Les noms de marque dans les unités phraséologiques : enjeux méthodologiques d’un corpus franco-espagnol non normé. Quelle place pour l’intelligence artificielle ?

Brands are part of our lexiculture and their names are constantly circulating in our everyday discourses. The phenomenon of genericization of brand names leads to the emergence of new phraseological units (PU) that speakers establish to codify meanings and achieve linguistic economy or to bring more expressiveness (such as avoir une tête de Playmobil in French and tener un pelo Pantene in Spanish).

To date, few researchers have focused on PU containing a brand name. The reason is undoubtedly the lack of available data, since the lexicographical treatment of these PU is almost nonexistent. These expressions mainly appear in oral communication or on social networks, grounds conducive to phraseological innovation.

This study aims to analyze the methodological challenges involved in the constitution of a non-standardized corpus of PU containing a brand name in French and Spanish. The orthotypographical variability of brand names, coupled with the lack of available resources, makes this corpus a challenging winning combo for the linguist. The use of artificial intelligence has been explored to extract these PU (semi-)automatically, but these language models reveal their limitations when applied to such heterogeneous, non-standardized corpora.