With the improvements in artificial intelligence, machine translation is constantly developing and improving. So then, to what extent can we use this tool? When does the intervention of a human translator become crucial? In this article you will find out everything you should know about machine translation, how it works, and when it is acceptable to use it.

To begin with, machine translation is the process of translating a document from one language to another by putting it through an machine translation software.

How does machine translation software work?

In order to understand the possibilities of machine translation, you must first be aware of the different types of machine translation that exist. We have listed 3 types below:

RBMT – Rule-based Machine Translation

This consists of a combination of grammar rules, and words from the dictionary. In terms of a specific field, it will use a specialised dictionary.

SMT – Statistical Machine Translation

This type of translation does not have knowledge of grammar rules. These statistical programmes “learn” to translate by analysing large quantities of text for each translation (source and target language). They can be used for specific fields by integrating all the specialised terms in the given field.

NMT Neural Machine Translation

This approach uses a network of artificial neurons to predict the word sequence probability by model sentences.

When should you use machine translation?

It is an option when there are no specific style requirements, when you need a translation to be performed in a “mechanical” manner. To determine whether you should use it, you should consider the following factors:

Is the word count of the document for translation important?

With thousands of words to translate quickly and the additional specialised field, machine translation can be better for a translator.

Is the document written in a brief manner?

If there are specific technological terms in the source text, without a specified style, then using machine translation is much more feasible. The software will recognise if the document contains field-specific vocabulary with less risk of confusing the terms.

Who or what is the document for?

If you are translating a document for your employees, in other words, a low-stakes situation, then machine translation is appropriate.

The main rule for choosing machine translation or not is the value of the source text. If it is important, it is better to have it translated by a professional translator.

Here are some examples of documents that are suitable for machine translation:

  • Product descriptions,
  • brief technical documents
  • FAQs
  • Regulations, which are available on every website and every type of document, but almost nobody read them.
  • Download our Guide: Translating your documents

What are the limitations of machine translation?

For now, a professional translator is preferable to even the most advanced machine translation software. When style plays an important role in the document reception, in terms of convincing, persuading, or reassuring, you should have it translated by a specialised translator in the field.

Machine translation can be used for technical instructions. However, these translations should always be proofread in order to minimise errors.

Furthermore, you should always keep in mind that machine translation does not always mean guaranteed quality in every language. For example, machine translation works very well for translation from French to English, but it does not work as well when translating French to German.

And now for the most important rule. Forget about Google Translate, this translation tool produces texts of random qualities and uses previous translations as reference. You cannot be guaranteed confidentiality with this machine translation tool.

The future of machine translation

In 1960, the beginning of machine translation, having access to a machine cost more than having a professional translator work on your document. Nowadays, many translation agencies use the tool. They may use it for specific parts of a translation or even for the entire document.

The rise of neural machine translation developed by Google and Microsoft, is full of promise for the future. This technology can understand the meaning of a word in a specific professional context.

However, this technology still has its disadvantages. They cannot effectively deal with linguistic combinations yet. In the coming years, the process will perform better, with even more languages available.

Here at ACSTraduction, we work in collaboration with our clients to find the best translation option for them. We also deal with all of the existing translation solutions.

We choose the best translation option for our clients in accordance with their objectives and their needs, as well as their budgets.