Sweden’s multilingual language policy through the lens of Turkish-heritage family language practices and beliefs

1.1 Family language policy and language transmission

Family language practices concerning heritage-language maintenance and shift may stem from deliberate management efforts, but language practices are also known to be volatile and to evolve in ways that are not consciously planned. The term family language policy (FLP) covers both such evolving language practices and parents’ conscious planning efforts (e.g. King et al. 2008; Lanza and Lomeu Gomes 2020; Schwartz 2010). FLP is a relatively recent concept at the interface of child language acquisition and sociolinguistics, where FLP at first mainly referred to “explicit and overt planning in relation to language use within the home among family members” (King et al. 2008: 907), such as making the decision to follow a one-parent/one-language strategy. FLP was soon expanded to also include implicit and covert family language practices, as studying these practices sheds light on the de facto language policy (Lanza and Lomeu Gomes 2020). For groups with a history of migration, family members and the family home typically constitute the core domain for heritage language use.^[2] Children growing up in multilingual homes will learn the majority/societal language one way or other, but not necessarily learn to speak the minority/heritage language (de Houwer 2007), unless family language policy pays special attention to that language.

Classic sociolinguistic theorising on language transmission (e.g. Fishman 1970) has put forward a three-generation model of language shift, according to which the first generation of immigrants (G1) adds the new societal language to their native language. Their offspring (G2) grow up bilingual with both the heritage and the majority language. G(eneration)3 grows up largely monolingual in the majority language. The shift to the societal language and loss of the heritage language is thus hypothesized to be completed by G3. Today we know that things are not quite as simple as suggested by this assimilationist three-generation model, since empirical studies document considerable variation in the extent and tenacity of heritage-language maintenance across families and communities. Influenced by a multitude of factors, processes of acculturation and language shift can be faster – or slower, stretching over more than three generations (Schwartz 2010; Spolsky 2012). Also, not all speakers are straightforwardly classified as G1, G2 or G3, but might be in-between generations (Pearson 2007): Some immigrated as children and were schooled in the new language setting, which may promote language shift, others grew up with one second-generation parent and the other parent being a first-generation newcomer who revitalized the use of the heritage language in the family. In models by e.g. Clyne (1991) and Fishman (1991), the family is seen as a bulwark against outside pressure, it is “the most common and inescapable basis of mother tongue transmission, bonding, use and stabilization” (Fishman 1991: 94).

Parents with a history of migration are not entirely free in making choices about family language use. Deeply embedded cultural values and identity issues influence parents’ language beliefs, attitudes and practices when bringing up children. But these are also subject to social pressures, including the language ideology of the host society, via its institutions, educators, health professionals, the media, and public discourse. The interplay of these societal and individual factors may lead to conflicts in the decision-making and implementation of FLP in heritage-language homes.

Interactions between family members in the home are generally seen as central for heritage language transmission, maintenance and development. Parents (or other primary caregivers) are key agents, and their language practices, both concerning the amount of input and patterns of interaction, can be instrumental in whether a young firstborn or only child becomes an active user of the heritage language or not (Hoff et al. 2014; Lanza 1997). According to large-scale survey studies, the highest chance for a child to become an active speaker of the heritage language is when parents exclusively or predominantly use this language with the child (de Houwer 2007; Jin et al. 2017).

At the same time, parental choices alone will not guarantee language upkeep, as parents and child do not exist in a vacuum. Parents go to work and need to socialise and adapt to function outside the home. Sooner or later, children will enter formal schooling and may also enter daycare services prior to school, and spend a substantial portion of their waking hours being socialised in the majority language. The educational system, as well as a child’s peer group, play a major role in shaping child language use. Even in households where parents strongly advocate heritage-language upkeep, a child will bring the majority language and aspects of mainstream socioculture into the home (Hoff et al. 2014; Süverdem 2022). By using the majority language in the home, the child invites more input in this language, triggering a self-reinforcing cycle (Pearson 2007: 400−401). In turn, parents will use the majority language more than before in parent-child interactions, even if this may happen unconsciously (Hoff et al. 2014; Prevoo et al. 2011). The child is thus an important agent in shaping and modifying family language policy; s/he may put up resistance to parental decisions and practices, and affect change.

Siblings, extended family and friends may also be closely involved in the raising of a child. With a single child, parents have a measure of control over home language use, but when siblings arrive, family language dynamics change. Parents cannot generally control which language siblings choose to speak with each other. Siblings have been found to interact in the majority language to a greater extent than what is found for parent-child interaction (Barron-Hauwaert 2011; Bezçioğlu-Göktolga and Yağmur 2022; Bridges and Hoff 2014; Kheirkhah and Cekaite 2018). School-aged siblings pass on the majority language to their younger siblings, but they also speak it to their parents, which in turn leads to a greater use of the majority language in the home, at the expense of the minority language. Siblings can thus contribute to language shift. By contrast, grandparents and other relatives are often seen as key agents promoting heritage language and heritage culture upkeep (Et-Bozkurt and Yağmur 2022; Süverdem 2022).

Parents may make deliberate efforts to boost exposure to the heritage language in the home, for instance by consistently choosing to communicate with their child in that language and/or making use of certain insisting discourse strategies from early on (Lanza 1997), by using minority-language media, or by bringing an extra native-speaker caregiver into the household. Joint picture/story book reading activities, singing and promotion of literacy is also common, as is the establishment of other enjoyable family cultural traditions or rituals strongly associated with the minority language (Schwartz 2010: 181−183). Whilst no particular family language policy has been demonstrated to guarantee that the child will become an active heritage-language speaker, “research does indicate that lack of attention to language planning in the home may lead to language shift” (King et al. 2008: 916).

Parents may also attempt to choose their children’s external environments very carefully, for instance by seeing to it that the child keeps close contact with relatives and minority-language speaking friends, by extended visits to the heritage country, by choosing to live in a neighbourhood with a high concentration of minority-language speakers, by sending the child to heritage-language classes and/or to a bilingual (pre)school instead of a monolingual mainstream one. Even so, such efforts are not just a matter of parental choice. For instance, the host country may not offer bilingual schooling in the minority language, and the family may not have the resources to be able to afford books, private language tuition, travel, or the loss of earnings that can be a consequence of stay-at-home parenting.

The social, economic and political conditions in the host country and its dominant language ideology will also impact family language policy. Mainstream society may promote, tolerate or repress the maintenance and development of heritage languages. Political decisions regarding language policy and their implementation can influence the beliefs formed and decisions made by family members concerning their everyday language use (Bezçioğlu-Göktolga and Yağmur 2022; Spolsky 2012). The authorities may officially support multilingualism and minority language rights, but how is this implemented in practice? Does the de facto language policy encourage language maintenance or shift? For instance, the use of heritage languages in preschools, school corridors, classrooms and schoolyards may be openly welcomed, tolerated or expressly forbidden. Mother-tongue tuition may be promoted, offered grudgingly only to some, or not at all; it may be included in (or excluded from) the curriculum and ascribed academic value (or not). In the educational and health systems, teachers, daycare staff, pediatric nurses and language therapists may counsel families for or against the use of the heritage language in the home. Such advice may increase or allay language anxiety in parents (Bezçioğlu-Göktolga and Yağmur 2018; Pulinx et al. 2017; Sevinç and Dewaele 2018), and thus the agency of educators and health professionals may impact FLP. Policies and attitudes towards ethnolinguistic groups and heritage languages need not be constant but may fluctuate over time.^[3] Family language practices, management efforts and beliefs do not emerge independently and are not isolated from society.

1.2 Aim and research questions

This paper explores the family language policies (FLP) of Turkish-heritage families in relation to Swedish multilingual language policy. As research on FLP in a Turkish-Swedish context has been lacking so far, it is worth putting an empirical, descriptive focus on family practices and beliefs. The following research questions are asked:

1. What are the predominant family language practices, taking into account variation in family constellation and background?

2. Which beliefs and attitudes do the parents have towards multilingualism and heritage-language maintenance?

3. How do family language practices and/or beliefs change over time?

4. Do the parents receive professional advice concerning their children’s multilingualism, and if so which?

5. How well do the family language practices and beliefs, as well as the advice given to parents, fit with Swedish mainstream language policy?

The paper is structured as follows. Section 2 characterises the Swedish context with regard to welfare, early education and school and language ideology, concerning official policy documents and implementation, with a special focus on mother-tongue tuition. Section 3 outlines the method of the study, including the sociolinguistic background of the participants. Section 4 reports empirical findings from the parental questionnaire survey and from the follow-up interviews and observations. Section 5 discusses the observed family language practices and beliefs with reference to some earlier findings for the same ethnolinguistic group in other settings, and in particular, in relation to Swedish state-level language policy.

2.1 Welfare, early education and school

The Swedish welfare system is generally regarded as highly developed and relatively successful in handling poverty and deprivation. Immigrants enjoy full access to health, schooling and social services. Early-years education is comprehensive and available regardless of social and economic situation. All children and adolescents who are registered residents of Sweden, including temporary residence-holders (Education Act 2010; Skolinspektionen 2013), have the right to education through publicly funded preschooling and schooling, as well as recreational childcare after school hours.

In Sweden, most parents work outside the home, and institutional childcare is widespread. According to recent figures (Statistics Sweden 2019b), 94 % of all three-to-five-year-olds attend preschool. Indeed, preschool (förskola) starts at a very young age (age one or two), and most children, including children of migrants, attend preschool for a major part of the day (six to 8 h/day, 30−40 h/week). Preschools are bound by the national curriculum (Skolverket 2018) to actively foster language, general cognitive and social skills, although there is variation in how this curriculum is put into practice. Preschool (for children between age one and six) is followed by förskoleklass (Grade 0), a preparatory year for primary school proper (starting at age 7). School is compulsory for nine years (Grade 1−9, age 7−15), though most pupils also complete three years of practically or academically oriented upper secondary education (Grade 10−12).

It is generally assumed that children from families with higher socio-economic status (SES) more easily get access to the types of learning experiences that stimulate language and other aspects of cognitive development. As preschool in Sweden is comprehensive and accessible from an early age irrespective of parental background or income, opportunities may be more equal than in some other countries. Children from less stimulating home learning environments may get a boost at preschool, and some differences between children from different SES backgrounds may potentially be levelled out.

Preschools and schools are generally run in Swedish. For children who have a home language other than Swedish, early and extensive preschool attendance may promote Swedish language proficiency. With time, they may develop a preference for Swedish over the minority language, which in turn may affect language use in the home environment.

2.2 Language ideology/policy in Sweden and home language education

Swedish is the principal and official language by law (Language Act 2009, Språklagen) and has long enjoyed majority sociocultural status. At the same time, official language policy at the state level encourages multilingualism and the development and upkeep of home languages other than Swedish. Minority language rights were protected in the 1974 constitutional reform (Justitiedepartementet 1974, Ch. 1, §2): “the possibilities for ethnic, linguistic and religious minorities to preserve and develop a cultural and social life of their own should be strengthened”. Thirty-five years later, the Language Act (2009, §14) reconfirmed these minority language rights, stating that persons with other first languages “are to be given the opportunity to develop and use their mother tongue”.

Already in 1977, the Home Language Reform introduced home language education for minority-language children. The current Education Act (Skollagen, from 2010) enshrines children’s rights to mother-tongue tuition (MTT, modersmålsundervisning), i.e. classes devoted to developing oral proficiency, and later, literacy, in the minority home language. MTT is offered as an elective subject as part of the school curriculum in primary and secondary school (i.e. not as an extra-curricular activity organised by volunteers, minority organisations, or foreign agents, as in many other countries). According to the national curriculum (Lgr11), MTT “should give pupils the opportunity to develop knowledge in and about their mother tongues”, and “develop their cultural identity and become multilingual” (Skolverket 2019: 87). The corresponding support for preschoolers is called mother-tongue support (modersmålsstöd). The national curriculum for preschools (Lpfö18) states that “children with a mother tongue other than Swedish should be given the opportunity to develop both their Swedish language and their mother tongue” (Skolverket 2018: 9).

Unusually from an international perspective, preschoolers and school-age pupils are entitled to home language education by law. Such legislation harks back to the pluralistic, multicultural politics of the 1970s that stressed the rights of individuals to their language and ethno-cultural heritage. Reference was made to the UNESCO recommendation that “[i]t is axiomatic that the best medium for teaching a child is his mother tongue” (UNESCO 1953: 11). A widely disseminated book urged that the threat of ‘semilingualism’ (halvspråkighet) should be staved off by fostering both the minority home language and Swedish (Hansegård 1968).^[4] A government-appointed scholarly expert commission also recommended this; all political parties agreed on the importance of immigrant children mastering their home language, as this would help them in learning Swedish (Borevi 2002: 209−210; Salö et al. 2018). The aim was ‘active bilingualism’ (aktiv tvåspråkighet). There was general agreement that MTT would have positive effects for the individual as well as for society at large. It also meshed well with the idea of an egalitarian and multicultural Sweden. MTT thus became a cornerstone of official state language policy.

MTT has continued to receive broad and vocal support from Swedish academics, especially from language scholars, and politicians have tended to accept these expert opinions (Salö et al. 2018; SOU 2019:18). In addition to the humanitarian viewpoint that MTT is beneficial for the development of cultural identity, it is argued that proficiency in the L1 (boosted by MTT) will strengthen the acquisition of (L2) Swedish, referring to Cummins’ (1979) threshold and interdependence hypotheses. MTT proponents also increasingly claim that there are additional benefits linked to MTT, such as improved skills in general school achievements and literacy, and try to back this up with statistics (Ganuza and Hedman 2018, 2019; Skolverket 2008; SOU 2019:18). In part, such arguments are put forward to fend off recent far-right political attacks.^[5] Yet focusing purely on utilitarian arguments may overshadow the symbolic function of MTT, namely to signal to pupils, families and society that minority languages are valued in themselves.

As mentioned earlier, the Multiculturalism Policy Index (2021) evaluates the multicultural policies of 21 Western countries regarding immigrant minorities, and one of its parameters is bilingual education, which also includes MTT. Sweden has held the top score on this parameter since the beginning of the index (1980), and initially was the only country to do so. If we take legislation and state-level language policy documents as a measure of the dominant language ideology in Sweden, multilingualism, the use of minority home languages and MTT in schools are openly promoted. Yet if we take the actual implementation of MTT as a yardstick for language policy today, the picture becomes more blurred (Borevi 2002; Ganuza 2019; Hyltenstam and Milani 2012; Salö et al. 2018).

When home language education was introduced in 1977, it was relatively well-funded, with earmarked state funds, where the municipality received a fixed share of full-time-equivalent teacher salary for each child enrolled in home language education. In 1990, the Swedish educational system was decentralised, and schooling responsibilities, including MTT, were transferred to the municipalities and lie with them to this day. Previously earmarked state funding for MTT was abolished, and the local authorities could set their priorities differently. This has given municipalities and schools extensive room for acting in ways that do not necessarily adhere to the official language ideology of the state.

Whilst the Education Act (2010: 800 §7) says that MTT must be offered nation-wide to all children growing up in families with a home language different than Swedish, conditions have tightened considerably during the past decades. Children are only eligible for MTT if the mother tongue, e.g. Turkish, is spoken in the home on a daily basis, and the child has basic knowledge of the language. Also, the School Ordinance (2011: 185, Skolförordningen, Ch. 5 §10) allows local authorities to opt out of MTT if there are fewer than five pupils for a particular language in the district or if there is no suitable teacher.

For budgetary or other reasons, municipalities do not always strive whole-heartedly to offer MTT. A survey by the Swedish Language Council (Spetz 2014) found that MTT provision varies greatly across the country, and that only 25 % municipalities offered MTT to all pupils who were eligible and had applied for MTT (Spetz 2014: 28−29). The report pointed to constant problems with implementation and marginalisation. For instance, classes may be offered at inconvenient hours or only offered in one school, making it cumbersome for children from other schools in the municipality to travel in to be able to attend. Quite often, pupils entitled to MTT do not get taught because class size is too small or because no suitable teacher can be found.^[6] Another complaint voiced by families and MTT teachers is that children of very different proficiency levels and/or ages are combined into one class to make up the numbers, making effective pedagogy difficult. MTT is poorly integrated with other school subjects and other aspects of school life, and MTT teachers shuttle between schools and are rarely included in regular staff meetings. Altogether, this signals that MTT and minority home languages have a lower status than other school subjects, including foreign languages (English, French, German, Spanish). Over the years, the hours afforded by municipalities to MTT classes have been cut down to 30−60 min/week.^[7] The provision of preschool mother-tongue support was already slashed (from 64 % to 12 %) during the 1990s, when earmarked state funding was discontinued (Nordenstam 2003). A few municipalities still offered some mother-tongue support during the 2010s, but have discontinued it since. In sum, there is a discrepancy between the legal status of home language education and its practical implementation. This discrepancy may send out signals different from official state language policy.

The Swedish National Agency for Education (Skolverket) registers how many pupils are eligible for MTT. During the academic year of 2021/2022, 28.6 % of all pupils in Grade 1−9 (age 7−15) were entitled to MTT and 58.5 % of these pupils were offered and attended MTT (Skolverket 2022).^[8] Turkish is one of the languages with the highest MTT attendance. Every year roughly 7,500 pupils are eligible for Turkish MTT in Grade 1−9, and slightly more than half of them, 53 %–57 %, also receive it (figures vary somewhat from year to year). Turkish is amongst the 10 largest MTT languages in Grade 0 (förskoleklass), and for many decades until 2016, Turkish was also amongst the ten largest MTT languages in Grade 1−9.

Apart from MTT classes, pupils are also legally entitled to ‘mother-tongue study guidance’ (studiehandledning på modersmål), if needed. This means that a MTT teacher will sit in during lessons in other school subjects, to translate and help the child understand subject matter. The extent to which this service is provided varies.

MTT classes and mother-tongue study guidance are official language policy, but so is Swedish as the default medium of instruction. Whilst there is no nation-wide language policy that would advocate the sole use of Swedish on school grounds, small-scale ethnographic studies nevertheless indicate that certain schools try to enforce a monolingual local language policy and reprimand pupils for using other languages than Swedish. For instance, Ganuza and Hedman (2017) observed how teachers in some schools explicitly asked bilingual students not to speak their mother tongue in the dinner hall and in the school corridors, relegating the use of minority languages to MTT lessons only. However, schools and individual teachers seem to vary greatly in the extent to which they tolerate or repress the use of the mother tongue. We are not aware of any larger, systematic studies of language management in schools in Sweden.

Preschools are legally required to provide opportunities for developing the minority language (Skolverket 2018: 9). It is not quite clear how this is to be implemented though, as there no longer is any requirement to offer mother-tongue support to preschoolers, nor any earmarked state funding for it. Multilingual practices certainly do occur in preschools, due to linguistically diverse child intake and multilingual staff hirings in many urban areas. Some children might thus have one or two staff members who speak and/or understand their heritage language. However, this is not the same as having MTT teachers in preschools. Ethnographic work suggests that in some municipalities, preschools are still served by visiting MTT teachers for some languages with a high intake of children with the same heritage language, whilst regular staff in practice vacillate between promoting and inhibiting certain language uses (e.g. Puskás 2019).

3.1 The project

As part of a larger research project on child bilingualism in Sweden, BiLI-TAS,^[9] data was collected from 105 Turkish/Swedish children age four to seven and their parents. There were 27 four-year-olds, 23 five-year-olds, 27 six-year-olds and 28 seven-year-olds, including two who had just turned eight years. Even though the primary objective of the larger project was to measure and compare the children’s language skills in Turkish and Swedish, parental questionnaire and interview data provide valuable insights on underlying family language policies, which are investigated in the present paper.

3.2 Data collection

Turkish-speaking families were recruited by contacting around 200 preschools and schools in urban areas of eastern central Sweden, as well as via mother-tongue teachers, places of worship, cultural associations, community centres and word of mouth. In the end, the participating children came from 50 different (pre)schools in the metropolitan region of Greater Stockholm and nearby larger cities (where the majority of Turkish speakers in Sweden are located).^[10] Families received oral and written information about the project in both Turkish and Swedish. Informed parental consent was obtained in writing. Families could terminate their participation at any time. The children carried out a range of language production and comprehension tasks (not reported here, see Bohnacker 2020; Bohnacker et al. 2016, 2020, 2022; Bohnacker and Karakoç 2020; Öztekin 2019).

The parents filled in a five-page paper-and-pencil questionnaire in Turkish or Swedish, and those who disliked writing were interviewed via telephone in Turkish. The questionnaire had been developed by the project team for several languages and piloted with Turkish-speaking parents, clinicians, MTT teachers and community members. It contained 36 questions (some yes/no, some estimation scales, and some open-ended questions), targeting the child’s language development, age at first regular exposure to the respective language, (pre)school attendance, parents’ language and educational backgrounds, and family language use in the home, including the parents’ language(s) spoken with each other and to the child, the child’s language spoken to the parents and to the siblings, language use between siblings, extended family etc. Parents also estimated the proportion of daily language input to the child (for details see Öztekin 2019). We also queried language-related activities for both languages in and outside the home, such as storytelling, joint book reading, singing, media consumption and home language education.

Questionnaire data was available for all 105 children, though in some cases, parents had left certain questions blank. Responses were collated in spreadsheets and anonymised. They form the database for the questionnaire survey results, which are reported in detail in Bohnacker (2022). They are summarised below and also analysed with regard to parental history of migration for the first time here.

For the longitudinal follow-up, ten of the originally four-year-old children were seen again two years later when they were six years, and they did the same language tasks (reported in Öztekin 2019).^[11] Their (pre)school learning environment was observed as well. During a home visit, a native Turkish-speaking researcher observed the child in their family environment and took field notes. This researcher also carried out a structured face-to-face interview with the parents in Turkish concerning language practices, management efforts and beliefs, and what had changed over the past two years. For instance, we inquired in detail about language use between family members, whether the parents were content with the child’s language use, and asked about any advice the parents had received from others regarding bilingualism. The parents also filled in an abbreviated questionnaire, which included 17 of the questions from the questionnaire two years earlier, targeting aspects of language use that may have changed. Responses were collated in spreadsheets and lengthier answers were thematically grouped. Some observations from the follow-up were reported in Öztekin (2019). The data have since been analysed further by the author and PI of the BiLI-TAS project. A digest of the answers from the interviews and questionnaires is provided below. These results are discussed here for the first time in relation to language policy.

3.3 The children and their families

Nearly all children (94 %, 99/105) in the large sample grew up in two-parent households; six lived with single parents. Most children had siblings (92 %, 96/105). Nearly all children (90 %, 95/105) were born and had lived in Sweden all their lives, only few (10 %, 11/105) had immigrated. By contrast, most parents were born in Turkey (70 %, 146/210), only 22 % had been born in Sweden, 3 % in a third country (and for 6 % this information was missing). In more than 90 % families, both parents had Turkish roots. Only few children (7 %) had two parents who were born in Sweden.

Another way to characterise the children’s backgrounds is not to focus on where the parents were born but where they were raised, and group them into generations.^[12] Table 1 shows that slightly more than half the children (52 %, 55/105) belonged to G(eneration)2, with two first-generation parents who had immigrated to Sweden as adults. Nineteen children (18 %) were G3, with two Sweden-raised parents. Around one third of the children (30 %) fell in between generations (“G2.5”) and either had two Turkey-born parents who came to Sweden as children, or had one Sweden-born parent who set up family with a newcomer from Turkey.

The breakdown by generation suggests that many children in the sample may have Turkish-L1 parents and only few two Swedish-L1 parents. Our questionnaire explicitly asked about languages. Most parents (72 %, 156/210) considered Turkish as their native language (L1), where for 70 %, Turkish was their only L1, for 2 % Turkish alongside Swedish, and for 1 % Turkish alongside Kurdish. Only 4 % parents reported Swedish as their L1, and 19 % reported a different L1, in most cases, Kurdish (Kurmanji). In 15 % of the families (16/105), both parents considered themselves non-native Turkish speakers (L1 Kurdish), but reported that they regularly spoke Turkish in the home.^[13] We chose not to exclude these families as they are an integral part of the Turkish-speaking community in Sweden. The different language constellations may however affect family language practices (see Results section).

Not surprisingly considering the parents’ L1s, nearly all children (93 %, 98/105) had been continuously exposed to Turkish from birth. 4 % were exposed to Kurdish from birth and to Turkish from age one to two, and for 3 %, this information was missing. Regular exposure to Swedish started before age 3;0 for 81 % (85/105) of the children, typically via preschool; this was the case for G2, G2.5 and G3 children alike. Children with later ages of onset for Swedish were usually born in Turkey and had immigrated to Sweden with their families.

Whilst most children were bilingual, 12 % (13/105) grew up in trilingual homes. Six of these children actively used a third language. Seven more children were exposed to a third language from one or both of their parents (mostly Kurdish), but did not speak it themselves.

Concerning socio-economic status (SES), parental educational levels ranged from less than six years of primary education to doctorates. Parental education levels varied similarly for G2, G2.5 and G3, except that all Sweden-raised parents had attended secondary school. The majority of parents had completed upper secondary school (with 12–13 years of schooling) but had no tertiary education. Most parents (72 %, 152/210) were in paid employment outside the home.^[14] Parental occupational levels varied from unskilled labour to senior professional, with a preponderance of service workers, craft workers, clerks and technicians. Most families lived in urban multicultural low-status neighbourhoods.

The backgrounds of the ten families participating in the longitudinal follow-up were representative of the larger sample. Most children (9/10) grew up in two-parent households, all ten were born in Sweden. Parental educational levels varied from primary school to postgraduate degrees. Most parents (17/20) were Turkey-born, and three were Sweden-born (including two with Turkish roots). Table 1 shows that most children (7/10) were second-generation, with both parents raised in Turkey and having migrated to Sweden as adults. Only one child was G3, where both parents had grown up in Sweden. Two children were in-between generations (G2.5). Nearly all (19/20) parents considered themselves native speakers of Turkish: Most (15/20) stated that Turkish was their L1 (including one bilingual Turkish/Swedish parent); 4 parents regarded both Turkish and Kurdish as their L1s, and one parent’s L1 was Swedish (this was the only parent in the sample without Turkish roots). In the three families where Kurdish was spoken alongside Turkish, the children understood some Kurdish, but were described as not being active users of Kurdish.

The sociolinguistic background of the families in the present study is in line with previous observations of a strong pattern of endogamy in the Turkish-heritage community in Sweden and elsewhere (e.g. Backus 2004; Yağmur 2016; Aktürk-Drake 2017, 2018; Yağmur and van de Vijver 2022). Setting up family with a spouse from the same country of origin (Turkey) may support the upkeep and transmission of the home language to the next generation. Interestingly, nearly one fifth of the parents in the present sample reported Kurdish, or Kurdish and Turkish, as their native language(s). In these families, children may thus be growing up in a trilingual home environment, which is likely to affect family language policy.

4.1 Parental language use and parent-child interaction

Parental language use in the home is a reflection of family language policy, whether it is consciously planned or not. In the large-scale survey (Bohnacker 2022), the majority of parents reported that they spoke almost only or mostly Turkish to each other (61 %) and also to their child (63 %). Very few parents (2 %) spoke mostly Swedish to each other or to their child (see also Öztekin 2019). However, when broken down into generational subgroups, we see a clear difference in the parents’ use of Swedish in the home: It is predominantly the G3 children’s parents who were raised in Sweden that communicate with each other and their child to a large degree in Swedish, see Table 2.

Unsurprisingly, Turkey-raised Turkish-L1 parents spoke (almost) only Turkish in the home. Not evident from the generational breakdown above is the fact that some Turkey-born/-raised parents also reported speaking a language other than Turkish (often Kurdish) to each other (11 %) and to the child (7 %). Only two couples, both Sweden-born, spoke mostly Swedish to each other (2 %). The remaining parents spoke both Turkish and Swedish to each other (14 %) and to the child (30 %); interestingly, these parents were nearly always raised in Sweden and also had a high level of education. Overall, far more parents were speaking at least some Swedish with the child than with each other.

In the longitudinal study as well, most parents (17/20, born and raised in Turkey) reported that they communicated with each other only or mostly Turkish, and mostly spoke Turkish and sometimes Swedish to their child, just as they had done two years earlier. In three families, the parents spoke both Kurdish and Turkish to each other. In two other families, one parent communicated in Turkish and Swedish with the child. These reports were confirmed during the home observation visits, when parents spoke mainly Turkish, watched Turkish TV channels and spent time with Turkish-speaking friends and relatives. One household stood out from the rest though: Here the native Swedish father spoke mostly Swedish to the child (G3), and the mother both Swedish and Turkish. This was the only mixed-language couple in the sample.^[15] The parents stated that they were now making made a conscious effort to speak more Turkish at home to ‘rescue’ the minority language.

Concerning child language use, the majority of children (65 %) in the large-scale survey were reported to speak almost only or mostly Turkish to both parents. 12 % (12/105) spoke only/mostly Swedish to both parents, 18 % spoke both Turkish and Swedish to their parents, and 5 % also a third language. Child-to-parent language use may be indicative of a generational shift of language preference, and this is confirmed by the breakdown in Table 3: Nearly all children who only/mostly spoke Swedish with their parents were in fact third generation, born to parents who themselves had been raised in Sweden. Still, most of the four-to-seven-year-olds, including slightly more than half of the G3 children, were reported to speak Turkish with their parents to a considerable degree.

As Table 3 shows, only a very small proportion of the G2 and G2.5 children were reported to communicate only/mostly in Swedish with their parents (and/or siblings, see next section). Interestingly, nearly all of these children grew up in trilingual homes, with Kurdish or another minority language besides Turkish. Though the small group sizes preclude meaningful statistics, it appears that such a trilingual language setting may speed up the process of language shift to the majority language Swedish.

In the longitudinal follow-up, the majority of children (9/10) at age four and also at age six were reported to speak almost only or mostly Turkish to both parents, except for one child who interacted only in Swedish with one parent and in Turkish and Swedish with the other parent (this was the mixed-language couple mentioned above). Thus, child-to-parent language choice mirrored parent-to-child language use (see also Öztekin 2019: 49−50). In general, compared to two years earlier, the parents noted only small changes in the child’s language choice towards them, if any.

In the interviews, we also asked the parents about interaction strategies, e.g. whether it happened that the parent spoke Turkish and the child responded in Swedish, and if so, what the parent would do, whether s/he would then switch to Swedish. Three parents said this situation did not occur (or very rarely) because the child always spoke Turkish with them. Most other parents acknowledged that such situations did occur, but they handled it in different ways. Some answered that they insisted on continuing in Turkish, that they rephrased or explained in Turkish and/or asked the child to say it in Turkish. One mother adapted to the language choice of the child, and in such situations tended to switch to Swedish herself. Several parents said that they did not behave consistently (“it depends”). Still, except for one family, parental language use with the child appears to have been predominantly in Turkish.

4.2 Siblings and grandparents

Compared to parent-parent and parent-child interactions, language use between siblings was somewhat different. In the large-scale survey, communication between siblings was reported to be only/mostly in Turkish for only 38 % of the children. 45 % of the children communicated with their siblings in both Swedish and Turkish, and 16 % only/mostly in Swedish.^[16] These overall percentages hide a generational pattern which becomes evident by the generational breakdown in Table 3: The children who interact with their siblings mostly in Swedish are nearly all third-generation (with two Sweden-raised parents): 39 % G3 children speak mostly Swedish with their siblings, compared to only 10 % in the G2 group. As mentioned above, the few G2 and G2.5 children who mostly spoke Swedish to their siblings tended to grow up in trilingual homes where also another minority language, such as Kurdish, was spoken.

In the longitudinal study, only one child spoke only/mostly Swedish with her sibling. As this G3 child grew up in a household with a native Swedish-L1 father, she was exposed to a lot of Swedish anyway. Her younger sibling spoke hardly any Turkish but only understood some. Therefore, the siblings spoke Swedish with each other, especially when the parents were out of earshot. In the longitudinal study, none of the ten children spoke only Turkish with their siblings, though half were reported to use ‘mostly Turkish’. Already at age four, about half the children had been interacting with their siblings equally in Turkish and Swedish and the same was reported in the questionnaire two years later when they were six years old (Öztekin 2019: 50). The interviews revealed however that the children in three households now spoke more Swedish compared to before, and in one family the child’s language use with siblings and parents had shifted towards more Swedish after a third baby was born. These observations are important clues to understand language shift in daily life.

We also asked about additional Turkish input providers such as grandparents and other extended family members. Grandparents have been characterised as gatekeepers, supporting the maintenance of the heritage language and culture. In the large-scale survey, the large majority of children (87 %) were reported to hear Turkish from extended family. The remaining children did not, and interestingly, they mostly had parents whose L1 was not (exclusively) Turkish and/or who had not grown up in Turkey but in Sweden or in a third country. It is not surprising then that these parents’ parents did not speak Turkish to the child.

In the longitudinal follow-up, we wanted to explore the issue of grandparents in greater detail. During the interviews, it emerged that all ten children spoke Turkish with their grandparents (this had not been explicitly queried in the questionnaire).^[17] It is well known that communicating with beloved relatives or friends who do not speak the majority language (Swedish) can be a powerful motivation for the continued use of the minority language. In some families, grandparents were involved in taking care of the children, and this required intergenerational communication in Turkish, as the grandparents were not always proficient in Swedish. Some of the children also had other Turkish-speaking relatives in Sweden, though their language use with these relatives in Sweden varied (Öztekin 2019: 51). For example, although one child spoke mainly Turkish with his relatives, for those who were more fluent in Swedish, he switched to Swedish. Two children had Kurdish-speaking relatives as well. These children understood some Kurdish but could not speak any, according to their parents. The child’s language use with the extended family showed that the child may adapt to the language of the interlocutors in ways that differ from their language choice with their parents or siblings.

4.3 Language-related activities in the home

Joint activities such as singing, storytelling and shared book reading in a particular language may also be reflections of family language policy. Apart from boosting socioemotional connections, it is well known that such activities influence the cognitive and linguistic development of the child and spur the child’s interest in language and literacy. In the questionnaire, parents filled in how often they did certain activities together with their child, and in which language (never, twice a month, 1−2 times per week, nearly every day).

In the large-scale survey, for 64 % of the children, parents reported that they regularly sang or listened to Turkish songs with their child (more frequently than in Swedish), and for 59 % that they regularly told stories to their child in Turkish (again more frequently than in Swedish). The remaining parents said that they rarely or never carried out such activities, or left the answers blank. Most children (77 %) were reported to watch Turkish television and films and/or play computer games regularly (i.e. several times a week) with their parents, slightly more so than for Swedish. We could not detect any generational tendencies here.

The frequency of shared book reading varied considerably between families (for details, see Bohnacker 2022). 66 % parents answered that they looked at picture books together with the child or read for the child in Turkish nearly every day or once or twice a week. Book reading in Swedish was also common. When little or no literacy-related language activities with the child were reported (“never” or “twice a month”), both parents disproportionally often had very low levels of education (up to primary school). Sweden-born bilingual Turkish/Swedish parents reported throughout that they read Turkish books with their (G3) child nearly every day. Interestingly, in these families parent-child and child-sibling interactions were predominantly in Swedish. The extra book reading activities in Turkish may thus have been a deliberate attempt by the parents to counteract the effects of the ‘onslaught’ of Swedish elsewhere.

In the longitudinal follow-up, we also asked about activities that the parents did with their child. As in the large-scale survey, watching television and films together was frequent in both Turkish and Swedish, though some families reported that they nowadays watched more Swedish television and cartoons than two years earlier. Joint book reading in Swedish occurred in all ten households to some extent. Joint book reading in Turkish was less common; some families did it nearly every day and others never or only rarely. Interestingly, the families had not changed their reading habits from when the children were four years. The frequency of singing and listening to songs as well as storytelling in Turkish and Swedish also varied widely across families, but again remained remarkably stable within a family over time.

Concerning joint book reading, reminiscent of the larger sample, parents with very low education tended not to read to their children, especially not in Turkish. This issue was explored further in the interviews, where we also asked about visits to the library. Public libraries in Sweden are free of charge and have branches all across the city; they are usually well-stocked with children’s books, including selections of minority-language books. Three families, all with very low education (primary school), reported that they never went to the library with their child. These parents used audio stories on the smartphone for their child to listen to as bedtime stories. One Kurdish/Turkish family with low parental education did regularly go to the library to borrow Swedish children’s books and read to their child. All other families (with at least one parent of mid-to-high education) regularly went to the library with their child to borrow Swedish books, and some also Turkish ones. In addition, these families also brought Turkish books from Turkey to read with the child. These are conscious efforts to support the minority language.

Whilst Swedish libraries provide access to high-quality children’s books independent of family income, particular books of choice might not be accessible via the library, for instance books in Turkish. These will need to be purchased from Turkey, and low-income households may not always have the resources to do so. In fact, only two families said that they owned “many” Turkish children’s books; interestingly, in both of these households, one parent was a teacher. Apart from cultural capital and financial resources, there are also constraints of time. Some parents worked very long hours, leaving them little time and energy to spend together with their child on joint activities such as book reading.

4.4 Heritage language education

Another conscious management effort to maintain and develop the heritage language is sending one’s child to mother-tongue tuition (MTT). In the larger sample, 54 % of the children attended Turkish MTT, typically for 40−60 min per week. The oldest children attended MTT much more often (age seven: 82 %) than the younger children (26 %–59 %). Non-attendees had not necessarily opted against home language education; some municipalities simply did not offer Turkish classes, or none for preschoolers.^[18]

Children are only entitled to municipal MTT in one language. Thus, when a trilingual family (e.g. Kurdish/Turkish/Swedish) opts for Turkish, this means opting out of MTT in another language. Here it is interesting to note that children from Kurdish/Turkish homes attended Turkish MTT as frequently as children from Turkish-only homes.

In the longitudinal follow-up, most parents had enrolled their child in municipal MTT classes (7/10), ca 1 h/week. At the individual level, MTT attendance had changed compared to two years earlier. Some children who did not attend at age four were now enrolled at age six, whilst others no longer attended. Some parents explained that this was because of problems with the practical setup of MTT, e.g. involving travel to a venue across town, inconvenient scheduling after school hours, and/or a mix of proficiency levels in one class. They also critiqued the lack of MTT offerings, especially for preschoolers. Several parents pointed out that the municipality had recently abolished mother-tongue support. One (high-SES) family had hired a private tutor instead.

4.5 Preschool language practices

In the large sample, most children had started preschool early, at 23 months on average, and were exposed to Swedish there. Children attended preschool for 32 h per week on average (range: 6−48 h). We could not discern any generational differences here. School-age children had previously been to preschool. As preschool is not compulsory in Sweden, enrolling one’s child could be considered a conscious parental language planning effort towards more Swedish. However, most parents cannot afford to stay at home with their child beyond the guaranteed parental leave, even if they wanted to, but must seek paid employment. Early and extensive preschool attendance thus need not signal parental “affirmative action” for Swedish, but could also be due to social and economic pressures.

In the longitudinal follow-up, all ten children had attended preschool from an early age (from 12−27 months, 26−40 h/week). At six years, most (8/10) attended Grade 0 of primary school, one child attended Grade 1, and one was still at preschool. For most children, estimated daily exposure to the two languages had changed over time. Whilst at age four many children were reported to have relatively even exposure (40:60, 50:50, 60:40) to Turkish and Swedish during the day, by age six most were reported to be exposed to 80 % Swedish, probably as a result of schooling. For the child with a Swedish-L1 parent, daily exposure to Swedish had even increased from 80 % to 95 %.

Nearly all schools were located in neighborhoods with children from many different language backgrounds. In some schools, a large proportion of the intake appeared to be children of migrants with varying levels of proficiency in Swedish. Only one child (the one with an L1-Swedish parent) attended school in a largely monolingual Swedish setting. Two attended a school with a bilingual English/Swedish profile.

Whilst the main language of communication inside all schools was Swedish, we observed that (pre)school staff did not always have a very high proficiency in Swedish. Only for three children were the teachers and other staff predominantly native-like speakers of Swedish. We also observed that some staff spoke Turkish or other languages to their colleagues during school hours and to the parents, e.g. when children were collected from school. These observations suggest that multilingual practices in school are commonplace. The parental interviews revealed that at their child’s earlier preschool, all ten children had been able to speak Turkish with at least one staff member, as well as with some schoolmates, something that had not been evident from the questionnaires. Some parents confirmed that at their child’s new school, there was a Turkish-speaking staff member who spoke Swedish and Turkish with their child. Several children also had Turkish classmates. However, four children no longer met any Turkish-speaking staff, and one family reported that whilst there were two Turkish-speaking teachers, “it is not allowed to speak Turkish at school”.

In sum, the use of the minority languages, including Turkish, appears to be encouraged or at least tolerated on (pre)school premises. In one school however, a language policy is implemented that discourages the use of Turkish. These language practices at school may influence family language policy.

4.6 Parental language beliefs and attitudes

In the large-scale questionnaire study (Bohnacker 2022), parents were asked which language(s) they considered to be the most important for their child to become proficient in: Swedish, Turkish, both Turkish and Swedish, or Other. Parents had clear opinions on this and hardly ever left this question blank (2 %). Sometimes spouses disagreed with each other on this question. A large majority of parents (72 %) considered Swedish and Turkish to be equally important. A further 8 % considered Swedish, Turkish and a third language equally important. The third language was usually Kurdish or another home language. A small but considerable group (14 %) considered Turkish more important than any other language. Two parents regarded Kurdish or both Kurdish and Turkish as more important than Swedish. Only 3 % ranked Swedish highest. We could not detect any generational pattern behind this; first-generation and second-generation parents’ attitudes appear to be similar here. However, trilingualism and possibly ethnic affiliation may play a role: Of those who ranked Swedish highest, all reported a language other than Turkish as their native language. Of those who ranked Turkish highest, all were L1-speakers of Turkish.

In the longitudinal sample, all parents considered Turkish and Swedish to be equally important for their child to become proficient in. When their child was four years, one parent had additionally mentioned Kurdish as being equally important, and another parent emphasised English (not spoken in the family but useful as a global lingua franca). At six years, two parents mentioned Kurdish. In general, Turkish was valued very highly, alongside the majority language.

Several parents said they had established strategies to maintain and develop the heritage language in their children. Most communicated with their spouse in Turkish. Most also tried to communicate with their children as much as possible in Turkish. Whilst this may have come “natural” to Turkey-born L1-Turkish immigrant parents, parents who have grown up bilingual themselves, deliberately make the choice of transmitting the heritage language.

Most had close contacts with relatives and heritage-language speaking friends. Most sent their children to MTT, and one had previously employed a Turkish-speaking childminder (dagmamma), also to boost exposure to Turkish.

Some parents were very conscious about language management, and a few were working extra hard to provide a richer language-learning environment for their children by organising extracurricular activities in Turkish and visits to Turkey. These parents believed that such outside activities would make the children more interested in the language and confident in using Turkish at home. For instance, one highly-educated mother organised Turkish reading hours at the library with other families. Interestingly, this was the mother in whose home the least Turkish was spoken, as her spouse was a native speaker of Swedish. This family reported that they now were making extra efforts to read and speak more Turkish at home.

In the interviews we asked the parents an open-ended question why they considered the heritage language so important (this had not been queried in the questionnaire). Interestingly, parents never spontaneously mentioned that language was a way of transmitting cultural traditions and values and strengthening ethno-cultural identity. Only one parent stressed that the language was important for keeping in touch with relatives. Instead, several parents, especially highly educated ones, explained that it was “good to be a bilingual”, that the children “should be world citizens”. Most striking however was the wide-spread belief that “if my child learns to speak the mother tongue well, s/he can learn other languages better”. This comment was made by parents of very different backgrounds, echoing the rationale behind official Swedish language policy.

4.7 Language anxiety and seeking advice

In the large-scale survey, parents were asked whether they had ever felt any anxiety about their child’s language development and why, and whether they had sought the help of professionals, such as educators or speech-language therapists, in this matter. Most families (81 %) said they had not been anxious, but 18 parents had felt some anxiety in the past, mostly because of late onset of speech or pronunciation difficulties. Seven of these 18 parents had been anxious because of the child’s bilingualism. Some worried about how the child would be coping with Swedish in (pre)school, as the family mainly spoke Turkish at home, whilst one parent feared that the child’s Turkish language was stagnating. In 13 cases, the parents had taken their child to a Swedish speech-language clinic. The language therapists there were reported to have judged the child’s language development as normal, thereby allaying the parents’ fears.

In the longitudinal study, when directly asked, only one of the ten families expressed any anxiety about their four-year-old child’s language development. Two years later, three parents reported that they had been concerned, one for the child being a late talker and ‘mixing’ languages, one parent for the child’s Swedish, and one for the child’s Turkish being patchy. In the interviews, we also asked whether the parents were satisfied with their child’s language use in the family and proficiency in Turkish, and two thirds of them were. The others were not quite satisfied, saying that their child had too many hesitations and word-finding difficulties in Turkish, was shifting towards Swedish, or was not learning ‘proper’ Turkish. We could not detect any generational or L1-related patterns behind these responses.

In the follow-up interviews, we asked whether the parents had ever sought advice on bringing up their child bilingually. Nearly all families (8/10) had done so and been informed by Swedish professionals that their child should learn to speak the mother tongue (Turkish) really well, as this would help the child to learn other languages as well, including Swedish. School teachers were explicitly supportive of the child’s Turkish development and advised the parents to provide as much input as possible in Turkish. More than half of the parents mentioned that they had consulted teachers, other educational experts, speech-language therapists and/or books in order to understand bilingual language development. They had been encouraged to speak Turkish with the child and maximise exposure to Turkish in the home (“mother tongue first!”, “teach your child Turkish really well!”). No parent reported that they had been told to use Swedish with the child.^[19]