Quality Translation for a Multilingual Continent - Priorities and Chances for European MT Research

Recent progress in translation technology has caused a real boost for research and technology deployment. At the same time, other areas of language technology also experience scientific advances and economic success stories. However, research in machine translation is still less affected by new developments in core areas of language processing than could be expected. One reason for the low level of interaction is certainly that the predominant research paradigm in MT has not started yet to systematically concentrate on high quality translation. Most of the research and nearly all of the application efforts have focused on solutions for informational inbound translation (assimilation MT). This focus has on the one hand enabled translation of information that normally is not translated at all. In this way MT has changed work and life of many people without ever infringing on the existing translation markets. In my talk I will present a new research approach dedicated to the analytical investigation of existing quality barriers. Such a systematic thrust can serve as the basis of scientifically guided combinations of technologies including hybrid approaches to transfer and the integration of advanced methods for syntactic and semantic processing into the translation process. Together with improved techniques for quality estimation, the expected results will drive translation technology into the direction badly needed by the multilingual European society.