11 Dec /17

Machine translation

Machine translation - Word of the day - EVS Translations
Machine translation – Word of the day – EVS Translations

Several years ago, if today’s word was uttered, it would most likely be met with scepticism; however, in just a few short years, we have seen machine translation, as well as the technology behind it, become faster and more reliable.

While we may think of the concept of machine translation as being modern, or even futuristic, the groundwork actually comes from far back in the past. The idea itself was notably first proposed in 1629 by the French philosopher and scientist Rene Descartes, who championed the concept of a universal language with ideas in different languages sharing a common symbol.

Making Descartes idea applicable, British computer scientist Andrew D. Booth first mentioned the concept of using computers to translate documents as early as 1946; moreover, American Warren Weaver expanded this idea into his 1949 work, (Memorandum on) Translation.

As for the first known use of the phrase machine translation, that can be traced back to James Whitney Perry’s 1952 work, Machine Translation of Russian Technical Literature, where he matter-of-factly states that: “Machine translation of Russian scientific literature was simulated.”

Fast-forwarding to the late 1990s, with computing power doubling every 2 years and technology becoming more affordable, the first provider that most of us became accustomed to was AltaVista’s Babelfish, which boasted a then-impressive 500,000 requests a day in 1997.

15 years later, after the advent of Google and the concept of Big Data, Google Translate was working with approximately 65 languages and translating enough text to fill 1 million books on a daily basis.

Within the last years, thanks to Skype’s Live Translation feature and Google’s recently released Pixel Buds, machine translation appears to pull us ever closer to a sci-fi world of universal linguistic understanding.

Neural machine translation, aided by artificial intelligence and deep learning technologies, is coming out as a potential game changer to impact translation accuracy and localisation efficiency, yet, machine translation is still far away from reaching the level of quality and message conveyance of actual professional human translation, and not to forget the issue with data protection and security.