Using a ‘Family Language Policy’ lens to explore the dynamic and relational nature of child agency

This article contributes to a dialogue between childhood studies and the sociolinguistic subfield ‘Family Language Policy’ (‘FLP’). The article argues that the two fields provide complementary vantage points for exploring child agency. It explains a revised version of a model I developed to conceptualise child agency in FLP, consisting of four intersecting dimensions: compliance regimes; linguistic norms; linguistic competence and generational positioning (Smith- Christmas, Handbook of home language maintenance and development . De Gruyter Mouton, pp. 218– 235, 2020a). The article examines two conversational excerpts as


INTRODUCTION
The question of what agency means and how humans achieve agency in their various social milieu is the subject of much discussion within the social sciences (e.g. Bourdieu, 1997;Giddens, 1979;Taylor, 1985). From these discussions emerge a conceptual dichotomy, with 'agency' positioned in opposition to 'structures', yet simultaneously, an understanding of the reflexive relationship between the two-that is, their role in mutually shaping each other. As Giddens and Turner (1987, p. 8) put it, 'agents, action, and interaction are constrained by, yet generative of, the structural dimension of social reality.' The field of childhood studies has been formative in advancing the concept of children as agentive beings (e.g. Alanen & Mayall, 2001;Prout, 2005). However, the treatment of child agency within childhood studies has also been critiqued on a number of fronts. Critiques lie in the lack of conceptual theorisation (Valentine, 2011), especially in terms of the side-lining of 'structures', as well as the treatment of 'children have agency' as a mantra within the field (Gallagher, 2019;Tisdall & Punch, 2012), and also for the tendency to read agency as 'a mere expression of "resistance" or "resourcefulness"' (Abebe, 2019, p. 8). Another critique is the opaqueness of the concept of 'competence' within discussions of child agency (Moran-Ellis & Tisdall, 2019). From these critiques arise the need to frame adult-caregiver relationships in terms of 'interdependence' (Punch, 2001), and the imperative to examine how child agency arises from practices over time and space (Bollig & Kelle, 2016). Child agency, therefore, must be seen as fluid and relational, especially in terms of role of generational positioning in shaping the ways in which children can enact their agency in everyday interactions (Esser et al., 2016).
These latter concepts resonate with an evolving understanding of 'child agency' within the relatively new sociolinguistic subfield 'Family Language Policy' (henceforth 'FLP; ' King et al., 2008;Luykx, 2003; see Curdt-Christiansen & Huang, 2020 for a recent overview), which has grown out of the coalescence child language acquisition studies and language policy research. Initially, the focus of this field has lain in elucidating the processes which lead to language maintenance-that is, caregivers transmitting a minority language (which refers to the lesser used language in the children's immediate sociocultural environment) to their children. This work has shown that linguistic practices which align with a pro-minority language FLP often result in successful language maintenance (Altman et al., 2014;Lanza, 1997). This emphasis on language maintenance can account for the tendency for child agency in FLP to be seen as subversion, that is, the child resisting the parameters of the FLP, which tends to equate to the child's use of the majority language (Smith-Christmas, 2020a). The other main arena for FLP explorations of child agency has lain in children's potential to disrupt the power differential along generational lines through the child's greater linguistic competence in the majority language, a common situation in transnational families (Antonini, 2016;Crafter and Iqbal, this volume;Revis, 2016). Like childhood studies, in FLP, there also exists the tendency to view child agency as either '"resistance" or "resourcefulness"' (Abebe, 2019, p. 8). Like childhood studies, a further critique of FLP's treatment of child agency is the lack of conceptual definition. A number of FLP studies (e.g. Bergroth & Palviainen, 2017;Fogle & King, 2013;Gyogi, 2015;Said & Zhu, 2019) invoke linguistic anthropologist Ahearn's (2001, p. 112) definition of agency as 'the socioculturally mediated capacity to act', yet there is little discussion within the field on the intricacies of applying this definition to child agency per se. One key difference between FLP and childhood studies, however, is that while childhood studies have been critiqued for its treatment of child agency as a mantra. FLP can be critiqued for the opposite: in spite of early work which emphasised the importance of child agency (e.g. Fogle, 2012;Luykx, 2003), child agency has remained at the periphery of the field until recently. I see FLP and childhood studies as offering complementary vantage points to exploring child agency. In addition to the complementarity inherent in FLP's focus on 'structure' (i.e. language maintenance) in contrast to childhood studies' emphasis on agency, there is also complementarity in terms of scope: while childhood studies' scope is wide, taking in the entirety of children's experiences with a variety of social actors in different contexts, FLP's scope is relatively narrow, as it centres primarily on the child's linguistic experiences in relation to family life. This relatively narrow focus I believe has the potential to help balance the more abstract theoretical concepts explored in childhood studies. For instance, in FLP, we tend to examine 'competency' through a linguistic lens-a lens which entails many layered complexities and challenges, but for which nonetheless exists at least tacit understanding of what it means to be a 'competent' speaker of a language as well as rubrics by which to assess a speaker's competency at a particular age (see Fletcher & MacWhinney, 1995 for an overview). Further, the focus in FLP on language practices provides a clear way to explore the concept that agency arises from practices as they evolve over time and space (cf. Bollig & Kelle, 2016).
To the best of my knowledge, until this special issue, there has been very little dialogue between childhood studies and FLP. My own previous work on child agency in FLP would have greatly benefitted from insights from childhood studies, but I am ashamed to say that, like other FLP researchers, I tended to fail to look much beyond applied linguistics (and occasionally developmental psychology) in my previous explorations of child agency in FLP. I see this special issue as an important step in rectifying this oversight and to contribute to a much-needed dialogue between the two fields. In embarking on this endeavour, I will first introduce a model I developed to conceptualise child agency in FLP, then apply this model to analysing selected conversational excerpts as a means to illustrating the dynamic and relational nature of language choice and child agency in family interactions.

CONCEPTUALISING CHILD AGENCY IN FLP RESEARCH
As explained in greater depth in Smith-Christmas (2020a), in conducting a review of child agency in FLP research, I developed a model to conceptualise how children's linguistic choices in family interactions can be considered agentive. Although it was not my necessarily intention to build a model in reviewing the FLP literature on child agency, I felt stymied by what I perceived as a lack of conceptual clarity on how agency was applied to children's language practices. I, therefore, set about building a model as a means to not only to synthesise the literature, but to help me grapple with some of the conundrums I encountered. In devising this model, I anchored my understanding of child agency within developmental psychologist Leon Kuczynski's (2002, p. 9) definition of agency as 'individuals as actors with the ability to make sense of the environment, initiate change and make choices' and FLP researcher Lyn Fogle's (2012, p. 41) key question of 'at what point can children have an influence on the construction of family language policies?' Analysis of FLP literature led to me argue that analysing how children's linguistic choices can be considered agentive should take four dimensions into account: compliance regimes; linguistic norms; linguistic competence and generational positioning, as outlined in the figure below. The concentric circles in the model serve to embed the child's individual linguistic choices in the accretion of family practices (inner circle) and the interactional milieu of wider society (outer circle). The double-pronged arrows show the reflexivity between these different layers, thus emphasising how children's linguistic choices are both shaped by and, in turn, play a role in shaping different interactions over space and time ( Figure 1).
As Punch (2002, p. 126) describes, there are many reasons why a child may make a particular choice and it can be hard to discern which factor is most influential. Language choice is no exception, and it should be emphasised that the four different dimensions in the model are intended to be read as porous and intersecting. Indeed much of the model came to fruition through an inductive understanding of the interrelated nature of the dimensions to categorise them as 'dimensions' in the first place. As previously mentioned, in FLP, child agency has often been seen through a resistance lens, that is, the child going against the FLP. The easiest way to begin to conceptualise child agency in FLP, therefore, was to start with a compliance/non-compliance paradigm (see Kuczynski & Hildebrandt, 1997). In FLP terms, compliance is usually couched as the child speaking Language X (which, for the purpose of this paper, is equivocal to the minority language). Although indeed caregivers may sometimes explicitly say 'Speak Language X' to the child, in FLP, 'Speak Language X' usually takes more subtle forms. For example, as Lanza (1997) illustrates in her landmark FLP study, the caregiver may not answer the child if he or she uses the 'wrong' language; alternatively, the caregiver may repeat the word in the 'right' language. Over time, the accretion of these compliancy regimes may lead to certain linguistic norms, which are conceptualised as the shared expectations that interlocutors have when interacting with each other. Determining at what point these accretions may constitute a norm-and how this impacts how we read certain linguistic choices as agentive or not-became a key question in building the model. One of the conundrums I wrestled over relates to Gafaranga's (2010) work on the Rwandan community in Belgium, which is tends to be cited as one of the earliest examinations of language choice and child agency in child-caregiver interactions (e.g. Gyogi, 2015;Revis, 2016). In Gafaranga's work, caregivers address the children in Kinyarwanda (the minority language) and the children answer in French (the majority language), but caregivers either (a) do not mark the child's choice as non-compliant, but simply continue speaking in Kinyarwanda while the child continues to speak French or (b) switch to French themselves. These two scenarios have become norms within this community, which begs the question: to what extent can the children's use of the majority language French be read as 'resistance' and to what extent can it be read as a norm of interaction?
Similarly, the issue of the children's linguistic competence-in other words, the child's command of grammatical structures and lexical items-illustrates other conundrums. In Gafaranga's case, the children have lower linguistic competence in Kinyarwanda compared to French, which further begs the question: to what extent can the child's reply in the majority language be seen as agentive if the child does not have the linguistic competence to answer in the minority language, even if the child wanted to? In FLP, linguistic competence is also often inextricably linked to generational positioning, the fourth intersecting dimension in the model. In my original model, I referred to this as 'power dynamics', but after engaging with the childhood studies literature, I feel that 'generational positioning' better encapsulates what this dimension entails, as my primary interest lies in how the generational order shapes language choice and especially how the child's greater competence in the majority language (a common occurrence in transnational families) can subvert the expected generational roles. As Revis (2016, p. 9) illustrates with her example of a caregiver exclaiming "They were the ones who translated for many people and I felt like 'I'm not the child! I'm the mother in this house'," instances where children perform language brokering tasks can reconfigure expected generational roles, sometimes causing friction within the family.

F I G U R E 1 Child agency in FLP research
In further re-examining this model in light of childhood studies discussions of child agency, I realise that one potential weakness with my model is its dichotic starting point, as it was very much developed to help explain the conundrums associated with determining whether a particular language choice was agentive or not. However, I believe that not only is the model still useful for synthesising the main issues in discussing child agency within FLP, but that the model highlights the relational, negotiated nature of child agency. I argue that by considering the intersectional dimensions of the model, as well as the reflexivity between the concentric circles (the family and society), the model provides a means for discussing how agency is co-constructed in conversation, and how this contributes to the family's FLP. To illustrate this point, I will focus on two conversational examples which I believe best illustrate the intersectional nature of the four dimensions and how it impacts the FLP. The following, therefore, will describe the research context and methodology, then move on to the analysing specific conversational examples.

RESEARCH CONTEXT AND METHODOLOGY
The analysis in this article is situated in data collected as part of the Marie Skłodowska-Curie Actionfunded project 'Languages, Families and Society' (LaFS grant no: 794800). The project focuses on six families in the Republic of Ireland (total population 4,761,865 people as of the 2016 Census), which was chosen as a case study site in part because of its rather singular position as a nation in Europe where, since Irish independence from Great Britain in 1922, the autochthonous minority language Irish has national status. However, despite high levels of institutional support for Irish, families who use Irish as their primary home language exist as minority of the population, as only 73,803 people report use Irish on a daily basis outside the education system (see also Ó hIfearnáin, 2013 for analysis on FLP in an Irish context). Another reason why Ireland was deemed a fruitful case study location for this FLP study is that immigration to Ireland-which historically has been a country of emigration-is a relatively recent phenomenon. The largest group of non-Irish-born nationals are Poles (122, 515 people) and 135, 895 people use Polish at home, 108, 698 of whom were born outside of Ireland and 27, 197 of whom were born in Ireland (Central Statistics Office, 2016). These numbers, particularly in terms of the number of Polish speakers born in Ireland, appear to indicate a high degree of language maintenance (see also Connaughton-Crean, 2020; Machowska-Kosciak, 2020).
As the aim of LaFS is to gain an in-depth picture of each families' linguistic experiences, the project takes a case study approach to looking at two families who speak Irish as a home language; two families who speak Polish as a home language (as well two families who speak Kurdish as a home language, for which data collection is still ongoing). The families all live within commuting distance of Galway City (total population: 76,953 people) in the west of Ireland, in which 2,344 people report using Irish daily outside the education system and the most-spoken language besides English or Irish is Polish (4,310 speakers), with Polish nationals (3,905) accounting for the third largest nationality in Galway after Irish and UK nationals. LaFS seeks to gain a detailed picture of language in the families' daily lives without having to employ the 'deep' ethnography I have implemented in other studies (e.g. Smith-Christmas, 2016, where I lived with the family). Thus far, cumulatively the families have provided over 24 hr of audio and video data through the project's main modes: first, a multimodal 'language diary', where families are given an iPad and asked to record anything that they feel encapsulates their day-to-day experiences of 'language', which can be in the form of audio and video recordings, photographs or 'traditional' written diary reflections. Family members then upload these language diaries to a secure OneDrive folder they share with me. The second form of data collection is audiorecorded visiting sessions, where I the researcher discuss the family's linguistic experiences and invite them to reflect their language experiences. The project received ethical approval from the National University of Ireland, Galway Research Ethics Committee (Reference: 19-AUG-19). One parent signed the consent form on behalf of the family, but as best practice, I would always seek all family members' assent in making the recordings during each visit. I would subsequently upload the audio recordings of the visits in the shared OneDrive folder, thus allowing family members the opportunity to listen back to the recordings and ask for any part to be deleted if they wished. The caregivers also tended to ask for the children's assent when making the recordings for the family's language diaries. All names have been changed to protect anonymity and I have also been intentionally vague on the families' exact locations for this reason as well. At the time I undertook the analysis for this article, approximately 6 hr of recorded data, consisting both of my audio-recorded visits as well as the families' own audio and video recordings, had been transcribed and were included in the corpus of spoken data.
The two data excerpts analysed in this paper are taken from two families' interactions as part of their language diaries. The choice to focus on these two excerpts was motivated by an initial coding process (conducted using NVivo) where I analysed the corpus of transcribed recordings and identified particular excerpts which appeared to best illustrate one or more of the dimensions discussed in the model. In looking through this sub-corpus, I realised that two of the examples I had identified appeared anomalous in that the children's use of the majority language actually worked in tandem with the caregivers' reification of the prominority language FLP. These two conversational excerpts are analysed in this paper using a microinteractional approach to language alternation (see Auer, 1984), which has been the central analytic framework employed in my previous similar work on language in family interactions (e.g. Smith-Christmas, 2016). In taking this approach, the focus lies in the finegrained details of the conversation, complemented by my knowledge of the family's wider language practices and the sociocultural milieu in which they move. I feel that the seeming anomaly of the children's use of the majority language actually reifying the prominority language FLP provides a good way to illustrate the interrelated nature of the four dimensions of the model, and in doing so, allows me to show the contextual and relational nature of agency.
The first excerpt is taken from an interaction in the family I have given the pseudonym the 'Wieniawski family' and I am indebted to fellow FLP researcher Dr. Maria Obojska to giving me an introduction to the Wieniawskis. The Wieniawski family consists of the mother, father, both born and raised in Poland and their two daughters, Anita (aged ten) and Kasia (aged eight), who were born in Ireland. The Wieniawskis live in an average income Galway suburb, and both daughters attend a nearby English-medium school. I first met the family in September 2019 and over the course of a year they provided a snapshot of their family life through their language diary, totalling approximately four hours of audio recordings and videos, in addition to photographs and 'traditional' written diary entries. I was able to meet with them six times, resulting in approximately 6 hr' of audio recordings. It is very clear that the Wieniawskis practice a very pro-Polish FLP: Polish is strongly their home language and the parents expect that the daughters will respond to them in Polish. Although not speaking Polish myself as the researcher has had certain disadvantages, one silver lining has been that it has afforded the daughters the opportunity to demonstrate their rich metalinguistic skills through their explanations of the recorded material, and Anita and Kasia even took part in some of the transcribing process with me. The rest of the transcriptions and accompanying translations have been completed by freelance transcribers in Poland.
The second excerpt is from an interaction involving a family I have given the pseudonym the 'Clancy family.' I met the Clancys in January 2020 through a call for participants I sent through various Irish language organisations. The Clancy family consists of the mother, the father and three boys: Cian (11 years old); Daithí (10 years old) and Aodán (6 years old). The Clancys live in a more rural area of the Galway suburbs which is classified as a 'Gaeltacht' area, which means that the three boys all attend Irish immersion school. The father can be classified as a 'new' speaker of Irish (see O'Rourke & Walsh, 2020) in that he was not socialised with Irish in the home as a child, but became a fluent and daily speaker of Irish through education and other more informal language opportunities. The mother does not speak Irish, as she was born and raised abroad and moved to Ireland after the couple was married. Because of COVID-19, I was only able to visit with the family officially twice in-person and then once online (thus totalling about an hour and a half of audio recordings), as while online visits worked well with other families, unfortunately, the Clancys' broadband connection did not appear reliable. However, I had met with the father prior to conducting the study and we had a very in-depth conversation about his language practices; as well, through our mutual involvement in Irish language activities, I have had the opportunity to observe the father's Irish language activism-inaction. The Clancy's language diary consists of 40 separate written entries compiled by Cian, all of which are in Irish, which are generally short sentences, though some are longer reflections and involve recounting whole conversations. Most entries are accompanied by photographs and short videos. The father also added a written diary entry and two videos, totalling approximately 10 min. I admittedly, therefore, do not have as in-depth picture of the Clancys' family life as I do of the Wieniawskis, but from the data that I do have, it is very clear that the father makes a very concerted effort to use Irish with his sons. As well, the fact that I speak Irish means that I was able to observe how the family's language practices unfolded in situ during the visits, whereas for the Wieniawskis, I of course knew they were speaking Polish, but was unable to understand their intrafamily conversations. The data were transcribed by a freelance transcriber in Ireland and I have provided the translations myself.

THE WIENIAWSKI SISTERS: ASKING PERMISSION TO USE THE MAJORIT Y LANGUAGE
The Wieniawski family practises a pro-Polish FLP and there are very few instances in the corpus in which the daughters appear to 'break' this pro-Polish FLP by using English. There are, however, some exceptions, one of which is when the daughters are discussing aspects of their school day with their parents. Both parents say that they apply more leniency to the girls' English language use in this instance, as they say they think that it is sometimes challenging for the girls to talk about concepts/events that occurred through the medium of English. However, it is clear from the corpus that although the parents may indeed apply more leniency in these instances, they still reify a pro-Polish FLP, either by encouraging the girls to elaborate on the English language content using Polish, or providing the Polish language equivalent for the English language item (see Smith-Christmas, 2021). The second instance in which the Wieniawski sisters go against the grain of the pro-Polish FLP is in their interactions with each other, as they report often using English when speaking together. The use of the majority language among siblings is not uncommon and can be a potential precipitating step for language shift within the family (Kopeliovich, 2013;Mirvahedi & Cavallaro, 2020). The following example-in which the sisters describe how they were inspired by a TikTok video to do something that involved putting glue for fingernails on their lips-illustrates the premise that the Wieniawski sisters often use English together. They use English both in rendering the constructed dialogue (i.e. saying what they said in an earlier conversation) and use English to argue with each other in the immediate interaction. However, although they are using the majority language, the following analysis shows that their actions within this sequence reify the FLP rather than go against it: Here, we see the multilayered and interdependent nature of how child agency is enacted through language and the relationship of this agency to the FLP. In Turn 1, Anita asks her father's permission to speak English, and it is only after he signals his acquiescence that she uses English in Turn 3, using a very funny, stylised form of English to do so (which is how she appears to enact the funny 'posh ditsy' role, that I observed her take on occasionally. It should be noted that her use of 'done' is seen to be a potential localised feature, not a linguistic transfer feature). By explicitly asking her father's permission, Anita signals that she understands the parameters of the FLP: if she would have codeswitched to English without prior approval, her utterance would have constituted a breach of established interactional norms. In making this request, therefore, Anita is reflexively reifying the norm of Polish-as-language-of-interaction within the family, thereby also contributing to the pro-Polish FLP. These observations are important within the scope of FLP because of the aforementioned dichotic lens which is often applied to language choice, especially the de facto tendency to see the child's use of the majority language as a form of resistance and therefore agentive. In this instance, Anita's use of English is not considered a form of resistance, as it actually serves to reify the FLP. Therefore, the agentive nature of Anita's use of the majority language is embedded in her knowledge of norms and how they relate to compliancy; and how her ability to manipulate these norms solidifies her active participation in the pro-Polish FLP. Furthermore, in this instance, her request also reflexively strengthens her role as child vis-à-vis her father, as it indexes the generational order, that is, because of the power differential, children must seek permission for norm-breaching behaviour. Conversely, we also see the dimension of linguistic competence-and its potential to challenge generational power relations-in this excerpt. As explained in greater detail elsewhere (Smith-Christmas, 2021), Anita and Kasia are considered the 'experts' in English vis-à-vis their parents: although the parents are fully competent English speakers, they nonetheless consider their daughters the 'native' English speakers. Thus, the father's Turn 4 is seen to be an acknowledgement of his daughters' competency, both in terms of language and also knowledge of youth culture associated with technology (e.g. TikTok). Therefore, in this short excerpt, in addition to compliancy regimes and linguistic norms, we also see the interplay of linguistic competence and generational positioning.
This particular example also illustrates the multilayered and dynamic nature of FLP, and how this relates to child agency. Here, we see evidence of two parallel norms: first, the pro-Polish FLP with the parents and then the intersibling norm of English. Siblings enactment of their own norms are usually read as agentive acts (see Smith-Christmas, 2020b); however, the way in which this agency is situated in the conversation means that it does not challenge the pro-Polish FLP. Not only has permission been obtained to use English, but the English that is used in Turns 3 and 7 is marked as temporally removed from the conversation at hand. The stylised, creative way in which Turn 7 is rendered and the father's own use of English in repeating 'lip change' in asking if the act of putting nail glue on one's lips is called a 'lip change' further sets it apart from the immediate conversation; in other words, Anita is positioned as not actively breaking the FLP by using English in these utterances, but simply reporting what transpired in the original language in which it was said. The switch back to Polish after speaking in English further reifies Polish-as-language-of-immediate-interaction. Thus, we see the nesting doll nature of FLP: a norm within a norm, and from this vantage point, we see the multilayered, negotiated and dynamic nature of the FLP as it plays out in language practices. Now we will turn to the Clancy family for another example of the apparent anomaly of English use serving to reify the pro-minority language FLP and how this illustrates the brothers' agency in the FLP-making process.

THE CLANCY BROTHERS: LAPSES IN COMPETENCY AS SITES OF PRO-MINORIT Y LANGUAGE FLP REIFICATION
In comparison with the Wieniawski sisters, the Clancy brothers appear in the more nascent stages of their linguistic competence journeys; the eldest boy, Cian appears to be the most competent speaker of the three boys, and this is attributed to the fact that he has spent the most amount of time in Irish immersion education. Overall, the Clancy brothers' use of the Irish bears much similarity to children in other autochthonous minority language situations, where although they attend an immersion minority language school, the wider peer group and societal use of the minority language impacts their own language use, and consequently, their competency. This is even the case in families where multiple caregivers use the minority language to the children (Smith-Christmas, 2016) and thus, the fact that the Clancy brothers only have one Irish-speaking caregiver does not necessarily account for what appears to be their asymmetrical competence in Irish and English. What is striking about the Clancy brothers, however, is that unlike other contexts (e.g. Gafaranga, 2010), this apparent asymmetry does not seem to mitigate against their use of Irish with their father, nor do they appear to resist their father's frequent corrections of their Irish. These two premises are encapsulated in the following excerpt taken from my notes on my first family visit: Very clear that Irish is the established language of interaction between the father and the children. Father will often correct their Irish, but this does not seem to bother the boys at all-they willing [ly] repeat what he has corrected. For instance, at one point YB [Aodán] said 'Tá sé (X)' instead of 'Is (X) é' and the father corrected him. YB [Aodán] happily corrected his utterance and carried on.
In this excerpt, I note how the father corrects Aodán's grammatically deviant use of the copula (which in this instance is considered contact-induced, i.e., uses an English language, as opposed to Irish language, grammatical structure) to the correct form ('Is X é' 'It is a (X)'). Not only do these type of in-situ corrections have the potential to cause friction in child-caregiver relations, but as the goal of any caregiver correction is for the child to modify their behaviour (linguistic or otherwise), corrections are often seen as a means to curtail a child's agency, especially as it reinforces the traditional generational power asymmetry between caregivers and children. What is apparent from the Clancy brothers' interactions with their father, however, is that the father's practice of correcting their Irish has become established as a norm and that, as noted in this excerpt, the boys willingly correct their utterances to align with the lexical and grammatical forms suggested by their father. It should be emphasised that the brothers do not appear to sense repercussions if they fail to correct their utterances; rather, parallel to Fogle's (2012) observations of Russian adopted children in the US, the existence of the 'correction norm' allows the Clancy brothers to enact their agency by actively creating contexts for their own language learning. In addition to the grammatical corrections, another facet of this 'correction norm' is that the boys' use of an English language lexical item is often met with their father's use of its Irish language equivalent. Thus, if the child speaker does not know the Irish word for something, he can then rely on the father in a scaffolding-style approach. An example of this is given below, which is taken from a video where the father is driving the car and the three brothers are in the back seat. They are all playing the game 'I spy with my little eye' and Cian has just gained the floor. Daithí, however, intervenes and turns the topic of conversation towards his chewing gum. However, he appears to face difficulties in terms of constructing his desired utterances and requires his father's assistance in formulating his subsequent statements: In this excerpt, the father's Turns 10, 12 and 14 align either with what Lanza (1997) terms the 'expressed guess strategy', where, using the minority language, the caregiver reformulates what they think the child has said in the majority language, or the 'adult repetition strategy', where the adult repeats the child's majority language utterance in the minority language. As Lanza discusses, these strategies are integral to reifying a pro-minority language FLP and thus, as emphasised earlier, are seen as part of the 'structures' element of FLP. What is relevant to our discussion of the interdependent nature of agency, however, is Daithí's role in this process. It is clear that his use of English in saying 'chewing gum' in Turn 9 is not because he is resisting his father's use of Irish, but rather, that he lacks the competence to do so. As mentioned earlier, my recent treatment of child agency and FLP (Smith-Christmas, 2020a), I discuss the conundrum that linguistic competency can pose to conceptualising agency; if for example, it is clear that the child lacks the linguistic competency to answer in the minority language, can their utterance still be considered agentive? I argue that even though the children appear to lack the choice to use the minority language by virtue of them not possessing the requisite linguistic competence, ultimately, their choice is guided by norms and thus, evidences the interdependent nature of agency in child-caregiver interactions. For example, in Gafaranga's (2010) case, in making their utterances in the majority language, the children were drawing on their past experiences (that there were no negative repercussions for use of the majority language), which in turn was contributing to language shift within the community. With 'Chewing Gum', it is a similar situation but with an opposite outcome: here, Daithí is using the majority language with the knowledge that his father will provide him with the Irish language equivalent. Daithí's signalling that he lacks the lexical item 'chewing gum' opens an avenue for his father to supply the lexical item, and thus, again, parallel to Fogle's (2012) work on the agency of Russian adoptee children, the child is creating his own context for language learning. Daithí's repetitions of his father's utterances provide a framework in which he may acquire this lexical item and associated expressions, and he and his father work together over several conversational turns in constructing what Daithí has intended to say from the outset: that it is easy to blow bubbles with his chewing gum. By appealing to his father's linguistic competence, this utterance also reifies the expected generational paradigm, as it places the adult in the expert role and the child in the novice role, much the same way as in the last example, Anita indexed the adult-child power differential through her request for permission.

Conversational excerpt 2: Chewing gum
Daithí's lapses in competence, therefore, appear to contribute to language maintenance, and like the example from the Wieniawski sisters, appears to be a further way in which the child's use of the majority language can reify the pro-minority language FLP. This example also highlights the child's use of language in achieving particular interactional goals: here, Daithí has successfully gained his father's attention, which previously has been directed at his brother due to the nature of the game. This is very similar to an example by Said and Zhu (2019), where the child uses a lapse in competency to garner the father's attention, thus showing the intersection of children's competency and agency in the creative ways they use language to achieve their interactional goals. Therefore, while the caregiver has laid down the structures (i.e. frequently invoking the repetition strategy, thus creating the norm that if a child does not know a particular lexical item or grammatical construction, it will be supplied), it is the child's active participation and manipulation of these structures to suit certain interactional purposes (i.e. vying for the caregiver's attention) that bring this into fruition in everyday interactions. Thus, reifying the pro-Irish FLP is the remit of both caregiver and child.

CONCLUSION
The purpose of this article has been to explore child agency from an FLP perspective and in doing so, build on the shared interest between childhood studies and FLP in the relational nature of child agency in child-caregiver interactions. In doing so, it has centred on what at first glance seems to be an anomaly: how the child's use of the majority language plays a role in reifying the pro-minority language FLP. It has focused on this conundrum as a way to illustrate the interrelated nature of compliance regimes, linguistic norms, linguistic competency and generational positioning, as proposed in the model I developed for examining child agency in FLP (Smith-Christmas, 2020a). We have also seen how the children's agentive use of the majority language in contributing to the pro-minority language FLP may relate to other goals, such as vying for caregivers' attention or creatively rendering a particular utterance. The article has also shown the multidimensional nature of the way in which other norms-such as language choice in siblings' interactions with each other-play in how families deal with evolving realities.
In undertaking this exploration, I have centred on the child's majority language use because it is often de facto seen as agentive, as agency in FLP tends to be read through a resistance lens, that is, going against the grain of the pro-minority language FLP as laid down by the caregivers. However, the analysis shows that in these two instances, the children's use of majority language use actually serves to reify the pro-minority language FLP. Rather than dismiss these examples as non-agentive because they align with the pro-minority language FLP, I have teased out the intricacies which illustrate how these particular instance of language choice are the culmination of different practices over time and space, and how they manifest in the child's agency both being shaped by and shaping their interactional milieu, and especially the generational dimensions of these specific instances of language choice, for example, language use with caregivers versus language use with siblings. There naturally are limitations to what the two examples discussed on this paper can elucidate in terms of the broad and complex topic of 'child agency', but hopefully the focus on the dynamic and negotiated nature of language practices in the family has provided a concrete way to illustrate how child agency arises relationally (Esser et al., 2016) and through the evolution of certain practices over time and space (Bollig & Kelle, 2016). I feel that more research situated in concrete examples of the relational nature of child agency will contribute to a deeper understanding of this complex subject and that these discussions can contribute to fostering a fruitful relationship between childhood studies and FLP.

TRANSCRIPTION CONVENTIONS
Times New Roman italics is used for Polish and Irish language speech. English speech is represented by Times New Roman bold. Courier New is used for translations of the Polish and Irish language speech.