This visual-world eye-tracking study examined the prosody and verb semantics integration in native and non-native Mandarin speech perception. Native Mandarin and English listeners' eye movements were recorded and they were asked to click on the object named in the second sentence while they listened to Mandarin sentences within a brief discourse context. These sentences varied in three binary factors: prosodic accent, target object information status, and semantic match between verb and target object. Results showed a complex interaction between L1, prosody, and verb semantics. When processing both old and new information, Mandarin listeners interactively combined high-level verb semantic cues with low-level prosodic cues across diverse contexts, such that effects of verb semantics were exaggerated under contrastive focus. English listeners showed a similar pattern with old information as Mandarin listeners, but with new information they showed less integration of different speech cues than Mandarin listeners. These results suggest that non-native listeners can adopt native-like strategies in integrating prosodic and semantic speech cues, but only when the information status of target words is familiar, and hence less tasking to process.