SCOLINTER: A trilingual corpus. The example of word segmentation [SCOLINTER: Un corpus trilingue. L'exemple de la segmentation en mots]

This article presents a longitudinal, trilingual (French, Spanish and Italian) school writing corpus. Its aim is to compare the evolution of literacy skills between these three languages in primary school. Actually, only CP (grade 1) has been entirely treated but the same pupils will be followed up to CM2 (grade 5). The exploitation of this corpus proposes a tool-based approach based on comparisons between the version of the texts produced by the pupils and a so-called ‹standardised› version. This approach makes it possible to produce a linguistic description of the achievements and problems that pupils face in their texts. In particular, this article focus on segmentation between words, highlighting both the frequency of the phenomenon and the categories of words in which it occurs. The comparative analysis reveals the presence of similar characteristics in the trilingual corpus