Acculturation and Language Orientations of Turkish Immigrants in Australia,France,Germany,and the Netherlands

Abstract

The authors examined acculturation and language orientations among Turkish immigrants in Australia (n = 283), France (n = 266), Germany (n = 265), and the Netherlands (n = 271). They expected that the countries with the least pluralistic climate (France and Germany) would show the lowest level of sociocultural adjustment and the highest level of ethnic orientation and language use; the opposite was expected in Australia, as the country with the most pluralistic climate; and the Netherlands would have an intermediate position. The predictions were largely borne out. The language orientation measures yielded a (symbolic) language value factor and a (behavioral) language preference factor. In all countries Turkish identity was a positive predictor and mainstream identity a negative predictor of both the language value and preference factor. Mainstream and Turkish identity showed stronger negative correlations in the less pluralistic countries. It is concluded that immigrants showed the least maintenance and the most adjustment in Australia, which is the country with the least pressure to assimilate.

Keywords

acculturation orientations identity integration policy self-identification Turkish immigrants

Acculturation research is flourishing. It is probably fair to say that we are slowly moving away from a first wave of studies of single immigrant groups in specific countries that may show a low external validity in other case studies. It has been argued that we need to move to comparative acculturation studies in which acculturating groups are followed over time (longitudinal studies), or in which these groups are compared within a single country, or studies in which immigrants from a single ethnic background are compared across countries (Sam & Berry, 2006). In line with this recommendation, the present study compares Turkish immigrant groups in four countries: Australia, France, Germany, and the Netherlands. These countries have salient differences in policies and general climate vis-à-vis multicultural and multilingual issues (as further explained below). We are interested in the social-psychological and sociolinguistic ramifications of these differences. Sociolinguistic aspects involve proficiency in the host language, language usage norms, registers, preference, and perceived value of heritage language as well as attitudes toward this language (Kang, 2006). Drawing on these immigration climate differences, we test hypotheses about cross-cultural differences in social-psychological and linguistic aspects of identity. We also test the cross-cultural invariance of a model that links mainstream identity, ethnic identity, and sociolinguistic variables.

Sociolinguistic Perspective on the Role of Language in Acculturation

Sociolinguistic studies have examined the linguistic consequences of immigration. Language maintenance or shift emerges as a result of language contact. The type of interaction between the majority and minority language speakers influences the linguistic outcome. There are a number of models in the sociolinguistic literature for the investigation of language maintenance and shift as documented by Clyne (1991). Models developed by Bourdieu (1982), Edwards (1992), Giles, Bourhis, and Taylor (1977), and Smolicz (1981) identify various factors that are important in language maintenance (or shift). These factors are usually divided into two related categories: those affecting a speech community and those affecting individuals within a speech community (Kipp, Clyne, & Pauwels, 1995). Group factors include size and distribution of an ethnic group, the policy of the host community toward minority languages, the position of the language within the cultural value system of the group, and proximity or distance of the minority language to or from the majority language. Birthplace, age, period of residence, gender, education, marriage patterns, prior knowledge of majority language, reason for migration, and language variety are considered to be relevant individual factors (Kipp et al., 1995, p. 123). We might add factors at the individual level such as language choice and socialization patterns, the interaction of which varies quite fundamentally across linguistic contexts. Different speech communities or individuals can behave in different ways under similar circumstances.

Retention of the heritage language is more likely when the ethnic group is larger, is more concentrated in certain areas, and the heritage language is held in higher esteem by its speakers. It has been suggested that

a minority group that possesses a publicly stigmatized identity, that has few legislative means at its disposal with which to secure its interests, that lives in a society characterized by an assimilatory ideology, and that is disadvantaged in relation to the majority with respect to economic and educational resources, could be expected to be less likely to maintain its language over time. (Hyltenstam & Stroud, 1996, pp. 569-570)

The type of interaction between the minority and the majority is another factor to be considered. One of the basic premises of the present study is that language maintenance, shift, and loss of minority groups do not occur in a political and social vacuum; acculturation orientations of the immigrant groups and language policies of the receiving societies have an effect on language use and adaptation patterns. A stronger ethnic orientation is taken to be accompanied by more ethnic language retention and a more positive attitude toward the language. The social function of language is regarded as a mechanism of social integration in the acculturation process. A close analysis of language from the perspective of sociology of language and sociolinguistics can give a detailed account of the role of language in social interaction and in the construction of ethnicity in the acculturation of immigrants (Koenig, 1999). By means of its communicative and symbolic function, language contributes to social integration of immigrants; in addition, language can be an important part of heritage identity.

Psychological Perspective on the Role of Language in Acculturation

From a psychological perspective, language can play multiple roles in acculturation. We discuss here the role of host and heritage language proficiency for sociocultural adjustment and ethnic identity.

Host language proficiency

Psychological studies have addressed the role of proficiency in the host language for sociocultural adjustment (e.g., Masgoret & Ward, 2006; Vedder & Horenczyk, 2006). Positive relations have been reported between host language proficiency, sociocultural adjustment, and host identity across various Western societies (Ataca & Berry, 2002; Clément, Noels, & Deneault, 2001; Jasinskaja-Lahti, 2008; Ward & Kennedy, 1993). The association may be generated by multiple pathways, most notably the opportunity to become acquainted with the host culture, establish relationships with mainstreamers, identify with the host group, and access important resources of the new host culture, such as school and job opportunities.

Heritage language proficiency

Proficiency in the heritage language has often been studied in relation to ethnic identity. The pathways through which ethnic language is related to adjustment to the ethnic culture are identical to those of the mainstream language just mentioned (access to the culture, networks, and resources). Yet the role of ethnic language in acculturation is more ambiguous than the role of the host language. There are two opposing views on the role of ethnic language. On the one hand, knowledge of the heritage language can help to maintain the immigrant’s ties with the ethnic culture, which in turn can facilitate psychological adjustment (Virta, Sam, & Westin, 2004). The stronger roots in the ethnic culture may facilitate sociocultural adjustment through ethnic community resources (also known as social and cultural capital) and may also facilitate adjustment to the host culture (Ait Ouarasse & Van de Vijver, 2004; Phinney, Horenczyk, Liebkind, & Vedder, 2001). For example, Fuligni, Kiang, Witkow, and Baldelomar (2008) emphasize the role of knowledge of the heritage language in Asian and Latin American adolescents in the United States to explore their ethnic roots and to have a sense of belonging to the culture of their ancestors. Knowledge of the heritage language is not taken to impede participation in U.S. society. On the other hand, there are studies in which knowledge of the host and heritage cultures (and accompanying identities) show opposite patterns. Clément and Noels (1992) found that more knowledge of the mainstream language and stronger ingroup feelings toward the majority were accompanied by a loss of knowledge of the heritage culture and a weaker ethnic identity.

The unclear status of the role of the heritage language in acculturation stems from two related sources. The first is related to the discussion about the endpoint of acculturation: Does acculturation always lead to complete assimilation, as claimed by authors like Gordon (1964), or is integration (with a bicultural identity and bilingualism) a long-term sustainable alternative, as the currently popular view in cross-cultural psychology holds (e.g., Berry, 1997; Phinney et al., 2001)? Language attrition provides a good example. The common model of language assimilation, developed in the United States, is the three-generation model of language assimilation: “The immigrant generation makes some progress but remains dominant in their native tongue, the second generation is bilingual, and the third generation speaks English only” (Waters & Jiménez, 2005, p. 110). The model generally is in line with empirical data, which show immigrant groups to shift to the mainstream language within three or four generations (Gonzo & Saltarelli, 1983), yet the model does not deal with the considerable variation across ethnic groups in speed of language assimilation. For example, Turkish immigrants are often able to maintain their language better across generations than predicted by the model (Extra & Yağmur, 2004), whereas Dutch immigrants to Australia and New Zealand showed much quicker language attrition (Clyne, 1992). The second source of vagueness has to do with identity. Language usage is an important behavioral marker of ethnic identity (Ashmore, Deaux, & McLaughlin-Volpe, 2004), but ethnic groups differ considerably in the symbolic value they attach to the heritage language. Whereas the Turkish language is viewed as a core marker of identity among Turkish immigrants in Western Europe (Extra, Yağmur, & Van der Avoird, 2004), speaking Berber is less central to the ethnic identity of Moroccan immigrants in the same countries; religion is a more important marker of their ethnic identity. As a consequence, the role of the ethnic language in the process of acculturation can differ across immigrant groups.

It can be concluded that the host language plays an important role in sociocultural adjustment, but that the heritage language plays a more ambiguous role in acculturation than the host language. We do not know under which conditions heritage language proficiency may (or may not) help to foster adjustment to the mainstream culture. Finally, there is a caveat that applies to both host and heritage language proficiency. The causal status of language in acculturation is far from clear. Language proficiency and other language variables have been used as antecedent, intervening, and outcome variables. A similar confusion underlies the mixture of language attitudes, skills (self-perceived or actual), and behaviors that have been used in combination in measures of acculturation. Our study aims to clarify the role of ethnic and host language in the process of acculturation.

Multiculturalism Ideologies and Heterogeneity in the Four Countries

In the literature, four clusters of state ideologies shaping integration and language policies of immigrant receiving societies are identified (Bauböck, Heller, & Zolberg, 1996; Bourhis, Moïse, Perreault, & Senécal, 1997; Koenig, 1999). This ideological clustering model distinguishes pluralist, civic, assimilationist, and ethnist ideologies. In the pluralist ideology, the state provides support for language classes and cultural activities to promote mother tongue maintenance alongside second language proficiency. Maintenance of ethnic group norms and values is accepted. A civic ideology expects that immigrants will adopt the public values of the mainstream society. The state neither interferes with the private values of its citizens nor provides any provisions for the maintenance or promotion of linguistic or cultural values of minorities. An assimilation ideology expects linguistic and cultural assimilation into the mainstream society. In the name of homogenization of the society, assimilationist language policies aim at accelerating language shift. An ethnist ideology shares most aspects of assimilation ideology, yet there are ideological and institutional barriers for immigrant minorities to be accepted legally or socially as full members of the mainstream society. Naturalization laws are helpful for distinguishing ethnist ideologies. The principle of ius sanguinis underlies acquisition of citizenship in such countries, such as immigration of the Aussiedler who after the unification of the East and West Germany could immigrate in Germany as they were recognized as ethnic Germans. On the basis of the state ideologies briefly described here, Australia, the Netherlands, France, and Germany broadly fit the four clusters of pluralist, civic, assimilationist, and ethnist ideologies, respectively. This should not be seen as a static categorization. In line with social developments in a society, especially in times of economic crises, there can be shifts in ideologies as well.

Australia

In terms of four clusters of state ideologies, Australia is shown to have a pluralistic model (Van Oudenhoven, Ward, & Masgoret, 2006). Australia is an immigration country and even though some surveys and opinion polls show widespread support for cultural pluralism in Australia, public opinion regarding multiculturalism and cultural diversity is ambivalent, especially concerning some non-European immigrant groups (Dandy & Pe-Pua, 2010). Nevertheless, Australia has an official policy of multiculturalism and cultural diversity is valued. The state provides funds and facilities for the maintenance of heritage languages and cultures of the immigrants. Among the Australian-born immigrants the desire to identify with Australia and even to assimilate into (Anglo-)Australian culture is high. Based on earlier studies of Turkish immigrants in Australia, it is clear that especially the second-generation Turks identify strongly with Australia (Yağmur, 1997). In line with the pluralistic policies of Australia, strong first language maintenance and successful sociocultural integration would be expected for different generations of Turkish immigrants.

France

Like many other West European countries, the nation-state ideology and maintenance of nationhood rooted in a commonly shared notion of cultural unity underlies French language policies (Archibald, 2002; Laroussi, 2003). Rather than promoting linguistic and cultural pluralism, French policies explicitly opt for integration and linguistic assimilation of immigrants. A Commission on Nationality was set up in 1987 to transform immigrants into French nationals. The Commission took a number of measures to develop the legal framework for achieving the assimilation of immigrants into the mainstream society (Lapeyronnie, 1990, cited in Archibald, 2002). Mastery of French was seen to be the most fundamental aspect of the acculturation process, because language is considered to be the overarching value to achieve social cohesion and national unity in France. Given the circumstances, one would hardly expect first language maintenance among younger immigrants. However, studies on Turkish immigrants have reported relatively low levels of linguistic and cultural assimilation (Rollan & Sourou, 2006; Tribalat, 1995; Yağmur & Akinci, 2003).

The Netherlands

Issues of integration, unemployment, school dropouts, and criminality are associated with immigrant groups in the media in the Netherlands. Even though such media representation may be poorly founded, Brands, Crone, Leurdijk, and Top (1998) demonstrate that immigrants are mostly associated with problems in the Dutch media, with the Turkish and Moroccan groups getting the largest share. The prevalent attitude in the Netherlands is against the maintenance of immigrant minority languages. Even in times of solid policies promoting multiculturalism in the early 90s, Extra and Verhoeven (1993) indicated that:

It is a common Dutch attitude that ethnic minority families should give up their home language and should switch to Dutch, and that ethnic minority children should spend all their energy on second language learning instead of wasting time on first language maintenance. In this conception, multilingualism is seen as a problem, not a resource. (pp. 22-23)

After a long period of support for integration both among policy makers and the public, Dutch policy makers’ approach is currently identified as “assimilationist” (Arends-Tóth & Van de Vijver, 2003; Van Oudenhoven, Prins, & Buunk, 1998; Van Oudenhoven et al., 2006). In this respect, the current trend in the Netherlands is highly similar to the “English only movement” in the United States (Barker et al., 2001). Just as in this movement, the Dutch-only tendency aims at limiting the use, maintenance, promotion, and salience of immigrant minority languages. The Netherlands is characterized by civic nationalism.

Germany

The distinction between civic and ethnic nations is one of the most common categorizations in the study of interethnic relations (Bourhis et al., 1997; Koenig, 1999). In this ideological clustering, ethnic (cultural) nations are based on common heritage, language, national territory, religion, customs, and history, whereas civic nations are based on a historic territory, laws and institutions, and the legal and political equality of all citizens. The ethnic nation is based on a common descent; a common language and customs are the key elements of the ethnic nation. The ideology of monolingualism is very strong in the ethnic state. Koenig (1999) suggests that the ethnic nation state has been based on policies of homogenizing culturally and linguistically diverse populations. Germany is identified as a typical ethnic nation encouraging monolingualism among the people in its territory (Bourhis et al., 1997). In this respect, the sociocultural distance between the mainstream German and immigrant Turkish is expected to be large.

Germany has ethnic nationalism. The ethnic-civic division is relevant for our research. Whereas civic conceptions of nationhood promote interethnic tolerance and a positive attitude toward immigrants, ethnic definitions of the nation are seen as contributing to xenophobia and racism (Janmaat, 2006). Even though Dutch policies are shown to fit civic nationalism, integration policies require the subordination of ethnic loyalties to the dominant mainstream culture and national identity in order to achieve full citizenship in the mainstream society (Phalet & Kosic, 2006).

Turkish Communities in the Four Countries

The Turkish community is the largest immigrant group in Germany (2.5 million) and the Netherlands (400,000). It is one of the largest communities in France (390,000) and somewhat smaller in Australia (90,000), yet across the four countries the immigrant communities have a common background (e.g., they originate from the same rural areas in Turkey; first-generation immigrants were mostly less educated and mainly males) and demographic characteristics in terms of concentration patterns, age composition, and immigration history (for more details, see Yağmur, 2004, for Germany; Yağmur, 2009, for the Netherlands; Yağmur & Akinci, 2003, for France; and Yağmur, De Bot, & Korzilius, 1999, for Australia).

Hypotheses

Our hypotheses start from the premise by Bourhis et al. (1997) that the multicultural ideology and climate has psychological correlates in immigrant groups. More specifically, we propose that a more pluralistic climate is associated with a stronger orientation toward the mainstream identity and language. We tested the following hypotheses:

There is a relation between the acculturation climate of receiving societies and integration patterns of Turkish immigrants. We expect that a more pluralist climate is positively related to a majority culture identity and to better sociocultural adjustment (including linguistic adjustment). More specifically, we expect that Turkish immigrants in Australia show the strongest mainstream identity and adjustment, followed by immigrants in the Netherlands, followed by immigrants in Germany and France (no differences are expected between the two latter groups).

More educated immigrants are more positively oriented toward the mainstream culture than less educated immigrants.

There is a relationship between ethnic identification and both the attitudes toward and use of the mainstream language. More specifically, there is a negative relationship between ethnic identification and the perceived value and actual use of the mainstream language.

Method

Participants

The total sample comprised 1,089 informants (575 female and 510 male). These participants included persons residing in Australia (283), France (266), Germany (265), and the Netherlands (271). The educational level was quite diverse, ranging from primary education certificate (17.7%) and upper secondary level (39.1%) to tertiary education (43.2%). A gender split for each country is presented in Table 1. Participants were selected on the basis of their willingness and their availability. In Australia, information about the survey was sent through different channels. A number of Internet sites targeting Turkish-Australians presented information about the survey and gave the link of the online survey and the telephone number of the first researcher. Besides, four Turkish language newspapers issued in Melbourne and Sydney provided the same details for their readers. Finally, Turkish consulates in Melbourne and Sydney sent out e-mails to the Turkish nationals and organizations. Two large community gatherings were visited in Sydney to distribute the questionnaires in person by the first researcher who also recruited Turkish speaking individuals in cafes, restaurants, and businesses in Auburn, a major Turkish concentration suburb in Sydney. Finally, 50 informants were recruited by a local Turkish newspaper reporter in Melbourne. In France, data were collected with the help of Turkish organizations mainly active in Lyon, Paris, and Rouen. In Germany and the Netherlands, similar procedures were followed to recruit informants. Turkish scholars from Hamburg and Bielefeld University also supported data collection in the German context. The Turkish Federation of Community Organizations provided the addresses of their members whom we could mail the questionnaires. In the Netherlands, accessing the Turkish speakers was relatively easier and straightforward due to the long-established community network of the researchers.

Table 1.

Gender Distribution of Informants for each Country

Country	Female	Male	Total
Australia	127	156	283
France	135	131	266
Germany	140	125	265
The Netherlands	173	98	271
Total	575	510	1,085

Measures

The scales, comprising 212 questions, are primarily based on the studies by Arends-Tóth (2003), Kang (2006), Phinney (1990), Verkuyten (2007), and Yağmur (1997). The biographical section included 12 questions on topics such as age, gender, marital status, birth country of the respondent and parent of the respondent (if married, partner’s country of birth as well), number of years in the immigration country, place of residence, highest diploma obtained, profession, and frequency of visits to Turkey.

The Multicultural Ideology Scale consists of 10 questions designed to assess attitudes toward linguistic and cultural diversity. The scale is based on Berry and Kalin’s (1995) instrument, which was used among Turkish-Dutch informants by Arends-Tóth (2003). An example of an item on pluralism is “<Australian> people should recognize that the <Australian> society consists of groups with different cultural backgrounds” (“Australian” was replaced by the corresponding adjective in other countries); an example of an item about language maintenance is “Ethnic minorities should be helped to preserve their cultural heritage in <Australia>.” Answers were given on a 5-point response format ranging from strongly disagree (1) to strongly agree (5).

The Ethnic and Mainstream Identification Scale consists of 21 questions designed to assess feeling of having ethnic and mainstream identity. The questions in this section are related to cultural, linguistic, social, ethnic, and religious components of ethnic identification. Examples of items are “I feel Turkish because I speak Turkish, (I am a Muslim, etc.)” and “I feel <Australian> because I speak English (I know <Australian> mentality, etc.).” All the questions in this section are rated on a 5-point response format ranging from strongly disagree (1) to strongly agree (5).

The Ethnic and Mainstream Identity Scale consists of 17 questions designed to assess orientation to Turkish and mainstream identity. All of the questions have an endorsement format, which asks participants to rate each statement in terms of how strongly they agree or disagree with it. Examples are “I am proud that I am a Turk” and “I am happy that I am <Australian>.” The section ended with the question, “All in all, do you feel more Turkish or more <Australian>,” having a 5-point response format: only Turkish (1) to only Australian (5).

The Ethnic and Mainstream Behavior Scale consists of 22 endorsement format questions designed to assess (a) attitudes toward ethnic and mainstream cultures; (b) affiliation with respective cultural groups; and (c) preferences with regard to food, music, activities, and media. The participants are asked to rate each statement in terms of how strongly they agree or disagree with, for instance, “I live in accordance with Turkish cultural norms and values” and “I live in accordance with <Australian> cultural norms and values.” All the questions in this section are rated on a 5-point response format ranging from strongly disagree (1) to strongly agree (5).

The Islamic Beliefs Scale consists of 18 questions designed to assess the extent of religious identification. Both symbolic identification and observance of religious practices were assessed. Examples of items are “I am a Muslim,” “I know Islamic rules very well,” and “I fast during Ramadan.” Questions were rated on a 5-point response format ranging from strongly disagree (1) to strongly agree (5).

The Ethnic and Mainstream Social Network Scale consists of 23 endorsement type questions designed to assess the structure of social networks of the informants, such as (a) spare time activities, (b) type of social interaction in the neighborhood, and (c) degree of contact with the ethnic and mainstream friends. The participants are asked to rate each statement in terms of how strongly they agree or disagree with, for instance, the following: “There are many Turks in the suburb I live” and “When I have personal problems, I share it with my Turkish friends.” All the questions in this section are rated on a 5-point response format ranging from strongly disagree (1) to strongly agree (5). The section ends with the question “All in all, are you more in contact with the Turkish or with the <Australian> people?” which has a 5-point response scale: only Turkish (1) to only <Australian> (5).

The Ethnic and Mainstream Cultural Norms Scale consists of 19 endorsement type questions designed to assess the attitudes of the informants regarding (a) Turkish/Mainstream norms and values, (b) vitality of respective groups, (c) degree of institutional support for cultural maintenance, and (d) degree of exclusion experienced in the mainstream community. The participants are asked to rate each statement in terms of how strongly they agree or disagree with, for example, the following: “The Turks in <Australia> can act together as a group” and “<Australian> people usually think negatively about the Turks.” All the questions in this section are rated on a 5-point response format ranging from strongly disagree (1) to strongly agree (5).

The Language Use, Choice and Preference Scale consists of 50 questions in 5 subsections: language register spoken with different interlocutors, such as mother, father, and siblings; language register spoken to informant; language use; language preference; and language choice across topics. The participants are asked to respond to language use or choice questions in a bipolar response format, for instance: “In which language do you interact mostly with your mother?” The responses are indicated on a 5-point bipolar response format ranging from always <English> (1) to always Turkish (5). The benefits of using bipolar formats in such research are discussed extensively by Kang (2006).

The Attitudes to Turkish Language Scale consists of 20 questions designed to assess attitudes toward Turkish language in various domains. The participants are asked to respond to each question in terms of how strongly they value Turkish language, for instance, “How important is Turkish to find a job?” and “How important is Turkish to rear children?” All the questions in this section are rated on a 5-point response format ranging from not important (1) to very important (5).

All scales with the exception of the language scales contained a mix of negatively and positively worded items. Item presentation was randomly rotated within scales.

Procedure

Accessing Turkish communities in the four national contexts was achieved with the help of Turkish cultural organizations. In all national contexts, the section on religious identity was found to be highly personal (and sensitive) and a number of informants refused to fill that part in.

All questionnaires were prepared in Turkish and in the national language of the country in which the participants lived. Questionnaires were prepared in both Turkish and the mainstream language of the immigration context given the expected differences in language proficiency of different generations. As expected, most second-generation participants filled in the questionnaires prepared in the mainstream language. Internet technology was also used by offering the questionnaire online. Some informants used online version of the questionnaire in the Netherlands, France, and Australia. Analyses did not reveal significant effects of administration mode on target variables; therefore, data of the two modes were merged.

The language, culture, and identity scales were administered in two ways. In all countries except Germany, informants had the choice to fill in the online version or paper version that was delivered to Turkish organizations. The informants filled in the questionnaires in their own time and returned them directly to the researchers. The questionnaires took approximately 25 minutes to complete. Some of the less educated informants reported difficulties in understanding complex language used in the questions.

Data Analysis

The data analysis comprised four parts. First, we established the internal consistencies and examined the structural equivalence, addressing the question to what extent an instrument measures the same underlying construct(s) in each group. We conducted exploratory factor analyses, followed by target rotations, and the computation of factorial agreement (Tucker’s phi) to assess structural equivalence. Values of Tucker phi of .90 and above are usually taken to point to equivalent instruments (Van de Vijver & Leung, 1997). Second, differences in scale means were tested in a MANOVA with country and education as independent variables. We report values of (partial) η², denoting the proportion of variance accounted for by an effect. Third, we employed path analysis to examine the associations between identity and Turkish language variables; a multigroup analysis was conducted to test the invariance of parameter estimates across samples. Given the large number of scales relative to the number of participants, we used factor analysis to reduce the number of variables in the path model. The path analysis modeled four variables: Turkish identity, mainstream identity, Turkish language preference, and perceived value of Turkish language. Fourth, we used a MANOVA to test for country and education differences in the four variables of the path model.

Results

The results are presented in three parts. Internal consistencies and structural equivalence analyses are addressed first. Structural equivalence involves the question to which extent a scale measures the same construct in each country. The second part examines country and educational differences in scale means in a multivariate analysis of variance (MANOVA). The third part addresses relations among scales using structural equation modeling as well as country and educational differences.

Internal Consistencies and Structural Equivalence

Values of Cronbach’s alpha were computed for each country and for the combined data. Results are presented in Table 2. Most scales showed internal consistencies well above .70. The median alpha across all scales was .86. Furthermore, the internal consistencies did not differ systematically across countries. Somewhat lower internal consistencies were obtained for Multiculturalism (alpha for global sample = .59) and Norms regarding the Mainstream Language and Culture (alpha = .65). As an inspection of the factor loadings did not reveal aberrant values in any country, we decided to retain the scales.

Table 2.

Psychometric Properties of the Scales: Internal Consistencies and Factorial Agreement Indexes

	Internal Consistencies					Factorial Agreement
Scale	AUS	F	G	NL	All	Median	Min.	Max.
Multiculturalism	.58	.49	.58	.73	.59	.93	.91	.98
Feeling Turkish	.85	.86	.86	.85	.86	.99	.99	1.00
Feeling Mainstream	.72	.79	.82	.73	.78	.97	.94	.99
Turkish Identity	.77	.77	.86	.78	.79	.99	.97	1.00
Mainstream Identity	.74	.78	.74	.72	.77	.99	.96	1.00
Turkish Behavior	.82	.89	.90	.89	.87	1.00	1.00	1.00
Mainstream Behavior	.86	.87	.81	.82	.86	.99	.99	1.00
Religious Beliefs	.95	.92	.94	.91	.93	.99	.99	1.00
Turkish Network	.84	.82	.86	.83	.84	.98	.97	1.00
Mainstream Network	.89	.85	.85	.82	.86	1.00	.99	1.00
Norms With Regard to Turkish Language and Culture	.72	.74	.72	.75	.73	.99	.97	.99
Norms With Regard to Mainstream Language and Culture	.70	.67	.62	.59	.65	.95	.93	.97
Language Register Spoken With Different Interlocutors	.88	.87	.90	.88	.88	.99	.98	1.00
Language Register Spoken to Informant	.83	.78	.81	.76	.80	1.00	.99	1.00
Language Use	.93	.90	.91	.86	.91	1.00	.99	1.00
Language Preference	.96	.95	.95	.93	.95	1.00	1.00	1.00
Language Choice Across Topics	.93	.92	.91	.88	.91	1.00	1.00	1.00
Private Value of Turkish	.73	.65	.70	.72	.75	1.00	.99	1.00
Societal Importance of Turkish Language	.95	.84	.83	.91	.90	.98	.97	.99
Attitudes Towards Turkish Language	.87	.82	.83	.86	.85	1.00	.99	1.00

Note. Internal consistencies (Cronbach’s alpha) are reported for each country and for the overall data set (“All”). Factorial agreement indexes represent values of Tucker’s phi. Values of .90 and higher are taken as evidence supporting the equivalence of the scale. Reported values are based on all six pairwise comparisons. AUS = Australia; F = France; G = Germany; NL = the Netherlands; Min. = lowest value of Tucker’s phi across the six comparisons; Max = highest value.

Structural equivalence was examined using the procedure outlined by Van de Vijver and Leung (1997). For each scale, an exploratory factor analysis is conducted for each country. The agreement of factor loadings across countries was evaluated by means of Tucker’s phi, which is an index of factorial agreement; a value of phi was computed for each of the six pairwise country comparisons. The results provided strong support for the equivalence of the scales; no values lower than .90 were observed and the median value was .99. It can be concluded that even the scales with the relatively low internal consistencies measured the same underlying constructs in each country.

Country and Educational Differences in Scale Means

A MANOVA was used to test differences between countries (4 levels: Australia, France, Germany, and the Netherlands) and levels of education (3 levels: up to lower secondary education, completed secondary education, and completed tertiary education); scale scores were the dependent variables. Most main effects and interactions were highly significant, which is not surprising given the relatively large sample size; therefore, we focus more on effect sizes. Table 3 presents the mean scores and effect sizes.

Table 3.

Scale Means (for Each Country and Education) and Effect Sizes (η²) in Multivariate Analysis of Variance With Country and Education as Independent Variables

	Country					Education				Interaction
Scale	AUS	F	G	NL	η²	Low	Med.	High	η²	η²
Multiculturalism	3.21_a	3.33_b	3.13_c	3.07_c	.038***	3.31_a	3.20_b	3.12_c	.016***	.005
Feeling Turkish	3.89_a	4.04_b	4.09_b	3.68_c	.017***	4.34_a	4.04_b	3.65_c	.101***	.010
Feeling Mainstream	2.56_a	2.31_b	2.09_c	2.32_b	.052***	2.17_a	2.28_a	2.44_b	.023***	.012
Turkish Identity	4.05_a	4.32_b	4.04_a	4.16_a	.020***	4.40_a	4.20_b	3.98_c	.046***	.018**
Mainstream Identity	2.78_a	2.30_b	1.83_c	2.55_d	.142***	2.05_a	2.25_b	2.62_c	.052***	.013*
Turkish Behavior	4.14_a,b	4.24_b	4.13_a	4.20_a,b	.004	4.35_a	4.16_b	4.12_b	.019***	.006
Mainstream Behavior	3.16_a	2.78_b	2.63_c	3.04_d	.079***	2.55_a	2.82_b	3.14_c	.084***	.028***
Religious Beliefs	3.45_a	3.99_b	3.79_c	3.68_c	.043***	4.04_a	3.82_b	3.50_c	.056***	.007
Turkish Network	3.87_a	4.09_b	3.94_a	3.89_a	.015**	4.13_a	3.95_b	3.86_c	.017***	.008
Mainstream Network	2.98_a,b	3.01_a	2.73_c	2.89_b	.014**	2.59_a	2.89_b	3.04_c	.052***	.020**
Norms With Regard to Turkish Language and Culture	3.98_a	4.00_a	3.95_a	4.19_b	.011**	4.05_a,b	3.97_b	4.07_a	.002	.013*
Norms With Regard to Mainstream Language and Culture	2.99_a	2.83_b	2.46_c	2.76_b	.084***	2.64_a	2.76_b	2.82_b	.005	.011
Language Register Spoken With Different Interlocutors	3.45_a	3.78_b	4.08_c	4.16_c	.023***	4.31_a	3.88_b	3.95_b	.053***	.030***
Language Register Spoken to Informant¹	3.97_a	3.71_b	4.10_c	4.06_a,c	.023***	4.29_a	3.85_b	3.92_b	.049***	.052***
Language Use¹	3.13_a	3.09_a	3.43_b	3.04_a	.019***	3.77_a	3.14_b	2.96_c	.098***	.031***
Language Preference¹	3.38_a	3.13_b	3.60_c	3.50_a,c	.023***	3.94_a	3.28_b	3.30_b	.067***	.021**
Language Choice across Topics¹	3.23_a	3.13_a,b	3.51_c	3.07_b	.017***	3.86_a	3.25_b	2.98_c	.110***	.040***
Private Value of Turkish	3.27_a	3.20_a	3.22_a	3.90_b	.103***	3.40_a,b	3.30_a	3.48_b	.004	.023***
Societal Importance of Turkish Language	2.28_a	2.40_a	2.39_a	2.65_b	.021***	2.44	2.43	2.41	.002	.005
Attitudes towards Turkish Language	3.29_a	3.28_a	3.68_b	3.61_b	.047***	3.60_a	3.44_b	3.43_b	.007*	.015*

Note. AUS = Australia; F = France; G = Germany; NL = the Netherlands. The three columns with η² as header represent the proportion of variance accounted for by the main effect of country, education, and their interaction, respectively. Means with identical subscripts did not differ significantly in a post hoc test (Least Significant Difference).

Higher scores refer to more frequent use or preference of Turkish.

p < .05.**p < .01. ***p < .001.

The effect size for country differed considerably across the instruments. The only large effect size was found for mainstream identity. The immigrants to Germany reported the lowest scores (M = 1.83), followed by the Turks in France (M = 2.30), the Netherlands (M = 2.55), and Australia (M = 2.78); post hoc tests revealed that all these means differed significantly from each other. Turkish identity showed the same means in Australia, Germany, and the Netherlands, whereas in France the mean was significantly larger. Moderate effect sizes were found for the private value of the Turkish language (with much higher scores for the Turkish immigrants to the Netherlands than for the other groups) and two other mainstream-related scales: Mainstream Behavior and Norms with regard to Mainstream Language and Culture. Both scales showed the lowest scores among the immigrants to Germany and the highest scores among the Turkish-Australian immigrants. Other scales yielded small effect sizes, with the exception of Turkish behavior; like the measure of ethnic identity, this scale revealed a combination of significant though practically negligible cross-cultural differences and high scores. In summary, we found the largest cross-cultural differences between Australia and Germany (with lower adjustment scores for the Turkish-Australians) and larger cross-cultural differences on adjustment scales (i.e., scales dealing with aspects of the mainstream culture) than on maintenance scales. The patterning of the means of Table 3 largely follows the hypothesized order of the first hypothesis, with the exception of the scores of the French group. We hypothesized that the French and the German samples, being exposed to the least favorable climate, would show the strongest ethnic identity and the least usage of the mainstream language. However, the pattern in the French group deviated from this prediction in that the mainstream language is spoken in this group more frequently than we expected. The more frequent usage of French may be a consequence of the French language assimilation policy. We return to this topic in the Discussion section.

Educational differences were on average as large as country differences (mean proportions of variance accounted for were .04 for both main effects). Yet the distributions were slightly different. No large effect sizes were found. Norms with regard to Turkish Language and Culture, Societal Importance of Turkish Language, Private Value of Turkish, Norms with regard to Mainstream Language and Culture, and Attitudes towards Turkish Language showed negligible relations with education. The strongest relations (with medium effect sizes) were found for Language Preference, Mainstream Behavior, Language Use, Feeling Turkish, and Language Choice across Topics. Furthermore, the means of all variables with small or medium effect sizes showed a perfect rank order with education level; higher levels of education were associated with stronger mainstream attitudes and behaviors and weaker Turkish attitudes and behaviors. These findings support the second hypothesis, which predicted a positive relation between educational level and mainstream attitudes and behaviors.

The average effect size of the interaction of country and education was small. Still, it is important to note that the most salient interaction components (explaining between 3% and 5%) were found for the variables dealing with actual language use (Language Register Spoken with Different Interlocutors, Language Use, Language Choice across Topics, and Language Register Spoken to Informant). The interaction was mainly due to the deviant pattern in the French sample; compared to the three other countries, the language behavior scales showed stronger relations with education in France. Turkish-French with lower levels of education talked more Turkish and more educated Turkish-French talked more French than immigrants with similar levels of education in other countries. This finding may (again) be due to the language assimilation policy in France, which has advocated the use of French in both the public and private sphere. This policy, which is stronger in France than in the other countries, has mainly affected the highly educated immigrants.

Path Analysis of Scale Relations

Preliminary analyses

There were a relatively large number of scales (20 scales) relative to the number of participants for each country (about 250 for each country). In keeping with the rule of thumb that for each free parameter there should be at least 10 observations (Kline, 2004), we decided not to include all scale means. Two types of scales were included in our study: identity and language. Both consisted of two types of scales: The identity scales involved mainstream and Turkish identity, whereas the language variables involved preferred usage and perceived value.

Correlational and exploratory factor analyses were used to confirm the expected positive interrelations of scales within the clusters. Correlations of the scale intercorrelations within each of four sets of scales were in the expected direction and significant in all but one case. Exploratory factor analyses confirmed the presence of a relatively strong first factor in each of the four analyses. Turkish identity comprised the following scales: Feeling Turkish, Turkish Identity, Turkish Behavior, Religious Beliefs, and Turkish Network (first factor explained 63% of the variance); mainstream identity was represented in Feeling Mainstream, Mainstream Identity, Mainstream Behavior, and Mainstream Network (66% explained); (Turkish) language usage in Language Register Spoken with Different Interlocutors, Language Register Spoken to Informant, Language Use, and Language Preference and Language Choice across Topics (75% explained); finally, Perceived Value of Turkish is represented in Norms with regard to Turkish Language and Culture, Norms with regard to Mainstream Language and Culture, Perceived Societal Importance of Turkish Language, and Attitudes towards Turkish Language (scores on mainstream norms were reversed) (36% explained). Scores on the four new constructs were computed by averaging the standardized scale scores that make up the construct (the standardization, done for each scale across all participants, was needed to correct for differences in means and standard deviations of the constituting scales).

Path analysis

Given our interest in the associations between identity and language, we chose a path model in which identity variables predict language variables, yet we appreciate that a longitudinal design would be needed to make more conclusive statements about the causal relations and possible feedback loops between the two sets of variables. We tested a model in which the two identity variables were correlated exogenous variables that predict the two language variables; relations between the two language variables are accounted for in the model by their common dependence on identity variables (see Figure 1). Moreover, we were interested in the invariance of the model across the four immigrant groups. A standard procedure for invariance testing was followed, as implemented in AMOS (Arbuckle, 2008). Results of the fit tests are given in Table 4. A test of configural invariance (postulating the same pattern of zero and nonzero paths without any cross-cultural equality constraint on paths or variances) yielded a fair fit (see Table 4). When the path coefficients were held constant across the countries (measurement weights model), the fit test revealed favorable results, Δχ²(12) = 17.21, ns, and ΔCFI = .007. However, imposing additional equality constraints led to a considerable reduction in fit. Therefore, we retained the measurement weights model as the most restrictive model with an adequate fit. The most salient parameter estimates of this model are given in Figure 1. The model holds that the identity and language variables are related in a similar manner across the countries. As could be expected, Turkish identity was positively associated with both Turkish language preference (β = .25) and perceived value of Turkish (β = .38). Mainstream identity was a negative predictor of both preference for Turkish (β = –.31) and perceived value of Turkish (β = –.16). All other parameters of the model differed across countries. The most interesting of these parameters are the correlations between the Turkish and mainstream identity. These correlations were negative and significant (p < .05) in all countries but differed in size. In the Australian, r(283) = –.18, and Dutch, r(271) = –.22, samples, the correlations were weaker than in the French, r(266) = –.44, and German, r(265) = –.52, samples.

Figure 1.

Multiple Group Path Model of Identity and Language Variables

Table 4.

Fit Indices of Path Model

Model	χ² (df)	Δχ² (df)	AGFI	RMSEA	CFI	ΔCFI
Configural invariance	21.70*** (4)		.902	.064	.978
Measurement weights	38.91** (16)	17.21 (12)	.957	.036	.971	−.007
Structural weights	98.63*** (25)	59.72*** (9)	.912	.052	.908	−.063
Structural residuals	172.11*** (31)	73.48*** (6)	.864	.065	.824	−.084

***

p < .001.

Country and educational differences in identity and language variables

A final MANOVA addressed country and educational differences in the four identity and language variables of the path analysis. The means and effect sizes are presented in Table 5. The means in the cells can be interpreted as deviations from the grand mean of zero (as the data were standardized). The effect sizes of the country differences were small (though significant, p < .001) for Turkish identity and Turkish language preference and moderate for mainstream identity and perceived value of Turkish. It is interesting to note that the immigrants to France combined the strongest Turkish identity (of all groups) with the lowest preference for using Turkish. If the combination would be a consequence of the language assimilation policy, it would suggest that the policy leads to more frequent use of the mainstream language but that the generalization of assimilation effects to other domains cannot be assumed. The means of mainstream identity and perceived value of Turkish are in line with this interpretation; the immigrants to France combined a relatively weak mainstream identity with a low perceived value of Turkish. The Australian sample showed a combination of a relatively well developed mainstream identity and relatively low scores on scales dealing with the Turkish identity, language, and culture. The German sample showed the opposite pattern. The means of the Dutch sample were close to zero (suggesting a pattern of average scores), with the exception of its relatively high score on perceived value of Turkish.

Table 5.

Scale Means (for Each Country and Education Level) and Effect Sizes (η²) in Multivariate Analysis of Variance With Country and Education as Independent Variables

	Country					Education				Interaction
Scale	AUS	F	G	NL	η²	Low	Med.	High	η²	η²
Turkish identity	−.17_a	.26_b	.02_c	−.10_a,c	.016***	.47_a	.08_b	−.27_c	.065***	.007
Mainstream identity	.40_a	−.04_b	−.48_c	.12_b	.089***	−.47_a	−.11_b	.31_c	.076***	.023***
Turkish language preference	−.04_a	−.28_b	.25_c	.02_a	.021***	.65_a	−.12_b	−.19_b	.095***	.040***
Perceived value of Turkish	−.34_a	−.27_a	.06_b	.56_c	.071***	.14_a	−.10_b	.03_a	.006*	.023***

Note. AUS = Australia; F = France; G = Germany; NL = the Netherlands. The three columns with η² as header represent the proportion of variance accounted for by the main effect of country, education, and their interaction, respectively. Scores have been standardized for each scale. Means with identical subscripts did not differ significantly in a post hoc test (Least Significant Difference).

p < .05. **p < .01. ***p < .001.

Educational differences were significant for all variables, although the differences for perceived value of Turkish were very small. The other variables showed a medium effect size; the rank order of the means across the educational levels was perfect for these variables. The scores of the participants with medium educational levels were closer to those with high education levels than those with low educational levels, so the group with low education levels stood out by the combination of high scores on Turkish identity and Turkish language preference and low scores on mainstream identity. It can be concluded that we confirmed the findings of the previous MANOVA in that education is positively related to adjustment and negatively to cultural maintenance. The interaction between country and education showed small effect sizes (and was even nonsignificant for Turkish identity). The interactions were mainly due to the stronger relation between education and the dependent variables in the French sample than in the other samples. The general tendency that with higher levels of education, immigrants are more focused on the mainstream and less on the Turkish language and culture is most pronounced in the French sample.

Discussion

We set out to investigate Turkish immigrant acculturation and language orientations in four different countries (Australia, France, Germany, and the Netherlands) to shed light on the relations between identity and language and on the interactions between receiving society orientations and immigrant responses. To start with the former, combining social-psychological and sociolinguistics perspectives provided new insight into the role of language in the acculturation process. In line with our conceptual framework and hypotheses, there was a strong relationship between ethnic identification and the value of both ethnic and mainstream languages. We distinguished between a symbolic language value factor and a behavioral language preference factor. We found that immigrants with a stronger Turkish identity valued and used the Turkish language more. The stronger the identification with the ethnic group, the higher the perceived value of ethnic language. Education turns out to be an important predictor in valuing the mainstream language; more educated informants are more inclined to value mainstream language compared to less educated informants in all national contexts. Mainstream identity works in the opposite direction; immigrants with a stronger mainstream orientation use and value the Turkish language less.

Given the differences in ideological orientations and acculturation climate of the receiving countries, it was expected that Turkish immigrants would show stronger patterns of integration in Australia and, to a lesser extent, the Netherlands, linguistic assimilation in France, and ethnic and cultural separatism in Germany. Our findings support Bourhis et al.’s (1997) ideological clustering model. We indeed found that Turkish immigrants have a high level of identification with the mainstream culture in Australia, while they have much lower identification levels in France, Germany, and the Netherlands. Turkish informants in Australia showed a combination of a relatively well-developed mainstream identity and relatively lower scores on scales dealing with Turkish identity, language, and culture. The German sample showed just the opposite pattern, higher levels of ethnic identification and lower levels of mainstream identification.

Turkish immigrants in France showed highly intriguing patterns. Even though they identified strongly with the Turkish group, they neither perceived the value of Turkish to be high nor had a strong preference for using Turkish in interaction with others. In the Netherlands, informants perceived the value of Turkish to be very high and identified with both the mainstream and the in-group. Less educated Turks show fairly similar patterns of acculturation orientations in all four contexts, whereas better educated Turkish immigrants show varying patterns of sociocultural integration. Educated Turks in Australia show solid patterns of linguistic and cultural integration, whereas the educated Turks in France show strong signs of only linguistic assimilation. Turkish immigrants in the Netherlands and Germany exhibit comparable acculturation orientations. Nevertheless, the distance between the German identity and Turkish identity turned out to be larger than the one between Turkish and Dutch identity. A policy focus on language assimilation is often defended on the basis that language is the key to understanding and participating in a culture. Our study suggests that the policy is effective in France, at least for the better educated groups, in that these groups often use the mainstream language in private situations. However, our study also suggests that language assimilation is not necessarily associated with a stronger mainstream identity. Speaking French is not the same as feeling French in this group.

It may seem counterintuitive that the mainstream identity is strongest and the Turkish identity weakest in the country (Australia) that embraces multiculturalism and that puts least pressure on immigrants. We would contend that such multiculturalism policies may create or confirm a climate in which immigrants do not feel that their ethnic identity is challenged and, therefore, they feel less need to protect it. In countries with a less friendly acculturation climate, ethnic and mainstream identities are more incompatible (as shown by their stronger negative correlations). Multiculturalism is an area in which well-intended but poorly conceived policies may reach the opposite of their original intentions. Pressure to adjust to the mainstream can reach the opposite goal if the policy is not accompanied by mainstreamers who support the assimilation of immigrants by providing them with a sense of belonging to the mainstream group.

It could be argued that the country differences in identity we observed are due to self-selection bias as immigrants with different backgrounds filled out the questionnaires in the three countries or that the countries attract different immigrant groups from Turkey. The most salient difference in background characteristic between the samples in the four countries is level of education. We conducted an ANOVA with country as independent variable and educational level as dependent variable. Country was significant (though with a small effect size). Immigrants in the Netherlands had the highest level of education, followed by Australia, France, and Germany. We cannot confirm or rule out that this pattern is affected by self-selection. However, we controlled for education in our analyses of identity and language and found that country effects were present “over and above” educational effects. So even if there would be self-selection and the immigrants with lower levels of education would be more likely to go to France and Germany than to Australia and the Netherlands, our study indicates that education cannot explain all the differences in identity and language use and preferences. Self-selection in the choice of immigration country may have other consequences. Thus, it may well be that countries that are viewed as offering more opportunities for upward social mobility attract immigrants that are more highly motivated to start a new life and to invest in the new culture. Australia might be viewed more as such a country than the other three because of its long immigration history. We cannot rule out the possibility that this aspect of self-selection has affected the results of our study.

Turkish immigrants were fairly similar in other background characteristics. Therefore, we think that the receiving country’s integration ideology and pluralism climate and Turkish immigrants’ level of interaction with the host society explain the differences in Turkish immigrants’ acculturation orientations. Notably, the fact that Australia has been an immigration country for a longer time might add to higher identification levels of Turkish-Australians with the host society. In the same vein, the differentiation of indigenous versus immigrant is more pronounced in the European context than in Australia, which lowers the group boundaries between the communities. Further evidence for the more liberal multiculturalism climate in the Netherlands (as compared to France and Germany) comes from a recent comparison of European immigration policies (Huddleston & Niessen, 2011).

Turkish and mainstream identity are predictors of important language variables in all groups; the two identities seem to play an opposing role, with the Turkish identity being associated with a stronger focus on Turkish and the mainstream identity associated with a weaker focus. These findings corroborate the third hypothesis. Our data suggest that both identities play an opposite role in language attitudes and behaviors, but it is important to note that the two identities are neither fully dependent nor fully independent. In the acculturation literature, it has become common to argue that adjustment to the mainstream society is not necessarily accompanied by loss of the ethnic culture. Our findings on identity are in line with this view, yet it should be emphasized that we do not find evidence for the sometimes claimed independence of ethnic and mainstream acculturation. The present study provides evidence that these correlations are negative, but vary across cultural contexts. The obvious implication from this finding is that it is counterproductive to just assume (in)dependence of identities; (in)dependence should be a matter of empirical scrutiny.

Our study suggests a possible patterning of these correlations. An inspection of Figure 1 suggests that the correlations between Turkish and mainstream identity are negative in all samples, yet the correlations were stronger in France and Germany than in Australia and the Netherlands. A multigroup model with equal structural weights (the path coefficients) and equal covariances for Australia and the Netherlands on the one hand and for France and Germany on the other hand yielded a good fit, χ²(24, N = 1,037) = 73.17, AGFI = 947, RMSEA = .043, CFI = .939. We propose that the stronger negative correlations of mainstream and ethnic identities can be expected in less inclusive and more assimilation-oriented climates of acculturation. If our results could be cross-validated, it would imply that ethnic and mainstream identity are opposites (and bicultural identities are less likely) in contexts in which there is more pressure to adjust to the mainstream. In this line of reasoning, it would only be in more inclusive societies that the two identities are independent. Although we endorse the often observed need to assume conceptual independence of mainstream and ethnic identity (and to assess both types of identity in acculturation studies), our study suggests that even in the most inclusive society of our study, Australia, the two types of identity show a significant, negative correlation.

The discourse about multiculturalism has changed considerably in all countries of our study. Media coverage of immigration often refers to negative aspects, such as criminality and unemployment among immigrants. The question could therefore be asked whether the validity of the present results would not have a limited time span. We think that these results are not very susceptible to discourse changes. The first and most important reason is that we have found in partly cross-sectional and partly longitudinal studies in the Netherlands that the support for multiculturalism was not affected by 9/11 (in 2001) and two political murders on well-known opponents of multiculturalism in the Netherlands (2004 and 2006) (Breugelmans, Van de Vijver, & Schalk-Soekar, 2009). Attitudes toward multiculturalism depend on deeply seated values like views on discrimination and the tolerance for heterogeneity, which do not easily change. Also, reported cases of racial discrimination have not substantially increased in the last 10 years in the Netherlands (Boog, Dinsbach, Van Donselaar, & Rodrigues, 2010). The second reason is that multiculturalism policies do not change strongly either. For example, a recently completed study of migrant integration policies found much evidence for the stability of multiculturalism policies and maintenance of the rank order of countries on multiculturalism indicators in multiple domains such as education and labor market opportunities (Huddleston & Niessen, 2011).

It is a limitation of the present study that we dealt with self-reports of language use and did not measure actual language use (as far as overt language use was concerned). It would be interesting to see whether the same two factors (of language preference and value) could also be found if there would be more emphasis on actual language behavior. Furthermore, our study was limited to Turkish immigrants. Extensions to other immigrant groups, both Western and non-Western, are still to be established.

On the basis of our analysis of Turkish acculturation orientations in the four national contexts, we can conclude that there is a close connection between the degree of acculturation and integration ideology of the receiving societies. Turkish immigrants in Australia appear to better integrate into the mainstream society compared to their counterparts in the European context. There can be a number of reasons that we cannot account for on the basis of our data; however, the clear difference between Australian and European discourse regarding immigrants and immigration might play a major role in the acculturation process. The public discourse about immigrants and asylum seekers is remarkably negative in Europe (Crowley & Hickman, 2008). As a result, immigrants’ position as outsiders is strengthened and managing migration and promoting social cohesion appear to be a greater challenge for policy makers in most European nation states. In addition, because most European countries have not considered themselves as countries of immigration, coming to terms with social and cultural changes as a result of diversity becomes much harder compared to Australia, which is characterized and accepted as an immigration country. Samers (1998) even claimed that in some European countries ideological construction of nationhood continues to be significant in the social construction of social exclusion of immigrants. Our data clearly show differences in the degree of Turkish immigrants’ sense of belonging in the four immigration contexts. These country differences may also be related to the psychological climate vis-à-vis multiculturalism experienced by immigrants in their everyday lives. Populations in countries with a long history of immigration such as Australia may well be more welcoming for immigrants. It is no surprise that the sense of belonging was highest in the Turkish-Australian group. Our study clearly shows that country and policy characteristics matter for immigrants and for the way they deal with the new country and its population. A faster adjustment to the mainstream society seems to be reached by using a carrot (like accepting immigrants as belonging to the country) than by a stick (like enforcing assimilation).

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Ait Ouarasse

Van de Vijver

F. J. R.

(2004). Structure and function of the perceived acculturation context of young Moroccans in the Netherlands. International Journal of Psychology, 39, 190-204.

Arbuckle

J. L.

(2008). Amos 17.0. Crawfordville, FL: AMOS Development Corporation.

Archibald

(2002). Immigrant integration: The ongoing process of reform in France and Québec. In Baker

S. J.

(Ed.), Language policy: Lessons from global models (pp. 30-58). Monterey, CA: Monterey Institute.

Arends-Tóth

(2003). Psychological acculturation of Turkish migrants in the Netherlands: Issues in theory and assessment. Unpublished PhD dissertation, Tilburg University.

Arends-Tóth

Van de Vijver

F. J. R.

(2003). Multiculturalism and acculturation: Views of Dutch and Turkish-Dutch. European Journal of Social Psychology, 33, 249-266.

Ashmore

R. D.

Deaux

McLaughlin-Volpe

(2004). An organizing framework for collective identity: Articulation and significance of multidimensionality. Psychological Bulletin, 130, 80-114.

Ataca

Berry

J. W.

(2002). Psychological, sociocultural, and marital adaptation of Turkish immigrant couples in Canada. International Journal of Psychology, 37, 13-26.

Barker

Giles

Noels

Duck

Hecht

Clément

(2001). The English-only movement: A communication analysis of changing perceptions of language vitality. Journal of Communication, 51, 3-37.

Bauböck

Heller

Zolberg

(Eds.). (1996). The challenge of diversity. Integration and pluralism in societies of immigration. Aldershot, UK: Avesbury.

10.

Berry

J. W.

(1997). Immigration, acculturation and adaptation. Applied Psychology: An International Review, 46, 5-68.

11.

Berry

J. W.

Kalin

(1995). Multicultural and ethnic attitudes in Canada: An overview of the 1991 national survey. Canadian Journal of Behavioural Science, 27, 301-320.

12.

Boog

Dinsbach

Van Donselaar

Rodrigues

P. R.

(2010). Monitor Rassendiscriminatie 2009 [Monitor race discrimination 2009]. Rotterdam, the Netherlands: Landelijk Expertisecentrum van Art.1.

13.

Bourdieu

(1982). Les rites d’institution [The institutional rites]. Actes de la Recherche en Sciences Socials, 43, 58-63.

14.

Bourhis

R. Y.

Moïse

L. C.

Perreault

Senécal

(1997). Towards an interactive acculturation model: A social psychological approach. International Journal of Psychology, 32, 369-386.

15.

Brands

Crone

Leurdijk

Top

(1998). Media en migranten [Media and migrants]. Amsterdam, the Netherlands: University of Amsterdam.

16.

Breugelmans

S. M.

Van

Vijver

F. J. R.

Schalk-Soekar

S. G. S.

(2009). Stability of majority attitudes toward multiculturalism in the Netherlands between 1999 and 2007. Applied Psychology: An International Review, 58, 653-671.

17.

Clément

Noels

K. A.

(1992). Towards a situated approach to ethnolinguistic identity: The effects of status on individuals and groups. Journal of Language and Social Psychology, 11, 203-232.

18.

Clément

Noels

K. A.

Deneault

(2001). Interethnic contact, identity, and psychological adjustment: The mediating and moderating roles of communication. Journal of Social Issues, 57, 559-577.

19.

Clyne

(1991). Community languages: The Australian experience. Cambridge, UK: Cambridge University Press.

20.

Clyne

(1992). Linguistic and sociolinguistic aspects of language contact, maintenance and loss. Towards a multifacet theory. In Fase

Jaspaert

Kroon

(Eds.), Maintenance and loss of minority languages (pp. 17-36). Amsterdam, the Netherlands / Philadelphia, PA: John Benjamins.

21.

Crowley

Hickman

M. J.

(2008). Migration, postindustrialism and the globalised nation state: Social capital and social cohesion re-examined. Ethnic and Racial Studies, 28, 1-23.

22.

Dandy

Pe-Pua

(2010). Attitudes to multiculturalism, immigration and cultural diversity: Comparison of dominant and non-dominant groups in three Australian states. International Journal of Intercultural Relations, 34, 34-46.

23.

Edwards

(1992). Sociopolitical aspects of language maintenance and loss. In Fase

Jaspaert

Kroon

(Eds.), Maintenance and loss of minority languages (pp. 37-54). Amsterdam, the Netherlands / Philadelphia, PA: John Benjamins.

24.

Extra

Verhoeven

(Eds.). (1993). Immigrant languages in Europe. Clevedon, UK: Multilingual Matters.

25.

Extra

Yağmur

(Eds.). (2004). Urban multilingualism in Europe. Immigrant minority languages at home and school. Clevedon, UK: Multilingual Matters.

26.

Extra

Yağmur

Van der Avoird

(2004). Methodological considerations. In Extra

Yağmur

(Eds.), Urban multilingualism in Europe: Immigrant minority languages at home and school (pp. 109-132). Clevedon, UK: Multilingual Matters.

27.

Fuligni

A. J.

Kiang

Witkow

M. R.

Baldelomar

(2008). Stability and change in ethnic labeling among adolescents from immigrant families across the high school years. Child Development, 79, 944-956.

28.

Giles

Bourhis

R. Y.

Taylor

D. M.

(1977). Towards a theory of language in ethnic group relations. In Giles

(Ed.), Language, ethnicity, and intergroup relations (pp. 307-344). London: Academic Press.

29.

Gonzo

Saltarelli

(1983). Pidginization and linguistic change in emigrant languages. In Andersen

R. W.

(Ed.), Pidginization and creolization as language acquisition (pp. 181-197). Rowley, MA: Newbury House.

30.

Gordon

M. M.

(1964). Assimilation in American life. New York: Oxford University Press.

31.

Huddleston

Niessen

(2011). Migrant Integration Policy Index III. Brussels, Belgium: British Council and Migration Policy Group.

32.

Hyltenstam

Stroud

(1996). Language maintenance. In Goeble

Nelde

Stary

Wölck

(Eds.), Contact linguistics (pp. 567-578). Berlin, Germany: Walter de Gruyter.

33.

Janmaat

J. G.

(2006). Popular conceptions of nationhood in old and new European member states: Partial support for the ethnic-civic framework. Ethnic and Racial Studies, 29, 50-78.

34.

Jasinskaja-Lahti

(2008). Long-term immigrant adaptation: Eight-year follow-up study among immigrants from Russia and Estonia living in Finland. International Journal of Psychology, 43, 6-18.

35.

Kang

(2006). Measurement of acculturation, scale formats, and language competence: Their implications for adjustment. Journal of Cross-Cultural Psychology, 37, 669-693.

36.

Kipp

Clyne

Pauwels

(1995). Immigration and Australia’s language resources. Canberra, Australia: AGPS.

37.

Kline

R. B.

(2004). Principles and practice of structural equation modeling (2nd ed.). New York: Guilford Publications.

38.

Koenig

(1999). Social conditions for the implementation of linguistic human rights through multicultural policies: The case of the Kyrgyz Republic. Current Issues in Language & Society, 6(1), 57-84.

39.

Lapeyronnie

(Ed.). (1990), L’intégration des minorités immigrées [The integration of immigrant minorities]. Paris, France: Adri.

40.

Laroussi

(2003). Les langues des populations issues de l’immigration maghrébine dans l’agglomération rouennaise: Maintien ou déperdition? [The languages of the Maghreb immigration in the city of Rouen: Maintenance or loss?] In Caitucoli

(Ed.), Situations d’hétérogénéité linguistique en milieu scolaire (pp. 11-41). Rouen, France: Collection Dyalang, Presses Universitaires de Rouen.

41.

Masgoret

Ward

(2006). Culture learning approach to acculturation. New York: Cambridge University Press.

42.

Phalet

Kosic

(2006). Acculturation in European societies. In Sam

D. L.

Berry

J. W.

(Eds.), The Cambridge handbook of acculturation psychology (pp. 331-348). Cambridge, UK: Cambridge University Press.

43.

Phinney

J. S.

(1990). Ethnic identity in adolescents and adults: Review of research. Psychological Bulletin, 108, 499-514.

44.

Phinney

J. S.

Horenczyk

Liebkind

Vedder

(2001). Ethnic identity, immigration, and wellbeing: An interactional perspective. Journal of Social Issues, 57, 493-510.

45.

Rollan

Sourou

(2006). Les migrants Turcs de France, Entre repli et ouverture [The Turkish migrants in France, between closure and opening]. Pessac, France: Maison des Sciences de l’Homme d’Aquitaine.

46.

Sam

D. L.

Berry

J. W.

(Eds.). (2006). Cambridge handbook of acculturation psychology. Cambridge, UK: Cambridge University Press.

47.

Samers

(1998). Immigration, ‘ethnic minorities’, and ‘social exclusion’ in the European Union: A critical perspective. Geoforum, 29(2), 123-144.

48.

Smolicz

J. J.

(1981). Core values and cultural identity. Ethnic and Racial Studies, 4, 75-90.

49.

Tribalat

(1995). Faire France. Une enquête sur les immigrés et leurs enfants [Making France. A study on immigrants and their children]. Paris, France: La Découverte (“Essais”).

50.

Van de Vijver

F. J. R.

Leung

(1997). Methods and data analysis for cross-cultural research. Newbury Park, CA: SAGE.

51.

Van Oudenhoven

J. P.

Prins

K. S.

Buunk

B. P.

(1998). Attitudes of minority and majority members towards adaptation of immigrants. European Journal of Social Psychology, 28, 995-1013.

52.

Van Oudenhoven

J. P.

Ward

Masgoret

A.- M.

(2006). Patterns of relations between immigrants and host societies. International Journal of Intercultural Relations, 30, 637-651.

53.

Vedder

P. H.

Horenczyk

(2006). Acculturation and the school. In Sam

D. L.

Berry

J. W.

(Eds.), The Cambridge handbook of acculturation psychology (pp. 419-438). Cambridge, UK: Cambridge University Press.

54.

Verkuyten

(2007). Religious group identification and inter-religious relations: A study among Turkish-Dutch Muslims. Group Processes & Intergroup Relations, 10, 341-357.

55.

Virta

Sam

D. L.

Westin

(2004). Adolescents with Turkish background in Norway and Sweden: A comparative study of their psychological adaptation. Scandinavian Journal of Psychology, 45, 15-25.

56.

Ward

Kennedy

(1993). Psychological and socio-cultural adjustment during cross-cultural transitions: A comparison of secondary students overseas and at home. International Journal of Psychology, 28, 129-147.

57.

Waters

M. C.

Jiménez

T. R.

(2005). Assessing immigrant assimilation: New empirical and theoretical challenges. Annual Review of Sociology, 31, 105-125.

58.

Yağmur

(1997). First language attrition among Turkish speakers in Sydney. Tilburg, the Netherlands: Tilburg University Press.

59.

Yağmur

(2004). Language maintenance patterns of Turkish immigrant communities in Australia and Western Europe: The impact of majority attitudes on ethnolinguistic vitality perceptions. International Journal of the Sociology of Language, 165, 121-142.

60.

Yağmur

(2009). Language use and ethnolinguistic vitality of Turkish compared with the Dutch in the Netherlands. Journal of Multilingual & Multicultural Development, 30, 219-233.

61.

Yağmur

Akinci

M.- A.

(2003). Language use, choice, maintenance and ethnolinguistic vitality of Turkish speakers in France: Intergenerational differences. International Journal of the Sociology of Language, 164, 107-128.

62.

Yağmur

De Bot

Korzilius

(1999) Language attrition, language shift and ethnolinguistic vitality of Turkish in Australia. Journal of Multilingual and Multicultural Development, 20, 51-69.