The basic hypothesis of the research is that a computer analysis carried out on corpora of texts transmitting specialist knowledge could give the opportunity to verify the actual presence in such texts of particular features such as the tendency towards nominalisation, a higher density of lexical items (or 'open class words'), the preferred use of certain verb tenses, a higher frequency of the passive voice and the presence of subject-specific collocations. The paper highlights the specific relevance of collocation for areas such as translation training and terminology management. Indeed, one of the proposed applications of the project is the development of a tool for the use of monolingual target-language corpora as translation resources. The aim is to give both trainee and professional translators a repertoire of subject-specific phraseological units whereby they could conform to the stylistic requirements of a given specialist domain in the TL.