The Verbmobil System currently under development in Germany would input speech in a source-language dialogue, translate it, and output synthesized speech in a target language, all in real time under the conditions of face-to-face dialogue. This is a study of the field and possible problems.