The Persistence of South African Educational Inequalities: The Need for Understanding and Relying on Analytical Frameworks

It is acknowledged that educational inequalities persist in South Africa 25 years after the advent of democracy in 1994, but the debate about the causes and solutions to poor education performance continues. Could the education system be fundamentally improved and mitigate somewhat the socio-economic inequalities from one generation to another? This article argues that the debate will benefit from the use of theoretical frameworks about the role of education in society, especially since many analyses do not engage explicitly with theories. This article summarises the theoretical debates to remind scholars about the need to compare and explore the arguments of different theoretical paradigms on the role of education and assist sociology of education teaching. It concludes with analytical examples, which argue that the best explanatory power comes from the post-structuralist theoretical position.


Introduction
Many scholars have argued and demonstrated that 25 years after the 1994 transition to democracy, the relationship between inequality of educational outcomes and poverty in South Africa remains strong. South Africa is known for its low gross domestic product per capita, as a large share of its population lives in poverty, high unemployment levels (27% is the formal rate, but it is more likely to be around 40%), and high levels of inequality (with a Gini coefficient of 0.66, which has remained consistent). In a similar manner, the education system is known for its bimodality, with many noting that around 80% of schools perform badly compared with the functional 20% of schools, and 80% of the maths matric distinctions come from the top 200 (out of 6,600) secondary schools. Many quantitative researchers reveal strong inequalities in educational outcomes: of 100 children who started Grade 1 in 2007, only 51 made it to matric, 40 of the 2018 matric cohort passed and 17 received bachelor's passes. Only 40% of those who began Grade 1 passed matric. In no-fee (quintiles 1, 2 and 3) schools, less than one in five Grade 9 learners reached the low international benchmark in Trends in International Mathematics and Science Study (TIMSS) 2015 in maths and science, whereas in quintile 4 and 5 schools the proportions were about 60%, and in independent schools around 80% (Van der Berg and Gustafsson 2019).
Quantitative research on South African education has been very useful in documenting different school variables and their association with educational outcomes. It shows that there has been some improvement. Van der Berg and Gustafsson (2019) note that cognitive school performance has improved significantly in the past 45 years. Whereas, in 1970, 45% of the 21-to 25-year-olds completed their primary schooling and 10% obtained matric, 95% completed primary schooling and 52% obtained matric in 2016. Furthermore, the national matric pass rate has also improved from the low 53.4% in 1995 to 81.3% in 2019. The government would argue that this shows there has been reasonable, if not satisfactory, progress in education since 1994, but the educational inequalities inherited as a result of years of discrimination during apartheid are hard to counter and cannot be undone overnight.
These figures and assessments point towards a central question underpinning this article: Could the education system mitigate somewhat the socio-economic inequalities passed down from one generation to the next? Put another way, has the performance and unequal nature of the education system improved much in the past 25 years and how could it improve its effectiveness and efficiency? Why did the many significant education reforms not transform the education system and its performance? Education policy analysts explain the impact of our education reforms in different ways. Some point to the symbolism of these policies, which is not considered implementable (Jansen 2002); others note that they were too often borrowed from overseas to keep up with the global competition without sufficiently understanding how to adapt them to suit the conditions and realities on the ground (Spreen 2001). A few others explain that they were often pro-middle class and intended to be so by the emergent middle class among government and department officials.
This article contends that many writings, including the recent book edited by Spaull and Jansen (2019) on the enigma of educational inequalities, focus their analysis on the nature and form of educational inequalities in different areas of the system since 1994. But they rarely engage explicitly with educational theories, which are needed to explain better the causes of these persistent inequalities and poor efficiencies in the education system. It is against this background that this article proposes to examine the debate on the causes of the education system's poor performance since 1994 and to identify solutions to improve its performance and persistent educational inequalities. This article aims to clarify the underlying reasons behind scholars' analyses of the different causes and suggested solutions to improve our education system, and concludes that there is a better explanatory analytical framework, which identifies the causes and offers solutions for our education system.
What follows is of course not new but is written to remind scholars of the importance of comparing and engaging with the arguments of different theoretical paradigms around the role of education and to assist sociology of education teaching.

The Need for Analytical Frameworks
Various arguments and positions have been developed by scholars to interpret the persistent educational inequalities and this article posits that it is important to unpack the different arguments by referring to three distinct schools of thought. These analytical frameworks will reveal why scholars identify different reasons or enabling or inhibiting factors and conditions to explain the poor educational performance and significant educational inequalities. This article distinguishes three "ideal-type" analytical frameworks for clarity purposes, but this does not imply that South African policy analysts are rigidly using one or the other; often they are eclectic and use a combination, though one can always detect in their analysis a dominant theoretical framework.
The article will show how and why each "ideal-type" analytical framework focuses on different factors or conditions to explain educational inequalities and will explain why the third analytical framework is considered most useful. This is because this framework takes into account the influence of both structures and agencies on the nature and evolution of the education system and explains that the power play was not sufficiently directed at challenging seriously the educational inequalities in favour of the disadvantaged.

The Liberal Functionalist Framework
The liberal functionalist position argues, after Parsons, that countries need an effective education system that functions to produce a competent and well-differentiated labour force through a meritocratic education system to meet labour market demands.
Adherents argue that education is neutral and contributes to reduce socio-economic inequalities and, in that respect, it is important to identify the school-based factors that work against that goal to allow individuals from all socio-economic backgrounds to work hard and succeed through the education system. This position would explain the inequalities perpetuated by Bantu education as an irrational, amoral and inefficient education system.
Van der Berg and his research team of economists of education from Stellenbosch were also interested to understand the binding constraints of our poorly performing education system since 1994. Their research focused on the inefficiencies of the system and the impact of specific school-based factors on the poor education outcomes of the majority ( Van der Berg et al. 2016). Their research identified the main problems at different levels of the education system that need to be addressed. This is not to say that they do not acknowledge the negative influence of societal factors and socio-economic inequalities such as home poverty, home language, parents' low levels of education, unconducive family settlement and structure. However, their analysis (as well as the 2018 World Bank report) assumes that the education system could improve by addressing the various systemic inefficiencies in order to rupture the reproduction of inequality and broaden educational opportunities for all, which will hopefully reduce the socio-economic inequalities for the next generation.
At departmental level, the poor provincial capacity to plan and administer as well as the weak district monitoring and support of schools have been acknowledged. Gustafsson and Taylor (2013) noted other institutional inefficiencies. Through value-added school production functions and fixed-effects models, they show the need for more efficient use of non-personnel funds by the authorities, with a special focus on educational materials, as these represent serious obstacles to a well-functioning system.
Other analysts focus on the school-level factors to show how school development has not been properly addressed since 1994. They note in particular the weak district support system for poorly educated/qualified principals as well as the poor and sometimes dysfunctional school leadership and its inability to ensure time-on-task in the classroom with teachers arriving and teaching on time and at their right level (Van der Berg et al. 2016). This issue is what causes some of these scholars to privilege the other side of the "change coin", the serious lack of school and teacher accountability (Taylor 2002).
At the classroom level, many scholars show the differentiated access to teaching and learning resources in schools, which hinder or enable a more conducive instructional environment. They also identify the poor professional knowledge of teachers (CDE 2013;Shalem and de Clercq 2019;Taylor 2019;Taylor and Vinjevold 1999) and their weak pedagogical instruction regime (Fleisch 2018). Another major inhibiting classroom factor is the unrealistic and overambitious outcomes-based education (OBE) curriculum policy and the difficulty of implementing it in poorly performing schools with teachers who struggled to understand and implement it (Jansen and Christie 2000). This is because many teachers in disadvantaged schools are not fully qualified and could not cope with a change of curriculum framework with little content and a different pedagogical method. This was overambitious and inappropriate to the majority of teachers and their classroom realities and conditions. Many school-based policies had a differentiated effect on various classrooms and teachers but, overall, disadvantaged teachers struggled the most. The exceptions here, according to Gustafsson (2005), are the technical efficiencies achieved through the reduction of learner repetition, facilitated by the policy prohibiting more than one repetition per schooling phase.
The language of learning and teaching (LoLT) policy was identified as problematic in that it stipulates that schools and their school governing bodies (SGBs) should preferably adopt as a language of instruction the local mother tongue of learners in the Foundation Phase. A period of three years with English or Afrikaans as First Additional Language is acknowledged internationally to be far too little to achieve academic fluency in the future LoLT from Grade 4 onwards (Roberts and Schollar 2011). This was seen as a serious obstacle, especially because statistics showed that many Grade 3 or Grade 6 learners could not read properly for meaning for that level/grade (Pretorius and Spaull 2016), a fundamental shortcoming for further learning.
Finally, these liberal functionalist scholars of the "ideal" type look at one political dimension that, they acknowledge, exists outside the education system and impacts negatively on educational performance. The weak social compact or cohesion between stakeholders, such as teacher unions, schools and communities (Gustafsson and Taylor 2013;Roberts and Schollar 2011) is seen as a serious obstacle. Their analyses often blame the South African Democratic Teachers Union (SADTU), which they show interferes with and organises their disadvantaged teacher members to boycott supervision and compulsory hours of teacher development (Van der Berg et al. 2016).
In brief, liberal functionalist scholars choose to focus on and analyse various departmental or school-based factors that hinder the efficiency of the system. These factors are conceived of in a technical manner and are believed to need better policy or programme redress strategies.
The research of these scholars who rely mostly on the liberal functionalist theoretical perspective is very similar and falls under School Effectiveness Research (SER), which became popular internationally in the 1980s and 1990s in drawing attention to various school-based factors that were correlated with education outcomes. Such SER scholars and human capital analysts in South Africa (Van der Berg and his team) who argue that enhanced levels of education raise people's and countries' productivity, stimulating economic growth and eradicating poverty (Psacharopoulos 1994), believe that their research can assist policy makers and educationists in identifying the important systemor school-based factors that need addressing. Their analyses, as Allais, Shalem, and Cooper (2019, 113) argue, position schools "as a factor independent of other social determinants, and treated as an independent policy variable which can be used to bring about the required social changes". Over the past three decades, SER has interacted with School Improvement Research (SIR), which focuses on the improvement paths and priorities of different schools, and both have evolved and improved each other over the last few decades. However, they have remained firmly located within the liberal functional paradigm, which assumes that education is and should be neutral to contribute best to the socio-economic and labour system and to upward social mobility on the grounds that education enables individuals to benefit from the improvement of school-based factors that maximise opportunities for all (Thrupp 2001).
An interesting weakness of the mainly quantitative research located in the liberal functionalist analytical perspective is that it can only establish correlations between certain variables and educational outcomes but cannot explain exactly how these schoolbased variables impact on educational outcomes. This is problematic because they recommend programmes targeting these quantitative variables to improve educational outcomes but cannot be sure how these improved variables work on educational outcomes.

The Marxist Functionalist Framework
The second school of thought is the Marxist theoretical perspective. Scholars who adopt this perspective argue that education is not neutral but functions rather as a political and ideological instrument to reproduce socio-economic inequalities through unequal educational opportunities provided to the differentiated population. Their studies focus on broader structural issues and reveal how differentiated socio-economic and education structures and factors reproduce inequalities in the next generation. In other words, their analysis does not focus on school-based factors (like the liberal position), but rather on all the unequal structures in society and in education. As early as the 1970s, Bowles (1972) produced large-scale evidence to show that the Unites States' class-differentiated population accessed and completed their education in such a way as to ensure classbased inequalities are transferred to the next generation (Bourdieu 1979;Bowles 1972;Moore 2004). Scholars who adhere to this theoretical framework, illustrated by the work of Van der Berg et al. (2011a) and Van der Berg and Gustafsson (2019), would explain the bimodal distribution of quality schooling in South Africa as a mirror of the harsh and unequal economic reality in South Africa.
In South Africa, such scholars argue that the post-1994 state did not seriously attempt a true reorganisation of the socio-economic structure to assist with the reduction of inequalities (Fine 2016). After 1994, a more political than economic transfer of power occurred, and the economic growth and investment strategy and policies did not target or subvert sufficiently the dominant monopolies (Hirsch 2005). The neo-liberal form of the economic growth paths adopted since 1996, when the Reconstruction and Development Programme was abandoned, in line with the global economic world and its increased socio-economic polarisation, did not promote economic growth for redistribution or an economic development strategy that created jobs for the poorly skilled and kept low the unemployment figures. Allais, Shalem, and Cooper (2019) note that, as a result, "the socio-economic conditions that structure learners' lives in South Africa today, as well as those of their families and communities, reproduce [strongly unequal] social and economic relations" in ways that make it unlikely that the education system can improve equality of learning and educational outcomes. For them, widespread poverty and socio-economic realities are the most substantial factors that explain inequality of educational outcomes. Allais, Shalem, and Cooper (2019) argue that in South Africa the relationships between education, poverty, and inequality are dominated by the fact that educational inequality is driven by poverty. Poverty structures learning and teaching in poor schools, in three ways: 1) learners are not well-prepared for schooling and their homes do not function as a second site of acquisition; 2) very poor schools do not have a reservoir of material and cognitive resources by any comparative measure to what rich schools have; and 3) school management in poor schools does not mediate the bureaucratic demands on teacher time to cover the curriculum in time frames which are very hard to achieve. (Allais, Shalem, and Cooper 2019, 113-14) These scholars locate the major causes of inequalities mainly within the political economy of the country rather than within the education system. The logic of this position is that the reduction of educational inequalities will not happen without some major restructuring in the unequal socio-economic structures.
The argument of Marxist education scholars in South Africa (Chisholm, Motala, and Vally 2003;Christie and Collins 1982;Jansen 2003;Kallaway et al. 1984;Shalem and Hoadley 2009) is that equal opportunities for all do not fundamentally exist because of the way unequal power relations and socio-economic inequalities are structured. They also argue that the education system is fundamentally biased against the poor and the working class in the way it is structured, accessed and completed. Their assumption is that the education system is never truly meritocratic as it functions above all to reproduce socio-economic inequalities through educational inequalities, making these inequalities more legitimate. The differentiated education system and structures ensure that unequal educational opportunities are provided differentially to reproduce the socio-economic inequalities. This occurs through various means, such as unequally distributed educational resources, the uneven quality of different principals and teachers, socially biased knowledge in the curriculum, among others, all of which contribute to low educational outcomes for the poor majority.
Thus, they show how the current bimodal school system is reproduced through the lack of support and targeted strategy to counter the poor quality of many teachers in disadvantaged schools, the poor culture of teaching and learning in many black schools and the insufficient money and resources provided to poor schools. With poor state resources, a lack of qualified and competent teachers and schools that are overstretched by the influx of learners of poor socio-economic status, socio-economic inequalities continue to translate into differentiated school performance and inequalities (Van der Berg et al. 2011b). This position foregrounds the socio-economic factors of learners' backgrounds as an obstacle to the reduction of educational inequalities.
They also argue that many education reforms work against the majority of the teacher and learner population. The OBE curriculum could only benefit competent and wellqualified teachers with resourceful classrooms. The aim of the 1996 Schools Act of keeping some of the middle-class learner population in the public school sector did not help much in minimising the polarisation between the public (and private) middle-class schools with a reasonable quality performance and high school fees and the majority of public schools. The capacity of schools to raise and fix high fees was often used by middle-class schools to maintain quality by hiring additional teachers and resources to allow the sons and daughters of the growing African middle class to access them (Motala and Carel 2019). In fact, the physical expansion of middle-class public schools came about mainly because of the initiative of the middle class rather than the government's funding (CDE 2013). Other analysts identify the problem in the minor budgetary redistribution to disadvantaged schools confined to non-personnel expenditures (Motala and Carel 2019) as an indication of a class-based financial strategy. Other examples will include the poor and inappropriate district support to improve teacher quality among disadvantaged schools (Shalem and De Clercq 2019).
Thus, these scholars who rely mostly on the Marxist theoretical perspective emphasise the unequal structures in the economy and education as main obstacles to the significant reduction of educational inequalities. They also criticise the SER and SIR findings for ignoring the importance of unequal non-school structures in influencing unequal educational outcomes as well as how non-school factors influence and interact with school factors in ensuring differentiated educational opportunities. The advantage of their analysis is that they attempt to show how non-school and school variables or factors discriminate against the poor majority and how this results in the persistence of educational inequalities.
The Post-Structuralist Framework 1 According to the third school of thought, post-structuralism, the analysis of inequalities should not be examined only through the unequal socio-economic and educational structures because structures cannot explain or determine fully what individual people can do and achieve. Proponents of this framework agree, in the case of South Africa, that the unequal socio-economic structure remained broadly untouched after 1994, inhibiting the reduction of socio-economic as well as educational inequalities. However, their argument is that individual and collective agencies are possible and can act to work 1 There are some varieties within the post-structuralist position, and this article will limit itself to its main characteristics in relation to the other two frameworks for clarity. within or to undermine the power of these unequal structures. Some will argue that the post-structuralists combine elements of the analysis of the liberal functionalists and Marxists. It is more accurate to say that the post-structuralists recognise individual or collective agencies within the constraining unequal structures and the state (which can be pressurised to represent the interests of various organisations) and that structures cannot fully determine outcomes because individual or collective agencies can work and contest to strengthen or undermine these structures.
How can the Marxist position explain, for example, the 1994 change of government? A Marxist analysis would show that there was a change of structures, which was the result of the various anti-apartheid struggles both inside and outside the country. However, a post-structuralist would show that the proponents of economic restructuring were not strong enough, especially since the unions were unable to develop significant challenging strategies or opposition to the neo-liberal forms of economic growth, which continued to accentuate and reproduce structural inequalities (Fine 2016).
Another example, which the Marxist position would struggle to explain, is the Soweto 1976 uprisings and the anti-apartheid struggles of the 1980s in education that forced the government to initiate some changes in the educational arena. Regarding the post-1994 unequal educational structures, one has to explain to what extent these were challenged and opposed by collective agencies or organs of civil society including school stakeholders (De Clercq 2010). Without the strong exercise of power by collective agencies, the post-1994 education system could not improve its inputs, processes and/or outputs in a more socially just manner. This is why the post-structuralist position on the reasons for the poor education performance of the majority of schools, 25 years after 1994, is located both inside and outside the education system, in school-and non-school-based factors, especially with the constraints existing within the unchanged political economy of the post-1994 era. In comparison with the Marxist analysis, the post-structuralist analysis goes beyond the reproduction of socio-economic and educational structural inequalities and examines the contestations, struggles and collective agencies within the various arenas of education to understand how and why some structures were strengthened or undermined by different sustained struggles.
Although the government claimed the education reforms promote equality and quality for all, the situation on the ground is different as the education system continues to function as a dual system, with 20% of the schools providing quality schooling and the remaining 80% consist of mostly poorly performing schools (see next section for further analysis). By looking at the level of opposition or resistance that school stakeholders and their organisations mustered to attain more socially just school factors or conditions, these analysts (Sayed and Motala 2012) argue that it was not enough to force the government to promote more meaningful education redress and quality for all. This is not to deny the few successes that have been attained such as the improvement of the physical infrastructure of some schools, no-fee status for 60% of schools and funding redistribution subsidies for disadvantaged schools. Teachers' working conditions also improved, mainly due to union militancy, and in particular their salary increases were, on the whole, in line with public servants' salaries, higher than the annual inflation. All these improvements show that the government felt under pressure to make some changes towards quality education in favour of the majority. However, these redress measures were not very substantial because, as the article will show in the next section, many reforms and initiatives in different areas of education did not sufficiently benefit disadvantaged schools, teachers and learners, but they did strengthen middle-class schools. This will be explained by examining how the unequal power relations in society became reflected in the education arena as well as how the strength of collective agencies was often not there to undermine the unequal structural educational inequalities.

Analysing Existing Educational Inequalities: The Advantage of the Post-Structuralist Position
This section will show how the post-structuralist analysis explains, in the author's view, more satisfactorily the impact of education reforms on the poor and persistent education inequalities, compared with the liberal and Marxist positions. As mentioned above, the scholars who rely mostly on the post-structuralist perspective argue that the post-1994 education reforms need to be analysed as a reflection of the shifting power relations and the power play (or contestation) that developed in this era (De Clercq 2010). There were strong power dynamics in the education sector after 1994 as well as contestations and struggles of different magnitudes in various areas of the education system. Let us first note some elements in the post-1994 political context. The strong collective agencies from organs of civil society that had mobilised in the pre-1994 era were demobilised and surrendered their power to a developmental state, which they thought would represent their interests. By the time education reforms were designed and introduced by the department to effect redress and move towards quality education for all (as the government claimed), many analysts warned that most education reforms were biased against the majority of disadvantaged schools and teachers (De Clercq 2010).
At the level of collective organisations such as SADTU or civic associations, one must mention that these all suffered from the flight of many people in their leadership positions to government, and their new leadership did not appear to manage to develop sufficient sophisticated strategies to mobilise their members to continue demanding and mobilising for pro-poor education policies.
Let us now look at the unequal structures and contestations or agencies in some of the major changing aspects of the post-1994 education system. At the level of the school, one of the first policies formulated, the 1996 Schools Act (RSA 1996), was, inter alia, about democratising schools and school governance. This was mainly in response to the many pre-1994 struggles against the bureaucratic oppressive school system, but there were also some compromises towards middle-class schools in the finalisation of the Act (Sayed 2002). The school governing bodies, constituted by a majority of parents, were said to be designed to make schools and their principals accountable managerially. School governing bodies were expected to act as overseers of the school budget and manage relations between different school stakeholders (Woolman and Fleisch 2009). However, the post-structuralist position would show that SGBs became a new site of power relations where various school stakeholders conflict and struggle for their interests as opposed to the interest of the school as a whole. As it emerged, according to the Department of Education Ministerial Report (DoE 2003) on SGBs, many disadvantaged schools were unable to develop democratic governance that could assist with school development. The report mentions that many schools with poor governance arrangements and operational efficiency had parents who did not manage to play their budgetary monitoring role or principals and chairpersons of SGBs who colluded to rule and govern as they wished. Another more serious report finding from the Volminck 2016 Ministerial Report (DBE 2016) was the abuse of power by the South African Democratic Teachers Union (SADTU), whose members sat on school appointment committees, as it ensured through its members that the SGBs allocate promotions and other senior positions first to their members, sometimes asking these new appointees to pay for their new jobs.
In contrast, the SGBs of many middle-class schools were assisted because their members had worked in parent-teacher associations (PTAs) in the pre-1994 era. For many of them, parents were there to bring more financial resources as well as competences and networks to assist with the schools' financial, governing or development needs, even if it is true that black parents did struggle to be recognised as equal participants in some of these schools (Woolman and Fleisch 2009). In contrast, the liberal functionalist position would argue that the Schools Act was appropriate for all schools but that more effort was needed to support and capacitate the SGBs of the poor schools and their boundaries (Heystek 2011). The Marxist position would argue that the unequal power relations and structures in education and in these schools were reflected in the relations between the school stakeholders. In contrast, the poststructuralist analysis better illustrates what happened in poor and middle-class schools through its focus on structures, power relations and agencies. One could argue in this light that democratic school governance should not have been introduced as a priority in the post-1994 era, mainly because the conditions on the ground and the unequal power relations in many poor schools were not conducive for the strengthening of school democracy, accountability and the relationships between different stakeholders.
Another school-based policy that was introduced was the 2001 White Paper 6 on Inclusive Education, which, in line with the international human rights discourse, was supposed to assist broader equal opportunities of access to all. It divided schools into schools for special needs, schools with learners with average learning barriers that were given additional posts for education specialists, and "ordinary" schools with learners with minor barriers. As Engelbrecht et al. (2016) and Makoelle (2012) note, "ordinary" disadvantaged schools lacked the resources, efficient administrative systems and suitable educators to ensure that the schools with learners with minor learning barriers could fulfil their role. In addition, the training of educators to deal with learners with learning barriers was rather inadequate, and the school-based support teams did not have much support from the district-based support teams.
This situation renders the implementation of this policy biased against the majority of poor schools. Only the middle-class public and private schools could recommend to parents of learners with minor learning barriers to consult with and benefit from the assistance of an education psychologist whom they could pay (Walton et al. 2009).
In this instance, the post-structuralist analysis notes that the policy was borrowed from overseas but did not suit the conditions on the ground, which were worse in poor schools, because it was conceptualised and implemented in a class-based manner that was not in favour of disadvantaged schools compared with middle-class schools. This position is in contrast with the liberal position, which would foreground the importance of continuing building the institutional capacity of districts and schools (Heystek 2011), while the Marxist position would emphasise the unequal structures militating against the implementation of this inclusion policy that appears rather liberal and desirable for all (Carrim 2011).
At the level of the classroom, teacher development and accountability, which are seen in the international literature as important change tools (Fullan 1982), did not improve sufficiently to ensure all teachers' improvement. Yes, the National Policy Framework for Teacher Education and Development (DoE 2006), the Integrated Quality Management System (ELRC 2003) (DHET 2011) were passed to privilege teacher development for redress before teacher accountability could come into effect. These policies appear in line with the demand of teachers' unions, and SADTU in particular, which worked hard for the adoption of the 1998 Development Appraisal System, according to which teachers had to be developed and supported before they could be made accountable. This argument is in line with Elmore's (2005, 297) request for reciprocal accountability, which he defines as "every unit of increased performance that the system demands carries with it an equal and reciprocal obligation on the part of the system to provide access to an additional unit of individual or organizational capacity, in the form of additional knowledge and skill".

and the Integrated Strategic Planning Framework for Teacher Education and Development in South Africa
However, these development-oriented policies required a lot more human and financial resources to enact adequate and meaningful development for disadvantaged teachers, and these were not mobilised and provided by the state. Many teachers complained of the lack of departmental capacity for meaningful teacher support as well as the department's greater interest in enforcing principals' and teachers' accountability, especially in poorly performing schools. This lack of departmental capacity to support in particular poor schools explains why the culture of teaching and learning of many disadvantaged schools and the professionalism of many of their teachers did not improve much after 1994. Beyond these problematic policies, Taylor (2007) concludes in his research that most school improvement initiatives and programmes initiated by NGOs (at times with some provincial departments) in the first 20 years of democracy did not impact positively on school performance. This lack of meaningful teacher development was a major problem for the disadvantaged schools. As Taylor explains, most teachers' knowledge and competences in poor schools remained inadequate and often well below the grade that they were supposed to teach (Taylor 2007;2019). Fleisch (2006) and Taylor (2002) attributed the problem first to the lack of teacher accountability in these schools. This was a problematic position for struggling schools and teachers who needed more initial support. Fortunately, Taylor later changed his position and recognised the need for better teacher support given the poor pre-or inservice professional development programmes and the lack of departmental or university training capacity.
In terms of this issue, the post-structuralist analysis would show how the state, which declares itself pro-poor schools, set itself up for failure. In fact, for a long time, NGOs, which work mainly with poor schools and teachers, made similar mistakes in their conceptualisation and implementation of programmes. The post-structuralist position would blame the poor leadership of the state and NGOs, which did not sufficiently take into account the poor conditions and realities of teachers in designing appropriate development programmes for these schools (De Clercq 2013). One could also argue that the government made the situation worse by developing ambitious and inappropriate school-based policies that stretched departments beyond their capacity to support such changes. The leadership of the unions, in contrast, appeared to prefer to blame the department or NGOs rather than actively participate in the reconceptualisation of more appropriate development programmes. This analysis contrasts with the liberal analysis, which would have foregrounded the importance of continuing building the institutional capacity of the departments and schools (Van der Berg et al. 2011a). The Marxist position, on the other hand, would have emphasised the unequal structures and policies that militated against appropriate pro-poor teacher development policies and programmes. The question that the poststructuralist analysis needs to answer is, why did the unions, and SADTU in particular, which started well with their membership mobilisation for the 1998 Development Appraisal System, not continue to fight for appropriate forms of teacher development and accountability? Could it be that the union leadership did not manage to challenge the departmental policies to ensure that teacher development programmes are appropriate and meaningful for the challenges faced by disadvantaged teachers?
Another important and related school-based reform, which was motivated by the pre-1994 struggles against the nature and form of Bantu education, is found in the curriculum arena. The 1998 OBE curriculum, revised in the 2002 Revised National Curriculum Statement, was borrowed mainly from Australia and was not appropriately adapted by the team of researchers in charge who did not comprehend sufficiently the teachers' realities to ensure that such curriculum could be implemented with the conditions on the ground (Spreen 2001). This OBE curriculum was said by curriculum analysts (Jansen and Christie 2000) to be much too ambitious in that it changed radically both the way teachers had to teach content and develop pedagogy. While middle-class schools and their teachers were sufficiently qualified to use and benefit from this new curriculum to improve their teaching and learning, most teachers from disadvantaged schools struggled to comprehend let alone implement this curriculum in their classrooms. In addition, these teachers were not inducted or supported appropriately to understand how to implement this complex curriculum. It is interesting to note here that the teams of curriculum designers allocated for each phase and subject appear not to have taken into account the conditions in the classrooms nor the fact that such contentless curriculum frameworks were not appropriate to assist poorly qualified teachers to work out an appropriate sequence and pacing for their subjects. These designers may have been pushed by senior officials in the Department of Education responsible for curriculum to design specific frameworks and their outcomes to align the South African curriculum with the international trend that favoured outcomes-based curricula to respond to the needs of the 21st-century economy and society. There may have been a few SADTU representatives on these curriculum design teams, but the curriculum experts appointed by the department dominated and would have overruled any SADTU fundamental opposition.
This means either that SADTU could have fought harder to ensure that the new curriculum was developed in such a way that disadvantaged teachers could understand and implement it realistically and appropriately in their classrooms (Rogan and Grayson 2003) or that it was unable to develop campaigns and mobilise its members against this unrealistic curriculum. It never managed to organise against the challenges the new curriculum raised for the majority of its members and instead waited for the implementation stage to ensure that its members boycotted classroom supervision and any form of accountability because of inadequate district support.
By 2010, the then Minister of Education was bold enough to admit that the curriculum policy was too ambitious and problematic and that it needed to change. A Commission of Enquiry was commissioned, which recommended a template for the new curriculum policy. This resulted in the introduction of the Curriculum and Assessment Policy Statements (CAPS) in 2012 with much more explicit content and assessment to be covered each week, abandoning once and for all the idea of a content-less curriculum.
This was a significant advance, even if teachers in disadvantaged schools still found CAPS too ambitious and problematic in its pacing and learner-centred pedagogical approach. Criticisms about the pacing and sequencing (Christie and Monyokolo 2018;Schollar 2018) and the complaints of many disadvantaged teachers led more recently to the introduction of large-scale interventions for poorly performing primary schools to assist with the implementation of CAPS. Indeed, under pressure because of the now better measured poor learners' outcomes (especially in Grade 3 and 6 as revealed in the Annual National Assessment results), a few provincial departments, in partnership with education NGOs, decided to assist with the pedagogical and assessment routines of CAPS. They provided poorly performing primary schools with scripted lesson plans, better readers and teaching resources as well as coaching of teachers (Fleisch 2018). These large-scale pedagogical or instructional interventions, according to recent quantitative evaluations (Fleisch 2018), appear more meaningful and appropriate to what these struggling poor schools and their teachers need to improve their understanding and teaching of CAPS, especially at the Foundation Phase because of the persistent poor reading ability of learners of disadvantaged schools (Pretorius and Spaull 2016).
The government presented the OBE curriculum as assisting with the competitive demands of the global labour market and enabling learners to be at the centre of teaching and learning and learn at their individual pace. This is what the liberal position would agree with, but it would have recommended better and more meaningful support and development to improve future implementation (Pudi 2006). The Marxist position would emphasise that such a globally aligned OBE curriculum was biased against the poor disadvantaged schools and teachers (Spreen 2001). It was designed to impress the global world by showing that South Africa was determined to improve and modernise its education system-at least for the minority of the learner population who would be well-educated to fill the highly skilled positions of the formal global economy. In contrast, the post-structuralist position would foreground the same arguments as the Marxist position with the addition that curriculum was not an arena or site where power play or contestations by the leadership of disadvantaged teachers and their organisations did not occur. It would emphasise the top-down way in which the government planned and introduced these curriculum policies (there was no serious pilot over a few years) and how this excluded the possibility of legitimate contestation by unions and researchers who had tracked and illustrated the difficult and inadequate curriculum implementation in poor schools (Jansen and Christie 2000).
The fact that the Minister of Education acknowledged the disastrous curriculum implementation in the majority of schools shows a certain courage by the leadership (compared with previous ministers) that dared to appear critical of the previous OBE curriculum frameworks. The Department of Education acknowledged in 2010 that the earlier curriculum policies were likely to undermine the teaching force (mainly in disadvantaged schools) and the school graduates who would not be well-prepared for the need to demonstrate independent, critical and problem-solving skills in the labour market.
What were the consequences for educational outcomes of all these problematic and class-based policies and programmes? As mentioned earlier, the matric pass rate improvement has occurred mainly as a result of pressures on Grade 12 teachers (and subsequent support programmes for teachers and learners) given the public visibility of the yearly matric results. And it should not be forgotten that there are many different ways to read and analyse the improved matric pass rate. Some researchers and educationists argue that the improved matric pass rate should be questioned and not trusted at face value as there are many ways in which schools could retain learners in the Further Education and Training (FET) phase so as not to tarnish their Grade 12 results. Other researchers looked at the decreasing number of registered maths and science matric candidates and graduates, especially in poorly performing schools, while others pointed to the rather low level (30%) of pass at matric level.
The liberal position would foreground the improvements in general, praising (as the Department of Basic Education would) the efforts of individual schools, teachers and learners, without delving deep into their bimodal pattern and how educational outcome inequalities have remained strong over the years. The Marxist position would go beyond the national average pass rates and detail the discrepancies over the years between the rich middle-class public and private schools and the public schools of quintiles 1 to 3 (Motala and Carel 2019). Their explanation would emphasise the structural inequalities and the fact that there had not been adequate pro-poor support programmes. The poststructuralist position would also foreground the persistent inequalities between middleclass and poor schools, but they would additionally point to the need for civil society to pressure the Department of Basic Education to develop more pro-poor redistributive policies (Sayed and Motala 2012), and the existence of specific schools that show leadership and agency to improve their matric results despite the odds and offer an analysis of their strategies and processes (Christie 2001). A recommendation could have been for the leadership to develop programmes that could empower poor schools, their teachers and learners to benefit from more conducive policies and programmes targeting specifically the conditions in these schools (such as the recent large-scale instructional improvement programmes in poor primary schools).
Finally, let us mention the poverty gap: learners from poor socio-economic backgrounds are likely to encounter many serious obstacles in attending, following and benefiting from their primary schooling. Drawing from the work of researchers in psychology, sociology, anthropology, linguistics, economics, the health sciences, and mathematics education, Fleisch (2008) documents the depth and scope of the challenges facing primary education, both outside as well as inside schools. He showed how poverty, poor health, poor resources and language of teaching were factors influencing the academic achievement of poor learners in reading, writing and mathematics. In short, the poverty gap is about the vicious cycle of the poor socio-economic status of learners and their parents, which is compounded by the poor quality of their schooling, and explains the poor chances for these learners to benefit from schooling.
The liberal position recognises the poverty trap (Van der Berg et al. 2011a). They understand it to be the result of societal inequalities which, they believe, can be minimised by more appropriate and efficient educational strategies or support programmes that enhance the efficiency of the system and its people. The Marxist position mainly emphasises the inescapable influence of societal inequalities, which are reinforced and compounded by structural educational inequalities. For example, Shalem and Hoadley (2009) show, on the basis of an analysis of data sets on inequalities-at the societal level, at the level of the homes and communities of the children that attend school, at the level of the schooling system and at the level of teachers-that there is an intractable pattern of accumulation of educational disparity among teachers. The poststructuralist position, while accepting the compounding impact of societal and educational inequalities on the poverty trap, also examines how the state is or can be pressurised by organs of civil society into adopting and implementing more pro-poor support programmes to compensate for the situation of disadvantaged learners (Equal Education 2018).

Conclusion
To sum up, this article has shown how the post-structuralist analysis, which takes into account the influence of unequal socio-economic and educational structures and agencies on the nature and evolution of the education system, enables us to understand better why socio-economic and education reforms were not sufficiently directed at challenging seriously these inequalities. What it identifies are the reasons behind the broad maintenance of class-based socio-economic and educational inequalities (with some minor exceptions). It recognises the absence of significant socio-economic restructuring, but also examines the level of sustained organised opposition and leadership in certain contestable arenas of the education system to warn that the state needs to be substantially challenged and pressurised to revert the impact of its classbiased educational reforms and programmes. This is not to say that no positive change has occurred or that the leadership committed to more socially just education did not achieve much. Rather it is recognised that there were pre-1994 anti-apartheid struggles that fought fiercely to ensure that the post-1994 government would improve the education system by effecting some financial redistribution and improving poor schools' physical infrastructure and resources. However, these redress measures were in no way sufficient to improve the quality of the education system for all, and in particular disadvantaged schools. In this sense, the state did not continue to be pressurised by the education sector to ensure that its education reforms and programmes of redistribution and support benefitted the disadvantaged majority.