Data mining and first language acquisition: a case study on a French corpus