Education and work: children’s lives in rural sub‐Saharan Africa

This paper proposes a dynamic conceptual framework – the edu-workscape – for understanding how rural children in sub-Saharan Africa navigate three key gendered social arenas: the household, school and workplaces. Focusing on school, in particular, the paper highlights the violence, harm and labour that occur there, and argues that learning, work and harm co-exist across all three institutional domains, and in context, and should therefore be considered holistically.

This paper explores children's lives within the nexus of education and work in sub-Saharan Africa (SSA). It achieves this through a selective review of the education literature as it relates to work and harm, and presentation of a conceptual framework for understanding this relationship. There is a particular emphasis on rural children, both because of their greater involvement in agriculture -which is the focus of Action on Children's Harmful Work in African Agriculture (ACHA) (the programme of which this paper is part) -and because they are more likely not to be enrolled in school (UNESCO 2015;Ministry of Education 2018). Similarly, we pay particular attention to Ghana, which was to be the location of the programme's initial fieldwork. 1 The need to better understand the relationship between education (formal schooling in particular) and child work in SSA is especially urgent for a number of reasons. First, the region has the fastest population growth globally, and a disproportionate number of young people: 2 43 per cent of the region's 1.1 billion people are currently under the age of 15, and a further 20 per cent are aged between 15 and 24 (UNDESA 2019). Second, SSA has the highest number of young people estimated to be involved in child labour (ILO 2017a(ILO , 2017b. It also has the greatest number young people out of school (UNESCO-UIS 2019): around 98 million children of school-going age -almost one-thirdare not enrolled in school, with participation rates having stagnated, as they have globally, since around 2008 (ibid.). Added to that are the many millions who may be enrolled but not attending regularly, and those attending but not learning. UNESCO figures for 2017 show that in SSA, 85 per cent of boys and 95 per cent of girls of primary school age were 'not achieving minimum proficiency standards' (ACPF 2018). As the impact of Covid-19 exacerbates socioeconomic inequalities, these figures for out-of-school children will rise, as will the gendered burdens of work many will have to take on (United Nations 2020).
Children least likely to be in school are female, from the poorest households and over-age for grade of entry; other factors associated with non-participation in schooling include living in a remote, rural area, or a conflict-affected region; coming from an ethno-linguistic minority or nomadic community; or having a disability (UNESCO 2015). Since SSA as a region contains many of the world's poorest countries (UNDP 2018a), even children who are in formal education frequently combine schooling with unpaid and/or paid work (Orkin 2012;ILO , 2017aILO , 2017bUNESCO 2015;UCW 2017). The main employment sector in SSA is agriculture, which employs more than half the total workforce, so many children are inevitably also engaged in agricultural work (ILOSTAT n.d.), involved in both subsistence and commercial farming, predominantly on familyowned/smallholder farms (FAO 2015).
Globally, financial constraints are often cited as the main reason why many children are either not in school at all, or are attending sporadically and juggling schooling and work (UNESCO 2015).
In Ghana, too, poverty has been identified as an important driver of non-attendance at school -although distance to school is also a major factor in rural areas (Ministry of Education 2018). However, as we elaborate further in sections 2 and 3, the realities are inevitably more complex and contingent, and subject to change, with wide variations across and within countries. Thus, the importance of context -in its multiple configurations and manifestations -is paramount.

Aim and scope of the paper
The aim of this paper is to explore children's lives as they relate to the education-work nexus in rural SSA, with a particular focus on Ghana. This we achieve initially through a selective and purposive summary of the education literature that highlights some of the key issues. We then elaborate an analytical framework with which to better understand the dynamics of education and work in rural children's lives.
First, we situate the issue of children's access to schooling within the broader discourse of child rights and the global development agenda, before identifying the children most likely not to be in school, and the reasons why. We then explore some of the key tensions between work and schooling identified in the literature, which we relate specifically to the lives of rural children. In section 4, we present an analytical relational framework that aims to support a more holistic understanding of the dynamics of children's lives as they relate to schooling, work and harm. This we term the 'edu-workscape', which includes household, school and work environments. We elaborate on each of these in turn in sections 5-8. In section 9 we highlight the problematics surrounding three key issues (context, childhood and gender) that are crucial to debates on children's work and education, and are highly pertinent to policy interventions aimed at increasing educational participation.
To inform this paper, we read more than 500 journal articles, reports, impact assessments and book chapters. These arose from searches of 'child work/labour', 'household chores' AND one or more from 'education', 'access', 'quality' and 'learning outcomes' (plus other similar terms) AND sub-Saharan Africa or Ghana (and other similar terms) in a number of databases. These included ERIC (Education Resources Information Center), Web of Science, Google Scholar, Education Sub-Saharan Africa (ESSA) and the ACHA programme Zotero database. We also trawled DFID and EPPI reviews, and the websites of Young Lives, 3 Gender and Adolescence: Global Evidence (GAGE), 4 the International Labour Organization (ILO), the United Nations Children's Fund (UNICEF) and various non-governmental organisations (NGOs), such as Save the Children and the International Cocoa Initiative (ICI). This was in addition to literature with which we were already familiar regarding gender and education, and other studies that we came across in reference lists while reading. In particular, we sought out qualitative and ethnographic work often overlooked in rigorous or systematic literature reviews. Although geographically there is an emphasis on Ghana, we also draw substantially on the abundant literature on Ethiopia, especially from Young Lives and, to a lesser extent, GAGE, which have generated a wealth of insightful studies.

Concepts and definitions
The relationship between schooling and children's work, which we explore through a selective review of the literature, is characterised by tensions. These are exacerbated by conceptual adherence to terms that have been framed in the global North (Nieuwenhuys 1994(Nieuwenhuys , 1998Twum-Danso Imoh 2013), and which are often used as oppositional binaries: child/adult; in school/out of school; traditional/modern; rural/urban; harmful/not harmful; female/male. Although useful for the purposes of making national and international comparisons, these binaries can oversimplify our understandings of realities on the ground as they 3 Young Lives is a longitudinal study of childhood poverty and transitions to adulthood, following the lives of 12,000 children in four countries (Ethiopia, India, Peru and Vietnam). 4 GAGE is a longitudinal study of 20,000 adolescents in six countries (Ethiopia, Rwanda, Bangladesh, Nepal, Jordan and Lebanon) focusing on 'what works' to improve their lives at specific stages of adolescence. 5 For discussion of these terms see, for example, Bourdillon, Crivello and Pankhurst 2015;Maconachie, Howard and Bock (2020), ACHA paper on Theorising 'Harm' in Relation to Children's Work; or DFID's rigorous review on The State of the Evidence on Child Labour, School Attendance, Learning and Child Well-being (Aslam et al. forthcoming).
are gradually solidified into universal categories and truths through the dominant discourse of global development (Boyden 1997;Twum-Danso Imoh 2013). We return to some of these contested terms and binaries later in the paper.
For the moment, we outline our usage of three key terms for this paper: 'education', 'child labour' and 'harm'. While we recognise 'education' to be a broad term related to all types of learning -for example, informal, non-formal and formal -the focus in this paper is on schooling (particularly basic education: primary and junior secondary). There are three reasons for this: first, educational provision in SSA (as elsewhere) since colonial times has primarily been structured through formal education; second, educational research and educational interventions in SSA have, until relatively recently, primarily been directed at this area of education; and third, debates surrounding the tensions between education and child labour are generally articulated with regard to formal schooling.
'Child labour', too, we recognise to be a contested and value-laden term. However, it is not the aim of this paper to (re)define and/or measure child labour; 5 rather, we align ourselves with academics who prefer the more neutral term 'children's work' (e.g. Boyden, Bourdillon, Woodhead), alongside other ACHA authors (e.g. Sabates-Wheeler and Sumberg 2020; Maconachie, Howard and Bock 2020). That said, we also use the term 'child labour' (sometimes distinguishing it from child work, as the ILO does) as it appears in the various studies reviewed, elaborating on the authors' understandings where necessary. Similarly, we employ the term 'child' to signify someone under the age of 18, as the term is predominantly understood in the literature, though we revisit this critically in section 9.
The notion of harm, since it is central to ILO definitions of what counts as hazardous child labour or the worst forms of child labour, is also of concern to this paper. Maconachie et al. (2020: 7) note that 'despite its voluminous writings, the ILO has itself never formally defined "harm"'. Without tying ourselves to a specific definition of harm, we would agree that 'the concept of harm is ambiguous, relative and contextual and it may be unhelpful (and even problematic) to present harm as an "objective" concept that can be defined, measured, and assessed with discrete criteria' (ibid.: 17).
It is also worth noting that in much of the education literature (outside the area of social protection), the discussion is more about 'violence' rather than report on school violence worldwide (UN 2016), 'harm' occurs only five times, whereas 'violence' has more than 400 mentions. Although it is beyond the scope of this paper to explore this relationship in depth, we return to the issue in section 6.
2 Education For All

Educational access
Access to full-time education for young people under the age of 18 is a human right. It is enshrined in the UN Convention on the Rights of the Child (1989) and the African Charter on the Rights and Welfare of the Child (1996, and implemented in 1999), and recognised as a global development priority that is reinforced through commitments to the global Education for All (EFA) goals, the Millennium Development Goals (MDGs) and the Sustainable Development Goals (SDGs). The MDG agenda, adopted worldwide up until 2015, specified completing a cycle of compulsory free primary (later expanded to basic 6 ) education; in the SDG on education (SDG 4), this commitment has now been expanded to a full cycle of primary and secondary schooling. Since access to full-time education is globally acknowledged to be a human right, any denial of that right is, by definition, a harm. Thus, from the outset, children's work -if it takes time away from children's educational activities (be they in class or at home doing homework) -is positioned as antithetical to, or at the very least, in competition with schooling (Myers and Boyden 1998;Bourdillon, White and Myers 2009). Yet many contextually situated studies demonstrate how child work makes essential contributions to household welfare and/or paradoxically may provide the means to attend school in the first place (e.g. Okyere 2012;Orkin 2012;Wambiri 2014;Pankhurst, Crivello and Tiumelissan 2016). Nevertheless, in the dominant development discourse (epitomised by the United Nations Development Programme (UNDP), ILO, Overseas Development Institute (ODI) 7 and World Bank, for example) child work/labour is predominantly framed negatively as an impediment to achieving EFA (e.g. Guarcello, Lyon and Valdivia 2015). The following opening statement in an ODI survey report on child labour and education in Dhaka, Bangladesh, exemplifies this: Universal secondary education and quality learning have been adopted by the international community as goals for 2030. Child labour, which remains endemic in many poor countries, represents an obstacle to both goals. It keeps children out of school, hinders effective learning and denies children an opportunity to acquire the knowledge and skills they need to escape poverty, and that their countries need to drive inclusive growth and human development. (Quattri and Watkins 2016: 9) Over the past 30 years, in pursuit of universal primary (and then basic) education, school enrolments have mushroomed across SSA, albeit unevenly, as demonstrated by improving enrolment and survival rates, diminishing repetition and drop-out rates, and narrowing 'gender gaps' (UNESCO 2015). However, in resource-constrained contexts, including in parts of Ghana, increases in school participation have not necessarily been matched by improvements in school quality and learning opportunities (Akyeampong 2009;Lewin and Akyeampong 2009;Woldehanna and Araya 2016). As the 2004 EFA Global Monitoring Report acknowledged, 'the focus on access often overshadows the issue of quality. Yet quality stands at the heart of Education For All' (UNESCO 2005: 18; see sections 5 and 6). Moreover, as we argue later in the paper, the non-provision of 'inclusive, equitable and quality education' (as articulated in SDG 4) to children who are in school, is as much a denial of their right to education as non-participation in schooling is to children who are out of school.
A great deal of research effort has focused on identifying precisely which children are not in school, 8 though as we point out in section 3, the 'in-school' and 'out-of-school' binary is itself problematic. The Global Initiative on Out-of-School Children has summarised the principal barriers and bottlenecks to children's participation in school (see, for example, UNICEF and UNESCO-UIS 2014: 38). Child work/ labour -however it might be defined -has been widely identified as one of these main barriers. risk of exclusion are migrants and fostered children: in West Africa, an estimated 20 per cent of 8-12-year-olds migrate independently for educational reasons; and in Ghana, an estimated 12 per cent of children are currently residing in foster households (Darko and Carmichael 2020). Children from nomadic communities, who account for around 20 per cent of the population in East Africa (UNESCO 2015), are another group with low access to education, as are children with disabilities (ACPF 2018), and/or children from minority ethno-linguistic groups (Smits, Huisman and Kruijff 2008). Otherwise, in very broad terms across SSA, survey studies have shown that children are more likely to be out of school if they come from rural (rather than urban) areas, from poorer (rather than wealthier) families, are girls (rather than boys), and are older (rather than younger) (UNESCO 2014). Inevitably, patterns vary considerably across and within nations, states, communities and even households, as national-level aggregates mask substantial internal disparities and inequalities.

2.2
Importantly, however, figures for out-of-school children are huge underestimations; neither one-off

Box 1: Snapshot of children's schooling and work patterns in Ghana
Ghana enjoys relatively high school participation rates at primary and secondary levels in comparison to many other SSA countries (UNDP 2018a), though net enrolment rates (NER) have plateaued at primary and junior secondary levels since 2013-14, at just over 90 per cent and 50 per cent respectively, and repetition rates are high. Even so, over 450,000 children are out of school, mostly from the poorest households and within the three northern regions (UCW 2016; Ministry of Education 2018), where child labour is said to be more concentrated (Krauss 2016). Completion rates vary substantially, with lower rates for boys than girls, for children from poorer families, children from rural areas, children in northern regions, and children from particular linguistic groups. School attendance also decreases with age. Although nationally there is broadly gender parity in enrolment at both primary and secondary levels, regional disparities exist, with far more girls out of school in the Northern and Western regions, while the Upper East and Upper West regions have lower male enrolment. Children with disabilities are also less likely to be in school (Ministry of Education 2018), as are pastoralists (Mfum-Mensah 2003).
The most recent dedicated child labour report found that over a quarter of all 5-17-year-olds both attend school and are engaged in paid employment, and around one in five are deemed to be involved in child labour (Ghana Statistical Service 2014). * The majority of children engaged in economic activity are involved in agriculture, with higher percentages in rural areas (ibid.). This is unsurprising given that agriculture (including fishery and forestry) provides paid employment for nearly two-thirds of the working adult population in rural areas, with slightly higher figures across the board for men rather than women (Ghana Statistical Service 2019). That said, small-scale artisanal mining is increasingly being combined with agriculture, especially during the 'hungry season' as rural families increasingly struggle to survive on agriculture alone (Hilson and Garforth 2012).
annual school enrolments (as recorded in Education Management Information System (EMIS) data) nor household survey attendance data (which only require children to have been at school once over a set period of time) reflect the amount of time that children are actually at school, or in the classroom engaged in meaningful learning (Fair 2016). Moreover, in many contexts in SSA, EMIS data are often inconsistent, unreliable and incomplete (Husein, Saraogi and Mintz 2017), making it even more challenging to gauge who is 'in' or 'out of' school.

Why are children not in school?
Within the dominant statistical research on school access, as exemplified by World Bank and UNESCO Global Monitoring Reports, reasons for not being in school are generally divided into supply-side and demand-side factors. Supply-side factors include: a lack of schools nearby, which also relates to school travel time and costs and safety concerns, especially regarding young children and girls; a shortage of classrooms, water and other necessary infrastructure; and inadequate supplies of teaching and learning materials and Many of these same 'inputs', however, are also commonly identified indicators of school quality (see sections 5 and 6), which, when inadequate, reduce demand for schooling (UNESCO 2005;Hunt 2008). Additional 'push' factors encompass issues to do with curriculum and pedagogy, disciplinary regimes, as well as informal teacher-student and peer interactions (UNICEF and UNESCO-UIS 2014; Bashir et al. 2018), which we address in more detail in sections 5 and 6.
Demand-side factors affecting school access in SSA include: economic hardship (which we return to below), which tends to be exacerbated in large households with high numbers of school-age children (UNESCO 2015); child or family health issues, including hunger; early marriage and pregnancy; perceptions of the low value of education, due variously to its poor quality (see sections 5 and 6), lack of relevance to local socioeconomic realities (Tafere and Pankhurst 2015;Jones et al. 2019) or absence of employment opportunities for school graduates (UNESCO 2015; Bashir et al. 2018).
Across the continent, however, poverty and the high cost of education (in terms of both direct and indirect opportunity costs, such as the loss of a child's labour) have been identified as the main reason why children are out of school in the very poorest communities (UNESCO 2015;UNESCO-UIS and UNICEF 2015;Bashir et al. 2018). To address this, most governments in SSA have abolished school fees, at least for primary education (ACPF 2018), and are increasingly committed to providing more resources for educational development (UNESCO-IIEP 2011). They are also struggling to keep pace with expanding school-going populations while simultaneously investing in trying to improve quality. However, costs to households still persist in the form of parent-teacher association (PTA) levies, school development funds, exam fees, uniforms, stationery, textbooks and transport, which means that completing even a cycle of primary schooling is unachievable for children in the very poorest households (UNESCO 2015). As a result, children from poor families have to work both to help themselves and the family or household satisfy basic needs, including paying for their schooling (Boyden 1994;Okyere 2012;Wambiri 2014). In this way, work appears to be in competition with schooling (Bourdillon et al. 2009). Yet, as we discuss in section 3, the relationship between work and children's schooling is much more complex and contingent, as much of the Young Lives data show (see Bourdillon, Crivello and Pankhurst 2015).

What is being done about widening access?
Programmes designed to increase participation in basic education are premised on the assumption that the main reason why children are out of school is because families cannot afford to educate them, and/or (to a lesser extent) because parents are unaware of the value of education, particularly with respect to girls (UNESCO 2015). Aiming to address the economic constraints poor households face, and reduce their reliance on children's labour, social safety net strategies designed to widen accesssuch as school feeding programmes, scholarships for girls, fee waivers and cash transfers -are spreading across Africa, 10 undergoing particularly rapid expansion in Ghana, Kenya, Senegal and Tanzania (Beegle, Coudouel and Monsalve 2018).
School feeding programmes have been particularly effective at increasing participation, especially in poorer areas, where enrolments are usually low (Gelli, Meir and Espejo 2007;Bundy et al. 2009;Jomaa, McDonnell and Probart 2011;Adelman, Gilligan and Lehrer 2018), and have been shown to contribute to improved test scores (within the project's limited time period) in a few instancesfor example, in Kenya and Senegal (Bashir et al. 2018). Cash transfers have also consistently been associated with improved school participation (Snilstveit et al. 2016;Bashir et al. 2018) as studies across a range of countries have shown that even very poor families increase their household spending on education as a result of the extra income (Devereux et al. 2008;Davis et al. 2016;Fisher et al. 2017). This has also been the case with the Livelihood Empowerment Against Poverty (LEAP) programme in northern Ghana, although it has also been noted that not all children necessarily benefit equally within households (Roelen, Karki-Chettri and Delap 2015).
To add to this, there is very little evidence on the long-term impacts and sustainability of such programmes in ensuring sustained access (Snilstveit et al. 2016), especially if funded by external donors for a limited period. Edmonds and Shrestha's (2014) brutally blunt paper, entitled You Get What You Pay For: Schooling Incentives and Child Labor, is a cautionary tale in this respect. The researchers returned to the site of a one-year project aimed at combating child labour in a carpet-weaving factory in Nepal, 16 months after the programme had ended, to find that all the educational gains accrued during that year had been erased, and the children were back weaving carpets.
An assessment of cash transfer programmes in six countries in SSA, including Ghana, showed that the impact on children's work patterns is never straightforward (Fisher et al. 2017). Although such interventions may increase access, they do not necessarily reduce student absenteeism, nor reduce 'child labour' (in this case, paid employment outside the home) as children may earn money after school and at weekends. Rather, they often signal a shift to unpaid work on the family farm as families increase their assets, such as livestock, which generate further work (Fisher et al. 2017;de Hoop, Groppo and Handa 2020).
Public works programmes (in which households offer adult labour in return for cash and/or food) can have similarly ambiguous impacts on children's educational participation and their patterns of work, as children often get drawn into gendered paid and unpaid labour as a result (see Devereux 2000;Devereux et al. 2008;Tafere and Woldehanna 2012;de Hoop et al. 2020). In Ethiopia's flagship Productive Safety Net Programme (PSNP), for example, although the extra income was found to reduce girls' and boys' workloads outside of the home and reduce the time spent on agricultural work, school attendance rates for younger girls dropped and they spent more time on substitute domestic work (Hoddinott, Gilligan and Taffesse 2011). In addition, some children were taken out of school to substitute adult labour on the programme (Tafere and Woldehanna 2012;Jones et al. 2019). Similarly, on a road construction programme in Zambia, income from adult earnings (and children's work during school holidays) helped pay school costs. Yet, in addition to substituting adult labour on the project, the relatively high wages on offer persuaded some children to miss school altogether to work on the programme, even without parental consent (Devereux 2000). Thus, even where programme objectives have been attained, there may be complex knock-on effects -both intended and unintended, and positive and adverse -on children's workloads and school attendance (Devereux 2000;Cornwall and Aghajanian 2017).

Complementary education -adapting education to working rural children
Whereas programmes such as cash transfers, school feeding and public works attempt to overcome the economic constraints that may prevent children from attending school, accelerated learning programmes (ALPs) 11 attempt to address many of the institutional features of formal education that keep children out of school. In contrast to the rigidity of mainstream government schooling, which is ill-fitted to rural lifestyles (which we discuss in section 3), several features of ALPs appear to acknowledge, to some extent, working children's lives and identities outside school.
Generally, an ALP is a condensed national curriculum completed in a reduced time period (six years into three, or three years into one) sometimes followed by a transition into the public education system, which may be supported to some extent for the initial year. Features include: reduced contact hours (in terms of both years in school and hours in the day), which substantially reduces the opportunity costs of schooling; and adapted curricula that often include life skills and livelihoodrelated content, which may be more related to working children's needs (Longden 2013). The less punitive disciplinary regimes of ALPs are also likely to appeal to working rural children, who are disproportionately punished for late-coming and absenteeism in mainstream schools due to work and household commitments (Dunne et al. 2013a;Pankhurst, Negussie and Mulugets 2016;Isimbi, Umutoni and Coast 2017;Devonald, Jones and Yadete 2020), which may contribute to dropout (see section 6).
ALPs have arguably shown the greatest promise in addressing both access and quality concerns in education, and improving participation, experiences and learning outcomes for out-ofschool rural children in SSA (see Baxter and Bethke 2009;Hartwell and Casely-Hayford 2010;Manda 2011;Longden 2013;Akyeampong et al. 2018aAkyeampong et al. , 2018b. However, their relative success in terms of increasing participation and learning outcomes is leading some education ministries, including in Ghana and Malawi, to take over these complementary programmes and try and learn from them, in order to improve the quality in mainstream schools (Longden 2013), which will encourage greater sustained access.
Moreover, there is evidence of some children in ALPs achieving comparable or better grades than children in mainstream schools, and broader learning outcomes that may help students in their home, community and working lives even if forced to drop out of school again (Humphreys et al. 2017;Akyeampong et al. 2018a). Thus, although ALPs still prioritise formal learning over work, there is at least some recognition of the wider context and the complex and competing time and work demands in the everyday lives of working children.
3 The relationship between education and work

Education and work in competition
The dominant discourse that circumscribes debates about the relationship between schooling and children's labour, as articulated by the likes of the ILO and the World Bank, sustains a hierarchical education-work binary, which positions education above work (Jonah and Abebe 2017). This is problematic for a number of reasons. First, it reflects neither the situation within the lives of children, nor the views of children themselves (Woodhead 1999; Twum-Danso Imoh 2013; Jonah and Abebe 2017). In addition, it assumes that school is a benign space for education and learning, and therefore the proper place for children to be (Myers and Boyden 1998;Bourdillon et al. 2015) whereas work is often assumed to deny children this right. Further, this is the default rationale even when it is acknowledged that the education on offer is sub-standard, and that children both need and want to combine schooling with work. Driven by international anti-child labour campaigns and policy, much of the research carried out on children's work by international organisations such as the ILO, UNICEF or the World Bank is motivated by the assumption that children are better off in full-time education, rather than in work (e.g. Quattri and Watkins 2016; ILO 2017a). This attitude is exemplified in the most recent regional ILO brief on Africa: 'There is an on-going need for investment in what we know works in getting children out of work and into the classroom -and keeping them there' (ILO 2017b: 8).
In this influential literature on child labour, work is frequently understood narrowly as 'productive work' related to income generation, which excludes much domestic work (often described as 'household chores') that is considered to 'constitute a "non-economic" form of production' (ILO 2017a: 17). Although 'household chores' have belatedly been included for the first time in ILO's global and regional reports on child labour (ILO 2017a, 2017b), they are considered separately -in addition to figures on child labour -and not as an integral part of children's workloads. We consider this further in section 7.
Within this context, in the global South, a number of multi-country studies and large-scale surveys (including randomised control trials) have compared the school attendance of 'working' and 'non-working' children (itself a problematic dichotomy, given the limitations about what counts as work), and concluded that generally, children who work above a particular threshold of hours and/or intensity of work are less likely to be 'in school' (UCW 2017;ILO 2017a). Yet the classification of children as being either 'in-school' or 'out-of-school' children constitutes a further unhelpful binary, especially since being 'in school' (be it enrolled or attending, according to the survey definition) may not necessarily involve much time within the school grounds, let alone in a classroom (Tafere and Pankhurst 2015). However, as more longitudinal mixed-methods research (e.g. the Young Lives programme) and qualitative and ethnographic studies (e.g. Berlan 2009; Okyere 2012) have shown, the reality of being 'in school' is inevitably more nuanced. Working children from poor rural backgrounds, in particular, may move in and out of being physically at school within a single day, or over much longer periods of time, depending on a range of factors, and learning-related activities may only constitute a small portion of that in-school time (Humphreys et al. 2015;Tafere and Pankhurst 2015).
A second common assumption is that work prevents children from persisting and/or achieving well in school. While work may be a major factor, evidence from more contextually situated studies in SSA suggests that 'problems with schooling are often the result of other difficulties that children and their families face, and cannot simply be blamed on work' (Bourdillon et al. 2015: 6). For example, rural households dependent on agriculture are especially vulnerable to environmental shocks, such as drought or flooding, which may also disrupt children's schooling (Rose and Dyer 2008;Tafere 2014 Crucially, the poor quality of schooling available in many rural areas is central to determining whether children will attend or persist in school, and whether children learn once there (Odonkor 2007;UNESCO 2015). So too is the perceived benefit of schooling by children and parents/ caregivers when viewed within the context of local employment opportunities (Hashim 2004;Jones et al. 2019). Many poor households are making strategic livelihood choices regarding their children's schooling based on a complex interaction of changing factors. These include available finances, and the multiple and varied socioeconomic needs of the whole household (itself dependent on the number, age, gender, birth order, health and abilities of its various members), and the quality and likely future benefits of education (Boyle et al. 2002;Nekatibeb 2002;Hashim 2004;Webbink, Smits and de Jong 2012;Abebe 2011;Maconachie and Hilson 2016). These are all liable to change as circumstances change. In Ghana, for example, the fact that the annual income for a household head who is self-employed in agriculture is the same whether they have no formal schooling or have completed basic education is likely to be an important consideration (Krauss 2016).
In addition, young people may themselves make their own choices and choose work over education at different times, especially if the schooling on offer is of low quality and they are failing to learn . Thus, although the many work demands put on children may prevent them from attending school, it does not follow necessarily that if these children were not in work, they would persist in school. The salience of 'truancy', 'lack of interest' or 'didn't want to go' as survey answers given by children for missing and/or dropping out of school suggests this may well be the case (e.g. Tafere and Woldehanna 2012;ILO-IPEC 2013;NPC 2015).
In contrast to the limited learning opportunities afforded by low-quality schooling, adults and children may consider children's work and home environments to afford better learning opportunities, which in certain contexts may improve their future livelihood opportunities more than the schooling on offer can (Myers and Boyden 1998;Admassie 2003). In a similar vein, the harm that is often attributed to many types of what the ILO calls 'hazardous work' 12 may actually be just as evident within the school environment (Berlan 2004), or even the home (Odonkor 2007). Ampiah and Adu-Yeboah's (2009) interview-based study on 90 school dropouts in Ghana provides evidence of the former. Although critical incidents related to poverty and/or the need to work were the most commonly cited tipping points that finally kept the children out of school permanently, children identified a combination of out-of-school and (predominantly) in-school factors that led to dropping out, including harmful school processes such as indiscriminate corporal punishment and forced manual labour for teachers, as we discuss in section 6.

Schooling and fit with rural lifestyles
A further manifestation of the hierarchical relationship between children's education and work is the assumption that school priorities take precedence over children's work and other aspects of their lives. The rigid structures of schooling demand that children's work, including within the household, should fit round the regimes of schooling (Bourdillon et al. 2015;Humphreys et al. 2015).
In practice, schools are not flexible enough to meet the needs of working children; rather it is the working children who are required to be flexible to fit in with the school system and bear the burden of trying to combine their various responsibilities. (Bourdillon et al. 2015: 8) Thus, if working children (or their families) do not adjust schedules to enable them to attend school, then work is blamed for their loss of education (Bourdillon et al. 2009). Since almost all children in SSA are engaged in some form of work -whether in the home or outside, unpaid or paid, or in the formal or informal sectors (Boyden 1994; Marsh and Kleitman 2005; Tafere and Pankhurst 2015) -school's lack of flexibility concerning annual, weekly and daily timetables has profound effects on children's ability to attend school regularly and their likely learning outcomes (Delap and Seel 2004;Orkin 2012;Humphreys et al. 2015). That said, research in Ethiopia reports the occasional exception, with some schools adopting flexible timetabling during harvest seasons or adjusting their weekly timetable to accommodate market days. One school even provided childcare services on market days so that girls who usually missed class to look after siblings could attend (Frost and Rolleston 2013).
In addition, double-shift schooling, which is common in rural areas, offers slightly more flexibility as the half-day timetable gives children more time to contribute to their households' livelihoods (Orkin 2012;Wambiri 2014;Pankhurst et al. 2016).
However, generally, as Caroline Dyer (2013) notes concerning the marginalisation of nomadic pastoralists within mainstream schooling: Schools have extreme difficulty in accommodating patterns of learner attendance that do not adhere to everyday, taken-forgranted, institutional norms, such as fixed school calendars and timetables, a set annual enrolment point, cohort-based teaching and progression on an annual cycle, often linked to examination success. (Dyer 2013: 608) In Ethiopia, for example, the nationally standardised main school holidays coincide with the grainproducing regions in the north, but do not suit the cash-crop calendar in the south (Abebe 2011). The clashing annual cycles mean that children returning to school mid-year after an absence for seasonal work may be told to return at the beginning of the next academic year, and repeat the grade (Colclough et al. 2003;Ananga 2011;Orkin 2012). This may discourage the child from re-entering school (Hashim 2004;Wouango 2013), sometimes due to feelings of shame or embarrassment at having to repeat and/or study with younger children (Ananga 2011;Hamenoo et al. 2018). In addition, grade repetition adds to both the direct and opportunity costs of education, and means that the child will be a year older, and therefore less likely to stay in school (Lewin 2009;Jones et al. 2019).
Similarly disadvantaged by school's annual cycles are Africa's many migrant children, who (either independently or with a family member) move location for whatever reason, but are not allowed to start a new school mid-year. Since (as we highlighted earlier) an estimated 20 per cent of 8-12-year-olds in West Africa migrate independently to seek educational opportunities (Darko and Carmichael 2020) or work opportunities to help fund their own or their siblings' education (Hashim 2005;Hashim and Thorsen 2011), this lack of flexibility in school admission times (and curricular organisation) is denying education to many.
That said, some types of work may be more compatible with school timetables than other types. Evidence from Ethiopia, for example, suggests that while they are minding cattle, boys may be able to study, whereas girls doing household chores cannot (Rose et al. 1997;Orkin 2012).
The dissonance provoked by the spatial and temporal disconnects between rural lives and formal education are compounded by other tensions. These include the curricular irrelevance of much of modern schooling -despite various governments' curriculum reforms -to the lives and needs of rural communities (Boyden 1994;Nekatibeb 2002;Human Sciences Research Council et al. 2005;Kanu 2007; Yeboah and Daniel 2019), including nomadic pastoralists (Odachi 2011;Dyer 2013; Ng'asike 2019). Such curricula take no account of the skills and achievements that children have learned outside school (Bourdillon et al. 2010) while schools position working students as childlike irrespective of the adult responsibilities they have outside school -as breadwinners, mothers or carers, for example (Dunne and Ananga 2013).
Multiple other micro-level school practices, which we elaborate further in section 6, exacerbate the incompatibility of children's work with education as schools often fail to show any flexibility towards the many constraints facing working children and their households (Orkin 2012 (2007), in her research on education in cocoa-growing areas in Ghana, terms the 'damage risk costs' of schooling.
In conclusion, contextually situated research provides insights into the complex ways in which education and work interact and the multiple influences and demands on children's time in their everyday lives. In the next section, we focus on three key social arenas, namely the household, school and work environments -which we term the 'edu-workscape' -which are central to children's education and work experiences.
4 The edu-workscape: learning, labour and harm As we illustrated in the previous section, whether or not rural children can realise their right to education depends on a host of diverse factors including the high direct and indirect costs of schooling, the changing socioeconomic needs and social dynamics of households, the lack of fit between school and community calendars and timetables, the demands of employers, and the quality of schooling available. The interactions between these various factors place competing demands on children's time in three key social arenas: the household, workplaces and school. In order to better understand these dynamics, we propose a conceptual model which we term the 'edu-workscape' (Figure 1).
This triangular relational matrix not only includes the workplace, the school and the household, but it also centralises the child. It offers a re-conceptualisation and contextualisation of the relationships between children's work and education. It acknowledges that both learning and work take place within each of the institutions of the edu-workscape, and that the potential for harm exists in all three domains. Further, the matrix highlights the fact that children's experiences are shaped by social relations within and between home, school and work in specific contexts.
The edu-workscape foregrounds the dynamic interplay within and between each institution and emphasises how the child navigates and experiences this nexus.
There are multiple tensions between the three social arenas: schools (embedded within education systems), households (with their multiple, often changing, family and community configurations) and workplaces (in multiple locations including agri-business and commercial farms). These are also subject to conditions in the wider economic, social, temporal and spatial contexts. This conceptual framework represents the complex and overlapping relations that are the nexus of children's lives. It is intended to provide some analytical clarity to help understand how children navigate their daily lives between these three key institutions and the conditions, constraints and opportunities offered by them within different and changing wider contexts. Given the relational and contextual contingencies, childhoods are never singular or uniform but rather complex and diverse. As we discuss later, they are experienced through the symbolic and material significance of a range of social variables that include gender as well as histories of migration, settlement and work in different contexts (Nieuwenhuys 1994(Nieuwenhuys , 2010Bourdillon 2006;Abebe 2009;Burman and Stacey 2010;FAO 2014;Jonah and Abebe 2017;Rai, Brown and Ruwanpura 2019). This means, for example, that gender practices within community and household contexts will produce operational understandings of appropriate work (domestic, unpaid, paid and/or hazardous) for both girls and boys in ways that delimit and define childhoods, future social trajectories and ongoing engagements in social and economic life. Schools too often reaffirm these understandings of gender, through their formal and informal processes, such as gender-differentiated teacher-student classroom interactions or allocation of school responsibilities. This gendering will also be intersected by ethnicity, religion, age, disability and migration status, among other variables, to produce complex and relationally sensitive 13 See, for example, UNESCO (2005UNESCO ( , 2015; Bashir et al. (2018), on SSA. edu-workscapes within which childhoods are constructed and lived out.
Understanding how children navigate the edu-workscape offers the potential for more nuanced and socially sensitive insights into their work, wherever it takes place, as well as its excesses and harms in context. It allows us to consider more holistically the total burden of work on children in the school, household and workplace, as well as recognising the learning across the three institutions. Further, judgements about potential harm to children should not only be focused on work, but should similarly refer to the whole terrain of the edu-workscape, in context.
In the next four sections, we explore the three social arenas of the edu-workscape. As our main concern is with schooling, we elaborate this more extensively in sections 5 and 6. We focus in particular on the more qualitative aspects of schooling, both the formal and the hidden curriculum, which have a significant part to play in the ways that children experience and navigate the edu-workscape in their daily lives. This is followed by consideration of the household and the workplace (sections 7 and 8 respectively).

Introduction
The introduction of the SDGs brought a change in emphasis from educational access in the MDGs to quality and equity. The target of ensuring inclusive and equitable quality education in the SDGs was an explicit recognition that both equity and quality in school are critical to full educational participation and to improved learning outcomes. This was accompanied by a recognition that 'in many countries, it is the poorest children who receive the poorest quality education' (UNESCO-UIS and UNICEF 2015: 42), and that schooling is as likely to perpetuate cycles of deprivation as it is to interrupt them (Rose and Dyer 2008;van der Berg 2008;Dyer 2013). Interestingly, low school quality has been attributed, in part, to the success of widening access to education. In poorly resourced contexts with insufficient school infrastructure -common to much of SSA -increased numbers of children attending school have resulted in overcrowded classrooms with an inadequate number of trained teachers and a lack of teaching resources (Chimombo 2009;Lewin 2009;UNESCO 2015;Bashir et al. 2018). This denial of good-quality education, as demanded by SDG 4, arguably does as much harm to children as their inability to attend school in the first place since, as Marina Odonkor (2007: x) so succinctly put it, 'Education simply does not have value unless it is of good quality' [original emphasis].
While the aim here is not to review in depth the numerous quality issues that plague many schools in the region, 13 and particularly those in poor rural areas, consideration of school quality is nevertheless crucial to understanding the education-work-harm dynamic. First, there are many aspects of schooling that can cause harm to children, ranging from the more obvious physical violence of corporal punishment and bullying to the less visible symbolic violence, such as the curricular exclusion of particular languages and cultures (Antonowicz 2010; Mafela 2014). In addition, across the continent, schools -and rural schools in particularare also often sites of physical labour, adding to the burden of work that children are already undertaking in the household and/or in places of employment (as we elaborate in sections 7 and 8). Further, the lack of decent-quality educational provision in many schools is a major push factor in persuading parents not to send their children to school in the first place, or in driving children out of school and into other activities, such as paid employment (Tafere and Woldehanna 2012;ILO-IPEC 2013).
The most obvious evidence of poor-quality schooling lies in the low learning outcomes of most children across SSA, despite slight improvements in some countries (UNESCO 2015;ACPF 2018;Bashir et al. 2018). Notwithstanding significant variation between and within countries, an estimated two in five pupils in Africa leave primary school without being able to do simple arithmetic or read or write a simple sentence (ACPF 2018). More specifically, as highlighted in the Introduction, UNESCO figures show that 85 per cent of boys and 95 per cent of girls of primary school age and 63 per cent and 59 per cent respectively at junior secondary level were 'not achieving minimum proficiency standards' (ACPF 2018). In Ghana, despite outperforming other West African countries in terms of investment in education, learning outcomes too are 'an area of concern', particularly in poorer households, in rural areas, and in the three northern regions, where attainment lags well behind (especially for girls) (Ministry of Education 2018). The inequitable regional distribution of funds exacerbates the north-south divide, which contributes to the regional unevenness in provision of quality schooling (Akyeampong 2011;Blampied et al. 2018).

Factors affecting school quality
Quality issues across SSA relate to a variety of interlinked factors, such as poor infrastructure, including a lack of (or dilapidated) classrooms, furniture, teaching materials, potable water and sanitation facilities (UNESCO 2005). Typically, wealthier households have access to betterresourced schools (Bashir et al. 2018). Crucially, there is also a severe shortage of suitably qualified teachers as the rapid expansion of schooling across SSA has not been matched with a similar expansion of a well-prepared teaching workforce (Mulkeen 2005;UNESCO 2015;Bashir et al. 2018). Remote rural areas in almost every country suffer most in this regard due both to inequitable recruitment and deployment processes, and because teachers generally prefer to live in or near urban areas (Mulkeen 2005 of female teachers in rural areas is particularly detrimental to female enrolment (Stromquist, Klees and Lin 2017).
Teacher shortages are exacerbated by teacher absenteeism and late-coming, which are endemic across the region (Bennell and Akyeampong 2007;Bold et al. 2017;Bashir et al. 2018), including in Ghana (Blampied et al. 2018;Ministry of Education 2018), and are often the key quality concerns expressed by students and parents, which may turn them against the schooling on offer (Boyle et al. 2002;Ananga 2011). This is especially notable in rural areas, where teachers may absent themselves for several days each month to get their salary cheque (Mulkeen 2005;Bennell and Akyeampong 2007). Teacher absenteeism is likely to be a central factor in why interventions to widen access may not necessarily decrease student absenteeism (e.g. Fisher et al. 2017) as children tend to miss school on days when they know there is no teacher to teach them (Ananga 2011).
Arguably, however, it is the quality of interactions between teachers and learners that is of the greatest importance in determining the quality of learning (Dembélé and Miaro-II 2003 Bashir et al. 2018), often in sub-optimal physical teaching and learning conditions in a language in which they (and/or the learners) may not be proficient (Lauwerier and Akkari 2015; Centre for International Teacher Education 2016).
Indeed, language-in-education policies and practices in the classroom are critical quality issues. They are pivotal in determining learners' educational experiences, their learning outcomes (Bashir et al. 2018) as well as their sense of identity and belonging (Benson and Kosonen 2013;Odugu 2017). Many students (especially in rural areas) learn in a language to which they have limited, if any, exposure outside school -a fact which is often linked to learner absenteeism, high repetition rates, poor learning and exam performance, and eventual school exclusion or dropout (Benson 2004;Alidou et al. 2006;Salami 2008;Smits et al. 2008 In broader terms, the curriculum has been subject to post-colonial critique that traces problems with relevance and quality to the colonial foundations of education, which have excluded indigenous knowledge and values in a system rooted in colonial 15 Article 30 states that: 'In those States in which ethnic, religious or linguistic minorities or persons of indigenous origin exist, a child belonging to such a minority or who is indigenous shall not be denied the right, in community with other members of his or her group, to enjoy his or her own culture, to profess and practise his or her own religion, or to use his or her own language.' 16 See, for example, Mirembe and Davies (2001) Compounding the school quality issues discussed above relating to the infrastructure, resources and formal curriculum are those relating to the informal or hidden curriculum in the daily operation of schools. Its influence is just as critical to how students experience school, their learning, whether they stay in the system, and their formal learning outcomes, as we illustrate in the next section. There is, therefore, now greater awareness within the education literature that while schools can be sites of learning, they can also be sites for violence, with the potential to cause children harm, damaging their physical and mental health, social relations and future livelihood opportunities (Harber 2004;Leach and Mitchell 2006;UN 2016). With regards to education, specifically, school violence is a major contributor to poor learning experiences and attainment, student absenteeism and eventual dropout (Harber 2004; Antonowicz 2010; Ananga 2011; UNESCO 2019).
As highlighted in this paper's Introduction, in mainstream education literature (outside the field of social protection), the term 'violence' is more commonly discussed, rather than 'harm'. In SSA, gender violence, corporal punishment and bullying are the three areas of obvious harm that have come to greatest prominence in the educational literature over the past couple of decades, propelled by high-profile international campaigns for their global eradication, usually framed within the discourse of child rights. These different forms of violence can all be conceptually linked within an analysis of school as a gendered institution operating within broader gendered societal norms (Connell 1987), and within violent historical processes, such as colonialism and apartheid (Harber 2004;Epstein and Morrell 2012; Adzahlie-Mensah 2014). Yet corporal punishment and bullying, in particular, are more often studied separately, within gender-neutral discourses of 'victim' and 'perpetrator/bully', with gender only considered as a categorical variable ).
As we elaborate below, many of these forms of school-based violence are acutely felt by children from poorer households and those in rural areas.
To better understand why, we apply our analytical edu-workscape to link these children's poor socioeconomic circumstances and gendered burdens of out-of-school work to gendered patterns of violence and labour in school. These collectively contribute to harmful educational experiences and poor learning outcomes -factors which are likely to push them out of school, or may dissuade them from entering in the first place.

Discipline, corporal punishment, and bullying
Though contexts vary, the colonial institution of modern schooling is remarkably consistent across SSA in its promotion of aggressive masculinities, male entitlement, and authoritarian teacher-student relations (e.g. Tafa 2002; Ngakane, Muthukrishna and Ngcobo 2012). This gender hierarchy is sustained, to a large extent, through punitive disciplinary regimes, which are still the norm in SSA, despite government and NGO attempts to bring about change, such as through UNICEF's child-friendly schools initiative. Central to school regulation is corporal punishment, backed up by a system of prefects and monitors, who also often have the authority to physically discipline their peers (Morrell 2001;Humphreys 2008;Dunne et al. 2013a). Although corporal punishment is now illegal in schools (and sometimes in other social arenas) in most SSA countries, there has been 'little progress' on its elimination (ACPF 2014(ACPF , 2018. This is largely due to the fact that it is widely accepted as a legitimate disciplinary practice by teachers, parents/ caregivers and children themselves -provided it is done in moderation and with the aim of correcting errant behaviour (Twum-Danso Imoh 2013; Masko and Bosiwah 2016). However, as the evidence presented below illustrates, its application is often excessive and children are frequently punished for issues outside their control. This is generally the case with children who are poor and are late or absent from school due to heavy household and external work commitments. Other physically demanding and humiliating school punishments are also commonly practised: children may be made to squat, frog-jump round the school compound in the sun, or kneel on hard floors and maintain particular uncomfortable positions, such as holding their hands above their head. Such punishments frequently take place during lesson time, thereby exacerbating the harm they cause by taking further potential learning time away from the child.
In general terms, boys tend to experience more corporal punishment at school. Younger boys tend to be beaten more than older boys, whereas girls and older boys may receive more verbal reprimands, which are often insulting and denigrating (Tafa 2002;Dunne et al. 2005;Soneson 2005aSoneson , 2005bHumphreys 2012;Bakari 2013;Pankhurst et al. 2016). Evidence from several countries in SSA also suggests that corporal punishment is generally experienced more often and more severely in schools in poorer communities and in rural areas (Soneson 2005a(Soneson , 2005bHumphreys 2008; Adzahlie-Mensah 2014), where teachers work in more challenging conditions and more often find themselves unable to cope (Sherry 2008 Jones et al. 2019), which children (or parents/ carers) may object to more than the punishment itself. As one girl in a study in Zambia complained: 'I was punished for coming late to school. I cleaned the toilets and cultivated in the garden; that is the punishment I was given. Should have punished me after class' (Soneson 2005a: 16).
In addition to the widespread violence by teachers against students across SSA, a recent global synthesis of prevalence studies confirmed that the region also has the highest reported rates of school bullying, with just under half of all students saying that they had been bullied, while just over a third reported physical attacks or being involved in physical fights -though there is wide variation across countries (UNESCO 2019). Physical bullying was most commonly reported by boys rather than girls (25.4 per cent to 18.7 per cent) but one in ten boys and girls complained of sexual bullying. Bullying rates for Ghana were high for both girls and boys, with around three in five experiencing school bullying. Bullying or 'teasing' is commonly reported by rural children from poor families who cannot afford the proper school uniform (Human Sciences Research Council et al. 2005). Girls too are often 'teased'/bullied by boys if they leak blood at school during menstruation, resulting in regular absenteeism and, in some cases, dropping out of school completely (Fehr 2010;Sommer et al. 2014;Coast et al. 2019). Peer violence is often compounded by teacher complicity in ignoring its occurrence, thereby normalising violence in schools Dunne et al. 2005).

Sexual violence and the sexual economy of schooling
Although studies on gender violence in schools may include consideration of corporal punishment and bullying, they have predominantly focused on sexual violence against girls Leach et al. 2014). Only more recently have national prevalence surveys among children and youth indicated that many boys also suffer sexual violence in school, though largely from their peers (UNESCO 2019). For example, over a quarter of male respondents in Nigeria (26.6 per cent) said that their first experience of sexual violence was perpetrated by a classmate or schoolmate (ibid.).
Evidence too exists of (predominantly) male teachers demanding sexual favours from girls (or, to a lesser extent, boys) in exchange for preferential treatment, including better grades and/or monetary assistance (Jones and Norton 2007;Antonowicz 2010;Postmus et al. 2015). A review of reports and studies on gender violence in West and Central Africa also reported boys 'procuring' girls for teachers in exchange for reduced fees (Antonowicz 2010). In some cases, however, girls may actively seek out sexual relationships with teachers as part of the sexual economy. As with 'sugar daddies' outside school, such relationships often enable girls to pay school expenses, and may be encouraged by the family (Jones and Norton 2007;Antonowicz 2010;Parkes et al. 2013). In such ways, education is implicated in sexually exploitative activities, which the ILO would likely categorise as belonging to the 'worst forms of child labour'.
The ill-effects of these multiple forms of violence experienced by children in schools in SSAfrequent corporal punishment, bullying, and verbal, emotional or sexual violence -include increased anxiety, preventing children from participating in class (for fear of being wrong and risking further punishment) (Feinstein and Mwahombela 2010) and depression (Cluver, Bowes and Gardner 2010). School violence can also affect students' concentration and learning, resulting in poor attainment (Talwar, Carlson and Lee 2011;Stein et al. 2019). It can lead to truancy, absenteeism, and eventual dropout (Soneson 2005a(Soneson , 2005bAnanga 2011;Dunne et al. 2013b;Acquah, Wilson and Doku 2014;Kibriya, Xu and Zhang 2015;Pankhurst et al. 2016). In addition, two studies in Ghana have associated school bullying at secondary school level with suicidal tendencies (Asante et al. 2017;Baiden et al. 2019). Sexual violence against girls may result in unwanted pregnancy, which generally spells the end of their formal schooling. Even where school policies allow young women to continue, bullying by classmates, economic constraints and/or lack of childcare usually make it impracticable (Wekesa 2011).

Child labour in school
Although the ILO and World Bank have belatedly recognised the importance of taking into account children's gendered domestic workloads (World Bank 2005; ILO 2017a) when measuring and quantifying child labour, schools as sites of labour have yet to be given much consideration. The underemphasised, gendered child labour regimes of schools in SSA that we report on below add to the burden of work that many children are already experiencing at home and in the workplace, and to the potential harm that schools can cause children.
They therefore need to be taken into account in debates about the tensions between child work and schooling, and when making judgements about school quality.
School-based studies in various national contexts in SSA, including in Ghana, have noted how children can spend anything from an hour to a whole day cleaning various parts of the school, toiling on the school farm, or fetching water and sand for school construction projects (Boyle et al. 2002;Berlan 2004;Benavot and Gad 2004;Odonkor 2007;Seidu and Adzahlie-Mensah 2010;Ananga 2011;Casely-Hayford et al. 2013;Humphreys et al. 2015;Adonteng-Kissi 2018). Classroom and school cleaning, in particular, is often officially timetabled.
Given that scheduled teaching time may be no more than 3-5 hours per day, especially in doubleshift schools, the time children spend labouring represents a considerable proportion of school time.
Practical components of curriculum subjects can also comprise physical work. Examples include picking up litter in Environmental Studies in Botswana (Silo 2013), or labouring on the school or teacher's farm for Physical Education lessons in Ghana (Seidu and Adzahlie-Mensah 2010), or in Agriculture classes (Riedmiller andMades 1991, cited in Maïga andKazianga 2016). This is in contrast to the valuable agricultural learning that can be done with supervised activities, where children and their families may also get to share the benefits (e.g. Okiror, Matsiko and Oonyu 2011, in Uganda).
Often it is schools in the poorest rural communities that receive the least government funding (Bashir et al. 2018) and therefore require more community and household labour (which inevitably includes children contributing) to build new classrooms and furniture and/or carry out school repairs (Swift-Morgan 2006; Essuman and Akyeampong 2011). Even the generally advantageous school feeding programme can result in extra labour demands on children. Studies in northern Ghana have noted how children may miss lesson time and walk long distances to fetch water and firewood for cooking the meals; in addition, schools that are short of eating utensils have to reuse them, which causes the whole process to reduce class time (Sulemana, Ngah and Majid 2013;Salifu, Boateng and Kunduzore 2018). A study of two rural schools in Tanzania found that children routinely had to bring water and firewood with them to school or, more commonly, got sent to fetch water for the school several times a week during school hours.
Refusal to comply could result in punishment, which in turn could drive children from school (Levison, DeGraff and Dungumaro 2017).
In Ghana, there is also ample evidence that schools can insist on students undertaking paid labour on farms and/or bringing sand and stones for construction to generate income to help run the school (Berlan 2004;Ananga 2011). Teachers also often use students as unpaid workers on their own farms and in their homes (Berlan 2004;Hashim 2004;Odonkor 2007;Ananga 2011;Seidu and Adzahlie-Mensah 2010;Casely-Hayford et al. 2013;Maconachie and Hilson 2016), which some students and/or carers resent (Ampiah and Adu -Yeboah 2009). This additionally puts girls (in particular) at risk when cleaning male teachers' accommodation (Antonowicz 2010; Shumba and Abosi 2011). Evidence from Central and West Africa indicates that sometimes communities and schools (or school staff) have agreements on using children's labour as a form of remuneration or incentive for teachers (Hashim 2004;Antonowicz 2010).
As Berlan's (2004) rural ethnography in a cocoaproducing area of Ghana highlighted, the labour in school can be more arduous and dangerous than work carried out on family-owned cocoa farms. In school, children were clearing the ground with machetes in the hot sun whereas on the cocoa farms there was more shade and less strenuous work carried out under greater supervision.
Parents (and/or the children themselves) have stated on occasions that they would rather benefit from the work themselves than give free labour to teachers and schools (Odonkor 2007; Seidu and Adzahlie-Mensah 2010; Ananga 2011). The following opinion voiced by an older Ghanaian boy sums up student resentment, which was widely reported: I do not like it when the teachers require us to bring sand and stones to school. We don't know what they do with the money… I also need money, so I prefer to work and make money for myself instead of making it for the teachers. Frequently, school labour replicates the gendered patterns of labour within the household, with girls and younger children shouldering the greater burden (Antonowicz 2010), although in some cases school tasks can be less gender-typed (see, for example, Casely-Hayford et al. 2013, in Ghana). Generally, girls tend to spend longer sweeping inside the classroom, collecting firewood and water (especially if school meals are cooked on site) (Sulemana et al. 2013) and cleaning latrines, whereas boys will move furniture or do heavier manual jobs outside, such as digging trenches, cutting down trees, and cleaning the compound (Boyle et al. 2002;Dunne et al. 2005;Antonowicz 2010;Casely-Hayford et al. 2013;Leach et al. 2014). The inequitable gendered distribution of school labour has been reported as a source of resentment by female students in various studies: We are always bullied by teachers. You can find ever time teachers telling us you girls go and fetch some water, or you girls go and clean the toilets, you girls go and do that, but boys never told to go to fetch water neither to clean toilets, but they are left playing football. All these we are doing during the classroom learning activities. Boys are not given these duties and they are telling us we have to do so because we are the females, these are the duties which traditionally are done by women at home. (Stein et al. 2019: 95) Labour tasks may also be allocated as punishments, as explained earlier, such as cleaning toilets (for girls), digging trenches or weeding (for boys) or working on the farm (Soneson 2005a(Soneson , 2005bPorter et al. 2011;Dunne et al. 2013;Levison et al. 2017), which may also cause parental resentment (Adonteng-Kissi 2018). Given the lack of water in many rural schools, and the fact that animals (or humans) might have defecated in the classroom, the allocation of toilet cleaning to girls has serious health implications (Humphreys et al. 2015).
In summary, as the evidence in sections 5 and 6 indicates, many rural schools in SSA have serious issues with educational quality (however quality might be defined). Rather than a safe haven from the presumed harm of work elsewhere, they are also sites of multiple forms of violence and harmboth sanctioned and unsanctioned -and spaces for physically demanding gendered work. Collectively, these common features of rural education in SSA can drive students out of school altogether. Moreover, the evidence shows that these multiple harms are likely to be greater for children from the poorest households as schools' inflexible timetables, poor-quality teaching, and harsh, inequitable disciplinary regimes compound the challenges that working children face when attempting to combine schooling with work and other social commitments.

The household -learning (gender) work
The second institutional setting in our edu-workspace is the household. Until relatively recently, research on child labour produced by organisations such as the ILO and World Bank have taken scant account of the work done in households, which is integral to the child's experiences of their edu-workscape. Unpaid work within the home makes an important contribution to sustaining the household and is integral to the daily life of many children ( Children's contribution to household labour is part of an intergenerational social contract which both adults and children value (Abebe 2011), but is under constant negotiation and in tension as children grow up and try to balance collective responsibilities and obligations with individual needs and desires in a changing environment (Hashim 2004;Tafere 2013;Kassa 2016). By helping parents and carers with various types of domestic work, including sibling care, children also free up the adults to undertake more paid employment (Pankhurst et al. 2016).
Despite the household being a significant site for child work, especially for girls (Chant and Jones 2005;World Bank 2005;Webbink et al. 2012), it is only relatively recently that household chores have been categorised as 'work' by the ILO (ILO 2017a).
Since girls tend to spend longer on such domestic tasks, it was belatedly recognised that by not including domestic labour in estimations of work, girls were being disadvantaged (World Bank 2005; ILO 2013), though it is still a contentious issue (Koolwal 2019). Even the term 'chore' suggests something less serious than 'work'. In underplaying the importance of domestic labour, the dominant discourse on work and labour reinforces the gender hierarchies within the household that intensify a gendered/sexual economy in which girls and women constantly struggle against systematic subordination (see section 9 for further elaboration).
The intensity of household work increases with age and the various tasks become more genderdifferentiated (Admassie 2003;Abebe 2011;Tafere and Pankhurst 2015;Young Lives 2018). Young boys and girls may both collect firewood and water, for example (Orkin 2012;Kassa 2016;Levison et al. 2017) whereas as they get older, boys tend towards farming while girls' domestic load increases in cleaning, cooking, care work, and household management, as well as paid and unpaid farm work (Abebe 2011;Orkin 2012;Pankhurst et al. 2016). It is worth noting too that domestic work, such as lighting fires (often with kerosene) and cooking, can be just as hazardous as paid work (Odonkor 2007).
Allocation of work is dependent on sibling order, household member configurations (including fosterage, for example) as well as temporary or permanent changes in household composition (through migration, for instance) such that gender work stereotypes may not always be strictly adhered to (Hashim 2004;Robson 2004;Heissler and Porter 2013). In general, however, girls take the greater domestic work burden which, along with expectations of marriage and child-bearing, frame judgements about what work and education is appropriate for females (Tetteh 2011;Egan 2013). While this may prepare children for stereotypical gendered work opportunities in both the domestic and 'productive' arenas of adulthood (Hashim 2004;Abebe 2011), it also has direct implications for schooling.
Different kinds of household work are more or less compatible with schooling. For example, as mentioned earlier, studies in Ethiopia have shown that boys minding cattle may also be able to study whereas girls doing household chores often cannot (Rose et al. 1997;Orkin 2012). On the other hand, in other contexts, the relative flexibility of domestic labour compared to the inflexible timetables of some paid employment might enable some girls to attend school more readily than boys (Moreira, Rabenevanana and Picard 2017). However, the greater household need for girls to care for siblings or complete their chores before attending school often results in them arriving late and being punished (Porter et al. 2011;Dunne et al. 2013a), as highlighted earlier, in section 6.
In poor communities with no access to water and no electricity, there will be increased time demand for household chores like fetching water and collecting firewood and less time for formal education (Porter et al. 2011;Webbink et al. 2012). Ironically, schemes aimed at providing income-generating opportunities for adult women, such as Ethiopia's PSNP and Rwanda's Vision 2020 Umurenge Programme (VUP), can actually increase the amount of time that children, and girls in particular, have to spend on substitute childcare and domestic chores (Jones, Tafere and Woldehanna 2010;Roelen 2015).
The household is also the primary site for learning. Through participation in work, children use their mental and physical abilities beyond those required in school and this can build children's confidence, self-esteem, social status, and belonging (Boyden 1994;Woodhead 1999;Mortimer 2010;Thorsen 2012;Aufseeser 2014;Bourdillon 2017;Young Lives 2018). In addition, the collaborative contribution of children to their household is reputed to encourage pro-social behaviour and a willingness to help (Coppens et al. 2016;Bourdillon 2017).
Gender relations, interacting with age and seniority (and other social markers of identity), are central to the social dynamics of household and community life. The social learning through household work instantiates these social hierarchies (Heissler and Porter 2013), and is often regulated through the use of corporal punishment. This is permitted in the African Charter on the Rights and Welfare of the Child (1990), provided it is 'with humanity and [is] consistent with the inherent dignity of the child'. Indeed, physical disciplining is regarded as a necessary process of socialisation into responsible adulthood and a sign of love (e.g. Archambault 2009;Frankenberg et al. 2010;Twum-Danso Imoh 2013;Chuta et al. 2019). If children return home late from school, or insist on going to school rather than doing chores, they may be beaten in the home (Hashim 2004;Soneson 2005aSoneson , 2005bMurphy et al. 2020) or deprived of food (Pankhurst et al. 2016;Adonteng-Kissi 2018). As part of this socialisation, as the children get older they, in turn, physically discipline their younger siblings for similar misdemeanours (Twum-Danso Imoh 2013; Pankhurst et al. 2016;Isimbi et al. 2017 Nevertheless, as with corporal punishment at school, although many children (especially younger ones) are accepting of such punishment (Pankhurst et al. 2016), some children may be bitter, resentful and angry when punishments are seen to be excessive and/or undeserved (Frankenberg et al. 2010; Twum-Danso Imoh 2013) and they may run away from home temporarily (ACPF 2014), or may migrate permanently to escape relationships that they consider to be abusive (Hashim 2004;Murphy et al. 2020).
Notwithstanding the significant learning that takes place within households -not least about gender identity and position -domestic work reinforces gendered hierarchical relations through social regulation in forms of physical, psychological, and symbolic violence (see section 9). These are often normalised as part of 'growing up' in households, schools and community life and are rarely included in notions of harm. Contextual understandings of these conditions of daily life within households and their bearing on schooling and life outside -which collectively comprise our edu-workscape -are essential to enhancing our understanding of the relationship between education and children's work (Bray 2003;Abebe and Bessell 2011).

Workplaces -learning at work
In this section we focus on the third element of the edu-workscape -the workplace. Much of the literature motivated by anti-child labour campaigns is concerned with the extent and nature of children's work and its harmful effects, particularly on children's health and perceived denial of their right to education (e.g. Tulane University 2015). Yet, while there are clearly contexts and conditions in which children's work may be detrimental to both, the overall picture is inevitably more nuanced. Importantly, in SSA, including in Ghana, the notion of working hard (in both domestic and waged labour) is part of being a 'good child' (Tafere 2013;Jonah and Abebe 2017) and a reciprocal intergenerational social pact between adults and children that is central to children's transition to adulthood within household and community life (Berlan 2004;Tafere 2013;Kassa 2016).
Research studies have shown that many children manage to combine work -of some sort -and schooling (Orkin 2012 . However, combining school and work becomes harder as children get older and their opportunity costs increase, and as they progress from primary to secondary school (Lewin 2009). Moreover, to attend secondary school, rural children often have to travel further and may have to pay for boarding, which adds to the already high costs of secondary education (Ohba 2011;Porter et al. 2011;Jones et al. 2019).
As in the home, however, the different kinds of waged work available have a significant bearing on the compatibility of work and schooling. In parts of Ghana, boys may have to give up the whole day to go fishing or farming, whereas girls may be able to engage in petty trading outside school hours (Ananga 2011). Whether work can be broken up into small blocks of time outside school hours is also a factor. This was found to be possible in Ethiopia when children worked for individual farmers, or helped on the family farm, where they could finish the work after school. In contrast, piecework tasks on commercial farms, fishing or herding cattle (if cattle need to be taken far from home to find pasture) also required a whole day, and therefore competed with schooling (Orkin 2012).
In Madagascar too, among semi-nomadic fishing communities, boys may be fishing all day, eventually undertaking more lucrative seasonal fishing migrations, prohibiting any further schooling, whereas girls do part-time octopus fishing on the local reef, which, along with their domestic labour, they can combine to some extent with school (Moreira et al. 2017). However, as discussed in section 3, the compatibility or incompatibility of certain types of work with schooling depends as much on the flexibility in the structures and processes of schooling, as on the nature and location of the work undertaken (Orkin 2012). As with domestic chores, children's ability to juggle paid/unpaid work and schooling on a daily basis will also depend on the various contextual factors highlighted in section 3.
The increasing demands on children's time can lead them on the gradual path to school dropout as attendance becomes more irregular and it becomes harder to keep up with school work (Ananga 2011; Pankhurst et al. 2016). The Young Lives data in Ethiopia also indicated that dropout was most commonly precipitated by illness or injury, either of the child or a family member, which forced a change in the child's role in the household livelihood strategy (ibid.). This resonates with other studies (e.g. Ampiah and Adu -Yeboah 2009 in Ghana) where illness or some other shock led to children leaving school to engage in substitute paid or unpaid labour or care work. It then becomes increasingly challenging for these older/over-age working children to find a pathway back to education (Hunt 2008;Dunne and Ananga 2013).
For many of the poorest children, schooling would not be possible if they did not work (Woodhead 1999;Orkin 2012;Bourdillon et al. 2015;Jonah and Abebe 2017). Income generated from children's work may be used to support their basic needs, which often includes the costs of their own schooling and/or that of other family members (Abebe 2011;Orkin 2012;Thorsen 2012;Wambiri 2014;Tafere and Pankhurst 2015;Bourdillon 2017;Jonah and Abebe 2017;Young Lives 2018).
In contexts of rural poverty, children's work often enables them to contribute to household income (Young Lives 2018), which can be a source of pride (Jonah and Abebe 2017). Beyond the vital economic contribution to sustaining their household, children are learning vital work-related practical and social skills (Woodhead 1999; Abebe and Bessell 2011). Working alongside parents or other household adults, be it farming or fishing, is often an informal apprenticeship (Krauss 2016;Moreira et al. 2017) or situated 'education' (Dyer 2013: 606) -a preparation and induction into becoming valuable adult members of society, as well as providing opportunities to develop entrepreneurial skills (Abebe 2011;Tafere and Pankhurst 2015;Bourdillon 2017;Young Lives 2018;ILO 2020). Learning about their interdependence in both their social and natural environments is important for young people's future livelihoods (Bolin 2006).
Within households too, skills acquired by children through work have been transferred to their parents so they can improve household incomes (Taylor et al. 2003). With evidence that working children have less-educated parents (Dar et al. 2002;Moreira et al. 2017), such reciprocal learning may begin to break down the intersecting cycles of poverty, poor education, and harmful child work.
Several studies indicate the learning synergies for children engaging in both school and work, with some suggestion that moderate work may be linked to improved attainment of low-performing students (Staff and Mortimer 2007). Work experiences can be drawn upon to learn new vocabulary, explain abstract concepts, and to understand some aspects of the curriculum, such as science, reading and writing, geography, mathematics, communications, social studies, food, nutrition and health (Nunes, Carraher and Schliemann 1993;Taylor et al. 2003;Aufseeser 2014;Banerjee et al. 2017).
There is some limited evidence that work provides children with greater confidence, skills, and connections to enable them to develop small businesses to generate funds and savings to support future studies (Marsh and Kleitman 2005;Tafere and Pankhurst 2015).
The complementarities of work and education have also been the focus of substantial private and non-government sector activity (Spielman et al. 2008; Jjuuko, Tukundane and Zeelen 2019). These are largely short courses delivered by extension service providers, farmer training centres, farm institutes, input supplying companies, and NGOs (Wallace 2007;Kirui and Kozicka 2018). The intake consists predominantly of those who have missed out on basic education, and the courses provide opportunities for skills acquisition and employment to improve social and economic inclusion. Different configurations of training and work often emphasise practical application, modern agricultural methods, mobile technology use, and agri-businesses (Kirui and Kozicka 2018). Evaluation studies indicate the positive impact of combining education and work, highlighting higher yields from farming and the potential for wider social and educational impacts (Akintoye, Adekunle and Kintomo 2012;CFS 2015).
Yet while there are plenty of potential benefits of combining work and education, it should not be forgotten that even putting aside the ILO's disputed classification of 'hazardous jobs' and the 'worst forms of child labour', violence, exploitation, and harm can be present in the workplace, as in the household and at school. Children can be beaten for not doing a job satisfactorily or can have their pay withheld (Orkin 2012;Pankhurst et al. 2016).
This choice and/or obligation for rural children to work takes place in a changing context of global, national, community, and socioeconomic realities, age-related gendered social and economic responsibilities within the household, and the likelihood of very little learning going on in school. Harsh discipline or abuse can be a feature of life in the household, in the workplace, and at school. Indeed, as our earlier discussions have illustrated, harm can occur across the edu-workscape, in the schools, workplaces and households that comprise the complex social arena of children's lives (Woodhead 1999).

Moving forward
In section 4 we offered the edu-workscape (see Figure 1) as a means to reconceptualise the field of children's work. The relational triangular matrix presents a complex social terrain through which children experience and navigate their lives, as the following quotes aptly illustrate: [Children's] work concurrently involves interdependent realities of survival, socialisation, participation, abuse and exploitation. (Abebe and Bessell 2011: 765) The relationships between child work and schooling are multi-directional, complex, and often specific to particular situations. (Bourdillon et al. 2009: 110) The focus in the paper has been on schooling in specific relation to children's work and harm. Our review, however, has thrown certain conceptual fragilities into relief, which relate to methodological aporia in the field. In this section, we will elaborate some of these conceptual and theoretical problematics, especially as they pertain to the contingent and relational matrix of the edu-workscape. We begin with discussions of context, childhood, and gender in advance of proposing areas and methodologies for further research. Differences in the definition of key concepts that frame the field (for example, chronological age boundaries of 'childhood') also imply particular kinds of social relations across different contexts. For example, the legal separation in international conventions of work and school for those under 15 suggests, firstly, that childhood is a time of not working, and secondly, it implies the separation of school from work. As we have stated earlier, neither of these conditions applies in large sections of the global South, where children (under 15) are often engaged in work (including within the school) and do combine schooling and work (Agbu 2009).

Understanding context
The notion of 'context' itself is multi-layered and relational, encompassing household, community, national, and international levels, including connections, disjunctions, and tensions within and across these overlapping and permeable categories. Exploration of the education-work nexus in context calls for research to be socio-historically and politically situated (within histories of colonialism and violent conflict, for example), paying attention to international and local economic and political forces (such as globalisation, neoliberal economic policies and the development of agri-industries), and sensitive to changing and competing cultures, traditions, and identities.

Childhoods
Conceptions of the child are central to our concerns with education and with child labour. Many critical studies point to a 'banal developmentalism' that refers both to approaches to economic and social development in the global South as well as in the conceptual framing of childhood (Nieuwenhuys 1998;Burman 2010). The overbearing 'Western' developmental models are seen in the projection of singular, linear, staged trajectories of both capitalist economies and of childhood (Agbu 2009). This 'banal developmentalism' is associated with the production of the child as a 'modern' subject through a Northern cultural imaginary characterised by 'affective investment in childhood', the normalisation of children at 'play', and the pathologisation of children at work -all saturated with moral opprobrium and occluded assumptions around household organisation, gender, wealth, class, race, etc. (Burman 2010: 14;Burman and Stacey 2010;Egan 2013).
A universalised linear chronology used to define age-related stages is a key reference for many of the international development goals. Age life-stage hierarchies are instrumentalised to define the extent and kind of work appropriate to particular ages in ways that effect a separation of childhood from work. It is this conceptual separation that is reiterated in the opposition between work and schooling. The latter is deemed to be an appropriate activity for children defined by years of age, even in contexts where birth age is not always recorded or known (Berlan 2009), and where children carry out adult roles and may not self-identify as children (Abebe and Ofusu-Kusi 2016 (Abebe and Bessell 2011), as it obscures the gendered processes and inequalities at work. These contextual and relational perspectives represent a shift in theoretical ground away from conceptions of the child in atomised behaviourist terms making individualistic rational choices about their lives and livelihoods (Bourdieu and Wacquant 1992).
Research on childhoods outside the strictures and assumptions of the international norms and legislative frameworks then has the potential to open up empirical and theoretical spaces to generate better understandings of multiple edu-workscapes in a wide range of different local contexts (Nieuwenhuys 2010;Okyere 2013; Twum-Danso Imoh 2011).

Gender
Gender relations are central to the social dynamics of household and community life. As we have illustrated in section 6, gender differences and inequalities are (re-)produced in the institutional life in schools. This normalisation of gender hierarchies and the subordination of women are echoed in the gender regimes in multiple arenas of social life, including workplaces (Connell 1987;Butler 1997;Oyěwùmí 2005;Lugones 2007). Significantly, these often harmful social experiences of gender subordination are normalised within household, school, and community life. Understanding the particularities of just how gender (as it interacts with other markers of identity such as age, socioeconomic status, location, ethnicity, religion, etc.) operates in different local social contexts is critical to addressing the gendered division of labour and the production of gender identity narratives that sustain these fundamental inequalities.
Turning more specifically to work, it is important to highlight that in the dominant development discourse, waged labour is used as the key definitional category and as a default against which all other kinds of work (formal versus informal, paid versus unpaid) are constructed. That said, such distinctions are, in practice, more blurred (Dar et al. 2002) in the agriculture sectors in SSA, where most children work (IMF 2015;OECD and FAO 2016;UNDP 2016). More than 85 per cent of work in Africa is in the informal sector, which involves large proportions of women, young people, and those with low educational outcomes (ILO 2018 Further, women's reproductive work, acknowledged as central to capitalist economies (Pateman 1988;Butler 1997;Oyěwùmí 1997), remains largely out of sight even though this reproductive potential overshadows the ways that girls navigate their edu-workscapes and future lives. Taken together, many quantitative data and analyses sustain the invisibility and misrecognition of women's and children's work Mies 2014;Prügl 2020).
As a whole, women remain over-represented in insecure, low-paid, part-time positions (McDowell 2014;ILO 2017c). The problematics of what is defined or measurable as work, however, apply to both women and men in precarious rural economies in SSA. Indeed, the ILO (2017c) has highlighted that almost 70 per cent of those with jobs across SSA and especially women and young people are in 'vulnerable' employment. This is defined as meaning they are working within the household or on their 'own account'. This 'own account' work is highly gendered, as girls face higher expectations than boys that they fulfil domestic chores and work on the household farm alongside their schooling (Bourdillon et al. 2015;Crossouard, Dunne and Szyp 2021). In addition, boys generally have greater and more lucrative employment opportunities than girls (Abebe 2011;Okyere 2013), who turn to 'hawking' and 'hustling', which may include sex work (Dunne and Ananga 2013). The reported prevalence of 'transactional sex', often to pay for education, is in part because most better-paid jobs are not available to girls (Leach 2003;Jones and Norton 2007;Petroni et al. 2017). To add to this, anticipated futures of marriage and childbearing intensify a gendered/sexual economy in which girls and women constantly struggle against systematic subordination. In these precarious conditions, the ways that girls navigate their edu-workscapes are subject to the vagaries of economic and social life, gendered expectations, and the opportunity costs of schooling.
It is evident that decontextualised survey research that relies on quantifying girls' and boys' work hours, or comparing enrolments, dropouts or attainment outcomes, cannot begin to engage with complex, contingent, and ever-changing scenarios of social and cultural life within which children work or the complex interactions between the home, school, and various workplaces.

New research spaces: the edu-workscape
In this final section, we consider opportunities for further research emerging from our review. We have argued strongly that we need a dynamic conceptual framework that takes account of the work, learning, and potential harm that can take place across three key social arenas -school, household, and workplace -and which situates these interactions within broader contexts. This re-conceptualisation also demands: a shift away from dominant discourses of development to better explore and understand the dynamics of local contexts; provision of space for more fluid, contingent, and emergent constructions of childhood; and an increase in the visibility of multiple forms of learning and work, especially those submerged in assumptions around gender.
Through the edu-workscape presented (Figure 1, section 4), we suggest a conceptual framing that focuses on the social geographies of children's lives as they move within and between the social relations of the school, workplace and household. These three interlocking domains are in play within the contingent and changing dynamics of the wider context.
The framing provided by the edu-workscape has multiple implications for how we explore and address children's work. So, for example, as elaborated earlier in this section, the notion of childhood related to chronological serial time may be neither universal nor relevant for all contexts in SSA. The proliferation of expectations of timestaged hierarchies of progress that reverberate in development discourses (national and educational) are a theoretical imposition. While we may wish to retain 'childhood' and other similar constructs, it remains important to understand them as such, while remaining open to engage with alternative theoretical structures and assumptions emergent in the field. In the discourses of rights, development, and education, a shift is required from generic universalised references to children, to make visible the ways in which these are highly gendered and multiply intersected by socioeconomic status, ethnicity, race, age, location, migration, and other markers of social identity. These all make a difference to how and when children access education, and to their experiences within the school system.
Several national governments in SSA have instituted complementary basic education (CBE) to provide programmes of accelerated learning for those who, for a range of reasons, have missed out on education. Early evaluations suggest positive opportunities for alternative forms of schooling to deliver educational rights to poor children in ways that better accommodate and support the livelihood needs and rhythms of their households. The implicit acknowledgement of children's contribution to households suggests a multifaceted understanding of poverty not only in economic terms but also for the time-poor. Research evidence indicates, however, that household and community needs rarely impinge on the organisation and operation of schools. For example, children from poor households, and especially girls, are likely to have greater demands on their time, often making them absent or late for school, resulting in punishment. Poorer health associated with poverty can lead to low concentration and slower learning, which may again lead to punishment. Further to this, as elaborated in section 7, far from being a sanctuary, the school often requires children to work and suffer violence. Beyond the physical and psychological harms of corporal punishments or sexual abuse, violence also includes more invisible harms of gender socialisation or having one's culture and language excluded from the curriculum. This raises critical questions about the normalisation of violence in schools and other arenas in the edu-workscape, and the point at which they can be considered as harm. The ways 'work' and 'harm' are stipulated for different children within different contexts and institutions appears to be an important set of issues to explore. These are likely to elucidate not only the worst forms of child work but also relate to improving working conditions in general for adults and children alike.
Successful efforts to plan, regulate or intervene in child work depend on the development of contextually specific understandings of children's lives. The implications here are for a broader range of research methodologies. Policy-influencing research on education and children's work in SSA, as elsewhere, has been dominated by quantitative data from randomised control trials, quasi-experimental studies, regression analyses of household and EMIS data, and impact evaluations (Cornwall and Aghajanian 2017). The World Bank's Facing Forward: Schooling for Learning in Africa (Bashir et al. 2018), which reports exclusively on quantitative data, is a case in point. These studies have been matched with a plethora of 'systematic' or 'rigorous' literature reviews on key policy areas. Some only include quantitative research (e.g. Guerrero et al.'s 2012 systematic review on improving teacher attendance). Yet, as one such rigorous review reflected, many of the 'high quality' (predominantly quantitative) studies paid no attention to the effect of context (Unterhalter et al. 2014). That said, there has been a more recent move towards the inclusion of qualitative data in 'mixed-methods' research, including in impact evaluations (Cornwall and Aghajanian 2017) -regarding cash transfers in SSA, for example (Davis et al. 2016;Fisher et al. 2017). The studies on working children's lives from the mixed-methods research of Young Lives (Young Lives n.d.) offer more contextually situated insights. As a longitudinal research programme, studies also captured the importance of changing contexts and circumstances on children's lives. At the same time, as more cautious qualitative researchers have noted, much 'mixed-methods' research often 'favours the forms of analysis and truth finding associated with positivism' (Giddings and Grant 2007: abstract).
Thus, while acknowledging the contributions of mixed-methods research, this paper highlights the need to invest in more exploratory qualitative research. In particular, spatial analyses can help untangle the complex social geographies of the edu-workscapes in children's lives within their community and broader contexts (Twum-Danso Imoh, Bourdillon and Meichsner 2018;Dunne, Humphreys and Bakari 2020). Such research will require more negotiated, participatory, collaborative, and emergent methodologies. There is a need to understand what is going on in all its complexities and to gain insights into how and why this is the case from multiple local perspectives, including those of children. Against the pre-ordained rigidity of survey research, the scope of these studies may be narrowed and widened according to the research focus, its motivating questions, and the processes of research as researchers enter and engage with people in the field. Iterations with more macro-data will also be informative. The edu-workscape that we have elaborated provides a framework for undertaking reflexive social research, which is vital for re-shaping conceptual, theoretical, methodological, and empirical approaches to the study of children's education, work and harm.