What makes speech sound non-native?

Abstract This study investigates how monolingual listeners evaluate the perceived nativeness of bilingual speech in the first (L1) and second language (L2), and which pronunciation cues underlie these judgments. English and German listeners rated the L1/L2 speech produced by L1 English-L2 German late bilinguals, alongside monolingual controls, and provided open-ended comments on perceived non-native features. Results showed that bilinguals were rated as less nativelike than monolinguals in both languages, with substantially greater deviations in the L2. Analyses of listener comments revealed a clear cross-linguistic dissociation: listeners relied more on prosodic cues when evaluating bilinguals’ L1 speech, and more on segmental cues in L2 pronunciation. These findings show that the relative weighting of perceptual cues differs depending on the overall nativelikeness of the speech signal. By combining nativeness ratings with explicit listener comments across both languages, this study provides new insights into the perceptual basis of L1 and L2 nativeness.