📝 By Laura Grant and Ro Manoim
South Africa got a sharp wake-up call in 2023 when the results of an international reading literacy survey found that 80% of our Grade 4s couldn’t read for meaning. We came stone last in a group of 57 countries that did the test.
Ten years earlier in 2013, Peru got a similar shock when the poor performance of its public schools in an international reading and mathematics assessment made national media headlines.
Peru acted quickly to start to turn things around.
Jaime Saavedra, Peru’s former minister of education, now at the World Bank as Director of Human Development for Latin America and the Caribbean, described how his country made ‘tangible short-term gains’ at a meeting of the 2030 Reading Panel in February.
It focused on four lines of action.
- Teachers: Improve the social standing of teachers because “teachers are the central partners in any education reform process”.
- Learning: Make quality interventions to improve teaching methods, strategies and resources. For Peru, this included providing learning material in “19 native languages”.
- Effective management: “It needs a high-quality bureaucracy to drive change at scale”, from school management upwards.
- Improve school infrastructure.
What can we learn from PIRLS?
The most well-known and widely reported evidence that South Africa’s children are not learning to read at school is the International Association for the Evaluation of Educational Achievement’s (IEA’s) 2021 Progress in International Reading Literacy Study (PIRLS).
South Africa is one of only three African countries that participates in PIRLS, the other two are Morocco and Egypt. Both of them did slightly better than South Africa, which is not great because they are lower-middle income countries whereas South Africa is classified as an upper-middle income country and should, in theory, have the resources to do better.
Brazil and Türkiye, which are upper middle-income countries, both scored better than South Africa, although they were also in PIRLS’ bottom 10.
Most of the participating countries are in Europe, with a few in Asia, such as Hong Kong, Taiwan, Macao, Singapore, Kazakhstan, Uzbekistan, Azerbaijan, and the Middle East, such as Israel, Saudi Arabia, Dubai, Bahrain, Iran, Jordan, Oman. Only one country in South America participated, Brazil. Peru’s shock reading results were from a different survey, the Programme for International Student Assessment (Pisa), it does not take part in PIRLS.
For the PIRLS study, the minimum result indicating that learners can “read for meaning” in easy texts is the “low” benchmark of 400 points. This entails retrieving basic information sufficient to answer straightforward questions about the text.
The median percentage of Grade 4 learners around the world reaching this minimum benchmark was 94%, but only 19% of South African learners managed it. Only 6% of South African learners exceeded this skill level and reached the intermediate benchmark (+475 points), meaning they can begin to interpret and integrate information about the text. Just 3% of our learners reached the “high benchmark” (+550 points), which means they proved their ability to interpret and evaluate complex texts.
This is not the first time South Africa’s learners have done badly in international reading surveys. In 2019, President Cyril Ramaphosa set a target that in 10 years “every 10-year-old will be able to read for meaning”.
To be fair, the 2021 PIRLS took place immediately after the Covid-19 pandemic, which disrupted schooling, but it’s still doubtful that the target will be met.
You might even ask why South Africa participates in PIRLS at all if it’s just going to make us look bad?
At a seminar on reading literacy in 2023, former education minister Angie Motshekga answered that question when she said PIRLS establishes a global standard for reading comprehension and “as a developing country, we are still on a journey to reach these international benchmarks”.
South Africa stopped doing local standardised assessments that measure how well learners read in primary schools a decade ago when the Annual National Assessments (ANAs) were halted. One of the reasons was that the teacher unions had raised objections.
In any case, the optimistic view here is that even if our learners don’t do that well in international tests, at least there is a will in government to do something about it.
It could also be seen as a positive sign that a quote by management consultant Peter Drucker, “What gets measured gets improved”, has been popping up in Department of Basic Education slide decks recently.
Local benchmarks
In 2022, a year after the famous PIRLS assessment was done, another survey was carried out. This time, a local one, which is being touted as a replacement for the Annual National Assessments. Called the South African Systemic Evaluation (SASE), it involved 56,650 learners from 1,688 schools. It looked at the reading and mathematics abilities of Grade 3, 6 and 9 learners across the country.
The Department of Basic Education released the results of the SASE only in December 2024.
The way the SASE measures reading proficiency is not exactly the same as the PIRLS assessment so they’re not directly comparable; however, the results show a very similar and worrying pattern in that only 20% of learners in Grade 3 were able to achieve the level of reading skills they are expected to reach in that grade.
In the SASE, the reading skills and knowledge learners are expected to be proficient at are divided into four performance levels.
The first level, named “emerging”, is where learners are just beginning to develop the skills required for grade 3-level reading. The next level up, known as “evolving”, is where learners are beginning to construct and adapt what they have learned. The third level, called “enhancing”, is where learners demonstrate that they actually have the required skills, are able to apply those skills and show they are moving towards independent learning. At the highest, “extending” level, learners show an advanced understanding of the knowledge and skills required, apply their knowledge in creative ways and can learn independently.
Learners need to have “enhancing”-level skills to meet the requirements of Grade 3.
Only one in five of the Grade 3s who took part achieved that level.
It’s important to recognise that it’s not all learners at all schools who can’t read. In the PIRLS tests some learners did well, while others did very badly.
Language, geography and socio-economic circumstances have a role to play in shaping children’s reading skills.
What does language have to do with it?
South Africa is one of the few countries that tested in multiple languages for PIRLS 2021. The assessment material was translated into all 11 official languages.
Afrikaans and English test scores were much higher than the other nine languages. They were the only two languages where the average scores were relatively close to 400, the minimum required to show an ability to read for meaning in easy texts.
Learners who completed the assessment in African languages scored below the national average of 288 points. The ones who scored the lowest were assessed in Xitsonga (223), Sepedi (216) and Setswana (211).
The results of the local SASE survey showed a similar pattern in the Grade 3s tested. Where the language of learning and teaching at school is English or Afrikaans, learners were better at reading.
Just over a third of the Grade 3s who were taught in English (36%) and Afrikaans (38%) showed the reading proficiency level expected for Grade 3.
Those taught in isiZulu (13%), isiXhosa (15%) and Siswati (12%) did the next best. Of the children taught in the other six languages less than 10% showed the required reading skills.
About half of the children taught in isiNdebele, Sepedi, Setswana and Xitsonga achieved only the first level, or “emerging” reading skills. Those are the same three languages that performed poorly in the PIRLS study, with the addition of isiNdebele.
Mother tongue as language of instruction
Grade 3 is an interesting time to test children for reading ability in South Africa. Children are taught in one of the 11 official languages (ostensibly their home language) in their first years of school, from Grade R to Grade 3, which is known as the foundation phase.
From Grade 4 the “language of learning and teaching”, or language of instruction, becomes predominantly English or Afrikaans, although there are moves to change this and extend home-language instruction.
Seventy-five percent of the Grade 3s in South Africa’s public schools are taught in their home language, according to the Department of Basic Education.
Research shows that there are benefits in teaching young children foundational reading and mathematics skills in their home language, even if the results of the SASE survey don’t appear to hold that up.
If learning to read in a home language does not mean a child will score higher in reading assessments then there is clearly a problem in the education system, the teaching, and/or the testing methodology, and language is central to all of these areas.
Professor Abdeljalil Akkari, an expert in education at the University of Geneva, argues that “pre-primary is the educational sector which has the greatest need to be based on local pedagogy, traditions and cultures”.
Some notable testing issues pointed out by people in the sector is how transposing and translating a European test into African languages may create more issues than it solves.
An example given by researchers at the University of Pretoria is how the isiZulu version of the PIRLS test needed to use the foreign words in translations such as “i-Hammerhead shark”. They show that due to translations, the isiZulu and English texts used in PIRLS aren’t totally equivalent, resulting in a harder test for the isiZulu schools compared to the English schools.
If you look in more detail at the home language of Grade 3 learners versus the language of instruction at schools a more nuanced picture emerges.
About a third of South Africa’s Grade 3s are taught in English, even though English is the home language of less than 10% of them. Forty-four percent speak isiZulu or isiXhosa.
Not surprisingly, 98% of the Grade 3s whose home language is English are taught in English at school; 92% of Afrikaans-speaking children are taught in Afrikaans. The picture is different for African language speakers.
Children whose home language is isiNdebele are the least likely to be taught in their home language at 50%, according to DBE data. Sesotho-speakers fare marginally better at 52%. Two-thirds of children who speak Xitsonga (66%) and isiZulu (66%) and more than 70% of the children who speak the remaining five languages were taught in their home language: isiXhosa 75%, Siswati 73%, Setswana 72%, Sepedi 76%, and Tshivenda 73%.
However, isiZulu-speaking children were better at reading than the Xitsonga-speaking children, according to the SASE results. And even though children who speak Sepedi at home are highly likely to be taught in Sepedi at school, they were the second-least proficient at reading.
Simply being taught in one’s home language is clearly not the only factor at play.
Provincial differences
There are six languages that are of particular concern because more than 40% of Grade 3 learners managed to achieve only the most basic performance level in their reading skills in the SASE tests. They are Sepedi, Setswana, Sesotho, isiNdebele, Tshivenda and Xitsonga.
Those languages are predominantly spoken in the four lowest-performing provinces: Northern Cape, Mpumalanga, North West, and Limpopo, says Nwabisa Makaluza, a researcher at Stellenbosch University, who contributed an advisory note for the Reading Panel 2025 Background Report.
In the PIRLS test scores, the provinces that performed poorly were basically the same, although the Eastern Cape replaced the Northern Cape in the bottom four.
In SASE’s four lowest-performing provinces a high percentage of Grade 3 learners at public schools are taught in their home language, for example, 87% in the Northern Cape, 72% in Mpumalanga, 79% in North West and 92% in Limpopo.
Whereas in Gauteng, only two in every five learners (43%) are taught in their home language. No home language is truly dominant in Gauteng. The most commonly spoken language is isiZulu, but only one in four Grade 3s speak isiZulu at home.
Gauteng is the most linguistically diverse province. More than 20,000 Grade 3 learners speak Sesotho, Setswana, Sepedi and English at home, more than 10,000 speak Xitsonga, Afrikaans and isiXhosa.
This diversity makes teaching in all 11 official languages a complicated affair, requiring multilingual teachers trained to teach foundation phase learners in multiple languages.
Despite its linguistic diversity, Gauteng’s Grade 3 learners did better in SASE reading tests than all but those in the Western Cape.
On the other hand, one language dominates in KwaZulu-Natal, where close to 90% of the Grade 3s speak isiZulu, and in the Eastern Cape, where 86% speak isiXhosa.
In the poorly performing provinces such as North West, close to 80% of learners speak Setswana. In Limpopo, well over half (58%) of the learners speak Sepedi at home. In Mpumalanga, SeSwati and isiZulu are the two main languages spoken by 60% of learners.
Looking at this from the perspective of the poorly performing languages, close to 70% of the Sepedi-speaking Grade 3s, 80% of the Xitsonga speakers and more than half of the Tshivenda-speaking children are in Limpopo.
Most of the isiNdebele speakers are in Mpumalanga and Limpopo. Most of the Setswana speakers go to schools in North West.
Not enough African language teachers
South Africa’s universities are not producing enough teachers to meet the demand for foundation phase teachers who can teach in African languages, according to a Department of Basic Education report by education economist Martin Gustafsson.
The most recently available data, which was for 2018, shows the languages with the biggest undersupply of teachers are Sepedi, isiXhosa and Setswana. But only three languages are producing enough teachers for the foundation phase: Tshivenda, siSwati and isiNdebele. These are the three languages that are spoken by the smallest portions of the population. And, interestingly, isiNdebele-speaking children are the least likely to be taught in their home language, according to DBE data.
“Some African languages are producing as little as 20% of the required number of language of learning and teaching-specific teachers,” states the report.
The language in which children are taught to read is just one factor. There are historical factors, such as the channelling of resources during apartheid to white schools where English and Afrikaans were the languages of instruction. Thirty years later many of those schools remain better resourced. The socioeconomic environment of learners also plays as part, poorer children may have limited resources both at school and at home and their guardians may have less time to read books.
Socioeconomic factors
Why are South African children who were tested in English or Afrikaans doing so much better than those who tested in other African languages?
This data can be misleading and give rise to the misconception that English-speaking children are better at reading. But, in fact, most of the children who completed the PIRLS test in English have a different home language. Which points to the likelihood that parents may choose to send their children to English schools because they are better resourced.
As part of the PIRLS, learners were asked some context questions, one of which showed that only about a third of the Grade 4s who participated in the English version of the test always speak English at home and nearly one in 10 children never speak English at home.
Of 9,842 South African schools surveyed in the PIRLS study, 73% are categorised as socioeconomically disadvantaged (based on school principals’ reports). African language schools are far more likely to be categorised as disadvantaged, whereas about a third of English schools were categorised as advantaged, much more than schools of any other language.
Looking at the reading achievement results in relation to the socioeconomic condition of the school points to the impact that an under-resourced school can have on a child’s reading ability.
While this is a reality in most countries according to the PIRLS Report, the problem is particularly significant in South Africa, where socioeconomic strata very often affect the language of schooling and testing.
Poor schools
Public schools in South Africa are divided into five quintiles which reflect the socioeconomic circumstances of the communities the schools service. Schools in the poorest communities are classified quintile 1 and those in the more affluent communities are classified quintile 5. Schools in quintiles 1 to 3 are the so-called “no-fee schools”, where learners do not have to pay school fees, and schools in quintiles 4 and 5 are fee-paying schools.
The results of the SASE show that there is a stark difference between the reading skills of Grade 3 learners in the lower quintile schools and the quintile 5 schools.
Children who go to quintile 5 schools are far more likely to be be able to read at the level required for Grade 3 and the reading skills of most of the children in the lower quintile schools were found to be below the required level.
Learners from the poorest backgrounds, who attend the no-fee schools (quintiles 1, 2 and 3) have the least access to resources. This highlights the inequality in South Africa’s education and the need for targeted interventions and resource allocations in no-fee schools, researchers say.
Something else to bear in mind is that girls do better at reading than boys across all quintiles, but particularly in poorer schools. Interventions should be made to support boys’ reading development.
Access to learning material
“Children learn better and are more likely to pursue their subsequent studies when they have begun their schooling in a language that they use and understand,” explained Professor Abdeljalil Akkari.
South Africa’s education policy states that the language of learning and teaching must be the learner’s ‘home language’, but as explained by Sinethemba Mthimkhulu, and other Pretoria University researchers, it is the school that chooses which language is to be regarded as the home language for their learners so in many cases the official home-language of the learner is not actually their mother tongue.
In addition, educational resources are primarily designed for English-speaking learners. The actual language profile of the country is not at all reflected in textbook publications.
Since the Covid-19 pandemic, many countries have incorporated digital learning into their schooling. The 2024 SA Book Publishing Survey shows that 1,130 new digital textbooks were published in English, over 600 in Afrikaans, and less than 300 were published in all the other South African languages combined.
Even more worrying is the lack of new print textbooks being published in Sepedi, Setswana, SiSwati, Sesotho, isiNdebele, Xitsonga and Tshivenda.
It’s not only textbooks, other reading materials also show an English and Afrikaans dominance in a country where two in five people speak isiZulu and isiXhosa.
The National Reading Barometer, through the National Reading Survey, found that access to books in home languages is still a huge problem in South Africa.
The survey found that 72% of parents who read with their young children would prefer to read in an African language.
It also found that schools are the most important source of reading materials in South African households. In many cases (40%), the books that adults read with their children at home are school textbooks, 33% are fiction books.
Looking at all books in general, less than 10% of book sales are for African language books, according to data from the latest South African Book Publishing Industry Survey.
In the period from 2021-2024 less than 1% of books sales in South Africa were isiNdebele or siSwati books, and Sepedi and Sesotho publications each accounted for only 1%.
isiZulu publications account for just 3% of these book sales and, although English is the home language of less than 10% of the population, English books made up 80% of the total book revenue, the book publishing industry survey shows.
Two out of three households (63%) do not have any fiction or nonfiction books at all (this excludes bibles, magazines, textbooks etc). Most speakers of Xitsonga, isiNdebele, and Tshivenda don’t have a single book in their language at home, and over 40% of Setswana and Sesotho speaks don’t have any books in theirs, according to the 2023 National Reading Survey findings.
Let the children read
Despite the immense problems with reading, inequality and lack of resources these reading surveys also reveal a shining light of hope, which is that South Africa’s children actually like reading.
Along with the PIRLS reading test were various surveys, for the parents, school teachers and principals, as well as the children themselves. In the children’s questionnaire one of the questions asked is whether they enjoyed reading. Over 70% of South Africa’s children enthusiastically said they enjoyed reading, the 11th highest percentage of the 57 countries participating in the survey.
And in an “enjoyment of reading” index, PIRLS found that 90% of the South African children like reading to some extent, and 50% of those like reading “very much”.
Not an unsolvable problem
Jaime Saavedra, who was Peru’s education minister when the country took steps to change its poor reading and maths results, had a simple yet profound closing message for his presentation at the 2030 Reading Panel workshop.
Public officials need to take their jobs seriously. They have an immense responsibility because they are “making decisions that can define the careers of thousands of people and the experiences of millions of children”, he said.
“If you are not afraid of that responsibility, you might not know what you are getting into.”
Progress is possible, even in the short term. But you need a sense of urgency, he said, because “as we speak, there is a child in a classroom who is not learning. For many, we are already late.”
This project was completed with the support of the Henry Nxumalo Foundation for Investigative Journalism