Telling network stories: researching migrants’ changing social relations in places over time

This article is situated at the nexus of migration research and qualitative social network analysis (SNA). While migration scholars often engage with networks simply as metaphors, I go further by examining how a thorough engagement with qualitative SNA can contribute to migration research in at least three key ways. First, exploring changing relational ties over time and across different places, including transnationally, I demonstrate that qualitative SNA offers new insights into how migrants make sense of these dynamic relationships. Second, following Dahinden (2016), I examine how using networks as a data collection method can help to unsettle the a priori ethnic lens in researching migration. Moreover, building on the pioneering work of network scholars such as Mische and White, I aim to make a methodological contribution by analysing how social networks are co-constructed as stories and pictures in research encounters.

Migration researchers have long been interested in the role of social networks in the trajectories and experiences of migrants (Boyd 1989). Nonetheless, 'social network methodology has rarely been at the forefront of such studies until relatively recently. Instead, social networks were often used either metaphorically or only describing dyadic relations' (Bilecen et al. 2018: 1). Rather than being simply self-explanatory, it is necessary for researchers to consider and clarify what they mean by 'networks' and also how this concept is applied in research contexts (Knox et al. 2006). As Mehra et al. (2014: 312) argue, 'social networks lead a famously dual-existence … on the one hand, recurring and relatively stable patterns of interaction and sentiment connecting individuals to each other … on the other hand, social networks are also mental (re)constructions of social relations, some real, some imagined. ' Indeed, as observed by early network scholars such as Burt (1982) and Krackhardt (1987), the network data collected by researchers are not so much an accurate reflection of 'reality' as the product of perception and cognitive processes. My work does not treat networks as fixed entities to be captured and measured but rather employs qualitative methods to understand networks as fluid, dynamic and somewhat nebulous webs of relationships. As argued elsewhere, the questions we ask, the research tools we use and the analytical framework we apply, shape how networks are presented and represented in the research process (Ryan and D'Angelo 2018;Ryan et al. 2014).
The contribution of this article is at the nexus of qualitative social networks analysis (SNA) and migration studies. In it, I consider how I engage with participants in processes of co-constructing networks to make complex, diverse and dynamic relationships visible in particular ways -through words and pictures. Hence, I suggest that 'telling network stories' offers new insights for migration scholarship by analysing how migrants narrate and make sense of their relationships within and between places, including transnationally. Furthermore, I consider the extent to which 'network stories' can provide important insights into the ethnic composition of social ties and hence, following Dahinden (2016), go beyond the a priori approach to ethnicity that has so dominated migration research (Glick Schiller et al. 2006).
During my research, I have accumulated a considerable corpus of data through longitudinal qualitative methods involving follow-up interviews with the same participants over many years. In this article I aim to deepen the analysis and provide richer understandings of that data by presenting two case studies. Thomson (2007) recommends case studies for presenting longitudinal qualitative data because they enable researchers to discuss change over time through a focused analysis of repeated interactions and exchanges. As Neale et al. (2012: 8) observe, research case studies offer us a good way to 'condense cumulative data into a meaningful narrative that carries interpretation and analysis'.
The article begins by presenting my conceptual framework of 'telling network stories' drawing on Mische and White (1998), Plummer (1995) and Mason (2004). Then after describing my research methods and introducing my two case studies, in the data section I address my two overarching research questions. First, how do migrants narrate the dynamic processes of making new ties, as well as sustaining existing ties, both locally and transnationally? Second, what role does ethnicity play in stories of migrant networks, and what can this reveal about the salience of identity? In the conclusion, I consider how qualitative SNA, especially the conceptual framework of telling network stories, and qualitative, longitudinal data contribute new understandings of social networks in migration.

Narratives and networks: telling stories
Network scholars have long been interested in the role of discourse, especially storytelling, in understanding interactions in networks (Knox et al. 2006). According to Ann Mische and Harrison White (1998: 695), 'discourse is the stuff of social networks' and went on to explain (Mische and White 1998: 703) that: Social networks can be defined as sets of actors jointly positioned in relations to a given array of ties (for example, friendship, advice, co-work, church membership, political alliance, business transactions, information exchange and so on). Each type of tie is accompanied by a set of stories (along with associated discursive signals) that are held in play over longer or shorter periods of time.
However, it is insufficient simply to argue that networks are 'constituted by stories' (Mische and White 1998: 695). Drawing on Goffman (1959), Mische and White state that wider socio-structural contexts mediate those discursive processes. There is an 'absence of an adequate theoretical understanding of the co-mingling of network relations and discursive processes' (Mische and White 1998: 695). Part of the reason for this has been the epistemological and methodological stand-off between structuralist approaches of network analysts and the more interactionist leanings of the sociolinguists (D'Angelo and Ryan 2019). Hence, there has been insufficient understanding of how networks are 'embedded in domains' (Mische and White 1998: 703). While Mische and White (1998: 718) apply quantitative methods, there is growing interest in the possibilities for qualitative research on social networks (Heath et al. 2009;Ryan et al. 2014).
While scholars such as Knox et al. (2006) and Mische and White (1998) are interested in how storytelling occurs within networks, I go further and, using qualitative methods, examine how networks themselves are constructed as discursive devices through processes of telling stories and drawing pictures. I aim, therefore, to contribute not only to network theory and methodology but also to consider how this can enhance migration research. In particular, I am interested not only in how migrant networks are constituted by stories but also how these are embedded in particular domains. One domain that has been relatively under-researched is the interview encounter itself. I am keen to understand how networks are co-constructed within the specific domain of an interview. In so doing, I draw on qualitative methods to explore rich, complex and dynamic interpersonal encounters but always mindful that these are situated in wider socio-structural contexts.
In recent decades there has been a proliferation of writing about the narrative analysis of research interviews (De Fina 2011;Mason 2004;Rainbird 2012;Ryan 2015a;Souto-Manning 2012). The notion of telling stories, including the interview as a story, has been particularly influential (Plummer 1995). Within the field of linguistics and language research more generally, there is interest in how words are used to communicate meaning and tell particular kinds of stories in specific contexts (De Fina 2011). Narratives are 'interpretative devices through which people represent themselves, both to themselves and to others' (Mason 2004: 165). Thus, interview stories perform personal work -spelling out who I am and how I relate to other people (Ryan 2015a).
However, that is not to suggest that the interview narrative can be understood just as an individual story. Personal narratives are 'grounded in changing webs of relationships' demonstrating 'the significance of context, contingency, constraint and opportunity' (Mason 2004: 166). Thus, it is important to contextualize narrative form, structure, content and meaning in wider socio-cultural frameworks (Souto-Manning 2012). We do not construct narratives purely of our own making, our stories interact with and are shaped by the wider narratives circulating in the society around us. For migrants especially, these wider narratives may be significant and reflective of particular public and political discourses about immigration. As Azarian (2017: 693) argued, story tellers draw upon a 'menu of perspectives, definitions and interpretations' situated within the wider symbolic structures.
The relationship between agency and structure in storytelling has been analysed by Ken Plummer (1995). His work revealed the contingencies that shape story makingthe who, what, where, when, why and how of narratives. He argues that these questions can be answered on four inter-connected levels. First, the socio-historical; how the narratives are situated in relation to wider historical factors and patterns in that society. Second, cultural; what cultural frames and dominant assumptions shape how narratives are constructed and told. Third, contextual; to what audience and in what sort of encounter is the narrative being related. Finally, personal; what are the specific motivations, experiences and reflexivities that shape the stories being told. In this way, narratives can be understood as stories situated in time and place, but also connecting the present with a sense of the past. While they are presented as individual and personal ('I'), they also relate to a sense of the collective ('us' or 'them').
The relationship between the interviewer and interviewee -both of whom perform the roles of performer and audience -shapes how stories are told, interpreted and understood. 'Storytelling is a prime site for identity negotiation' and, in the interview encounter, 'the kinds of identities people present crucially depend on who they understand their interlocutors to be' (De Fina 2011: 30). My work is informed by a critical reflexivity (Pillow 2003) that is designed to expose the complex and dynamic processes behind data collection and so pays attention to how networks are constructed in the interview encounters between myself and the participants (Ryan 2015b). In analysing 'network stories', I am interested in how networks are co-constructed by the interviewer and interviewee through words and visual images within the interview contexts. Network data are not collected in a vacuum. The questions we ask and visual tools we use shape the kinds of data that participants present to us (D'Angelo and Ryan 2019). I reflect further on these methodological contexts in the following section.

Methods
I have long been interested in migrants' relationships and attachments as a way of understanding feelings of belonging and dynamic processes of embedding in different places, including transnationally, over time (Ryan 2007;Ryan et al. 2008). The value of a social network approach for understanding transnational relationships has been highlighted by several scholars (for example, Dahinden 2005;Schapendonk 2015). As Lubbers et al. (2018) argue, the theory and methods of social network analysis have much to offer our understanding of the structure, content and meaning of transnational relations. However, Lubbers et al. (2018) go on to note that the transnationalism literature has paid insufficient attention to the processes of change over time. They conclude by calling for more longitudinal migration research.
I use rich in-depth interviews, followed up over time to generate longitudinal qualitative data, as well as visual methods in the form of a sociogram (Ryan et al. 2014;Tubaro et al. 2016). The use of sociograms to visualize networks can be traced back to the pioneering work of Moreno and Northway in the 1930s and 1940s (Freeman 2000). I used a target sociogram, influenced by Northway (1940), to map ego networks (Ryan et al. 2014). This followed a simple paper-based design adapted from Barry Wellman (see image in Chua et al. 2011) andHersberger (2003). Participants wrote down the names of their contacts on a target diagram consisting of concentric circles divided into quadrants labelled friends, family, work, neighbours, hobbies, others (see Figures 1 and 2). As an ego-net design, the sociogram focused primarily on the participant's ties. I did not ask participants to visualize transitivity, the ties between alters, though this often came up in the interview narrative.
This type of visualization, especially when drawn directly by the respondent and combined with interviews, can add valuable details about network size, structure and interpersonal closeness (Tubaro et al. 2016), while also prompting memories and stories about particular relationships (Hogan et al. 2007). As discussed later in the case studies, comparisons of the interviews conducted with and without a sociogram, involving the same participant, show how new data can emerge when the visual tool is used. One criticism of sociograms is that they offer a snapshot view of network composition and therefore are unable to capture dynamism over time (Conway 2014). In my experience, as discussed elsewhere (Ryan and D'Angelo 2018;Ryan et al. 2014), combining a sociogram with a biographical interview allows that dynamism to unfold through the interaction between the visual tool and the interview discussion.
Interviews usually began by asking about migration processes; when, how and why people migrated. Inter-personal relationships were often central to this story and references to friends, partners and relatives were woven through the narrative before the sociogram was introduced -about 15 minutes into the interview. Initially, participants were simply asked to add their social and familial connections to the sociogram according to degrees of closeness, within the concentric circles and across the four quadrants. I explained to participants that 'closeness' in this case meant emotional, rather than geographical, closeness. So, they should also include people to whom they felt close even if these were geographically distant relationships. Thus, I sought to capture transnational as well as local ties. I provided just enough instructions to make the process meaningful but I avoided placing a priori definitions on key terms (Widmer 2006). I wanted participants to describe what they understood by 'close' or 'important' relationships.
When collecting data on how participants perceive their socially significant relationships, researchers should not place limits, a priori, on who is considered relevant (Widmer and La Farga 2000). Rather than a prescribed list, such as a name generator, and to avoid imposing constraints, I used 'free listing' to allow participants to decide who to include (Widmer 2006). Hence, I did not place any limits on how many or how few alters should be named. Thus, the size and boundaries of the network were left to the discretion of the participants. Across my datasets, this free listing approach has resulted in sociograms that can look quite different with varying numbers of alters. Unlike quantitative SNA, which seeks to measure across sociograms, hence requiring greater comparability, my qualitative approach to SNA allows for variability and individuality in how data are co-constructed and visualized. I treat the sociogram as a picture embedded in the interview text. Therefore, the visual image cannot be analysed or understood without the accompanying interview. Any other researcher wishing to use one of these images would also need to read the interview transcript to make sense of both pieces of the story -words and picture.
Once the sociogram was completed, usually signalled by the participant saying 'I can't think of anyone else', I began to ask specific questions about particular relationships such as 'if you had a personal problem who would you feel able to speak to'? This sometimes prompted memories of other people not previously included in the sociogram, highlighting the challenge of recall in network research (Merluzzi and Burt 2013).
Visualization has been described as a means of making invisible social relationships visible (Conway 2014). However, it is important to note that the sociogram is never an entirely accurate record but rather reflects how a participant is feeling on the day and which relationships occur to them in that particular frame of mind (Heath et al. 2009), thus reminding us of the salience of cognitive processes (Krackhardt 1987). Moreover, sociograms are not neutral tools for collecting pre-existing network data; on the contrary, the design and layout of the tool shape what data are depicted and collected (Ryan et al. 2014;Tubaro et al. 2016). Hence, sociograms do not simply make existing relationships 'visible' but rather contribute to the co-construction of network data.
The sociogram-interspersed with discussion -took the participants approximately 20 minutes to complete. Then the interview continued around other related topics, including future plans. As the interview progressed many participants remembered additional contacts and added them to the sociogram. Hence, completing the sociogram was an iterative process through the interaction between interviewee and interviewer. Although interviews and sociograms are both qualitative methods, these are distinct data collection tools which, as argued below, elicit different kinds of information. Using visual, as well as narrative, techniques allowed different stories to be told suggesting the complexity, multi-dimensionality and fluidity of social relationships.
Data were analysed using thematic coding. A priori codes were generated from the original research questions but new codes were also allowed to emerge from the data. Each pair of sociogram and transcript were analysed together recognizing picture and story as co-constructions that interacted to generate different but complementary data.
Introducing Irene and Ewa 1 I have selected two participants whom I have followed for several years. Each participant was encountered on at least four separate occasions over a period of approximately ten years. This involved several in-depth interviews, one of which included a sociogram. 2 As a qualitative researcher, I am not aiming for representativeness. These two case studies are each unique in their own way, but nonetheless also reflect wider patterns found across my datasets of migration research.
Ewa is a university graduate in her thirties who initially moved from Poland to London in the early 2000s. I first interviewed her in 2006 as part of an ESRC-funded study on Polish migration to the UK. 3 That interview took place in a café. Although she initially experienced deskilling in the British labour market, through hard work, continual efforts and additional study, she eventually achieved a position as a statistician that was commensurate with her qualifications. When I interviewed her in 2014 she was married, had two children and was working in the research department of a large organization in London. The second interview, in which the sociogram was completed, took place in her home, sitting at her kitchen table.
Irene is a lawyer in her forties who originally moved from France to London, as a student in the early 2000s. Having qualified at a British university, she briefly returned to France but then decided there were more opportunities to build her career in London. She was first interviewed in 2011 and again in 2013, when the sociogram was completed, as part of an ESRC-funded study on highly skilled migrants in the City of London. 4 At that time she was married, living in South East London and on a career break while bringing up her three children. Both interviews took place at her home and the sociogram was also completed at her kitchen table.
By coincidence, both women are married to Englishmen. This was not a factor in recruitment and only emerged later during the interviews. Indeed, Ewa was not in a relationship when originally interviewed in 2006.
The following sections are organized around the two key research questions, as explained earlier. I first analyse the women's stories of making new friends in new places, while simultaneously maintaining existing ties transnationally. Then I examine how issues of ethnicity, nationality and identity were threaded through the narratives. In each section, I consider what we can learn about the meaning and dynamics of migrants' social relationships within places and over time.

Stories of making new friends in new places
Researchers have long been interested in how migrants make new ties in new places (Gill and Bialski 2011;Schapendonk 2015). In our first interview, in 2006, I asked Ewa how she had made friends in London. She told me that upon arrival she knew no one in London, apart from a friend of her brother's who met her at the train station but then offered no further help: he 'said "hi, welcome to London" and that was all. I really was alone.' Ewa reiterated that she didn't 'have anyone'. This statement is significant and challenges the common perception that migrants have ready access to networks upon arrival in cities like London. Ewa needed to make new friends from scratch and the residential charity, where she initially worked and lived as a carer provided ample opportunities for a 'social life': she now 'could make friends, no problem'. Ewa observed that these were mostly migrants 'so they value friendship'. This observation points to the salience of reciprocity. Migrants cannot simply make friends just as they choose. Networking is not only about opportunities to meet new people, but also requires shared interests and reciprocity (Ryan 2011). Thus, other people need to be 'open' and motivated to reciprocate in forging social ties with migrants. As illustrated in Ewa's story, it is likely that other migrants may be especially motivated to reciprocate as they too are attempting to establish new friends in new places. This observation becomes especially salient, later on, when we consider the challenge of making 'English friends'.
In the first interview, 2011, Irene also explained how she made new friends when she arrived in London. However, she added that opportunities to make and sustain friends changed over time and through key life events (Bidart and Lavenu 2005). Irene initially moved to London as a student and observed that it was easy to make new friends at university: 'you do make a lot of friends at university.' But later on, in the workplace, while it was easy to socialize and network, it was difficult to make real friendships: 'making friends was a much slower progress during my working years, very slow.' Irene observed how having children completely changed her opportunities to make new friends. Motherhood, she explained, 'opens up a completely different world' of childbased sociality that, as a migrant, she 'hadn't realized existed' in London.
As noted elsewhere (Bidart and Lavenu 2005;Scott and Cartledge 2009;Wellman 1984), networks change through a person's life course and children, in particular, change network composition, especially for mothers (Ryan 2007). Both women moved to London when they were single and were well placed to observe how their friendship ties had changed over time, especially since becoming mothers. While Irene was already a mother during our first interview, the case of Ewa is especially interesting because her situation altered dramatically between the two interviews. Therefore, it is useful to consider how those changes were narrated in her network stories.
Reflecting critically on the questions I asked in the first and follow-up interviews, I became aware that I foregrounded friendship making in the first round of interviews. However, when completing the sociogram, in the second interview, participants tended to begin with their family ties. Hence, it is important to consider how the sequencing of interview questions shapes data.
In interview 2, Ewa began by writing the names of her family, friends and some work colleagues who had become friends on the inner circles of the sociogram. Stating that she is only really interested in people who are 'quite close': 'I don't care about people far away' (she means in outer circles -see Figure 1) 'they either come quite close to me or I don't care.' She went on to explain: 'I put a lot of energy in friendship. It must be worth it.' As Azarian (2017: 692) notes, 'a story is primarily a justifying narrative' that seeks to explain and rationalize a particular line of action. The visual image prompted Ewa to justify her strategies of relationship making.
Irene likewise began to populate the sociogram with her closest ties, particularly her family. However, she also reflected aloud on how to include everyone: 'lots of family … let's not put them all.' While naming all her closest ties, especially among family and friends, she wrote collective categories for slightly more distant ties such as 'business colleagues' and 'the Players' a local amateur dramatic group (see Figure 2). While working on the sociogram, she repeated her earlier observation that social ties changed over time: 'but they change over the years as well.' Talking aloud, she added: 'that was very different because it was pre-family', namely, before children. Indeed, Ewa's changing life circumstances from interview 1, when she was single and childless, to interview 2 by which time she was married and had two children, formed a key organizing frame for her network narrative. Completing the sociogram, she reflected aloud on how she got to know people in her new suburban neighbourhood. Her husband was surprised that she quickly got to know their neighbours. She explained that as she was on maternity leave, and at home during the day, she regularly received postal deliveries for neighbours and thus got to know them when they collected their parcels. She described how having young children completely changed relationships within a locality: 'I think it is a very different way of settling in a country if you do or don't have children.' Since becoming a mother she has made new friends at playgrounds, in toddler groups, at nursery, and so forth.
Repeat interviews, especially combining narratives and visual images, highlight the dynamism of ties over time. Migrants' networks far from being static continue to evolve over time in ways that require continual effort within specific opportunity structures based on mutuality and reciprocity (Ryan 2011). Moreover, as well as local ties, transnational relationships cannot be taken for granted but continually evolve.

Narrating transnational ties
The sociogram prompted Ewa to think and reflect on her connections and gradually she added people to the outer circles. As it emerged, the composition of her visual network seemed to surprise her: 'oh my goodness, it's all England.' Looking at the picture of her social ties on the paper sociogram, she paused and remarked aloud 'just thinking if I still maintain really close friendships in Poland.' After some reflection, she added some people: 'actually, I'm not fair, I should put them in here' (adding cousins/aunts in Poland). Nonetheless, while adding extended family, she seemed to need to justify why there were so few friends in Poland on the sociogram: 'when I went to Poland I used to spend time with friends but it became less and less regular.' She explained the gradual fading of these transnational friends through lack of time: 'my time is quite short in Poland and you have to choose.' Ewa explained that she had to prioritize visiting her relatives during her Polish trips: 'we are very close.' She emphasized that family is 'the centre of my life'. Her ties to transnational relatives appeared to have increased after the birth of her two children. In a subsequent email communication, Ewa told me that the interview had caused her to reflect: 'our interview made me think how much I still rely on my Polish links … it's my Polish family that is at the heart of my social links. … As well, in any troubles, I reach back for Poland (childcare arrangements etc.)' (email from Ewa 2014). This self-initiated email reflection from Ewa is worthy of further comment. Clearly, she had been thinking about the interview long after it was completed and I left her house. She wanted to share her reflections. It seemed important to her that the information she had presented about her social relationships was meaningful. As argued elsewhere (D'Angelo and Ryan 2019), participants usually take the research process seriously and want their network data to fit their sense of themselves. This may be even more so when the data are visualized and appear on paper before their eyes in the interview encounter. How network data look may be surprising to participants and prompt reflections. I will return to this point in the section below.
When completing the sociogram, Irene was also reflective about her transnational relations: 'I suppose French friends in France are more distant.' Like Ewa, Irene also observed that ties to friends in France had weakened over time, while ties to relatives remained strong. Irene takes her children to France every year on holiday with extended family in a summer house owned by her mother. For Irene, this was partly about giving her children a chance to enjoy the kinds of family holidays she experienced as a child, as well as connecting with relatives in France. Moreover, holidays in France were also about 'passing on the heritage' as the children were immersed in French language and culture.
Therefore, both women illustrate how transnational ties can be differentiated. As Michael Eve (2010) observed, long distance relationships are more likely to be maintained over time if they form part of strongly connected groups, such as extended families. By contrast, ties to individual friends may take more effort to maintain and so gradually fade. As illustrated by Irene and Ewa, using a social network approach including visualization, can reveal interesting aspects of how different transnational relationships are negotiated over time. Hence, we can see how the salience of transnational ties, particularly to parents, may increase over time as grandparents take on caring roles and become conveyors of cultural heritage, family memory and identity.
Having highlighted the ways in which the women narrated their changing networks, both local and transnational, over time, in the next section I look more deeply at how opportunities but also obstacles to networking are narrated and what this suggests about the ethnic composition of ties.

The ethnic composition of networks
As Dahinden (2016) has argued, social network analysis may be useful for overcoming the ethnic lens in migration research. When introducing the sociogram to my participants, I gave no instructions on the ethnicity of social ties. Instead, I waited to see if this emerged spontaneously.
As noted, in the first interview, Ewa stated that most of her initial friends in London were migrants. As she did not ethnicize them, I asked probing questions about their nationality. She then highlighted their diversity; they were Canadian, Japanese and Irish, but not Polish: 'I never look for Polish friends.' As with many other migrants across all my studies, Ewa observed the particular difficulties associated with making friends with English people. She remarked that it is quite hard to 'enter their circle'. This begins to suggest some obstacles in the way of forging new social relationships.
During that first interview, although describing herself as 'quite Polish in my habits', such as eating Polish food and attending a Polish church, Ewa did not see herself as part of a Polish community in London: 'I don't belong to this group.' Far from perceiving co-ethnics as sources of support, Ewa stated that since Poles in London were often competing for the same jobs and accommodation, they were reluctant 'to support each other'. Plummer (1995) reminds us that stories do not exist in a vacuum but are shaped by specific socio-temporal structures. Therefore, it should be remembered that my first interview with Ewa took place in 2006, just two years after Poland's accession to the EU, when hundreds of thousands of Polish migrants arrived in London. At that time, the dominant, rather negative, public discourses in the UK associated Polish migrants with construction workers, hard drinking, hard working men living in overcrowded, poor quality accommodation (Garapich 2008). As a highly educated young woman, Ewa perhaps sought to distance herself from such negative stereotypes of Polish migrants.
In interview 2, when populating the sociogram, Ewa once again did not ethnicize her social ties. She simply added alters without stating their nationality or ethnicity. When asked some probing questions about the ethnicity of her closest connections, there emerged marked consistency with her first interview. Although one or two of her closest friends in London were Polish, she reiterated that she did not seek out Polish people. She repeated that she is 'open-minded' and gets along with all sorts of people: 'I'm settling easily.' However, as she described meeting new friends through her children's activities, Ewa observed that these tended to be 'non-English' women. I was curious and asked what nationalities these women were 'Slovak, Romanian, Brazilian, you name it, Australian'. She noted that it was easier to 'connect with mums who are non-English'. I asked why that was the case. She explained that 'non-English' mothers have something in common, not having grown up in the UK, they don't understand some local cultural references. This may be especially apparent at toddler groups where English songs, nursery rhymes and stories prevail. Migrant mothers may feel cultural outsiders. Hence, they are drawn together as they 'sort of all accept each other'. Echoing a point made eight years earlier in the first interview, Ewa still found it difficult to make friends with English people, despite the fact that her husband is English.
Similarly, Irene noted her difficulty making English friends, claiming that the English are 'extremely superficial'. This echoes remarks made by other participants that, for example, the English are very polite but difficult to read. So it is very hard to get to know them properly and establish a meaningful friendship (Ryan et al. 2014). Using a critically reflexive lens (Pillow 2003), it is important to consider how the dynamics of the research encounter between interviewer and interviewee may shape the data collection process. It is significant to note therefore in most cases, and certainly in the two case studies under discussion here, that the interviewees were aware that I am not English. This may have allowed them to express views that they would have felt less comfortable stating before an English interviewer.
Quite spontaneously in our first interview, Irene stated that since having children, she was 'naturally' drawn to French mothers. Becoming a mother in London made her feel a strong need to connect with other French mothers. She explained it was 'extremely important' for her to socialize with French mothers and have her children play with other French-speaking children. These new French mothers, whom she actively sought out at toddler groups, in parks and play activities, have now become her 'best friends'. 'I'll always be French, I suppose.' This observation highlights the ways in which networks can be constructed to fit particular views of one's identity (Ibarra and Deshpande 2007). All of this emerged in the first interview. I wondered if introducing a visual tool would make any difference to how Irene's network was coconstructed in the second interview. In other words, what is gained by using a sociogram?
In completing the sociogram, Irene populated the friendship quadrant quite densely with nine alters tightly packed in this small space (see Figure 2). As noted, I used a free listing technique, giving participants minimal instructions, to observe how they described alters. At this point, Irene had not mentioned the ethnicity of these friends so when she had finished the sociogram I asked her about the friendship ties. Irene replied, 'a lot of them are French, if not all of them.' However, then something very interesting happened. Irene laughed and disrupted the usual interview process by asking me if I had a 'hidden agenda'. She then quickly added two other alters -a Greek and an Australian woman. She emphasized that her networks were actually very 'multicultural'. This incident is very curious and begs the question why she felt the need to add non-French women.
I suggest that introducing the sociogram in the second interview, to visualize a network, creates a tangible image on paper in front of the interviewer and interviewee and elicits different data from the previous interview. The visual tool confronts participants with an image of their network that may be surprising and even uncomfortable. As De Fina (2011) observes, storytelling is a site for identity negotiation. We seek to create a narrative that is comfortable and that fits our sense of self. When combining interview narratives with visual pictures, participants may strive for consistency and reduce potential inconsistencies. Throughout both interviews, Irene presented herself as 'open minded', embracing the multi-culturalism of London. She chose to live in south-east London, and avoided the French 'bubble' of west London. In fact, she was quite critical of those who chose to live in that highly French quarter of London: 'these people are living in France -that is not me.' Nonetheless, the tangible, visual image on the sociogram suggested a different story of predominantly French social relationships. When this became visible, Irene quickly sought to change it by adding two non-French women.
This observation suggests that while networks may be a good data collection tool to challenge the a priori ethnic lens, so dominant in migration research, we need to be cautious about seeing networks as co-constructions within interview encounters. Rather than capturing an accurate account of network structure and composition, the final output is shaped by complex cognitive and perceptual processes (Krackhardt 1997), as well as by the dynamic interactions between interviewer and interviewee.
As migrants, both women had to forge new ties in new places. As highly educated, professionals who spoke English fluently, they had opportunities to socialize with a wide range of people. Nonetheless, in the interview narratives and especially through the sociogram, both women presented their networks very differently. Ewa was adamant that she didn't need other Poles and mainly made friends with migrants from other nationalities. This was a consistent feature across both her interviews. Irene, by contrast, suggested that her social ties had changed dramatically over time and became more French after having children. These observations suggest how migrants, even given similar opportunities for networking, may make different decisions about who to befriend and this is shaped, in part, by personal preferences and the importance of ethnic identifications (Ibarra and Deshpande 2007). Nevertheless, both described difficulty in making English friends. Thus, not just opportunities and shared interests but also reciprocity are required to forge new ties (Ryan 2011). While participants interpreted this as English 'superficiality' or 'reserve', from an SNA perspective it can be interpreted somewhat differently. Migrants share interests in making new friends, but the 'native' population already has long established friendship and familial ties and hence is less motivated to make new friends.

Conclusion
In this article, I have sought to contribute at the nexus of migration and social network research by considering how migration research can benefit from the tools and vocabulary of qualitative SNA (Bilecen et al. 2018). Rather than simply taking social networks for granted, migration research needs to achieve more critical insights into the meaning, composition and dynamism of social ties (Ryan 2011;Schapendonk 2015). Moreover, instead of treating networks as fixed entities to be captured, measured and analysed through the research process, I have drawn on classic network theory (Mische and White 1998), to understand networks as 'discursive devices' that are located within particular socio-structural contexts. In so doing, I have sought to develop that work in three specific ways.
Firstly, using qualitative longitudinal data, I revealed changing relational ties over time and across different places, including transnationally. Repeated interviews with the same participants allowed me to collect deeper and richer data. Furthermore, interviews that are repeated over an extensive period of time can facilitate insights into changing relationships through the life course in contexts of unfolding opportunities, mutuality and reciprocity. This is helpful not only in understanding how migrants perceive and construct local ties in the destination society, but also how their transnational relationships may evolve over time. As illustrated by the two case studies in this article, transnational ties are differentiated and dynamic. Both women narrated strong and enduring ties to their families, including extended kinship groups, facilitated by regular visits. However, in each case, it is noteworthy that ties to friends in the country of origin appeared to have weakened after more than ten years of migration. This finding adds further evidence to the observation by Eve (2010) and Lubbers et al. (2018) that transnational ties are more likely to endure if they form part of dense clusters rather than single ties to disparate individuals.
Second, building on the work of Dahinden (2016), I have considered how using social networks as a method may help to unsettle the ethnic lens (Glick-Schiller et al. 2006) in migration research. Using interviews and sociograms together, I have demonstrated how the complexities of ethnic composition may gradually become apparent as social ties are presented through oral and visual methods. Nonetheless, my work in London illustrates how interviewees may seek to present a network that feels comfortable and fits with the wider multi-cultural discourses of this global city. Furthermore, my analysis suggests how migrants may encounter obstacles in their efforts to forge new ties, especially with the local 'native' population; requiring a justification narrative. This underlines Ken Plummer's observation that personal stories are shaped by wider contextual discourses. Hence, I suggest that while there is clear value in not beginning our research from a priori ethnic categories, nevertheless, a social networks approach may be useful in understanding how migrants apply their own ethnic lens in making sense of social ties.
Third, I have contributed to qualitative social network research by responding to the call by Knox et al. (2006) for more critical engagement with the concept of social networks. Adopting a critically reflexive approach, I have explored the dynamic, interactive processes by which networks are co-constructed through stories and pictures in interview encounters. Thus, building on the earlier work of Krackhardt, I argue that network data collected by researchers is shaped by processes of perception and selfpresentation (D'Angelo and Ryan 2019). This observation provides a more nuanced understanding of how migrants construct and present their network narratives and, hence, may provide important insights into how they make sense of, justify and explain their relationships. Hence, I suggest that rather than making invisible relationships visible, social network research is actually implicated in the co-construction of relationality, especially through visualization tools. Comparing across interviews done with and without a sociogram, I have suggested how the visual tool helped to elicit different kinds of data. Other researchers have noted how participants find their network visualization highly interesting (Hogan et al. 2007). However, my work also reveals how participants found the visualization surprising. They reflected on the image as it emerged in front of them and continued to add extra alters over the course of the interview until the picture and story of their relationships began to feel more comfortable to them. Moreover, both participants discussed in this article took further steps to clarify their networks, Irene very deliberately added new alters to change the diversity of her friendship group, while Ewa emailed me to highlight further the salience of her long-distance family ties.
To sum up, in this article I argue that adopting a critically reflexive approach to qualitative SNA illustrates how networks are used as discursive devices to make sense of differentiated, dynamic and dispersed relationships. However, this process is not spontaneous. Rather, it is shaped by the tools we use and the questions we pose. In telling network stories, participants seek to construct a coherent narrative with which they are comfortable. Therefore, I propose that, instead of taking networks for granted, using these insights can help migration scholars to gain deeper understanding of how networks are presented, constructed and interpreted within the research process.