Over the last few months 3 companies offering mainstream corporate software have built translation support into their products. All chose to develop their own technology rather than integrate tried-and-true language industry software.

• MadCap (US) released the “ultimate authoring suite,” including Lingo, an XML-based, Unicode-based translation memory tool integrated with Flare (its online help authoring product) and Blaze (its upcoming publishing tool). The company has made multilinguality an integral part of the document development process. It claims that Lingo “eliminates the need for file transfers in order to complete translation…[so] documentation and localization professionals no longer have to risk losing valuable content and formatting… Documentation professionals can deliver a consistent user experience — in print, online, and in any language.”
• Cross Media Solutions (Germany) will soon release a translation tool for its SKATE catalog publishing application. The company wanted to reduce the cost of producing catalogs in different languages, so it built its own inline translation tool using TMX and XLIFF. The resulting XLIFF-EDIT component directly reads the SKATE database format, lets authors and translators edit it directly, and permits them to apply previously translated database terms throughout the entire database catalog.
• After years of cramming every innovation into its acrocheck product, acrolinx (Germany) has been busy repackaging its flagship tool into separately purchasable modules. The next release, focused on a range of information quality management improvements, is scheduled for release in the coming months. Among its solutions (we’ll write more in a future posting) is a nifty set of linguistic extensions for cross-lingual checking, intelligent indexing, evaluation, and a tool for finding micro-clusters of reused terms. This last one, named Intelligent Reuse, promotes reuse at the sentence or phrase level, similar to authoring memory but with linguistic intelligence to make it more precise — and thus improve translation matches. While acrocheck has traditionally focused on improving consistency, efficiency, and quality across the information life cycle within a language, its extensions provide these capabilities in and across multiple languages.

What intrigued us about these recent announcements was the developers’ decision to build functionality traditionally associated with the language technology industry — translation memory and terminology — into mainstream content-centric products. They follow the path of software vendors like AuthorIT (Localization Manager), DocZone (translation memory), Ektron (XLIFF inside), thebigword’s LanguageDirector, and Tridion. Some of these pioneers developed their own translation technology, while others like DocZone and thebigword have integrated tools from smaller suppliers like XML Intl, and Tridion has embraced SDL’s technologies. In the cases of Madcap, Cross Media, and acrolinx, all three skipped traditional CAT tools (SDL/Trados, Atril, WordFast), new-wave translation memory (Elanex, Kilgray, Lingotek, XML Intl), and even academic and open source offerings. They found the commercial tools wanting, too expensive, or too complex.

You’ll see more of this “translation inside” happening driven by more use of enablers like XLIFF, machine translation hooks from productivity suites like Office 2003/2007 and Lotus Symphony Beta 4, and improved translation options from Google and others. Traditional TM tools will complicate themselves out of this embedded market through yearly upgrade cycles at high price points with ever more features and complexity. Developers with no historical ties or loyalty to traditional translation memory tools like Trados would consider a browser-based TM or small plug-in component, a standardized segmentation scheme, bi-directional TM transfer, and a low- or no-cost of ownership. We expect that these translation components will come from smaller commercial, open source, and even academic institutions that have no installed base to cannibalize.