Predicting word or sentence structure from prosodic cues (e.g., pitch, pauses, or duration) is an important mechanism in speech processing. The mechanism is readily employed by native listeners but considerably less so by second language learners except with extensive training. This prompts the question of how flexible and adaptive the predictive system is. Can listeners adjust to accommodate, for example, diverging predictive cues in a different dialect? The present paper tests this for prosody-based word structure prediction in mainland Scandinavian. Neurophysiological and behavioural results suggest that listeners from non-standard dialect areas can in fact predict word forms in the standard variety even when the predictive cues and their validity differ considerably. This suggests a certain degree of adaptability in the predictive system, potentially depending on factors like familiarity, prestige, and exposure onset.