Monthly Archives: March 2018

The Comprehension Iceberg: Developing reading benchmarks in African languages


For the last few years we have been working on a project exploring reading in three African languages; isiZulu, Xitsonga and Northern Sotho (Sepedi). This was part of an ESRC-funded study looking at “Leadership for Literacy” and schools in challenging contexts. This week we published the first of a series of working papers on some of our results. The paper is titled “Investigating the comprehension iceberg: Developing empirical benchmarks for early grade reading in agglutinating African languages  (2018) and is co-authored with colleagues and friends Prof Elizabeth Pretorius (UNISA) and Mpumi Mohohlwane (DBE), who is now also doing her PhD at RESEP.

In the paper we argue here that we need to move beyond a repetitive focus on low comprehension outcomes; this is simply the tip of the iceberg. Below the surface there is widespread evidence that most children have not acquired the basic ‘tools’ for reading success – the ability to accurately and fluently decode letters and words and move from an effortful activity to an automated skill. Knowing that 78% of Grade 4’s cannot read for meaning in any language is important and sobering information, but probably not as helpful as knowing which components of reading children are struggling with and why teachers are struggling to teach them.  If children and teachers are ‘falling at the first hurdle’ (Taylor, 1989) does it make sense to focus on the seventh or eighth hurdle and ask why learners and teachers are not making it over?

One of the important contributions of the paper is that we try to estimate benchmarks for these three African languages. While many benchmarks exist in English there are currently no benchmarks in African languages and unfortunately you cannot just version the English benchmarks for the African languages because the language structures (orthographies) are so different. This is easiest to explain using an example:

Screen Shot 2018-03-13 at 11.08.10 AMSo in isiZulu the first sentence is only 3 words while in Northern Sotho (Sepedi) it’s 13 words. We go on to show that there are also large differences in ‘required fluency’ across these languages and that accuracy rates and speed differ between conjunctive and disjunctive languages – see below.

Screen Shot 2018-03-13 at 10.54.39 AM

“Comparison across the languages shows that accuracy seems to be more important for fluent reading in isiZulu than in Northern Sotho or Xitsonga. The isiZulu learners reading at 21 wcpm or faster are reading with 95% accuracy or higher. In contrast, 95% accuracy is only associated with reading at 51 wcpm or faster in Northern Sotho and 31 wcpm or faster in Xitsonga. One of the reasons why Decile-1 learners are reading so slowly is that they are making mistakes on every second or third word. The fastest Northern Sotho readers (wcpm=107) and Xitsonga readers (wcpm=91) in the sample made no mistakes whatsoever.” (p12)

This can also be seen visually:

Screen Shot 2018-03-13 at 10.55.10 AM

We then use this information and some rudimentary comprehension outcomes to benchmark the different languages:

Screen Shot 2018-03-13 at 10.54.55 AM

We’d encourage those interested in this type of work to read the full paper – if you have any comments or suggestions we’d love to hear from you! Please include them in the comments below.


The importance of learning to read in mother-tongue is widely acknowledged in the linguistics literature yet reading acquisition in African languages remains under-researched and under-theorized. While numerous studies have highlighted the low levels of comprehension among learners reading in African languages in South Africa, little has been done to understand what lies beneath this ‘comprehension iceberg.’ In this paper we present new empirical evidence on reading outcomes and the subcomponents of reading for 785 Grade 3 learners across three languages (Northern Sotho, Xitsonga and isiZulu), drawn from 61 primary schools in South Africa. This is the largest sample of such learners to date. Using an adapted EGRA-type assessment we assessed letter-sounds, single-word reading, non-word reading, oral reading fluency and oral comprehension. From this data we present results on fluency, accuracy and comprehension and how these might relate to each other in these morphologically rich agglutinating languages. We also show that there are large differences in reading subcomponents between languages with conjunctive and disjunctive orthographies. Our results suggest that there are minimum thresholds of accuracy and oral reading fluency in each language, below which it is virtually impossible to read for meaning. These are 52-66 wcpm in Northern Sotho, 39-48 wcpm in Xitsonga and 20-32 wcpm in isiZulu. We argue that there is a strong need for empirical language-specific norms and benchmarks for indigenous African languages and present our benchmarks for these three languages as a move in that direction.