While there are plenty of machine translation tools out their to help with getting a basic understanding of what has been written in a foreign language, there are still many problems with machine translation. Some time ago I was working on a project which would use computer mediated technologies as a way of bridging translaters and interpreters with clients based on many different factors which in the end would do one thing. ensure high-quailty translation or interpretation for the client on a mobile device, kiosk or desktop based device. Without the hiccups which are so common with machine translation. The project was mothballed for a number of different reasons – though it was a good exercise in better understanding machine translation, voice recognition and the state of these technologies globally.
Speaking a number of languages and having been directly involved as a translator and editor I can just imagine the challenges that Microsoft is facing with its “live translation” feature on Skype, which was presented at Re/code’s Code Conference last week, held from May 27-29 in Rancho Palos Verdes, CA, by Gurdeep Singh Pall, a Microsoft Vice President.
From the image below, I can ascertain a bit about this product even though I myself was not at Pall’s presentation. As you can see the bottom of the screen, you can see the phrase “What brings you to london?” This suggests two things. The first is that this technology will be a hybrid one – depending on voice recognition of the source language for input and the output or target language being printed out on the other users display. The non-capitalization of the world London also suggests that the technology doesn’t have the “smarts” to ascertain a number of nuances. These naunces can be important in different situations, and non-verbal cues can be different between different languages, regions and cultures. Without these “smarts” the technology will not be picking up on the other seven aspects or non-verbal cues which must be considered by professional interpreters: intonation, tone of voice, vocally produced noises, body posture, body gestures, facial expressions and the pause.
I can understand that this tool will make it easier for a number of very simple tasks and may be developed at a better level in languages that share similar linguist and semantic structures, though I feel that there could be many major gaffes. I spent three years in a top law firm often editing many legal documents translated from Ukrainian or Russian to English, there were often many times that incorrect translation could cost someone thousands if not millions of dollars and this being the case I could not agree more with one of the skeptics of this technology Ken Erickson, a business anthropologist at the University of South Carolina’s Darla Moore School of Business.
When you are talking about this kind of electronic translation, good as it is, it “lulls you into a sense of comfort where you should be not so comfortable,” said Erickson. For over the last twenty-five years I have tried to keep up with developments in machine translation and then the first popular web-based application introduced by Digital Equipment Corporation and SYSTRAN known as Babel Fish in 1997. Since then many other web-based machine translation tools have appeared – Google Translate, and Bing Translation. Some of the experience which Microsoft has gain in the development of Bing Translation will somehow be integrated to function with Skype Translator, which is to be launched later this year according to Pall.
As in the past when suggested that journalists consult with their company’s librarians/information specialists in the case of this still to be released Microsoft Skype product I will again concur with Erikson. “If you want to negotiate a contract, you better not rely on something like this,” Erickson says. In such a case spend the money to hire a professional translator – machine translation still has a long way to go!
Source of inspiration: Marketplace