Alex Ross recently wrote a piece for the Wall Street Journal basically suggesting that in 10 years time, we will have something like a Babelfish.
I did not agree much with the piece for various reasons but I have been busy for the last few weeks and frankly, this discussion (podcast based) by three interpreters demonstrates possible gaps in understanding what is required here, and where our technology actually in reality more than I have time to draft at the moment.
Way back in the history of computer science, computer programmers who were unfamiliar with the world outside the English language created an encoding system that couldn’t handle accented charactors. One of the points that struck me about this is that a significant amount of the progress in NLP is for recognising English…but we need significantly more work amongst a lot more languages.
Many things have been forecast for some time down the road. Those things rarely arrive, or if they do, they arrive in an entirely different shape to what we predicted.
I will say that I don’t agree that interpreters sell themselves as being dictionaries on legs – I’ve always seen interpreters as a bridge.