Annotation de la continuité référentielle dans un corpus scolaire – premiers résultats

The Scolinter research project investigates the writing proficiency of primary school students in France, Italy, and Spain. The eponymous corpus consists of a large longitudinal corpus of comparable children’s writing in these three languages (Ponton et al., 2021). The aim of this project is to create a NLP tool tailored to this corpus to assist researchers in the linguistic description of phenomena relating to cohesion and textual coherence, in particular referential continuity. To meet these goals, the first stage of this research consists in designing a model and choosing an annotation format meeting these goals. This paper reviews the research on anaphora, coreference and referential continuity, presents the specific features of the Scolinter corpus, suggesting methodological pathways for further work.