Abstract Research syntheses have demonstrated that pronunciation instruction works, which means that whether instruction is effective is no longer an open question. Instead, contemporary intervention research has shifted to investigating how instruction can be further optimized, asking targeted questions about the instructional features that catalyze learning. In this paper, I examine the concept of instructional optimization, focusing on anticipated effect sizes (gains). I outline a four-pronged empirical approach to provide robust data for designing optimal pronunciation interventions. First, I describe the need for replication studies, which provide insight into the precision and stability of effects across distinct research samples and contexts. Second, I advocate for a systematic approach to study design. In such an approach, which is closely tied to the principles of replication, one or two variables are manipulated at a time, leading to a set of maximally comparable studies that lend insight into the impact of specific variables. Third, I explain the need to situate instruction within a longitudinal perspective to examine how robust and durable instructional gains are. Finally, I turn to adaptive approaches, where the surface format that instruction takes is highly variable and responsive to learner needs while the adaptive decision tree that generates the form is fixed and replicable.
Charlie Nagle (Mon,) studied this question.