Towards Epistemic Justice: Constructing Knowers in Multilingual Classrooms

Abstract

In this study of a postcolonial school, we expand understandings of epistemic justice from the perspective of language, addressing issues of know-ledge, understanding, and participation in communicative practices. We suggest that monoglossic language-in-education policies constitute a form of epistemic injustice by diminishing learners’ ability to make epistemic contributions, a capacity central to human value. We further suggest that translanguaging in formal school settings generally promotes epistemic access rather than epistemic justice, leaving value hierarchies and relations of knowing unchanged. Conversely, this study presents linguistic ethnographic data from a three-year project where learners could choose their language of learning to Grade 6 and use all languages in subject classrooms. We analyse how a Grade 6 learner used laminated, multilingual stances to construct others as knowers, negotiate epistemic authority, and promote solidarity. We argue that she thereby constructed new decolonial relations of knowing and being. Moreover, the shift from monolingual to multilingual episteme, which substantially improved performance overall, enabled new social, epistemic, and moral orders to emerge from below, laying the basis for greater epistemic justice.

1. INTRODUCTION

This paper aims to expand understandings of epistemic justice by addressing it from the perspective of language. In particular, it explores how language-in-education policies and practices that embrace a multilingual episteme can construct more egalitarian, just, and ethical conditions for learning. Epistemic justice relates to issues of knowledge, understanding, and participation in communicative practices (Fricker 2007; Santos 2014). It is concerned with relations of knowing: those that construct (or fail to construct) others as knowers and, more importantly, as producers of knowledge. In this paper, we understand epistemic injustice as inextricably bound to coloniality, the racialized structures of power and prescriptions of value that survive colonialism and are kept alive in contemporary structures of governance (Quijano 2000).

From a decolonial perspective, language is profoundly implicated in epistemic justice. There is a compelling international consensus that a solid foundation in a familiar language is a strong predictor of educational success (e.g. Bamgbose 1984; Collier and Thomas 2017; Ouane and Glanz 2011; The World Bank 2021).¹ Yet a majority of educational systems worldwide still implement monolingual, effectively colonial, models of education. These models are found not only in former colonies that chose an ex-colonial language for economic and/or nation-building reasons but also in countries of the global north with indigenous and/or minority language populations and increasing migration-driven diversity.²

The consequences are, almost without exception, a perpetuation of linguistically structured inequalities, evident in substantial differences in educational outcomes for speakers of marginalized languages (Ouane and Glanz 2011; Lo Bianco 2016).³ We suggest that such monoglossic language-in-education policies and practices constitute a form of epistemic injustice in that they can inhibit or prevent learners from making epistemic contributions, that is, from conveying knowledge to others (Fricker 2007). This kind of ‘testimonial’ injustice is ‘a wrong done to someone specifically in their capacity as a knower […] in a capacity central to human value’ (Fricker 2007: 1, 5).⁴ From a socio- and applied linguistic point of view, it perpetuates colonial ideologies of who it is that can legitimately know and through what language. In this way, it sustains social perceptions that judge second language speakers as deficient or less credible (Veronelli 2015). In classrooms, the harmful operation of these colonial ideologies results in epistemic exclusion, silencing, or resistance (Kerfoot and Tatah 2017; Kiramba 2018).

In contrast, this paper explores the ways in which a transformative language-in-education policy can contribute to epistemic justice by countering exclusion, reducing harm, and promoting participatory parity. It draws on data from a three-year Linguistic Ethnography (LE) during the implementation of the Western Cape Language Transformation Plan (2007) in 16 South African schools. This Plan provided that, in contrast to prevailing practices worldwide, multilingual primary school learners could choose any of the three main regional languages as medium of instruction up to Grade 6: isiXhosa, Afrikaans, or English.⁵ A crucial addition to this visionary but essentially monoglossic official policy was made by those working with teachers in pilot schools: learners were encouraged to use all language varieties in their repertoires in any classroom, thus changing from a monoglossic to a heteroglossic episteme.

The paper, therefore, goes beyond current work on translanguaging in education, showing what happens when multilingualism is ‘unmarked’ (Menezes de Souza 2021)—seen as the norm rather than an exception—and identifying multilingual classrooms as crucial sites for the promotion of epistemic justice. It suggests that translanguaging in formal school settings is for the most part geared towards a monolingual outcome, that is, towards accessing knowledge in an official language. This unidirectional impetus means that translanguaging generally remains an affirmative rather than transformative strategy in Fraser’s (1995) sense, leaving underlying hierarchies of value and relations of knowing unchanged and promoting epistemic access rather than justice.

Conversely, unmarking multilingualism can illuminate the potential of learners as epistemic agents. It can legitimize semiotic practices which may intersect with or be embedded within those legitimated in institutional policies and practices (cf. Stroud 2018) and simultaneously nurture sustained processes of multilingual knowledge-building. This paper makes such processes visible. It thereby also addresses the lack of research on multilingual children as knowers, focusing explicitly on peer mediation processes.⁶ Here it brings together and expands work on epistemic and affective stances (e.g. Heritage and Raymond 2005; Goodwin and Kyratzis 2007; Stivers et al. 2011; Cekaite 2013) to show how, as learners take up various stances towards the task in progress, they constitute one another not only as epistemic actors but also as social and moral actors.

The paper draws on observational, interview, and audio-recorded interactional data to illuminate emergent social, moral, and epistemic orders in one of the project schools. We show how a Grade 6 isiXhosa-speaker, Ensha, informally took on the role of facilitator and co-constructed knowledge with her peers using her first language (L1), English, Afrikaans, and other elements of her semiotic repertoire. We analyse the ways in which she used laminated multilingual affective and epistemic stances to construct others as knowers, negotiate epistemic authority, and promote solidarity. We suggest that, in so doing, she constructed new relations of knowing and being, a new ethical project.

We start with a brief contextual background and then discuss the affordances of LE for illuminating the ongoing negotiation of relationships to knowledge and other knowers. We then present an analysis drawing on analytical tools from IS and Appraisal theory to identify the ways in which epistemic, social, and moral orders are simultaneously constructed.

Throughout, we understand languaging, rather than translanguaging, to be at play. Maturana’s (1990) concept of languaging has ‘change, ongoingness at its center’, capturing the dynamism and flexibility of everyday semiotic practices (Veronelli 2015: 121). In this sense, it is consistent with a multilingual episteme, one that sees ‘multilinguality’, or the multilingual capability, as an essential condition of what it means to be human (Agnihotri 2014). While translanguaging is often described as transcending ‘the boundaries between named languages, language varieties, and language and other semiotic systems’ (García 2009; Li 2018: 9), it retains at its core the notion of moving across bounded entities. Translanguaging has, in many contexts, been shown to have strategic value in helping to challenge entrenched classroom hierarchies of value and validate linguistic identities (Lin and He 2017; Alby and Léglise 2018; Hernandez Garcia and Schleppegrell 2021; Makoe 2021). However, for our purposes, languaging is better able to carry the onto-epistemological challenge of rethinking, or unthinking, language-in-educational contexts. It contains both Halliday’s (1985) ‘meaning potential’ and the radical heterogeneity suggested by Thibault (2017), building on Love (1990). This heterogeneity involves ‘the interaction of processes on many different time-scales, including neural, bodily, situational, social, and cultural processes and events’ (Thibault 2017: 76). Such considerations are essential from a decolonial perspective, as is the need to incorporate a plurality of interpretive resources and expressive practices (Medina 2012).

We suggest that the emergent processes of languaging and knowing identified in this paper are decolonial in that they contribute to transforming the logic of language and learning from below and to imagining possible futures. We end by offering a definition of epistemic justice in education.

2. BACKGROUND: LANGUAGE, RACE AND EPISTEMIC JUSTICE IN SOUTH AFRICA

2.1 The epistemic harm of contemporary language-in-education policies

As indicated above, bi- or multilingual education is not merely a policy or linguistic issue, but an epistemic one (Menezes de Souza 2017). In South Africa, in the General Household Survey of 2018, only 8.1 per cent of all respondents reported English as a first home language (Statistics SA 2021). While the format of the survey did not allow for bi- or multilingual homes, it can be estimated that at least 75 per cent of learners are learning through an unfamiliar language (Taylor and von Fintel 2016). Although the post-apartheid Language-in-Education Policy (1997) recognizes 11 official languages and promotes functional multilingualism (Heugh et al. 1995; Heugh and Stroud 2020), most schools change to English as language of learning at Grade 4, if not earlier, making this in effect a subtractive, monoglossic policy. This policy has been preserved by a complex mix of historical and economic factors whereby African languages are consistently devalued, either as relics of apartheid education or as unable to carry the knowledge necessary for scientific and economic participation in a globalizing, neoliberal economy (Alexander 1989, 2012; Prah 2015; Stroud and Kerfoot 2021). Attempts to develop bi- or multilingual education beyond Grade 3 are choked by such language ideological factors but primarily by a lack of infrastructure and resources, both human and capital, and a lack of political will, the seeming inertia of a postcolonial elite (cf. Fanon 1961). This institutional inertia contrasts strongly with the everyday vibrancy of linguistic practices outside and, often invisibilized, inside the classroom. As urban schools face increasing diversity in learner populations, education authorities lag behind in imaginative responses to promoting the linguistic well-being of all learners, maintaining the essentialized ties between language and ethnicity on which the apartheid system was founded, and perpetuating social and epistemic harm (cf. Oostendorp 2022).

A direct consequence of ineffectual policy implementation is evident in the Progress in International Reading Literacy Study (PIRLS 2016) which found that 78 per cent of South African Grade 4 children could not read for meaning in any language (Howie et al. 2017).⁷ In 2019, only 37 per cent of learners that had started in Grade 1 12 years earlier completed their schooling and only 16 per cent obtained a pass for university entry (UMALUSI 2015; Department of Basic Education 2020; van der Berg et al. 2021). These figures suggest that, by the end of their schooling, the majority of those learning through a second or additional language gain only the most superficial ability to engage with school knowledge, that is, to negotiate it, appropriate it, transform it or transmit it effectively to others.⁸ This form of linguistically based epistemic injustice is a ‘tracker prejudice’ ‘systematically connected with other kinds of actual or potential injustice’ (Fricker 2007: 27). It lays learners open to long-term trajectories of economic and sociopolitical exclusion and disadvantage, along with reduced confidence in their own epistemic worth.

2.2 The Western Cape’s Language Transformation Plan

In contrast, the learners in the school studied here were not obliged to change to English as language of learning in Grade 4. As described above, they were part of the Language Transformation Plan (LTP 2007), a pilot project to enhance epistemic access and challenge raciolinguistic hierarchies of language and value (see Alim et al. 2016). The project was supported by a two-year bilingual teacher in-service programme run jointly by the Language Education Department of the University of the Western Cape and an NGO, the Project for the Study of Alternative Education in South Africa (PRAESA).

Literacy results improved dramatically. Eight of the 16 schools opted to write the Grade 6 assessment in their chosen language (isiXhosa) and after two years achieved ‘almost four times higher than their scores for 2005’ (Western Cape Education Department 2008). This significant improvement in outcomes resulted in plans for a roll-out of this policy to all Western Cape schools. However, a change of political leadership in the Western Cape led to a unilateral shutdown.

This paper illuminates one of the effects of this change of episteme on ways of knowing, speaking, and being. It presents findings from a four-year LE conducted during the implementation of this plan. It analyses the translingual negotiation of knowledge through the co-construction of affective stances indexing solidarity and epistemic stances indexing respect and open-mindedness. We suggest that the shift to a multilingual episteme enabled the construction of new relations of knowing and the emergence of a decolonial ethics of care.

3. LINGUISTIC ETHNOGRAPHY: INVESTIGATING THE CONSTRUCTION OF KNOWERS

The study was designed as an LE which combines IS and ethnography. IS, grounded in the work of Gumperz (1982a,b) can illuminate the intricacy of communicative processes in which there are considerable differences in access to cultural, linguistic, and symbolic capital (Rampton 2006) while ethnography has counterhegemonic potential which can help to destabilize conventional truths (Blommaert 2006).

LE is thus ideally suited for investigating the ways in which knowers are constructed in a multilingual classroom. It provides tools for analysing how epistemic and affective alignments are discursively established across speakers and turns (Jaffe 2009). At the same time, it offers concepts such as participation framework, footing, and contextualization cues (Goffman 1981; Gumperz 1982b) which enable an analysis of how speakers are included or excluded, their epistemic stances taken up or side-lined, their desired social identities enabled or circumscribed. Appraisal, a heteroglossic Bakhtinian framework for analysing evaluative meaning-making, extends these analytical possibilities (see White 2003; Martin and White 2005).

To gain insight into these processes, we collected observational, interview, and audio-recorded interactional data among multilingual 10–13 year olds in a multigrade English class (Grades 4–6). This paper focuses on one of the ten key participants aged 10–13 years in the project. Participants were selected to ensure equal representation by gender and from the two major language groups in the school—Afrikaans and isiXhosa. Both authors carried out observations; the second author followed participating children to different locations in school and home settings. Learners carried pocket recorders during interactions in classrooms and on playgrounds. Accordingly, neither researcher was present during the interactions. Additionally, researchers had been in the school for five years by the time this data were collected, so their presence was largely unremarked.⁹ The corpus of classroom data consists of 40 hours of audio-recordings complemented by interviews; interviews were initiated in English but followed language switches as initiated by learners or their caregivers. Recordings were audio rather than video, as the latter was found to be too intrusive. As a result, only field notes could testify to the use of embodied semiotic resources in meaning-making. Given the time-consuming nature of transcribing and translating multilingual data, the opportunity for retrospective participant checking was sometimes lost. However, translations and interpretations were checked with home language speakers, both lay and academic. Transcription conventions are given in Appendix 1 in Supplementary Data. Multilingual informed consent forms were signed by teachers and by all learners in the class and their caregivers.

3.1 Setting

The school studied here is in Delft, a large ‘township’ on the periphery of Cape Town with a uniformly low socio-economic status. Only 27 per cent of those aged 20 years and older had completed Grade 12 (City of Cape Town 2013).¹⁰ At the time of the research (2008–2012), the school consisted of 70 per cent black African learners with 60 per cent ‘coloured’ teachers and 80 per cent ‘coloured’ administrative staff including the principal.¹¹ This profile reflected that of the surrounding municipal ward in which 67 per cent were African (Statistics SA 2013). South African studies of language policy and practices have generally focused on formerly ‘White’ schools as sites for engagement with entangled ideologies of whiteness and linguistic value. These schools, now highly diverse but often with white teachers and staff, expect Black learners to acculturate to established monoglossic, ‘anglonormative’ linguistic, and cultural hierarchies (McKinney 2016). In the school studied here, however, the White English-speaking ‘Other’ was absent from the site. Moreover, this was a new school: no single group was positioned as holding greater claims to legitimized knowledge (cf. Soudien 2007).

The data in the next section are from a large, multigrade, English-medium Economics and Management Science class the year before the start of the LTP. The use of other languages was, therefore, covert in contrast to that in the subsequent section.

4. ENSHA: SURVIVING IN THE ENGLISH-MEDIUM CLASSROOM

Ensha was 13 when these data were collected. She was from an isiXhosa-speaking family, was learning English as a subject, and had acquired Afrikaans from her peers. For the first five years of her schooling, she chose an isiXhosa medium stream where she successfully completed each year, a lively participant in class who clearly enjoyed learning. However, once she reached Grade 6, her mother, perhaps influenced by broader neoliberal ideologies of the low value of African languages, changed her to the English-medium stream. She subsequently failed and had to repeat the grade.

In contrast to previous years, observations during Ensha’s first year in Grade 6 showed her as a quiet and non-participating member of the class. When asked why she no longer raised her hand in class, she responded as follows:

Ensha: Maybe the teacher asked me the question [….], if I am telling the answer,the students will laugh at me.

Basirat: Do you think they laugh because you are telling the wrong answer or the way you tell the answer?

Ensha: The way I tell the answer.

Here her ability to be recognized as a competent knower was compromised by her inability to articulate that knowledge in a particular language. This lack of recognition was compounded by a fear of humiliation, embodied in self-silencing, a response to the institutional erasure of her linguistic and knowledge resources which made the subject position of ‘competent knower’ impossible (see also Kerfoot and Hyltenstam 2017).

When explicitly called on by the teacher, she would put a finger on her lower lip and look down at her feet, or look around the class above people’s heads. If another student tried to suggest an answer while she was struggling, she would exclaim forcefully in isiXhosa ‘Intoni?’ (What?). When particularly hard-pressed, she would push her chair violently into her desk or knock her desk sideways, the materialization of epistemic harm moving into the body.

This linguistic marginalization can be seen as a form of epistemic exclusion, ‘an infringement on the epistemic agency of knowers that reduces her or his ability to participate in a given epistemic community’ (Dotson 2012: 24) and a denial of part of what it is to be fully human, so that a person ‘may be, quite literally, prevented from becoming who they are’ (Fricker 2007: 4).

As a result of these experiences, even after successfully passing Grade 6 the second time in English, Ensha stated unequivocally that she would have preferred to continue with isiXhosa as medium. Her mother agreed.

The next set of data from the same classroom was collected in the first year of the implementation of the LTP where all languages were considered legitimate resources for learning.

5. NEW WAYS OF KNOWING AND OF BEING

Despite Ensha’s negative experience the first time, her failure and subsequent repetition of Grade 6 had interesting consequences for her academic identity: she began acting as facilitator among the Grade 4 and 5 learners in her multigrade English class. During reading or break time, she called Afrikaans and isiXhosa L1 learners together and set up simulated classrooms in the playground.

In the classroom, as documented throughout the year, she frequently took the role of knowledge mediator for her Grade 6 peers. In the six excerpts that follow, we see her scaffolding the concept of ‘economic flow’ for two learners, Thulani (male) and Zozo (female), who had entered Grade 6 for the first time and were one and two years younger, respectively. The group knew each other well and worked together frequently. All spoke isiXhosa at home but also English and Afrikaans with friends at school.

The task was a three-stage activity in which groups had first to define a flow, describe something that flows, and then draw arrows on two diagrams. The teacher, a participant in the LTP bilingual in-service education programme, briefly introduced the task, reminding learners that they could use any language. The excerpts below come from one long, focused exchange in which learners tried to determine the nature of the task set out by the textbook and to identify appropriate answers. At the same time, they continually positioned themselves and others with respect to the epistemic order.

As was often the case, Ensha moved transmodally between the English textbook, the diagrams, isiXhosa, and features of an urban vernacular register, simultaneously deploying various embodied meaning-making strategies: in this way, register meshing (Gibbons 2006) involved complex forms of multilingual trans-semiotizing (Lin 2019; McKinney and Tyler 2019). Our focus here is on the ways in which Ensha simultaneously negotiated epistemic authority, constructed her peers as knowers, and promoted solidarity. For reasons of space, the main focus is on Ensha but uptake by other members of the group is central to the analysis.

In what follows, we use the resources of IS and Appraisal theory to analyse how Ensha’s epistemic and affective stances are accomplished interactionally and multisemiotically.

Language transcription key

• textisiXhosa¹²

• textAfrikaans

• textEnglish

• texttranslation from isiXhosa

• texttranslation from Afrikaans

For more transcription conventions, see Appendix 1 in Supplementary Data.

Extract 1.

Turn	Name	Recorded data	With translation
12	Ensha	(Reading the textbook placed so all can see) What does the economic cycle flow walk through? ¨Hayi! I don’t know ↑mna uNumber one↑mna (3) ¨↓Hayi bo!	What does the economic cycle flow walk through?No way!I don’t knowas for me, numberoneme (3) No, man!
13	Zozo	Ebesxelele nje uMiss ngoku ebesithi=	Miss just told us now, she said=.
14	Ensha	=Hey, hey, he::y, uNumber one.

Ensha’s opening move in Extract 1 focused the group’s attention on the task. Both Zozo and Thulani had previously stated that they did not know how to go about it. Remarkably, she claimed not to know, although she did, thus allowing Zozo to provide information in response (13). As Heritage and Raymond (2005: 34) point out, ‘under conditions where both speakers have putatively equal access to a referent state of affairs, first speakers may downgrade initial assessments […] while second speakers respond with declaratives. These two practices cooperate to cancel the epistemic implications of the first and second positioned status of their contributions’.

Exclamatives such as hayi, hayi bo, and hey are typically used to encode a judgement or evaluation of events (cf. Eggins and Slade 1997: 89). Here Ensha strongly foregrounded her own lack of knowledge, aligning herself with Thulani and Zozo’s epistemic stances and setting the epistemic frame as one of inquiry. The exclamative fronting of hayi and then the use of the emphatic pronoun mna in both left and right dislocated positions (as for me, me) had an intensifying effect. As Moore and Snell (2011: 97) point out, pronoun tags can ‘fulfil interpersonal functions which go beyond emphatic expression of referential meaning’ and can play a role in maintaining group norms, here setting the tone as one of informality. The choice of isiXhosa as code here also carries an affective charge, lightening the key.

In this case, downgrading her own epistemic status served to put Ensha on the same footing as others, consistent with a concern to safeguard and promote the interests of those involved. ‘I don’t know’ (12) functioned here as a ‘prepositioned epistemic hedge’ or ‘forward-looking stance marker’ (Weatherall 2011: 317) that washed over the interaction as a whole. In addition to its epistemic effect, the pragmatic motivation behind the production of I don’t know is often a concern to save the face of self and others (cf. Tsui 1991). This stance marker thus had both epistemic and social effects. The repetition of ‘hayi bo’ at the end of the turn and her exclamation in (14) increased the epistemic force so that the whole turn was saturated with doubt.

This epistemic stance taken by Ensha is repeated throughout the interaction using a variety of semiotic and interactional strategies, as further discussed below.

Extract 2.

?

21	Ensha	(animating the teacher’s instruction again) Describe something that is flowing and give an example.
22	Zozo	¨Mna andiyazi le, andiyazi le.	As for me, I don’t know this one, I don’t know this one.
23	Ensha	Khawujonge ((not clear)) People that are working are making the flow of money. Makabhale ‘flow of money’.	Have a look. ((…)) People that are working are making the flow of money.We should write‘flow of money’.
24	Thulani	Yenza ‘flowing of money’.	Do it‘flowing of money’.
25	Ensha	Okay.
26	Thulani	Describe kengoku.	Describethen
27	Ensha	(Checking the instruction again) Describe something that is flowing and give an example. Imali iyangqinelwa?	Describe something that is flowing and give an example. Does everyone agree with money

After some further debate on whether to ask the teacher for clarification, the interaction continued in Extract 2 with Ensha re-animating the teacher’s instruction (21). This move had the effect both of refocusing the group and opening the floor for others’ input.

When Zozo claimed a lack of knowledge, Ensha pointed to a diagram on the next page using a directive ‘Have a look!’ (23) to draw the others in. Here the pragmatic purpose of the directive is a focuser/attention getter rather than a command: it opened an ‘interactive focus on speaker-provided information’ (Schiffrin 1987: 267). She then suggested they should write ‘flow of money’ as a solution (23).

Although the high obligation modal ‘should’ would seem to be a marker of superior epistemic status, this modal formulation still opens up the dialogic space to alternatives. It ‘explicitly grounds the demand in the subjectivity of the speaker—as an assessment by the speaker of obligation rather than as a command’ (Martin and White 2005: 111). The implicit directive is thus construed as contingent and acknowledges that the speaker is a participant in a dialogic exchange (Martin and White 2005). The use of ‘we’ rather than ‘you’ further mitigated this directive, formulating it as a joint course of action.

Thulani immediately agreed and prodded Ensha to move on and describe it (24–26). Before doing this, Ensha questioned whether everyone in the group agreed (27), once again constructing the others as of equal epistemic status.

However, in Extract 3, Zozo made a repeated counter-suggestion of ‘rain’ (28), animating the teacher’s example and indicating the drawing in the textbook of what appears to be a cloud dripping onto a hilltop.

Extract 3.

28	Zozo	Rain. Rain.
29	Ensha	Okay kaloku rain. Uyaphambana ngo! Uzakuqala ngo activity two? Ngu activity lona.	Okaythen, rain.You are mad! Are you going to start withactivity two?This isactivityone, this one.
30	Zozo	Ndithetha le kaloku mna,¨nantsi (pointing at the text).	I am talking about this one, me,¨here.
31	Ensha	(looking at Thulani and asking for his take on the suggestion). Khawuthethe¨wena	¨You speak
…
36	Thulani	Describe something that is flowing and give examples (still reading the question)(5) (looking at the next page, pointing at different examples) Imali (1) imali (1) imali(1) apha kufunwa into e-economic.	Describe something that is flowing and give examples Money, money, money, here they want something economic
37	Zozo	Bhala ‘rain’, rain is flowing.	Write‘rain’, rain is flowing
38	Ensha	(looks up at Zozo, a smile in her voice) Uzakuyibhala wena wedwa lo rain is flowing.	You are going to write it all by yourself thatrain is flowing
39	Thulani	Let’s give examples.

Ensha initially expressed support for this suggestion (29)—‘Okay fine’—a minor clause that ‘encodes a following-up reaction, and typically position[s] the speaker as a compliant supporter of prior interaction’ (Eggins and Slade 1997: 95). Then, inscribing one of only two explicit judgements in the Extract, she exclaimed ‘You are mad!’ (29 in isiXhosa), casting doubt on Zozo’s epistemic capacity, but keying it as a joke through a light tone (see also 38, 84, 52). Moreover, even this light-hearted appraisal was swiftly modulated in the next move when Ensha suggested that Zozo could be looking at the wrong activity. Zozo defended herself by pointing at the textbook whereafter Ensha, rather than overtly disagreeing further, changed footing and sought Thulani’s opinion: ‘You speak’ (31). Here she used an imperative, but the stress on ‘you’ and a falling intonation keyed it as encouragement rather than command. It functioned as a form of repair, a way round the disagreement with Zozo, but also as a way of involving all participants as producers of knowledge.

Thulani aligned with Ensha in suggesting that the answer should be something to do with money, pointing to the next page in the textbook as evidence (36). Zozo disaligned, using an unmitigated directive, signalling certainty and a strong epistemic claim—‘Write rain, rain is flowing’ (37). This directive elicited a further teasing response from Ensha that she, Zozo, would have to write that answer all by herself (38).

Thus, although Ensha disagreed with Zozo’s proposition, she invoked rather than inscribed her negative appraisal (cf. Martin and White 2005: 53), mitigating its epistemic weight by keying it as a joke. The humorous frame and the smile in her voice tempered the force carried in the strong obligation modal ‘you are going to’ and the intensifier ‘all by yourself’. Humour, ‘a semantic resource related to appraisal and involvement’ (Eggins and Slade 1997: 155), enables interactants to negotiate potential points of conflict, in this case, to make light of what was perhaps a serious point of disagreement.

Thulani (39), in what could be seen as a face-saving move, did not explicitly align with either speaker.

Extract 4.

42	Ensha	Or ↓ singenzi lonto like senze i-electricity.(10) Awuyazi wena i-electricity iqala phi izotsho apha?	Or ↓we don’t do that. Like,let’s doelectricity. (10)Don’t you know where electricity starts before it comes here?
43	Thulani	Unyanisile, unyanisile.	That’s true, true

Rather than escalating the debate, Ensha made a different suggestion: electricity (42)(Extract 4). Her use of the English ‘or’ to open the move indexes a new example of economic flow. ‘Or’ looks both backward and forward in a text, offering interactants a choice between previous or upcoming alternatives. ‘Or’ thus prompts the ‘exchange of responsibility for the maintenance of conversation’ (Schiffrin 1987: 181). Having more than one example to illustrate ‘flow’ strengthened Ensha’s position as a knower but also offered her hearers the choice as to which was most acceptable.

In addition, by realizing her countering move as a suggestion and using the inclusive ‘we’ and ‘let’s’, Ensha positioned all as involved and left the possibility for disagreement open. Together her use of ‘or’ as discourse marker, pronouns and mood choices tempered the degree of obligation (Eggins and Slade 1997: 99) and opened possibilities for engagement.

In the second part of her move, the discourse marker like is multifunctional, a marker of new information and focus as for Thulani above (D’Arcy 2017: 14–15), but here it is also discursively contrastive in highlighting a different option. At the same time, Ensha’s use of this local urban register feature established solidarity with her interlocutors and indexed a relatively low level of investment in her suggestion.

After a pause to offer a turn transition, she probed: ‘Don’t you know where electricity starts before it comes here?’ (42). Negative interrogatives are ‘strongly designed for “yes” answers’ and thus assert epistemic supremacy. Here though it functions more as a prompt, a reminder of common knowledge and negotiation of common ground. Thulani read this as such and acknowledged that they did know (43). Both agreed to work with this idea.

However, Ensha, perhaps sensing that the flow of electricity would be difficult to describe, made a further suggestion of clothes (47), elaborating the flow from sheep to shop in detail (49a,b) after Thulani agreed.

Extract 5.

47	Ensha	Or ngeyempahla siyenze nge mpahla (5) Xakusenziwa impahla kuthathwa ilaphu, imaterial.	Orclothes, let’s use clothes. When clothes are made, Cloth, material is taken.
48	Thulani	Ewe.	Yes.
49a	Ensha	Mos kuthathwa iship13mos né?? (short untranscribed aside about a boy at the next table)	So they take a sheep, right, don’t they?
49b		Nenze ijezi ngelantuka nge (2)animals(1)izilwanyana. (3) Then kengoku bazise eshop eshop kengoku zithengwe. That is a flow leyo né? Iyafana ngathi yiflow?	They make a jersey with something, with(2) animals (1) (translatesanimalsinto isiXhosa). (3) Thenthey are taken to shops where they are sold. That is a flow, that, right? It is just like ↑flow.

Once again signalling a new option through ‘or’, Ensha used a range of strategies to maintain a delicate balance of epistemic status and social solidarity. In 49a, she opened with the discourse marker ‘mos’, which carries the sense of ‘as we know’. Mos indexes shared knowledge. It thus both constituted her interactants as equal in epistemic status and strengthened the participation framework. Its repeated use after the main clause, followed by the Afrikaans tag né, (isn’t that so?) augmented her other-focused stance. The tag né both seeks agreement and signals readiness to relinquish the turn, downgrading her asserted rights and tempering the degree of investment in the proposition.

In 49b, after describing the flow, she ended with a further double tagged declarative, ‘that is flow, that, right?’ (49b), both tags, one in isiXhosa and one in Afrikaans. The right dislocated demonstrative tag, ‘that’, confirmed and clarified the referent and also emphasized her contribution (Moore and Snell 2011: 102). This stance is once again modulated by ‘right’ with an upward inflection, checking understanding and seeking alignment with her proposition.

Epistemically, the pragmatic effect is therefore one of seeking confirmation rather than stating a position. It reduces the authoritativeness of the claim, leaving open the option of disagreement. Nevertheless, although this tag is in principle expansive and hearer-focused, it generally assumes that others will align. In fact, Ensha does not pause for others to respond but repeats the proposition, this time as a declarative ‘It is just like flow’. ‘Just’ acts as an intensifier, strengthening her stance, augmenting her claim.

Such a declarative in some frameworks would be considered neutral and monoglossic, as encoding the speaker’s right to know and to assert what is being declared, rights which are grounded in the assumption that the speaker knows something that the others do not (Heritage 2009). However, such a ‘bare’ assertion can also be associated with the opposite assumption that speaker and audience operate with the same knowledge, beliefs, and values (White 2003), a heteroglossic construal which is more in line with a participation framework geared towards solidarity.

Thulani and Zozo both accepted this idea. Ensha pretended to be taken aback (52), rounding her eyes and exclaiming in isiXhosa with mock surprise:

Oh, bethunani, ndiyakwazi awuzikuvuma wena!

‘Oh,come on, I know you are not going to agree!’

The epistemic particle ‘oh’ functions as a contextualization cue, priming the hearers for a change of state (Heritage 2002), a signal in this case of an upcoming ‘nonserious’ thought which ruptures the expected sequence/action structure. It also indicates an ‘upcoming emotional or evaluative utterance’ (Tree and Schrock 1999: 281), which in this case invoked disbelief, an epistemic stance which is intensified by the exclamative form of the move. Here, Ensha made an ironic metapragmatic comment on their respective roles in the epistemic order as enacted during the interaction. By this use of humour, she implicitly accorded her interactants equal epistemic status.

In the rest of the interaction, Ensha continued to try and determine exactly what they are expected to do. After the teacher asked: ‘Where are you, people?’, Ensha (87) remarked seemingly out of the blue:

Bendizakuthi kuThulani makahambe ayobuza kuMiss ababantu bacheba igusha uba zintoni xazibizwayo sizokwazi ukuthi zimanufacturer.

(I was going to say to Thulani he must go ask Miss what those people that shear sheep are called so we can call them manufacturers).

This is a clear upward shift from the specific example they have been discussing towards a more general abstract proposition with greater semantic density (Maton 2013). Within SFL this would be an example of a move towards grammatical metaphor in which processes are turned into things, especially important in disciplinary knowledge. This is a model use of nominalization to build an argument where greater abstractness and technicality project greater epistemic authority.

There is no take-up as the learners move on to drawing arrows on the picture.

6. CONSTRUCTING KNOWERS THROUGH NEW RELATIONS OF KNOWING

As we have shown above, Ensha made use of a wide range of inclusive, hearer-focused, and dialogically expansive strategies which simultaneously worked to modulate her epistemic authority, legitimize others as knowers, and promote solidarity.

In her appraisal of others’ epistemic contributions, two features stand out. First are the ways in which she turned potentially contractive, categorical assertions into heteroglossic opportunities for engagement, for example, through code and/or register meshing, mood choices, option markers such as ‘or’, hedging, tags, prosody, gesture, gaze, eyebrow lifts, and other embodied resources. In these ways, utterances that could be construed as challenging or fending off alternative positions (dialogically contractive) were rekeyed to index an openness to alternative positions (dialogically expansive) (cf. White 2003).

Second is the marked absence of explicit realizations of appraisal. Rather, appraisal is construed implicitly, invoked rather than inscribed, and mitigated through humour. Similarly, there is no evaluative lexis in relation to knowledge contributions (except ‘mad’), which is swiftly downplayed. Also lacking are explicit discourse markers of concession or counter-argument and modals of certainty. Minimising overt appraisal in this way works against hierarchical relations of knowledge. Along with the almost complete lack of interruptions or overlaps, and a lack of force or intensity in constructing epistemic stances, these absences served to maintain relations of solidarity and strengthen a participation framework in which all had equal status as knowers.

The dialogic relationships created in this way embody an emergent and decolonial ethics of knowing, one premised on solidarity, respect for others, and open-mindedness. Such values are associated with the relational ethics of, for example, ubuntu (Etieyibo 2017: 313), a Nguni Bantu term meaning ‘humanity’, sometimes translated as ‘I am because we are’. Ensha’s multisemiotic practices were central to the construction of this ethical epistemic order (see also Makalela 2016; Ndhlovu and Makalela 2021 on ubuntu translanguaging).

A further significant aspect of Ensha’s practices is that the most extended use of the schooled register is in isiXhosa, not English, evidence of the fact that she was able to resemiotize knowledge acquired in English the previous year into isiXhosa, without support, thus belying the belief that teachers are unable to use home languages for content subjects without specific training. It is also worth noting that Ensha’s turn 49b is an example of presentational talk in isiXhosa—a fluent explanation using a discipline-specific register. In other turns, her ‘schooled’ register was characterized by use of the passive form, complex and compound sentences, increased semantic density and abstraction. Here lie the flashes of knowledge that Ensha possessed but that was rarely signalled overtly.

It is thus striking that Ensha uses isiXhosa for all significant knowledge-building work (elaborating, exemplifying, evaluating). There is only one instance of direct translation (49b), the most ubiquitous practice in translanguaging research. A large body of recent work demonstrates the potential of translingual practices for more inclusive and equitable education. Yet, while often valuable in legitimizing multilingual resources and enhancing access to knowledge, translanguaging in formal school settings is for the most part intended to achieve a monolingual outcome. It remains a route to knowledge in an official language (see also Heugh 2015, 2021). As a function of this monolingual episteme, evidence of translanguaging tends to be limited to single words, phrases, or short sentences. Very often these are translations of concepts or instructions, ‘reproductional translanguaging’ (Probyn 2015; Krause and Prinsloo 2016) rather than knowledge-building practices (but see Duarte 2019; Krause 2021). This ultimately unidirectional impetus means that translanguaging remains an affirmative rather than transformative strategy, tweaking rather than transforming institutional arrangements. The ‘new reality’ (García and Leiva 2014) that emerges is fragile and fleeting: ‘a surface level-change in practices to redress an injustice based on what is, in effect, ethnolinguistic racism’ (Block 2018: 15). It leaves underlying colonial hierarchies of value and relations of knowing unchanged (Stroud and Kerfoot 2013, 2021; Heugh 2021). Without broader changes in institutional ideologies and structures, translanguaging will remain ‘a multilingualism of monovocality’ (Stroud and Kerfoot 2021: 29), promoting epistemic access rather than epistemic justice.

As demonstrated here, multilingualism when unmarked, that is, understood as a norm and taken for granted as the basis of everyday interactions, holds the promise of more egalitarian, just, and ethical conditions for engaging Others.¹⁴

7. CONCLUSION

The paper aimed to investigate the potential of multilingualism both ‘as a resource and a transformative epistemology and methodology of diversity’ (Stroud and Kerfoot 2013: 396, 2021). We have argued that the lack of opportunity to learn in a familiar language at school beyond Grade 3, if at all, constitutes a persistent and systematic form of epistemic injustice, ‘a wrong done to someone specifically in their capacity as a knower’, a capacity which is key to human value (Fricker 2007: 1; see also Kerfoot 2014; McKinney 2015). By contrast, the paper suggests that Ensha’s social and pedagogical practices constitute an emergent project of epistemic justice, defined here as an ethical project of reversing epistemic exclusions, mitigating epistemic harm, and seeking parity of epistemic authority for linguistically marginalized speakers and knowers.

Previously, in her first year in the English-medium Grade 6 class, Ensha was not recognized as a ‘knower’—she was silenced and mocked by her peers, and her access to knowledge was reduced. She was thus wronged in her capacity as a subject and producer of knowledge, a form of epistemic injustice, a denial of part of what it is to be fully human (Fricker 2007; Veronelli 2015). However, in her second time in Grade 6, by spontaneously taking on the role of knowledge mediator or facilitator, she erased the possibilities of these experiences for others—an example of what Maldonado-Torres (2007) calls Fanon’s ‘receptive generosity’: ‘Was my freedom not given to me then in order to build the world of the You?’ (Fanon 1967: 231–232).

While her pedagogical practices could be seen as incorporating both instructional and regulative elements in Bernstein’s (1996) sense, we suggest that the ways in which she constructed relations of knowing represents a different model of learning, one based on different assumptions and values, a dialogic and collaborative process of engagement with knowledge which is deeply ethical. This example of ‘alternate ways of knowing and being human’ (Santos 2014) was only made possible by a radical shift in classroom and school language ideologies, unmarking multilingualism and subverting the normative order.

This study has thus shown an example of a ‘multilingual’ episteme in action in a formal classroom context (see also Kerfoot and Bello-Nonjengele 2016). It has shown how policies and practices that embrace a multilingual episteme, giving equal value to all languages and linguistic varieties as epistemic resources, can point the way to new relations of knowing and new paths to knowledge.¹⁵ These dynamic, emergent practices are forms of linguistic and epistemic citizenship (Stroud 2001, 2018; Stroud and Kerfoot 2021). They have the potential to interrupt hegemonic regimes of language and learning, begin the long process of undoing epistemic harms, and promote a decolonial ethics of care. Using a southern, decolonial lens that assumes multilingualism as a starting point and learners as agents of social and cultural production, can ‘produce different knowledge and produce knowledge differently’ (Lather 2013: 635). It can illuminate the emergence of politically fragile yet institutionally robust social, epistemic, and moral orders from below, orders that can lay the basis for conditions of greater epistemic justice.

Footnotes

1

A familiar language is understood as a language spoken at home and/or more broadly in everyday interactions. This term includes home and family languages but recognizes that, in many highly multilingual contexts, education in a language used at home is not possible but education in a wider community language, familiar to children and spoken outside the home, is.

2

South and North are understood metaphorically: ‘a South […] also exists in the global North, in the form of excluded, silenced and marginalized populations’ (Santos 2012: 51). For this reason, south and north are not capitalized.

3

While a home language advantage is context-sensitive (Nag et al. 2019: 1) and influenced by other dimensions in the home language and literacy environment, for example, books at home and adult literacy practices, a global majority of learners face the learning conditions outlined here.

4

There are other deeper, interconnected forms of epistemic injustice which Fricker (2007) calls ‘hermeneutic injustice’ and which are closer to the epistemic harm identified by Mignolo (2009) and Ndlovu-Gatsheni (2018). These forms have to do with the dominant, often northern, interpretative frames that can obscure ‘significant aspects of [individuals’] social experiences from collective understanding’ (Fricker 2007: 154). While the processes identified in this paper point to the potential for reversals at this deeper level, their investigation is beyond the scope of the paper.

5

The languages spoken in Cape Town exist in many forms and carry complex and shifting indexicalities, entangled with histories of colonialism, race, power, and resistance. ‘Standardized’ varieties of isiXhosa, Afrikaans, and English are taught in schools, with English as the main language of schooling. Various urban vernaculars continue to evolve, combining features of different languages to suit speakers’ purposes. Histories of migration mean that many other languages are also spoken.

6

This paper focuses on multilingual educational contexts where the majority of learners are required to learn in a language they do not speak at home. The goal in such contexts is monoglossic, that is, proficiency in the official language only. A small body of research on learners as knowers has been carried out in two-way dual language contexts (e.g. Gutiérrez et al. 1999; García-Mateus and Palmer 2017) but these contexts differ in that both languages are valued and developed in the school.

7

In England this figure is 3 per cent, in the USA 4 per cent, in Chile 13 per cent, in Iran 35 per cent (Howie et al. 2017: 55).

8

We use the terms second and additional language in line with South African educational terminology. They do not capture the fact that many learners come from bi- or multilingual homes. Similarly we use ‘named’ languages and racial categorizations in the ways that those who use them, including the state, do. These terms are contested, labile signifiers.

9

Author 1 is a White South African and so very visible in this school. In order to minimize the potential for bias, she observed from corners of playgrounds and classrooms. Moreover, because of the previously authoritarian school regime associated with whiteness, she played no part in interviews or classroom recordings. Nor did she have any direct engagement with participants, except when initiated by them. Author 2 is a Black South African of Nigerian origin who spent many years in an isiXhosa-dominant region of South Africa and communicates well in this language. After five years, she was very familiar to teachers, learners, and many families.

10

The last census was in 2011.

11

Under apartheid, the designation ‘coloured’ was a category constructed for all those of ‘mixed’ heritage, including descendants of Indonesian and Malay slaves as well as the Khoe-San. Because it incorporates a number of culturally distinct groups, the word is generally written today with a lower case ‘c’. Black, like White, can be capitalized when indexing a political stance; in such cases, it often incorporates people classified as Indian or coloured. At other times, the lower case is used to avoid essentialization. The present government retains the former apartheid racial categories for policies of redress and equity.

12

We have chosen to make isiXhosa the unmarked language to signal its function as the primary epistemic resource.

13

Phonetic transcription of the pronunciation of sheep in a South African English variety. IPA:/ʃɪp/

14

The importance of multilingual practices as the central mediating tool has long been highlighted by scholars in the South (Agnihotri 1995, 2014; Heugh 1995, 2021; also Gutiérrez et al. 1999 in dual language classrooms). Others have drawn attention to the need to engage with teachers, caregivers, and communities to enrich understandings of the value of multilingual resources (see e.g. Chimbutane 2020; Guzula 2021).

15

It is crucial to note that this shift in the forms of language considered legitimate for learning does not exclude acquisition of the ‘standard’ form of one or more national or official languages. Both goals form part of ‘functional multilingualism’ (Heugh et al. 1995; Heugh and Stroud 2020; Van Avermaet et al. 2018).

Acknowledgements

We are indebted to Kathleen Heugh, Christopher Stroud, David Karlander, Margie Probyn, and Ben Rampton for invaluable comments and continued inspiration. We also thank three anonymous reviewers for helping to strengthen the paper. Many others have contributed profoundly to our knowledge and understanding on this project, notably, our collaborators on the Language Transformation Plan at the Project for the Study of Alternative Education in South Africa (PRAESA), Neville Alexander, Xolisa Guzula, Ntombizanele Mahobe, and at the University of the Western Cape, Vuyokazi Nomlomo and Peter Plüddemann, as well as Ruth Versfeld and, of course, the young participants and gifted Grade 6 teacher whose classroom made this paper possible. We are grateful to Shabiera Khatieb Patel, Ferial Gafieldien, Kashifa Janodien, Nathalie Hattingh, Lonwabo Haya, Bulelani Teteni, and Someka Ngece for transcriptions and translations, and to Sivuyile Lhoza and Sibonile Mpendukana for rechecking key sections at short notice. Finally, we salute the visionary leadership of the then Western Cape Minister of Education, Cameron Dugmore, and Anne Schlebusch, special advisor to the Language Transformation Plan in the Western Cape Education Department.

Funding

This research was supported by the National Research Foundation (NRF) of South Africa under grant 62314, the Swedish Riksbankens Jubileumsfond (RJ) Project no. SAB17-1020:1, the Faculty of Humanities at Stockholm University, and the Wallenberg Foundations through a Fellowship at the Stellenbosch Institute for Advanced Study (STIAS). We thank them for making this research possible.