In order to broaden the agenda for scientific research in machine translation, it is proposed to conduct a workshop that will bring together investigators currently at the forefront of this field along with key scientists in related areas of natural language processing. The program has been designed to bring out the link between evaluation and future directions in MT research. Evaluation is seen not as a results-oriented activity but rather as an intellectual exercise that focuses attention on areas in which research is needed. The program includes the evaluation of related NLP research . In addition, three entire sessions will be devoted specifically to addressing the issues that MT science is currently facing: "Contributions of Related NLP Research to Future Directions in MT," and "Future Directions in MT." Each of these sessions will be a discussion based on a series of specific topics prepared in advance by a study team to be designated for this purpose, which will guide the debate. A summary report will be prepared for general dissemination. This activity comes at an important time in the development of MT science. Fresh insights are needed, especially from areas of NLP that incorporate the latest linguistic research in discourse theory. The evaluation process, a spontaneous movement that is gaining momentum in the MT field throughout the world, helps to bring to light the research issues that hold the key to process.
