Introduction

Mozambique is a multilingual and multicultural society where, in addition to Portuguese, the official language, over twenty-four Bantu languages are spoken. In the last 47 years, the proportion of Mozambicans with a Bantu language as first language (L1) has been decreasing, while the proportion of speakers of Portuguese as L1 or second language (L2) has increased considerably.

In view of the attested loss of Bantu language speakers, this study seeks to understand the role of Family Language Policy (FLP) in the process of language shift to Portuguese. FLP is here understood “as explicit and overt planning in relation to language use within the home among family members” (King et al., 2008, p. 907). The study follows a critical, multi-layered approach that recognises the national, institutional and interpersonal layers of language policy work, where FLP is located. Within this framework, it is assumed that multiple language policy actors and agents engage in a variety of processes at different but interlinked layers that shape actual language policy activities (Hornberger & Johnson, 2007). Assuming, in line with Gomes (2018, p. 61), that “language practices and social reality are dialectically and recursively entangled”, this option entails exploration of alternative approaches to FLP, ones that, among other things, question Fishman’s (1991, 2006) view of family and intergenerational transmission as central for language maintenance, and move beyond Spolsky’s (2004, 2012) tripartite model of language policy as applied to FLP. Research based on a critical, multi-layered view of language policy falls within the ecology of language paradigm, which is an appropriate approach to the study of FLP, as family language policies and practices are shaped by, and also shape community, institutional and societal decisions. This is particularly true in postcolonial contexts such as Mozambique, where local languages co-occur in diglossic relationships with powerful former colonial languages, often leading parents to choose these dominant languages for primary socialisation and formal education of their children. We thus consider that language shift in postcolonial Mozambique is best understood when individual and family decisions on language use are appreciated against socio-historical, ideological and political forces, unveiling “the conflicts that family members must negotiate between the realities of social pressure, political impositions, and public education demands on the one hand, and the desire for cultural loyalty and linguistic continuity on the other” (Curdt-Christiansen, 2013, p. 1).

Consistent with the ecological approach to language policy, this study seeks to understand how bilingual parents’ language policies and practices on bi/multilingualism are shaped by their personal biographies and social trajectories, as well as by historical and contemporary ideologies and discourses on multilingualism and social order in Mozambique. The study found that FLP decisions are mainly driven by a combination of instrumental, sociodemographic and politico-ideological factors, often in conflict with local sociocultural desires.

This article may contribute to FLP research by foregrounding the use of critical and multi-layered lenses to expand our understanding of multidimensional the dilemmas and challenges faced by families in the management of bi/multilingualism in postcolonial contexts, which, in many cases, lead to language shift to ex-colonial languages. The study may also have implications for language policy making and implementation, in particular for language-in-education policy, reinforcing the call for multi-layered efforts to secure bi/multilingual conviviality and maintenance.

FLP and language shift

Recent studies on FLP have focused on how language use is explicitly or deliberatly managed within the home among family members and how that micro level family management influences and is influenced by macro level socio-historical, ideological and political processes (e.g. Canagarajah, 2008; Curdt-Christiansen, 2013; Duff, 2020; King & Fogle, 2013; Spolsky, 2004, 2012). These interrelated concerns substantiate the importance of adopting critical, multi-layered approaches to FLP, which allow not only to understand how the family plays a key role in processes of language maintenance and shift, but also how the family negotiates its language maintenance responsibilities with macro level pressures (Canagarajah, 2008; Curdt-Christiansen, 2013; Duff, 2020). In fact, a plethora of studies have shown that “the forces underlying language shift or language maintenance in families have largely been located in the macro level” (Higgins, 2018, p. 311).

Language shift in Africa

Language shift refers to the process whereby a community gradually abandons its original language in favour of another, often a high status language of a socially or economically dominant community (Fishman, 1964). According to Batibo (2005, p. 89), the process is initiated when a speech community ceases to transmit its language to its descendants as a result of the “weaker position of the speakers socio-economically or demographically”.

Studies on language shift have tended to focus mainly on minority language immigrant communities (e.g., Canagarajah, 2008; Fishman, 1991; Gal, 1979; Stoessel, 2002) and little is known about language shift in Africa (e.g. Batibo, 2005; Bodomo et al., 2009; Coleman, 2011; Igboanusi, 2009; Letsholo, 2009), despite the fact that this continent has a highly complex linguistic situation. In order to account for these differences, Bodomo et al. (2009) propose the concept of ‘multilingual language shift’, which covers different types of language shift in highly multilingual contexts in the new nations of Africa (p. 358). Referring specifically to West Africa, Igboanusi (2009, p. 300) distinguishes between three major patterns of language shift: (1) shift from indigenous languages to the ex-colonial languages, namely English, French and Portuguese; (2) shift from minority languages to the dominant West African languages; and (3) shift from minoritised languages to regional linguae francae. Batibo (2005, p. 98) stresses the importance of this latter pattern, pointing out that, in many African communities, the ex-colonial language is restricted to a small proportion of the elite and “the only prestigious language would therefore be the prominent lingua franca, L2.”

Although Southern Africa is also a multilingual region, there are some differences regarding the scenarios in which language shift can occur. For example, in Angola and Mozambique, Igboanusi’s (2009) pattern (3) is not common, and, in the absence of national linguae francae, pattern (1) is clearly dominant, with shift from several indigenous Bantu languages to Portuguese.

Scholars have identified several factors that interact in complex ways to cause language shift. For Batibo (2005, p. 93), “the main cause of language endangerment—and by implication language shift and death—is the pressure that the weaker languages experience from more powerful or prestigious languages. This pressure may be caused by demographic superiority, socio-economic attractions, political predominance or cultural forces.”

As far as African countries are concerned, among these general factors, it is worth highlighting some specific aspects of the process of language shift. With regard to political factors, language policy and language-in-education policy are particularly relevant. For instance, in Portuguese speaking countries such as Mozambique and Angola, the privileged position of this language, associated with the tendency to exclude African languages from formal domains, has contributed greatly to the process of language shift to Portuguese (Chimbutan, 2018; Inverno, 2018). Attitudinal factors (e.g. negative perceptions in relation to the low status of many African languages and associated cultures) have also favoured the shift to ex-colonial languages. As to demographic factors, linguistically mixed marriages and internal migration due to urbanisation or military pressure as a result of wars can also play a significant role in language shift.

The range of factors presented above shows how macro-level forces influence language shift and condition language policies and practices within families. Within this setup, families are often trapped between the maintenance of their ancestral languages and cultures, on the one hand, and the fulfilment of educational, socioeconomic and political desires and priorities, on the other hand (cf. Canagarajah, 2008; Curdt-Christiansen, 2013; Mufwene, 2004).

A critical, multi-layered approach to FLP

The recurring evidence of the power of macro level factors to influence language shift led scholars to problematise Fishman’s (1991, 2006) emphasis on the centrality of the domain of family and intergenerational transmission for language maintenance (e.g. Canagarajah, 2008; Spolsky, 2004). While acknowledging the importance of family efforts contributing to language survival, including through intergenerational transmission, scholars have claimed that the processes involved are better understood when those micro level efforts are embedded within macro level realities (e.g. Canagarajah, 2008; Duff, 2020; King & Fogle, 2013; Spolsky, 2004, 2012). As Canagarajah puts this:

“…the family is not a self-contained institution that can adopt its own strategies and devices for language transmission, as Fishman theorizes. It has to negotiate its linguistic responsibilities with other social and economic pressures (that is, historical, social, and institutional forces)” (Canagarajah, 2008, pp. 170–171).

This multilevel analysis of language processes is the core of critical, multi-layered approaches to language policy. In a broad sense, these approaches revolve around the assumption that language policy activity occurs on different layers (Hornberger & Johnson, 2007; Ricento & Hornberger, 1996). It is from this multi-layered approach that Ricento & Hornberger (1996, p 408) advanced their metaphorical representation of research into language policy as the unpeeling of an onion and as a lens to visualise a “schema of agents, levels, and processes […] that together make up the LPP [Language Planning and Policy] whole”. Within this framework, it is assumed that multiple language policy actors and agents engage in a variety of processes at different but interlinked layers that shape actual language policy activities (Hornberger & Johnson, 2007). This framework recognises the national, institutional and interpersonal layers of language policy work, where FLP is located.

Research based on the multi-layered view of language policy falls within the ecology of language paradigm, which has come to be widely regarded as a useful approach for critically addressing current multilingual policy demands (Hornberger, 2006; Ricento, 2000). Within this paradigm, it is assumed that languages, in multilingual contexts, come in contact with each other and with the social, cultural, economic and political environments in which they are used (Hornberger, 2002; Kaplan & Baldauf, 1997). As a consequence, “language planning activity cannot be limited to one language in isolation from all the other languages in the environment” (Kaplan & Baldauf, 1997, p. 271) and must take into account a wider range of contextual factors implicated in the status and use of those languages.

This study is located within the critical, multi-layered approach, which we find appropriate to research FLP in Mozambique. As a postcolonial multilingual context, where local Bantu languages co-occur in diglossic relationships with Portuguese, this country emerges as a site of tensions where competing factors influence parents’ decisions about the choice of language(s) for primary socialisation and formal education of their children.

Research context

Colonial and postcolonial language policies

In essence, the same language ideologies and policies that prevailed during colonial rule have been maintained in postcolonial Mozambique.

In most of its colonial presence in Mozambique, Portugal adopted overt and de jure assimilationist ideologies and policies. The Portuguese assumed that one of their missions was to ‘civilise’ the natives by spreading the Portuguese language and Portuguese/European cultural values (e.g. Meneses, 2018).

Consistent with this colonial assimilationist ideology, Portuguese was imposed as the language of formal functions, including formal education. In contrast, the use of Bantu languages in formal domains was in general forbidden, allegedly to ensure that the ‘indigenous’ population acquired the Portuguese language and associated cultural values properly (Chilundo et al., 1999). One of the consequences of this language ideology was the stigmatisation of Bantu languages and cultures, even by their own speakers, which remained “restricted to the informal, home domains and to ideas of tradition and the local” (Stroud, 2007, p. 30), and strategically used for evangelisation purposes. This perception reflects the colonially manufactured diglossic distribution of languages in Mozambique, in which Portuguese has been constructed as the language for official and specialised functions, and Bantu languages as languages for informal functions and ethnic attachment (Chimbutan, 2011, 2021; Stroud, 2007). In line with the assimilationist ideology, Portuguese language and associated culture are perceived as superior and as gateways for socioeconomic mobility.

The language ideologies and policies adopted in the colonial era continue to influence the policies and practices currently pursued in Mozambique. As a matter of fact, faced with the challenge of managing the multi-ethnic and multilingual situation of Mozambique, at independence, the FrelimoFootnote 1 government opted for retaining Portuguese as the official language of the country. In contrast, no official status was granted to Bantu languages, which continued to be confined to informal domains and functions.

Following the language ideology adopted during the liberation struggle, Frelimo perceived Portuguese as the neutral language of integration and modernisation. This ideological perspective was epitomised by the declaration of Portuguese as the language of “national unity”. Within this ideological setup, multilingualism was conceptualised as the seed source of tribalism and regionalism, which should be combated vigorously.

In line with this monoglotic language ideology, Portuguese was defined as the sole language of formal education (Firmino, 1998; Katupha, 1994). Accordingly, the same policy proscribed the use of Bantu languages in schools and students caught breaking this policy could be punished. Pedagogically, the use of Bantu languages was perceived to be detrimental to the students’ acquisition of the Portuguese language.

The privileged status and symbolism ascribed to Portuguese through the State language ideology and language-in-education policy drove and continues to drive many parents, in particular urban and middle class parents, to socialise and formally educate their children exclusively in Portuguese, thus paving the way to language shift.

However, a socio-political shift from a State-centered socialist orientation to a democratic one in the early 1990s opened spaces for some form of multilingual ideologies and discourses. While Portuguese maintained its dominant status as the sole official language, for the first time in Mozambican history, it is enshrined in the revised 1990 Constitution that the State promotes the development and increased use of Bantu languages in public life, including in formal education (cf. República de Moçambique, 1990, Article 5). This shift shows that, at least in terms of legal provisions and political discourses, Mozambique began to witness some form of embracement of multilingualism. Stroud (2007, p. 42) uses the term “re-traditionalization in the modernization of Mozambique” to refer to the State’s embrace of local languages and associated cultures, a process that has, however, meant “the return or invigoration of ‘traditional/colonial’ categories of diversity, which include the oppositions tribal-ethnic-nationalist” (Chimbutan, 2018b, p. 16). The introduction of bilingual education in selected primary schools in 2003 after pilot initiatives in late 1990’s (cf. Chimbutan, 2011; Benson, 2000), and the legal use of Bantu languages in other formal arenas, including in local government institutions, illustrate the exploration of the current spaces for multilingualism and multiculturalism in Mozambique. However, language practices and attitudes point to the prevalence of the monolingual bias, with the Portuguese language retaining its powerful status, including in bilingual education (e.g. Chimbutan, 2021).

Sociolinguistic profile of mozambique

As mentioned so far, there are over twenty Bantu languages spoken in the country, in addition to Portuguese, and also a few foreign languages (FL), which include English, Asian languages and other African languages.

Data from the 2017 national population census indicate that Bantu languages are spoken by nearly 82% of the population (about 80% as L1) whereas Portuguese is spoken by 58.1% of the citizens (Instituto Nacional de Estatística [INE], 2019). A comparative analysis of the results of the censuses carried out after independence (1980, 1997, 2007 and 2017) shows that the proportion of Mozambicans with a Bantu language as L1 has decreased by 19.6% points between 1980 and 2017 (Figure 1).

Figure 1
figure 1

Distribution of speakers of Bantu languages as L1 in 1980, 1997, 2007 and 2017 (Chimbutan, 2012; Firmino, 2000; INE, 2019)

Full size image

Although the pattern captured in Figure 1 applies to all 11 provinces of the country, the level of loss of speakers of Bantu languages as L1 between 1997 and 2017 varied considerably among the provinces: In 2017 the proportion of lost speakers of Bantu languages as L1 ranged from 38 to 93%. With the exception of Maputo city and Maputo province, which had a loss of speakers of Bantu languages accounting for 37 and 24 percentage points respectively, in the other provinces the loss was between 2 and 13 percentage points.

In contrast with Bantu languages, the proportion of speakers of Portuguese has increased considerably after independence (Figure 2). While in 1980, 5 years after independence, only 24.4% of the population could speak Portuguese as L1 or L2, this proportion rose to 58.1% in 2017. A comparative analysis of the results of the last four censuses shows that the proportion of Mozambicans with Portuguese as L1 or L2 has increased by about 34 percentage points between 1980 and 2017.

Figure 2
figure 2

Distribution of speakers of Portuguese as L1 and L2 in 1980, 1997, 2007 and 2017 (Chimbutan, 2012; Firmino, 2000; INE, 2019)

Full size image

With the exception of Maputo city and Maputo province, which had an increase of about 37 percentage points of speakers of Portuguese as L1, in the other provinces the proportion of this category of speakers only had an increased between 3.2 and 12.8 percentage points between 1997 and 2017.

The increase in the proportion of speakers of Portuguese in Mozambique is closely related to political and ideological factors, in particular the privileged status and symbolism ascribed to this language. It is also associated with sociodemographic and socioeconomic factors, such as migrations from rural to urban areas, and a substantial increase in access to formal education, primarily conducted in Portuguese, with the consequent expanded use of the language, including in informal domains. The advancement of the Portuguese language and decline of Bantu languages have led scholars to conclude that a process of language shift from Bantu languages into Portuguese is going on in Mozambique in particular, in urban areas (Firmino, 2002).

Participants and procedures

This study builds on results of the first research project on the behaviour of Mozambican bilingual population, L1 Bantu—L2 Portuguese, carried out in 2015–2019.Footnote 2 The project aimed to survey two generations of three cities of the country (Maputo city, Xai-Xai and Quelimane): Young university students born after the independence of the country (1975) and middle class parents of this generation. For this article, we partially explore the results of the survey to the parents.

The option for middle class participants was based on the assumption that, in general, this is the social class in which the decision on the language(s) to be transmitted to the next generation is a real dilemma. The same dilemma is not often faced in upper class families where most parents opt for Portuguese transmission, often without problematizing this option, or in lower class families, especially those in rural areas, who tend to conduct their lives almost exclusively in Bantu languages. In other words, it is likely that it is in middle class families that the factors constraining parents’ decisions on the choice of language(s) for primary socialisation of their children most stand out.

The survey to the parents’ generation was based on semi-structured focus group interviews, a ‘sensitive method’ that allows to explore and understand the complexities of the processes taking place within the family (Schwartz, 2010, p. 185).

The interviews were carried out in 2016, involving groups of 4–8 participants, at different sites: primary schools, universities, and churches. The duration of interviews ranged from 60 to 120 min. These were conducted in Portuguese and/or in the participants’ L1/Bantu languages. In fact, most of the discussions occurred bilingually, with participants and moderators switching from Portuguese to Bantu languages (mainly, Changana, Ronga and Chuabo) and vice-versa. All interviews were audiotaped for transcription. Ordinary writing conventions in Portuguese and in Bantu languages were used for the transcriptions. The interviews or parts of interviews in Bantu languages were translated into Portuguese.

The interviews were conducted in a cordial atmosphere, to allow participants to express their opinions as openly as possible. The moderators had a list of open-ended questions on different topics. In order to specifically understand the process of language shift to Portuguese, the participants were invited to talk about their choices of languages for primary socialisation and formal education of the children, as well as about the motivations (social, cultural, economic, political, etc.) behind those choices.

Thematic analysis was used to approach interview data. Thematic patterns were identified based on sensitising categories established in previous studies on FLP, and also on the objectives set for this study. Related to the focus of this article, among others, we identified three major themes: factors for the choice of Portuguese as the primary language of socialisation and formal education; implications of this choice at individual, family, community and societal levels; and strategies proposed by the participants to mitigate language shift into Portuguese. As part of the method, the thematic patterns captured were systematically revised throughout the analytical process. The insider status of the researchers in the Mozambican community has contributed to interpreting the data with more insight.

This article focuses on those parents who chose to socialise their children in Portuguese. A more extended analysis covering different patterns of language socialisation captured in the wider study can be found in Chimbutan et al. (2022), within the framework of new materialism.Footnote 3 Out of the 31 participants, with ages ranging from 46 to 82, 14 were female, and 17 were male. Most of them worked in the area of education (about 60%) or public administration (about 20%). Other professions (about 20%) included a variety of occupations (agricultural technician, nurse, engineer, housewife, head of residential block).

Most parents (68%) had Bantu languages as their L1, about 20% were L1 speakers of Portuguese and a small minority (12%) were simultaneous bilinguals Bantu language/Portuguese. The most spoken Bantu languages as L1 were Changana, Chuabo and Ronga. Other Bantu languages included Chope, Tonga, and Muninga. While these parents opted for socialising their children in Portuguese, considerable percentages of them declared to speak either in a Bantu language and Portuguese (about 40%) or in a Bantu language (about 32%) with their partners; and just about 25% used Portuguese only in the same interactional context. This means that in many families the children were in some way also exposed to Bantu languages.

Family language policy and language shift in Mozambique

Ecological factors shaping portuguese-only family language policy

The language choice pattern that leads to language shift to Portuguese by urban and middle class bilingual families is anchored in monolingual colonial and postcolonial language ideologies and policies, that have contributed to construct positive attitudes towards this language and its associated cultural values, to the detriment of Bantu languages, which have remained linked to informal functions.

The answers of the participants about the reasons for the choice of Portuguese as the language of primary socialisation and formal education of their children clearly show that this option is primarily driven by instrumental (43.5%), sociodemographic (36%) and politico-ideological (20.5%) factors. Instrumental factors are associated with the parents’ desire to ensure the academic success of their children and, ultimately, their future access to social and economic opportunities, including better opportunities to obtain well payed jobs. Sociodemographic factors have to do with intermarriages among speakers of different languages, and residence in multilingual urban areas or rural-to-urban migration. Politico-ideological factors are linked with assimilationist language policies and practices adopted in colonial and postcolonial Mozambique, which have privileged Portuguese to the expense of Bantu languages. While instrumental factors reoccur more frequently in the participants’ accounts, the three sets of factors are intertwined.

Regarding instrumental factors, and more particularly the parents’ desire to ensure the school success of their children, see the following extract:

When I was born… the first language I spoke was Chuabo, a dialect. Only seldom, at school, you know… I spoke Portuguese… Well, just to say things such as ‘this is a pen’. So, sometimes I would bang my head asking myself: “But what is a pen?!” This is because, it was Chuabo that I learnt when I was an infant. When I got to grade 5, I could not yet speak Portuguese well, because we spoke Chuabo at home. Therefore, when I had my children, for them not to go through the same problems, I started to communicate with them in Portuguese and up to now. (39MJ/F/Quelimane/2016)

Mrs. MJ argues that she had opted for socialising her children in Portuguese to avoid that they went through the same communication and learning challenges she faced in school. This rationale was pervasive in the study, as also expressed in Mr. BP’s account: “…we looked forward (…) to see whether they ran faster than we did, you know! We faced difficulties for using two languages. So, we tried a short cut…” (05BP/M/Maputo/2016). This suggests that, by opting to transmit Portuguese to their children as early as possible, these parents were striving to provide ‘good parenting’, drawing on their own struggling language learning and schooling experiences (King & Fogle, 2006).

According to other accounts, the instrumental choice of Portuguese was based on popular beliefs according to which the earlier a child learns an L2, without interference of its native language, the better.

In the teacher training course, in terms of methodologies, the demand was for us to teach the children but using the Portuguese language. (…) When I had my first children, obviously, I started to use Portuguese expressions in their education. That was to avoid that they had difficulties in speaking Portuguese when one day they progressed to another level. (43RP/M/Quelimane/2016)

This extract shows how Mr. RP’s condition as a teacher influenced his preference for transmitting Portuguese to his children, thus preventing them from facing difficulties in speaking this language in school and beyond. The account subsumes the language-in-education policy that prescribes Portuguese as the language of teaching and learning, as well as mirrors the teaching and learning theory of maximum exposure to the target language. In fact, during his pre-service teacher training, Mr. RP learned that teachers should “…teach the children but using the Portuguese language”, meaning that they should teach in Portuguese and use this language maximally.

The option for Portuguese is also associated with the participants’ perception that this language and the related cultural capital acquired through schooling are gateways for upward socioeconomic mobility and success.

Portuguese is a language that I like so much that I sent my children to school to learn it. (…) Why I did that? This is because our native language does not harness graduate or doctorate degrees. It is just a language that is part of our tradition and one that they should not forget. In contrast, with the Portuguese language, they can get jobs, progress in their professional carriers, earn good salaries, and take care of me, as I am old. (…) The Portuguese language can help them in their future. (24MM/M/Xai-Xai/2016)

Although perceiving his native language as a symbol of tradition and ethnolinguistic identity, Mr. MM, like other parents, opted for the socialisation of his children in Portuguese in order to ensure that they have access to all high domains, thus guaranteeing their socioeconomic success in the future. As Mr. BP put it: “We looked forward.” (05BP/M/Maputo/2016). Chimbutan et al. (2022) highlight this instrumental association between education outcome and socioeconomic rewards as a key factor for the choice of Portuguese as the language of primary socialisation and schooling in Mozambique. This pattern of language choice and the motivations behind it are consistent with those captured in other post-colonial multilingual contexts, as documented, for example, in Coleman’s (2011) edited volume.

The choice of Portuguese has also been influenced by sociodemographic factors such as mixed marriages involving couples with different Bantu L1, as well as by migration to urban settings. In a context of great population mobility and urbanisation, interethnic marriages are becoming quite common in Mozambique. As a result, the choice of Portuguese as the family language can be regarded as a coping mechanism adopted by intermarried families to deal with the ethnolinguistic diversity and with the urbanised environments where they often reside. The following extract illustrates the operation of this factor:

Well, we speak Portuguese at home because… given the coexistence of two dialects, two mother tongues. I am Chuabo and she is Lomwe. So, we did not know which language we should transmit to our children. In any case, the first language they acquired was Portuguese. (52FF/M/Quelimane/2016)

As can be understood, pragmatic reasons led Mr. FF’s family to opt for Portuguese to the detriment of the couple’s native languages, Chuabo and Lomwe, which he refers to as ‘dialects’. In families with mixed couples such as this, Portuguese is perceived as the unifying language, as also clearly expressed by Mr. MA, a native speaker of Lomwe, who was married to a native speaker of Sena: “The Portuguese language is what unites us in the home” (49MA/M/Quelimane/2016). This perception echoes, at least in part, the macro-level politico-ideological construct of Portuguese as the language of national unity, as also stressed by participants in the study:

In the past, the Changana used to marry among themselves, the Chope married among themselves and so on. However, today we marry to anyone we love. We are not in love with the language, but with the person. (05BP/M/Maputo/2016).

Population mobility into multilingual urban areas is another sociodemographic factor influencing families’ choice of Portuguese, as illustrated in this extract.

My first employment was in Inhassunge, my home village. It was also there where I got married. When I got married… the first language that my elder children acquired was Chuabo (…) Here in Quelimane I ended up having one child only, who doesn’t know any Chuabo. He only knows Portuguese. (46LA/M/Quelimane/2016)

This account shows how the move to an urban setting led Mr LA’s family to shift their language transmission pattern—from Chuabo into Portuguese. Different from the somehow ethnolinguistically homogeneous home village, the heterogeneous urban context is an important driver of this shift in families arriving or living in urban areas. The urbanization effect on language shift in Mozambique is more apparent in Maputo City and Maputo Province, the most urbanised settings in Mozambique, where in the 2017 national census about 49.5% and 62.2% of the inhabitantes, respectively, declared Portuguese as their L1 (Chimbutan et al., ftc.). The data seem to substantiate Firmino’s (2002) conclusion that a language shift from Bantu languages to Portuguese was taking place in the inner city of Maputo within both the older generation and its descendants, a phenomenon that may have expanded to the outer city.

The analysis of the effects of demographic factors on families’ language choices substantiates findings from studies on other African contexts which point out that increasing urbanisation, social mobility and intermarriage practices are correlated with language shift in the homes, from minoritised into dominant languages, often linguae francae and/or inherited colonial languages (e.g. Batibo, 2005; Igboanusi & Wolf, 2010).

The instrumental motivations and the impact of urbanisation on the choice of Portuguese are intrinsically tied to politico-ideological forces. Particularly relevant are colonial and postcolonial diglossic language ideologies and policies that have contributed to construct Portuguese as the high status language, in contrast with Bantu languages, restricted to informal functions and perceived as reservoirs of tradition. The co-operation of instrumental and politico-ideological forces on the parents’ option for Portuguese is made apparent in the following extract:

I, in my family, when I have a child, I first teach Portuguese to her/him. Why? Because I was born at the time of the Portuguese colonial rule. I teach Portuguese to the child so that s/he can grow up and learn from her/his siblings and from other children or to make her/his life easier when s/he starts schooling… this is because we were under colonialism and they said that the traditional language or Ronga, as I am Ronga, was dog’s language. At that time Portuguese was the legitimate language. (18IM/F/Maputo/2016, our emphasis)

In addition to the desire to facilitate school integration and socioeconomic success, Mrs. IM’s option for Portuguese was also informed by the colonial assimilationist ideology that constructed African languages as ‘dog’s languages’, i.e. as languages of less human people. Therefore, as attested in other postcolonial and also in diasporic contexts (e.g. Canagarajah, 2008; Coleman, 2011; Duff, 2020; Shen et al., 2020), since Bantu languages are perceived as not conducive to educational and economic rewards, many urban and middle class families neglect the transmission of these languages to new generations.

Language shift to portuguese: parents’ challenges and dilemas

In urban and middle class bilingual families, the parents’ decision to transmit Portuguese at the expense of Bantu languages is often made amid sociocultural dilemas and challenges, as they are aware of the communication limitations and sociocultural loss stemming from their Portuguese-only option. As illustrated, these parents are often still attached to their ancestral languages and cultures, which they assume should be passed on to their children, mainly for the sake of ethnic identity, cultural preservation, and intergenerational communication.

As a consequence of families’ option for Portuguese, many urban and middle class children do not speak Bantu languages, and are barely exposed to the associated culture in the home. It is worth mentioning, however, that parents who adopted this policy are now regretful, and condemn themselves for not having educated their children bilingually—in Portuguese and in their native language(s):

Well, if I had predicted at the time that my daughter would need Ronga in the future, I could have taught her this language at the same time as Portuguese. (…) So, I think it is important to have Portuguese for future professional purposes, but that traditional language is also important as a source of traditional wisdom and as a medium of communication with that group of people who cannot speak Portuguese. (01EM/F/Maputo/2016)

Mrs. EM’s daughter is a medical doctor in a public hospital. She is struggling to communicate with patients who cannot speak Portuguese, but Ronga, the native language of her own mother. Mrs. EM finds her daughter’s situation regretful, as she acknowledges that if she had known that her daughter would need Ronga in her future, she could have educated her bilingually—in Ronga and in Portuguese. Mrs. EM now perceives Ronga not only as a communicative resource, but also as a vehicle for accessing traditional wisdom.

A similar feeling of regret was expressed by other parents, such as Mr. BP, who said that by educating their children straight in Portuguese, “We tried a short-cut, but the short-cut led to this, isn’t it?” In the longer account from which this extract was taken, Mr. BP expressed concern about the cultural consequences of this option. The use of the pronoun ‘this’ subsumes his narrative about his children’s inability to speak their ancestral language and lack of access to the wisdom associated to that language.

Overall, parents are aware of the danger that the hegemony of Portuguese poses to Bantu languages, perceived as reservoirs of ancestral traditions and symbols of ethnolinguistic identity:

… language symbolizes individual identity. So, being symbol of identity, it also affirms our culture. When we overlook our language, we are automatically overlooking our culture. (37IL/F/Quelimane/2016)

Feelings of regret associated with cultural loss and communication breakdown between family and community members as a result of FLP privileging the transmission of dominant languages have also been attested in different postcolonial and diasporic contexts (e.g. Canagarajah, 2008; Duff, 2020; Wong Fillmore, 1991).

Acknowledging that language policies adopted by families are critical to the maintenance and development of Bantu languages, the participants are supportive of measures that could be taken in order to raise the social status of Bantu languages and mitigate the process of language shift. When invited to talk about specific actions, although recognising the role of the churches as institutions of use and transmission of these languages, parents mainly trust top-down strategies, driven by the government. They believe that, without the government support, families and civil society groups would not be able to rescue Bantu languages on their own. As expressed by Mr. AS: “Families are already weak.” (34AS/M/Quelimane/2016) In line with this subaltern stance, parents reiterated the call for the government to officialise Bantu languages and expand their use in education. As discussed in the next section, this stance may be a consequence of participants’ experiences with postcolonial socialist governance which was based on State-centered decision-making and tended to inhibit agency at individual, family and societal levels.

Understanding family language policy and language shift from a critical, multi-layered perspective

The evidence produced in this article substantiates the view that FLP decisions and language transmission patterns are greatly shaped by instrumental, sociodemographic and politico-ideological factors. As shown, the process of language shift to Portuguese in urban and middle class families has been mainly triggered by the perceived and de facto educational and socioeconomic rewards associated with this language, the increasing urbanisation and associated growing number of intermarriage practices, and its powerful political status. As argued, these pressures operate in an interrelated way and are entwined with family histories and trajectories. These are the conditions often regarded as the main causes of language shift in postcolonial Africa (e.g. Batibo, 2005; Igboanusi & Wolf, 2010; Igboanusi, 2009).

The complex and interrelated way instrumental, sociodemographic and politico-ideological factors of language shift operate in the Mozambican urban context underscores the value of approaching FLP from a multi-layered perspective informed by the ecology of language paradigm. It also destabilises Fishman’s (1991, 2006) perspective on the centrality of the family domain and intergenerational transmission for language maintenance, by substantiating the view that the family is not a static and self-contained institution, but a dynamic one that operates in line with other social institutions and economic conditions (Canagarajah, 2008).

This view helps to understand why family language ideologies and language transmission patterns may change as the families move from rural to urban settings, as they transition from a lower to a higher social status, and as national policies move from strict monolingual orientation to some form of multi- or plurilingual orientation. Based on other multilingual contexts (cf. Canagarajah, 2018; Coleman, 2011; Duff, 2020; Mufwene, 2004), it has been found that this adaptive nature of the human kind concurs to explain why families shift from their native languages to dominant languages, particularly inherited colonial languages and/or linguae francae in the African continent. As noted earlier, Mozambique does not have a national lingua franca, which may explain why the shift is often from the native Bantu languages into Portuguese, the ex-colonial language. The increasing proportion of Mozambicans who have declared Portuguese as their L1 in national censuses (from 1.2% in 1980 to 16.6% of the population in 2017) can be considered, at least in part, as evidence substantiating this pattern of shift. This pattern differs from that attested in countries with widely spoken native languages, as it is the case of Tanzania (e.g. Foster, 2023) and Botswana (e.g. Letsholo, 2009). In Botswana, for example, the shift is more often from minoritised native languages to Setswana, the lingua franca, although also to English, the inherited European language (e.g. Letsholo, 2009).

The evidence in this article suggests that the shift from Bantu languages to Portuguese in Mozambique is chiefly a phenomenon associated with increasing urbanization of the country. Census data seem to substantiate this finding, at least in part. As a matter of fact, in the 2017 national census, 39% of the population living in urban areas declared Portuguese as their L1, against only 5.1% of the population in rural areas (Chimbutane et al., ftc.). Although not all urban L1 speakers of Portuguese have lost their ancestor Bantu languages, the percentage of this category of speakers (39%) may be taken as an indicator of a language shift process in urban settings. As shown, this process is more apparent in Maputo City and Maputo Province, the most urbanised settings in Mozambique.

We should acknowledge, however, that the education background and the social status of the parents in this study may have influenced this general tendency to plan for the use of Portuguese as the family language and the language of education of their children, as also found in other African contexts. For example, focusing on FLP involving African migrants within Africa and considering the temporality dimension of language planning, Anthonissen & Stroud (2022, p. 119) found that while “socially secure families have time and material resources to invest in FLP”, in socially vulnerable migrant families “much planning and eventual fixing of an FLP is done implicitly, unreflectively, and therefore also covertly”. Although this study did not focus on migrant families, the parents in our sample seem to fit well in the category of ‘socially secure families’ (Anthonissen and Stroud 2022).

As in other multilingual contexts, preference for the dominant language in Mozambique does not necessarily mean loss of native languages and associated cultures, as families are often required to negotiate between broader social, economic and political pressures, and the desire to maintain and keep loyal to their ancestral linguistic and cultural values (Canagarajah, 2008; Curdt-Christiansen, 2013; Duff, 2020; Shen et al., 2020). In fact, the evidence produced in this study indicates that, while tending to invest in their children’s proficiency and academic success in Portuguese, parents also emphasise the identity value of their native languages and the need to maintain them. This suggests that, in principle, parents often recognise the need of both Portuguese and Bantu languages for their children, although associating them with different functions and affordances. As argued in Chimbutan (2021), this positionality substantiates Kusch’s notion of duality, which, contrary to dichotomy, entails avoiding taking extreme positions and instead maintaining the fragile balance between the opposites, so that the world does not destroy itself (Holas, 2017, p. 69, paraphrasing Kusch, 2008). Therefore, participants’ accounts suggest that urban and middle class parents take bi-/multilingualism in Portuguese and Bantu languages as part of their citizenry and also that of their children, although there is often a mismatch between their family language policies and practices.

Despite this dual positionality, as speakers of minoritised languages, the participants in this study also tended to “succumb to market forces, to survival instincts, social momentum, and opportunism, and shift to dominant languages” (Duff, 2020, p. 7). That is, affective and sociocultural attachments to their native Bantu languages seem not to be strong enough to transmit and maintain them, as socioeconomic forces often speak louder, particularly in urban settings. It is within this dilemmatic situation that some parents feel regretful when their children do not speak or are not proficient in their own native languages, which often subsumes cultural loss and intergenerational communication breakdowns (Canagarajah, 2008; Duff, 2020; Wong Fillmore, 1991).

As noted earlier, while acknowledging that the family is the prime institution for promotion and transmission of Bantu languages to new generations, the participants tended to place excessive weight on top-down efforts to maintenance and development of these languages. This positionality seems to be a consequence of the postcolonial socialist governance which was aimed at forging a kind of citizens who assumed that they could not act without the State’s guidance and sponsorship. Assuming this ideologically manufactured subaltern, non-agentive status, the participants seem not to be aware of the power of bottom-up efforts to secure transmission and maintenance of Bantu languages. As advocated by Lüpke and Storch (2013, p. 307), the existence of a ‘home base’ (village, neighbourhood, house or social group) providing “the opportunities for maintaining and creating communities of practice and social networks in a given language ecology” can prove to be key for the vitality of African languages. This view is in tune with Stroud’s (2001) suggestion that, instead of waiting for the State to determine their language use and rights, communities of speakers of minoritised languages can “seize power over the discourses and representations of language that define them” (p. 249) and exercise linguistic citizenship. Bottom-up processes such as these allow community voices and agencies to be heard in otherwise institutionally disadvantaging situations (cf. Chimbutan, 2018b; Foster, 2023).

Concluding remarks and the way forward

In view of the attested loss of Bantu language speakers in postcolonial Mozambique, this study has sought to understand the role of FLP in the process of language shift to Portuguese, within the ecology of language paradigm. While also considered Spolsky’s (2004, 2012) contribution to language policy research, we tried to move beyond that by adding to the studies that take a critical perspective to FLP. Focusing on middle class urban bilingual parents’ language policies and practices, the study found that FLP decisions are mainly driven by a combination of instrumental, sociodemographic and politico-ideological factors, often in conflict with sociocultural desires of the participants, particularly the desire to preserve their native languages and cultures and to ensure intergenerational ties. Pressured by these factors, which gravitate around the socioeconomic and political power accorded to Portuguese, middle class parents choose to invest in the transmission of this language to their children at the expense of Bantu languages, which is contributing to pave the way for language shift, especially in urban settings.

While, as mentioned, this article may contribute to FLP by foregrounding the value of using critical and multi-layered lenses to understanding family language choices in postcolonial contexts, further research on the process of language shift in Mozambique is certainly needed. Among others, there are two threads of research that can be pursued. The first thread concerns the need to investigate the boundaries and scope of the ongoing process of language shift from Bantu languages into Portuguese. Such investigation may be based on a survey of the sociolinguistic profile of L1 speakers of Portuguese who can also speak Bantu languages as well as their level of competence in those languages. A research perspective like this may allow not only to deepen our understanding of the boundaries of this pattern of shift from Bantu languages into Portuguese, but also to examine the dimension of bi/multilingualism Portuguese-Bantu languages in Mozambique. Regarding the latter dimension, although in a different sociolinguistic context, Posel & Zeller (2016) found that, contrary to conclusions from previous studies (e.g. de Klerk, 2000; Kamwungamalu, 2003), the increase in English language use among black South Africans is associated with increase in multilingualism rather than with shift from Bantu languages into that language. The second thread has to do with the need to attain a broader scope of the outcomes of FLP in Mozambique, which may include the exploration of other patterns of language transmission (e.g. families that opt for primary transmission of Bantu language(s) or for bi/multilingual transmission—Portuguese and Bantu language(s)), as well as the investigation of shift from minoritised Bantu languages into other dominant Bantu languages, as illustrated in Chimbutan, et al. (ftc.) and also documented in studies on other African contexts (cf.Batibo, 2005; Foster, 2023; Igboanusi, 2009; Letsholo, 2009). This broader research perspective would contribute to further substantiate Bodomo et al. (2009)’s notion of ‘multilingual language shift’.