September 13, 2018

Will Machine Translation replace the art of linguistics?

Machine translation is just getting better and better. From crude rules-based, word-by-word software to new tech that thinks like a real person, you could be forgiven for thinking we no longer need human translators. But is that really the case?

As ever, tech giant Google finds itself way ahead of the intelligence curve. Since it first launched its Phrase-Based Machine Translation (PBMT) tool in 2006, the software has developed considerably. Its new development, Google Neural Machine Translation (GNMT), aims to mimic the functioning of a human brain. It treats the whole sentence as a group, rather than just a series of words that follow one after the other. The initial outcome was phenomenal; producing nearly human-level quality of translations and highlighting a significant improvement from its previous form.

pasted image 0

Source: https://ai.googleblog.com/2016/09/a-neural-network-for-machine.html

The benefits of such sophisticated software are vast, and have the potential to drastically improve the way we communicate with people from around the world. On a superficial level, it simply gives a more natural flow and interpretation of languages, just as human translation does. But, unlike the traditional form, it is more economical and time efficient. From a business perspective, it is also able to develop databases which will remember terminology specific to each brand, making it easy to keep brand phrasing and language consistent. Perhaps most importantly it means that by automating workflows, business have more power over their spending and use of time.

In practice, machine translation can be hugely influential in expanding a business’s reach. Research has shown that over 70% of customers would rather buy a product that is offered in their own language, meaning that website translation is paramount to attracting a global customer-base. Similarly, other digital content and self-service processes must also be localised to gain trust, show consideration and ultimately enhance the customer experience. Translated social media conversations and chat-bots help the business stay relevant, responsive and local to each individual user; a simple, cost-effective way to provide key information and real-time updates. It proves that the business cares about its customers, and wants to connect on a day-to-day basis.

Looking to the future, machine translation will be able to nourish real connection with consumers, a wider understanding of cultural differences, and a truly global customer-base in which language barriers are a thing of the past. That said, we can’t get too complacent just yet… machine translation is still known to produce errors ranging from minor to the comically huge – sometimes to the point of no return:

pasted image 0 (1)

Source: https://interestingengineering.com/30-best-and-worst-google-translate-fails-that-will-make-you-cringe-forever

Machine translation technology believes that language is just a series of words and sentences strung together according to specific rules that can be decoded. At its core, yes, translation is just code-cracking. But in practice, it is much more than that. This is where the technology meets its match and humans come back into play. We still need humans to provide quality control and sensitivity because machines cannot get their heads around it yet (pun intended). But what’s to say that they won’t be able to in years to come?

For now, we should probably leave the machines to automating a brand’s language, or just to the lower-stakes, everyday conversations. It will always be handy when we need to find out where the nearest bar is on holiday, or when we need to give that colleague in France a quick-fire email reply. Who knows how the software will adapt? While we wait, let’s leave the poetry and literature, or the pitches and legal-contracts to the pros.

Posted by

Team Curious

0 comments Add a comment

No comments yet


Leave a Reply

Your email address will not be published. Required fields are marked *