This study examines which factors forecast the synchronization process across speakers in dialogue at the lexical and gestural plane. Cumulative priming explains best alignment at the first level, whereas at the second, this is best accounted for by how chatty speakers are, by whether or not two gestures overlay, and whether the gestures emerge towards conversational conclusion. A comparison with affiliated studies demonstrates that high-level, referential synchronization and low-level, behavioural synchronization appear to be ruled by distinct codes.