Multilingual education in early years in Luxembourg: a paradigm shift?

ABSTRACT There is a call for multilingual pedagogies including the use of literacy in several languages in early childhood education. However, many practitioners find it difficult to challenge the dominant language ideologies and are unsure of how to develop literacy practices in multiple languages. This paper is based in Luxembourg where a multilingual programme has been implemented in early childhood education in 2017. The research project examines the language used in daily communication and literacy activities of educators and parents in day-care centres, as reported by educators in two online-questionnaires. The findings show that the educators and parents use multiple languages when communicating, singing and reading with children in the centres. In addition to French and Luxembourgish which dominate, they use five other languages. Their reported multilingual practice reflects their beliefs that speaking and reading in several languages promotes language learning. However, while the programme is multilingual, a range of home languages are marginalised. The educators produce a language hierarchy in the centres in which the parents reproduce. While collaboration with parents can be effective in bringing home languages into day-care centres, educators need to be aware of language hierarchies and ideologies.


Introduction
In response to the demand to enhance the quality of early years education and better ensure access to the languages of instruction as part of achieving educational equity, multilingual programmes have been implemented in early childhood education in some European contexts, for example in some federal states of Germany, Switzerland, and Luxembourg. Some research findings indicate, however, that the focus continues to be on the majority language and that pre-schoolers rarely use their home languages in centres (Kratzmann et al., 2017).
The present exploratory paper provides an insight into the ways in which educators in Luxembourg address the linguistic diversity in early childhood education and foregrounds some language ideologies at play. In this small multilingual country in Western Europe bordering Germany, France, and Belgium, a new policy called for multilingual education in 2017. The overall aim of the policy is to facilitate children's integration into Luxembourg's diverse society and give them equal access to school (Simoes Lourêiro & Neumann, 2020). Based on the national framework for non-formal education (MENJE & SNJ, 2017), the programme has three pillars: language education, partnerships with parents, and networking with cultural, social and educational institutions. It provides one-to-four-year-olds with playful opportunities to encounter linguistic and cultural diversity, and helps them develop skills in Luxembourgish and French through conversations, songs, and regular literacy activities in several languages (p. 4).
The data of this paper derive from the mixed-method project collaboration with parents and multiliteracy in early childhood education (COMPARE, 2020(COMPARE, -2023 which investigates, firstly, the development of multiliteracy and collaboration between parents and educators in daycare centres, and, secondly, the influence of multiliteracy activities and collaboration on children, parents and educators. This contribution presents the language use of educators and parents in daily conversations and literacy activities, as reported by the educators in two online questionnaires completed in 2020. The findings describe the situation three years after the change of the language policy and could help policy-makers identify appropriate next steps. Internationally, they help understand the role of educators and parents in developing multilingual practices from the earliest stage of education. Importantly, they show that while collaboration with parents helps the educators to address linguistic diversity in the daycare centres, educators need to tackle language ideologies to develop more inclusive practices.

The multilingual education programme in Luxembourg
Multilingualism is an everyday reality in Luxembourg where residents speak a range of languages apart from the three official ones -Luxembourgish, French, and Germanon account of the high proportion of immigrants (Kirsch & Seele, 2020). In 2019, 47% of residents did not have Luxembourgish citizenship (STATEC, 2020). Luxembourgish was the language spoken most frequently at home (53%) followed by French, Portuguese, German, English, and Italian, depending on the citizenship of the residents. While 31% of them spoke multiple languages at home and 74% worked in a multilingual environment (Reiff & Neumayer, 2019). The most frequent languages used were French, English, and Luxembourgish. In the education sector, however, language use differs with residents speaking mainly Luxembourgish, French, German, English, and Portuguese (Reiff & Neumayer, 2019). Focusing on school-aged children, the Ministry of National Education, Childhood and Youth (MENJE) informed that 65% spoke a first language other than Luxembourgish in /19 (MENJE, 2020. While Luxembourgish is considered to be the language of integration, German and French play a central role in the trilingual national curriculum. German is particularly important because it is the language the children become literate in through formal schooling. To promote social justice and reduce the persistent inequalities in the attainment of children with a migrant background and low socio-economic status, the government took several measures including the 2017 language policy that called for a multilingual programme in formal and non-formal education institutions in early childhood education and care (ECEC). In Luxembourg, ECEC is divided into two sectors but regulated by the same ministry (Kirsch & Seele, 2020). The formal one is organised within the official school system and comprises an optional preschool year for three-year-olds and the two-year compulsory preschool for four-to six-year-olds. The non-formal education sector for children aged zero to four (e.g. daycare centres) is split into non-profit organisations and private, for-profit organisations. The formal sector employs mainly teachers; the non-formal one qualified educators, typically with a vocational degree. While the language use of adults and children was regulated in the formal sector by a policy asking for Luxembourgish, the daycare centres in the non-formal sector could decide on their linguistic profile and pedagogical approaches. The non-formal sector being largely private, much depended on the staff: 50% of the educators do not speak Luxembourgish and French dominates (Honig & Bock, 2017). Owing to the absence of official regulations, the linguistic landscape of the non-formal sector diversified. Luxembourgish came to dominate in the state-funded daycare centres and French in the private ones. Other institutions were officially multilingual with French, Luxembourgish, German, or English being used in everyday communication. Unsurprisingly, several ethnographic studies have indicated that multilingualism is a reality in the everyday practice of different types of structures in the non-formal sector. Nevertheless, many educators, even in bilingual centres, hold on to monolingual norms as well as policies of separating and excluding languages (Neumann, 2015;Seele, 2016).
The existing language diversity in the formal and non-formal sector was regulated with a change to the Children and Youth Act in 2017. Influenced by the development of multilingual programmes on an international level (García et al., 2017), Luxembourg witnessed a paradigm shift that resulted in a multilingual education policy. Teachers in the formal education sector and educators in the non-formal one are required to value children's home languages and develop some skills in Luxembourgish and French, although the weighting of these languages differs according to the setting (Kirsch & Seele, 2020). Furthermore, the professionals are asked to develop partnerships with parents because this 'collaboration can contribute to the development of children's home languages' (MENJE & SNJ, 2018, p. 111). In particular, professionals should 'engage in continuous exchange and make offers that actively involve parents in the everyday life of the centres' (p. 111). The educators are also encouraged to invite parents into the centres to tell stories or sing songs, for example in the children's home languages.
The policy was intensely debated in 2016 and 2017. The findings of a questionnaire completed in 2016 by teachers and educators indicate that the 44 participants held negative or ambiguous attitudes towards multilingual education (Kirsch & Aleksić, 2018). Since then, the new multilingual programme has been integrated into the compulsory national framework for non-formal education in childhood (MENJE & SNJ, 2018). Over the last years, the National Youth Service (SNJ) produced guidelines, videos, and conferences, and provided specialised training in multilingual education to 700 educators who became the référents pédagogiques.
This contribution investigates the ways and the extent to which the situation has changed since 2017. It explores whether the multilingual education policy has been transformative with regard to the range of languages educators and parents use in daily communication and literacy activities in the daycare centres. In doing so, it also addresses the extent to which educators and parents have overcome the traditional monolingual ideologies (Weber, 2014).

Multilingual education programmes: ideologies, practices, and activities
To put the situation of Luxembourg into a wider context, it is necessary to examine multilingual programmes in ECEC elsewhere.

Language ideologies, hierarchies, and multilingual pedagogies
Monolingual language ideologies in ECEC settings manifest when the dominant language is privileged over the minority language (Bergroth & Palviainen, 2016;Kratzmann et al., 2017) and when professionals separate languages according to hierarchies. They may be influenced by the ideology of the standard language and embrace the language hierarchy established by their nation-state and education system. At the top of this hierarchy are usually the national language(s), followed by the foreign languages, often English, taught at school. Community languages are at the bottom of this hierarchy (Ellis et al., 2011). Given such strong monolingual ideologies, it is difficult for teachers to draw on the children's highly diverse linguistic and cultural resources in the classroom. This is, however, necessary for education to promote equal opportunities and social justice (European Commission, 2011;García et al., 2017).
Multilingual pedagogiesalso called translanguaging pedagogies by García and her team (García et al., 2017)challenge dominant ideologies and practices and encourage social participation and social justice. They aim to leverage the students' resources and contribute to raising their achievements. In order to be transformative and contribute to individual and societal change, it is important that educators value the students' diverse resources, challenge traditional monolingual practices and design their curriculum and teaching in a way that students are exposed to several languages in meaningful interactive activities (García et al., 2017). Such flexible bilingual practices need to be carefully monitored and planned to guarantee equal participation (Hamman, 2018). Furthermore, they need to be based on 'an ethical and responsible theory of flexible multilingual education' (Weber, 2014, p. 7).

Literacy activities in multiple languages
Early literacy experiences at home and in ECEC influence children's language and literacy development and their school achievements (Farver et al., 2013;NICHD, 2005). Studies with monolingual children learning English have shown that shared reading or dialogic reading stimulates their oral language development. However, for this to happen, the adult and the child need to interact. When children engage in verbal conversations with the adults, they can develop vocabulary, knowledge of print and story comprehension (Torr, 2019;Wasik et al., 2016).
Studies on multilingual literacy activities in ECEC are rare. Research has been carried out mainly in contexts where English is introduced as a foreign language or in bilingual contexts with multilingual children. Two examples are a case in point. In Spain, Andúgar and Cortina-Peréz (2018) showed that teachers organised playful communicative activities in English that promoted participation such as circle time, storytelling and songs. In preschools in Sweden, songs were found to contribute to the linguistic and communicative development of multilingual children learning Swedish (Kultti, 2013).
A few studies come from the field of multiliteracies spearheaded in Canada. The aim of this innovative pedagogy was to develop multimodal, multicultural and multilingual approaches that engaged all students in a range of literacy practices and included parents (e.g. Taylor et al., 2008). Findings in primary and secondary schools showed that children perceived themselves as capable learners whose resources were legitimated at school. Teachers drew more frequently on children's resources and parents became more aware of their role in supporting their children's biliteracy development. Currently, two multiliteracy projects run in Germany. The projects Family Literacy and Rucksack are rooted in intercultural education and aim to implement multiliteracies in and outside school. They promote children's literacy skills and familiarity with books from preschool and encourage parent involvement (Salem et al., 2020). A study in Luxembourg with data collected before the 2017 language policy showed that such collaborative literacy activities in multiple languages were rare (Kirsch, 2019;.
In sum, this literature has shown that multilingual pedagogies have found their way into ECEC, but that they can only be transformative if educators are aware of the status of the languages used and prepared to challenge monoglossic ideologies and practices. While there is some research on multilingual approaches in ECEC, few are based in the non-formal sector of education. The present study is based in daycare centres in Luxembourg and seeks to understand the ways in which educators address linguistic diversity and move away from the dominant monolingual language ideologies. The research questions read as follows: (1) What languages do educators report using in daily communication and literacy activities with children? (2) What languages, as reported by educators, do the parents use with children when they offer literacy activities in the daycare centres?

Methodology
The present paper from the project COMPARE draws on two questionnaires with educators, analysed descriptively.

Questionnaires for the educators
The online questionnaire on the literacy activities reported on in this contribution was developed by the second author. It contains 58 questions divided into four sections. Each section was adapted from questionnaires used in several studies that investigated students' home literacy environment (Burgess et al., 2002;Farver et al., 2013). Section A is composed of questions on literacy activities in educators' institutions including reading books, telling stories, singing songs, doing rhymes or engaging in alphabet games (28 questions; e.g. 'How often do your read books to the children?'). Section B contains questions on the educators' beliefs about language development (14 questions; e.g. 'Communication in more than one language in the institution will hinder children's language development'). Section C is about the educators' reading activities at home (8 questions; e.g. 'How often do you read for fun and pleasure?') while Section D concerns the educators' professional situation (8 questions; e.g. 'How many years have you worked in early childhood education?'). The online questionnaire on collaboration between educators and parents, developed by the first author, is composed of five sections that contain 118 questions in total. The questions in Section A ask educators to give information about their professional situation (14 questions; e.g. 'What language(s) are spoken most frequently by the adults in your institution?'). Section B contains questions on educators' personal situation (7 questions; e.g. 'Which language(s) do you prefer to speak?'). Section C tackles the topic of the educators' reasons for collaborating, aims, and expectations, as well as the importance of collaboration between educators and parents (24 questions; e.g. 'Collaboration is important because parents are experts and partners with equal rights'). Section D is about current situations involving educators and parents, including passing conversations, personal meetings, parent cafés, and activities offered by parents as well as factors influencing the collaboration between educators and parents (38 questions; e.g. 'How frequently do parents/ family members tell stories to children in the day care centre? In which language(s) do parents usually sing/dance with children in the day care centre').
The questions in these sections draw on the work of Betz et al. (2017), Lengyel and Salem (2016) and Reynolds et al. (2017). Finally, Section E addresses the desirable collaboration between educators and parents (35 questions; e.g. 'How often do you wish that, in the future, you exchange information in passing conversations with parents?').

Procedure and sample
The questionnaire on collaboration was online from 31 January to 16 March 2020; the literacy questionnaire from 20 April to 22 June 2020. Having received a list of almost 700 specialised educators (référents pédagogiques) and of 341 d care centres from the MENJE and the SNJ, the authors sent out emails to these educators, asking them to complete the survey and forward the link to other members of staff. Three reminders were sent weeks later. In addition, completion of the questionnaires was mandatory for seventy educators from 35 d care centres who took part in a professional development organised by the project COMPARE.
After the data had been 'cleaned'participants were removed if they had responded to fewer than 50% of the questions -289 questionnaires on collaboration and 452 on literacy were analysed. The second questionnaire was answered by more educators, possibly because of the topic and the closure of the centres due to the COVID-19 pandemic. Of the educators who completed the literacy questionnaire, 239 or 53% (vs. 54% who completed the questionnaire on collaboration) worked in non-profit organisations. In both questionnaires, the majority were specialised educators in multilingual education (53% vs. 59%) and almost 60% had been working in a day care centre for less than ten years (59% vs. 51%).
Finally, it is important to consider the home languages of the children and the educators. In both questionnaires, French followed by Luxembourgish and Portuguese were the most frequently used languages by the educators at home, while English, German and Italian were spoken less. Considering the home languages of the children, the educators reported a wide range which testifies to the language diversity in Luxembourg (MENJE, 2020). Table 1 presents the languages that more than 25% of the educators reported the children used at home. Furthermore, 10-20% of educators reported the following six home languages, displayed in decreasing frequency: Russian, Chinese, Dutch, Albanian, Bulgarian and Czech. The similarity of the findings in both questionnaires is noteworthy.

Multilingualism in practice
The educators report that they themselves and the parents used several languages in daily conversations and literacy activities which indicates that they have moved away from the monolingual ideologies which previously dominated in ECEC.

Language use in daily conversations in the day care
The educators appear to be open to the daily use of multiple languages. The responses in the literacy questionnaire showed that 83% disagreed or strongly disagreed with the statement that communication in more than one language in the institution hinders children's language development. This belief seems to be reflected in the reported practices. The educators were asked in both questionnaires to indicate their language use in daily communication with the children. While the percentages slightly differ according to the questionnaire, the findings presented in Table 2 clearly illustrate that French and Luxembourgish are the dominant languages while other languages are used in addition. The frequent use of French and Luxembourgish (above 80%) reflects the linguistic profiles of the for-profit and non-profit organisations of the sector (Honig & Bock, 2017) and is in line with the national framework for non-formal education (MENJE & SNJ, 2018).
The multilingual education programme also requires educators to value children's other home languages. They can do this in various ways, for instance by allowing children to use their home languages or using these themselves, providing they know them. As seen in Table 2, the participants report using mainly Portuguese, English, German, Italian and Spanish apart from Luxembourgish and French. The reported practices are therefore multilingual. The comparison of the educators' reported language use at home and at work, presented in Figure 1, highlights some differences which may suggest that the educators make a special effort to use multiple languages in the daycare centre. For example, 17% of the participants in the literacy questionnaire indicate that they use Portuguese at home; 33% in the daycare centre. Figure 1 also provides an overview of the most frequently used languages of the educators and the children, as reported by the educators, indicating some parallels. The linguistic needs of children speaking Luxembourgish, French, Portuguese, German, English, Spanish and Italian may be addressed by educators who speak these languages. As several multilingual educators collaborate in a team, it is likely that some of them speak some of the children's home languages. By contrast, other home languages such as Arabic, Polish, Serbian/Croatian/Bosnian, Russian or Chinese (Table 1) are absent. These languages are less widely spoken in Luxembourg and by the educators and their absence marginalises these languages.

Literacy activities in multiple languages
The educators were also asked in the literacy questionnaire to report on the range of literacy activities they carried out such as reading books, singing songs or playing alphabet games. In what follows, the focus lies on singing songs, reading books and telling stories with or without a book. While 86% reported singing daily, 65% mentioned reading books and 62% telling stories daily.
The educators believed that reading books and telling stories in the main language of the centre promoted language development. While 75% agreed or strongly agreed that this practice influences the development of that particular language, 59% agreed or strongly agreed that it was beneficial for the development of all of the children's languages. The educators had similar positive views on reading in languages other than the main language of the centre. Here, 70% agreed or strongly agreed on a positive effect on the development of all languages. These positive beliefs were reflected in the reported multilingual practices.
Asked about language use, the educators indicate singing songs, reading and telling stories mainly in French (98%, 96%, 84%) and Luxembourgish (73%, 66%, 51%), thus, in the languages emphasised by the national framework. In addition, they mentioned the use of Portuguese, English, German, Italian and Spanish in relation to songs (33%, 47%, 36%, 12%, 7%) and to a lesser extent in activities around books. The reported use of Spanish and Italian, but of English and German, in particular, increases in songs compared to their use in oral conversations. While several languages continue to exist in literacy activities, thereby gaining in status, other home languages continue to remain invisible. Figure 2 shows the frequency of the languages reported by the educators. Languages reported less than 1% have not been presented.
Moreover, Figure 2 alludes to the status of languages in activities around books. French is the language used most widely across all types of oral and print activities. Luxembourgish, Portuguese and English are used less and the use of Italian and Spanish is rare in In which language(s) do you speak? In which language(s) do you sing?
In which language(s) do you tell stories? In which language(s) do you read? reading and storytelling. By contrast, the frequency of German, an official language in Luxembourg and the language of instruction at school, increases. This shows that literacy acts as a filter that reinforces the status of French and Luxembourgish, while also foregrounding German, which was less visible in daily communication. By contrast, Portuguese, English, Italian and Spanish are rare and other home languages are not mentioned. Their use, however, is required if all home languages are to be valued. Collaboration with parents, a pillar of the multilingual programme, may offer opportunities to bring in more home languages. Collaboration, surveyed in the second questionnaire, is, therefore, given some consideration next.

Language use in collaborative activities with parents
The questionnaire on collaboration indicated typical events in which educators and parents engaged together including passing conversations, personal meetings, parent cafés, and collaborative activities. The educators were asked to identify the ones they carried out and indicate the frequency. In what follows, the focus lies on three collaborative activities: singing, reading stories, and, finally, engaging in activities such as artwork or joining the children for a walk. Of the 289 participating educators, 34% reported that parents never engaged in such activities, 44% that they did so once or twice a year and 17% that they did so every few weeks. The educators were also asked to indicate the parents' language use with children in these three types of activities (Figure 3). French was the language used most frequently in daily communication (62%), singing (56%) and reading (57%), followed by Luxembourgish (53%, 46%, 48%). The other three languages most used in these three activities were Portuguese (27%, 27%, 28%), English (19%, 21%, 26%), and German (18%, 16%, 20%). Very rarely, educators mentioned other home languages, in this case, Serbian/Croatian/ Bosnian, Chinese, Arabic and Hindi. Figure 4 shows in addition that according to the educators, the parents rarely used their home languages in collaborative activities in daycare centres: 21% communicated, 15% sang and 19% read stories more than once a month in their home language.
These findings indicate that the parents bring the same languages into the centres as are already present and that these languages are not necessarily their home languages (Figure 4). This may suggest that parents choose (or were perhaps are asked to use) a language which is considered legitimate in the daycare centre. As educators reported, parents rarely brought in languages less widely spoken in Luxembourg. The status of the more dominant languages may therefore have been reinforced. Figure 5 juxtaposes the language use of the educators, reported in the literacy questionnaire and the language use of parents, reported by educators in the collaboration questionnaire. This juxtaposition suggests that the pattern of language use of the parents almost mirrors that of the educators in daily conversations and songs, although the educators indicate a more frequent use of English songs. French, Luxembourgish and German are used frequently by the educators and the parents, according to the former. Italian and Portuguese, however, are used less frequently by the educators in literacy activities compared to daily communication and songs. By contrast, the parents, as reported by the educators, use these languages in all activities. Other home languages have rarely been mentioned. The language hierarchy created by educators, seems to be reproduced by the parents.

Discussion
The following section discusses the findings in relation to multilingualism, language hierarchies, collaboration, and finally, educators and parents as agents of change.

ECEC, a multilingual sector with open-minded educators
The questionnaires have shown that the non-formal education sector in Luxembourg is truly multilingual: the educators, parents and children report on the use of multiple languages in the day care centres. This finding corroborates earlier ethnographic studies (Neumann, 2015) as well as the national statistics that show that residents use multiple languages at work (Reiff & Neumayer, 2019;STATEC, 2020). By contrast, the language use in the day care centres differs from the wider education sector. This result highlights the sociolinguistic specificities of the non-formal sector as well as the linguistic requirements generated by the diverse clientele. The educators appear to be open to multilingualism, similar to the professionals reported by Kratzmann et al. (2017) and Portolés and Martí (2020). These positive attitudes are likely to be related to the fact that the participants live in a country where multilingualism is both valued and a daily reality. Compared to the data from 2016 (Kirsch & Aleksić, 2018), the educators also appear to have opened up to multilingual education and moved away from the prevailing monolingual ideologies in the sector. They believe that the use of multiple languages in daily conversation and literacy activities is beneficial for the development of children's entire linguistic repertoire. A similar positive belief was held by prospective preschool and primary teachers in Spain. They understood that learning several languages is not harmful and disagreed with models of language separation (Portolés & Martí, 2018). Given that beliefs about language use and language learning are influenced by the country, educational policies, pedagogical practices and working experiences (Başöz & Çubukçu, 2014;Li & Walsh, 2011), one can assume that the multilingual programme has to some extent influenced the educators' beliefs and multilingual practices. For instance, the educators surveyed in 2021 reported more activities with books than those surveyed before the implementation of the programme (Kirsch et al., 2020). Still, in 2020, 18% of professionals read stories every other day and 12% once a month. This percentage shows that literacy activities have not become a daily routine yet although they are firmly established in ECEC (Andúgar & Cortina-Peréz, 2018;Torr, 2019;Wasik et al., 2016).

Languages hierarchies
While the educators embrace the use of multiple languages, a language hierarchy is in place (Ellis et al., 2011). This finding echoes other studies where ECEC practitioners indicated a clear preference for high status languages (Kratzmann et al., 2017;Portolés & Martí, 2018). The dominant use of French and Luxembourgish in Luxembourg is expected owing to the new policy and the linguistic profiles of the educators. The place of German is noteworthy, however. German was not included in the language policy (MENJE & SNJ, 2018). This omission had been widely criticised in public debates in 2017. How could the multilingual programme in ECEC ease the transition to primary school and promote social justice (Simoes Lourêiro & Neumann, 2020) if German, the main language of instruction and the language of literacy at school, was not given any consideration? Albeit not firmly rooted in the programme, the language has made its entry into the non-formal education sector through the educators and the parents who report its use in literacy activities. Literacy, therefore, acts as a filter that reinforces the status of French and Luxembourgish while also foregrounding German.
German, together with Portuguese (a dominant community language) and English (a global language), occupies the middle section of the language hierarchy. At the bottom figure Italian and Spanish, used less frequently. The language hierarchy which is created in the daycare centres appears to be reproduced by the parents. Other home languages less widely used in Luxembourg are mentioned once or twice by the educators and there is a danger of marginalisation.

Collaboration with parents
Collaboration with parents is a pillar of the multilingual programme (MENJE & SNJ, 2018). Educators are encouraged to exchange regularly with parents, invite them to offer language activities in the day care centres and develop a parent advisory council. A study in 2016 showed that educators rarely engaged in collaborative activities such as participating in art work, singing and narrating stories. Of the 99 educators, 41% reported that parents never engaged in such activities, 43% that they did so once or twice a year, and 8% that they did so every few weeks (Kirsch, 2019). In 2020, fewer educators report not engaging in such activities and more doing so every few weeks. The educators in the 2016 and 2020 questionnaires may not be the same, therefore, we can only tentatively suggest that the implementation of the national framework may have contributed to improving collaboration.
Parents in Luxembourg are said to bring in the same five languages that have been given a legitimate space into the centres. While the programme is multilingual, it may only be transformative and fulfil its promises of social justice if educators reflect on the existing language hierarchies, actively challenge them (García et al., 2017;Weber, 2014) and develop partnerships with all parents (Salem et al., 2020).
Parents do not have the same social position as educators in the non-formal sector and, therefore, may not suggest literacy activities in home languages, unless they are encouraged to do so. They may be particularly reluctant if their literacy cultures are more oral than print-based and differ from those that have a legitimate place in the centres. Family literacies depend on the parents' background and lifestyle (Poolman et al., 2017) and may differ from those in ECEC institutions. According to Cairney (1997), the purpose of collaboration is not 'to coerce, or even persuade, parents to take on the literacy definitions held by teachers. Rather, it is to enable both teachers and parents to understand the way each defines, values and uses literacy as part of cultural practices' (p. 65). This requires interest and respectful dialogue. Apart from the parents' and the educators' biographies, socio-economic status, language competence and experiences, effective collaboration depends on the educators' open-mindedness, interest, resourceorientation and reflexivity and the way in which they give multicultural aspects due to consideration (Lengyel & Salem, 2016;Reynolds et al., 2017). These very aspects have been addressed in multiliteracies projects based on intercultural education (Salem et al., 2020;Taylor et al., 2008).

Agents of change
The educators and parents in Luxembourg have become agents of change (Bergroth & Palviainen, 2016). Their agency is visible in the spaces they give to languages less visible in the national framework. The educators emphasised German in literacy activities, a language omitted in the framework. This shows that they have set new priorities and have begun to develop new practices. Similarly, according to the educators, parents engaged in storytelling activities in Italian and Portuguese, languages that the educators seemed to have filtered out in literacy activities.
In order for multilingual education to be transformative, educators and parents need to challenge dominant practices, overcome the language hierarchy they currently produce and bring in marginalised languages and cultures. These findings confirm that the implementation of a policy is not top-down but that actors interpret and transform policies to suit their needs (Palviainen & Curdt-Christiansen, 2020).

Conclusion
Multilingual approaches in ECEC exist in some European countries. Luxembourg, where a new programme was implemented three years ago, is, therefore, an exciting hub for doing research of international significance. The findings of this paper have shown that policy-makers, educators, and parents have been actively developing multilingual approaches which take account of children's linguistic backgrounds and develop their repertoires. Not all languages are visible as yet and the sector needs to act to avoid the marginalisation or obliteration of languages and cultures used less widely in Luxembourg. The understanding that various people speak various languages is not enough. Educators need to adopt a more inclusive, heteroglossic perspective (García & Otheguy, 2020;Weber 2014). Initial training and professional development courses with a focus on intercultural education may help educators become more aware of their values and language ideologies, reflect on these, and develop the multilingual stance necessary to design a learning environment where children connect home and institutional languages (García et al., 2017;Lengyel & Salem, 2016). While collaboration with parents is key to introduce more languages into the non-formal sector, the role of the parents, as outlined in the policy, needs to be revised as it cannot be limited to linguistic input. Further research is needed to examine the space occupied by German and English as well as home languages, and the agency of educators, parents, and children.
While this study has provided insights into the multilingual programme in Luxembourg and contributed to a better understanding of the actors and language ideologies, it has also some limitations. The paper is based on two questionnaires and, therefore, can only indicate reported beliefs and practices. A questionnaire with parents and observations in three centres will further our understanding of collaboration between the educators and the parents, and offer insights into multiliteracy activities in the daycare centres. Efficient collaboration could help all actors bridge the gaps between language and cultural practices at home and in educational institutions, and contribute to the children's linguistic as well as emotional, social, and cognitive development.