Automatic Rating of Machine Translatability.

We describe a method for automatically rating the machine translatability of a sentence for var- ious machine translation (MT) systems. The method requires that the MT system can bidi- rectionally translate sentences in both source and target languages. However, it does not re- quire reference translations, as is usual for auto- matic MT evaluation. By applying this method to every component of a sentence in a given source language, we can automatically iden- tify the machine-translatable and non-machine- translatable parts of a sentence for a particular MT system. We show that the parts of a sen- tence that are automatically identified as non- machine-translatable provide useful information for paraphrasing or revising the sentence in the source language, thus improving the quality of the final translation.