IS DESCRIBING LANGUAGE MERE BUTTERFLY COLLECTION? ON EPISTEMOLOGY, STATISTICAL LANGUAGE MODELS, AND CORPUS

This work focuses on the philosophical tension between conventionalists and rationalists, present in the history of science from pre-Socratic period to contemporary Linguistics. R also takes into account the relevance of statistical models rather than considering language phenomenon as an elegant representation of the expression of thought. For that purpose, we take some conventionalists' accounts of language throughout philosophy history and we also analyze computer scientist Peter Norvig's recent essay on whether language science should really take the path linguist Noam Chomsky considers as the only one to a scientific approach. Therefore, in this work, we attempt to take empirical facts and, therefore, pure description, to portray linguistic phenomenon as an evidence of a conventional use rather than a rational creative human impetus. Finally, we also claim that Corpus Linguistics grounds Norvig's arguments on language science.