Family language policy in Swedish-English families: Rhizomatic conceptualisations

1 Introduction

Research on language acquisition in multilingual transcultural families is an empirical topic which can trace its lineage back to at least the early part of the twentieth century, when Ronjat (1913) employed a diary study to document his son’s bilingual development in French and German. In the following decades, similar studies continued to be popular within the field, while developmental psycholinguistic approaches ultimately dominated from the 1970 s. The present study concerns itself with what has become known as the field of family language policy (hereafter FLP) (King, Fogle, & Logan-Terry, 2008). FLP can be seen as emerging from these earlier research traditions, but importantly, coincides with an approach to the study of family multilingualism which is grounded within sociolinguistic method and theory (King, 2016). This approach builds on Lanza’s (1997) seminal study, which used sociolinguistic and discourse analytic means to address a classic psycholinguistic question regarding early language differentiation in bilingual children. The sociolinguistic underpinnings of the topic arise from two subfields, namely, language socialisation (Ochs & Schieffelin, 1984) and language maintenance and shift (Fishman, 1991). In addition to the abovementioned foundations, FLP has drawn heavily upon Spolsky’s (2004, 2009) model of language policy, which focuses on language practices, language ideologies, and language management. Increasingly, however, the field has come to encompass theoretical perspectives from “contemporary language policy research [...] through which language practices may be seen as de facto grassroots language policy” (Lanza & Lomeu Gomes, 2020, p. 154).

The centrality of the family within the study of language and society as a whole has been highlighted by Hazen (2002), who highlights the role of the family in language variation, and Blommaert (2019, p. 6), who writes that through studying language in families, one is in fact studying “society in its very complex concreteness”. Although traditionally theorised to be a private domain (Fishman, 1991), Lanza and Lomeu Gomes (2020, p. 165) explain that the family is better conceptualised as a space that lies somewhere on a continuum from the private to the public. Bourdieu (1991, p. 61) writes that the family is one of the two principal domains where the recognition, production, and reproduction of ‘legitimate language’ takes place (the other being the formal education system), while Curdt-Christiansen and Huang (2020, p. 174) describes the family as a “microcosm of a macro-society, reflecting the larger sociocultural environment in which they are situated”. Ultimately, permeating discourses from the wider society shape language ideologies, which in turn shape language practice in the family. Equally, the family as a unit influences the individual, who then takes their place within the larger society and reproduces those linguistic forms and ideologies which were negotiated and co-constructed within the family domain, resulting in a circular relationship between language in the family and language in society.

FLPs are influenced by a complex web of multiscalar social processes. A broad body of research on the field of FLP has highlighted the entanglement between socioeconomic, sociopolitical, sociocultural, and sociolinguistic factors with interactional practices in the home domain (Curdt-Christiansen & Huang, 2020). However, most current studies restrict their aims to investigate a narrow range of phenomena in relation to FLP and thus do not typically attempt to approach a broader understanding of the evident complexity that is found (Mirvahedi, 2020). Furthermore, despite the dismantling of predefined categorisations of language and identity found in much contemporary sociolinguistic work (see e. g. Heller, 2007, Léglise, 2017), there is often a certain taken-for-grantedness regarding the central concepts of family, language, and policy within the field (Lanza & Lomeu Gomes, 2020, p. 166)^[1]. I will also argue that the concept of agency is equally taken for granted. In addition, many research designs take a predetermined approach and privilege certain policy-makers within the family. Such research designs undermine the idea that FLPs are in fact “multiple individual policies that include individual ideologies, management approaches, and practices within a single family unit” (Hirsch and Lee, 2018, p. 890). These structuralist approaches can lead to an inability to observe unforeseen interconnections between actors, agential forces, lived experiences, societal discourses, and semiotic resources as relates to the family domain. Additionally, many FLP studies have been described as ‘snapshot’ studies, which are based on descriptions of a FLP at a given point in time (Palviainen, 2020, p. 237). Such studies do not critically consider FLP as a dynamic system which is constantly in flux and intimately connected to conceptualisations of space and time.

In this article, I aim to show how a comprehensive understanding of FLP within Swedish-English families may be achieved through the implementation of a rhizomatic approach (Deleuze & Guattari, 1987; Pietikäinen, 2015; 2018), which allows for the tracing of the multiscalar heterogeneity that is a FLP. A rhizome is “a pragmatics composing of multiplicities or aggregates of identities” (Deleuze & Guattari, 1987, p. 15) and a connection of “any point to any other point” (p. 21). The metaphor of the rhizome articulates a view of the world as a complex and non-hierarchical network. It rejects essentialist categorisations and sees social life as fluid and enmeshed with the environment. By conceptualising a FLP as a rhizomatic policy system, an open-ended constellation of relevant, non-hierarchical elements may be considered. I therefore argue and aim to show that a rhizomatic approach to the study of FLP is one of the ways that might allow the field to move past taken-for-granted concepts, to allow for a critical perspective on space and time to be taken, and to allow for consideration to be given to how agency emerges not from within a single autonomous family member, but rather from the “collaboration, cooperation, or interactive interference of many bodies and forces” (Bennett (2010, p. 21). The approach aims to trace the historically situated processes which underpin the meaning and value attributed to linguistic forms at a particular space and time. In being inherently non-hierarchical and non-predeterministic, this approach aims to view FLPs through a lens that is quite different from many other studies.

The data drawn upon comes from a project featuring four Swedish-English transnational families. These families are composed of parents from differing sociocultural backgrounds; in each family, one parent was raised in Sweden and the other was raised in the United Kingdom. Each participating family has two or more school-aged children. This population differs from many previously researched bilingual family configurations. Although FLP research has grown to encompass a wide range of languages and contexts globally (Lanza & Lomeu Gomes, 2020: 161), as regards to English in particular, past studies have often investigated bilingual families residing in countries in which English is the primary societal language. In these international settings, where a ‘foreign’ language does not have parity in terms of prestige with the local language, a preference for the societal language and resistance towards the minority language has often been found in children in bilingual family contexts (Caldas, 2012). In Sweden, while English is not generally spoken as an everyday societal language, it is nevertheless afforded a great deal of status and privilege, it does have “a near ubiquitous presence”, and the population has generally “high levels of communicative competence” in English (Henry, 2016, p. 443). The setting explored here thus makes for a context where majority versus minority language dynamics are different than in many other settings and is therefore an interesting site for examining the relationship between societal pressures and language use in bilingual families. More explicitly, it is interesting to explore whether resistance towards English is found in children in Swedish-English families in Sweden or whether ideologies surrounding English prevent such resistance (see also Nandi, 2023 on power and resistance). Additionally, as relates to the English-origin parent, it would be relevant to know how migrating to a society which has a general willingness to speak English with migrants (Bolton & Meierkord, 2013) affects the individual’s desire or perceived capability of learning Swedish, and importantly for this study, how this then affects language policy within the home.

The departure point of the analysis in this study is to investigate several interactional episodes within the abovementioned families and consider how these interactions are entangled within a complex web of interlinked processes. In order to accomplish this, the interactional data is combined with biographical interviews as well as from retrospective video analysis with the families through stimulated recall procedures (Santiago Sanchez & Grimshaw, 2020). More specifically, I attempt to investigate how the language practices within these families do not emerge spontaneously, but instead are deeply rooted within categories of identity, which are themselves embedded within interconnected historical, social, economic, and political contexts (Pietikäinen, 2018. p. 185). Each episode is regarded as a temporary assemblage of thought and action. I regard a FLP to be a rhizomatic policy system (Hult, 2015, p. 225) which is itself built from the connection of multiple of these temporary assemblages.

2 Multiscalar complexity and agency in family language policy

Spolsky’s framework of language policy highlights the inherent multiscalar complexity in FLPs, referencing how differing social processes affect the three components of language policy (practice, ideology, and management). The value of the language policy lens is that it goes beyond the ‘how’ and considers the ‘why’ in relation to language practice. How multilingual families interact can be analysed in isolation through various interactional research paradigms, but my view is that language policy allows for the detailed analysis of interactional patterns in combination with the analysis of the conceptual and material factors which underpin those practices. The presupposition that wider social environments influence FLPs has been well documented empirically (e. g. Curdt-Christiansen & Huang, 2020; Van Mol & De Valk, 2018). Family internal factors such as the number of children in a family, the age of children, and the marital status of the parents have also been discussed as factors which may have effects on FLPs (Caldas, 2012; Ochs & Scheiffelin, 1984; Macleroy Obied, 2010). Spolsky’s model has, however, recently been critiqued in relation to his statement that “language policy is all about choices” (Spolsky, 2009, p. 1). Lomeu Gomes (2020, p. 62) and Hiratsuka and Pennycook (2020) have suggested that language policy is, in fact, not all about choices. They instead suggest that the reality of practically getting things done in busy family life “might supersede the intention of prioritising the use of this or that named language” (Lomeu Gomes, 2020, p. 62). This does not, however, signify that family members never make explicit choices about language policy. Some parents, for example, do make an explicit choice regarding a FLP (e. g., to implement a one-parent-one-language policy), and then do systematically follow through with that policy in interaction. Rather, the position criticises that within Spolsky’s model, there is an “inherent overvaluation” given to “the role of choice in contexts where, in fact, language practices might be shaped by other contingencies” (Lomeu Gomes, 2020, p. 62). In line with non-predeterministic rhizomatic thinking, the degree to which a language practice is a conscious choice or not^[2] should be critically considered based on the evidence presented and not taken for granted in advance of this.

This article takes the position that language practices are the actualisation of language policies, conscious or not. This position corresponds to recent language policy scholarship in which language practices in themselves are viewed as a de facto grassroots language policy (Lanza & Lomeu Gomes, 2020, p. 154; Van Mensel, 218). I consider language policy as the underlying mechanism which produces these practices. From this perspective, I see language policy as the coalescence of lived experiences, material elements^[3], circulating conceptual discourses (such as ideologies of communication), space, and time, which influence language practices. Within this conceptualisation, agency is viewed as emerging from the momentary assemblage of these diverse things, concepts, and places; agency is distributed, porous, and relational, existing not within individuals, but rather diffused across multiple entities (Bennett, 2010)^[4]. Although an acceptance of the complexity of agency in FLP has been discussed (Smith-Christmas, 2020: 231), the posthumanist and new materialist lens on agency that is adopted in the present article is a radical departure from previous conceptualisations of agency within the field. However, I wish to stimulate a debate as to how viewing agency in this way might provide analytic insights which would not have been revealed otherwise, and in this article, I aim to exemplify such findings in my own analysis.

3 Rhizomes and assemblages

The concept of the rhizome has been adopted in much recent sociolinguistic scholarship. Pietikäinen (2015, p. 210), investigating multilingual dynamics in Sámiland, sees rhizomatic approaches as being capable of tracing “the changing trajectories and circuits of language resources”, while also being able to “capture the connectivity and interaction between and across the resources”. Similar descriptions and applications of the rhizome can be found in Leppänen and Kytölä (2016, p. 168), who investigated multimodal social media practices, and suggest that “rhizomatic structures create connections between individuals, communities, or nodes of social action that are, by default, non-linear, non-chronological, non-hierarchical, and non-binary”. Others have adopted rhizomatic thinking to exemplify critical sociolinguistic research methods (Heller, Pietikäinen, & Pujolar, 2018), classroom language practices (Prinsloo & Krause, 2019), and translanguaging (Heltai, 2021). A rhizome, in Deleuze & Guattari’s (1987) terminology, is a structure or system that is ultimately made up of many individual assemblages. Assemblages emerge from the interactions between their constitutive parts (DeLanda, 2016, p. 21); they are not just an ad hoc collection of components but are more than the sum of the parts (Pietikäinen, 2021, p. 4).

The rhizomatic approach proposed in this article is fundamentally a discourse analytic approach. The discourses investigated operate at various scales. One level of discourse relates to the interaction order within multilingual families. Drawing on Goffman’s work (1967, 1983) the interaction order calls attention to the interpersonal scale and considers how interaction is socially situated and co-constructed by interactants. Analysis of the interaction order further allows for the in-situ observation of what it is possible to talk about (and when), what is included and what is excluded (Pietikäinen, 2018). Specifically, within FLP contexts, a pertinent question to ask may be “why that now?” about an utterance, or rather “why are certain semiotic resources deployed in this specific context?” These questions relate to the apparent norms of interaction (Hult, 2015, p. 225) and habitualised practices within particular spatial configurations.

Other discourses operate at a more conceptual level. These discourses include circulating language ideologies as well as societal discourses which are made relevant by the participants. Typical ideologies revealed from past research have concerned the acceptability of language mixing (K. S. Pietikäinen, 2014), linguistic legitimacy (Wright Fogle, 2013), and the supposed efficacy of a particular strategy in raising bilingual children (Barron-Hauwaert, 2004). Also of relevance is what has been described as ‘neoliberal language ideology’. Such ideologies might relate to the perceived value of a certain language on the global market, or more everyday pressures where parents might consider if they are spending too much money or energy on attempting to raise multilingual children (Curdt-Christiansen, 2020, p. 189). Other more general societal discourses may invoke what it means to be a ‘good parent’ (King & Fogle, 2006). These may relate to beliefs around what a good parent supposedly does with regards to, for example, literacy practices, such as reading a child a story or helping a child with homework (Roberts, 2022). In addition to conceptual discourses, there also exist material discourses. These are more explicitly related to space and may involve how spaces in the home are physically laid out and how material is assembled in these spaces. I view material to consist most obviously of the human family members, but in line with a non-anthropocentric reconceptualisation of agency (Braidotti, 2013), it may also include other animate or inanimate objects.

A key element of a rhizomatic approach is to consider these discourses as being interconnected, continually changing over time and space, and historically embedded. Here, I bring attention to the concept of Spracherleben, the lived experience of language (Busch, 2017), as well chronotopic identities (Blommaert & De Fina, 2016), which have been described as the rearranging of the self and the associated linguistic repertoire at specific points in space and time due to the influence of life trajectories and experiences (Soler & Roberts, 2019). Such conceptualisations have typically been viewed as concerning the individual scale, but I view them also in relation to the interpersonal scale due to the parallel, intimately connected life trajectories that are often seen in members of the same family.

4 Data and method

The data analysed in this article comes from a larger study on the interactional practices of Swedish-English families living in Sweden. The project plan was reviewed by the Swedish Ethical Review Authority in May 2018. Families were then approached to participate in the study through social media channels. Upon agreeing to participate, the four families who feature in this article were sent cameras and asked to self-record instances of their everyday lives. After finishing their recordings, I visited the families in their homes and undertook interviews based on their lived experiences of language and their biographies more generally. These were followed by stimulated recall procedures. The interviews were primarily conducted in English and took place with the entire family, but the children came and went as they pleased. In the stimulated recall, the families were shown instances of what they recorded and were asked to elaborate on what they were seeing. The parents in the families consented to participate in the study after receiving information (in English and Swedish) in accordance with the Swedish Research Council’s (2017) guidelines. According to Swedish legislation, parents must consent on behalf of their children if they are under fifteen-years-old. However, specially written project information packs were prepared for the children in these families, and parents were asked to discuss participating in the study with their children before consenting. Table 1 presents details of the families who participated.

The participant families were selected on the basis that they shared a number of similarities, such as all having children who were similar ages, but differed in other ways, such as being from different socioeconomic backgrounds and living in different areas. These were conscious choices in order to explore varied family types. One single-parent family was chosen to include a family type which is rarely considered in family language research contexts, although common in many societies.

The choice of having participants self-record themselves was a methodological consideration to reduce the effect of the researcher being present in filming, as was the choice of conducting interviews after the filming and not before in order to not influence participants further. Despite these efforts, it is undeniable that many things have been indexed by the researcher and this project before the recording even starts. It is therefore more reasonable to accept that these data are collaboratively produced between the researcher and the participants (Talmy, 2011, p. 27). For a more detailed and reflexive account of the data collection procedure, see Roberts (2023).

The analytic procedure began by attempting to discover the habitualised practices observed in the video recordings. Each recording was tagged and annotated using ELAN, a piece of software which allows for multiple layers of annotation to be assigned to time-based media. The annotation scheme used codes each interaction based on the material context (who/what is present and what is the spatial configuration), as well as meta information about the interaction order (what is being talked about). In addition, salient features of the sequential organisation of the discourse are marked, drawing on conversation analytic and interactional sociolinguistic frameworks. Within the stimulated recall, participants were asked to elaborate on these habitualised practices, but were also shown what appeared to be deviant cases in which their language practices were less straight forward to understand. The interviews were thematically coded in line with the phenomenological tradition (Busch, 2017), which emphasises an emic perspective, focusing on how participants experience a phenomenon in a way that is meaningful to them. The purpose of this coding is to identify recurring patterns of meaning and to categorise these into themes. Following these steps, I attempt to trace the connections between the different data types, paying particular attention to how they intersect with scales of space and time. Each interaction can be viewed as a temporary assemblage of the various discourses enmeshed within the wider socio-material environment. Within this study, I present a number of interactional episodes which I have determined to be underpinned by a complex web of social processes, and can be considered as nodes that are “particularly dense intersections of multiple threads” (Heller, Pietikäinen, & Pujolar, 2018, p. 2).

5 Context-sensitive repertoires and their historical underpinnings

Throughout the data, it is shown that all participants draw upon a multilingual repertoire to varying degrees. Participants orient to ‘English’ and ‘Swedish’ as separate linguistic entities in the interviews (cf. Jaffe, 2007), and the self-recordings show that all participants use features associated with both depending upon the context. Depending on the spatiotemporal configuration, participants move between practices that index an ideology of parallel monolingualism (Heller, 1999), while in other times and spaces, they enact a repertoire that is more heteroglossic and polycentric in nature (Pietikäinen, 2010; 2013, Makoni & Pennycook, 2007). I will use ‘English’ and ‘Swedish’ at times when describing language practices as this is what is made relevant emically by the participants and does not signal a subscription to languages as “neatly separated, labelled and counted” (Lomeu Gomes, 2020, p. 17). I now aim to show how the linguistic repertoire which a participant draws upon in a specific timespace configuration is intimately connected to their own, and their fellow participants’, historical bodies. This observed repertoire can be considered as a context-sensitive ‘practiced language policy’ (Bonacina-Pugh, 2012).

It can be seen throughout the self-recorded data that in the Andersson, Clarke, and Pearce families, there is a clear enacted preference towards a monolingual English repertoire when all family members are present. The implicit preference for English in these three families is also found in all parent-parent interactions where the parents are alone. A link can most obviously be drawn here between their lived experience of language (Busch, 2017), and their personal historical trajectories to the present. The parents in these three families began their relationships in English owing to them meeting in English language contexts. Four of the parents met and lived together in England, while two parents met on the Internet and later moved to England together. Moreover, the three English-origin parents report a continued proficiency asymmetry between themselves and their partners, despite them having lived in Sweden for several years in some cases. That is to say, possibly owing to the socioeducational position of English in Sweden, the Swedish-origin parents are more proficient in English than the English-origin parents are in Swedish.

The norms of interaction between parents and children in these three families also follow an asymmetrical pattern. English-origin parents enact a clear English-only policy with their children, and the children reciprocate the same policy in the opposite direction. This is deviated from only on rare occasions. With the three Swedish-origin parents and their children, the interaction order is much more diverse, and ranges from an enacted Swedish-only policy in specific contexts, to the enactment of a fluid translingual repertoire (Hiratsuka & Pennycook, 2020) in other circumstances, and an English-only policy in yet others. The default within spatial configurations which only feature the Swedish-origin parent and children is, however, a monolingual Swedish medium (Gafaranga, 2010). A key factor which influences these language practices is the presence of the English-origin parent. If the English-origin parent is alone in a room with the children and the Swedish-origin parent enters, the medium of interaction typically does not change from a monolingual English medium. However, if the opposite happens, then the medium of interaction typically changes from Swedish to English. This closely relates to an observation made by Bourdieu (1991, p. 52) that “the relation between [...] people may be such that one of them has only to appear in order to impose on the other[s], without even having to want to, let alone formulate any command”. This matter was highlighted in one of the stimulated recall sessions. The moment in question comes from a self-recorded episode from the Andersson family, in which the Swedish-origin father and his three daughters are playing a card game. The activity began in Swedish but became more translingual as the game went on. I replayed the family this episode and asked about their perspectives on these kinds of practices. Excerpt 1 starts with me asking the father if he gives much conscious thought to this enactment of a translingual repertoire.

Excerpt 1 – Andersson stimulated recall

Tim: ((Looking at Andreas)) Do you think about how you’re always, sort of, switching?

Andreas: No, I wouldn’t have thought so. No.

Anna: When I’m not here do you switch?

Emily: No, we speak Swedish. Usually.

Bianca: When I’m upstairs with Emily, we speak English.

Emily: ((Pointing at self and Bianca)) Yeah, we speak English.

Anna: They speak English together, but when they speak to Francesca, they often speak to her in Swedish.

Emily: Yeah, we switch with her.

Anna: Yeah, so Francesca is the most Swedish of any of these ((pointing at Emily, Bianca, and Francesca)).

Andreas: Tricky one this. It could be because we’re actually playing a card game, just the four of us, so we could have done it in Swedish, but you were in the room ((pointing at Anna)).

Although Andreas initially answers “no, I wouldn’t have thought so” to the question, it is suggested by Emily that Andreas and the daughters actually typically speak Swedish together when Anna (the mother) is not present. He then elaborates towards the end of Excerpt 1 that it is Anna who functions as a catalyst for the enactment of a repertoire which features more English characteristics, “we could have done it in Swedish, but you were in the room” (cf. Bourdieu, 1991, p. 52). Interestingly, this excerpt also touches on sibling language practices. Bianca states that she and Emily often speak English together, which is confirmed by Emily. This is however not the case between Francesca and Emily or Bianca though as “when they speak to Francesca, they often speak to her in Swedish,” which is confirmed by the self-recorded data. What is observed in the video recordings and reported by this family, then, is a highly complex system of languaging, which is governed by nonrandom spatial configurations. A particular spatial configuration leads to interactants invoking various participation frameworks (Goffman, 1981) and frames (Gumperz, 1982), which in turn sanction a particular repertoire in that context. These participation frameworks and frames are intimately entangled with the chronotopic identities of all participants, with their historical bodies, as well as with pervasive conceptual discourses in place. Each of these components come together to create a temporary assemblage of thought and action.

The link between the family’s historical trajectory and the current practiced language policy amongst these three daughters is very apparent. Emily (age 11) and Bianca (age 9) were in fact born in England, but the family had moved to Sweden by the time Francesca (age 7) was born. The family reports initially having a rather monolingual FLP, despite some aspirations towards a one-parent-one-language approach. It was when the family moved to Sweden that Swedish became more prevalent in the home. Andreas reports during the interview that he made a more conscious effort to use Swedish from this point—his chronotopic identity changed due to a new timespace configuration (Blommaert & De Fina, 2016). It can be inferred that this shifting of his chronotopic identity was influenced by new situated conceptual discourses, which altered based upon the spatial environment in which the family was embedded. A conceptual discourse that prescribed English monolingualism as an acceptable outcome prevailed in England, while a dislocation from that environment resulted in new discourses which promoted Swedish. Equally relevant was the increased Swedish proficiency of Emily and Bianca, which can at least partially be attributed to their exposure to the Swedish education system and society at large.^[8] The sociolinguistic milieu in which Francesca spent her first years, then, was much different to the one which Emily and Bianca inhabited.

A strong connection between lived experiences and sibling language practices is found in all other families as well. In the Clarke family, the children typically use an English-focused repertoire with each other, which can be linked to an initial upbringing in England. In the Pearce and Thorén family, though, the children almost exclusively use a Swedish repertoire with each other, and the children in these families have no experience of living outside of Sweden. The connection between the external environment on FLPs and practices is clear to observe (Curdt-Christiansen & Huang, 2020; Roberts, 2021). A systematic literature review conducted by the author suggests that much past work on FLP has focused on language practices between parents and their children (see also Aronsson 2018 on the parent-child dyadic bias), while less often focusing on interaction between siblings. With a rhizomatic approach, a more holistic view of all possible constellations of interactants is considered as it does not assume the parent-child relationship to be more important than any other relationship in the family. The analysis also shows how the rhizomatic FLP system must be considered as being formed from the connection between past and present temporalities. As will be argued below, the speculative future also forms a part of this structure.

6 Flexible repertoires arising from lived experiences of language

Although the children in the Thorén family are observed using a Swedish-only repertoire with each other at nearly every opportunity, the practiced language policy with Sarah, the single-parent, English-origin mother, is much more complex. Sarah has what can be regarded as the most flexible language practices of anyone in the data. Here, I first aim to show how Sarah enacts this diverse repertoire in three interactional moments. I then aim to show how these norms of interaction can be understood in relation to her lived experience of language, which shares limited parallels with the parents in the other families. Excerpts 2 and 3 are from self-recorded episodes that occur while the whole family are sitting together after eating dinner. In Excerpt 2, Sarah and her eldest child, Liam (age 12), are discussing what is happening at ‘Tibble’, the local football team for which Liam plays. Before this excerpt begins, Sarah and Liam had been conversing in English.

Excerpt 2 – Thorén recording

Liam: [...] then I have a football match with my local team.

Sarah: How does it feel now that, um, like your team’s changed that half of your team players have gone to Eneby?

Liam: Ja, alltså de har nollfemmor, det är inte våra, vi har bara förlorat en spelare tekniskt sett.

[Yea, so they have zero-fivers, they aren’t ours, we’ve only technically lost one player.]

Sarah: Ja, men en spelare till Eneby.

[Yea, but one player to Eneby.]

Liam: Fast vi har fått fyra pers, eh Liam, William, Elias, alltså vi har fått några nollfemmor, nu är vi det äldsta laget.

[Although we have got four people, eh, Liam, William, Elias, so we have got a few zero-fivers, we’re the oldest team now.]

Sarah: Do you think you’ll join with Eneby or do you think you’ll stay at Tibble?

Liam: Jag kommer stanna på Tibble.

[I’m going to stay at Tibble.]

Sarah begins in English, asking Liam about how he feels since some of the players have left for another team, ‘Eneby’. The way which she articulates the word ‘Eneby’ follows a stress pattern associated with Swedish. In the next turn, Liam switches entirely to Swedish. The motivation for this change is uncertain, but Sarah and Liam retrospectively suggest that it is perhaps induced due to the delivery of Sarah’s turn final utterance. Nevertheless, Sarah’s lack of evaluation of this language alternation indicates that Liam’s response is unmarked and unproblematic. Sarah’s next turn shows her moving into Swedish with “ja, men en spelare till Eneby”, which Liam sees as unproblematic as he does not interactionally attend to this language alternation in any way, and his next turn continues in Swedish. Following this is Sarah’s turn, where she returns to English, “do you think you’ll join with Eneby or do you think you’ll stay at Tibble”. Finally, it is Liam’s turn, in Swedish, where he says he will stay at his current club. Sarah’s code choice in this excerpt corresponds with Lanza’s (2007, p. 56) ‘adult code-switching’ discourse strategy, whereby the parent uses both languages in interaction with their child. Sarah, as well as Liam, can be seen here adopting parallel bilingual medium of interaction Gafaranga’s (2010), with neither party explicitly nor implicitly requesting a medium-switch. When shown this clip during the stimulated recall, I asked Sarah if she ever consciously considers her language choice, and she reveals that she “has to work harder on talking English than talking Swedish”. Below, I will argue that this belief is clearly a product of her life trajectory and of her lived experience with language.

In Excerpt 3, the Thorén family are discussing which jobs the children would like to do when they grow up.

Excerpt 3 – Thorén recording

Sarah: Ja men, Astrid, när du blir vuxen vad vill du jobba med då?

[Yea but, Astrid, when you become an adult what do you want to work as?]

Liam: Vill du vara polis kanske?

[Maybe you want to be a police officer?]

Astrid: Nej. Jag kan va en doktor om jag får...

[No. I can be a doctor if I can...]

Sarah: Vill du va en doktor?

[Do you want to be a doctor?]

Astrid: om jag får ge sprutor på a::lla. ((laughing))

[if I can give injections to everyone.]

Sarah: Då får du ge sprutor. Det kan man göra om man är en sjuksköterska också.

[You can give injections then. You can also do that if you’re a nurse.]

If you’re a nurse.

Astrid: Man ska lägga i sängen.

[You should lay in bed.]

Sarah: ((Looking puzzled)) Ah, patients lie in bed. That’s not their job. That’s what they do when they’re ill. My job is I’m a nursery schoolteacher, and Kalle’s job is he’s a property technician, and dad’s job is he’s a salesman.

In the first two turns, Sarah and Liam are pursuing a response (Pomerantz, 1984) from Astrid. Astrid ultimately exclaims that she could become a doctor if she was allowed to give injections to everyone, which she finds humorous. These initial exchanges all take place in Swedish. Next, we see a translingual turn from Sarah. She says, in Swedish, that you can also [give injections] if you are a nurse “det kan man göra om man är en sjuksköterska också”. She then turns to English with “if you’re a nurse”, making the point and its reiteration more visible. This is a word for word translation of part of her Swedish utterance. Sarah is again orienting here to a bilingual medium of interaction (Gafaranga, 2010), although throughout this particular episode as a whole, the children orient to a monolingual Swedish medium. In the next turn, Astrid indicates that “man ska lägga i sängen” (you should lay in bed), which initially does not seem to make sense to Sarah, as she looks at Astrid with a puzzled expression. This is likely due to Astrid using lägga (lay) instead of the more typical ligga (lie). Sarah then tries to clarify the situation by explaining that it is patients who lie in bed, which is not their job, but it is what happens when they are ill. She goes on to clarify what a job is by stating that she is a nursery schoolteacher, Kalle (Sarah’s partner who lives elsewhere) is a property technician, and the children’s biological father is a salesman. This entire turn was done in English.

Finally, in Excerpt 4, Sarah, Amanda, and Astrid are rolling out gingerbread dough in preparation for baking gingerbread cookies. This is a typical activity that many people in Sweden do near Christmas.

Excerpt 4 – Thorén recording

Astrid: Va ska jag göra?

[What should I do?]

Amanda: Pepparkakor.

[Gingerbread.]

Sarah: Rulla den tunt.

[Roll it out thin.]

Roll it out so it’s nice and thin.

Figure 1

Astrid (left), Sarah (centre), and Amanda (right) making gingerbread.

The interaction was taking place in Swedish before this point and continues in Swedish here with Astrid’s first turn “va ska jag göra” (what should I do?), to which Amanda responds “pepparkakor” (gingerbread). Anna then produces a multilingual turn, which is reminiscent of the turn in Excerpt 3 where Sarah translated to “if you’re a nurse”. Here is a similar action directed at Astrid where Sarah translates and expands on the original Swedish utterance, “rulla den tunt” to English, “roll it out so it’s nice and thin”. A double pedagogical motive could be seen here. On the one hand, the translation functions as a strategy of bringing the English meaning into focus, while on the other hand, it functions as a way to emphasise the action that is required (i. e., to roll out the gingerbread dough). This turn has an embodied component, as seen in Figure 1, where Sarah makes a circular motion with her hand in combination with her utterance. When adopting a rhizomatic approach, modalities beyond speech should also be given due consideration.

Although this family orients to Swedish and English as separate entities in the interview, in Excerpt 2, they can be seen combining the languages in creative ways, where both languages come together in interaction, which is completely unmarked and unproblematic for the family. This results in a co-constructed repertoire which blends Swedish and English into a unit where the apparent differences between Swedish and English, and their status as separate named languages, become interactionally irrelevant. This translingual family repertoire (Hiratsuka & Pennycook, 2020) functions as a resource for communication, which is successful and perfectly capable of conveying the desired social action in this context. Excerpts 3 and 4 show Sarah moving between English and Swedish in single utterances, and here a more deliberate orientation to English and Swedish as separate entities is observed, indicated by her translations from Swedish to English. It is interesting to note that Sarah’s translations were not explicitly requested by any of the children.

Stimulated recall data do reveal a pedagogical motive for this kind of discourse strategy. Sarah is trying to ensure that her children become proficient in a number of different domains in both Swedish and English, and thus uses translation into English as a linguistic model aimed at the children. In the interview, Sarah mentions that she is satisfied with her eldest and youngest children’s abilities in English and Swedish, but she expresses regret towards the language competences of her middle child, which she attributes to her own historical language practices. She says that with the first child, she adopted a stricter language policy, where she focused more on English, but after the second child was born, she says that “Swedish was so natural to me that talking English again, it was a huge effort and I didn’t really understand the importance of it.” However, upon the birth of her third child, she reemphasised bilingualism as an important outcome for her children’s futures, as was also unanimously discussed in the interviews by all families. A link can be drawn here to conceptual societal discourses, in particular, the preference for raising bilingual families and the link between bilingual parenting as good parenting (King & Fogle, 2006). Although bilingualism is oriented to as the preferred outcome for all families, the role of English and Swedish in these children’s speculative futures is not equal. Swedish represents the local, the domestic, and the ordinary while English represents the international, the global, and the possible in terms of future social, educational, and career opportunities. As the father from the Clark family puts it, “we think it’s important they [the children] maintain their English because Swedish is not the most global of languages”. All parents adopt an explicit neo­liberal language ideology (Curdt-Christiansen, 2020, p. 189) to a certain degree and are well aware of the economic power of English as a global language, but equally, the potential for English to function as a vehicle of education internationally and thus social mobility. The parents, however, emphasise that the preferred variety of English is British English. Sarah states in the interview that she wants her children to “sound English” and that she hates it “when they sound American”. Here she draws a connection between language and identity, stating that language is “not just communication” but also a “culture thing”. She says she wants her children to “feel British and English as well as Swedish” drawing an explicit link between language and identity. Anna from the Andersson family, on the other hand, emphasises her desire that her children speak “correct English” and says that if the children were to speak a mix of English and Swedish, she would “correct them because I don’t want them to learn Swenglish”.

Returning to Sarah, I see the entanglement of her lived experience of language and the observed interaction order. Sarah discusses how she moved to Sweden in her early twenties, at a much younger age than the other foreign-born parents in this study. She says that she had always been interested in languages and had studied German previously. It was in fact on a language exchange to Germany where she met the Swedish father of her children. Sarah reports having had quite an easy time when learning Swedish and soon after started working in Swedish and using Swedish as an everyday means of communication. Here, a link can be drawn between Sarah’s experience and Pujolar & Puigdevall’s (2015) concept of the linguistic muda, a specific biographical juncture in which a shift in language use takes place. This can also be seen as a shift in her chronotopic identity (Blommaert & De Fina, 2016). Sarah’s ability to use the many different forms and registers of Swedish which make up her Swedish repertoire is indexed by her undertaking of university studies in Swedish, by her current work life in a monolingual Swedish environment, and through the observation that her children do not react negatively to her speaking Swedish in any way. In many of the other families, when the foreign-born parent uses Swedish, this is readily mocked and ridiculed by the children.

7 Human and non-human family configurations

Thus far, I have considered interactions which take place between human actors. I now turn to two instances where the interaction order within the home is influenced by a spatial configuration that contains inanimate objects. Through interaction, these initially inanimate objects acquire “the curious ability of inanimate things to animate, to act, to produce effects dramatic and subtle” (Bennett 2010, p. 6). The effects I will focus on relate to the ability for inanimate objects to influence the enactment of otherwise invisible linguistic repertoires of family members.

Excerpt 5 takes place while the Pearce family are preparing to eat dinner. At the start of this sequence, the mother and the father of the family are in the kitchen (seen behind the wooden doors in Figure 2). They are discussing some mundane details of getting the dinner ready, which occurs in English. Elsa (age 3) is in the living room by herself with a toy horse (see Figure 2, left). What occurs next is that this toy horse starts making a neighing noise and starts moving around. This is due to the horse being electronic, and Elsa interacted with the horse in a way that induced this animated state. Following this, Elisabet, the mother, shouts from the kitchen “Elsa! No horses at the table!” This invokes the family’s expected norms of interacting in this space (the dining room) at this socially defined time (dinner time). The immediate future is of the upmost importance for this interaction. The family evidentially has a policy which mediates what is acceptable behaviour during dinner. John, the father, then enters to try and enforce this policy with “Elsa, put horsie away”. He then turns to Swedish while looking at the horse with “Hej då horsie. Hej då”, he waves at the horse (Figure 2, right), and finally returns to English with “bye bye”.

Excerpt 5 – Pearce recording

((Toy horse starts shaking and neighing)).

Elisabet: ((From kitchen)) Elsa! No horses at the table!

((The horse falls off the bench and under the table.

Elsa goes under to the table to get the horse)).

John: ((John enters from Kitchen)) Elsa, put horsie away.

Hej då horsie. Hej då. ((John waves at the horse))

[Goodbye horsie. Goodbye.]

Bye bye.

Elsa: Bye bye.

John: Bye bye.

Figure 2

Elsa and horsie (left). John saying goodbye and putting away horsie (right).

Throughout the video data, John is almost never observed using Swedish. However, here, in this assemblage of time, space, material, and social expectations, he resorts to Swedish. He draws on a wide semiotic repertoire in order to accomplish this social action. The repertoire he draws on combines a ‘baby talk’ register in English and Swedish with embodied action in order to get Elsa to put the horse away. These resources may be directed at Elsa, rather than the horse, but one can also observe an affective component in this assemblage regarding the horse. The horse is not simply put away without comment, instead, a ‘farewell ritual’ is enacted (Goffman, 1967). John and Elsa orient to the horse as having face needs that must be addressed, and thus they must appropriately say goodbye.

Another example of non-human interaction was found in the Clark family. In this sequence, the Clark family interacts with a Google smart speaker. The smart speaker is a new acquisition for the family. During this excerpt, the camera is pointed at the dining room table, where Karin (the mother) and Oliver (the son) are sitting. The interaction starts with Martin (the father) summoning Emma (the daughter) to come and see what he is doing.

Excerpt 6 – Clark recording

Martin: Emma come here I wanna show you something!

...

Hej Google. Hur mycket är klockan?

[Hi Google. What time is it?]

Google: Klockan är fem över sex.

[The time is five past six.]

Martin: Hej Google. Vad är väder?

[Hi Google. What is weather?]

Google: Just nu i [town name] är det noll grader celsius och klart.

[Right now in [town name] it is zero degrees celsius and clear.]

Martin: Jag pratar svenska nu!

[I speak Swedish now!]

Emma: Hej Google. Varför pratar du svenska?

[Hi Google. Why do you speak Swedish?]

Google: Jag lär mig mer och mer varje dag som går.

[I learn more and more every day that goes by.]

((Emma laughs))

The smart speaker requires the command ‘Hej Google’ (Hi Google) in order to start ‘listening’. I therefore will refer to the smart speaker as ‘Google’. In the first interaction between Martin and Google, he asks, in Swedish, what time is it? Google responds, in Swedish, that it is five past six. Then, Martin asks Google “vad är väder?” The action that Martin is trying to accomplish is presumably to find out what the current weather is. What he says, though, literally translated, would be “what is weather?” This is not an idiomatic way of asking for the weather in Swedish, indicating Martin’s unfamiliarity with this everyday Swedish register. Nevertheless, Google has no problem in interpreting Martin, and gives the current weather forecast. Upon hearing this, Martin exclaims “jag pratar svenska nu!” (I speak Swedish now!) Here, Martin is making reference to his past self, who apparently was unable to speak Swedish, but the interaction with Google shows that Martin’s current self was actually able to successfully use resources from Swedish in order to accomplish an action. The accomplishment, and the associated affective dimension, is achieved in collaboration with Google. Martin’s perceived success in communication has not resulted from an interaction with another human being, but rather an interaction with a non-human, an example of ‘thing-power’ (Bennett, 2010). Martin’s interaction with Google can be seen as an assemblage formed through the interconnection of the interaction order, enacted as the two question-answer adjacency pairs; his historically situated self, his previous status as one unable to speak Swedish; and on conceptual discourses which relate to what it is to speak a language.

The non-hierarchical rhizomatic approach should not offer a privileged position to only certain types of interaction between certain interactants. As has been shown, interactions between human and non-human actors can reveal the otherwise invisible ability of an actor to deploy specific semiotic resources. This kind of data can contribute to a greater understanding of the resources that participants have access to and thus can give important information for the researcher attempting to build a picture of a FLP. Questions may be asked as to why a family member is able to use semiotic resources associated with one language within these types of interactions, while in other spatial configurations, such resources seem to be inaccessible.

8 Discussion and conclusion

The analyses above begin to build a picture of how the elements within the rhizomatic structure that is the FLP in these families come together in momentary assemblages. Although each assemblage was unique, each was formed through the interaction between socially constructed spaces, the evocation of past, present and future timescales, material elements, affects, agentive forces, and semiotic resources. The availability of acquired semiotic resources is mediated by the configuration of the other elements in the assemblage. Table 2 is a non-exhaustive list of the entangled elements which were shown to be relevant in understanding the interactions considered within the analysis.

Table 2 illustrates the complexity involved within FLPs. Listed are just some of the elements which were found to be relevant within in a small set of family interactions. The link between each element ranges from weak to strong; the forces are different in each of the families. Certain elements have a relatively inconsequential relationship with each other, while other elements are intimately bound together. It is through the tracing of these circuits that their role within a FLP can be understood.

Historically, studies on language within multilingual families have privileged the position of the parent over the child and have situated the agency to decide over language practices with the parents. More recently, the view that children are agents in their own right and influence language regimes has received much acceptance within the field (see, e. g., Smith-Christmas, 2020). I argue that the field should continue the expansion of how agency is conceptualised within the family domain and should begin to take into consideration how agency is dispersed and distributed across a wide range of entities, including the nonhuman. The analyses presented in this article highlight the ability of initially inanimate objects to become animated (i. e., ‘Horsie’ and ‘Google’) and produce effects and affects. It was shown that these actants were able to interact with human beings in a way which led to them deploying non-typical linguistic repertoires. Without these non-human entities, such repertoires would not have been visible. In line with rhizomatic thinking, a posthumanist, non-anthropocentric view of agency would seem to provide a more accurate frame (at least in certain cases) that shows how family members and their language practices are “enmeshed with the environment and technology” (Nayar, 2014, p. 4). Perhaps less obvious, but equally important, is the agentive capacity of conceptual discourses. For instance, Sarah’s desire to be a good mother through the raising of children who are bilingual clearly affects her language practices, as seen in Excerpt 3 and 4. What also seems to affect her language practices and her interpretation of conceptual discourses is that she inhabits a single parent household. An explicit focus on research designs which recruit only idealised families would have potentially excluded this single parent family, and thus several of the insights gained from this study. Many of the connections explored within this article were gained through the research design adopted within this project, which drew on diverse data types. Without a combination of data sources and perspectives, only a small fragment of these assemblages would have been able to be observed and understood.

A rhizomatic approach to FLP exemplifies how ongoing interactions between family members are intimately entwined with both past and future temporalities. The full semiotic repertoire which family members have at their disposal is historically situated; it is the product of their past interactions, their lived experience of language. Their knowledge of when specific semiotic resources are allowed to be deployed and when they are to be restricted is also due to their lived experiences (c.f., Busch 2021). The future is equally important. The near future dictates many of the interactions observed: getting ready for dinner, joining a new football team, bringing gingerbread to a Christmas gathering. The distant future also plays a mediating role in language policies and practices. Through speculation and future-oriented narratives (Pietikäinen, 2021, p. 7), the parents in these families all report the importance of bilingualism for their children’s futures. Bilingualism expands beyond named languages to encompass a myriad of pragmatically employable codes, registers, and modalities to be used for various speculative future pursuits. English, or rather, specific forms of English linked to global possibilities, are seen in particular as an important piece of knowledge capital (Braidotti, 2019), which the English-origin parents are able to imbue in their children; it is reasoned that a failure to raise a bilingual child in a Swedish-English family is a failure of parenting (an ideology grounded within a situated conceptual discourse). This, combined with the asymmetrical language proficiency observed between parents within this context, makes it unsurprising that the default policy within two-parent Swedish-English families seems to be a preference for English within spatial configurations that contain the English-origin parent. It can be theorised that the continued asymmetrical language proficiency results from the interaction of several components, namely, an ideology (and lived reality) that high levels of Swedish proficiency are not needed to live and survive in Sweden, the perceived difficulty of learning a new language as an adult, the English proficiency of the local population, the acceptability and ability of English to function in everyday communication, as well as in some cases, a potential that in the long term, the families may move elsewhere, which again is interconnected with the potential of English as a global lingua franca.

A rhizomatic approach is a way to understand complexity within FLPs and to consider the connections between elements which may have been unexpected. It allows for investigations where the researcher is forced to consider their preconceived ideas about language and how people experience the world. It demands a more encompassing view of agency, which goes beyond the most obvious actants and considers how agency is distributed and relational. This study has only exemplified a small number of temporary assemblages of thought and action. In order to build a holistic picture of the rhizomatic policy system that is a FLP, multiple assemblages from the same family should be analysed, as well as the connectivity between such assemblages. It is from analysing a multiplicity of such assemblages that one can assess the reoccurring discourses which seem to be most relevant and carry the most force in any FLP. FLP as a research field offers much potential for understanding topics relating to language in society more broadly (language shift, language maintenance, language socialisation, literacy etc.) as well as more overarching issues relating to, for instance, education, equality, identity, and migration. However, my view is that if FLP as a field is to meaningfully contribute to larger societal issues, then we must attempt to more comprehensively understand the multitude of interconnected factors upon which a FLP is founded. It is only through engaging with this complexity that accurate depictions and understandings of FLPs can be formed. In this article, I have argued that this complexity can be engaged with through the adoption of a rhizomatic approach and by conceptualising FLP as a rhizomatic policy system.