Jump to ContentJump to Main Navigation
Growing up in Diverse SocietiesThe Integration of the Children of Immigrants in England, Germany, the Netherlands, and Sweden$

Frank Kalter, Jan O. Jonsson, Frank van Tubergen, and Anthony Heath

Print publication date: 2018

Print ISBN-13: 9780197266373

Published to University Press Scholarship Online: May 2019

DOI: 10.5871/bacad/9780197266373.001.0001

Show Summary Details
Page of
date: 26 May 2020

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Chapter:
(p.219) 9 Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants
Source:
Growing up in Diverse Societies
Author(s):

Jörg Dollmann

Frida Rudolphi

Meenakshi Parameshwaran

Publisher:
British Academy
DOI:10.5871/bacad/9780197266373.003.0009

Abstract and Keywords

Proficiency in the language of a new country is perhaps the most important precondition for the successful integration of immigrants in various other integration aspects, like educational and vocational success, interethnic relations and ethnic identify formation. Explaining ethnic disparities in linguistic integration therefore has the potential to aid our understanding of ethnic differences along various other integration dimensions. In the present contribution, we first demonstrate substantial heterogeneity of adolescents’ language proficiency in four European countries depending on their ethnic origin and their migration history. In order to further understanding these differences we examine very different individual and family factors that can be hypothesised to influence language learning processes. Besides an influence of social background on language learning, we show that ethnic specific factors such as language use in the family are at least partly relevant for the language acquisition process.

Keywords:   language, language proficiency, language use, immigrants, ethnic minorities, exposure, incentives, efficiency, L1, L2

9.1 Introduction

PROFICIENCY IN THE LANGUAGE of the receiving country is one of the most important preconditions for the successful integration of immigrants in various aspects of life (Esser 2006) and is therefore of key importance for young persons with an immigrant background. Having a good command of the language of the receiving country substantially enhances educational success (see Campbell et al. 2001 and Durham et al. 2007 for the importance of early language skills). Furthermore, language proficiency is important for positioning in the labour market and has a positive influence on earnings (Dustmann 1994; Chiswick & Miller 1999; Bleakley & Chin 2004). It also affects social integration, providing channels through which immigrants can make contact with friends and spouses (Esser 2006; Martinovic et al. 2009; Schlueter 2012), and is important for identity formation. A good command of the language of the sending country suggests a stronger sense of belonging to one’s country of origin, and a weaker sense of belonging to the receiving country (Phinney et al. 2001). Explaining ethnic disparities in linguistic integration is therefore crucial for our understanding of ethnic differences in other integration dimensions.

Foreign-born (‘first-generation’) youth undoubtedly face linguistic challenges when they move to a new country with a main language other than their mother tongue, especially at the beginning of their stay (Heath & Kilpi-Jakonen 2012; see also OECD 2010). However, it is much less obvious whether—and why— native-born youth with foreign-born parents (‘second generation’) should have worse language skills than majority youth. Nevertheless, there are results indicating that they are not fully on a par with their peers whose parents are native (p.220) born (Heath & Kilpi-Jakonen 2012; see also OECD 2010). In this chapter, we contribute to previous research on language acquisition among immigrants that predominantly focused on adults, especially on first-generation adult immigrants (e.g. Chiswick & Miller 1995; van Tubergen & Kalmijn 2005; 2009), broadening the scope to include second- and even third-generation adolescents. We ask whether the explanations suggested in previous research account for differences in language skills among adolescents, including youths who themselves often have no direct migration experience. As a starting point, we use a theoretical framework of the process of language acquisition among foreign-born adults (e.g. Chiswick & Miller 1995; van Tubergen & Kalmijn 2005; 2009) that emphasises three important aspects of learning a language: exposure to the language, motivation and efficiency.

Using this framework as a point of departure, we address three basic research questions. First, do we find language differences between adolescent immigrants of different generational status? A key issue of linguistic integration is whether second- and third-generation minority youth acquire language skills on a par with those of majority youth. Still rarely studied is the extent to which gradual integration of language skills occurs. Secondly, does language acquisition vary between different ethnic groups, as indicated by country of origin, and between different countries of destination? For example, we will compare the language proficiency of children of immigrants in Sweden, whose contact with the language of the receiving country before migration was most probably limited or non-existent, to that of children of immigrants in England, who are more likely to have had contact with and incentives to already learn the world language in the sending country. Thirdly, to what extent do factors commonly proposed to be relevant to explaining language differences among first-generation adult immigrants also contribute to the understanding of language differences among younger, second- and third-generation immigrants? Do we find an impact of exposure to the language spoken in the receiving country, as indicated by language usage within the families? Do motives of remigration or emigration affect language acquisition among children of immigrants? Does efficiency in learning a new language, attributable, for example, to differences in the linguistic distance between the languages of the sending and the receiving countries, affect the language-learning process? If so, do differences in these resources account for varying language skills between origin groups?

The chapter will proceed as follows. We will first present some theoretical considerations about preconditions for language acquisition that may account for ethnic group differences in destination-language proficiency, and results from previous research. We will then empirically test the outlined hypotheses, ending with a conclusion.

9.2 Theoretical Considerations and Previous Research

The acquisition of language skills is a specific form of learning and an investment in human capital, particularly with regard to a new host-country language (p.221) (Chiswick & Miller 1995; 2001; van Tubergen & Kalmijn 2005; Esser 2006; van Tubergen 2010). There are numerous theories of language acquisition within the fields of psychology, linguistics, economics and sociology. Despite the heterogeneous theoretical traditions, each of these disciplines emphasises three key concepts as crucial to the process of learning a language: exposure, efficiency and incentives (Chiswick & Miller 1995; 2001; Esser 2006; van Tubergen 2010). Systematic differences in these three conditions between different immigrant groups and receiving contexts may affect willingness to engage in learning the language of the receiving country and the level of difficulty involved, which may contribute to ethnic group differences in language skills.

9.2.1 Exposure

One simply needs to be around a language to be able to learn it. Exposure to the language of the receiving country (L2) refers both to the sending and to the receiving context (Chiswick & Miller 1995; 2001). Some immigrants had already been exposed to the host-country language in their country of origin, prior to migration. This holds true especially if the primary language is the same in the country of origin as in the country of destination, like English in Jamaica or German in Austria, but also if the language of the receiving country is an official language in the sending country, as English is in India. It is also relevant if the host-country language is de facto the official (if not primary) language of the sending country, such as English in Bangladesh, or if there is a connection between the (language) history of the receiving and that of the sending country via the colonial past (Chiswick & Miller 2001). Ethnic groups with origins in countries with overall higher exposure to L2 in the sending country (and as an extreme: where L2 is the same as the language of the sending country, L1) are therefore expected to have better language skills than learners with a heritage in countries where exposure to L2 is usually less prevalent.

Exposure to L2 in the receiving country occurs, for example, through everyday contact with users of the language, through reading, watching TV, listening to the radio or through education or language courses. One important factor determining the extent of exposure to the language spoken in the receiving country is length of stay. Results consistently support the finding that first-generation immigrants tend to have higher language skills the longer they have stayed in the receiving country (e.g. Dustmann & Fabbri 2003; Jirjahn & Tsertsvadze 2004; van Tubergen & Kalmijn 2009). However, there are also reasons to expect the exposure argument to be relevant for native-born children with parents or grandparents who migrated to the receiving country, that is, second- or third-generation immigrants (Becker 2011). For example, differences between native-born children in exposure to the language of the receiving country may be due to different language patterns within the household. It is common for native-born students with foreign-born parents to speak a different language in the home. Among pupils with strong immigrant background (first or second (p.222) generation), a considerable share in the four European countries that we study speak a language other than the Programme for International Student Assessment (PISA)-assessment language at home (OECD 2010: figure 11.4.9 and table 11.4.4). Different language environments within families or in neighbourhood communities (with varying levels of ethnic segregation) imply differences in the language usually spoken within these contexts (e.g. Chiswick & Miller 1996; van Tubergen & Kalmijn 2009), which may contribute to differences in exposure to and opportunities to learn L2. Furthermore, other activities, such as different kinds of media consumption, can involve more or less usage of L2. As Chiswick & Miller (1996) show, the availability and usage of foreign-language newspapers seems to correlate negatively with proficiency in L2. Summing up the exposure arguments, one can assume that a more frequent use of the main language spoken in the country of origin, and therefore a less frequent use of the main language spoken in the country of destination in non-school situations, may be associated with lower language skills in L2.

9.2.2 Efficiency

Language learners need efficiency in their development of language skills. Efficiency in learning a new language strongly depends on one’s age upon coming in contact with it (Long 1990; Newport 1990; 2002). For immigrant children, one can therefore assume that a young age on arrival will have a positive effect on their language acquisition. Since we are studying a school cohort, and thus have very little variation in age in the data, this corresponds with our expectation for exposure regarding length of stay: the earlier pupils immigrate, the higher their expected proficiency in L2.

Furthermore, general cognitive skills improve the ability and efficiency to acquire new language skills (Esser 2006). Assuming an equal distribution of cognitive skills across different ethnic groups, one cannot expect them to be an important factor in explaining ethnic differences in language skills. However, differences in the selectivity of immigrants into receiving societies exist (Ichou 2014; Engzell 2015). The selectivity of immigrants may depend on the motive to migrate: labour market immigrants are commonly assumed to be more favourably selected than refugees, while family-tie immigrants lie between the two (e.g. Chiswick & Miller 2015). Among labour market immigrants, host countries may attract highly or poorly skilled workers from sending countries, which is another issue of selectivity that could be important for language-learning processes. Importantly, selectivity likely occurs not only with regard to observed characteristics, but also to unobserved ones affecting children’s language skills (e.g. parents’ cognitive skills or their reading habits).

Recent research on language acquisition among adult immigrants emphasises the importance of their educational background prior to migration as one major factor determining their efficiency in learning a new language (Chiswick & Miller 1995; Esser 2006). This rationale is, however, more problematic for (p.223) students with direct or indirect migration biographies. For youth, educational success (in terms of track placement or grades) may not only be a resource for the language-learning process but also a consequence of their language proficiency. Nevertheless, educational and socioeconomic resources in the family context are crucial for young students. Resources within the family provide individuals with knowledge and practical skills that are not only essential for educational achievement but, more generally, also for language-learning processes (Esser 2006). For example, the daily availability of reading material or routine and competent linguistic behaviour within the family increases not only exposure to L2, but also provides a stimulating environment that increases efficiency in learning new concepts and new terminologies (Chiswick & Miller 1995). Consistently demonstrated is the relationship between parental education and children’s linguistic development (e.g. Dustmann 1997; Calvo & Bialystok 2013; Hoff 2013). Differences in students’ educational and socioeconomic backgrounds may therefore contribute to differences in their efficiency in learning a new language.

Finally, the efficiency in learning a second language may depend on linguistic distance—the extent to which languages differ from each other (Chiswick & Miller 2004). For example, English, like many Western European languages, belongs to the Germanic branch within the Indo-European language family. This means that immigrants from Western European countries may find it easier to learn English because it is similar to the language(s) in which they are already proficient. Immigrants from countries using languages from other branches, or even from other language families, may find it comparably more difficult though, because English is not similar to their first language. Results by Heath & Kilpi-Jakonen (2012) indicate that the reading literacy of 15-year-old students tends to be lower among immigrant children originating from linguistically dissimilar countries, and more so for those from non-Western than for those from Western countries. Consequently, we expect that children with a heritage from countries in which the language branch and language family are similar to those of the destination country’s language tend to have better language skills than children from countries in which the language difference is greater.

9.2.3 Incentives

Language learners need some kind of incentives or motivation, such as a wish to cope well at school, make friends, enjoy hobbies/sports or (at least later, as adults) get established in the labour market, in order to put in effort. One argument maintains that the motivation to learn a second language depends on the anticipated length of stay in the receiving society (Chiswick & Miller 2015). Investments may be more likely among those who intend to stay longer in the receiving country than among those with strong return orientations (Dustmann 1999; Isphording & Otten 2014; Chiswick & Miller 2015). The motivation to learn a new language is probably lower among those who do not intend to stay (p.224) long or who expect to join an ethnic-group-specific labour market rather than the general labour market of the receiving country (Chiswick & Miller 2001). However, this argument is mainly proposed and empirically studied among first-generation adult immigrants who intentionally migrated to a certain receiving country. Whether differences in migration motives of parents affect the language learning of their children, who did or did not migrate themselves, is less studied. The same caveat holds true for the role of remigration motives: Can we expect the (often probably vague) motive to leave the receiving society in a few years’ time to influence the language-learning process among adolescents?

The motivation to invest in a language may vary depending on its general utility and productivity (Esser 2006). One way to assess the general value of a language is by means of its communicative value, or Q-value, mainly representing the extent to which others in the world speak this language (de Swaan 1999; 2001). For example, English has the highest Q-value of all languages, most people speaking it as either their first or second language. The motivation to learn the language of the receiving country therefore may depend on its prevalence and utility beyond the receiving country. Immigrants migrating to England may be highly motivated to acquire English-language skills, because the language is spoken not only in the receiving country but also in many other parts of the world. Swedish, in contrast, does not have a high prevalence outside of Scandinavia. Its lower language utility may lower the motivation of immigrants to Sweden to learn Swedish. Again, this is an argument initially proposed for adult immigrants, and it is an empirical question whether we will find this relationship also among adolescents, especially those of the second and higher generations. Even if we do not expect children to be well aware of these aspects of languages, their parents may be. Thus, we expect the possible effect to run mainly through parents’ considerations. It is likely that the language used by the family in the home will have an influence on the language preferences and proficiencies of members of that family related to the power of its communicative value.

Financial costs can also be a factor in the language-learning process and may reduce the motivation to invest in learning a new language. The most likely direct economic cost factors are tuition fees or expenditures on course materials for language-learning programmes (Esser 2006; van Tubergen 2010). Since the costs are more easily borne by well-off families, immigrant groups that are financially better off are more likely than poorer groups to invest in such learning processes. We can expect this to vary in importance across survey countries, because the occurrence and levels of tuition fees differ.

From such considerations, among all different kinds of conditions connected to the concepts of exposure, efficiency and incentives, the general resources related to families’ socioeconomic background seem to play a major role in their language-learning process. These resources are of general importance as they are not only suited to examining possible differences between immigrant groups in the acquisition of a new language, but can also be fruitfully applied to explain language-learning processes among the majority population. Such (p.225) general resources matter for any kind of language-learning process, regardless whether it be learning one’s mother tongue or any additional second or third language (Esser 2006). For example, the positive role of an advantageous family background, with daily availability of a stimulating language environment, has consistently been shown to be important for the language-learning process among children and young adults without an immigrant background (Payne et al. 1994; Rodriguez et al. 2009). To summarise, systematic differences in the discussed parameters—exposure, efficiency and incentives—between natives and immigrants may likewise help to explain ethnic differences in language skills in a receiving country.

9.3 Empirical Findings

How do language skills vary between immigrant groups and youth of different generational status across the four European countries studied? The analyses in this chapter rely on results from language tests carried out during wave 1 of the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU), which measured children’s lexicon (Kalter et al. 2016). During this test, respondents had to choose correct synonyms or antonyms for a specific number of words out of a list of possible alternatives in a particular space of time. In the following analyses, language test scores are rescaled to have a mean of 0, and a standard deviation of 1, for each country separately.

In Figure 9.1, we show mean group differences in standardised language test scores for some of the most important immigrant origin groups in each of the four survey countries. Test scores of majority pupils are set to zero and serve as a reference point in our comparisons (reference category: ‘REF’).1 The results indicate a large heterogeneity in language skills across countries of origin. Although most—but not all—minority groups are disadvantaged in relation to majority youth, the magnitude differs remarkably. The origin country groups standing out as particularly disadvantaged are Eastern Europe in England, Morocco in the Netherlands, and Somalia and Iraq in Sweden. Youth of Turkish heritage have substantially lower language test scores than the majority in all three countries with a large Turkish group. For the most disadvantaged groups, the differences amount to around 0.8 to 1.3 of a standard deviation, which are substantial disadvantages.

Minority disadvantage in language skills clearly differs between host countries too. The overall smallest ethnic differences are observable in England (with the exception of the East European group), while the largest, most consistent minority disadvantages are observable in Sweden. These cross-country differences may be (p.226)

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Figure 9.1 Language test scores (standardised) shown by survey country and origin group.

Note: Reference group (REF): Students without a significant migration background. Weighted means. N=4,155 (EN); 5,004 (GE); 4,207 (NL); 4,801 (SW). Abbreviations: ASIA: Asia; B&H: Bosnia and Herzegovina; CAR: Caribbean; E EUR: Eastern Europe; FIN: Finland; IND: India; ITA: Italy; IRA: Iraq; KOS: Kosovo; MENA+: Middle East and North Africa; MOR: Morocco; NWS EUR: Northern, Western and Southern Europe; OTH: Other; PAK: Pakistan; POL: Poland; RUS: Russia; S-AFR: Sub-Saharan Africa; SER: Serbia; SOM: Somalia; SUR: Suriname; TUR: Turkey.

due to differences in the Q-value of the different languages spoken in the survey countries, although we cannot test this assumption more directly. We also observe interesting differences for specific ethnic groups across the CILS4EU countries. While Eastern Europeans in Germany do comparatively well, they are—as already mentioned—the group with the largest disadvantages in England.

Group differences may be due to the variety of the composition of minority groups, for example, in their length of stay within and across host countries. Figure 9.2 focuses on differences in exposure to the host-country language, showing language test scores by immigrant generational status. The results reveal two important things. First, they indicate that the language deficit among adolescents of immigrant origin declines over time, in line with results of previous research, visible in the second, third and fourth uppermost bars of the figure. Germany is an exception with its similar disadvantage level for the youngest age group on arrival and youth who arrived at age 6 to 10. These differences in language tests scores may result in part from differences in the ethnic composition of adolescents and their families, that is, different ages at arrival (we are analysing cross-sectional data and do not follow individuals over time). Nonetheless, the descriptive pattern is a rough indicator of increasing language skills over a longer stay in the receiving society. (p.227)

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Figure 9.2 Language test scores (standardised) shown by survey country and generational status.

Note: Reference group (REF): Students without a significant migration background. Weighted means. N=4,155 (EN); 5,004 (GE); 4,207 (NL); 4,801 (SW).

Secondly, it reveals that native-born children of foreign-born parents, those of ‘second-generation’ minority origin, score lower on the language test than the majority population without any significant migration background does, with the exception of England. This important descriptive finding suggests that the lower educational performance shown in previous research for subsets of these ethnic minority groups (Kristen & Granato 2007; Jonsson & Rudolphi 2011; Jackson et al. 2012) could be due in part to their shortage of language skills. Furthermore, those of the second generation are disadvantaged to a similar or greater extent than immigrant youth who arrived before they had reached school age, which could again be the result of differences in the composition of different immigrant cohorts.

When focusing on adolescents with even more distant migration histories, that is, third-generation youth, subsumed under the bar for ‘some migration background’, disadvantages disappear in the Netherlands or turn into slight advantages, as in the case of England. However, in Sweden and Germany we still find ethnic gaps of up to almost a third of a standard deviation for this group. Pupils with one foreign-born parent and one native-born parent with an immigrant background (children of a ‘transnational marriage’) also have a language disadvantage in Germany, while this is substantially smaller in the Netherlands and Sweden and non-existent in England. Finally, and as expected, the penalty for late arrival is substantial across all four countries, but appears to be less so in England. The disadvantage in language skills amounts to as much as around one standard deviation (England), or substantially more, among the latest arrivals.

As already discussed, differences between origin groups may be partly an effect of different migration histories, and differences between immigrant cohorts may be due to the fact that ethnic groups simply represent different generational (p.228) groups (see Chapter 3). Therefore, we consider both factors jointly in multivariate analyses. However, the variables ‘ethnic group’ and ‘generational background’ have—by definition—the same reference category, that is, respondents without a strong migration background. Therefore, estimating coefficients for the variables simultaneously requires some adjustments to the set-up: extension of the reference category for one of these variables in order to avoid a perfect overlap (see Chapter 3). In the present chapter, we add immigrants from Northern, Western and Southern Europe (NWS EUR, shown in Figure 9.1) to the reference category of respondents without a significant migration background (shown in Figure 9.2 as ‘REF’). Coefficients for the ethnic groups are now interpreted as effects relative to Northern, Western and Southern Europeans of the same generational status, while the variable ‘generational status’ refers to members of Northern, Western and Southern European countries relative to the population without a strong migration background. This approach seems reasonable, as across all countries the Northern, Western and Southern Europeans are the group whose performance almost equals that of the majority without a significant migration background.

In Figure 9.3 results are overviewed for country of origin (net of generational status) (for the complete model see Model 1, Tables A9.1–A9.4 in the Appendix). Predicted values from the first model (‘gross difference’) reveal that in Sweden pupils with an immigrant background have an overall strong minority penalty in

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Figure 9.3 Predicted language test scores (standardised) for origin groups, shown by survey country.

Note: Reference group (REF): Students with Northern, Western and Southern European background. Estimations are based on linear regressions, Models 1 and 2 in Tables A9.1–A9.4 in Appendix. N=3,736 (EN); 4,722 (GE); 4,012 (NL); 4,532 (SW). SES = socioeconomic status.

(p.229) host-country language skills, while England is at the other end of the scale. Within countries, the same overall pattern as in Figure 9.1 is observed. In England, the most disadvantaged ethnic minority groups (from Eastern Europe and Pakistan) have language test scores that are around one half of a standard deviation lower than those of Northern, Western and Southern Europeans, whereas larger disadvantages are observable in the other survey countries. In Germany, minorities from Turkey and Serbia are most disadvantaged (approximately 0.6 to 0.7 of a standard deviation), while in the Netherlands immigrants from Turkey and Morocco achieve substantially lower (around 0.7 of a standard deviation) language test scores. In Sweden, minority youth from Somalia, Turkey and Iraq, in particular, perform significantly worse on the language test than the majority population does (all exceeding 0.8 of a standard deviation in their disadvantage).

To what extent are the observed minority disadvantages attributable to unequal distributions of resources and to other factors that we can expect to be important in learning a language? To answer this question, we add control variables commonly assumed to affect language skills and to have an unequal distribution between minority groups, as well as between minorities and the native majority population. These include cognitive test scores (indicating academic skills), measured by means of a standard, language-free cognitive ability test (Weiß 2006) during wave 1 of CILS4EU; parental education, categorised as below (upper) secondary education, (upper) secondary education or university education; parents’ occupational status, based on the highest ISEI score (International Socio-Economic Index of Occupation Status) in the family; and economic hardship, measured by means of a question in the parental questionnaire on the respondent’s ability to raise a specific amount of money by tomorrow. For both latter variables, a missing category is used in the analysis to reduce the number of missing values.

These factors consistently correlate with language proficiency in all four countries (see Model 2 in Tables A9.1–A9.4 in the Appendix). With better-educated parents and a higher occupational status, children’s L2 skills rise significantly, in line with the assumption that a more favourable parental socioeconomic situation provides a more learning-friendly environment as well as the efficiency to learn L2. The students’ cognitive skills, another important aspect of efficiency, also correlate with language test scores in the hypothesised way. Furthermore, economic resources correlate positively with destination-language skills, which may partly reflect increasing opportunities to capture the manifold economic costs in the language-learning process, thereby contributing to higher levels of L2.

Once we take differences in these factors into account, the language test scores of many minority groups are no longer significantly lower than those of Northern, Western and Southern European students (see estimates ‘controlling for SES [socioeconomic status] + cognitive skills’ in Figure 9.3). For example, when social background and cognitive skills are considered, only students with a Turkish heritage do worse in Germany. Note that some groups do worse only after consideration of individual and family resources important for the language-learning process, such as students with Middle East and Northern Africa (MENA+) (p.230) heritage or from Asia in England. Their opportunities to learn English therefore seem to be better than the test scores they actually achieve, at least in comparison to Northern, Western and Southern Europeans.

To summarise, and disregarding the mentioned deviations from the main pattern, a substantial part of gross minority disadvantage is attributable to differences in cognitive skills and in parental resources (education, occupational status, economic hardship), measures indicating different opportunities for and restrictions to learning a new language, in terms of exposure, efficiency and economic prospects.

Figure 9.4 shows the generational status effects (net of country of origin) with and without controlling for SES and cognitive skills. In line with Figure 9.2, the gross differences show a clear late-arrival penalty in all four destination countries, with a disadvantage among students arriving at age 11 or older that varies from around 0.7 (England) to as much as almost 1.5 (the Netherlands) of a standard deviation. The disadvantage of immigrant youth who arrived at age 6 to 10 is much less, and in the Netherlands even non-significant. Furthermore, the results do not indicate significant disadvantages for Northern, Western and Southern European students who immigrated at an early age (5 years of age or younger) or second-generation youth (aside from the German case). A striking finding is the comparably good achievement of students whose parents are inter-ethnic couples (‘Child of intermarriage’).

We do not expect individual skills and resources to account as strongly for generational-status differences as for country-of-origin differences, because different origin groups are more likely to vary in these characteristics compared to different generational groups within the same origin group. A comparison of the

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Figure 9.4 Predicted language test scores (standardised) for generational status groups, shown by survey country.

Note: Reference group (REF): Students without a significant migration background. Estimations are based on linear regressions, Models 1 and 2 in Tables A9.1–A9.4 in Appendix. N=3,736 (EN); 4,722 (GE); 4,012 (NL); 4,532 (SW).

(p.231) gross results and the model controlling for cognitive skills and parental resources in Figure 9.4 does not contradict the expectation. The main pattern is small differences, although in three of the survey countries (all but England) some of the penalty for first-generation immigrants is attributable to the measured resources.

As can be seen from Model 2, controlling for SES and cognitive skills, the general resources already explain many of the ethnic disadvantages between different origin groups. However, for some groups disadvantages still exist, which shifts our focus to differences in ‘specific ethnic’ factors that may be responsible for these remaining ethnic gaps in language proficiency. Thus, we restrict the sample to students with an immigrant background and exclude pupils of majority origin, that is, students without a significant migration background.2

The immigrant-specific Model 3 includes the same variables as does Model 2, but estimated only for children of immigrants. Added in Model 4 are other factors that we can expect to be important for language learning among children of immigrants: stay tendency, often described as an important factor indicating adult immigrants’ motivation to learn a language, and use of the second language in the household, indicating exposure to the destination language in activities outside the school. Usage of another language than the language of the receiving country is measured as a total score combining four variables based on survey questions asking about the frequency of use in different contexts: in the family, at the computer, when listening to music and when watching TV. Furthermore, we examined the potential role of the linguistic distance between the first and the second language, expecting that a shorter distance will increase the efficiency in learning a new language. Finally, we considered the geographic distance between the country of origin and the country of destination, assuming that immigrants from more distant countries show lower return tendencies and therefore a higher motivation to learn L2. However, the results of these two latter variables are not included in the models, as these factors did not significantly correlate with the outcome.

Before turning to the results of these additional, ethnic-specific variables, it has to be noted that the results from the general resources model (Model 2) estimated only for children with at least some migration background (Model 3, cf. Tables A9.1–A9.4 in the Appendix) basically confirm the role of these resources in explaining differences in language proficiency also among minority youth. Interestingly, the explanatory power of these variables seems to be higher in the minority-specific model, especially in England and the Netherlands.

Regarding the ethnic-specific factors included in the multivariate analyses (Model 4, cf. Tables A9.1–A9.4 in the Appendix), less frequent use of another language than the language of the receiving country correlates positively with language test scores in England, Germany and Sweden, while in the Netherlands (p.232) we find no such pattern. In Germany, the results for stay tendency point in the opposite direction from the one hypothesised. Language skills of students who are undecided about their future and indicate that they don’t know whether or not to remain in the receiving society at age 30 tend to be higher than the language skills of students who are sure they will stay in Germany. In none of the four countries do we find positive effects on the language skills for those who are sure of their future stay in the respective society.

Figures 9.5 and 9.6 demonstrate how ethnic and generational gaps change upon consideration of these additional variables (cf. Tables A9.1–A9.4, Models 3–4). As seen in the figures, inclusion of ethnic-specific factors further decreases ethnic gaps and differences between generational groups, but to a much lesser degree than the general resources considered in Model 2 do. Additionally, disadvantages for some ethnic and generational groups persist. For example, in Sweden almost all immigrant groups except Somalis, Kosovars and Southern Africans still do worse than Northern, Western and Southern Europeans.

The same holds true for the different generational groups, now contrasted with third-generation children of immigrants (‘some migration background’) as the reference category. Here, the consideration of ethnic-specific factors seems to matter most for first-generation immigrants (with the exception of the Netherlands), but only to a small degree. However, even when considering the

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

Figure 9.5 Predicted language test scores (standardised) for origin groups, shown by survey country: immigrant-only models.

Note: Reference group (REF): Students with Northern, Western and Southern European background. Estimations are based on linear regressions, Models 3 and 4 in Table A9.1–A9.4 in Appendix. N=1,694 (EN); 2,320 (GE); 1,254 (NL); 2,039 (SW).

Ethnic Differences in Language Skills: How Individual and Family Characteristics Aid and Prohibit the Linguistic Integration of Children of Immigrants

(p.233) Figure 9.6 Predicted language test scores (standardised) for generational status groups, shown by survey country: immigrant-only models.

Note: Reference group (REF): Students with some migration background. Estimations are based on linear regressions, Models 3 and 4 in Table A1–A4 in Appendix. N=1,694 (EN); 2,320 (GE); 1,254 (NL); 2,039 (SW).

language environment in the families and the stay tendency, most recently arrived first-generation immigrants still achieve comparably poor language test scores.

9.4 Conclusion

Language skills of young persons of immigrant origin are very consequential for different aspects of their everyday life and their integration success in other dimensions, such as successful navigation through the educational system, the labour market, inter-ethnic friendships, romantic relations and marriages and development of host-country ethnic identity. Explaining differences in language skills is thus key to our understanding of intergenerational integration in a wide sense.

This contribution shows that language skills among 14-year-old children with an immigrant background are heterogeneous in England, Germany, the Netherlands and Sweden. Children who migrated themselves face particular difficulties in learning the main language spoken in their destination country (L2). As expected, the greatest disadvantages are observable for those most recently arrived, at age 11 or older. With the exception of England, this also holds true for students of foreign-born parents who themselves were born in Germany, the Netherlands or Sweden. Therefore, language difficulties are relevant not only for foreign-born students, but also for second-generation immigrants, while in all of the four countries disadvantages for third-generation children of immigrants either have disappeared (England, the Netherlands) or have almost disappeared (Germany, Sweden). In addition to disparities in language skills due to different (p.234) lengths of stay, we also show that language proficiency clearly differs between ethnic origin groups, both within and between the receiving societies. Some groups fare comparably well, like students with a Northern, Western or Southern European heritage, while others face more difficulties, such as those of Eastern European origin in England, of Turkish origin in Germany and the Netherlands, and students originating from Iraq or Somalia in Sweden. Although results indicate gradual integration of language skills among foreign-born youth, linguistic integration does not seem to be an ‘automatic’ process completed equally by all groups at the same pace, but rather one with differences between groups due to the group-specific availability of the preconditions necessary for language learning.

To understand language differences better, we focused on differences in the exposure to L2, the efficiency in learning L2 and differences in the incentives to invest and costs of investing in learning L2 as potential influence factors. We apply a theoretical framework usually considered important in explaining language skills among—predominantly adult—first-generation immigrants and ask whether these preconditions also are relevant for language learning among younger immigrant cohorts with different migration histories. Concerning factors that are related to the efficiency in learning a language, we find a strong positive relationship between individuals’ cognitive skills and their language skills, but also between language skills and parental educational background, which may serve as an indicator of efficiency and of exposure to a language-learning-friendly environment. Finally, economic preconditions seem to matter, which again may be due partly to more general socioeconomic effects, but in turn also could relate to the opportunity to capture any costs that may occur in the language-learning process, such as those for course materials or extra lessons. However, these measures did not account for all of the ethnic disadvantages in L2 proficiency. The initially observed differences—although significantly reduced—persisted for some ethnic groups in our multivariate analyses, while the initial disadvantages for others were entirely due to differences in the factors of exposure, efficiency and incentives.

In analyses focusing on the subpopulation with an immigrant background, we concentrated more directly on arguments proposedly important for exposure, and asked whether the frequency of usage of L1 may hinder proficiency in L2, a relationship confirmed in three of the four survey countries. Furthermore, a shadow of the future in terms of a pronounced stay tendency does not seem to play a role in the motivation to learn the language of the receiving country—at least not among the observed adolescent population. This also holds true for another factor that could affect return migration motives: the geographic distance between the sending and the receiving country. Furthermore, for our adolescent population, the argument about linguistic distances does not seem to apply. Though again, differences in the proposed mechanisms capturing exposure, efficiency and incentives could not explain all language disparities observed among different ethnic minority groups. Consequently, factors other than the ones examined here also seem to matter for differences between different ethnic minority groups and between the majority population and some minority groups.

(p.235) To sum up, the theoretical framework to understand differences in language acquisition among first-generation adult immigrants also can be applied fruitfully—at least partially—to explain differences in language proficiency among second- and third-generation adolescents. Some of the factors usually capturing the general concepts of efficiency, exposure and incentives seem to be consequential for young immigrants and children of immigrants as well. Here parental education and social background or adolescents’ cognitive skills seem to matter most, emphasising the importance of a positive language-learning environment, especially in the early years in the life course. Other factors, like the stay tendency in the receiving society—among adult immigrants usually connected to a higher motivation to learn L2—seem to be less important for children of immigrants. This may be because the language-learning processes in both groups (children of immigrants versus adult immigrants) differ in their timing. Adult immigrants actually decide to learn a language immediately before or after migration, and also consider such factors as return tendencies, while among immigrant children the language-learning process may be less a rational decision than a ‘natural’ process, given the parental and sociocultural background into which they were born. Therefore, focusing more explicitly on these early conditions of language development within and outside the family may increase the general framework’s ability to explain differences in language proficiency among second-or third-generation immigrants in more detail.

References

Bibliography references:

Becker, B. (2011), ‘Cognitive and Language Skills of Turkish Children in Germany: A Comparison of the Second and Third Generation and Mixed Generational Groups’, International Migration Review, 45: 426–59.

Bleakley, H. & Chin, A. (2004), ‘Language Skills and Earnings: Evidence from Childhood Immigrants’, Review of Economics and Statistics, 86: 481–96.

Calvo, A. & Bialystok, E. (2013), ‘Independent Effects of Bilingualism and Socioeconomic Status on Language Ability and Executive Functioning’, Cognition, 130: 278–88.

Campbell, J.R., Kelly, D.L., Mullis, I.V.S., Martin, M.O. & Sainsbury, M. (2001), Framework and Specifications for PIRLS Assessment 2001 (Chestnut Hill MA, Boston College).

Chiswick, B.R. & Miller, P.W. (1995), ‘The Endogeneity between Language and Earnings: International Analyses’, Journal of Labor Economics, 13: 246–88.

Chiswick, B.R. & Miller, P.W. (1996), ‘Ethnic Networks and Language Proficiency among Immigrants’, Journal of Population Economics, 9: 19–35.

Chiswick, B.R. & Miller, P.W. (1999), ‘Language Skills and Earnings among Legalized Aliens’, Journal of Population Economics, 12: 63–89.

Chiswick, B.R. & Miller, P.W. (2001), ‘A Model of Destination Language Acquisition: Application to Male Immigrants in Canada’, Demography, 38: 391–409.

Chiswick, B.R. & Miller, P.W. (2004), ‘Linguistic Distance: A Quantitative Measure of the Distance between English and Other Languages’, IZA Discussion Paper No. 1246.

(p.236) Chiswick, B.R. & Miller, P.W. (2015), ‘International Migration and the Economics of Language’, in B.R. Chiswick & P.W. Miller (eds), Handbook on the Economics of International Migration (Amsterdam, Elsevier), 211–69.

De Swaan, A. (1999), ‘Language and Culture in Transnational Society’, European Review, 7: 507–18.

De Swaan, A. (2001), Words of the World: The Global Language System (Cambridge, Polity Press).

Durham, R.E., Farkas, G., Scheffner Hammer, C., Tomblin, B. & Catts, H.W. (2007), ‘Kindergarten Oral Language Skill: A Key Variable in the Intergenerational Transmission of Socioeconomic Status’, Research in Social Stratification and Mobility, 25: 294–305.

Dustmann, C. (1994), ‘Speaking Fluency, Writing Fluency and Earnings of Migrants’, Journal of Population Economics, 7: 133–56.

Dustmann, C. (1997), ‘The Effects of Education, Parental Background and Ethnic Concentration on Language’, Quarterly Review of Economics and Finance, 37: 245–62.

Dustmann, C. (1999), ‘Temporary Migration, Human Capital, and Language Fluency of Migrants’, Scandinavian Journal of Economics, 101: 297–314.

Dustmann, C. & Fabbri, F. (2003), ‘Language Proficiency and Labour Market Performance of Immigrants in the UK’, The Economic Journal, 113: 695–717.

Engzell, P. (2015), ‘How Migrant Selectivity Shapes Second-Generation Educational Careers in Sweden’, paper delivered at Popular Association of America conference.

Esser, H. (2006), Sprache und Integration: die sozialen Bedingungen und Folgen des Spracherwerbs von Migranten (Frankfurt am Main, Campus).

Heath, A.F. & Kilpi-Jakonen, E. (2012), ‘Immigrant Children’s Age at Arrival and Assessment Results’, OECD Education Working Paper No. 75.

Hoff, E. (2013), ‘Interpreting the Early Language Trajectories of Children from Low SES and Language Minority Homes: Implications for Closing Achievement Gaps’, Developmental Psychology, 49: 4–14.

Ichou, M. (2014), ‘Who They Were There: Immigrants’ Educational Selectivity and their Children’s Educational Attainment’, European Sociological Review, 30: 750–65.

Isphording, I. E. & Otten, S. (2014), ‘Linguistic Barriers in the Destination Language Acquisition of Immigrants’, Journal of Economic Behavior & Organization, 105: 30–50.

Jackson, M., Jonsson, J.O. & Rudolphi, F. (2012), ‘Ethnic Inequality and Choice-Driven Educational Systems: A Longitudinal Study of Performance and Choice in England and Sweden’, Sociology of Education, 85: 158–78.

Jirjahn, U. & Tsertsvadze, G. (2004), ‘Bevölkerungsanteil und Sprachkenntnisse von Migranten’, Review of Economics, 55: 142–62.

Jonsson, J.O. & Rudolphi, F. (2011), ‘Weak Performance—Strong Determination: School Achievement and Educational Choice among Ethnic Minority Students in Sweden’, European Sociological Review, 27: 487–508.

Kalter, F., Heath, A.F., Hewstone, M., Jonsson, J.O., Kalmijn, M., Kogan, I. & van Tubergen, F. (2016), ‘Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU)’. ZA5353 Data File Version 1.2.0, Cologne, GESIS Data Archive. DOI: 10.4232/cils4eu.5353.1.2.0.

Kristen, C. & Granato, N. (2007), ‘The Educational Attainment of the Second Generation in Germany: Social Origins and Ethnic Inequality’, Ethnicities, 7: 343–66.

Long, M.H. (1990), ‘Maturational Constraints on Language Development’, Studies in Second Language Acquisition, 12: 251–85.

(p.237) Martinovic, B., van Tubergen, F. & Maas, I. (2009), ‘Dynamics of Interethnic Contact: A Panel Study of Immigrants in the Netherlands’, European Sociological Review, 25: 303–18.

Newport, E.L. (1990), ‘Maturational Constraints on Language Learning’, Cognitive Science, 14: 11–28.

Newport, E.L. (2002), ‘Critical Periods in Language Development’, in L. Nadel (ed.), The Encyclopedia of Cognitive Science (London, Macmillan), 737–40.

OECD (Organisation for Economic Co-operation and Development) (2010), PISA 2009 Results: Overcoming Social Background—Equity in Learning Opportunities and Outcomes, vol. 2 (Paris, OECD Publishing).

Payne, A.C., Whitehurst, G.J & Angell, A.L (1994), ‘The Role of Home Literacy Environment in the Development of Language Ability in Preschool Children from Low-Income Families’, Early Childhood Research Quarterly, 9: 427–40.

Phinney, J.S., Romero, I., Nava, M. & Huang, D. (2001), ‘The Role of Language, Parents, and Peers in Ethnic Identity among Adolescents in Immigrant Families’, Journal of Youth and Adolescence, 30: 135–53.

Rodriguez, E.T., Tamis-LeMonda, C.S., Spellmann, M.E., Pan, B.A., Raikes, H., Lugo-Gil, J. & Luze, G. (2009), ‘The Formative Role of Home Literacy Experiences across the First Three Years of Life in Children from Low-Income Families’, Journal of Applied Developmental Psychology, 30: 677–94.

Schlueter, E. (2012), ‘The Inter-Ethnic Friendships of Immigrants with Host-Society Members: Revisiting the Role of Ethnic Residential Segregation’, Journal of Ethnic and Migration Studies, 38: 77–91.

Van Tubergen, F. (2010), ‘Determinants of Second Language Proficiency among Refugees in the Netherlands’, Social Forces, 89: 515–34.

Van Tubergen, F. & Kalmijn, M. (2005), ‘Destination-Language Proficiency in Cross-National Perspective: A Study of Immigrant Groups in Nine Western Countries’, American Journal of Sociology, 110: 1412–57.

Van Tubergen, F. & Kalmijn, M. (2009), ‘Language Proficiency and Usage among Immigrants in the Netherlands: Incentives or Opportunities?’, European Sociological Review, 25: 169–82.

Weiß, R.H. (2006), CFT-20R. Grundintelligenzskala 2 (Revision) (Göttingen, Hogrefe).

(p.238)

Table A9.1. Multivariate analysis (OLS regression) of language test scores: England

Model 1

Model 2

Model 3

Model 4

b

SE

b

SE

b

SE

b

SE

Gender (ref: male):

   Female

0.015

(0.050)

0.015

(0.041)

−0.072

(0.071)

−0.094

(0.072)

Age in years

6.768***

(2.066)

5.012**

(1.906)

6.273**

(2.225)

5.089*

(2.015)

Age in years (squared)

−0.226***

(0.069)

−0.168**

(0.064)

−0.207**

(0.074)

−0.167*

(0.067)

Origin group (ref: maj./NWS EUR):

   EEUR

−0.517*

(0.226)

−0.372*

(0.147)

−0.333*

(0.157)

−0.218

(0.170)

   CAR

−0.175

(0.115)

−0.090

(0.086)

−0.105

(0.086)

−0.163

(0.088)

   MENA+

−0.267

(0.166)

−0.395**

(0.128)

−0.394**

(0.138)

−0.376**

(0.129)

   PAK

−0.502***

(0.117)

−0.260**

(0.092)

−0.176

(0.090)

−0.101

(0.091)

   S-AFR

−0.170

(0.121)

−0.100

(0.116)

−0.117

(0.113)

−0.139

(0.113)

   ASIA

−0.212

(0.127)

−0.342***

(0.088)

−0.350***

(0.095)

−0.284**

(0.087)

   INDIA

−0.015

(0.129)

−0.099

(0.112)

−0.046

(0.080)

0.012

(0.082)

   OTH

0.013

(0.099)

0.009

(0.093)

−0.011

(0.095)

−0.057

(0.090)

Generational status (ref: majority—Model 1/2; ref: some migration background—Model 3/4)

   Arrived age 11 or older

−0.683***

(0.165)

−0.678***

(0.127)

−0.864***

(0.131)

−0.660***

(0.142)

   Arrived age 6-10

−0.532*

(0.219)

−0.519*

(0.209)

−0.737***

(0.209)

−0.561**

(0.171)

   Arrived age 5 or younger

0.058

(0.148)

−0.101

(0.137)

−0.298*

(0.145)

−0.225

(0.157)

   Age at arrival not known

−0.401

(0.339)

−0.424

(0.321)

−0.613

(0.346)

−0.500

(0.382)

   2nd generation

0.204*

(0.093)

0.158*

(0.067)

−0.070

(0.064)

0.013

(0.073)

   Child of transnat. marriage

0.175

(0.092)

0.119

(0.075)

−0.126

(0.078)

−0.088

(0.083)

   Child of intermarriage

0.355***

(0.105)

0.259'

(0.085)

0.086

(0.084)

0.095

(0.091)

   Some migr. background

0.270***

(0.084)

0.211***

(0.075)

Parental education (ref: lower sec):

   Upper secondary

0.113***

(0.038)

0.043

(0.087)

0.055

(0.083)

   University

0.193'

(0.047)

0.254

(0.075)

0.258***

(0.072)

Highest parental occ. status (ISEI)

0.004'

(0.001)

0.004

(0.002)

0.004***

(0.002)

ISEI missing (ref: no):

   Yes

−0.203'

(0.078)

−0.378

(0.094)

−0.361***

(0.097)

Economic solvency (ref: no): Yes

0.255'

(0.094)

0.328

(0.133)

0.286***

(0.142)

Economic solvency missing (ref: no): Yes

0.104

(0.091)

0.262

(0.130)

0.272***

(0.132)

Cognitive skills (range 0-27)

0.082'

(0.006)

0.080

(0.010)

0.081***

(0.010)

Stay tendency (ref: yes):

   No

−0.060

(0.120)

   Don't know

−0.052

(0.076)

LI language use

(score, range 1 'always'-4 'never')

0.203***

(0.060)

Intercept

−50.525***

(15.449)

−39.286

(14.305)

^9.295

(16.735)

^1.284**

(15.044)

No. of obs.

3,736

3,736

1,694

1,694

R2

0.082

0.288

0.373

0.387

Note: Design weighted, accounting for clustering; standard errors (SE) in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

ASIA: Asia (except India and Pakistan); CAR: Caribbean; E EUR: Eastern Europe; INDIA: India; MENA+: Middle East and North Africa plus Afghanistan; NWS EUR: Northern, Western and Southern Europe; OTH: Other; PAK: Pakistan; S-AFR: Sub-Saharan Africa.

(p.240)

Table A9.2. Multivariate analysis (OLS regression) of language test scores: Germany

Model 1

Model 2

Model 3

Model 4

b

SE

b

SE

b

SE

b

SE

Gender (ref: male):

Female

0.165***

(0.041)

0.194***

(0.032)

0.151***

(0.056)

0.146*

(0.057)

Age in years

−0.976

(0.943)

−0.005

(0.650)

1.624

(1.174)

1.265

(1.136)

Age in years (squared)

0.024

(0.030)

−0.003

(0.021)

−0.058

(0.037)

−0.046

(0.036)

   Origin group (ref: maj./NWS EUR):

   ITA

−0.436***

(0.151)

−0.191

(0.135)

−0.198

(0.137)

−0.192

(0.145)

   EEUR

0.046

(0.151)

0.073

(0.156)

0.072

(0.156)

0.062

(0.155)

   POL

−0.041

(0.142)

0.009

(0.137)

0.022

(0.138)

−0.008

(0.139)

   RUS

−0.106

(0.128)

−0.092

(0.120)

−0.104

(0.125)

−0.105

(0.129)

   SER

−0.617***

(0.160)

−0.271

(0.151)

−0.288

(0.153)

−0.219

(0.159)

   MENA+

−0.391***

(0.134)

−0.178

(0.126)

−0.192

(0.128)

−0.173

(0.132)

   TUR

−0.729***

(0.131)

−0.473***

(0.121)

−0.484***

(0.127)

−0.426***

(0.138)

   S-AFR

−0.346*

(0.167)

−0.092

(0.148)

−0.136

(0.154)

−0.165

(0.152)

   ASIA

−0.031

(0.214)

−0.052

(0.188)

−0.039

(0.195)

−0.032

(0.190)

   OTH

−0.128

(0.146)

−0.046

(0.165)

−0.051

(0.162)

−0.017

(0.173)

Generational status (ref: majority—Model 1/2; ref: some migration background—Model 3/4)

   Arrived age 11 or older

−0.956***

(0.179)

−0.852***

(0.162)

−0.505***

(0.165)

−0.316

(0.166)

   Arrived age 6-10

−0.409***

(0.143)

−0.467***

(0.153)

−0.198

(0.151)

−0.004

(0.157)

   Arrived age 5 or younger

−0.272*

(0.137)

−0.295*

(0.120)

−0.075

(0.128)

0.076

(0.142)

   Age at arrival not known

−0.767***

(0.238)

−0.562***

(0.191)

−0.316

(0.221)

−0.131

(0.235)

   2nd generation

−0.322*

(0.135)

−0.255*

(0.126)

−0.053

(0.116)

0.125

(0.123)

   Child of transnat. marriage

−0.109

(0.177)

−0.087

(0.143)

0.114

(0.126)

0.257

(0.135)

(p.239)    Child of intermarriage

0.244*

(0.120)

0.150

(0.116)

0.334***

(0.109)

0.379***

(0.110)

   Some migr. background

−0.165

(0.148)

−0.198

(0.154)

(p.241) Parental education (ref: lower sec):

   Upper secondary

0.061

(0.042)

0.068

(0.071)

0.074

(0.070)

   University

0.274

(0.055)

0.346***

(0.097)

0.351***

(0.086)

Highest parental occ. status (ISEI)

0.005

(0.001)

0.005***

(0.002)

0.004*

(0.002)

ISEI missing (ref: no): Yes

−0.031

(0.102)

−0.061

(0.109)

−0.057

(0.100)

Economic solvency (ref: no): Yes

0.173

(0.043)

0.150*

(0.061)

0.130*

(0.059)

Economic solvency missing (ref: no): Yes

0.081

(0.059)

0.052

(0.087)

0.029

(0.086)

Cognitive skills (range 0-27)

0.090

(0.006)

0.077***

(0.008)

0.075***

(0.007)

Stay tendency (ref: yes):

   No

0.136

(0.094)

   Don't know

0.142*

(0.068)

LI language use (score, range 1 'always'-4 'never')

0.159***

(0.037)

Intercept

9.757

(7.285)

−1.146

(5.066)

−13.016

(9.223)

−10.870

(8.903)

No. of obs.

4,722

4,722

2,320

2,320

R2

0.165

0.361

0.392

0.407

Note: Design weighted, accounting for clustering; standard errors (SE) in parentheses;

(*) p < 0.05,

** p < 0.01,

(***) p < 0.001.

ASIA: Asia; E EUR: Eastern Europe (except Poland, Russia and Serbia); ITA: Italy; MENA+: Middle East (except Turkey) and North Africa plus Afghanistan and Pakistan; NWS E: Northern, Western and Southern Europe (except Italy); OTH: Other; POL: Poland; RUS: Russia; S-AFR: Sub-Saharan Africa; SER: Serbia; TUR: Turkey.

(p.242)

Table A9.3 Multivariate analysis (OLS regression) of language test scores: Netherlands

Model

1

Model

2

Model

3

Model

4

b

SE

b

SE

b

SE

b

SE

Gender (ref: male):

Female

0.131

(0.109)

0.137

(0.100)

0.266***

(0.075)

0.267***

(0.074)

Age in years

−5.419*

(2.287)

^.331*

(2.169)

−5.367**

(2.053)

−5.625**

(2.051)

Age in years (squared)

0.170*

(0.076)

0.137

(0.071)

0.169*

(0.068)

0.177**

(0.068)

Origin group (ref: maj./NWS EUR):

   EEUR

−0.142

(0.193)

−0.043

(0.164)

−0.051

(0.175)

−0.083

(0.171)

   CAR

−0.228

(0.151)

−0.276

(0.154)

−0.324*

(0.162)

−0.336*

(0.163)

   SUR

−0.323

(0.177)

−0.452**

(0.157)

−0.444*

(0.196)

−0.454*

(0.203)

   MENA+

−0.261

(0.192)

−0.370

(0.192)

−0.305

(0.193)

−0.287

(0.194)

   MOR

−0.676**

(0.226)

−0.580**

(0.200)

−0.468*

(0.207)

−0.469*

(0.205)

   TUR

−0.691***

(0.165)

−0.644***

(0.163)

−0.606***

(0.158)

−0.611***

(0.165)

   S-AFR

−0.409*

(0.202)

−0.320

(0.196)

−0.269

(0.212)

−0.276

(0.219)

   ASIA

0.020

(0.189)

0.008

(0.205)

−0.002

(0.214)

−0.023

(0.223)

   OTH

−0.071

(0.252)

0.008

(0.225)

−0.028

(0.226)

−0.044

(0.233)

Generational status (ref: majority—Model 1/2; ref: some migration background—Model 3/4)

   Arrived age 11 or older

−1.465***

(0.154)

−1.290***

(0.162)

−1.104***

(0.200)

−1.070***

(0.201)

   Arrived age 6-10

−0.119

(0.205)

−0.131

(0.191)

0.013

(0.243)

0.019

(0.242)

   Arrived age 5 or younger

0.212

(0.153)

0.206

(0.148)

0.373

(0.202)

0.383

(0.196)

   Age at arrival not known

−0.667*

(0.273)

−0.581*

(0.276)

−0.481

(0.305)

−0.476

(0.312)

   2nd generation

−0.057

(0.187)

0.016

(0.177)

0.114

(0.181)

0.122

(0.192)

   Child of transnat. marriage

0.157

(0.250)

0.127

(0.209)

0.198

(0.239)

0.208

(0.234)

   Child of intermarriage

0.314**

(0.120)

0.235

(0.131)

0.336

(0.185)

0.338

(0.183)

   Some migr. background

−0.012

(0.206)

−0.099

(0.207)

(p.243) Parental education (ref: lower sec):

   Upper secondary

−0.132

(0.128)

−0.023

(0.124)

−0.031

(0.122)

   University

−0.048

(0.135)

−0.106

(0.162)

−0.132

(0.161)

Highest parental occ. status (ISEI)

0.007***

(0.001)

0.008**

(0.003)

0.008*

(0.003)

ISEI missing (ref: no): Yes

−0.274*

(0.123)

−0.397***

(0.112)

−0.394*

(0.112)

Economic solvency (ref: no): Yes

−0.234*

(0.097)

−0.100

(0.121)

−0.092

(0.124)

Economic solvency missing (ref: no): Yes

−0.156*

(0.079)

−0.091

(0.131)

−0.085

(0.134)

Cognitive skills (range 0-27)

0.061***

(0.008)

0.053***

(0.009)

0.053*

(0.010)

Stay tendency (ref: yes):

   No

−0.025

(0.093)

   Don't know

0.100

(0.083)

LI language use (score, range 1 'always'-4 'never')

0.029

(0.074)

Intercept

42.928*

(17.274)

32.897*

(16.514)

41.071**

(15.601)

42.926*

(15.584)

No. of obs.

4,012

4,012

1, 254

1,254

R2

0.078

0.167

0. 358

0.360

Note: Design weighted, accounting for clustering; standard errors (SE) in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

ASIA: Asia; CAR: Caribbean; MENA+: Middle East and North Africa (except Morocco and Turkey) plus Afghanistan and Pakistan; MOR: Morocco; NWS EUR: Northern, Western and Southern Europe; OTH: Other; S-AFR: Sub-Saharan Africa; SUR: Suriname; TUR: Turkey.

(p.244)

Table A9.4. Multivariate analysis (OLS regression) of language test scores: Sweden

Model 1

Model 2

Model 3

Model 4

b

SE

b

SE

b

SE

b

SE

Gender (ref: male):

Female

−0.114***

(0.035)

−0.096**

(0.029)

−0.072

(0.045)

−0.063

(0.045)

Age in years

5.771*

(2.318)

3.529

(2.080)

0.114

(2.809)

0.128

(2.767)

Age in years (squared)

−0.200*

(0.078)

−0.120

(0.071)

−0.006

(0.095)

−0.006

(0.093)

   Origin group (ref: maj./NWS EUR):

   FIN

−0.223

(0.121)

−0.229*

(0.096)

−0.202

(0.099)

−0.228*

(0.098)

   E-EUR

−0.459***

(0.111)

−0.339***

(0.096)

−0.332

(0.098)

−0.333***

(0.096)

   B&H

−0.557***

(0.133)

−0.322**

(0.117)

−0.370

(0.119)

−0.333**

(0.120)

   KOS

−0.660***

(0.124)

−0.248*

(0.110)

−0.260

(0.116)

−0.170

(0.122)

   MENA+

−0.613***

(0.109)

−0.340***

(0.090)

−0.361

(0.094)

−0.346***

(0.093)

   IRA

−0.804***

(0.129)

−0.466***

(0.110)

−0.482

(0.117)

−0.450***

(0.113)

   TUR

−0.821***

(0.142)

−0.465***

(0.124)

−0.512

(0.130)

−0.482***

(0.129)

   S-AFR

−0.357*

(0.150)

−0.087

(0.130)

−0.079

(0.137)

−0.102

(0.136)

   SOM

−0.869***

(0.133)

−0.156

(0.123)

−0.169

(0.134)

−0.175

(0.134)

   ASIA

−0.303**

(0.112)

−0.205*

(0.099)

−0.204

(0.100)

−0.220*

(0.097)

   OTH

−0.398**

(0.124)

−0.268**

(0.095)

−0.265

(0.102)

−0.249*

(0.103)

Generational status (ref: majority—Model 1/2; ref: some migration background—Model 3/4)

   Arrived age 11 or older

−0.972***

(0.147)

−0.780***

(0.125)

−0.886

(0.142)

−0.728***

(0.166)

   Arrived age 6-10

−0.272*

(0.123)

−0.221*

(0.096)

−0.329

(0.105)

−0.191

(0.109)

   Arrived age 5 or younger

0.085

(0.117)

0.092

(0.100)

0.006

(0.107)

0.080

(0.106)

   Age at arrival not known

−0.337

(0.202)

−0.337

(0.203)

−0.297

(0.194)

−0.242

(0.195)

   2nd generation

−0.132

(0.097)

−0.110

(0.077)

−0.188

(0.079)

−0.092

(0.082)

(p.245)    Child of transnat. marriage

0.057

(0.110)

0.216*

(0.094)

0.127

(0.102)

0.185

(0.105)

   Child of intermarriage

0.243***

(0.093)

0.215***

(0.081)

0.116

(0.087)

0.149

(0.089)

   Some migr. background

0.044

(0.119)

0.094

(0.088)

Parental education (ref: lower sec):

   Upper secondary

0.053

(0.051)

0.046

(0.072)

0.036

(0.070)

   University

0.103

(0.053)

0.146*

(0.073)

0.138

(0.073)

Highest parental occ. status (ISEI)

0.004***

(0.001)

0.003*

(0.001)

0.003*

(0.001)

ISEI missing (ref: no):

   Yes

−0.355***

(0.079)

−0.351*

(0.088)

−0.359*

(0.087)

Economic solvency (ref: no): Yes

0.155***

(0.048)

0.234*

(0.075)

0.240*

(0.077)

Economic solvency missing (ref: no): Yes

−0.035

(0.045)

0.066

(0.072)

0.077

(0.075)

Cognitive skills (range 0-27)

0.078***

(0.003)

0.074*

(0.005)

0.071*

(0.005)

Stay tendency (ref: yes):

   No

0.057

(0.063)

   Don't know

0.084

(0.055)

LI language use

(score, range 1 'always'-4 'never')

0.111*

(0.032)

Intercept

−41.108*

(17.104)

−27.284

(15.328)

−1.788

(20.842)

−2.306

(20.551)

No. of obs.

4,532

A,,532

2,039

2,039

R2

0.150

0., 358

0.384

0.391

Note: Design weighted, accounting for clustering; standard errors (SE) in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

ASIA: Asia; B&H: Bosnia and Herzegovina; E EUR: Eastern Europe (except B&H and Kosovo); FIN: Finland; IRAQ: Iraq; KOS: Kosovo; MENA+: Middle East (except Iraq and Turkey) and North Africa plus Afghanistan and Pakistan; NWS EUR: Northern, Western and Southern Europe (except Finland); OTH: Other; S-AFR: Sub-Saharan Africa; SOM: Somalia; TUR: Turkey.

Notes:

(1) Note that in this chapter, the third generation is, unlike in most other chapters in this book, not part of the ‘majority’ reference group but included in the ethnic minority groups, i.e. those students with at least ‘some’ migration background.

(2) The reference category for ethnic groups is again NWS EURs, while the reference category for generational status is third-generation immigrants (immigrants with ‘some migration background’).