Sentiment analysis is an automatised technique of analysis aimed to measure the “polarity” and the “subjectivity” of large corpora of messages. The case study of the present paper consists of a selection of Pope Francis’ tweets on ecological, social, religious themes and the relative polemic replies. In the degree of agreement/disagreement in response to a tweet, the referential function is not relevant; the emotive and conative functions prevail. The political strategies aimed at corroborating or refuting claims in terms of “fact checking” seem not relevant to these forms of communication based on personal enunciation, on the relation between the two simulacra “me” and “you”, and on the manifestation of one's own comment with respect to a topic. Furthermore, the techniques aimed at detecting the presence of hate speeches to apply, possibly, a precautionary censorship are lexical-sensitive, and fail to consider the context in which words co-occur. Finally, the paper presents a technique of analysis based on quantum information retrieval which can provide new insights on the relation between hashtag, address sign, topic, and reply.