The African continent is well-known for societal and individual multilingualism (Makalela 2015; Ndhlovu & Makalela 2021). Despite this indigenous multilingualism, it is the colonial languages English, French and Portuguese that dominate education (Guzula 2018; McKinney 2017; Ngũgĩ 1986). From the fourth year of formal schooling, African language dominant children in sub-Saharan Africa experience a traumatic severance from their familiar language used in teaching, learning and assessment in favour of education in the colonial language. At the point when children begin to specialise academically in the fields of natural science, mathematics and social science, they find that their most familiar languages are excluded from the learning experience. In effect, this excludes children from full participation in knowledge acquisition and creation at school, a situation which has been described as ‘epistemic exclusion’ (Kuchah et al. 2022). The detrimental effects of these language policies on learning for African children have been well described (Antia 2021; Guzula 2018; McKinney 2017; Probyn 2009; Stroud & Kerfoot 2021) and increasingly interpreted as part of the wider scourge of the coloniality of language and racist language ideologies (Makoni 1998; Ndhlovu & Makalela 2021; Ngũgĩ 1986; Tyler 2023; Veronelli 2015).

Veronelli (2015) has described the coloniality of language as having two pillars. Firstly, colonisers interpreted colonised peoples as being deficient in true language capabilities and positioned them as mere communicators. Secondly, the speech of colonised peoples, such as Africans, was codified by missionary linguists within the logic of Western language theory and therefore scholars have argued that standardised African languages exist as colonial inventions (Makoni 1998). These registers are often far removed from how speakers use language in daily life. A South African scholar, Kell (2022:188), has extended this analysis to the domain of literacy in particular, arguing that it is based largely on English literacy and that the ‘coloniality of literacy’ is a dominant force in education.

One of the outworkings of the coloniality of language in the South African schooling context is the endurance of the ‘text frontier’ (Tyler & McKinney 2024) in African language use in schooling. While teachers and learners have long engaged in oral translanguaging (García & Wei 2014; Makalela 2015) or code-switching (Adendorff 1993; Probyn 2009) in classrooms, there has been a near-universal avoidance of the use of African languages or bilingualism in texts. Learning materials and assessments are provided in English only beyond Grade 3 (Guzula & Tyler 2025) and classroom walls and children’s exercise books are equally monolingual in English in the content subjects (Krause-Alzaidi 2023; Tyler 2023). One of the ironies of the infamous Bantu Education system was that it included the development of learning materials in African languages for the first 8 years of school. However, the mandatory use of African languages as languages of teaching and learning became aligned in the public imagination with the low quality of education, which was designed to prepare black¹ children for a life of servitude to their white masters. The negative attitudes towards African language learning materials still remain an ideological barrier to their use in education today, even as the Department of Basic Education embarks on MTbBE. Furthermore, the high status of English is well established within a milieu of anglonormativity: the expectation that everyone will be proficient in English and is deficient if they are not (McKinney 2017).

In their article entitled ‘Translanguaging and Literacies’ García and Kleifgen (2020) describe a pernicious injustice against children, which occurs as a result of the exclusionary force of anglonormativity:

McKinney (2022) describes this lack of confidence in the literate subject position in her study on note-making in a Grade 4 science class in Cape Town. McKinney describes the children’s rote note-taking as ‘safe writing’ resulting in the foreclosure of opportunities for the children to learn science deeply or for ‘making sense of text’ (McKinney 2022:14). As their familiar linguistic resources were excluded from their sense-making in text, the teacher resorted to a focus on securing the fixed wordings (Lemke 1990) of the science topic to be rote-learned in preparation for the assessment.

Arguably the most egregious manifestation of monolingual English literacy norms in schooling is in testing practices. Only answers which conform to assessors’ notions of standard English are accepted on standardised as well as local school-designed tests even if the non-English answers include correct content. Prinsloo and Krause (2018) describe the preparation for and administering of these tests as ‘a rather painful parody of effective engagement’ (p. 165). The results of these tests serve to sediment the bimodal schooling results in South Africa where, in the main, English and Afrikaans speaking children (predominantly White) succeed and African language speaking children (Black) do not (Christie 2020). McKinney and Set (2023) have shown how the schism between heteroglossic classroom practices and monoglossic assessment is prevalent in Namibia also.

Leaving behind monolingual English literacy practices, I now move to a discussion of how biliteracies have been theorised, including as part of decolonial practice in education.

Traditional biliteracy development, which assumes the goal of literacy to be reading and writing in two separate named languages only, is deemed particularly inappropriate for emergent bilingual children (García 2009) who are speakers of non-dominant languages. Going beyond traditional biliteracy models, this study shows children engaging in activities employing separate named languages as well as fluid ubuntu translanguaging (Makalela 2015) or meshed registers (Gibbons 2006) for simultaneous biliteracy development (Benson 2022; Guzula 2018). For bi/multilingual learners, especially emergent bilinguals, confidence in being literate is fostered by having regular opportunities to express oneself in writing using familiar linguistic and other semiotic resources while also developing new resources for literacy. Biliteracy was originally conceptualised broadly as ‘any and all instances in which communication occurs in two (or more) languages in or around writing’ (Hornberger 1990:213). Hornberger later developed the notion of the continua of biliteracy. The theory of continua recognises the ‘highly complex and fluid systems’ in use during any biliteracy activity (Hornberger & Link 2012:264), building on the notions of flexibility, fluidity and dynamism already well established in bilingual language scholarship (e.g. García 2009). In line with the continua of biliteracy, García and Kleifgen (2020:1) argue that ‘bi/multiliteracies cannot be fully understood as simply the use of separate conventionally named languages or separate modes’. Features of named languages can co-exist in parallel in multilingual texts (Sebba 2013); or features of named languages can be combined in a meshed register (Gibbons 2006). Sebba proposes a framework for multilingual text analysis in which he shows how multilingual texts can be presented in relation to the following categories: linguistic properties, language-spatial relationships and language-content relationships (Sebba 2013). These relationships have consequences for meaning, which form part of multilingual text authors’ design decisions.

Scholars have argued that a critical Southern perspective on multilingualism(s) is key to the decolonial project (Ndhlovu & Makalela 2021; Stroud & Kerfoot 2021). Part of this perspective includes bringing Southern epistemologies to the fore. Ndhlovu and Makalela (2021) assert that African ‘ways of knowing’, or epistemologies, should be centred when designing learning. Taking into account the ways of knowing that children bring to the classroom will allow them to be positioned as embodying a ‘knower’ subjectivity (Kerfoot & Bello-Nonjengele 2023). Children gain confidence if their prior knowledge, including linguistic knowledge, is valued. Positioning children as knowers in a Southern context requires embracing a multilingual episteme (Kerfoot & Bello-Nonjengele 2023) where all linguistic resources are valued in classrooms. A central area in which the knower subjectivity is expressed is in reading and writing for learning. Children in school need to experience becoming ‘literate beings’ (García & Kleifgen 2020:564). Emergent literate beings are beginning to control reading and writing in a variety of styles for a variety of purposes. For racialised bilinguals, such as the children in the present South African case, this literate subjectivity will be catalysed by reading and writing experiences, which draw on their full and emerging semiotic repertoires (Lin 2015). They therefore need to be given opportunities to develop a range of biliteracies. The bua-lit language and literacy collective in South Africa have described these opportunities as ‘rich literacies’ (bua-lit language and literacy collective 2018) and have distinguished these from the impoverished literacy practices, such as safe writing, offered by the dominant monolingual, ‘back-to-basics’, skills-dominated literacy paradigms being used to try to address South Africa’s poor literacy results on international tests.

Despite the dominance and official validation of monolingual English literacy practices, research has also traced biliteracy practices. In the following literature review, focused on studies from the Global South, I trace biliteracies in learning materials and biliteracy practices in classrooms. These activities involve the potential for children to both read and write in two or more languages.

Bilingual learning materials vary in terms of genre. Milligan, Clegg and Tikly (2016) and William and Ndabakurane (2017) reported on the use of language-supportive textbooks in Rwanda and Tanzania, respectively. Extracts from Biology and Mathematics textbooks written in English show occasional bilingual Kinyarwanda/English and Swahili/English glossaries among other language-supportive strategies. While bilingual pedagogies were implicit in the materials (e.g., learners were instructed to have discussions in Kinyarwanda), the use of African language text was limited to the glossaries. Abdulatief and Guzula (2024) studied one science teacher in the Western Cape who had been enrolled in a professional development programme, which included in-class mentoring. She employed multilingual science glossaries and Google Translate to enable her learners to create bilingual content posters for learning and wall display and presented her board notes bilingually. Charamba (2023) studied the practices of a primary science teacher in the Free State who created a cumulative English–isiZulu bilingual chart for the science topic under study. The teacher and his Grade 5 learners contributed to it throughout the school term and displayed it in the classroom. Makalela (2015), in his study of a Grade 6 bilingual Sepedi/English literacy intervention in Limpopo, observed that both Sepedi and English appeared in the print environment inside the classroom. Both ‘parallel’ and ‘dynamic’ bilingualism was present in children’s texts, which were displayed in multilingual literacy corners, giving legitimacy to the children’s biliteracies.

Biliteracies can be developed by teachers through the writing activities, which they set for their learners and through how learners recruit bilingual resources spontaneously in their writing. High school learners in Cape Town were shown to engage successfully in learning science through writing translations of isiXhosa science definitions into first standard English and then a familiar informal register of isiXhosa (Tyler 2023). In Kenya, Kiramba (2017) analysed writing by Grade 4 children for English and Kiswahili composition tasks. She found that while the task required children to write strictly in one language only, translanguaging in the compositions was common and allowed the children to display knowledge in English, Kiswahili and Kimeru and to renegotiate identities. Tyler and McKinney (2024) presented data from a mathematics classroom where a teacher set an activity for children to write down their thinking during a calculation. The children drew freely on both English and isiXhosa and translanguaging to represent their thinking. Charamba and Zano (2019) reported on a bilingual English-Sesotho classroom in the Free State province of South Africa. Science learners discussed and wrote their answers in their home language and then translated their answers into English (the school’s medium of instruction). The groups would then read their responses (in English) to the rest of the class. These learners were also given multilingual learning materials in Sesotho and English.

The culmination of biliteracy pedagogy is bilingual assessments. Although this component of a fully integrated bilingual education model is often the most resisted (De Backer, Van Avermaet & Slembrouck 2017), there are examples of innovative practice to be found. Charamba (2021) studied the difference in performance in Grade 5 Zimbabwean children when they wrote a monolingual English test versus a bilingual English/Shona test. The children performed significantly better on the bilingual test and also demonstrated their linguistic flexibility by answering the bilingual test in either English or Shona or a translanguaged register. Antia (2021) argues for the inclusion of African languages in multilingual examinations in South Africa for epistemic justice, by showing how South African learner research participants engaged in bilingual English/Afrikaans co-languaging when completing a school-leaving examination.

While the studies reviewed above have pointed to the efficacy of including biliteracy practices in learning for emergent bilingual children, a gap emerges in the literature where little is known about the range of these practices and the texts, which are produced in classwork and assessment. The need to fill this gap is particularly urgent in primary science, which is a key focus subject for MTbBE in South Africa. The present case study addresses this gap.

Ethical clearance to conduct this study was obtained from the Human and Social Sciences Research Ethics Council of the University of the Western Cape (No. HS22/6/32). Adult participants and children gave written assent and the parents gave written consent on behalf of their minor children. All participants’ identities, as well as that of the school, were kept confidential through assigning pseudonyms.

There were monolingual English science textbooks in a cupboard in Ms Azania’s classroom when we arrived for our fieldwork in January 2023. Ms Azania reported, however, that she did not hand these out to the children because they did not know enough English yet to be able to use them. In contrast, we observed the children reading iSayensi Yethu independently on a number of occasions during the intervention. In this section I will describe the different kinds of bilingual text, which fostered biliteracies in the children, beginning with the bilingual textbook (Figure 1). Sebba (2013:113) argues that in analysing bi/multilingual texts, the visual and spatial relationships as well as the linguistic properties of the text need to be considered. Drawing on Sebba’s framework, I pay attention to language-spatial relationships, language-content relationships and linguistic mixing types of the textbook to show the diversity of biliteracies the children were exposed to in reading the textbook.

The introduction to the textbook is provided first in isiXhosa and then in English. The Grade 4 science teacher is addressed as follows in the English version:

Here the authors of iSayensi Yethu refer to the linguistic mixing types (Sebba 2013) of the textbook, including both monolingual units (‘the languages appear in separate texts’) and mixed units (‘languages are used together in one sentence’). An example of a mixed unit, otherwise known as a translingual text, is:

The purpose of the mixed units was to present the content in a more familiar, spoken-like way (Guzula & Tyler 2025). Presenting mixed units in the same textbook as monolingual units was intended to provoke discussion in classrooms about different kinds of language for learning (i.e., critical multilingual awareness) as well as to provide a model of how children may write. In terms of the language-content relationships, the textbook includes equivalent, disjoint and overlapping texts (Sebba 2013). An example of an equivalent text is a glossary of scientific terms presented in both languages. Here the same content is presented twice, for example, ‘iimpawu = characteristics’ (bua-lit language and literacy collective 2022:2). An example of disjoint texts is in the information text on cows where different content is presented in English and isiXhosa:

An example of overlapping texts is found in the explanation of once-living things where only some of the content from English is repeated in isiXhosa:

In terms of language-spatial relationships, most prominent was the code preference system, which alternated. In headings, English was given preference, appearing above isiXhosa, but in the ‘did you know?’ boxes isiXhosa appeared above English. The authors designed this intentionally so that neither language dominated the textbook (Guzula & Tyler 2025).

In addition to the bilingual textbook, Ms Azania printed posters with key words from the topic in both languages and fixed these to the classroom walls (Figure 2). These posters produced a print-rich environment (Makalela 2015) in both languages. The posters consist of coloured card printed with text in isiXhosa and English. Ms Azania also affixed bilingual, laminated science posters to the walls produced by a local publisher and donated by the project. Using these durable, high-quality resources to present biliteracy on classroom walls increases the status of African languages and biliteracy for learning. This is in contrast to another school environment in Cape Town where African language text was displayed using non-durable, inexpensive resources such as cardboard and hand-written text and printing was reserved for English text (Tyler 2023:46).

We observed high levels of participation in all classroom activities, and Ms Azania reported that participation had increased with the advent of the bilingual materials in comparison with the same period the previous year. Bilingual talk was common and was engaged in by the teacher and the children, who took longer turns and chose their language of response freely. Different kinds of pedagogical translanguaging (Gorter & Cenoz 2017), such as varying input and output languages and guiding children to use specific languages for specific activities were part of the lessons. It is beyond the scope of this article to analyse these oral practices, but certainly they reinforced the biliteracies.

Apart from the bilingual materials themselves, which guided the children to use both isiXhosa and English (and in some cases any other language) in writing, there were many examples of biliteracies in Ms Azania’s pedagogy. Ms Azania wrote bilingually on the board using equivalent texts (Sebba 2013). In the example in Figure 3, she contextualises the contributions made by the children in both the language in which they spoke (isiXhosa) and in English. In the boardwork, isiXhosa is privileged, appearing on the left, as this is the language in which the children offered the examples. In the colourful posters (Figure 2), English is privileged, appearing at the top. In this way she legitimises the children’s voices as well as teaches them the orthography of the less familiar equivalent English words.

With the support of the bilingual materials, Ms Azania guided the children to write in both languages in their workbooks (Figure 4 and Figure 5). Figure 4 shows a child labelling an activity in which he is identifying pictures of living and non-living things. In his book, he has written a list of seven life processes in both English and isiXhosa. In this case, he has followed the instructions to write each piece of information in both languages: an example of equivalent bilingual texts (Sebba 2013). The writing in Figure 5, however, shows a different pedagogical instruction given in the booklet where learners can respond by writing examples of living and non-living things in any language they choose. This child knows the names of things around her in English (fly, worm) and isiXhosa (ibhabatane, intethe) and uses these names to display her developing scientific knowledge. It is significant that the child chooses his most familiar language, isiXhosa, to display his emergent, hesitant ideas. Indeed, the child may have found a dead butterfly and grasshopper in the playground and grappled with the conceptual challenge of whether these were living or non-living things, consequently placing them in the ‘not sure/andiqinisekanga’ column.

During the course of the study of living and non-living things, I tentatively broached the possibility of bilingual written assessments with Ms Azania. According to general practice and the language policy of the school, these were usually written monolingually in English. We wondered together whether the head of the school science department, the principal and the education department official would allow the children to read and write a bilingual assessment. Aided by the careful introduction of the project, our requests to the powers-that-be landed positively, and we began working on the method of preparing bilingual assessments. The extract found in Figure 6 is taken from the summative test, which the children wrote at the end of the first half-year of Grade 4.

Ms Azania designed the bilingual test to pose each question first in the children’s most familiar language, isiXhosa, and then in English. However, the languages were not strictly separated, and flexibility was included. For example, the formal isiXhosa word for ‘fridge’ is ‘isikhenkcezisi’ but this word would be unfamiliar to the children in their daily lives. They would be more familiar with the word ‘fridge,’ and so Ms Azania accommodates this and includes it in brackets. In this way she also provides a vocabulary learning opportunity in isiXhosa. Biliteracy of this nature underscores the importance of language and content subject teachers working together in bilingual schools to consolidate new information across subjects. When writing the bilingual tests, learners had the option of answering each question in whichever language they chose. In some cases, learners chose one language and answered the whole test in that language; in other cases, learners answered some questions in English, some in isiXhosa and some in mixed units (see Figure 6).

Question 4.1 in Figure 6 requires the children to engage in a core scientific activity: prediction (Lemke 1990). By giving the children freedom to respond using their full linguistic repertoire, the teacher enables the children to focus this scientific activity rather than diluting their efforts by requiring them also to produce a complete English answer. For many children at this age and level of prior exposure to English, it would be impossible to produce a comprehensive monolingual English answer. Drawing on their full linguistic repertoire enables them to develop the cognitive and written genre skills necessary to make scientific predictions.

The child in Figure 6 has responded to the question using both English and isiXhosa resources. The second line of her answer justifying her choice of Bongi’s seeds, ‘kuba isefestileni and kukho ilanga elinobushushu’ [because they are on the windowsill and there is warm sun there], includes the English feature ‘and’ which is sedimented into the child’s linguistic repertoire and is unmarked, aligning with the concept of ubuntu translanguaging (Makalela 2015). The teacher awarded full marks for the answer, whereas in the previous year she would be obliged to award zero as the answer was not given in standard English.

A total of 42 children who wrote the bilingual test responded using both English and isiXhosa resources, with isiXhosa being used more than English. This is to be expected as only a summer holiday separated the children from a prior monolingual isiXhosa language of instruction. In the bilingual test group (class 4G), 36 of the 42 children (86%) achieved a passing grade, and in the English-only group, only 64 out of 126 children (51%) achieved a passing grade. While this study is qualitative in design, does not utilise an experimental control group and considers a classroom to be an open, dynamic system, this descriptive 35-percentage-point difference in pass rates suggests a possible relationship between the bilingual test and the learner performance that warrants serious attention.