The letter M addresses the titanic battle in the translation world between human and machine:
Machine translation (MT) – Ever since Google Translate was launched in 2006, Internet users have become accustomed to being able to translate documents pretty much instantly from one language into another at zero cost. This service has been a boon to millions of people who want to get the general gist of a text. Google Translate and similar systems combine sophisticated software with a huge database of existing translated material (of variable quality), which is searched for similar text strings that can be adapted to fit. Melding technology and human-made translations in this way delivers much better results than older systems, where the machine did all the work.
So should translators start looking for another job? Well, no, actually, because even the smartest machine translation system will struggle to express things in a natural, idiomatic manner and even small errors, such as an incorrect choice of preposition, can significantly distort the meaning of a text. Any document for publication, or where a correct understanding of the content is crucial, will still require human translation skills.
In fact, translators also benefit from MT technology. They can produce a quick-and-dirty draft translation and then use some of the time saved for detailed checking and polishing. This will tend to work best with technical texts, though – experience suggests that using MT on something like a press release offers few benefits because a great deal of rewriting is required to make the text sound snappy and persuasive.
The potential downside for human translators is that clients may one day routinely use machine translation and then just ask a translator to “edit” the output when a better result is needed. The client will think they are saving the translator work – and thereby saving money – while the translator may end up using very little of the machine output and feeling unfairly treated.
To sum up, this isn’t really a battle at all. It’s not about “machine bad, human good” – both have their strengths and their role to play. Choosing the appropriate option, or best combination, is key.