Thanks to artificial intelligence, there has been a shift from statistical machine translation to neural machine translation. Statistical machine translation generates translations based on static models. Formal documents that have little to do with how we talk to each other every day.
In the translation world, artificial intelligence using neural network technology is making translations more and more human.
How? By gathering much more informal data through, for example, social media messages with the aim of converting sayings, proverbial expressions, regional terminology, slang and other spoken elements more accurately from one language to another.
Post-editing: the future of translation
Over the years, many technological aspects have gradually changed the way translators work. The emergence of post-editing is probably one of the most significant changes and the future of translation.
The post-editing process consists of reviewing and revising a text that has been translated by advanced translation technology, or machine translation. Post-editing has been used for some time, mainly by large companies in the IT sector.
Post-editing: the best of both worlds
Looking at the history of translation, we can safely say that post-editing will be increasingly used in the future. Mainly in technical and legal fields.
Artificial intelligence and human expertise
Post-editing is a combination of:
- The exceptional speed of machine translations
- Advanced translation machines can produce thousands of words within minutes and convert them into reasonably accurate and usable translations in virtually all languages.
- And the human touch of a professionally trained translator.
The human touch adds the nuances to a translation that a machine does not have. Such as sensitive cultural aspects, word jokes and tone of voice.
What types of text is post-editing suitable for? Post-editing is ideally suited for technical, legal and general business translations, such as:
- blog articles
- product descriptions
- general terms and conditions