“ Creating the illusion of speaking Romanian well ” : Hungarian speakers ’ teaching and learning the majority language in Romania

: The case study of the article is translanguaging as an educational strategy in preparation for the graduation exam in Romanian language and literature in a Hungarian school in Miercurea Ciuc/Csíkszereda, Romania. Romanian language competence scores are at the bottom of national rankings in this Hungarian-majority town in Szeklerland. Students who speak a minority language have their knowledge of the majority language evaluated in the graduation exam in Romanian language and literature based on the same criteria as ﬁ rst-language speakers ’ , which has strong implications for their participation in Romanian society. The main research question of this ethnographically informed article is how translanguaging happens in a classroom where students ’ ﬁ rst language is being used with the aim of facilitating performance in their second language. The article argues that in the classrooms where the research was conducted, translanguaging is a strategy that negotiates between students ’ educational needs in the local environment and the expectation espoused by the state to perform as if they were monolingual Romanian speakers. Similarly, students use translanguaging to strategize between the curricular expectations and their language performance. Yet, I argue that in this case study the emancipatory potential of translanguaging is limited due to ethnolinguistic hierarchies that remain unchallenged.

in most of the other parts of Transylvania, Hungarians live among the Romanian majority (Figures 1 and 2).
As in the Szeklerland, including Csíkszereda, the administrative seat of Hargita county, the linguistic environment is mainly Hungarian, and the majority of students come from monolingual Hungarian milieu, they have low exposure to the majority language outside school. Their "acquisition of Romanian solely depends on the institutionalized teaching of Romanian" (Fazakas 2014: 347). However, due to the ineffective methods of teaching Romanian in classrooms, the result is that a large proportion of Hungarians in the Szeklerland shows a low level of bilingualism in terms of effective communication skills in Romanian (Benő and Szilágyi N 2005;Csata 2016a;Fazakas 2014;Horváth 2005;Horváth and Toró 2018;Kiss 2011). Regarding the knowledge of Romanian in Transylvania among the Hungarian-speaking population, around 16% self-reported having major difficulties (12%) or hardly speaking it at all (4%) (Horváth and Toró 2018). However, even though there are wide and well-developed Hungarian minority institutional networks that make it possible to live in a "Hungarian world" in Transylvania (Brubaker et al. 2006), for certain types of social mobility, integration into "Creating the illusion of speaking Romanian well" Romanian language and culture is needed. Language knowledge and discrimination based on it "creates and removes opportunities for certain speakers" (De Korne and Hornberger 2017: 248). Not being a first-language speaker of Romanian creates the experience of a "school failure" and impedes students' social and economic position (Tódor 2015) in several ways: their higher education choices and prospects (Péntek 2011), position in the labor market (Csata 2016b(Csata , 2018Kiss 2011), and drive for emigration (Beretka et al. 2018;Brubaker et al. 2006;Waterbury 2010).
The case study of translanguaging with the aim of preparing students for the graduation exam in Romanian language and literature in 12th grade (last year of secondary education) classrooms in a Hungarian secondary school in Csíkszereda facilitates the conceptualization of translanguaging. Unlike most research that explores translanguaging in migrant contexts, where at least one of the languages used is widely distributed, most often English, I deal with a setting that has an autochthonous minority in East Central Europe and where neither of the two languages used are major European languages. In this case study, like in most cases of translanguaging, there is a hierarchy of languages, which is in the focus of my research. The case study makes possible to analyze how translanguaging is used as a negotiating strategy for both the teacher and the students: for the former between the top-down policies of teaching and testing the majority language and the bottom-up educational and communicative needs, and for the latter between the expectations of the curriculum and the graduation exam they are faced with and their Romanian language skills. This dynamics show the interplay between the official and the underlying "unofficial" levels of language policy (Wodak and Savski 2018). What is specific in the case I discuss in this article, however, is that translanguaging has a particular goal too: to enhance students' language skills in their weaker language in a testing situation. Therefore, I pose the question how translanguaging takes place in a classroom where students' first language (Hungarian) is being used with the aim of facilitating their performance in a second language (Romanian) at the graduation exam. I argue that even though translanguaging offers teacher and students a possibility to transgress the monolingual bias of second-language education and caters to students' educational needs, it is not used to its full potential because of the monolingual bias that majoritylanguage speakers, as well as minority speakers, are socialized into. This language ideology regards monolingual Romanian to be the norm for all speakers, be their first language Romanian or a minority language.
The article will introduce the locality in which the study took place, presenting the education system in Romania, with special focus of minority education, and the teaching and learning of Romanian within that. The aim of outlining these is to direct attention to language hierarchies in Romania that contextualize the case study in general, to the reasons for the low performance of Hungarian-speaking students from the Szeklerland in Romanian language, and to the importance of the graduation exam in Romanian language and literature for them in particular. Then I will go through the theoretical premises on which the research is based, namely literature on educational policies of teaching official languages, teachers' agency, language testing, and translanguaging. Next, I will explain my choice of sources and methodology, as well as my position with regard to the research. The main part of the article is the discussion of the case study of the classroom strategies for the preparation of the graduation exam in Romanian language and literature in three of Gabi's 12th-grade classes in a Hungarian secondary school in Csíkszereda. The article ends with conclusions about the significance the case study has on debates regarding translanguaging and majority language teaching and learning.

Minority education in Romania and teaching
Romanian to minority language speaking students Primary education in Romania starts when students are six years old, and lasts for eight years. After that, students can enroll into secondary schools, which are either theoretical (grammar), technological, or vocational schools (Toró 2013). At the end of the 12th grade, all students in Romania have to pass the graduation exam in several subjects, including Romanian language and literature. The official national passing rate of the graduation exam in Romanian language and literature has been between 55 and 76% (Papp Z. et al. 2018). The scores from the graduation exam count in towards the admission to universities. Art. XIII of the Constitution of Romania (1991) recognizes only Romanian as the official language of the country (see also Marácz 2015;Kiss et al. 2018), and the Law on National Education prescribes that all citizens have the duty to learn Romanian (Legea 2011; see also Papp Z. et al. 2018). This means that starting from kindergarten and then throughout primary and secondary education, students in schools with minority language instruction learn the official language of the country. Yet, those who belong to the Hungarian, German, Ukrainian, Serbian, Slovak, or Czech national minorities have the right to education in their first language from kindergarten to secondary school, and in some cases at university level, too (Legea 2011; see also Papp Z. et al. 2018;Péntek 2011).
With regard to bilingual education, García and Kano (2014) make a typology of educational models according to how the majority language is being taught to minority language students: it can be 1) a taught subject; 2) the language of instruction; 3) the teacher's context; or 4) the context of the school. Classes with Hungarian as the language of instruction in Romania can be characterized as type 1, and within that following a maintenance type of bilingual education, organized either in separate schools or in schools with parallel classes in Romanian and in a minority language. In Miercurea Ciuc, the former is the norm, with several schools in town providing Hungarian-language instruction in all subjects (Benő and Szilágyi N 2005;Kiss 2011;Tódor 2019;Tódor and Dégi 2018).
In the past decades, the social capital accessible through the knowledge of the country's official language has undergone major shifts in Romania. For the generation who took part in education during the state socialist period, Romanian was the language of social mobility: education in Romanian was a way to get employment and be able to participate in the mainstream society. The end of Ceauşescu's regime and Romania's eventual EU integration resulted in large numbers of Hungarians from Romania migrating to study and work in Hungary or Western Europe. The influence of the kin-state (Brubaker 1996;Pogonyi 2017;Waterbury 2010) and Hungary's politics to transborder Hungarians has also changed, investing in programs to encourage young people to stay in their countries of origin through funding universities and offering scholarships (Takács et al. 2013). As a result, students in secondary schools in Transylvania today have the possibility to study in Hungarian language in the region or they can pursue higher education in Hungary, as well as elsewhere in Europe, in a foreign language. Even if students continue on to higher education in Romania, the majority of them will study in Hungarian 3 (Papp Z. et al. 2018). Also unlike their parents' or Gabi's generation, current secondary school students in Csíkszereda use foreign languages, especially English rather than Romanian as the global lingua franca for travel, popular culture, international communication, and in certain institutions also for communication between Romanians and Hungarians (Csata 2016a;Marácz 2015). Hence, for many Hungarian students from the Szeklerland, the official language of the country they live in has a low linguistic market value and is often merely a school subject.
Parallel to the linguistic market value of the knowledge of Romanian, the legislative framework of its education as a second language, or a non-native language (limbă nematernă in Romanian) has changed too. Until the 1995 Law on Education, article CXX, § 1 , all students, regardless of which language of instruction they had at school, were learning Romanian according to the same curriculum (Kiss 2011;Marácz 2015), not taking into "account the students' linguistic and cultural background" (Tódor 2015: 26). Conversely, all teaching material was in Romanian language only. 4 Teachers of Romanian are trained in Romanian language and literature programs at all major universities in Romania, and there is no differentiation of the study program based on students' first language or whether they would be teaching Romanian as a first or as a non-native language. The 2011 Law on Education stipulates that in all phases of public education, students studying in a language of instruction that is not Romanian should study the subject Romanian language and literature according to curriculum designed for them and from textbooks other those used by their peers for whom it is a first language (Legea 2011; see also Tódor 2015. However, implementation is often inadequate (Horváth and Toró 2018). As reforms are introduced gradually and no special curriculum, textbooks, and testing methods have not been made so far for secondary school students (Tódor 2015).
Therefore, the gradual introduction of the differentiated curriculum means that the program of the graduation exam in Romanian language and literature for the generation who finished their secondary education in 2020 (the 12th graders of my case study) and for those who will have the exam in the forthcoming years does not differentiate evaluation criteria based on the first language of the students; it only has minor differences between the humanities theoretical track and pedagogical vocational education as opposed to the science theoretical track and all other tracks of vocational and technological education. The Program lists the competences the exam tests: (1) correct and adequate use of the Romanian language in different communication situations (use of strategies and rules of oral expression in monologue and dialogue, use of forms of written expression compatible with the communication situation, identifying the stylistic functions, reception of the meanings, and use of language acquisition in the production and reception of various oral and written texts); (2) appropriate use of comprehension and interpretation strategies, modalities of thematic, structural, and stylistic analysis in the reception of literary and nonliterary texts (identifying the topic, analyzing the linguistic structures and elements of composition of the narrative, dramatic, and poetic text, comparison of perspectives in texts, and interpretation of texts at first sight); (3) contextualizing the texts studied by reference to the period or to cultural and literary movements (identifying and explaining the relationships between literary works and the cultural context in which they appeared, and building an overview of the Romanian cultural phenomenon, e.g. the interwar or the modern period); (4) argumentation of opinions in various situations (identifying argumentative structures in literary and non-literary texts, arguing a point of view on an issue under discussion, and comparing and evaluating different arguments to formulate one's own judgment).
The Program also lists the canonical authors whose work students should know. With regard to language and communication, these language skills need to be applied in the exam on the phonetic (e.g. correct pronounciation of neologisms, accents), lexico-semantic (e.g. derivatives, compounds), morphosyntactic (e.g. inflectional forms, grammatical agreement), orthographic and punctuation (e.g. capitalization, dividing), and stylistic-textual level (e.g. archaic and regional elements of texts, colloquial language, speech figures) (Programa 2020). The above detailed specifications provide a good overview of what skills and knowledge students should acquire by the end of their secondary education, but it also shows that these are features that do not test functional communication skills in everyday situations but take them for granted. To perform well in the Romanian language and literature exam, test takers need to be able to conduct literary criticism and have detailed knowledge of the periods, genres, canonical authors of Romanian literature and their works. Papp Z. and his colleagues show that in the graduation exam in Romanian language and literature, the scores of Hungarian speaking students are only 1-3% below the national average. With regard to the 2013 graduation exam, Toró (2013) reports larger differences between the national average scores and those of students attending Hungarian schools, but he also draws attention to the fact that students in professional and vocational schools generally score lower than those in grammar schools, regardless of the language of instruction. It is a general tendency among students who attend schools with Hungarian language of instruction that those who fail the graduation exam largely do so because of low scores in Romanian language and literature (Papp Z. et al. 2018;Tóró 2013). Effectively, this means that with investing great efforts in terms of energy, time, and money for private tutoring (for those who can afford), the majority of students from the Szeklerland do pass the exam.
It was at the time of my fieldwork that the move of the Hungarian Civic Party (Magyar Polgári Párt -MPP) to abolish the graduation exam in Romanian language and literature for students who attend school in a minority language was univocally rejected at the Romanian Parliament with even the Democratic Alliance of Hungarians in Romania (Romániai Magyarok Demokratikus Szövetsége -RMDSZ), the largest Hungarian party in Romania, voting against it (Kulcsár 2019). However, as a student in Miercurea Ciuc shared his opinion with me, "there is too much fuss around it". His classmate agreed: "We're going to get a 5 [the lowest passing grade on the scale from 1 to 10] anyway." Thus even though the issue of Hungarian-speaking students taking the graduation exam on the same terms as their Romanian-speaking peers is a topic extensively discussed and criticized in the Hungarian public in Transylvania, there is a wide consensus among adults and youth alike that the graduation exam in Romanian is and will remain to be a milestone in the education of Hungarians in Romania.

Theoretical framework: majority language education through translanguaging
In nation-states, full command of the country's official language helps linguistic minorities' social mobility, inclusion into the dominant culture, politics, economy, "Creating the illusion of speaking Romanian well" and education (Shohamy 2006). Linguistic competence reflects the power relations of a society (Bourdieu 1977(Bourdieu , 1991. However, Gal emphasizes that language not only reproduces existing hierarchies between groups but constructs them actively, too (2016). Hence, the official language of a country is both and ideologically charged and a practically important issue as it provides (or fails to provide) a means of integrating minority members into the mainstream society (Kiss 2011). In Romania, Hungarian speaking students acquire the official language of the country mainly in educational settings but also through informal means. The peculiarity of this case study lies in the fact that even though Hungarian speakers are a linguistic minority on national level, their exposure to the majority language is limited because on the local level, in the Szeklerland, they are a majority and have little contact with Romanian language. "Schools are important sites of social and cultural reproduction, and […] accomplish state agendas" (Heller 2006: 16). School teachers are those who mediate between top-down language management and language practices in the classroom. In performing these duties, they are bureaucrats, "soldiers" of the system (Shohamy 2006), but they can also act reflexively and critically. Teachers are "not just another cog in the policy wheel" but active problem solvers dealing with students' actual learning in the classroom (Mohanty et al. 2007: 212; see also Filipović and Vučo 2012;Jaspers 2015). If we look at language education policy through Hornberger and Johnson's metaphor of the onion (2007), teachers are the chefs stirring it on a moment-by-moment basis as a "direct response to realities on the ground" (García and Menken 2007: 256). Teachers' agency in this view corresponds to ways pedagogical choices and decisions are made, dependent on sociocultural factors, the teachers' ideologies, and the students' needs (Le et al. 2020). It is, however, a capacity constrained by the confinements of the structure and reliant on others' power to act (Weinberg 2020). In the case of the classroom in Csíkszereda where I did my research, the teacher, Gabi, employed pedagogical strategies found to be the most suitable to prepare the students for the graduation exam in Romanian language and literature. In accordance with own language ideology, these teaching strategies through which Gabi implemented the curriculum and prepared students for the exam were a negotiation between the students' needs arising from the local socio-linguistic situation and the national language and educational policies.
Apart from implementing the curriculum, teachers are responsible for preparing students for language tests. Such tests structure the relationship among languages and create hierarchies among citizens through defining the prestige and status of the language tested, stating what is correct and what is not in a language, and suppressing linguistic diversity (Shohamy 2006). Even though modern states officially promote multilingualism, they usually do not value the linguistic diversity within their borders. This unspoken drive to homogeneity excludes and marginalizes speakers who do not conform to the monolingual language norm (Blackledge 2008;Jonsson 2017). In their language use, bilingual speakers are constructed as a problem, and they are expected to use "correct" language and comply to "double monolingualism", that is, monolingual standards and competence in both languages they use (Heller 2006). Gogolin (1997) calls such an approach that sets monolingualism as the norm for all members of the nation-state the "monolingual habitus". In this vein, language diversity in the classroom is limited, with language policy insisting "that languages should be kept separate in the learning and teaching of languages" (Creese and Blackledge 2010: 104; see also Cummins 2007;Paulsrud et al. 2017). Being a "one size fits all" (Shohamy 2006) test, in the Law on Education, the graduation exam in Romanian language and literature is seen as the norm for all students regardless of their first language or their educational profile. Conversely, the exam in the minority language, that is the students' first language if they attend school in a language other than Romanian is conceptualized as an additional exam (Legea 2011). The expectation to achieve results as if Romanian was their first language marks minority language test takers as incompetent language users. Having to invest extra efforts to perform, the test puts them in a disadvantaged position compared to their majority language speaker peers.
Contrary to this approach, translanguaging insists on the permeable use of languages in the classroom. While language contact practices can also been termed codeswitching, there are differences between the two terms, and the use of translanguaging in this case study is substantiated by several features of this term. First, while the focus of codeswitching is on the separation of languages, translanguaging emphasizes their interrelatedness (García 2009;Goodman and Tastanbek 2020;Jonsson 2017;Otheguy et al. 2015). Second, unlike codeswitching that emphasizes the code, that is, the identifiable languages, translanguaging focuses on the participants and their practices (Creese 2017): it is a concept that "shifts the focus of the analysis from languages in contact to the speakers who are communicating" (Jonsson 2017: 20; see also Creese 2017;Jakonen et al. 2018). Third, as opposed to code-switching focusing on language practices as indicators of proficiency, translanguaging connects practice to making meaning (Goodman and Tastanbek 2020). Fourth, translanguaging is a goal-driven perspective commonly applied to educational settings and is not a value-neutral concept. As a language education strategy, translanguaging is "an arena for negotiation as well as resistance" (Shohamy 2006: 92). It has the potential to transgress and destabilize language hierarchies and interrogate linguistic inequalities (García and Kano 2014;Otheguy et al. 2015). Starting out from these characteristics of translanguaging, this case study sheds lights on how the teacher and the students employ this educational strategy in the classroom. Gabi's students were using Hungarian and Romanian in an interrelated way, which gave them freedom to express themselves and relate to the class content with more understanding. Yet, translanguaging was employed with the aim of preparing students for a monolingual Romanian exam in which Hungarian-speaking students started from a more disadvantaged position in the hierarchy of language competence, which limited the emancipatory potential of this strategy.
Language education practices are always situated locally and globally at the same time, as well as in the social, the political, and the economic realms (Pérez-Milans 2015; see also Pennycook 2010); therefore, one shall go beyond the official domain to see real language policies. Canagarajah (2006) sees the focus of linguistic inquiry on the micro-level and the everyday. Thus, ethnography becomes a means "to understand in more mobile, fluid, and contextual ways how language resources are mobilized for different ends" (Pennycook 2006: 69) and it aims to problematize and question language policies and their role in the society (Blommaert and Jie 2010). Therefore, ethnography is the primary method of this research because it serves to unpack the difference between official language policy and its implementation on local level and to analyze how students' bilingualism is acted out in Gabi's classes in Miercurea Ciuc.

Methodology and positionality
The sources of this ethnographic research can be divided into two concentric circles. The methods of collecting data from the "inner circle" of the research includes participant observation in Romanian language and literature classes in a secondary school in Miercurea Ciuc. In order to gain "insights from the inside" (Canagarajah 2006: 156), I attended Romanian language and literature classes of three parallel 12th-grade classes for a week, one of which has five, and two have four classes per week. The classes were taught by the same teacher, Gabi, and in all of the observed classes, the students were analyzing the same poem as part of the preparation for the graduation exam. I "linguistically shadowed" (Dewilde and Creese 2016) 5 Gabi between the classrooms and the teachers' room, and we had extensive conversations before and after the classes, during breaks, over meals and coffee. The topics we discussed included classroom practices, pedagogic decisions, and the wider setting of education, language, culture, and politics of Miercurea Ciuc and of Transylvania in general. This ethnographic approach enabled me to grasp multilingual practices in the Romanian class. Because it was the first time Gabi was teaching the 12th grade, I could partake in an exploration of the ways to answer students' educational needs. When analyzing data, I assigned parts the interview transcripts and my fieldwork notes as belonging to certain categories. I had some of these on my mind prior to starting the research, such as "mixing languages", "translating", "attitudes to Romanian", while others emerged after the fieldwork period when interpreting the data. They contained among others "translanguaging" or "teaching test-language". These were mostly labels that were specific to the case study and that lead me to use particular theoretical concepts and formulate the main question of my research. This process was not one-way though: I constantly moved back and forth between the data and the categories, refining them and finding other relevant examples, to tighten the focus of the research and arrive to a more detailed interpretation of the data.
All research is shaped by the interests of the researcher, the funders of the project, its envisioned readership, as well as the power relations between the researcher and the members of the community studied (Canagarajah 2006). This is particularly true for interpretative approaches and for research where the researcher relies on data collected from participants in a context he or she is not fully familiar with. Critical ethnography in researching language policy seeks to shed light on and challenge power hierarchies between various social groups (Wodak and Savski 2008), especially between majorities and minorities, indigenous and colonized peoples (Johnson and Ricento 2013). In doing so, it stipulates that researchers are aware of how their "own historically and socially situated subjectivity shapes different stages of the research process, especially when they are working closely with educational practitioners and students in local schools and classrooms" (Martin-Jones and da Costa Cabral 2018: 85; see also Wodak and Savski 2018). In line with this, my position in the research can be seen as that of a "halfie researcher" (Abu-Lughod 1991;Subedi 2006) in two ways. I have an East Central European cultural identity but come from a "Western" institution and write for an English-speaking audience. On the one hand, the fact that Hungarian is my first language, which I shared with the teacher and the students, as well as the fact that I myself come from a minority background though from a different country (Serbia), created a common ground with Gabi and the students, despite the age difference between us and the expert position I occupied. My own minority (and) educational experience, academic but also personal, enabled me to grasp a segment of the realities of the students and teacher and to be accepted as part of the classroom. I performed this closeness (Karabegović et al. 2020) by sitting among students in class and asking questions on occasion, making comments, or answering Gabi's inquiries. Like other researchers doing classroom ethnography, I was primarily an observer to the classes, but occasionally a participant as well (Hodges 2020). While my highly limited knowledge of Romanian restricted my options of reading on the topic, it was an asset in the classroom: students did not feel ashamed of their errors in Romanian in front of me, especially because I occasionally took the risk of attempting to say something in that language. Therefore, being a researcher who is only "half-bilingual", who has a very limited repertoire of one of the languages used in the classroom, enabled me to experience the subject matter of the classes and the learning process together with the students.
The wider circle of my data consists of the conversations I had prior to the research and after it with my research subjects and with professionals in the subject from Transylvania. Before the two-week-long fieldwork geared specifically to this research, I had visited Csíkszereda twice, once for a week and once as a shorter visit of a few days. In the first one of these visits, I met Gabi and the students in the classes with whom I later conducted the research. In order to gain wider insights into the topic of my study, I also conducted semi-formal interviews with other teachers of Romanian language and literature, students, parents, an author of a textbook of Romanian for primary school minority language students, and experts in the field of Romanian as an official language and Hungarian minority politics in Csíkszereda, Kolozsvár, and Marosvásárhely.
I also communicated with Gabi after the fieldwork via email and social media as a follow-up on issues of the graduation exam. 6 I took fieldwork notes in classes, and of the conversations with adults outside the classrooms, I not only took notes but also taped them. Gabi and the professionals I interviewed signed consent forms, and students gave oral consent to my participation in classes, which was also agreed upon by the school principal. The final draft of the article was also approved by Gabi.
5 Discussion of the case study: thinking Hungarian, writing Romanian

Strategizing languages for the exam
Pedagogical solutions are always localized and often ad hoc and "messy" (Jaspers 2015). In the classes I visited, the way Gabi and the students were preparing for the Romanian graduation exam was by proceeding from understanding the text to be 6 In mid-March 2020, while I was working on this article, the students of the school where I conducted research, like almost all schools in Europe, switched to digital learning due to the Covid-19 pandemic. Twelfth graders continued the preparation for the graduation exams online, then in June, they took the exams regularly. At the end of June, I received the news from Gabi that over 90% of them successfully passed the graduation exam in Romanian language and literature.
analyzed via mainly the medium of their first language, Hungarian, to the final product of the analysis, an essay that would be solely in Romanian language. However, a Hungarian speaker who has neither knowledge of Romanian nor of the context would have difficulties understanding what was going on in the classroom. Eventually, it was the combination of the languages that made possible the comprehension and the completion of the task (Creese and Blackledge 2010). For instance, the class used the following sentences: (1) A curcubeu a boldogság szimbóluma.
When Hungarian phrases are translated to English directly, while the translation of the Romanian words are put in brackets, the sentences would look like the following: (1) The curcubeu [rainbow] is the symbol of happiness.
(2) What is the idea poetica [poetic idea]? What we can see in the above examples is that within a single sentence, students and teacher merged words belonging to two registers that they associated with two separate languages. Romanian words were phonologically integrated into Hungarian sentences, making words such as "prima strofát" (3), "leoiacát", and "iubirevel" (5) not easily assignable to one or another named language. The choice of what words were used in Hungarian and what in Romanian depended on the students' knowledge: those that they had access to easily were in Romanian, while those they did not know or could not remember in Romanian were in their first language. While languages used in the classroom are named for convenience (Romanian and Hungarian) both by the research subjects during fieldwork and the researcher in the article, Otheguy and his collaborators remind us that languages are social constructs and "cannot be defined in terms of a set of essential lexical or structural features" (Otheguy et al. 2015: 286). As Goodman and Tastanbek argue, "through a translanguaging lens, what has previously been called alternation, mixing, hybridity, and fusion of codes is better understood as languaging practices that transcend named language boundaries" (2020: 11). Furthermore, the fact that the practice has a clear goal marks this translanguaging: for the students, it was being able to express themselves freely and concentrate on the subject matter of their utterance. Through seeing the above examples as translanguaging, it is possible to focus on the students as language users and on their way of strategizing educational expectations and their language skills. Translanguaging happens in every classroom, but it is the local relationships between languages and their speakers that define it (García and Kano 2014). As I gradually understood, what was said in which language in the classroom was not random but the result of the teacher's pedagogical decisions. Gabi prompted the students to use as many Romanian words related to the text they were analyzing and to the exam structure as possible. Shohamy calls this way of providing students with shortcuts to the exam "test-language" (2006) and argues that when the content of the test de facto becomes the content of the curriculum, it narrows down students' knowledge (2001) . Already used to this way of learning, the students were quick to adapt the learned phrases and use them in sentences that in most cases mixed Hungarian and Romanian. Gabi encouraged them to speak with each other in class, share ideas that came to their minds in no particular order of speaking and in no particular language as long as they respected the topic. "It's like playing with plasticine together in kindergarten", said a student describing the classes. This gave confidence and a voice to students (García and Wei 2014), involved them (Creese and Blackladge 2010), and enabled all of them to participate in the learning process regardless of their level of knowledge of Romanian. Similarly to students using translanguaging to strategize between the expectations set in the curriculum and the graduation exam they are faced with and their actual linguistic knowledge, the teacher was negotiating between teaching in line with the curriculum and fulfilling students' educational needs. In facilitating translanguaging in class, Gabi went against all three aspects of the monolingual principle of second/foreign language teaching: the direct method that claims that instruction should be carried out exclusively in the target language, the "no translation" assumption that discourages translation between the first language and the second language, and the "two solitudes" approach that strictly separates the two languages (Cummins 2007).
Bilingual speakers "draw on any kinds of resources useful and accessible to them, with various degrees of fluency", depending on the individual speaker and the locality (Spotti and Blommaert 2017: 165; see also García and Kano 2014). In the case of dominant Hungarian multilingual students preparing for an exam in Romanian language and literature, it included resources from not only Hungarian and Romanian but also English and Latin. For example, Gabi reminded students remember that centrum is the same in Romanian and Latin, students deduced the meaning of the Romanian word cerc from the English "circle", ignorata from the English "ignore", Gabi translated descuiat into both Hungarian as nyitott and English as "open-minded" and teased students that their English was better than their Romanian. By doing this, Gabi was trying to draw on students' experience of having studied other languages in their learning of Romanian. Therefore, in line with multilingual pedagogy, the teacher, instead of attempting to keep languages separate from each other, used them as resources (Haukås 2016) in order to achieve the best possible results in the graduation exam in Romanian.
Clearly, in the 12th grade, the primary task for the students and the teacher was to prepare for the graduation exam. While most students claimed that the exam in mathematics was equally difficult, it was the Romanian class that was devoted to the exam preparations almost completely. "I will burn the notebook after the exam", said a student. "Pass it on to someone next year instead", answered the teacher. As Gabi claimed, and as I could compare to my visits to the classes in which I conducted research the previous year when the students were in 11th grade, there was a major shift in the classroom practices in the 12th grade. In the previous years, especially the last two leading to the 12th grade, the mode of teaching Romanian to students who have another first language was moving towards the communicative paradigm . However, in the 12th grade, students I observed found themselves in the "old paradigm", with little time for anything else other than to prepare for the graduation exam. Gabi attempted to break out of the tight framework of the graduation exam preparations occasionally, both during the classes when they were analyzing literary works and in a few separate classes devoted to "playing games", as the students called it. During classes, Gabi would prompt students to deduce the meaning of a word from something assumed to be familiar to them. For instance, when a student asked what fertil meant, instead of translating it to Hungarian, the teacher said: "Remember, in the store, in the gardening department, there is the insecticide and the fertilizator [fertilizer]". Gabi also jokingly referred to perioada fertila [fertile period], saying that "such an intensive course before the graduation exam would be useful for you". Gabi also tried to rely on students' intuition about what is correct: when a student said primul secuenţă for "first sequence", Gabi asked: "Doesn't it burn your tongue?". To the student's negative answer, Gabi said the correct phrase: prima secuenţă. Similarly, if the "feeling" failed the students and someone could not differentiate between the use of este and sunt, the third person singular and plural of the verb "to be", the teacher explained its use in Hungarian.
Requested by students, Gabi occasionally allowed students to play games instead of analyzing texts. In such class I observed, students divided themselves into groups and could choose between playing High-low, Activity, or another guessing game. Despite Gabi's instructions to use only Romanian, they mostly spoke to each other in Hungarian but stuck to the rule in a performative sense in that their final answer was always in Romanian. In this, they sought each other's help as well as requested translations from Gabi. Building on their existing knowledge of languages, they used Romanian words with a similarly sounding equivalent in Hungarian or English because they could easily and quickly be recalled, such as for instance electronic, plastic, and silicon when guessing the material of an object. The atmosphere in the classroom was playful and relaxed, and from the observer's point of view, there was a mixture of Hungarian words, Romanian words, as well as a lot of gesticulation. "I like it when they are cheerful. It's quite rare in Romanian classes", commented Gabi, referring to the high workload students have to deal with in regular Romanian classes. In one group, the phrase that needed to be guessed was "Milky Way", and students were struggling to remember it in Romanian. They tried with "milk", lapte, and Gabi encouraged them to remember how to say "way", but instead of cale, its synomym drum was what they could recall, until a student remembered the whole phrase, calea lactee from Costache Negruzzi's historical novel Alexandru Lapusneau, a literary work they had had to read and analyze in one of the previous grades. Similarly, the word curcubeu [rainbow] was recalled from the poem Leoiaca tânărâ, iubirea, and on the occasion when the answer was "bird", instead of the word pasăre, a student remembered how to say a particular type of bird, lark, in Romanian ciocârlie, from the same poem. The direction of learning was in these cases the opposite of what many decision-makers in language policy assume, namely it went "from literature to the street" instead of vice versa. As the interactions and institutions through which it could be learnt, such as the neighborhood, the television, the radio, the army, or the work collectives, have changed or vanished in the past decades, the belief that the country's official language can be acquired by all members of the society as a "native language" also got refuted (Shohamy 2006).

From double monolingualism to translanguaging
The monolingual bias of institutions, the school being a crucial one, is "a key agent in the exclusion of linguistically diverse populations" (Piller and Takahashi 2011: 374). Second-language speakers have to prove their knowledge of the majority language "against the measures developed by the dominant group, who use the agencies of the state (schools, bureaucracies, language academies, the media) to describe as what counts as linguistic competence" (Heller and Duchêne 2012: 5;see also Shohamy 2006;Flubacher and Yeung 2016). If assessed according to the monolingual standard as the norm, second-language speakers are seen as using a language that does not naturally "belong" to them and thus cannot compete with first-language speakers in language use (García and Kano 2014). A student whose first language is Romanian or who attends school in a minority language but in an area with high exposure to Romanian may have difficulties performing the tasks related to the graduation exam in Romanian because they do not have sufficient knowledge of the subject matter. However, a person with limited knowledge of Romanian living in the Szeklerland encounters language specific problems, those that are related not only to the subject matter but to understanding the task and to being able to express oneself. Therefore, in the classes I observed, Gabi chose to move away from the monolingual habitus and employ teaching strategies that combine elements from Romanian and Hungarian languages. This was a pedagogical decision in line with the teacher's language ideology (Ganuza and Hedman 2017): Gabi's "socio-political motivation to legitimize the use of more than one language in communication" (Jakonen et al. 2018: 32) and believing that students' educational needs are best suited by placing emphasis on understanding the content. García calls such teachers as having a translanguaging stance because they trust that through this practice their students can make meaning for themselves (2020). However, it is important to see that being a first-language speaker of Hungarian, Gabi had the choice of using this pedagogical strategy and to transgress the principles of monolingual second-language teaching. Tódor's reasearch demonstrates that the majority of teachers teaching Romanian to Hungarianspeaking students in areas where Hungarian is the dominant language believe in the positive effects of speaking the students' first language (Tódor 2005). Yet, in Transylvania, as elsewhere, teachers' strategies may be very different depending on their pedagogical decisions and their knowledge of the students' first language (Tódor and Dégi 2018).
For understanding how translanguaging works, it is crucial to see that it is a result of deliberate planning by the teacher (Ganuza and Hedman 2017). In Gabi's classroom, it was using Hungarian language in combination with Romanian that "Creating the illusion of speaking Romanian well" made the content of the curriculum more accessible for the students. Baker (2001) sees the possibility to facilitate students' understanding of a second language by using their first language as an educational advantage because it may suit students' needs better than any of the professionally orthodox approaches. However, the pedagogical routine Gabi chose was admittedly an experimental one and did not follow any pre-set didactical approach. In settings when minority-language speaking students need to produce academic texts in a colonial or dominant language of the state, such a "hidden practice" (Laihonen and Tódor 2017;Shohamy 2006) is seen as an "illicit pedagogical strategy used widely by teachers and students" without recognition in either language education policy or in scholarship (García and Wei 2014: 124). As Creese and Blackledge (2010: 105) note, based on research of numerous case studies in bilingual classrooms, "moving between languages has traditionally been frowned upon in educational settings, with teachers and students often feeling guilty about its practice". Gabi expressed insecurities to me about what would happen if an education inspector visited the class, which has not happened yet in the teacher's career but could be expected as part of a regular procedure. In line with this, when I mentioned Gabi's method to a teacher in a Cluj-Napoca where Hungarian-speaking students are much more exposed to Romanian and thus have less difficulties in preparing for and passing the graduation exam in Romanian language and literature, she saw it as cheating. Unlike Gabi, this dominantly Romanian speaking teacher used solely Romanian with the students and did not devote an entire year for the exam preparation but moved ahead with the prescribed curriculum. Gabi saw the advantages of speaking solely in Romanian in class, believed that from a communicative standpoint it would be more useful for the students to use the target language only, and in fact perceived it as the norm expected from all Romanian teachers, but explained the choices that led to not doing it: too little time and too much material to teach and learn.
As Jonsson (2017) argues, translanguages includes situations in which the participants know a very limited number of linguistic features that they can use only under given circumstances, but they employ their entire register in order to achieve their communicative aim. In the classroom in Miercurea Ciuc, such was the students and teacher's practice of shuttling between Romanian and Hungarian in order to convey the meaning. This type of translanguaging, even though can be termed as "natural" in the sense that it was encouraged by the teacher to help the understanding of the subject matter to the class (García and Kano 2014; Williams 2012), did not arise from the students' everyday language practices but was generated by the specific task they had to perform: produce an academic text in Romanian at the graduation exam. Written communication is usually associated with more power and control. Therefore, while translanguaging was allowed and encouraged in class, the essay that was to be the end-product of the analysis had to be written in monolingual literary Romanian. "It's easier to look for similes in Hungarian because if you don't understand the word in Romanian, then …", complained a student to me. In an attempted balance between learning with understanding and learning by heart, Gabi was trying out a method and hoping it would yield results at the exam: asking the students after every class to write down what they discussed in class in academic Romanian at home in a form of a miniessay. They had to analyze the title, the artistic devices, and each stanza separately for homework. Following the structure of the exam, Gabi set a word limit for each part and encouraged students to try to keep within a certain time limit too in which they would write the mini-essay. Gabi also reminded the students how many points they can get for such an exercise in the graduation exam. The instructions were test-related and often very practical, such as "if you manage to make this part longer, you don't need to write so many words in other parts." Gabi explained to me that that the students would receive handouts with the answers to the possible exam questions personally written in literary Romanian (and as I could see later, when I received it as well, with Hungarian translations of the main parts of the analysis) in the form and length appropriate to the exam, but only at the end of analyzing the subject matter in class. Gabi claimed that the aim of this was to make students realize that they could do it themselves too, part by part. As the teacher said, "keltjük az illúziót [we are creating the illusion] that we speak Romanian well, and not only that, but literary Romanian, and we know how to analyze literature". By seeing this translanguaging practice as a kind of a deception, Gabi in fact invoked double monolingualism as the norm for the students. This model states that in the classroom, "participants treat languages as bounded entities […] from which departures can be sanctioned as a 'mix'" (Jakonen et al. 2018: 46). Thus, in Gabi's classes the monolingual habitus was challenged neither at the graduation exam nor in the classes. Instead, Romanian classes provided its temporary suspension: students' first language, a minority language could be used to facilitate the preparation for the graduation exam and was a valued linguistic resource. The Romanian classes with Gabi as a teacher became "multilingual spaces that, even if briefly, alleviated some of the pupils' linguistic problems" (Jaspers 2015: 111). Yet, even though these spaces proved effective in solving everyday communicative situations, they can only be short-lived, and for both the teacher and the students, translanguaging proved to be a negotiation strategy between policy and practice rather than resistance to or change of the language education policy. Outside the classroom, the Romanian monolingual norm prevails and dominates language education policies and language hierarchies in which the dominant language is valued more than minority languages, and multilingualism is envisioned as double monolingualism, which means that it can be contested only on local level (Liddicoat and Curnow 2014).

Conclusions
Taking the assumption that social processes are embedded in and constructed by specific localities (Pennycook 2010) as a starting point, the article focused on teaching and learning Romanian, the official language of Romania in the case of Miercurea Ciuc in the Hungarian-majority Szeklerland region. The ethnographically informed case study was that of a Hungarian secondary school in the town where students in the final year of their secondary education were preparing for the graduation exam in Romanian language and literature. Through the analysis of translanguaging in the classroom, this article aimed to contribute to the understanding of the uses and limitations of translanguaging as a pedagogical strategy, especially in a dominantly minority language context where its aim was to facilitate students' performance in a majority language testing situation.
Using Hungarian, the students' first language, to facilitate comprehension of the subject matter despite the monolingual bias of the curriculum and mobilizing students' passive knowledge of Romanian were the main methods used by Gabi, the teacher whose classes I observed. Such practices show the value of translanguaging as well as the agency that teachers can have in de facto language policies by acting locally in the classroom. Therefore, I have argued in the article that the translanguaging Gabi and the students were engaged in to prepare for the graduation exam in Romanian language and literature is a negotiation strategy between top-down education policy of teaching the country's official language and bottom-up language practices answering to the students' needs in the specific locality of Csíkszereda. The article has demonstrated how translanguaging works in a classroom in a school in East Central Europe with minority language instruction where students prepare for language testing in the majority language on a first-language level and has unpacked not only the linguistic but also the social dynamics of the two languages and their speakers.
The graduation exam in Romanian language and literature requires an essay of literary values written in Romanian. Such an approach constructs the language practices of second-language speakers as deficient. This hierarchy of speakers is the result of a language ideology that sets monolingual Romanian and multilingualism and double monolingualism as the norms and excludes those who do not possess the knowledge and skills to perform according to these standards. Translanguaging as used in Gabi's Romanian language and literature classes is a local negotiation of the monolingual habitus. Gabi and the students' translanguaging practices have an emancipatory potential in the local classroom setting because they make it possible to include both Romanian and Hungarian in the learning process, facilitate understanding, and encourage students' confidence in the preparation for the graduation exam. However, linguistic and social hierarchies existing outside the classroom and the "hidden agenda" (Shohamy 2006) of majority language education policy is to establish and maintain the dominance of the majority language and its speakers. This prevents translanguaging educational practices in this classroom from achieving wider linguistic and social equality or being an agent of structural change. Instead, they remain "a nameless and untheorized method" (Canagarajah 2012: 259). Language education policy that rests on the monolingual habitus is not challenged in Gabi's classroom but reflects the ethnolinguistic hierarchies present in Romanian society. However, as the case study of the Hungarian 12th graders from Miercurea Ciuc shows, translanguaging can and does serve as "a vision and ideology of what education could be if students were to be given experiences that de-link from the […] matrix of power" (García 2020). Translanguaging can thus be an inspirational pedagogical strategy for teachers, students, policy makers, and researchers in language education in various ethnolinguistic contexts.