Young children as actors of institutional language policies and practices in day care centres

As children’s agency in influencing institutional language practices is often not carefully reflected in early childhood education curricula, the objective of this paper is to offer meaningful insights about how institutional language policies are both reproduced and transformed by children’s everyday use of language. For this purpose, we will combine conceptual resources from social theory, sociolinguistics and childhood studies in order to analyse children’s linguistic behaviour by applying a structure-agency perspective as a relational approach. Drawing on data from ethnographic field research within institutional day care centres in Luxembourg, our findings demonstrate that the status of children as actors in institutional language practices is strongly connected to institutional policies as a structural condition. However, this does not mean that children just enact these language policies, because they are actors of both maintaining, undermining and alternating them. In this respect, especially the translanguaging of children and caregivers plays a crucial role in the Luxembourgish context as it allows to build a bridge between the official institutional language policy and the individual linguistic repertoires. Considering the goal of establishing a plurilingual environment in early childhood education which now is paramount to the educational language policy of the Luxembourgish government, this article suggests that translanguaging practices should be considered as one of the key starting points to create a plurilingual ecology in and through everyday practice in the day care centres.


Introduction
Luxembourg can be described as both a super-diverse and a multilingual society. Historically, multilingualism in Luxembourg emerged from a triglossic national language ecology which diversified further in the slipstream of different migration waves since the end of the 19th century (MENJE 2018 a: 109). 1 This is reflected by the large use of translingual communication practices applied by speakers drawing on different linguistic repertoires simultaneously, flexibly and strategically in processes of making sense and meaning (Kirsch and Aleksić 2018;Neumann 2015). This goes beyond the mixing of and the switching between languages, which is also common in Luxembourg, and includes the three official languages Luxembourgish, French, German, but also other ones like English, Portuguese, Spanish and Italian. The challenges regarding multilingualism in Luxembourg arise particularly within the context of the Luxembourgish education system. The diverse school populationand especially those with an immigrant backgroundare confronted with an institutionalised trilingual curriculum. Given that students´multilingual competencies are assessed separately on monolingual constructs of native speakers, language competencies frequently lead to educational inequalities and school failure (Simoes Lourêiro et al. 2019). Promoting multilingual competences already in the early years thus encompasses a prerequisite for educational success (MENJE 2018a). Responding to these challenges emerging from super-diversity, the educational policy programme of éducation plurilingue 'plurilingual education', was introduced in 2017 in the Luxembourgish formal and non-formal settings of Early Childhood Education and Care (ECEC). This new plurilingual program aims to prepare children for the language situation in Luxembourg, in order to facilitate their integration into society, arousing their curiosity for language learning and especially to ease their transition into the Luxembourgish education system.
Curricula and language promotion programmes in ECEC mostly put emphasis on what caregivers and educators can do in order to implement a multilingual language environment in the institutional everyday life. However, although research in applied linguistics shows that engaging with children in child-centred play stimulates linguistic interactions and development (Glupczynski Spencer et al. 2011: 117;Kirsch and Aleksić 2018: 151), dealing with a linguistic and cultural super-diversity in ECEC poses a major challenge in the pedagogical practice (List 2018;Neumann 2015: 23). This is not least demonstrated by several ethnographic field studies in multilingual or bilingual crèches (Benz 2017;Brandenberg et al. 2017;Seele 2016;Schwartz 2010). Research is focused on language practices initiated or guided by the adults in these institutions, while the multi-or translingual conversation strategies applied by children themselves are often overlooked (Brandenberg et al. 2017).
In this line, drawing on the study by Glupczynski Spencer et al. (2011) on linguistic diversity in early childhood, we encounter a structure-agency divide claiming that although policy as a macro-level agent plays a crucial role regarding the programmatic shift in ECEC, the actual change within the field is mutually determined by micro-level agents such as early years practitioners and children; though the impactparticularly of the latteris often not acknowledged. In consequence, very little is known about children's agency (0-4 years old) in terms of how they are shaping and influencing the multilingual environment in ECEC settings. Thus, this paper intends to explore the following question: How do children contribute to the constitution of language practices in institutional day care centres with their particular language repertoires? This question is twofold in the sense that it deals with (1) children's language use and translingual strategies (see García 2009) between their home language(s) and the institutionalised language policy in the day care centres on the one hand, as well as with (2) the influence of children's language practices on the linguistic environment of day care centres on the other hand. These questions will be addressed by drawing on data and findings from ethnographic field research in Luxembourgish day care centres with children under the age of four. The aims of this paper consist of grasping insights about how ECEC institutions deal with and can benefit from the language repertoires and practices of childrenboth in interaction with practitioners as well as within peer interactions among children themselves. Although the data we will refer to in the following are stemming from case studies dating back to the time before the programme of early plurilingual education had been introduced, they nonetheless give valuable examples and insights concerning the question of how children as actors are influencing the language environment in day care centres. These insights can be especially useful when reflecting on the challenge of how the new programme of plurilingual education could be filled with life by taking into account in how far children exert influence on the language use and environment in ECEC settings.
2 Young children's agency and language policies/practices Social scientists have looked at both structure and agency in isolation as well as in mutual dependency (Edwards 2016: 73). However, it was especially the term agency which has been given particular attention in contemporary theoretical discourses. This also constitutes one of the core issues of the research agenda in the international and interdisciplinary field of childhood studies (James and Prout 1997: 8). In social theory, the conceptualisation of relational approaches to agency (Emirbayer and Mische 1998) has become particularly prominent in recent years. This can be related to an attempt to overcome the classic structure-agency dichotomy as well as the macro-micro division of the 'social', which is, for example, one of the basic assumptions of traditional socialisation theory. The debates in social theory also influenced both the discussion on agency in childhood studies (Esser et al. 2016) and research in the area of applied linguistics. The latter primarily deals with the construction of this notion and the impact of agency on the reproduction or transformation of embedded structures (Ahearn 2001;Schwartz 2018). In addition, the primacy of agency constitutes a common framework in educational practice-theory (Biesta and Tedder 2006), which is also reflected in studies referring to bi-and multilingual education policy (i. e. Schwartz 2018). However, when it comes to the question of how different forms or grades of agency should be systematically determined in terms of the impact of actors, agency is still very differently conceptualised across various academic disciplines or even within the same field of research. These conceptualisations include terms such as "oppositional agency, complicit agency, agency of power, agency of intention, etc." (Ahearn 2001: 130), a fact whichfollowing Ahearn (2001)calls attention for scholars to clearly define how this term is being adopted and approached. Based on Giddens' (1984) concept of structuration, this article does not propose a dichotomous but a dualistic and relational perspective on structure and agency, claiming a co-dependency between these two analytically relevant dimensions of a social order. Following Ahearn (2011: 112), agency is referred to as agents' "socio-culturally mediated" actions that produce and reproduce structure while these actions derive from conditional capacities linked to the position and opportunities of actors in a given social context. Structure is seen as a both confining and enabling framework of social rules and resources (rule-resource sets, i. e. policies, structural power and constraints such as sanctions), which, according to Giddens, are neither completely independent of agents, nor do they have full control over their actions. So, the relation of structure and agency can be conceptualized as a recursive process including both the reinforcement of structures and their transformation. While Giddens rejects the primacy of agency, claiming that the institutionalisation of structural elements leads to routinized social practices, he nonetheless puts emphasis on agents' ability to transform a "structured-praxis" by means of their social position and influence or their practical knowledge regarding internalised social structures (Edwards 2016). Given the dual and recursive constellation of structure and agency, by focusing on young children's agency as a key concept in this paper, we thus adopt an ecological notion of agency that reflects Giddens' (1984) concept of structuration. This means that we will focus on children as "actors-in-transaction-with-context" and "actors acting by-means-of-an-environment rather than simply in an environment" (Biesta and Tedder 2006: 19; see also Emirbayer and Mische 1998). However, in difference to Giddens, we do not necessarily assume that agency as a socio-culturally mediated ability to appear as an actor in a certain social context requires a specific level of consciousness, intentionality and cognitive capacity depending on age or developmental status. Such an assumption is especially problematic when it comes to the empirical question of observing the agency of crying babies or very young children in general and would mean a relapse into an essentialist conception of children´s agency. This conception has been largely criticised in sociology of childhood during the last years as it tends to construct agency as a pre-social feature of human individuals (see Esser et al. 2016;Oswell 2013). Thus, similar to the research of Bergroth and Palviainen (2017) which also deals with the methodological challenge of the intentionality and reflexivity of children´s agency in regard to language policies, we analyse agency in our observational data rather as an effect of relations emerging from a concrete social context than as the origin of such relations or an outcome of a certain capacity of an independent individual. In this sense, relational approaches are considered as based on a flat ontology which allows to analyse the agency of children as the situational and practical accomplishment of a social order children are themselves actively contributing to (James, 2009).
Shedding light on actor's actions (practices) that derive from a responsiveness to their embedded structures (i. e. environment or context) helps to understand the interweavement between both structure and agency as well as the process of reproduction and transformation of social practices (Edwards 2016;Schwartz 2018). In this perspective, analysing children's agency in regard to language practices in ECEC services is of particular interest, as these settings often represent children's transition into a new language environment (Boyd and Huss 2017; see also Björk-Willén and Cromdal 2009;Cekaite et al. 2014;Karrebaek 2011;Kultti 2013, andKultti 2014). In regard to our analysis, this means that we are dealing on the one hand with children's linguistic repertoires, and on the other hand with institutional language policies and practices. However, instead of treating both sides as different and separated realities, we will pay attention to children's enacted linguistic repertoires as an integral part of everyday language use and policies in the ECEC institutions. The term institutional language policy in educational contexts refers to "statements of goals and means for achieving them that constitute guidelines or rules shaping language structure, language use, and language acquisition" (Toleffson 2008: 3). However, language policies can have an explicit and/or an implicit status (Baldauf 2005), so that they do not only exist in the form of written documents but also as a locally practiced language policy (Bonacina-Pugh 2012). Speaking of practiced language policy implies that we do not draw a sharp distinction between institutional language policies and practices of language. The latter often have an institutionalized status so that we rather regard institutional policies and practices as only analytically distinguishable sides of the same coin. Against this background, in the course of our study, the written and official institutional language policy thus represents the structural condition under which the children's and the caregiver's agency comes into force and, in turn, takes influence on and is influenced by the everyday institutional order of language use in terms of a practiced language policy.
In the past three decades, the concept of agency in the broader field of childhood studies has stimulated a vast amount of often small-scale empirical studies focusing on children's everyday micro-lives in order to reveal their capacity of meaning making, acting independently orat leastof influencing their social environment (Leonard 2015: 128). Considering the more specialised area of research on multilingualism, the outset in research on the relationship between the agency of plurilingual preschool children and their language practices in bi-and multilingual institutional settings has also demonstrated the role of children's agency in conceiving of and coping with their own and others' language practices Fogle and King 2013;Ahearn 2001; see also Almér 2017;Cekaite and Evaldsson 2017;Puskás and Björk-Willén 2017). Studies on language policies in early childhood education (e. g. Bergroth and Palviai-nen 2017; Kirsch 2018; Schwartz, Kirsch and Mortini 2020) further show that pedagogical practices construeas stated by Edwards (2016: 80) -"educational rhetoric-reality gaps" in regard to the implementation of educational policies (structured rules, such as language policies), due to individual (e. g. agents' actions) and structural (e. g. agents' environments) factors. In other words, in order to analyse the issue regarding the complexity of theory-in-practice, one must consider the institutional (temporal, cultural and structural) context to construct a holistic understanding concerning the realities of practices. In line with Giddens' notion of structurationthat defines pedagogical practices as behavioural routines deriving from the dynamic interplay between structure and agency (Edwards 2016) -Kirsch (2018) explains this phenomenon by means of agency being the ability to impact the policy-enactment and development by supporting or undermining governmental intentions through their interpretation of established political structure.
With the discussion on children's agency in relation to institutional language use and policies, this article aims to extend the state of the art of current research, which mainly focuses on children's agency in free play where the presence and influence of adults is weak. This is because research on children's agency traditionally assumes that their agency is more transparent within peer interactions regarding the use of bi-and multilingual practices (i. e. Mourão 2018). By analysing children's contribution in the constitution of language practices, this paper not only focuses on children's agency in reproducing or counteracting established structures, but also on their influence in modifying or directing practiced institutional language policies (see also Jaspers and Verschueren 2011). 3 The research context: Early childhood education in Luxembourg and the challenges of a multilingual society During the past ten years, Luxembourg has witnessed an enormous transformation in the sector of ECEC before compulsory schooling at the age of four. This transformation is twofold as it includes both a vast increase of day care facilities and places for young children, as well as significant efforts by the state to guarantee a high quality of services in the sector (Honig 2015;Neumann 2018). One of the most significant developments in terms of quality was the introduction of the Cadre de référence nationale sur l'éducation non formelle des enfants et des jeunes 'National Framework for Non-Formal Education in Childhood and Adolescence', which clearly identifies the educational mission of the day care sector. Furthermore, this document explicitly defines educational goals and guidelines in terms of a pedagogical approach to be applied when working with young children (see MENJE 2018a). Thus, for the first time, Luxembourg established an early childhood education curriculum for preschool children comparable to similar curricula and programmes in other OECD countries such as Germany, Austria, UK or Norwayto mention just a few.
While the national curriculum already included several sections dealing with the issues of linguistic diversity and plurilingual education (see e. g. MENJE 2018 a: 24), the implementation of the programme éducation plurilingue 'early plurilingual education' in 2017 has increasingly strengthened these issues and now forms an integral part of the national curriculum for preschool-age non-formal education. This programme is based on a new policy guidance with three main pillars: (1) familiarising children with Luxembourgish and French as well as valorising and supporting children's home language(s); (2) creating a network between the fundamental institutions within the ECEC sector such as schools and day care centres, as well as local and national welfare services; and (3) partnering with children's families (MENJE 2018b).
This recently implemented programme of early plurilingual educationwhich represents the first official language policy in the ECEC sector in Luxembourgimplies great challenges for the structures and professionals. Several ethnographic field studies in Luxembourg have shown that, in previous years, most of the structures were relying intuitively on an institutional policy of promoting a certain language (declaring either Luxembourgish or French as main language) in order to come up with the linguistically diverse environment represented in the everyday life of the different centres (see Neumann 2012, andNeumann 2015;Seele 2016). In addition, these studies have also shown that the aim to promote a certain language practically was often linked to a didactical approach based on a monolingual norm of language use which is guided by the assumption to "use only one language at a time" (Jørgensen 2008: 168). This was the case despite the fact, that there neither was a compulsory language curriculum for the ECEC sector prescribed by the state nor any regulation privileging one of the three official languages over another (Seele 2015: 264). Moreover, these ethnographic studies have demonstrated that the monolingual norm was even applied in officially bilingual centres as they still separated languages in the sense of a "double monolingualism" (Jørgensen 2008: 163), such as when applying strategies like one faceone speech . Against this background, the programme of early plurilingual education can also be seen as a reaction aiming to overcome the monolingual norm of language use in day care centres, which stands in sharp contrast to the language diversity in the social reality of the country.
In light of the foregoing, this paper will focus on language practices in the ECEC settings before the establishment of the framework for early plurilingual education. This will be done by raising the question of how far the institutional strategies and language policies shape and are shaped by language practices the children apply in institutional everyday life among themselves as well as in conversation with adults. With the practices and conversations of children at the centre, the paper draws attention to their agency in contributing to the enrichment of the linguistic environment of ECEC settings. Hence, the paper intends to offer a complementary perspective to the focus pedagogical guidelines routinely apply, which are much more concentrated on the possible activities of adult caregivers and parents than on those of children.
This perspective will be unfolded and illustrated based on data and findings from ethnographic research on language practices in Luxembourgish day care centres conducted in the course and the framework of the project "Realities of Education and Care. Quality and Qualification in Flexible Structures of Daycare in Luxembourgian Maisons Relais pour Enfants" (see Honig et al. 2013). In general, the project dealt with the practice of education and care for children under the age of four in the day care sector of Luxembourg by examining local, institutional everyday life with the methods of ethnographic fieldwork. This included participant observation, audio recordings and videography as data collection strategies in six selected childcare centres during different research phases between 2010 and 2015. All phases incorporated several weeks of intensive fieldwork in which the researchers placed themselves approximately three times a week at different sites within the institutional setting of the day care centres. For contrasting purposes, the sample of the sites for field research has been selected during the research process on the basis of the following criteria: location of the centres (urban/rural), number of children inscribed (less than 50/more than 100), group structure (age span of groups), sociocultural environment of the centres (privileged/non-privileged), legal form of the provider (public institution/private and non-commercial/private and commercial). 2 Data analysis was based on field notes and protocols from participant observation, field interviews with children, parents, adults and caregivers, audio and video recordings as well as photo-graphs, documents (e. g. local concepts, annual reports), and artefacts (e. g. pieces of handicrafts by children and adults). 3 Data analysis was based on an inductive process by applying coding strategies of Grounded Theory (open, axial and selective coding) in order to categorize different forms of pedagogical practices understood as activity complexes relevant to the educational ambitions of the centres. As the main study revealed language practices in the centres as one of the most important arenas of "doing education" in Luxembourgish ECEC, the project resulted also in several case studies which paid special attention to the function of the language use by adult caregivers in these practices (e. g. Neumann 2012, andNeumann 2015;Neumann and Seele 2014;Seele 2016). For the purpose of this paper, these case studies have been revisited and central sequences were analysed again with a shifted focus towards children´s agency. 4 The analysis was based on the following questions: In how far do children appear as actors in the sequences, how do they exert influence on the ongoing situation? How is this linked to the language use of children and how is their language use linked to the institutional language policy of the centre? Not least: How do caregivers react or reflect upon the language use of the children and how does this result in a restricted or an extended form of children´s agency? 4 Children's agency in shaping language environments: Examples from field research The following collection of sequences from field research gives insights into children's agency in influencing everyday language practices in day care settings which are characterised by a diversity of languages spoken by different actors. In examining language practices in a multilingual day care environment, we focus on both interactions between children and practitioners, as well as among peers.
With these sequences, we intend to demonstrate in how far children's language use is linked to institutional language policies while children are influencing the accomplishment of these policies at the same time. This means that children's agency manifests itself in both reproducing and undermining or transforming the explicit or implicit institutional language policy. Across all four presented obser-3 A detailed overview of the methodology and different findings of the whole study has been published in Honig et al. (2013). 4 All the sequences in the following have already been presented in former publications but were previously analysed under the totally different perspective on the language use of the adult practitioners. Here, it is the first time they are analysed with a special focus on children´s agency.

Young children as actors
vations, we will thus outline the importance of the acknowledgment of young children's agency in shaping the linguistic ecology in ECEC settings.

Translanguaging and language learning opportunities in conversation among peers
Despite the monolingual norm many day care centres were applying until the programme of early plurilingual education came into effect in 2017 (e. g. Luxembourgish as the colloquial language; see Kirsch and Günnewig 2015), the following excerpt shows that multilingual practices seem to nevertheless constitute a natural occurring phenomenon among preschool children as a matter of course beyond this institutional policy and the interactions between the children and the practitioners.
Annabelle (2 ½ years) opens a picture book and slides her fingers over the animals depicted there. She looks at Pierre (2 ½ years), points to one of the pictures and tells him excitedly and in French: "C'est un cochon!" (French, 'That's a pig!'). Pierre looks at her with big eyes and says in Luxembourgish: "Weider!" (Luxembourgish, Go further!). Annabelle slides her fingers a bit further, points to the next picture and says: "Une chèvre!" (French, 'A sheep!'). Pierre laughs and says, again in Luxembourgish: "Nach weider!" (Luxembourgish, 'More further!') (Neumann 2015: 31).
This sequence took place in a day care centre in the morning during a free play phase, that is, independent from caregiver-led activities. Both children, Annabelle and Pierre, seem to be engaged in a meaningful free play interaction that offers a distinct language learning and development opportunity as they are using their own language repertoires in a translingual form of communication. This results in a situation including more than just one language and without any explicit form of translation. In doing so, they go beyond the official language policy of the day care centre which emphasises Luxembourgish as the colloquial language for children as well as for adults. 5 A picture book, as a multimodal text, serves in 5 Although there was no official language policy by the state for ECEC institutions before 2017, the day care centre decided Luxembourgish to be the main language of everyday communication. The fact, that Luxembourgish is regarded the main language of everyday communication of both children and adult caregivers is for example assumed in the annual reports of the day care centre's groups (see Seele 2016: 55). Furthermore, this was also given as information in the admission interview with parents and also regularly communicated to children in everyday interaction. In order to characterize the everyday language use in the centre, Seele (2015: 266) summarizes that the caregivers "speak Luxembourgish with the children as a matter of course. Retrospectively, they justify this case as an interactive mediator of multilingual practices, linking pictures and words to create meaning (Mourão 2016). However, whereas Mourão's (2016, and approach to picture books refers to a picture-word dynamic in which children's agency relies on their engagement in regard to an interaction with an adult reader and the interplay between the visual illustrations and the written words, this instance outlines the interweavement between pointing to and expressing the images within peer interactions, based on children's language repertoires. Similar to the findings of Chaparro-Moreno et al. (2019), this clause represents peers' agency as a means of language learning and teaching. In this line, Pierre is thus given the opportunity to stretch his language repertoire by being exposed to a foreign language, whereas Annabelle's agency is represented as a form of pedagogical practicewhich is usually performed by a pedagogical instancebased on her interactional approach as a way of teaching. Such settings of child-initiated, play-based activities constitute a structural framework within the ECEC services that enhances children's agency by enabling more peer interactions Chaparro-Moreno et al. 2019;Mourão 2018); it is this takeover aspect of the teaching-role that is essential in supporting children's agentic contribution to the institutional language practices. Therefore, Annabelle appears as a kind of a collaborator who supports the educational mission of the institution. In essence, we acknowledge therefore the potential of fostering preschool children's language repertoire through dynamic peer interactions in free play situations, regardless of the different language profiles. This is most relevant when considering that playin particular free playhas been placed at the centre of the early childhood curriculum for over a century and has a number of forms; however, it is child-initiated play that is considered essential in children's development (Mourão 2014).
The next sequence represents a similar case of children's agency within a contrasting framework. Regardless of child-initiated, play-based activities, children also clearly perform agency in educator-led activities by conforming to the institutional language policy in relation to a change of interaction. The following field note was taken during a common activity in the morning when six children accompanied by the caregiver Ilona were baking bread in the kitchen and playing with the pastry while preparing it.
Interestingly, Alessandra (3 years and 4 months) and Carmen (3 years and 10 months) started talking French to each other during the activity. They both speak it quite fluently, this practice with different reasons, ranging from conservative attitudes about preserving the 'mother tongue', to the pragmatic need to have one common language that everyone understands, and, finally, to the future-oriented argument of preparing the children for school". often repeating after each other and making up a fanciful game about the pastry. They say, for example: "C'est le papa kangourou. Et ça, c'est la maman. Mon petit chouchou." ['This is daddy kangaroo. And this is the mommy. My little darling']. Ilona (caregiver) comments to me: "Unsere beiden Französinnen." ['Our two French girls', in German] However, she doesn't say anything to the girls themselves, just continuing to speak Luxembourgish with them. And they also answer her in Luxembourgish. Only one time, Alessandra says something in French to Ilona, but Carmen translates it immediately. Moreover, the girls also teach each other new vocabulary. For instance, when Ilona forms a serpent out of the pastry, Alessandra comments to Carmen: "Ça, en français, c'est un serpent." ['That's a snake' in French.] Carmen repeats: "C'est un serpent" ['That's a snake'.] [...] (Seele 2016: 115-116). By dissecting this sequence, we ascertain three different situations of children's performance of agency, in which the caregiver-led activity is marked by turn-taking roles in terms of language practices.
First, whereas the pedagogical activity is offered in Luxembourgish, both children interact nonetheless in French with each other. Although the practitioner clearly prioritises the Luxembourgish language by complying with the monolingual norm of language use and promotion of the institution, a more permissive and inclusive approach to multilingual practices is adopted by enabling the children to use their language of choice instead of constraining their agency, which shapes the linguistic environment within the institutional setting.
Second, although both children speak French to each other, they conform to the language practices of the practitioner by using Luxembourgish as well during their interaction with her. Even when Alessandra says something in French to Ilona, Carmen immediately translates it into Luxembourgish. In this line, the children's flexible and strategic shuttling between French and Luxembourgish indicates their agency that derives from individual and structural factors in relation to a clear notion of appropriate language use within the change of settingson the one hand, with a peer, and on the other hand, with the practitioner. Linguistically, this also includes code-switching between French and Luxembourgish, but, from a perspective on the social position of the children in the setting, the shuttling between the two languages is also an act of sense-making as it allows them to demonstrate their knowledge about the institutional language policy which expects them to use Luxembourgish. Therefore, according to Wei (2018), the language use of the children as well must be considered as an example of translanguaging. Drawing on Giddens' (1984) concept of structuration, children's translingual behaviour in interaction with the practitioner indicates their agency in reproducing a structured-praxis that derives from the institutionalisation of the monolingual norm in Luxembourgish, and relates to routinised language practices. This is especially highlighted following the immediate translation into Luxembourgishas a form of translanguagingon Carmen's behalf, whereas Lux-embourgish in this case represents an institutionalised language norm, particularly in the communication with the pedagogical staff members.
Third, similar to the previous transcription, the children in this instance expand their language skills by taking an active role in teaching each other new vocabulary in French during the caregiver-led activity. Due to the permissive perspective on multilingual practices on the practitioner's behalf, social interactions among peers are enhanced, which in this case leads to to a joint excercise of language teaching and learning. Alessandra actively uses her language repertoire to shape Carmen's language skills, whereas by repeating after her peer, we could assume that Carmen as a young learner savours this active language learning process.

Young children transforming institutional language policies
Even though Luxembourgish constitutes the colloquial language within most of the ECEC settings in this project's sample, the following example serves to illustrate the influence of children's language practices in shaping practitioners' language use. It provides insights that multilingualism is represented not only in regard to the diversity of language repertoires of staff and children in the day care centres, but also concerning the language practices among and with the children, in which children's agency prevails over the institutionalised monolingual structure.
Melanie, one of the educators, asks me what I have found out in regard to the languages here in the kindergartenhow many they (caregivers) would for instance speak, if two or three (languages), and probably all mixed up. I answer that mainly Luxembourgish and French are spoken, but also German, Portuguese, Italian and other languages. Melanie looks very surprised. "Really? Italian as well? And Portuguese? But I do not, do I? I actually always try to talk Luxembourgish with the children. But sometimes when they (children) answer in French, I also just keep talking in French without even realizing it. I only realise later: Oh, I am speaking French right now". I tell her that I have also often observed how one person would say something in French and another one would answer in Luxembourgish, whereas both understand each other without any problem. She then repeats again: "Yes, but it always gets so confusing" (Neumann and Seele 2014: 359; translation by the authors).
Melanie's reaction towards the confrontation of multilingual practices within the day care centre seems initially surprising to her as she states that she uses language thoughtfully, trying to reproduce the monolingual norm of promoting Luxembourgish as a colloquial language with the children. This is a rather common phenomenon within the early stages of the education system in Luxembourg, in which the promotion of the Luxembourgish language aims to form a national identity by legitimising and preserving the use of Luxembourgish within a classroom setting as the dominant language (Kirsch 2018;Neumann 2012). However, in spite of the structural power of the monolingual norm which is the central strategy to realize the official institutional language policy of promoting Luxembourgish, translingual behaviour seems nonetheless to be present. This could be explained by the familiarity with translanguaging within daily practices in the context of Luxembourgish society. In this line, children thus actively contribute to the constitution of the institutional language practices by affecting practitioners' language use based on their own linguistic repertoires. They thus undermine the institutional language policy to some extent, and, in so doing, influence the realised language ecology. Furthermore, although the educator in this instance states an unease with translingual practices that do not conform with the institutional language policy, taking children's agency into account regarding their language usein this respect as collaborators -, and thus enabling plurilingual practices within an environment of cultural and linguistic super-diversity, could eventually lead to a pressure release on the practitioner's behalf.
In the course of the last sequence we outline another form of children's agency which deals with behaviour of resistance. Although resistance is regularly overlooked in educational discourses about children's agency, for example, in the context of participation (Neumann et al. 2019), it can express a remarkably strong agency. Considering the first sequence again, in which Annabelle and Pierre are examining a picture book together and Annabelle tells him the names of the animals in French, it is interesting to see what happened in the moment the caregiver Monique entered the stage.
Pierre (2 ½ years) turns his head upwards and looks at the caregiver Monique, who is about to put various toys on a shelf. She looks to Pierre, bends down a bit and points to one of the drawings in the picture book: "Wat ass dat?" ['What´s this?', in Luxembourgish). Annabelle (2 ½ years) looks at her and answers: "C'est un lapin." ['It's a rabbit', in French). Monique points to the picture again and asks a little bit more resolutely: "Wéi heescht dat?" ('How do you call this?', in Luxembourgish). Annabelle remains silent, whereupon Monique tells her: 'So Hues!' ('Say rabbit!', in Luxembourgish). Annabelle answers again: "Un lapin" (Neumann 2014: 253; translated by the authors).
Whereas the caregiver, as in the previous examples, not only conforms to the colloquial language, which is Luxembourgish, but also insists on it, Annabelle repeatedly refuses to switch from French to Luxembourgish and insists on her part to use the French name for rabbit. In line with her tone which reveals that she is not joining the children´s role play, the caregiver Monique takes herself more the role of a strict teacher who commands the children 'to perform' in Lux-embourgish. While the other examples before show, that the use of languages other than Luxembourgish is tolerated to some extent, this sequence makes clear that there are also strong requests to perform in Luxembourgish which are addressed to the children. As the caregiver actually does not friendly explain how the animal is called in Luxembourgish but just asks to "say" it in Luxembourgish this rather points to the need to commit and to demonstrate the commitment to the language promotion goals of the institution than to a pedagogically well reflected strategy. This means, that the promotion of Luxembourgish isat least on a performative levelnot just left to chance. However, even under the conditions of resistance, the practice remains translingual, because this conflict is carried out in both languages and the two speakers follow each other in what has been said in the other language. As witnessed in Heller (2008), in this sequence it can be clearly seen that the boundaries between languages and the differences in the social positioning of the speakers are not unquestionably given but are constantly re-created and negotiated. Moreover, it is precisely the child's resistance that highlights the power of this monolingualisation strategy, which hegemonises one of the two languages, and also results in a distinctive positioning of the speakers (see also Kuhn & Neumann 2020). However, the child's agency is thus demonstrated in this case through the possibility of asserting themselves in their own language. According to Sahlins (1981), this can be understood as the actor's resistance that derives as a consequence of the collisional dynamics between cultural understandingswhich in this case can be referred to as the child's linguistic habitusand the institutional language policy (Ahearn 2001). In terms of Giddens' (1984; see also Section 2) concept of structuration, resistance as an intended action relates to agency in terms of the transformation of structure. Although the situation in this example does not illustrate the extent to which such resistance practiced by the child will result in the transformation of the institutional language policy, it is not unlikely that an accumulation of such resistance can lead to a structuration of the language ecology described in the previous example as "confused".

Final discussion
Starting from a theoretical perspective on children's agency, the purpose of this article was to analyse the extent to which: (1) the children act upon the institutional language policy and alternate their language use within their repertoires; (2) the children actively contribute to the institutional language ecology. The main objective was to show how language practices at an institutional level can be influenced by language practices through children's agency.
Our findings offered various insights into children's agency in regard to their contribution to the constitution of the language practices within the day care centres in Luxembourg: In terms of the first question, which focused on children's language use in relation to the structural language environment, it appears that our findings indicate that the children especially align with the institutionalised language policy in communication with the practitioners. Although this might not be surprising at first sight, moving beyond the data and bringing in the context under study, this result should not be taken for granted and deserves to receive some attention. Given that both children and practitioners encompass a multilingual repertoire and could also communicate in a language other than Luxembourgish, and keeping in mind that there is no state regulation that determines a monolingual policy in ECEC services, aligning with the practitioners' language practicesas, for instance, in our example through switching from French to Luxembourgishshows that children act upon the institutionalised language policy. Despite the multilingual setting, an underlying monolingual norm of language use drives children to alternate their language to Luxembourgish in communication with the practitioners as a sort of strategy of language policy management in which language alternation is used as a means to align in interactional roles . As for this instance, the structure frames the social roles and determines the legitimised language for communication within a specific setting and interaction. Hence, this finding can be regarded as exemplary for children's agency in contributing to the maintenance of a monolingual norm. This brings us to the second question, which looks at children's contribution to the plurilingual institutional language ecology with their own linguistic repertoires.
Similar to the research of Boyd, Huss, and Ottesjö (2017) on children's agency in creating and maintaining language policy in practice in a bilingual preschool environment in Sweden, our examples show that the young learners not only conform their multilingual language practices to an institutionalised monolingual norm within the ECEC settings, but also contribute to the constitution of those as actors. This becomes chiefly visible on the level of language practices realised by children in the everyday life of the centres. One of our findings highlights children's agency in peer interactions as resources for their language development (see also Chaparro-Moreno et al. 2019); they contribute to the institution´s language practices by taking on a teaching role, drawing on a shared or even distinct L1 language. This mechanism seems to be well-explained in research, noting that children usually alternate to L1 within peer interactions as a preferred language for socialisation, whereas L2 (as a colloquial or instructional language) is mostly used for institutional means (Morton and Evnitskaya 2018). Furthermore, although research on children's agency mostly focuses on free play situations, our findings provide in-sights that this turn-taking role of "young peer teachers" may not be exclusive to child-initiated, play-based activities, but is also present in educator-led activities. Moreover, in regard to the practitioners' language use, our results show that in spite of a monolingual norm, children's language practices based on their own linguistic repertoires seem to have a direct or an indirect impact on the practiced institutional language policy by establishing an environment of translanguaging. This claim can be associated with the first argument presented in the previous chapter, where not only children align with practitioners' language practices, but also vice-versa. Additionally, children's acts of refusing to alternate their language use and, so to speak, to not align to the institution's colloquial language provide an even stronger indicator of children's agency and their productive participation in shaping the language practices at an institutional level. Nonetheless, our study also highlights the importance of practitioners' attitudes on multilingualism as a potential factor for an active performance of children's agency in terms of multilingual practices within the ECEC settings.
Focusing on children's agency in terms of language use in day care, similar to the research of Bergroth and Palviainen (2017) on agency of bilingual children which however analyses children´s linguistic practices in regard to an official language and educational policy in ECEC settings, our findings demonstrate that agency manifests itself in both reproducing and undermining the institutional language policy of the centres. Within a structure-agency perspective, we understand children's agency in translanguaging thus as the engagement in practices of meaning and sense-making that are characterised by the flexible use of different languages (Morton and Evnitskaya 2018), and derive from their linguistic repertoires (Wei 2018) in relation to the structural settings/activities. From a practical perspective, this means that practitioners' language use is being shaped by the children's agency on the one hand, whiledespite of the children not being explicitly informed about the monolingual norm in Luxembourgishchildren's language practices are being framed by the practitioners' monolingual language use on the other, which in turn is influenced by the structured language policy.
Recent studies (Jaspers and Verschueren 2011; Kirsch 2018; Schwartz 2018) have critically discussed some shortcomings of a monolingual norm of strict language separation within an existing heteroglossia, claiming it to narrow students' linguistic and cultural resources as well as to generate effects of social exclusion for some of them (Neumann 2012). In this line, the presented results illustrate the influence of children's as well as practitioners' use of translanguaging on the institution's everyday language environment because translanguaging functions as a means to transform official institutional language policies into a situationally enacted and practiced language policy on the level of everyday communication. If, as the results of our paper show, children are both actors of maintaining and transforming institutional language policies and practices, this should be carefully considered during future language planning in ECEC institutions.
Regarding the current educational ambitions of establishing a plurilingual policy in day care facilities for preschool-age children in Luxembourg, translanguaging should be seen as one of the key strategies both to create a plurilingual language environment in and through everyday practice in the day care centres as well as to recognise children's agency. Such a strategy of flexible language use has recently come to attention within discussions about translanguaging as a pedagogical concept (Vogel and García 2017), which is currently being debated in distinct bi-and multilingual contexts (Morton and Evnitskaya 2018). In line with this, the chance of achieving such a goal very much depends on the freedom and possibilities of the children to bring in their own linguistic repertoires by themselves. This article thus suggests a more permissive and inclusive language policy that encourages and values children's agency and flexible use of their linguistic repertoires. As such, challenges of language learning, participation, inclusion and social justice could be interrelated (García and Kleifgen 2018;Hélot 2018). Accordingly, with this paper we aim to further the acknowledgment of children's roles in language policy enactments, a theme which seems to be relatively under-researched. Finally, we argue for a notion of language policy being not only a top-down but also a bottom-up process, even within the ECEC sector.