Benchmarking a pathway out of learning poverty

Image: Julia M Cameron on Pixabay.

“Early reading is the basic foundation that determines a child’s educational progress through school, through higher education and into the workplace. All other interventions – from the work being done to improve the quality of basic education to the provision of free higher education for the poor, from our investment in TVET colleges to the expansion of workplace learning – will not produce the results we need unless we first ensure that children can read” (Ramaphosa, 2019).

Although more children in low- and middle-income countries are in school and staying in school longer than ever before, a large proportion are not acquiring the fundamental skills required to build the necessary human capital to support sustainable growth and poverty reduction. The World Bank and UNESCO Institute for Statistics introduced the concept of Learning Poverty to highlight this learning crisis and provide an early indicator of progress, or lack thereof, towards sustainable development goals. Learning Poverty is defined as the percentage of 10-year-olds in a country who are either out of school (schooling deficit) or are in school but do not meet a minimum level of reading proficiency (learning deficit). Reading proficiency at age 10, as defined by the Global Alliance to Monitor Learning, means “Students independently and fluently read simple, short narrative and expository texts. They locate explicitly stated information. They interpret and give some explanations about the key ideas in these texts. They provide simple, personal opinions or judgements about the information, events and characters in a text[i].” Reading proficiency is not only easily understood but a lack thereof is also “usually a clear indication that school systems are not well organised to help children learn in other areas such as maths, science and the humanities[ii]”

One of the key challenges in addressing the learning crisis is that there are many low- and middle-income countries that do not participate in the international or national assessments that are necessary to measure learning poverty. Nevertheless, of those that do participate, South Africa has one of the highest rates of learning poverty in the world, with 8 out of 10 children in Grade 4 considered learning poor pre-COVID-19. Given the near universal enrolment in primary school, learning poverty in South Africa is almost entirely due to learning deprivation among children who are in school. Localised early grade reading studies and systemic evaluations from the Western Cape indicate that learners lost between one to two years of schooling due to the pandemic-related disruptions[iii]. The learning poverty headcount is almost certain to rise when new assessment data becomes available at the end of this year.

Figure 1: Learning poverty rate across the world

Source: World Bank Education Statistics.Reading is a complex and hierarchical process and at the root of the very poor outcomes for reading comprehension at Grade 4 are gaps in the fundamental skills that are essential for learning to read in the Foundation Phase (Grades R to 3). If we are to map a pathway out of learning poverty, we need a series of grade-specific learning poverty lines that allow us to identify learners who are at risk of learning poverty at age 10 early on in their schooling experience. Such learning poverty lines, or benchmarks, would clearly articulate the expected proficiencies at each grade and map out what a successful trajectory to reading proficiency by age 10 looks like. They could also potentially signal to teachers where to shift their instructional focus to meet learners at their reading level. At a provincial and national level, such benchmarks would facilitate the monitoring of reading development and provide early indications of progression towards education and sustainable development goals.

Grade-specific reading benchmarks for oral reading fluency (i.e., the number of correct words read per minute) exist in English[iv]. However, these benchmarks cannot simply be transferred to all South African languages due to phonological, morphological, and orthographical differences between languages. The example in Table 1 highlights the considerable variation in average word lengths for selected South African languages. Clearly, simple comparisons of fluency across languages are uninformative. Moreover, the trajectory of reading development and accuracy-fluency and fluency-comprehension relationships depend on the linguistic features of each language. Despite increasing acknowledgement of the importance of gaps in foundational skills, there has been very limited research into these relationships in African languages.

Table 1: Average word length for example early grade text by language[v]

Language Text Total words Average word length Total single syllable words
Sepedi Go be go na le mosepedi yo a bego a na le tlala. O fihlile motseng wo mongwe a kgopela dijo. Go be go se na yo a bego a na le dijo 33 2.9 23
Xitsonga A ku ri ni mufambi loyi a ri na ndlala. U fikile emugangeni a kombela swakudya. A ku nga ri na loyi a ri na swakudya. 26 3.4 17
isiZulu Kunesihambi esasilambile kakhulu. Safika emizini omunye sacela ukudla. Abantu babengenakho ukudla. 10 8.5 0

Source: Spaull, Pretorius, and Mohohlwane (2020) based on the Molteno Institute for Languages and Literacies Vula Bula Stone Soup story.

Recognizing the need for language-specific benchmarks, the Department of Basic Education (DBE) is leading efforts to establish language specific benchmarks for foundational reading skills in all of South Africa’s official languages. This endeavour began in 2019 with a consultative design process that brought together academics, reading practitioners, funders and international benchmarking experts. The outcome of this process was a road map for establishing benchmarks that identified three approaches – re-analysis of existing data, extending upcoming data collection for ongoing evaluations to support benchmarking and, where necessary, collecting data specifically for benchmarking.

Alongside this scoping process, the DBE was working in close collaboration with a team of researchers from SALDRU (University of Cape Town), RESEP (University of Stellenbosch), NORC (University of Chicago) and UNISA to establish early grade reading benchmarks for the Nguni languages[vi]. This work pulled together existing large scale reading assessment data from four different reading and literacy studies[vii] conducted between 2016 and 2019 to produce a dataset with multiple assessment points for nearly 16,400 unique learners across 660 no-fee schools. The approach to establishing benchmarks was guided by a combination of insights from the data, reading development theory, expert linguistic knowledge of the Nguni languages, and an understanding of curriculum demands and system realities. An outcome of the empirical work was methodological innovations that had several technical and contextual advantages over traditional benchmarking methods[viii].

In 2021, the DBE expanded the planned data collection for the ongoing Early Grade Reading Study in the North West province to provide data for the establishment of Setswana benchmarks. New data collection allowed for a rigorous design process for the assessment tools with multiple rounds of piloting. Researchers from SALDRU and RESEP worked closely with Setswana language specialists and the DBE to refine the tools and finalise the benchmarks[ix] At the same time, SALDRU and RESEP researchers compiled a dataset on English First Additional Language (EFAL) combining reading assessments for around 23,000 unique learners from over 900 schools across six studies[x]. These data were used to establish grade-specific EFAL reading benchmarks from Grade 2 through to the end of the Intermediate Phase (Grade 6)[xi]. The DBE then approached SALDRU researcher, Cally Ardington, in her role as principal investigator of evaluations of the Funda Wande literacy and numeracy interventions in Limpopo and the Western Cape. With funding from Funda Wande and the Zenex Foundation, planned data collection for the evaluations was broadened in scope to include Sepedi and Afrikaans benchmarking. Once again, new data collection allowed for a broad collaborative process involving Sepedi and Afrikaans language specialists, the DBE and SALDRU.

By the end of 2023, benchmarks for all South African languages will have been established and released by the DBE. Analysis for English Home Language has started, piloting in preparation for Xitsonga data collection is under way and funds are being raised for Tshivenda. Effective collaboration between government, funders, research organisations and African language specialists has not only accelerated the benchmarking agenda, but also resulted in methodological innovations, established best practices and supported capacity building. That said, the value of benchmarks lies in their use. With support from the Zenex Foundation and J-PAL Africa, SALDRU and the DBE are currently conducting a pilot study with teachers from 40 schools across four provinces to explore how best to support teachers in effectively using the newly established benchmarks in their classrooms.

Beyond benchmarking, a significant contribution of this work has been the collation and harmonization of data across all available early grade reading studies. These data have been used to estimate COVID-19-related learning losses[xii], examine deficits in foundational reading skills and map out reading development trajectories. Together with the DBE and other researchers, SALDRU is in the process of putting together a grant application to i) create a living database to serve as a repository for existing and future studies; ii) develop and support a research agenda around these data; and iii) expand the group of researchers working on learning deficits through a range of capacity building activities. We believe that these data are an essential resource if we are to effectively address Learning Poverty and make progress towards sustainable development goals.


[i] World Bank 2020: 11

[ii] World Bank 2020: 5

[iii] Ardington et al. 2021b; Kotze et al. 2022; van der Berg et al. 2022

[iv] Hasbrouck and Tindal 2006

[v] The English translation of the text is as follows: “There was a stranger who was very hungry. He came to a village and asked for food. Nobody had any food.”

[vi] Ardington et al. 2020

[vii] The four studies were the Story Powered Schools (SPS) evaluation; the Early Grade Reading Study II (EGRS II); the Funda Wande (FW) evaluation; and the Leadership for Literacy (LFL) project (Ardington et al., 2019; Ardington and Meiring, 2020; Department of Basic Education, 2019; Taylor et al., 2019).

[viii] Ardington et al. 2021b; Kotze et al. 2022

[ix] Wills et al. 2022

[x] The studies were the Early Grade Reading Study (EGRS I) in North West province and the related Reading Support Project (RSP); the second Early Grade Reading Study (EGRS II) in Mpumalanga; the Leadership for Literacy (LFL) project in higher and lower performing no-fee schools across KwaZulu-Natal, Gauteng and Limpopo; the Story Powered Schools (SPS) impact evaluation in KwaZulu-Natal and the Eastern Cape; and the Funda Wande (FW) evaluation in Limpopo province (Menendez & Ardington, 2018; Wills & van der Berg, 2020; Ardington & Henry, 2021; Department of Basic Education & University of Witwaterstrand, 2020).

[xi] Wills et al. 2022

[xii] Ardington et al. 2021b



Ardington, C., Hoadley, U., & Menendez, A. (2019). Impact Evaluation of USAID/South Africa Story Powered School Program Endline Report. Chicago: NORC at the University of Chicago, USAID.

Ardington, C. & Meiring, T. (2020). Impact Evaluation of Funda Wande Coaching Intervention Midline Findings. Cape Town: SALDRU, University of Cape Town.

Ardington, C., Wills, G., Pretorius, E., Deghaye, N., Menendez, A., Mohohlwane, N., et al. (2020). Technical Report: Benchmarking early grade reading skills in Nguni languages. Stellenbosch: ReSEP, Stellenbosch University. Cape Town: SALDRU, University of Cape Town. Chicago: NORC at the University of Chicago. Languages Technical Report V06.pdf?ver=2020-10-28-141752-997

Ardington, C., Wills, G., & Kotze, J. (2021a). COVID-19 learning losses: Early grade reading in South Africa. International Journal of Educational Development, 86 (October 2021), 102480.

Ardington, C., Wills, G., Pretorius, E., Mohohlwane, N., & Menendez, A. (2021b). Benchmarking oral reading fluency in the early grades in Nguni languages. International Journal of Educational Development, 84, 102433.

Ardington, C.  & Henry, J. (2022). Funda Wande Eastern Cape Evaluation. Baseline Report. Cape Town: SALDRU, University of Cape Town

Department of Basic Education. (2019). The Secondary Early Grade Reading Study. Year 1 Report. Learner performance after the first year of implementation. Pretoria: Department of Basic Education.

Department of Basic Education and the University of the Witwatersrand. (2020). Early Grade Reading Study 2017-2019, Waves 1-4 Merged [dataset]. Version 1. Pretoria: DBE and Wits [producers], 2020. Cape Town: DataFirst [distributor], 2020. DOI:

Hasbrouck, J., & Tindal, G. A. (2006). Oral reading fluency norms: A valuable assessment tool for reading teachers. The Reading Teacher, 59(7), 636–644.

Kotze, J., Wills, G., Ardington, C., Taylor, S., Mohohlwane, M., Nuga-Deliwe, C. (2022). Learning losses due to the COVID-pandemic. Advisory Note for the Reading Panel.

Menendez, A & Ardington, C. (2018). Impact Evaluation of USAID/South Africa Story Powered School Programme – Baseline.

Ramaphosa, C. 2019. President Cyril Ramaphosa: State of the Nation Address 2019. 20 June 2019. South African Government. Available at:

Spaull, N., Pretorius, E., & Mohohlwane, N. (2020). Investigating the comprehension iceberg: Developing empirical benchmarks for early-grade reading in agglutinating African languages. South African Journal of Childhood Education, 10(1), a773.

Taylor, N., Wills, G., Hoadley, U., 2019. Addressing the ‘leadership conundrum’ through a mixed methods study of school leadership for literacy. Research in Comparative and International Education 14, 30–53.

Van der Berg, S. and Hoadley, U. and Galant, J. and van Wyk, C. and Böhmer, B., 2022. Learning Losses from Covid-19 in the Western Cape: Evidence from systemic tests. RESEP Working Paper. University of Stellenbosch

World Bank. (2020). Ending Learning Poverty. What will it take? World Bank. Washington.

Wills, G. & van der Berg, S. (2020). Measuring School Leadership and Management and Linkages with Literacy: Evidence from rural and township primary schools in South Africa. Educational Management Administration and Leadership. 45 (5) 708-731. DOI: 10.1177/1741143220915923

Wills, G., Ardington, C., Pretorius, E., Pooe, E., & Ramagoshi, R. (2022). Benchmarking Early Grade Reading Skills: Setswana and English First Additional Language. Technical Report. Johannesburg: Khulisa Management Services.