The construction of languages has always been related to linguistics. Most of these initiatives address real scientific questions but from a non-academic point of view. The fact that Ferdinand de Saussure’s own brother, René de Saussure, wrote a theoretical essay on the construction of the Esperanto word (de Saussure, 1914) is an amusing illustration of this. In this paper, we propose a method inspired by experimental archaeology. The experiment consists in trying to obtain an artefact similar to the one observed using this or that construction method. An equivalent approach in linguistics would be the generation of linguistic systems based on explicitly formulated principles. Trying to generate similar systems pushes the linguist to explicitly define the principles that are needed and to explore all their consequences. In this context, we show that the use of notions induced by the observation of natural languages leads to a certain degree of circularity and that it is therefore more interesting to explore a priori principles based on very general assumptions.