Do Neural Language Models Show Preferences for Syntactic Formalisms?

ACL 2020