Reconciling diverse findings in studies of phonetic convergence

Abstract

Phonetic convergence occurs both when individuals interact in conversation, and when listeners rapidly repeat words presented over headphones. Results from multiple studies examining phonetic convergence offer an array of often confusing and disparate findings. Reconciling such diverse findings is difficult without a clear rationale for engaging in one acoustic measure over another. The current paper proposes a paradigm that models perceptual and acoustic measures together. Measures of multiple acoustic-phonetic attributes were compared with a perceptual measure of phonetic convergence in a shadowing study. Although convergence was not significant in any acoustic measure alone, the combination of acoustic attributes predicted perceived phonetic convergence on an item-by-item basis. Because perceptual measures integrate across multiple acoustic-phonetic dimensions, future studies of phonetic convergence should use perceptual tasks to calibrate the relative contribution of individual acoustic-phonetic parameters.