Intergenerational co-residence – the case of Turkey

An important context for the study of intergenerational co-residence in a specific immigrant group is thus the prevalence of such living arrangements in the home country. Indeed, the three-generational extended household has traditionally been considered a cultural ideal, especially in rural parts of Turkey (Abadan-Unat Reference Abadan-Unat1986; Aytac Reference Aytac1998). Whilst such intergenerational co-residence has never been more common than nuclear households among the population as a whole (Timur Reference Timur and Allman1978), it holds particular importance at two points in the family lifecycle: for young newlywed couples, who may spend the first years after marriage living with the parents of the groom, and in old age, when frail and/or widowed parents may live with adult children. Thus, more than half of Turks (in Turkey) over the age of 65 years live in non-nuclear households, predominantly in the form of co-residence with adult children (Yavuz Reference Yavuz2009). Furthermore, especially in little-educated rural segments of the population, the youngest son and his wife may indeed live with the husband's parents from their marriage onwards (Bastug Reference Bastug, Liljeström and Özdalga2002). Such extended households have been termed ‘traditional’ ones, in contrast with extended households that are formed in response to specific needs, e.g., those of older family members later in the family life cycle (Aytac Reference Aytac1998). The significance of intergenerational co-residence for older individuals must also be understood in relation to limited Turkish state provisions for income support in old age and a general lack of public care, including limited access to care homes (Saka and Varol Reference Saka and Varol2007).

Calling attention to the theme of ‘linked lives’ (which family members live together), the vast majority of extended households in Turkey (95%) are patrilocal, i.e. parents co-habiting with sons rather than with daughters (Aykan and Wolf Reference Aykan and Wolf2002: 408). In patrilocal households, the co-resident son's wife – the daughter-in-law – is generally expected to provide much of the domestic labour (Fisek Reference Fisek and Kagitcibasi1982; Yakali-Camoglu Reference Yakali-Camoglu2007). Such gendered expectations may be retained within the context of migration: in a German interview study with 30 adult children of original Turkish labour immigrants (the respondents were in the 25–35 years age group), the topic was expectations regarding care for parents when they grew old. In the interviews, the men generally implied that their wives – i.e. the older immigrants’ daughters-in-law – and not the actual sons (or daughters) themselves would become the main care-givers (Lorenz-Meyer and Grotheer Reference Lorenz-Meyer, Grotheer, Arber and Attias-Donfut2000: 200).

After this brief discussion of intergenerational co-residence in the country-of-origin context, the next section explains the data underlying the present analysis.

Quantitative method: administrative registry data

The first data source was quantitative, utilising registry data from Statistics Denmark. Such registry data are available because all legal residents in Denmark (i.e. both immigrants and natives) are assigned a unique personal identifier that is linked to a wide array of information from authorities, including where and with whom individuals live. The advantage of using such registry data is that it contains data on the full population of a given group and avoids potential non-response bias. Regarding age, the analysis included the 1,801 Turkish immigrants in Denmark who were between the ages of 65 and 74 years old in 2011.Footnote ³ These data were quantitatively compared with the situation amongst the 543,835 Danes in the same age group (Statistics Denmark 2016).

These registry data allow a detailed mapping of different types of extended households, revealing, for example, whether older individuals live with sons or daughters and whether these adult children have spouses, children, or both, of their own (Liversage and Jakobsen Reference Liversage and Jakobsen2016).

Qualitative method: interviews with ageing Turkish immigrants

The second data source was interviews with members of 24 households containing at least one older immigrant. Address information from Statistics Denmark made it possible to locate households with at least one Turkish-born resident who was between 70 and 72 years of age in 2013. Letters about the project in both Turkish and Danish were sent to such households in different parts of Demark and were followed by phone contact to provide further information and, if possible, arrange for interviews. As access to low-educated elderly immigrants is often difficult (Gardner Reference Gardner2002; Zubair and Victor Reference Zubair and Victor2015), only approximately half the letters led to interviews; the other immigrants were either unreachable or unwilling to participate in the research investigation for health or other reasons.

All interviews were carried out in Turkish or Kurdish – a prerequisite as the majority of the respondents spoke only limited Danish, and many of the older women hardly spoke the host country's language at all. The interviews combined semi-structured and life-history approaches, in which the respondents were encouraged to talk about their lives and were asked specific questions, including about with whom they lived, what help they received, and their views on their present-day and possible future living arrangements (Bertaux Reference Bertaux, Humphrey, Miller and Zdravomyslova2003).

As earlier research on Turkish first-generation immigrants has shown that the use of a rigid methodological approach can seriously damage rapport with respondents and thus jeopardise interview quality, the interview was approached in a flexible way (Mirdal Reference Mirdal1984, Reference Mirdal2006). Thus, if a spouse wanted to participate in a household interview, he or she could do so. This flexibility led to married couples being interviewed together in 15 of the 24 households. In nine households, the older migrants (mainly widowed ones) were interviewed on their own. In a few cases, another family member – often a daughter or a daughter-in-law – also participated in parts of the interview. Such family member participation arose either from family visits interrupting the interview or from family members living in the same household as the initially contacted immigrant.

While the primary respondents were 70–72 years old, their spouses could be either younger or older. Hence, the full age spectrum ranged from one wife (from a second marriage) who was in her mid-fifties to a husband in his early eighties. Regarding class background, the majority of the respondents had very limited schooling: most of the men had been schooled for five years and most of the women had no schooling at all. This observation aligns with Turkish survey results showing that in 1970, 69 per cent of Turkish men and just 40 per cent of Turkish women were able to read and write (Özbay Reference Özbay and Kagitcibasi1982: 135). Educational background is relevant as in Turkey, the less-educated live in extended households more often than better-educated members of the population (Aytac Reference Aytac1998).

All interviews but one (due to respondent preference) were taped and transcribed in full (in Turkish). Aided by NVivo software, the topic of analysis arose from the fact that intergenerational co-residence was an important topic in the majority of the interviews and an issue that older immigrants living in nuclear households also discussed. Based on this topical interest, the transcripts underwent a directed content analysis (Hsieh and Shannon Reference Hsieh and Shannon2005) that applied a coding scheme focusing on the concrete details regarding living arrangements, the reasons given for their existence and how such living arrangements were evaluated. To protect anonymity, respondent names have been altered. Spouses who participated in the interviews are referred to as ‘[pseudonym]’s husband/wife’. Interview excerpts were translated from Turkish into English by the author.

Extent of intergenerational co-residence amongst older Turks in Denmark

As a point of departure, the analysis mapped the percentage of 65–74-year-old Turkish immigrants (N = 1,801) who live in some type of extended household and compares it with that of their native Danish counterparts. Households containing either non-adult children (of which there were very few) or children below the age of 25 years who had no partners or children of their own were classified as nuclear households.Footnote ⁴

The registry data analysis shows that in the selected age group, 23 per cent of the Turkish immigrants in Denmark live in extended households with adult children (Figure 1). On the one hand, this proportion far exceeds the 2.7 per cent of their older Danish counterparts who live in such households. On the other hand, the immigrants’ co-habitation rate falls far below the level observed among their Turkish peers who never emigrated: survey data show that 55 per cent of 65–74-year-old Turks live in non-nuclear households (Yavuz Reference Yavuz2009).

Source: Register data from 2011 (Statistics Denmark 2016).

Figure 1. Household types for the 65–74 years age group.

Figure 2 shows a further sub-division of the 23 per cent of older immigrants who live in intergenerational co-resident households. The figure shows that the predominant type of intergenerational co-residence among older Turkish immigrants in Denmark is living with sons and the wives and (later) children of those sons. Such families conform to the Turkish cultural ideal of patrilocal extended families (Aykan and Wolf Reference Aykan and Wolf2002) and make up 59 per cent of the extended households. In sharp contrast, only 6 per cent of extended households are parents living with daughters who have partners, children, or both, of their own. This finding corresponds with findings from Turkey, where co-residence with daughters is also rare and may occur when, e.g. a daughter is an only child (Aykan and Wolf Reference Aykan and Wolf2002). The second most common type of intergenerational co-resident households – and of particular interest in the present analysis – is the 20 per cent of extended households that include single sons older than 25 years. The remaining 15 per cent of the households contain either single daughters or a combination of sons and daughters.

Notes: 1. There may be other single children too. 2. These may be families with both a single son and a single daughter over 25 years, or families with two children, both of whom have partners, children or both.Source: Calculations based on register data from 2011 (Statistics Denmark 2016).

Figure 2. Types of intergenerational co-resident households amongst 65–74-year-old Turkish immigrants in Denmark.

For comparison, among the 2.7 per cent of the older native Danes who live with adult children, more than half of this small percentage co-habits with unwed adult sons. The proportion of Danes living with adult children (sons or daughters) who are married is negligible.

To sum up, the registry data analysis documented that the co-habitation rate of the older Turkish immigrants in Denmark falls roughly midway between the levels found in the countries of origin and of destination. The registry data also showed that most of the older immigrants (around three-quarters) live in nuclear households, i.e. as couples or on their own. In the next sections, I analyse the qualitative interviews to better understand the processes underlying the observed pattern of intergenerational co-resident living. I frame the discussion in terms of the four themes from lifecourse research (Elder Reference Elder1994).

First theme: the social meaning of co-residence in old age

Relevant to the discussion of the social meaning of co-residence for older immigrants, Matthäi (Reference Matthäi2004) developed a typology based on interviews with older, single immigrant women in Germany. Matthäi distinguishes among three types of family relations. The first type is ‘(extended) family community’, which resembles family relations in the traditional societies from which most of the interviewed immigrants originated. The second type is more ‘modern – nuclear – family arrangements’, stressing affection rather than instrumental support in intergenerational relations. The third type is ‘fragile family arrangements’, in which older women live alone and perceive intergenerational relations as dissatisfactory due to a discrepancy between the older women's traditional family expectations and the individualisation processes of the host country (Matthäi, quoted in Baykara-Krumme Reference Baykara-Krumme2008). This typology thus calls attention to both a polarity between immigrants who are happy to live in either extended or nuclear families and to the existence of older immigrants who find themselves dissatisfied with living in nuclear families due to unfulfilled desires to live in extended families.

In the interviews conducted for this paper, these three types of family relations and living arrangements were also observed. Some respondents expressed happiness with living in ‘modern’, nuclear family arrangements (Matthäi's second type), attributing importance to the independence of both themselves and their children (see also Liversage and Mirdal Reference Liversage and Mirdal2017). Such respondents emphasised that they considered extended households ‘old fashioned’ and out of tune with modern life in Denmark. As one widow, Pinar – who was one of the better-educated respondents – put it:

Pinar lived in a flat of her own and had no plans to move in with the younger generation. Regardless of not co-habiting with her two daughters, who – in the Danish dual-earner context – both worked, Pinar nevertheless received their regular support. Thus, living in a ‘modern’, nuclear fashion did not necessarily entail a lack of intergenerational contact and support.

Similarly, several other respondents from nuclear households had a high level of interaction with their adult children. This interaction was often facilitated by the older family members living very close to, and often next door to, one or more of their children – a way of life that could reduce the felt need to share accommodation fully. Indeed, in 11 out of the 18 nuclear households identified from the qualitative material, the generations lived very close together (see Table 1), in several cases with respondents living within the same flight of stairs in a multiple-storey building as one or more adult children. Families who live in such close proximity can be described as ‘functionally extended families’ (Bastug Reference Bastug, Liljeström and Özdalga2002). This phenomenon is widespread in Turkey and may be more common than both distant living and intergenerational co-residence, especially within cities (Aytac Reference Aytac1998). Such living arrangements may provide families with both substantial daily contact and the privacy of separate dwellings and may reduce the need for intergenerational co-residence.

Table 1. Respondent characteristics

Notes: 1. Woman was second wife. 2. Came to Denmark as refugees. 3. Spouses’ placement in parentheses indicates that they did not participate in the interviews. n.a.: not available.

The interviews also show, however, that some respondents attributed a strong positive value to co-residence in old age. Returning to Matthäi's typology, such respondents could either belong to the fortunate first type who lived in extended households, or they could belong to the dissatisfied third type, unable to live in the way they desired. As one couple exemplifying the latter type put it:

These two spouses, who were in rather poor health, in fact received substantial day-to-day help from the younger generation, even though they did not have any of their five children as neighbours. The couple nevertheless expressed a longing to live within an extended family – and not with a daughter and her husband but in the ‘ideal’ arrangement, with a son and a daughter-in-law. Believing that they would have lived in such a household had they remained in Turkey, the couple today had to contend with living on their own. The desolation that this couple expressed echoes the sentiments of some Pakistani and Bangladeshi elders who see peers in the community being ‘abandoned’ by their children due to, for example, daughters-in-law not wishing to live with and care for them (Victor and Zubair Reference Victor, Zubair, Karl and Torres2016). As all of Matthäi's three types were represented in the source material, the interviews thus show not only that the social meaning of co-residence in old age varied substantially among the respondents but that some respondents in fact desired co-resident living but had been unable to have these desires fulfilled.

Second theme: linked or interdependent lives

The quote from Ahmet and his wife in the preceding section exemplifies one of the multiple ways in which lives are linked: older immigrants’ abilities to live in extended family households depend not only on their own wishes but also on the wishes and actions of others. Hence, co-residence with a son and his wife – the ‘culturally ideal type’ – also depends on the desires and actions of this younger generation. Such households (59% of the extended households according to the registry data) made up four of the 24 households visited for interviews. These four extended households were all of the ‘traditional’ type (Aytac Reference Aytac1998); they were established years ago, when the then-young co-resident son married, in all cases to wives from Turkey who had arrived as marriage migrants. The sons in question were also amongst the younger ones in the large families, corresponding to the traditional Turkish practice that the youngest sons and their wives support the son's parents (Bastug Reference Bastug, Liljeström and Özdalga2002). In contrast, among Korean immigrants, for example, it is commonly the oldest sons and their wives who live with parents (see Kim Reference Kim2012).

Older respondents living in such ‘ideal’ extended households generally expressed both happiness and gratitude for living this way. As Mohammed's wife told us:

In some of these extended households, we heard of considerable sharing of material and non-material resources, e.g. with the elderly mother and the daughter-in-law both contributing to household chores and to the care of young children. Hence, depending on the needs and abilities of different family members, intergenerational support could flow downwards as well as upwards. If the health of the older immigrants deteriorated, however, their contribution to household chores could be replaced by the need to receive daily care and support. Returning to the concept of linked lives and considering the gendered nature of care, living with a mother- or father-in-law in poor health could thus place a substantial care burden on a daughter-in-law. Such was the case in the extended household in which the 71-year-old widow Sürreya lived. Sürreya told us that in the past two years, her health had declined:

Sürreya's daughter-in-law was present during part of the interview as she had returned from work. About the present-day situation, with her mother-in-law requiring increasing care (especially for personal hygiene), the daughter-in-law said the following:

Given Sürreya's needs, the Danish welfare state offered free home help. While immigrant families’ combinations of family and state-supported care for the elderly is documented elsewhere (Lan Reference Lan2002; Khvorostianov and Remennick Reference Khvorostianov and Remennick2015), Sürreya did not accept this combination and rejected the Danish home helper. The daughter-in-law thus had to work a particularly arduous ‘second shift’ (Hochschild Reference Hochschild1998); she was expected to both go to work and to care not only for her mother-in-law but also her husband and their two minor children. From Sürreya's perspective, however, she lived in the type of household that was desirable in old age, where she could receive the expected daily help from a daughter-in-law. Sürreya's preference for hands-on care from a member of the younger generation thus contrasts with the views of older European immigrants in Luxembourg, whose expectations regarding hands-on care are oriented towards either spouses or professionals (Karl, Ramos and Kühn Reference Karl, Ramos and Kühn2016).

While older relatives who required care would most likely also present a family challenge if they lived in households of their own, the provisions of support might well have been more widely distributed among different family members, rather than primarily falling upon the daughter-in-law.

Third theme: agency and intergenerational relations

As evident in Sürreya's present-day household, the gendered and generational hierarchy in such families may place a burden on daughters-in-law. Many of the female interviewees had themselves experienced the subordinate daughter-in-law position in extended families in the 1970s Turkish countryside. Hence, Belgin told the following story about living with her parents-in-law before immigrating to Denmark:

Such extended family living arrangements arose not only from cultural preference but also from the economic scarcity in Turkey at the time – a scarcity that often also instigated the men's emigration. Additionally, after arriving in Denmark, daughters-in-law found that their subordinate position could shape how their lives unfolded. As Sabahattin's wife said about the time shortly after she had joined her husband in Denmark:

According to this quote, Sabahattin's wife had been unable to resist the family's (rather than simply the mother-in-law's) demands. Consequently, she had to live with and care for her mother-in-law for four years. Sabahattin's wife's actions thus exemplify what Kandiyoti (Reference Kandiyoti1988) calls following the ‘rules of the game’ in a given type of patriarchal context. A central element of a younger woman's acceptance of her weak positioning is that ‘the life cycle in the patriarchally extended family is such that the deprivation and hardships she experiences as a young bride are eventually superseded by the control and authority she will have over her own subservient daughters-in-law’ (Kandiyoti Reference Kandiyoti1988: 279). Modernisation processes may, however, change the family dynamics within a lifetime:

While Kandiyoti wrote from the context of a rapidly modernising Turkish society, migration from a lesser to a more developed part of the world may also instigate a ‘breakdown of classic patriarchy’. Hence, in the Danish context, the economic independence of the younger generation and the lower level of economic dependency on parents for survival in a welfare state partly explains why the proportion of older Turkish immigrants living in extended families is only approximately half that observed in Turkey (Yavuz Reference Yavuz2009).

That intergenerational co-residence could be a contested issue between mothers-in-law and daughters-in-law, with the two generations of women attaching very different meanings to such living arrangements, was also evident in a story told by the respondent Zahide. She herself lived in a ‘culturally ideal’ extended family (with a son, daughter-in-law and grandchildren) and told the following story about another Turkish immigrant family in Denmark. A son in the family had married a wife from Turkey. The young wife arrived from Turkey as a marriage migrant and moved in with her husband and her parents-in-law. The newlywed couple soon had two children. Then, the following events reportedly occurred due to the daughter-in-law's unhappiness with the intergenerational co-residence:

This short story can be read as a ‘moral tale’ in which the narrative plot contains a clear moral argument (Mattingly and Garro Reference Mattingly and Garro2000). According to the story, the daughter-in-law initially followed the prescribed path, moving from her home in Turkey and in with her new husband and his parents. Over time, however, and possibly affected by the norms and practices of the surrounding Danish society, the woman sought to exert agency and move into a nuclear household with her spouse and children. As the husband was linked to both his wife and his mother, his stance on the wife's desire to move became crucial. As told by the older woman Zahide, it is unsurprising that the storyline evolves with the son's rejection of the wife's dissent (vis-à-vis the structures of classical patriarchy), with him instead vowing to never leave his mother.

According to Kandiyoti, in the Turkish context of ‘classical patriarchy’, older women have a vested interest in suppressing romantic love between youngsters to ‘keep the conjugal bond secondary and to claim sons’ primary allegiance’ (1988: 279), to maximise the protection and support such older women can rely on. That such strong mother–son bonds may indeed challenge the bond between a son and his wife is well documented in studies of Turkish families (Bastug Reference Bastug, Liljeström and Özdalga2002; Olson Reference Olson and Kagitcibasi1982). In fact, in Zahide's story, the wife's desire to dismantle the extended family resulted in her husband siding with his mother and with the wife being expelled from her own marriage. As divorce in Turkey and amongst Turkish immigrants is considered a negative family event (Akpinar Reference Akpinar2003; Liversage Reference Liversage2012a, Reference Liversage and Charsley2012b), the story's positioning of divorce (especially for a couple with children) as preferable to a son's leaving his mother adds weight to the importance narratively attributed to retaining practices of intergenerational co-residence.

The concluding line – ‘they are marrying him [again] this year’ – constructs the older parents, rather than the son himself, as agents in his (arranged) remarriage. In effect, the story ends with the subordinate daughter-in-law about to be replaced by another woman who most likely will better ‘know her place’ and not challenge the practices of intergenerational co-residence.

Similar stories placing sons between their parents’ desire to retain intergenerational co-residence and ‘disobedient’ wives pushing for nuclear living are also reported as widely occurring in Turkey (Yazici Reference Yazici2012). The circulation of such stories in both Turkey and the immigrant diaspora can be read as an indication that the retention or demise of extended families is a contested issue whose continued existence is under pressure and thus cannot be taken for granted.

Fourth theme: the context of historical times and places

The fourth of Elder's themes is the context of historical times and places. Having already discussed extended family living in the country-of-origin context, this section focuses on the context of Denmark.

According to the literature, two important factors to consider when investigating extended household living amongst immigrants is the economic situation in the investigated group and the possibilities for extended families to live in the available housing stock.

Regarding the importance of the economic situation, the literature shows that one rationale behind immigrants living in extended rather than nuclear households is that individuals with few means can reduce their living expenses through economies of scale (Glick Reference Glick1999; Glick and Hook Reference Glick and Hook2011; Gurak and Kritz Reference Gurak and Kritz2010). Regarding the economic situation of the investigated group, studies show that rates of poverty amongst older Turkish immigrants are high; in fact, they are much higher than in the majority population due to both marginalisation from the labour market earlier in life and to Danish pension rules that disadvantage immigrants (Jakobsen and Pedersen Reference Jakobsen and Pedersen2016). Thus, while just 1 per cent of older Danes (in the 65–74 years age group) are classified as poor according to Organisation for Economic Co-operation and Development measures, a full 29 per cent of their Turkish immigrant peers are classified as poor. As two older women say about their difficult economic situation:

These two women both lived in nuclear households and were not able to share expenses with their adult children; however, older immigrants living in extended households also told of having very limited means. The unfavourable socio-economic conditions in which a large percentage of older Turkish immigrants find themselves thus most likely contribute to the creation or perpetuation of extended family living arrangements.

A second contextual factor that may affect intergenerational co-residence is the available building stock. As intergenerational co-residence is highly unusual in the individualised Danish society, flats in Denmark's publicly owned rented housing estates (where the majority of the older immigrants from Turkey live) cater almost exclusively to nuclear households. In fact, the widow Halime, quoted above, lived alone when she was interviewed. Until a decade ago, however, she had lived with her husband, son and daughter-in-law. When the son and his wife had their second child, however, the flat became too small; with no access to larger accommodation, the ‘culturally ideal’ extended family split up. Hence, a lack of suitable dwellings may decrease older immigrants’ abilities to live in intergenerational co-resident households.

A third contextual factor – to which the literature pays scant attention – is the host country's immigration regime. Returning to the importance of context in terms of both time and place, Denmark in the 1980s and 1990s had a very open immigration regime in terms of marriage migration. The majority of the older immigrants’ children (80–90%) who came of age at this historical time indeed married spouses found in Turkey (Celikaksoy-Mortensen Reference Celikaksoy-Mortensen2006; Liversage Reference Liversage2013). As extended households are important at the point in the family lifecycle when children become adults and marry, 80 per cent of the 18–24-year-old brides who moved from Turkey to Denmark due to marriage in the 1990s began life co-residing with their in-laws (Liversage and Jakobsen Reference Liversage and Jakobsen2010).

In 2000–2003, however, Danish rules of family migration were tightened substantially. To both reduce marriage migration and decrease parental influence regarding offspring's marriages, a rule stipulated that both partners in a couple had to be at least 24 years old for a spouse to obtain a marriage migration visa to Denmark. Another rule – of particular importance to the present study – specified that for couples to gain a visa for one spouse, they had to start life together in a dwelling of their own and not in a parental household (Liversage and Rytter Reference Liversage and Rytter2014).

Kerim, who lived with his wife, had indeed encountered the consequences of this rule. The couple had an only child – a daughter. When she married a spouse from Turkey a decade previously, the following events occurred:

Consequently, Kerim and his wife lived in a nuclear household rather than in the extended household they desired. These visa regulations affected which women could become part of extended households and make them into ‘ideal’ ones; for linguistic and other reasons, the respondents considered only young women with a Turkish (and not, e.g., a Danish) background as appropriate spouses for co-resident sons, but sons who wished to continue living with their parents after marriage could no longer seek wives from their town or village of origin. Their marital options were thus limited to women of Turkish origin already living in Denmark. This altered legal framework may partly explain why the second most common type of intergenerational co-resident households are those with unmarried sons – a phenomenon I discuss in the next section.

A ‘transitional type’ of co-residence?

According to the quantitative analysis, in 20 per cent of the extended households, 65–74-year-old immigrants live with unmarried sons – thus, in households that do not afford the same access to female labour from the younger generation as the culturally ideal type of household. In the qualitative material, two of the 24 households contained unmarried sons who remained at home at mature ages. In at least two other households, these sons had lived at home for years but had ended up leaving the parental home while still single. Sabahattin and his wife were one couple with whom an unwed son – in his late thirties – had lived until shortly before the interview. Regarding their son's prolonged unwed status, Sabahattin's wife said the following:

Evidently, the formation of a ‘traditional’ type of extended household requires two steps: first, a son must remain at home, and second, he must marry a wife willing to move into the parental household. While Sabahattin's family initially succeeded in the first step – having a son remain at home – the second step remained uncompleted. While Sabahattin's son might more easily have found a spouse in a Turkish village, partly because marriage to him would be a ticket to Europe (Timmerman Reference Timmerman2008), the Turkish women in Denmark whom Sabahattin's son courted clearly stated that moving in with his parents after the wedding was not an option. As a consequence, this son remained a bachelor for years on end.

The previous statement from Sabahattin's wife also indicates that a son's remaining at home may be accomplished through a strong emotional tie between the mother and the son – a bond that was also visible when Mohammed's wife stated earlier in the paper that ‘my son would never have the heart to [leave me]’ and in the ‘moral tale’ told by Zahide. When Sabahattin's son finally left the parental home, he might well have improved his chances in the marriage market, but the move made Sabahattin lament: ‘They have all left … It is only us left here, in this big house’. While this elderly couple had expected to live in an extended household, and while the old women herself had ‘done her dues’ as a daughter-in-law when she returned to Turkey to care for her mother-in-law years ago, the couple did not succeed in filling their flat with the desired family members. Sabahattin's wife was thus one of the immigrant women who had ‘paid the heavy price of an earlier patriarchal bargain, but [were] not able to cash in on its promised benefits’ (Kandiyoti Reference Kandiyoti1988: 282).

The 5 per cent of older Turkish immigrants who co-habit with unwed sons may thus represent a ‘transitional type’ of extended household, shedding light on the processes through which household patterns evolve over time in the context of migration. Changing patterns may not come about only from a general pattern of children adapting to the more individualised host country society (Baykara-Krumme Reference Baykara-Krumme2008). Instead, sons may express continued support for retaining established patterns of support towards older parents (Lorenz-Meyer and Grotheer Reference Lorenz-Meyer, Grotheer, Arber and Attias-Donfut2000). In contrast, a more reluctant participant may be the (potential) daughters-in-law. One element in explaining the observed pattern of extended household living may thus be the combination of national rules that bar daughters-in-law from Turkey from entering such households and the seeming unwillingness of Turkish women in Denmark to move in with in-laws. The immigration regime may thus also affect patterns of co-residence and the number of older immigrants who come to belong to Matthäi's ‘fragile’ family type – older immigrants who find themselves unable to become part of a desired type of extended household.