Among the all too rare theoretical research leading to a semantics of emotions, we find the « grammatical » approach elaborated by Ludwig Wittgenstein in his Philosophical Investigations (1953). Presently, it remains relatively unknown that the semantic analysis of emotions played the role of a paradigmatic example which produces the idea that ordinary language is made up of different “language games” governed by complex structures. Founded on a cross-cultural and cross-linguistic perspective which is based primarily on German and English, the theoretical model proposed by Wittgenstein explores – starting from a very broad range of emotions (an exclamation, a complaint, a scream, fear, pain, doubt, desire, distress…) – the specificity of their intonation, the choice of lexicon and the syntactic turns of phrase able to express them in well-defined surroundings. But Wittgensteinian thought, which contributed to what would become known as analytical philosophy, largely escaped the interest of linguists. Therefore, the aim of this communication is to do justice to the rich and intricate analysis that is concealed within this approach by showing that it allows us to perceive the semantics of emotions from a new angle. Thus, in order to highlight the way in which the expression of emotions constitutes a language game, I‘ll begin by adopting a general, theoretical point of view that will question the relationship between the semantics of emotions and learning, experience, ostensive explanation, informative content, the descriptive aspect and the cognitive dimension. Then, I shall undertake an analysis of the specific use of the lexicon and utterances used to signify emotions. This study will allow the emotional language game to become more concrete by clarifying the meaning of terms which occur and interfere within it. In addition, this examination will allow us to define the game’s parameters by pointing out certain semantic deviances (i.e. does it make sense to say “I doubt that I’m in pain”?). Finally, based on the analysis on the specific case of expressing pain, I ‘ll show that the originality and interest of Wittgenstein’s theoretical approach also lies in the inverse relationship he establishes between semantics and grammar. To give a name to an emotion amounts to preparing the grammar of this word, or to showing the place this new word will occupy, in such a way that it is up to semantics to structure grammar.