The future is here. We have self-driving cars, smart homes, and software that can translate anything from any language into any other language. That’s right – it looks like the babel fish from Douglas Adams’ magnum opus is here and nobody needs to learn any languages anymore! Translators have now become obsolete and are all undergoing a rebranding as pastry chefs – apparently the only field left where we can’t be replaced by AI that does our job better than we could ever hope to.
But wait, that’s not what’s actually happening, is it? Translators are still very much a thing and they’re earning good money for their services. In fact, people wanting accurate document translations still have to do it the old-fashioned way, because Google Translate apparently doesn’t entirely cut it when it comes to actual translations of human-generated content. And then there are those eerie instance where the technology just goes haywire and the translation turns out to be a garbled mess.
So what’s wrong?
Well, Google is actually a fairly smart tool. And that’s all it really is – a tool. It’s hardly of any use to you if you don’t know how to use it, and you definitely won’t trust a hammer – no matter how advanced – to build your deck for you. We’re here to tell you exactly why Google Translate doesn’t seem to deliver on all of its promises all the time.
First, the good
Before we get to the reasons why Google Translate isn’t doing its job for you, let’s go over how smart the tool really is. There have been plenty of updates to the software lately that have made the Google Translate experienced smoother than ever before, but even in the past it operated on a fairly well-thought out principle.
Google’s main strength is its access to a large body of data. A lot of this data is textual, and a lot of that data includes official documents which are fairly formulaic – within and between languages. How the GT algorithm worked (and still does to a degree) is that it looked up the phrase that you typed in within its many documents and compared it to its version in a similar document written in your target language. This, of course, meant that phrases that are more formulaic and more frequently encountered in official documents were translated with more accuracy while more abstract sentences could get pretty messy.
Times have changed since then, and Google Translate has gotten smart – or rather deep. With the advent of AI, machine and what is commonly referred to as “deep neural networks”, Google Translate can actually be even more accurate with its translations as ever before. With the added repertoire of input from users, as well as the ability to learn from them, the translations produced are now much closer to a regular, human-made, accurate translation. Of course, as long as the translated piece in question is a short fragment.
So what exactly are the problems with this?
Language is still too nuanced
There’s something that artificial intelligence and other learning algorithms are still lacking in – the experience of being human. The thing about language is that intrinsically cultural, and culture is, as far as we know, still a uniquely human experience. What this entails is that there are far more layers to a language than just a linear string of words that has a direct equivalent in a different language – there are many pragmatic aspects to them, such as a bunch of words that mean basically the same thing but in slightly different contexts, ro strings of words that make up a meaning that is completely different than what the actual words mean in a literal sense.
Combine that with just the fact that on both a lexical and a grammatical level most languages are completely unrelated to each other, and it’s no wonder a machine can’t always accurately predict what exactly is being said. Think about how some languages have grammatical gender while others do not, how some always need to use pronouns while others can skip them because all the information necessary is in the form of the verb. For all of these examples, humans would have no problem translating them because they can directly read, or at least infer, the context – and for that, you need knowledge that is completely cultural.
You speak a niche language
Now, that said, even considering the limitations of Google Translate, you’ll still have a better experience with it if you’re translating to or from English, German, French, or Chinese than you will with any other language. While it’s true that the technology allows you to translate over a hundred languages, not all of those languages are equal in the eyes of Google.
First of all, there’s the issue of the available documents mentioned before. English, German and French have the benefit of being historically widely spoken and being some of the most prominent languages in the EU – with EU documents being so prevalent in Google’s database, it’s no wonder they’d be in abundance. Chinese is listed there due to the sheer number of native speakers, of course, who are at this point more and more frequently using the Internet. It’s only logical, then, that this breadth of usable content will make it easier for Google Translate to create viable translations.
Things get complicated further if both your source and target language are more niche, as then you pretty much never get a direct translation. Since finding matching documents might be harder if you’re translating, for example, from Urdu to Swahili, the software simply uses English as an intermittent language. While this might seem like no big deal, there is a lot to be said about things being lost in translation. Even when done by an experienced human, there is a lot of nuance to language that simply gets lost with translation. If one more step is involved in the process, that only serves to increase the losses further.
You’re not using it right
The final thing that bears mentioning – and is probably the most important of them all – is that a lot of people simply aren’t using Google Translate right. Sure, it’s advertised as this babel fish solution we mentioned earlier – you can just type in your email and, just like magic, you’ll have a ready translation fresh out of the oven – just like mom used to make. And while that will certainly give you a better understanding of the content than if you were to read it in a language that’s completely foreign to you, mistakes happen, as you should know already.
That’s because Google Translate’s true potential lies not in it as a replacement to a real-life translator, but as an assistant to them. It is a tool that makes translations much faster – it provides you with a rough equivalent of the original document in seconds, giving the translator an easy way to compare both language versions in real time and move on from there. It may be smart and it may bring the world closer together, but as we’ve stated at the beginning, a tool is only useful to someone who knows how to use it.
It turns out translators are not becoming obsolete at all – far from it. In fact, they might just be more valuable now, with such an advanced tool in their hands. Google Translate is an amazing invention that only keeps getting smarter, but if you don’t know what you’re doing, you are likely to end up with some big mistakes. If you just want to see what something means and would otherwise be unable to understand it, GT will serve you perfectly fine, but if you want to actually translate any document, don’t skimp on money – hire a translator and avoid big blunders.