Exploratory Simulations for Confirmatory Analysis: English Auditory Word Learning

The relationship between phonological similarity and sub-lexical frequency in auditory word learning has been studied in the experimental literature but given much less attention in terms of modeling. The current work takes a dynamic cognitive network approach to simulating trial-level output to subjects that vary according to their sensitivity to phonological similarity (modeled via positional phoneme density), sub-lexical frequency (phonotactic probability), and the task effect of recency. After an introduction to the model's components, I demonstrate how a cognitive model that can account for subject- and trial-level phenomena can be used to simulate multi-level datasets in what is called an exploratory simulation for confirmatory analysis (ESCA). I discuss how the capacity to weigh the pros and cons of experimental parameters through effect sizes and statistical power generated in an ESCA has implications to educators and researchers.