The foundation of any good relationship is communication. This is true of friendships, marriages, communities, and civilizations. Being able to articulate nuanced thoughts and feelings is the only way by which we gain a deeper understanding of those around us. Throughout history, language has been the primary means used by humanity to communicate with one another. But we don’t all speak the same language, and so barriers remain. And while the urge to understand one another my date back to the dawn of time it is the advent of science-fiction that pushed the concept of a universal translation machine to the forefront of human imagination. Though unlike spaceships and androids, the concept of a universal translator came not from the wilds of our imagination but rather as a matter of practicality. In serialized science fiction, where the protagonists encounter a new alien species every week there needs to be some sort of plausible explanation for why they understand human language without devoting half of every episode sitting around learning how to speak to one another. Thus, the concept of an advanced computer program or organic microbe or psychic ability that instantly translates alien languages and enables communication. Of course, after seeing it on TV or reading about it in their favorite book there are plenty of inventors, programmers, and linguists who set out to create their own Universal Translator. Those efforts are what we are here to discuss in this edition of We Can Do That Now.
Believe it or not, for once real-world inventors have it a bit easier than their fictional counterparts. After all, they don’t need to account for never before encountered alien dialects. No, they get off easy, needing only to provide an instantaneous translation of the six thousand odd languages and dialects located right here on planet Earth. The key is not every linguist on the planet typing up every possible translation from every known language to every other known language. Language is too complex for that and a near-infinite number of combinations of words exists. What is required is machine learning, artificial intelligences that are capable of learning. Why is this important? Language is all about context cues. When a computer is trying to translate audio and it hears “bear” it may translate the word as “bare” because it doesn’t recognize from context that one is correct, and one isn’t. But machine learning allows it to be taught the difference and to apply what it has learned to its next translation. But these machines need hundreds of millions of data points to become accurate enough to be useful, which is easy when you are translating English to French but not so easy when trying to translate Swahili to Farsi.
The race to be the first to perfect this technology is on, with companies like Google, Microsoft, and Facebook all getting in on the action. Video chats that translate in real-time and specialty headphones that can link you with anyone speaking one of eighty-plus languages who has another activated pair. Airports like JFK and LAX have already rolled out Google’s Nest Hub at their information desks to facilitate conversations with international travelers. And various apps can now be downloaded to your phone that offer limited translation services.
Universal Translators still have some growing to do as a technology, and they won’t be translating alien languages for us anytime soon. But, as far as terrestrial communication goes, it’s safe to say that science fiction has become science fact. Universal Translators, We Can Do That Now.