Conceptualisation of family and language practice in family language policy research on migrants: a systematic review

Family language policy (FLP) is increasingly recognised as a distinct domain of language policy concerned with the family as an arena of language policy formulation and implementation. While FLP is a relatively new research area, its conceptualisation of family and language practice requires re-examination due to social changes and technological developments, including the expansion of digital communication within families and the rise of globally dispersed families a product of global migration and transnationalism. In this systematic review of migrant FLP research, we investigate how the notions of family and language practice are conceptualised in research. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, we identified a total of 163 articles for analysis. Our analysis reveals that the majority of studies were conducted in nuclear families, i.e., those consisting of a father, a mother, and one or more children. Studies also tend to conceptualise the family as fixed and physically located in one place. Paradoxically, around half of the studies acknowledge the presence of geographically dispersed family relations, but this does not necessarily affect their conceptualisation of what comprises a family. Language practice was conceptualised as physical and face-to-face communication in 51% of instances, with only 11% incorporating an analysis of digital communications. Based on our review, we recommend that FLP researchers researching migrant families reconceptualise the family as geographically dispersed and language practice as digital and multimodal when necessary. Such a reconceptualisation will help researchers understand the hitherto underexamined contributions of dispersed family members and multimodal digital communications in migrant FLP.


Introduction
Over the last few decades, the family has emerged as one of the most important domains for language management and practice (Clyne & Kipp, 2011;Pauwels, 2016), making it a critical site of language policy-making and implementation. Many studies have contributed to the vibrancy of family language policy (FLP) research as a distinctive subfield of language policy (Curdt-Christiansen & Gao, 2021;Smith-Christmas, 2022). FLP can be defined as a process in which individual family members, especially parents, try to regulate the use of specific language(s) that may help or hinder the maintenance of their cultural ties or cultural heritage (Caldas, 2012).
Most studies in early language policy research focused on public domains such as state, school, and workplace (King et al., 2008), paying limited attention to private domains such as the family (Robinson et al., 2006;Wan & Gao, 2021;Wiley & Wright, 2004). The relevance of private domains has, however, been recognised in major conceptualisations of language policy, such as the conceptualisation of FLP as a process by Spolsky (2004), where he stated, "language policy in the family may be analyzed as language practice, ideology and management" (p. 43). Spolsky's (2004) conceptualisation of FLP comprises language ideology (i.e., beliefs/views about HL maintenance), language management (efforts to maintain the language) and language practice (the actual use of the language). Early FLP research focused on the "policy" aspect of FLP, for instance, how FLP emerges from the interactions between research on child language acquisition and language policy (King et al., 2008). While relevant research draws on a variety of related fields of inquiry, such as heritage language maintenance (Lee, 2006) and home language maintenance (Shen et al., 2021;Tseng, 2020), the term "family language policy" has been increasingly used to define this body of research in publications (Lanza & Gomes, 2020). Lanza and Gomes (2020) note that "there has been an increase over time of publications with various degrees of intensity over the past decade" (p. 159) and they have recorded an almost nine fold increase in the number of publications using the term "family language policy" from 2008 to 2018.
The salience of FLP in language policy research has been driven by an overall increase in global mobility, which has required many migrant families to make challenging decisions regarding their language practice, especially concerning the maintenance of the heritage language (HL) among their children in their host contexts. In the backdrop of the rising awareness of FLP concepts such as family may need to be reexamined and perhaps reconceptualised because global migration and transnationalism have significantly changed the nature of what one may call a "family". For instance, the number of migrants worldwide has increased, both as an absolute figure and as a fraction of the global population, rising from 192 million (2.93%) in 2005 to 281 million (3.6%) in 2020, with Europe, North America and Oceania seeing some of the most significant increases in immigration (Institute of Migration, 2021). This means more families are becoming globally dispersed and are not physically confined to one location.
In addition, language practice is one of the key components of language policy (Spolsky, 2004). The growth of this globally dispersed population has coincided with a parallel growth in internet use, with the number of users growing from just 413 million in 2000 to 3.4 billion, or almost 60% of the world's population, in 2016 (Roser et al., 2022). Geographically dispersed families increasingly rely on digital communication technologies such as Skype and WhatsApp to interact with their relatives and friends in other countries (Palviainen & Kędra, 2020). Thus, for instance, in 2014, Skype accounted for approximately 40% of the entire telephony market (Worstall, 2014), and in 2011, according to the New York Times, "170 million users each month connected for more than 100 min on average. In the last year or two, video use has surged, now accounting for 40 per cent of Skype's traffic, with 170 million users each month connected for more than 100 min on average" (Lohr, 2011).
The COVID pandemic and associated travel restrictions have only helped to accelerate this trend (Hatoss, 2023). In addition, many senior family members, such as grandparents, have recently begun using digital communication technologies, making it easier for younger migrant family members to maintain contact with each other (Taipale, 2019). Taipale (2019) refers to families that make extensive use of technology to remain in contact across significant distances as "digital families" (p. 14), defined as "one form of distributed extended family, consisting of related individuals living in one or more households who use at least basic information and communication technologies and social media applications to stay connected and maintain a sense of unity despite no more than occasional in-person encounters between them" (Taipale, 2019, p. 14). Families may communicate using different media, exchanging voice, text, video, and images via various services like video calling, text messaging, and email (Taipale, 2019, p. 90). These multimodal communications may help sustain family bonds and maintain the HL (Lexander, 2021). Language practice is thus neither confined to face-to-face settings nor situated in any one physical location but is instead embedded within the broader context of communication involving multimodal digital tools. The use of digital media for communication has also happened within a broader context of increasing electronic literacy (Lee, 2006), the use of various internet based, digital HL learning tools such as weblogs (Lee, 2006) and the use of HL related apps (Little, 2019) among immigrant families.
The question thus arises as to whether the conceptualisation of family and language practice in the existing migrant FLP research appropriately reflects social changes and technological developments. One recent systematic review attempted to examine the concept of the family in the FLP literature on dispersed families, concluding that the concepts of ideology, management, and practice within FLP should be expanded to accommodate the realities of transnational families (Hirsch & Lee, 2018). Nevertheless, several unknowns remain. For instance, it is not known how common it is for families represented in the current migrant FLP literature to have dispersed members contributing to language maintenance in some form. Not much is known about who these family members are (in terms of relationships), how contact is maintained with such dispersed family members, or to what extent digital communication modes are used in the maintenance of that contact, i.e., to what extent they may be considered "digital families" as discussed above. Indeed, little is known about the types of families (e.g., nuclear families) that the broader migrant FLP literature has surveyed to date. If digital communication modes are an increasingly important component of language maintenance in FLP, then it is necessary to understand the extent to which the migrant FLP literature has already engaged with this idea relative to traditional face-to-face modes of language transmission in FLP. Such unknowns point to a literature-wide issue which can be addressed through a systematic review exploring how the notions of family and language practice are conceptualised and operationalised in existing studies and how the migrant FLP literature should respond to shifting understandings of family and language practice. Our review addressed the following research questions: 1. How is the concept of family constructed in migrant FLP studies? 2. How is the concept of language practice constructed in migrant FLP studies?
By 'FLP studies', we refer to the FLP-related publications included in this systematic review.

Methodology
We employed a PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) approach to conduct this study. PRISMA is a standardised protocol for conducting systematic reviews and delivering transparent reporting (Liberati et al., 2009). A systematic search strategy with pre-selected search terms and selection criteria was used based on the PRISMA guidelines, checklist, and flow diagram. Our review identified the studies via five databases: Education Resources and Information Center (ERIC), Linguistics & Language Behaviour Abstracts (LLBA), Psych Info, Scopus, and Web of Science. While the term "family language policy" has been in use at least since the early 2000s (Hollebeke et al., 2020), the tripartite FLP framework of language ideology-managementpractice (discussed in the Introduction section) was proposed by Spolsky in 2004 (Spolsky, 2004). Thus, our search was confined to the period from January 2004 through to December 2022. "famil* language polic*"AND (migrant* OR immigrant* OR transnaƟonal* OR aliens) AND ("language pracƟce*" OR "language management" OR "language belief*" OR language ideolog* OR language planning OR language strateg* OR language aƫtude*)

Procedure
The search terms were determined based on Spolsky's (2004) framework (Figure 1), in addition to our exploration of existing FLP systematic reviews (Hirsch & Lee, 2018;Hollebeke et al., 2020) and discussion between the first and second authors.
We designed these terms to capture the FLP literature concerning transnational immigrant or migrant families. Since this research aims to examine conceptualisations of family and language practice within the FLP literature, our search terms were also designed to extract literature from within the FLP domain. While other scholars have used a broader set of terms (Hirsch & Lee, 2018;Hollebeke et al., 2020), our approach to FLP in this research is limited to studies that explicitly use the term, "family language policy" as FLP has become an established, well defined subfield of inquiry in language policy research (Lanza & Gomes, 2020).
Wildcard (*) characters were used to maximise article capture as multiple variations of some words may exist. Note that for the rest of this article, we use the term "migrant" to refer to both migrants and immigrants for notational ease. The distinguishing factor between the two is the intention of the latter to stay permanently. We consider a family to be a "migrant" if at least one parent of the children being studied is a migrant. Thus, if the father lives abroad and the mother and child do not, or if the entire family has migrated and has extended family members elsewhere, in both cases, the family would be considered a migrant family. This definition allowed us to examine the likelihood of the use of digital modes of communication between the child and other family members. Our search terms were applied to the five databases listed above. The search was designed to return articles which contained the search term anywhere within them. As can be seen in Figure 2, our search yielded 592 results, including duplicates. After automatic and manual deduplication, 535 manuscripts remained. These manuscripts were screened based on several criteria. First, manuscripts in languages other than English were excluded. Next, manuscripts that were not peer-reviewed (e.g., conference papers) were excluded, following protocols followed in other FLP systematic reviews (Hollebeke et al., 2020). After this screening process, 482 peer-reviewed manuscripts remained. All 482 retained articles were further screened in a second stage to determine inclusion. Manuscripts not involving bi/multilingual immigrant families (123) were removed since we aimed to investigate conceptualisations of family and language practice among bi/multilingual immigrant families. Since this selection criterion requires the selected research to be conducted among families who are immigrants and who are also bi/multilingual, the vast majority of studies that were excluded in this stage were studies that were not conducted among immigrants. This stage of screening reduced the number of manuscripts to 359.
Of these 359 manuscripts, 71 studies were excluded as secondary research and review papers. Thus, only articles presenting the results of primary research were retained. The remaining articles were then scanned to see if they discussed the topics significant to our review, that is, FLP, family, or language practice; the 29 manuscripts that did not address these topics were excluded. Finally, an in-depth screening was implemented for relevance to the research questions, and 96 manuscripts were rejected as irrelevant. This resulted in a final set of 163 articles.
In the initial screening, three authors (Authors 1, 4, and 5) screened all the articles based on the selection criteria, as presented below.

Selection criteria
To summarise, the following selection criteria (Figure 2), based on the PRISMA guidelines, were applied: (i) peer-reviewed articles and peer-reviewed book chapters published in English only; (ii) research population confined to bi/multilingual migrant families; (iii) research papers limited to primary research; (iv) studies mentioning the terms FLP or family or language practice; and (v) studies relevant to the research questions.

Methods
Various data points of interest were extracted from the final set of reviewed studies. This included the number of participants, study design, and various demographic features of the study participants, such as education and country of origin. The family types in each study were coded independently by the first and second authors. If differences in coding choices were found between the two authors, these were resolved through discussion using an open coding process and the coding was finalized together with the second and third authors. Notes were made on whether each family represented in the literature had dispersed relatives and, if so, who these dispersed relatives were. Visits to home countries and vice versa were noted. To understand how language practice is conceptualised with regard to the different modes of communication, a set of predetermined codes (closed coding) was used, with the remainder of the coding procedure being similar to that for families. Language practice is coded as follows: (a) traditional HL practice:-that does not involve communication, such as reading HL books; (b) traditional HL communication practice:involving traditional communication using postcards etc. or face-to-face (physical) communications; (c) digital HL practice:-that involves the use of digital modes of HL learning (but not communication), such as watching television or online videos in HL that do not involve communication with another person; and (d) digital HL communication practice:-that includes digital communication, mediated through various digital modes of communication, such as Skype. Unique instances of most items, such as the number of times digital HL communication practice was mentioned, were coded. Thus, one research paper could have multiple instances of a specific item, such as HL communication practice. Since this is a systematic review that seeks to establish how families and language practice are conceptualised in the literature, the results were tabulated as numbers and percentages, and modal statistics (i.e., the most frequently reported statistics across all reviewed papers) were reported when necessary. Such an approach is common when a literature review seeks to establish a baseline, first-time estimate of the phenomenon in question, for example, see Hirsch and Lee (2018).

Results
Before proceeding to our findings, it is necessary to provide a summary of the analytical approaches and research designs used by the reviewed studies since this has a bearing on their generalisability (for quantitative studies) and trustworthiness (for qualitative studies). In addition, it is essential to understand the demographics of the participants represented in the reviewed studies so that any findings can be interpreted within their appropriate contexts.
The most common analytical approach was thematic analysis (27% of included studies), followed by quantitative methods (16%), discourse analysis (13%), and grounded theory (9%). However, around 23% of the research papers did not clearly mention a specific design or analytical approach. Various data collection methods were used, with interviews (45%), questionnaires/forms/tests/surveys/assessments (22%), and direct observation of participants (13%) being the most common. Note that by the term 'questionnaires', we mean a set of questions that may be posed during the interview process or manually handed to a set of participants, with generally full participation or close to full participation. In contrast, by 'surveys', we imply lists of questions that are sent out by various means, with a certain fraction being returned to the researcher. Other methods were also used, such as focus-group interviews, stimulated recall, field notes, and linguistic assessment.
The number of families represented in each study varied from one to 258, with the median being four. The number of study participants ranged from a minimum of one to a maximum of 500, with a median of 11. The families included a minimum of one child to a maximum of eight children. The parents were from 60 different countries of origin, with China (11%), Russia (6%), and Korea (5%) being the most common. The studies also involved around 30 different countries of residence, with the United States of America (USA) (17%), Australia (10%), Israel (8%), and Canada (6%) being the most common. The language most frequently spoken by parents was reported as English (21%), followed by Spanish (8%), Chinese (7%), and Russian (7%). The most commonly reported (statistical mode) minimum length of residence of migrants in the current host country was 10 years, while the statistical mode of maximum length of residence was 20 years. The range varied from 6 months to 50 years.
Around 29% of the studies involved both parents. Data were collected from children in about 22% of studies, and only mothers were sampled in 22%. A smaller number of studies targeted specific groups. For instance, one study specifically targeted adolescents and another targeted young adults. Also, around 5% of studies targeted single-parent families, and 7% targeted just fathers.
Below, we discuss our results in the context of our research questions.

The concept of 'family'
Four types of families were represented in the reviewed FLP literature,-nuclear, extended, extended digital and nuclear digital ( Figure 3), with some studies including more than one type. We start by discussing the most commonly reported family type: nuclear families.

Nuclear families
Nuclear families form the vast majority of families examined (73%). Our investigation identified variations within these nuclear families with regard to HL and family composition. For the remainder of this section, we use italics to indicate a specific family type.
In the studies examined in this review, parents in nuclear families generally spoke at least two languages, one being their HL and the other being the language of the community to which they had immigrated. As such, two types of nuclear families can be distinguished. The first, of which 32 instances were found, are nuclear bilingual inter-married families or mixed families, in which parents from different nations spoke different languages. These included Russian-Estonian and Russian-Spanish families maintaining Russian as an HL in Estonia and Spain (Ivanova & Zabrodskaja, 2021), Russian-German families in Germany (Brehmer, 2021), and Portuguese-German families in Germany (Costa Waetzold & Melo-Pfeifer, 2020). Families with parents from different cultures and countries tended to interact with dispersed relatives (Lexander, 2021).
The vast majority of families are, however, of the second type, where both parents speak the same HL in addition to the language of the dominant community. We call these nuclear bilingual families. For example, one study examined families living in Germany in which parents spoke both Portuguese and German (Costa Waetzold & Melo-Pfeifer, 2020). Kopeliovich (2010) described a bilingual Russian-Hebrew nuclear family in Israel with eight children. The family had immigrated to Israel from Russia with four children and had another four children in Israel. This family settled in Jerusalem along with other Russian-speaking migrant families. Kopeliovich (2010) explained that these families possessed a multicultural ideology focusing on maintaining Russian as their HL in Israel. Other examples of nuclear bilingual families include Albanian families in Greece (Chatzidaki & Maligkoudi, 2013) and Pakistani families in the UK, including those with one British-born parent and one Pakistani-born migrant parent sharing the same HL (Curdt-Christiansen & La Morgia, 2018).
While social scientists sometimes distinguish nuclear families from single-parent families, we mention them here since they share similarities with nuclear families. As the name implies, in single-parent families, the child or children lived with one parent who was solely responsible for their child(ren)'s HL maintenance. For instance, Brehmer's (2021) study involved several Russian single mothers in Germany who were divorced or separated. Another study investigated Russian singlemother families in Israel, focusing on child-rearing and language (Zbenovich & Lerner, 2013). Three instances of such families were found.
Certain other interesting variations of nuclear families were also observed. One study described a nuclear family in the context of international adoption (Fogle & King, 2013). In such an international adoptive family, the parent(s) needed to learn the children's HL (through a language course or by employing other resources) to prepare for adoption. This was seen in a study of three international adoptive families in the USA: the English-speaking parents attempted to use the children's HL (Russian) in daily conversations after adopting Russian children (Fogle & King, 2013). In another instance, sojourning families lived temporarily for a few years in a foreign country and thus had to also plan for their return to their home country. In a study by Bahhari (2020), 10 Saudi-Arabian families in Australia maintained their HL (Arabic) in their home during their sojourn in Australia. The connotation of the term "sojourn" varies in the literature, with one author defining home visits to Taiwan during school holidays in Australia as sojourns (Eisenchlas et al., 2021). A final yet intriguing instance of nuclear families is vulnerable families, such as those represented in a study of HL maintenance among Mandarin and Cantonese-speaking parents in the USA with children with autism (Yu & Hsia, 2019). The study found a lack of HL-based language support for children with autism in the community and a shift in FLP priorities from HL learning to be able to communicate in English after an autism diagnosis.

Extended families
While nuclear families form the bread and butter of FLP research, some studies examined extended families (16%). In these families, children and parents lived together with other adult relatives, such as grandparents, uncles, or aunts, in the same physical place. Extended families offer an interesting basis for FLP research, with quite a few studies highlighting the significance of extended family members for HL maintenance. Examples included an Italian family in Canada consisting of a child and their parents and grandparents (Corsi, 2020) and a Chinese family in Singapore, in which the grandmother lived with the family, took care of the grandchildren, and helped them maintain their HL (Ren & Hu, 2013). Studies involving extended families most frequently reported grandparents as the extended family member living in the home (26 studies), with 31 reporting a variety of other extended family members. One study each reported the presence of uncles and aunts. Extended families may use multiple languages, which may provide a further layer of language support for children Curdt-Christiansen (2016) provided examples from Singapore of grandparents acting as linguistic gatekeepers in multilingual extended families, including a three-generation trilingual Chinese family, a bilingual Indian family, and a bilingual Malay family. The trilingual Chinese family had two English-speaking children living with their English-speaking parents, aunt and other family members. All family members spoke Mandarin, and all except the children spoke Hokkien. The aunt was the homemaker and the caregiver for the grandmother, uncle, and children when the parents worked. The bilingual Indian family was a family of teachers, in which the Tamil-speaking grandparents took care of the children while the parents taught at a school. Other members of the household included a Tamil-speaking uncle and a Filipina helper. This family is representative of an extended Indian migrant family living in Singapore, with family members mainly speaking Tamil (their HL) and English. The bilingual Malay family included Malay-speaking grandparents, two aunts who were university students, and an Indonesian helper living with the two children and their parents. Two important features to note from these families are the presence of multiple languages within the families and the promotion of a positive attitude towards the HL and other languages in the children through exposure to these languages. Some researchers have noted that families often tend to address the absence of extended relatives in their midst through the use of digital communication tools (Robertson et al., 2016). Conversely, the presence of relatives within the physical realm of the home reduces the likelihood of digital communication with dispersed relatives (Robertson et al., 2016).

Dispersed family
While extended families have many family members living in close proximity, nuclear families also retain connections with their dispersed extended family members. Indeed, one characteristic evident in many nuclear families was the practice of home visits to dispersed family members. Home visits to family abroad were found in around 47 studies, with five of these studies reporting (reciprocal) visits from family abroad to the focal family.
For example, Cape Verdean families in the USA consider it essential for children to visit Cape Verde and speak to grandparents in Cape Verdean Creole (Kaveh & Sandoval, 2020). Kim et al. (2015) observed three Korean parents in the USA who similarly considered home visits an essential aspect of language maintenance as home visits allowed children to practice their HL (Korean) with their dispersed family members. This trend points to the fact that nuclear families have dispersed relatives with whom they seek connection, which is also relevant to the use of digital communications, as discussed later.
The migrant FLP literature engages well with the concept of a dispersed family, or as introduced briefly earlier, geographically dispersed families who maintain contact with each other physically or digitally (Yeoh, 2009). As mentioned, most studies examined in this systematic review appear to be aware of the existence of dispersed family members, with 65 of 163 papers (40%) mentioning some form of a dispersed relative and 46 of 163 (28%) mentioning home visits. However, this engagement does not necessarily extend further into how these dispersed family members maintain contact with each other beyond physical visits and what effect this has on HL maintenance. It is very likely, however, that dispersed family members interact with each other in ways other than through physical visits (such as through digital means).

Digital families
A small number of studies explored what can be conceptualised as digital families, with two studies examining FLP within a digital nuclear family, with the father living away from the family (Lee & Pang, 2021), and 17 studies examining digital extended families, with various extended family members maintaining contact with the focal family (which was almost always a nuclear family) through digital means, such as Skype and WhatsApp (Gharibi & Seals, 2020;Hua & Wei, 2016;Lexander, 2021). Digital families are discussed further in the 'Digital HL Communication Practice' section below.

Other noticeable family features
Most studies investigated first-or second-generation migrants. Very few studies reported on the educational backgrounds or professions of family members. Among those that did, the modal value (most frequently reported statistic) of the minimum level of education reported among parents was a few years of schooling. In contrast, the mode for the maximum level of education was postgraduate education. While both professional/white-collar and trades/blue-collar professions were reported, the former were twice as numerous in the data. It is possible that parents with higher education are more concerned about their children's HL proficiency and use. These results may also represent a selection bias, with better-educated parents being more likely to participate in such studies (Li, 2015).

The concept of 'language practice'
The concept of language practice in the reviewed studies is closely related to the conceptualisation of the family. Since most studies conceptualise a family as a group living in the same physical location, language practice is also conceptualised as occurring within the home or the same physical location. Indeed, language practice does not have to involve communication, and may be done without the help of digital tools. We therefore begin our discussion by exploring studies that examine some of these means of language practice that do not involve direct communication. We use the term 'digital HL practice' for practice that involves digital media, and 'traditional HL practice' for that which does not.

Traditional and digital HL practice
Around 20% (57 of 281) of all instances of language practice mentioned in the examined papers involved traditional HL practice and did not include any communication or use of digital media (Figure 4). These include reading in HL and language study, such as children reading picture books in their HL (Portuguese) in Germany (Costa Waetzold & Melo-Pfeifer, 2020; Kirsch, 2012;Kirsch & Gogonas, 2018) and children practising their HL by reading textbooks on their own (Curdt-Christiansen & La Morgia, 2018). Parents also attempted to maintain their HL by reading newspapers (Russian) (Kang, 2015;Mori & Calder, 2017;Schwartz, 2008).
Digital tools provide unique and interactive opportunities to practice HL; thus, the FLP literature has extensively engaged with this topic. Around 51 instances of digital HL practice were found, accounting for 18% of all observed practices. Television, video, and radio remained common forms of practising and learning HL. Some studies showed how children's HL maintenance is influenced by watching films and cartoons in the HL (Eriksson, 2015;Hua & Wei, 2016). For example, children in some Greek families in Luxembourg watched Greek television (TV) channels to support HL maintenance (Kirsch & Gogonas, 2018), and children in Chinese families in Canada watched videos in their HL (Li, 2015). Russian children in Israel listened to the radio in their HL (Schwartz, 2008), and some Hakka Chinese children in Malaysia listened to Hakka radio programs at the suggestion of their fathers (Xiaomei, 2017). Browsing the Internet for HL material was also reported. One study, for instance, reported Russian children in Canada browsing websites in their HL (Makarova et al., 2019). Fuentes (2020) provides an example of HL practice using digital media among two Sinhalese families in the USA to illustrate their HL maintenance. However, these two transnational Sinhalese families had different motivations for maintaining their HL. The Kola (pseudonym) family maintained their HL (Sinhalese) because they had the long-term goal of returning to their home country of Sri Lanka. In contrast, the Nil (pseudonym) family emphasised HL maintenance out of fear of HL loss in the USA. Alongside their traditional HL practices, these families developed strategies to maintain some digital HL practices. The children in the Kola family practised their Sinhalese by reading Sinhalese newspapers online, while the children in the Nil family practised their Sinhalese through active engagement on Sinhalese websites. The children of both families regularly watched Sinhalese TV programs. Both families thus sought to maintain their HL (Sinhalese) through digital HL practice, thereby maintaining their Sri Lankan identity in the USA (Fuentes, 2020).
However, HL practice cannot remain confined to activities that do not involve communication, and thus, the FLP literature has also examined HL practices that involve communication. As such, we discuss traditional and digital means of language communication practice next.

Traditional and digital HL communication practice
The majority of studies in this review conceptualised language practice as faceto-face communication located within one physical location, such as the home. Thus, 51% of instances (143 of 281) of language practice instances identified in the reviewed studies consisted of traditional HL communication practice. As an example of traditional HL communication practice, we can turn to Eriksson (2015), who highlights how 15 Russian migrant families and 15 Latvian migrant families settled in Ireland maintained their HL using traditional means. These families described using traditional HL practices at home with their children, for instance, having a "Russian-only" or "Latvian-only" policy at home. They also taught their children Russian or Latvian and read HL texts with them. Some parents also gave dictations in HL (Russian) to their children at home, played memory games with them to develop Russian vocabulary, and invited Russian teachers into their homes to develop their children's HL skills. The previously discussed home visit strategy was also used, with some parents and children going to their home country, i.e., Russia or Latvia, to visit their friends and relatives during vacations to maintain their HL. Also, grandparents living in the families' homes in Ireland played a significant role in HL maintenance. All 15 children in this study communicated with their grandparents only in their HL (Russian). The HL practice strategies used by the Russian families in this study for their children were more formal (for example, involving dictation practice) and frequent compared to those of the Latvian families.
Other examples of traditional HL communication practice observed include children communicating orally in their HL at home with their parents (Gomes, 2019;Gu & Tong, 2020;Wilson, 2020), speaking in their HL (Mandarin) with their mother and nanny at home (Yu & Hsia, 2019), practising their HL (Russian) at home with their grandmother (Zbenovich & Lerner, 2013), or practising HL (e.g., Persian) with siblings (Kheirkhah & Cekaite, 2018). Still other examples include mothers reading storybooks in the HL to their children (Kirsch & Gogonas, 2018) and reading to their children for pleasure, as seen in the reading of Russian, Korean, and Japanese books by parents (Kang, 2015;Mori & Calder, 2017;Schwartz, 2008).
As discussed earlier, the family is not confined to one physical location, and neither is language practice. Dispersed families have existed since long before digital means of communication were available. These families have often retained contact through various traditional communication modes. One study reported traditional HL communication practice wherein the child maintained communication with their grandparents by sending postcards in the HL (Hakyoon & Myoungeun, 2020). Al-Sahafi (2015) reported communication between children, local family members, and dispersed family members, including grandparents, via the telephone, a communication mode that perhaps lies within both the traditional and digital realms. However, since most reviewed studies that mention the telephone (Brehmer, 2021;Gharibi & Seals, 2020;Lexander, 2021) do so in the context of mobile phones, and because at the beginning of the first decade of the 21 st Century, many traditional telephone calls are now routed through digital networks, we consider phone calls (voice calls) to be a form of digital communication.
Another option for retaining contact is physically visiting dispersed family members. Thus, some children communicated with their dispersed family members in their HL during their visits to their home countries (Bahhari, 2020;Chowdhury & Rojas-Lizana, 2020;Makarova et al., 2019;Tran et al., 2021). However, physical visits are expensive and time-consuming, albeit a more personal means of retaining contact and the development of digital modes of communication have allowed the retaining of contact among dispersed families and the practising of HL when opportunities arise (Taipale, 2019). Thirty instances of digital HL communication practice were found, representing around 11% of all instances of communication practice. Studies often treated digital HL communication practice as an afterthought in comparison to other traditional HL practices, and very few engaged directly with questions related to digital FLP (Lexander, 2021;Revis, 2021). Still, studies documented digital communication in HL between children and relatives (such as grandparents) overseas through video calls, voice calls, and other digital tools such as WhatsApp (Bahhari, 2020;Chowdhury & Rojas-Lizana, 2020;Hua & Wei, 2016).
A unique study by Lexander (2021) explored four Senegalese families' digital language practices in Norway. This study also reported digital communication between family members living under the same roof, unlike other studies in this space. In this study, digital extended families used a range of social media tools (including WhatsApp, Messenger, Short Message Service (SMS), Viber, Snapchat, Facetime, and Skype) to interact with their family members -both located physically together and dispersed -in their HL (Wolof) and other known languages. One family in this study was the Norwegian-Senegalese Coly (pseudonym) family, which consisted of a Senegalese mother and four children in their teens and twenties. Norwegian was the primary language of communication in this family. The children had learnt some English at their Norwegian schools, on TV, and on the Internet. Three of the family's four children had learnt Arabic from Koranic schools. In addition, Awa (the eldest child) had learnt French at her school. Lexander (2021) highlighted the modes of digital communication and the languages Awa used with her immediate family members for everyday interactions: she communicated with her mother and her siblings digitally through regular phone calls in their HL (Wolof) and through SMS messages in Arabic and sometimes in English using Facebook Messenger. With her dispersed family members, Awa used other modes of communication that are popular in Senegal. Her communication with her uncle in Senegal was mainly in English and written through Messenger. With her two aunts living in Senegal, Awa would speak and write in English and Wolof, using Messenger and Facebook with one aunt and Viber and WhatsApp with the other. Awa also communicated with her friends in Senegal through Messenger (in English), Skype and WhatsApp. Overall, her linguistic repertoire with families and friends, whether physically proximate or dispersed, was diverse, consisting of Wolof (HL), English, French, and Arabic, with some minor conversations in German and Norwegian. These languages were transmitted over multiple media; thus, for instance, she used WhatsApp, Messenger, Skype, Viber, Snapchat, and FaceTime when communicating with a friend in Germany. Conversations and language practices among the multilingual family members were thus negotiated or managed, and their relationships and HL were maintained in a polymedia environment (Lexander, 2021) that included digital communication modes.
Digital voice calls were reported even with the proliferation of more interactive video and multimodal communication services. There were reports of children interacting with their dispersed family members through mobile phone calls (Brehmer, 2021;Gharibi & Seals, 2020); one child from a Hispanic family in New Zealand spoke in their HL (Spanish) over the phone with dispersed members (Navarro & Macalister, 2016). An essential aspect of digital FLP communication is that while many migrants are situated within developed countries in the Global North, their dispersed families may be located in relatively underserviced rural areas of the Global South, where despite the ongoing proliferation of mobile and data services, connections can be poor and communication intermittent. One example appears in Revis (2021) study, which describes digital communications between a migrant family in New Zealand and their dispersed family in rural Columbia. Connectivity issues caused the family to be able to video chat only twice a year despite their desire to communicate more frequently.

Discussion and conclusion
This systematic review has identified the diversity of family types and language practices represented in contemporary migrant FLP studies. Nuclear families accounted for more than half of all families identified in this review. This is unsurprising since the father-mother-child(ren) conceptualisation of a family physically located in one place is the norm in FLP and elsewhere (Oyewumi, 2002). While 40% of all reviewed papers mention dispersed relatives in some form, they are typically considered separate and distinct from the focal family, and while their contribution to the child's language practice may be examined as an afterthought, their role within the conceptualisation of family is generally not considered in the FLP literature. The FLP literature, thus, broadly fails to account for the dramatic changes in the concept of what constitutes a family that have occurred in recent years due to geopolitical, sociocultural, and technological changes.
With regard to the families examined in the reviewed studies, our research highlights issues with access, equity and diversity in FLP research. For instance, none of the 163 reviewed studies, focused on lesbian, gay, bisexual, or transgender migrant families. No study involved families that have members with disability (i.e. vision or hearing impairment). There was also a significant selection bias towards welleducated migrants and those employed in white-collar professions. Issues with access to digital communication services were also identified among those with dispersed family members who were either not tech-savvy or lived in poor rural areas of the Global South (Brehmer, 2021;Revis, 2021). Finally, a third of all FLP studies focused on Chinese, Korean, and Russian migrants, and there seems to be a lack of studies concerning members of other large migrant groups, such as South Asians (Indians, Bangladeshis, etc.). These issues deserve attention in future research exploring language practice in the domain of the family.
The phenomenon of dispersed relatives raises the issue of language practice through digital communication in FLP. However, our analysis reveals an asymmetry between traditional and digital modes of communication, with traditional modes being both observed and investigated more often (51% vs 11%). Indeed, similar to traditional HL practice not involving communication (20%), digital HL practice not involving communication has been investigated more frequently (18%) than digital HL communication practice. While one possible explanation is that traditional HL communication practice may be far more prevalent than digital modes of language communication, it may also be that digital language communication practice, when present, is less likely to be reported in FLP studies. Another factor that may explain the relatively low prevalence of digital language communication practice in the reported studies is that although the technologies for such practices have existed for longer, their popularity has only taken off in the last decade, especially during the COVID-19 pandemic. Research conducted during the COVID-19 pandemic may not have been published at the time of data collection for this systematic review. Nevertheless, we believe that given the increasing importance of digital communication tools, there is a need for future FLP studies to investigate the role played by these tools in HL maintenance.
We identified a range of studies that mention the use of digital communication. Video conversations between children and grandparents were a common theme in many of these (Chowdhury & Rojas-Lizana, 2020;Nakamura, 2019). While some of these studies mentioned issues with digital literacy and access (especially, in the case of access, in rural areas) as barriers to utilising these technologies (Brehmer, 2021;Revis, 2021), most studies agreed on their usefulness. According to Lexander (2021), mobile phones provide the opportunity for migrants to enjoy personal HL conversations in the privacy and comfort of their rooms. In contrast, media such as Skype may serve as tools for parent-monitored conversations between children and dispersed family members (Chowdhury & Rojas-Lizana, 2020). Different digital communication media serve different purposes and may complement each other in providing various semiotic resources through which meaningful communications may be implemented (Lexander, 2021). Thus, text, video, images, and voice may be transmitted over different digital platforms, together providing semiotic resources for communication; this is very different from the face-to-face language practice that the majority of the FLP literature investigates.
The salience of grandparents in HL maintenance (Curdt-Christiansen, 2016) was clear across numerous papers included in this review, and is associated with both traditional families physically located in one place and families with dispersed family members, and with both traditional and digital modes of communication. While our research identified a variety of dispersed relatives with whom the HL was maintained, grandparents were the extended family member most likely to be physically present with the parents and children, and thus to play a role in the child's HL maintenance; they were also the dispersed family member most commonly discussed in the literature. Visits to grandparents in their home countries were commonplace (Eriksson, 2015). Not surprisingly, as discussed previously, grandparents were also the dispersed family members with whom children communicated most regularly in HL using digital communication media such as Skype. Given the salience of grandparents in both traditional and digital FLP, there is a surprising lack of attention to this topic in research. Future FLP research thus needs to focus on research questions related to this unique aspect of FLP.
While the findings above deserve further attention, we must also treat them with caution. Although our review uses the PRISMA approach to systematic reviewing and a comprehensive set of search terms, there is a possibility that we may have missed some studies on the topic due to any of several potential reasons, including the non-comprehensiveness of the searched databases and the specificity of the search terms; this is to be expected of any systematic review (Greenhalgh & Peacock, 2005). For example, our review used search terms limited to "family language policy", since our motivation was to engage with key concepts, including family and language practice, within studies that use the term "family language policy". Future reviews may include additional terms to broaden their coverage of FLP research. Given that we have reviewed a set of 163 studies identified from five databases, it is unlikely that the argument for the reconceptualization of family and language practice in FLP research would be undermined by the omission of papers due to our use of search terms. In addition, our review investigated the prevalence of certain FLP-related practices and characteristics within a comprehensive set of studies, which does not necessarily reflect the prevalence of these data points in the population. In other words, for example, the fact that 11% of the FLP cases described in the reviewed literature involved digital communication does not mean that 11% of all FLP communications among migrants are, in fact, digital; our statistics simply reflect what was found in the surveyed studies. It may also be possible for researchers to examine the distribution of dispersed families and the overall prevalence of digital communication media use among dispersed families. However, this would likely be an exercise outside of the sociolinguistics or language policy domain.
Finally, we would also like to use this opportunity to call for more research on FLP-related language learning outcomes from the use of digital communication media. The importance of these media can only be appropriately understood if their effects on children's HL learning can be understood, both quantitatively and qualitatively. Nevertheless, despite these limitations, our review is one of three existing systematic reviews on FLP and complements the other two studies on this topic, which focus on FLP outcomes and transnational families in migrant FLP studies (Hirsch & Lee, 2018;Hollebeke et al., 2020).

Recommendations
This review raises a few critical questions to be addressed in FLP. First of all, the results show that the concept of family is not homogenous and that there are wide variations within this concept. Seventy-three percent of the families represented in the surveyed studies were typical father-mother-child(ren) nuclear families. Studies on the effect of FLP on outcomes (Hollebeke et al., 2020) need to be cognisant of this variability, and FLP researchers should demonstrate greater sensitivity to the diversity of families in which FLP unfolds. It is essential to recognise that the family is a socio-cultural construction and that family members negotiate and inhabit roles constructed on cultural, social, and political bases, which profoundly mediate their involvement in managing language practices within the domain of the family (Gharibi & Seals, 2020).
Second, FLP research needs to demonstrate a critical awareness of different modes of language practice. While the vast majority of FLP language practice is still traditional, there is evidence that digital HL communication practice is occurring and probably increasing in frequency and significance. Future research should focus on exploring the various aspects of digital HL communication practice and be cognisant of such communication occurring even when the research is focused on traditional FLP. Researchers can incorporate these shifting conceptualisations of language practice by utilising research tools such as multimodal analysis and online ethnography, which acknowledge the multimodal, semiotic nature of digital communication among dispersed families (Kress, 2001).
To summarise, the shifting nature of the family due to global migration needs to be acknowledged, and we believe that future research on FLP should engage with the reality of dispersed family members and their role in FLP. This is best realised through a conceptualisation of language practice as multimodal and digital and as providing semiotic resources for communication in addition to being face-to-face and physical.
Funding Open Access funding enabled and organized by CAUL and its Member Institutions. University of New South Wales, Research, Priyanka Bose, Training Scholarship, Priyanka Bose.

Declarations
Conflicts of interest The review work was financially supported by a Research and Training Scholarship and a Waratah Scolarship that the first author (Bose) received at University of New South Wales, Australia. The corresponding author (Gao) is a member of the journal's editorial board. We have no other competing interests to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.