Translavic cooperates with renowned providers of translation technology and is up to date with the latest innovations in the field of machine translation. You can contact us for machine translations and for the post-editing of machine translations.
Request a free quote
To what extent can translation engines take work off the hands of human translators? At Translavic, we prefer to see people and machines working hand in hand and complementing each other. Our goal is to deliver the high-quality translations that our clients expect from us. Handing over the process completely to machines is not a viable option, but translation technology is making great strides forward and translation machines can be a useful tool in the translation process.
Well-trained linguists are required to build, train and assess translation engines. After all, only professionals who know how to translate in a particular language combination, can assess whether the data used to feed and train the translation engine is of the desired level. And the subsequent assessment of the quality of the translations produced by the engine can only be carried out by specialists.
When to use translation engines
Machine translation can be an interesting option if you:
regularly translate texts in the same language combination(s) in a specific field (e.g. user manuals, technical documentation, web shop texts);
translate large volumes of text within a short period of time;
have a large volume of text where a ‘rough’ translation is sufficient in the first instance, so that you have an idea of the scope and can decide which parts are eligible for a more extensive translation process.
The great advantage of a translation engine is its ability to translate thousands of words in just a few minutes. However, the output in its raw form is not suitable for publication and still requires post-editing by a competent translator. Below, you can read more about how post-editing works.
Machine translation is probably a less appropriate option when converting your content into another language requires a lot of creativity and literal translation is not suitable, as is the case with marketing materials, for example.
Whether your content lends itself to machine translation depends heavily on the language combination, the type of text, the specialism and the time available. After analysing the source texts and running a test project, we can determine together with you as the client whether the use of machine translation is worthwhile, and which translation engine produces the best results. Would you like to know what we can do for your project with help of machine translation? Please feel free to contact us and we will be happy to explore the possibilities together with you.
Neural machine translation
Artificial intelligence (AI) is becoming increasingly common in many aspects of our daily lives. Just think of the virtual assistant on our smartphones, the chatbot that helps us with the order when we visit a web shop, or the recommendations that apps like Spotify and Netflix make based on our preferences for music or films and series.
AI is also used for machine translation. In recent years, neural machine translation (NMT) has taken off. This form uses a neural network in several layers that – rather like the human brain – can recognise patterns, learn and make decisions. The great advantage of NMT over its predecessors, such as rule-based machine translation and statistical machine translation, is that it looks at the sentence as a whole in translation, able to take the context into account. As a result, the translated sentences are smoother, sound more natural and the grammatical structures are more often correct than previously.
This does not mean, however, that NMT output can just be taken for granted. The main pitfall of NMT is that the translated sentences often sound so good that any mistranslations, omissions and unwanted additions that the engines can produce, are easily overlooked. It is therefore important that the machine translation is always checked by a trained post-editor.
With custom NMT, a neural translation engine is trained for a specific field of expertise. Using translation memories, targeted terminology and other relevant multilingual data, the engines can be continuously fed and trained to perform better with specific content. The machine can also learn from the corrections of previous mistakes.
If you would like to know more about the translation technology used by Translavic, please feel free to contact us. We will be happy to discuss the most suitable translation solution for your content.
Request a free quote
Post-editing of machine translations
The output of translation engines must be assessed, and corrected where necessary, by trained post-editors. The desired quality level and the extent to which the post-editor has to intervene in the machine translation are determined together with you as the client. Your target group and the desired style and tone-of-voice can also be taken into account.
If you only want to be able to understand what is in the text via the translation, then perhaps a light post-editing, in which only translation errors are corrected, will suffice. However, it may also be the case that style is important for your product. Then the machine output undergoes a full post-editing, which not only takes into account the correct translation and meaning, but also the complete linguistic and stylistic correctness of the target text. Whatever your wishes, we will provide translations that match your goals.
Post-editors always work in their native language. A text translated from English to French by machine is therefore checked by a French native speaker who has experience in post-editing English-French machine translations.
Are you interested in our machine translation and/or post-editing services? Feel free to contact us or request a free quote.
Request a free quote