Dynamic definite descriptions, implicit arguments and familiarity