Abstract
Reading with comprehension is an essential lifelong skill which starts developing long before children start schooling. It is an essential skill used throughout an individual’s life and is a complex process facilitated by a variety of knowledge and skill sets. According to the Simple View of Reading (SVR), the two main skills that enable reading with comprehension are decoding and linguistic comprehension. While the PIRLS results indicate that South African learners perform very poorly in reading comprehension, the causes of their poor performance are yet to be closely examined in many African languages including Sesotho. This study adopted a two-pronged approach to probe possible contributors to poor reading comprehension among Grade 4 Sesotho learners. Firstly, it assessed the foundational reading skills of decoding (letter-sound knowledge), reading fluency, oral reading and written comprehension to examine the relationship between these variables. Secondly, the study analysed the fidelity of translation of the two PIRLS texts used to assess reading comprehension, a narrative and information text and their accompanying questions from English to Sesotho. This analysis served to identify possible translation issues which may have affected text comprehension.
The study is a correlational mixed method that quantitatively assessed a purposive sample of 241 Grade 4 learners from 5 quintiles 3 and 4 Soweto schools and qualitatively examined fidelity of translation of the PIRLS passages used for reading comprehension assessment. Tests were used to assess letter-sound knowledge (LSK), oral reading fluency (ORF) and oral reading and written comprehension.
The LSK test comprised two subtasks which assessed Sesotho simple- and complex letter-sounds respectively. Each learner had to sound out the simple and complex letters from a chart for one minute and errors were subtracted from the total number of letters sounded, to establish how many letters correct per minute learners could identify.
The learners’ ORF was determined by letting them read from a 132-word text titled Tshoswana le leeba (Ant and dove). For this test, learners read for one minute and errors were recorded. Thereafter, learners read for an extra two minutes. Learners were then orally asked questions and were made aware that they should seek answers from the text. To score the test, the number of words incorrectly read was subtracted from
v
the total number of words read in a minute. The learners’ reading comprehension was furthermore tested using six oral questions.
For the written reading comprehension, the PIRLS narrative and information texts were accompanied by 15 and 17 questions respectively. The questions included constructed response and multiple-choice formats and tapped into four comprehension processes or levels of comprehension as specified by PIRLS.
The results showed that Sesotho Grade 4 learners could only correctly identify on average 26 correct letters per minute (clspm). They struggled to identify both simple and complex sounds, although performance on complex letter sounds was much lower. Letters were either skipped, misidentified by confusing them with similarly contoured letters or letter patterns, or sounded out in English in some cases. The mean for ORF was 30.5 words correct per minute (wcpm) indicating a slow and inaccurate reading pace for the grade. Performance on both oral and written comprehension was low, at 38% for the oral reading comprehension, 34% for the narrative and 26.6% for the information text. The results of the written comprehension revealed that only a small percentage of learners was able to answer literal questions and the numbers plummeted in the next three levels of comprehension as learners struggled to make straightforward inferences and answer integrative and evaluative questions. In sum, the learners did not have adequate foundational decoding and fluency skills to support reading comprehension. All assessed variables showed significantly strong to very strong relationships, oral reading fluency was impacted by poor knowledge of letter-sounds where learners inaccurately read words and struggled to identify various letter-sounds contained in them. Inaccurate decoding affected reading accuracy and fluency, which in turn affected reading comprehension.
Pertaining to the qualitative aspect of the study, the findings were that at both macro and meso levels the two text genres were equivalently translated. However, there were some inequivalences in the micro features in both texts which include inaccurately translated words and phrases. For example, in the narrative text, the noun ‘flood’ was translated by using a metaphor noka e ja ditlhokwa instead of the equivalent noun morwallo and the word ‘believe’ was translated as dumela instead of kgolwa in the information text. There were also minor errors in the translation of questions or parts of texts related to some questions. In these cases, learners’ low performance could be
vi
attributed to translation errors. However, generally learners performed poorly on most questions not affected by translation inequivalences, which suggests that poor reading comprehension stemmed mainly from poorly developed foundational decoding and fluency skills.