Family language policy in refugee-background communities : Towards a model of language management and practices

As interest in the field of family language policy is burgeoning, an invitation has been issued to include more diverse families and language constellations. This article responds by presenting family language management data from Ethiopian and Colombian refugee families living in New Zealand. As part of the researcher’s ethnographic involvement in both communities, data was obtained through participant observations, interviews with parents and children, and recordings of naturally-occurring interactions between family members. Findings from both communities differ greatly: While many Ethiopian families used explicit management for their children to speak Amharic in the home, Colombian families tended to prefer laissez-faire policies as they did not direct their children’s language choice. Nevertheless, their children typically spoke Spanish, their heritage language. As a theoretical contribution, a model is developed to coherently present the caregivers' choice of language management and their children’s typical language practices. This model helps to uncover similarities and dissimilarities across families and communities. Since families typically moved through different management and practice constellations over time, the model also assists in identifying recurrent family language policy trajectories. The article concludes by drawing practical attention to the need and best timing for informing recent refugees about options and resources concerning intergenerational language transmission.


Introduction
The study of family language policy (FLP) is receiving burgeoning interest as globalisation is introducing superdiverse constellations of multilingual speakers and families.Managing home language use involves decision-making about the ways in which language contact situations are reflected and expressed in family communication.It is the task of family language policy research to uncover, describe and analyse the different factors involved in home language choices.
Family language policy in refugee-background communities 41 King (2016) outlines the historical context in which FLP studies developed.She names classic diary studies in which linguists typically documented the bilingual development of their own children (e.g.Ronjat, 1913) and studies addressing psycholinguistic issues (e.g.De Houwer, 1999), partly from a more sociolinguistic approach (Lanza, 2004) as the types of studies pertaining to the first and second phase respectively of FLP research.In a third phase, she outlines that similar issues continued to be addressed with an emphasis on collecting data that was to be analysed qualitatively, such as through parental interviews and recordings of naturally occurring home interactions.During this phase, a detailed definition arose of FLP, labelling it as "explicit (Shohamy, 2006) and overt (Schiffman, 1996) planning in relation to language use within the home among family members (King, Fogle, & Logan-Terry, 2008)" (King, 2016, p. 2).The challenge for the present phase of research, King (2016, p. 2) argues, is the inclusion of different family types, such as displaced families, and the addition of more diverse language constellations.Including dispersed families in the mix of participating families appears to be a reasonable call in the face of the current refugee crisis.At the end of 2014, just before the current state of emergency, 19.5 million people already lived in settings outside their home country primarily because of war or other humanitarian tragedies in their own countries (UNHCR, 2015).Governments tend to focus on the refugees' quick integration in society and the labour market (e.g.Internal Affairs, 2014), and language learning features as one of the primary tasks in this process.FLP studies in these displaced communities are particularly interesting, not least because, given the often large number of refugees from the same source country, they offer the potential to investigate the individual family in the context of their larger ethnic group.Further, while these families' situations may be comparable to those of socio-economicallyadvantaged migrants in some aspects, with families often facing similar home language decisions, their situations tend to be more delicate and vulnerable in other aspects.For example, rather than leaving based on economic consideration or a desire for adventure, existential reasons such as war and persecution are by definition the cause of their forced migration (see UNHCR, 1967).Moreover, though this may be individually different, they embark on life in a new country with less access to resources and other socio-economic means, no immediate access to employment and often a lack of proficiency in the language used in their country of settlement.Thus, their starting position for establishing a new home may be more demanding than for other migrants.
While there are only few accounts of refugees' language transmission efforts, Hatoss ' (2014) research provides an exception, highlighting added challenges for the Sudanese families in Australia in her research: Most families implemented and enforced home language rules to teach their children Dinka, but encountered challenges in the process, for example when the children had been born in a transit country and had lost one of their parents in the war (2014, p.183).Potentially equipped with little cultural capital for their current context, the refugees' position in society is also likely to impact at least initially on their home language strategies, as they may not be able to readily access information about maintaining a minority language nor other material available to assist them with home language transmission.Yet, research findings have underscored that it is exactly at the family level that the most important language maintenance efforts tend to take place (Fishman, 1991), and that intergenerational language transmission requires effort and dedication (Okita, 2002).Therefore, the implementation of FLP merits continued scholarly attention.

Literature Review
FLP studies have evolved over the last decades, covering a wide range of topics that are analysed with an extensive array of methodologies.The overall findings seem to suggest that, just as human actors are complex and may perform contradictory actions, also language management and language practices tend to be multifaceted, fluid and shaped by external circumstances and internal motivations.After providing a brief introduction to Spolsky's model of language policy, the following sections outline findings concerning families' language management and ensuing language practices.
According to Spolsky's (2004) model, which has been widely used for characterisations of FLP (see Schwartz, 2008;Curdt-Christiansen, 2009;Kopeliovich, 2010), language policy can most fruitfully be divided into the three components of language beliefs, language practices and language management.Language beliefs include the beliefs and ideologies people hold about the language(s) concerned, language practices refer to people's actual and observable language behaviour, and language management stands for any attempts to modify the existing language practices.Applied at the family level, this model offers a valuable analytical abstraction for categorising and summarising the complex relationships and practices that emerge when describing home language maintenance efforts.
The three components of the model may not necessarily correspond to each other: For example, given that beliefs about the minority language are often positive in families from a minority background, it seems surprising that in some cases language transmission is not or only partly successful.One reason to account for this is that there tends to be a gap between the parents' stated goal and their actual practices (see e.g.Yu, 2010).Although many parents want their children to speak the heritage language, they signal acceptance when their children use the majority language and sometimes do so themselves.
An important linking concept to account for this gap between beliefs and practices was introduced by De Houwer (1999) who utilised the notion of 'impact belief' as a crucial intervening factor for whether parents socialised their children into using the minority language.'Impact belief' refers to the parents' conviction of being able to "exercise some sort of control over their children 's linguistic functioning" (1999, p.83).Without doubt parents may positively affect children's language development without any deliberate intent, as discussed below.Nevertheless, parents who deliberately plan their interaction strategies may be more successful in contributing to their children's acquisition of the heritage language, particularly in minority language context where the heritage language typically receives less support (De Houwer, 1999;Ghimenton, 2013;Pérez Báez, 2013).
Parents with a strong impact belief tend to be attentive to their language use and even use metalinguistic means to convince children to speak the desired language.Chumak-Horbatsch argues that a parental impact belief "is accompanied by strategies such as home language rules and praising/punishing children's language behaviour" (2008, p. 5).While a mother in one study who threatened to withhold food from her son if he did not speak the language she wanted (Fredman 1995in De Houwer, 1999, p. 89) furnishes a rather drastic example of deliberate language management, usually the means parents employ are less radical and may range from consciously modelling the preferred language to explicitly telling children to speak a certain language.For instance, Walker (2011) describes the case of a Peruvian mother married to a monolingual English-speaker who only spoke to her son in Spanish and expected him to reply in Spanish.This extended even to situations where other monolingual English-speakers were present and where she reportedly continued to address him in Spanish and afterwards repeated her statement in English for overhearers (Bell, 2001).Altogether, her language practices and management instantiated her conviction that her son should learn Spanish.
In contrast, if immigrant parents do not have an impact belief, they might leave language choice completely to their children (Lanza, 2007, p. 52). De Houwer (1999) argues that majority language use in the home is potentially due to the lack of an impact belief.For example, a group of Zapotec immigrant parents to Los Angeles believed that they could not influence their children's language behaviour (Pérez Báez, 2013).As a consequence, they stopped using Zapotec with their children and language shift occurred within only few years.Likewise, Kulick's (1992) influential study showed that parents from a village in Papua New Guinea were surprised at their children's shift to Tok Pisin, seemingly completely disregarding the fact that they hardly ever used their own language with their children.
The research suggests that an impact belief, in relationship with other factors, such as the families' sociolinguistic ecology and the quantity and quality of their language input, seems to be the intervening variable between positive minority language beliefs and home language management and practices conducive to bilingual development.Parents who consider it their responsibility to influence their children's language use appear more likely to use management and practices that encourage their children to use the minority language.
It will serve well to highlight three types of language management that have arisen in the discussions of FLP data from several countries: explicit management, implicit management and laissez-faire policies.Spolsky (2009, p. 25) describes explicit management as verbal interventions demanding the use of a particular language.Some research has suggested that most successful transmission usually occurs where there is explicit management and the caregivers have deliberated on strategies for teaching their children their minority language (Kasuya, 1998;King, Fogle & Logan-Terry, 2008).For example, Kasuya (1998) explored the language use of families who lived in the US and wanted to raise their children to speak Japanese and English.Based on Lanza's (2004) parental discourse strategies, she investigated parental responses to their children's use of the 'unwanted' language.One of her conclusions was that those parents who were most explicit in telling their children which language they wanted them to speak were typically most successful at achieving the desired result.These explicit management strategies are likely to be accompanied by implicit management, i.e. child-directed activities supporting this management.
Implicit management does not entail verbalised instructions, but instead seeks to engage in activities that tend to be conducive to language maintenance, such as book reading in the minority language.Evidence for the effectiveness of implicit language management in the absence of explicit management was provided by Kopeliovich (2009) in her study in Israel, who showed that a father's attempt to introduce Russian to his children by connecting the language to interesting activities seemed to be more welcomed by the children than the mother's verbalised requests to the children to speak Russian.The father preferred to spark the children's enthusiasm for Russian culture with planned literacy activities, which he conducted individually with each child, reading poems and books in both Hebrew and Russian and using both languages equally for vivid discussions of these works.Although he explained that he was not primarily interested in transmitting Russian to his children, the various activities he conducted with his children ultimately contributed to their increased Russian competence.Language transmission in this case seems to be a by-product of another superordinate goal.
Similarly, yet perhaps a result of more deliberate planning, children in Hispanic families in Canada tended to use more Spanish if the parents introduced the language in the home through book reading, storytelling, quality interactions and the provision of learning materials for the children (Tamis-LeMonda & Rodriguez, 2008).
Different to the explicit types of management and the implicit types that nevertheless deliberately promote the use of a particular language are what Curdt-Christiansen (2013) refers to as 'laissez-faire policies'.The term was used by Curdt-Christiansen (2013) in her work with Singaporean Chinese families.She ascribed the term laissez faire attitude to mothers in her research who did not interfere with their children's language choice while providing routine homework support, with the result that their children predominantly spoke English, a potential hegemonic language in Singapore.Generalising this to a certain extent and linking it to the wider context, Caldas (2012) suggests that most families lack conscious language management because it is the families' embeddedness in "history and circumstances" (2012, p.351) that predetermines language choice.That this unreflected language use by caregivers may not lead the children to actively take up their minority language is supported by Curdt-Christiansen's (2016, p.11) observation, supporting arguments brought forth by Ó hIdearnáin (2013), that in her data "habitual linguistic practice […] failed to build a 'language reproduction' line".These studies show that not all language practices are shaped by explicit rules and may instead arise out of an "unmanaged" situation (Spolsky, 2004, p. 8).Overall, descriptions of the ways in which families adopt these different management types produce diverse pictures of family language policies which "lie along a continuum ranging from the highly planned and orchestrated, to the invisible, laissez-faire practices of most families" (Caldas 2012, p. 352).
It is crucial to note in this process that language management may involve various actors.Fogle and King (2013), two authors who have repeatedly framed the field of FLP, draw attention to the fact that, on the one hand, caregivers may overtly state a policy and be credited with decisions about home language use.On the other hand, however, this explicit policy may be implicitly negotiated in the background and undermined through the practices of individual actors (Hornberger & Johnson, 2007).Children in particular have been shown to take an active part in negotiating initial decisions, even revoking the language management that their parents implemented (X, 2016).A long-term ethnographic account of a family on the Isle of Skye highlighting this aspect is furnished by Smith-Christmas (2016).Based on Gafaranga's (2010) concept of "Talking language shift into being" as well as her detailed observations, she explored the interactional dynamics in this family, who was motivated to implement a Gaeliccentred FLP.Contrary to the caregivers' wishes, however, the children in the family instituted a dual-lingual mode, in which the adults initially spoke Gaelic, but let the children reply in English, and occasionally switched to English themselves.Wong-Fillmore (1991, p. 338) reports that immigrant children in her US data even switched to English in cases where their parents had difficulty to understand them if they did not speak the minority language.Overall, then, language management may be susceptible to change in response to the children's agency.
The literature review has highlighted different components of FLP based on Spolsky's (2004) model of language policy and has shown that the relationship between language beliefs, language management and language practices, which involve several different actors within the family, is not always linear; instead, explicit parental language management and laissezfaire policies ensue a number of different child language practices, and an impact belief is suggested to mediate between the individual components.
The nature of language management (explicit vs. laissez-faire) and its effects on the children's language practices has been a recurrent theme in the study of FLP, and, responding to King's (2016) call, this study seeks to expand on it by investigating language dynamics in displaced families, focusing on Amharic-speaking Ethiopian and Spanish-speaking Colombian refugees in New Zealand.Through a three-year long ethnographic approach, the present study draws on the advantages of more detailed insight into the interactional dynamics of these families.
On a theoretical level, what becomes clear in King's description of the historical context and current positioning of FLP is that there is a diversity of approaches with which the field is being explored and which, according to King, may put the integrity of the field of FLP in jeopardy.Modelling family language policy, as this paper seeks to do, has the potential to provide coherence to findings gained in a variety of contexts using different approaches, render visible some of the underlying patterns and provide explanations that go beyond individual accounts.

Background
New Zealand currently has an annual quota of 750 refugees that enter the country in six intakes a year.Governmental panel teams interview individuals and families in selected third countries where these are registered with the United Nations High Commissioner for Refugees (UNHCR).If these families fit the categories within the country's immigration scheme, they are then taken to the Mangere Refugee Reception Centre in Auckland, and attend an English language training and cultural introductions for six weeks together with refugees of other origins.Subsequently, they are resettled to other cities in New Zealand with the attempt to house the ethnic communities closely together to offer them the chance to assist one another in their settlement process.
Decisions as to the particular source countries of refugees are revised and adjusted at regular intervals.As a result of this geographical narrowing of countries of origin, members of the same ethnic communities enter the country at the same period, enabling them to build community in New Zealand and offer support for their fellow country people.The fact that some of New Zealand's refugees form a more or less homogenous group in terms of origin will be relevant for the proposed model.
The Ethiopian and Colombian participants in this study were drawn from two of these refugeebackground communities.While Ethiopians were first taken into New Zealand in the 1990s, there have been follow-up intakes through which family members have been able to join the former wave of refugees.Thus, the community has had the opportunity to become established and to structure themselves, mostly in terms of strong networks between members of either Protestant or Ethiopian Orthodox Churches.By contrast, the Colombian families were brought to New Zealand only from 2007 onwards and have therefore had a relatively shorter period of time to build community.At the time of the census in 2013, 1245 Ethiopians and 654 Colombians self-identified as such on the census.
I conducted the research in Wellington, which is the capital of New Zealand with a population of about 240,000.At the time of research, 165 Colombians and 237 Ethiopians reported to live in the greater Wellington area.While the Ethiopian community showed great cultural diversity, including Oromians, Tigrayans and Amharas, my focus was on those that spoke Amharic as a mother tongue.In the absence of official statistics, community leaders agreed that the Amharicspeaking community comprised approximately 100 members.Thus, both the Ethiopian and Colombian communities are small as compared to the total population; yet the Colombian community in particular is growing steadily.

Methodology
My methodology is situated within the approach of linguistic ethnography (Rampton, 2007).I conducted participant observation between 2012 and 2015 as I attended refugee events, Ethiopian and Colombian church groups and services, and was invited to join a soccer team of female Colombian refugees.I also visited a few families to observe their home interactions and recorded one hour of a mother's interactions with her children during one of these visits.These events provided for regular, at least weekly, contact with the Colombian community during the course of the research and weekly contact with Ethiopian participants for the course of a few months, followed by less sporadic visits and encounters.Through the constant interactions particularly with the Colombian community I was able to negotiate my role on an insideroutsider continuum to be a friend to the community.
The data from the observations provided valuable insight for planning the semi-structured indepth interviews.These were conducted with 15 caregivers from the Colombian community and 14 caregivers from the Ethiopian community, and with eight and nine of their children respectively.Primary caregivers were mostly mothers, but also uncles, aunts and grandmothers assumed this role in four families.The total number of participants constituted about half of the research population (caregivers with at least one child under the age of 12) in the Colombian community at the beginning of data collection, and seemed to cover the majority of the relevant Amharic-speaking Ethiopian families, as was suggested to me by a number of community members during data collection.Interviews were conducted in English with Ethiopian participants and in Spanish with Colombian participants, and relevant interview passages were subsequently translated for the write up.
Table 1 provides an overview of the family compositions and educational background of the families.While the majority had studied English at school in Ethiopia, only few participants reported speaking more than a few words of English upon arrival.Information gained from interviews paints an initial picture and provides insight into language ideologies (see Blom & Gumperz, 2007;Ghimenton, 2013), but the interview setting nevertheless functions as 'social practice' (Talmy, 2011) where the data is co-constructed together with the researcher.These reported practices therefore need to be considered with caution.To further complement reported data, recordings of naturally-occurring speech between children and parents may open the backstage where language choice is negotiated in family interactions (Genesee, Nicoladis & Paradis, 1995;Lanza, 2004).Hence, three mothers from the Colombian community were asked to audio record their naturally-occurring home interactions with their children.These recordings, yielding another two hours of engaged parent-child interactions, were described and relevant parts were transcribed orthographically and, where necessary, translated.They provided more dynamic insight into language management and practices in the families.
While long-term ethnographic work and interactional data underlie this article, I here focus predominantly on larger patterns of language management and practices across the two communities in order to provide a broader basis for developing a model of family language management and practices.I describe patterns of Ethiopian families using explicit minority-language management and Colombian families using laissez-faire policies and management in favour of English.

Family language policy in the Ethiopian community
The relative majority of families in the Ethiopian community used Amharic-only management and their children also typically used Amharic in the home.In the interviews, caregivers typically conceptualised their language use as active teaching and modelling.One mother of two young children, a girl and a boy who had recently started attending kindergarten, describes her efforts in the following way: We speak Amharic, but the boy starts to switch now.But I say 'no, speak in Amharic'.Some of the language that he can't understand I explain it in English and then I go back to Amharic again 'this is what it means in Amharic'.So when he asks me in English, I say next time you want to say this, say it in Amharic, like this.So I teach him, I make him repeat it and next time he says the Amharic word.
This mother seems to have an impact belief (De Houwer, 1999) because she sees herself as a model of correct language choice for her son.More concretely, she outlines her opposition to her son's use of English as she redirects him to use Amharic.She reports that she uses Amharic with him most of the time, and explains concepts in English whenever he does not follow her, just to return to Amharic again and model language use.This series of exhorting, teaching, modelling and having him repeat constitutes an important part of her Amharic language management.The following sections outline five characteristics of families that used explicit Amharic management with their children following suit.
First of all, some families who used explicit Amharic-only management were characterised by having low English proficiency.In this sense, it seemed to be a pragmatic choice to tell their children to speak only Amharic because this was the only language for communication, and therefore essential to secure ongoing rapport between family members.Nevertheless, low parental English proficiency did not mean that parents necessarily implemented Amharic language management.Other parents with only little knowledge of English used no language management, or asked their children to speak English so that they could practice the language at home with their children.Neither was low parental English proficiency the only factor that encouraged the use of Amharic-only management, as is shown by the use of this type of management by parents who could easily converse in English.
Second, Amharic-only management seemed to draw strength from parents' beliefs about ethnic identity, language and religion.All caregivers in this scenario strongly voiced their pride in being Ethiopian and having the background of a rich and strong culture.Their culture seemed intricately intertwined with religion, with the Ethiopian Orthodox Church instantiating the relationship between the Ethiopian Orthodox religion and Amharic, the language which is used alongside the traditional church language Ge'ez in church services in the Wellington community.In this sense, Amharic use in the church and for religious purposes more generally exceeded instrumental value, and represented the rich connection between Ethiopian history and Christianity.Caregivers repeatedly emphasised the strong link between beliefs about ethnic identity, language and religion (see X, 2017).
Third, caregivers who had implemented explicit language management shared an understanding of the following two points of concern.First, all agreed that their children had sufficient access to English so that there was no lack of input for them to learn the societal language.This was certainly due to the fact that the Ethiopian community was comparatively small, and that the older children's classmates, despite being from several different countries, had established English as their lingua franca.Thus, English was within easy reach for all children.As a second point, parents exhibited an impact belief and shared the recognition that the only place for their children to learn Amharic was the home.One caregiver explicitly emphasised that 'English they will learn it anytime, but once they forget the culture and the language it's gonna be very hard for us to teach them again.'Amharic was only marginally represented in public life in Wellington so that participants frequently encountered other members of society who were not familiar with the language.This general unawareness of the language by others was accompanied by a lack of institutional support for Amharic which could have taken the form of official Amharic classes or play groups in the language.Families who used Amharic management highlighted these two points most clearly.
Fourth, determined implementation of Amharic-only management typically required discussion among caregivers.At times, this discussion took place before the child was born, and caregivers had a vision for their language management strategies from the beginning.At other times, however, language management decisions were made only at a later stage when the caregivers were no longer satisfied with the status quo.The families' FLP thus underwent a dynamic evolution in response to linguistic and non-linguistic factors, as will be discussed with the help of the model below.
Fifth, the decision to use Amharic-only management in the home entailed other choices.One was to "modify their children's language environment" (Spolsky 2012, p. 7) by shielding the children from English-language influences.A few mothers deliberately stayed home themselves to take care of their children instead of sending them to day care where they would have been exposed to English at an early age.They desired to equip the child with a solid foundation in Amharic.One family asserted that their three-year-old son interacted with children at the playground in English and sometimes watched English TV shows, but that they tried to ensure that the majority of his interactions took place in Amharic.Often this necessitated a strict separation between private and public domains for language use.For instance, in another family, any interview question about language use to express different feelings (happiness or anger) were reciprocated with the question: Do you mean at home or outside?'Home' in this case characterised the domain of Amharic language use, whereas the public domain was exclusively English.
In summary, Amharic-only management seemed to be a popular option among the families, and it drew strength from the perceived close association between Amharic and Ethiopian culture and religion as well as from the declared lack of opportunities to learn the language in society.It also tended to follow discussion about language choice between the caregivers and ensue deliberate attempts to modify the children's linguistic environment.

Family language policy in the Colombian community
This section provides details about laissez-faire policies and English management used in the Colombian community.The majority of Colombian participants featured a laissez-faire policy.This was exemplified by the following comment from one mother, whose son grew up speaking Spanish, but had just begun speaking more English in the home as a result of his exposure to the language in day care:

No hay regla, tra-intentamos hacerlo para ayudarme a mí con el inglés, pero no funcionó [laughs] [...] entonces ya, dejamos que fluya así.
There's no rule, we tried to do it in a way that he would help me with my English, but it didn't work [laughs] [...] so yea, we just let it flow.
After mentioning her futile attempts to speak English with her son, this mother seems to indicate resignation and a decision to refrain from further management attempts.
Interestingly, when discussing their motivations for leaving their children to choose which language they wanted to use, the caregivers presented two different strands of arguments.Half of the families using a laissez-faire policy had not paid any particular attention to language use.These caregivers typically shared a low education level, low socioeconomic background and low proficiency in English.Like all other Colombian families that were interviewed, these mothers were keen for their children to learn English and saw it as a great achievement if they did so.However, the default language of the home was Spanish for practical reasons, and the understanding that their children would never forget Spanish was widespread.The other half of the caregivers who used a laissez-faire policy showed some degree of recognition that home language choice may be important for their children's development of Spanish language proficiency.While they had not yet introduced explicit Spanish-management, they considered doing so in the future.
It is thought-provoking to note, however, that caregivers typically focused on their older children in their deliberations about language use.While the older children in these families were the ones that I interviewed and that continued to speak Spanish, their younger siblings seemed to understand Spanish, yet be more inclined to use English.For example, in one of my home observations, one single mother of six children asked her three-year-old daughter pásame estos zapatos ('pass me these shoes'), and her daughter followed the request, simultaneously saying shoes.Another mother, Cristina (C), lamented that she found her five-year-old son José's (J) Spanish very difficult to understand, which was exemplified during our interview when he entered the living room to ask for some paper (M = interviewer; translated passage on the right side): Following this episode, we continued our conversation in Spanish, and José and his three-yearold brother played in the background using English.While Cristina had explained that her son encountered difficulties with pronouncing words in Spanish after spending two years in New Zealand without any Spanish-speaking friends, some other caregivers stressed that their children were unlikely to forget Spanish even if they promoted English learning at home, yet often conceded at the end of our conversation that their youngest children could potentially fail to learn Spanish.
That younger children needed to be included in language management strategies became visible in the reports of some families describing how their youngest children typically communicated with each other in English.This was corroborated by my observations, which showed that when playing games some of the younger children would either use English or a mixture of English and Spanish.The caregivers responded to this language choice with reserved pragmatism: Whereas the use of Spanglish in the US, though widespread, is not necessarily encouraged or accepted by all, participants in my study typically had no reservations about mixing the two languages, nor did they generally show dissatisfaction with their children's use of English with each other.
Despite not having language rules, all caregivers who held a laissez-faire policy continued to speak Spanish with their children most of the time.Still, instances of translanguaging were also common in some caregiver-child interactions, as suggested by the following conversation between a mother (M) using a laissez-faire policy and her four-year-old son David (D), recorded just before his bed time: Although David shows good command of Spanish in other family interactions and was, apart from peripheral influences, first fully exposed to English only a year prior to the recording, his language choice in this passage is undoubtedly influenced by his strong exposure to the language in day care.Both mother and child here take turns initiating language changes, with David using English most of the time and his mother acknowledging his language choice, carrying on the conversation partly in Spanish (corresponding to Lanza's 2004 move-on strategy and repetition strategy), partly in English (Lanza's code-switching strategy).Overall it can be said, however, that families whose younger children typically used English increasingly felt that their children no longer understood Spanish and they therefore had fewer expectations of them to use Spanish.
Based on a number of observations and discussions with community members, it appears that the wish to learn English to integrate into the majority culture was an important factor influencing Colombian families' FLP trajectories.Especially in the first few months after arrival, the families experienced great cultural differences and conceptualised fast English learning as a means to participate in social and economic life in New Zealand.When their children attended day care, many were tempted to use their children as conversational partners to practice their English skills.However, data from this research shows that these attempts always failed, and even in cases where children used English with each other, they refused to speak English with their mothers upon her request.One mother refers to this saying: A veces he tratado que hablemos inglés en casa, pero "ay no, no me hable, que no le entiendo".
Sometimes I've tried for all of us to speak English in the house, but "oh no, don't speak to me, I don't understand you".
Conceptualising to English as the linguistic capital of New Zealand (Bourdieu, 1977) and use of the language as essential to access various types of social, educational and economic opportunities, this mother further engaged with creative language socialisation attempts and used media, games and general persuasion to provide her children with opportunities to speak English even in the home.For example, she reported transforming her children's enjoyment for watching movies into a language learning opportunity (Lin & Siyanova, 2014): Family language policy in refugee-background communities 53 [...] pues, ver una película y una serie animada es una diversión para ellos, como entretenimiento, porque nosotros no salimos a ningún lado [...] entonces lo único que puedo hacer es -si es una película, la vamos a poner en inglés para que se involucren bastante con el idioma.
[...] well, watching movies and cartoons is fun for them, like entertainment, because we don't go anywhere.[...] so the only thing I can do is that when we watch a movie, I can put it on in English so that they can get involved with the language.
These comments convey the intentions of this mother to introduce English as a home language albeit for limited functions.Given that her family had been resettled to a more peripheral suburb and did not own a car, which in her opinion impeded them from going to more places where English was spoken, she refers to her attempts to bring the societal language into the home.Still, her efforts were countered with resistance, and both children and mother confirmed that they habitually spoke Spanish.
This section has provided insight into the motivations for using laissez-faire policies and English management, and given examples of how older and younger children responded to these two types of management.The following section presents a model that fruitfully subsumes these different factors.

A model of family language policy
This study has drawn on Spolsky's categorisation of language policy into language beliefs, language management and language practices.The re-occurrence of certain types of management and practices across communities suggests that it may be beneficial to provide an overall framework as a point of reference for further descriptions of family language management and practices in other contexts.
When suggesting a model for reflecting a reality that is complex and filled with idiosyncratic decisions, we must heed the famous 20th-century statistician George E.P. Box' warning; in his words "[a]ll models are wrong, but some are useful" (Box, 1976, p.197).What may be called a disadvantage is the simplification inherent to models, which let complex situations involving a variety of idiosyncratic behaviours appear orderly and simplified.In this case, a model of language management and practices in which families' linguistic behaviour corresponds to scenarios will never be able to appropriately reflect the multifaceted decisions and behaviours of transnational families in all their detail.Depending on the ways in which the model is used, with the responsibilities lying with each individual researcher, this may have its dangers in that family language policy may be interpreted as static, as when a family is assigned to a certain scenario not only for a point in time but for their entire migration process, or when language choice patterns are in danger of being interpreted as fixed and determined rather than as tendencies and constructions under the influence of wider societal forces.
On the other hand, a model for language management-practice combinations has the potential to become a valuable asset for scholarly explorations of family language policy.In the last ten years, there has been a plethora of studies about family language policy, many of them stressing a number of different aspects; framing the diverse studies with a look towards understanding greater patterns highlights parallels and connections between the individual case studies and offers the basis for comparison.Further, highlighting how specific management-practice scenarios are adopted by the families researched is likely to provide a degree of abstraction, allowing for deeper insight through the recognition of underlying similarities and shared individual traits and behaviours across families which, amidst all idiosyncrasies, may provide illuminations as to the reasons and explanations for the families' adoptions of specific language management styles and their children's typical language choices.Box renders this in the following words: "For such a model there is no need to ask the question "Is the model true?".
If "truth" is to be the "whole truth" the answer must be "No".The only question of interest is "Is the model illuminating and useful?" (Box, 1979).In this spirit, the proposed model is hoped to contribute to the field of family language policy by offering to unite the insights gained in different (case) studies and provide greater illumination and sources of explanation.
The following key considerations were taken into account for developing the model: First, while language beliefs can shed light on a family's motivations for home language maintenance, it is their language practices and management that constitute the observable components of their FLP.They should therefore be at the centre of the model.Second, one may wonder whether language management should refer to parental efforts or include the children's attempts at managing home language practices.While children have been shown to play an important and active part in the FLP process, focusing on the caregivers' management seemed most appropriate for the purposes of the model, and corresponds to the traditional understanding of language management being the responsibility of the caregiver.Caregivers may reflect on their practices in metalinguistic ways by discussing their language management strategies before the birth of the child and also repeatedly during the parenting process.Third, language practices, conversely, most appropriately referred to the children as the future bearers of language maintenance.Making room for fluid language choices as positioning devices, practices here refer to the child's typical language use.
Parental management strategies and the children's language practices are grouped together into six different scenarios (see Figure 1).Language management is situated on an axis ranging from explicit management to use the heritage language, via a laissez-faire policy to explicit management to use the majority language.Relating to practices, the two categories are the children's typical use of either the heritage language or the majority language.

Figure 1. Model for categorising management-practice scenarios in family language policy
The model consequently divides management and practices into six scenarios.For example, Scenario D applies to families using a laissez-faire policy in which the children typically speak the majority language.If one was looking for what may be called by some a success scenario where caregivers have implemented management in favour of the heritage language and the child typically follows this management, Scenario A would provide a more detailed overview.
Since the model was meant to describe management and practice scenarios of families within ethnic communities, it needed to provide a way to reflect the number of times a scenario was chosen within a community.Thus, it needed to incorporate quantitative information.This was achieved by using luminosity of colours, with darker colours for a scenario representing a higher incidence of this scenario in the community1 .
Figure 2 gives an overview of the distribution of management and practice scenarios across families in the Ethiopian community.Scenario A (explicit management to use Spanish, child typically uses Amharic) applies to the relative majority of the families in the community and therefore has the darkest degree of luminosity.Only one family has Amharic management with their child not following suit (B).Laissez-faire policies are also used by some families, as is indicated by the luminosity of Scenarios C and D. Children in these cases typically use English (D).One family has implemented rules to speak English, yet their child has continued to speak Amharic most of the time (E).There is no family that has successfully asked their children to speak English at home (F). .This seemed to be a consequence of their recent migration and the perception that Spanish was still present in many areas of their family and community interactions.In most of these families, the children typically still used Spanish (Scenario C).In a few families, the children had already begun to make English the default language of the home (Scenario D).
Only a few families used explicit Spanish-only management, and for those who did so, the children all typically spoke Spanish (Scenario A).One mother exhorted her children to speak English, but these typically used Spanish (Scenario E).In a few cases, a family initially had no management and their children typically spoke Amharic (Scenario C).Usually triggered by the child's entry into the institutional system of the country, this monolingual situation was challenged and English increasingly became part of family interactions (Scenario C).In response to their children's changing language use and proficiency (increased knowledge of English and potentially less familiarity with Amharic), families instituted Amharic-only management, with their children following suit (Scenario A).It was advantageous in these situations if the caregivers had provided their children with a solid foundation in Amharic during their early years, so that the children could fall back onto the knowledge they had acquired when most of their input still occurred in Amharic.This trajectory is represented by the red arrows in Figure 4.In other families, however, the child's lack of familiarity with Amharic meant that Amharic-only management resulted in no success as the child continued to speak English (Scenario B), and the parents eventually settled for a laissezfaire policy (Scenario D).This is represented by the dotted lines in Figure 4.While the situation in the families may certainly be more complex, reflecting the diverse linguistic and nonlinguistic influences, these trajectories highlight important intermediary steps in the development of FLP over time.Provided that the goal is the maintenance of the minority language, the occurrence of these trajectories would highlight the importance of promoting the minority language from an early stage in order to secure a solid foundation for the child.
While the Colombian community is still at the beginning of their migratory experience, a shift in language management can already be noted for a number of mothers, who initially tried to direct their children to use English in the home (Scenario F).Given that these attempts failed, i.e. their children continued to speak Spanish, the caregivers shifted to using a laissez-faire policy, and their children typically used Spanish (Scenario C).Based on data from the younger children discussed above, it is conceivable that their FLP would move towards Scenario D as a next step, but further research is needed in this area.
Overall, the model seeks to provide a coherent way of relating language management to language practices, with the potential to furnish additional quantitative information and to trace families' language policy trajectories over time.

Conclusion
This study has responded to King's (2016) call by presenting home language maintenance data from two refugee communities in New Zealand and provided insight into family language management decisions and language practices in several families.
First, findings about FLP differed significantly between families in the Ethiopian and Colombian community.The majority of families from the Ethiopian community had an impact belief (De Houwer, 1999) and used explicit home language management.Caregivers recognised that their children would have enough exposure to English outside the home.They also realised that the home was the only place where they would be able to provide an environment for their children to learn Amharic.Caregivers often discussed home language choice and modified their children's language environment (Spolsky, 2012) to strengthen their Amharic proficiency.These results resemble Hatoss' (2013) results from the Sudanese Dinkaspeaking community in Australia.
Regarding the relationship between parental language management and children's language practices, children in Ethiopian families with a minority language-only policy were more likely to use their minority language in the home.Conversely, children in families with a laissez-faire policy were more likely to typically use English.This result corresponds to previous findings by Kasuya (1998) and King, Fogle and Logan-Terry (2008) who suggest that explicit management is more likely to lead the children to speak the minority language.
Colombian families, conversely, had a strong tendency to use laissez faire policies (Curdt-Christiansen, 2013) based on a belief that their children would continue to speak Spanish.
Children in these families tended to speak Spanish.While this does not reflect results from the Ethiopian community concerning the link between parental management and children's practices, another reason to account for the children's language choice may be their recent migration and the hitherto dominant socialisation they had received in Spanish.Younger children, however, were more likely to use English with one another.While caregivers typically had no strong reservations about this, some voiced concern about their younger children's Spanish proficiency.Still, a few Colombian caregivers attempted to introduce English into the home, partly to provide practice for their children due to what they perceived to be limited opportunities to be engaged in an English-language environment outside the home, partly to benefit from their children's linguistic expertise and improve their English.
Second, while non-refugee migrant families of a higher socio-economic standing are often familiar with the language of the country they are migrating to and may be able to plan ahead for their children's heritage language education (see e.g.King & Fogle 2006), the refugee families in this study were in a different situation: they were chosen for resettlement to an English-speaking country, yet typically had no prior knowledge of English.Consequently, some caregivers considered their children a source of access to the majority language, which was expressed particularly clearly in the recently arrived Colombian families, who considered implementing strategies that would make English the language for intra-familial communication.
Third, the article has presented a model to illustrate the relationship between parental language management and children's language practice.The axis describing child language use has two points of reference, each focusing on children's typical language use with their parents.The other axis referring to parental language management instantiates three forms of management.Two forms stand for parents' explicit intervention in their children's language use by prescribing the use of a specific language to them -either their minority language or the societal language.The third form represents parents' decision to use no language management at all, thereby adopting a laissez-faire policy (Curdt-Christiansen, 2013) in which the child is left in charge of his/her own language choice.Given the complexity of family language scenarios, the model provides an abstraction of parental language management and their children's language practices.For studies of minority families who are part of a larger ethnic community, the model further offers the benefit of providing a visualisation for the frequency of occurrence of each scenario in a specific community using different luminosity of colour.Applying this model to FLP studies in different communities means that parallels and connections may be uncovered against a common framework.
Fourth, the dynamic trajectory of FLP outlined in this article points to a practical conclusion.It was shown that the child's entry into the institutional system of the society tends to trigger the need for more explicit language management; yet the children can only comply if they have obtained a solid grounding in their heritage language at a young age.Therefore, practical intervention, possibly in the form of information brochures and advice sessions, may be required early on before the child enters day care to secure a solid foundation in the heritage language.In New Zealand, appropriate timing for this could be the refugees' arrival in the Mangere Refugee Reception Centre, where valuable social and linguistic foundations are laid for their subsequent stay in the country.This type of linguistic advice may also be included in the programme of organisations providing support to new parents.
One aspect that asks for further investigation is that the model offers the potential for a visual comparison of reported versus observable behaviour.Especially in the current academic climate exhibiting a growth of ethnographic research, such cross-checks of what families state they do and their actual language practices (see Ghimenton, 2013;Curdt-Christiansen, 2016) may furnish interesting insight into ways in which families construct and conceptualise their own FLPs vis-à-vis their actual practices.
Figure 2. Management-practice scenarios in the Ethiopian community

Figure 3 .D
Figure 3. Management-practice scenarios in the Colombian community

Figure 4 .
Figure 4. Trajectories of management-practice scenarios in the Ethiopian community

Table 1 .
Participants' family compositions and education levels