Neural machine translation (or NMT) has revolutionized translation. While this new system can assuredly save you money, it has to be used in the appropriate manner to ensure the quality of the final product. Today, we are going to go over how NMT post-editing works and why it is worth it.
The quality of the output of machine translation depends largely on the quality of the source text and the language pair. The more straight forward a text is, the better the raw machine translation will be. Likewise, if the source and target languages are linguistically similar, such as two romance or Germanic languages, the raw translation will be better than that of two extremely different languages. Due to these limitations, raw neural machine translation isn’t ideal for anything beyond “gisting,” i.e. trying to understand the basic meaning of a text.
Since machine translation isn’t perfect, LGO offers neural machine translation post-editing. After a text has been machine translated, experienced translators and linguists proofread and correct the raw translation. This is a relatively new process, so we are constantly training and supervising new specialists.
Though there are levels of machine translation post-editing, for materials that are published or widely distributed, we suggest heavy editing. Such materials include technical documents, annual reports, contracts, legal documents, and more. Heavy editing ensures that less obvious errors, like ambiguity and voice, are fixed and that your translated document reads as if a human translated it.
If you think that your project could benefit from machine translation post-editing, don’t hesitate to contact us. Our team can get you a quote within one hour!
