Introduction
Today, many Western countries face two salient phenomena that profoundly change the way societies are organized: population ageing and the globalization of migration (Torres
2013). Consequently, increasing numbers of people age in a foreign land. Due to an accumulation of risk factors, such as low socio-economic position (Reijneveld
1998), poorer health conditions (Denktaş
2011) and facing difficulties that are associated with international migration such as discrimination and social exclusion (Silveira and Allebeck
2001), older migrants are considered socially vulnerable (Cela and Fokkema
2016). Exemplary of this social vulnerability is the finding that older migrants are lonelier than their native peers (Fokkema and Naderi
2013; Victor et al.
2012). Regarding loneliness among this population, there is a specific need to not only study structural factors leading to loneliness, such as low socio-economic position, as has often been done (Victor et al.
2005). Instead, more focus is desired on how belonging from a migrant’s perspective, with all sociocultural precariousness that comes with migration, explains variation in loneliness among older migrants (De Jong Gierveld et al.
2015).
Inspired by Berry’s (
1980) model for acculturation strategies, we distinguish between different forms of belonging from a migrant’s perspective. Belonging can take place within the country of settlement, but can also be oriented towards the country of origin, i.e. a transnational belonging. An understanding of how older migrants give meaning to a sense of belonging in an increasingly transnational world is still largely unexplored (Torres
2013). We aim to fill this knowledge gap and study how a sense of belonging functions as protection against loneliness. First, we explore profiles of belonging among older migrants. Second, we determine how these profiles are associated with loneliness. We study Turkish and Moroccan older migrants in the Netherlands, a receiving country for many labour migrants in the 1960s and 1970s. Turkish and Moroccan migrants now form the main migrant groups in the Netherlands, together with Surinamese and Antillean migrants.
Belonging and loneliness
Departing from loneliness and well-being literature, belonging centres around having social attachments to and interactions with other people (Baumeister and Leary
1995). Elsewhere, belonging is posed as a vital mental health concept and defined in more generic terms: belonging requires a personal involvement in a “system” or “environment” (Hagerty et al.
1992). This is where one can say: “We belong together” or: “I am one of them.” For migrants, belonging is hardly self-evident. After migration ties to family, friends and community in the country of origin are put under pressure (Treas and Batlova
2009). Insecurity about how to socialize and about social expectancies in the new country obstructs the development of a new social network (Watt and Badger
2009). In migration studies as well as in human geography, belonging is brought in relation to a feeling of being “at home” (Yuval-Davis
2006), in which “home” represents a “symbolic space of familiarity, comfort, security and emotional attachment” (Antonsich
2010, p. 650). The significance of home arguably becomes stronger in the context of migration: the sheer fact of being in a different place with different customs and social norms compromises the extent to which one feels a sense of belonging and might lead to uncertainties regarding identity (Lee et al.
2010). Lynd (
1958, p. 210) simply put it this way: “Some kind of answer to the question: ‘Where do I belong?’ is necessary for an answer to the question: ‘Who am I?’” A lack of belonging can result in loneliness, deprivation, feeling an outsider and valuing life as unfulfilled and shallow (Verkuyten
2004). Loneliness is a situation experienced by an individual as one where there is a dissatisfying quality or quantity of personal relationships (De Jong Gierveld
1998). This definition concerns the subjective evaluation of actual social relationships and interactions. The notions of belonging mentioned above, tapping into social embeddedness, belonging
somewhere and being
someone are combined when we bring “national” and “transnational” belonging in relationship with loneliness in this paper.
Although related, the concepts of belonging and loneliness are not the same (Hagerty et al.
1992). Loneliness is the negative evaluation, or feelings of “missing,” of a particular situation and results from a deficiency to effectuate certain standards regarding social relationships (De Jong Gierveld
1998). Belonging is a description of the social world and not a subjective evaluation. For instance, a person may belong to a certain group (“I am one of them”), but this is not an evaluation in itself, and can thus still experience loneliness. We explain variations in loneliness by studying different forms of belonging (Hagerty and Patusky
1995). We articulate belonging using two components: first, behavioural practice, like social interactions (Anthias
2013) and second, imagined belonging, reflecting being part of a bigger whole, and being at home (Antonsich
2010; Hagerty et al.
1992).
Belonging to larger society and belonging to the “own group”
Migrants’ sense of belonging can be multifaceted. We use Berry’s work (
1980) on acculturation to illustrate some options (although simplified). Acculturation is the process whereby groups with distinctive cultures engage in contact (Berry
2005). Berry (
1980) distinguishes four responses to this intercultural contact from a migrant’s perspective, along lines of orientation towards one’s own group and towards the mainstream population. These responses are integration, assimilation, separation and marginalization (Table
1) and contain different implications for well-being (Berry
2005; Sonn
2002): migrants who maintain their own identity and preserve their own culture are generally better off than those who do not (LaFromboise et al.
1993; Phinney et al.
2001).
Table 1
Four acculturation strategies
We added to this model our own focus of study: a sense of belonging. We draw a parallel on the rows in the model and show in Table
1 that having a strong (+) sense of belonging to the “own group” is corresponding to a relative preference (+) for “maintenance of heritage culture and identity” in Berry’s (
1980) model. In the columns, something similar happens. With “relationships sought among groups,” Berry aims to capture an attitude that prefers to actively seek contact with, and participate in larger society, which also encompasses “other” (minority) ethno-cultural groups. Again, we draw a parallel: relative preference (+) for engagement in larger society is analogous to having a strong (+) sense of belonging to larger society. In our model, the outcomes shown in Table
1 do not refer to a position in society where one is (un)able to participate economically or socially, or where adaption of cultural norms and values takes place (Berry
2005). They are, instead, an indication of an individual experience of being socially embedded (behaviour) and being at home (imagination), whatever the position in society. Thus, “integration” in Berry’s terminology, i.e. participating in society, in our model means a feeling of belonging to the own group as well as to larger society.
A sense of belonging is associated with less loneliness (De Jong Gierveld et al.
2015; Prieto-Flores et al.
2011). Hence, we propose that marginalized older migrants (those belonging nowhere) are loneliest. Older migrants who have a separated and assimilated sense of belonging are expected to be less lonely than older migrants who face marginalized belonging, because both have a strong belonging towards either larger society or the own group. Least lonely, we argue, are those integrated, because of their strong belonging to larger society as well as to the own group. Hypothesis 1 is: older migrants with a marginalized belonging experience the most loneliness, followed by those with assimilated and separated belonging, whereas older migrants with an integrated belonging are least lonely.
Transnational belonging
Berry (
2005) focuses on acculturation
within the country of settlement. In our model, a sense of belonging to one’s own group could be directed to both the own group
in the country of settlement (Table
1; national belonging) and the own group
in the country of origin (transnational belonging).
The complex whole of affiliations and connections that migrants employ during their lives, thereby linking their societies of origin and settlement, was termed “transnationalism” (Glick Schiller, Basch and Blanc-Szanton
1992). In the field of transnationalism, behaviour is distinguished from imagination. There, other terminology is prevalent. “Ways of being” (behaviour) is referred to as actual border-crossing behaviour and social relations. “Ways of belonging” (imagination) signals an identity component: the transnational way of life is a central element of the self, through memory, imagination and nostalgia (Levitt and Glick Schiller
2004). We do not follow this terminology, but differentiate between transnational belonging as behaviour and as imagination. Combining the two to indicate a transnational orientation is not new to the field of transnationalism (Boccagni
2012b; Levitt and Glick Schiller
2004). We argue that (a lack of) belonging
within country borders, as we have discussed using Berry’s (
1980) acculturation model, is complemented by another dimension: transnational belonging. Transnational belonging
crosses borders between countries of origin and settlement, behaviourally or imaginary.
Transnational belonging acknowledges the importance of place and could therefore play a decisive role in explaining loneliness. Surely, being in the country of settlement and immersing in the own group within this country is not the same as travelling back and forth
between both countries, keeping in touch with people
in the country of origin and being involved in that
place of origin. Place attachment has caught scientific interest and refers to peoples’ attachment to physical locations (Gustafson
2001; Rubinstein and Parmelee
1992). Place attachment increases with old age and gains importance for its impact on well-being in older age (Buffel
2015; Wiles et al.
2009). Homesickness is a common reaction to geographical relocation and captures missing social bonds, but also missing places (Vingerhoets
2005). Therefore, although people may have migrated together with their spouse and children and reside among other migrants from the same community, homesickness for childhood neighbourhoods or countries of birth still occurs (Baldassar
2008). For first-generation older migrants, this “longing for a place” is significant. They were born and raised in a different place and vivid memories thereof leading to nostalgia for the old country in its social and physical sense (Baldassar
2008). Many travel back and forth on a regular basis (Baykara-Krumme
2013). Moreover, approaching retirement age, some consider returning to the country of origin permanently (De Haas and Fokkema
2010), possibly reinforcing place attachment. A sense of belonging to the own group within the country of settlement will probably not obviate feelings of homesickness for a place. Visiting the country of origin, either behaviourally of imaginary, might be more able to fulfil this yearning for a locality or could prove to have a protective function against the psychological ill effects that acculturation poses.
To resume, we propose that next to belonging within the country of settlement, a transnational sense of belonging is a dimension of belonging, which does not blindly go along with the extent to which one feels to belong to the own group in the country of settlement or the larger society of settlement. As for other forms of belonging, we expect that transnational belonging protects against loneliness. This leads to Hypothesis 2: A transnational belonging has a protective effect on loneliness above belonging to larger society in the country of settlement and the own group in the country of settlement.
Results
National belonging
In the latent class analysis, five clusters are the best fitting solution, which are interpretable in terms of Berry’s (
1980) categorization; the fifth shows similarities with two strategies. Table
2 shows the cluster means for each variable, by acculturation category.
Table 2
Sample and cluster means per belonging variable, by acculturation strategy
National belonging: to larger society
|
Behaviour |
Dutch proficiency (3–12) | 7.40 | 2.40 | 6.22 | 6.54 | 8.88 | 8.77 | 5.66 | 53.4*** |
Social organizations (0–1) | 0.27 | 0.45 | 0.17 | 0.18 | 0.39 | 0.56 | 0.00 | 23.7*** |
Contact frequency with Dutch nonkin (1–4) | 2.42 | 0.94 | 1.49 | 1.51 | 3.11 | 2.89 | 2.80 | 160.9*** |
Imagination |
Cultural distance (3–12) | 7.59 | 2.50 | 8.91 | 8.79 | 6.15 | 5.57 | 9.33 | 70.0*** |
Self-identification Turkish/Moroccan and Dutch (0–1) | 0.07 | 0.26 | 0.09 | 0.07 | 0.07 | 0.12 | 0.00 | 2.2 |
Self-identification Dutch (0–1) | 0.12 | 0.33 | 0.16 | 0.06 | 0.14 | 0.26 | 0.00 | 7.4*** |
National belonging: to own group
|
Behaviour |
Mosque attendance (1–6) | 4.48 | 1.96 | 1.86 | 5.73 | 5.76 | 1.92 | 5.84 | 771.8*** |
Contact frequency with Turkish/Moroccan nonkin (1–4) | 2.77 | 0.86 | 2.14 | 2.35 | 3.13 | 2.53 | 3.53 | 56.7*** |
Imagination |
Cultural identity (4–16) | 8.55 | 2.62 | 7.14 | 7.20 | 9.50 | 8.91 | 9.67 | 24.4*** |
Self-identification Turkish, Kurdish, Moroccan or Berber (0–1) | 0.77 | 0.42 | 0.74 | 0.82 | 0.75 | 0.54 | 1.00 | 13.4*** |
Transnational belonging
|
Behaviour |
Frequent contact with children in Turkey/Morocco (0–1) | 0.07 | 0.25 | 0.11 | 0.07 | 0.04 | 0.08 | 0.05 | 0.8 |
Frequent contact with extended family in Turkey/Morocco (0–1) | 0.21 | 0.41 | 0.17 | 0.18 | 0.26 | 0.23 | 0.17 | 1.1 |
Visiting frequency Turkey/Morocco (1–3) | 1.73 | 0.80 | 1.72 | 1.81 | 1.69 | 1.46 | 1.99 | 4.6** |
Medical care in Turkey/Morocco (0–1) | 0.10 | 0.29 | 0.07 | 0.13 | 0.10 | 0.06 | 0.11 | 0.9 |
Imagination |
Considering return migration (0–1) | 0.26 | 0.44 | 0.21 | 0.32 | 0.29 | 0.19 | 0.25 | 1.3 |
Feelings of loss (0–5) | 3.40 | 1.49 | 3.41 | 3.55 | 3.48 | 2.72 | 3.72 | 5.5*** |
In the marginalization cluster scores indicate low belonging to one’s own group, but also low belonging to larger society. For instance, participation in activities in conjunction with the own language or cultural group is lowest and comprehension of Dutch is poor. In addition, frequency of contact with Turkish/Moroccan, as well as with Dutch nonkin, is lowest. One cluster, which we termed marginalization/separation, shows most similarity with the marginalized cluster, but with a little extra focus on the own group. Only mosque attendance differs substantially, with the marginalized/separated cluster indicating a higher score. Also, respondents in the last mentioned cluster identify themselves as Dutch to a lesser extent than the marginalized cluster.
The integration cluster contains average scores on many variables as well, but is characterized by a relatively strong sense of belonging to the own group (mosque attendance and cultural identity levels are high), as well as to larger society. Dutch proficiency is high, cultural distance to Dutch larger society is low, and there is frequent contact with Dutch nonkin, as well as with nonkin in the own group. Scores in the assimilation cluster are indicative of a sense of belonging to larger society. Scores are highest on the self-identification dummies that indicate belonging to Dutch larger society and mosque attendance is low. Further, cultural distance is lowest and participation in organizations with many Dutch members highest. The separation cluster conveys strong belonging to the own group. All respondents identify themselves as Turkish, Kurdish, Moroccan or Berber. They go to the mosque the most and participate frequently in activities in conjunction with the own language or cultural group. The separation cluster experiences great cultural distance and has most contact with Turkish/Moroccan nonkin, in comparison to the other clusters.
Transnational belonging
Next, we employ analysis of variance on the indicators for transnational belonging (Table
2). The five clusters differ in their visiting frequency to Turkey/Morocco and their feelings of loss. Those in the separation cluster have the highest frequency of visiting the country of origin, followed by those in the marginalization/separation, the marginalization, the integration and the assimilated cluster. For feelings of loss a similar pattern was observed. Thus, two of the six indicators are not independent from Berry’s categorization, but four are. Consequently, the results are not unambiguous with respect to whether or not transnational belonging is a different dimension of belonging. Because correlations between the six aspects are low (|r| ranges between .01 and .12), we additionally suggest the indicators are not unequivocally measuring transnational belonging as one concept.
National and transnational belonging and loneliness
In the regression of loneliness, the separation cluster functions as reference category for the clusters of belonging (Table
3). Supporting Hypothesis 1, the results from this analysis and subsequent analyses with other categories of reference (results not shown) reveal that the marginalization and the marginalization/separation clusters are lonelier than the integration, assimilation and separation clusters; there are no differences among the two and three clusters, respectively.
Table 3
Regression of loneliness on national belonging clusters, transnational belonging and control variables (N = 461)
Constant | 4.54 | 0.36*** | 3.47 | 0.63*** | 12.64 | 3.27*** |
Marginalization (vs. separation) | 1.54 | 0.51** | 1.63 | 0.51** | 1.16 | 0.52* |
Marginalization/separation (vs. separation) | 1.81 | 0.49*** | 1.82 | 0.48*** | 1.36 | 0.47** |
Integration (vs. separation) | −0.02 | 0.46 | 0.04 | 0.46 | −0.01 | 0.44 |
Assimilation (vs. separation) | 0.13 | 0.51 | 0.46 | 0.53 | 0.42 | 0.53 |
Frequent contact with children in Turkey/Morocco (0–1) | | | −0.06 | 0.59 | −0.32 | 0.57 |
Frequent contact with extended family in Turkey/Morocco (0–1) | | | −0.42 | 0.37 | −0.26 | 0.35 |
Visiting frequency Turkey/Morocco (1–3) | | | −0.13 | 0.19 | −0.26 | 0.18 |
Medical care in Turkey/Morocco (0–1) | | | −0.01 | 0.50 | −0.11 | 0.48 |
Considering return migration (0–1) | | | 0.43 | 0.34 | 0.33 | 0.33 |
Feelings of loss (0–5) | | | 0.35 | 0.10*** | 0.24 | 0.10* |
Age (55–66) | | | | | −0.01 | 0.05 |
Female (vs. male) | | | | | −1.02 | 0.33** |
Level of education (1–3) | | | | | −0.60 | 0.25* |
Having a paid job (vs. no paid job) | | | | | −0.29 | 0.36 |
Married (vs. not married) | | | | | −1.13 | 0.36** |
Physical functioning (0–28) | | | | | −0.05 | 0.03 |
Self-rated health (0–4) | | | | | −0.57 | 0.15*** |
Moroccan (vs. Turkish) | | | | | −0.87 | 0.31** |
Length of residence in the Netherlands | | | | | 0.00 | 0.02 |
R
2
| 0.06 | | 0.09 | | 0.22 | |
F change | 7.6*** | | 2.6* | | 7.8*** | |
In Model 2, we add variables for transnational belonging. Associations for the clusters do not change substantially. Of the transnational variables, only feelings of loss are independently associated with loneliness. A higher level of transnational belonging (as it is embodied by feelings of loss) contributes to more loneliness, as we propose in Hypothesis 2. Visiting Turkey/Morocco, which significantly differed between the clusters (Table
2), is not associated with loneliness.
Model 3 brings control variables to the fore. Still the marginalized marginalized/separated remain lonelier than the separated, although the difference is smaller compared to Models 1 and 2. Feelings of loss also continue to be associated with more loneliness. Being born in Morocco decreases the likelihood of reporting loneliness compared to being born in Turkey. The same goes for being female, being married, having a higher education and having better self-rated health.
Discussion
In this study, we investigated whether migrants’ belonging impacts loneliness. We studied three dimensions of belonging: to larger society and to the “own group” in the country of settlement, and transnational belonging. Latent class analysis on the first two dimensions of belonging identified five clusters: marginalized, marginalized/separated, integrated, assimilated and separated migrants. We found evidence that transnational belonging is a separate dimension of belonging but results also indicate that it is partly an extension of belonging to the own group in the country of settlement. Hence, we conclude that we have not unravelled whether transnational belonging is a different dimension of belonging.
We found some support for Hypothesis 1: migrants in the marginalization and marginalization/separation cluster are lonelier than migrants in other clusters that have a stronger sense of belonging, i.e. integration, assimilation and separation. Yet, the integration cluster did not show less loneliness than the assimilation or separation clusters. This means that some form of belonging to a certain group is important as a protection against loneliness, but that this does not necessarily has to involve the larger society of settlement. A strong sense of belonging to the own group (separation) is just as effective against loneliness as is integration (or assimilation).
Hypothesis 2 proposed that transnational belonging also functions as a protective mechanism against loneliness. We did not find support for this hypothesis. Where visiting frequency of Turkey/Morocco played a role in distinguishing different levels of transnational belonging between the clusters, it did not decrease loneliness. Even more so, the only significant effect we found for indicators of transnational belonging pertains to feelings of loss and was directed oppositely: more transnational belonging increases loneliness. We further find that the additional effect of personal socio-demographic resources reduces the risk of loneliness, which has been found in studies among nonmigrants (De Jong Gierveld
1998).
That transnational belonging does not decrease loneliness might be explained theoretically. Scholars have argued that being ingrained in two places makes for “betwixt and between” identities (Grillo
2007), or “double absence” (Sayad
1999). Where the relation between nearby social relationships and loneliness is rather obvious, the same apparently does not hold for distant ones. In previous research on ICT-based “co-presence” in transnational relationships, it was concluded that the golden standard remains to be face-to-face social contact (Baldassar
2008). This could mean that a transnational lifestyle contributes to homesickness and loneliness, instead of alleviating it, and causes “uprootedness,” i.e. a diminished sense of belonging.
There appears to be a second explanation for why feelings of loss result in more loneliness. Even though the variable feelings of loss captured an orientation towards the country of origin conceptually, the scale contained negatively presented items that were closely related to the concept of loneliness, such as “I miss people…” Feelings of loss were expected to be telling of transnational belonging in combination with the other variables (such as visiting the country of origin), but by itself it may have tapped into loneliness.
This brings up some methodological issues. Against the backdrop of a lack of large-scale survey data among older migrant populations (Fokkema and Naderi
2013), the LASA data provided us with a unique opportunity to systematically study older migrants’ loneliness. Nevertheless, innate to secondary data analysis as we have employed here, this has also brought some limitations regarding the relation between variables and concepts. For instance, perhaps our measurements of transnational belonging are not exhaustively grasping the concept. However, extensive research on this topic has not resulted in true consensus or clear articulation of what transnational belonging exactly entails, let alone offered unambiguous operationalization of the concept (Boccagni
2012a). Moreover, the data available did adhere to many of the elements that were mentioned in the literature in relation to transnational belonging, as described in the measurements section. Framed into the body of knowledge on explaining differences in loneliness, distinguishing an interlinked behavioural and imagined component in belonging (Hagerty and Patusky
1995; Levitt and Glick Schiller
2004) was helpful.
Further, we relied on cross-sectional data, which implies that the findings in this study are only a snapshot of reality. Belonging to a group of people, to a place and to a plethora of other social categories is not a stable feature over time (Yuval-Davis
2006). This means that the dimensions of belonging are not “end states” or ultimate destinations, nor do they relate to each other steadily over time. It further means that we were not able to test causal mechanisms. We interpreted the results congruent with the theoretical foundation for belonging to lead to loneliness (Baumeister and Leary
1995) instead of the reversed.
The acculturation model, gaining wide popularity in social science as well as in the public debate since Berry’s (
1980) use of it, has been widely critiqued (Rudmin
2003). It was put forward as too simplistic (Schwartz and Zamboanga
2008), as presenting an erroneously static and homogenous interpretation of culture with the fourfold categories opted by Berry as end states of a linear process (Hermans and Kempen
1998), thereby leaving no room for the dynamics and complexity of human development, nor for its interaction with context (Bhatia and Ram
2009). Despite valid critiques, the acculturation model succeeds to identify factors that are prominent in migrants’ experiences (Burholt
2004). Especially when explaining loneliness among older migrants, there have been repeated calls to not only study migrant’s vulnerability for loneliness (as a group), but to look specifically at differences—among others, in sociocultural embeddedness—within this group and thus explain how loneliness varies among migrants (De Jong Gierveld et al.
2015; Fokkema and Naderi
2013; Wu and Penning
2013). Moreover, we did not aim to present larger society and the own group as socially cohesive or homogenous. We approach the model from the viewpoint of individuals finding a sense of belonging and a “home” in society, rather than adapting to a specific set of cultural norms. The fact that we find groups interpretable of Berry’s (
2005) categorization also suggests that its merit in reflecting social reality is still tenable.
In the Netherlands, Turkish and Moroccan migrants have experienced conditions that are commonly found among migrant populations in Europe and beyond: on average they have a lower socio-economic position and worse health than the native population. Due to this dominant pattern of disadvantage and their similar migration backgrounds as guest workers and subsequent family reunifications, we did not study Turkish and Moroccan migrants separately (Denktaş
2011). The focus on people with a migratory past for whom a sense of belonging is not obvious applies to both migrant groups. Moreover, both are known to have strong ties with their country of origin (Fokkema et al.
2016). However, it might be the case that a sense of belonging to larger society and to the own group, as well as a transnational belonging differs between Turkish and Moroccan migrants, which would obscure our results. For example, we were left with the finding that Turkish migrants are lonelier than Moroccan migrants, which was similarly observed in a previous Dutch study (Uysal-Bozkir et al.
2015).
Despite these limitations, this study has various innovations. We have connected a theory on loneliness provoking factors with theoretical insights on migrants’ social integration. Therefore, it provides new insights on variations in loneliness among migrants, which has repeatedly been called for. Second, we added a transnational lens, which, until recently, has mostly been neglected in research on ageing (Torres
2013). Third, we empirically added a transnational dimension to a two-dimensional acculturation model, something that was previously suggested by Van Oudenhoven et al. (
2006). Although theorizations and qualitative research on transnational orientations are manifold, we find mixed evidence for transnational belonging as a separate dimension of belonging and that this was not vital for understanding differences in loneliness. Yet, transnational belonging is apparent for all forms of belonging distinguished in this paper. Consequently, more attention and prominence should be given to transnational belonging in empirical research among older migrants in general, and further understanding on how it relates to well-being specifically.