In this paper we present a new approach to the expression of certainty and uncertainty in scientific experimental articles. This will permit to ascertain the validity of knowledge extracted from biological literature and used to automatically populate a domain ontology. We argue that lexical terms such as show, find, observe... express a semantic category different from the one characterized by markers such as demonstrate, validate, support... We name the latter category “confortation” as it conveys a notion of strengthening and we propose five other semantic categories: lack of knowledge, objects of study, hypothesis, observations, and general knowledge. This last category and the linguistic phenomenon of reported speech are respectively examined as consensual truth and as knowledge reported from identified scientific sources.