Implementation of Educational Policies for Migrants: A Case Study in the Swiss Canton of Geneva

The Swiss context presents a specific pattern being a confederal country relying on a subsidiarity principle. Thus, the cantons operate in a framework of a reduced power of central authority (Boulenger et al., 2012; Revaz, 2020) and are autonomous regarding education policies at local level (Akkari, 2019). However, since the adoption of a new law in 2005, the Federal Council, cantons, municipalities and cities officially collaborate on migration policy (Chifelle, 2018; Facchinetti, 2012). In this particular context, the confederation developed migration policies giving guidance to cantonal governments. Our research aimed at analyzing a confederal policy guidance on 1 We would like to extend our warmest thanks to Karelle Stiassny (HEP Vaud Switzerland), who has contributed greatly to our thinking. Education Policy Analysis Archives Vol. 29 No. 69 Special Issue 2 migrants’ education and its interpretation in the canton of Geneva. In particular, we examined the ”decoupling” between general guidance from the Confederation and the implementation at local level. For this purpose, we studied the implementation of a cantonal program named L’école des mamans (mothers’ school) dedicated to prepare migrants’ families for their children enrolment in primary school. Our main results show that there is a distortion between policy objectives and the implementation phase. We also observe a model of resistance to change with no modification of actors’ practices towards migrants’ parents despite the new integration policy guidance.


Implementation of Educational Policies for Migrants: A Case Study in the Swiss Canton of Geneva
Contemporary societies have experienced large movements of people. The statistics speak for themselves: it is estimated that there are around 250 million migrants around the world (Wihtol de Wenden, 2018). Switzerland is no exception. Foreign nationals make up more than 20% of the Swiss population, although they are not evenly spread across the country. Swiss-born foreign nationals account for 37.5% of the foreign population. There is more migration to Geneva canton than to other cantons. Foreign nationals make up 40% of the local population there, with Swissborn foreign nationals accounting for 60% of all foreign nationals. It should be noted that these rates cannot easily be compared across countries, as they are influenced by citizenship laws, which vary from one country to the next. In Switzerland, citizenship is determined primarily based on jus sanguinis. This means that a second or third generation young person born in Switzerland from two foreign parents is considered a "foreign national born in Switzerland." This means that people can remain foreigners for several generations (even if a recent vote has simplified access to Swiss nationality for the third generation). Therefore, "foreign nationals born in Switzerland" are not generally considered as migrants. The latter have experienced a migratory event in their recent biography (the term "newly arrived migrant" refers to anyone who arrived in the country within the last five years).
The migrant population in Switzerland, like in the rest of the world, refers to a variety of migrant profiles whose classical categorizations are not sufficient to describe the diversity of situations (Wihtol de Wenden, 2010). For example, at the legal level, migrants are subject to categorizations assigned by states that grant them a status but that do not necessarily reflect the reality of their migratory journey (Barbou des Places, 2010). This also reveals an asymmetry as soon as they obtain a migratory status. The social dimension -which has similar effects-will be more strongly expressed in integration policies, as we shall demonstrate.
Switzerland also stands out because of the country's federal system. The federal government has very limited power (Boulenger et al., 2012;Revaz, 2020), and the cantons are autonomous when it comes to a large number of public services, including educational policies (Akkari, 2019). In terms of migration-related policies, a new law was adopted in 2005 requiring the federal government, cantonal governments and municipalities to work together (Chifelle, 2018;Facchinetti, 2012). As a result, the federal government draws up guidelines for migration policies and the cantonal governments are tasked with implementing them.
Our study seeks to analyze the federal government's policy guidelines concerning the education of migrants and how those guidelines are interpreted in Geneva canton. We studied a cantonal initiative aimed at helping migrant families prepare their children for primary school. The initiative is known as L'école des mamans in French, which can be translated as the "school for moms." More specifically, the project is aimed at mothers of allophone families who have recently arrived in Switzerland. As we will see, most of them are in situations of economic precariousness in Geneva, which does not necessarily reflect their economic and social situation in their country of origin.
The project has two main objectives: to support mothers by giving them French classes and information on the Geneva school system in order to create a stronger link between migrant families and schools, and to help integrate children into the school system through pre-school activities. From our analysis, we will show that the way in which the institution conceives and defines what is (or should be) the integration of migrant populations is based on a classic vision of integrationwhich primarily involves the acquisition of the canton's language towards an economic integration-.Through our analysis of how this initiative is run, we will question the concept of decoupling and how it applies in Switzerland with regard to school-related migrant policies. We will analyze the reasons for this decoupling by focusing on educational organizations' actual practices. Finally, our article provides an analysis on three levels: the macro level (i.e., federalism), the meso level (i.e., the cantonal and municipal levels, and the canton's Department for Public Education 2 ) and the micro level (i.e., how local groups interpret policies and put them into practice) and to think of the way local actors interprete federal legislation in the field of migration and integration.

Migration-related Policies in a Federal Context
In the early post-war period, there was strong demand in Switzerland for foreign laborers, whose legal status was usually that of seasonal workers. One of the aims of the country's migrationrelated policy was to maintain migrants' temporary status as seasonal workers to ensure that they would then return to their home countries. Starting in the mid-1960s, a rise in xenophobia had an impact on Switzerland's migration policy, and there were successive attempts to limit the number of foreign nationals allowed to enter Switzerland for work (Piguet, 2017). During the oil crisis, demand for foreign workers dropped in Switzerland, like elsewhere in the world, before picking up again in the period from 1985 to 1992, which Piguet (2017) describes as the country's second wave of immigration. It was also during this period that integration began to emerge as a social issue. The 'three circles' model, which the federal government began using in the 1990s, became part and parcel of subsequent migration policies. This model differentiated between migrants considered capable of assimilating and those whose 'cultural distance' was too great.
Switzerland's evolving migration policies over the post-war period show that there have been hesitations and changes in policies and legislation. They also show the influence that social representations of foreign nationals have had on these policies (Meuwly, 2018). Like in any federal system, legislative powers in Switzerland are spread across different decision-making levels, from the federal government to the cantons and then the municipalities, which represent the smallest decision-making unit. Switzerland is therefore extremely decentralized, and the principle of subsidiarity means that it is the smallest possible unit that takes action on the ground (Meuwly, 2018). The federal government has authority when it comes to foreign policy, security, customs and the currency, as well as federal legislation and defense. All other powers lie with the cantons or the municipalities. However, in 2005, the federal government adopted the Swiss Federal Foreign Nationals and Integration Act (FNIA), which took effect in 2008. The FNIA sets out the conditions under which foreign nationals can enter and stay in Switzerland. But it also seeks to define what integration means and how it can be achieved.
The FNIA sets out four criteria for integration: respect for public safety, security and order; respect for the values of the Swiss federal constitution; language skills (this is at the core of local integration measures); and participation in the country's economy life. Geneva's cantonal integration program (PIC II), which is the canton's current mechanism for implementing the FNIA, is managed by the cantonal councilor in charge of the Cantonal Department for Security and the Economy (and not the councilor responsible for social cohesion, 3 for example). In the FNIA, integration is defined essentially based on the extent to which foreign nationals residing lawfully in Switzerland participate in working life. It focuses on each foreign national's individual responsibility to integrate, and affirms the need for openness in their regard on the part of the Swiss population (Art. 4 of the FNIA). Finally, command of the canton's language 4 is seen as essential to foreign nationals' integration. As we will see, these various aspects are reflected at the cantonal level. The primary aim of the FNIA is to set out the responsibilities at each level of government and ensure at least a minimum of cooperation between municipalities, cantons and the federal government. Lastly, it defines which groups the integration mechanisms are aimed at, and more specifically the priority group, which comprises migrant women, young people and children. The 2018 Swiss Federal Ordinance on the Integration of Foreign Nationals adds that integration mechanisms are aimed at individuals with integration needs (Art. 8 of the FNIA) but does not specify the criteria for determining those needs.
Cantonal integration programs are the local mechanisms for implementing the FNIA and set out more details about how integration is to be achieved. In Geneva, PIC I (2014-2017) heavily (and necessarily) reflected that legal framework. At the cantonal level, the focus was on integration as a way of ensuring social cohesion. But, at the federal level, responsibility is placed mainly on the individual. PIC 1 stipulated that: "Anyone residing in Switzerland must comply with the law and respect public order, aspire to be financially independent and respect the cultural diversity of the country and its inhabitants. That is to be achieved by being involved in Swiss society and respecting all members of that society. Anyone who does not follow this fundamental principle or who deliberately hampers integration efforts will face punitive measures 5 " (PIC I, 2014, p. 2). The subsequent program (PIC II), which has been in place since 2018, more explicitly sets out how the federal directives will be implemented at the cantonal level. The language objective at the federal level, for example, is to ensure migrants can master one of the national languages enough to communicate on a daily basis and in a way that is adapted to their work situation. The related cantonal measure involves conducting an assessment of foreign nationals' needs in terms of learning 'French for integration purposes.' PIC II has mechanisms that seek to achieve three objectives: to provide key information; to ensure acquisition of the cantonal language and provide access to training; and to include the children of migrant families in daycare facilities with a view to helping them socialize and reducing inequalities.
An initial analysis of the policies governing integration mechanisms shows that they are coherent in terms of the view of integration (i.e., individual responsibility), the related values (i.e., following the rules and social cohesion) and the fundamental importance attached to knowing the cantonal language and contributing to the country's economy. The initiative that we will study, the "école des mamans," is theoretically fully aligned with that vision.
The federal and cantonal laws are largely consistent, yet we will see that local mechanisms do not always reflect the spirit of the law as the French class (one of the main components of the studied initiative) for example that are beyond the federal objectives. We will use theories on the sociology of public policy to account for this disconnect.

Theories on the Sociology of Public Policy
There is a question that researchers have been trying to answer since the 1970s: How can we explain the gap between public policy design and implementation? According to Hassenteufel (2014, p. 98; following the work of Pressman & Wildavsky, 1973), one of the key reasons for this inconsistency is that policy has to be implemented at multiple levels. Research conducted into federal institutions in the United States has shown that the greater the distance between the decision-making authority and the authority tasked with implementing policy, the greater the gap between policy design and implementation. The study by Pressman and Wildavsky (1973) looked at how a federal subsidy policy to get minority communities back to work was not implemented despite strong initial consensus. They observed that the gap between decision-making and implementation was primarily due to the multiple hierarchical levels between the original policymakers and those tasked with implementation. More specifically, they reported that it is the multiplicity of interactions that widens the gap even when those involved agree with the key objectives of the initial policy decision.
The study by Falkner, Treib, Harlapp and Leiber (2005), which looks at how European Union directives are incorporated into national law and implemented at the national level, shows that there is a gap between the directives and their implementation in various European Union countries. That disconnect can be explained by the systems of representation in each Member State, and whether or not they are pro-European Union.
The concepts of public policy coupling and decoupling were established by Weick in the 1970s in order to analyze the linkages between the different hierarchical levels of educational organizations. The concepts are a useful way of analyzing the functional role of decoupling, but do not take a normative approach (i.e., they do not state whether coupling and decoupling are good or bad). According to Lessard and Carpentier (2015), educational organizations do not have to react to every change in their environment. In fact, decoupling or loose coupling enables those involved in education to adjust and maintain the autonomy they need in their work and also gives room for them to resist certain policies. In a decentralized federal system, like the one in Switzerland, the issue of coupling and decoupling is extremely relevant. Local stakeholders are used to making decisions locally, so how do they react to, interpret and manage federal requirements? Lessard and Carpentier (2015) highlight another point: There tends to be greater decoupling when several frameworks are juxtaposed. In our case study, the federal requirements set out in the FNIA raise issues concerning: migration policies, which have traditionally been the responsibility of the federal government; integration, which is usually the purview of social workers; and educational issues, which are cantonal prerogatives. The fact that these three groups of issues -migration, social work and education -are put in competition could be an initial explanation for the decoupling.
Once we have established that decoupling can and does occur in educational policies, we still have to look at the gap that emerges between public policy design and implementation and try to understand the reasons and scale of the gap and any variations in it.
The content and even the formulation of the policy could also explain the gap -or the decoupling -since the implementers on the ground may not agree with the content or consider it illegitimate. Researchers who have looked at this issue have observed that the same policy can be implemented in different ways depending on the groups involved. In his work on street-level bureaucracy, Lipsky (1980) highlighted the role of civil servants on the ground, who have some degree of discretionary power and can therefore have a direct impact on the beneficiaries of the public policy: "The actions of these agents can contribute to reorienting the public action itself by modifying its aims 6 " (Hassenteufel, 2014, p. 106). Lipsky therefore considers them to be policymakers too. A study by Spire (2007) into the administrative management of requests for asylum in France came to the same conclusions. Spire observed the civil servants working at the asylum application counter within a prefecture. He demonstrated that the lack of official guidance allowed the long-serving officers to impose case management practices on newer arrivals. While he identified several types of behavior among these officers, their practices were in fine similar: "Depending on their social background, their seniority within the service and their convictions, they may adopt different accommodating tactics, but as they need to be united in order to make the work less arduous, they standardize their practices and neutralize any attempt at resistance 7 " (Spire, 2007, p. 21). In this situation, the agents do have discretionary power; the legal criteria set by the government are deliberately vague in order to meet a dual political requirement, that of ensuring respect for human rights while also controlling migration.
This theoretical framework helps us understand that the gaps and decoupling are not necessarily the result of ineffective organization but that they are also the result of individual actions, autonomy and a negotiation of the legitimacy of policy decisions. The "3I" approach complements this conclusion. This theory, developed by Palier and Surel (2005), brings together contributions from neo-institutionalist and cognitive theories. The researchers consider that public policies cover three analytical areas: ideas, interests and institutions. The ideas are "understood to be cognitive frameworks that materialize in the reflections, discourse and behaviors of those involved" (Revaz, 2020, p.32). Looking at the interests involved in public policies allows us to analyze the actions of educational stakeholders based on the cost-benefit that it represents for them. Interests can be both personal and collective. Finally, the institutional aspect takes into account the fact that actions take place in specific institutional settings that are governed by codes, behaviors and habits. These can impact people's social representations, since the three areas are interdependent. There can be dissonance between these three areas, and that can also help to explain the gap between policy design and implementation.
Returning more specifically to the migration policy defined in the first part, we might find another possible interpretation of this gap. The divergent conceptions of what integration is can be an explanation. Until 2005 and the application of the LEI, federal policy was a migrant admission policy (mainly around the mechanisms of quotas and the right of entry to the territory). It then becomes an integration policy. We enter a new era in which the question of integration of foreigners is conceptualized at the federal level. This vision gives priority to the acquisition of one of the three national languages which should allow for economic integration. The Swiss context is peculiar as it has a unique relationship to its three national languages with nearly 63% German speakers, less than a quarter French speakers and 8% Italian speakers. According to J. Robin (2005), the founding myth of Switzerland is that of the linguistic and cultural plurality which is the basis of national unity and identity. However, behind this idea of unity may hide strong oppositions between linguistic units, which do not always overlap with the territorial units that are the cantons (Robin, 2015). Thus, the requirement for newcomers to master the language as a prerequisite for social and professional integration must be understood in this singular conception of integration. This vision reflects a "classic" conception of integration processes according to which the processes of integration have three main characteristics. It is an individual (even psychological) process. It is a natural, unconscious and linear process and as such inevitable. And above all it implies a convergence of the ways of being migrant populations and of doing migrant populations towards the mainstream (for a synthesis of the work of the Chicago school see for example Safi, 2011). According to legislators, language therefore appears as the main criterion for integration. Segmented assimilation theory, however, allows us to understand how these processes are complex and less linear than the classical view suggests. This theory comes from the United States and is also relevant in the Swiss context. Yet language is only one element among others participating in the integration process. Felouzis, Charmillot and Fouquet-Chauprade (2016) have shown that second-generation pupils in Switzerland are part of a process of inferiorizing assimilation. Wanner et al. (2002) define four modes of integration: structural integration, formation of ethnic minorities, marginalization and assimilation, which are quite close to the three typical ideal processes described by segmented assimilation.
In our research, we will see that the way in which integration is conceived varies between the conceptions of the confederation and those of the local actors.

Public Policies on Integration
Federal provisions are therefore translated into cantonal texts which must in turn entail a concrete implementation through mechanisms that cover all the aims set out in the texts. The Office for the Integration of Foreigners of the Canton of Geneva is in charge of developing and financing these systems (several dozens). Some are aimed at early childhood structures, others are developing assistance for professional integration, and still others relate to the health of migrants, for example.
The ICP II (following the sub mentioned ICP I for the period 2018-2021) states several cantonal objectives including French classes for migrants and early childhood services to migrant families to ensure equal opportunities. One of the measures encourages activities to develop a coeducation and notably to strengthen parental skills at a crucial transition (i.e. first years of schooling). The text always refers to « parents » or « families » but never specifically targets "mothers." The ICP II also dedicates a section to the training of civil servants to fight against racism and discrimination.
The "école des mamans" project is part of two new trends in educational policies. The first one refers to policies of "parent support programs." The second one refers to the work to promote cooperation between schools and families. On the first one, parenting support policies have become categories of public action (Chauvière, 2008). Research on these policies shows that they are not very conducive to the development of reciprocal relations and in fact present an asymmetry of power between parents and professionals. They "presuppose that (it is the) (parental) skills that pose a problem above all other thing " (Garcia, 2014, p.238). In this regard, Giuliani (2009) noted that the use of "parent" instead of that of "family" -until then consecrated-shows a reversal of logic emphasizing the capabilities of parents in opposition to a "deficit view" of the family. It is therefore quite revealing that the program studied is only intended for mothers of families (and not fathers), contrary to parenting policies which "tend to neutralize relations between the sexes" ( Cardi, 2007, p.27). And on the other hand it puts forward the term "mother" emphasizing not the status of parent of these women but accentuating the non-autonomous aspect and incapacity of these mothers . Isn't the project then that of transforming these potential "bad mothers" into "good mothers (Cardi, 2007) against a background of culturalist prejudices and stereotypes?
Parents were for a long time kept at arm's length by schools, but from the 1990s onwards parents have become key partners for schools, especially in addressing academic underperformance (Giuliani, 2019). Payet (2017) refers to a double partnership with parents, one that is both institutional and pupil-centered. The institutional partnership is based on parents' rights to participate in school governing bodies, while the pupil-centered partnership focuses on pupils' academic education in particular (Payet, 2017;Périer, 2012).
For migrant families, the partnership tends to focus on the pupil. Perregaux (2006Perregaux ( , 2008Perregaux ( , 2010 looked at the specific role that the schooling of the eldest child plays in the cultural integration of families and their relationship with the school as an institution. The organization of the family changes when family norms are challenged by institutional norms. Parents often feel invalidated relative to their child in education and say that they would like to learn the institutional norms and be recognized as partners in their own right (Montandon & Perrenoud, 1987). In a study carried out in Switzerland, Ogay (2017) demonstrates that migrant parents have to interpret both the school's physical and symbolic boundaries in order to understand their place with regard to the school. The study shed light on the asymmetric relationship between teachers and parents, which is directly linked to the status of teachers, who are considered experts, and that of parents, who are considered to be executors of teaching requirements. As such, there is no reciprocity in the family-school relationship: studies have shown that teachers have high expectations of parents based on deeply rooted norms of what a good pupil or a good parent is. Despite this, no -or very littleinformation is shared with parents concerning the practices and codes of the school (Delay, 2013;Payet, 2017;Périer, 2012). The family structure must therefore adapt to the school and to the process of learning how to be pupils and parents of pupils.
The "école des mamans" program was created in 2009 in a school in the center of Geneva after school actors noticed that the ties between families and the school were weak. During initial parentteacher meetings, the teacher and the school's principal noticed that many of the mothers were not native French speakers, were often alone in Geneva and came from disadvantaged backgrounds. They also observed that it was these types of families that rarely came into the school, did not attend meetings and did not fill out school documents. The "école des mamans" project was therefore created in response to these two issues. The French classes are free, but from the outset, the project founders set up a system in which the mothers would give something in return. Rather than making use of the mothers' professional or other skills, the entire system was based exclusively on the mothers preparing meals for teachers and institutional actors. Another important part of the project, which seems now to have been set aside, was that the project founders realized that the school tended to have negative perceptions of these families, and the teachers' interpretations and judgments of certain behaviors were not conducive to a successful partnership. The project founders therefore intended to provide teacher training in ethnopsychiatry. This school of thought is an approach combining an anthropological approach and psychoanalysis. We will develop it later in this article.
The "école des mamans" program was initiated in this one pilot school, and then other écoles des mamans were created using the first school as an experimental model, as well as a set of specifications, which we analyzed for our study. This overview of the program for migrant mothers (or more specifically non-French-speaking mothers from disadvantaged neighborhoods) brings to light an important point: the project founders' initial aim was to improve the link between the school and the families by working with the mothers and with teachers to change their perceptions and lay the foundations for reciprocal and balanced communication. At the beginning, the project was therefore deeply rooted in the school, in order to facilitate the integration of these families into a school system whose codes they were not familiar with. From the funding partner's point of view, it is clear that the program was fully aligned with the canton's integration program, as it achieved the objective of language acquisition (French in this case) and provided daycare for the children not yet at school.
Concretely, the program is run through small structures within some of the primary schools in Geneva canton 8 . It is aimed at mothers of migrant families with children who are at school or who will soon start school. School's principal or teachers themselves usually create the écoles des mamans, however they have to be run by associations in order to receive funding from the cantonal integration office 9 and the City of Geneva. The cantonal integration office is in charge of implementing integration policies at the cantonal level by translating federal guidelines into local mechanisms. At the time of our study, there were four écoles des mamans. Based on the framework put forward by the Department for Public Education, 10 which was the project's initial funding partner, the écoles des mamans have two main objectives. The first one aims at helping non-French-speaking mothers from disadvantaged backgrounds to more easily follow their children's education by giving them a better understanding of the Geneva canton's school system. The second one aims at helping them to learn French in order to integrate into social life in the canton. The project aims to strengthen the link between school and families by bridging the gap between the mothers and their children's education and to provide activities that those mothers can participate in to reduce their isolation and make use of their skills. The project includes French-as-a-foreign-language classes, with between one and three classes a week during school time. A daycare facility is available during the French classes for mothers with smaller children.
By analyzing this program, our aim in this article is to answer the following questions: How is a federal policy to promote the integration of foreign nationals implemented at the local level? If policy texts are drawn up at the macro and meso levels, will this prevent decoupling with the practices on the ground, as is often observed with public policy? Or is there still decoupling at the implementation level? And if there is, how can it be explained?

Methodology
This study was conducted using a mixed qualitative approach involving the analysis of legislative texts, semi-structured interviews and observations. The various officials texts studied are listed in the table below. They represent different levels of decision: cantonal government, cantonal integration office, local associations.
The methodology has been built to assess the compliance of the initiative l'école des mamans with the official texts of the cantonal office of integration on a double perspective. On one hand, it analyzes the compliance of the project with the official texts on a theoretical ground and on the other hand, it analyzes the project's implementation by local actors.
After an analysis of the official texts, interviews and observations, data have been analyzed through a thematic grid according to three main dimensions: project coordination, vision of the project and strategical objectives. Gathering and analyzing data in the texts and on the field has allowed us to put in perspective public policies at macro and micro levels.
At the time of the research, the initiative was implemented in four different schools in the canton of Geneva. The interviews and observations have been carried out along a period of four months in the four schools. As stated before, every école des mamans is part of one primary school 8 Schools with high rates of migrants population. 9 The Office for the Integration of Foreign Nationals, set up under the cantonal integration program, works to promote social cohesion and the integration of foreign nationals based on the values of the Geneva and the Swiss federal constitutions, through projects and initiatives conducted in partnership with local associations and organizations. 10 Geneva canton's Department for Public Education, Training and Young People ran the program until the cantonal integration office became the project's new funding partner. with a population characterized by low socio-economic background families. The same categories of actors have been interviewed in every schools: institutional coordinators, école des mamans project coordinators, schools' principals, school's teacher, French teachers dedicated to the initiative, daycare managers dedicated to the initiative and mothers. During the research, 38 interviews were carried out including four focus group with mothers (29 mothers) for a total of 68 persons interviewed. The observations have been carried out in five French classes as well as five sessions of daycare during the French classes, including at least one in every schools.

Table 1
Legislative texts analysed The interviews and observations were conducted based on the central themes of the program. We chose to use, as the basis for our work, the themes set out in the project's initial specifications, which the Department for Public Education submitted to the associations setting up one école des mamans. There are two reasons for using the specifications as the framework and common thread throughout the interviews and observations. First, the specifications were drawn up and used at the start of the project and therefore provide information on how the project evolved relative to its initial goals and design and its current implementation. Second, they cover all aspects of the program, both in terms of project management (e.g., identifying needs, association support, budget, monitoring and evaluation) and in terms of the program's vision and strategic goals (e.g., target group, goals, services in return and networking). In addition to these themes, we looked at the origin, sense, extent and evolution of the project and its relevance. One grid was developed for the interviews and another for the focus groups involving mothers taking part in the program.

Total: 38 interviews and 68 interviewees
Two observation grids (one for the French classes and one for the daycare facilities) were drawn up to ensure that the observation criteria were the same for the different places visited.

Table 3
Description of the Observations

French classes
Five teachers Five sessions

Participated in three meals
We identified the people to interview based on their involvement at either the institutional or operational level. The institutional level refers to those working either within the municipal authorities or the funding partner. The municipal authorities are responsible for ensuring that the cantonal integration policy is implemented, while the funding partner is the project's sponsor. The operational level includes those involved on the ground (i.e., principals, teachers, association workers and the mothers themselves). Observations took place during the French classes, with one observation per teacher. In the same way, the daycare facilities were visited while the French classes were being observed. Taking part in the meals provided direct and participatory observations. The observation has been carried out during three different meals in three different schools (the last school did not plan a meal during our observation period). We participated in these meals like any other participants such as the schools' principals and schools' teachers for example. Thus, the analytical grid was created based on three themes that were identified as relevant: project management, the project's strategic goals, and the project's impact.

A Theory-grounded Project in line with Federal Policy Guidelines but Decoupled while Implemented
The project's main aim is to enable mothers to learn more about the Geneva school system in order to strengthen the ties between schools and families (in this case, mainly through the mothers). This therefore goes beyond the primary objective set out in ICP II, which prescribes the development of "French courses for communication and integration." The literature shows the need to develop postures and projects to remove the implicit school rules, a condition for effective collaboration between parents and education professionals. Périer (2012) thus explains: "reaching the status of partner involves understanding and appropriation of the 'user manual' and rules of interaction" 11 (p. 87). Passing these rules on would appear to be a prerequisite for any balanced collaboration between schools and families.
Lahire (1995) also shows that academic success is linked to the level of harmony that exists between school and family life, in other words whether there is dissonance or consonance in "the forms of social relations in interdependent networks." 12 The "école des mamans" program, which enhances the mothers' understanding of the system and the people involved, helps to build the foundations for a partnership between schools and families.
The école des mamans could be part of this movement to seek symmetry and equality in information. Yet we will see that its implementation moves it away from a balanced partnership. We observed a change between the project's initial vision (as per the specifications) and the funding partner's vision (as per the cantonal integration office's guidelines and PIC II) and variations in how the project was implemented in different écoles des mamans (based on the articles of incorporation and reports drawn up by the associations). The initial specifications explain why there might be this disparity in the documents gathered from different écoles des mamans: There was room for the implementers at different levels to interpret the project requirements. There were also varying perceptions of the project's goals: some considered the focus to be on the mothers, while others thought the focus should be on the mothers and the children.

Were the Project Goals in line with the Policy Guidelines and with Local Practices?
All of the associations studied used the pilot school as their model when creating their écoles des mamans. The project founders' initial aim had been twofold: to promote links between migrant families and the school through the mothers by improving communication (i.e., through French classes, information about the Geneva school system, and social integration within the neighborhood) and to provide teachers with training to help them change their representations of migrant families. As we have said, the project founders had noticed that teachers often had a prejudiced vision of migrant families. That was why training in ethnopsychiatry was initially offered to the teachers. The aim of that training was to help teachers understand the cultural issues at stake and reduce prejudices. The teachers who had taken the training felt that it had had an impact on their way of thinking about the parental involvement of these families, as it had questioned their narrow vision of those families. However, we can mention here the risks of such training with educational actors and social and associative workers. The possible exacerbation of cultural differences -to explain the behavior of these families-entails a risk of essentialization and reification of the culture and overdetermination of the cultural dimension (Fassin, 2000). The objective of such an action lies on the contrary in a need to deconstruct stereotypes and prejudices attached to the real or supposed culture of these families. The current systems no longer offer any training for the professionals involved, nor for the teachers of the school in which the system is located. This shows that this project, initially designed in close interaction with the schools, has changed in nature. Even though the french classes are physically located in primary schools, they have in fact only a relatively weak link with the establishments, in certain schools the French lessons are the occasion to read the official documents of the school or to answer at the teacher's request for a meeting, some mothers are sometimes invited to go and read poems in class.
Yet, during our interviews, some people from the schools still spoke of migrant families using derogatory language: the beneficiaries didn't make enough effort to learn the language and venture out of their community even though they were responsible for their own integration, and institutional stakeholders should require more from them. 13 During interviews, most of the mothers assert they come to the program to learn French. When asked about their project after l'école des mamans (that lasts only 2 years), they all answer that they want to continue learning French to be more autonomous in their daily life and to be able to help their children with school homeworks. They emphasize the fact that the group level in French was heterogeneous and that most advanced students had to learn the basics over and over with no opportunity to further improve their level. They also stress they found it unacceptable for women not to come on a regular basis: Katryn 14 from Kosovo told us "either you take it seriously or you do not come. Other women are waiting to be part of the program." All of them told us that for them it was more than a French class. It was an opportunity to develop social relationships especially for those newly arrived in Switzerland or with no family or friends from the same community. This result shows that women are aware of the issues at stakes for their own integration (especially at the social level) and that actors oversimplify the reality by reducing their vision of women migrants through the scope of their own norms and prejudices.
Other actors think that the program should not be open to fathers, using gender prejudices as their argument: fathers are better integrated in society and already speak French, and they wouldn't let their wives take French classes if they were mixed 15 . It should be noted that the focus groups with the mothers involved in the program only very partially corroborated these hypotheses. As a matter of fact, we have observed two different discourses. The first one gives the same conclusion. Women told us they feel more comfortable to be in class with women only. For example, Fatoumata from Burkina Faso said "my husband would not like me to be in class with other men" or Salma from Morocco, "if our husbands are there we do not want to talk" and finally Nuray from Turkey says: "my husband has a tough mentality." However, other women of the program wonder why men are not part of the project as school should be a shared preoccupation for both parents. For example, Sara from Togo told us "my children keep asking me why daddy is not at the école des mamans, I told them he should come too." The discourse of the field actors (project coordinators and French teachers) does not reflect the diversity of these women profiles and illustrates a reductive thinking scheme. They overinterpret women motivation or personal situation on the basis of a normative behavior they should have, according to their own criteria.
Another notable difference was the daycare facilities for the younger children when their mothers were taking French classes. This was initially designed as a merely logistical service so that the mothers could attend their French classes. But in addition to the mother-centered focus, the new funding partner (i.e. the Cantonal Integration Office) introduced a socio-pedagogical component for the children in order to get them ready for school and facilitate their integration into the local school system. In terms of implementation, we observed that the different écoles des mamans did not implement this component in the same way. Some of the projects did not accept this paradigm shift and continued simply to provide a daycare facility. Others set up a pre-school facility similar to the pre-school system run by the City of Geneva, which included work on the social and educational integration of children, in line with federal, cantonal and local directives on the education of migrant children. During our inquiries, we observed the daycare facilities of each association. While the play areas are similar, the way the children are taken care of is different. In some facilities, the children were left to play under the supervision of adults. In others the children's time was more structured, with games, learning activities and discussions: they would arrive, have some free or structured playtime depending on their age, spend some time on benches to talk in French, sing songs, read a story or play a group game, and then have snack time.
The interviews and observations also revealed that the French classes were conducted in two very different ways. In some projects, the classes were very academic and centered around a teaching manual and reading, vocabulary and grammar exercises. The classes were very passive and took primarily a lecture format. In other projects, the classes involved a variety of materials to help the mother learn to communicate with the school and other organizations in the neighborhood. The classes covered grammar and vocabulary and included practical exercises using actual documents received from the school. The mothers would, for instance, analyze children's books so that they would be better equipped to help their children with their homework. Use of this impromptu teaching material can be considered an actionable teaching approach that uses task-based language learning to help the learner become a social player (Rosen, 2009). We observed numerous exchanges in these classes. The French-as-a-foreign-language teachers and mothers co-constructed the learning experience: the mothers brought documents from school that they wished to have explained to them, or they asked the teachers to create templates they could use for school (e.g., an absence letter, or a request for a meeting).
We therefore observed different practices on the ground in both of the project's components. This can be explained by another observation. The projects were set up based on two different management approaches: for two of the projects, an association linked to each school had been created specifically for the project, while the other two projects were managed by an existing association that works with migrants on a number of projects but is not directly linked to the schools. During the interviews, these two approaches were discussed: for some, it was crucial that the association be part of the school to ensure the mothers and their children could be closely supported and to involve teaching committees. They considered that the teachers were best placed to know the families in question and the challenges regarding the relationship with the parents. For others, an external association seemed like the natural choice given the extent of the administrative work. In these cases, the school principals believed that the project had to be managed by an outside association to ensure its continuity.
Our observations revealed the impact of these two types of management: the daycare component was not implemented in the same way, and the quality of the French classes for the mothers was not the same. Paradoxically, when the schools managed the associations, the quality of the French teaching was lower and the daycare component did not include a socio-pedagogical approach to facilitate the children's transition to school. On the other side, the projects run by the migrant association worked to prepare children for school by forging close partnerships with preschool facilities in the city, and provided high-quality French classes to the mothers. There are two opposing approaches here: for some, the aim of the French classes is to educate the mothers, and the daycare is simply a logistical tool, while for others the French classes are an integration tool, and working to integrate the children is a separate goal.
These findings show that, despite the initial shared goals, which were aligned with the official texts governing migrant education, each project was implemented differently because of varying points of view across the associations.

Implementation by Local Players: Prejudices and Paternalism
Our study has shown the different ways in which the "école des mamans" project is implemented depending on who was in charge of implementation. There are two reasons for the diverging approaches: the project guidelines are vague in places, leaving room for interpretation, and there are no project monitoring tools. Two findings (the selection of beneficiaries and what they had to provide in return for the French classes) illustrate the underlying tensions that can occur when there is too much room for interpretation, especially concerning projects for migrants, as social representations of this group can be strong.
In the project guidelines, the beneficiaries are selected based on the following criteria: non-French-speaking mothers who are either isolated or in a disadvantaged situation. But the terms "isolated" and "disadvantaged" 16 are not defined. In reality, the project managers interpret this as they see fit and select the mothers exclusively on the basis of their being non-French-speaking, which they consider by default to mean they are also disadvantaged (these projects are run in impoverished neighborhoods). While this lack of precision in the guidelines does not necessarily cause difficulties for those running the projects, it can lead to the risk of categorizing the non-French-speaking women involved in the project. There is a risk of social categorization (all of the mothers are by default perceived as disadvantaged). There is a risk that their pre-migration education in their country of origin is not taken into consideration (i.e., their education and qualifications in their language of origin are not taken into account). Finally there is a risk that these women's skills are not recognized (except for their cooking skills, for instance).
In reality, the mothers that we met came from various social and economic levels and considered themselves disadvantaged to varying degrees. And these two factors were not necessarily correlated: mothers from more advantaged backgrounds may feel isolated, while others from more disadvantaged backgrounds may be integrated into their neighborhood or in regular contact with family and friends, who are generally from their community, while others arrived in Switzerland alone and then got married and don't have a network in the country. 16 Isolation and précarité are the terms used in the original French.
The beneficiaries get free French classes and in return they have to prepare a meal for the project team and the teachers of the school involved. This exchange raises two questions: Does there really need to be something in return? Are there not other ways to make use of the women's skills in a less stereotyped way? To answer the first question, we need to look at the integration processes that migrant populations have to go through. These processes are costly for the families, both symbolically and socially. The idea of having to give something in return sidelines the work already being done by these families and is probably rooted in a psychologization of the process. Of course, what they provide in return could be considered a form of gift designed to maintain the balance between the different parties -a form of reciprocity.
The issue of gender stereotypes also came up in our interviews. The institutional partners were not in favor of this system of providing something in return, which they considered contradictory to the country's integration-related values. They believe that the mothers should be recognized simply as parents. Those on the ground, however, think of it first and foremost as "an activity that the women enjoy doing" that enables the implementers to discover dishes from the mothers' countries of origin: "the idea is to adapt to the mothers' culture to avoid a separation." 17 Their words focus on the fact that the women's cultures of origin value cooking as a skill. These different perceptions need to be thoroughly deconstructed. The view that the supposed cultures of origin highlight women's domestic skills means that there is a risk of cultural reification, with the women remaining locked in roles that are not valued socially.
Furthermore, our observations showed that the format of the meals puts the mothers in a situation that reflects the implementers' social representations of them and even infantilizes them. Meal preparation consists in elaborating a handwritten invitation for the school teachers (sometimes with colors and illustrations similar to a child exercice). Then, on d-day, mothers prepare their meal at home and bring it to schools. Before starting to eat, they have to present and then describe the dish that they have cooked and some also have to prepare another activity (e.g., reading a poem in French or playing a French-language game). Most of them told us that during the meal they do not really talk to the teachers but mainly with other mothers of the program because "they all have more or less the same level in French and they feel more at ease" says Leyla from Syria. In reality, this format creates asymmetry in the relationship between the mothers and the other people present at the meals. The findings of our interviews and observations are directly aligned with the literature studied. The literature demonstrates that associations, like social workers (Garcia 2014;Giuliani, 2009), tend to show paternalism towards migrant women (after the spoken presentation by these mothers, the other people there reacted by saying "they're so sweet," 18 an expression usually used to describe a child) and to give them activities or make use of skills that maintain them in their family role (i.e., looking after the children, cooking and doing housework; Choffat et al., 2014). During our interviews, those involved repeatedly talked about the mothers thanking them, being very grateful, expressing their gratitude and being appreciative to the French teachers. This view is aligned with the idea that the institutions take a paternalistic approach to the female beneficiaries of the program. When we talked with the mothers, they also told us how grateful there were for the teachers' involvement. They are able to identify their own progresses, Fatiha from Sudan says "I feel more comfortable when talking to my son's teacher" or Leyla from Syria said "I am not scared anymore to go to the doctor or to ask for my way in the city center." Yet they do not acknowledge their progresses as the result of their own merit but rather as a result of the dedication of the French teachers. Regarding the meal, most of them told us "they love it" and that cooking was one of their favorite activity.
These results show how they are submitted to paternalism and how they are accepting a model treating them as children. In addition, there is a folklorizing vision of the culture of origin of these mothers that can also be observed in the school field. The preparation of meals is not specific to this project: we can find it in certain educational activities in schools or, for example, within the framework of development projects around the idea of citizenship in "neighborhoods" (Palomares & Rabaud, 2006). The analyzes made by Palomares and Rabaud of the latter show that the use of cooking activities by "mothers" is part of paternalistic postures. Following the example of the école des mamans, the authors identify speeches according to which these mothers are locked in the private sphere and this kind of projects is an opportunity "to leave their homes" locked in their roles of mother and their domestic role.
Krasniqi Malaj (2015) adds that society in general considers these women to be low-skilled. Yet, the interviews with the mothers showed that they come from very varied social and professional backgrounds and that they have many other skills that they could have made use of in exchange for the French classes. Many of them, for instance, speak several languages. And while many of them are stay-at-home mothers, some had previously held jobs in accounting or finance, and others did admin for their husbands' companies. The mothers could have done other things than cooking in return for the French classes, such as working with teachers on plurilingual projects (some mothers were already involved in this type of project).

Discussion and Conclusion
In a federal system, legislative mechanisms and powers are fully integrated across several different hierarchical levels. In Switzerland, the idea of integration, and how it should be achieved by local mechanisms, is consistent across federal legislation and cantonal policies. In other words, the classic vision of integration conveyed in federal texts is found in cantonal texts and aims. However, observations in the field show a locally variable discrepancy between the objectives set and the way in which they are implemented.
Rather than ensuring that implementation is adapted to local needs, those goals bring to light contradictory ideas of integration processes and prejudices about the mothers' culture. In this article, we have demonstrated not only that decoupling occurs, but also the many forms it can take. Sometimes, those involved seem to rework the vaguer parts of the legislative texts, which they don't always agree with; and sometimes their own attitudes towards migrants distance them from the spirit of the law without their representations explicitly being called into question. At the end of our reflections, it therefore seems to us that the case we are studying here is more a matter of loose coupling than of decoupling. The LEI's format, which is ultimately not very prescriptive, allows the development of flexibility and interpretation for local actors. In addition, there are weak monitoring and management tools at the cantonal level. This is in line with the principle of subsidiarity on which Swiss law is based, as we explained at the start of the text. On the other hand, this explains the various forms that the devices have taken and the developments since their creation giving local implementers a lot of flexibility in defining the mechanisms. The implementers are numerous and have a large amount of autonomy, which means there is a lot of room for the texts to be interpreted differently, especially as nothing is done in advance to address the implementers' attitudes and prejudices. This can be seen in the ways in which the different projects are run and what is offered to the mothers and their children. As a result, the paternalism and prejudices regarding migrant families continue.
Palier and Surel's '3I' approach allows us to understand how the three dimensions occupy a respective variable weight. At the institutional level, the guidelines are relatively consistent, although the view of integration is not always in line with the implementers' representations. We might add that these differences in how the projects are run can be explained by the lack of any kind of monitoring mechanism and the fact that the projects are run very locally (by associations).
In terms of interests, the local implementers focus more on the experimental, do-it-yourself approach used at the outset than on the changes that were brought in when the project was taken over by a new funding partner. And as we have already noted, there are major tensions between the professional identities of the groups involved: social workers, teachers and association members.
Finally, the 'i' for ideas has a strong weighting, as we have seen with regard to the implementers' perceptions of these families. Integration conceptions oscillate between two visions. At federal and cantonal level, official texts take a linear approach based exclusively on the individual who is summoned to integrate, notably through language acquisition -for economic integration purposes -but without taking into consideration the efforts of reshaping/restructuring of the host society. At the local level, integration conceptions go beyond the issue of language. However these conceptions tend to reify the culture of individuals by assuming that it is their duty to educate, or even 'normalise' (Giuliani, 2009;Garcia, 2014) migrant mothers considered too far removed from Swiss culture (itself fantasised) and from school culture.
In order to understand the conceptions and postures vis-à-vis migrant populations, we can apply the analytical categories of Grignon and Passeron (1989) in Le Savant et le Populaire (Felouzis & Fouquet-Chauprade, 2013). The first category focuses almost exclusively on the social and symbolic domination of migrant individuals (or population with immigrant backgrounds). In an extreme form of this conception, migrants are solely perceived through their experience of racism and discrimination, at the risk of promoting a miserable picture of these individuals and denying them any potential for action. The second category, on the contrary, values culture, solidarity, networks and community. We did not talk much about this category in our research, as it did not lend itself to this type of analysis 19 . However, we know to what extent the previously settled community of origin is an important resource from an identity, symbolic and material perspective. For example, Mathieu Ichou shows how "'neighbours' take on the role of 'providential individuals' who provide support towards the schooling of children, participate in 'the creation of social capital, which is academically profitable for the children of immigrants" (Ichou ,p. 245).
Similarly, Claire Schiff (2015) shows how the community of origin constitutes an economic resource and a resource for integration into the community to Sikhs newly arrived in the Paris region. These two examples thus show the capacity for action of individuals and the forms of solidarity that are essential to integration processes, which are not always to be found in institutions or politics, but also in the community. However, these aspects should not be overestimated either, at the risk of lapsing into relativism and culturalism.
Finally, we demonstrate that local initiatives go beyond federal guidelines and requirements. In this regard, it would be more adequate to refer to loose coupling rather than decoupling. French classes were designed by the Confederation as a prerequisite to migrant mothers' integration; they became education schemes for migrant mothers where they are subject to a situation of asymmetry with regard to social workers and the educational institution. Similar observations are made for other populations in precarious situations. The specificity of women migrants' situation lies in the perception by local actors who see their culture of origin as a deficit position.
At the beginning, the project was meant to include work with the local implementers and schools on their negative representations, but that component of the project was left aside, giving the stereotypes and prejudices free rein, which created asymmetrical relations between local implementers and the mothers. When dealing with issues such as the integration of migrant families, where approaches and representations can vary so greatly and even clash, there needs to be prior work to address those representations for implementation to be successful. Providing training and deconstructing stereotypes is central in this regard. Training and deconstruction of stereotypes is essential but not sufficient. The intent of the legislator is not neutral either. On the contrary, it vehicles a certain conception of migrant families and of the integration process. Let's not be mistaken: this scheme does not target "all" families newly arrived in Switzerland or in the Geneva area. The Geneva Canton welcomes important migrations of staff members from international organization (WTO, UN) who are not targeted by these policies. Thus, it suggests that these policies are focused to a population with "integration requirements," as stated in the law. What are these "integration requirements"? What are the criteria applied to identify those migrants who have higher "integration requirements" than others? What definition of integration is presupposed? What does it imply with regards to the Swiss society, its evolution and its ability to welcome migrants? These forces are at work in society and require more than a reflection on local actors' prejudices. It's about a more comprehensive political project. Readers are free to copy, display, distribute, and adapt this article, as long as the work is attributed to the author(s) and Education Policy Analysis Archives, the changes are identified, and the same license applies to the derivative work. More details of this Creative Commons license are available at https://creativecommons.