Union Formation and Partner Choice in a Transnational Context: The Case of Descendants of Turkish Immigrants in France1 Nadja Milewski Institute for Sociology and Demography, University of Rostock Christelle Hamel French National Institute for Demographic Studies (INED) This paper examines the transition to a first union of descendants of Turkish immigrants in France. We use data from the project The Integration of the European Second Generation, 2007, and apply eventhistory techniques. We find that descendants of Turkish immigrants who grew up in France enter a first union earlier and more often in a direct marriage than do young adults without an immigrant background. We then describe the type of union in more detail and estimate the likelihood of a transnational partner choice, that is, between a young adult born in France of Turkish immigrant parentage and an immigrant from Turkey. We pay attention to social factors such as education, city of residence, and to cultural factors such as the rules of affinity in Turkey and the attachment to the norm of virginity at marriage as factors that orient partner choice. Finally, we discuss what anthropological methods could contribute to this research. 1 The TIES project is coordinated by Maurice Crul and Jens Schneider from the Institute for Migration and Ethnic Studies (IMES) at the University of Amsterdam. The French part of the TIES project was supervised by Patrick Simon and Christelle Hamel at the National Institute for Demographic Studies (INED). The French fieldwork was funded by the Volkswagen-Stiftung, INED, and the city hall of Paris. We are grateful for the discussions with the participants of the TIES project and in particular to Helga de Valk, Patrick Simon, Arnaud Bringe, and Stephanie Condon as well as for the valuable language editing by Catriona Dutreuilh. The first author’s work in the TIES project has been funded within a Marie-Curie Research Training Network of the European Commission. This paper was first presented at the 2nd Workshop of the EAPS Working Group on The Anthropological Demography of Europe ‘‘Beyond national boundaries: Migration and transnationality in Europe’’, Paris, December 2007. 2010 by the Center for Migration Studies of New York. All rights reserved. DOI: 10.1111/j.1747-7379.2010.00820.x IMR Volume 44 Number 3 (Fall 2010):615–658 615 616 International Migration Review INTRODUCTION Large-scale immigration has been changing the face of Western Europe and its demographic trends in recent decades. The immigrant populations and the migration flows are themselves changing. Whereas the immigration of workers coming mainly from former colonies and from Southern European countries dominated from the 1950s until the beginning of the 1970s, family-related migration developed from the 1970s, as well as refugee and asylum-seeking migration from 1980. Immigrants contribute to make Western European countries culturally more diverse, and immigrants’ descendants born in the countries of settlement constitute a growing share of the young adults living in the metropolitan areas today. Descendants of immigrants have to manage the differences between their parents’ culture of origin and the culture of the society in which they have been raised. In this multi-cultural context, union-formation behaviors can be seen as a highly appropriate topic of research for the study of the way immigrants’ children combine the norms and values of two socio-cultural areas. Yet, little is known about the matrimonial behaviors of the women and men of immigrant parentage in Western Europe. Our paper aims at expanding this knowledge by focusing on the transition to a first union of young adults of Turkish descent living in France. We concentrate on descendants of Turkish immigrants because they form the most dispersed immigrant group living in today’s Europe and because most of the studies find considerable differences between Turkish immigrants and the respective host populations concerning family-formation trends and cultural maintenance. Moreover, France ranks second as destination country of Turkish emigrants after Germany (Manço, 2005; De Tapia, 2006). It has, however, not been studied in detail yet as researchers working on immigrants with quantitative data generally concentrate on the bigger immigrant groups living in France. The perspective of research we follow in this paper focuses on the combination of qualitative and quantitative analysis, and on cultural patterns as an explanation for practices, in line with the anthropological demography perspective (Bernardi and Hutter, 2007; Johnson-Hanks, 2007). We actually think that this combination of qualitative and quantitative methods is particularly relevant to understanding matrimonial practices of descendants of immigrants who live in a multi-cultural Union Formation and Partner Choice in a Transnational Context 617 and a multi-national context. Union formation is a classic topic of demographic research that especially focuses on the age at first union and on the type of union (marriage or cohabitation) and, for anthropologists, that pays attention to the rules of affinity in countries where the choice of partner is regulated by norms other than individual choice. For the study of matrimonial practices in a migration context, the anthropological demography approach is particularly relevant and useful. This is the reason why we will concentrate on these three dimensions of matrimonial behaviors: the timing and type of first union (marriage or cohabitation) as well as the partner’s origin with respect to his ⁄ her migration background. The structure of the paper is as follows: We start by sketching the history of Turkish immigration to France and by giving an overview of the current patterns of union formation in both France and Turkey. We then review the scientific literature regarding union formation and partner choice and summarize research that has been carried out on the union formation of young adults of Turkish origin living in other Western European destination countries, as well as of other young adults of immigrant origin living in France. Following this, we develop our research hypotheses, by focusing on the respective contributions of quantitative and qualitative research on these same topics. Our study applies to a new and original dataset that describes the living conditions of young people of Turkish descent in France union. The survey is part of an international project called The Integration of the European Second Generation (TIES). It is the first time that a quantitative survey has been carried out on the Turkish-descent population in France. BACKGROUND OF THE RESEARCH The Turkish Population in France Around 3.6 million Turkish citizens live abroad, of which 3.2 million live in Europe (De Tapia, 2006). Turks form the most dispersed immigrant group in Western European countries. There are 4,959,000 first-generation immigrants in total in France, representing 8 percent of the total population in 2008. The main immigrant groups in France are Algerians, Moroccans, Portuguese, Italians, and Turks. The number of Turkish 618 International Migration Review immigrants is 222,000 (Borrel, 2006; INSEE, 2008b). Turkish immigrants started settling in France at the end of the 1960s, but this immigration flow increased in the 1980s because of political disturbances in Turkey. Turkish immigrants in France are on average less educated than they are, for example, in Germany, as the West German recruitment strategies for so-called guest workers included a selection of foreign workers based on their levels of qualification, which was only the case in a later period in France (De Tapia, 2006; Insel et al., 2003). Thus, immigrants from Turkey to France are more likely to come from rural areas, especially from central Anatolia. Immigration flows from Turkey are currently increasing faster than flows from Algeria and Morocco. Among first-generation immigrants from Turkey, 27 percent have French nationality, compared to 40 percent of the entire immigrant population living in France (INSEE, 2008b). This low percentage may be partly because of the fact that the Turkish immigration is relatively recent. Indeed, 62 percent of Turkish immigrants are below age 40 (INSEE, 2008b), so their children are only now reaching adulthood. Hence, union formation is becoming a concern in their life and an important issue for both themselves and their family. Union Formation in France and Turkey What do we know about practices of union formation in France and in Turkey, and what are the similarities and differences between them? As in the case of Turkish immigrants coming to Western Europe, two distinct family-culture patterns are brought into contact as the family structures vary considerably between these two regions and nuptiality behaviors in the two countries are very different. Among these differences are the universality of marriage, a relatively young age at leaving the parental home, and a very low prevalence of non-marital cohabitation in Turkey (Reher, 2004; Sunar and Fisek, 2005). Nevertheless, some slow changes are under way: according to the 2003 Turkish Demographic and Health Survey, the median age at first marriage was 21 for women aged 25–29 in 2003, while it was 19.2 for women aged 45–49 the same year, which corresponds to a 2-year increase in 20 years (HUIPS, 2004:91). This relatively young age at union formation is correlated with the low mean level of educational attainment. By contrast, France – like other European countries – has experienced substantial changes in union formation practices during recent Union Formation and Partner Choice in a Transnational Context 619 decades that contrast sharply with the Turkish matrimonial pattern. Demographers have identified three major trends there. The first is the delay in the transition to adulthood, as defined in terms of leaving the parental home, completing education, and being financially independent. The second is the increase in the ages at first union and at first birth. The third major change is the remarkable decline of marriage. These three phenomena are strongly interrelated. Until the beginning of the 1970s, young people left the parental home when they got married and started to cohabit with their spouse. At that time, fewer than one out of five couples started to cohabit without being married (Prioux, 2003). The massive increase in the number of children attending secondary school and following higher education led to an increase in the age at leaving parental home, getting a job, and forming a union, while marriage started to be seen as an overly conservative institution. Among the cohorts born between 1944 and 1955, half of the individuals had already formed their first union at age 24.6 years while the cohort born in 1971 had formed their first union at age of 26 (Prioux, 2003). In 1996, nine couples out of ten started living together without being married. Moreover, a large proportion of couples are unlikely to get married at all (Toulemon, 1996). Using the French Generations and Gender Survey, Sebille confirms these trends and indicates that the postponement of the transition to adulthood is still an ongoing process as some young French adults start their life in a partnership by forming long-term stable couples without cohabitation (Sebille, 2009). She also notes that although the age at leaving the parental home is quite stable, the reasons for leaving home have nonetheless changed radically, as young adults no longer leave their parents home to get married but to study. They start their adult life by a long period of living alone, with financial support from their parents, and possibly being in a non-cohabiting couple. Descendants of immigrants face a challenging situation, in particular when it comes to union formation, as the choice of a partner and of living arrangement are fields where culture, values, and norms dominate the scene. Portes and Zhou (1993:75) emphasize that ‘‘Growing up in an immigrant family has always been difficult, as individuals are torn by conflicting social and cultural demands, while they face the challenge of entry into an unfamiliar and frequently hostile world.’’ The bigger the cultural differences between the immigrants and the majority population, the more acute the conflicts are liable to be. Women and men of immigrant parentage have to find their pathway to adulthood by choosing from or between 620 International Migration Review elements of two cultural heritages under societal circumstances that may be in contradiction with cultural traditions. Foner (1997) calls this challenge a ‘‘bicultural adjustment’’. Given these differences between France and Turkey, studying the patterns of first union formation of immigrants’ descendants is particularly pertinent and ties in closely with the perspectives and purposes of anthropological demography. In order to put our research questions we start with a review of the scientific literature on union formation and partner choice in the Turkish population in Turkey and in Europe, paying attention to the respective contributions of both quantitative and qualitative methods. Turkish Immigrants’ Union-Formation Practices in Emigration Countries The only available dataset on careers and life conditions of immigrant descendants in France that allows the study of matrimonial behavior is the ‘‘Mobilité géographique et insertion sociale’’ survey [Geographical Mobility and Social Integration Survey] conducted in 1992.2 Descendants of Turkish immigrants in France were not included there as they had not reached the respective ages by that time. As a result, our knowledge from quantitative surveys on this population is rather limited and does not concern matrimonial behavior but rather education and employment (Simon, 2003). For this literature review, we therefore look mainly at data regarding Turkish immigrants’ children in other European countries as well as data regarding other young adults of immigrant parentage living in France. We organized this review of the scientific literature by focusing on the main topics explored by the quantitative surveys: first the origin of the partner, second the timing of the first union, third the type of union (marriage or cohabitation). Partner’s Origin and Migration Background. The international literature has concentrated primarily on the homogamy ⁄ heterogamy of couples with respect to their nationality or their migration background, and religious 2 A new national survey on Immigrants and descendants of immigrants, titled Trajectoires et Origines: enquête sur la diversité des populations de France, coordinated by French National Institute for Demographic Studies and by French National Statistical Institute was conducted in 2008. Union Formation and Partner Choice in a Transnational Context 621 affiliations. The choice of partner is influenced by three major domains, namely individual preferences, the norms for interactions of the social groups to which the (potential) partners belong, and the structure of the marriage market (Kalmijn, 1998). With respect to national origin or migration background, the patterns of mixed marriages vary between national-origin groups, destination countries, and period. Evidence for national homogamy between spouses has been found both in actual behavior as well as in studies on preferences. With a rate of more than 80 percent, the Turkish population abroad shows particularly high rates of homogamy as compared to immigrants of other origins. This result applies to both the first and the second generations. In the countries where data are available on Turkish-origin children born and raised in Western Europe, it appears that the rate of transnational marriages with a Turkish immigrant is lower than for the first generation but higher than for other young adults of immigrant parentage (Tribalat, 1996; for France; Roloff, 1998; Milewski, 2003; for Germany; Lievens, 1999; and Timmermann, 2008, for Belgium; de Valk, 2008, for The Netherlands; Lucassen and Laarman, 2009). The endogamous partner choice of first-generation migrants with respect to the country of origin appears somewhat self-explanatory as, for example, marriage arrangements may have been made prior to the migration of either of the partners, and the structure of the marriage market for the first immigration waves was usually characterized by a very imbalanced sex ratio (cf. Lievens, 1999). In addition, labor migrants may have perceived their stay abroad as temporary. Therefore, a spouse from the same country of origin may appear more reasonable. But even for those immigrants who arrived single in France, the rate of mixed marriages remains low. In 1992, the ‘‘Mobilité géographique et insertion sociale’’ survey has shown that among immigrants originating from Turkey who were single at their arrival in France and over 15 years old, only 19 percent of men and 10 percent of women married a person born in France. These percentages were even lower for immigrants who came before age 15 to France (12 and 5 percent, respectively; Tribalat, 1996). By contrast, their descendants may neither face an imbalanced marriage market nor insecurity regarding their stay in the country. Furthermore, a language barrier does not exist for them. Hence, three main possibilities in partner choice for descendants of immigrants emerge: a person whose parents were both born in the country of destination (a ‘‘native’’), a person who moved himself ⁄ herself to this country as a first- 622 International Migration Review generation immigrant, and another second-generation person (a person born in France whose parents originate from Turkey or who moved with the parents as a child; of course, a person from a third country can also be a potential partner; the literature shows, however, that this kind of couple is exceptional (cf. Milewski, 2003; Safi, 2008). International findings as well as studies on immigrants and their children in France are still quite rare, but show different results for different immigrant groups at different destinations. Whereas descendants of Algerian immigrants in France3 and of Moroccan immigrants in Belgium have an almost even distribution of the three union types (but with strong gender differences as mixed marriages are less frequent for women than for men; Tribalat, 1996; Lievens, 1999), descendants of Turkish immigrants living in Germany, Belgium, and the Netherlands appear to prefer a spouse from Turkey (Lievens, 1999; González-Ferrer, 2006; Timmermann, 2006; de Valk, 2008; Lucassen and Laarman, 2009). In general, both quantitative as well as qualitative studies suggest that this specific behavior can be attributed to the Turkish family ties. It appears that chain migration may play a role here, as well as the structure of the marriage market at destination: the smaller this population is, the bigger may be the trend to choose a Turkish immigrant as a spouse. The high proportion of homogenous couples may also be related to the religious affiliation of the immigrants and their descendants as most of them are of Muslim faith and may have a preference for a partner of the same religion. Mixed-marriages may, however, more often concern Turkish-origin men than women – as this is a common pattern in immigration contexts for both the first and the second generations (e.g., Kalmijn, 1998; Lievens, 1999; Milewski, 2003; Filhon and Varo, 2005; de Valk, 2008) – as the Muslim doctrine does not allow the marriages of a Muslim woman to a non-Muslim man whereas the opposite is possible. Lievens (1999) finds in Belgium that low-educated Turkish-origin men born in Belgium or moved at very young ages and higher educated women are more likely to have an immigrant spouse. For the latter group, however, he sees a different motivation for the immigration of a spouse: 3 The rate of mixed marriages of young adults of Algerian parentage with a partner of French parentage is high: among young men, 50 percent live with a partner whose two parents were born in France and this rate is 24 percent for women (Tribalat, 1995:77). By contrast, the rate of marriages with an immigrant is 36 percent for women and only 17 percent for men. Union Formation and Partner Choice in a Transnational Context 623 He assumes that higher educated women are more assimilated to the western life styles (here: Belgium) and may therefore want to escape the initial marriage phase of living with the husband’s family (as is common in Turkey) and therefore choose a partner from Turkey. Achieving ‘‘modern’’ goals by ‘‘traditional’’ means is the term for this behavior. Evidence for this has been found mainly in qualitative studies (in France, e.g. Autan, 1995), but in a similar quantitative analysis for Turks in Germany González-Ferrer (2006) did not find evidence for this hypothesis. In our study, we examine whether Turkish immigrants’ descendants living in France are more likely to follow the demographic patterns observed for other Turkish second generations in other destination countries or rather the patterns of descendants of other immigrant groups living in France by looking at the unions formed by Turkish descendants from France and first-generation immigrants from Turkey. We hypothesize that Turkey has a dominant influence as a marriage market for the second generation living in France. Timing of the First Union. Although the age at the first union is a classic demographic indicator for the study of transition to adulthood and for family formation, this indicator is not widely used when it comes to children of immigrant parentage. The focus on the origin of the partner as an indicator of integration seems to make the other aspects of union formation life fade into the background. For Germany, Milewski (2003) has shown that the on average younger age of immigrants from Turkey coincides both with a low frequency of pre-marital cohabitations and lower education, as compared to Germans. Yet, as we already indicated, the formation of the first union takes place relatively early in Turkey and coincides in general with marriage, while, by contrast, young adults in France form their first union much later than in Turkey. This suggests that this indicator might be a key information source, indicating both the values that the individuals uphold and their living conditions. For instance, it has been shown concerning descendants of Algerian immigrants living in France, that their entry into a first union is later than that of French-origin women and men. This unexpected postponement is partly because of high levels of unemployment. The combination of a lower level of education and exposure to racial discrimination reduces their access to the labor market and therefore postpones couple formation. This also indicates that Algerians’ descendants follow the norm whereby a household must not be founded before at least the man (if not both 624 International Migration Review partners) has obtained a stable employment situation.4 What is the situation for descendants of Turkish immigrants living in France? Type of First Union. Marriage is one of the core elements of the strong family patterns in Turkey. Only about 2 percent of all Turkish women never marry. Marriage is both strongly interrelated with leaving the parental home and with the first childbirth (Ergöcmen and Eryurt, 2004). Consequently, almost all Turkish immigrant women of the first generation living in Western European countries (BMFSFJ, 2000, for Germany) are married. The high rate of married women in the Turkish migrant populations can, however, not only be traced back to marriage behaviors in Turkey, but must also be related to the conditions under which international migration is possible. West Germany, France, The Netherlands, and other Western European countries restricted immigration from non-EU (EG) member states after 1973 to family reunion. A family-related move to any of these countries from Turkey is therefore only possible for persons who are married to someone who holds a residence permit or is a citizen of the respective destination country. At the same time, a transnational marriage strengthens the ties between Turkish emigrant communities and Turkey (e.g., Münz, Seifert, and Ulrich, 1997, for Germany; Régnard, 2006, for France; Timmermann, 2006, 2008, for Belgium). For descendants of Turkish immigrants in France we assume that a marriage is of less importance in terms of legal conditions. This aspect may, however, gain importance when the partner comes from Turkey. In these cases, a high likelihood of marriages (instead of non-marital unions) can be expected. Consequently, pre-marital cohabitation cannot be an option for these couples. For women and men whose partner is of French origin or has at least spent more time already in France, we assume that the frequency of cohabitation is higher, but still much lower than for French-born people as the marriage rate in the French population is in general very low, and we are looking at the first union in our paper. Studies on descendants of Turkish immigrants in other destination countries show that these young adults would like to cohabit first and marry later, 4 In 1992, among young adults of Algerian parentage, aged 25–29, only 29 percent of men and 55 percent of women lived with a partner, compared to 38 percent of men and 65 percent of women of French parentage (Lefranc and Thave, 1995). Union Formation and Partner Choice in a Transnational Context 625 too, as for example is the case in The Netherlands (de Valk, 2006, 2009). Studies on actual behavior show, however, that a direct marriage was still the very dominant pattern of union formation of Turkish immigrants and their descendants, at least in the 1990s (Lievens, 1999). Therefore, we assume that the cultural heritage as brought from Turkey gives marriage as the prevailing norm for union formation, especially in cases of a transnational partner choice. Therefore, we work with the competing hypothesis that we will still find a very large share of direct marriages among first unions of Turkish descendants in France. Matrimonial Practices in the Turkish Community As we have seen, quantitative surveys are a useful tool to shed light on some standard demographic measures of union-formation practices, such as the timing of first union, and to measure the prevalence of cohabitation or marriage, and mixed marriages. But these data generally say little about the social context in which these practices occur, or about the meaning of these practices for individuals. Here, we present some qualitative data that highlights the matrimonial practices in the Turkish population both in Turkey and in Europe. We find it desirable that scientists take into account knowledge gained from qualitative surveys, especially in the phase of drafting the questionnaire. This practice is getting more common today, and the review of qualitative knowledge also constituted the background that oriented the design of the TIES survey regarding partnership histories. Our study illustrates which insights from qualitative research have entered into the quantitative TIES-research design and how this information can be used in the interpretation of the findings. We concentrate here on elements of partner choice which refer to values and cultural practices. The Rules of Affinity in Turkey and in the Emigration Context. Historically, Turkish society was organized into tribes, also called segmentary societies by anthropologists. Paul Stirling (1965) made a very dense and rich ethnographic description of the Turkish example of this type of organization in his book titled Turkish Villages. The segmentary system in Turkey is based on both matrilineal and patrilineal lineages, and is a heritage of the central Asian nomadic society and of the Muslim religion. This segmentary organization of society is widespread from Asia to North Africa. In Turkey, its political influence is now restricted to the level of 626 International Migration Review villages, as the centralized state was set in place in the beginning of the 20th century (Gökalp, 1987). In segmentary societies, partner choice is determined by belonging to a lineage. The rules of affinity make the marriage of a man with the daughter of his maternal uncle (nomadic heritage) or with the daughter of his paternal uncle (Islamic heritage) a preferential marriage (Gökalp, 1989, 2006; Ilcan, 1994). Given that there is a political, economic, and social hierarchy of lineages, these marriages in the parentage are of crucial importance: they are seen as the most prestigious marriages, as well as the most homogeneous. Gökalp (1994) quotes a report written by Timur (1972) from a demographic survey conducted in 1968 among 4,500 families, which shows 36 percent of respondents claimed to be married to a ‘‘close relative’’. Using the 2003 Turkish Demographic and Health survey, Koc (2008) indicates that out of the 8,075 ever-married interviewed women, consanguineous marriages accounted for 22 percent of the total, which indicates that despite a sharp decrease, preferential marriage remains an active practice. This bears witness to the enduring existence of segmentary organization, at least in the family sphere. Agricultural reforms, industrialization, and the development of the tertiary sector weakened the segmentary system and therefore the pattern of the marriage between cousins: these new sectors of activity now provide more employment than agriculture and most of the population lives in cities (Vergin, 1985). These changes are precisely what caused both rural exodus in Turkey and emigration towards European countries. In other words, the parents of the Turkish second generation in France originate from a rural society that was facing crucial transformations and a risk of disappearance when they emigrated and they married in accordance to the rules of affinity of the segmentary society. This led Tribalat in 1992 to include a question on the family ties between partners in the questionnaire of the Geographical Mobility and Social Insertion Survey. In 1992, marriages with a cousin or another person of the extended family represented 27 percent of men’s marriages and 23 percent of women’s marriages among Turkish first-generation immigrants living in France (Tribalat, 1996:81). Considering qualitative data, Gökalp (1994) suggests these changes in the Turkish society lead to a reinforcement of preferential marriage practices in Turkey. Given the importance of this kind of marriage in Turkish society, the question of whether unions in the kinship also exist in the population of the Turkish second generation is especially legitimate. This led the Union Formation and Partner Choice in a Transnational Context 627 designers of the TIES survey to include a question on the family ties between the spouses in the questionnaire. Intergenerational and Gender Relations and Partner Choice. As a consequence of both the preferential marriage and the young age at marriage in Turkey, marriage arranged by the parents is a widespread practice. Thus, Turkish immigrants who live in France belong to a generation who were married by their parents in a context where individual choice of the partner was not an option. According to the Geographical Mobility and Social Insertion Survey, the rate of arranged marriages by parents among Turkish immigrants who were already married when they arrived in France was 51 percent for women (46 percent with their consent and 6 percent without their consent) and 38 percent for men (36 percent with their consent and 3 percent without their consent). These results show that parents’ intervention in the partner choice was a common practice for this generation raised in Turkey. Conversely, in France, the partner choice is supposed to be based on a romantic attraction, free of any parental pressure. Sociologists have shown that today, parents implement educational strategies to ensure that their children choose a partner of the same social rank, resulting in strong social homogamy within couples (Bozon and Héran, 2006). The definition of ‘‘a good marriage’’ (we would rather say today ‘‘of a good partner’’ as marriage has disappeared) is interiorized, and parents no longer directly choose a partner for their children who thus have the illusion of free choice. Even if the social determinants that lead to the choice of a partner invalidate the idea that ‘‘free choice’’ actually exists, it is obvious that the choice of the partner is now individual. In the migration context, parents may not share the same point of view on partner choice as their children born and socialized in a European country. Qualitative research emphasizes that parental influence on partner choice is of high importance in Turkish immigrant families. Conducting qualitative research on partner choice among descendants of Turkish immigrants in Germany, Straßburger (2003) defines several forms of marriage arrangements by parents that range from informal introduction of potential candidates to very binding negotiations. In a qualitative survey conducted in the beginning of the 1990s with Turkish families living in a village in the center of France, Autan (1995) examined matrimonial practices and noted that both young female and male Turkish immigrants who arrived at a young age in France (before 14 years old) conform to 628 International Migration Review their parents’ expectations by marrying their cousin or at least someone who originates from the same region of origin in Turkey. Note that Autan’s work deals with young people who were born in Turkey and who married a short time after their parents settled in France. Is her observation still valid 15 years later concerning the young people born and raised in France and at least 20 years after the parents’ arrival on the French territory? To what extent do they arrange the marriage of their children or try to influence their marital choice? To further our knowledge in this area, some specific questions not generally included in quantitative surveys on integration of immigrants’ descendents, were included in the TIES questionnaire. They allow a better understanding of parents’ role in their children’s partner choice. Those questions concern the place where the partner was first met and the parents’ role in the respondent’s decision to marry. Sexuality: A Key Issue in Transnational Marriages. In segmentary societies, extramarital sexuality, especially pre-marital sexuality, constitutes an offense to the family honor. The threat of shame and dishonor leads to strict control over both men’s and women’s sexuality. The feeling of shame that results from extramarital sexuality is a social sanction and the means by which women are socially controlled (Peristiany, 1966; Ozyegin, 2009). Given that marriage is almost universal in Turkey, sexuality outside marriage is still looked on as a violation of morality. Nevertheless, since the 1980s, the rapid economic and political changes, as well as the European integration process, have challenged the cultural values relating to gender roles and sexuality. In 1999, on the initiative of the Turkish feminist movement, a decree made it illegal for officials to request a virginity test without the consent of the women (Ozyegin, 2009). Today, both young men and women are calling for a sexual liberalization, considering that notions of virginity and sexual honor are anachronistic, but according to Ozyegin’s qualitative research, these young adults have to face their parents’ disapproval of premarital sexuality. Up to the end of the 1960s, in France, the events that marked the transition to the adult sexuality (in a stable couple), that is, leaving the parental home, getting married, and experiencing first sexual intercourse, occurred at the same time (Girard, 1974). A quantitative survey on sexual behaviors of French people has shown that this pattern is now restricted to a small share of the population: only about 7 percent of men and 11 Union Formation and Partner Choice in a Transnational Context 629 percent of women aged 18–24 in 1992 had their first intercourse on or after their wedding day (Bajos and Spira, 1993). In the migration context, the difference between the two countries inevitably creates a situation where individuals are submitted to contradictory injunctions. Turkish parents and their descendants may not be able to ignore that the norm of female virginity has become old-fashioned and is now even stigmatized in France. Consequently, young adults of Turkish origin have to face and resist the pressure of other young adults of their age while at the same time considering their parents’ opinion and submitting to their authority. We believe that sexuality is a key issue for examining how the values governing matrimonial practices evolve in the migration context. This led the French team of the TIES survey to include questions on sexuality in the questionnaire,5 both on attitudes and on practices that address the issue of the norm of virginity. It is probably the first time that such a question has been included in a quantitative survey. The TIES Survey – Overview of Sampling, Data, Method, and Indicators Data on Descendants of Turkish Migrants in the TIES Survey. The TIES is an international survey whose aim is to examine how the receiving countries integrate descendants of immigrants and to evaluate their ‘‘integration models’’. For this purpose, it was necessary to conduct the survey among a group that is widespread in Europe in order to control the effect of the origin. Therefore people of Turkish origin were chosen, and a standardized questionnaire was used in each country. The aim was to achieve a representative sample of the Turkish second generation in seven countries. The target sample size was 500 descendants of Turkish immigrants, who were born in the country of destination, and 500 descendants of the native population in each country, aged 18–35. Furthermore, second-generation persons of Moroccan and ex-Yugoslavian origin were included in a few countries. The fieldwork was first conducted in The Netherlands in 2006. France is the second country to implement the survey in the field. The interviews were 5 The question on the age at the first intercourse was included only in the French questionnaire of the TIES survey, but not in the other countries participating in the TIES project. 630 International Migration Review conducted there in the second half of 2007 (TIES, 2008). (The other countries included in TIES are: Germany, Belgium, Austria, Switzerland, Sweden, and Spain.) What is an immigrant? What is a descendant of an immigrant? In this paper, we use the definitions given by the French High Council for Integration, a government body that gathers scientific information on immigrants in France. Its definition is now used by the French National Statistical Institute and by the French demographers and statisticians (INSEE, 2008a) and was used to design the sample. An immigrant is a person who did not have French nationality at birth and whose country of birth is not France. Someone who currently has French nationality may be an immigrant if s ⁄ he was born abroad and did not have French nationality at birth. Hence, all foreigners are immigrants, but not all immigrants are foreigners. Someone who was born abroad and had the French nationality at birth is not considered as an immigrant. In French statistical analysis it is generally considered that a descendant of an immigrant is a person who was born in France and who has one or two parents born abroad. A descendant of two immigrants has a foreign nationality at birth if neither parent is naturalized, but this person generally becomes French at age 18. If one of the two parents is naturalized, the child has two nationalities at birth. Grand-children of immigrants are invisible in French statistics as they are systematically French citizens at birth. The TIES survey is designed to be conducted in two cities in each of the countries involved, as immigrant families are concentrated in urban areas. The aim of this urban sampling is also to examine the impact of the city context on respondents’ life courses. In France, the two cities are Paris, the capital, and Strasbourg (close to the German border). In the French case, Turkish families still form a small group and their children born in France constitute a tiny proportion of the whole population. Turkish immigrants are not present everywhere on the French territory. They live in three areas: Paris and its suburbs which is home to 64,000 of them, the North-East of France (Alsace) close to the German border where their number is about 28,000, and the South-East (Rhône-Alpes) where they total 37,000. Although they are much more numerous in the South-East than in the Alsace, they represent a much larger proportion of the immigrant population in Alsace than in the South-East (INSEE, 2008b). That is the reason why Paris and Strasbourg, the biggest city in the Alsace, were chosen to conduct the survey. Union Formation and Partner Choice in a Transnational Context 631 The Impact of the French Sampling Method on the Measurement of Mixed Couples. In France, the census collects individual information on the country of birth, which makes it possible to identify immigrants, but not their descendants because they were born in France (Simon, 2003). Moreover, there was no other sample frame available. For example, in the Family History Survey (survey Enquête Histoire familiale, 380,000 respondents, conducted in 1999) in which the parents’ country of birth is recorded, there were only 344 descendants of Turkish immigrants. Considering that they were dispersed throughout France and that their age range was much wider than that of our sample, this survey could not be used as a sampling frame for the TIES purpose. To get around this obstacle, an original sampling method was designed, that has some consequences on the counting of ‘‘mixed couples’’. This method worked in two steps. First, a German institute (Humpert and Schneiderheinze), that is specialized in the onomastic method (the recognition of the affiliation of a patronym or a surname to a language) made a selection of Turkish-sounding names in the phone book in order to build a reduced sample frame, compared to the whole phone book that would have been too large. Some 10,000 subscribers were selected, that is, 10,000 households for which there was a high probability of finding someone who was born in Turkey or who was of Turkish origin. Then, a screening questionnaire was administered in order to record the socio-demographic characteristics of each member of the household (age, sex, country of birth, and parents’ country of birth, level of education), in order to select potential respondents, that is, those aged 18–35 born in France and with one or both parents born in Turkey. Given that in most of the cases the respondent who answered the screening questionnaire was not the one selected for the survey, the interviewer had to call a second time in order to present the survey to the potential survey respondent and to get an appointment to conduct the face-to-face interview. At the time of the first call, the interviewer frequently noted down the selected person’s mobile phone number when s ⁄ he had one, to get in touch with him ⁄ her directly. All the interviewers were Turkishspeakers: this linguistic proficiency enabled them to talk easily to the first respondent when s ⁄ he was not fluent in French, which was sometimes the case when the first respondent was the mother of the selected person. The method of the Turkish-sounding-name selection contains two biases. Children of Turkish-migrant women who are married to a man who is not of Turkish origin do not appear in the sample frame as they 632 International Migration Review have acquired their husband’s last name and therefore do not have a Turkish name any more. Female descendants of Turkish immigrants who are married to a man whose last name does not sound Turkish do not appear in the sample frame either, which theoretically increases the proportion of women who are married to either a Turkish immigrant or to a descendant of Turkish immigrants. This raises the question of the TIES data reliability for exploring union formation patterns. Several arguments lead us to believe that these biases have a low impact. First, in 1999, only about 5 percent of Turkish immigrant women were married to a man who was born in France because most of the women in this group where already married before they migrated (Borrel and Tavan, 2003; INSEE, 2005:61). Second the proportion of children of these mixed couples amounts to less than 4 percent of the total number of children of Turkish immigrants (this figure has been calculated from the Family History Survey, 1999. This survey, however, does not provide information on the country of birth of the partners’ parents). Although these persons are effectively excluded from our sample, the representativeness of the sample is not greatly affected. We may expect the second bias to be more problematic, as the rate of ‘‘mixed marriages’’ is generally higher among descendents of immigrants than among immigrants. But in the Family History survey, 80 percent of female descendants of Turkish immigrants are married to Turkish immigrants, and 20 percent of them are married to a man born in France, but whose father may have been born in Turkey. In these cases, the women are not lost from the sample frame. The number of women whose parents are Turkish immigrants and who are married to a man born in France of French parents is under-represented in the TIES sample, but we know that they represent fewer than 20 percent of married women of Turkish immigrant parentage in France. In order to reduce this bias, we designed the following procedure: The respondents living in the household of each Turkish-sounding-name person were asked if there was a Turkish immigrant in the household who had a daughter who had left home and married a man whose name was not Turkish, that is, who had a husband whose father was not born in Turkey. In this case, the young woman was added to the sample. Unfortunately, this technique does not identify women (married to a man whose name does not sound Turkish), whose parents live in an area other than Paris or Strasbourg. Young women of Turkish immigrant parentage who are married to a man whose father was not born in Turkey finally represent about 8 percent of the Union Formation and Partner Choice in a Transnational Context 633 sample of married women in the TIES survey. We cannot say whether they are under- or over-represented, but we can affirm that this figure is consistent with the proportion of young men married to a women of French parentage. Indeed, we know that the rate of transnational couples is generally higher for men than for women. For instance, in 1993 the Geographical Mobility and Social Insertion Survey showed that 15 percent of female descendants of Algerian immigrants aged 20–29 and 24 percent of men of the same background and same age were married to a spouse whose parents were both born in France (Tribalat, 1995:77). In the TIES survey, the rate of men who are married to a woman whose two parents were not born in Turkey is 25 percent. The bias has been partly corrected, even though we do not know whether the proportion of young women married to a man whose father was not born in Turkey (about 8 percent) should be closer to or farther from 25 percent. (Table A1 in the Appendix summarizes the different steps of the screening procedure.) Among the 10,550 Turkish-sounding names selected, 3,633 were numbers of companies or were no longer available and 294 other numbers were continuously engaged. Therefore, only 6,917 calls led to a contact with someone. Some of the respondents immediately said that no one in the household was in the target group either because there was no one in the age bracket or because the family was not of Turkish origin, which both may correspond to the reality or may have been a polite way to refuse to participate in the survey. Finally, 37 percent of these first respondents agreed to answer to the screening questionnaire, that is, to describe their household, but 268 of them gave up before the end, considering that the questionnaire was too long or that the questions were too personal. Out of the 2,231 fully completed screening questionnaires, 50 percent corresponded to households in which (or through which) at least one person in the target group could be contacted, that is 1,628 potential respondents. Only one person was chosen per household, that is, 1,114 individuals. The women no longer living at their parents’ home, who were married to a man whose father was not born in Turkey, had priority for the selection. As most of the respondents to the screening questionnaire were not the selected person, a second call was made to contact them directly. Out of the 1,114 selected persons, 88 percent were successfully reached. The rate of refusal of the face-to-face interview was 49 percent. The description of the sampling procedure shows how complex it was to identify young adults of immigrant parentage in France. Further- 634 International Migration Review more, the absence of data on the total population of Turkish immigrants’ descendants prevents us from calculating weights to extrapolate our results to this population as a whole. ANALYSIS The whole sample consists of 500 descendants of Turkish immigrants (282 women and 218 men) and 351 young adults of French parentage (188 women and 163 men). Our analysis proceeds in three steps. We first estimate the transition to a first union of young adults of Turkish descent and of French descent. Then, we look at the type of union and describe further characteristics of the first unions of young Turkish and French descendants. Thirdly, we concentrate on transnational unions of the Turkish second generation, describe them and give some determinants. Transition to a First Union Method. We analyze the transition to a first union for the whole sample, separately for women and men. In this part, a union is defined as the formation of a shared household, without differentiating between married couples and non-marital cohabitations. We apply event-history techniques: The first statistical tool used to describe the patterns is Kaplan-Meier survival estimates. These calculations give an estimate of the share of women who enter a union, and of how quickly they do so. A crucial indicator here is the median age. Therefore, this tool can be used even when not all women of the study population have formed a union yet. Second, piecewise-constant intensity regression models are estimated as a form of indirect standardization, as suggested by Hoem (1987) (cf. Hoem, 1993; Blossfeld and Rohwer, 1995). Monthly information on age of the respondent and age at union formation can be used. If monthly information is not available because of missing answers, we impute such events to have occurred in June (this applies to one case regarding the month of birth and to 77 cases concerning the month of union formation). The model can be formalized as follows: ln li ðt Þ ¼ yðt Þ þ Rl al x il þ Rm bm w im ðt Þ; where li(t) denotes the hazard of a first union for individual i at process time t and y(t) represents the baseline log-hazard. The process time for Union Formation and Partner Choice in a Transnational Context 635 the transition to a first union is the time as the respondents turned age 15. The end of the respective process time is either at the formation of the first household or at the interview when no union is reported (censoring). The term wim(t) represents the effect of a time-varying variable (educational enrolment). The term xil denotes the effect of a time-constant covariate (such as birth cohort, religious affiliation, place of residence). Covariates. The covariates we use for the first part of the analysis – transition to a first union – are the following ones: Place of residence: The cities sampled are Paris and Strasbourg with an almost even share of the respondents of the both groups in the two cities. We assume that the transition to a first union is faster in Strasbourg than in Paris as the ‘‘traditionality’’ of the Alsace region is higher than in the capital. Birth cohort: This variable has the following categories: 1971–1975, 1976–1980, 1981–1985, and 1986+. The sample of the descendants of Turkish immigrants is relatively young: Only about 28 percent of them were born before 1981. This compares to 50 percent in the French-origin sample, that is, every second respondent was at least 27 years old at the interview. We assume that younger cohorts enter a union later than older ones, thus following more the Western European patterns. Education: We use the highest level of education attained. For those who are still in education, this variable gives the highest degree of schooling or university education that the respondent has completed before his or her current education. We group the categories as follows: ‘‘vocational higher secondary level’’ (which corresponds to the following French diploma: Certificat d’aptitude professionnelle, Brevet d’études professionnels, Baccalauréat professionnels), ‘‘academic higher secondary level’’ (Baccalauréat general ou technique), and ‘‘tertiary education’’ (university and vocational diploma in higher education level: diplômes de l’enseignement supérieur). We assume that higher educated persons enter a first union later than lower educated ones. Religion during childhood: The survey captures information related to religion at different points in time, that is, during childhood and at the time of the interview. We use the question ‘‘Were you raised in a particular religion?’’ We call this covariate ‘‘religion during childhood’’. Almost 80 percent of the Turkish immigrants’ descendants were raised in the 636 International Migration Review Sunni Islamic tradition, whereas almost half of the respondents of French parentage were raised in the Catholic Church. Few respondents reported being raised in other Muslim school (such as Shia or Alevi) or in the Protestant or Orthodox Churches. As the share of these cases is each very small, we use the categories ‘‘Christian’’, ‘‘Muslim’’, and ‘‘Jewish, other, none’’. We assume that persons who were raised in a religion are more likely to follow the pattern of early entry into a union in the form of marriage. Number of siblings: We can also include information on the respondent’s family of origin. We use here the number of siblings. We also include the educational background of the parents. For mothers and fathers, we built a dichotomous variable for each that refers to higher education. It distinguishes between primary ⁄ lower secondary education and educational attainment that is higher secondary or tertiary. We assume that immigrant descendants with higher educated parents are more likely to follow the French than the Turkish demographic patterns. Table A2 in the Appendix gives an overview about the sample. Results. As our sample is very young, fewer than 50 percent of our respondents had experienced a first union at the survey time. The proportions were about 33 percent in the group of Turkish origin (115 women and 52 men) and about 47 percent in the group of French ancestry (96 women and 68 men). The transition rates to a first union are higher among the women in the Turkish-origin group. 50 percent of the women have entered a union before age 24, whereas the median age is about 25 years in the French origin group. By age 30, only 11 percent of women of Turkish ancestry have not formed a union. This compares to 23 percent of women of French ancestry who had not formed a union by that age. (The differences between the groups are significant on the 5 percent level; see Figure I.) Men enter a first union about 2 years later than women. For men, the transition to a first union is not significantly different for the youths of Turkish descent and those of French ancestry. The median ages are 26 and 27 years, respectively. The share of men not having experienced a first union at age 30 are 35 and 37 percent, respectively (see Figure I). Turning to the intensity-regression analysis for women, the first step is the comparison between the group of Turkish origin and the group of French parentage, controlling for the age of the women (Model 1 in 637 Union Formation and Partner Choice in a Transnational Context Share of persons without union Figure I. Transition to a First Union 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 15 20 25 30 Age in years Turkish-descent w omen French-descent w omen Turkish-descent men French-descent men Table 1). We find significantly higher first-union risks (+35 percent) for Turkish-descent women than for French-descent women. Model 2 introduces birth cohort and place of residence. As expected, younger cohorts are less likely to enter a first union than older ones, and this is the case in both origin groups. Interestingly, the place of residence does not play an explanatory role for the French-descent women, but it does for the Turkish descendants. Women in Strasbourg have a 75 percent higher chance to enter a first union than those in Paris (results of separate modeling are not shown here). These two variables do not, however, explain the differences between the Turkish- and French-origin groups. Model 3 adds the educational attainment, and this variable explains the group differences. As expected, women who have an academic secondary education or tertiary education enter a first union later than those who have a vocational education. Adding this variable to the model removes the differences between the groups, thus indicating some compositional differences with a higher share of descendants of Turkish immigrants having completed only vocational education. Note that we have used the completed education. We may thus underestimate the share of the persons who were still in education by the time of the interview. We worked in a preliminary analysis also with time-varying indicators of educational enrolment and educational 638 International Migration Review TRANSITION TO A TABLE 1 FIRST UNION OF WOMEN, RELATIVE RISKS Variable Model 1 Model 2 Model 3 Model 4 French ancestry Turkish ancestry Birth cohort 1971–1975 1976–1980 1981–1985 1986+ Place of residence Paris Strasbourg Educational attainment Vocational secondary level Academic secondary level Tertiary education Religion during childhood None (or other) Christian Muslim Number of siblings 0 1 or 2 3+ Father with higher education No Yes Age <20 20–25 25–30 30–35 1 1.4* 1 1.5** 1 1.2 1 0.9 1 1.3 0.6* 0.2** 1 1.2 0.7** 0.2*** 1 1.2 0.7** 0.3*** 1 1.4** 1 1.3* 1 1.3 1 0.5** 0.5*** 1 0.5** 0.5*** 1 1.0 1.3 1 0.90 0.90 1 0.8 0.002*** 0.009*** 0.011*** 0.007*** 0.002*** 0.008*** 0.008*** 0.006*** 0.003*** 0.016*** 0.004*** 0.006*** 0.004*** 0.019*** 0.019*** 0.014*** Source: Calculations based on TIES France 2007. Note: Significance: * = 5%; ** = 1%; *** = 0.1%. track currently attained. They showed, however, similar results for the diploma, and union-formation risks were naturally very low in times of enrolment for both groups. Therefore, we decided to work with the simple education indicator. So far, the results indicate that the differences between the two groups regarding the transition to a first union are due to differences in the social characteristics (mainly the level of education) rather than the Turkish cultural heritage of the migrant descendants. Model 4 adds three further covariates that refer to the socio-cultural background of the respondents. They have a significant impact on the transition to a first union only if the education of the respondent is not considered. Controlling for education, these variables lose their impact. Women with more Union Formation and Partner Choice in a Transnational Context 639 siblings, and women with a relatively low educated father have higher union-transition risks than women with less than two or no siblings and a father with higher education. The mother’s education does not matter at all, and the religion during childhood does not add much to the model as there is a high correlation with the ethnic-group variable. For men, the analyses were done using the same covariates. Their effect on the first-union risks was similar to that for women, and no significant differences appear between the two groups of origin (therefore, results not displayed here). Type of First Union Now, we have a closer look at the type of first union. A total of 167 descendants of Turkish immigrants have entered a first union and 164 persons of French ancestry. When we distinguish by union type, a striking difference occurs: In the Turkish-descent group, about 80 percent of the first unions are marriages, whereas only 2 percent of the first unions are marriages in the French-origin group (see Table 2). For descendants of Turkish immigrants, the most frequent union type is a direct marriage. About 70 percent of the Turkish-descent women married and formed a household when they married or within 3 months of marriage. This compares to an about 56 percent of Turkish-descent men. There, the share of direct marriages is somewhat smaller, but still the dominant union formation type. By contrast, direct marriage in first union formation does not exist in the sample of respondents of French parentage. The reasons for leaving the parental home tie in with these findings. More than 80 percent of Turkish-descent women left the parental home for union formation. About 10 percent had lived on their own after leaving the parental home. Though men of Turkish descent are less likely to leave the parental home for union formation than women, with about 56 percent, union formation is the most frequent option nevertheless. By contrast, about four-fifths of the French-descent respondents left the parental home in order to live on their own. Table 2 shows also the meeting place of the partners. Family networks and vacation in Turkey are of highest importance in meeting the future partner for Turkish-descent women and men, whereas for Frenchdescent interviewees friends, school, and the public sphere are more important for contacts. We must stress that the proportion of partners 640 International Migration Review CHARACTERISTICS OF THE TABLE 2 FIRST UNION OF TURKISH-DESCENT WOMEN AND MEN (%) AND FRENCH-DESCENT Women Men Turkish- French- Turkish- FrenchDescent Descent Descent Descent Union type Direct marriage Marriage with pre-marital cohabitation Cohabitation Married, but no information on pre-marital cohabitation Meeting place of partners School Work Friends Association Public place, dancing, vacation elsewhere Parents introduced us Through someone in parents’ network Family celebration Vacation in Turkey Other Missing values Reason for leaving parental home Marriage Cohabitation Independent living Education Work Parents returned Problems with parents Still with parents Missing values Origin of partner + partner’s age at immigration Born in France, French parentage Born in France, Turkish parentage Born in Turkey, immigrated aged under 18 Born in Turkey, immigrated aged 18+ Born in Turkey, immigration age not known Born in France, immigrant parentage (other origin) Born in another country, immigrated aged under 18 Born in another country, immigrated aged 18+ Born in another country, immigration age not known Missing values N 71 10 11 8 0 0 98 2 56 21 21 2 0 1 97 1 7 4 12 3 6 2 4 13 28 7 13 15 5 28 2 9 0 1 0 NA 3 36 8 4 25 0 8 6 10 6 15 2 17 15 7 24 3 16 1 1 1 NA 1 29 76 7 3 7 0 1 0 5 1 7 26 7 41 11 NA 5 1 0 37 19 6 10 10 0 2 17 0 0 18 13 43 24 NA 1 1 0 10 9 13 56 9 3 0 0 1 1 115 75 NA NA NA NA 7 8 3 5 1 96 25 17 12 31 7 6 0 0 1 0 52 78 NA NA NA NA 3 13 4 0 1 68 Source: Calculations based on TIES France 2007. Note: NA–not applicable. introduced by parents is quite rare: only 2 and 6 percent for men and women respectively, which indicates that arranged marriages directly by parents is a practice that is tending to disappear in the migration context. Instead, family celebrations, the parents’ network of friends, and vacation Union Formation and Partner Choice in a Transnational Context 641 in Turkey are of bigger importance for the Turkish second generation. This may indicate still a large influence of the extended family.6 Concerning the origin of the partner, we find that the majority of the partners of the Turkish-descent respondents are of Turkish origin or are immigrants from Turkey (81 percent in total). Again, differences appear between women and men, with women being more likely to have a Turkish immigrant or immigrant descendant as a partner than men (87 and 67 percent, respectively). For French-descent interviewees, we find that about 20 percent of the partners are not of French origin. Characteristics and Determinants of a Transnational Union of Turkish Immigrants’ Descendants Method. The unions of Turkish immigrants’ descendants can be classified into two main categories: transnational unions and other unions. We define a transnational union as a union between a person of Turkish descent who was born and raised in France (i.e., our respondents from the TIES survey) and a person from Turkey who immigrated to France at age 18 or older. The category of other unions includes both partners born in France and of Turkish origin and partners who are of French origin, that is, who were born in France without immigration background. We focus on the determinants of entering a transnational union versus any other union type by applying a logistic-regression model. The model can be formalized as follows: ln Xðx i Þ ¼ b0 þ bj x i ; where X(xi) represents the odds ratio of an individual i for having a transnational union and bj represents the impact of a covariate j on the odds. 6 Note that this variable has relatively large shares of missing values. This is caused by the survey design as this information was only asked for the current partner or, in the case of single persons, for the previous partner, but not for the first partner. As about half of the French respondents were no longer in their first union at the time of the survey interview, the information for the first partner is not available. 642 International Migration Review Covariates. The covariates we used for the first part of the analysis – transition to a first union – are the same ones that we use in the analysis on the transition to a first union. In addition, we constructed several indicators that are specific to the Turkish-descent group such as school attended in Turkey and language raised in. These indicators did not, however, show much variation in the sub-sample and could therefore not be included in the analysis. Two indicators, that we use in addition, are ethnic networks during school years and the fathers’ home region in Turkey (we only use the information on the father as it correlates strongly with the mother’s origin). For the networks, we used the question on the national background of the three best friends in secondary school. Results. In order to estimate the likelihood of having a transnational union we proceeded in two steps. First we ran the models separately for women and men and then in a joint model. We decided to present the only joint model here as almost none of the covariates had a significant impact on men’s partner choice and their sample size is very small anyway. Women are much more likely (see Table 3) to have a partner from Turkey than men. For women, we find that those living in Strasbourg are much more likely to have a transnational partner than women living in Paris. For both sexes, there is no clear trend by birth cohort (see Table 3, Model 1). What does matter – again – is the education of the respondent. The higher the educational attainment, the lower the likelihood of a transnational partner choice. This variable also explains the differences between persons from Paris and Strasbourg, indicating compositional differences especially for the women in these two cities. Social networks in school and the father’s region of origin in Turkey only played a role in bivariate tests. When controlling for the place of residence, they do not contribute to explaining the patterns. This is related to the fact that residents of Strasbourg are more likely to have a network that consists of more Turkish than other friends. For the fathers’ region of origin, we did not find any impact. In bivariate statistics, persons with a father from the Aegean region or the Black Sea were more likely to have a transnational union, but this effect vanishes when controlling for place of residence and education. How can we explain the city differences? The characteristics of the immigrants living in Strasbourg are different from those who live in Paris. As Paris is the capital, it is obviously more open to diversity and to new behaviors. Its population is also much more educated than in the rest of the country, a factor which has a strong impact on norms and representa- Union Formation and Partner Choice in a Transnational Context DETERMINANTS OF HAVING A 643 TABLE 3 TRANSNATIONAL UNION OF TURKISH DESCENDANTS (ODDS RATIOS) Variable Sex Men Women Place of residence Paris Strasbourg Birth cohort 1971–1980 1981+ Educational attainment Vocational secondary level Academic secondary level Tertiary education Number of siblings 0–2 3+ Religion during childhood Muslim Other or none Friends during secondary school 0–1 Turkish 2–3 Turkish Father’s region of origin in Turkey Aegean or Black Sea Anatolia Other region Model 1 Model 2 1 2.8** 1 3.0 1 2.0* 1 1.2 1 0.8 1 0.8 1 0.4* 0.1*** 1 1.5 1 0.8 1 1.1 1.4 0.6 1 Source: Calculations based on TIES France 2007. Note: Significance: * = 5%; ** = 1%; *** = 0.1%. tions, especially regarding sexuality, gender, and matrimonial behaviors. Paris appears to be more attractive for educated people, and this is true for the French population and for immigrants too. On the contrary, Alsace is one of the most traditional areas in France, because of the strong influence of religion. Several surveys on values of the French population have shown that the stronger the religious beliefs, the lower the permissiveness regarding sexuality (e.g., acceptance of pre-marital sexuality, of abortion, and of homosexuality; Lambert, 2002). Alsace is indeed the only area in France where the principle of ‘‘laı̈cité’’ (separation between the state and the Church) is not applied: for instance, there are religion courses in the public schools, which are forbidden everywhere else. This particularity is a heritage of Alsace’s former annexation to Germany, a country where citizens finance Church maintenance. This specificity has an impact on the behaviors of both immigrants and the French population. Therefore, a comparison of the Turkish religious associations in both Paris and Strasbourg and of their influence on the Turkish community 644 International Migration Review would probably help us to understand the differences. An in-depth study about the impact of religion courses at school on representations of sexuality and marital life could shed light on transnational marriage, as these marriages are linked to more conservative conceptions of sexuality. The last step in our analysis is a description of the characteristics of the various union types of Turkish descendants. We carried out chisquared tests using indicators that are frequently mentioned in the literature as playing a role in cultural maintenance within Turkish communities. We asked if these indicators are more frequent in transnational unions than in other union types (see Table 4). We find the importance of virginity confirmed in both actual behavior and attitudes. In transnational unions and in other ‘‘Turkish’’ unions (i.e., a union with a partner of Turkish origin raised in France) more than half of the respondents did not have premarital sex. The share of persons without premarital sex is lower in the category of other Turkish unions and the lowest share is in mixed unions where the partner is of French or other origin.7 Pre-marital cohabitation occurs in only 12 percent of the transnational unions, which can probably be traced back to legal conditions. We should note, however, that pre-marital cohabitation is even less frequent in other Turkish unions and that also in other unions the share is less than a third. This indicates that there is still a norm of marriage within the Turkish community. Unions between relatives (that is, mainly between cousins) are more frequent in transnational unions than in other Turkish unions. And finally, two questions refer to the involvement of parents in the union formation. First, it was ascertained by the following question: ‘‘Has pressure ever been placed on you by your family or your family-in-law to renounce your marriage?’’ The responses show that almost 30 percent of parents put pressure on their children to not get married when the partner was not of Turkish origin. Second, the question was asked: ‘‘Has pressure ever been placed on you by your family or your family-in-law to encourage your marriage?’’ Here, the frequency of parents’ involvement was with about 35 percent the highest in transnational unions. This stresses that the parents’ influence in union formation is still relatively high, although 7 The French questionnaire includes a question on the age at the first sexual intercourse. As it is not a run-of-mill query, the interviewers should not read the question. They had to show it to the interviewee who had to enter his or her response in the computer without giving the response loudly. This technique was implemented in order to preserve the respondents’ intimacy. This question was put at the very end of the questionnaire. Union Formation and Partner Choice in a Transnational Context CHARACTERISTICS OF THE UNIONS Transnational Union TABLE 4 TURKISH-DESCENT WOMEN OF Other Turkish Union Pre-marital sex*** Yes 41 49 No 59 51 Married before age 20*** Yes 36 17 No 64 83 Premarital cohabitation*** Yes 12 6 No 88 67 Missing values 0 27 Partners are relatives*** Yes 25 6 No 75 94 Parents put pressure to renounce marriage*** Yes 14 20 No 86 65 Missing values 0 15 Parents encouraged marriage*** Yes 35 15 No 65 70 Missing values 0 15 Attitudes: ‘‘Pre-marital sex for women is not acceptable’’*** ‘‘I agree’’ 49 48 ‘‘I disagree’’ 51 52 Attitudes: ‘‘Pre-marital sex for men is not acceptable’’*** ‘‘I agree’’ 33 28 ‘‘I disagree’’ 67 72 N 80 54 AND 645 MEN (%) Mixed Union with French or Other Partner Total 91 9 53 47 0 100 23 77 28 16 56 66 34 0 100 14 86 28 16 57 19 65 16 16 28 56 25 60 16 91 9 41 59 3 97 33 25 75 167 Source: Calculations based on Ties France 2007. Note: Significance: *** = 0.1% (chi-squared tests). the parents may not directly introduce the future partners in the first place (as shown in Table 2). Another question asked if the respondent was married against her ⁄ his will. About 5 percent of all Turkish-descent persons in couple said that they actually did not want to get married to their partner (not displayed in the tables). SYNTHESIS AND NEW PERSPECTIVES FOR THE STUDY OF TRANSNATIONAL UNIONS Synthesis of Findings Let us briefly review our findings from the point of view of our starting questions. 646 International Migration Review Transition to a first union: We find our hypothesis confirmed that the entry into the first union occurs earlier, but only for women of Turkish descent compared to women of French descent. This is mainly related to compositional differences between the groups. For men, however, there are no differences between youth of Turkish descent and respondents of French parentage. We can conclude that union formation of Turkish descendants in France occurs somewhat later than in Turkey, but there is no similarity with the pattern of Algerian descendants in France whose first union formation is even more postponed than that of other young adults. Type of union: Our analysis clearly identifies marriage as the dominant form of union, in most cases direct marriage, as expected, even more so for women than for men. Migration background: Our analysis shows the importance of Turkey as a marriage market in the sense that more than half of the partners in the TIES sample came for union formation from Turkey to France. We cannot confirm, however, that there is a higher likelihood of transnational unions for lower educated men or higher educated women. Homogamy of origin and endogamy: Descendants of Turkish immigrants in France show a relatively high rate of homogenous marriages as far as the national ⁄ migration background of the partner is concerned. This is consistent with international findings and also ties in with ethnographic observations: the prevalence of both the endogamic marriage rule (with relatives) and of the religious homogamy rule. Furthermore, our hypothesis of higher homogamy of origin among women than among men was confirmed. Taking these findings together, we see a large influence of the Turkish family culture in union formation. There remains, however, an open question from our analysis: How can we explain this strong influence of the Turkish pattern on union-formation behavior for young adults, especially young women who grew up in France? At the individual level, what are the reasons that motivate the choice of a Turkish immigrant rather than a young man born in France as a spouse? Intergenerational Relations and Gender Roles Expectations as Key Issues in Partner Choice We have seen that for women, forming a transnational union is strongly correlated to the existence of family links between partners. Marriages Union Formation and Partner Choice in a Transnational Context 647 between relatives represent 25 percent of transnational couples. This result suggests that the partner originates from the parents’ town or village of origin, which is probably true for other transnational unions too. This maintenance, in the migration context, of the endogamic and preferential marriage rules reveals that Turkish immigrant parents, especially those who come from rural areas, have kept strong links with their family members remaining in Turkey, strong enough for their daughters to marry their cousin who lives in Turkey. Given that Turkish immigrants have themselves been married in accordance with these rules, that they feel a commitment to their family remaining in Turkey and that they also face attitudes of rejection from French society, we can easily understand that some of the parents do wish to maintain strong relations with their origin and to encourage their daughter to marry a spouse originating from their home region.8 But to what extent is this wish shared by their daughters, and if it is, how does it pass down from one generation to the other? To what extent do the Turkish immigrants’ daughters subscribe to the Turkish norms and values? This raises the issue of transmission of values, and the issue of conflicts and negotiations between generations. The hypothesis of parental pressure is not really confirmed: only 2 percent of the women and 6 percent of the men mentioned that their parents had introduced them to their spouse, which shows that the practice of a marriage that is directly arranged by the parents no longer dominates in the migration context, but about 20 percent of respondents claimed that they suffered from parental pressure against their marriage and 25 percent experienced a strong encouragement for their marriage. Turkish parents probably adopt other strategies to orient their children towards a Turkish partner without proposing or imposing one, such as explaining how a marriage with a man of their region of origin would please them, or trying to create opportunities for their daughter to meet young men during holidays in Turkey, while strictly controlling their nights out in France. This pleads for the addition of some new questions on parental control over social life and leisure practices in further quantitative surveys. 8 The same process of weakening of tribes as a starting point for rural exodus and emigration occurred in North Africa. In a qualitative research about affective and sexual life among descendants of North African immigrants in France, Hamel (2003) observed that if in certain cases, which seem to be the less frequent, parents want their children to respect the preferential marriage inside the agnatic lineage or at least in the matrilinear branch, in other cases parents emigrated in order to escape the pressure and the control of the group on their own sexual and affective life (Hamel, 2003). 648 International Migration Review Like all young people, descendants of immigrants have to find a compromise between their parents’ expectations and their own wishes. Do they negotiate with their parents and how do they do it? Are they in a position to reject their parents’ expectations? Some of the young women of North African origin interviewed by Hamel (2003) in France were facing similar parental expectations and explained they did not want to risk a break-off with their family by choosing a partner of French origin. They clearly stated that even though they could fall in love with a ‘‘French’’ young man, they preferred to orient their choice towards a partner acceptable for their parents, that is, a partner of North African origin. They argued that older sisters who left home at 18 or 20 years old to have a free choice have experienced very hard material conditions of life and also huge emotional distress. Thus, in order to understand how the partner choice occurs, intergenerational relations must be examined in terms of their affective, material, and power dimensions. The affective and economic cost of breaking off of relations with family in the event of strong disagreement partly explains why daughters adapt their choice to their parents’ expectations. But this can not be sufficient to understand the matrimonial choices of these young women of Turkish descent. Some of them may share their parents’ values and expectations and see advantages in the choice of a partner originating from Turkey. In her ethnographic survey conducted in the beginning of 1990s with Turkish families living in France, Autan (1995) notes that both young female and male Turkish immigrants who arrived at a young age in France (before 14 years old) have conformed to their parents’ expectations regarding the origin of their partner. For women, Autan (1995) observes that a marriage with a Turkish immigrant enables them to transform gender relations in their couple (as our results show, this is independent from educational attainment). Indeed, the law regarding family reunification requires the partner living in France to have an income higher than the legal minimum income of a full-time job. In their case, the traditional virilocal pattern of residence does not apply, so they are not subjected to the control of their family-inlaw. Their husband is also dependent on his spouse for all administrative decisions because of the language barrier. These considerations may help young women to accept a marriage with a man born in the region of origin in Turkey, even though the relations with their husband in France are not so easy to manage: their husbands have to adapt to a society where gender roles and norms are different, and this can create misunderstandings Union Formation and Partner Choice in a Transnational Context 649 and conflicts. Do these factors also motivate female descendants of Turkish immigrants to choose a spouse in Turkey? What is the role of both men’s and women’s representations of gender roles on their choice of a partner? What are the expectations of the immigrant partner who marries someone living in Europe? Thus the study of intergenerational relations and of gender roles constitutes a key issue in future research for understanding why transnational unions are so frequent. Sexuality and Relation with the French Environment: The Competition of Values We saw that the maintenance of the norm of virginity before marriage for women is linked to a transnational marriage. Hamel (2006) found among North African women in France that they had developed three kinds of discourse about virginity: a first group reacts to the strong stigmatization to which they and their parents are subject by claiming their attachment to virginity and their wish to get married with a man of North African origin while claiming their desire to choose their partner themselves. They explain that their parents supervise their choice of friends and their social life, but they also criticize the ‘‘Western pattern’’ of sexuality, considering that although ‘‘French’’ women are more free, they are not always respected by French men. They claim to use virginity as a way to select respectful partners. Claiming their attachment to virginity is also a way to reassure their parents and to postpone marriage, which enables them to pursue their education. The second group is composed of young women who decided not to follow this norm, which was a long and difficult decision, for they had to fight the feeling of betraying their parents’ confidence. Like the women of the first group, they refuse their parents’ stigmatization but they also testify that losing their virginity was a way of appropriating their body and a kind of release from female subordination. Women of the third group left their parents’ home when they were 18 years old. They broke off with their parents because they wanted to be free to choose their partner, because they were against the virginity norm and, most of all, because they were afraid of proposals of marriage. Their decision led them to a very precarious situation: it stopped their education, which restricted their ability to find a stable job and some of them became homeless. Several years after their departure, they were regretting their decision to leave. This qualitative research has shown that claiming 650 International Migration Review the attachment to virginity is sometimes used by young women to get their parents to allow them to study or to choose their husband by themselves, even if it means choosing a husband in North Africa. The attitude of Turkish immigrants’ daughters toward virginity and their low rate of marriage to a man born in France can probably be interpreted as the result of negotiations and of compromises between parents and their children. CONCLUSION AND OUTLOOK Although the study presented here is based on a relatively small sample we are able to contribute to the discussion on the role of culture in shaping union formation by including cultural indicators. Indeed unionformation behaviors are a key topic of research for demography and anthropology. We showed that the objective of combining the knowledge produced by these two disciplines has influenced the design of the TIES survey. This combination appears to be highly relevant for interpreting the high rate of transnational unions: as we have shown it is strongly correlated to the attachment to the norm of virginity and to endogamic marriage. We found maintenance of Turkish patterns, but also their modification. We can conclude that there seems to be a development towards a multi-cultural society, in which several cultural patterns cohabit. But we must also stress that union-formation behaviors of descendants of Turkish immigrants are a syncretism of two cultures of reference in the sense that their behaviors are different from those that currently occur both in Turkey and in France and are at the same time the heritage of these two cultures. The continuing existence of strong family ties within the Turkish population does not fit either the classical assimilation theory (Gordon, 1964) or the segmented assimilation theory (Portes and Zhou, 1993) because the values and behaviors are directed to specifically Turkish roots and not to a French or an immigrant ‘‘mainstream.’’ We used the anthropological approach to explain and highlight the individual choices. This unquestionably allowed us to better understand the high rate of transnational marriages, but we do know, as anthropology has shown so many times, that culture and social norms, especially regarding issues such as affinity, gender relations, and sexuality, are constantly evolving. This is even truer in a migration context. Hence, we do believe Union Formation and Partner Choice in a Transnational Context 651 that a historical and comparative perspective should be adopted by taking into account the history of the Turkish immigration flow in France and the impact of the French national context. Our comparison with the descendants of North Africans showed that the descendants of Turkish immigrants are distinguished by a more pronounced preference for an immigrant partner. This difference, despite a common religion, along with the existence of the segmentary system in North Africa and of the same rules of affinity, invites us to follow the historical perspective to interpret and understand both the high rate of transnational marriages and its correlation with the attachment to virginity. Indeed, these two groups slightly diverge from each other, for the share of transnational marriages is much higher among young people of Turkish origin. The North African migration flow is also much older than the Turkish one. After several years in France, North African immigrants generally abandon the idea of returning home. Hence, their attachment to the values of the country of origin decreases as well as their feeling of being obliged to take account of the opinions of the family back home. To what extent may the high rate of transnational unions be considered as a characteristic result of a recent settlement, whatever the country of origin? This suggestion pleads for a comparison of families who arrived at different periods in time, which will be possible only several years from now. One of the objectives of the TIES project is to investigate the impact of the country of ‘‘destination’’. Hence, we will have the opportunity to compare the union formation behavior of descendants of Turkish immigrants in France to that of other Turkish-origin persons in The Netherlands, Germany, Austria, Belgium, Sweden, and Switzerland. The comparison with Germany will be highly instructive, as Germany is the country in Europe were the Turkish immigration flow is the largest and the oldest, which implies that the second generation in Germany forms a wide marriage market which may reduce the rate of transnational unions. 652 International Migration Review APPENDIX OVERVIEW OF THE TABLE A1 TIES SCREENING PROCEDURE IN FRANCE Turkish-sounding names Successful calls Respondents who agreed to answer the screening questionnaire Completed screening questionnaires Households in which no one is in the target group Households in which there is at least one person in the target group Households in which there is an immigrant whose daughter or son is in the target group, and lives somewhere else Households in which there is both one person at least in the target group and an immigrant whose daughter or son is in the target group and lives somewhere else and is married to a man whose father was not born in Turkey Total number of persons in the target group Including the daughters no longer living in their parents’ home, who are married to a man whose father was not born in Turkey Selected persons for the survey (one per household) Including the daughters no longer living in their parents’ home, and married to a man whose father was not born in Turkey Persons impossible to contact (after 20 attempts of calling) Wrong numbers (a mistake occurred when the first respondent delivered the selected person’s phone number or when the interviewer noted it down) Successful calls (the interviewer had a contact with the selected person) Selected person who refused the face-to-face interview Selected persons who accepted the face-to-face interview but did not come to the appointment for the face-to-face interview Usable completed interviews Including the daughters no longer living in their parents’ home, who are married to a man whose father was not born in Turkey DESCRIPTION OF THE SAMPLE TABLE A2 TRANSITION FOR THE TO A 72 1,628 16 1,114 5 110 21 983 372 110 500 4 FIRST UNION (%) Women First union Yes No Birth cohort 1971–1975 1976–1980 10,550 6,623 2,499 2,231 1,117 1,001 41 Men TurkishDescent FrenchDescent TurkishDescent FrenchDescent 41 59 49 51 24 76 42 58 9 23 21 31 6 18 23 26 Union Formation and Partner Choice in a Transnational Context DESCRIPTION OF THE TABLE A2 (CONTINUED). SAMPLE FOR THE TRANSITION TO A Women TurkishFrenchDescent Descent 1981–1985 1986+ Place of residence Paris Strasbourg Educational attainment Vocational higher secondary level Academic higher secondary level Tertiary education Religion during childhood None (or other) Christian Muslim Number of siblings 0 1–2 3+ Father with higher education No Yes Missing values Mother with higher education No Yes Missing values % N 653 FIRST UNION (%) Men TurkishDescent FrenchDescent 28 40 27 21 28 47 21 29 45 55 49 51 56 44 50 50 46 29 26 16 21 63 50 26 24 24 24 52 19 2 79 46 54 1 25 2 73 42 57 1 5 50 46 12 70 19 7 56 37 14 66 20 70 24 6 16 76 9 71 26 3 10 82 8 77 19 4 100 218 20 76 4 100 188 74 23 3 100 282 17 76 7 100 163 Source: Calculations based on TIES France 2007. Note: Shares do not add up to 100% because of rounded figures. REFERENCES Autan, C. 1995 ‘‘La tradition au service des transitions. Le mariage des jeunes turcs dans l’immigration.’’ Migrants-formation 101:168–179. Bajos, N., and A. Spira 1993 Les comportements sexuels en France. Paris: La documentation française, 352 p. Bernardi, L., and I. Hutter 2007 ‘‘The Anthropological Demography of Europe.’’ Demographic Research 17:541– 566. Blossfeld, H.-P., and G. Rohwer 1995 Techniques of Event History Modelling. New Approaches to Causal Analyses. Mahwah: Lawrence Earlbaum Associates. 654 International Migration Review BMFSFJ – Bundesministerium für Familie, Senioren, Frauen und Jugend (Hg) 2000 Familien ausländischer Herkunft in Deutschland. Leistungen, Belastungen, Herausforderungen. Sechster Familienbericht. Berlin: BMFSFJ. Borrel, C. 2006 ‘‘Enquêtes annuelles de recensement 2004 et 2005. Près de 5 millions d’immigrés à la mi-2004.’’ INSEE première 1098:4. ———, and C. Tavan 2003 ‘‘La vie familiale des immigrés.’’ In France. Portrait social. Ed. INSEE. Paris: Pp. 109–124. Bozon, M., and F. Héran 2006 La formation du couple: textes essentiels pour la sociologie de la famille. Paris: La Découverte, coll. Grands Repères. Classiques. De Tapia, S. 2006 ‘‘Les Turcs expatriés en 2005–2006: combien sont-ils ? Où sont-ils ? Les étrangers de Turquie: combien sont-ils ? Où sont-ils ?’’ Revue européenne des migrations internationales 22(3):229–251. Ergöcmen, B. A., and M. A. Eryurt 2004 ‘‘Other Proximate Determinants of Fertility, in Hacettepe University Institute of Population Studies.’’ In Turkey Demographic and Health Survey 2003. Ed. Hacettepe University Institute of Population Studies. Ankara: HUIPS, Ministry of Health General Directorate of Mother and Child Health and Family Planning, State Planning Organization and European Union. Pp. 89–96. Filhon, A., and G. Varo 2005 ‘‘Les couples mixtes, une catégorie hétérogène.’’ In Histoire de familles. Histoires familiales. Les résultats de l’enquête famille de 1999. Ed. C. Lefèvre, and A. Filhon. Paris: INED. Pp. 483–501. Foner, N. 1997 ‘‘The Immigrant Family: Cultural Legacies and Cultural Change.’’ International Migration Review 31(4):961–974. Girard, A. 1974 Le choix du conjoint: une enquête psycho-sociologique en France. Paris: Presses universitaires de France, 201 p. Gökalp, A. 1987 ‘‘‘Le dit de l’os et du clan’. De l’ordre segmentaire oghouz au village anatolien.’’ L’Homme 27(102):80–98. ——— 1989 ‘‘Population, parenté et nouvelles pratiques matrimoniales en Turquie.’’ In Le prix de l’alliance en Méditerranée. Ed. J. P. John. Marseille: Editions du CNRS. Pp. 145–155. ——— 1994 ‘‘‘Mariage de parents’: entre l’échange généralisé et le mariage parallèle. Le cas de la Turquie.’’ In Epouser au plus proche. Inceste, prohibition et stratégies matrimoniales autour de la méditerranée. Ed. P. Bonte. Paris: Edition de l’école des hautes études en sciences sociales. Pp. 439–452. Union Formation and Partner Choice in a Transnational Context 655 ——— 2006 ‘‘Conjoints et stratégies matrimoniales dans l’immigration.’’Cahiers d’étude sur la méditerrannée orientale et le monde turco-iranien 21. <http://cemoti.revues.org/ document563.html>. (Accessed 14 July 2010). González-Ferrer, A. 2006 ‘‘Who Do Immigrants Marry? Partner Choice Among Single Immigrants in Germany.’’ European Sociological Review 22(2):171–185. Gordon, M. 1964 Assimilation in American Life. Oxford: Oxford University Press. Hamel, C. 2003 L’intrication des rapports sociaux de sexe, de ‘race’, d’âge et de classe: ses effets sur la gestion des risques d’infection par le VIH chez les jeunes Français descendant de migrants du Maghreb, Mémoire de thèse. Paris: EHESS, 720 p. ——— 2006 ‘‘La sexualité entre sexisme et racisme: les descendants de migrant-e-s du Maghreb et la virginité.’’ Nouvelles Questions Féministes 25(1):41–58. Hoem, J. M. 1987 ‘‘Statistical Analysis of a Multiplicative Model and Its Application to the Standardization of Vital Rates: A Review.’’ International Statistical Review 55(2):119–152. ——— 1993 ‘‘Classical Demographic Methods of Analysis and Modern Event-history Techniques.’’ IUSSP: 22nd International Population Conference, Montreal, Volume 3: 281–291. HUIPS – Hacettepe University Institute of Population Studies 2004 Turkey. Demographic and Health Survey. Ankara: HUIPS, 217 p. Ilcan, S. M. 1994 ‘‘Marriage regulation and the rhetoric of alliance in northwestern Turkey.’’ Ethnology 33:273–296. INSEE 2005 Les immigrés en France, 153. <http://www.insee.fr/fr/ppp/sommaire/IMMFRA05. PDF>. (Accessed 14 July 2010). INSEE 2008a <http://www.insee.fr/fr/default.asp>. (Accessed 14 July 2010). INSEE 2008b Enquête annuelle de recensement de 2004 à 2006. <http://www.insee.fr>. (Accessed 14 July 2010). Electronic file: donnees-socio-demo-etranger-immigres. xls. Insel, A., S. Gürsel, and H. Levent 2003 ‘‘Les déterminants économiques et sociaux de la migration des travailleurs turcs vers l’union européenne.’’ Migrations Etudes 115:1–12. Johnson-Hanks, J. 2007 ‘‘What Kind of Theory for Anthropological Demography?’’ Demographic Research 16:1–26. Kalmijn, M. 1998 ‘‘Intermarriage and Homogamy: Causes, Patterns, Trends.’’ Annual Review of Sociology 4:395–421. 656 International Migration Review Koc, I. 2008 ‘‘Prevalence and Sociodemographic Correlates of Consanguineous Marriages in Turkey.’’ Journal of Biosocial Science 40(1):137–148. Lambert, Y. 2002 ‘‘La religion en France des années soixante à nos jours.’’ Données sociales: 2002– 2003 Edition. Paris: INSEE. Pp. 565–579. Lefranc, C., and S. Thave 1995 ‘‘Les enfants d’immigrés. Emancipation familiale et professionnelle.’’ Insee Première 368:1–4. Lievens, J. 1999 ‘‘Family-Forming Migration from Turkey and Morocco to Belgium: The Demand for Marriage Partners from the Countries of Origin.’’ International Migration Review 33(3):717–744. Lucassen, L., and C. Laarman 2009 ‘‘Immigration, intermarriage and the changing face of Europe in the post war period.’’ History of the Family 14:52–68. Manço, U. 2005 ‘‘Turcs d’Europe et la Turquie en Europe: deux intégrations en une.’’ Confluences Méditerranée 52:81–91. Milewski, N. 2003 ‘‘Partner Selection by Immigrants in Germany: The Impact of Religious Affiliation and Education on Age at Marriage.’’ Anthropologie 41(3):291–294. Münz, R., W. Seifert, and R. Ulrich 1997 Zuwanderung nach Deutschland. Strukturen, Wirkungen, Perspektiven. Frankfurt, New York: Campus Verlag. Ozyegin, G. 2009 ‘‘Virginal Facades: Sexual Freedom and Guilt among Young Turkish Women.’’ European Journal of Women’s Studies 16(2):103–123. Peristiany, J. 1966 Honour and Shame: The Values of Mediterranean Society. London: Wedenfeld and Nicholson, 266 p. Portes, A., and M. Zhou 1993 ‘‘The New Second Generation: Segmented Assimilation and Its Variants.’’ Annals of the American Academy of Political and Social Science 530:74–96. (Interminority Affairs in the U.S.: Pluralism at the Crossroads). Prioux, F. 2003 ‘‘L’âge à la première union en France: une évolution en deux temps.’’ Population 58(4–5):623–644. Régnard, C. 2006 Immigration étrangère en France en 2005. Rapport annuel de la direction de la population et des migrations. Paris: La documentation française, 260 p. Reher, D. S. 2004 ‘‘Family ties in Western Europe: persisting contrasts.’’ In Strong Family and Low Fertility: A Paradox? New Perspectives in Interpreting Contemporary Family and Reproductive Behavior. Ed. G. D. Zuanna, and G. A. Micheli. Dordrecht, Boston, London: Kluwer Academic Publishers. Pp. 45–76. Union Formation and Partner Choice in a Transnational Context 657 Roloff, J. 1998 ‘‘Eheschließungen und Ehescheidungen von und mit Ausländern in Deutschland.’’ Zeitschrift für Bevölkerungswissenschaft 23(3):319–334. Safi, M. 2008 ‘‘Inter-mariage et intégration: les disparités des taux d’exogamie des immigrés en France.’’ Population 63(2):267–298. This French journal is translated in English: Intermarriage and assimilation: disparities in levels of exogamy among immigrants in France, Population 63(2): 267-298. Sebille, P. 2009 ‘‘Un passage vers l’âge adulte en mutation?’’ In Portraits de familles - L’enquête Etude des relations familiales et intergénérationnelles. Ed. A. Régnier-Loilier. Paris: INED Collection «Grandes Enquêtes». Pp. 315–340. Simon, P. 2003 ‘‘France and the Unknown Second Generation.’’ International Migration Review 37(144):1091–1119. Stirling, P. 1965 Turkish Villages, 1st ed. (This old book which is no more published has been numerized by the Center for social anthropology and computing of the University of Kenterburry: <http://lucy.ukc.ac.uk/TVillage/StirlingC9.html>). (Accessed 14 July 2010). Straßburger, G. 2003 Heiratsverhalten und Partnerwahl im Einwanderungskontext. Eheschließungen der zweiten Migrantengeneration türkischer Herkunft. Würzburg: Ergon Verlag (Doctoral Thesis). Sunar, D., and G. O. Fisek 2005 ‘‘Contemporary Turkish Families.’’ In Families in Global Perspective. Ed. J. L. Roopnarine and U. P. Gielen. Boston: Pearson. Pp. 169–183. TIES 2008 <http://www.tiesproject.eu/>. (Accessed 14 July 2010). Timmermann, C. 2006 ‘‘Gender Dynamics in the Context of Turkish Marriage Migration: The Case of Belgium.’’ Turkish Studies 7:125–143. ——— 2008 ‘‘Marriage in a ‘Culture of Migration’. Emirdag Marrying into Flanders.’’ Europêan Review 16:585–594. Timur, S. 1972 Türkiye’ de aile yapı̈sı̈ [Structure of the Turkish family]. Ankara: Hacettepe University publication. Toulemon, L. 1996 ‘‘La cohabitation hors mariage s’installe dans la durée.’’ Population 3:675–715. Tribalat, M. 1995 Faire France. Une enquête sur les immigrés et leurs enfants. Paris: La Découverte, 231 p. ——— 1996 De l’immigration à l’assimilation. Enquête sur les populations d’origine étrangère en France. Paris: La Découverte, 302 p. 658 International Migration Review de Valk, H. 2006 Pathways into Adulthood. A Comparative Study on Family Life Transitions among Migrant and Dutch Youth. Utrecht: Thela Thesis (Doctoral Thesis). ——— 2008 ‘‘Union and Family Formation.’’ In The Position of the Turkish and Moroccan Second Generation in Amsterdam and Rotterdam. The TIES Study in the Netherlands. Ed. M. Crul, and L. Heering. Amsterdam: Amsterdam University Press. Pp. 143– 168. ——— 2009 ‘‘Parental Influence on Union Formation Preferences among Turkish, Moroccan and Dutch Adolescents in The Netherlands.’’ Journal of Cross-Cultural Psychology 38(4):487–505. Vergin, N. 1985 ‘‘Social Change and the Family in Turkey.’’ Current Anthropology 26(5):571–574.
© Copyright 2026 Paperzz