This blog post is based on the Machine Translation Literacy project infographics developed by Lynne Bowker, Professor at the Department of Languages, Linguistics and Translation at Université Laval in Québec, Canada.
2024-11-18
News
Using machine translation is easy. But using it critically requires thought.
Are you already aware of the risks of machine translation? Then this post is not for you. But if you find yourself thinking “What’s wrong with using Google Translate?”, then you should probably continue reading.
Machine translation (MT) is a tool that automatically translates a text from one language to another, like Google Translate or DeepL Translator. And MT is older than you might think! It’s become popular since Google Translate was released in 2006. But researchers have been working on it since just after World War II. In recent years, combining MT with deep-learning based approaches like neural machine translation or large language models (LLM) has made rapid progress.
Machine translation tools are simple to use. Just open the tool in your browser, type in your text, choose your languages, click “translate”, and you’ve got a translation! Well, sort of.
“Professional translators have spent years studying and practicing to get to the point where it looks easy.” says Lynne Bowker, Professor at Université Laval specializing in machine translation literacy.
“Computers can imitate some of the work of translators, but they don’t actually understand the texts – they only process them, using techniques such as pattern matching and calculating probabilities.”
Using an MT tool may be easy, but using it critically requires some thought. Thinking about whether, when, why, and how to use machine translation are part of the term “MT literacy”. It comes down to being an informed and critical user of this technology, rather than being someone who just copies, pastes and clicks.
Machine Translation Literacy is a project led by Lynne Bowker. The goal is to inform about how MT systems process information and to teach how to interact with MT tools, before or after the translation phase, to get the best results.
So what can you do if you need to use machine translation? These are three steps you can take to ensure the best quality outcome.
Ask yourself: Is machine translation a good choice for this text?
“Just because you had a great results the last time you used a machine translation tool, there is no guarantee that the next time will be a success,” says Bowker.
“One of the most important skills to develop if you plan to use machine translation is risk assessment.”
Tools like Care to Translate have been consciously developed to reduce the consequences that an incorrect translation can have in healthcare settings. It has also taken sensitive patient data into account. By using the app's medically verified phrase library, you are ensuring that culture and context is taken into consideration when communicating with patients.
For lower-stakes tasks - go ahead and use machine translation
For higher-stakes tasks - use machine translation with caution
A good quality translation begins with a good quality original text. Make sure the text you want to translate is clear and well written. You can improve the output text by improving the input text.
Machine translation is a helpful tool on many occasions, but check and edit the translated text before sharing it. And make sure the readers know the text was machine translated, and by which tool. Knowing the source of the translation lets readers determine how much they want to trust the content.
“Alerting other people that it is a machine translated text is very important because it will allow them to decide how much trust they should place in that translation.” Bowker explains.
“Hopefully the other people will also have developed their own machine translation literacy skills, but if they haven’t, you can help to educate them about the potential risks.”
Each MT tool has been "trained" with different materials, so results may vary. You've probably tried Google Translate, but what about DeepL, Naver Papago, Bing Translator or Systran? Compare several and see which works best for you.
“Of course, we are not personally able to train the free online versions of tools such as Google Translate or ChatGPT,” Bowker points out.
“But a hospital or clinic could potentially invest in a tool that has been customized for medical translation, and this could help to improve the translation quality.”
In the teaching resources of the Machine Translation Literacy project, you’ll find lots of great material to read and share. In the research output you’ll find videos of lectures, articles and much more.
And if you’re in need of translation assistance for higher-stakes tasks, feel free to use our medical translation app Care to Translate, a digital phrase library where all phrases are medically, culturally and contextually verified by native speakers with medical expertise.
