Structure and Sound: How Similarity and Consistency Jointly Shape Language Learning and Change

This study investigates how two core pressures in language change—structural consistency and phonological similarity—interact to shape learners’ representations of linguistic systems. Using a 2x2 artificial language learning experiment, 80 adult participants were exposed to one of four spoken artificial languages that varied in the consistency of plural marker conditioning (consistent vs inconsistent) and the phonological similarity of the alternating markers (similar vs. distinct). Results revealed that learners’ encoding and reproduction of linguistic input are jointly shaped by perceptual constraints and structural cues. Learners were less accurate in repetition when markers were similar, particularly when they were also inconsistently conditioned, suggesting that structural consistency protects against confusability. When producing the markers themselves, participants regularized more in the similar-marker conditions, regardless of consistency, yet, crucially, they discovered and maintained consistent input structures more readily when markers were distinct, suggesting that perceptual similarity can erode structural learning. In addition, participants’ repetition error patterns aligned with attested pathways of sound change, including a bias toward devoicing and diphthong shifts analogous to those found in natural language sound changes. We argue that language change cannot be fully understood without considering how input properties interact with cognitive biases. Overall, this study highlights the value of experimental paradigms for isolating mechanisms of change while also pointing to the complex interplay of structure and sound in shaping language evolution.