Enunciative Praxis and Enregisterment in the Domain of Generative Artificial Intelligence

Abstract

This article focuses on Generative Artificial Intelligence (GenAI) in the area of text-to-image translation. I address the question of stereotyped visual styles using the theory of enunciative praxis formulated in continental semiotics (and, more specifically, in so-called Paris School Semiotics). I relate this theory with the concept of enregisterment in linguistic anthropology in order to describe the way in which visual forms undergo a process of sedimentation and typification and then are transposed to feed into other forms. The first section of the article presents the way in which a painting (as a stratified form of other forms), as well as a corpus of interrelated paintings, may be studied through the methodology of enunciative praxis and enregisterment; the second section develops the study of databases and algorithms as tools for a renewed approach to semiotic theory of enunciation and visual language.