Machine translation based on neural networks

Unlike what we see in devices and browsers, a professional translation carried out by automated systems ALWAYS has at least one linguist involved in the process who oversees it.

Machine translation optimises mechanical tasks, but does not replace the work of a translator. It speeds up times and therefore saves costs.

However, not all texts and not all languages can be machine translated if we are to ensure that the message of a highly complex document is correctly conveyed in the target language. For this reason, we always recommend that the translation be carried out by a qualified translator from the very start.

iso9001

Did you know that?

CLINTER is one of the very few companies in Spain (and also one of the first) to obtain UNE-ISO 18587:2017 certification, which ensures high-quality work with machine translation tools.

What are the advantages of machine translation?

Machine Translation (MT) or NMT (Neural Machine Translation) is a translation tool based on neural networks that helps the professionals involved and benefits the client.

While it is not AI, AI does work in a similar way

Machine translation analyses textual elements and detects how they relate to each other, not only at word level, but also at sentence and context level. It uses technology based on neural networks that can calculate the probability of a text being translated in a specific way, depending on the memory or database that feeds it.

Provides consistency and avoids errors

When working with a translation memory and a glossary adapted to your needs, MT can ensure that the text to be translated is always handled in the same way and not differently each time. This is especially useful for technical texts where the message must be accurately conveyed throughout the entire document.

Saves time, reduces costs

While the final translation will be always reviewed by several professional translators, including machine translation as an initial step speeds up the linguist’s work. This means lower costs for the end customer.

The more accurate the initial machine translation, the less time a professional will need to revise and adapt the text (post-editing).

Machine Translation is not for all texts

There are limitations to using machine translation tools that have to do with the languages involved, the complexity of the document, the type of language, and the purpose of the text. European languages work well with automated translation engines, especially when it comes to technical texts in any field: engineering, medical, legal, etc. There is greater scope for precision with these text types.

However, in marketing or literary texts where the essence of the message matters more than how it is written, or in languages where nuance and idiomatic expression have a stronger influence, it may not be worth adopting machine translation processes.

What’s a machine translation process like?

Would you like to know how this state-of-the-art tool is used? We explain how:

The first step is to check the text to be translated to make sure that machine translation can be used. The client’s requirements, the source and target languages, and the text type, are checked: this includes terminology, scope and wording.

When possible, the text will be pre-translated and prepared for the “human” part of the translation process.

A professional linguist specialising in the subject matter will revise (post-edit) the result of the machine translation to ensure that the text is coherent, free of errors, and adapted to the context for which it is requested, and complies with the UNE-ISO 18587:2017 standard

Get in touch with CLINTER for your translations and make the most of the latest technologies in professional translation processes.