from being the centre of coordinative discourse on the future of education to being the centre of communicative discourse on what changes were needed for the future of education within the context of an agenda set by transnational coordinative discourse.
During this same period, state policy shifted from supporting the societal role of the school as constituting a common frame of reference in the 1980s, such that the need for individualisation was addressed within a common school organisation, to directly contradicting this political argumentation and, instead, arguing for free school choice and the authorisation of private school organisers, including a voucher system for the distribution of resources, at the beginning of the 1990s. This normative displacement was promoted as a necessary response to a pluralistic society comprising individuals with different needs and preferences (Wahlström 2014b). In the international discourse of lifelong learning and the agreement of key competencies within the OECD and the EU, the individual was placed at the centre, and the school seen as having an obligation to 'deliver' solutions to each individual's needs. Proponents argued that such an individual focus would be impossible within a common public school. As Bill (1991/92:95, 8) notes, though '[a] uniform system cannot offer the variety and flexibility as the future requires of the school system' , a 'stimulating competition' among schools can contribute to increasing the quality of the school system. This normative displacement of foreground ideas from a collective to an individual perspective of schooling was simultaneously based on a discourse of international human rights, which argued that 'The right and opportunity to choose a school and to choose the children's education is important in a free society. This basic principle has also been expressed in a number of international conventions to which Sweden has committed itself ' (Bill 1991/92:95, 8).
In response to this displacement in normative foreground ideas, the cognitive ideas formulated through the extensive reform agenda of the 2000s emphasised high-quality education across all schools and enhanced monitoring of knowledge results and assessments in compulsory schools to support the international competitiveness of the Swedish school system. Reforms initiated by this line of thought included the implementation of a state school inspectorate and the implementation of recurrent national knowledge tests in compulsory school. The cognitive ideas contained in the new School Act (2010) and the new curriculum for compulsory school (2011) were largely motivated by Sweden's declining performance in international knowledge tests, including the Progress in International Reading Literacy Study (PIRLS), the Trends in International Mathematics and Science Study (TIMSS) and the Programme for International Student Assessment (PISA). The PISA surveys, in particular, played a prominent role in supporting the cognitive ideas underlying a series of implemented reforms. PISA 2012 revealed a sharp downward trend in the performance of Swedish 15-year-olds in all three core test subjects over the last decade, representing a larger decline than that of any other OECD country during the same period (OECD 2015). In response to this disappointing result, the Swedish Ministry of Education and Research invited the OECD to conduct a review of the quality of school education in Sweden, with a primary focus on compulsory education. The OECD review (2015) presented a wide range of proposals for increasing the quality of the Swedish school system, largely motivated by various PISA survey results. This review, in turn, served as the basis for the 'School Commission 2015' , which was tasked with presenting proposals to increase knowledge achievement, improve education quality and increase equivalence in Swedish schools (Swedish Ministry of Education 2015). The close connection and coordination between the international and national arenas implies that national policy actors implemented a strong communicative and persuasive discourse when implementing their reform ideas within the national arena.
Consequently, cognitive foreground ideas for institutional change in education are shaped by common discourse coalitions rooted in close relations between international policy arenas, such as the OECD, and national policy arenas. The initiative to invite the OECD to evaluate the Swedish school system was taken by a coalition government comprising four non-socialist parties. When the report was presented in 2015, however, the government had changed, and the report was received by a coalition of the Social Democrats and the Green Party. Therefore, though it was a non-socialist government that gave the assignment to the OECD, it was a Social Democratic/Green Party government that established the Swedish School Commission and linked the OECD's proposals to the Commission's assignment. However, there was no linear transfer from international policy to the national curriculum, the Education Act or guidelines from authorities. Instead, the policy proposals from the OECD report were to be recontextualised to fit Swedish traditions and norms in the report from the Swedish School Commission. Some of the Commission's proposals may be recognisable as OECD recommendations, but they will not be identical. Thus, in the Swedish case, there seems to be a stable coordinate discourse concerning the programmatic foreground ideas that dominate the national policy arena. Rather than differing on these ideas, the political parties, supported by media and other arenas for public debate, contest the cognitive ideas of policy solutions: that is, the concrete proposals for change.

The first and second local levels
The concept of lifelong learning and its associated competencies are the main features of the knowledge economy. During the 2000s, there was a continuous shift in focus from learning goals to 'knowledge requirements' and from schools as arenas for citizenship learning to schools as 'delivery agencies' for students who are now perceived as customers in a specific education quasi-market (Biesta 2005;Young 2008). The market-and competition-oriented rhetoric gained traction in Swedish policy debates during the 2000s, reinforced by the declining performance of Swedish students in PISA surveys and the resulting image of a Swedish school system in crisis. The Swedish compulsory school system appeared to one in need of more explicit demands and firmer systematic governance, accountability and evaluations on all levels. This was the normative, coordinative background discourse against which the cognitive discourse of a new curriculum reform for compulsory school was communicated to policy actors and professionals at local levels by the government in 2011.
In Lgr 11, a technical-instrumental form of curriculum and a neoconservative view of curriculum content were brought together in a combination made possible by predefined content and assessment standards, called 'knowledge requirements' (Sundberg and Wahlström 2012). In the neoconservative tradition, which is often embedded within education organisations themselves, schools are responsible for transmitting predetermined subject content for new generations to learn. However, over the last few decades, this traditional view of the task of schools has been challenged. Policy representatives have advocated a more technical-instrumental form of curriculum, in which the basic needs of the national economy and labour force are placed in the foreground. Thus, education has become a means to an end rather than an end in itself (Young 2008). The national and local interpretations of the concepts proposed by the EU and the OECD, such as standards, competencies, accountability and diversity, have produced a neoconservative idea of school subjects as the organising principle for the Lgr 11 curriculum. In this sense, Lgr 11 diverges from the EU's transnational education discourse, which, instead, promotes a cross-curricular approach to basic skills and subjects. Therefore, instead of perceiving competencies as an overarching term, Lgr 11 subordinates and adapts the term 'competencies' to specific subject-based knowledge and skills (Nordin and Sundberg 2016).
In a theory-based evaluation of reform implementation, the results indicate a strong coordinative discourse within each school included in the survey regarding how to adapt to the new curriculum and how to adapt the new curriculum to current local teaching traditions. Simultaneously, there is a distinct communicative discourse between the NAE and teachers regarding how to interpret and implement the Lgr 11 curriculum through websites and national guidelines. The Lgr 11 curriculum is a means for transmitting a communicative discourse concerning both the content and the conveyance of ideas to teachers. A vast majority of surveyed teachers 4 regarded Lgr 11 as either important or very important for their perceptions of their assignments as teachers, indicating that the pervasiveness of this discourse only a few years after the reform's initial implementation (Wahlström and Sundberg 2015).
Soft governing by means of guidelines, recommendations and support material has had a significant impact on teachers' curriculum work and teaching practices (Wahlström and Sundberg forthcoming). The 'tightening up' of different parts of the curriculum (e.g. to strengthen the alignment between different subject content areas and assessment criteria) has been part of a coordinative discourse about how to apply and implement the curriculum. This increased centralisation of control is, however, double-edged. On one hand, teachers perceive that increased control creates clarity in their mission, legitimising their teaching on the basis of the curriculum and focusing their teaching subject content (which was a major reason for the reform). On the other hand, teachers express a conflict between governance and professionalism. In curriculum policy texts, Swedish teachers are mainly positioned (and perceived) as curriculum implementers and, to a lesser extent, curriculum developers.
The enacted curriculum is strongly focused on technical questions concerning 'how' and 'when' and is less focused on pedagogical questions regarding 'what' and 'why' . Such results can be interpreted and explained by a strengthened coordinative discourse among actors at different levels of the reform process. Case studies at local levels of schools and classrooms identify how traditional background ideas and versions of teaching with teacher-led wholeclass instructions and recitations are highly present in classroom curricula and discourses (Wahlström and Sundberg forthcoming).

A concluding discussion
In this article, we argue for discursive institutionalism as the bridge that connects comparative policy studies with curriculum research. We introduce a framework for policy studies from a local perspective by integrating core elements and concepts from DI and CT as a way to overcome risks of one-sidedness through an overemphasis on either 'transnationalism' or 'nationalism' . By contextualising a theory-based evaluation of the Swedish Lgr 11 curriculum reform that considers all three analytical levels -local (first and second level), programmatic and societal -we show that policy transfer is dependent on and interacts with coordinative and communicative institutional discourses (Wahlström and Sundberg forthcoming).
Since the early 2000s, a strong normative coordinative discourse coalition formed by transnational and national Swedish politicians and experts within the OECD and the EU has promoted an education policy discourse based on the premise that each national education system must prepare its students to be competitive in a global knowledge economy. This discourse coalition was challenged by a communicative discourse coalition of Swedish non-socialist politicians who advocated institutional foreground cognitive ideas of subject-based knowledge requirements and 'order in school' that shaped the Lgr 11 curriculum reform on the programmatic level. The consequences of the struggle between these partially opposing institutional forces manifest in a CT analysis on the programmatic level, where such core concepts as accountability and competence can be attributed to a coordinated transnational/national institutional ideology and the concept of 'knowledge requirements' can be identified as drawing from a nationally based institutional foreground understanding of the purpose of schooling. Whereas DI contributes discursive analyses of the exchange of ideas and the character of the discourses by which these ideas are communicated, CT investigates the enactment and consequences of institutional discourses in national official policy documents, national curricula, versions of classroom curricula and classroom discourses, taking aspects of power, knowledge concepts and social reproduction into account.
In Sweden, compared to the former Lpo 94, Lgr 11 can be understood as an institutional second-order change. This means that the aim of school education has shifted through a displacement from a normative coordinated discourse on democracy to a cognitive communicated discourse on knowledge as the school's main goal, without any change to the institutional foundation of school governance by objectives and results.
Schools as institutions are formed by the norms and rules that enable and constrain school activities. These norms and rules are made possible or impossible by the coordinative and communicative discourses of actors inside and outside the institution: that is, of politicians, education experts, teachers, students and citizens (see Schmidt 2012). Education institutions, which comprise norms and values formed simultaneously and often contradictorily across different arenas, are the basis for education reforms, but institutions cannot govern the schools themselves; instead, they formulate the playing field. School organisation is crucial for governing a school system, and the school system is organisationally maintained by actual national and local structures expressed through curriculum content, national systems for knowledge tests, school inspectorates, economical resource distribution systems, legal regulations regarding the authority to provide education, etc. Thus, whereas DI can contribute with analyses of institutional changes and resistance, CT can contribute with analyses of the intersections between institutional changes and organisational enactments to support a better understanding of the meaning of the interactions among institutions, organisations and social actors in the shaping of education reforms. Closer cooperation between CT and policy studies would most likely benefit from DI's concepts of different types of ideas and discourses, as well as its interest in social actors. Moreover, via the concept of recontextualisation, processes of policy import and export are problematised as practices of translation, contestation and negotiation when discourse elements move across discourses. Any assumption of linearity from one level to another can, thus, be questioned, and the transference of education policy can be unravelled and re-understood by studying local actors in their institutional and organisational contexts and by distinguishing between 'persuasive' and 'deliberative' communicative discourses.
The theory of discursive institutionalism, complemented by CT's concept of recontextualisation, highlights the agency and translation processes of different kinds of institutional elements when evaluating the effects of implementation in different contexts. This Swedish case study demonstrates the crucial role of theoretical concepts in analysing the seemingly technical and national character of curriculum reforms. When powerful transnational discourse coalitions advance their positions by defining key reforms linked to national curricula and national evaluation systems, it becomes increasingly important for researchers to combine concepts and theories from comparative policy studies and curriculum theory studies in order to trace discourses and effects across different national education reforms.

Notes
1. For an overview of the Nordic curriculum tradition, see Gundem, Karseth, and Sivesind (2003). 2. Discursive institutionalism can be defined as both distinct from and complementary to the other three new institutionalist approaches: rational choice institutionalism (rationalist interests), historical institutionalism (path dependence) and sociological institutionalism (cultural framing) (Meyer and Rowan 2006;Schmidt 2008). 3. The present analysis is based on Sundberg and Wahlström (2012), Wahlström (2014aWahlström ( , 2016, Wahlström and Sundberg (2015) and Nordin and Sundberg (2016). The referred policy analyses concern transnational educational policies related to the Swedish curriculum reform for compulsory school (Lgr11). The analysis considers approximately 25 documents from the European Commission and the Council of the European Union, approximately five documents from the OECD and approximately 15 national policy documents. The articles are related to the research field of comparative education, and the reading of the policy documents is based primarily on critical discourse analysis. 4. The survey was part of a quantitative study of primary and lower secondary school teachers' (n = 1887; response rate = 62%) understandings and performances of the enacted curriculum reform in 2013. It covered teachers in grades 6 and 9 in all subjects. The initial survey study was followed up by a qualitative interview study (n = 18).