This article presents a stratified cluster-randomized trial that tested whether a common National Curriculum Development Centre (NCDC)-aligned blending routine would work equally well in two closely related Ugandan languages namely Luganda and Lusoga, or whether orthography-specific adaptation was needed. Twenty-four Primary 3 classes were recruited across Luganda- and Lusoga-speaking communities and randomized within language to either a 12-week supplemental blending condition or a business-as-usual comparison. The routine was built around the familiar gradual-release sequence of teacher modelling, guided practice, paired work, and individual response turns. It was anchored directly in the P3 Luganda teacher guide and pupil book for the Luganda arm and matched in sequence and lesson purpose for the Lusoga arm. The study was framed by the Script-Dependent Hypothesis, which predicts that reading development varies with the way print maps onto speech. Mixed-effects models showed that the intervention improved oral blending and segmentation in both language groups, but gains were larger in Luganda, especially on blending accuracy and pseudoword decoding. Observation data also showed that Lusoga teachers more often paused to reteach how to handle multiletter and prenasalized patterns. The findings suggest that the NCDC methodology itself is not the problem. The gradual-release structure appears to travel well. What did not travel unchanged was the instructional grain size of the blending tasks. The article argues for a careful middle position: keep the NCDC pedagogy, but adapt word selection, chunking routines, and practice pacing to the orthographic and phonological demands of each language.