Exploring the injustices perpetuated by unfamiliar languages of learning and teaching: the importance of multi-angle, learner-focused research

This paper argues for the importance of foregrounding learners’ experiences in language-in-education research, particularly in sub-Saharan Africa and other postcolonial contexts where there is an unfamiliar language of learning and teaching. Standing firmly on the shoulders of decades of research that compellingly demonstrates a range of ways in which the use of an unfamiliar language is detrimental to classroom practice and learning outcomes, we suggest that there are yet further negative consequences that are currently under-researched. We argue that combining observation of learners with methods that create space for learners to explain their experiences in their own words enables important new insights into how epistemic injustices intersect with broader structural injustices in learners’ lives. Our proposition is informed by our work and research in a variety of contexts but draws most heavily from qualitative research conducted with young people in primary and secondary schools in Tanzania, Rwanda and South Africa. Our conclusions demonstrate how learner-focused research could importantly and beneficially extend the evidence base that is available to support calls for changes to language-in-education policy and practice.


Introduction
Globally, millions of children are learning using an unfamiliar language of learning and teaching (LoLT) that is not widely used in their everyday lives (UNESCO 2016, World Bank 2021).This is most common in sub-Saharan Africa and other postcolonial contexts where dominant languages, such as English, often persist as a continuation of colonial language policy and are viewed as central to global economic competitiveness and development (Tikly 2016).These language-in-education policies have been widely criticized for positioning children who struggle to learn through an unfamiliar LoLT as somehow deficient -either linguistically or cognitively (McKinney 2017, García et al. 2021, Phyak and Sah 2022, Kalyanpur 2022).However, the Project for the Study of Alternative Education (PRAESA), under the directorship of the late Professor Neville Alexander and, in collaboration with director Nicki Westcott, neatly turned this perception on its head.It produced a DVD (Westcott 2004) which combined learner interviews with classroom observations to powerfully show the significant communicative and emotional challenges that White, English-speaking Grade Six learners at an affluent private school in Cape Town faced when they were taught a single Science lesson in isiXhosa, a language they were learning for conversational purposes only.While this one research study provoked debate and discussion in South Africa about the unfair impact of learning in an unfamiliar LoLT, this paper is borne from shared frustration that learners across sub-Saharan Africa, and beyond, continue to struggle and suffer from language policies and practices that are 'immoral' (Spolsky 1977, p. 20) and unjust.
This paper makes the case for centring learners' experiences in research about the use of an unfamiliar LoLT, particularly in sub-Saharan Africa and other postcolonial contexts.Existing language-ineducation research has compellingly demonstrated a range of challenges faced when using an unfamiliar LoLT, particularly highlighting difficulties in communication and understanding (Brock-Utne et al. 2010, Ouane and Glanz 2011, Milligan et al. 2020).Alongside this, there have been initiatives that demonstrate what learners are capable of when enabled and supported to use a familiar language in learning (Nomlomo 2009, Langenhoven 2010, Desai 2012, Makalela 2015).However, we argue that there is a need to develop a more comprehensive evidence base about the impact of language in education that recognizes learners' knowledge and agency.Moreover, we assert that this shift in focus is necessary for research to fully attend to the multiple ways in which the persistence of language policies that prioritize unfamiliar LoLTs is a significant source of global injustice.Although arguments about the injustice of the predominance of ex-colonial languages are powerfully made in debates about language-in-education (Brock-Utne 2012, Tikly 2016), many of these critiques are mainly focused at the macro-level and remain disconnected from the evidence gathered at the micro-level in classrooms.We are concerned that this may limit the potential for justice-based arguments to affect policy and practice, and thus we seek ways to connect evidence and theory to strengthen the call for change.
In advocating for a foregrounding of learners' experiences in LoLT literature, we also argue that we must take a multi-angle approach.We make this case on methodological and theoretical grounds.Methodologically, we argue for the use of multiple research methods that enable an observational perspective to be combined with learners' explanations of their own experiences.Theoretically, we argue that this combination of methods is necessary to expose the multiple layers of injustice that young people face when learning in an unfamiliar LoLT.We draw on Fricker's (2007) discussion of epistemic injustice -in particular, the concepts of hermeneutical and testimonial injustice -to demonstrate how some forms of inequity and oppression may be hidden if we approach classroom experiences from a single angle.Moreover, we highlight ways in which this multiangled approach enables us to see how LoLT-related injustices intersect with, and in some cases exacerbate, existing structural injustices in children's lives (Anderson 2012, Milligan 2022).We present this argument using examples from two studies in Rwanda and Tanzania but build upon many years of combined experience in designing and conducting LoLT research, including Desai's co-leadership of the Language of Instruction in Tanzania and South Africa (LOITASA) Project between 2002and 2011(Brock-Utne et al. 2010, Desai 2016).

Literature review
We are not alone in calling for more research that centres learners' perspectives and experiences.Nor is the study of language-in-education the only area where we, and others, have significant concerns about learners' experiences of education being overlooked.We join a range of colleagues in their criticism of approaches to education research and improvement that focus on measurement of learning outcomes (narrowly understood as literacy and numeracy scores) and standardized, decontextualized approaches to teacher effectiveness (Barrett 2011, Alexander 2015, Mitchell and Milligan 2023, Schweisfurth 2023).Instead, we look to examples of work that demonstrate that focusing on learners' perspectives can offer much richer understanding of the wider context that learners are trying to navigate, and the range of challenges that impact upon their capability to engage with education (Jones 2011, DeJaeghere 2017, Anangisye 2020, Buckler et al. 2022).These broad, multi-dimensional perspectives are particularly important if we are to understand how different individual and structural factors intersect either to support young people's aspirations, or to form 'a net of intermeshing structures of exclusion' (Unterhalter 2021, p. 159).However, these kinds of broad, multidimensional, learner-focused approaches have been much less present in language-in-education research in postcolonial contexts, perhaps at least in part due to the immediacy of concerns about communication challenges in an unfamiliar LoLT and the clear implications for classroom practice.

Learners and LoLT research
Repeatedly, research has found that the use of an unfamiliar LoLT poses a significant obstacle to classroom communication, limits access to curriculum content and, thus, has a negative impact on learning outcomes (Qorro et al. 2008, Ouane andGlanz 2011).A significant amount of the research in this area has relied heavily on classroom observation and analysis of classroom interactional patterns, with a focus on teacher strategies and perspectives (Afitska et al. 2013, Webb and Mkongo 2013, Ssentanda 2014, Sibomana 2022).For example, studies have identified that rote learning, 'safe-talk', and teacher-controlled code-switching predominate when learners and teachers are struggling to negotiate learning through an unfamiliar language (Mwinsheikhe 2009, McKinney et al. 2015, Kiramba 2019).In fact, study after study highlights the restricted nature of learners' contributions when an unfamiliar LoLT is used (Vuzo 2010, Ngwaru 2011, McKinney et al. 2015, Kiramba 2018, Westbrook et al. 2023).By contrast, in multilingual classrooms where a familiar language plays a significant role, studies have found much greater participation of learners (Lavoie 2008, Qorro et al. 2008, Guzula et al. 2016, Kiramba 2019).This has led to initiatives to develop pedagogies that include familiar languages in classroom discussion (Probyn 2015, Makalela 2019, Vaish 2019, Erling et al. 2021).
There are two main areas of LoLT research that more actively involve learners beyond observing their classroom participation.In the first group of studies, learners have been engaged in languagebased assessments or tasks.The findings from these studies have powerfully and usefully illustrated that learners' knowledge of the unfamiliar language is insufficient for the language demands of the curriculum or available textbooks (Roy-Campbell and Qorro 1997, Desai 2012, Rea-Dickens and Yu 2013, Barrett et al. 2014, Mligo and Mwashilindi 2017) or to allow them to express themselves and demonstrate their understanding (Brock-Utne and Desai 2010).However, as the goal of most of these studies has been to gather robust evidence of the severity of the language problem, the focus has been on task-performance rather than learners' perspectives.
There is another group of studies that engage learners' perspectives through questionnaires, interviews and focus groups, but the focus is predominantly on language attitudes and preferences (Rubagumya 1989, Mafela 2009, Dyers and Abongdia 2010, Nyamubi 2016, Getie 2020).Although some of these studies include opportunities for learners to voice their opinions at length, a lot of the questioning is much more restrictive, and the interpretation of data often lacks nuance.Learners' perspectives have been labelled as 'perplexing' and 'contradicting', as young people report that they struggle to use the unfamiliar LoLT, whilst simultaneously expressing a preference for retaining that language in its current role (Senkoro 2005, p. 12, Kinyaduka and Kiwara 2013, p. 94, Davis et al. 2015).Although some researchers have pointed to this data as evidence of the power of language ideologies that determine language-related values (Desai 1999), there can be a tendency for such contradictions to mean that learners' perspectives be dismissed.Rarely has research sought more in-depth explanations from learners themselves about the reasons why answers to different questions might appear to be in conflict with one another, or with researchers' observations of classroom experiences.
Although not the first study of the role of the LoLT in sub-Saharan Africa, perhaps the most comprehensive research thus far has been the LOITASA study, conducted in South Africa and Tanzania in two phases: 2002-2006and 2007-2011(Brock-Utne et al. 2003, Qorro et al. 2008, Brock-Utne et al. 2010, Desai et al. 2013).The project took a quasi-experimental approach that compared subject lessons taught, and supported by learning materials, in English and familiar languages (isiXhosa in South Africa and Kiswahili in Tanzania).In South Africa it took the form of a longitudinal study looking at learners from Grade 4-6.Classroom observation data were collected alongside data from learner tasks and progress was monitored longitudinally.In addition, interviews were conducted with teachers, learners and parents.These interviews played a significant role as learners and their parents recounted their experiences of using a familiar language as the LoLT.In particular, these perspectives highlighted that the use of English as the LoLT presented both practical and emotional challenges for both learners and their parents, who were frustrated that they were unable to support their children with learning in English.The wealth of different forms of evidence generated through the LOITASA project has offered a valuable snapshot of learners' experiences of language-in-education, combining multiple methods to develop rich and multi-dimensional understandings (LOITASA and Westcott 2006).However, the extensive and in-depth nature of this research project has not yet been replicated elsewhere, and its findings have, unfortunately, not been taken on board by schools, policy makers or education departments.
There are a limited number of studies that recognize the relevance of students' lives beyond schooling in developing a richer understanding of the role of LoLT.For example, some studies have included factors such as learners' exposure to the unfamiliar LoLT outside of school and measures of socio-economic status in their quantitative analysis of learners' school-based achievements (Smith 2011, Rea-Dickens and Yu 2013, Nyamubi 2019).Although they have identified important relationships between these factors and learners' academic performance, the nature of the research design and analysis has not allowed for more in-depth exploration of individual circumstances or cross-referencing with specific learners' experiences in classroom observations.At the other end of the methodological spectrum, there are a small number of studies, including several ethnographies, that include a range of qualitative methods to develop a richer, socially-situated picture of learners' experiences (Hamid 2010, Ngwaru and Opoku-Amankwa 2010, Chimbutane 2011, Bhattacharya 2013, Tamim 2021, Speciale 2022, Kuchah and Milligan forthcoming).Through these studies we are often introduced to individual learners, offering the opportunity to view young people's classroom encounters alongside information about their home lives and sometimes also in conjunction with insights into their imaginations and aspirations.This type of research also often highlights inequalities and injustices, as the widened gaze encompasses multiple dimensions of learners' experiences as well as acknowledging the local, national and global forces that shape the study context.

Foregrounding learners' experiences and epistemic justice
When we approach the issue of language-in-education using the lens of epistemic justice, it highlights that the inclusion of learners' voices alone, however creatively and carefully we design the space for those voices, may not be enough to enable us to identify and understand all the forms of oppression that learners experience.This paper uses the related concepts of testimonial injustice and hermeneutical injustice to argue for combining research methods that foreground learners' accounts with observation-based methods that may enable us to see examples of oppression that learners would not, or cannot, label as injustices.We see the lens of epistemic justice to be particularly important given the broader complicity of language-in-education policies in the 'epistemic monoculture' (Masaka 2019) of post-colonial schooling systems and the linguicide (Ngūgī 1986(Ngūgī , 2009) ) of languages in countries across the Global South.As Esch (2011, p. 235) explains, the consistent promotion of English and French as the medium of education in Cameroon led to a situation where 'the concomitant assumption that local languages -and their speakers -were essentially deficient constituted a case of injustice of an epistemic nature'.
This points to the essential role of language in the processes of testimonial injustice, in which a hearer determines a speaker to lack credibility based on their prejudice toward the identity or communication mode of the speaker (Fricker 2007, McIntosh andWilder 2023).There is some evidence that the communication mode for learners and the 'credibility marker' of confidence when speaking the unfamiliar LoLT is a source of testimonial injustice in classrooms in Ghana and Tanzania where teachers equate the learners who are able to speak more confidently in English with those being academically more able (Opoku-Amankwa 2009, Walker 2020).This in turn may mean that these learners get more opportunities to speak and practise the unfamiliar LoLT and by contrast, other learners are silenced, marking them as 'unreliable sources of knowledge' (Masaka 2019, p. 302).Multilingual classroom practice, by contrast, is offered as a corrective as learners have the opportunities for their pre-existing knowledge(s) and experiences to be recognized and used in classroom learning (Milligan 2022, Kerfoot andBello-Nonjengele 2023).
The questioning of an individual's credibility also relates to hermeneutical injustice whereby an individual may start to doubt their own knowledge.As Fricker (2007, p. 162) explains, this puts someone at an unfair disadvantage for making sense of their own experiences and making something 'communicatively intelligible'.Walker et al. (2020, p. 80) further suggest that 'unequal participation in generating social meanings' contributes to 'hermeneutic marginalization of a person or group and exposes how dominant discursive resources fail (or refuse) to comprehend the experiences of the oppressed' (80).Within the context of learning in an unfamiliar LoLT, the 'dominant discursive resources' prioritize dominant, 'global' languages and position the 'fallacy' that the best way to learn a language is to use it as the sole LoLT as an 'undeniable "truth"' (Phillipson 1992, p. 185, Brock-Utne 2012, p. 787).Against these powerful narratives, young people may have limited discursive resources for making sense of their struggles with the LoLT.There are, thus, clear implications for research that simultaneously seeks to centre learners' testimonies as credible sources of knowledge while also critically engaging with, and challenging, language-in-education policies and practices as powerful sources of injustice.
We also argue that the value of an epistemic justice lens can be heightened by placing it within an explanatory framework of broader patterns of structural injustices, for example, in relation to gender, race and socio-economic disadvantage.We see this as particularly important given that one impact of the focus on teachers and classroom dialogue in LoLT research has been that learners' classroom participation (or silence) has been decontextualized from their broader educational experiences and life circumstances.Anderson (2012) suggests the importance of considering how initial structural injustices -for example, access to education -extend the potential for marginalization on epistemic grounds.Kerfoot and Bello-Nonjengele (2023, p. 5) similarly argue that an initial linguistically-based epistemic injustice 'lays learners open to long-term trajectories of economic and socio-political exclusion and disadvantage, along with reduced confidence in their own epistemic worth.'We show further in our findings that the relationships between epistemic and social injustices are not so sequential but rather a complex intersection within young people's everyday experiences of learning, both within and outside of school.This is important since epistemic injustices do not stand alone but are embedded in wider social, cultural and economic structures and practices.This in turn necessitates methods that enable understanding of the broader life circumstances of young people beyond what can be seen in the unfamiliar LoLT classroom.

Methodology
This paper draws on data from two discrete language-in-education studies that focused on different questions and used different combinations of research methods.The first study, led by Adamson, explored students' negotiations of their language environment in two secondary schools in Tanzania.The second study, led by Milligan, explored girls' experiences of English-medium education in Rwanda.Neither study was specifically designed to be 'epistemically just' in terms of its methodology (for example see Walker et al. 2020), however, both studies were informed by a commitment to ethical research, including issues relating to positionality and power, and a theoretical interest in language and social justice, understood through the lenses of the capability approach (Adamson 2021) and epistemic (in)justice and exclusion (Kuchah et al. 2022, Milligan 2022).In designing both studies, we were cognizant of the historical role that observation has played as a tool for both creating and reinforcing colonial hierarchies and injustices (Denzin and Lincoln 2005).Moreover, we were conscious of the ways in which research methods have reproduced the power of dominant, neoliberal, Western forms of knowledge (McIntosh and Wilder 2023).These were key reasons for including multiple research methods in both studies, including approaches that would foreground young people's agency and experiences.This paper, and our fuller understanding of the importance of the concept of epistemic justice in research, is the result of looking back and reflecting on these findings and the ways in which they were shaped by the design of, and subsequent adaptations to, the research process.Our discussions and thinking about the intersections between LoLT and learners' broader lives and social contexts are also influenced by Desai's work, both in research and practice, which is embedded in multilingual, disadvantaged communities in South Africa (Desai 2012, 2016, Milligan et al. 2020).
It is important to acknowledge that our argument for a multi-angled research approach that combines different methods could be understood as a form of triangulation.However, we are certainly not advocating the use of multiple methods as a technicist approach to confirming or validating research findings, one common understanding of the concept of triangulation, particularly found in mixed-method research and research within the postpositivist paradigm (Hammersley 2008, Denzin 2012, Vogl et al. 2019).Rather, we look to multiple methods as valuable for 'allowing for new or deeper dimensions to emerge' (Jick 1979, p. 604).In both studies that contribute data to this paper, we designed combinations of methods that would enable us to access richer, contextualized information about learners' holistic educational experiences.
The study in Tanzania took an ethnographic approach to explore students' experiences and negotiations of language in two secondary schools in the Morogoro Region.The first was a well-established, large, urban school that offered both lower and advanced secondary levels, Forms 1-6. 1  The second school was less than 10 years old, offered only lower secondary Forms 1-4 and was in a rural location, 35 km from town.The field research took place over 10 months and a range of methods were used to generate data.The dataset included: 51 lesson observations; fieldnotes that recorded informal discussions and participant observation between lessons and in the wider space in and around school; and 31 group interviews that included the voices of 146 students.In addition, there was a group of 10 student researchers who conducted their own interviews and co-designed and facilitated two participatory workshops.The student researchers also supported the explanation and interpretation of emerging research findings.The researcher spoke both English and the Tanzanian lingua franca, Kiswahili, and learners were free to use either language of their choice.The study was designed in line with the ethical process at the Institute of Education, University of London in 2014.It also had support from the Department of Education at the University of Dar-es-Salaam and was granted a research permit from the Tanzanian Commission for Science and Technology (COSTECH).
The data from the Rwandan study comes from the second part of a mixed-methods case study of girls' experiences of learning in English Medium Basic Education in four schools in the Burera, Kirehe, Nyarugenge and Ruhango districts of Rwanda (see also, Uworwabayeho et al. 2021, Kuchah et al. 2022, Milligan et al. 2023).The qualitative data generation took place in 2021, undertaken between 48 girls, in the last years of primary and lower secondary school, 2 and two Rwandan female researchers -Aline Dorimana and Aloysie Uwizeyemariya.The 48 girls were identified by the headteacher as broadly representative of the year groups in terms of their academic achievement, likelihood to dropout or transition to further levels of education and attendance.The six case study girls in each year group at each school were observed in four lessons through dual lesson observations with one researcher documenting the lesson content, key activities that took place, and noting the teacher's approach to inclusion and language use and the other researcher focusing on the six girls' engagement and talk in each lesson.All girls also took part in narrative interviews in their home language of Kinyarwanda.The data generation was significantly affected by the impact of Covid-19 and subsequent school closures, delayed from 2020 and with methods adapted in response to relevant risk assessments and to prioritize the time and wellbeing of both the girls in the study and the Rwandan researchers.For example, original plans to undertake photovoice research were amended to short notes that the girls took about their learning and follow up interviews where the notes were the prompts for the semi-structured discussion.The lessons were also video recorded and transcribed by Adamson in collaboration with Dorimana and Uwizeyemariya.The thematic analysis was a collaborative process between Milligan, Adamson, Dorimana, Uwizeyemariya and three further colleagues (Alphonse Uworwabayeho, Kuchah Kuchah and Terra Sprague).Original plans to co-analyse and build theoretical conclusions with the 48 girls were also curtailed by Covid-19 and funding deadlines.The study received ethical clearance at the University of Bath and the University of Rwanda.

A multi-angled approach can shift how we understand and interpret learners' experiences
In both studies, if we had relied on researcher-led observation alone, we may have misinterpreted learners' experiences.In the Rwandan study, interviews with girls revealed that several of the lessons observed were not typical of learners' day-to-day experiences in those subjects.Alice (Ruhango district) explained that she enjoyed History and Citizenship because the teacher 'teaches very well, he translates from English to Kinyarwanda or sometimes mix the two languages to help us understand well'.However, researchers had observed minimal use of Kinyarwanda by most teachers, so the conversations with girls were central to understanding that this was not normal.Both the Rwandan researchers and previous researchers in other contexts have suggested that teachers may alter their language use under these circumstances because the use of a local, familiar language contravenes prescribed language policy (Probyn 2009).
In the Tanzanian study, conversations with learners were crucial for developing a fuller understanding of classroom silences.Adamson has written about understanding reasons for learners' lack of response to teacher questions in detail elsewhere (Adamson 2022a).She notes that her initial assumption was that learners did not understand the content or the question, and thus couldn't respond.However, this was importantly challenged by students, who explained that they often did have some level of understanding, particularly when the teacher had translated content into Kiswahili.Instead, they suggested that it was feelings of fear and shame that were the most powerful reasons for their silence.Both these examples demonstrate that traditional observation methods alone could have strongly perpetuated testimonial injustice.The socio-emotional dimension of the use of an unfamiliar LoLT was important in both studies, but is severely under-researched.Its significance may have remained unacknowledged here too if methods were not included that created space for young people to be understood as credible commentators on their own experiences of education.
In a second example from Tanzania, we see the power of mechanisms of hermeneutical injustice.When initially asked about the role of language(s) in their education and future aspirations, learners' firmly, and unanimously, asserted the importance of English, with common phrases and sentiments such as: 'English is the most important thing is the life of a student' [Form Two student, rural] and 'schools should put rules in place that force students to use English language in all locations' [student re-sitting Form Four, urban].However, this narrative appeared in tension with the language choices and behaviour observed on a daily basis.This 'disjuncture' between learners' (and teachers') 'professed attitudes and practices' has also been observed by Desai in her work in South Africa (Desai 2016, p. 346).In the Tanzanian context, this prompted Adamson to ask more, and different, questions.She found that, although the high prevalence of Kiswahili use in school was, in part, explained by learners' struggles with English, there was another, important, explanation.Learners regularly asserted that 'Kiswahili ni lugha yetu' (Kiswahili is our language), contrasting this with 'Kizungu' (language of the European/foreigner) as an alternative term to refer to English.One student in Form Four at the rural school explained students' strong preference for Kiswahili, saying 'English isn't their real language'.This is a powerful example where the broader discourse around the importance of English and its use as the sole LoLT obscured the fact that learners valued multiple languages simultaneously, and that Kiswahili played a key role in learners' senses of identity and belonging.Even in the case of the last student who appears to be challenging the primacy of English, it should be noted that she made this comment in English, using the language as a marker of credibility to set herself apart from other students who only used Kiswahili.Thus, although the student is sympathetic to the language situation of others, she still reproduces the dominant language discourse works to epistemically marginalize her peers (Anderson 2012).
These three examples show that, without the use of multiple methods that approached learners' experiences from multiple angles, and, importantly, created space for their own explanations, in both studies we could easily have misinterpreted learners' realities and perpetuated epistemic injustice.In addition, our research shows that the use of a multi-angled approach in LoLT research is crucial for broadening the focus and including information about learners' lives that has not traditionally been considered central to understanding the role of an unfamiliar LoLT.

A multi-angled approach helps us to understand the relationship between experiences of LoLT and broader life circumstances
Using interviews, informal discussions and participatory activities to talk to young people about their experiences of education, many identified home and community-based factors that constrained, or enabled, their capability to learn.In the Rwandan study, which was focused on girls' experiences, many girls talked at length about their lives outside school and the gendered expectations to take on significant home chores.Other girls gave in-depth accounts of the financial and familial precarity and violence that characterized their home lives.These sorts of accounts are not unfamiliar in research looking at gender equity and education (Jones 2011, Warrington andKiragu 2012).However, this kind of detail is much less commonly included in LoLT research.This may be because, at first glance, factors such as levels of chores or insecure home lives might seem unrelated to the language being used in the classroom.However, as researchers we must ensure that the way that we focus our own attention and the choices we make about boundaries in our research do not compound testimonial and hermeneutical injustices by excluding potential dimensions of learners' experiences that do not immediately seem relevant to us, but are considered relevant by learners themselves.In our research it became evident that a wide range of factors, which have also been identified as relevant to girls' broader capability to learn (Cin andWalker 2016, Tao 2018), should play a much more prominent role in LoLT research.Not only does the use of an unfamiliar LoLT add an additional, and highly significant, obstacle to engagement in learning, but this interacts with other challenges that learners are simultaneously experiencing and negotiating.
From girls' accounts, it was clear that the additional time and effort to try to understand the unfamiliar LoLT needed to be found outside of lessons.Vestine from Nyarugenge district in Rwanda explained the importance of this out-of-school revision: I must revise all lessons when I reach home so that when I get back to school, I am on page … this is because when the content becomes huge, you may not know where to start and become lazy, so I like to make sure that I have done revisions before it becomes hard for me There were a small number of girls who were able to protect time and space to revise curricular content and practise English vocabulary and were able to attend early morning teacher-led, quiet revision sessions before school started.By contrast, other girls talked at length about the limitations on their time and repeated sources of distraction, at school, at home and on the way between the two.These girls spoke of attempting to memorize content, often with words in English that they did not understand the meaning of, so that they could try to keep up at school.However, Charity from Burera district in Rwanda noted that even memorizing content was challenging: Sometimes when I am studying and fail to translate in Kinyarwanda, I just memorize the content, it is actually hard to memorize something in English that you don't even know the meaning in Kinyarwanda.
The importance of these opportunities for learning outside of school was also echoed in the Tanzanian study.When asked whether she believed she could succeed in the national secondary school examinations, a girl in Form Two at the rural school replied: ' … if my life at home were to change.Because if I were to have time to study, if I were to get money for going to tuition to help me a little, I would be able to progress.But like this … I don't know … ' Girls' accounts of their lives outside of school, thus, revealed stark differences in individual circumstances and the levels of responsibility that different girls carried for activities additional to their learning (see also Milligan et al. 2023).Moreover, we found that these differences in out-of-school experience mapped against differences in learners' classroom experiences.In both studies, we were able to develop an understanding of the ways that individual learners experienced the classroom and how they participated in discussions and activities.In Tanzania this was possible through the length of time spent in school and the selection of particular groups that were observed repeatedly.In Rwanda, the dual-focused classroom observation design meant that one researcher was solely focused on the experiences of a small number of case study girls who were tracked through four different subject lessons.This revealed differences in different individual learners' engagement and interaction in the classroom.For example, in the Rwandan study, we found that the types of talk different girls were engaged in differed quite dramatically, and thus the extent to which they were epistemically excluded (for more detail see Kuchah et al. 2022).Being able to match these observations of individual learners' classroom experiences with accounts from interviews and other activities allowed us to look for patterns and connections between the in-school and out-of-school domains.
One area of learners' experiences that was clearly shaped by life circumstances outside of school was the socio-emotional impact of challenges in the classroom, particularly the use of an unfamiliar language.In both Tanzania and Rwanda, learners described feelings of shame about what they perceived to be their shortcomings in relation to the LoLT.Learners also talked about feeling humiliated or 'discouraged' by the comments and reactions of others, through criticism, mockery and laughter at language-related mistakes.There are numerous examples from both studies of learners describing negative emotions.In Ruhango district in Rwanda, Nazou explained: 'losing self-confidence troubles me during learning at school, when I raise my hand, I feel people can laugh at me or when I say something and others laugh'.These experiences were certainly not limited to girls.A male student in Form Four at the rural school in Tanzania stated: 'So they break a person's heart … even if you are interested in speaking, you shouldn't speak … because you are afraid'.
In both contexts, gender was found to be a factor that influenced how learners experienced and responded to these emotions.In Tanzania, although both boys and girls stated that they disliked the common practice of laughing at mistakes, it was both observed and discussed that boys had the option, to some extent at least, to take control of this laughter by playing the fool.By contrast, girls particularly felt that laughter was a judgment on their broader intelligence and ability.However, a range of other out-of-school factors, including the level of support and encouragement for learning at home, access to a family member who speaks English, access to financial resources to pay for additional materials and tuition for learning English, and the availability of time, space, and a sense of personal safety, also influenced how resilient learners were to these negative emotions.Those who faced the greatest challenges outside of school were most likely to be silenced by negative emotions in the classroom.
In writing with Rwandan colleagues, we have conceptualized the intersection of language with wider gendered life circumstances with the Kinyarwanda phrase 'umuzigo w'inyongera' (Milligan et al. 2023).This literally translates as a 'double-burden' but significantly, the word 'umuzigo' reflects a burden that weighs heavy on the heart.This helps to understand the range of burdens that these learners articulated -from being weighed down by the burden of household chores, to the weight of parental expectation to perform well at school, to the socio-emotional weight of feeling shame to speak in class -and how an unfamiliar LoLT may add more weight to an already heavy load.It was also observed in both Rwanda and Tanzania that, despite the recognition of a wide range of factors that made it more challenging for both girls and boys to participate and flourish in the classroom, learners tended to individualize and blame themselves for failing to 'ignore' and overcome these difficulties.For example, when Vestine (above) equated 'not knowing where to start' when approaching the 'huge' amount of content with being 'lazy'.Or when a Form One student at the urban school in Tanzania (who had attended a private, Englishmedium primary school) said of her classmates (who had attended government, Kiswahilimedium primary schools): 'they are not improving because … they are not serious in their studies … if they really want to know English, just let them try' (see also Adamson 2022b).
The power of discourses relating both to the unquestionable value of English and the neoliberal, 'responsibilitization' of individuals for their own wellbeing and success (DeJaeghere 2017, p. 30) contribute significantly to the perpetuation of hermeneutical injustice as learners themselves rarely acknowledged the structural constraints they were subject to.This finding leads us to advocate even more strongly for a learner-focused, multi-angle approach that can begin to address the notable absence of information about learners' out-of-school lives in LoLT research, begin to build more complex accounts of learners' heterogenous experiences, and both identify and redress instances of epistemic injustice.

Discussion
The findings from our studies, briefly discussed here and in our wider research (Desai 2010, 2012, 2016, Kuchah et al. 2022, Adamson 2022a, 2022b, Milligan et al. 2023), clearly demonstrate the ways that learning in English not only affects children's access to the school curriculum but also silences them.Within this 'Anglonormative' (McKinney 2017) and 'epistemic monoculture' (Masaka 2019), children are positioned as unreliable sources of knowledge because they cannot express their knowledge in English.By contrast, the multi-angle research design that we have proposed in this paper has potential to reposition learners as credible sources of knowledge about their own experiences.Firstly, the research projects were designed to welcome learners' contributions in Kiswahili and Kinyarwanda, which were familiar languages for the learners in these studies.This point may seem obvious, but we note that this is not always the case, particularly in research conducted with secondary school learners.Secondly, this paper responds to the lack of in-depth opportunities provided in the existing LoLT literature for learners to explain their experiences and perspectives.In both these ways, we argue that the approach can help to counter the testimonial injustice that children may encounter in monolingual classrooms by taking seriously children's own perspectives that may otherwise be considered as uncredible.
However, validating learner testimony alone is not enough to clearly evidence and challenge the structural injustice of being forced to learn in a dominant, unfamiliar language.This is particularly important in the Rwandan and Tanzanian contexts we have discussed where the narratives about both the value of English and the importance of using English as the sole LoLT persist as 'undeniable truths' (Brock-Utne 2012, p. 787).Children in these studies did not view challenges with the LoLT as forms of structural injustice.Rather, narratives about the importance of English prevented learners from acknowledging the value they attributed to other languages.Moreover, their struggles with the LoLT were individualized and internalized as personal failings that carried feelings of shame.This is a clear example of a 'situated hermeneutical inequality' as shared language ideologies determine that the place of English in education is unquestionable, preventing learners from 'making sense of an experience which it is strongly in her interests to render intelligible' (Fricker 2007, p. 147).Here, the combination of observations, alongside conversational methods, was particularly important for us to be able to see the aspects that young people do not, and may not be able to, articulate.
Many authors have argued that there is a jarring disconnect, created by an unfamiliar LoLT, between what happens in the classroom and children's broader life experiences, knowledges and languages (Rubagumya 2000, Vuzo 2010, Adamson 2021).Being able to connect across these spaces is one of the strongest contributions of a learner-focused, multi-angled research approach.Like some of the multi-method and ethnographic studies that we have already acknowledged (Chimbutane 2011, Bhattacharya 2013, Tamim 2021, Speciale 2022), this combination of looking at and listening to young people's experiences both inside and outside of the classroom results in a more critical understanding of the impact of learning in a dominant, unfamiliar language.It enabled us to learn about how classroom-based epistemic injustices intersected with learners' broader life circumstances to compound existing inequalities.These included characteristics such as gender, which although widely recognized as contributing towards broader educational inequalities, has seen very limited acknowledgement as a factor in LoLT research (Milligan and Adamson 2022).However, we also learned about important inequalities in terms of learners' access to material resources, time to practise English and revise core curricular concepts, and positive, supportive relationships with adults.This, in turn, helped to explain the silences we observed in the classroom.
Focusing on learners and broadening the scope of LoLT research is, therefore, particularly important for understanding the experiences of the most marginalized learners.It has been repeatedly observed that policies that require the use of an unfamiliar LoLT marginalize speakers of minority languages (Gandolfo 2009, Reilly et al. 2022).We certainly do not contest this point.However, we note that, in the two studies discussed in this paper, learners' familiar languages, Kiswahili and Kinyarwanda were the majority languages in their countries.Yet, the unfamiliar LoLT devalued shared linguistic resources, epistemically excluding all learners.This exclusion did happen to different extents for individual learners, but these differences in experiences were, at least in part, connected to differences and inequalities in learners' circumstances outside of school (Milligan et al. 2023).By creating spaces for learners to talk about their experiences both in school and beyond, we learned that the use of an unfamiliar LoLT acts as a factor of marginalization and exclusion for everyone, but that it is experienced as particularly challenging by those who are already facing significant barriers to engagement in learning.These are also often the learners who are most silent in the classroom.As such, these are the young people who are at risk of being marginalized or rendered invisible by research designs that do not extend beyond the classroom and fail to recognize the connections between experiences in and out of school.

Conclusion
Undoubtedly, a range of different types of evidence is needed to both understand the impact of the use of an unfamiliar LoLT and to affect change in policy and practice.In this paper we have acknowledged the important contributions made by existing research, but we have also argued that learners' perspectives and experiences are currently underrepresented.In doing so, we have demonstrated the importance of taking a multi-angle approach that combines learner-focused classroom observation with other qualitative methods that enable learners to explain their experiences from their own perspectives and situate them in the context of a broader picture of their educational and life circumstances.This approach responds to the fact that what happens in a classroom is influenced by structural factors across multiple levels of society (Bartlett andVavrus 2014, Tikly 2016), and that these factors and norms may result in injustices that learners do not, or cannot, name as such.At the same time, it recognizes that individual learners bring to the classroom with them different experiences, knowledges and resources that shape the ways they engage with learning.Undoubtedly there is a need for more research to understand how these different combinations of factors interact with personal characteristics and differences in classroom environments, but to do this we must first recognize the importance of foregrounding learners' experiences in LoLT research.We also note that we have focused predominantly on learning experiences and there is scope for more research that explores injustices in assessment and learners' outcomes.
We acknowledge that this type of research is neither straightforward, nor does it respect tight time constraints.It is no coincidence that several of the studies we have acknowledged as offering a broader account of learners' experiences are ethnographic.However, the Rwandan study discussed here demonstrates that research designs that consciously and intentionally focus on individual learners can achieve this school-life connection within much more limited time periods.We also note the importance of flexibility in research design and allowing methods to respond to, and be shaped by, the realities, preferences and initiatives of learners themselves.Our heightened understanding of the importance of epistemic justice, not just as a concept for understanding our findings, but also as a guiding principle for research, will undoubtedly influence our approach going forward.In particular, we will consider possibilities for reflexive, co-constructed approaches to classroom observation.
In concluding this paper, we are also reminded of Walker et al.'s (2020) assertion that research that is methodologically designed to attend to epistemic injustice will remain 'non-ideal'.Although engagement in research may go some way towards young people seeing themselves as credible contributors, this, in itself, will not bring about significant change to the broader, unjust structures within which young people live and learn.Instead, we hope that a body of powerful qualitative data will be generated through this multi-angle, learner-focused approach that will importantly grow the arsenal of evidence that can be used in the effort to finally shift stubborn language-in-education policies (Qorro 2013).In doing so, this approach seeks to restore humanity to learners who have been unfairly portrayed as passive and deficient, despite their daily efforts to succeed in an education system that is built on barriers.

Notes
1. Young people in Tanzania are typically 13-14 years old when joining Form 1, but the urban school, in particular, had a number of overage learners.A number of students were required to repeat Form 2 due to unsatisfactory performance in end-of-stage examinations.2. In Rwanda, the two year groups included in the study were P6, which is the final year of the primary stage, and S3, which is the third and final year of lower-secondary.Girls in P6 included in the study were between 11 and 15 years old and S3 between 15 and 19.Some of this range in ages is explained by class repetition.