Ever since machine translation entered the scene, the translation process has become easier than ever. Machine translation allows you to enjoy high-quality translations at an affordable price while also reducing turnover time.
However, a strange question keeps popping up with regards to machine translation; the question of post-editing. Is post-editing necessary even after a project has been machine-translated?
Well, let’s take a look. But, before we really go into the topic, let’s try and understand what post-editing actually is and what purpose it serves.
Key Terms Explained
Post-Editing, as you can probably guess from the term, refers to the editing of content after it has already been machine-translated. The purpose is quite obvious – ensuring that the translation is up to the agreed-upon standards. Post-editing is different from regular editing in that it specifically refers to the editing of machine translated content. Editing, on the other hand, refers to fixing errors in human-translated content.
We also have something called “light post-editing”, which is basically the utilization of basic human skills to edit machine translated content. The focus of light post-editing is to make the content more understandable. On the other hand, complete post-editing refers to a more complex form of editing where the machine-translated content is edited to, not only become more understandable but also be stylistically accurate and industry-appropriate.
The next key term we’re going to talk about is “Raw Output”. Raw Output is nothing but the translated content that comes out of a machine translation system. There are, generally, two components to raw output. Firstly, you have segments that have been translated by leveraging a translation memory or TM and raw translations done by the machine translation system.
Of course, there are exceptions – if the translation has been carried out by a non-customized or commercial machine translation unit, the raw output as a whole is likely to be made up of bilingual data.
The point here is that the raw output is usually a mix of literal translations from the machine translation unit and accurate translations from the TM system. The mix-up can lead to a lot of common idioms, domain-specific terms, and colloquialisms to get lost. This is exactly why post-editing after machine translation is very necessary.
Case in Point
For instance, let’s look at the term “howdy!”. This is a term that is used in a very specific context. However, when you machine-translate content that uses such terms, the actual application of the term is sure to go wrong and even alter the meaning of the words around that term.
Machine translation is a literal kind of translation and it does not factor in sentiment or context.
This is where human-linguists are still very much needed. You can create all the rules and glossaries you want, but some things are just better done when humans are involved. Humans understand sentiment and context. They get the sub-text.
So, it’s naturally better to have a human post-editor making sure that your content is accurate and meaningful at the end of the day. That way, the finer nuances of the content you’re translating can be carried forward.
Apart from the fact that post-editing ensures accurate translations, it also helps streamline future machine translation projects. For example, we can learn about the more common errors that are likely to show up.
This will ensure that the system is in a constant state of improvement, gradually reducing the error rate for future projects
In conclusion, it’s now quite obvious that the value of post-editing is very high when it comes to machine translation. You simply cannot replace human knowledge with machine knowledge, at least for now.
Ready to go global? Sign up for a 2 week trial here and we’ll reach out to help.