Machine Translation Post Editing

When translators started using personal computers as their workstations, the spread of computer software tools like Trados, SDL, Wordfast and others spread and begun increasing their productivity. This also created criticism from the hard to change line of thought that suggested translators should not use productivity tools as it “decreased” creativity and discouraged learning through research.

Today, we as translators have many resources that can help complete jobs faster, with more accuracy, consistently and also maybe a little bit more monotonously. Until… Machine translation (MT)… or better said Neural Machine Translation (NMT).

There is not much of a difference between the line of thought that said using Trados and other productivity tools was our doom and what is happening with the recent spotlight on Neural Machine Translation.

Machines can translate a vast amount of text relatively quickly and cheaply. However, can we actually use the output from machine translation?

Well, yes… it’s OK if you are translating a blog or an article and just want to get a sense of the content, but with material that can potentially be risky, the odds are that you would want a professional translator involved. If you want to keep in touch with your Facebook friend from Japan… machine translation works. Do you want to fully understand a hospital informed consent form? Machine translation won’t be enough.

Nevertheless, the examples above might be a bit black and white and MT is all about shades of gray. We believe that it can be used as a potential productivity tool in certain instances.

Machine translation might come as a solution to those businesses who are looking to increase productivity in their content translation, as it allows doing it in more languages, more quickly and less expensively.  However, the process of machine translation does not end by using only an engine tool with artificial intelligence. The translation will always be imperfect, that’s why you need to rely on a professional translator to go over the contents.

So, which requirements should the post editor meet? It will vary from one project to another, but the editor has to be able to analyze the text as a whole. Does not need to change each section of text, but rather quickly decide what needs to be readjusted. Sometimes, terminology errors need to be corrected, others it might be the punctuation and format. Take notice that machine translation is carried out by a machine that might not understand the tone of the text. That is why the post editor needs to check if the final text can easily be read and if it has a fluid and natural tone.

Some tips:

  • Don’t make too many corrections.
    • Have a clear idea of what is expected of you.
    • Don’t lose too much time to thought. Do you have a doubt on a particular segment? Delete it and retranslate it from scratch!
    • Use all the tools available: spell-checkers, glossaries, translation memories, and QA tools.
  • Always work towards the quality the client is expecting.
    • Don’t get discouraged. Practice will make you a real post-editing pro!
Machine Translation Post Editing

Not Found

Apologies, but no results were found for the requested archive. Perhaps searching will help find a related post.