It is known that a phrase may have multiple meanings. Phrases such as “green tea cup” may be interpreted with two different meanings—a “green-colored tea cup” or a “cup of green tea.” Then how people know the exact meanings of apparently syntactically ambiguous linguistic expressions? We propose that gesture that accompanies speech may help disambiguate syntactically ambiguous structures. The present study investigated whether the difference in phrase structures influences the production of gestures. Participants produced gestures as they produced a Japanese four-word phrases. We examined all possible synchronization patterns of speech and gestures. We found, for the first time, gestures tended to synchronize with the chunks of words that form a constituent in syntactic structures. Our study suggests that gestures may play an important role in disambiguating syntactically ambiguous phrases. This could be a reason why humans have continuously used gestures even after they acquired a powerful tool of language and why today, they still produce language-redundant gestures.