Mainstreaming or retrenchment? Migration-related diversity in Dutch and Flemish education policies

This article analyses how states adapt generic policies to the increasing diversity that characterises contemporary European societies. More particularly, it zooms in on how migration-related diversity is mainstreamed into education policies in the Netherlands and Flanders and why we observe different policy trends in these two cases. We find that the focus on migration-related diversity largely faded in Dutch education policies in the period from 2000 to 2014. In Flanders, this trend towards ‘migration-related diversity retrenchment’ is less prevalent during this period, even though a similar evolution has started to take place more recently. These findings present a puzzle, as the most evident explanation for diversity retrenchment, namely the increasing politicisation of migration and diversity, cannot account for this difference since the Netherlands and Flanders are characterised by similar degrees of politicisation of migration-related diversity. Our findings thus call for an exploration of underemphasised explanations for diversity retrenchment. We show that the diverging degree of diversity retrenchment can be explained by the presence or absence of a sub-state nationalist project and diverging degrees of neoliberal retrenchment policies. Sub-state nationalism seems to have temporarily offered a buffer against the neoliberal retrenchment of migration-related diversity.

Page 2 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 or a shift of responsibilities for the governance of migration-related diversity across policy domains (Martinelli, 2014;Scholten & van Breugel, 2018;Westerveen & Adam, 2019). Rather than treating immigrant integration as a separate institutionalised policy domain, within a mainstreaming strategy, the governance of migration-related diversity is addressed as an integral part of generic policy domains. Yet, so far only a few studies have focused on how immigrant integration is addressed in generic policy fields such as education, housing, and employment and on how governments coordinate immigrant integration across policy sectors (i.e. horizontal coordination, see Spencer, 2011;Scholten & van Breugel, 2018;Westerveen & Adam, 2019) or across governance levels (i.e. vertical coordination, see Adam & Hepburn, 2019;Adam & Caponio, 2018).
In this article, we explore whether, how, and why migration-related diversity is mainstreamed into education policies in the Netherlands and Flanders (Belgium). With the disappearance of centralised immigrant integration policies and the aim to bring immigrant integration 'into the mainstream' , it becomes important to study how migration-related diversity is addressed within the field of education that has always been considered a crucial field for immigrant integration (Rijkschroeff et al., 2005). Flanders and the Netherlands have historically gone through similar policy developments in the field of immigration and immigrant integration (Jacobs & Rea, 2012). Moreover, the cases are characterised by similar educational systems, resulting from a tradition of pillarisation in which groups with different religious or philosophical convictions could establish their own schools (Franken & Vermeer, 2019;Loobuyck & Jacobs, 2010).
The article proceeds in four steps. In the first section, we describe the conceptual framework. Bringing together the literature on mainstreaming as a public policy strategy and the literature on multicultural education, we propose an innovative set of four indicators to assess the mainstreaming of migration-related diversity into education policies: 'targeting' , 'monitoring' , 'content integration' , and 'diversity in the school system' . These four indicators account for both the incorporation of migration-related diversity in mainstream policy fields as well as the adaptation of mainstream institutions to migration-related diversity. In the second section, we present our theoretical framework, explaining policy trends in the mainstreaming or retrenchment of migration-related diversity. In the third section, we explain our research design and method of analysis, followed by our comparative findings on whether, how, and why migration-related diversity is mainstreamed into Dutch and Flemish education policies. We demonstrate that the focus on migration-related diversity largely faded in Dutch education policies in the period from 2000 to 2014, while in Flemish education policies this trend towards the retrenchment of migration-related diversity is less prevalent during this period. Taking into account some recent changes in Flemish education as well as integration policies, our comparison shows that Flanders moves in the direction of migration-related diversity retrenchment at a different pace than the Netherlands. We explain this diverging pace of diversity retrenchment by pointing to the presence of sub-state nationalism in Flanders and diverging degrees of neoliberal retrenchment policies in the Netherlands and Flanders. More specifically, we suggest that sub-state nationalism has, at least temporarily, offered a buffer against the process of retrenchment. In the conclusion, we draw together the findings of our study and propose suggestions for further research.
Since the early 2000s immigrant integration policies in Europe would have been marked by a so-called 'multiculturalism backlash' (Vertovec & Wessendorf, 2010;cf. Kymlicka, 2012) and subsequent 'assimilationist turn' (Joppke & Morawska, 2003), away from group or community-oriented approaches and with an increasing (symbolic) emphasis on adaptation. More recently, scholars have pointed at a shift towards a 'mainstreaming' approach with a focus on broad immigrant integration or citizenship programmes, distributed over several policy areas (Martinelli, 2014;Scholten & van Breugel, 2018;Westerveen & Adam, 2019). What are the minimal conditions to speak of such a mainstreaming approach in the field of education? To develop a conceptual framework for assessing the mainstreaming of migration-related diversity in education policies, we first turn to the policy literature on mainstreaming (e.g. Booth & Bennett, 2002;Priestley & Roulstone, 2009;Semmel et al., 1979). This literature conceptualises mainstreaming as a two-sided process. On the one hand, mainstreaming involves a transition from separate policy measures that target specific groups to the incorporation of these target groups in mainstream policies. On the other hand, mainstreaming requires that policy programmes are adapted to accommodate the needs of specific groups in the mainstream (see Boucher & Maclure, 2018). Moreover, the literature emphasises the need for active coordination of mainstreaming across policy sectors or levels.
To adapt our set out of indicators for assessing migration-related diversity mainstreaming to the field of education, we borrow from the literature on multicultural education. This literature focuses on different approaches towards adapting schools and other educational institutions to ethno-cultural diversity, as to reach educational equality (Banks, 1993, 3). We can make a distinction between more particularist and more universalistic approaches to multicultural education (Eldering, 1996). Particularist approaches are solely targeted towards ethnic minority students, while universalistic approaches target all pupils in a given context. A further differentiation can be made between approaches that focus on remedying educational disadvantages (e.g. language deficits) and approaches that focus on the promotion and preservation of cultural diversity (e.g. promoting bilingualism) (Eldering, 1996, 318-321). The most widespread and well-known conception of multicultural education is the reform of school curricula to integrate content on ethnic minorities, known as content integration (Banks, 1993;Bennett, 2001). Furthermore, multicultural education can also comprise changes in teaching styles, teaching staff, and school structures. For example, multicultural education can entail granting equal rights to minority groups to establish private schools or allowing for reforms in the school system to improve education outcomes for disadvantaged groups (e.g. reducing academic tracking).
Bringing together the literature on mainstreaming and multicultural education, we distinguish four different indicators for migration-related diversity mainstreaming in the field of education. First, we will look at the targeting of migration-related diversity within educational policies. Targeting is a central element of policymaking and essential for an effective redistribution of goods and services. A key question in targeting is which categories are defined (and which are avoided) to construct the target group of redistributive policies, a contested dilemma within immigrant integration policies and studies (see De Zwart, 2005).
For the targeting of migration-related diversity similar dilemmas arise, for example, should students be targeted universally, or should students with a migration background be distinguished? Moreover, if students with a migration background are considered a separate target group, which categories are used to distinguish them?
Second, we will look at the monitoring of migration-related diversity in education policies, as the reverse end of the targeting approach. This includes the presence or absence of 'ethnic monitoring' , or the collection of data disaggregated by ethnic categories (such as birthplace, nationality, and language) as well as the monitoring of specific policy measures related to migration-related diversity, such as citizenship education. Together with targeting, monitoring practices can shed a light on how migration-related diversity is addressed and assessed in the generic policy field of education (Simon et al., 2015;cf. Booth & Bennett, 2002).
Third, in line with the literature on multicultural education, we consider the extent to which content on migration-related diversity, such as minority religions or languages, are integrated into general school curricula. Within so-called content integration, different subapproaches can be distinguished. Teachers can, for instance, utilise examples from different cultural groups (such as holidays and names) in their teaching, but leave the general curriculum intact, or they can reform the whole structure of the curriculum to inspire students to address social problems (Banks, 1993).
Fourth, we will assess whether there is a focus on migration-related diversity in the school system, outside the curriculum. For this indicator, we evaluate to what extent changes are made in teaching styles, teaching staff, and other school structures to adapt to migrationrelated diversity in schools. For example, to what extent and how are teachers trained to deal with migration-related diversity in the classroom or how is migration-related diversity dealt with in the enrolment system of schools?
The first two indicators relate to how students with a migration background are addressed (or not) in the field of education, through targeting and monitoring. The latter two indicators look at how the education systems are adapted to accommodate the needs of students with a migration background, by assessing the integration of content on migration-related diversity in the curriculum and the focus on migration-related diversity in the broader school system. Together, these dimensions form a holistic framework for assessing migration-related diversity mainstreaming in the policy domain of education, encompassing both the incorporation of migration-related diversity in education policy, as well as the adaptation of educational institutions to migration-related diversity. The conceptual framework is portrayed in Fig. 1.
We will use this framework with its four indicators to assess whether and how migrationrelated diversity is mainstreamed into the educational sector in Flanders and the Netherlands. In the next section, we discuss the literature on interventionism versus retrenchment, and sub-state nationalism to formulate our expectations on how these factors might affect the relative presence or absence of migration-related diversity mainstreaming in Dutch and Flemish education policies.

Integrationism versus retrenchment
The political context in which policies are formed may be a factor in whether and how migration-related diversity is mainstreamed. Research has, for example, demonstrated the important role of party politics in explaining the (non-)adoption of multicultural policies (Westlake, 2020). Beyond left-right divides and the rise of extreme-right political parties, which undoubtedly play a role in both our cases, we want to highlight the less-studied factor of neoliberal retrenchment politics. Understood as a particular kind of social and political rationality, neoliberalism involves a logic of individual responsibility of citizens and decreased public societal responsibility for what are perceived to be 'individual problems' (see Brown, 2006). Yet, the literature on immigrant integration policies stresses a tendency towards interventionism that would be inherent to 'integrationism' (Favell, 2014) and ideas of state-led multiculturalism (Banting & Kymlicka, 2013). This interventionism reflects a strong belief in what governments can do to solve 'integration problems' . Favell (2014) argues that this interventionism is the foundation of the prevalence of 'integration policy models' in various European countries. He argues that immigrant integration policies are often constructed around a clear belief in the legitimacy of, and potential for, government intervention in the socio-economic and socio-cultural position of migrants, reflecting a strong belief in rational societal steering. Schinkel (2017) adds that the very concept of 'integration' echoes a belief that 'integration' is a problem that can be associated with migrants in particular and requires state intervention. At the same time, the literature also refers to the rise of more neoliberal approaches to immigrant integration, suggesting less government steering, less focus on groups or communities, and more on the socio-economic side of 'integration' (Favell, 2014). Informed by a neoliberal rationality, immigrant integration would be left to society and, in particular, to the market. This neoliberal approach to immigrant integration is often associated with increasing global economic liberalisation as well as the growing influence of the European single market. Kymlicka (2015), for instance, speaks of 'neoliberal multiculturalism' as a new type of multiculturalism that is no longer focused on tackling the marginalisation of minorities in society but rather on stimulating minorites to utilise their cultural capital to succeed in the market. Neoliberalism has entered the field of migration and immigrant integration through ideas of 'earned' or 'deserved' citizenship

Content integration
Diversity in the school system for migrants and minorities, including the accompanying welfare benefits (Kymlicka, 2015;van Houdt et al., 2011). In immigrant integration policymaking specifically, neoliberalism is manifest in government retrenchment, or a declining belief in what government policies can or should do, and thus a deinstitutionalisation of integration policies. The responsibility for immigrant integration shifts from the state to the individual migrant, which is also apparent in certain types of civic integration policies (van Houdt et al., 2011). In terms of mainstreaming, we would expect a clear focus and active coordination of migration-related diversity in generic policy fields under the influence of an integrationist logic. Under the influence of a neoliberal logic, to the contrary, we would expect diversity retrenchment, with a delegation of responsibilities to individuals, rather than mainstreaming. Considering that the Netherlands is shown to be marked more strongly by neoliberal retrenchment politics as well as the multiculturalism backlash (Jacobs, 2004;Kuipers, 2006), we expect less interventionism or targeted policy measures in the Netherlands than in Flanders.

Sub-state nationalism
An additional factor that might explain the existence and extent of mainstreaming of migration-related diversity is the presence or absence of a sub-state nationalist project. Scholars have demonstrated how sub-state nationalism is an important factor to take into account while explaining the development of immigrant integration policies in sub-state nations (e.g. Barker, 2010;Adam, 2018;Manatschal et al., 2020;Zuber, 2022). Immigration presents a specific challenge for sub-state nationalist movements, which can best be described by the so-called 'legitimation paradox' (Adam, 2013;Jeram et al., 2016). On the one hand, including migrants into the sub-state nation's community could reduce the cultural homogeneity that is needed to legitimise the claims for regional distinctiveness (Gagnon & Iacovino, 2007), since migrants often seem to opt for integration into the majority language group (e.g. English in Quebec, Spanish in Catalonia, French in Flanders). On the other hand, excluding migrants could discredit the nation-building processes of sub-state nations because they could then be regarded as intolerant and illiberal (Kymlicka, 2001). Accentuating the tolerant credentials of the sub-state nationalist project is necessary for sub-state actors since sub-state nationalisms have continuously been depicted as inherently 'ethnic' and less 'advanced' than the more civic conceptions of nationalism of modern nation-states (Franck, 1997;Hollinger, 1995;Ignatieff, 1993). Due to this stigma, sub-state nationalist movements have to demonstrate their liberal credentials if they want their autonomy project to appear legitimate to external actors (Barker, 2010).
Beyond the need to legitimise the sub-state nationalist project externally, sub-state nationalists also need to legitimise their claims for autonomy internally, towards their own population. They need to show that they do something with the obtained autonomy and that they do things better than the policymakers at the central state level. This need for internal legitimacy stimulates public interventionism in all policy areas. One could, therefore, argue that sub-state nationalism offers a counterbalance against neoliberal retrenchment, as the region always needs to 'do' more (and better) than the central state, and thus leads to interventionism rather than retrenchment.
Page 7 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 Besides the search for internal and external legitimacy as a driver for multiculturalist and interventionist policies, researchers have pointed at the relevance of historical narratives of past cultural oppression (Adam, 2013;Jeram & Adam, 2015). Jeram and Adam (2015) showed that Basque and Flemish sub-state nationalist actors used references to the historical cultural oppression by the Spanish and Francophone elites respectively, to delegitimise assimilationist policy choices and motivate pro-diversity policy choices. Lastly, concerned with the prestige and survival of their minority language, minority regions are generally particularly interventionist on language learning for immigrants or ethnic minorites (Hambye, 2009;Westerveen, 2020). In terms of migration-related diversity mainstreaming, we would expect more targeted policy measures and interventionism in cases where sub-state nationalism is present (Flanders) than in cases where sub-state nationalism is absent (the Netherlands).

Cases, data, and analysis
For this study, we have selected the cases of the Netherlands and Flanders, the Dutchspeaking region of Belgium. Even though the Netherlands is a unitary state and Flanders a region or community within a state, we consider these cases comparable for our study of education policies, since the Flemish Community (forming the Flemish Government together with the Flemish Region) has exclusive competence over education policy 1 and immigrant integration policies. The Netherlands and Flanders furthermore have experienced similar patterns of immigration and have developed comparable multiculturalist immigrant integration policies in the past (Jacobs, 2004;Vermeulen & Penninx, 2000). In addition, considered typical consociational democracies (Lijphart, 1969), political evolutions within both cases are similar with the notable difference that in the Netherlands one far-right party (Lijst Pim Fortuyn, LPF) briefly entered a government coalition (in 2002) and one coalition was only possible with the support of the extreme-right party Partij voor de Vrijheid (PVV) (in 2010), while in Flanders the cordon sanitaire against extreme-right party Vlaams Belang has remained in place. Moreover, the Dutch and Flemish educational systems are similarly organised. Both education systems are based on the principle of free choice of education for parents and in both cases subsidies are provided to faith-based schools (Franken & Vermeer, 2019). Consequently, the Netherlands and Flanders are characterised by a pluralist school system in which education is largely considered a public task. For this research, we focus on primary and secondary schools as they form the two major, generic educational institutions.
For the analysis, we have collected policy documents on migration-related diversity in education. Our data include all policy documents from the Dutch and Flemish governments that set out policy strategies for migration-related diversity in education between 2000 and 2014. This is an important period because migration-related diversity has become highly politicised in both the Netherlands and Flanders since the early 2000s, among others spurred by the growing success of extreme-right parties in both cases (see Coffé, 2005;Pellikaan et al., 2007). Our time frame covers several legislative sessions and is sufficiently long to sketch interesting trends regarding migration-related diversity Page 8 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 mainstreaming and retrenchment in Dutch and Flemish education policies. Since we aim to study the attention for migration-related diversity in a context of immigrant integration mainstreaming and assess the role that education policies are expected to play in this process, we only analysed documents that deal directly with education. These criteria resulted in a sample of 62 documents (32 for the Netherlands and 30 for Flanders, see Additional file 1). The content of the selected documents was analysed by the use of qualitative content analysis (Schreier, 2014). We coded the relevant extracts of the documents according to a thematic coding scheme, based on our four indicators for migration-related diversity mainstreaming (i.e. targeting, monitoring, content integration, and diversity in the school system). For each of the four indicators, we inductively explored the dominant policy approaches as apparent in the documents, focusing on the kind of policies proposed, who the policies are aimed at and their proclaimed goals. We subsequently compared the patterns in coding throughout time and across the two cases to identify trends in the mainstreaming of migration-related diversity into education policies. Thereby our analysis is focused on the official policy discourse regarding migration-related diversity in education, or 'policy on paper' . While we acknowledge that policies might be adapted during their implementation and thus might look different on the 'ground' , this focus allows us to show if and how migration-related diversity is addressed in Dutch and Flemish education policy. By tracing the framing of migration-related diversity in the official policy discourse, we reconstruct the main trends in the governmental approaches towards migration-related diversity in education policies.

Migration-related diversity mainstreaming in Dutch and Flemish education policies
Below we will discuss the findings for Dutch and Flemish education policies. This discussion is structured according to our four indicators of migration-related diversity mainstreaming in education: targeting, monitoring, content integration, and diversity in the school system. The discussion is followed by a comparison and explanation of the main policy trends in each of the cases.

Targeting
The Netherlands The Netherlands has had a long history of targeted policies for specific ethnic or cultural minorities (the so-called Minorities Policy from the 1980s). However, since the 1990s, such a group-oriented approach made place for a more neoliberal approach targeted at the socio-economic integration of individual migrants. In the early 2000s, Dutch educational policies were aimed at dissolving the 'educational disadvantage' of students in a vulnerable position, targeting disadvantaged students with and without a migration background. Migration background, language comprehension, and the broader 'educational context' at home were considered important factors that can lead to educational disadvantage of students (Staatssecretaris OC&W, 2000a, 2002. Parents with a migration background were also specifically targeted for language classes and special civic integration courses, which were coupled to educational programmes for their children (Staatssecretaris OC&W, 2000b, 25). Furthermore, Dutch schools were required to "adjust (…) to diverse nationalities and learning-and development needs of the students" (Staatssecretaris OC&W, 2000a, 4). These educational disadvantage meas-Page 9 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 ures were considered to emancipate the students and thereby indirectly foster integration. However, over time, the disadvantage policies were decoupled from migration background. This started with the abolishment of the mother tongue instruction classes ('Onderwijs in Allochtone Levende Talen') in 2003 by Christian democratic Minister of Education Maria van der Hoeven. Moreover, while previously the migration background of the student's parents played a role in determining funding for schools, this was abolished as of 2004. Thereby targeted migration-related diversity policies in education were ended and the extra funds for schools were re-allocated to a broader defined target group. From 2006 onwards a further move from targeted to generic policies is visible in Dutch education policies. In 2006, Minister van der Hoeven replaced the former disadvantage policies with a broad 'inclusive approach' , explicitly targeted at all students and linked to a broader 'shared citizenship narrative' (Minister en staatssecretaris OCW, 2006). This shift towards the abolishment of targeting was continued by Christian democratic Minister of Education Marja van Bijsterveldt, who explicitly replaced it with a generic focus on 'quality' (Minister OCW, 2011a). As of 2011, the finance structure for schools was again adjusted. Instead of targeted financing, schools now receive a 'lump sum' , meaning that the finances are not earmarked, and the definition of priorities and target groups is decentralised to the schools (Tweede Kamer der Staten-Generaal, 2011). These reforms mark a further retrenchment of central state interventions on migration-related diversity in Dutch education policy.
Flanders The Flemish policy documents include various measures that specifically target students with a migration background, albeit formally only through linguistic indicators. These targeted policies include additional Dutch language classes (i.e. on top of the normal curriculum) for those who do not speak Dutch at home (Vlaamse Overheid, 2013, 7). Moreover, regular testing is introduced to assess the individual language capabilities of students (Vlaamse Overheid, 2005a, 47-48). All these measures are aimed at avoiding language deficits and consequent unequal opportunities in education (Vlaamse Regering, 2011, 5). Good knowledge of the Dutch language is considered a prerequisite for educational and professional opportunities as well as a "condition for social self-sufficiency and integration (…)" (Vlaamse Regering, 2011, 14). Lastly, language is viewed not only as the common language of instruction but also as 'a language for a common identity' (Vlaamse Regering, 2011, 14). Related to this, is the discussion about mother tongue instruction ('Onderwijs in Eigen Taal en Cultuur') in Flemish education policies. Whereas-similar to the Dutch mother tongue classes-originally it was aimed at facilitating guest worker families' return, later on, this policy continued to exist within a more multicultural framework. However, throughout time, the number of projects as well as the number of schools that participate in these projects and the available funding had declined. In 2011, after increased contestation, social democratic Minister of Education Pascal Smet decided to cut all funding for mother tongue instruction. In the same year, Minister Smet revised the language policy for Flemish schools. The 2011 language note ('talennota') stipulates that secondary schools can offer classes in a fourth foreign language (next to French, English, and German) in any of the official languages of European Union countries as well as Brazil, Russia, India, and China. From 2014 onwards, the restriction to languages from these Page 10 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 countries was lifted, whereby schools can choose which foreign languages they want to offer in addition to French, English, and German. Against the wishes of Flemish nationalist parties N-VA and Vlaams Belang, this revision enabled schools to offer languages of important minorities in Belgium, such as Turkish and Arabic. A second type of targeted policies concerns the funding policies that apply to Flemish schools. In Flemish education policy, targeted funding is based on both socio-economic indicators and linguistic indicators. The financial resources of schools are based on a weighting system, which allocates extra teaching hours to schools on the basis of equal opportunity indicators ('gelijke onderwijskansenindicatoren'). In primary education, these indicators are the language spoken at home as well as the attribution of a school allowance and the mother's level of education (Vlaamse Overheid, 2011, 55;Vlaamse Overheid 2009, 46). In secondary education, two more indicators are added, namely temporary or permanent placement of the student outside the family and a parent belonging to the travelling population. However, there is long-term disagreement between Flemish political parties about these additional funds. Flemish nationalist party N-VA has argued for a universal basic allowance per child, while other political parties rather plead for less autonomy for schools in how they employ the additional funds.
As opposed to the targeted measures discussed above, from 2010 onwards documents also speak about an aim of 'non-targeting' every now and then. Instead of separate targeted policies for each relevant target group, the said objective is for the general Flemish education policy to become more inclusive towards all groups. This is referred to as 'explicit but no exclusive targeting' in which general policies should take the 'characteristics of minority groups into account' (Vlaams Parlement, 2011, 15). From 2010 onwards, the policy discourse thus seems to slightly move away from policies that are specifically targeted towards ethnic minorities. However, in terms of specific policies and in comparison to the Netherlands, targeted policy measures such as the weighting system have not been abolished in Flemish education policy until today.

Monitoring
The Netherlands Monitoring in Dutch education policy is primarily commissioned through the annual Education report ('onderwijsverslagen' published by the Inspectorate of Education). In addition, the annual Integration Report also includes a designated section on the educational performance of students with a migration background (alternately published by The Netherlands Institute for Social Research and Statistics Netherlands). In these reports, the performance of students with a migration background (first and second generation) is compared to the overall student performance, along with socio-economic indicators and occasionally highlighting specific ethnic groups. This reflects a broader Dutch practice of systematically compiling data on people with a migration background based on someone's country of birth (first generation) or the place of birth of at least one of the parents (second generation) and drawn from formal government registries (so not a census).
Furthermore, the biennial 'Cohort Study Primary Education' is used to determine the disadvantage of students by random sampling. The outcomes are used to recalibrate the disadvantage policies mentioned above. The Minister of Education, Culture, and Science refers to the Cohort Study Primary Education as an indicator to readjust Page 11 of 20 Westerveen et al. Comparative Migration Studies (2022) , 2006). The citizenship programme partly followed from the advice from the same Inspectorate to develop education that explicitly focuses on citizenship and societal development (Minister OCW, 2011b).
Flanders Similar to the Dutch case, the Flemish government monitors the educational performance of students based on different indicators and through different types of (international) studies. First, it collects data on the socio-economic background of students and monitors how this is related to their school career. Second, Flanders also monitors the educational performance of 'allophone' ('anderstalige') students by regularly assessing their level of Dutch (particularly those going to schools close to the 'language border' in Belgium) (e.g. Vlaamse Overheid, 2013, 66). Moreover, it differentiates, in its administrative data, between students who speak Dutch at home and students who do not. The language spoken at home is collected through a questionnaire filled in by the parents of the student. Not only the number of allophone students is monitored, but language is also frequently referred to as an important explanatory factor in the educational performance of students (Vlaamse Overheid, 2014a, 13-14). Third, data about the nationality of students are registered. This registration allows the Flemish government to monitor, amongst others, the participation of (a part of the) students with a migration background in pre-schooling and higher education.

Content integration
The Netherlands Regarding the integration of migration-related diversity in the curriculum, in 2004 the Dutch government introduced 'citizenship building' as an educational priority, assigning an explicit role to schools in the formation of 'active citizenship' (Minister OCW, 2004a). Integration and the opportunity for students of different backgrounds to meet were formulated as part of the citizenship approach (Minister OCW, 2004a. Although some (legal) parameters were set, the concrete development and implementation of the citizenship training is left primarily to the schools themselves.
In 2006, a focus on the promotion of 'Dutch and European citizenship' and the counter-acting of radicalisation are added to the goals of citizenship education (Minister en staatssecretaris OCW, 2006), alongside continued support for dialogue and anti-segregation. Furthermore, a 'societal internship' is introduced in secondary schools (Minister OCW, 2009). In a 2009 evaluation, the Inspectorate reports that, despite its early stage, many schools have made (first) efforts to develop a citizenship approach. In 2013, it is concluded that although meeting the legal minimum standards of the citizenship approach, the development has stagnated (Tweede Kamer der Staten-Generaal, 2013). Thus, while 'citizenship education' , as a potential instrument for content integration, is set up in the Netherlands, the implementation of the approach seems to lack behind.
Page 12 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 Flanders With respect to content integration, the Flemish government puts great emphasis on languages. More specifically, the promotion of multilingualism is presented as one of the objectives of Flemish education: "In an ambitious language policy, the aim is that every child, every youngster speaks Dutch excellently, and in addition [has] a sound knowledge of two or more foreign languages" (Vlaamse Overheid, 2012a, 30). However, this statement contradicts the common practice in Flemish schools of punishing pupils for using French or other languages (see Agirdag, 2017), as well as the Flemish government's strong emphasis on improving the Dutch language skills of pupils with a migration background as the solution to ethnic inequalities in education (Westerveen, 2020). Like the Netherlands, a second focus in content integration is citizenship education. In Flanders, citizenship education has been part of the cross-curricular attainment levels since 1997. Different goals are set for citizenship education, including dealing with diversity and strengthening democracy and social cohesion (Vlaamse Overheid, 2005b, 137-138). However, where and how citizenship education is integrated into the curriculum is largely left up to the school (cf. De Groof et al., 2010).
Finally, there are also efforts to adapt the curriculum to a diverse classroom and enhance respect for diversity. The 'interculturalisation of education' is considered 'a fully-fledged basis of the Flemish anti-discrimination policy' (Vlaamse Overheid, 2000, 37). However, it is not exactly clear what this interculturalisation would entail. Again, schools are given the freedom to give substance to this. The integration of cultural diversity in the Flemish curriculum might thus be limited to a policy on paper.

Diversity in the school system
The Netherlands Moving our focus from the curriculum to the broader school structures, we find that, throughout the years, policy documents mention different efforts to adapt to a diverse school environment, by inter alia adjusting teacher training to educational disadvantages in schools (Staatssecretaris OC&W, 2000a). In 2004, the Dutch ministry for education observes that the strong language focus in education "does not meet the specific learning needs of students with a migration background and native students with a weak language command" (Ministers van OCW en VI, 2004, 4). Furthermore, (all) teachers are trained to work with a diverse student population (Tweede Kamer der Staten-Generaal, 2007). In 2006, mentors and role models are deployed to support students with a migration background (Minister en staatssecretaris OCW, 2006). Later this is explicitly linked to diversifying the teaching staff by first attracting more teachers with a migration background (Full color 1) and, as a follow-up project, improving the workand learning climate at schools for teachers in training with a migration background (Full color 2) (Minister en staatssecretaris OCW, 2006, 10).
In addition, education is considered an instrument for desegregation. Up to 2004, school admission policies were oriented towards mixing students of different backgrounds to enhance immigrant integration (Minister OCW, 2004a). In 2005, dispersal policies are still mentioned as an important immigrant integration measure, but now Christian democratic Minister of Education Maria van der Hoeven says this should be decided locally and can only be guided by educational disadvantage, not ethnicity, revising her former approach (Minister OCW, 2005). Throughout the years, education and Page 13 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 anti-segregation are slowly decoupled and in 2009 the Dutch government (under the ministership of social democrat Ronald Plasterk) explicitly distances itself from anti-segregation as a priority in education policies (Tweede Kamer der Staten-Generaal, 2009).
Flanders When we zoom in on the adaptation to migration-related diversity in the Flemish education system, we detect some attention for language and migration-related diversity in teacher training. The Flemish policy emphasises the handling of linguistic diversity and, more importantly, 'teaching proper Dutch' (Vlaamse Overheid, 2007, 60). Between 2002 and 2008, documents also speak of actions aimed at 'diversifying the teacher force. ' For example, it is mentioned that 'allochthones' , people with a handicap, and older employees are underrepresented among the teaching staff (Vlaamse Overheid 2005b, 104). To arrive at a better representation of these three groups in the teacher force, 'targeted actions' are planned. However, the responsibility is, again, delegated to the schools (Vlaamse Overheid, 2005b, 104). After 2008, we no longer find mention of these actions, nor of the aim to diversify the teacher force. Lastly, just like the Dutch documents, Flemish policy documents also talk about an aim of countering segregation and arriving at a more 'balanced composition' of schools (Vlaamse Overheid, 2011, 114). This balanced composition includes a social mix (i.e. students with different socio-economic backgrounds) as well as a cultural mix (referring to 'multicultural schools'). However, language seems to be the dominant focus here: "(…) [language] (home language) rather than ethnic origin is an important item here" (Vlaamse Overheid, 2002, 33).

Comparing and explaining trends in migration-related diversity mainstreaming and retrenchment
When comparing the two cases, it becomes clear that while in the Netherlands several targeted programmes in education have been dismissed since 2003, similar programmes have stayed in place in Flanders, even though political developments in the two cases are comparable. In the Dutch case, policies targeting students with a migration background, including mother tongue instruction and targeted funding, have been renounced. Instead, Dutch education policies have moved towards a generic 'quality' approach, in which students are individually held responsible for their educational performance. This indicates a decreasing state intervention in the Netherlands regarding migration-related diversity in education policy. In contrast, and despite recurring discussions around the use of targeting, targeted policy measures such as targeted funding have not been abolished in Flanders until today. Although Flanders and the Netherlands develop different targeting strategies, remarkably they both have continuous ethnic monitoring practices in the period under research (cf. Westerveen & Adam, 2019). In terms of adapting the school curriculum and broader school system to diversity, Flanders and the Netherlands also adopt similar approaches. In both cases, citizenship education forms part of the curriculum. However, the content and form of citizenship education are mostly defined by individual schools without explicit criteria defined by the governments. It is thus hard to assess to what extent migration-related diversity is addressed in citizenship education, although, at least on paper, it seems a bit more embedded in the case of Flanders with a focus on Page 14 of 20 Westerveen et al. Comparative Migration Studies (2022) 10:31 multilingualism as well as a limited focus on interculturalisation. With regard to the broader school system, both cases show similar efforts to diversify the teaching staff and adapt teacher training to the diversification of the student populations. However, in both cases, the focus of policies is rather on immigrants and minorities' deficits and the remedying of disadvantages than on how migration-related diversity can enrich schools. Furthermore, in both Flanders and the Netherlands de-segregation and mixing are mentioned as policy objectives. The objective of de-segregation has, however, been abandoned in the Netherlands in 2009, again indicating retrenchment rather than interventionism for the Dutch case. In Flanders, the aim of mixing and countering segregation (particularly of pupils with different mother tongues) is maintained during the studied time frame (2000 to 2014). Yet, more recently, the Flemish government has announced plans to discontinue its anti-segregation measures in primary and secondary education. Table 1 summarises the comparative findings for the Netherlands and Flanders. Based on this comparison, it can be concluded that, in the period from 2000 to 2014, there is a trend towards retrenchment rather than the mainstreaming of migration-related diversity in Dutch education policies, while this trend is less prevailing in the same period in Flemish education policies. The existing literature mainly points to different degrees of politicisation of migration-related diversity due to the electoral success of extreme-right parties to explain the retrenchment of multiculturalist immigrant integration policies. Even though the cordon sanitaire has prevented the electorally successful Flemish extreme-right party Vlaams Belang from becoming a coalition partner in the Flemish government and such an instrument is not present in the Netherlands, both cases are similarly characterised by the growing popularity of extreme-right parties (Coffé, 2005;Pellikaan et al., 2007). To explain the diverging trend in the mainstreaming of migration-related diversity in the Netherlands and Flanders, additional factors, therefore, need to be added to the explanatory cocktail.
First, the varying degree of neoliberal retrenchment politics in the Netherlands and Flanders (Kuipers, 2006) can help explain why the Dutch government has abandoned many of its targeted policy measures in education, while these continue to exist in Flemish education policy. A neoliberal rationality would inform the Dutch government to retreat from redistributive policies targeting migrants and minorities. It shifts the focus away from state intervention and towards the individual responsibility of citizens to integrate and participate in society. This is reflected in the generic (non-targeted) focus on 'qualitative' education in which it is each student's individual responsibility to perform well in school. The present but more limited influence of neoliberalisation in Flanders can explain that this shift in thinking has not yet resulted in significant policy reforms with regard to migration-related diversity in Flemish education. Second, the presence of sub-state nationalism in Flanders also helps to explain why the retrenchment of migration-related diversity policies is more limited in Flanders than in a central nation-state, such as the Netherlands. Flanders is a prototypical minority nation. 2 Research has shown that such minority nations use education as well as immigrant integration policies to further their sub-state nationalist projects (e.g. Keating, 2005). In particular, the boundary drawing activities of the nationalist project between 'us' and 'them' inspire interventionism on the cultural dimension of immigrant integration (Adam, 2013: 17). Moreover, as explained above, internal legitimacy seeking of the sub-state nation promotes policy interventionism, while retrenchment possibly delegitimises the claim for more autonomy, or the defense of obtained autonomy. The search for internal legitimacy of the Flemish government thus favours state intervention with regard to migration-related diversity in education policies, rather than migration-related diversity retrenchment. In addition, the Flemish nationalist struggle for the continued relevance of Dutch, in a country in which many immigrants are tempted to use or learn French, informs interventionist policies on langue learning. In the policy documents of the Flemish government on education studied here, we observed an overwhelming focus on the importance of the Dutch language in Flemish education. Whereas in the Netherlands retrenchment got the upper hand, replacing former targeted language policies in education by generic policies about the 'quality of language' and a focus on the student's individual responsibility, in Flemish education targeted language policies were maintained, reflecting more active state intervention on the Dutch language capabilities of students.
Furthermore, the search for external legitimacy of sub-state nations as discussed above might be even more relevant for Flanders than for other sub-state nations (like Scotland, Catalonia, and Quebec), as Flemish nationalism is still associated with past Nazi collaboration (Wauters, 2005). The Flemish nationalist project became increasingly (re-) associated with racism at the moment it had finally obtained more autonomy. During the 1990s, when Flanders inaugurated its own parliament, the electoral results of the extreme-right and sub-state nationalist party Vlaams Blok (now Vlaams Belang) had been on the rise for several years. The following Flemish governments acknowledged that the success of this party strenghtened the perception of an intolerant Flanders outside it borders and felt the pressure to uphold an inclusive version of Flemish nationalism. 3 The electoral success of Vlaams Blok also pushed the original Flemish nationalist party Volksunie, out of which Vlaams Blok originated, to become ever more pro-migrant Westerveen et al. Comparative Migration Studies (2022) 10:31 as to 'imagine' the Flemish nation differently (Adam & Deschouwer, 2016). This external legitimacy seeking reflex in Flanders can help explain the stronger mainstreaming of migration-related diversity in Flemish education policies compared to the Dutch case.

Conclusion and discussion
In this article, we analysed and compared the mainstreaming of migration-related diversity into Dutch and Flemish education policies between 2000 and 2014. We contribute to the existing literature in migration studies by providing a case study of the adaptation of generic policies, here education, to the increasing diversification of European societies. We proposed an innovative conceptual framework with four indicators to assess the mainstreaming of migration-related diversity in education policies, namely: targeting, monitoring, content integration, and the inclusion of diversity in the school system. Although the Netherlands and Flanders have traditionally been viewed as similar cases when it comes to their immigrant integration policies, our analysis reveals diverging policy trends regarding the mainstreaming or retrenchment of migration-related diversity in education. Except for ethnic monitoring, the focus on migration-related diversity has largely faded in Dutch education policies, thus leading to migration-related diversity retrenchment. By contrast, in Flanders, the approach to migration-related diversity has remained relatively stable over the period 2000 to 2014. This confirms earlier research that looked at the multiculturalism backlash, which has argued that formerly multiculturalist policies have largely been abolished in the Netherlands, while they have stayed in place in Flanders (Banting & Kymlicka, 2020). These diverging trends in the Netherlands and Flanders cannot be explained by the increasing politicisation of migration-related diversity, which is similar in both countries. Instead, we have shown that diverging degrees of neoliberal retrenchment, as well as the presence of sub-state nationalism in Flanders, can yield insights into the diverging trends regarding migration-related diversity mainstreaming or retrenchment in Dutch and Flemish education policies. In the case of the Netherlands, this led to decreasing government interference on migrationrelated diversity in education, while in Flanders the combination of these two factors led to continued government intervention on migration-related diversity in education. Yet, what must be noted is that our analysis only covers policy developments up to 2014. More recent developments, such as the fading focus on the diversification of the teaching staff, the 2018 reform of the Flemish enrolment decree (which abolished antisegregation measures) as well as the growing contestation of the weighting system, indicate that Flanders might be moving towards a similar trend of migration-related diversity retrenchment in education as observed for the Netherlands. In addition, a recent reform of the Flemish civic integration policy also point in the direction of neoliberal retrenchment politics. Nevertheless, Flanders seems to move in this direction at a different pace than the Netherlands, as might be explained by the interplay between sub-state nationalism and neoliberalism. While sub-state nationalism might not completely hold back neoliberal diversity retrenchment, it does seem to slow down the process. Because of the conservative and exclusionary stigma attached to sub-state nationalism and Flemish nationalism particularly, Flemish sub-state nationalist actors always had to legitimate themselves by showing their liberal and tolerant credentials. However, with the current normalisation of the extreme-right within European nation-states (Wodak, 2019),