Potential Elitism in Selection to Bilingual Studies: A Case Study in Higher Education

1 Introduction

English Medium Instruction, or EMI, is the educational practice of using English as the vehicular language (in non-English-speaking contexts) to teach any sort of content except the language itself. Given an already lengthy tradition and recent rapid expansion, EMI has become an important field of research (Macaro & Tian, 2020). As would be expected, much of this research has focused on the fields of linguistics and pedagogy, where (in broad strokes) several lines of particular importance include: an analysis of the challenges faced by educators and students, especially in relation to minimal English skills (Butler, 2014; Hellekjær, 2010); the effectiveness of EMI programs, both in terms of improvements in the vehicular language (Martínez-Agudo, 2022a) and in relation to course contents (Anghel, Cabrales, & Carro, 2016; Dallinger, Jonkmann, Hollm, & Fiege, 2016; Hernández-Nanclares & Jiménez-Muñoz, 2017); assessments made of EMI courses by teachers (Aguilar & Rodríguez, 2012) and students (Aguilar & Rodríguez, 2012; Coyle, 2013); and issues of diversity (Madrid & Pérez-Cañado, 2018). Furthermore, and following Macaro and Tian (2020), EMI constitutes a fruitful domain for investigations into various educational policies and their implementation across different scales, ranging from the national level to how individual educators interpret these policies. This encompasses their implementation at the institutional level and the requisite curricular adaptation of content and methodologies.

Going beyond the strictly linguistic and pedagogical aspects of such programs, many authors have sought to evaluate bilingual studies from a social perspective. The impetus and interest of our research fall within the latter current, in the social inequality that bilingual studies may generate or reinforce, when the potential segregation of students whose English is not sufficiently strong at the beginning of university studies moves them to opt for exclusive instruction in the official language of their institution, thereby resulting in a diploma that may be deemed of lesser value, to the extent that “teaching through one single language is seen as second-rate education” (Lorenzo, 2007, p. 35).

This controversial idea of whether the presumed elitism of bilingual programs is myth or reality (Pérez-Cañado, 2020) has become the subject of its own scientific debate. As noted by Fernández-Sanjurjo, Blanco, and Fernández-Costales (2018), studies are lacking that examine the extent to which socioeconomic differences in families might lead to differences in student performance within bilingual programs. Such an oversight poses a scientific problem and might be contributing “to the creation (or enlargement) of a divide between an elite of young multilingual learners and those who could not access bilingual education” (Fernández-Sanjurjo et al., 2018, p. 17). Moreover, the reported scarcity of studies on the impacts of socioeconomic factors on student access and on grades obtained in bilingual studies is more pronounced at the university level, as the vast majority of published works have thus far focused on primary and secondary education.

The present research intends to help cover this gap detected by various authors, taking as our case study the degree in International Studies conferred by the Universidad Autónoma de Madrid (UAM) in Spain. In this course, there are two groups of students: those who study entirely in Spanish and those who pursue a bilingual program of subjects in both Spanish and English. The research will focus on assessing the influence of specific socioeconomic factors within the family, in conjunction with the disparate exposure of students to English during secondary education, on the selection and self-selection of students into bilingual or monolingual groups. Furthermore, the study will explore the outcomes achieved by each of these two groups. Ultimately, an analysis will also be conducted regarding the perspectives of both students and professors on the potentially divergent assessments that society and the individuals involved may make regarding the outcomes achieved by each group.

To this end, we have designed three interlinked research questions:

1. RQ1: Does socioeconomic level have an impact on the selection of a bilingual program by a student or on his/her acceptance into such a program by the institution?

2. RQ2: Is there a correlation between belonging to one group or the other (bilingual or monolingual) and the results obtained over the university career?

3. RQ3: Do those involved – teachers and students – consider that the social value of these studies will be equal, regardless of whether the degree has been studied in the bilingual or monolingual group?

The first two questions are intended to evaluate (based on our case study) the potential social elitism of bilingual studies at the university level, whether due to the weight that socioeconomic factors may play in the initial selection for the bilingual group or in light of the improved results that bilingual students can ostensibly obtain, as derived from relatively better training opportunities. The third question seeks to examine the potential role of bilingual studies in the reproduction of inequalities; should the hypothesis of segregation based on status be accepted (and verified), and given that a bilingual diploma will receive greater social recognition (and therefore greater economic rewards) than a monolingual diploma, then bilingual programs at public universities would both fail to fulfill a purpose of social rebalancing and act as a lever that reinforces prior inequalities.

2 Literature Review

EMI refers to the practice of using English as a vehicle for instruction at educational centers in areas where the majority population does not speak English as a first language (Macaro & Tian, 2020). According to Macaro and Tian, this concept refers fundamentally to the use of English as a vehicular language in higher or university education; for primary and secondary education levels, the preferred term is Content and Language Integrated Learning (CLIL). The concept of CLIL was coined in the 1990s (Marsh, 1994) in reference to educational methodologies in which a subject was taught in a foreign language, focusing on both the content and the language itself.

However, unlike CLIL methodologies focused on both content and language, there is little scientific evidence that higher education professors using the EMI method pay particular attention to the use and development of English by their students (Macaro & Tian, 2020). This has led some authors to propose more integrative methodologies grouped under the acronym ICLHE, for Integrated Content and Language in Higher Education (Macaro & Tian, 2020). As Ruiz-Madrid and Fortanet-Gómez (2023) indicate, this term appears to vindicate the “I” of “Integrated” in a university context where neither professors nor students are native English speakers. In any case, while the term ICLHE continues to be used, EMI is apparently gaining a presence in the literature expressly focused on higher education (Ruiz-Madrid & Fortanet-Gómez, 2023).

As articulated in the introduction, the purpose of our study is neither conceptual nor terminological. It does not center on the more directly linguistic aspects of English usage either but, instead, focuses on specific social implications associated with its use as the language of instruction in higher education. Therefore, we generally address all types of bilingual programs, following the definition by Barrios (2022): “all types of provisions in which a foreign language is used to teach subjects in the curriculum, other than language lessons themselves” (p. 183). Of course, as noted above, most of the prior research considered here has been focused on primary and secondary education and thus likely uses the acronym CLIL.

Aside from the problems of terminology and definition mentioned above, the entirety of the scientific evaluation of these bilingual programs is apparently subject to what Fernández-Sanjurjo et al. (2018) describe as a “pendulum effect.” Following an initial period of “unbridled enthusiasm” (Pérez-Cañado, 2016, p. 10), in which such programs seemed a panacea for all educational challenges, a number of authors soon began to address certain problems (both theoretical and applied) observed in bilingual methodologies. Here, the controversy upon which we direct both theoretical and analytical focus is that of the egalitarian or elitist character of these bilingual programs.

As regards this controversy around potential elitism, among those opinions most frequently cited in favor of bilingual programs is that of Marsh (2002) in a report for the European Union, stating that “[e]galitarianism has been one success factor because the approach is seen to open doors on languages for a broader range of learners” (2002a, p. 10). This author further remarks that CLIL programs have allowed “gender mainstreaming in terms of male and female performance in language learning” (Marsh, 2002, p. 11). The main argument behind this report by Marsh (2002) is that such programs offer educational opportunities within public systems that were previously available only to those who could afford private education, whether formal or informal (through private classes). Under this argument, not only would bilingual education not be elitist – at least in the public system – but it would represent a crucial element in reducing inequality in terms of access to education.

Marsh’s aforementioned ideas in defense of the egalitarianism and equity of bilingual studies refer essentially to the potential inequality of the moment of access to education in a foreign language, whatever the social status of a student. Fernández-Sanjurjo, on the other hand, suggests that bilingual studies may generate a second type of inequality, adding to the problem of access to the issue of whether the methodology “provides a suitable learning setting for students with different backgrounds (compared to traditional monolingual learning approaches)” (Fernández-Sanjurjo et al., 2018, p. 18). That is to say, following the determination of access, such programs could fail to adequately address the diversity of possible learning situations, including that of the initial level in the language of instruction.

In addition to the problems of initial access and attention to diversity is the possibility of further inequality at the end of the process, should bilingual education offer better educational results than those obtained in a monolingual program. Thus, we may distinguish between two debates around the possible elitism of such methodologies, although the two may overlap in most situations.

1. The problem of whether bilingual programs are equally accessible to all students regardless of family social class, gender, nationality, prior academic training, place of residence, etc.

2. The problem of educational outcomes obtained, depending on access conditions as well as varying perceptions of these results by society when, within the same degree, a bilingual course with English coincides with a monolingual course in the official language of the institution.

The problem of access is directly related to the selection or self-selection of students, to the extent that this is conducted by the institution or by the student or his/her family. A situation of self-selection would occur in the case of those who, perhaps justifiably, believe that they are not sufficiently trained to follow a bilingual program, or whose families cannot help in this regard (whether educationally or financially). Among works that most clearly cite the elitism of bilingual programs, those of Paran (2013) and Bruton (2011, 2013, 2015) can be highlighted for their argument that students are (self-)selected at the beginning of the process, implying unequal conditions for those admitted to monolingual programs only.

The first aspect to be taken into account when dealing with the selection of students is that of the criteria according to which selection is made. Pérez-Cañado (2020), who has analyzed selection processes in several European countries, finds that the most common criteria are the level of the language of instruction and knowledge of the subjects to be studied (a combination of the two being common).

As Lorenzo, Casal, and Moore (2010) indicate, behind these apparently objective selection processes may be hidden aspects of social class and parental choice. The latter can be crucial and linked to the educational levels of the parents, given that not all families are capable of perceiving the importance that such education may have for their children’s future. Whether or not they are aware of the potential value of bilingual education, families of lower status or education will be conditioned in their choice by their own inability to assist their children in their studies, either directly (due to low educational or English levels) or economically (due to the impossibility of accessing private classes, courses abroad, etc.). These factors, considered here as educational markers of social class that can establish class bias in a selection process, will be recalled in the empirical phase to determine the potential elitism of the program under study. In relation to this combination of educational and economic factors, Butler (2014) concludes her research on students in East Asia by noting that “parents with higher SES [socioeconomic status] tended to adjust their assistance according to their children’s changing needs; parents with lower SES tended to remain controlling and often failed to respond to their children’s changing needs to foster their self-competence and self-determined motivation” (Butler, 2014, p. 16).

In light of the aforementioned, one conclusion might be that, whatever the openness of a public bilingual system, the socioeconomic level of the parents will serve as a significant marker in determining access by students to that program. In this sense, Fernández-Sanjurjo et al. (2018) conclude that students enrolled in bilingual programs come from environments where the English language is prioritized, to the extent that parents see it as an added value in their children’s education.

Following these arguments, several authors have rejected the assessment of Marsh (2002) mentioned above. Bruton qualifies CLIL systems as being a kind of gatekeeper, emphasizing that they perform subtle pre-selection of students (Bruton, 2013). In the same vein, Paran (2013) argues that bilingual programs attract the most motivated, intelligent, and linguistically competent students, giving them greater opportunities than the rest. In any case, the final question is whether students are categorized based on their merits or whether “selection is done – to a greater extent – on the basis of socio-economic status of the family (and the learners’ motivation, intelligence, and linguistic proficiency to a lesser extent)” (Rascón Moreno & Bretones Callejas, 2018, p. 116).

Hüttner and Smit (2014) contest this discriminatory vision of bilingual systems, arguing “that CLIL in itself is a negligible factor compared to other, system-inherent sources of privileging some groups on socio-economic grounds” (2014, p. 161). This argument focuses on aspects that these authors consider inherently discriminatory in educational systems, including high fees and private institutions that advantageously coexist with public ones.

This notion of CLIL as an elitist system has also been empirically questioned by Pérez-Cañado (2016). According to her analysis of different empirical approaches, she concludes that “CLIL and non-CLIL classes have been found to be homogeneous on four different variables: verbal intelligence, motivation (where four factors have been considered: will, anxiety, disinterest, and self-demand), socioeconomic status, and extramural exposure to the foreign language” (Pérez-Cañado, 2016, p. 19). Her conclusion is that these results disarm “the belief – in our context, clearly unsubstantiated – that the most intelligent, motivated, and socially privileged students are those found in CLIL streams” (Pérez-Cañado, 2016, p. 19).

Given the importance of this issue and the scientific disagreement around it, it seems more prudent to adhere to the position of Hüttner and Smit when they point out that “educational discrimination (…) would require a lot more focused attention, political as well as academic” (Hüttner & Smit, 2014, p. 162).

The second aspect to be addressed regarding equality vs elitism in bilingual studies must focus on the question of achievement, i.e., on comparative analysis of the grades obtained by students in monolingual and bilingual programs. If we accept the hypothesis of initial segregation, and also that bilingual students later obtain more favorable academic results, then these systems would not only assume existing social inequality but also reproduce it, and possibly legitimize it. As occurs in many aspects related to these educational methodologies, clear scientific disagreement again prevails.

Paran (2013) points out that the CLIL system benefits only what he calls “great achievers” in elite schools and contexts. In a similar vein, Anghel et al. (2016) highlight that divergent results have been identified not only between CLIL and non-CLIL students but also, within the former group, differences are observed in favor of those with higher socioeconomic levels. These authors find a correlation between socioeconomic level and achievement in bilingual programs (Anghel et al., 2016).

In contrast to these affirmations and interpretations, it has been argued that the socioeconomic level of the family is indeed a significant variable when measuring the educational results of students, but that this is not peculiar to bilingual programs: “In general, students whose parents had a higher educational level outperformed those whose parents had lower levels of education, both in CLIL and non-CLIL” (Pérez-Cañado, 2020, p. 6). Therefore, this would not constitute a negative factor when evaluating such methodologies, which might conceivably be blamed for not sufficiently struggling against socioeconomic discrimination in education but not for promoting it.

To conclude this review, we should repeat that most of the studies cited are based on analyses of educational programs at the primary or secondary levels of education; the paucity of research at the university level is evident as it corroborates one of the few existing works (Aguilar & Rodríguez, 2012). One point in favor of bilingual university education is that existing studies appear to find no significant variations in the outcomes between groups in English and those in a native language (Aguilar & Rodríguez, 2012).

3 Methodology

The case study through which we address our research questions is the degree program in International Studies offered by the Department of Philosophy and Letters (Facultad de Filosofía y Letras) of the UAM (Spain). Launched in the 2014–2015 academic year, this program has at the time of writing been in effect for nine academic years.

The chief characteristic of this degree, making it an ideal case for our purposes, is that its students are divided into two groups: one entirely monolingual, in Spanish, and another bilingual, with more than a third of the subjects taught in English and the rest in Spanish, although this proportion may vary slightly depending on the year. This mode of organization implies that there are three types of subjects within the degree: those taught in Spanish for all students; those taught in Spanish only for students in the monolingual group; and those taught in English only for students in the bilingual group.

At the beginning of the program, students can request affiliation to the bilingual group if they possess sufficient certification in English (B2 level, according to the Common European Framework of Reference for Languages or CEFR) or else pass a test given by the degree program’s own administration. According to the data provided by the department, during the 2021–2022 academic year, an average of 46 students were enrolled per subject, distributed between the monolingual and bilingual groups in a relatively balanced way.

In our analysis, we worked with three methodologies and three types of data, with the first being a survey of students conducted to produce quantitative data. Questions were asked about the educational and language levels of students’ parents as well as the students’ formative experience during secondary education, especially as regards increased possible exposure to English (both considered educational markers of social class).

These anonymous surveys were conducted in Spanish during the school term in the second semester of the 2021–2022 academic year. The survey was carried out during class time in one of the subjects taught in both Spanish and English, with the permission of the professor. In each of the groups surveyed, express permission was requested orally. In total, from among the four courses and the two groups (bilingual and monolingual), 131 questionnaires were considered valid (two, submitted practically blank, were discarded). This meant coverage of just over 70% of the total students enrolled in the degree at the time of this empirical work (at 185, according to data from the UAM’s Department of Philosophy and Letters). To determine the effect on the choice of bilingual and monolingual options of variables related to the academic level of the parents and the educational opportunities of the students during their high school studies, logistic regression models were carried out at the univariate level.

Second, to compare the results obtained by the bilingual and monolingual groups, analysis was made of the grades obtained by the students in seven of the eight completed academic years. It was not possible to use data from the first academic year (2014–2015), as the students’ records did not differentiate between the bilingual and monolingual groups. These data were provided by the Department of Philosophy and Letters of the UAM, with corresponding guarantees of anonymity. With these data, two indicators were calculated that facilitated the comparison of the groups: the success rate (percentage of students who passed each subject in the first examination) and the average grade (calculated according to the Spanish system of 0 minimum to 10 maximum).

Finally, semi-structured interviews were conducted with the professors instructing the bilingual group (by email, following initial contact in person or by phone) to obtain qualitative information on their experience as well as their assessments of the studies and their students. At the time of research, 19 professors at the university taught subjects in English (some subjects being shared by more than one professor). All were contacted, but ultimately only nine responded to the request for an interview.

4 Results

Within the International Studies program at the UAM, the only criterion for selection to the group that undertakes a portion of studies in English (following passage of the minimum cut-off mark) is the academic level of a student in that language at the beginning of the course. Therefore, the only way to prove elitism in the bilingual group from a social perspective would be to demonstrate a connection within this group between the level of English and the socioeconomic status of the family, affording some students certain opportunities to improve their English skills, which others of lower status would not have had. This would indicate a class bias in the selection of students.

To this end, the first factor to be taken into account is whether or not (self-)selection has occurred, i.e., if some students who had wanted to join the bilingual group were not admitted. The answer to this question is clearly yes since approximately 20% of the students surveyed in the monolingual group stated that they requested entry into the bilingual group but were not admitted. Moreover, 11.6% of respondents indicated that they did not request the bilingual track because they knew their level of English to be insufficient. In other words, in addition to a selection made by the institution, some negative self-selection was made by the students themselves, as mentioned in the Literature Review. Furthermore, a small segment of the monolingual group (6.9%) began their studies in English but soon switched to monolingual, largely for reasons of language level. Thus, we find that just over 38% of students in the monolingual group remained outside the bilingual program either through selection or by self-selection.

4.1 Educational Markers of Social Class and (Self-)selection to the Bilingual Program

To test whether this (self-)selection effectively implies class bias, the survey asked students (in both the bilingual and monolingual groups) about their access and use during secondary education of what we have termed educational markers of social class (see left column in Table 1). The aim was to detect a possible correlation between the availability during high school of a series of opportunities to increase exposure to English, which might explain their language levels and, consequently, their acceptance to the bilingual group at university. The results obtained after statistical analysis are reflected in Table 1.

Table 1

Effect of academic variables on the choice of the bilingual stream by students in the IS degree at UAM

	Bilingual stream, n (%)		Univariate logistic regression
	No	Yes	OR (IC 95%)	p-value
Type of high school
Public	28 (45.2)	34 (54.8)	1
State-funded (concertada)	24 (53.3)	21 (46.7)	0.72 (0.33–1.56)	0.404
Fully private	9 (42.9)	12 (57.1)	1.10 (0.41–2.98)	0.854
Bilingual studies in high school
No	40 (55.6)	32 (44.4)	1
Yes	20 (36.4)	35 (63.6)	2.19 (1.07–4.49)	0.033
Official English level certification
No	18 (90)	2 (10)	1
Yes	43 (39.8)	65 (60.2)	13.61 (3.00–61.64)	0.001
Full course abroad
No	45 (52.9)	40 (47.1)	1
Yes	16 (37.2)	27 (62.8)	1.90 (0.90–4.02)	0.094
English course abroad
No	39 (58.2)	28 (41.8)	1
Yes	23 (37.1)	39 (62.9)	2.36 (1.16–4.79)	0.017
Private classes in English
No	27 (44.3)	34 (55.7)	1
Yes	34 (49.3)	35 (50.7)	0.82 (0.41–1.63)	0.568
OR: odds ratio IC: confidence interval

Source: Own elaboration based on student survey data.

Note: Bold indicates statistically significant values due to having a p < 0.05.

The educational marker of social class presenting the greatest impact on being selected for the bilingual group (lower p-value) is the possession of an official English level certification. Those with certification (of any type or level) were 13.61 times more likely to be selected than those without. This is not surprising, since to be selected for this group, one requirement is to have just such a certification (minimum level B2, in accordance with the CEFR), although this can be substituted by a test implemented by the program. In the case of those who obtained a level certification during secondary education, both the linguistic requirement (having at least an average level of English) and the administrative requirement (being able to prove it) were met; hence the high correlation.

A more detailed analysis of the types and levels of certifications achieved (Figure 1) reveals that students in the bilingual group had more certifications and also possessed a higher level of certification. This reinforces the idea of English certification as an educational marker of social class that can prove certain class bias within the bilingual group, since the majority of (self-)selected students were those with an average level or higher (Advanced and Proficiency), often explained by reinforcements received outside compulsory education.

Figure 1

Official English certifications according to (bilingual/monolingual) track. Source: own elaboration based on student survey data.

The second of the variables being considered as a potential marker of social class, which can predict (self-)selection to the bilingual group – albeit with a much lower level of significance than the previous (p-value = 0.017) – is that of having taken English courses in an English-speaking country. Those who did so were found 2.36 times more likely to be (self-)selected than those who did not, and the result is that those (self-)selected for the bilingual group were almost twice as many as those who had not taken languages in English-speaking countries (62.9% vs 37.1%),

Finally, the third variable (an educational marker of social class) with statistical significance as a predictor of (self-)selection to the bilingual group was that of having completed secondary education in a bilingual program (OR = 2.19, p = 0.033, Table 1). As in the above case, practically two-thirds of students in the bilingual group (63.6%, compared to 36.4% in the monolingual course) completed their secondary studies in a program of this type.

The two latter variables (having taken English courses in an English-speaking country and having studied in a bilingual high school program) – the impacts of which are similar as a predictor of (self-)selection to a bilingual program – may engender controversy and potentially divergent interpretations in terms of their consideration as markers of class bias in these programs. Taking English courses abroad might be presumed a clear educational marker of social class, to the extent that such stays are paid for by the families’ educational budgets. On the other hand, the fact of having studied secondary education in a bilingual program could be valued either as a factor of inequality (following Bruton and Paran) or as a leveling factor (following Marsh), given that in the Autonomous Community of Madrid, where our focus of analysis is located, the bilingual model of secondary education is integrated into the public school system.

The potential impact of family status on (self-)selection to the bilingual group – and, therefore, the proof of possible elitism of the program analyzed – is diminished when considering the null statistical relationship of the final three variables (educational markers of social class) to be considered as predictors of integration into the bilingual group. No statistical correlation is found between having been (self-)selected to the bilingual group and having completed secondary education in a private or subsidized educational institution (p-value greater than 0.05 in both cases, Table 1). In other words, having a family able to pay for a private education (in which attention to languages may be greater than in public school) does not imply an advantage for integration into the bilingual group. Likewise, the variables of having completed an entire course abroad or having regularly attended private English classes (Table 1) show no relationship with ultimate integration into the bilingual group (in both cases, the p-value is greater than 0.05).

It must be taken into account that these three variables are among the investments in education that present the highest costs for families; therefore, they may be viewed as very clear educational markers of social class. Nevertheless, among the variables shown to be clear predictors of integration into the bilingual group, only the fact of having taken English courses in English-speaking countries seems to carry the same significance as an educational marker of social class as those previously mentioned.

5 Discussion and Conclusions

In this section, we establish a discussion between the scientific work outlined in the literature review (again, based fundamentally on analyses at the primary and secondary levels of education) and the quantitative data resulting from this research (addressed in the previous section). Opinions from a third source will be added to this discussion: qualitative information, likewise produced for this research and collected through semi-structured interviews with professors who teach the bilingual group.

The first argument verified by our case study is that at the UAM program, a kind of negative (self-)selection is perceived for those students who hope to integrate into the bilingual group but cannot do so. At the same time, a kind of “positive self-segregation” has been detected, which may conceivably lead a student to forgo, and even to challenge, the bilingual education model. This refers to students who, despite a sufficient level of English accreditation (or else thinking themselves capable of passing the program’s internal language level test), do not apply to the bilingual group for various reasons (perhaps expecting better results or greater “comfort” levels in their native tongue and thus preferring to study in Spanish). This notion of “positive self-segregation” has not been found in the reviewed literature, but evidently, such students should not be considered mere “remnants,” as in the terminology of Bruton (2013, p. 593).

In order to provide evidence for the debate on potential elitism, the discussion should revolve around whether selection based on the English level (a division among students that is explicitly manifested in the structure of the degree program) is rooted in a socioeconomic rationale. As mentioned, the argument in favor of elitism would be that students are not selected through strict academic meritocracy but in a way mediated by the economic capacities and academic levels of their families, who have offered them more and better opportunities for exposure to English (Bruton, 2013, 2015) in the years leading up to university.

The best argument that can currently be offered in favor of the theory of elitism would be based on the positive correlation between having been selected for the bilingual group and two variables considered potential educational markers of social class: having completed at least one language stay in an English-speaking country and/or having completed high school studies in a bilingual program. Interestingly, a significant number of professors interviewed from the bilingual group draw the same logical connection: most students in the bilingual group have a high English level due to relatively intense exposure to the language, as compared with other Spanish students, and this is feasible, in their opinion, thanks to a relatively high socioeconomic level.

Any of the aforementioned variables (such as stays abroad) would be considered a clear marker of social class (Bruton, 2015; Fernández-Sanjurjo et al., 2018). However, as yet undetected is any debate around the point at which such variables as the socioeconomic level of the family or the potential capacity to access relevant academic services would begin to produce a class bias and establish patterns of social inequality in access to public education.

More controversial is the debate on the variable of bilingual education during secondary schooling, a potential educational marker of social class that shows a clear correlation with integration into the bilingual track of the university degree. The statistical connection demonstrated in this analysis indicates elitism around the bilingual group of International Studies at the UAM only when accepting the further premise that bilingual programs at the primary and secondary levels are intrinsically discriminatory.

This is the line taken by authors such as Bruton (2015) and Paran (2013), who argue that CLIL or bilingual programs do not meet the obligation of public education to respond to diversity and to thereby offer equal opportunities to all students, whatever their initial situation (including language level). If this argument is accepted, then the statistical connection indicated above would clearly indicate certain segregation or elitism in the (self-) selection of students to the bilingual study track in our case study. As shown by the statistical correlation, greater opportunities for access would thus be offered to the English (bilingual) program, not only for those whose socioeconomic status has afforded greater possibilities for exposure to English (by taking courses abroad), but also for those who (for similar reasons) have been able to develop their pre-university secondary education in bilingual programs.

For other authors, bilingual education at the primary and secondary levels is not only non-discriminatory (and thus unlikely to generate class bias at higher levels of education), but in fact offers clear opportunities for social equity by allowing quality instruction in English to socioeconomic strata that could not otherwise afford it (Fernández-Sanjurjo et al., 2018; Hüttner & Smit, 2014; Marsh, 2002). In the words of Fernández-Sanjurjo et al. (2018), “CLIL and non-CLIL groups can be regarded as homogeneous, and no statistical differences can be found according to SES” (socioeconomic status) (p. 19). The same conviction was shown by some of the UAM professors interviewed, who considered that bilingual education (whether at university or lower levels) is a factor in social leveling, extending an opportunity (at public prices, however high) to students from lower socioeconomic backgrounds.

Without entering into a discussion of whether bilingual programs at the primary and secondary levels are discriminatory or not, we can highlight certain indications that appear to support the conclusion that – strictly in the case under analysis – any potential segregating effects would be quite limited. The first of these is the null statistical relationship between having been (self-)selected for the bilingual group and having completed high school in a private or subsidized institution.

Second, also found to be null, is the statistical relationship between the educational levels of a student’s parents and the opportunities offered (such as studies abroad, private education, private classes, etc.) for achieving a high level of English. Indeed, although some aspects linked to better education in the English language do offer improved chances of a student being accepted into the bilingual group, it remains unproven whether this issue is necessarily linked to family status.

This is consistent with some of the empirical work seen in the literature for other educational levels, such as that of Pérez-Cañado (2016), in which she finds no differences between CLIL and non-CLIL students in relation to “extramural exposure to the foreign language” (p. 19). Our conclusion is, however, subject to objection due to methodological issues stemming from the limitations of this work, such as the possible weakness of the parents’ educational levels as a (proxy) predictor of the family’s socioeconomic level. Nevertheless, this factor has been used in other research of this type (Alonso et al., 2008; Martínez-Agudo, 2022b; Rascón Moreno & Bretones Callejas, 2018; Sirin, 2005).

Even so, given the data here obtained, it follows that students with parents of low or medium educational levels did not, in this case, suffer discrimination in access to sufficient English instruction (and thus in potential access to the bilingual track) as compared to students from families with high educational levels. This marks a clear contrast with work carried out in other contexts and at other levels of the educational system, where a connection was indeed observed between the educational level of the family, the socioeconomic status, and the level of exposure of the children to foreign languages, with the conclusion that “children of parents from different socioeconomic backgrounds and with different educational levels are not homogeneously distributed in all types of school” (Muñoz, 2008, p. 589).

Because only certain educational markers of social class have proven good predictors of integration into the bilingual group, and given that no statistical connections have been established between the parents’ educational and language levels and the students’ possibility of greater exposure to English, it cannot be said that the students, in this case, are distributed in a pattern similar to that indicated by Muñoz (2008). Indeed, the most that can be said is that, despite some indications, in the absence of additional in-depth and consistent research, elitism cannot be proved in the selection of students for the bilingual group in the UAM International Studies degree. The qualitative responses of professors teaching in this program reveal similar ambivalence, with a near-even distribution between those who consider the bilingual-track students to represent an academic and idiomatic elite (derived somewhat from social elite status) and those who claim not to have perceived a segregating effect.

As regards the weight that the educational and language levels of parents may exert on the English levels of their children (and thus the chances of being selected for a bilingual group at university), the specialized biography presents studies of this issue at different levels and in different contexts (Fernández-Sanjurjo et al., 2018). For example, Muñoz (2008) points out that “the educational level of parents has a significant influence on children’s foreign language learning success (see for example the PISA report in Europe 2003)” (p. 589). Martínez-Agudo (2022b) confirms this relationship, establishing 65% as the proportion of students in CLIL programs whose parents possess a university degree (our data are very similar to those reported by this author). In any case, we must cite the clarification made by Pérez-Cañado (2020) in her exhaustive research, in the face of such empirical evidence: “this is clearly the case of all students, whether they are following CLIL or non-CLIL programs” (p. 15). In other words, although the educational levels of parents do clearly influence educational decisions, as well as the results obtained by their children, in no case has this effect been found to be caused by CLIL methodologies or bilingual programs.

Although our analysis has detected a connection between the educational and language levels of the parents and the selection for the bilingual group, this has only been established in the case of the father and not of the mother. No similar findings have emerged from the literature on bilingual education, although some general works have reflected on the potentially distinct roles that the educational levels of a father or mother may play in both the educational level and achievement of a child (e.g., Korupp, Ganzeboom, & Van Der Lippe, 2002; Minello & Blossfel, 2017). This is part of a much broader debate around gender and behavior models not directly relevant to this work; the most that can be said here is that in the case analyzed, (self-) selection to the bilingual group in relation to the parents’ educational levels has been found to be sensitive to the parents’ gender.

Moving ahead to a discussion of academic results, it seems evident that in this case no special selection was made of what Paran (2013) terms “great achievers.” Indeed, the indicators used to measure achievement in the two groups do not reveal that bilingual students obtained better grades than their monolingual peers at the end of the degree program. This is consistent with other findings in the scientific literature, including examples attending to the university level in Spain (Aguilar & Rodríguez, 2012). Moreover, Anghel et al. (2016), Lorenzo et al. (2021), and Pérez-Cañado (2016) report homogeneous results between students from CLIL programs and monolingual groups at other, lower educational levels.

In clear contrast, among the professors interviewed, a strong current of opinion considered that students in the bilingual group can indeed be considered “great achievers,” in keeping with Paran (2013). Among those who expressly maintained this opinion, a significant portion felt that this higher grouping of bilingual students manifests not only in terms of knowledge and results but at the attitudinal level, indicating greater competitiveness.

Certain agreement prevails between the quantitative and qualitative results offered by the professors as regards the differing assessments that society will likely make of the studies undertaken and therefore the differing rewards (in both prestige and earnings) that graduates may receive, depending on their adherence to a bilingual or monolingual cohort. That is to say, both students and professors consider that the monolingual track will be viewed to a certain extent as a lower-level degree (Lorenzo, 2007), in contrast with the real achievements attained by both groups, which are very similar.

Although the students’ reasons for entering the bilingual group include both instrumental motivations (aimed at greater socioeconomic returns from their studies) and inherent motivations (the intrinsic pleasure of improving one’s English and using it as a means for communication and culture) (Somers & Llinares, 2021), the economic instrumentalization of education is difficult to transcend. This is especially clear in the case of knowledge of English, viewed largely as an instrument rather than a vital manifestation of culture by a neoliberal vision of education that only “sees language proficiency as the main causal factor in economic success” (Paran, 2013, p. 335).

One possible explanation for the results obtained, where clear (self-)selection was detected but without clear patterns of discrimination for socioeconomic reasons, may lie in the educational level at which this experiment was carried out: higher education, compared to an overwhelming majority of research and reflections focused on primary and secondary education. Thus, it is possible that at the university level, no clear segregation exists between students in the bilingual and monolingual groups because the university is itself selective; that is, selections based on status have already taken place. As Piketty (2022) points out in one of his latest works, the fact that public higher education is often nearly free (which is not the case in Spain, much less in the Community of Madrid, with the country’s highest prices) does not prevent it from causing social selection: “The pursuit of extensive training in institutions of higher learning in the absence of an adequate system of minimum income represents a considerable investment for people with lower incomes. Moreover, these students have not always had access to the preliminary training, the social codes, and the networks that provide opportunities for certain courses of study” (Piketty, 2022, p. 178). In other words, as Bourdieu theorized half a century ago, the greater or lesser endowment of a certain cultural capital should be added to the economic availability and/or the assessment of the opportunity cost of going to university (Bourdieu, Boltanski, & de Saint Martin, 1973) that allows the potential student to consider, or even just imagine, access to certain levels and academic spaces.

Therefore, it would be difficult to detect discrimination at the university level, even in the public system, if such has already occurred. Furthermore, a significant percentage of those who have attended private institutes during secondary (high school) education decide to continue their studies at private universities and/or to follow educational paths with a different profile than the one under analysis. In other words, if (self-)selection occurred before reaching the educational level, type of institution, and area of study that comprise our case study, then it is very likely that we are obtaining a level of student homogeneity (in both family status and academic performance) that is a product of prior biases.

Another discussion worthy of attention is whether, from the perspective of segregation and equal opportunities, separation for academic reasons (in this case, language level) carries the same meaning in primary and secondary education as it does in higher education. Even under the assumption that primary and secondary bilingual programs are inherently elitist, as some authors maintain, neither the meaning nor the impact of segregation for academic reasons would seem to be the same as regards university education. Indeed, as certain professors remarked in interviews, this would serve a leveling purpose by being offered at a public university at public prices – as has been claimed by various authors focused on the lower echelons of education (Fernández-Sanjurjo et al., 2018; Hüttner & Smit, 2014; Marsh, 2002).

It must further be taken into account that such segregation (for academic reasons) already occurs for students in both the bilingual and monolingual groups, as compared to those not accepted into the degree program. Prior to any separation based on the language of instruction, a previous selection process is made for academic reasons, the criteria of which focus on the grades obtained during secondary education.

We therefore fully agree with Pérez-Cañado (2020) when she states that the relevant conclusions in her own study are once again “mixed, if not contradictory” (p. 7). Likewise, as relates to our research questions and to novel doubts raised in this Discussion section, we find that new and more specific research will be necessary before the questions prompted by this educational model can be answered.