One question that always comes up regarding machine translation is quality. In particular, people always seem to be asking: which is better, human or machine translation?
In terms of quality, human translation remains the winner, of course. Machine translation has yet to reach human translators’ creative flair, and their critical eye for linguistic precision.
But they also come with certain challenges—time, effort, and cost. This limits the kind of applications that it could be put to, especially in a fast-paced business setting.
This is where machine translation steps in. Machine translation is a quick, scalable solution that works perfectly for high-volume and time-dependent tasks.
The quality of MT has improved significantly over the past few years, to the point that it has reshaped the whole industry. There are many new applications and use cases for MT today that were unthinkable just over half a decade ago. And with the help of post-editing, those use cases have only continued to grow.
In general, machine translation can fulfill three different purposes, each with different expectations regarding quality:
We’ll go over each one in more detail below, with some insight on how much post-editing may be needed in each case.
If you don’t know what machine translation post-editing is, you can check out our guide to machine translation post-editing here: What is MTPE?: An Introductory Guide For Businesses
In the context of assimilation, the purpose of MT is simply to help the user understand a text in a foreign language. Often there may be text you want to access that isn’t available in your language. This can include news articles, lecture or podcast transcripts, or blog posts, among other things.
MT is useful here as it can help you get a quick look and general understanding of what the text is saying.
In this case, the quality of the output text doesn’t need to be highly polished, just of adequate quality that the gist of the text can be understood. Depending on the specific purpose, light MTPE or even raw machine translation may be enough.
For dissemination purposes, the output text may need to be of a higher quality. Dissemination refers to when the goal is to produce text in a foreign language. In this case, MT works best for documents that are of a short-term or transitory nature, such as newsletters or memorandums, that need to be translated into multiple languages. Most of these will only require light post-editing.
If your documents are of a more complex or permanent nature, then you might want to consider full post-editing instead.
Machine translation can even, in a limited capacity, fill in for the role of interpreters, thanks to the advent of speech-to-text technology. If two people or a small group of people want to communicate with each other but don’t have someone to interpret for them, there are various apps they can use right on their phones to communicate with each other. Google Translate and Microsoft Translator both have speech-to-text options, and Baidu Translate is available for the Chinese market.
However, this solution works best in more casual settings, where people can take the time to clarify things with each other when the MT isn’t clear enough. For a business setting, having a professional do the work of interpreting is still the best option.
Machine translation still has some way to go before it can truly compete with human translators in terms of quality. But that doesn’t mean it doesn’t have its uses, and as you can see there are many applications for MT that suit different purposes. It’s all about knowing what your purpose is, and knowing which machine translation solution is the best fit for it.
© Copyright 2023 Tomedes All Rights Reserved.