MACHINE TRANSLATION & POST-EDITING
Machine translation has been around for over a decade now, but it has only recently reached the necessary quality levels to be a part of the translation process and not a hindrance to it.
Final human-level quality is still several years away, if not decades, but we can now use Neural Machine Translation (NMT) to increase productivity, improve consistency and reduce costs for those projects that require impossibly-fast turnarounds and have different quality requirements.
The fact that we can now train machine translation engines and incorporate client translation memories and glossaries means a higher degree of customization, with a final result that is a vast improvement over the results we were seeing five years ago.
How can we help you?
The first step is to determine if machine translation is the correct option for your material. This is not always the case, since there is content that does not lend itself well to this type of service. It is crucial that this be acknowledged before the text goes through a machine translation engine in order to avoid the painful editing process of a poor-quality output.
The second step is choosing an engine and then incorporating all linguistic requirements, glossaries, style guides, and specific instructions into the engine before the text is processed.
Once our post-editors receive the output generated by the machine translation engine, they will proceed with the review process in order to correct and fix sentences and specific terms that were not properly localized by the engine. This process is called Machine Translation Post-Editing (MTPE) and can be performed by either one or two linguists.
This process can be subdivided into two categories, based on the quality requirement level and the amount of linguists involved:
a) Light MTPE (Machine Translation Post Editing): it produces perfectly understandable text, but without much focus on consistency and style. There is only one linguist involved and the output is mostly used for internal purposes.
b) Full MTPE (Machine Translation Post Editing): the final result is equivalent in terms of quality to text translated and fully edited by human translators. It involves the work of two linguists (translator & editor) and, once completed, the final output is ready for publication.
