Machine translations… will human translators become extinct?
There seems to be a bit of a commotion in the world of translating. Big investors and tech companies are storming the translations services market. There is an increase in machine translations and more and more agencies are switching to PEMT (post-editing machine translations). Some translators are really enthusiastic, others shudder with dread at the picture of such a future. So what’s the best way to go when as a company, you want to have your website, specifications or other kind of text translated?
Translation systems by companies such as Google, DeepL, Systran and Bing have continued to develop in recent years. The tools are becoming smarter, faster and more skilful. Not only can a translation be completed at a breathtaking pace, it’s also a relatively cheap solution. So it’s a smart, cheap, fast and progressive solution! But how do the advantages of machine translations compare to those of human translations? Here are the facts.
How usable is a tool such as Google Translate?
Machine translations are particularly effective for some uses. For example, a translation app makes it very easy to wish someone a happy birthday when you’re abroad. During a holiday, it’s an easy way of looking up the translation of a dish on the menu. These kinds of applications can also be helpful when you receive an email in a foreign language, for example. It’s handy to translate the text with an app or translation programme to give you the gist of the message. Automatic – computer-generated – translations can certainly be a fast and effective aid to communication. They’re extremely suitable for brief texts which don’t have to be published or distributed. A machine translation offers another advantage in addition to speed: it’s a cheap way of translating text.
Google Translate is the best-known example of a translating machine. There’s no-one who hasn’t ever googled the translation of a word or sentence. The German DeepL, a counterpart of Google Translate, claims to be more accurate because of improved deep-learning algorithms and a source database from the Linguee search service. The fact that DeepL performs better is said to be due to its experience with Linguee, DeepL’s other service. Linguee started up in 2009. It enables users to search for bilingual texts.
The digital language/translation revolution doesn’t stop there: an American tech company has developed a real-time interpreter – an app on your phone directly translates the spoken word into your native language. Skype has also introduced a live translation application which allows people to communicate without any problems even though they don’t speak each other’s language. On YouTube, you can have the voice in a video translated into subtitles.
Did you know that the translation services market is estimated to be worth 43 billion euros? Google, Apple, Facebook, Salesforce, Microsoft and Amazon are all investing heavily in translation technology.
The latest development in machine translation is neural machine translation. Where neural translating systems differ from static machine translations is that they use artificial neural networks. These networks imitate the workings of our brains and are able to see connections between works which are not adjacent to each other in a sentence. All the big players in the market (Google, Bing, DeepL, Facebook, Systran etc.) are working on the development of these translating systems. The technology is evolving at a breathtaking speed. Translators also make use of automatic translation suggestions and artificial intelligence to raise their productivity.
Despite all the technological developments, it’s true to say that a translation made by a machine is almost never of the very highest quality. It’s also true that the majority of clients really doesn’t find ‘not quite right’ quite right. They want texts which are 100% right.
Automatically translated texts often look a bit contrived, strange or unclear. They frequently require correction and refining. They may differ per sentence in tone of voice, spelling preferences or forms of address. Such translations certainly don’t amount to good calling cards for your own organisation. There is a quick way of checking whether a machine translation is suitable for your requirements. Once you have entered your text and translated it with the aid of the machine, have the translation translated back into the original language. When there are discrepancies between the two, it will be easier to see that not all texts are suitable for machine translation or that they result in ‘false friends’.
What does a machine translation give you?
Should you choose a machine translation, a human translation or a combination of both? This depends entirely on the demands you make of your text. Language is filled with ambiguities, word play, hidden meanings and other pitfalls which machines are unable to read or understand. In the case of longer texts, nuance and context are crucial for a qualitatively good and reliable translation into another language. A machine translation can be efficient but it will always lack linguistic accuracy and authenticity and the ability to read between the lines. And these are exactly the elements that a human translator recognises, understands and applies. The machine is skilled in displaying words in another language but is, up till now, incapable of detecting things like irony and humour. If you, as a company, have done your utmost to make a beautiful website with powerful, inviting texts, obviously you want to see that in the translation too. Marketing or advertising texts often contain plays on words, highlight a cultural aspect or have a specific focus. The point will be missed if you don’t get the translation right.
A translation computer will always choose the most obvious words and it lacks subtlety. Machines will not recognise the fact that certain words can add to the power of a text and reflect an identity, brand or organisation. The result is that those words are replaced by more generic examples which don’t necessarily embody the message either correctly or completely.
The translator as editor: post-editing
Machine translations offer possible ways of supporting human translators. By means of post-editing, the translator acts as an editor for the translating machine. The translator checks the translations and makes changes where the translating machine has gone wrong. That sounds ideal! However, in practice it provides little profit in terms of time and money. It frequently takes a human translator just as long to optimise a machine translation as it would have done to translate the text themselves.
In terms of consistency in texts too, it’s wise to use a human translator. When a translator works with a CAT tool (software which allows you to save your own translations efficiently) they accumulate a translation memory for each specific type of text. In other words: they create translation memories which remember how the human translator translated a word or phrase previously. Using translation memories, translators can consistently use client-specific professional jargon and apply any preferences in choice of words or terminology.
Localisation and SEO also play a part
A good and accurate translation is a crucial element in the business operations of companies operating internationally. Since every language contains culture-specific aspects, localisation of translated texts is essential. Localising a text involves translating content from the source text in such a way that the culture-specific translation aspect is not lost. Native speakers have the most extensive understanding of a local language. Localising a text means adapting it to the local language, from the point of view of a culturally sensitive and skilled native speaker. Translating machines do not have a function which has both the congenital and the learned fluency of a language. Culture is preserved in language, something a tool such as Google Translate is simply incapable of understanding.
Localisation is an essential concept for every successful company when creating multilingual content. A modern marketing strategy offers the target group completely customised content. Clients tend to choose products and services which are adapted to the client’s location and culture. So it’s really important to adapt to your target group in order to get the best possible results from your practice. For examples of this, read our article on cultural differences between the Netherlands and Germany, which gives tips on how you can put this to good use when doing business.
In the case of online texts, human translation offers yet another advantage over a machine translation: translating machines lack the SEO skills for the source text. It’s inadvisable to use machine translation when international SEO tactics need to be applied.
If a text is aimed at publication, we currently believe that engaging a human translator (whether to translate or to post-edit) is the only way to be absolutely sure that your text comes across to your target group exactly as you intended. Good translators – with their own specialisms such as medical, legal or technical texts – are scarce.
Only a human translator offers a reliable and high-quality result if the character of your text is to be conveyed in full. And you can be sure that you’re getting your message translated meticulously, in a fitting context and with the correct terminology.
Chuck a sausage in the frying pan or go out for a romantic dinner?
Some of the larger providers in the translation services market compete fiercely on price. As with other products, that can be a reason to choose those providers. And this is often totally dependent on the objective. Sometimes you chuck a sausage in the frying pan, other times you may go to the fish and chip shop and if you really want to make something special of it, you’ll go out to dinner at an excellent restaurant, specialised in ‘haute cuisine’. In other words, sometimes you put a text into a translation app, other times you choose a cheap option and if it’s really important, you go to a specialist. It all depends on what you want to achieve with your texts at that moment.
Exito works with an eye for detail, supports fair rates and is consequently able to continue to provide its customary quality. We have at our disposal a carefully selected network of native speakers, each of whom has their own expertise in a variety of specialist professional fields. That way, we always find the right translator, copywriter or editor for each project and we’re able to offer the best quality within the deadlines set.
Do you need a reliable translation which is ready for your immediate use? Please don’t hesitate to get in touch with us!