The present study aimed to obtain a comprehensive understanding about how Chinese Mandarin-speaking children performed prosodic decoding of syntactic ambiguity under the visual and audio modality and to what extent their manipulation of implicit prosody differed from explicit one. Altogether 31 Mandarin-speaking children were recruited and required to resolve different types of syntactic ambiguity when reading and perceiving a sentence. Accuracy and reaction time in the visual and audio modes were measured and analyzed. The results showed: first, comparatively most children were better at applying explicit prosody to resolve most syntactic ambiguity than implicit one, though a positive correlation was found between the two. Secondly, syntactic ambiguity with an attachment structure increased the processing cost for most children, i.e. they spent more time on prosodic decoding, which agreed with Derivational Complexity Hypothesis. Finally, regardless of the syntactic ambiguity types, most children consistently held a bias to decode the pause at earlier-occurring prosodic boundaries under both audio and visual modality, which offered an interesting perspective from Late Closure Strategy.