Traditionally, website translation involved long manual processes. Hiring a professional translator was just one aspect, the other side was ensuring that translated content was displayed on your site – taking up valuable developer resources.
The whole process was painstaking and for many business owners, reason enough to delay a global business expansion. After all, if your website isn’t displayed in multiple languages, you have little chance of appealing to those in new markets as they won’t be able to understand your offering.
While the traditional way of translating a website can still be used, website translation tools make translating websites effortless. But how do they exactly work to deliver such results?
The technological trifecta
Like most technologies, translation technology always introduces something new every year. However, experts believe the tools in use today are based on three leading technologies.
Machine translation (MT)
The first translation tools were machine-based, meaning they conducted translations with rules and conditions inputted in the device beforehand. Old as this technology may seem (first introduced in the 1950s), MT remains widely used today for its speedy translation, extensive language options, and lower operational costs.
One significant factor in MT’s continued use is artificial intelligence, giving rise to the neural-based MT. By creating a neural network that can mimic the functions of a human brain, developers can input vast quantities of linguistic data. As a result, translation tools can translate whole paragraphs with acceptable, if not high, accuracy.
In fact, long gone are the days of MT making laughable mistakes, neural-based MT is leading the way for many marketing teams looking for fast and accurate website translation.
Computer-assisted translation (CAT)
CAT involves employing translation software, which may seem similar to MT. However, whereas MT is fully autonomous, CAT works with human users to deliver more accurate translations. In this setup, machine-based translation plays second fiddle to the extensive contextual knowledge of human translators, generally as reference material.
One downside of CAT is its limited application. Unlike MT, which can benefit anyone with enough know-how, CAT is designed for expert translators.
Translation Management System (TMS)
A TMS combines the best of both worlds, particularly the automated efficiency of MT and human understanding present in CAT. The machine does most of the heavy lifting while the user reviews the result and makes creative changes as necessary. A TMS may also go by other names like a globalization management system (GMS).
This technology goes beyond translating content. Companies that conduct transactions in foreign markets find TMS tools useful because of their ability to streamline workflows and develop effective business strategies. It won’t be unusual for TMS software to feature functions to enhance multilingual search engine optimization.
One such website translation tool leading the way in the industry combines neural-based MT with professional translators and full-editing control allowing you to translate the whole of your website in minutes. Weglot, a WordPress translation plugin, also translates any website technology, both translating and displaying the content of your site.
Will human translators be obsolete?
Improvements in the results that translation technology gives are all but apparent. According to one industry source, based on edit distance, neural MT systems are getting better at their jobs by 3% to 7% every year. Edit distance refers to how much a developer has to change a translation system’s code to deliver a result that rivals human translation.
These figures may appear insignificant, but these incremental improvements compound every year that technology improves. The more accurate translation technology grows, the less need for humans to review their results, at least in theory. But if the technologies discussed earlier are any indication, it’ll take a while before translation technology can be completely independent of human intervention for several reasons.
First, multiple studies found that neural MTs tend to “hallucinate” or yield results that don’t come from the source material. These hallucinations may be grammatically sound, but they barely make any sense in context. Businesses and organizations using translation software can’t afford such outputs, namely when translating crucial documents.
Second, automation has yet to reach its zenith despite significant strides in multiple industries. Even as humans won’t be translating as much in the following decades, they’ll still have a hand in creating better translation technologies. Their ability to understand contexts more deeply than a machine will remain an advantage.
Translators and other relevant professionals may have to upskill to adapt to the up-and-coming technologies. A report by the World Economic Forum states that three out of ten companies all over the globe say they’ll need to retrain their workforces. Regardless, the costs involved are a small price to pay compared to breaking down language barriers.
It’s undeniable that translation technology has come a long way, from relying on predetermined input of rules to using what’s essentially working electronic brains. The technologies involved have opened plenty of new doors toward new opportunities. The next milestone is anyone’s guess, but people, for now, are glad that they can switch between languages within a few clicks.
- How Translation Technology Enables Multilingual Content - June 10, 2022