Clinical and experimental findings on the bilingualism support two fundamental assumptions: 1) languages mastered by a bilingual use the same neural substrates, and therefore a common brain network; 2) a control mechanism is needed to avoid mixing between the languages. In addition, it seems that these neural substrates are modulated by the proficiency of the languages. This thesis proposes to clarify some aspects related to the influence of the used language (first or second), as well as to the second language proficiency, on the brain mechanisms involved in producing and understanding language. To this end, this work examines two aspects on a population of translators with the help of functional magnetic resonance imaging (fMRI): a) the dissociation between neural substrates involved in language selection and in cognitive control; b) neural substrates involved in written word comprehension, with special emphasis on semantic processing and on early word recognition during visual processing.