R. examines four hypotheses related to the semantic description of human languages. The first one states that sentences contain instructions for interpreters on how to build the senses of possible utterances, using their knowledge and beliefs about the situation of utterance. According to hypothesis II, these instructions concern the point of view required for interpretation. Hypothesis III argues that these points of view can be described using topoi, topical fields and topical constraints, whilst the latter, according to hypothesis IV, are conventionally associated with the words of a natural language, thus reflecting ideology, belief or competence.