Most people have probably tried using Google Translate to translate an individual word or a short sentence while on holiday. In a professional context, however, machine translation may not always be sufficient. For this reason, it is essential to consider whether machine translation or human translation is the best fit for a given translation task.
The solid approach to translation
Human translation can be viewed as a traditional approach to translating texts from one language to another. The process is time-consuming and not without costs, but you can be sure that the best linguistic solutions will have been applied. A human translation picks up on the finesses in the source text, and carries them over to the target language – ensuring that the translation of your original text will hit the mark with your target group and convey the intended message in the target language!
Human translations are crafted by skilled professionals who translate into their native language. As a result, you can be sure that any linguistic niceties have been translated correctly, and that the translated text clearly conveys the intended message. By choosing human translation, you can be confident that the language in the target text will be fluent and natural, and that specific terms and phrases have been translated correctly. As such, the tone of the original text will have been retained and the cultural creativity will not have been lost through simplification.
What is machine translation?
Machine translation is an automated process where translation software translates a given text from one language to another. Thoroughly developed technology helps deliver the translation almost instantaneously, and with a high degree of precision. This type of translation is a smart solution if you need to understand or communicate a text that need not be absolutely perfect in all regards. However, it is important to be aware that machine translations cannot take into account cultural finesses and linguistic contexts to the same level as a human translation.
The development of machine translation stretches back many years, and the technology involved continues to open up new opportunities. The famous Georgetown IBM experiment was conducted in 1954, when it proved possible to use a computer to translate more than 60 Russian sentences into English. However, the process was considered too complex, as the machines available at that time were unable to contain the huge volumes of data required.
Several decades later, in the early noughties, the requisite software and hardware had been developed sufficiently to complete simple translation tasks successfully. There are several forms of machine translation available today, but neural machine translation is probably the most widespread.
Types of machine translation:
- Rule-based machine translation
This is the oldest form of machine translation, where predefined linguistic rules assist the machine in translating text from one language to another.
Rule-based machine translation is seldom used today. -
Statistical machine translation
This method draws on analysis of words and sentences to create statistical models which can generate translations. Statistical machine translation is based on existing human translations, which it uses to “learn” how to translate texts. -
Neural machine translation
Neural machine translation utilises artificial intelligence to deliver a fast and accurate translation. It is far and away the most commonly used form of machine translation today.
Which approach should you choose?
Are you unsure of which form of translation to choose?
There are several aspects to consider, but a good place to start is to think about what type of text you need to have translated. Professional documents and specific trade texts should be handled by professional translators, while more standardised texts may be better suited to machine translation, where the finished text can be delivered quickly.
Read our blog post about when it is best to use machine translation here.
Combine machine translation with human post-editing!
If you’re considering using machine translation on a text, it’s a question of finding the right balance. That is precisely why World Translation gives you the option of combining machine translation with post-editing by our skilled, professional staff.
Raw machine translation
Raw machine translation means that no editing or proof-reading is performed on the translated text. The text is delivered as generated, and no consideration is given to possible differences in cultural interpretation or semantics. This form of translation is suitable for internal documents, emails and brief summaries, for example.
Light post-editing
A lightly post-edited text has been reviewed by a professional, whose task is to correct errors in content. This results in a text that is grammatically correct, and which conveys the intended message.
Full post-editing
Full post-editing is the best option if you need a fluent, professional text. Here, a human translator checks every aspect of the translated text – including grammar, style and semantics. This approach to machine translation provides the highest quality.
It is clear that there are advantages to machine translation, but also that it should only be used for specific translation assignments. This means that it is always a good idea to talk to your translation agency before making your final decision. Your agency can advise you and answer any questions you may have, guiding you to choose the right service.
