Differentiated embedding: Polish migrants in London negotiating belonging over time

ABSTRACT Developing on Granovetter’s classic work on embeddedness in systems of social relations, this paper proposes the concept of ‘differentiated embedding’ to explore how migrants negotiate attachment and belonging as dynamic temporal, spatial and relational processes. When Poland joined the EU in May 2004, the large flow of migrants to the UK was perceived by many migration researchers as heralding a new form of transient mobility associated with short-term, temporary and circular migration, and high levels of transnationalism. Relatively little attention was paid to how these migrants were integrating in local contexts. Based on 20 in-depth interviews and network mapping with Polish migrants, resident in London for a decade, I examine why participants extended their stay and how their decisions were shaped by interpersonal relationships locally and transnationally. London as a ‘superdiverse’, global city offers place-specific opportunities for building networks and developing processes of embedding. Nonetheless, a focus on networks risks overlooking the wider structural context in which migrants live and work. Thus, I argue, there is a need for a differentiated concept to capture the nuanced interplay of structural, relational, spatial and temporal embedding. This concept not only captures multi-scalarity and multi-sectorality but also levels of belonging and attachment.


Introduction
Initially, as probably most people say, we set up for ourselves one yearto see how things are progressing, where we're going to be, if we're going to like it, and then we're going to make our minds up and decide what to do. (Dominik 1 ) In line with the findings of other studies (Eade, Garapich, and Drinkwater 2006), the vast majority of my participants described their initial migration plans as somewhat short term. Like Dominik, above, most people spoke about coming for a year: 'I was planning to stay for a year, only for a year' (Patryk). Nevertheless, the 20 people in my study had gradually, sometimes unconsciously, extended their stay over time. I am curious to understand why and how temporary plans turn into longer term stays. I want to explore the processes that lead to extensions of migratory periods. In their narratives, some people appear indeterminate and short-term plans, have stayed and settled in Britain. Their experiences suggest that while mobility rights may provide options for circularity, transience and temporariness, these rights also confer opportunities to adjust migration plans and extend the stay. The impact of Brexit on curtailing mobility rights is likely to be highly significant in shaping settlement plans, but is beyond the scope of this paper.
Having observed migration pattern among waves of Irish migrants, I was curious to see if post-accession Polish migrants would exhibit similar patterns. I was unconvinced by the widespread assertions among migration scholars that intra-EU mobility rights would herald new forms of migratory movement (for a discussion see Glorius, Grabowska-Lusinska, and Kuvik 2011). My initial research with Poles ) already began to find evidence of family reunions, family formations and growing numbers of Polish children in British schools, suggesting patterns of longer term stays were beginning to emerge.
That is not to suggest that extending the stay can be understood through an integration versus transnational dichotomy. As other researchers have noted, settling into the new country does not necessarily mean losing connections with the home country (Erdal and Oeppen 2013). It is necessary to explore negotiations across multiple sites simultaneously (Erdal and Ezzati 2015;Grzymala-Kazlowska 2016). In this paper, drawing on previous work , I propose embedding as a conceptual device for understanding dynamic, complex, multidimensional and spatially differentiated processes of attachments.
Embeddedness has been widely used across a range of disciplines to explain migration: 'features of social embeddedness are among the most influential factors for migrant settlement, onward movement and return' (Korinek, Entwisle, and Jampaklay 2005, 794). However, embeddedness has been described as a vague and 'fuzzy' concept, lacking in precision and clarity (Hess 2004). There have been calls for a clearer understanding of the qualities of embeddedness and the multidimensional nature of ties, as well as more research on the dynamism of this process over time (Hite 2003(Hite , 2005. Rather than a static notion of embeddedness, we can suggest the more active notion of embedding . In this paper, drawing on Granovetter's classic work on embeddedness in systems of social relations, I develop the concept of 'differentiated embedding' to explore how migrants negotiate attachment and belonging as interconnected temporal, spatial and relational processes. I suggest that a 'differentiated' notion of embedding is useful in understanding the dynamic processes through which migrants negotiate attachments and belonging to varied degrees in different social and structural settings. As noted by Grzymala-Kazlowska and  in the Introduction to this special issue, 'integration' has diffuse definitions across the academic literature, while within official policy discourse in the UK, and other European countries, it has become increasingly normative and dominated by an 'assimilationist stance' (Phillimore 2012, 529; and Introduction to this special issue). This paper engages with the themes of the special issue, as outlined in the Introduction (Grzymala-Kazlowska and Phillimore 2017), by presenting differentiated embedding as a way of thinking about the nuanced details of migrants' experiences of engagement with the people and places that make up their social world, and in a way that may mitigate often fixed and narrow concepts such as 'integration' .

Differentiated embedding
Embeddedness can be defined as 'social relationships that foster a sense of rootedness and integration in the local environment' (Korinek, Entwisle, and Jampaklay 2005, 780). However, many authors highlight the term's 'theoretical vagueness' (e.g. Portes andSensenbrenner 1993, 1321). As the concept has developed across many disciplines, its meaning has become more diffuse. Noting this 'fuzziness', Hess (2004) observed that each new application offers a different interpretation of who is embedded in what. He cites Jessop as saying that embeddedness is 'an increasingly popular but confusingly polyvalent concept' (Jessop in Hess 2004, 167). Although many authors begin by acknowledging the classic work of Polanyi in coining the concept (Beckert 2007), it is Granovetter who popularised the concept.
In proposing embeddedness, Granovetter sought to carve a path between under-socialised (Hobbesian) and over-socialised (Parsonsian) views of the choices people make in relation to economic behaviour: Actors do not behave or decide as atoms outside a social context, nor do they adhere slavishly to a script written for them by the particular intersection of social categories that they happen to occupy. Their attempts at purposive action are instead embedded in concrete, ongoing systems of social relations. (1985,487) While focusing mainly on economic behaviour, Granovetter's work has been particularly influential among migration scholars. Decisions about where and when to migrate as well as processes of settlement have been shown to be embedded in social networks (Korinek, Entwisle, and Jampaklay 2005;Nannestad, Svendsen, and Svendsen 2008). However, Beckert (2007) argues that the focus on embeddedness within social networks risks overlooking wider, structural (macro) dimension of society. As Portes and Sensenbrenner argue it is necessary to specify just how 'social structure constrains, supports, or derails individual goal-seeking behaviour ' (1993, 1321). As noted, Hess (2004) is critical of the 'fuzziness' of embeddedness as a concept. While acknowledging the role of networks (the relational), he also highlights two other key dimensions that comprise embeddedness: societal (an actor's belonging within wider socio-political structures) and territorial (the ways in which an actor is located in particular places). Taken together these three dimensions of embeddedness (relational, societal and territorial) are closely knitted together and form the space-time context of socio-economic activity (Hess 2004). Thus, in discussing migrants' embeddedness it is essential to pay due attention to the structural opportunities, but also obstacles, that may be encountered.
Geographers like Hess (2004) and others (Findlay and Stockdale 2003;Robinson 2010) help to add a spatial lens to the concept of embeddedness. It is necessary to acknowledge the materiality of place. The resources and opportunities available to migrants may depend in part on the socio-economic, cultural and physical particularities of the local contexts in which they live and work (Hickman, Mai, and Crowley 2012). In other words, different 'place effects' (Robinson 2010) present different opportunity structures. While my focus here is largely on social networks, I am cognisant of the importance of socio-economic structures within specific territorial contexts as highlighted throughout this paper. In the context of intra-EU migration, it is interesting to consider how opportunity structures may be racialised in specific ways (Garner 2006) and how nominal whiteness can become refracted in different ways within particular contexts (Fox, Moroşanu, and Szilassy 2012). However, that discussion is beyond the scope of this particular paper.
In contributing to theoretical understanding, this paper challenges notions of embeddedness as a static, achieved state. Granovetter also warned against exaggerating the degree of embeddedness, suggesting that 'culture is not a once and for all influence but an ongoing process, continuously constructed and reconstructed during interaction ' (1985, 486). Thus, Granovetter was sensitive to the dynamics of embeddedness.
As Jon Mulholland and I argued at length elsewhere , embedding needs to be understood as a dynamic process. Maintaining relationships requires effort (Bourdieu 1986) and thus one cannot assume that, once established, embeddedness remains fixed. Individual biography reveals the dynamism of relationships through the life course as an interplay between the spatial and the temporal (Findlay and Stockdale 2003).
In this paper, I suggest embedding is not only contextual and dynamic but also differentiated. Given varied place-specific opportunities, embedding may be negotiated differently across particular sectors of society. Korinek, Entwisle, and Jampaklay (2005) identify four sectors in which migrants may embed differently: household, workplace, neighbourhood and wider community. Although there may be overlap across these sectors, they can also be experienced quite differently arising from varied opportunities to negotiate attachment and belonging. For example, a migrant may be actively embedding in the labour market (workplace sector) but not feel any sense of connection in the residential area (neighbourhood sector). Taking this notion of sectors further, my research used a visual tool, described below, to analyse how migrants negotiate embedding to varying extents with particular people across different settings.
Migrants are not only negotiating embedding at the local level in the destination society, of course, but also are usually connected to spatially dispersed places and people. Thus, 'migrants' personal communities may be far-flung and may traverse geographic social settings' (Korinek, Entwisle, and Jampaklay 2005, 797). Embedding as a dynamic process can be extended to take account of this transnational dimension and Hess's work is significant in emphasising the 'multi-scalarity' of relationships (2004,181). Thus, transnational mobility need not represent dis-embedding (Hess 2004, 176). However, as recent research shows, maintaining personal ties across dispersed sites can be difficult over time (Ryan, Von Koppenfels, and Mulholland 2015).
In contributing to theory building, I argue that a differentiated conceptualisation of embedding may help to go beyond a simplistic, one-dimensional, 'all or nothing' view of migrant 'integration'. In addition, I suggest, the concept is nuanced and not only captures multi-scalarity and multi-sectorality but also depths of attachments and belonging. In understanding and distinguishing depths of embedding, the work of Hite (2003Hite ( , 2005 is relevant. She argues that 'relational embeddedness is still assumed to represent a single, dichotomous construct'one is either embedded or one is not (2003,13)leading to the erroneous conclusion that all embedded ties are alike. Her work highlights different degrees of embeddedness across various social and relational settings.
To understand these dimensions, it is necessary to pay greater attention to the nature of the relationships within specific arenas. Hite identified key relational characteristics associated with different degrees of embeddedness including loyalty, obligation, sociality, trust, ease, effort, frequency and duration (2003,22). She then constructed a spectrum of typologies ranging from 'hollow' and 'functional' through to 'full' embeddedness. According to Hite, 'hollow' is defined by low levels of frequency, short duration and limited trust;'functional' represents increased trust and social ease, while 'full embeddedness' is associated with high levels of trust, loyalty and obligation. She argues that embeddedness may develop through these typologies over time by increasing effort, frequency and duration of contact (2005). Although Hite was focusing primarily on the economic sector and business relationships, her approach and typologies have wider application (see . I build on her work by examining embedding within different sectors. Using sociograms, I consider the nature of relationships within these sectors to explore the differentiated embedding in which migrants were engaging.
In this section, I have argued that rather than embeddedness as a static, achieved state, it may be more helpful to use the concept of embedding to capture the dynamics and contextuality of temporal, spatial and relational processes. In addition, I have argued that embedding needs to be understood as differentiated. Rather than a simple, static binary of embedded or not embedded, there are varying degrees of attachment and depths of trust and reciprocity between actors within various social domains. The paper develops over four sections in which I present my data and then conclude by reflecting upon the usefulness of embedding as a conceptual framework for understanding the dynamism of migrants' varied attachments over time especially as the implications of Brexit begin to unfold. But first I begin with a brief methods section in which I explain my narrative and visual research design.

Methods
Building upon my earlier work on Polish migrants ), in 2014, on the 10th anniversary of EU enlargement, I undertook new qualitative research involving 20 migrants who had been resident in the UK since EU accession (Ryan 2015). This was not intended to be a representative sample. Participants were recruited using a range of techniques including convenience sampling through Polish networks in London as well as snowballing through Polish contacts. The criterion for selection was that participants needed to have arrived in London approximately 10 years earlier. The majority of the participants (17/20) were women. The average age was 36 years. The mean year of arrival was 2005, with the majority of participants moving to the UK between 2004 and 2007. Thirteen were married, five divorced and two were single. There was an even split between those with and without children, 10/20. All but one participant were graduates. Most arrived from Poland as graduates, though many did further study post-arrival in London (see research report, 2015).
In order to generate richer data on relational embedding, I decided to use a combination of interviews and sociograms so both oral and visual data were collected. I used a simple paper-based target sociogram in combination with biographical interviews to explore how participants created, sustained and changed social relationships over time. The simplicity of Mary Northway's original 'target' design sociogram makes it ideal for collecting data 'in an intuitive and easy way' (Carrasco et al. 2008, 9). My sociogram consisted of three concentric circles divided into four quadrants (friends, family, work, neighbours/hobbies/other) and was adapted from Hersberger (2003; see also Tapini, forthcoming). Defining network boundaries is a considerable challenge for network researchers (Clark 2007, 18). I sought to avoid delimiting network boundaries by allowing participants to decide how many contacts they wish to add. Clearly, this type of sociogram is not a neutral tool for capturing a pre-existing network (Ryan and D'Angelo 2017). Its layout and design are important as these may influence how the data are represented (Huang, Hong, and Eades 2006, 3). In addition, the questions asked by the interviewer may influence how the network is visually depicted and verbally explained (Ryan, Mulholland, and Agoston 2014).
Completing a sociogram in the context of a biographical interview provides an important opportunity to consider how networks are represented differently using these two distinct data collection techniques. The interview began by asking about migration processes. Interpersonal relationships were usually central to this story. Friends, partners and relatives were woven through the narrative. I also asked how relationships evolved and changed over time. Thus, networks were discussed in some detail before the sociogram was introduced. This usually occurred about 15 minutes into the interview and participants took approximately 15-20 minutes to complete the sociograminterspersed with discussion and prompts. The sociogram was left on the table while the interview moved on to other topics, such as future migration plans. It is noteworthy that many participants went back to the sociogram throughout the rest of the interview to add extra alters.
I conducted an integrated analysis in two phases. Firstly, I conducted a narrative analysis of a complete interview transcript and sociogram, focusing on how a participant tells his/her story in words and images. The second phase of the analysis was conducted in NVIVO and involved thematic coding across the full dataset to identify the extent to which specific nodes were shared among participants. This enabled an analysis of particular patterns of networking, such as what factors facilitated new social connections and challenges of sustaining old ties in contexts of migration. Thus, I suggest that just as visual and narrative data are collected together, there is a strong rationale for analysing them together through an integrated method. This analysis captures the dynamic interplay between how people talk about and visualise their social ties.
Place-specific opportunity structures everybody I know from Poland, every time they come over here, they're like 'just go there for a year, two years' so it's very easy to come over but it's very difficult to go back … once that year's over, it's like another year, another year, another year, another year, yeah so … cos life is easier here. (Patryk) Several people described their migration as almost accidental, happenchance, holidays that gradually extended into longer term, though often indefinite, stays. The process of coming to London was frequently narrated through stories of friends who had already or were about to travel to Britain. Marika's story was typical of these 'happenchance' migration narratives. Having failed to secure a university place on her preferred degree programme in Krakow, she decided, at short notice, to accompany her friend who was going to London. She described her decision as 'really, really quick' because 'if I will think about it for a bit longer I probably wouldn't do it' (Marika). She anticipated staying for a year to learn English and then reapplying to a university in Poland. So moving to London was a way to 'kill time' and have 'a bit of adventure'.
The narrative of adventure was common among those who arrived when young and single. Angelika similarly described her decision to move to London as an adventure, associated with parties and fun, adding: 'I didn't actually plan to stay that long.' Like several other young arrivals, Patryk had decided to take a 'gap year' from his university studies in Poland. He intended to make some money in London: 'I just wanted to go and may be have fun for a year.' Initial intentions of participants appear to fit with notions of transience, temporariness, circularity and liquid migration (Glorius, Grabowska-Lusinska, and Kuvik 2011). Nevertheless, all of them still live in Britain. As Erdal and Ezzati (2015) note, despite uncertainty of initial migration plans, over time migrants may engage in processes of settling in so initial return intentions may not reflect actual return behaviour. Part of the reason for extending the stay may be changing projects often associated with place-specific opportunity structures. Young people such as Patryk, Sonia and Marika, who originally moved to London for fun, discovered opportunities to study for free in the UK, before the introduction of university tuition fees. In interviews it was challenging to undercover precisely why and at what point these participants had decided to explore study opportunities in London. As Amit and Riss note, motivations for migration are multifaceted and 'may change over its course' (Amit and Riss 2013, 65) and participants may not be able to clearly explain why they made particular decisions at specific times. Sonia, having returned to London after a brief stay in Poland (so-called double migration White 2014), decided 'out of the blue' to investigate university programmes here: 'if I am educated here it always looks better, even if I go back home I will be able to get a better job there as well' (Sonia). Thus, deciding to embark on a three-year degree course in London was not necessarily a commitment to long-term settlement but a way to keep options 'open' including acquiring additional qualifications to facilitate return to Poland. However, as shown below, doing a British degree provided new opportunities for further embedding.
For these three young people, and other participants who completed university courses in Britain, acquiring a British degree led to additional in-place career opportunities while also further embedding them in specific educational and employment structures. Marika completed a science degree in London: I came here with some English knowledge, I spent a lot of time learning words, like specific, in science we've got a lot of procedures … learning all the words for each thing, it took me a long time. So it's like after 3 years when I got the confidence that I know scientific words right now, it wasn't even an option for me to just, you know, just go back to learn everything in Polish.
Like education, employment may be regarded as a 'means' as well as a 'marker' (Ager and Strang 2008) of embedding in a specific socio-economic context. Marika's sociogram depicted her embedding in the employment sector. Her work quadrant contained more ties than either her family or neighbour/hobbies sectors did. Like several other participants, particularly those without children (see the section below), work provided not only professional connections but also friendship and socialising opportunities. The ethnic composition of those ties tended to reflect the diversity or superdiversity of the working environment (Grzymala-Kazlowska 2016; Grzymala-Kazlowska and Phillimore 2017). Typical of many London work places, participants tended to work with colleagues from many nationalities. Hence, ties in the work sector of the sociogram were usually the most ethnically diverse. These professional and relational ties reflect but also reinforce embedding as place-specific structures provide opportunities to develop new local ties. Clearly, London as a global city provides a very specific social context for migrants. Research in other British cities also suggests that migrants may have particular opportunities to develop place-specific social ties (see Grzymala-Kazlowska and Phillimore 2017).
Investment in career development in the London context may have taken considerable time, resources and personal sacrifices. While participants who gained British university degrees appeared to have much faster and easier routes into the local labour market, acquiring entry-level jobs commensurate with their qualifications (Szewczyk 2014), those who had completed their studies in Poland often had longer and more challenging routes into professional occupations in Britain (Trevena 2011).
Gabi experienced initial de-skilling and worked in a restaurant while completing a number of training courses, such as IT, thus re-validating her cultural capital through acquiring locally recognised qualifications, and then got a job in a bank (for a fuller discussion, see Ryan 2016). She was not optimistic about her career prospects in Poland: 'it's all about contacts in Poland and for example the job I do here, how could I translate and transfer my knowledge from here to Poland?' Having studied and worked for so long in the UK, she was concerned about the potential risks of returning to Poland: 'I don't want to start from the beginning again.' The experiences of participants like Gabi and Marika, above, highlight the non-transferability of credentials and develop upon Erel's (2010) observation that cultural capital cannot be simply carried across borders in a 'rucksack'. While it is difficult to bring cultural capital to the destination country, it may also be difficult to bring newly acquired cultural capital back to the country of origin . This applied not just to knowledge acquired from degree courses but also work-based practices. This suggests how short stays and temporary mobilities can be gradually extended over time as migrants begin embedding in place-specific opportunity structures. This point illustrates the spatial and temporal, as well as the structural, dimension of embedding; participants like Gabi were gradually embedding in the British labour market but at the same time dis-embedding from the Polish labour market.
But that is not to suggest that embedding can be understood simply as a structural process. Participants' narratives and sociograms reveal the dynamic interplay of numerous factors. When Agnieszka initially arrived in London in 2004 she was working in a care home and felt frustrated by her low-level job. Over the following 10 years, she undertook several training courses and at the time of the interview in 2014 was working in the voluntary sector and training to be a psychologist. It was apparent that her work and professional identity had become very important to Agnieszka. Denoted by high centrality, she appeared to be deeply embedding within her employment sector and, through her ongoing training and 'supervision group', was deriving high levels of support and personal fulfilment. This is clearly illustrated in her sociogram which shows significant overlap across friendship and work sectors (with arrows connecting friends and work colleagues). The fact that her colleagues had become close friends suggested the density and intensity of her social networks. Drawing on Hite's (2005) typology, explained earlier, we see how shared interests, sociality and frequency of contact enabled Agnieszka to develop deep, trusting ties with colleagues which both reflect and reinforce embedding in professional relationships. As discussed below, relational embedding was often a key reason why people extend their stay and how they negotiate belonging in London.

Relational embedding
While acknowledging the structural and spatial dimensions of embedding, I am also keen to explore the relational aspects (through networks) and how these facilitate migrant embedding in different places. Martyna was one of several women whose trip to London was motivated by 'romantic reasons'. She had begun a holiday romance with a Polish man, based in London, during his visits to Poland. They carried on visiting each other for some time but when it was apparent that he did not wish to return permanently to Poland, she decided to 'give it a go' in London to see how they would get along together. Upon arrival, in 2005, she had no clear plans about how long she would stay or if the relationship would develop into something stable. Ten years later, she is married with two sons and feels happy with her decision to come to London.
But of course participants were not just embedding in romantic relationships. The sociograms also revealed the role of children in creating local friendship networks. For example, Ewa arrived in London as a young, university graduate in 2002 and experienced initial de-skilling through work in the care sector. Since then, having completed a number of training courses, she developed her career as a data analyst. Ewa married an English man, had two children and moved to an affluent suburb in south London. She described how having children completely changed her sense of belonging in British society: If you have children you have to participate in everything that's happening in society. You have to go to the same hospitals, the same playgroups, start the same schools, parent evenings, and you really get more and more understanding of what's happening. You have to participate … and then you are really settling in because you know your children will grow up here and they will start their families here so you need to be a member.
Other participants also spoke at length about how their social circles had expanded significantly through their children and their involvement in schools. Ewa also commented on the ethnic diversity of the social ties formed through playgroups and school. Thus, as discussed elsewhere , child-based sociality provides opportunities for migrant parents to engage with London's superdiversity.
Klaudia's sociogram (Figure 1) illustrated an interesting pattern of interconnected social ties. Having initially been a full-time mother, she gradually built up friendship and professional connections through her son's school. Starting as a volunteer at the school, she later stood for election to the board of governors: 'I was involved with my son's school, I was on the governing body and you know, getting to know the school and being really-really involved.' This involvement developed further when she started to work as a classroom assistant and then decided to re-train as a teacher. At the time of interview, she had recently secured a permanent teaching job in a local secondary school. School appears in several ways on her sociogram. Her friendships were formed through school. As a teaching assistant, it is hardly surprising that her work colleagues were also school-based. In addition, she had many ethnically diverse local neighbourhood ties through the parents of her son's friends. The overlaps and interconnections between these linkages were illustrated by her best friend 'AC' who appeared (with a connecting line) in both work and friendship quadrants. As Figure 1 illustrates, the four quadrants were fairly equally populated and most ties were located in the two inner circlesthis may again reflect the level of overlap between work, friends and neighbours around school. The sociogram appears to suggest an even level of embedding across the sectors. The combination of narrative and visual data suggests that Klaudia was engaged in 'full' embedding (Hite 2005) in London as her relational attachments evolved spatially and temporally.
The experiences of these participants suggest how local, child-centred activities can create new opportunities not only for social embedding in ethnically diverse local neighbourhoods through developing new friends but also to gain a deeper involvement in local activities and a better understanding of systems and structures in British society, and in some cases, leading to new career opportunities. However, migrants are not only involved in forming new relations in the local context, most are also actively involved in long-distance ties in the country of origin.

The dynamism of transnational ties
While most people maintained strong and enduring ties with family in Poland, their links to friends appeared to be less stable and less certain. Some participants put 'friends in Poland' in the inner circle of the sociogram, denoting a very close relationship. Klaudia: 'I've got my best friend, of course we are, you know, in touch all the time.' But the centrality of these ties did not necessarily mean regular communication. A noticeable pattern in the data was exemplified by Sylwia: For some people even if they did not see, or even contact, each other regularly, these friendships had a deep emotional attachment. They were rooted in childhood (Morosanu 2013) and maintained in 'one's heart' or in 'one's mind'. However, these long-distance friendships may be based on shared links from the past rather than the present. While some participants were confident that these long-standing ties were worth maintaining, others were less sure.
Several participants explained their dwindling friendship networks in Poland in terms of lack of time. Wiktoria stated: 'I haven't got many friends in Poland, I mean who I really keep in touch with … we're all busy with our lives.' Particularly on visits back to Poland, there was not enough time to catch up with everyone and most people prioritised meeting family rather than friendship groups. As Sonia noted: if I went back to my hometown there is a couple of friends … if I made a phone call they'd probably be like 'yeah, of course, come over and lets have a coffee and catch up' but I sort of never have time when I go back'.
Clearly, Sonia felt no incentive to maintain these hometown friendships. However, in the course of completing the sociogram, she remembered another friend: Actually I have one friend who is in Poland, and I am in touch with him, I just thought, maybe I'll put him sort of here (adds to sociogram) … that's the person I keep in touch but we don't sort of, we don't keep in touch like you would with friends here, so on a regular basis, so we kind of talk to each other every few month … So I'm putting him here (Sonia) So it seems the act of completing the sociogram triggered a memory, an association with someone in Poland. However, that relationship did not appear to be very important to Sonia, perhaps she added him because she felt the need to show one remaining friendship tie with Poland. As argued already, the sociogram is not a neutral tool for collecting data but actually shapes how people recall and represent their social relationships (Ryan, Mulholland, and Agoston 2014).
Like Sonia, Dominik also spoke about prioritising particular people during visits to Poland: Because once I'm going for say a week, I've got my parents, my brother, my niece, probably I should put my niece here as well (adding her to sociogram) and then I've got my really good friend who lives in Poland, and if you're going for a week, then you have 5, 6 or 7 people.
He did not have time to see other former friends: I recognised that we don't have much in common any longer. Because I am a different person than I've left 11 years ago Poland, and they are different people than I knew 11 years ago. You know, now they've got their families … and they've got their new friends.
Dominik reflected that not only had he changed, is a different person, but his former friends in Poland had also changed, they now had families, children, but also new friends. This is perhaps an acknowledgement that it is not only migrants who move, but former friends also 'move on' and make new friends, perhaps there is no longer a place for the migrant in their former friendship networks which may result in 'hallow' embedding (Hite 2005). As a divorced man with no children, Dominik's sociogram revealed his closest family ties were all in Poland. However, his closest friendship ties were almost all in London. This chimes with Hess's (2004) observation that migrants are maintaining and creating relationships across various geographical scales. But not all spatially dispersed ties are of equal intensity. On the sociogram Dominik's 'old friends in Poland' were located in the outer circleno longer close but perhaps included because of their association with the pastfor 'old time sake'. This illustrated the multidimensionality of his embedding as his relational ties were spatially dispersed and temporally dynamic.

Ambiguous embedding
As noted, earlier, the concept of differentiated embedding offers a way of understanding the complex, dynamic and uneven dimensionalities and spatialities of attachment. The ambiguities of embedding are particularly apparent in the case of Sylwia who spoke about feeling 'disconnected'. Although she had a well-paid, permanent job in her chosen profession, after years of study and hard work, the breakup of her marriage had left Sylwia in a state of uncertainty about her future. As a single, working mother, with two children, she clearly felt bereft of wider family support networks: 'I miss my family, I feel quite lonely here … I just feel very kind of, very disconnected.' Sylwia's sociogram clearly illustrated her sense of disconnectedness. In the words of Hite (2003), it suggests 'hollow' embedding. Of all the participants in my study, Sylwia's sociogram was among the sparsest (see Figure 2). Although a senior professional in a large organisation, she did not appear to have many ties to people at work and included only a few colleagues on her sociogram. Despite working in similar professions, in contrast to Agnieszka, discussed earlier, Sylwia did not appear to be embedding in any professional or workrelated networks. Her family quadrant contained only her two children in London and her parents and brother in Poland. Her neighbours/hobbies quadrant was completely unpopulated, which was unusual for participants with children. Working full-time and looking after her two sons, she did not have the time or perhaps opportunity to get involved in hobbies or local initiatives. Thus, despite living and working in superdiverse environments, migrants may lack the time, motivation or opportunities to build up new relationships, a point also noted by Phillimore, Humphris, Khan (2017).
Embedding has a strong subjective dimension (Davids and Van Houte 2008). One cannot simply measure degrees of embedding off visible markers such as employment. Feelings of belonging and attachment are crucial in negotiating embedding. This point was illustrated differently by Mateusz. When he arrived in London as a young, recent graduate, he made the pragmatic decision to train as a nurse because it enabled him to enter a skilled profession, while earning a salary during training. Now 10 years on, he felt frustrated: 'I never fully identified with this ethos of nursing … I don't feel like I've been made to do nursing.' Although Mateusz was unsettled in his current careerwhat could be described as 'functional' embedding (Hite 2005)he was exploring other career options and training opportunities, and saw his future in London: 'I feel like I aspire to be a Londoner.' This statement points to the identity aspects of embedding through feelings of identification with the place of residence. Mateusz saw becoming a Londoner as a gradual process, to which he aspired but had not yet fully achieved. Now married to a Polish woman with two young children he explained: 'I never had an adult life in Poland, this is where I made it for myself in a way, you know, this is where I've been happy, this is where my family is.' Mateusz had no plans to return to Poland. Hence, while he did not consider himself to be successfully embedding career-wise, on a personal/familial level he felt very settled in London.
The narratives of Mateusz and Sylwia illustrate differentiated embedding across multisectors (Korinek, Entwisle, and Jampaklay 2005) and spaces (Hess 2004) but each for different reasons. Sylwia could be seen as successfully embedding in the work sector due to her senior professional occupation. However, she was not embedding at all in her local neighbourhood sector as demonstrated by the complete absence of any local ties in her sociogram. By contrast, Mateusz was not embedding in his work sector but was evidently embedding in his local neighbourhood through strong friendship ties that helped him feel settled in London. Hence, although both had permanent jobs commensurate with their qualifications, their sense of attachment and belonging in London differed enormously. These findings call to mind earlier observations by Phillimore (2012) and also (Phillimore, Humphris, and Khan, 2017) about complex and uncertain 'interlinkages' between different domains of integration. I suggest that the notion of differentiated embedding allows us to think critically about the ways in which migrants may negotiate belonging across different domains of society. In this way, differentiated embedding captures the complex interplay of the personal/subjective (micro), relational (meso) and structural (macro) dimensions of migrants' experiences.

Conclusion
Insufficient attention has been paid to how intra-EU migrants negotiate belonging and attachments over time (Collett 2013;Grzymala-Kazlowska 2016;Koelet, Van Mol, and De Valk 2017). In attempting to address this gap in the literature, my work examines how intra-EU 'temporary stays' were gradually extended over time while, at the same time, remaining somewhat undefined; perhaps not permanent but clearly not transient either. This process provides a useful opportunity to examine the various factors that enable gradual extensions of the stay in the destination society as well as factors that may hinder return to the country of origin. In this paper, I have sought to contribute to theory building by proposing the concept of differentiated embedding to examine these complex, uneven and dynamic processes.
Although embeddedness emerged in economic sociology (Granovetter 1985), geographers have been influential in adding a spatial dimension to explore migrant negotiation of attachments across several places simultaneously (Hess 2004). In previous work, it was proposed that rather than a static, achieved state of embeddedness, it would be more useful to adopt the dynamic concept of embedding as a way of capturing the ongoing activity and effort involved in this process . While a normative notion of 'integration' dominates policy discourse and practice, usually conceived of as linear, unidirectional and monolithic (Phillimore 2012), we may suggest that embedding offers a less linear and less normative way of understanding multidimensional and multi-spatial interpersonal relationships over time . The innovative method of combining in-depth interviews with hand drawn sociograms (Ryan, Mulholland and Agoston 2014) allows for greater detail on content and structure of networks. These data are also useful in exploring the ethnic composition of social ties in superdiverse settings (Ryan 2016).
However, that is not to imply that embedding can be understood simply in terms of relationships and networks. A focus on social networks risks overlooking the macro dimension of society (Ryan and D'Angelo 2017). Clearly, interpersonal social ties are an important part of developing attachment in a new environment, but we cannot overlook wider structural contexts. As EU citizens, the participants in my study enjoyed rights to mobility and employment. Nonetheless, they also face many practical challenges such as language barriers, recognition of qualifications and translating credentials (Grzymala-Kazlowska, 2016). In addition, they encountered obstacles such as loneliness and the unfamiliarity of a new environment. It is apparent that London as a superdiverse, global city offered place-specific opportunities for building networks and developing processes of embedding. Nonetheless, it took time and effort to begin to adjust and navigate their new context.
I have proposed differentiated embedding as a framework for bringing together micro, meso and macro dimensions of belonging, relationality and opportunity structures. This differentiation is important not only for understanding various dimensions but also diverse depths or degrees (Hite 2003(Hite , 2005 of embedding. People need not necessarily embed to the same extent and in the same way across different domains of society. Using a visual tool and in-depth interviews, I collected data on how migrants navigate specific domains including employment, neighbourhood, familial and friendships ties both locally and transnationally. In this way, my findings show that rather than a simple, one-dimensional form of embeddedness, migrants are negotiating embedding to different degrees across various domains. My work also illustrates that embedding is neither unidirectional nor irreversible. Migrants do not simply continue to embed over time; on the contrary, it is apparent that life eventssuch as divorce or bereavementmay result in ambiguous or even reverse embedding (dis-embedding).
In keeping with the themes of this special issue, my analysis has implications for how we understand migrant experiences of settling in and negotiating belonging over time. Clearly, different groups of migrants have varied rights and entitlements and may encounter diverse barriers in settling into a new society, as will become even more apparent as the UK prepares to leave the EU and mobility rights are brought into sharp focus. I suggest that differentiated embedding may provide a useful analytical framework for researching how EU migrants, living in the UK, will respond to Brexit. Adding civic embedding to economic, relational and spatial embedding is likely to become especially important. Hence, this conceptual framework may help us to understand the dynamism, complexity, multidimensionality and multi-sectorality of migrants' decision-making in contexts of major structural change.
Note 1. The names of all participants have been changed.

Disclosure statement
No potential conflict of interest was reported by the author.