Less than a month since Microsoft demoed its latest version of Skype Translator Google has chipped in with its own update designed to break down language barriers. In December, Microsoft revealed a test version of Skype Translator that decodes Spanish and English in almost real-time for bilingual video calls.
Fast-forward a month, and Google has officially announced two new features for its Translate app. The search giant has confirmed speech-to-text translation will be added to the application, while Word Lens means users can point their smartphone camera at signs or text for on-screen translation, even without an internet connection.
Machine translation is getting serious
The competition between Google and Microsoft is heating up as both technology giants aim to solve the world’s language problems. And while machine translation, apps and software have been around for some time now, the standard has always been very low.
This remains the biggest challenge for Google, Microsoft and other names in the industry – not just translating speech or text, but accurately capturing the same meaning in another language. The term ‘lost in translation’ couldn’t be more fitting here, and the truth is technology may never truly crack the art of translation.
Fancy features, but still no closer to accurate translation
The fact is, machine translation alone simply isn’t enough to accurately capture the same meaning across language barriers. The latest versions of Skype Translator and Google Translate are genuinely impressive, but they also reinforce how far technology is from accurate translation.
It’s a tall order at that – not only because of the unique complexities in each language, but how humans communicate as well. It takes fluent language skills, deep cultural understanding and years of human interaction – even for professional translators – to convert from one language to another with confidence.
Context, lost in translation
According to Peter Gilliver, an Oxford English Dictionary lexicographer, the verb form of “run” has 645 different meanings. It’s a perfect example of how context plays a key role in communication and affects the meaning of, not just single words, but an entire sentence or conversation.
This means, for one word alone, a computer needs to learn 645 definitions, and also how to apply and distinguish between them, based on context. And this is just to understand the meaning of the original message, forget about translating it into another language!
No rise of the machines just yet
So the latest Google and Skype achievements are pretty exciting for what they are but, the truth is, they haven’t brought us any closer to accurate machine translation. In fact, more than anything, they have highlighted how far technology has to come, the importance of human translation and the fact machines may never be up to the challenge.