This paper proposes a novel approach to measure disclosure in patent applications using algorithms from computational linguistics. Borrowing methods from the literature on second language acquisition, we analyze core linguistic features of 40,949 U.S. applications in three patent categories...