Our Analysts' Insights

Every day, nearly three quintillion bytes of digitized data come into being – a quintillion is one thousand trillions – so that means 3 × 10¹⁸ (“The Calculus of Global Content”). This daily wave of new content supports interactions and transactions across the entire spectrum of human and machine activity – and localizing it is essential for many international business, governmental, and humanitarian activities. There is no sign of this daily growth in content volume slowing down – and with it comes gigantic projects to transform and translate it for other purposes and markets.

Daily Volume of Translation 

How much of that daily flood of content gets professionally translated? We calculated the daily translation output of LSPs based on our Global Market Study sizing model, factored in some realistic assumptions about what actually would be worth translating, applied that to the daily tsunami of 3 × 10¹⁸ bytes, and determined that much less than 1% of that quotidian amount could possibly get translated by a human.

It would take 1,000 translators working full-time (that is, eight hours a day) for 61,375 work years to process 0.01% of the total volume into just one language. Reducing the total volume even further, we estimated it would still take two billion full-time translators to render all relevant content into the economically important languages. For some languages, the number of translators required would exceed the total pool of speakers.

Human vs. Machine Translation 

How about using machine translation? The mainstream business press offers an increasingly confident – but fundamentally misguided – claim that today’s NMT is as good as, or has even surpassed, human translation. While machine translation could process vastly greater volumes, that increased availability from NMT comes at the cost of linguistic quality. 

Rather than write the obituary for human linguists, it’s time to confront the stark realities of this ever-growing pile of content. Few companies or governments are prepared to pay for − much less manage − all of the translation they’d like to do. And if they were ready to step up with the cash, maybe enrollment in foreign language studies and translations wouldn’t have dropped as dramatically as it has in recent years. 

The challenge for companies and governments is to apply the right translation modes to the right applications. Besides using ever-improving NMT software, it also means provisioning the few – and ever fewer – professionals who can translate with better tools so that they do it faster, better, and more efficiently. LSPs and in-house translation departments can pair CAT-equipped and MT-augmented humans with smarter software to:

• Give translators linguistic superpowers. They can process more words faster with modern computer-aided translation (CAT) tools that help them create higher-quality output that is consistent with the application or brand voice, uses the correct terminology, and manages the work. While classic tools – such as Déjà Vu and Trados – have been adding features, other well-funded start-ups such as Lilt and Smartcat and open-source efforts such as MateCat are attempting to redefine the relationship between human and machine. Mobile tools such as Stepes are pushing translation to the handheld devices that increasingly dominate content creation and consumption, and automated content enrichment capabilities in offerings from companies such as SDL, STAR, and Vistatec bring web-scale knowledge to translation. All of these move in the direction of augmented translation that puts the AI capability of machines at the fingertips of translators. 
   
• Equip the crowd with powerful tools. Some of those CAT tools can in turn bring para-professionals – that is, bilingual businesspeople or professionals such as lawyers, doctors, engineers, and designers – into the translator resource pool. They could supplement their day jobs with translation gigs in specific domains where their expertise gives them an advantage. Even “walk-ins” from society in general can handle some translation tasks. They all need powerful but easy-to-use tools to do the specialized translation that today’s MT shouldn’t process – and contribute to the pool of data to train the machines for what silicon can do today or that quantum neural-gel-matrices will handle in the future.
   
• Help curate the customer experience. Businesses cannot intelligently and sensitively engage with people throughout all the phases of the customer journey in a language that the customer isn’t comfortable with. Knowledge of and insight into linguistic, cultural, political, and other locale-specific details will come with humans doing the work. Over time, their work will contribute to deep training of MT engines. Because language and culture changes rapidly and non-deterministically, there will always be a place for humans working in these roles.
   
• Enable better human-computer interaction. Content types like marketing and verticals such as legal, life sciences, and financial services require more human oversight and touch than machines can provide. The shift to augmented translation will offload repetitive tasks and free up humans to focus on those aspects of translation that require their attention, making them more productive and valuable.
   
• Integrate MT into the translation workflow. Machine translation is just one component of a complex set of technologies required to get the right content to the right people in the right language at the right time. Our research shows widespread use of MT at LSPs. If the LSPs are making more money at more profitable levels, they will both be able to need and afford more tools to process the wider range of content they manage. This creates a virtuous cycle − MT could create more opportunities for human translators by getting their client’s customers into deeper stages of the journey.

In an environment where the popular business press and slick viral ads depicting cross-cultural romance promise that language is nothing more than a minor hurdle, investing in translator tools may seem like investing in rotary phones or cathode-ray-tube televisions. The fact of the matter is that human translation isn’t going anywhere for the foreseeable future, but translators and businesses need the capabilities offered by sophisticated but easy-to-use CAT tools. Far from being a waste, investment in them will deliver considerable advantages to users and help drive greater productivity.