Aside from the irresponsible journalism that propagated this story in the first place, the basis for the concept is fundamentally flawed. There cannot be such thing as a wristband a signer can wear that will translate their signed language into spoken language; why? Because signed language is not just on the hands! Signed language is on the face and the body as well. The grammar of signed language is made through eyebrow, mouth, cheek, and even nose movements. Signed language is made with head nods and shakes, head and body tilts, and even shoulder shrugs. Anyone who ever took an introductory course in ASL should know this.
There is one other important flaw in the concept of a gesture-to-speech translation machine, and that is the notion that there is one “sign language.” No, folks, “sign language” is not universal! No sir, no ma’am. Even if Google were able to take input from a human interface device located on a signer’s body–even if that included all the points on the face and body necessary to read signed language–Google would have to add hundreds of signed languages into their Google Translate engine. Language is culture-bound, just as gesture is culture-bound. I’d like to see how this supposed “Google Gesture” would translate the thumbs up gesture, which can mean something like “up yours” in countries other than the United States.
American Sign Language (note that the A in ASL stands for American; i.e., not universal) is a much richer and more complex language than people give it credit for; in fact, so are all the signed languages in the world. Until enough people learn to appreciate the sophistication, complexity, and diversity of signed languages, we will continue to swallow false stories like this hook, line, and sinker.
It’s great to see how people other than “interpreters” are implementing the “interpretation” of vague language for practical applications! Panos Alexopoulos, in his presentation Vagueness in Semantic Information Management, discusses how Internet engineers can design databases with search capabilities that can “interpret” what consumers mean when they say they are looking for, say, a “Big, modern restaurant.” (How many square feet is big? What year range or architectural and interior design qualifies as modern?) He discusses the challenge of developing algorithms that can translate vague search terms into specific results. Very interesting!