4 Misconceptions About Machine Translation

misconceptions.about.machine.translation

As the world is gradually embracing technologies that were seen as questionable in the past, such as AI and machine learning, the technologies themselves are developing and improving each day. One of the areas in which this is noticeable is language and how it is processed. For example, there are still a lot of misconceptions about machine translation and how it actually works.

 Where do misconceptions about machine translation stem from?

For a long time, the language was seen as a holy grail, something innate to humans, which could not be replicated by machines, due to the complexity and constant evolution. In the meantime, machine translation moved from statistical models to neural machine translation engines.

In other words, machine translation engines are not trying to simply compare the content to be translated to a huge library of previous translations, showing the best statistically correct translation. Machine translation engines are now trying to understand the content and the relation between the words.

This has given rise to a renewed interest in machine translation and how it can help automate localization processes, especially for huge amounts of content which need to be readily available to users in near real-time. Of course, as with any new technology, there are bound to be a lot of questions and misconceptions. Let’s try clearing up some of the misconceptions about machine translation today.

#1. Machine translation generally provides bad quality translations

First of all, there are many machine translation engines out there. The most famous one is Google’s machine translation engine, used by Google Translate. Other providers of machine translation solutions include Microsoft, IBM, and DeepL.

translators.dialogue

That being said, different machine translation engines support different language combinations and provide different quality. Google Translate, for example, currently supports the largest combination of language combinations, while DeepL supports a limited number of language combinations. You also need to be aware that translations from English to German will be better than translations from Swahili to Inuit.

#2. You can translate any type of content with machine translation

This is one of the biggest misconceptions about machine translation. Machine translation still cannot translate all types of content equally well.

localization.file.guide.for.beginners

Repetitive content, such as product lists, can be more easily translated than one page of content full of slang and marketing messages.

#3. Machine translation improves instantly

People often confuse machine translation with machine learning. Yes, the two are closely related, but they are definitely not the same thing. Machine translation engines always improve, but you will not see results immediately, especially if you are not teaching the engine how to translate in a more accurate way.

localization.testing

This is why translations from Swahili to Inuit will be less accurate, as there are not enough materials the engines can learn from. Of course, you can always choose to teach the machine translation engine and improve the results. If you have large amounts of content, which has been already translated in the language combinations you want to make use of, we offer the option to train the machine translation engine, which will then produce more accurate results.

This can also be applied to terminology, as you can also have lists of terms that you want the engine to use, reducing the number of corrections you need to make. This brings us to the last point on the list.

#4. Machine translation will take our jobs

This is one of the most popular misconceptions about machine translation. Whether machines will ever be able to produce human conversations, pass the touring test, and translate any type of content without human intervention is a topic for another debate.

automated.translation.forPMs

The current truth is that machines still need our help. For example, the company-specific terminology still needs to be checked and corrected. Translations of slang still need to be improved upon, and there is still no replacement for the human touch and the nuances translators need to implement for different types of target audiences, markets, content, etc.

Machine translation is meant to greatly increase the speed at which humans can work, and as a by-product, reduce the amount of money companies need to spend to localize their content.

Is machine translation for you?

If you want to ask more questions or check if machine translation is the way to go for your content, reach out to us. As always, we are glad to chat, and there are no strings attached.

Gosia
Written By:

Gosia loves copywriting and product translation. Additionally, she's a content marketing and lolcats junkie.

Add a Comment

Your email address will not be published. Required fields are marked *