Narrative Analysis as a Means of Investigating CLIL Teachers’ Meaningful Experiences

An ever-increasing need for a bilingual education in globalized societies have set new challenges for all stakeholders from ideological (monoglossic vs heteroglossic) as well as methodological perspectives. Teachers’ persistent interest in different forms of bilingual education has attracted us to explore the potential of Content and Language Integrated Learning (CLIL) as a means of attaining a bilingual education in the second decade of the 21st century, especially the professional development of teachers who work in the given context. In this study, narrative analysis is employed to investigate how teachers’ explicit meaningful experiences lead a teacher to become a CLIL teacher in the Estonian educational settings, and disclose the factors shaping this process. The results reveal a variation in the teachers’ meaningful experiences driven mostly by their context – the type of bilingual program, the status of the foreign language, school support for collaborative practices as well as a variation in the belief of what constitutes CLIL views on languages and personal pedagogical beliefs.

ilingual education has become an increasingly common and "natural" part of mainstream secondary education in many countries (Ritchie, 2012). But learning in another language is a much more complex cognitive, social and cultural process than just a translation of the subject content into another language (van Kampen et al., 2018), and thus needs more attention and awareness of all stakeholders involved in the process. Content and Language Integrated Learning (CLIL) is one of the most popular forms of bilingual education which, in varying degrees, is explicitly present in many monolingual contexts (Bravo-Sotelo, 2020;Cinganotto, 2016;Goris et al., 2019;Rumlich, 2020). Still, there is little empirical evidence of its construct and effectiveness and the discussion is often shrouded by misconceptions and misunderstanding (Pérez Cañado, 2020).
Research on CLIL in Europe clearly displays polarization. In CLIL related research, both educational and sociolinguistic praise (Llinares et al., 2012) and criticism (Paran, 2013) can be found. Supporting voices, seeing CLIL offering an advantage over non-CLIL to learners (Dale et al., 2010;Dalton-Puffer, 2011;Nikula et al., 2013), emerged while the European transnational ideological framework was evolving together with local political support (Mehisto et al., 2008). In the 2010s, the critics took over the discussion calling for more evidence which would show the effectiveness of CLIL (Bruton, 2013;Piesche et al., 2016). Today, both defenders and counterpart rivals agree upon the need for additional objective empirical research on the subject (Pérez Cañado, 2020).
Involving 3000 variables (Mehisto et al., 2008), mobilizing many stakeholders -learners, teachers, parents, educational institutions, the state and international communities (Baetens Beardsmore, 2009) -bilingual education is very context rooted. In Europe, many countries have longstanding successful CLIL programs (Austria, Germany, Finland, Spain, Netherlands, Italy) with a specific setting to each country. For example, Goris et al. (2019) highlight the differences between the Spanish CLIL implementation model (top-down initiatives with fairly optimistic results) and CLIL in the Northern Europe countries (bottom-up initiatives with less optimism about the CLIL approach). For a comprehensive overview of CLIL in different bilingual educational systems see Mehisto and Genesee (2015).

B
In Estonia, research has been conducted on the essence and differences between the national vs minority language education (Masso & Soll, 2014), but little research is available on CLIL involving a foreign language (Dvorjaninova & Alas, 2018). Historically, the medium of instruction in CLIL for the predominantly Estonian/Russian students here has been either English, German or French. Although, the teaching tradition is long, there are no teacher-education programs that focus specifically on CLIL teacher training. In fact, subject teachers and language teachers are trained separately. We know very little about how some language teachers end up teaching a subject in a foreign language or how a subject teacher starts teaching his/her subject in a foreign language, i.e., how they become CLIL teachers, and what being a CLIL teacher means to them. International research on CLIL teaching careers is scarce. This study is the first phase in a design-based research looking for a design for a teacher training programme supporting bi-or multilingual education in its broadest sense. Investigating teachers' meaningful experiences through their self-reflection the study aims to disclose how CLIL teachers perceive their working environment (Meristo et al., 2013), and how the environment has shaped them. Teachers' meaningful experiences are considered to be one of the key factors shaping teachers' professional agency and collaborative practices (Green & Pappa, 2020;Orland-Barak, 2017;Senge et al., 2014;Slabina & Aava, 2019). Investigating their meaningful experiences facilitates understanding how to support CLIL teachers' professional agency at school in a monolingual context. In order to understand the CLIL teachers working environment it is necessary to discuss the bilingual ecology of the context.

Bilingual Ecology
At the beginning of the 21st century, bilingual ecology can be described either from the perspective of monoglossic or heteroglossic ideologies, mobilizing different theoretical frameworks of bilingual education respectively (García, 2009). Traditional models of bilingualism, such as subtractive and additive, have a monoglossic tradition, whilst recursive and dynamic models conceptualized in the 21st century have emerged from Bakhtin's (1981) heteroglossic ideology. The latter, encompassing CLIL among other types of education, considers multiple language practices in interrelationship (García, 2009). The main argument advocating the need for new models of bilingualism comes from the understanding that bilingualism cannot be seen as a mathematical addition or subtraction of languages, but has to be viewed as a norm of language practices, such as translanguaging. Translanguaging is seen here in a broader sense of sociolinguistic activities encompassing the idea of code-switching, mixing and altering between languages and going beyond it by implementing a variety of bilingual communicative strategies such as translation and mediation (Nikula & Moore, 2019). Bilingualism is "plural, mixing different aspects of language behaviour as they are needed, to be socially meaningful" (García, 2009, p. 48). Although globally the understanding of bilingualism has shifted from monoglossic to heteroglossic, it has not yet been adopted in all contexts (García, 2009, p. 116) and by all levels of stakeholders. Some countries still have a predominantly monoglossic approach to bilingualism, concentrating on the development of language proficiency from the point of view of the dominant language (cf. Blackledge & Creese, 2010).
Emerging in Europe in 1994(Marsh et al., 2001, CLIL can be viewed alongside a variety of practices around the world where a foreign language is used to teach a subject (Graddol, 2006). Practitioners (Ball et al., 2015;Paran, 2013) distinguish between 'hard' or 'strong' CLIL and 'soft' or 'weak' CLIL. 'Hard' CLIL denotes contexts where the full instructional focus is on the curricular subject taught by a subject teacher and the medium of instruction is an additional language. 'Soft' CLIL has the emphasis on language development with just a part of the subject curriculum (usually delivered in L1) taught using an additional language. The nature of the coursehard or softwill probably set professional requirements to the teacher in the given context. Recent studies, however, emphasise the integrated nature of content and language as the defining element of CLIL, where the implemented pedagogies bring about that integration (Nikula et al., 2016;Paran, 2013). That, too, has implications for CLIL teachers' professional qualities.
Another important consideration is the teachers' view of the role of language in the CLIL context. Previous research shows that teachers' views of language have an impact on their choice of teaching methodologies (Borg, 2003). The view on language can be discussed from two points of view: formal and functional (Bovellan, 2014, p. 48). The formal view of language is inherited from structural linguistics (Bloomfield, 1914), focusing on the grammatical and lexical element forming the code. The functional view of language is based on the theory of language as communication and interaction (Halliday, 1973(Halliday, , 1975Hymes, 1972), and considers language use as a social activity rather than a pre-given entity or structure (Pennycook, 2010). Recent CLIL related research follows mainstream second language development trends and focuses on the broader understanding of the social perspective in language learning (Dalton-Puffer, 2013;Llinares et al., 2012;Morton and Llinares, 2017). The new generation CLIL approaches are grounded on such integrated theoretical approaches as systemic functional linguistics (Halliday, 1973), sociocultural theory of learning (Vygotsky, 1978) and second language acquisition (Block, 2003;Kasper, 2009). Language is viewed as a vehicle of content knowledge (academic language ability), as well as a means of acquiring and communicating that knowledge (basic interpersonal communication). CLIL teacher education should thus mean increasing one's proficiency in being able to develop both aspects.

Contextual Factors: A Case of Estonian Schools
As a CLIL setting depends on a number of situational (language policy, language status, diversity of population, learners), operational (teachers, whole school, curriculum) and outcome (diploma, examinations, skills) factors (Baetens Beardsmore, 2009), the contextual factors shaping the Estonian setting of CLIL need to be specified here.
CLIL in Estonia has two distinct linguistic focal points: CLIL fostering the national language (Estonian) and CLIL promoting the acquisition of a foreign language (English, German, French, Russian) (Maljers & Wolff, 2007, p. 53-54). These two types differ noticeably from each other operationally and situationally. CLIL fostering the national language is part of the country's language policy and concerns non-Estonian speaking learners with the aim of integrating the latter in the local community. CLIL through a foreign language involves the learners in the respective bilingual education regardless of their mother tongue. This paper explores the latter type of CLIL in the bilingual education settings of English, German and French. Historically, these are the main foreign languages included in the National Curriculum of the Estonian basic and upper secondary education. Although Russian is widely taught in Estonia both as a first and a foreign language, it was excluded from the current study because as the largest minority language of about 30% of habitants (Statistics Estonia, 2019) it brings along a political dimension and tends to overlap with the first type of CLIL defined above.
Different foreign languages vary in their status in the Estonian classrooms: English is very well represented, while German and French are noticeably less so (Sõstar, 2012). The latter two languages have a stable position, though, due to historic traditions. Although foreign languages are offered at both private and state school, the current paper will view municipal schools (government funded schools) only, as it is those schools that can more consistently be discussed in terms of teacher education and professional development.
As seen in the Annals of the Ministry of Education and Research (HTM, 2019), different forms of CLIL have been practiced in Estonian general education for close to a century. It is these contexts with a long-standing bilingual education focus that the current study investigates. It is conducted in the schools with advanced multilingual programmes promoting extended foreign language learning in the three aforementioned languages that were established in Estonia in the 1960s (HTM, 2019). There, foreign language instruction begins at the age of 7 or 8 and CLIL subjects -geography, biology, history, art or music -are added at the age of 13-14. At the upper secondary school level, CLIL subjects are a compulsory part of a bilingual module designed by the school.
Here, it is important to highlight Estonian schools' considerable autonomy in decision-making and implementation of teaching methods, materials and content. The state neither restricts nor promotes CLIL, leaving the choice and responsibility to the school management (Sõstar, 2012). In most cases, CLIL with foreign languages can be characterized by bottom-up initiatives (Maljers & Wolff, 2007) enhanced by teacher agency, which is seen as a teachers' pedagogical key resource that facilitates teacher-initiated changes and reforms in educational practices (Leijen et al., 2019;MacLellan, 2017;Orland-Barak, 2017). Promoting teacher agency is a part of the teachers' professional development starting from an early stage of the career. Additionally, there are many personal and contextual environment challenges (cf. Bandura, 1977;Day et al., 2007;Glatthorn, 1995;Green & Pappa, 2020;Meristo, 2016;Opdenakker & Van Damme, 2007;Sass et al., 2012). The personal environment encompasses self-efficacy beliefs, self-regulation, cognitive skills and attitudes; whilst the contextual environment includes the physical surrounding, organizational support and teacher community, but also building good relationships with students by creating a supportive and friendly atmosphere, establishing trust and sharing responsibility with students; as well as showing care towards them (Segolsson & Hirsh, 2019;Tschannen-Moran & Woolfolk Hoy, 2001). The aforementioned synergy is not self-evident, but rather seen as a result of a teacher's conscious professional growth that leads to a collaborative learning environment.
Rapid social and technological changes in the past decades have brought to light new educational challenges that require new approaches -21st century learning skills demand 21st century teaching (Ananiadou & Claro, 2009;Saavedra & Opfer, 2012;Sahin, 2009;Silber-Varod et al., 2019). These learning skills can be summarized under the three main categories: information and communication skills, thinking and collaborative problemsolving skills, interpersonal and self-directional skills. Competences needed in the 21st century challenge also teacher professionality. The 4Cs framework for CLIL, proposed by Do Coyle (2007), supports teachers well to adapt to the changing learning environment and fosters the acquisition of the aforementioned learning skills. Content, Cognition, Culture, and Communication are relevant regardless of the CLIL subject. Explaining the content and expressions by using appropriate language, as well as digital resources, and enabling interaction enhances information and communication skills, but also collaborative problem-solving skills. Cognition, i.e., understanding and analysing the content and situating it to the broader context and relating it to life-situations fosters thinking skills. Raising intercultural awareness through the positioning of self and 'otherness' promotes interpersonal and self-directional skills. Scaffolding strategies are essential to support the learning skills across these four Cdimensions (Mehisto et al., 2008). However, the extent of cultural component and the use of realia may vary and depends on the particular course (Javorčíková & Zelenková, 2019).
Given the complexity of the setting, a qualitative approach has been chosen to investigate the CLIL teacher professional shaping, as it can give us a useful key to understand the interrelationship between teachers' thoughts, actions, choices and consequences (Riessman, 2008).

Research Questions
This study aims to explore teachers' CLIL meaningful experiences as triggers of professional development with the help of narrative analysis. The narrative analysis aims at answering the following research questions: • Which experiences appear as meaningful in the CLIL teachers' professional development? • Which contextual factors are reported to shape the professional CLIL teaching career in the context where no explicit formal CLIL teacher training is provided?

Research Design
In the last few decades, research has been pointing out the potential of narrative inquiry in human and social sciences research (Meraz et al., 2019;Nasheeda et al., 2019;Polkinghorne, 1995;Riessman, 2008). A narrative allows an insight into individuals' personal experiences that can be studied within their unique life circumstances and their particular context (Riessman, 2008). Through narratives, it is possible to restructure perceptual experience of a person by rearranging the memories of the events of a life. Riessman (2008) maintains that "the events become meaningful because the individual chose to include them in the story". Narrative analysis is seen here as a tool of conceptual growth of main stakeholders (Coyle et al., 2018). As such, this method can be very fruitful to investigate experiential acquisition of CLIL teachers' competences during their professional career.

Participants
The study analysed the narrative responses of three motivated teachers coming from schools with different types of CLIL classes in three foreign languages (English, German and French). The choice of participants was motivated by the teachers' teaching experience in foreign language CLIL in three comparable institutional settings. The respective institutions are municipal schools which run a bilingual programme in one of the above-mentioned foreign languages. The participants were contacted individually through the professional network of the authors and informed about the aim of the study. Participation in the study was voluntary. The participants were interviewed, the interviews were then recorded and subsequently transcribed. All participants had the opportunity to check and comment on both the transcript and the final narrative. For ethical reasons, the names of the participants were changed. Table 1 presents the background and profiles of the participants and their contextual factors.

Data Collection
A focus group interview with the three teachers was carried out in March 2019, then transcribed and split into three parts corresponding to each participant's input (Riessman, 2012). Data about educational background, contextual factors, general teaching experience and specific CLIL teaching experience was gathered in December 2019. Narrative analysis of the interviews was conducted by two researchers in parallel, and the final narrative emerged from the triangulation of results between the four concerned researchers. Following the procedure proposed for narrative analysis, the current analysis was conducted in five phases specified below: 1. From focus group interview to accurate transcript (including emotions, e.g. smiles; utterances, e.g. hmmm); 2. From accurate transcript to each participants' transcript (holisticcontent approach, Earthy & Cronin, 2008); 3. From each participant's accurate transcript to storying (Nasheeda et al., 2019); 4. From storying to structured narrative analysis, restoring chronological plot, exploring in parallel CV and school archives (Nasheeda et al., 2019;Polkinghorne, 1995); 5. From narrative analysis to findings. The validity of the findings was achieved through triangulation. Multiple methods of systematic evaluation were used: thematic and structural analyses (Meraz et al., 2019); interview analyses by two researchers separately; consideration of background data explaining the participants' contextual environment (e.g. school context, CV-s, language policy).

Analysis
From among different methods in narrative analysis, we chose to follow narrative configuration principles developed by Polkinghorne (1995). The data were organized synthetically into stories according to the classic core narrative: introduction (abstract), orientation (who, what, were, when), complication (problem), evaluation, resolution and coda (Earthy & Cronin, 2008).
After synthesizing the narratives, the second stage was an analytical analysis of the results. It was inspired by the multimethod approach to narrative analysis (Nasheeda et al., 2019) and multiple methods of systematic evaluation (Meraz et al., 2019). Both represent Riessman's analytic approaches (2008,2012), hence thematic and structural analyses were applied.
Thematic analysis was used to explore participants' stories for meaningful experiences in CLIL classes. It focused on 'What was said?' to understand the deeper meaning of the stories from the holistic point of view. Many similar themes crossed teachers' narratives, but their impact on the story varied.
Structural analysis aimed to disclose the relations between teachers' professional growth and their experiences in CLIL. The chronological reconstruction of events revealed the challenges in CLIL classes as well as ways to cope with difficulties. The overarching question was 'How did you become a CLIL teacher?'

Results
The results are presented in two parts. Firstly, the results of the structured narrative analysis, introducing the process of becoming a CLIL teacher, i.e. the starting point, problems, solutions and current situation are shown in some detail. Secondly, the results of the thematic analysis, conveying the most meaningful experiences of each participant that emerged in their narratives through different aspects that were discussed are outlined.

Structured Narrative Analysis: CLIL Teachers' Professional Career
The results of structured narrative analysis are presented in the Table 2.

Thematic Analysis: Teachers' CLIL Meaningful Experiences
At the beginning, Kelly experienced parents' fear, uncertainty as to which teaching methods to choose and lacked quality materials. Later she created her own CLIL materials and developed her own teaching method for the CLIL courses. Now she is enjoying collaboration with students and has a trusting relationship with them. Kelly believes that the teacher's role is to be a mentor and a guide.
Max considers CLIL classes as a tool to develop the "real use of language", especially in basic school. In upper secondary school, CLIL is an added value for students' future professional life whereas content teaching is most important here. Occasionally, Max still experiences difficulties in CLIL, but he believes that being a native speaker is an advantage.
Linda values CLIL classes more than language lessons and values content learning over language learning. She needs to create CLIL materials and that leads to a huge workload at school. In general, she has had a positive experience with CLIL, even with linguistically weaker students, even within one year of course participation.

Discussion
All participants are in service as CLIL teachers. It is noteworthy that none of them had any training in CLIL methodology. Their understanding of being a CLIL teacher is based on their experiences that have been shaped during their professional path. All of them faced obstacles at the start, although the nature of those difficulties was different.
Kelly started teaching CLIL while she was a novice teacher. When recalling her first CLIL experience, she acknowledges the lack of materials as her most severe problem ('When I first started teaching CLIL, compiling materials was the hardest challenge.'). Her solution was to use a subject content coursebook designed for L1 speakers which did not take into account the local, i.e., Estonian, national curriculum and was linguistically not suitable for L2 speakers in her classroom. The unsuitability of the nonadapted materials may have been one of the reasons for the eventual change in her own role in the CLIL programme. While choosing or developing materials for a CLIL classroom, it is generally useful to rely on the Cummins (1984) matrix that allows the teachers to consider the relationship between cognition and language in the given materials. The texts and tasks, while being contextually meaningful and cognitively appropriately demanding for the learner, would also have to be linguistically accessible (Coyle et al., 2010). This assumption is supported by Ball et al. (2015) who highlight the importance to adapt materials for CLIL classes. Being affected by the parents' fear that the materials are unhelpful, which translated into their overall dissatisfaction with the CLIL course ('Some parents asked, "why this circus?" if you don't have a proper colourful coursebook to hand out.'), Kelly discovered the importance of appropriate instructional materials. López-Medina (2021) has summarised the research on the assets of using textbooks: organisation of content, support for the teacher and students, the resultant feeling of safety expressed by both, as well as such aspects as an attractive layout and cost-effectiveness. Thus, because coursebooks are seemingly convenient readily available tools for classroom use, teachers are quick to resort to their use in the classroom. López-Medina's (2021) survey of the related literature also demonstrates, though, that, however enticing, teachers ought to steer clear of indiscriminate use of coursebooks, as they do not meet the multifarious ideosyncartic needs of different educational contexts, nor should all their content be taught in the order suggested by the coursebook. Although textbooks for a CLIL context are hard to write for a multitude of reasons, there is an ever-increasing number of them being published and thus available for teachers to use. Rather than prohibiting teachers from using textbooks, she proposes that their choice should be informed. Thus, a checklist for CLIL textbooks has been put forward consisting of 60 criteria encompassing seven categories: content, cognition, communication, culture, language, integration and a category labelled general (López-Medina, 2021). Considering these would facilitate CLIL teachers finding context appropriate pre-developed teaching tools and save them from some very time-consuming materials development. But even with a very close match of the coursebook with the curricular requirements, it would still need the CLIL teacher to use the coursebook with discretion and supplement it with additional contextappropriate materials. Simultaneously, Kelly discovered the impact of another contextual factor in the programmethe role of parents and their perception of what constitutes a good education. Parents were concerned about the sustainability of the content knowledge ('The parents worried about the following school year, whether their children would cope with the content as the CLIL course was based on a foreign country's coursebook.'). Another driving force in the development was the discovery of a need for specific instructional strategies needed when combining content and language. Although very competent in content knowledge, Kelly lacked instructional strategies to transmit that content, as do most novice teachers whose self-efficacy is considered to be rather low at the beginning of the professional career (cf. Tschannen-Moran & Woolfolk Hoy, 2001), but the added challenge here was combining subject instruction with language instruction. Kelly's first CLIL experience was interrupted by her maternity leave. She initiated a new CLIL course about 20 years later which proved to be a source of much satisfaction. The reasons for this are manifold ranging from tailor-made materials to involving students in the course design (cf. table above) ('Collaboration with students is a pure joy.'). During the intervening years, she was still working as a teacher and her professional growth contributed to her self-reflection skills and enabled her to recall her second CLIL experience as meaningful. She used different methods of CLIL, taking into account students' age, interests and motivation ('Advanced students need a solid "backbone" and general knowledge of French history, while younger students' are filled with curiosity that needs to be guided.').
Max had been working as a language teacher for eight years before starting his CLIL career by initiating a CLIL geography course at the same school. The working environment was thus familiar. Although experienced as a teacher, he still faced many challenges such as lacking subject content knowledge and appropriate materials for instruction as well as collaborating little with colleagues ('At first, geography lessons were difficult, I started from zero. I had no experience and no connection to previous geography teachers.'). Despite the challenges, five years later, he initiated another CLIL course at upper secondary school level, CLIL economics. Interestingly, when describing his CLIL experiences, Max mostly focuses on the economy course, paying much less attention to his geography CLIL experience in basic school. Although continuing to teach it, this course appears to be less meaningful for some reason. It could be explained by what he considers important in attaining CLIL success. When discussing his principles of teaching, he highlights the importance of developing his students' oral skills, but since CLIL geography only takes place once a week, he feels he cannot contribute much there ('I only meet them once a week, it's just too little to communicate with students.'). CLIL economics has more contact hours and Max experiences palpable success that nurtures his teaching agency (Leijen et al., 2019), which in turn seems to have been a key motivator to establish a CLIL teachers' team at his school ('I see my role as a team leader.'). Also, he mentions being a native speaker as a strength that fosters his competence. Previous research (Moussu & Llurda, 2008) on diffrenecies between native (NSs) and non-native speaker (NNSs) teachers of foreign languages points out that there are advantages and disadvatages of both groups with different classrooms and learners. Indeed, in spite of his positive CLIL experience, teaching is still a source of recurring apprehension for Max ('There are still difficult groups to teach which makes me tense.'). He notices linguistically weaker students' occasional resentment of CLIL classes. It is not immediately clear what the source of frustration is but could indicate a need to hone methodological choices, e.g. the level of scaffolding tasks to make them manageable to the students which would hopefully lead to an experience of success and therefore satisfaction. The frustration could also signal a need to work on interpersonal relations. Here the relationships that need attention are both those among the students and those between the teacher and the students, as trusting relationships with students lead to a collaborative learning environment (Segolsson & Hirsh, 2019;Tschannen-Moran & Woolfolk Hoy, 2001).
Linda became involved in CLIL after having been a subject teacher of natural sciences for two years. The problems she faced differed from those of Kelly and Max. Her school was already practicing bilingual education following an IB curriculum; thus, the CLIL concept was not novel as it may have been in the two contexts discussed above. Linda's main concerns was and still is availability of teaching materials. Initially, the school provided some ready-made materials designed for similar contexts abroad. She realized, however, that effective CLIL materials are those needed to be adapted to the Estonian national curriculum as well as to students' needs in her particular context. She notes that the time spent on creating and remodelling teaching materials is a considerable additional workload ('Creating materials is slavery work.') which may deter teachers from developing their own materials. As a solution, she emphasizes the importance of sharing CLIL materials within the CLIL teachers' community, thus reiterating the finding of previous research (Ball et al., 2015). Her biggest concern, however, continues to be her students' soaring stress level at the beginning of a school year, caused by their initial contact with the CLIL course. As a novice CLIL teacher, she struggled to alleviate the situation with mixed results ('When I started with CLIL, I tested a lot of different ways, but many of them didn't actually work.'), but with experience, she resorted to such strategies as an individualised approach, use of L1 with linguistically weaker students, and content-oriented assessment. This approach is echoed in Tschannen- Moran and Woolfolk Hoy (2001) work who maintained that with experience, the CLIL teachers' self-efficacy improves through a better command of instructional strategies. Linda's reflection on her experiences leads her to perceive teaching within CLIL as a continuum of sustainability, i.e., she knows her students' prior content knowledge, takes that information into account, and relates her course outcome (i.e. what her students have achieved) to the next year's course content to guarantee a smooth transition.
Contrary to Kelly and Max, Linda's approach demonstrates that she is more content-focused than language-focused and considers her ability to meet subject content requirements a key factor in her CLIL course. For Kelly, the success lies in promoting students' motivation by providing supportive learning conditions (e.g. a positive atmosphere, trust). On the other hand, Max finds that CLIL is successful when 'the real use of language' occurs. The results of the study fully concur with the findings of Bovellan (2014) that the length of teaching experience in general or in CLIL does not lead to a clear conceptualization of integration in contentlanguage learning. The current study shows that any prospective CLIL training program should help to enhance teacher awareness of the need to better conceptualize their beliefs about subject-content pedagogies and language learning. It is a useful starting point for a CLIL teacher professional development path. Dale et al. (2018) have recently developed an analytical framework for language teachers in bilingual education which could be useful for this purpose.
The analysis of the respondents' narratives discloses some dynamics of personal interpretation of the meaning of CLIL and the respondent's professional identity. For Linda, who started teaching CLIL in the context of an internationally recognized bilingual programme and within an institutionally well-established tradition of practicing CLIL, the content learning dominates language learning ('Language is a tool to learn a subject', 'I make it clear at the beginning that I only assess content knowledge and not language.'). Thus, she follows a hard CLIL model, as described by Ball et al. (2015), but does not identify herself as a CLIL teacher. The use of hard CLIL is supported by the dominant status of the English language in the society and her students' daily exposure to it. Kelly and Max, on the other hand, teach less represented foreign languages in Estonia and their students' language contacts with both German and French outside the classroom are limited. Both of them have a strong CLIL teacher identity. Kelly changed her content focus completely after the first challenging experience and adopted a soft CLIL approach focusing on the language but bringing elements of content learning into the programme ('There was a need to increase a contact with the target language.'). Initially, Max practiced soft CLIL with basic school students, focusing mostly on developing his students' oral skills through content (prioritising language proficiency), but later, with economics in upper secondary school, he preferred the hard CLIL approach ('There is a benefit for students to learn the content that even 80% of Germans don't know [economics].'). Thus, teachers' understanding of CLIL is not fixed, but rather dynamic, stemming from either the demands of the setting or their personal beliefs.; thus, one teacher may apply both hard and soft CLIL without breaching his/her own principles.
The teachers' understanding of CLIL is connected with their view of language (Bovellan, 2014). Linda and Kelly have a formal view of language ('By the end of the 9th grade they have acquired such a level in English that later, students don't make grammar mistakes and there is no need to teach it any more, they only need specific vocabulary.'). Max, however, expresses a more functional view by stressing the importance of communication ('The language must be used.'). Their view of language, in turn, may have influenced their methodological choices (Borg, 2003).
Previous research on CLIL teacher training needs (Pérez Cañado, 2016) and teachers' adjustment to school environment highlights the role of cooperation between teachers (Meristo et al., 2013;Opdenakker & Van Damme, 2007). Surprisingly, the participants in this study said very little about cooperation. Full of enthusiasm, Kelly and Max pioneered the CLIL project alone at their school and may have neglected that aspect of materials' development in their narrative. Linda's school, however, was a well-established environment for bilingual education. So, it is perhaps no surprise that she mentioned cooperation in the context of development and sharing of materials. Max's school represents an advanced multilingual programme and thus supported his initiative to develop an entire bilingual programme in German, involving several colleagues and creating conditions for collaboration. This kind of collaboration turned out to be meaningful for Max ('One of my tasks is to create a team of CLIL, we are three now.'). Collaboartive discourse (Slabina & Aava, 2019) is very strong in Max's voice.
Collaboration among teachers as a meaningful experience seems to be enhanced if there is support to the individual teachers' CLIL efforts from the school board. Although, top-down initiated teacher collaboration may remain artificial and superficial (Vangrieken et al., 2015), the board support alone is insufficient if the aspect of collaboration is lacking as the Kelly's example illustrates -CLIL initiative at her school was abandoned when she took a break from teaching, but coupled with teacher pedagogical beliefs, the course became a success. Thus, the role of school administration is rather to provide teachers with a supportive atmosphere for collaboration and to encourage the emergence of self-selected collaboration partners (Krammer et al., 2018). Linda's and Max's experiences are cases in point here, they were able to choose their team members and the programs were backed by the school administration. But the same can also be seen in Kelly's case, where the individual's enthusiasm and interest proved to be successful with the help of the support by the school board. Besides increasing teachers' job satisfaction and self-efficacy (Vangrieken et al. 2015), teacher collaboration is crucial from students' perspective by serving as a real-life model facilitating the acquisition of collaborative learning skills. The best way to teach students team work and collaboration is to use the same methods in teaching them.
Behind these dynamics of professional identity, an enactment of professional agency can be seen as shown by Green and Pappa (2020). In Kelly's case, she was supported by the school board, though there were no colleagues to express interest in collaboration for CLIL. That may have been the reason why her maternity leave ended the CLIL classes at her school and it took nearly 20 years to start a new CLIL course. This was initiated by Kelly again when she finally had an opportunity to develop CLIL materials in cooperation with a colleague. The cooperation with a native speaker trainee became meaningful for her ('With our French trainee we developed CLIL materials and she contributed with her vision.').
The current study seems to show that Estonian CLIL is a hybrid case between Spanish and Northern European cases as described by Goris et al. (2019). It has the same model (bottom-up initiatives) of implementing CLIL as the Scandinavian countries have but the dynamics are closers to the Spanish trends (a more optimistic approach).
With all three teachers, there is significant personal investment in the CLIL course design and development. The findings show that practicing teachers, starting a CLIL course, need just as much scaffolding as the novice teachers at the beginning of their teaching career. Establishing a community to share best practices and resources might reduce the teachers' workload if the school develops as a learning organisation and supports cooperative practices at all levels (Meristo et al., 2013;Senge et al., 2014).

Conclusions
The study investigated CLIL teachers' meaningful experiences using qualitative analysis based on Polkinghorne's narrative configuration. This approach is innovative because it allows researchers to delve deep into CLIL teachers' beliefs about teaching and learning, the challenges related to subject and language integration, their subjective knowledge of CLIL and their perception of contextual factors that shaped them as CLIL teachers.
When looking at the meaningful experiences that CLIL teachers report to shape their professional development, a number of factors are highlighted: personal initiatives, school support, understanding of bilingual education, beliefs about an appropriate CLIL model, students' needs and their cognitive abilities and curricula demands. The analysis also showed that even if the experiences appeared to be different in different school and language context, there was considerable overlap in what seemed to be meaningful contextual factors.
Although in Estonia foreign language teachers and subject teachers can receive short-term training in CLIL methodology, it is not mandatory. Starting a CLIL teaching career is a completely new experience and comparable to what novice teachers experience when they enter the profession. The CLIL approach obligates teachers to conceptualize their teaching: define the content, instructional strategies, methods, assessment, etc. So, they need as much support and a sense of community as novice teachers. This understanding should lead teacher educators to review the concept of CLIL teacher training in the light of bilingual education taking place in monolingual contexts.
Overall, CLIL teachers' narrative analysis disclosed three main determining experiences: the role of an ideological framework for understanding the concept of CLIL; teachers' relative loneliness in practicing CLIL in the Estonian context; and the importance of a supportive school environment and collaborative practices. The level of collaboration between teachers may determine how they conceptualize and apply a CLIL approach in their teaching. Furthermore, this approach may also be an effective tool to support teachers' professional agency at school as a learning organisation.
The study, although small in scope, highlights a number of important considerations while investigating CLIL teachers' meaningful experiences in their professional development. Narrative analysis proved to be an effective tool to investigate non-tangible hard-to-reach features that guide the teachers' professional path.

Notes
This work was supported by the Republic of Estonia and the European Social Fund under the project "Development of Competence Center in Tallinn University" (2014-2020.1.02.18-0640).