Computer Aided Translation vs Machine Translation (TM vs MT)

We live in a world where the technology is changing at a very high speed. This also affects the translation and localization industry. However, this technology is often misunderstood by people outside our industry.

When we talk to our clients and friends about translation, they usually have only one thing in mind: Google Translate. Then we explain to them that translation is a human process and that even with Google tools, there are still people who make a living of translation. Then, we explain the advantages of using CAT (Computer Aided Translation) tools to work more efficiently, and again they tell us, “Ah, so you just prepare the file and then upload it to Google Translate, right?”. Then, you take a deep breath and explain the differences between a CAT tool and a Machine Translation engine. Here you have a simple explanation:

CAT tools help translators to make a consistent use of terminology. All translations are human-made and are stored in a repository called a Translation Memory (TM). Especially for localization and technical translations, when an update or a new version of a product is released, the TM is used to make the most of the old translations, so that the terminology is consistent in all versions of such a product. The terminology, however, can be changed due to marketing decisions or other reasons, so the TM should be updated to still make the most of it.

Machine Translation, however, in the first stage makes use of the human-made translations. These translations are used to train the MT engine and when a new text is to be translated, the MT engine translates the text either by using the translations that are more frequent (statistical approach) or applying grammar rules (rule-based approach). In both cases, the MT engine is the one doing the translation, taking bits and pieces from here and there and turning it into a sentence which should make sense. Note that we say “should make sense” because unfortunately there is still a lot to do in some languages before MT can be used for everything.

human translation versus machine translationWhen should you use one or the other? Well, it really depends on the purpose of your translation.

We suggest you to use human translation and CAT tools when you need a high-quality translation that goes beyond the language and takes into account the culture of the target country. Your audience will feel identified with your products and company philosophy. Human translation is especially useful in marketing texts and high-sensitive fields like medicine, where even a spelling error can lead to fatal consequences. Sometimes, for creative pieces of marketing, it is even better not to use any tools, as the same word can have different translations within the same text.

However, if you just want to give your audience an overall idea of what you mean, or if your document is highly technical, then machine translation could be a solution for a quick job. However, even if you use MT, you should still review and edit the job; there probably will be things to fix so that the MT engine is improved and get better translations next time. This process is known as Post-Editing.

At Jensen Localization we can help you with both human and MT translation. We can build an MT engine for you or we can Post-Edit your machine translated text, once we check the quality of the raw output and we agree on the level of quality you desire (sometimes it is so poor that it is faster to translate it again from scratch). Fell free to contact us if you need further information!


Alicia González is a translator specialized in software localization. After working for around 10 years as a translator, reviewer and Project Manager, she left the production side to take the role of Business Development Manager at Jensen Localization, where she advises companies on how to increase their global presence thanks to the use of foreign languages. She writes on this blog about the importance of translation and localization in international business, procedures and tools. She is also working on new Machine Translation developments.