The use of data-driven learning for investigating homophony and mastery of French spelling

Homophonie et maîtrise du français écrit : apport de l’apprentissage sur corpus

Maï Leray (1) , Henry Tyne (1)

1 CRESEM - Centre de Recherche sur les Sociétés et Environnements en Méditerranées

Mastery of spelling in L1 French is notoriously complex, in particular due to morphographic difficulties and grammatical homophony. This study investigates the use of data-driven learning in French primary schools for the homophonous set /sE/ (cest, sest, ces, ses, sais/t). The results show that whereas initially there is no difference in comparison to traditional teaching (i.e. following pre-test, fewer errors are made in general at the time of learning by both experimental and control groups), there is less attrition for the data-driven learning group, with some spellings progressing even at the post-test stage. We suggest that data-driven learning may be used for developing spelling in L1 French as learners are exposed to enriched input. The question of how a predominantly L2-based method can be applied to L1 pedagogy is raised.

Mots clés: Data-driven learning; spelling; French; homophony

Domaines: Linguistique; Education

Maï Leray, Henry Tyne. Homophonie et maîtrise du français écrit : apport de l’apprentissage sur corpus. Linguistik Online, 2016, Corpus, grammaire et français langue étrangère : une concordance nécessaire, 78 (4), ⟨10.13092/lo.78.2954⟩. ⟨hal-01401956⟩