Why don't people use character-level machine translation?

Jindřich Libovický, Helmut Schmid, Alexander Fraser

We present a literature and empirical survey that critically assesses the state of the art in character-level modeling for machine translation (MT). Despite evidence in the literat…