No globalization technology inspires more debate than machine translation (MT), specialty software that translates text from one language to another. Everyone wants the multilingual expertise of Star Wars’ C3P0 protocol droid or Douglas Adams’ improbable Babel Fish. But what most get is the often awkward if not comical output from the online BabelFish and FreeTranslation. In their minds, no MT will ever approach the quality of human translation (HT).

We are more optimistic about its prospects, viewing MT as an alternative to “zero translation” — that is, the usual option of no translation at all. MT fills the very big gap between what organizations would like to offer their international audience and what they can afford to translate. We expect to see continuing evolution of MT over the next three years, pushed by military, intelligence, and other government needs.

One of the companies pushing MT hard is Language Weaver, labeling its efforts “statistical machine translation software” (SMTS) to distinguish it from the older rules-based technology underlying most free translation systems. This probability-based approach employs a statistical analysis of patterns observed in large volumes of content. Simplistically speaking, systems computationally analyze this data, treating it as strings of characters, determining patterns, and exploiting these regularities. Funded by government contracts and then venture capital, Language Weaver claims to be the only supplier offering a commercialized statistical solution (although some European suppliers might question that).

Besides re-focusing the discussion around statistics, Language Weaver’s newest version addresses typical IT concerns about the lightweight, workgroup nature of most globalization software. Sprinkled throughout this press release are enterprise catch phrases such as scalability, load balancing, distributed processing, and a 100x speed increase. The release also answers anti-MT blasts with claims of higher quality, accuracy, and translation bandwidth of 500,000 words per minute. Finally, in a subtle response to competitors’ claims that Language Weaver is all about CIA-funded, Arabic-centric translation, the release notes an improvement in processing Latin-based languages. This press release does its job in positioning Language Weaver as an enterprise-class solution.

In another press release, Language Weaver announced that it hired industry outsider Kirti Vashee as vice president of sales and marketing. We recently met with Kirti to discuss his move from mainstream enterprise technologies (CMS, messaging, and storage systems) into the language technology business. As an expatriate living in the States with years of experience setting up and running international operations, Vashee sees a big opportunity in extending Language Weaver’s reach to markets on the receiving end of a flood of information and data.

What’s the bottom line on MT? We view this quickly evolving technology as part of the digital mindset, fitting seamlessly into the zeitgeist of process automation. It can ingest vast amounts of content that would otherwise never get translated and thus speed up the availability of information to decision makers — sitting in their corner offices or in bunkers in undisclosed locations. Closing the multilingual information gap is critical for business and government, and there will never be enough human translators to do that. MT will fill that gap for many companies.