A few weeks ago, Lionbridge, the biggest LSP in the business, signed a multi-year software technology exclusive to offer Software-as-a-Service (SaaS)-based, text-to-text language automation solutions to commercial clients based on IBM’s Real Time Translation Service (RTTS) technology.
The press is that it’s a game-changer. “The strategic agreement is expected to produce a real-time multilingual communication solution that can increase the efficiency of global business operations, and help organizations become more interconnected by enabling rapid exchange and understanding of critical information. The offering also will enable clients to access and take action on data and information that presently is unavailable due to language barriers. Unlike free-ware translation applications, the combined solution can be customized using each client organization’s existing content and configured for specific business processes to increase translation quality and availability.”
That smells like press release copy to me, and these guys get a lot more splash than we could get with the same offering. I’ve been working on hybrid real time translation for years, but the implementation that I can pull off with my limited resources has been pretty not-so-hot. Ok, so these guys know what they’re doing, but just how well can it be done? Or are they just better at selling bad translation than I am?
“The partnership will create a customizable, real-time multilingual communication platform by combining the capabilities of Translation Workspace™, Lionbridge’s cloud-based translation memory, and RTTS, IBM’s real-time translation technology. The combined solution is expected to broaden the use of automated translation technologies by reducing cost, quality and process barriers that inhibit effective use of machine translation. This will enable clients to cost-effectively translate the vast majority of content that is not being translated today – from eSupport, FAQs, manuals and technical documentation to blogs, instant messaging, user-generated content and chat.”
Best and most informed discussion on the implications are on Dave Grunwald’s GTS Blog. “Lionbridge should give their PR people a bonus for putting together such a great announcement. But is this a game changer for the industry? What do you think?”
Personally, it sounds like translation fail to me. I’m OK with bad MT, but I get concerned about offering multiple levels of quality on the human side.
But you’ll get more useful commentary from some of the heavyweights who’ve posted there. Kirti Vashee of Asia Online is the go-to guy for realistic discussion of MT, calls IBM “the poster child for MT failures.” György Smidt gets to what I consider the heart of the machine translation experience. “Rather than requiring you to consciously use a tool, you never even see the tool, which operates hidden in the background… this is really something new, not just a me-too bid by Lionbridge. There have been piece-meal attempts to do this by other companies, but nothing systematic enough to be used as a corporate communication tool.” Guilty as charged.
No one expects quality in real-time systems, even if they say they do. You know the service business iron triangle? Fast, cheap, good, pick two. After years trying to make money with machine translation, I’ve come up with the MT iron triangle: fast, free and easy, all of the above.