As I drive into work this morning, I find myself listening to a fascinating interview with a renowned scientist. However, I find it hard to concentrate on what she is saying, because the interviewer's Boston accent keeps getting on my nerves. "Change accent to Midwest," I say. For the next part of the interview, I'm in auditory nirvana, listening to both parties speak about a fascinating topic, and even better, in the accent I grew up with.
However, as I hear these voices that remind me of home, my mind wanders to my great uncle. He loved science, and always encouraged me to study hard at this subject in school. "Change second voice to Uncle Nicholas," I say. Now, this makes my drive an absolute breeze. Hearing Uncle Nick tell me about the latest advances in science makes it so much easier for me to absorb the information. It's a shame he cannot be alive to hear his own voice forming sentences spoken by others, but I'm glad he had the foresight to leave so many recordings behind. It's one of my favorite features of the language conversion software that's built into my car.
Okay -- you caught me. I don't really have this tool available. And I don't have a great uncle named Nicholas. And a thick Boston accent does not annoy me (well, maybe just a little).
But this scenario, while entirely invented, is completely conceivable. And I didn't even reach the part where, while I was driving, I sent the interview around to a friend in Japan (translated into Japanese text and shared via Facebook), to my cousin's iPhone (purged of political content on which we happen to disagree), and to a high school student I am mentoring (converted into video and sung by cast members from Glee).
Science fiction just didn't take it far enough. The supposedly elusive goal of automatically converting information from one language into another -- commonly known as machine translation -- was not the end goal. That was only the beginning of a content transformation revolution. What we're already seeing in society is a proliferation of options for rendering information into our preferred formats.
Google is already giving people the opportunity to change reading levels as they surf the web. Speech technology is advancing by leaps and bounds. This week, I had a Skype video call with my nephew in which he looked like the Easter Bunny, courtesy of a digital overlay that mapped his facial movements.
During a short commute like the one I described, a person's preferences for receiving information can change every few minutes. Throughout the course of a typical day, there might be hundreds of moments in which we would change how communication takes place if we had any choice in the matter. But the options currently available to the average person are not so flexible. Today, we settle for what we have, but tomorrow's possibilities for improving communication will be limited only by human creativity.
The improvements in automatic, real-time, inter-language communication that we're seeing today mark an important moment in history. But when this technology gets personal and affects the average person, that's when we'll really see expanded interest, widespread adoption, and integration. It will become part of our everyday lives, to the point that we'll even take it for granted. Imagine the implications for teaching children -- or adults for that matter -- when they can receive information in ways that are highly relevant and customized to their learning styles and preferences. Humankind's ability to process and use information will vastly improve as a direct result of language conversion technologies.
Someday soon, we won't talk about machine translation. We'll discuss "language preferences," of which translation will be just one little part. Until then, people all around the world will be waiting for these advances in technology to reach them and enrich their lives. It's the end of machine translation as we know it. Let's just hope we don't have to wait very long.
Follow Nataly Kelly on Twitter: www.twitter.com/natalykelly
Nathan Newman: As Alternative to Smartphone Tracking, We Need a Public Geolocation Map
Daniel Luzer: Higher Education's Technology Problem
Saskia Miller: My Grandmother Just Emailed Me From Her iPad 2! An Inter-Generational Tutorial
Computer people have been working hard at getting computers to understand real human language since the 1950's. Many people have concluded it is impossible. I disagree, but it is still far away.
Linguists and others have investigated translation and concluded that true translation between languages is flat out impossible. With this I agree.
I think that verbal communication with the computer, a la Star Trek, is about as far as we will ever be able to go. There is no Babel Fish.
But, of course, dreamers will always dream on - oblivious to the real world.
“Kelly seems to have no comprehens ion of what she is talking about other than a wish list.”
Based on your response, my guess is that Nataly knows a good deal more about this subject than you do.
“Linguists and others have investigated translation and concluded that true translation between languages is flat out impossible. With this I agree.”
I’m not sure where you are getting that. That is certainly not a mainstream consensus among linguists. It is, at most, a matter of debate. Considering the thriving field of translation studies, “flat out impossible” seems quite the overstatement. People generally don’t keep studying the impossible. But beyond that, can you define “true translation” as you are using it? Perhaps you have some uncommon definition in mind.
“I think that verbal communication with the computer, a la Star Trek, is about as far as we will ever be able to go. There is no Babel Fish.”
But this is, in essence, what Nataly describes: computer-mediated communication.
“But, of course, dreamers will always dream on - oblivious to the real world.”
Nataly may have painted a rosy picture, but a lot of what she talks about is already possible and in use. I happen to be what one might term a “machine translation skeptic”, but more text now is translated on a daily basis by machine translation than by all human translators combined. So clearly she has some comprehension of what she is talking about.
In that case, I'm afraid you are quite simply mistaken. There are already speech-to-speech machine translation systems. They are not perfect, but they work. The U.S. military has used limited systems of this nature in Iraq and there are working research prototypes for more complex ones. These systems are not perfect (in particular they have trouble dealing with differences in speakers’ voices and with non-grammatical input) but they are already usable if their users are willing to adapt a bit to their limitations.