EUROTRA: AN ASSESSMENT OF THE CURRENT STATE OF THE EC'S MT PROGRAMME
Doug Arnold, Louisa Sadler · 1992
In this talk we are going to give a view of what is happening, and is likely to happen in the 'Eurotra' programme, and its surrounding initiatives. To do this, and to make sure the paper has some relevance beyond the immediate future, we will also sketch out some of the history of the project, and describe the results that have been achieved. We will also give some pointers to the technical, and other literature on the programme. In outline, the history is this. The official launch of the Eurotra programme was in November 1982, though a certain amount of work had been done by various groups and individuals before this. The stated aims involved research and development (R&D) aimed at production of a 'pre-industrial prototype for the official European Community languages, covering a restricted text type, and domain' (nowadays, we would probably call this a 'feasibility of concept demonstrator'), but there were important secondary aims to do with the development of infra-structure to enable the development of Computational Linguistics (CL), Natural Language Processing (NLP) and computationally applied language studies throughout the member states. From the outset a key principle was the distribution of expertise across the community, and the work on individual languages and language-pairs was carried out by groups of researchers in some 17 University institutes organized into 'language groups' (one per language) across all member states. Fundamental work on the software, formalism, and common linguistic representations was coordinated by a 'Central Team' of between 10 and 20 researchers, drawn from the participating centres. Though originally envisaged as being of 44 months duration, various delays and extensions due to the accession of Spain and Portugal, (so that the original seven official EC languages