<p>Previous research has observed P600 but no N400 effects on grammatically well-formed but role-reversed sentences, prompting an examination of the syntax-based semantic composition. Multi-stream and single-stream models propose divergent meaning integration mechanisms and different functional interpretations of N400 and P600 components. Here, we test these competing hypotheses by examining how the ease of lexical retrieval and the plausibility of role-reversed sentences influence brain responses. We engaged Chinese native speakers in a sentence reading experiment while recording their brain potentials. The experiment manipulated lexical retrieval ease through priming the object nouns or not, and varied the plausibility of role-reversed sentences by employing the distinctive characteristic of Mandarin Chinese that allows plausible role-reversed sentences. Results revealed that priming significantly reduced the N400 responses, suggesting facilitated lexical retrieval processes. Implausible role reversals elicited greater P600 amplitudes compared to plausible role reversals and non-reversed canonical sentences, despite both reversed conditions featuring semantic-syntactic conflicts, suggesting that P600 is associated with semantic plausibility rather than syntax-semantic conflict monitoring. For plausible role reversals, we observed a sustained frontally distributed positivity, suggesting an elaborative analysis of events implied by non-canonical verb-argument structures. These findings support single-stream models of sentence processing, where the N400 indexes semantic retrieval and the P600 reflects integration efforts to build coherent sentence representations. This study is the first to separate semantic implausibility and syntactic-semantic conflict, providing insight into the neurocognitive mechanisms involved in syntax-semantic interface processing.</p>