Academic achievement in a language revitalization context: A study on the influence of language and socioeconomic factors

RUNNING HEAD: MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION Altavilla, J., Elosua, P., & Solano-Flores, G. (2023). Academic achievement in a language revitalisation context: a study on the influence of language and socioeconomic factors. Journal of Multilingual and Multicultural Development, 44(10), 1092–1107. Academic achievement in a language revitalization context: A study on the influence of language and socioeconomic factors This is an Accepted Manuscript of an article published by Taylor & Francis in Journal of Multilingual and Multicultural Development on 17 Jan 2021, available at: https://doi.org/10.1080/01434632.2020.1865382. It is deposited under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http:// creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 2 Academic achievement in a language revitalization context: A study on the influence of language and socioeconomic factors Abstract We examined the influence of language and socioeconomic factors on academic achievement in the Basque Autonomous Community (BAC): a multilingual society with a recent history of language revitalization. We analyzed scores obtained by Grade 4 and Grade 10 students on a diagnostic mathematics test administered in either Basque or Spanish in the BAC. We found that L1 Basque users who were of high socioeconomic status (SES) were the top performers across language groups. A larger percentage of high-SES L1 Spanish users earned top scores when tested in their home language. In contrast, low-SES L1 users of Spanish performed better when tested in Basque. These findings corroborate and expand upon findings from previous research conducted in the BAC (Elosua & Egaña, 2017; Flores, 2019; Santiago, Lukas, Joaristi, Lizasoain, & Moyano, 2008), which show that SES and language shape academic achievement in contexts of successful language revitalization efforts. Keywords: Basque, language revitalization, language variation, socioeconomic status MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 3 Introduction This paper explores how language and socioeconomic factors relate to academic achievement in multilingual societies that are undergoing language revitalization. Worldwide, approximately 2,500 languages are endangered or recently extinct (Moseley, 2010). Around the globe—and particularly in Europe—some endangered languages are being revitalized through the education system. This is the case for Irish (Ceallaigh, & Dhonnabhain, 2017; O’Sullivan, Bird, & Burns, 2019) and Welsh (Jones, 2016). In Spain, schools have been sites of revitalization for the autonomous languages of Catalan, Galician, and Basque (Lasagabaster, 2017). We examine academic achievement in the context of the revitalization of Basque. We focus specifically on the Basque Autonomous Community (BAC), a region in Spain where most L1 Basque users reside (Cenoz & Gorter, 2019). The knowledge generated herein is timely. Spain, following a global trend, has begun to use test scores to drive educational policy reform, measure accountability, and increase the transparency of political agendas (Verger & Curran, 2014). In recent years, scholars and policymakers have begun to examine the standardized test scores of students within the BAC (Flores, 2019; Gortazar, 2018). Given that students are, by and large, taught in the minority language in the BAC, it is imprudent to analyze academic achievement without accounting for the language revitalization context. It is particularly important to consider that high enrollment in Basque-medium schools— while indicative of revitalization—may belie underlying inequities. Since 1983, families in the BAC have been able to choose from three linguistic models (Sierra, 2008). Model A schools are those where Spanish is the language of instruction and Basque is taught as a subject. In Model B schools, content is taught in Spanish and Basque. Basque is the language of instruction (and Spanish is a subject) in Model D schools (Lasagabaster, 2017). As of 2017, 96% of primary- MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 4 school students and 91% of secondary students attended schools where Basque was one of the languages of instruction (Cenoz & Gorter, 2019). Although most students receive all or some of their instruction in Basque and take standardized assessments in that language, most BAC residents speak Spanish at home (Cenoz & Gorter, 2019; Gortazar, 2018). The misalignment between home language and language of instruction and assessment may impact academic achievement. Evidence from the United States suggests that students typically perform better on tests that match their home language (Abedi, 2002; Solano-Flores & Trumbull, 2003). This has also been shown to be true for students in the BAC (Elosua, 2016; Elosua, & Mujika 2014; Gortazar, 2018). Thus, Basque-medium schools—the most common of the three linguistic models of schools (Cenoz & Gorter, 2019) —may privilege the relatively few L1 users of Basque over L1 Spanish users. Socioeconomic status (SES) may intersect with language to exacerbate educational inequity in the BAC. Numerous studies of large-scale assessment data show that SES is positively correlated with academic performance (Akhtar & Niazi, 2011; OECD, 2016; Sirin, 2005; White, 1982). In the BAC, SES has been shown to interact with language of instruction or home language to influence achievement in math (Elosua & Egaña, 2017) and other subjects like science and reading (Flores, 2019). The relationship among language, SES, and academic achievement is especially critical to understand in regard to mathematics. Although mathematics is commonly assumed to carry a lower linguistic load than other subjects, it is, in fact, characterized by language-related challenges (Abedi & Lord, 2001; Solano-Flores, BarnettClarke, & Kachchaf, 2013; Halliday, 1978). Thus, in the study herein, we focus on mathematics achievement. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 5 Purpose of the Study The relationship among SES, academic performance, and language in the BAC is wellestablished (Elosua, 2016; Elosua & Egaña, 2017; Elosua, & De Boeck, 2019; Flores, 2019; Gortazar, 2018; Santiago, Lukas, Joaristi, Lizasoain, & Moyano, 2008; Vega-Bayo & Mariel, 2015). However, the ways in which those factors impact the relative distribution of mathematics scores has not yet been addressed. Using measures of mathematics achievement from a largescale test, we examine test score differences between students of different SES and different home languages tested in either Basque or Spanish. First, using traditional analyses of variance, we examine the statistical significance of mean score differences attributable to the main and interaction effects of SES and language of assessment, by home language. Second, using score distributions, we examine the relative frequencies of students, by socioeconomic strata, home language, and language of assessment. Together, these complementary approaches aid us in exploring this question: How are SES, home language, and language of assessment related to mathematics achievement among elementary and secondary students in the Basque Autonomous Community (BAC)? Before delving into our methodology and results, we summarize the sociolinguistic context of the BAC and describe the role that schools have played in the language revitalization process. Then, we explain the connection between language and score variation on mathematics tests. Context of Basque Revitalization Although Basque and Spanish have coexisted within the BAC for centuries, a host of forces have gradually shifted language use toward Spanish (Lasagabaster, 2017). Those forces have included (but have not been limited to) urbanization, industrialization, and, under Franco’s MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 6 dictatorship, political pressure for a monolingual state (De Luna, Zalbide, Darquennes, & Suberbiola, 2014; Zalbide & Cenoz, 2008). In addition, autonomous languages like Basque only have co-official status within their respective communities; Spanish is the official national language (Lasagabaster, 2017). Although social status of Basque relative to Spanish has gradually improved (Aldekoa and Gardner 2002), Basque is still an unlikely candidate for successful revitalization. Users of this language are scattered across three small regions (two in Spain and one in France). Furthermore, Basque is the only existing non-Indo-European language in Western Europe (Elosua & Egaña, 2017). While Basque has been in contact with Spanish and French for centuries, it bears little syntactical or morphological resemblance to either language (Cenoz & Gorter, 2019). This fact is critical to our study because the distance between Basque and Spanish means that taking mathematics tests in Basque may be challenging for L1 Spanish users. Another detail about the language revitalization context that is important to our study is that the standardized Basque (‘euskera batua’) taught in schools in the BAC coexists with many regional varieties (Hualde, & Zuazo, 2007; Zalbide & Cenoz, 2008). Students who speak a variety other than standardized Basque may encounter challenges in school. For example, users of regional varieties may not know mathematical terms in standardized Basque or may be unfamiliar with formal mathematical discursive styles in Basque. Local varieties may have lower status in schools, as compared to standard Basque (Silverstein, 2003). In other words, both L1 Spanish users and users of regional dialects of Basque who attend Basque-medium schools are faced with the double challenge of being tested in mathematics content while they are continuing to develop their ‘euskera batua’ skills. Basque Revitalization and Schools MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 7 Students in the BAC are now being taught in a language that their grandparents were forbidden to speak. During Franco’s dictatorship, public use of all autonomous languages was prohibited (Cenoz, 2008). Throughout that time, informal Basque-medium schools operated clandestinely, sometimes out of private homes. Material resources were scarce, in part because written Basque was not standardized until 1968, in the final decade of Franco’s regime (Lasagabaster, 2001). After Franco’s death and the re-establishment of the Spanish democracy, the revitalization movement was taken up primarily by the Basque government (Lasagabaster, 2001). Still, widespread formal education in Basque only began in 1982, after the Law of the Normalization of Basque guaranteed access to education in that language (Sierra, 2008). Today, although communities and families also play vital roles in the preservation of Basque (Aldekoa & Gardner, 2002; Elosua and Egaña, 2017), schools continue to be the main drivers of large-scale revitalization efforts (Gorter & Cenoz, 2011). Even in areas with many L1 Basque users, Basque is not widely used outside of school or the home (Cenoz & Gorter, 2019; Valadez, Etxeberria, & Intxausti, 2015). That is because Spanish and, increasingly, English are utilized in more prestigious social, political, and economic spaces (Cenoz & Gorter, 2019; Lasagabaster, 2017). In addition, overall numbers of Basque users in the BAC are still low (Cenoz & Gorter, 2019) and concentrations of Basque users are unevenly distributed across the region (Altuna & Urla, 2013; Cenoz, 2008; Nelde, Strubell, & Williams, 1996). For these reasons, schools are still critical to language revitalization. Sources of Score Variation on Mathematics Tests Researchers in the United States (Solano-Flores & Li, 2009) and in the BAC (Elosua, 2016) have found that score variation on mathematics tests is related to language variety. Language variety is a particularly critical factor to consider in the BAC because a standard written variety MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 8 of Basque was not adopted until 1968 (Hualde, & Zuazo, 2007). To this day, multiple regional varieties of Basque are still utilized (Cenoz & Gorter, 2019). Consequently, students who use Basque dialects at home may not learn standard Basque until they enter school (Zalbide & Cenoz, 2008). L1 Spanish students and their L1 Basque peers who use dialects of Basque at home might, therefore, struggle with mathematics test items written in standardized Basque. Socioeconomic status (SES) may also contribute to score variation on mathematics tests for students in the BAC (Elosua & Egaña, 2017; Vega-Bayo & Mariel, 2015). In the BAC, most students—regardless of SES or home language—are educated in Basque (Cenoz & Gorter, 2019; De Luna, Zalbide, Darquennes, & Suberbiola, 2014). One possible implication is that high-SES, L1 Basque users in Basque-medium schools may outperform L1 Spanish users from similar income levels on standardized mathematics tests. That may be due to the fact that Spanish users in those schools are neither instructed nor typically tested in Spanish. Methods Participants The database contained sociodemographic information and mathematics test scores for 18,298 students in Grade 4 and 16,981 students in Grade 10. We applied two inclusion criteria: (1) participants had to be born in the BAC, and (2) participants had to self-report either Basque or Spanish as their home language. Although immigrants make up an increasingly large proportion of the population in the Basque Country (Gortazar, 2018), we opted to exclude them from our analyses. Immigrants to Basque communities tend to be from lower socioeconomic backgrounds, are less likely to attend Basque-medium schools, and are more likely speak non-Romance languages at home (Cenoz & Gorter; Gortazar, 2018). For these reasons, the distribution of their mathematics scores may be MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 9 quite different from that of L1 Spanish or L1 Basque students. Indeed, those differences are so consequential that we believe a separate study is warranted, which examines the achievement of immigrant students relative to students born in the BAC. (Note 1) Variables Socioeconomic Status (SES) In this original database, students were categorized according to five SES levels, based on the Index of Socioeconomic and Cultural Status (ISEC). (Note 2) To facilitate interpretability, we aggregated cases into three categories: low-, middle-, and high-SES. The distribution of students by SES was roughly the same for Grade 4 and Grade 10: Approximately 40, 20, and 24 percent of the students were respectively low-, middle-, and high-SES. Home language (L1) Home language was either Spanish or Basque, as self-reported by students’ families. L1 is the language most commonly (always or almost always) used at home with members of the household. Our working database contained 12,036 L1-Spanish users and 4,025 L1-Basque users in Grade 4 (n=16,061), and 11,469 L1-Spanish users and 3,231 L1-Basque users in Grade 10 (n=14,700). Language of Assessment Language of assessment is the language in which students took the mathematics test. The language of assessment was determined by test center administrators, but typically corresponded to the linguistic model of students’ schools (Instituto Vasco de Evaluación e Investigación Educativa, 2012a, b). Indeed, test center administrators elected to test most students—93.52% of Grade 4 testing groups and 94.35% of Grade 10 testing groups—in mathematics in the language of instruction (Instituto Vasco de Evaluación e Investigación Educativa, 2012a, b). (Note 3) MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 16 composite SES level among all school types in the BAC. Most non-immigrant students at Model A schools are L1 Spanish users (Elosua & Egaña, 2017; Lasagabaster, 2001). Typically, as in our study, students at Spanish-medium schools are tested in Spanish (Instituto Vasco de Evaluación e Investigación Educativa, 2012a,b). What this suggests is that the statistically significant score differences and relatively large effect sizes that we noted for the main effect of SES on cumulative mathematics scores may be indicative of the unique challenges faced by low-SES students at Spanish-medium schools. Our analyses of the score distributions of L1 Spanish students provide further evidence that low-SES students who are taught and assessed in Spanish may be at a disadvantage. We found that low-SES L1 Spanish students who took the test in Spanish performed worse than those who took the test in Basque. This is not the pattern one would expect, especially for Grade 4 students, given the challenges inherent to learning mathematics while developing the language in which it is taught and assessed (Abedi, 2009; Hakuta, Buter & Witt, 2000; Murray, 2010; Ní Ríordáin & O’Donoghue, 2008; Schleppegrell, 2007). One plausible explanation is that, because low-SES L1 Spanish students tested in Basque were taught mathematics in Basque, they performed well when tested in that language. That hypothesis is supported by research in other contexts of revitalization, which shows that that L1 English students can perform well in Irish (Ní Ríordáin & O’Donoghue, 2011) and Welsh (Jones, 1998) if they are taught in those languages. Alternatively, school-level SES may contribute to differences in achievement among lowSES students tested in Basque and Spanish. Santiago et al. (2008) found that school-level SES had a larger impact on academic performance for students in the BAC than family-level SES. It could be that the Spanish-medium schools attended by low-SES L1 Spanish students may be under-resourced, as compared Basque-medium schools. Indeed, as was previously noted, Model MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 17 A public schools—populated primarily by L1 Spanish users—have, on average, higher percentages of low-SES students than other educational institutions in the BAC (Flores, 2019). What this means is that although students are tested in their home language, they may lack highquality instructional materials or teachers. Thus, despite the alignment between home language and language of assessment, which has been found to positively contribute to academic achievement in the BAC (Elosua, 2016), low-SES L1 Spanish students may not perform well on mathematics tests, if they attend low-SES schools. A third possibility is that low-SES L1 Spanish students are not being adequately supported, as compared to high-SES peers. Studies have found that, among L1 Spanish students, those who attend Spanish-medium schools exhibit higher levels of SES than those who attend bilingual or Basque-medium schools (Elosua & Egaña, 2017). If a Spanish-medium school is primarily populated by high-SES, L1 Spanish students, the needs of low-SES students may be overlooked. The latter group may not be provided with extra language supports in class or on tests, given that they are L1 users of the language of instruction and assessment. However, low-SES students may not be familiar with the language practices at school, which often reflect middle or upper-class norms (Silverstein, 2003). Thus, they may still require supports. An important limitation of the study is that, due to characteristics of the data available, we were unable to analyze the relationship between the students’ scores on the mathematics test and their scores on the Basque and Spanish language proficiency tests. Prior research has found an important relationship between performance on tests and literacy skills in the language of assessment (e.g., Kurdek & Sinclair, 2001; Ní Ríordáin, & O’Donoghue, 2011; Walker, Zhang, & Surber, 2008). While those findings were obtained with other populations, it is reasonable to assume that some of the score variation that we attribute to language or SES may be related to MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 18 students’ literacy skills in the language of assessment. Future research about achievement in the BAC should address the relationship between mathematics and language proficiency scores. For now, our results show that important differences in academic achievement still exist in the BAC, despite the successful revitalization of Basque. Together, our ANOVA analyses and our analyses of score distributions suggest that low-SES, L1 Spanish students may be at an especial disadvantage. It is critical, therefore, to consider how minority revitalization efforts may impact language groups differently, depending on SES. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 19 Acknowledgements This research was partially supported by XXXXXXXXXXXXXX, under grant XXXXXXXXXXX. The authors wish to thank XXXXXXXXXXXXXXXXXXXXXXXX for their comments on earlier versions of this paper. Declaration of Interest The opinions expressed in this paper do not necessarily reflect the opinions of our colleagues or those of the funding agency. We have no potentially relevant interests to declare. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 20 Endnotes Note 1 We also opted to exclude immigrant students because they comprised a relatively small proportion of our sample. Immigrant students represented 7 percent (n=1,131) of students in Grade 4, and 9 percent (n=1,369) of students in Grade 10. Note 2 SES was measured using the Index of Socioeconomic and Cultural Status (ISEC.) Each student was assigned a socioeconomic rank, ranging from 1 (the lowest level) to 5 (the highest level.) Those rankings were calculated by averaging the values associated with factors such as household income, parental occupation, level of parental education, number of books in the house, and access to the Internet (Flores, 2019). That average was converted into a descriptor of overall SES: Lower class, lower-middle class, middle class, upper-middle class, and upper class. Note 3 Students were grouped according to the linguistic model of their school. If test center administrators opted to test students in the language of instruction, all students took the mathematics test in the same language, which corresponded to linguistic model. If administrators elected to test students in their home language, students within the same test group could take the test in different languages. In those instances, the language of assessment was based on selfreported information about home language use. For example, within a single testing group from a Basque-medium school, L1 Spanish students could take the mathematics test in Spanish and L1Basque users could take it in Basque. In Grade 10, nearly 99% of students in Basque-medium schools took the test in Basque, 90.83% of students in Spanish-medium schools took it in Spanish, and 88.78% of students in MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 21 bilingual schools took it in the whatever language they received mathematics instruction (Instituto Vasco de Evaluación e Investigación Educativa, 2012b). Percentages for students tested in the language of instruction were similar for Grade 4 students: 97.01% for Basquemedium, 93.15% for Spanish-medium and 85.24% for bilingual schools (Instituto Vasco de Evaluación e Investigación Educativa, 2012a). Note 4 Information available to the researchers did not include the process of development and translation of the test, or whether Basque or Spanish was the language in which the test was originally developed. Also, the data available did not contain information on the center where each student took the assessment or information about students’ performance on the Basque, Spanish, and English components of the assessment. Hence, we could not identify the language used to test any individual student, nor could we relate measures of language arts proficiency with the mathematics scores. Note 5 As shown in Table 2, in Grade 4, 4% of L1 Basque students across the three SES levels (n=154 students) were tested in Spanish. Similarly, in Grade 10, only 3% of L1 Basque students (n=99 students) were tested in Spanish. These percentages reflect the fact that most L1Basque attend schools that teach and conduct assessments in Basque, or schools that utilize a bilingual model (Cenoz & Gorter, 2019). Note 6 We based the relative magnitude of the effect size associated with SES on Sirin’s (2005) meta-analysis of the literature about academic achievement and SES. Sirin concluded that, for fixed-effects models with student-level data, the average effect size for SES was .28. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 22 Note 7 There are two types of schools in the BAC: public and semi-public. Public schools are statefunded, while semi-public or “concertado” schools are privately-run, but publicly-funded (Elosua & Egaña, 2017; Lasagabaster, 2001). There are public and private varieties of Model A (e.g. Spanish-medium), Model D (e.g. Basque-medium), and Model B (e.g. bilingual) schools. Students at semi-public schools tend to be of higher-SES backgrounds than students at public schools, across linguistic models (Flores, 2019; Gortazar, 2018). MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 23 References Abedi, J. (2002). Standardized Achievement Tests and English Language Learners: Psychometric Issues. Educational Assessment 8(3), 231–257. Abedi, J. (2009). Validity of Assessments for English Language Learning Students in a National/International Context. Estudios Sobre Educacion, (16). Abedi, J., & Lord, C. (2001). The language factor in mathematics tests. Applied Measurement in Education, 14(3), 219-234. Akhtar, Z., & Niazi, H. K. (2011). The relationship between socio-economic status and learning achievement of students at secondary level. International Journal of academic research, 3(2). Aldekoa, J. & Gardner, N. (2002) Turning knowledge of Basque into use: Normalisation plans for schools. International Journal of Bilingual Education and Bilingualism, 5:6, 339-354. Altuna, O. & Urla, J. (2013). The Basque street survey: Two decades of assessing language use in public spaces. International Journal of the Sociology of Language, 2013(224), 209-227. Ceallaigh, T. Ó., & Dhonnabhain, Á. N. (2017). Reawakening the Irish language through the irish education system: Challenges and priorities. International Electronic Journal of Elementary Education, 8(2), 179-198. Cenoz, J. (2008). Learning through the minority: An introduction to the use of Basque in education in the Basque Country. Language, culture and curriculum, 21(1), 1-4. Cenoz, J., & Gorter, D. (2019). Minority Languages, National State Languages, and English in Europe: Multilingual Education in the Basque Country and Friesland. Journal of Multilingual Education Research, 9(1), 9. De Luna, I. M., Zalbide, M., Darquennes, J., & Suberbiola, P. (2014). The use of Basque in MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 24 model D schools in the Basque Autonomous Community. European Journal of Applied Linguistics, 2(2), 254-286. Elosua, P. (2016). Minority language revitalization and educational assessment: Do languagerelated factors impact performance? Journal of Sociolinguistics, 20(2), 212-228. Elosua, P., & De Boeck, P. (2019). Educational assessment issues in linguistically diverse contexts: a case study using a generalised linear mixed model. Language, Culture and Curriculum, 1-14. Elosua, P., & Egaña, M. (2017). Socioeconomic patterns and educational performance based on language-of-instruction models in the Basque Autonomous Community. Journal of Multilingual and Multicultural Development, 38(8), 686-698. Elosua, P., & Mujika, J. (2014). Impact of family language and testing language on reading performance in a bilingual educational context. Psicothema, 26, 328–335. Flores, A. (2019). Basque Nationalism and Education: The influence of Euskara on student achievement in the Basque Country. Independent Study Project (ISP) Collection. 3070. Retrieved from https://digitalcollections.sit.edu/isp_collection/3070 Gortazar, L. (2018). The PISA" shock" in the Basque Country: Contingent factors or structural change?. Studies on the Spanish Economy (FEDEA). Gorter, D. & Cenoz, J. (2011). Multilingual education for European minority languages: The Basque Country and Friesland. International Review of Education, 57, 651-666. Hakuta, K., Butler, Y. G., & Witt, D. (2000). How long does it take English learners to attain proficiency? Santa Barbara: University of California Linguistic Minority Research Institute. Halliday, M. A. K. (1978). Language as social semiotic. London: Edward Arnold. MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 25 Hualde, J. I., & Zuazo, K. (2007). The standardization of the Basque language. Language Problems and Language Planning, 31(2), 143-168. Instituto Vasco de Evaluación e Investigación Educativa (2012a). Informe de resultados y análisis de variables: Educación primaria. Bilbao, Euskadi: ISEI.IVEI. Instituto Vasco de Evaluación e Investigación Educativa (2012b). Informe de resultados y análisis de variables: Educación secundaria obligatoria. Bilbao, Euskadi: ISEI.IVEI. Jones, D. V. (1998). National curriculum tests for mathematics in English and Welsh: Creating matched assessments. Assessment in Education: Principles, Policy & Practice, 5(2), 193 211. Jones, M. (2016). Research briefing: Welsh-medium education and Welsh as a subject. National Assembly for Wales. Retrieved from https://senedd.wales/research%20documents/rs16048/16-048-english-web.pdf. Kurdek, L. A., & Sinclair, R. J. (2001). Predicting reading and mathematics achievement in fourth-grade children from kindergarten readiness scores. Journal of Educational Psychology, 93(3), 451. Lasagabaster, D. (2001). Bilingualism, immersion programmes and language learning in the Basque Country. Journal of Multilingual and Multicultural Development, 22(5), 401-425. Lasagabaster, D. (2017). Language learning motivation and language attitudes in multilingual Spain from an international perspective. The Modern Language Journal, 101(3), 583-596. Moseley, C. (Ed.). (2010). Atlas of the World's Languages in Danger. UNESCO. Murray, D. R. (2010). Irish-Medium Language Immersion Programs’ Effects on Mathematics Education. Journal of Mathematics Education at Teachers College, 1(2). Nelde, Peters, Miquel Strubell and Glyn Williams. 1996. Euromosaic. The Production and RUNNING HEAD: MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION Table 3. Fixed-design SES*Language of Assessment Between-Subjects ANOVAs Grade 4 Grade 10 Students with Home Language of Basque (n=4,025) Students with Home Language of Basque (n=3,231) Predictor SS df MS F p η² Predictor SS df MS F p η² SES 59722.625 2 29861.313 15.092 .000 .007 SES 73851.180 2 36925.590 16.081 .000 .010 Language of Assessment 5037.980 1 5037.980 2.546 .011 .001 Language of Assessment 15112.261 1 15112.261 6.582 .010 .002 SES x Language of Assessment 1325.525 2 662.763 .335 .715 .000 SES x Language of Assessment 3546.093 2 1773.046 .772 .462 .000 Students with Home Language of Spanish (n=12,036) Students with Home Language of Spanish (n=11,469) Predictor SS df MS F p η² Predictor SS df MS F p η² SES 2266738.573 2 1233369.286 566.498 .000 .086 SES 2578964.255 2 1289482.128 561.675 .000 .089 Language of Assessment 50580.011 1 50580.011 25.282 .000 .002 Language of Assessment 11930.156 1 11930.156 5.197 .023 .000 SES x Language of Assessment 11785.898 2 5892.949 2.946 .053 .000 SES x Language of Assessment 11301.045 2 5650.523 2.461 .085 .000 RUNNING HEAD: MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION Grade 4 Grade 10 Figure 1. Grade 4 and Grade 10 cumulative score quintiles: Percentage by SES 0% 5% 10% 15% 20% 25% 30% 35% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Percentage of Students 0% 5% 10% 15% 20% 25% 30% 35% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Percentage of Students MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 34 Grade 4 Grade 10 Figure 2. Grade 4 and Grade 10 cumulative scores by SES: Percentage by quintile 0% 10% 20% 30% 40% 50% 60% 70% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Percentage of Students 0% 10% 20% 30% 40% 50% 60% 70% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Percentage of Students MATHEMATICS ACHIEVEMENT AND LANGUAGE REVITALIZATION 35 Figure 3. Grade 4 and Grade 10 cumulative score distributions of Spanish users, by SES and language of assessment