5 Reasons Why the World Can’t Rely on Machine Translation (Including Examples)
Machines and artificial intelligence (AI) are cool, aren’t they?
By helping us with mundane household chores and finding information in a matter of seconds, they have certainly revolutionised our modern lives.
But do you know what search engines, online tools and other types of machines can’t do? Feel. And when it comes to selling products or services to an international audience, it’s the (lack of) emotion that makes or breaks the deal.
In the language world, we immediately think of tools like Google Translate in the context of machines and machine translation. For language professionals, the scary part is that Google Translate (and comparable tools such as Amazon Translate, Microsoft Translator, iFlytek, Apple Translate, etc.) has become a lot better in recent years. Can machine translation replace human translation though? Absolutely not.
Let’s look at some linguistic reasons why we can’t rely on machine translation and explore some real-life examples.
5 linguistic reasons why we can't rely on machine translation
I sometimes handle machine-translated material for language agencies. It’s my job then to “post-edit” the content, making it read as if it were written by a human.
Most recently, I worked on an editing task for a friend. The content had been machine-translated into English to save time. At first glance, I’ll admit, it wasn’t horrific. I didn’t scream and instantly close down my laptop. But: Going through the material and reading it out loud, I quickly realised there were plenty of adjustments to make.
1. Local differences
Complexity, local variations and ever-changing language make automated translations a difficult problem to solve.
Is your text supposed to address a UK or a US audience, for example? Machine translation tools have no clue. How could they? As is the case with many languages around the world, English isn’t the same everywhere. This results in inconsistencies such as visualise vs visualize, humour vs humor, metre vs meter etc.
It’s often these little nuances that make all the difference. Having a human translator handle your content with your specific target audience and its culture in mind shows that you’ve thought of them by putting in the extra bit of care.
Would you call this a bread roll or a tea cake? The debate between the US and the UK never ends.
2. Long sentence structures
Languages such as German and Dutch tend to have long words and complicated sentences. English sentences, on the other hand, are usually more straightforward and shorter. Amongst other things, that’s because English uses the active over the passive voice where possible and prefers verbs to nouns.
Using lots of long sentences, you run the risk that your readers doze off or move on. Your human translator can help you create content that’s straight to the point and keeps your readers interested in finding out more about what your business has to offer!
3. Off-brand words
Are you snappy and fun in your marketing materials? Or do you prefer a more elegant tone to show authority in your field?
Only a human translator familiar with your brand’s style, tone of voice and glossary (if you have one) will be sure to match your written materials with your brand voice and intent. To give you an example of possible off-brand words: With machine translations, you may end up with something such as "insanely large" when really, you know that doesn’t convey your brand’s tonality. A human translator may suggest something like "immensely large" or even just the simple version "large" instead – as long as it’s on-brand and reflects your business.
4. False friends and similar words
A false friend in a language sense describes “a word in a foreign language that looks similar to a word in your own language, but has a different meaning”.
Let’s just say that machine translation tools aren’t always your friend either: The text that I recently worked on had a heading that read produce. I knew that the context of this piece was video recording. That’s why the correct heading here was Production. Produce and production look like similar words, but couldn’t be more different in meaning: Produce is what you find in supermarkets (fresh fruit and vegetables), while production – in this case – refers to the technical process of creating (producing) video content.
Having specialised industry knowledge is key. Governments, businesses, and health services that need reliable translations still go for professional translation services for that reason.
My friend, who had asked me to help her with this editing task on her company’s behalf, messaged me after the first round of corrections. She agreed with me saying that the original text was nowhere near the quality it needed to be.
5. Reaching customers on a human level
When it comes to us humans, we need more than plain information to make a purchasing decision. Only once we feel in some way emotionally engaged or see a solution to our problem, we trust a supplier enough to buy a product or service from them.
To achieve this level of emotion through language often requires things like wordplay, (pop) culture references and humour.
Illustrating a lack of (human) awareness, here are some words and phrases that Google Translate struggled with:
What would Sir Mix-A-Lot sound like if he sang this in German? Haha!
I sure hope not!
Excuse my French.
Side note: If you enjoyed these, check out this hilarious YouTube channel for more Google Translate fails.
When does machine translation make sense?
When it comes to human vs machine translation, there can certainly be a happy balance: Machine translation tools can serve professional human translators.
For some types of content, it actually makes sense to apply machine translation to save time:
Content intended for company-internal use only (rather than public-facing content)
Documents with a quick “expiry date”, e.g. a meeting agenda (rather than a website)
Highly technical documents with lots of numbers (rather than creative materials)
A trained and experienced human translator can post-edit/review the material to get the most out of it here.
In fact, translators use machine-assisted tools. So-called CAT (computer-aided translation) tools can be efficient and increase consistency within and across documents. The human translator is still in charge and teaches the CAT tool new subject-relevant terminology through prolonged use.
Machine translation: friend or foe?
According to Jaroslaw Kutylowski (managing director of DeepL, a neural machine translation service), there’s still an “ongoing need for human processes in professions such as marketing and law, where tiny linguistic nuances often make a huge difference”. Also, there's no robot that could replace human creativity.
The dangerous assumption that learning a language is useless because “machines can do it for you” can lead to an overreliance on machines that could then lead to a decline in human knowledge. While machine translation can satisfy our need for speed, we, aka humans, ultimately need to have the last word.
Remember: Humans buy from and connect with other humans.