The truth is that sometimes, a project does not require the use of MT to be completed quickly and accurately. Other times, using MT is the only way to finish translating content by the established deadline. The best way to proceed is to have a set of criteria and a clear goal to help you answer the question: to MT or not to MT? Ultimately, the decision to use AI translation comes down to you.
If a project has vast amounts of content, using MT is likely the only way the project can be completed on time and on budget. Consider this: the average professional native translator can translate 2,000 words per day per language. How long do you think it will take a team of translators to complete a project that has millions of words? A very long time is the answer!
If you have to invest money in content that is not your main product/app/website, but more like your user documentation, technical manuals, FAQ, or HELP systems, then you should consider a long-term MT strategy with your preferred translation partner.
If the content you have to translate is going to be "consumed" fast, for instance, customer feedback, emails, consumer-generated reviews, etc., then you might consider using MT as the translations won't be "on the shelf" for too long. You do not need top quality for something that will be gone soon or will only be read once.
If you have a limited budget, or worse still, if you do not have a budget at all, then MT might be the only solution available to you. Think about translation as an investment that will lead to sales and generate revenue for your company.
MT significantly streamlines the translation process because the translation team begins to translate text that has been MT'ed already; they are not staring at a blank screen. This makes them faster as they have to post-edit the translated content, tweak it, and change it to make it more understandable to the intended target audience. This allows for large projects to run faster. It is often the only way to complete vast amounts of text by a strict deadline.
Despite huge improvements and the de facto use of neural machine translation, non-technical content tends to have nuances and can be loaded with meaning that a machine simply won’t understand. So, post-editing machine-translated content is always advisable. If you aim to get good quality (near human-like), then deploying MT and post-editing is undoubtedly the best option.
When it comes to machine translation, not all languages are equal. Some languages feature complex structures and many nuances in their vocabulary that will not always be captured by a machine translation tool. Therefore, some languages will require more post-editing effort (and more time and money) than others. Romance languages are the best performers. Scandinavian and Germanic languages will require some post-editing. And Slavic and Asian languages will require much more post-editing effort.
Ready to Try Machine Translation? Download our free eBook, Machine Translation and Post-Editing: Six Steps to Success.