The development and research of discourse-annotated corpora is a relatively new field, therefore Lithuanian researchers seek to supplement the existing corpora resources and look for ways to study discourse relations by linking and comparing them with their counterparts in other languages because in different languages discourse relations are realized by different linguistic means. The aim of the article is to present the developing available corpora resources drawing on the experience of foreign scholars and to discuss guidelines for translation research at the discourse level. Therefore, the article first deals with the possibilities of expressing discourse relations by using discourse markers as their linguistic realization in different languages, discussing possible choices of translators, taking into account discourse realtions in translation and the use of different linguistic means. The article presents the parallel multilingual corpus TED-MBD (