This paper presents a new framework to model an assessment process for a complex and multidimensional syndrome such as depression. Since the measurements of depression are inherently imprecise, we explicitly model the context of the assessment process, and we analyze various aspects of imprecision (syntactic, semantic, and pragmatic). The framework is based on fuzzy logic and semiotics. The fuzzy-logic approach allows for the representation of quantitative imprecision of the measurements and the semiotic approach allows for the representation of the qualitative imprecision of the concepts. We have applied this fuzzy-semiotic framework to two types of clinical measurements: the rating by the experts and the filling out of self-administered questionnaires. The proposed framework provides a conceptual foundation for the construction of a medical decision support tool. Keywords-- Assessment of depression, fuzzy logic, imprecision, modelling, medical decision, semiotics.