Translating books into other languages stretches as far back as written history goes. With the Rosetta Stone, we unlocked the secrets of Ancient Egyptian heiroglyphics. We’ve moved philosophical foundations of thinking from Latin and Greek bases into thousands of other languages, and books like the Bible have been translated into every possible tongue in the world.

Yet, all of these changes had one thing in common until recently: they were all adapted by humans who spoke more than one language. This took time, energy, and resources and, of course, dealt with the fact that human error could come into play in the translations.

Now, computers use machine translations to transfer text between languages almost instantly. Over the past decade, this facet of computer science has become an invaluable tool to millions of people as they visit other countries, communicate with foreign language speakers online, try to learn new languages, and, in the case of research, attempt to change their manuscript to that of the preferred journal publishing format.

While it’s been arguably invaluable, the reality is that sometimes, machine translations don’t consider things like style, dialect, jargon, and other language nuances. The result of an incorrect machine translation can negatively impact your journal article.

How Machine Translation Works For and Against You

Machine translation, by definition, is when content is automatically translated from one language into another. The language starts at a source and moves to a target language without human assistance. 

Computers have had the translation function since their initial application in the early 1950s. Since computers required so much processing power to function on a basic level, there was no way to house and store something as complex as translating languages until recently. When basic machine translation became part of everyday software and hardware, developers finally had the chance to do what they knew was possible: teach computers how to do machine translation.

When Google’s team of experts turned their focus to artificial intelligence (AI), and how it could be used to learn more about the neural system, an inadvertent bonus happened. They learned that their smaller designed machine translation engine could compete with larger computers and was continually improving the language quality of translations based on response and input.

The use of neural machine translation became Google’s pet model, and other development companies began copying the output. But, like any computer system, there is still a margin of error. What computers don’t understand is nuance, tone, dialect, and other important language components.

The more complex AI is, the better this becomes, but right now, when you entrust your academic manuscript to a machine translator, you’re taking a chance that they might or might not get your words right.

When to Use Machine Translation

Modern methods of machine translation use neural processes. You can choose multiple platforms, depending on factors like your budget, speed, and content.

The use of machine translation is beneficial when speed and volume are a factor. If you need to translate thousands or millions of words quickly, machine translation does this for you. The program generally doesn’t watch for comprehension and whether what it’s translating makes sense. The right program can do this, but if you’re using a cheap or free software app, you may be losing something in translation.

It’s also helpful to use machine translation when you have to switch from one language to another if both languages are somewhat common. Many platforms offer more than 50 languages and can confidently translate from the source to the target quickly. This swift turnover means lower translation costs for you and a faster turnaround when you’re ready to publish your work.

The key is to be proactive and on top of your translation, though. Never assume that the machine translator has clearly switched your ideas, theories, and other content, and that the words you’re using in your primary language translate into the same meaning in your target language. 

Before you start your machine translating process, do your research to find out which programs work better for academic language. Then, make your article as easy to translate as possible by avoiding jargon, using short sentences, and following other best practices. Go through your article and look for words that may not have a clear translation, such as necessary academic terms specific to your field. Then, find out the translation as best as possible, and create a custom-engine glossary tailored to your most frequently used text.

Opt for full post-editing when you send your work to a publisher and let them do the hard scouring for you. If any areas don’t make sense to a scientific editor, they may have been changed in translation. It’s better to catch them now than after the paper is released to your audience.

Now that your paper has been translated and published, follow its influence on your target demographic with Impactio. With Impactio’s data analytics tools, you can watch your work as it expands throughout your community and grow your scholarly reputation at the same time.