Practitioners’ language-supporting strategies in multilingual ECE institutions in Luxembourg

ABSTRACT Multilingual education is mandatory in early childhood education in several European countries. Scholars working in first, second and foreign language learning have shown the effectiveness of interaction-promoting and language-modelling strategies for language development. In addition, teachers in bilingual contexts have been translanguaging to foster language learning. To this day, few research studies have examined these strategies in combination and their deployment in multilingual contexts. The present study takes place in multilingual Luxembourg where 64% of the four-year-olds do not speak Luxembourgish. It focuses on three practitioners who took part in a professional development course on multilingual pedagogies and presents their language-supporting strategies as well as their translanguaging practices. The findings show that the practitioners use a range of strategies in Luxembourgish, French and home languages, in daily conversations, routines and language and literacy activities to address children’s linguistic needs.


Introduction
Early childhood education (ECE) has been described as an 'essential foundation for successful learning, social integration, personal development and later employability' by the European Commission (2011, 1). ECE institutions fulfil this function, among other ways, by acknowledging children's diverse linguistic repertoires and helping them improve skills in institutional languages and home languages, where different. Over the last decades, the language diversity of children in ECE increased as a result of migration and globalization and, as a consequence, the Council of Europe has promoted multilingual education in ECE. Multilingual programmes tend to focus on the development of language awareness, multilingual competence as well as linguistic and cultural tolerance, key factors in developing responsible citizens. Several European countries such as Germany (in some states), Switzerland or, recently, Luxembourg have implemented such programmes; other countries have not yet taken up the call. Research suggests that practitioners remain unsure of how to deal with language diversity, provide language support or develop language skills (Michel and Kuiken 2014). There is, however, a wealth of literature on effective language-supporting strategies in the field of second language acquisition and from intervention studies with monolingual and bilingual learners (Bond and Wasik 2009;Girolametto and Weitzman 2002;Hamre et al. 2012). These techniques include interaction-promoting strategies such as asking questions and languagemodelling strategies such as providing corrective feedback. ECE practitioners in Luxembourg have received professional development (PD) training on multilingual pedagogies including the use of these strategies (e.g. . The present article presents the language-supporting strategies as well as the language use of ECE practitioners in Luxembourg following a PD course. Luxembourgish, French and German are the official languages of this small country bordering Belgium, France and Germany. However, many more are spoken on account of the high number of immigrant residents. In 2019, 47.5% of the residents did not have Luxembourgish citizenship (STATEC 2020) and in the academic year 2016/ 2017, 64% of the four-year-olds did not speak Luxembourgish at home (MENJE 2020). While multilingualism is a social reality, monolingualism was still the aim of the early years' curriculum until 2017, at the time this study was conducted. Teachers were required to develop children's skills in Luxembourgish, perceived as the language of integration and the bridge to becoming literate in German in Year 1. To help practitioners in the formal and non-formal education sectors capitalize on children's multilingual resources, Kirsch designed the research project 'Developing multilingual pedagogies in early childhood education' (MuLiPEC). It was based on a socio-cultural perspective on language and learning and aimed to develop practitioners' knowledge, skills and practices related to multilingual approaches through PD. Particular attention was given to literacy activities, language-supportive strategies and translanguaging, that is the deployment of a person's entire semiotic repertoire to communicate (García and Otheguy 2020). This article sheds light on translanguaging and the language-supporting strategies of two teachers and one educator who worked in a preschool and an éducation précoce ('early education') during the academic year 2016/2017. The results show that the practitioners used a range of strategies to address children's linguistic needs in several languages throughout the day . The findings are important for researchers and teachers interested in implementing multilingual pedagogies and developing children's linguistic repertoires in multilingual contexts.

Developing languages in ECE: effective activities and strategies
Scholars working from cognitive and socio-cultural perspectives agree that children learn languages when they receive appropriate comprehensible input and use languages with adults and peers in interactive ways which may include actions, sensual experiences and culture-specific objects. For children to learn a language, they need to be cognitively, socially and emotionally involved and supported by more knowledgeable people who model and monitor language use (Swain, Kinnear, and Steinman 2010). To promote language learning, teachers need to create a stimulating linguistically and culturally sensitive learning environment that provides children with a range of opportunities to receive linguistic input and encourages them to talk. The following three sections review effective language and literacy activities and strategies.

Ensuring interactions in conversations and activities
In ECE contexts, teachers can provide input through face-to-face interactions in daily conversations as well as through language and literacy activities. Scholars agree that conversations are a key method to develop children's oral competence in preschool based on evidence of correlations between the children's vocabulary, the adults' language use and the quality of their talk (Bond and Wasik 2009;Dickinson et al. 2003). The teachers' use of language appears to vary with the learning context. In settings where young children develop their first language or learn an early foreign language, teachers have been found to use advanced language and a varied vocabulary (Bond and Wasik 2009;Edelenbos, Johnstone, and Kubanek 2006), whereas teachers in bilingual settings have used simple and repetitive language (Mifsud and Vella 2018). Classroom talk often occurs in routine activities where children use simple and contextualized language (Dunn, Beach, and Kontos 1994). Practitioners in bilingual classrooms (Castro et al. 2011a) and those teaching early foreign languages (Andúgar and Cortina-Peréz 2018;Edelenbos, Johnstone, and Kubanek 2006) can offer playful but structured learning environments with routines or circle time.
In order to develop language skills, teachers may also plan a range of language activities within a holistic, play-based and child-centred curriculum where they explicitly teach vocabulary through labelling words and then practise these in activities such as memory games (Alstad and Tkachenko 2018;Andúgar and Cortina-Peréz 2018). Dickinson and Smith (1994) found that preschool-aged children in a Head Start Programme rarely engaged in cognitively challenging tasks such as analysing or summarizing although these tasks, which require higher order thinking, promote vocabulary learning.
Finally, teachers may organize literacy activities such as sharing books. Book reading increases children's vocabulary and develops their oral language skills, particularly if teachers interact with children, for instance, through asking questions, expanding their utterances or organizing meaningful extension activities promoting language use (Barnes, Dickinson, and Grifenhagen 2017;Wasik, Hindman, and Snell 2016;Whitehurst and Lonigan 2001).
Interaction-promoting and language-modelling strategies Girolametto and Weitzman (2002) identified three types of conversational strategies that adults can use to promote language development: child-centred strategies (e.g. following the child's lead), interaction-promoting strategies (e.g. asking questions) and languagemodelling strategies (e.g. expanding utterances). The results of intervention studies with early years teachers in the United States and Australia (Girolametto, Weitzman, and Greenberg 2004;Hamre et al. 2012) and with teachers of second language learners in Germany (Simon and Sachse 2013) showed that teachers in the experimental group talked more to the children after their training than those of the control group and that children produced more language, developed their vocabulary, and engaged in more conversations with peers. Amongst the wide range of terms used to discuss these strategies in academic journals, this article has opted for 'interaction-promoting' and 'language-modelling'.
Interaction-promoting strategies encourage children to interact with adults and peers using more than one-word answers and to think about the language being used. Adults may, for example, facilitate turn-taking (Girolametto and Weitzman 2002) and encourage peer interactions (Castro, Espinosa, and Páez 2011b). The use of questions is the most widely researched type of these strategies (Dickinson, Darrow, and Tinubu 2008;Girolametto, Weitzman, and Greenberg 2004;Wasik, Hindman, and Snell 2016;Whitehurst and Lonigan 2001). Educators distinguish between open questions, closed questions and gap-fill-questions where children add words, agree or disagree. Other interaction-promoting strategies include repeating children's utterances (Hamre et al. 2012;Simon and Sachse 2013), giving information and offering explanations (Barnes, Dickinson, and Grifenhagen 2017;Girolametto, Weitzman, and Greenberg 2004). Finally, professionals promote talk if they help children use their limited language creatively such as through translanguaging (Andúgar and Cortina-Peréz 2018;Rowe 2018).
Language-modelling strategies build on children's utterances and transform these. This happens when adults provide feedback on the content and language use of children, thereby modelling sophisticated language and engaging children in high-quality dialogue (Bond and Wasik 2009). Two main types of strategies have been identified: expanding and providing corrective feedback. Expanding means rephrasing the children's utterance using a richer vocabulary, a more complex syntax or greater detail (Barnes, Dickinson, and Grifenhagen 2017). There are many different corrective strategies including recasts, implicit or explicit correction or metalinguistic information (Ellis and Sheen 2006). Practitioners may offer implicit corrective feedback by reformulating an incorrect sentence (Simon and Sachse 2013;Wasik, Bond, and Hindman 2008).

Translanguaging
Based on the analysis of 31 studies on interactions between teachers and multilingual children, Langeloo et al. (2019) show that drawing on children's home languages and cultural backgrounds is one effective means of offering emotional and instructional support. However, home languages other than school languages have had little space in the curriculum for ideological reasons among others. In bilingual contexts, home and institutional languages have often been separated. Arguing against this tradition, the 'theory of translanguaging' holds that languages are not separate and autonomous systems but part of one unique repertoire from which speakers select and combine features flexibly and dynamically (García and Otheguy 2020;Wei 2018).
Translanguaging has been conceptualized, among others, as a practice and a strategy. 1 In the present article, we understand translanguaging as the deployment of a person's entire semiotic repertoire. Studies in bilingual and multilingual ECE settings have tended to focus on the linguistic repertoire, showing that teachers and children use features of several languages flexibly and dynamically (Gort and Sembiante 2015;Mary and Young 2017;Palviainen et al. 2016). More rarely have researchers examined the deployment of extralinguistic resources such as mime, gestures or body movements (Gort and Pontier 2013;Mifsud and Vella 2018) and paralinguistic resources such as tone of voice, articulation and rate of speech (Gort and Pontier 2013;Mifsud and Vella 2018;Simon and Sachse 2013).
The review of 29 qualitative translanguaging studies in ECE by Pontier, Boruchowski, and Olivo (2020) reveals that teachers and children translanguaged in a range of typical ECE contexts such as play, show-and-tell, read alouds and other language activities. The teachers' unplanned and strategic shifts between languages facilitated children's communication (Garrity, Aquino-Sterling, and Day 2015;Kirsch 2018), promoted their meaning-making and learning (García 2011;Gort and Pontier 2013;Kirsch 2018;Palviainen et al. 2016) and supported their socio-emotional development and multilingual identities (Mary and Young 2017). Only one of the reviewed studies explicitly addressed the teachers' language-supportive strategies in translingual interactions. Gort, Pontier, and Sembiante (2012) explored the ways in which teachers used questions in English and Spanish read alouds and found that questions were inequitably distributed. While teachers asked different types of questions in both languages, the questions in the English sessions created more opportunities for extended dialogues.
In sum, this literature review has brought together typical ECE language and literacy activities that promote teacher-child interaction, language-supportive strategies and translanguaging. While most studies have been carried out in monolingual or bilingual contexts, few if any have analysed these strategies in multilingual settings and in combination with translanguaging. The present study addresses these gaps by asking: . In which situations and how often do the three practitioners in Luxembourg translanguage? . What interaction-promoting and language-modelling strategies do they use when interacting with children? . In what ways does the strategy use differ across the two classes and languages?

Methodology
The data from this article are drawn from the project MuLiPEC (2016-2019) which aimed to help practitioners develop multilingual pedagogies to capitalize on children's varied linguistic resources and develop their language repertoires. Forty-six practitioners who worked in the formal education sector (e.g. schools) and the non-formal sector (e.g. day-care centres) enrolled on a 15 h PD course in May/June 2016 designed for this purpose. Practitioners then self-selected to continue the PD during the academic year 2016/2017. Based on the size, location and type of institution, the researchers selected seven participants, three of which worked in the formal sector. They were observed, coached and took part in six network meetings. The aims of this long-term PD path were to deepen the practitioners' understanding of multilingualism and language learning, and help them implement activities in Luxembourgish, French and children's other home languages. The topics covered included perspectives on multilingualism, translanguaging, theories of language learning, literacy activities and language-supporting strategies . 2 The present article focuses on the three practitioners who worked in the formal education sector.

Participants
Two of the participants taught in an éducation précoce ('early education') and one in a preschool. Preschool is compulsory and caters to four-to-six-year-olds; early education is non-compulsory and caters to three-year-olds. Qualified teachers work in preschools while a qualified teacher co-teaches with a qualified educator in early education. The professionals follow the national curriculum which, at the time, focused on the development of skills in Luxembourgish. 3 The participants were the preschool teacher Ms Vivian 4 and the early education practitioners Ms Clara (teacher) and Ms Jane (educator). All three spoke Luxembourgish, German, French and English. Ms Vivian also spoke Spanish, Ms Clara Portuguese and Ms Jane acquired some Portuguese during the course of the study. The children came from diverse backgrounds. Thirteen home languages were counted among the 18 preschool children and five among the 11 children in early education. In September 2016, all but one of the preschool children could speak some Luxembourgish but none of the three-year-olds could. Some background information about the children can be found in the Appendix.

Data collection and analysis
The article draws its data from 80 video-recorded observations made over 32 visits by the principal researcher (Kirsch), a research associate and the doctoral candidate Simone Mortini. The video-recordings lasted between 2 and 20 min each and showed practitioner-child interactions in daily conversations during creative activities (e.g. art work), routines and circle time, and language and literacy activities (e.g. storytelling, memory games). Table 1 provides an overview of the number and types of videorecorded activities in each classroom.
All video-recordings were transcribed and descriptions of relevant non-verbal behaviour were added. In a first stage, the video-recordings were classified in relation to the adults' use of languages. Then, the percentage of multilingual activities where adults translanguaged was calculated. In a second stage, working inductively and deductively, the various types of interaction-promoting and language-modelling strategies were coded in all 80 activities (Girolametto, Weitzman, and Greenberg 2004;Hamre et al. 2012). In this article, the most frequent strategies are presented. Next, all strategies were added to a table representing the different routines and planned language and literacy activities. Each time an adult deployed a particular strategy, the instance was listed. In this way, the frequency of the strategies across the types of activities could be established. For example, Ms Vivian used open questions in 4 rituals, 7 creative and 13 language and literacy activities (e.g. 4 storytelling events, 9 language activities). In other words, she used them in 24 out of 40 or 60% of the recorded activities. Percentages were calculated to facilitate comparisons across classes. In addition to the observations, the practitioners had participated in 10 semi-structured interviews. While this article does not draw on these data which had been analysed thematically (Braun and Clarke 2006), it nevertheless presents several relevant quotes from four interviews to illustrate some findings.

Ethics
The project was based on democratic values and the researchers treated participants with respect and were sensitive to their linguistic and cultural backgrounds. Special care was needed as they worked with young children. Prior to the data collection, the research project was validated by the Ethics Review Panel of the University of Luxembourg and the National Commission for Data Protection. 5 Furthermore, it complied with the General Data Protection Regulation. The voluntary participation was unremunerated and participants had the following rights: information, interruption, access, rectification, objection, erasure, restriction of processing and data portability. Once informed consent was received, data were collected. All personal data were processed securely and confidentially. All names are pseudonyms. The data are stored on password-protected servers of the University of Luxembourg and will be deleted 10 years after the end of the project.

Findings and discussion
The practitioners in both classes created a language and literacy-rich environment through routines, reading books, telling stories, planned language activities and spontaneous conversations where children were given ample opportunities to talk and interact with each other. As described in greater detail in , translanguaging was a legitimate practice in both classes. The flexible language use of the practitioners and the language-supporting strategies will be presented next.

Translanguaging practices
The three practitioners used Luxembourgish, French, and at times the children's home languages, flexibly and dynamically throughout the day, independently of the type of activity observed (Kirsch 2020 Ms Vivian displayed pictures of a story she had narrated the previous day in Luxembourgish and encouraged the class to retell the story. The children participated using the language they were most confident in. Greg, a Spanish-speaking boy who had learned French in a day-care centre and Luxemburgish in early education, communicated in French and Luxembourgish (lines 2, 4, 7). He was in the early stages of acquiring Luxembourgish, a language he had refused to speak in early education. David, who spoke Portuguese at home and had developed some Luxembourgish in early education, tried in Luxembourgish (line 11). The children, like the teacher, made ample use of gestures. Ms Vivian listened attentively to children, observed their movements and accepted all answers irrespective of the language. She enacted her bilingual identity and translated from French into Luxembourgish. She offered the word 'hurt' in Luxembourgish (line 5) and translated the word 'forest' from French (line 8). Furthermore, she deployed interaction-promoting strategies in lines 1, 3, 5 and 10 (i.e. asking questions, encouraging listening) and language-modelling strategies in lines 8 and 12 (e.g. expanding, giving corrective feedback).
The two other practitioners, Ms Clara and Ms Jane, also let children express themselves multimodally and in the language of their choice. All three adults opened translanguaging spaces across activities and throughout the day (see Kirsch 2020;Kirsch and Seele 2020 for examples). They translated children's utterances and translanguaged to provide input, ensure comprehension and promote dialogue. In a storytelling activity in February, for example, Ms Clara said to the class: 'Dat sinn Comic, Comic mat deenen super-heróis, mat allen deenen Superhelder'. [These are comics, comics with these superheroes, with all these superheroes] (observation, 6 February 2017). She spoke Luxembourgish but mentioned the word superheroes first in Portuguese (in italics) and then in Luxembourgish while pointing to pictures. During an art activity on 20 September 2016, Ms Clara asked children to put on their aprons in Luxembourgish. When the three-year-olds did not understand, she repeated the word 'apron' in French or Portuguese according to the child's language background. This was their usual way of teaching new words and expressions. Ms Clara and Ms Jane developed knowledge of new words in 48% of the observed recordings and repeated words in 55% of them while Ms Vivian did so in 33% and repeated expressions in 40% of the video-recordings. These percentages show that the three-year-olds were introduced to more new words than the four-year-olds.
To encourage talk, the practitioners commented in Luxembourgish, French, Spanish or Portuguese when they deemed it helpful. The percentage of the video-recorded activities where the practitioners translanguaged was high: 59% in the preschool and 65% in early education (see also Kirsch et al. 2020, 13). As Ms Clara explained: 'It's our everyday life, saying things in their language and explaining' (Interview, 9 September 2016). Speaking about multimodal communication, Ms Jane mentioned that she communicated with words, mime and gestures 'because the children reacted better and learned more' (interview, 27 September 2017).
The practitioners in this study acted similar to practitioners in studies undertaken in the United States (García 2011;Gort and Pontier 2013;Gort and Sembiante 2015), in Finland and Israel (Palviainen et al. 2016;Schwartz and Gorbatt 2016) as well as in France (Mary and Young 2017) in that they were responsive to children's linguistic needs, moved flexibly between languages and used visualisation strategies including mime, gesture, movement, objects and pictures.

Interaction-promoting strategies
All practitioners understood the importance of interactions as seen in the excerpts below.
'It is important that we always have interactions. During storytelling too, as one may tend to have children listen (only).' (Interview Ms Jane, 30 April 2017) 'We have to involve them and give them the feeling that they are able to talk. And we have to try and understand them.' (Interview Ms Clara, 27 September 2017) The most widely used strategies to stimulate talk were questions as seen in Excerpt 1 (lines 1, 3, 5). Ms Vivian explained: 'I have become more aware of the ways I speak and ask questions. I realize now when I fall into my habit of yes-no-questions. I know that I need to let children speak more' (Interview, 14 September 2017). Excerpt 2 provides an example of the use of questions in a multilingual conversation between Ms Jane and a Portuguese child.
Excerpt 2. Use of questions in a literacy activity in the précoce (Observation, 3 February 2017) Ms Jane asked a closed question (line 1) and a confirmation question (line 3) following a child's answer in Portuguese (line 2). Ms Jane's confirmation question repeats the child's utterance in question form but translated into Luxembourgish. In this way, the question provides input in the target language, ensures that children who did not speak Portuguese understand the child's answer, and encourages all children to think and participate. One strategy fulfils a range of functions.
Closed questions were used more frequently than open questions and clarification questions in line with research findings (Dickinson, Darrow, and Tinubu 2008;Girolametto, Weitzman, and Greenberg 2004;Wasik, Bond, and Hindman 2008). The particular order of the questions is related to the children's language competence: closed questions demand less linguistically complex responses. Clarification questions as well as open questions compel children to think and express themselves more fully. This may explain why Ms Vivian, who worked with the older children in the preschool, asked more open and fewer closed questions than Ms Clara and Ms Jane (see Table 3).
Apart from asking questions, the practitioners tended to confirm that they attentively listened to children, either by nodding, uttering the word 'yes' or repeating parts of their own utterances or those of the children (see Excerpts 1, 2). In this sense, they were highly response to the children (Barnes, Dickinson, and Grifenhagen 2017) and acted similar to the professionals found in Hamre et al. (2012) and Simon and Sachse (2013). Ms Vivian was found to repeat children's utterances slightly more often than the practitioners in the précoce (Table 2). This may be the case because the older children already spoke more Luxembourgish than the younger ones and the repetition of their utterances provided additional input. None of the practitioners tended to encourage peer interaction at the beginning of the school year, although this strategy promotes language development (Castro, Espinosa, and Páez 2011b;Girolametto, Weitzman, and Greenberg 2004;Schwartz and Gorbatt 2016). Finally, all professionals encouraged children to speak using their entire linguistic repertoire (Pontier, Boruchowski, and Olivo 2020) and children translanguaged as shown in Excerpts 1 and 2.

Language-modelling strategies
All practitioners pronounced clearly and reformulated children's utterances by expanding them (Barnes, Dickinson, and Grifenhagen 2017) (see Excerpt 1, line 8). All three practitioners also rephrased their own utterances using a more varied vocabulary (Edelenbos, Johnstone, and Kubanek 2006;Bond and Wasik 2009). They adapted the vocabulary to the linguistic needs of the children and spoke in short sentences, particularly so the practitioners in the précoce. Excerpt 3 from a memory game shows how Ms Jane expanded on her short utterances by repeating words and adding them into one sentence. In this example, she explained the difference between two memory cards while pointing to the reindeers' faces.
Excerpt 3. Expanding in a language activity in the précoce (Observation, 12 December 2016) The three practitioners also gave corrective feedback (Hamre et al. 2012;Simon and Sachse 2013;Wasik, Hindman, and Snell 2016) (see Excerpt 1, line 12). Another example is shown in Excerpt 4, where Ms Vivian and the children were speaking about the fox in the Luxembourgish version of 'The Gruffalo' (Donaldson and Scheffer 2010 When a child mistakenly referred to a 'wolf' in French (line 1), Ms Vivian corrected in an explicit manner and explained in French (italics) that the animal was not a wolf but a fox (line 2). This explanation was useful as not all four-year-olds may know the difference. Once the child had repeated the correct word, Ms Vivian confirmed, nodding, repeated the information, this time in Luxembourgish, and the child repeated the word in Luxembourgish. This example shows the interplay of interaction-promoting and language-modelling strategies across languages.
As with the interaction-promoting strategies, there are differences in the practitioners' frequency of strategy use, which seemed once again to be related to the children's language competence. Ms Vivian used more corrective feedback and elaborated more frequently on children's utterances than Ms Clara and Ms Jane who used simpler language with the children who were in earlier stages of their language development (Excerpts 1-4). Finally, as before, language-modelling strategies were used both in multilingual conversations and the rarer monolingual ones.

Conclusion
This study has shown that three practitioners in ECE in Luxembourg translanguaged in a range of contexts (e.g. conversations, daily routines, language and literacy activities) similar to other ECE practitioners (Pontier, Boruchowski, and Olivo 2020). They thereby used several interaction-promoting and language-modelling strategies in Luxembourgish, French and the children's home languages. The use of similar strategies has been reported in contexts where second or early foreign languages were taught (Alstad and Tkachenko 2018;Andúgar and Cortina-Peréz 2018;Edelenbos, Johnstone, and Kubanek 2006;Mifsud and Vella 2018). However, these teachers did not use languages as flexibly as the practitioners in Luxembourg who frequently translanguaged (Kirsch 2020). The language-supportive strategies encouraged children to respond in a language of their choice (e.g. French and Portuguese in Excerpts 1, 2, 4) and enabled the practitioners to give children input in Luxembourgish (for instance, by translating words). Furthermore, the strategies are likely to promote language development. The practitioners in Luxembourg frequently elaborated on utterances and offered corrective feedback unlike the 52 Head Start teachers reported in Barnes, Dickinson, and Grifenhagen (2017) who seldom expanded statements. According to Barnes, Dickinson, and Grifenhagen (2017), these responsive strategies promote receptive vocabulary growth.  The practitioners' strategy use (in any language) depended on children's language background and linguistic needs (Mary and Young 2017;Schwartz and Gorbatt 2016), which demonstrates that the practitioners scaffolded the learning process (Swain, Kinnear, and Steinman 2010). Ms Vivian used longer sentences, choose interaction-promoting strategies that required more thinking, and gave more corrective feedback compared to Ms Jane and Ms Clara. Their simpler input was well adapted to three-year-olds who did not speak Luxembourgish at the start of the school year. The observations highlight both the interplay of various types of strategies and the manner the strategies were orchestrated to promote the development of multiple languages. This, in turn, highlights the practitioners' skills and knowledge which were based on a sound understanding of language learning (Swain, Kinnear, and Steinman 2010).
While the main limitations of the empirical part of the study arise from the small number of participants, the findings, nevertheless, contribute to the field of language learning because the research project examines strategy use and translanguaging in combination, across two age groups in ECE settings, and in a multilingual country. Furthermore, it brings together literature from first, second and foreign language learning.
Owing to the amount of natural data collected over one year and the theoretical underpinning of the paper, some recommendations can be made. At times of increasing linguistic diversity in schools, it is important that future practitioners develop a good understanding of theories of language learning as well as appropriate methods, language and literacy activities and language-supportive strategies, including the encouragement of peer interaction which was little used in the present study. Developing an understanding of the ways in which the teachers' and children's semiotic repertoire and languagesupportive strategies promote language development should be part of the teacher education curriculum and feature in PD courses. Future research studies could focus on strategic (planned and reflected) translanguaging, an aspect not explored in the present study, and on the exact relationship between translanguaging, language-supportive strategies and language development.
Multilingual education has become a reality. It is hoped that this study will encourage ECE practitioners to open up to translanguaging and reflect on their use of strategies to leverage children's semiotic repertoire for learning.

Notes
1. See Jaspers (2018) for an overview of different definitions. 2. Like the training sessions of Hamre et al. (2012) or Simon and Sachse (2013), the MuLiPEC training was collaborative and inquiry-based. Practitioners were asked to carry out activities, video-record them and discuss the recordings in groups. These discussions dominated both the initial 15 h course and the network meetings. Observations, interviews and a questionnaire were deployed to identify the influence of the PD on the practitioners' skills, knowledge and practices Kirsch 2020). 3. An overview of the complex ECE system in Luxembourg is presented in Kirsch and Seele (2020). 4. In the early years, children tend to call the practitioners either by their first name or by Ms and the first name. 5. The Ethics Review Panel of the University of Luxembourg and the National Commission for Data Protection validated the research project under the references ERP-16-003 MuLiPEC and TO11002.