On the referential/attributive distinction