By: Meredith McCormac, Director of Programs, Luminos Fund

When we launched the Bilingual Boost program, funded by the Gates Foundation, we set out to explore a challenging question: How can education systems help children successfully learn to read and understand two or more languages, including languages they may not yet speak or understand?

There are many factors at play, including what languages are involved, what children already know, how teachers teach, when an additional language is introduced, what materials are available, and the operational and political realities of the education system itself. This complexity is why, from the beginning, Bilingual Boost needed more than a strong implementing team. It needed people who would challenge us with hard questions, evidence, and experience—which is why we established the Bilingual Advisory Council (BAC), bringing together researchers, practitioners, policymakers, and experts in language, literacy, assessment, and education systems from across Africa and beyond.

Bilingual Boost is a three-year initiative to generate evidence on how children develop literacy skills across languages, currently working across Ethiopia, Ghana, Mozambique, and South Africa. The BAC was never intended to be a group that periodically receives updates and endorses our next steps. Rather, its role is to improve the work as it unfolds: interrogating assumptions, bringing new research and data into the project, challenging how we interpret evidence, and ultimately helping us translate findings into policy messages that are actually useful to teachers, ministries, and everyone else making decisions about how their children should learn.

After an initial virtual meeting earlier this year, the BAC came together in South Africa last month for its first in-person convening. Over two days, BAC members examined the theories of change behind our country studies, debated how literacy skills transfer across languages, challenged our research questions and assessment designs, and considered what evidence could emerge that policymakers actually need. We left with sharper thinking than we arrived with, which was exactly the point.

Making Evidence Useful

One of the clearest examples of the value of these discussions came through a rich conversation about the Biliteracy House,[1] a conceptual model for understanding how reading skills develop and transfer across languages when children learn to read in two languages.

New research presented to the BAC provides compelling evidence of a tipping point in biliteracy development: a threshold beyond which children’s decoding skills in their first language begin to transfer much more strongly to learning to read in a second language. The threshold varies across language pairs and writing systems, but it was evident across all of the language pairs studied.

Oral language remains one of the strongest predictors of reading development, but rather than reaching a particular threshold, it continues to develop and contribute to reading over time.

For policymakers and practitioners, findings like these raise an important challenge: how do we translate such nuanced evidence into practical guidance without oversimplifying it? The discussion highlighted two implications. First, strengthening children’s reading skills in their first language matters because there appears to be a point beyond which those skills transfer much more powerfully to second-language reading—even if the exact point differs by language and context. Second, oral language needs to be explicitly supported throughout the process because its contribution to reading remains important even without the same threshold effect.

For Bilingual Boost, this gives us a sharper consideration for our country studies: not simply whether skills transfer across languages, but when that transfer accelerates, how it differs across languages, and what that means for the sequence of literacy instruction.

Learning From Disagreement

Not every discussion at the convening converged on an answer. Some of the most useful exchanges were the ones where the BAC did not reach a consensus. In Mozambique, for example, the bilingual model builds children’s literacy in their first language while developing oral Portuguese before introducing formal Portuguese decoding later. Some BAC members questioned whether delaying Portuguese decoding would leave bilingual students at a significant disadvantage on Portuguese reading measures in Grade 3. Others made the case for first establishing a stronger literacy foundation in the child’s first language before introducing formal Portuguese decoding.

There was no neat resolution. Under Mozambique’s current bilingual model, children in a monolingual Portuguese pathway may have had years of Portuguese decoding instruction by the end of Grade 3, while bilingual students may have had only months. Comparing their Portuguese outcomes at that point may therefore tell only part of the story about bilingual literacy, even as ministries understandably need evidence within the timelines their systems use.

The BAC emphasized that the benefits of building strong language and literacy foundations through bilingual instruction may become more visible in later grades, as children transfer skills developed in their first language to literacy in a second language. This principle is well established in the evidence base, but Bilingual Boost offers an opportunity to examine how these longer-term benefits emerge across different bilingual models and contexts.

The discussion also reinforced that Bilingual Boost is not one experiment repeated four times. Our studies are asking related questions in very different linguistic and education contexts. Measures may need to differ, interventions may need to differ, and the most important policy questions may differ. Cross-country learning does not mean forcing every country into the same model. Our goal is to understand what can legitimately be learned across these different contexts.

Luminos students in Ghana complete a group activity during a lesson. (Photo: Obeng Baah for the Luminos Fund) 

Evidence Must Work in Real Classrooms

Across all these discussions, one important consideration kept surfacing: what happens between teachers and children in the classroom.

One of Bilingual Boost’s core research questions asks how large children’s vocabularies are in their first (L1) and second (L2) languages. To answer this, we are developing vocabulary assessments in each of the languages we are studying, drawing on language-specific word lists and expertise to identify words that are appropriate for children at different levels of vocabulary development. When BAC members tested our pilot vocabulary assessments, seemingly small design choices—ambiguous images, distractors, accents, how children experience getting an answer wrong—quickly became consequential. The exercise was a reminder that these details can affect what an assessment is actually measuring, and therefore how confidently we can interpret the results.

They raised a similar point about oral language instruction. While researchers increasingly agree on what works for decoding, oral language pedagogy can be harder to define. Teachers may be encouraged to promote conversation or ask deeper questions, but what does that actually look like in a crowded, multilingual classroom? One idea from the discussion was to identify a small number of practical, evidence-informed strategies and show what they look like when ordinary teachers implement them well in ordinary classrooms. That kind of translation—from a research principle to something a teacher can actually do, in their context—is central to what Bilingual Boost ultimately needs to contribute.

These were not just theoretical discussions: they are already informing our next steps, including refinements to our vocabulary assessments and how we support teachers to use evidence-informed oral-language strategies in multilingual classrooms.

If Bilingual Boost ultimately contributes to how education systems support children learning across languages, it will not be because we began with all the answers. It will be because we invited people to challenge our thinking and stayed open to what the evidence tells us now, and what we want to continue to uncover empirically, practically, and politically for supporting children to become strong bilingual readers.

To learn more about Bilingual Boost and the Bilingual Advisory Council, please visit our website.

Dr. Meredith McCormac is a Director of Programs at the Luminos Fund, where she manages a multi-year foundational literacy program designed for implementation in multilingual environments across multiple African countries. Previously, Meredith held leadership roles at RTI International and World Learning, directing several USAID programs focused on foundational learning. She also worked as a consultant for the Global Partnership for Education (GPE), partnering with ministries of education across East and Southern Africa to support their development of GPE grants. Meredith holds a PhD in International Education Policy from the University of Maryland and completed her doctoral research in Ethiopia as a Fulbright scholar. 

[1] Nakamura, P. R., & Holla, C. (2026). A tipping point for transfer: Evidence for bilingual reading transfer thresholds from 14 language pairs across five countries. In C. Lü, X. Li, & S. Ke (Eds.), Advances in second language reading and cross-linguistic transfer: In honor of Keiko Koda. Springer.