Rewarding integration? Citizenship regulations and socio-cultural integration of immigrants in the Netherlands, France and Germany

E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Rewarding Integration? Citizenship Regulations and the Socio-Cultural Integration of Immigrants in the 1 Netherlands, France and Germany Evelyn Ersanilli and Ruud Koopmans This paper compares the levels of socio-cultural integration of naturalised and non-naturalised immigrants in the Netherlands, France and Germany. Sociocultural integration is measured by host-country identification, proficiency and use of the host-country language, and interethnic social contacts. To increase crossnational comparability, we focus on immigrants from two rural regions in Turkey who migrated before 1975. Based on the assumption that easily accessible citizenship promotes socio-cultural integration, we test two hypotheses. First, whether naturalised immigrants display higher levels of socio-cultural integration than non-naturalised immigrants. Second, whether immigrants in countries with few preconditions for naturalisation show higher levels of socio-cultural integration. We find that naturalisation is positively associated with socio-cultural integration only in those countries—France and Germany—that have traditionally required a certain degree of cultural assimilation from their new citizens. Regarding country differences, we find that Turkish immigrants in France show higher levels of socio-cultural integration on all four indicators. For host-country identification, they share this position with Dutch Turks. The results show that limited cultural assimilation conditions tied to citizenship may be helpful in promoting socio-cultural integration, but also that the allowance of dual nationality does not have the negative effects that are sometimes ascribed to it. Keywords: Citizenship; Naturalisation Policies; Dual Nationality; Socio-Cultural Integration Introduction Since the turn of the century, several European countries—for example Denmark, the Netherlands and Germany—have introduced stricter socio-cultural integration requirements for naturalisation. Existing language requirements have been extended, with formalised tests that sometimes also include a cultural section. In debates on changing the requirements for naturalisation, two lines of argument are commonly used. Left-wing parties—such as the Social-Democrats and the Greens in Germany and the Netherlands—argue that the acquisition of citizenship stimulates integration and therefore access to citizenship should be easy. Granting immigrants citizenship means giving them a vested interest in society and signals acceptation. Conservative parties—such as the Christian Democrats in Germany and the Netherlands—tend to argue that citizenship 1 This paper has appeared as Ersanilli, Evelyn & Ruud Koopmans (2010) ‘Rewarding integration? Citizenship regulations and socio-cultural integration of immigrants in the Netherlands, France and Germany’ Journal of Ethnic and Migration Studies, 36(5), pp 773—791 special issue on ‘Migration and Citizenship Attribution: Politics and Policies in Western Europe’ edited by Maarten Vink . Copyright Taylor & Francis The article is available online at http://www.informaworld.com/smpp/content~db=all?content=10.1080/13691831003764318 E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 should be the end-point of integration and only awarded to those who have made a conscious choice for their new country and can fulfil high integration requirements. In this view, citizenship should be a reward for successful integration and also an incentive to integrate (see e.g. de Hart 2005; de Hart and van Oers 2006; Hailbronner 2006). Despite cross-national convergence in citizenship laws, several significant differences remain (de Hart and van Oers 2006; Howard 2005). These differences allow an examination of the relation between citizenship regulations and the integration of immigrants. The European Civic Citizenship and Inclusion Index considers regulations that allow naturalisation after three years of residence, without any supplementary requirements such as language tests, and allow dual nationality to be the most favourable for integration (Geddes and Niessen 2005), but is this justified? This paper addresses the question by investigating whether easily accessible citizenship indeed promotes higher levels of the socio-cultural integration of immigrants. We focus on socio-cultural integration because many countries have been implementing stricter sociocultural requirements for citizenship access and debates about dual citizenship also focus on socio-cultural aspects of integration such as identification and language proficiency (see e.g. Staton et al. 2007). The countries studied in this paper—Germany, the Netherlands and France— were selected because they have clearly different conceptions of citizenship and attendant nationality and naturalisation policies. Until the 1990s, Germany had citizenship based on ius sanguinis, with high barriers to naturalisation. Subsequently, the barriers were lowered somewhat and, in 2000, a limited degree of ius soli was introduced. Dual nationality is accepted only in a minority of naturalisations. The Netherlands has a stronger ius soli component and low requirements for naturalisation until 2003; dual nationality is mostly condoned. France has a strong ius soli component in its citizenship law and unconditionally allows dual nationality. However, it has traditionally imposed cultural requirements for naturalisation. International comparisons always pose methodological problems. The main problem is the lack of adequate comparative data (Favell 2003). Different countries use different statistical categories and the composition of the immigrant population varies across countries, in terms of the countries of origin, the regional origins within these countries, and the types (e.g. guestworker, postcolonial, refugee, family reunification) and timing of migration flows. Without controls for these factors, it is not possible to determine to what extent crossnational differences are due to differences in integration policies such as the nationality and naturalisation regulations discussed in this paper, or to migration patterns and compositional effects. Existing immigrant surveys cannot sufficiently circumvent these problems because information on the type and timing of immigration and on the regional origin of immigrants is usually lacking. Controls for country of origin are also problematic in these surveys because, as a result of the very uneven spread of immigrants across destination countries, representative surveys of the immigrant population contain many empty or nearempty cells on the country-of-origin variable. For instance, immigrant surveys in France usually contain only a handful of Turks, and surveys in Germany only a handful of Moroccans, which creates a shaky basis for statistical controls for countries of origin. To avoid these problems, the data used in this paper are based on a quasiexperimental design that focuses on one comparable and clearly circumscribed immigrant group in three destination countries—namely Turks originating from two rural regions in Turkey—who either themselves migrated before 1975 or are E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 the foreign-born children of these first-generation guestworkers. We exclude the second generation born in the host country because, in countries with strong ius soli elements in their citizenship law such as France and the Netherlands, virtually the entire second generation holds citizenship, and thus there is no empirical basis for a comparison of its naturalised to non-naturalised members of the second generation.1 In the following section, we discuss existing research on the relationship between naturalisation and integration, and formulate two hypotheses to test the claim that easy naturalisation promotes socio-cultural integration. We then provide an overview of naturalisation policies in Germany, the Netherlands and France. Subsequently, we discuss the data and the operationalisation of variables, and present the results of regression analyses with host-country identification, language use and proficiency, and social contacts with host-country ethnics as the dependent variables. We conclude that host-country identification is indeed enhanced by easily accessible naturalisation, but linguistic and social integration are not. Naturalisation is positively associated with linguistic integration only in those countries that have traditionally required a certain degree of cultural assimilation from their new citizens. However, we do not find any indication that the allowance of dual nationality would be detrimental to socio-cultural integration. Citizenship and Socio-Cultural Integration Bauböck et al. observe that in recent naturalisation policy developments ‘the concept of “naturalisation as a means of integration” is apparently being replaced by another paradigm of naturalisation as the “crowning of a completed integration process”’ (2006: 24). Both paradigms presume a link between naturalisation and integration, but in a different direction. Several studies have looked into the relation between integration and naturalisation. Some of them looked at measures of integration as determinants for naturalisation (e.g. Constant et al. 2008; Portes and Curtis 1987; Yang 1994), others at the effects of naturalisation on integration (e.g. Bevelander and Veenman 2006; Fougère and Safi 2008). Both types of study have found mixed results. In the United States, Yang (1994) found a positive relation between English competence and naturalisation, but Portes and Curtis (1987) found no significant relation between knowledge of English and the likelihood of naturalisation for Mexican immigrants. In Germany, Constant et al. (2008) found a positive effect of having close German friends both on the intention to naturalise, and on actual naturalisation, for immigrants from Turkey and Yugoslavia. However, in the Netherlands, Bevelander and Veenman (2006) found no significant relationship between contacts with Dutch natives and the odds of naturalisation for Turkish and Moroccan immigrants. If a relationship between socio-cultural integration and naturalisation exists, it is of course important to know the direction of this relationship. One way of testing this is using panel data (Portes and Curtis 1987). However, longitudinal studies within one country are not really suited to answer the question that is central in public debates about naturalisation—namely whether naturalisation with minimal or with strict requirements has the strongest positive impact on socio-cultural integration. Even if for a certain country it is established that naturalisation has positive subsequent effects on socio-cultural integration, it does not follow logically from this that lowering the requirements for naturalisation and thus increasing the number of naturalisations will have positive aggregate effects on socio-cultural integration, because the naturalisation effect may well depend on the strictness and type of criteria attached to naturalisation. Therefore cross- E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 national analyses that compare countries with different naturalisation regimes are necessary to complement existing single-country studies. Although the results of previous studies are inconclusive, not least because they lack a cross-national comparative component, for the sake of clarity we will take the view that easily accessible naturalisation promotes socio-cultural integration as a basis for formulating our hypotheses. If this view is correct, two things should follow. To begin with, immigrants who hold the nationality of the country of residence should display, compared to non-naturalised immigrants, higher levels of socio-cultural integration in the sense of stronger identification with the country of residence, higher language proficiency and usage, and more social contacts with host-country ethnics (H1). While finding that such an empirical pattern is necessary for accepting the claim that easy naturalisation promotes socio-cultural integration, it is not sufficient. A positive correlation between host-country nationality and socio-cultural integration would namely also fit the opposite view that strict naturalisation requirements stimulate socio-cultural integration. We must therefore also look at the data from a cross-nationally comparative angle. If the view that easy naturalisation promotes socio-cultural integration is correct, we should find that immigrants in countries with accessible citizenship regimes display higher levels of sociocultural integration than their counterparts in countries with restrictive citizenship regimes (H2). Naturalisation Policies in Germany, the Netherlands and France Of the three countries in this study, Germany has the most rigid naturalisation regime. It is probably the most cited example of an ‘ethnic’ citizenship regime. Reforms in 1991 and 1993 made naturalisation somewhat easier for both firstand second-generation immigrants. However, for the first generation, language knowledge as well as an ‘orientation towards German culture’ (Hinwendung zum Deutschtum) remained preconditions for naturalisation, although the strictness with which they were applied varied across the German federal states (Hagedorn 2001; Hailbronner and Renner 1998). Over the course of the 1990s the naturalisation rate increased slowly from 0.4 per cent in 1990 to a peak of 2.5 per cent in 2000 (see Table 1). Table 1. Naturalisation rates for all immigrants and Turkish immigrants in 1990, 2000 and 2005 by country Naturalisation rate, Naturalisation rate, Turkishall foreign- born¹ born¹ 1990 1995 2000 2005 1990 1995 2000 2005 Germany 0.4 1.0 2.5 1.7 0.1 1.6 4.01 1.9 Netherlands 1.8 11.4² 7.7 4.1 1.0 19.9² 4.7 3.5 1 3 1 3 France 1.7 4.6 4.3 0.6 5.9 5.1 Source: Authors’ own calculations based on SOPEMI 2000 and 2008. 1 Data for 1991. ² Data for 1996. ³ Data for 1999. The citizenship law that came into effect in 2000 lowered the residence requirement to eight years and abolished the requirement of identification with the German culture. At the same time, language criteria were formalised and a loyalty oath to the German constitution was introduced (Groenendijk et al. 2000; Koopmans et al. 2005). Though immigrants are still required to renounce their E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 previous nationality, the grounds for exemption have been broadened. Between 1987 and 1999 dual nationality was tolerated in 23 per cent of all naturalisations (170,000 cases). Since the 2000 citizenship law, the average rate has increased, largely due to the automatic granting of dual citizenship to refugees (Green 2005). In 2005 the toleration rate of dual citizenship for all immigrants was 47.2 per cent, though for Turks it was only 15.5 per cent. Contrary to France and the Netherlands, Germany does not allow recipients of welfare or unemployment benefits to naturalise, unless they ‘cannot be held personally responsible’ for this situation (Hailbronner 2006; Koopmans et al. 2005). In 2000 Germany also introduced a limited form of ius soli for the second generation. The implementation of the 2000 law did not, however, lead to a higher naturalisation rate. After the peak in 2000, the rate slowly declined again, and by 2005 it was down to 1.7 per cent. Several authors have suggested that the long processing time of citizenship applications deters people from applying (Green 2005; Koopmans et al. 2005), but the newly implemented language and civic integration tests are also likely to have played a role. Until 1994 the naturalisation rate for Turks was lower than the general rate—as low as 0.1 per cent in 1990. In the following decade the naturalisation rate rose, two factors contributing to this increase. Firstly military service as a precondition for being released from Turkish citizenship was more often accepted as grounds for allowing dual nationality. Secondly the Turkish government started to allow the reacquisition of the Turkish nationality after German naturalisation, and between 1993 and 1995 the naturalisation rate of Turks almost tripled (Joppke 1999). Many naturalised Turks reacquired Turkish nationality after receiving German nationality. With the 2000 nationality law, the voluntary acquisition of a foreign nationality resulted in the automatic withdrawal of the German nationality. Due to this new rule, an estimated 40,000 Turks lost their German nationality (Hailbronner 2006). The Netherlands is the country that the most clearly shifted from the citizenshipstimulates-integration view to the view that citizenship is a crown on successful integration. From 1983 to the mid-1990s the dominant view was that citizenship acquisition stimulates integration (de Hart 2007; Heijs 1995). Therefore the new Citizenship Act of 1985 lowered the requirements for naturalisation. Firstgeneration immigrants can obtain citizenship after five years of legal residence. Having a reasonable knowledge of the Dutch language and being accepted in Dutch society were requirements for naturalisation (van Oers et al. 2006) but, in practice, there was only a modest informal language assessment, consisting of a few oral questions on name, address, year of arrival and year of birth. Between 1983 and 2003 less than 2 per cent of applications were turned down on grounds of insufficient integration (van Oers et al. 2006). Until 2003, Dutch-born children of immigrants had an unconditional option right to the Dutch nationality when they come of age. Since 2003 the option right can be refused, based on the outcome of a public order investigation. In 1992 dual citizenship was introduced, which led to an increase in the naturalisation rate from 4.2 per cent in 1991 to 11.4 per cent in 1996 (see Table 1). The right to dual citizenship was, however, highly contested and, in October 1997, the obligation to renounce prior citizenship was reinstated (van Oers et al. 2006). Nevertheless, there are many exemptions to the renunciation obligation and the law is not applied very rigidly. In 2006, 62.7 per cent of applicants kept their original nationality (van Oers et al. 2006), a significantly higher share than in Germany. In 2003 a new act introduced a naturalisation exam that not only tests oral and written language skills at a much higher level than before but also includes questions on Dutch politics and society. People who qualify for Dutch nationality through option (the elderly, the Dutchborn and the spouses of Dutch citizens) do not have to fulfil an integration E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 requirement. The reform led to a decrease in naturalisations, because half of the applicants failed the exam (Bauböck 2006; van Oers et al. 2006). In 2005 the naturalisation rate was down to 4.1 per cent, which is, however, still average by European standards. Until 1992 the naturalisation rate of Turkish immigrants was slightly below the Dutch average—3.0 per cent in 1991. With the allowance of dual nationality, the naturalisation rate for Turks started to rise and peaked at 19.9 per cent in 1996. Between 1992 and 1997 almost 140,000 Turks—about half of all people of Turkish origin in the Netherlands—became Dutch citizens, compared to fewer than 14,000 between 1987 and 1992. Most of those who naturalised retained their Turkish nationality (Böcker 2004). This continued after the official reinstatement of the renunciation requirement. Between 1998 and 2006 the number of people holding both Dutch and Turkish nationality increased by almost 90,000. Böcker and Thränhardt (2003) calculated that in 2001 all naturalising Turks kept their Turkish citizenship. France has a strong ius soli tradition of citizenship and, as a result, secondgeneration immigrants automatically become French. For first-generation immigrants, naturalisation is possible after five years of residence. Applicants have to prove their language ability and sufficient assimilation, the latter being part of French nationality law since 1945. In the 1950s sufficient assimilation mainly meant sufficient language knowledge but, in the 1970s, when the number of non-European applicants for naturalisation increased, sufficient assimilation also meant accepting French values—wearing a headscarf, for example, was sometimes judged to be a sign of insufficient assimilation (Weil and Spire 2006). Between 1985 and 2003 about 25 per cent of applications were turned down. According to Weil and Spire (2006), 40 per cent of rejected applications (i.e. 10 per cent of all applications) had to do with insufficient assimilation, five times as many as in the Netherlands. A 2003 law introduced knowledge of the rights and duties of citizenship as one of the criteria for assimilation into the French community. Since 2005 French proficiency is determined in a 20–30-minute interview in an Assimilation Evaluation Office (Weil and Spire 2006). Despite low residence requirements and the full allowance of dual nationality, the naturalisation rate in France is not very high; 2.5 per cent in 1990, increasing to 4.7 per cent in 2000 (see Table 1). However, this is in part explained by the fact that—unlike in the Netherlands—the obtention of citizenship through ius soli by the second generation is not represented in the French naturalisation statistics. 2 A study by Tribalat showed that naturalisation among Turks was relatively rare. Of those who came to France before 1975, only 13 per cent of men and 17 per cent of women had acquired French nationality (Tribalat 1995). But, as in Germany and the Netherlands, their naturalisation rate rose during the 1990s, exceeding the average rate and reaching 5.5 per cent in 1999 and 6.1 per cent in 2005. Table 2 summarises the citizenship regulations in the three countries. All three have made changes to their citizenship legislation over the past decades and in all countries both views on the relation between citizenship and integration have been present in political debates. Nevertheless differences remain. France combines a short residence requirement and the allowance of dual nationality with fairly strong linguistic and cultural integration requirements and a strong ius soli for the second generation. Until 2003, the Netherlands had the lowest barriers to naturalisation, with a short length of residence, minor integration requirements and a de facto acceptance of dual nationality. Germany has the highest naturalisation requirements and allows dual nationality only in a minority of cases. Thus, if Hypothesis 2 about the effects of accessible naturalisation is correct, we should find that levels of socio-cultural integration are highest in the E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Netherlands, intermediate in France and lowest in Germany. The 2003 legislation change in the Netherlands is not likely to have had a big impact on the respondents of this study since they are long-time immigrants and were eligible for naturalisation well before 2003. Table 2. Overview of citizenship regulations Germany Netherlands France Residence requirement 10 years (8 yrs since 2000) 5 years 5 years Language requirement Yes Yes, simple test until 2003. Now oral and written test Yes Integration requirement Yes Yes, simple oral test; oral and written test + societal knowledge since 2003 Yes Exclusion of welfare recipients Yes, but with exceptions No No Dual nationality Not allowed, several grounds for exemption. Länder differ in applied rigidity Allowed between 1991–97; since 1998 mostly condoned Unconditional Ius soli Yes (since 2000, but only if nationality of parents is renounced) Yes (since 1985), can opt in at majority Yes , can opt out at majority (1993–98 manifestation de volonté) Data and Variables Several studies have shown that naturalisation rates vary between origin groups within the same country of residence (Bevelander and Veenman 2006; Diehl and Blohm 2003; Fougère and Safi 2008; Staton et al. 2007; Yang 1994). To get a clearer view of differences between countries it is therefore best to study the same immigrant group in each country. Turks are the largest group of thirdcountry nationals in the EU, accounting for approximately 20 per cent of all thirdcountry nationals (Groenendijk et al. 2000). The Turkish immigrant population of Germany currently amounts to over 2.5 million and makes up almost 3 per cent of the German population. France and the Netherlands have a significant, but much smaller Turkish population of about 350,000, making up respectively 0.5 and 2 per cent of the population. Turkish migration patterns to Germany, France and the Netherlands were fairly similar during the guestworker era, but started to diverge after 1975 due to different regulations for family migration and inflow of asylum-seekers (see e.g. Dagevos et al. 2006; Muus 2003). To minimise the effects of different immigration policies, the target population of our study is limited to migrants who arrived before 1975 and their Turkish-born offspring. Since Turkey is a country with large regional differences in prosperity, religious life, ethnic composition, degree of urbanisation and level of education, and the regional origins of the Turkish communities differ between countries, the target population is further limited to migrants from two rural regions in Turkey, South-Central and EastCentral Anatolia.3 It is important to emphasise that our survey did not aim to be representative for the Turkish populations in Germany, France and the E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Netherlands. The aim was to create a cross-nationally comparable sample, thus allowing a better test of causal hypotheses than would have been possible with representative samples, which would have amounted to comparing apples and oranges, without being sufficiently able to control for their different properties. The sample was mainly drawn on the basis of surname sampling from online phonebooks, based on a list of 30 stems of common Turkish surnames. Though phonebook sampling does not provide a perfectly representative sample, it is the best way to obtain a cross-nationally comparable sample. While in the Netherlands the official registration of ethnicity makes it possible to draw immigrant samples that include people holding Dutch nationality from official registries, this strategy is not applicable in France—where the registration of ethnicity is banned by law—or in Germany, where only foreign nationals are registered, and where, as in France, naturalised immigrants disappear in the statistical category of ‘Germans’. By using Turkish surnames, we were sure that people of Turkish origin had an equal chance of being sampled in each of the countries, irrespective of their nationality status. The choice for telephone directories follows from this sampling decision, as phone directories are the most encompassing listing of names that is available. Of course, some people are not listed in the phonebook and therefore could not enter our sample. However, this factor plays in all three countries, and therefore does not subtract from crossnational comparability. Again, it is important to emphasise that the aim of this survey was not representativeness, but comparability. We could have drawn a more representative sample in the Netherlands on the basis of official registries, but such a choice would have undermined cross-national comparability. In crossnational research the best strategy is not to make the optimal choice for each country individually, but to make the same choice in all countries, even if for some countries ‘better’ options were available. Nevertheless, to check for possible bias in this method, we employed two additional strategies in addition to phonebook sampling, namely by recruiting respondents during their summer holidays in the region of origin, and by asking respondents in one destination whether they could give us phone numbers of their parents or children, or of people from their region of origin in one of the other destination countries. Of course, these sampling methods may have their own bias. Therefore all analyses in this paper were controlled for sample type, indicating that there is no important sample bias, as none of the sample control variables attains significance in any of the analyses. Data were collected between November 2005 and June 2006 by means of a standardised telephone survey in all three countries. The survey was conducted by bilingual interviewers and, depending on the preference of the respondent, could be completed in either Turkish or the language of the country of residence. All respondents qualified for citizenship based on the length-of-residence criteria in their host country. Most previous studies used a limited operationalisation of socio-cultural integration. Yang’s data only allowed him to look at English language competence (1994), Constant et al. (2008) only at having German friends. Bevelander and Veenman (2006) used a more elaborate operationalisation by measuring identification with the host country, contacts with host-country nationals and levels of modernity. We measured socio-cultural integration with four indicators—host-country identification, language use and proficiency, and social contacts with host-country ethnics. Language is often cited as one of the most important aspects of integration, with insufficient proficiency seen as a threat to national cohesion and a cause of insufficient (economic) independence. We will examine the relation between the possession of citizenship of the host E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 country and proficiency and frequency in the usage of the host-country language. This latter was measured by asking respondents which language they spoke the most frequently in three different contexts, namely with their friends, partner and children: Turkish, French/Dutch/German, or both about equally often. The answers were converted to scores of 0 (mostly Turkish), 0.5 (equally often Turkish and French/Dutch/German) and 1 (mostly French/Dutch/German). A scale was constructed based on the means of the three items (Cronbach’s alpha .66). To measure host-country language proficiency, respondents were asked how often they experienced problems in understanding. Responses were measured on a 5-point scale ranging from ‘never’ to ‘always’. This scale was inverted so that a higher score means fewer problems and therefore a greater proficiency. Loyalty has always been an important part of citizenship. The quintessential immigration country, the United States, has therefore long demanded an oath of allegiance of its new citizens. Loyalty is operationalised as identification with host-country nationals (Germans, French, Dutch) and measured with three questions: To what extent do you feel connected to [group]?; To what extent do you feel [group member]?; To what extent are you proud of being [group member]? Cronbach’s alpha for identification with the host country is 0.78. As a final indicator of socio-cultural integration into the host society, we look at social contacts. Respondents were asked about the ethnic composition of the social group they went out with. The scores are 1 (predominantly Turkish), 2 (equally often Turkish and French/Dutch/German)4 and 3 (predominantly Dutch/German/French). The difference between naturalised and non-naturalised Turkish immigrants in each of the three countries is modelled by creating six dummy variables; naturalised and non-naturalised immigrants in each of the three countries. Nonnaturalised immigrants in Germany serve as the reference category since our hypotheses predict they will have the lowest level of socio-cultural integration. In this way, we can investigate simultaneously the difference between naturalised and non-naturalised immigrants within a country, and the differences between countries. In addition to the regressions with non-naturalised German Turks as the reference category, we also ran regressions with the other five categories as reference groups in order to be able to test the significance of the difference between each pairwise contrast, e.g. between naturalised and non-naturalised immigrants in France, or between naturalised immigrants in the Netherlands and Germany. We report these significance levels in the text. In addition, we controlled in the analyses for individual-level demographic factors that are known to influence socio-cultural integration and naturalisation (gender, generation, level of education, employment and marital status). Generation is added as a dummy that distinguishes between immigrants who migrated as adults (the first generation) and those who migrated as minors (the in-between or 1.5 generation). Three additional demographic characteristics were controlled for: region of origin, religion, and the relative size of the Turkish immigrant community (in the respondents’ place of residence, as this may affect socio-cultural integration).5 East-Central Anatolia is an ethnically and religiously more diverse region than South-Central Anatolia, and this can affect sociocultural integration. The same holds for religious denomination—the sample includes both Sunnis and Alevis. Alevis are often considered to practice a more liberal and humanistic form of Islam than Sunnis. Finally, we also control for sample type. The phonebook sample serves as the reference category. Descriptive statistics of all variables are included in Table 3. E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Table 3. Descriptive statistics of dependent and independent variables for the Netherlands, France and Germany Variable Dependent variables Host-country identity Freq. of speaking host-country language Host-language proficiency Social contacts Mean SD Min Max 2.54 0.29 3.59 1.61 1.00 0.29 1.21 0.58 1 0 1 1 5 1 5 3 Independent variables Germany, naturalised Netherlands, naturalised France, naturalised Germany, non-naturalised Netherlands, non-naturalised France, non-naturalised 0.12 0.23 0.15 0.17 0.05 0.26 0.32 0.42 0.36 0.37 0.21 0.44 0 0 0 0 0 0 1 1 1 1 1 1 Female 1.5 generation Education, none/primary Education, secondary Education, post-secondary Alevi East-central Anatolia Married Working Phonebook sample Holiday sample Snowball sample Share of Turkish immigrants 0.41 0.70 0.36 0.55 0.09 0.10 0.38 0.92 0.50 0.66 0.09 0.26 2.11 0.49 0.46 0.48 0.50 0.29 0.30 0.49 0.27 0.50 0.48 0.28 0.44 1.49 0 0 0 0 0 0 0 0 0 0 0 0 0.02 1 1 1 1 1 1 1 1 1 1 1 1 7.46 Results The percentages of naturalised citizens in our sample are presented in Table 4. The data show the expected pattern, with a high share of host-country citizenship possession in the Netherlands and lower shares in France and Germany. The second column shows the percentage of people with dual citizenship among those who naturalised. These data fit with the trends among immigrants in general and Turks in particular within the three countries as displayed in Table 1. In the Netherlands and France, about 90 per cent of naturalised Turkish immigrants retained their Turkish passport; in Germany only 24 per cent did so. We now turn back to the multivariate analysis set out in Table 5 in order to investigate how these different patterns of naturalisation have affected the sociocultural integration of Turkish immigrants. Table 5 shows the results of ordinary least-squares regressions with each of the four indicators of socio-cultural integration as dependent variables. Starting with host-country identification we find that, in the Netherlands, the difference between naturalised and nonnaturalised Turks is not significant. In France and Germany, however, there is a significant difference between immigrants who did and those who did not naturalise (p<.001, respectively p<.05). These results thus provide support, in E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 two of the three countries, for the first hypothesis, which predicted a positive relationship between naturalisation and identification. Turning to the cross-national differences addressed by Hypothesis 2, we see that identification with the host country is higher in France and the Netherlands than in Germany, regardless of naturalisation status (compared with nonnaturalised Dutch Turks p<.05; all other differences with Germany p<.01). Identification of non-naturalised immigrants in the Netherlands is higher than of non-naturalised immigrants in France (p<.05), but for naturalised immigrants, the difference goes in the opposite direction (p<.05). These results largely support our second hypothesis—immigrants in countries with accessible citizenship regimes display higher levels of identification with the host society. For frequency of speaking the host-country language, the results show a different pattern. In both France and Germany, naturalised Turkish immigrants speak the host-country language more often than those who did not naturalise. In the Netherlands, however, the difference is not significant. Again the first hypothesis is only partly confirmed. Regarding country differences, Turkish immigrants in France use the hostcountry language significantly more often than their Dutch and German counterparts, regardless of naturalisation status. Non-naturalised French immigrants even use the host-country language more often than naturalised Dutch immigrants (p<.10). The differences between the Netherlands and Germany are not significant. Contrary to identification with the host country, frequency of speaking the host-country language therefore does not display the pattern that Hypothesis 2 predicted, as Turkish immigrants in the Netherlands do not use the host-country language more than their German counterparts, and use it significantly less than those in France. Our second language variable shows a similar pattern. The third column in Table 3 shows that naturalised immigrants in Germany experience fewer problems with German than their non-naturalised counterparts. In France this difference is also significant (p<.05). However, the Netherlands again diverge from the expected pattern; there is no significant relation between problems with the Dutch language and possession of Dutch nationality. Again, we find support for Hypothesis 1 in Germany and France, but not in the Netherlands. Table 4. Possession of host-country nationality and dual nationality by country Germany Netherlands France Host-country nationality (% of total) 39.6 82.8 36.0 Dual nationality (% of naturalised) 24.4 91.5 90.0 E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Table 5. Unstandardised coefficients of OLS regression for four measures of sociocultural integration Host-country identification Germany non-naturalised Ref. Ref. Germany, naturalised .34* Netherlands, naturalised .73*** (.12) France, naturalised Freq. of using host-country language (.14) 1.06*** (.14) Host-country language proficiency Ref. Social contacts Ref. .09* (.04) .35* (.14) .08 (.09) .03 (.03) .13 (.12) .13 (.07) .17*** (.03) .40** (.14) .23** (.08) Netherlands, non-natural’d .88*** (.19) .00 (.05) -.04 (.19) .11 (.12) France, non-naturalised .50*** (.12) .09*** (.03) .09 (.12) .15* (.07) Female -.07 .03 .07 (.08) .02 (.05) Generation 1 Ref. Generation 1.5 .12 Education, none/primary (.08) (.02) Ref. Ref. Ref. .14*** (.03) .60*** (.11) .11 Ref. Ref. Ref. Ref. Education, secondary .37*** (.10) .15*** (.02) .97*** (.10) .20*** (.06) Education, post-secondary .29 .24*** (.04) 1.28*** (.16) .33*** (.09) Sunni Ref. Ref. Ref. Ref. Alevi -.04 .08** (.03) .01 South-Central Anatolia Ref. Ref. Ref. East-Central Anatolia .23** (.08) .04 (.02) -.01 (.08) .07 (.05) Married .14 (.14) -.26*** (.03) -.18 (.14) -.01 (.08) Working -.09 (.09) .04* .18* (.09) .13* Phonebook sample Ref. Holiday sample .10 (.13) .06 (.03) -.03 (.13) .00 (.08) Snowball sample .03 (.09) .01 (.02) .13 (.09) .06 (.05) Share Turkish immigrants .01 (.03) -.01* (.01) .00 (.03) -.01 (.02) (.20) .21*** (.05) Constant Adj. R2 N 1.50*** .13 (.11) (.16) (.14) (.02) Ref. .37 646 645 Two-tailed t-tests, *p < .05 ** p < .01 *** p < .001 (.14) (.05) Ref. 2.35*** (.20) 645 (.08) Ref. Ref. .40 .05 (.06) 1.16*** (.12) .09 626 Cross-nationally, we find that naturalised Turks have fewer problems with the host-country language in France than in the Netherlands (p<.05). Among the non-naturalised Turks we do not find any significant cross-national differences. Finally, we look at social contacts with host-country ethnics. In none of the three countries is the difference in the extent of interethnic contacts between the naturalised and the non-naturalised significant. In other words, we find no support for Hypothesis 1 in regard to this variable. Comparing across the three countries, we find that naturalised French and Dutch Turks have higher levels of interethnic E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 social contacts than the reference category of non-naturalised German Turks. If we hold nationality status constant, we find no significant country differences among naturalised immigrants. Among non-naturalised immigrants, the only significant difference is between French and German Turks. As far as the relatively low levels of interethnic contacts among German Turks is concerned, this result fits Hypothesis 2. However, the fact that interethnic contacts are somewhat more strongly developed among French than among Dutch Turks is not in line with it. Conclusions Despite convergence, citizenship legislation still varies between countries. Moreover, European immigration countries still carry the imprint of the more strongly divergent policies of past decades. As we have seen, these differences are reflected in naturalisation rates and in the prevalence of dual nationality among the naturalised. In the Netherlands which—at least until 2003—had the easiest access to naturalisation, the majority of Turkish immigrants have naturalised. In France the long-time presence of ius soli has led to a high degree of citizenship possession for the second generation but, despite a similarly short residence requirement and the full allowance of dual nationality, the naturalisation rate of Turkish immigrants is much lower than in the Netherlands. This is related to the much stricter linguistic and cultural assimilation requirements that applied to naturalisations in France compared to those in the Netherlands before 2003. Based on the assumption that easily accessible naturalisation promotes sociocultural integration, we formulated two hypotheses. The first implication of this assumption pertains to within-country differences, and states that naturalised immigrants should display higher levels of socio-cultural integration than those who did not naturalise. This hypothesis received support for the German and French cases regarding identification and language. Naturalised Turkish immigrants in France and Germany identified more strongly with the host country, used French or German more often and reported higher proficiency in it. In the Netherlands, however, Hypothesis 1 had to be fully rejected as we found no significant differences between the naturalised and the non-naturalised on any of the indicators of socio-cultural integration. This result reflects the absence in the Netherlands until very recently of significant linguistic and cultural assimilation preconditions for naturalisation. However, the absence of significant differences between the naturalised and the non-naturalised in the Netherlands also indicates that naturalisation has had no significant positive subsequent effects on the socio-cultural integration of those who became naturalised, as the argument that easy naturalisation promotes socio-cultural integration would have led us to expect. The second implication of the assumption that easily accessible naturalisation promotes socio-cultural integration pertains to cross-national differences, and states that levels of socio-cultural integration should be higher in countries with high naturalisation rates and minimal naturalisation requirements, along the lines of the best practices for naturalisation recommended by the authors of the European Civic Citizenship and Inclusion Index (Geddes and Niessen 2005). Hypothesis 2 therefore stated that levels of socio-cultural integration should be highest in the Netherlands, intermediate in France and lowest in Germany. This hypothesis could only be partly confirmed for identification with the host country, which was significantly stronger in France and the Netherlands than in Germany, both for the naturalised and the non-naturalised. The fact that levels of E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 identification with the host country are not higher in the Netherlands than in France does not fit the hypothesis, however. We find a similar, though less overt, pattern for interethnic social contacts, which are the most frequent among French and the least among German Turks, with Dutch Turks in between. For the language variables we found no support for the second hypothesis whatsoever. Turkish immigrants in France, and not those in the Netherlands, turn out to have the highest levels of host-country language proficiency and use. This latter result applies both to the naturalised and the non-naturalised French Turks. Even non-naturalised French Turks use the host-country language more often than naturalised Dutch Turks. Turkish immigrants in Germany and the Netherlands do not differ significantly regarding language use, but those in the Netherlands report somewhat less host-country language proficiency. The fact that Turkish immigrants in the Netherlands overall show the lowest levels of linguistic integration clearly contradicts Hypothesis 2. Combining the results regarding the two hypotheses we can conclude that there is little support for the assumption that low barriers to naturalisation promote socio-cultural integration. If we compare the naturalised to the non-naturalised within countries, we find that the positive relationship between naturalisation and socio-cultural integration is strongest in France, limited to linguistic integration in Germany, and entirely absent in the Netherlands. In other words, precisely in the country with the easiest access to naturalisation, we find the least evidence of a positive impact of naturalisation on socio-cultural integration. If naturalisation has an effect on socio-cultural integration at the individual level, this effect is limited to the two countries that have made naturalisation conditional on a certain degree of linguistic and cultural assimilation. The cross-national differences that we found did not provide much evidence for beneficial effects of naturalisation with minimal conditions, either. France, the country that has historically the most emphasised linguistic assimilation as a precondition for citizenship, is also the country where Turkish immigrants display the highest levels of host-language proficiency and usage. Conversely, the lack of emphasis on linguistic assimilation that long prevailed in the Netherlands has promoted lower levels of host-country proficiency and usage. Probably the language factor also plays an important role in explaining the higher frequency of interethnic social contacts among the French Turks. The only aspect of sociocultural integration where Dutch Turks did perform similarly to their French counterparts and at a much higher level than German Turks was host-country identification. This aspect of socio-cultural integration, with the sense of belonging and acceptance that is attached to it, is of course a not-unimportant dimension of integration and in that sense past Dutch naturalisation policies have at least achieved one of their aims. There is little reason to fear, however, that this positive effect will erode now that the Netherlands have made citizenship less easily accessible, particularly by introducing stricter language requirements. This policy shift brings the Netherlands close to the kind of naturalisation policies that France has long pursued, and as our results show, levels of host-country identification in France have not been harmed by such demands for assimilation to the dominant language. Apart from the result for host-country identification, there is a second reason why our results should not be taken as support for the view that socio-cultural integration is best promoted by very strict naturalisation requirements. Had we taken this assumption as the point of departure for formulating our hypotheses, we would also have found little support for it, as Germany, which clearly has had the most restrictive naturalisation regime, performs relatively poorly on all four indicators, particularly identification and interethnic social contacts. Our results E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 rather indicate that the French combination of short residence requirements, strong ius soli elements—which of course do not affect our respondents directly, but may affect them importantly through their children—coupled with certain demands of linguistic and cultural assimilation, has been the optimal mix for promoting the socio-cultural integration of immigrants. It is worth emphasising that this French mix includes the unconditional toleration of dual nationality, often framed in political debates as antithetical to socio-cultural integration. In full contradiction to this view, France—the only one among our countries that unconditionally allows dual nationality—is simultaneously the country where the positive effects of naturalisation on sociocultural integration are the strongest, including a comparatively strong sense of identification with the host country. By contrast, Germany is the country that has the strongest restrictions on dual nationality, but this has not made German Turks better integrated than their French counterparts on any aspect of socio-cultural integration. Of course, these results should be treated with some caution because our data pertain only to Turkish immigrants. Even though Turks are the most important immigrant group in Europe, this limits the generalisability of our findings. The naturalisation behaviour, as well as the determinants and consequences of naturalisation, may differ for other immigrant groups. It is therefore important to extend this kind of study to other immigrant groups and to other immigration countries. We see the cross-nationally comparative approach that we have followed in this paper as an important complement to single-country studies of the relationship between naturalisation and integration. However, future work should try to combine the strengths of cross-national and longitudinal approaches. This would require cross-nationally comparable panel data containing information on immigrants before and after their naturalisation, and on a comparable group of immigrants in the same country who did not naturalise. Acknowledgement We would like to thank Betty de Hart, Maarten Vink and the anonymous JEMS reviewer for their constructive comments and helpful suggestions. Notes [1] [2] [3] [4] In our sample only 2.6 per cent, i.e. two respondents, of the second generation in the Netherlands do not have a Dutch passport. In France only 4.7 per cent, i.e. six respondents, do not have a French passport. From 1993 until 1998, when the Pasqua law was in effect, the French-born second generation did not automatically receive citizenship but had to show a ‘manifestation de volonté’. During this period the second generation is present in the statistics. South-Central Anatolia consists of the provinces of Afyon, Aksaray, Karaman, Kayseri, Konya, Nevşehir, and Niğde. East-Central Anatolia encompasses Adiyaman, Amasya, Elaziğ, Malatya, Tokat, Tunceli and Sivas. The middle category includes a small number of respondents (n=49), who indicated that the majority of their social contacts were with members of immigrant groups other than Turks. We also ran the analyses excluding this group and found similar results as those reported below. The only exception is that, excluding these respondents, the difference in social E. Ersanilli & R. Koopmans [5] Journal of Ethnic and Migration Studies , 36(5): 773—791 contacts between naturalised Dutch Turks and non-naturalised German Turks is not significant. We calculated the number of Turkish immigrants (excluding the second generation) as a percentage of the total population within geographical units. The variable thus varies from 0–100. For the Netherlands, data for 2005 on the municipal level were taken from the Central Statistical Agency (CBS) website. For France, data from the 1999 census on the level of the commune were used. Data were not available, however, for communes with fewer than 5,000 inhabitants, so the percentage of Turkish migrants within the respective arrondissements were used. German data were taken from the only dataset with information on the country of residence instead of nationality—the Mikrozensus 2005—the Kreis or country level being the lowest level for which it will allow us to calculate the percentage of Turkish immigrants. References Bauböck, R. (2006) Migration and Citizenship: Legal Status, Rights and Political Participation. Amsterdam: Amsterdam University Press. Bauböck, R., Ersboll, E., Groenendijk, K. and Waldrauch, H. (eds) (2006) Acquisition and Loss of Nationality: Policies and Trends in 15 European States. Amsterdam: Amsterdam University Press. Bevelander, P. and Veenman, J. (2006) 'Naturalization and employment integration of Turkish and Moroccan immigrants in the Netherlands', Journal of International Migration and Integration, 7(3): 327–49. Böcker, A. (2004) 'The impact of host-society institutions on the integration of Turkish immigrants in Germany and the Netherlands'. Istanbul: Bogazici University, paper presented at a workshop on the integration of immigrants from Turkey in Austria, Germany and the Netherlands, 27–28 February 2004. Böcker, A. and Thränhardt, D. (2003) 'Einbürgerung und Mehrstaatigkeit in Deutschland und den Niederlanden', in Thränhardt, D. and Hunger, U. (eds) Migration im Spannungsfeld von Globalisierung und Nationalstaat. Wiesbaden: Westdeutscher Verlag, 117–34. Constant, A., Gataullina, L. and Zimmerman, K. (2008) Naturalization Proclivities, Ethnicity and Integration. Berlin: Institute for the Study of Labor, IZA discussion paper 3260. Dagevos, J., Euwals, R., Gijsberts, M., and Roodenburg, H. (2006) Turken in Nederland en Duitsland. De arbeidsmarktpositie vergeleken The Hague: Sociaal en Cultureel Planbureau. de Hart, B. (2005) 'Het probleem van dubbele nationaliteit. Politieke en mediadebatten na de moord op Theo van Gogh', Migrantenstudies, 21(4): 224–38. de Hart, B. (2007) 'The end of multiculturalism: the end of dual citizenship? Political and public debates on dual citizenship in The Netherlands (1980– 2004)', in Faist, T. (ed.) Dual Citizenship in Europe: From Nationhood to Societal Integration. Aldershot: Ashgate, 77–102. de Hart, B. and van Oers, R. (2006) 'European trends in nationality law', in Bauböck, R., Ersboll, E., Groenendijk, K. and Waldrauch, H. (eds) Acquisition and Loss of Nationality: Policies and Trends in 15 European States. Amsterdam: Amsterdam University Press, 317–57. Diehl, C. and Blohm, M. (2003) 'Rights or identity? Naturalization processes among "labor migrants" in Germany', International Migration Review, 37(1): 133–62. E. Ersanilli & R. Koopmans Journal of Ethnic and Migration Studies , 36(5): 773—791 Favell, A. (2003) 'Integration nations: the nation state and research on immigrants in Western Europe', Comparative Social Research, 22(1): 13–42. Fougère, D. and Safi, M. (2008) The Effects of Naturalization on Immigrants' Employment Probability (France, 1968–1999). Berlin: IZA. Geddes, A. and Niessen, J. (2005) European Civic Citizenship and Inclusion Index. Brussels: British Council. Green, S. (2005) 'Between ideology and pragmatism : the politics of dual nationality in Germany', International Migration Review, 39(4): 921–52. Groenendijk, K., Guild, E. and Barzilay, R. (2000) The Legal Status of Third Country Nationals who are Long Term Residents in a Member State of the European Union. Strasbourg: Council of Europe. Hagedorn, H. (2001) Wer darf Mitglied warden? Einbürgerung in Deutschland and Frankreich im Vergleich. Opladen: Leske. Hailbronner, K. (2006) 'Germany', in Bauböck, R., Ersbøll, E., Groenendijk, K. and Waldrauch, H. (eds) Acquisition and Loss of Nationality Policies and Trends in 15 European States. Amsterdam: Amsterdam University Press, 213–51. Hailbronner, K. and Renner, G. (1998) Staatsangehörigkeitsrecht. Munich: Beck. Heijs, E.J.M. (1995) Van vreemdeling tot Nederlander: de verlening van het Nederlanderschap aan vreemdelingen (1813–1992). Amsterdam: Het Spinhuis. Howard, M.M. (2005) 'Variation in dual citizenship policies in the countries of the EU', International Migration Review, 39(3): 697–720. Joppke, C. (1999) 'How immigration is changing citizenship: a comparative view', Ethnic and Racial Studies, 22(4): 629–52. Koopmans, R., Statham, P., Giugni, M. and Passy, F. (2005) Contested Citizenship. Immigration and Cultural Diversity in Europe. Minneapolis: University of Minnesota Press. Muus, P.J. (2003) 'An international comparison of migration and immigrant policy with respect to immigrants from Turkey and their participation in the labour market', in Hagendoorn, L., Veenman, J. and Vollebergh, W. (eds) Integrating Immigrants in the Netherlands. Cultural Versus Socio-Economic Integration. Aldershot: Ashgate, 17–40. Portes, A. and Curtis, J.W. (1987) 'Changing flags: naturalization and its determinants among Mexican immigrants', International Migration Review, 21(2): 352–71. Staton, J.K., Jackson, R.A. and Canache, D. (2007) 'Dual nationality among Latinos: what are the implications for political connectedness?', The Journal of Politics, 69(2): 470–82. Tribalat, M. (1995) Faire France: une grande enquête sur les immigrés et leurs enfants. Paris: La Découverte. van Oers, R., de Hart, B. and Groenendijk, K. (2006) 'The Netherlands', in Bauböck, R., Ersbøll, E., Groenendijk, K. and Waldrauch, H. (eds) Acquisition and Loss of Nationality Policies and Trends in 15 European States. Amsterdam: Amsterdam University Press, 391–434. Weil, P. and Spire, A. (2006) 'France', in Bauböck, R., Ersbøll, E., Groenendijk, K. and Waldrauch, H. (eds) Acquisition and Loss of Nationality Policies and Trends in 15 European States. Amsterdam: Amsterdam University Press, 187–211. Yang, P.Q. (1994) 'Explaining immigrant naturalization', International Migration Review, 28(3): 449–77.