We’ve talked a lot about the drawbacks of machine translation (MT) in earlier articles. It is still our opinion that MT will probably never rival the output of professional translators; but there are more and more cases in this digital age, with its explosion of content, where price and speed are more important than having the “best” translation. In these cases, MT, with or without human post-editing, can be a valid choice. MT technology has now developed to such an extent, and has been applied in enough cases, to show that it can be useful for a number of different business purposes.
Factors to consider in deciding about MT
In order to decide whether automated translation makes sense for any particular purpose, there are a variety of factors that you should take into account.
- The volume of material
- The repetitiveness of the material
- The frequency with which the material changes
- The purpose of the translation
- The speed with which the translation needs to be accomplished
- The languages involved
All of these factors should be taken into account in making a decision on whether to use an MT system for your materials. Large volume, repetitive copy, frequent changes, and a requirement for speed are all factors that suggest that MT may be the way to go. Widely used languages are also better candidates for MT since there are more likely to be already well developed translation engines out there. On the other hand, a need for fluency, accuracy and cultural appropriateness argues for professional translation.
Common uses of MT
Two uses of MT are well-established:
- Gisting: getting a general sense of what a document is about
- Large volume technical translation
The second use is often combined with initial authoring of the documents in a form of Simplified English or Controlled Language—sets of authoring rules that make the writing more consistent, less ambiguous, and simpler grammatically and syntactically. These documents are then run through MT and then post-edited by human translators to ensure that the translations are at least comprehensible.
Recent experience indicates that MT may also be beneficial in other circumstances. Our white paper (see link below) provides more examples of when MT may be advisable.