If you are not in the translation industry, these two terms are usually quite confusing, so let us give you a quick explanation that we hope you find useful when talking to your translation suppliers.
Since translation is a repetitive task, it can benefit from certain level of automation. And this is where different translation tools come into play. Your translation partner could tell you about the use of CAT or MT systems, but what does this mean?
The main difference lies in the fact that with a CAT tool, a professional translator is using a translation tool or app in order to speed up the translation process, whereas MT means that the translation is fully performed by a machine.
The difference is clear so now let’s analyze in more detail. `
Computer-assisted translation (CAT)
When using a CAT tool the source text is translated by a human translator using functions that can improve the speed and accuracy of the translation in the process. Computer Assisted Translation tools rely on translation memories, databases of translations that provide translators with suggestions and corrections based on translations previously done for that same client, area or field. They also include terminology management software and quality checkers. Essentially, CAT tools help translators work faster, consistently and with more precision. CAT tools are really useful and helpful when working on specialized areas or industries with very specific terminology, like legal or medical, but the tasks must be performed by professional and specialized native translators who master the use of the CAT tools, the source and target languages and the special subject terminology. CAT tools are not for everyone.
Regarding availability of these tools, there is a wide variety out there. Some can be found online for free or for a small fee but the most advanced and widely used ones are most costly. No matter the CAT tool used, the most important concept to remember is that the quality of translation depends on the human translator.
Machine translation (MT)
Machine Translation is a different story. When using it, texts are completely done by machines. No human translators are involved in the process.
There are two different distinctions in modern MT: statistical and neural based engines.
Statistic Based engines learn through statistical analysis of a bilingual text (corpora), a source text of translated material and a target text of untranslated material, generally provided by the developer or user, that is then used to automatically deduce a statistical model of translation.
Neural Based engines are the most modern approach to MT. Neural networks are designed to mimic how the human mind learns. These engines seek to understand the context of what is being translated to properly predict the correct word choice.
Since humans are not part of the translation process, the accuracy of the translated text is not perfect and human translators are hired for the post-editing phase. The results also vary depending on the language combination mainly because of the amount of data available to feed and train engines.
The advantages of MT include cost and speed. Machines work almost instantly, and many engines can be used for free. On the other hand, and in spite the great improvements made in the last few years, the accuracy of machine translation is still limited so it is not the best option for many source texts. We would say that MT is the perfect solution for high volume of generic text for internal use that require a quick delivery. It can be used for other contents but always polished by a human post-editor.
Conclusion
Now that you know the difference, you are ready to discuss, or at least to understand, the solution proposed by your translation partner for your localization needs. Just keep in mind the volume and nature of your text, the languages involved, the target audience and where is the content going, how quickly you need the text back, if there are any legal restrictions applied to this content and the available budget.
All options are out there and available for you. You just need to choose a translation system and a workflow that aligns with your needs, purposes and available resources.