To study targeted linguistic aspects, notably authorial reasoning and positioning, we have composed a French corpus of diverse written scientific texts that reflect both discursive sub-genres as well as a range of disciplines. We have selected disciplines that we found to be both representative and relatively available.
Three disciplinary families are included: Humanities (Linguistics, Psychology, Education, and Natural Language Processing), Experimental sciences (Biology, Medicine) and Applied or Technical sciences (Electronics, Mechanical Engineering). The selected sub-genres cover research articles, written conference correspondence or reviews, theses, and “mémoires d’habilitation à diriger des recherches”. The corpus contains 4.8 million words.
CONSULTE THE CORPUS (Firefox 2+ is recommended, Chrome and Internet Explorer are possible).
This corpus is uniquely available for non-commercial research under a Creative Commons licence, on the condition of mentioning the source, the ANR project after signature of a convention.
Format : XML – TEI
Character coding : UTF-8