Improving early grade reading outcomes in multilingual, English-medium classrooms is an urgent challenge facing South African education. While the national curriculum mandates that learners read with fluency and comprehension by the end of Grade 3, data from the Progress in International Reading Literacy Study (PIRLS) and local benchmarking studies show that the majority, particularly those in multilingual, under-resourced schools, do not meet this standard. The challenges are especially acute for learners whose home language differs from the language of instruction, yet enter ‘straight-for-English’ classrooms with minimal linguistic support. This literature review explores the nature of this challenge, the instructional components most strongly linked to literacy development, and the potential of classroom-based interventions such as QondaRead to offer scalable, cost-effective solutions. Despite wide consensus that early reading ability is pivotal to future educational success, South Africa’s national data reveal a bleak reality. According to PIRLS 2021, only 19% of Grade 4 learners can read for meaning, the lowest performance among participating countries (Roux, Van Staden & Tshele 2023). More recent national data reinforce this picture: the Funda Uphumele Consortium (2025) found that only 18% of Grade 1 learners met the benchmark for letter–sound decoding, and just 19% of Grade 3 learners reached the ORF threshold required for basic comprehension. The consequences are far-reaching: learners who do not master reading in the early grades remain perpetually behind, with academic disadvantage compounding each year of schooling (Khosa 2022). Fluency, the ability to read accurately, smoothly, and with appropriate expression at a pace that enables understanding, is a critical bridge between decoding and comprehension. Once decoding is automated, working memory is freed for making sense of the text. As Pretorius and Spaull (2016) argue, fluency and comprehension must be regarded not merely as pedagogical goals, but as social imperatives in a deeply unequal society. Evidence suggests that learners who have not reached expected levels of fluency in the early years are unlikely to improve without intervention or remediation (Bigozzi et al. 2017; Friedman & Kern 2009; Lonigan & Anthony 2000; Pretorius et al. 2016). In the South African context, the national CAPS makes clear that ‘by the end of Grade 3, readers are expected to read accurately on their own, at a steady rate, with comprehension and with enjoyment’ (Pretorius, Mohohlwane & Spaull 2020:2). Yet, despite empirical evidence that more than 80% of learners entering Grade 4 have not reached this curriculum goal, from Grade 4 onwards, the curriculum provides no explicit teaching of how to read. These challenges are particularly acute in under-resourced schools. In South Africa, 72% of learners attend no-fee Quintile 1–3 public schools (White & Van Dyk 2019). In such contexts, the national curriculum’s assumption that all learners will be reading fluently on entering Grade 4 remains aspirational at best (Long & Bowles 2024; Pretorius & Spaull 2016).

Language adds a layer of complexity. The South African Language in Education Policy affirms learners’ right to receive instruction in the language most commonly spoken in their homes, during the Foundation Phase (Grade R – Grade 3) of schooling (RSA 1997) although policy mandates a transition to English or Afrikaans as the LoLT from Grade 4 onward. The scale of this challenge is stark; about 7% of South African learners speak English at home yet 23% of Grade 1 learners enter the system in straight-to-English settings (Sibanda & Tshehla 2025; South Africa, Department of Basic Education [SA DBE] 2018). For those learning in an indigenous home language, the three years of basic English taught as a First Additional Language, typically 1–3 h per week, provides inadequate preparation for the transition to English as the new LoLT by Grade 4 (Sibanda & Tshehla 2025). From a cognitive perspective, acquiring early literacy in a second language places significant demands on working memory and processing efficiency. They must simultaneously learn to decode (break down) unfamiliar phoneme-grapheme relationships (the connection between written letters or letter combinations and the sounds they represent) and acquire academic vocabulary in a language they do not speak at home (Cummins 2000). This disconnect is rarely addressed in curriculum pacing, leaving multilingual learners behind from the outset.

Fluency development plays a pivotal role in bridging decoding and comprehension. The SVR (Gough & Tunmer 1986) and componential models (Duke & Cartwright 2021) highlight the need to integrate phonics, oral fluency, and vocabulary in a coherent instructional sequence. Multilingual learners benefit most when these components are explicitly scaffolded and reinforced through repeated reading, guided practice, and contextualised texts (Cummins 2000; Pikulski & Chard 2005). Without fluency, learners lack the automaticity needed for comprehension, further entrenching literacy gaps.

Effective reading development depends on integrating decoding, fluency, vocabulary, and comprehension instruction (Pretorius & Spaull 2016). Of these, fluency has emerged, from extensive research, as a key predictor of later reading comprehension (Pikulski & Chard 2005). Without fluency, learners cannot allocate cognitive resources to make meaning from text. As Pikulski and Chard (2005) note, fluency development enables readers to allocate attention to meaning rather than decoding, acting as a bridge between phonics and comprehension. The SVR (Gough & Tunmer 1986) and updated componential models (Duke & Cartwright 2021) identify fluency as a bridge between phonics and understanding. The SVR posits that reading comprehension arises from the interaction between two core processes: decoding and language comprehension. While influential, the SVR has been critiqued for underemphasising the sociocultural and cognitive challenges of multilingual contexts, particularly for learners acquiring literacy in a non-home language (Snow & Matthews 2016). For isiXhosa-speaking learners, for example, decoding in English requires not only grapheme–phoneme knowledge but also cross-linguistic transfer (using knowledge from the home language to support learning in a second language) and syntactic reorientation (adjusting to differences in word order and sentence structure between languages). This highlights the tension between models privileging decoding and those, such as Cummins’s Basic Interpersonal Communicative Skills (BICS) and Cognitive Academic Language Proficiency (CALP) framework ([1979] 2008), which foreground the layered progression from surface-level communicative fluency (BICS) to the deeper CALP required for comprehension. In practice, decoding-heavy approaches can accelerate word recognition and fluency yet risk neglecting the broader linguistic and academic language demands that ultimately enable comprehension. QondaRead responds to this by prioritising explicit instruction in phonics and daily fluency practice while also situating these within structured opportunities for vocabulary growth and meaning making, aiming to strengthen both dimensions of the reading equation. In addition, QondaRead aligns with updated componential models of reading (Duke & Cartwright 2021), which identify fluency as a critical bridge between decoding and comprehension. Fluent readers are able to read with automaticity, inflection, and accuracy, allowing cognitive resources to shift from word recognition to meaning making (Pikulski & Chard 2005). This is particularly vital for learners acquiring literacy in a second language, where decoding is complicated by unfamiliar vocabulary and syntax. Figure 1 illustrates the theoretical foundations of QondaRead components.

Research underscores the critical role of teacher practice in shaping learner outcomes, particularly in early grade literacy. Structured pedagogy interventions such as Early Grade Reading Study (EGRS) and Funda Wande have demonstrated that when teachers are equipped with accessible materials and practical routines, reading outcomes improve significantly (Hoadley & Boyd 2022). However, such interventions often require extensive coaching or external support, limiting scalability in under-resourced contexts. QondaRead was designed for full delivery by existing classroom teachers, requiring no withdrawal from the curriculum or additional time beyond the literacy block. Focus group feedback suggests that teachers valued the predictability and structure provided by daily phonics and guided reading routines, and many reported growing confidence in implementing evidence-based strategies. This reinforces the importance of instructional models that build teacher ownership and are feasible within everyday classroom constraints.

A growing body of research supports the claim that carefully structured interventions can deliver meaningful literacy gains in under-resourced settings. In a study by Long and Bowles (2024), learners at a no-fee Quintile 3 school in Makhanda participating in the Whistle Stop School pull-out intervention programme consistently outperformed both provincial and national benchmarks on measures of ORF and comprehension. Fluency is most often assessed using ORF measures. Learners read a given text for 1 min and the number of WCPM is recorded as their ORF score. These offer valuable diagnostic insights into learner progress particularly when measured against national benchmarks (Funda Uphumele Consortium 2025; Khulisa Management Services & SA DBE 2022). Internationally, similar results have been reported in small-group, phonics-based interventions (Elston, Tiba & Condy 2022), with guided and repeated reading shown to be effective across multilingual contexts (Hudson, Mercer & Lane 2000). Comparable multilingual interventions in other low-resource contexts provide useful points of reference. In Kenya, the Primary Math and Reading initiative demonstrated that daily, structured instruction in phonics and fluency across Kiswahili and English substantially improved learners’ decoding and ORF, with evidence of cross-linguistic transfer (Piper et al. 2018). Further research in Kenya underscores the mediating role of text reading fluency between decoding and comprehension, reinforcing its importance as a developmental bridge (Wawire, Liang & Piper 2023). In India, the Facilitating Reading Acquisition in Multilingual Environments (FRAME-India) study and the MultiLiLa project similarly showed that multilingual scaffolding and biliteracy approaches can accelerate foundational literacy when phonics and fluency practice are integrated into regular instruction (Hickey, Lewis & Baker 2020; Nag, Vaidya & Chiat 2014). More recently, classroom phonics programmes supported by digital tools, such as GraphoLearn Rime, have been found to enhance English literacy outcomes for children in low-resource Indian schools (Mishra, Dash & Singh 2024). Together, these findings suggest that QondaRead’s focus on systematic phonics and fluency instruction aligns with a broader evidence base demonstrating the effectiveness of such approaches in multilingual, resource-constrained settings.

While a growing literature documents early-grade reading interventions where the local African language is the LoLT, far fewer South African studies are designed for learners who speak an African language at home but are taught in English. Recent large-scale ‘structured pedagogy’ programmes are typically implemented by classroom teachers using lesson plans, readers and coaching (e.g. EGRS, Funda Wande). Against this backdrop, QondaRead is distinctive in targeting straight-to-English settings with a focus on English language fluency, embedding repeated-reading routines within everyday classroom practice, implemented by school personnel and infrastructure, supporting scalability (Hoadley & Boyd 2022; Mohohlwane et al. 2024). Through its use of classroom-embedded instruction, localised materials, and benchmark-aligned assessments, QondaRead represents an attempt to design a contextually responsive and relatively low-cost approach to supporting early literacy. This study contributes to the literature by reporting findings from the programme’s first full-scale implementation, using a mixed-methods design to generate evidence that may inform future efforts aimed at strengthening literacy provision in under-resourced contexts.

Including detailed costing data supports the central premise of this study, that structured, high-impact literacy interventions can be designed to operate at low cost within the constraints of public school systems. Very few literacy interventions in South Africa publicly report disaggregated per-learner costing, making QondaRead’s transparency an important contribution for policymakers and implementers seeking scalable models. The cost of materials per teacher varied by grade band. For Grade R and Grade 1, the package included a full set of graded readers with supporting materials, a complete single-sound phonics programme, and a data projector to enable delivery of lessons as designed. The total cost per teacher in 2024 was R30 000. These resources are expected to last for a minimum of five years. With an average class size of 35 learners, this translates to a cost per learner of approximately R172 (175 learners over five years). For Grade 2 and Grade 3, the per-teacher cost was lower, at R22 000, as it was limited to a set of graded readers per class. Using the same assumptions regarding longevity and class size, this equates to a cost per learner of approximately R126. These figures offer a transparent and replicable estimate of the per-learner cost of the QondaRead intervention, excluding training and personnel costs. Detailed financial reporting by other literacy interventions in South Africa remains limited; cost-effectiveness comparisons are often based solely on learning outcomes, such as effect sizes reported in randomised evaluations. For example, the Funda Wande programme has published strong learner gains but has not released publicly disaggregated per-learner cost data. In this context, QondaRead’s costing approach contributes rare and valuable insight into the real-world affordability of literacy interventions, which may serve as a useful benchmark for future programme comparisons.

This study employed a quasi-experimental, pre–post, mixed-method design to evaluate the impact of the QondaRead programme on early literacy development in multilingual under-resourced settings. Quantitative data were collected to measure changes in foundational reading skills over the course of a school year, while qualitative insights were gathered to explore teacher experiences, implementation fidelity, and contextual factors influencing programme delivery.

Although no control group was used, historical data from one school enabled a limited comparison with pre-intervention cohorts. This design allows for robust interpretation of change over time but limits causal inference, a recognised limitation of quasi-experimental studies.

The evaluation was conducted in four resource-constrained, English LoLT primary schools in Makhanda, a small city in the Eastern Cape province of South Africa. These schools serve predominantly isiXhosa-speaking learners and are situated in historically disadvantaged communities characterised by poverty, under-resourcing, and low academic performance. GADRA Education, a long-standing educational non-governmental organisation in Makhanda, led the intervention in collaboration with school staff and local funders. This city has a unique educational ecology characterised by strong cross-sector partnerships, data-informed interventions, and an explicit commitment to improving public education outcomes (Long & Bowles 2024; Msomi 2024; Ntlabezo & Westaway 2024).

The sample included 712 learners across Grades 1–3 from four primary schools implementing the QondaRead programme. Schools were selected through purposive sampling based on their participation in the QondaRead intervention and their classification as under-resourced schools where English serves as the LoLT.

All learners in these grades at the four participating schools were included in both the intervention and the pre–post assessment cycles. Specifically:

Approximately 60 learners were excluded from the final analysis because of absenteeism at either the baseline or endline testing. For the qualitative component, 12 Foundation Phase teachers (three per school) participated in focus group discussions. Teachers were selected purposively based on their direct involvement in implementing QondaRead across the three grade levels. All participants provided informed consent.

This study used secondary data obtained from GADRA Education with the necessary permissions. The study was conducted in accordance with GADRA’s internal data governance and ethical protocols, rather than through a separate university ethics review. It did not undergo a separate university Institutional Review Board (IRB) review because it was an internal, commissioned study by GADRA Education. Under applicable guidelines, it did not qualify as human subjects research requiring IRB oversight, as it used fully anonymised, retrospective data with no new participant interaction. No new informed consent was sought from the original participants, as the study relied on existing data that were fully anonymised and involved no direct contact or intervention, thereby meeting the usual criteria for a consent waiver (i.e., minimal risk and impracticability of re-consent).

All records were fully stripped of direct and indirect identifiers by the data provider prior to researcher access, ensuring that no student, teacher, or school could be identified in the analysis. Data use was strictly in accordance with GADRA Education’s data governance policies, and the dataset was handled on secure, access-controlled systems in line with standard data protection practices. No new data collection or participant contact occurred, so the study imposed no burden and raised no concerns about coercion or power dynamics. We recognise that students are a vulnerable population, but in this secondary analysis, their welfare was safeguarded by ensuring complete anonymity and requiring no active involvement on their part.

The SA DBE and the 2025 Funda Uphumele Consortium define minimum reading benchmarks for learners acquiring English as an additional language:

National performance remains low, with the 2025 survey reporting:

These statistics underscore the scale of South Africa’s literacy crisis and provide a basis for evaluating QondaRead’s relative impact.

Analysis of performance across the four participating schools revealed consistent gains. While learners with stronger baseline skills made the most progress in absolute terms, relative gains were largest among learners who started with the weakest reading fluency, indicating that QondaRead may help narrow early literacy gaps rather than widen them. For example, one teacher noted:

This shift corresponds with a rise in the proportion of learners achieving 40+ LSF from 0% to 29%, as shown in Figure 2. Similarly, a Grade 2 teacher’s comment that ‘learners became more confident reading aloud’ aligns with the Grade 2 average ORF increase to over 35 WCPM, well above the national average of 25 WCPM. The large effect sizes observed across all grade levels (Cohen’s d > 1.0) indicate that the improvements in reading fluency were not only statistically significant but also educationally substantial. In schools where learners began close to fluency, the programme accelerated their progress to well beyond national expectations. In lower-performing contexts, learners made measurable, statistically significant improvements that moved them closer to reading proficiency. These gains represent a critical step toward ensuring that learners transition successfully from ‘learning to read’ to ‘reading to learn’. The historical comparison with School D’s 2017 Grade 1 cohort provides further insight into the nature of these gains. The November 2024 cohort achieved a mean LSF score approximately 40% higher than the 2017 cohort at the same point in the year, with a significantly smaller proportion of complete non-readers and a larger share meeting the SA DBE benchmark. This pattern suggests that QondaRead’s implementation in earlier grades may have contributed to stronger school-entry skills, underscoring the potential for cumulative, cross-cohort effects in literacy interventions. In Grade 2 and Grade 3, learners approached or exceeded SA DBE fluency benchmarks. The effect sizes observed in this study far exceed what would typically be expected from normal maturation or classroom exposure alone. National studies, for example, show that Grade 1 learners usually make only modest literacy gains during the year (Khosa 2022; Taylor et al. 2019; Western Cape Education Department 2006). The much larger gains reported here, therefore, provide strong evidence of the intervention’s practical significance for improving foundational literacy. These outcomes offer a direct response to our core research question. They show that when appropriately designed and integrated into daily teaching routines, low-cost interventions like QondaRead can produce literacy gains that exceed those expected from typical classroom exposure alone, even for learners who do not speak the LoLT at home. The accelerated gains observed suggest that with appropriate classroom-based support, learners can make rapid literacy progress, even in highly constrained learning environments where they are not learning to read in the language of their homes.

In a national context where early reading stagnation persists, the Funda Uphumele Consortium (2025) reports that only 18% of Grade 1 learners meet the decoding benchmark, and just 19% of Grade 3 learners achieve the comprehension threshold of 50 WCPM. Against this backdrop, QondaRead schools demonstrate substantial divergence from this pattern, represented in Figure 3.

As shown in Figure 3, QondaRead learners in Grades 1–3 exceeded national averages by 15–20 WCPM on average, with 29% of Grade 1 learners reaching the LSF benchmark, and 41% of Grade 3 learners surpassing the 50 WCPM comprehension threshold, more than double the national average.

Unlike many externally delivered or small-group pull-out programmes, QondaRead was designed to be implemented within the existing classroom routine, requiring no changes to the school timetable or additional teaching staff and no extra administrative burden for teachers. This is particularly critical in low-resource settings, where scalability often fails as a result of dependence on external personnel or infrastructure. The programme’s success reinforces the growing evidence that literacy interventions embedded in daily instruction, when supported by structured materials, can be both effective and sustainable (Duke & Cartwright 2021; Pretorius & Spaull 2016). The simplicity and accessibility of the phonics and guided reading model used in QondaRead make it an attractive option for broader replication.

One of the most encouraging findings was that learners with the weakest baseline performance demonstrated the greatest relative improvement. This suggests that QondaRead may help narrow early literacy gaps rather than exacerbate them, an important consideration for equity in education. These outcomes speak directly to the aims of fostering inclusive, high-quality literacy learning in diverse and dynamic contexts. By meeting learners where they are, linguistically, cognitively, and socially, QondaRead helped to surface and strengthen their latent literacy potential.

Qualitative data underscored the value of teacher ownership and professional learning in successful implementation. Teachers described the programme as manageable, meaningful, and aligned with their instructional goals. They valued the structure and consistency it brought to their teaching practice, as well as the collaborative learning enabled by school-wide implementation. ‘We had a clear plan every day. That made the biggest difference’, noted a Grade 1 teacher, referring to the guided structure.

This supports the broader literature on responsive pedagogy and funds of knowledge, suggesting that effective literacy instruction is not simply about materials, but about enabling teachers to draw on their own professional knowledge and that of their learners to make meaningful instructional choices (Elston et al. 2022; Moll et al. 1992).

While the study provides robust early evidence of impact, several limitations must be acknowledged.

Without a counterfactual, it is difficult to definitively attribute improvements to the intervention alone. Although historical comparisons (e.g., School D’s 2017 cohort) provide some context, these are not equivalent to a causal research design. Future evaluations could benefit from incorporating quasi-experimental methods, such as a matched comparison group drawn from similar under-resourced schools not implementing QondaRead. Additionally, a difference-in-difference (DiD) analysis, comparing literacy gains over time between treatment and non-treatment schools, could offer stronger causal inference while remaining feasible in real-world school settings. These approaches would allow researchers to isolate the impact of QondaRead more robustly while respecting the ethical and logistical constraints of educational fieldwork.

The study was confined to a single city with strong NGO–school partnerships, which may not generalise to other provinces.

Sustainability and long-term outcomes remain unknown. Although short-term fluency gains were substantial, follow-up studies are needed to examine whether these gains translate into later reading comprehension, vocabulary development, and academic achievement in Grade 4 and beyond.

While national benchmark comparisons offer insight, differences in school context, assessment administration, and sample composition should be noted when interpreting divergences.

Future evaluations should explore whether short-term fluency gains hold over time and whether early literacy growth leads to sustained improvements in reading comprehension and academic success. Such studies should consider longitudinal tracking, randomised control designs, and implementation studies across varied provinces to assess adaptability and scalability at a systemic level.

The results of this study hold important implications for both policy and practice. First, they highlight the feasibility of low-cost, high-impact interventions that work within existing systems and timetables. Second, they call for investment in teacher-led models of literacy development that embed structured fluency and phonics instruction without displacing broader comprehension goals. Finally, they reinforce the urgent need for early and sustained intervention in under-resourced schools, particularly before the Grade 4 transition to English as the LoLT. These findings also offer clear policy signals. Structured decoding and fluency instruction should be mandated and resourced from Grade 1 in all English-LoLT, under-resourced schools, especially in multilingual contexts. Classroom-embedded delivery must be prioritised over externally driven, high-cost interventions, to ensure sustainability. National funding frameworks should incorporate per-learner costing standards, to enhance transparency and planning. Critically, language transition policies must be aligned with long-term fluency support to avoid exacerbating early learning inequalities. If scaled with fidelity, QondaRead-style programmes could help shift South Africa’s literacy trajectory, without overburdening teachers or education budgets.