It has been shown that speakers use prosodic cues to disambiguate the syntactic structure of a sentence and listeners are sensitive to such cues. But the distribution of prosodic boundaries has been reported to depend on the lengths of constituents as well as the syntactic structure of utterances. Hence, it is possible that listeners are sensitive to these alternative reasons (i.e., syntactic or length-related) for why a speaker might introduce a prosodic break (Clifton, Carlson, & Frazier, 2006). The present study of Turkish employs a phoneme restoration paradigm to investigate more closely the time-course of three factors (prosodic cues, syntactic Late Closure, and phrase length effects) in the comprehension of a late/early closure ambiguity. The results confirm a significant role of prosody in restoring missing disambiguating phonemes; listeners tended to maintain an analysis that was syntactically or prosodically favored on-line. Notably, they did not generally revise that decision in face of the rhythmic factor of phrase lengths. This may be because length contrasts become fully apparent only later in a sentence. This is supported by the fact that when tested post-sententially in a previous study of Turkish (Deniz & Fodor, 2017), it was found that phrase lengths as well as prosodic and syntactic effects did influence parsing decisions, indicating that all three sources of guidance were at work. By comparing these two studies, with their different methodologies applied to the same materials, we document here the novel finding that rhythmic phrase length effects are indeed delayed, as previously contemplated by Clifton et al.