The works of Georges Baudoux contain a curious mix of fact and stereotype. His short stories have been read as real historical and ethnographical accounts and his characters, real people according to O'Reilly, express themselves in a language peculiar to each ethnic group - the at times slangy French of the whites, "canaque" French, bichelamar etc. Not without talent, Baudoux succeeds in transcribing different accents, different ways of speaking, different languages and even manages to reproduce in some way the oral nature of Kanak languages. This work of "retranscription" thus gives a certain linguistic authenticity to the text. Nevertheless, as Hollyman points out, Baudoux's Kanak French is stereotypical. Baudoux reports what he hears in the New Caledonian bush but his observations sometimes seem deformed by his prejudices and his own representations. Is this the case for the other languages present in the work of Baudoux ? "Sauvages et Civilises", a short story in which we find all the stereotypes associated with Kanaks that have been around in the colony since the publication of the writings of Cook and d'Entrecasteaux, gives us the opportunity to analyse linguistically the speech of two "others": Socrates, the Reunion Creole living in a Kanak tribe and the Vietnamese cook working for the whites on board their yacht. While we notice a good number of authentic traits in these representations, we also see some stereotypes. What motivates Baudoux's representations? Where, indeed, can we situate the genesis of his representations?