Accentedness and Comprehensibility in Second Language Cantonese: An Investigation of Mandarin Learners' Read Speech

Research on accentedness and comprehensibility has largely focused on nontonal target languages, leaving it unclear whether established predictors of these two constructs apply to tonal languages with complex tone systems, such as Cantonese. We investigated 40 Mandarin-speaking learners' read speech in second language (L2) Cantonese by combining native listener ratings with 58 candidate predictors encompassing acoustic, sociolinguistic, and auditory sensitivity measures in a single analytic framework. Generalized additive mixed models were used to capture potential nonlinear relationships between predictors and ratings. Results indicate that accentedness and comprehensibility were strongly correlated (r = .79), indicating substantial but not complete perceptual overlap. Greater rhythmic variability (VarcoΔV) was associated with higher ratings across both constructs, whereas higher pause frequency predicted lower ratings. With regard to accentedness, additional predictors included longer residence in Hong Kong, target-like formant values for certain vowels, and native-like tonal patterns. By contrast, comprehensibility was linked to more extensive use of Cantonese, more positive language attitudes toward Mandarin, earlier age of acquisition, stronger sensitivity to amplitude rise time (a measure of auditory temporal processing), and a few vowel- and tone-related measures. These findings extend the accentedness-comprehensibility literature to a typologically distinct tonal language and suggest that the relative contributions of acoustic, sociolinguistic, and auditory sensitivity factors to listener judgments may vary across linguistic contexts, underscoring the need to broaden current models of L2 speech learning beyond predominantly English-based evidence.