A comment on "Why Do Children Think Words Are Mutually Exclusive?" (Brody et al., 2024)

An early emerging understanding of information structure in spoken language is demonstrated in Brody et al.'s (2024) study, contrary to the general assumption that young children innately assume new and familiar words to be mutually exclusive. Across 3 experiments, learners up to 2 years of age (N = 106) showed mutual exclusivity if the novel words were spoken with focus, an information-structural marker of contrast. We successfully computationally reproduced the reported results, ran additional robustness tests, and considered potential covariates that were recorded but not used in the original paper (i.e. age and gender). We also assess potentially disproportionate effects of influential participants via Cook's distance, and refit a model where their data are removed. Further, we consider alternative specifications of random effects (e.g., maximal models) to address convergence issues and obtain interval estimates of experimental conditions' means that account for random effects. Finally, we examine whether different optimizers address model convergence issues. Across our replication tests, we find that the original results are robust and key estimates remain significant. Of note, our influence analysis and Bayesian analysis showed stronger effects than those originally reported, likely attributable to inattentive/noisy responding to some trials. Overall, our replication efforts should provide increased confidence in the original effects.