‘No, I Don’t Like the Basque Language.’ Considering the Role of Cultural Capital within Boundary-Work in Basque Education

The aim of this study is to analyze the nature of multiethnic academic interactions in relation to theories of cultural capital and boundary-work. More precisely, it considers to what extent school structure is related to the cultural capital of students from different ethnic backgrounds and explores its relationship to Intergroup Contact Theory and identity. Methods include documentary analysis, participant observation, interviews, and focus groups conducted from an ethnographic perspective between 2015 and 2016. Based on data collected in a Basque school attended by a high proportion of immigrant students, intraethnic and interethnic student–student and student–teacher relationships, and inequalities within these, are analyzed. Results indicate that the distribution of students in different classes tended to be ethnically marked, as most immigrant students chose to attend classes that were taught mostly in Spanish, whereas most autochthonous students were enrolled in classes with a high Basque instruction. The study considers the effects of students’ language choices and concludes that Basque has implications for the theories of identity, cultural capital, and boundary-work, as learning Basque is an academic and implicit rule in Basque education and society.


Introduction
Most authors agree on the importance of the school as a key setting for teenage students' socialization, as pupils learn important social skills in their academic interaction (Bourdieu 2008a;Dubet 2010;Dubet and Martuccelli 1997;Eckert 2002Eckert , 2004Erikson 1989Erikson , 2000Irena et al. 2016;Stein 2004;Weber 2009). Following this idea, some authors argue that the structure of the school guides student interaction with other students and with teachers (Bourdieu 2008a;Passeron 1973, 1981;Dumais 2005;Dumais and Ward 2010;Lareau 2011Lareau , 2015. These claim that students' distribution influences their relationships and these interactions respond to the hidden structure of the academic institution. More precisely, they suggest that the education system reproduces inequalities among diverse social and/or ethnic groups which are present in society. The research that underpins this idea is that of Bourdieu (2008a;Passeron 1973, 1981), according to whom teachers unconsciously favour middle-class students because they hold the same cultural capital, or symbolic acquisitions. manners, speech, gestures, body language, and other social class markers encourage teachers' approval of middle-class and elite students over lower-class ones. Bourdieu designated these practices as habitus. Bourdieu's (1991Bourdieu's ( , 2008a proposal has been extensively criticized because it leaves little room for social mobility and his ideas have been referred to as deterministic (Jenkins 1982(Jenkins , 1992King 2000;Merolla and Jackson 2014). Indeed, although his theoretical tools serve to analyze social inequality in relation to social class and position, they do not explain other strategies that individuals pursue in a wider power-relations framework. More precisely, cultural capital explains how external forces determine a precise social position, but it does not take into account a person's agency. This paper will illustrate these alternative explanations, according to which individuals move away from the position they have been assigned.
Current authors have applied the concept of cultural capital in multiethnic environments using a wide range of techniques and perspectives in their analyses. In some of these studies, students' cultural capital, ethnicity, and schooling success have been compared (Baquedano-López et al. 2013;Dumais 2005; Dumais and Ward 2010; Jaffe-Walter and Lee 2011; Kisida et al. 2014;Lareau 2011;Merolla and Jackson 2014;Nawyn et al. 2012;Van Praag et al. 2015). For instance, Lareau (2011Lareau ( , 2015 bases her research on the Bourdieusien concept of cultural capital in the multiethnic U.S. education system, analyzing the inequalities between African-American and white students. In line with Bourdieu (2008aBourdieu ( , 2008b, Lareau (2011;Lareau et al. 2016) claims that the U.S. education system is directed by hidden norms. When students and their families understand these norms, also designated as the 'rules of the game', they have a better chance of being academically successful and climbing the social ladder. According to Lareau (2011), middle-class students tend to understand such rules better than lower-class students because they have learnt those at home, in consonance with Bourdieu's argument (2008aBourdieu's argument ( , 2008b on cultural capital acquisition. In line with the idea of the 'rules of the game' by Lareau (2011) and the reproduction of inequalities by Bourdieu (2008a), some authors claim cultural capital is ethnically guided. More precisely, these affirm that in daily interaction with students from distinct ethnic backgrounds, some teachers tend to favour local students, as these hold their same cultural capital, leading to ethnic inequalities (Martín Rojo 2010Stein 2004;Weber 2009). Similar studies focus on the Othering practices by teachers towards ethnic minority students in ethnically diverse schools (González 2010;Foley 2010;Martín Rojo 2010Pérez-Izaguirre 2015;Simmons et al. 2011;Stein 2004;Weber 2009). In this context, Othering practices refer to the process by which teachers treat minority students as the 'Others', which usually leads to discrimination of different ethnic groups who are not part of the dominant society.
Other studies focus on the precise 'rules of the game' of multiethnic education systems. For instance, Moskal (2016) suggests that language is fundamental in migrant students' cultural capital. Based on an ethnographic study, she presents the schooling experiences of first-generation Polish teenage immigrants in Scotland. Moskal (2016) highlights that not knowing how the Scottish education system works is an important element in first-generation Polish students' lack of success at school, and emphasizes that when students' families encourage English language learning in compulsory education, it increases their academic success. She also illustrates that these students and their English-speaking peers attend different classrooms, which worked against Polish students' English acquisition and socialization with local students. Lareau's (2011) contribution to the concept of the 'rules of the game' will be relevant in this paper to understand the malleability of the concept of cultural capital, and ethnicity will be considered alongside other cultural capital markers such as schooling performance and language, as highlighted by Moskal (2016).

Conceptualizing Boundary-Work, Intergroup Contact Theory, and Social Identity
After Barth (1976) established a comparative study on ethnic boundaries, ethnic relationships have been the focus of many authors. Barth (1976) proposed that ethnic boundaries exist and persist Soc. Sci. 2018, 7, 150 4 of 20 between different groups through interdependent interaction. He argued that ethnic boundaries guide each group's social life and interactions, but are not affected by each collective's cultural content. Wimmer (2008Wimmer ( , 2013 has made a more recent contribution to ethnic boundaries research and proposed a more complex theory of ethnic relations. He suggests that the nature of ethnic boundaries is defined both by a categorical and a cognitive dimension: the categorical dimension defines the 'Us' and 'Others' 2 , and the cognitive dimension indicates the behaviours with which we divide the world between 'Us' and 'Others'. When individuals or groups apply both dimensions, and their actions correspond, ethnic boundaries are formed. This paper will assess the applicability of Wimmer's (2008Wimmer's ( , 2013 concepts in the Basque setting, testing them against research participants' voices and actions. Based on these two dimensions, Wimmer (2008Wimmer ( , 2013 classifies five strategies that actors use for boundary making. The first, 'shifting boundaries through expansion', refers to the creation of an encompassing boundary, where actors incorporate existing categories into a new one. The second, 'shifting boundaries through contraction', is the opposite strategy, as it reduces the elements the boundary encompasses. The third, 'inversion', aims to re-order the hierarchy of dominated and dominant ethnic groups. The fourth, 'repositioning', takes account of individual strategies to change position within an existing hierarchy, which may include assimilation as a way to overcome a minority stigma. Finally, 'blurring boundaries' overcomes ethnicity as a principle of social organization and categorization, and other non-ethnic principles are used for social organization (Wimmer 2008(Wimmer , 2013. Wimmer (2008) also explains how diverse ethnic perspectives may end up in a consensus. In a divergent ethnic conflict, agreement between distinct groups and individuals appears if the interests of different parties engaged in the dispute overlap. Actors with different social positions and ethnic backgrounds may end up agreeing on an acrimonious matter if there is a partial overlap of interests. Wimmer (2008Wimmer ( , 2013 designates this as 'cultural compromise', according to which individuals and groups justify their own demands and actions, and represent their interests as public benefits. The novelty of Wimmer's (2008Wimmer's ( , 2013 concepts poses a challenge to previous theories on boundary making; however, his theory has been criticized as he "ethnicizes" non-ethnic social and cultural areas (Brubaker 2014;Song 2014). In other words, Wimmer (2013) overuses the notion of ethnicity and applies it to address diverse social phenomena that are not necessarily ethnic. This paper will assess the applicability of these concepts in the Basque setting, using research participants' voices and actions to test them.
A similar boundary-work study by Lamont and Molnár (2002) indicates that self-defined and externally identified boundaries are not necessarily ethnic, and may include other characteristics such as social position and gender. Lamont and Molnár (2002) propose two kinds of boundaries: symbolic and social boundaries. Symbolic boundaries are actors' conceptual devices that serve for defining reality and categorizing people, practices, and objects in time and space, which cause feelings of difference and similarity between groups.
According to the authors, when symbolic boundaries are socially accepted, agreed upon, and guide social interaction in important ways, they become social boundaries. These social boundaries objectify the differences defined in the symbolic boundaries and often lead to unequal access to resources for different groups. For instance, symbolic boundaries turn into social boundaries when processes of ethnic marginalization happen. When this occurs, there is a change in perception of distinct social or ethnic practices and access to resources becomes differentiated between diverse groups. The work on symbolic and social boundaries and inequalities by Lamont and Molnár (2002) is based on the premise that external forces transform social, ethnic, or gender differences into inequalities, as is Bourdieu's work (2008aBourdieu's work ( , 2008b. This paper will assess this perspective in relation to the ethnographic 2 Wimmer (2008Wimmer ( , 2013 uses the term 'Them' instead of 'Others' in his work. However, I chose to use the designation 'Others' to refer to the extensively researched 'Us/Others' binary, in line with most authors in the social sciences (Berg 2010;Simmons et al. 2011;Koefoed and Simonsen 2012;Sterzuk 2015). Soc. Sci. 2018, 7, 150 5 of 20 case presented and will propose alternative explanations based on individual and group strategies of boundary-making.
The concept of ethnic and social boundaries has been tested using students' voices to understand their interethnic and intraethnic relations and their connection to academic achievement, (Boda and Néray 2015;Dimitrova et al. 2017;Lund 2015;Munniksma et al. 2015;Thijs 2017;Van Praag et al. 2015). For instance, Lund (2015) describes the multiethnic Swedish education system and suggests that, at school, students are categorized in terms of their Swedish or immigrant background. This 'Us/Others' categorization is, according to the author, very present in society and in students' discourse and praxis, and becomes part of the students' schooling processes. Lund (2015) argues that Swedish and immigrant backgrounds are separated by boundaries intrinsic to the schooling system and enacted by actors. By analyzing the complexities of what has been designated as 'free school choice', Lund (2015) deconstructs its apparent freedom concluding that students' academic choices are socially, institutionally, and ethnically guided.
Such differentiation can also be conceptualized as the consequence of stereotyping and discrimination between distinct ethnic groups in a wider power-relations framework. Intergroup Contact Theory addresses these and proposes that contact improves relationships and reduces prejudice and conflict among different collectives (Bastian et al. 2012;Cech 2017;Janmaat 2015;Kanas et al. 2017;Pettigrew 2008;Pettigrew et al. 2011;Tredoux et al. 2017). In this work, I will only address multiethnic academic relationship, according to the objectives presented above. In line with the premise of Intergroup Contact Theory, in a study conducted in the U.S., Mikulyuk and Braddock (2018) found that ethnically diverse schools enhance social cohesion because allowing interethnic contact reduces prejudice, stereotyping and discrimination among groups. This evidence works as a defence of diversity against ethnically segregated schools in the U.S.
By contrast, other studies indicate that Intergroup Contact Theory should be more carefully regarded (Bastian et al. 2012;Janmaat 2015;Schellhaas and Dovidio 2016). For instance, in a study conducted in the Netherlands, Vervoort et al. (2011) show that in schools attended by a high proportion of ethnic minority students, majority-minority interethnic contact and friendship were relatively common, although the general attitudes each group held toward each other were less positive than in ethnically more homogeneous schools. In other words, according to this study, a high proportion of ethnic minorities at school seems to be related to more out-group negative attitudes among students.
Finally, an intrinsic element to interethnic contact is social identity, as each ethnic group holds an identity. I will rely on Jenkins (2008) to define identity as the capacity to discern who is who and what is what. This capacity is rooted in language and helps humans make different kinds of classifications on the 'Us' and the 'Others'. In other words, identity is the discursive construction of who belongs or not to a specific category (Durrheim et al. 2011). The idea by Jenkins (2008) also relies on a Meadean concept of identity, according to which subjects, or the 'Us', are constructed in interaction with the 'Others' (Mead 1982;Stryker and Stryker 2016). In multiethnic educational contexts, some studies have analyzed the negative (Weber 2009) and positive (Zuma and Durrheim 2012) avenues for intergroup relations related to these concepts. This paper will discuss these in relation to the theories of boundary-work.
In sum, the proposals by Wimmer (2008Wimmer ( , 2013, Lamont and Molnár (2002), and Lund (2015) suggest that the existence of boundaries is dependent on "Us/Others" categorizations and coherent praxis, which are related to the identity of each group. Nevertheless, while the studies by Lamont and Molnár (2002) and Lund (2015) rely heavily on the premise that institutions and societies generate unequal circumstances that determine the process of boundary-making, Wimmer (2008Wimmer ( , 2013 considers further elements to make sense of these, based on precise individual and group interests. In line with this proposal, Intergroup Contact Theory suggests that interethnic contact is positive to reduce discrimination and stereotyping between ethnic groups (Bastian et al. 2012;Pettigrew et al. 2011), but not all studies support this premise (Vervoort et al. 2011). This paper will test these perspectives in relation to the case study presented above.

The Basque Country and Basque Education System
The Basque Country is located at the Cantabric sea border and includes a part of southern France-Iparralde and northern Spain-Hegoalde. Consequently, the Basque language cohabits with French in Iparralde and Spanish in Hegoalde and is a minority language in both territories. My research took place in the BAC, which is a region located in Hegoalde, where both Spanish and Basque are official languages. Until the 1970s, Basque was not taught in schools, as a consequence of repression under Franco's dictatorship in Spain. It was not until the beginning of the 1980s that Basque acquired an official status in the BAC and now both Basque and Spanish are languages for instruction in the education system. Since the establishment of democracy in Spain in 1978, Basque nationalist movements in the BAC have promoted the revitalization of Basque. Indeed, Basque is not a neutral element in the Basque Country and, although a description of the cultural and linguistic history is beyond the scope of this article, it remains important to note that Basque is an ethnic marker for autochthonous people. In the education system, speaking Basque indicates Basque community-belonging (Echeverria 2003;Martínez 2014). The Basque education system is divided into three linguistic models that students choose between. In model D, instruction is in Basque, and Spanish is studied as a subject; in model B, both Basque and Spanish are languages for instruction; and in model A, Spanish is the language for instruction and Basque is studied as a subject. In some centres, each linguistic model corresponds to at least one classroom at each level in the public system. However, this is not always the case, as both Basque society and government administration promote Basque learning through the reinforcement of models B and D (Department of Education 2016). Hence, model A is less and less available in both public and private schools.

Methods
This study is an ethnographic analysis from an anthropological perspective, based on the results of fieldwork conducted in a Basque school attended by a high proportion of immigrant students, which I will designate as Udabia, 3 during the 2015/2016 school year. I first contacted Udabia in spring 2015, as some secondary education teachers recommended this school. According to them, Udabia's studentship was very diverse in terms of ethnicity, and academic performance. In line with the aims proposed, the techniques used to collect data regarding school structure, distribution of the cultural capital, boundary-work, and the nature of interethnic and intraethnic interactions were: documentary analysis, semi-structured interviews, participant observation, and focus groups. The chronological order of each of the events in relation to the research approach proposed at the beginning of the study is the following: First, a documentary analysis was conducted. It included a literature review on the Basque-based educational studies, newspaper articles referring to immigrant studentship, and analysis of reports published by the education administration in the BAC.
Second, following Spradley (1980) and Dewalt et al. (2011), I adopted a classical anthropological data-gathering perspective and conducted a nine-month participant observation in the classroom at Udabia with the highest immigrant studentship. This classroom corresponded to a 2nd Certificate of Secondary Education (CSE, corresponding to 8th Grade in the U.S. and Year 9 in the UK) level and I will designate it as 2G. An unexpected recurrent observation was students' claim that they did not appreciate learning Basque, which involved teachers' role in classroom interaction. The observation also extended to less ethnically heterogeneous 2nd CSE groups of students, from classrooms that I will name as 2H and 2I, in interactions with 2G's students during recess. The courtyard distribution of students during recess also indicated diverse patterns of intraethnic, interethnic, and boundary relationships.
Third, 36 semi-structured interviews were conducted with the Head of the School, teachers and students in 2G. Interviews with education stakeholders were conducted in Basque, as it is the usual vehicular language in Basque academic environments, while interviews with students were conducted in Spanish. Students openly asked me to do it in Spanish as they felt more comfortable and most of them did not speak Basque. Following Marvasti (2010), the proposed aim of these interviews was to explore subjects' discourses about their self in relation to others. Teachers and the Head of the School, all autochthonous educational agents, described how students related to each other during their lessons and how they behaved at school with other students and teachers. Most of these interviews also introduced how teachers felt and interpreted 2G students' negative attitude towards Basque learning. Semi-structured interviews conducted with students in 2G enabled me to include students' voices in the discourse, which was predominantly created around them.
Finally, three guided focus groups were undertaken with 2G students, in groups of 6-8 individuals. When pupils were asked about their relationships with other students, ethnicity was often at the root of their discussion. During these focus groups, students showed a tendency to elaborate more; indeed, they constructed phrases and interpreted meaning from their peers' remarks.
The four-stage data collection facilitated cross-referencing at different stages of the process. The combination of document analysis, participant observation, semi-structured interviews, and focus groups enabled me to compare, correlate, and validate the data, enabling effective recording of multiethnic interactions.
Data collected was codified using a computer-assisted qualitative data analysis software package, Atlas.ti. Data was collected, processed, and analyzed continuously in a cyclical process that enabled themes present in the literature review to appear, be modified, and re-classified. Finally, data was categorized from general to particular, the results of which indicated a link between the distribution of students, academic interactions, and Basque learning. These themes are analyzed in the results section as follows: • General distribution of students: It refers to the criteria guiding how students were distributed. Two main elements were taken into account: Distribution of students recommended by teachers and chosen by students in 2G and their families. 2nd CSE students' spatial distribution during recess.
• Academic interactions during Basque language lessons in 2G: It refers to peer interactions involving affinity, friendship, and language learning. It also includes student to teacher interactions when learning Basque. I chose Basque lessons as these immigrant students showed a high level of interaction both with other students and teachers. The elements taken into account were: The lower-level Basque lesson. The higher-level Basque lesson. Students who disliked Basque.

Sample
Udabia is a public school located in the BAC community, in a Spanish-speaking region that is surrounded by Basque-speaking areas called Mirebe. The main language for communication in the streets of Mirebe is Spanish, and Basque for a foreigner is sometimes perceived as unnecessary. I chose this location as Udabia is one of the few centres in the region with a considerable ethnic imbalance, both in comparison to other centres and among its own classrooms. The ethnic distribution imbalance was very prominent in the discourse of the educators at Udabia, who described the school as an 'immigrant's centre.' Indeed, in the 2015/2016 school year the centre was composed of 32 teachers and 207 students, of whom 37% were immigrants; 13 different classrooms corresponded to 4 different CSE grades, with 3 classes at each grade. The sample taken from Udabia corresponded to the students of the 2nd CSE (N = 53). At this level, one class employed model D and two classes employed model B. However, the two model B classes differed in their levels of Basque instruction. The distribution of students was the following: 2G (n = 19) was a classroom composed of 15 immigrants, one Spanish, 4 and three autochthonous students, of which two were Roma. The origin of immigrant students was: 13 Latinos, one Bulgarian, and one Portuguese. The majority of Latino students came from Ecuador and the others were Nicaraguan, Colombian, Peruvian, and Bolivian. 2G studied in a 'light' model B; the 'light' designation indicates that, although in theory the class was designated as a model B, it functioned as a model A. 2H (n = 17) was attended by 15 autochthonous students, of which three were local Roma, and two immigrants from Ecuador and Ukraine. 2H studied in a strong model B, which involved a higher level of Basque instruction. 2I (n = 17) was composed only of autochthonous students and studied in a model D.

Results
In the following subsections I will present the results of the documentary analysis, participant observation, interviews, and focus groups organized according to the themes presented above.

General Distribution of Students
In general terms, the results of the literature review ( ) concurred that immigrant studentship in the Basque education system tends to be concentrated in specific public schools. In the 2015/2016 school year, 62 schools shared the following characteristics: over 20% of their students were immigrants and those students were also concentrated in specific classrooms. This was the case in Udabia, as it had 37% immigrant studentship. In Udabia, like in these schools, autochthonous pupils tend to choose model D or model B with a high Basque instruction, whereas non-autochthonous learners often opt for model A or model B with a low Basque instruction. Consequently, an ethnic division often appears between classrooms due to autochthonous and non-autochthonous students' language instruction choices (Department of Education 2016; Elosegi 2008, 2010;Martínez 2014;Muñoz 2015;Septien 2006).
The classrooms composed of a majority of immigrant students usually have comparatively poor academic performance and low Basque learning proficiency (Department of Education 2016). In Udabia, observation of group 2G in comparison to their peers in 2H and 2I confirmed the data in the literature review; as overall, students in class 2G had an unsatisfactory performance and low achievement in Basque language. By contrast, most students in groups 2H and 2I had an average academic and Basque language performance.

Distribution of Students Recommended by Teachers and Chosen by 2G Students and Their Families
The Udabia Head of School, teachers, and counsellors mentioned in the interviews that they always recommended all students, regardless of their ethnicity and mother language, to take model D or strong model B classes, as they have a higher Basque instruction, and speaking Basque enhances academic success, mobility, and integration within the academic community in the Basque education system. As Basque is strongly reinforced at school, nearly all the curricular and extracurricular activities that public, and most private, schools in the BAC organize involve speaking and promoting Basque. According to both students and teachers in Udabia, although all students were encouraged to study Basque, immigrant students and their parents usually chose not to enrol their children in model D or strong model B classes for three main reasons: because they believed that Basque was too difficult; as in the streets of Mirebe they heard Spanish and thought studying Basque was not really necessary; and because, for most of the immigrant families who took part in this study, Basque was not a significant language outside the Basque Country.
An extreme case that I had a chance to observe was that of an Ecuadorean student in 2G who I will name Agustín. Agustín's father came to Udabia to demand that the centre only instruct his son in Spanish. The father, whose mother language was Spanish, did not believe that Basque was useful and complained about Basque lessons. Obviously, the Head of the School did not comply with his demand, but it created friction between the Basque teacher and student. As Agustín knew that his father supported him in his decision to not study Basque, he would often not comply with the institutional ruling to learn it. In the interview I conducted with him, he said the following: Interviewer: Why don't you study in model D? Agustín: Because if you travel abroad, Basque is not useful as it is only spoken here (in the Basque Country). We could say that Basque is not as widely known as other languages.
Interviewer: Do you enjoy learning Basque? Agustín: Well . . . it's difficult. I understand Basque but I can't speak it. Interviewer: Then, why don't you pay attention to the teachers' explanation during Basque classes?
Agustín: Because teachers bore me [ . . . ] sometimes I even like his (referring to the Basque teacher) lessons, but they're boring and it makes me want to provoke him.
In this extract, Agustín acknowledged the difficulty of learning Basque and openly expressed his desire to provoke the teacher during lessons. At the same time, he did not think he would benefit from enrolling in model D, in line with his fathers' ideas. However, the choice to enrol in a light model B prevented Agustín and his 2G classmates from building relationships with autochthonous students, a majority of whom were enrolled in classes 2H and 2I. Hence, language instruction choices created an ethnic boundary between classrooms and resulted in a lack of relationship between autochthonous and non-autochthonous students. Furthermore, students in 2H and 2I could take part in more extra-curricular activities than 2G in Udabia, as most of these use Basque as a vehicular language. Ethnic background was also a matter of conversation among pupils and teachers, and instructors admitted that classroom separation tended to create a pattern for student relationships. In an interview I conducted with María, a teacher in Udabia, she explained the following: Interviewer: Why do you think there is a tendency for immigrant and autochthonous students to attend different classes?
María: As we don't have a Basque speaking environment in Mirebe... it's much more difficult for them (immigrant students) to understand the importance of Basque. In Basque-speaking towns, students speak Basque during recess, students understand that Basque is necessary in their daily life, and enrol in model D with autochthonous students. However, students in Udabia enrol in secondary school but they had attended primary schools before, and some of them have a very low Basque proficiency [...] Often, immigrant students who are enrolled in a light model B in primary school, choose to be enrolled in a light model B in secondary school, [...] although that's going to change in the future, as the Basque administration tends to substitute model B with model D.
This extract illustrates that María was optimistic about the integration of future generations of immigrant students who were going to enrol in Udabia, as they were going to be integrated in model D with autochthonous students. However, she acknowledged a differential ethnic tendency to enrol in classrooms with distinct levels of Basque instruction at the time of the interview (early 2016). According to the observations conducted, such an ethnic division had an impact on student relationships.

2nd CSE Students' Spatial Distribution during Recess
Essentially, students tended to relate to peers from the same ethnic background during recess-so much so that Lara, a student in group 2G, claimed that the courtyard was divided by countries. Lara had arrived the previous year from Castilla, a Spanish region, and had enrolled in group 2G. In the interview I conducted with her, she admitted she had encountered many difficulties relating to her peers from other ethnic backgrounds. Her description was quite accurate, as during recess, 2nd CSE level students divided themselves along ethnic lines.
In December 2015, students were self-distributed the following way: Roma students tended to play behind the school facilities, Latino adolescents located themselves on the opposite side of the courtyard in a large group, and Spanish and autochthonous adolescents, mostly girls, walked around the main school building. Students of other European descent tended to be alone. Football and basketball provided the only occasion for interethnic relations, including autochthonous students, but only male students took part in these activities. In the rest of the cases mentioned, interactions were intraethnic. At that time, the Latino group encompassed students who had come from diverse Latin American countries, and ethnic boundaries were present when they interacted with other non-Latin American students.

Academic Interactions during Basque Lessons in 2G
When group 2G studied Basque as a subject, they were divided into two levels, while 2H and 2I students had an only-Basque language lesson. In 2G, the lower-level group was part of a project called HIPI-Hizkuntza Indartzeko Proiektuko Irakaslea-designed to socialize students in Basque, but most of the lesson was conducted in Spanish. As the other group in 2G had a higher level of Basque proficiency, the teacher could conduct the classroom in Basque, although students tended to answer in Spanish. Pupils in these groups of 2G did not consistently maintain relationships along ethnic boundaries. In the following section, both student-student and student-teacher relationships in 2G will be explained.

The Lower-Level Basque Lesson
The HIPI class was composed of seven students who were newly arrived in the BAC and had to learn Basque for the first time. Their common characteristics were their lack of knowledge of Basque, their ethnic background as non-autochthonous pupils, the use of Spanish as the vehicular language in their conversations, and their negative attitude toward Basque. The group was composed of seven students whose ethnicity ranged from Latino and Portuguese, to Bulgarian. Students chose their own seats in this class and tended to gather in terms of ethnicity and gender. The usual student attitude during this class was disruptive, as students tended to speak among themselves in Spanish and did not take notes, defying Gurutze's, the teacher's, orders. However, Gurutze could control the classroom dynamic quite easily, as the student-to-teacher ratio was very low. Figure 1 represents student distribution, according to more than 20 days of classroom observation: As Figure 1 illustrates, Miguel, David, Alejandro, and Roger took up positions at the left of the class, Myriam sat alone, and Ana and Lorena were close, on the right. Hence, all male students were located at the left of the room, while female pupils chose to sit on the right. Most of the interactions among students were between Miguel, David, and Roger. Miguel, who had arrived the previous year from Peru, was one of the most passive students and regularly claimed to be tired. David, who had come the previous year from Portugal, was often by his side. David mastered Spanish and was a proficient student, but in Basque sessions he tended to interrupt the class and not do the exercises because he was interacting with Miguel. Roger, a Nicaraguan student, interacted with Miguel and David and occasionally with a Bulgarian student, Myriam, with whom he showed a love/hate relationship. Myriam was on her own, except for these few interactions with Roger. Their occasional exchanges sometimes interrupted the classroom dynamic. Alejandro learnt quickly but did not make the effort to pay attention to the lesson and was disruptive; he interacted sporadically with his male peers. All of these student interactions were interethnic, as Latino students related to Portuguese or Bulgarian students. Myriam and David were the only European students in this class, as the rest were all newly arrived from Latin America. Ana did not pay much attention to the class, but as her friend Lorena did, she was more integrated in Gurutze's lessons. Both students are Latina, Ana from Nicaragua and Lorena from Ecuador; consequently, these interactions were intraethnic. answer in Spanish. Pupils in these groups of 2G did not consistently maintain relationships along ethnic boundaries. In the following section, both student-student and student-teacher relationships in 2G will be explained.

The Lower-Level Basque Lesson
The HIPI class was composed of seven students who were newly arrived in the BAC and had to learn Basque for the first time. Their common characteristics were their lack of knowledge of Basque, their ethnic background as non-autochthonous pupils, the use of Spanish as the vehicular language in their conversations, and their negative attitude toward Basque. The group was composed of seven students whose ethnicity ranged from Latino and Portuguese, to Bulgarian. Students chose their own seats in this class and tended to gather in terms of ethnicity and gender. The usual student attitude during this class was disruptive, as students tended to speak among themselves in Spanish and did not take notes, defying Gurutze's, the teacher's, orders. However, Gurutze could control the classroom dynamic quite easily, as the student-to-teacher ratio was very low. Figure 1 represents student distribution, according to more than 20 days of classroom observation: As Figure 1 illustrates, Miguel, David, Alejandro, and Roger took up positions at the left of the class, Myriam sat alone, and Ana and Lorena were close, on the right. Hence, all male students were located at the left of the room, while female pupils chose to sit on the right. Most of the interactions among students were between Miguel, David, and Roger. Miguel, who had arrived the previous year from Peru, was one of the most passive students and regularly claimed to be tired. David, who had come the previous year from Portugal, was often by his side. David mastered Spanish and was a proficient student, but in Basque sessions he tended to interrupt the class and not do the exercises because he was interacting with Miguel. Roger, a Nicaraguan student, interacted with Miguel and David and occasionally with a Bulgarian student, Myriam, with whom he showed a love/hate relationship. Myriam was on her own, except for these few interactions with Roger. Their occasional exchanges sometimes interrupted the classroom dynamic. Alejandro learnt quickly but did not make the effort to pay attention to the lesson and was disruptive; he interacted sporadically with his male peers. All of these student interactions were interethnic, as Latino students related to Portuguese or Bulgarian students. Myriam and David were the only European students in this class, as the rest were all newly arrived from Latin America. Ana did not pay much attention to the class, but as her friend Lorena did, she was more integrated in Gurutze's lessons. Both students are Latina, Ana from (1) Distribtuion of students (2) Key.

The Higher-Level Basque Lesson
The higher-level Basque class was composed of 12 students, most of whom were not newly arrived in the Basque Country. Some, whose parents were Ecuadorean, were born in the Basque Country and had attended a Basque public school alongside autochthonous children. Other students had arrived some years ago from different Latin American countries and had attended a Basque public school since. Two students were newly arrived in the Basque Country and were transferred from the HIPI class because their teachers thought they were able to study Basque at a higher level. Finally, there were two local Roma, and a Basque student. The distribution of students was as specified in Figure 2. Soc. Sci. 2018, 7, x FOR PEER REVIEW 12 of 21 (1) (2) Figure 2 represents student distribution during class. In these lessons, most students respected the classroom distribution they had been assigned by their teacher, which provided students with a better working environment. Like in the HIPI class, the students' attitudes were quite disruptive and except for the Basque and two local Roma students, they all had negative attitudes toward learning Basque. Students were either half-laid on the table or interacting with each other, interrupting Manuel's, the teacher's, lesson. The comparably larger class meant that Manuel struggled to control the classroom dynamic. I will introduce a few examples of classroom interactions to illustrate the intraethnic and interethnic relationships.
The two local Roma students, Jennifer and Kevin, had a love/hate relationship whereby they were constantly together but they often hit each other and argued, creating an escalating conflict in the classroom. Eguzkiñe, the only self-defined Basque student in 2G, was on her own and would only sometimes integrate with the Roma students. However, in general, she was marginalized by her peers, who would often make fun of her and sometimes even refuse to participate with her in assigned working groups.
The rest of the students were also partly ethnically divided: the Ecuadorean girls, Á ngela, Amaia, Maite, and Lur, interacted among themselves and sometimes with the other Ecuadorean-descended males, Agustín and Julio. Juan is Bolivian, and from February he did not interact with his peers, regardless of their ethnic background. Perla, a Nicaraguan student, and Lara, a Castillian student, interacted with each other in the Basque lesson, often openly complaining that Basque was too difficult. (1) Distribution of students (2) Key. Figure 2 represents student distribution during class. In these lessons, most students respected the classroom distribution they had been assigned by their teacher, which provided students with a better working environment. Like in the HIPI class, the students' attitudes were quite disruptive and except for the Basque and two local Roma students, they all had negative attitudes toward learning Basque. Students were either half-laid on the table or interacting with each other, interrupting Manuel's, the teacher's, lesson. The comparably larger class meant that Manuel struggled to control the classroom dynamic. I will introduce a few examples of classroom interactions to illustrate the intraethnic and interethnic relationships.
The two local Roma students, Jennifer and Kevin, had a love/hate relationship whereby they were constantly together but they often hit each other and argued, creating an escalating conflict in the classroom. Eguzkiñe, the only self-defined Basque student in 2G, was on her own and would only sometimes integrate with the Roma students. However, in general, she was marginalized by her peers, who would often make fun of her and sometimes even refuse to participate with her in assigned working groups.
The rest of the students were also partly ethnically divided: the Ecuadorean girls, Ángela, Amaia, Maite, and Lur, interacted among themselves and sometimes with the other Ecuadorean-descended males, Agustín and Julio. Juan is Bolivian, and from February he did not interact with his peers, regardless of their ethnic background. Perla, a Nicaraguan student, and Lara, a Castillian student, interacted with each other in the Basque lesson, often openly complaining that Basque was too difficult.
In this case, intraethnic relationships were present in two different and well-established groups: the Roma group, which was formed of the two Roma students who would only occasionally interact with Eguzkiñe, and the Ecuadorean group, who would interact among themselves. The rest of the students were either on their own, as Juan was, or would unite with other students of different ethnic backgrounds, as Perla and Lara did. However, this had not always been the case, as earlier in the 2015/2016 school year the Latino group had encompassed students from many Latin American countries. The cause of the separation in the Latino group had been a fight in January 2016 between two girls, Perla, a Nicaraguan, and Ángela, an Ecuadorean. After the conflict, Perla did not rejoin her Latino group of friends and Ana, another Nicaraguan student, followed her. The original Latino group continued, composed of the Ecuadorean students, who were a majority, whilst Perla and Ana created another group composed of other 2G pupils and students from higher CSE courses, who were not necessarily Latina. Perla later explained in a focus group that she did not know exactly what the problem had been with Ángela, but that Ángela and the rest of the Ecuadorean students no longer spoke to her. She decided not to pry because as she explained, 'you can find friends anywhere'. She felt she did not need her previous Latino group because she felt capable of making new friends.
In this case, an obvious boundary had been drawn between Ángela's and Perla's groups. There was a clear line that none of them would transgress: they would not interact with each other, and by extension, Perla's closest friends would not speak to Ángela's and vice versa. Although the conflict was not related to ethnicity, the criterion that was taken to separate both groups was country of origin: the Ecuadorean group constricted their boundaries and only a few other Latino-descended students were accepted under very strict circumstances. Perla started to relate to students from diverse ethnic backgrounds, although it must be noted that none of them were autochthonous students, and all were female and Spanish native speakers. As she mentioned in the interview I conducted with her, she expected affinity and an understanding attitude from her new friends. For her, these relationships involved speaking Spanish and using language as a tool for connecting through humour and deep conversations. In this last case, the ethnic boundary expanded and diverse ethnic backgrounds were accepted.

Students Who Disliked Basque
All 15 immigrant students and the Spanish pupil in 2G complained they had to study Basque and tried to prevent Basque lessons from advancing. Some of their alliances were built according to the objective of disrupting Basque lessons, encompassing a variety of ethnic backgrounds. Students who attended any of these Basque lessons were either passive and complained lessons were too difficult, or interacted with each other and interrupted the lesson. They did not show such a disruptive behaviour in other subjects. When students were asked about these attitudes, most claimed that they did not feel motivated to learn. An illustration of this attitude was revealed to me in an interview I conducted with Ana: Interviewer An important element revealed by Ana is her claim that she does not like Basque, which she then retracts, saying that she does not understand it. The underlying idea is that Basque is compulsory, but she does not feel inclined to learn it; however, the teachers perceived such claims differently. In all of the 15 interviews I conducted with educators, Basque learning was considered very important, especially with the Basque language teachers. Most of them openly expressed their irritation with student attitudes toward Basque, and a few of them felt directly affected by them. Those who felt sensitive about student behaviour and comments were open about their ethnic identity and they linked their Basque ethnic identity to the Basque language. I will use the example of a teacher, Antonio's, interpretation of students' claims: Antonio: Well, we must take into account that 2G students are teenagers, and as such, they must differentiate from us, the teachers, and say "Basque is not ours and it is not useful. You have been imposing it on us at school, but now we choose not to learn it anymore" [...]. For us teachers, Basque is important, for them it is not. They don't feel Basque.
In this excerpt, Antonio explains how he interprets immigrant students' comments when they claimed to dislike Basque. He distinguishes teachers' interpretation of Basque as an ethnic marker, and immigrant students' understanding of Basque as a school rule. When Antonio says 'For us, teachers, Basque is important, for them it is not. They (immigrant students in 2G) don't feel Basque', he touches on the importance of Basque for Basque people. If these students do not consider Basque important, it is because they do not feel Basque. Indeed, this is linked with his previous discussion of students as teenagers who differentiate between themselves and teachers by choosing not to comply with school rules. In other words, Antonio notes that immigrant students' choice not to learn Basque is an adolescent reaction to an institutional rule, which is different from the interpretation most Basque teachers made.

Discussion
In the case study presented, the structure of the education system was designed around the level of Basque instruction. As mentioned, Basque is not a neutral element, as it marks academic community-belonging and enhances the prospects of getting a well-paid job in adulthood in the BAC. It is an essential element and an implicit rule for achieving success in the Basque education system, in line with other ethnographic settings where mastering the local language enhances educational opportunities (Moskal 2016;Weber 2009). In Lareau's (2011) terms, Basque is part of the 'rules of the game', as it enables upward social mobility. As such, according to the data presented, it acted as a cultural capital marker: students who had a higher Basque proficiency performed well in other areas, whereas students who had a lower Basque proficiency performed poorly and could not fully integrate in the academic community. Furthermore, it tended to be ethnically marked: a majority of students who had a higher Basque proficiency were autochthonous, while immigrant students were a majority of those with lower Basque proficiency. The theory of cultural capital explains the school's distribution of students in relation to Basque instruction as an external element imposed on students that reproduces ethnic inequalities. However, it relies heavily on the premise that external forces determine such a distribution and academic success, but does not account for individual choices within such a structure. This study suggests that the reality is not so rigid, and immigrant students' choices to enrol in classes with low Basque instruction and then not fully comply with the institutional rule of learning it constitute a form of agency. Hence, in line with the data presented, hypothesis (a) is valid but not complete, as Basque instruction does act as a cultural capital marker; however, immigrant students chose to learn it as little as possible, against school advice.
It must also be acknowledged that classroom separation influenced minority and majority students' spatial distribution during recess. Data seems to suggest that by contrast to the premise of Intergroup Contact Theory, minority-majority cohabitation and contact did not bring inclusion or positive relationships. In fact, relationships between immigrant and autochthonous students were rare, even though they shared school space. What is more, a process of ethnic segregation took place: a majority of immigrant students studied in classrooms instructed in Spanish, while a majority of autochthonous students studied in classrooms mostly instructed in Basque and did not relate to each other during recess. In line with the conclusions by Vervoort et al. (2011), an unbalanced proportion of ethnic minorities in schools and classrooms brought about negative outcomes for majority-minority interethnic relations in this study. However, it must be noted that prejudice between majority and minority groups was not measured and more data on this respect would be needed to reinforce this argument in relation to Intergroup Contact Theory.
Following Lamont and Molnár (2002), I argue that symbolic and social boundaries were institutionally marked: whereas symbolic boundaries indicated the difference between mother languages, habits, and ethnic background in such a multiethnic environment, social boundaries objectified such differences in classroom separation according to the instruction of Basque. Indeed, students who studied most of their subjects in Spanish were a majority of immigrants and had fewer opportunities to be integrated in the academic community and activities than pupils who studied most of their subjects in Basque, who were a majority of autochthonous pupils. Hence, I argue that such ethnically unbalanced language choice became a social boundary that brought about discrimination affecting immigrant students. Nevertheless, in line with the critique of cultural capital theory, the theory of symbolic and social boundaries also considers that external forces produce unequal interethnic relations, and does not consider individual agents' influence, as most immigrant students chose to be enrolled in a classroom with a low Basque instruction attended by a high proportion of immigrant students, contradicting school guidance.
Moreover, the theory by Lamont and Molnár (2002) does not account for all intraethnic or interethnic associations that took place among pupils. Indeed, sometimes interethnic students' relationships were guided by intraethnic and others by interethnic criteria. A more accurate model to explain these is Wimmer's (2008Wimmer's ( , 2013, which accounts for the complex intraethnic and interethnic groupings that took place in this study. Depending on the precise interest of each ethnic group and individual, ethnic boundaries shifted through expansion, contraction or blurring. Among the complex dynamics that enabled intraethnic groupings among immigrant students was affinity. The reason for interethnic associations was the overlap of an interest, such as playing a sport, the need for friends, or having a common 'enemy'. An example of this last case was that of immigrant students' claim to dislike Basque. These students, regardless of their previous intraethnic or interethnic associations, united against the teacher when they had to study Basque. This union was sometimes explicit, when they openly claimed they disliked Basque, or implicit, when they showed a passive or very disruptive behaviour during Basque lessons. This indicates a cultural compromise, in Wimmer's (2008Wimmer's ( , 2013 terms, as students who would not usually relate to each other made a silent consensus to avoid learning Basque. This acted as an ethnic boundary-producer between students and teachers, as teachers were used to teaching Basque in other classrooms without facing these complaints. More precisely, immigrant students did not identify with Basque and consequently, were not motivated to learn it. However, teachers interpreted this attitude as the enactment of an ethnic boundary: as speaking and promoting Basque is directly linked to being part of the Basque community, openly claiming to dislike it involves rejecting community involvement. In other words, two identities were enacted: Basque identity as the dominant academic identity by teachers and autochthonous students, and non-Basque identity as non-Basque learners by immigrant students. This binary idea was present in the discourse by some teachers, who ended up in an Othering practice when treating immigrant students in the case presented. Hence, in line with the theories presented, identity was rooted in language enacted by an Othering discourse based on a differential concept of group belonging (Durrheim et al. 2011;Jenkins 2008). Group belonging was marked by speaking and learning Basque in this case study. This ultimately worked to the detriment of immigrant students, as they did not fully integrate into the Basque academic community. In this sense, hypothesis (b) is not completely accurate. When Basque language learning, boundaries between teachers and immigrant students were enacted if students claimed to dislike Basque or did not actively learn it. However, during those classes immigrant students in group 2G tended to bond regardless of their ethnic background to avoid studying Basque. In such moments, ethnic boundaries blurred and interethnic contact happened among immigrant students in 2G.
In line with this argument, teachers' and immigrant students' ideas differed in terms of the language-community affiliation. Whereas for immigrant students learning Basque was an institutional rule, for teachers, who were autochthonous, acquiring Basque was a matter of ethnic identity and affiliation. All in all, this indicates that the link between identity, ethnicity, and language determined each party's interpretation of an interaction and the interethnic or intraethnic groupings that resulted. It also suggests that the applicability of Wimmer's (2008Wimmer's ( , 2013 proposal to the nature of ethnic boundaries is dependent on the parties involved on each side of the boundary.

Conclusions
The aim of this ethnographic study was to analyze the nature of multiethnic academic relations and their link with the theories of cultural capital, boundary-work, and Intergroup Contact Theory. First, the case study analyzed shows how some elements of cultural capital theory correlate with Lamont and Molnár's (2002) theory of social and symbolic boundaries. Students were ethnically and unequally distributed according to the instruction of Basque; as such, I argue that Basque acted as a cultural capital marker. However, the goal of linguistic models does not aim to ethnically distribute students. In fact, this distribution has a double nature: it is established as an institutional rule of linguistic models to ensure Basque language acquisition, and students choose to study in one of the classrooms corresponding to the level of Basque instruction. In the case study presented, students were influenced by the external forces that distributed them at school according to the institutional rule of learning Basque. However, students also showed strategies to move within the structure: whereas autochthonous students chose to enrol in classes with a high Basque instruction, non-autochthonous students enrolled in classes with a low Basque instruction, against school advice. Such advice was indicative of the 'rules of the game' in that setting, as learning Basque in the BAC enables community affiliation and upward social mobility.
Second, in this work, data suggests that Intergroup Contact Theory does not apply in the relationship between autochthonous and immigrant students, as in spite of the high ethnic diversity and cohabitation at school, processes of ethnic segregation took place between majority and minority groups. However, more research would be needed to measure prejudice among majority and minority groups to build upon this argument. Third, this study also tested Wimmer's (2008tested Wimmer's ( , 2013 proposal on the strategies pursued by actors for boundary-making and concludes that it takes account of the shifts in boundaries created between distinct groups. Interethnic relationships occurred among immigrant students and these were especially noteworthy when immigrant students opposed learning Basque.
Fourth, this paper proposes that boundaries are dependent on the interpretation of each party in an interethnic relationship and introduces language as an essential element for defining the nature of those boundaries. The differing value placed on Basque by each party resulted in immigrant students' lack of integration in the academic community. It is also indicative of the identity held by each group in relation to Basque: Basque identity was defined by the speaking and learning of Basque, which enabled academic community integration, and non-Basque identity referred to a lack of Basque learning and opposition to its learning, which prevented their integration.
A limitation in this study is that it analyzes cultural capital, boundaries, Intergroup Contact Theory, and identity in terms of ethnicity and other aspects are not taken into account. For example, gender or teacher-student relationships in a classroom authority framework are not explored. Ethnicity and its relationship to language were chosen as they were the most salient elements in the relationships between students and teachers. Although a limitation, this finding also indicates new avenues for future research in other ethnographic contexts, i.e. the relationship between language learning, cultural capital, Intergroup Contact Theory and boundary-work in other multiethnic academic settings.
All in all, this paper contributes to the theory of the cultural capital within boundary-work and proposes that language is a fundamental element for the definition of boundaries, social integration, and identity in multiethnic academic environments. More precisely, this paper proposes that the boundaries between minority and majority groups can be defined in terms of language and have important consequences for upward social mobility.
Finally, this study provides evidence for education stakeholders in the BAC to foresee some of the problems that may arise when teaching Basque to ethnic minorities in centres with a high proportion of immigrant students. More precisely, by training teachers to understand the reasons for resistance to learning Basque, and informing immigrant students' parents of the importance of Basque for their children's overall educational achievement, the social integration and academic success of immigrant students' may be improved.
Funding: This work was supported by the research project entitled "Identidad en interacción: Aspectos ontológicos y normativos de la individualidad biológica, cognitiva y social", reference: FFI2014-52173-P, funded by the Ministry of Economy and Competitivity of Spain and the doctoral fellowship funded by the University of the Basque Country (UPV/EHU), reference PIF 2013.

Conflicts of Interest:
The author declares no conflict of interest.