MT Research Revitalized from The Hague?

At the crossroads of the major European languages, the Netherlands enjoys a healthy reputation for multilingualism. The Minister of Education, for example, would like to bring the use of English as the instructional language in university classes up to 20%, hoping, among other things, to attract more students in the post-1992 era. But, as usual, the matter is hotly contended.
    Language issues are also involved in current Dutch technology strategy discussions. Over the past decade, the Ministry of Economic Affairs has financed a number of Machine Translation-related studies, even coughing up fifty percent of the ECU 4 million that went into BSO’s Distributed Language Translation (DLT) project (see LIM nr. 1). Moreover, it put nearly ECU 10 million into such joint university projects as Speech Analysis and Synthesis, Knowledge-Based Systems, and Man-Machine Communication.
    Enter the 1990s. The government’s IT subsidy stream has been reduced to a trickle, while contributions to the giant JESSI megachip project were dramatically reduced from ECU 42 million to a paltry ECU 3 million. But now, after all this belt-tightening, there is a glimmer of hope for language technology. At the request of the Dutch Ministry of Economic Affairs, British consultants Butler Cox pinpointed key areas for IT research which have the potential to inject new life into the Dutch economy over the coming decade. They pre-selected twenty-seven mid-term technologies out of an array of sixty-one possibilities. Such technologies as parallel database engines, handwriting recognition and robot locomotion should be ripe for industrial exploitation around mid-decade — provided there is additional research — say Butler Cox.
    More interestingly, Bulter Cox spot some important trends. Computer Support for Cooperative Work (CSCW, or Groupware), for example, could result in some interesting multi-author wordprocessing environments. On the demand side, there is the voracious appetite of certain sectors of Dutch economy (agriculture, transport, trade, and the media) for new and improved applications of IT. Building on this technology push/demand pull base, Butler Cox shortlists three of the original sixty-one growth points areas as which deserve special incentives from the government: automatic product and transaction identification and tracking, distributed hypermedia, and — low and behold — Machine Translation.
    In evaluating MT, Butler Cox sees trends integrating MT into wordprocessing environments with interactive methods to disambiguate text sentences, new algorithms exploiting parallel processing architectures and, not surprisingly, the need for considerably more research.“The Netherlands has a significant research capability in the Machine Translation area,” write the authors.“In our view, MT research needs revitalizing and redirecting. It should seek to exploit powerful new technologies such as parallel processing and neural networks. Its objectives should be the delivery of limited tools with wide applicability…”. Powerful techniques, modest objectives: doesn’t that sound attractive?