Is Google Translate threatening the translation profession?
Updated: Mar 1, 2021
Andrew Bell is a medical translator working from Spanish, Portuguese and Catalan to English trading under the name Bell Johnson Translations. You can connect with him by using the following links:
If I had £1 for every time I was told Google Translate would put me out of a job, I’d be a very rich man right now. There’s no doubt that technology is constantly improving, and this includes machine translation (MT) engines like Google Translate, but will technology ever reach a point where it replaces human translators?
To answer this question, I think we first need to look at what MT engines like Google Translate actually do. Simply put, they provide translations of documents almost instantly in a variety of language combinations. If you need a quick translation, Google Translate can be a great option. However, translators probably wouldn’t recommend using MT, and this isn’t because we’re worried MT will steal our jobs, it’s because as language professionals, we pride ourselves in producing high-quality accurate texts, something that isn’t guaranteed when you use MT. That’s not to say you can’t get really good results from MT, because you can, but it really does depend on the type of text you're translating and the languages you are translating to and from.
In general, in the language industry, we say that MT is great for ‘gisting’. If you’re interested in understanding the main idea of a text, MT will almost certainly be able to convey that to you. There may be a few robotic sounding sentences or even the odd mistranslation here and there, but you should be able to understand the central argument. However, if you need an accurate and fluent translation, let’s say because you’re translating a legal contract, you don’t want to risk MT getting something wrong.
MT may also be better for some language combinations than others. Languages that are used more frequently around the world, and hence have more available data for MT engines, will inevitably produce more accurate and fluent results, for example, I imagine the results between English and French will be a lot more usable than the results between Georgian and Zulu.
The aim of this blog post isn’t to paint a bad picture of MT, because it truly is a remarkable tool when used correctly. Like I said before, it can produce very accurate results, and hence can be a very powerful weapon in the translator’s arsenal. In fact, as technology improves, MT will inevitably change the face of the language industry. Even now, there has been a rise in post-editing of machine translation (PEMT), which is where a translator corrects MT output. This service can only be carried out when the MT output is of reasonable quality, but when it is possible it can save the translator time and the client a lot of money. That being said, PEMT will never fully replace translation services, especially when confidentiality is key. Currently, when you use a MT engine, you are sharing your data with a third-party, and in many cases giving this third-party ownership of your data. Let’s say you translate the findings of a top-secret enquiry using MT, then the MT provider could potentially share these results before the data is officially made public.
I’ve only just touched on the topic of MT and there is certainly a lot more that I could say. To summarise, I’d like to say some texts are always going to need to be translated by a translator, but translators are certainly going to have to start getting used to sharing the spotlight with MT in the coming years.