Eighty-some years ago a Latin teacher by the name of Wren Jones Grinstead (and what a name it is) wrote a piece for the Classical Journal about the use of projects in the beginning Latin class.
By far the most interesting and provocative part of this most interesting article is the suggestion that we introduce our students to two characters whom they should always keep in mind: Romanus (who knows no English) and Barbarus (who knows no Latin). The argument goes that Language exists in a dualistic mode, between the first and second person, one aiming to be understood, the other to understand.
Now one of the chief reasons for the stiff and silly translations too often offered in our classes lies in the fact that the pupil does not visualize Romanus as the speaker of the Latin sentence, and then in turn make himself the utterer of the same thought to the English- speaking Barbarus in his own vernacular. Hence he is merely solving a puzzle, and his only criterion of success is the teacher's authority; whereas it should be found in the socialization of his own imagination. For the perennial query to the teacher, "Is this right?" the pupil should come to ask Barbarus (or Romanus, as the case may be), "Do you get me?"
This is shockingly good, and I plan to introduce these guys immediately.
(Grinstead, Wren Jones. "The Project Method in Beginning Latin." The Classical Journal 16.7 (1921): 388-398.)