We present a novel parser combination scheme that works by reparsing input sentences once they have already been parsed by several different parsers. We apply this idea to dependency and constituent parsing, generating results that surpass state-of-the-art accuracy levels for individual parsers.