Evaluating the causal impact of first-language reading instruction on the subsequent development of second-language reading proficiency in multilingual Ghana

• In Event: Laying the bricks for foundational learning through effective language of instruction decision-making

Proposal

In multilingual settings like Ghana, children often begin learning to read in their mother tongue (L1) before transitioning to a second language (L2), such as English. This approach is based on the cross-language transfer hypothesis, which suggests that mastering L1 skills supports L2 reading acquisition. However, most evidence has been correlational, leaving a gap in understanding the causal links between L1 and L2 reading skills.
This study aims to establish a causal relationship between L1 reading skills and subsequent L2 reading development. In practical terms, we investigate whether teaching children L1 reading before L2 reading is effective or necessary, focusing on both basic and complex reading skills, such as letter-sound recognition, decoding, and oral reading fluency.
We conducted the study in Ghana with a sample of 4,352 first- and second-grade students from 200 randomly selected schools. We used an instrumental variables (IV) approach, applying a randomized L1-focused reading intervention to create exogenous variation in L1 reading outcomes. The intervention targeted L1 instruction exclusively, ensuring compliance with the exclusion restriction. Of the 200 schools, 100 were randomly assigned to receive the L1-focused intervention, while the remaining 100 served as the control group. This randomization satisfied the IV’s relevance condition, allowing us to isolate the causal effect of L1 reading proficiency on L2 reading outcomes (see below for Stage 1 results). Children’s reading abilities in both L1 and L2 were measured using the Early Grade Reading Assessment.
The first-stage results show that the L1-focused intervention had significant impacts on learners’ L1 reading performance, satisfying the relevance condition of the IV. Learners in the intervention group improved in all L1 reading skills. In the second stage, we find only L1 letter-sound identification had a significant positive effect on L2 letter-sound identification. Improvements in L1 nonword decoding and oral reading fluency did not transfer to their corresponding L2 reading skills. A possible explanation for this limited transfer is that L1 and L2 in Ghana share nearly identical Roman alphabet orthographic features, particularly at the alphabetic level.
Together, our study provides causal evidence that L1-L2 cross-language transfer is inconsistent and occurs only at the basic alphabet level, with limited effectiveness in enhancing more advanced reading skills. We acknowledge that L1 instruction serves important purposes beyond language learning, such as promoting cultural and linguistic diversity. However, when focusing specifically on language acquisition, our findings suggest that delaying English instruction may not be necessary and could be introduced from the outset. Policymakers could balance L1 and L2 instruction to optimize learning outcomes while also considering cultural diversity.

Author

• Jeongmin Lee, University of Notre Dame