Analyse du corpus ORTHOTEL : apport du traitement automatique à la classification des déviations orthographiques
Véronique Aubergé, Nadia Ghneim, Rabia Belrhali · Langue française · 1999
Véronique Aubergé, Nadia Ghneim, Rabia Belrhali. Analyse du corpus ORTHOTEL : apport du traitement automatique à la classification des déviations orthographiques The aim of this study is to present organized statistical data extracted from a large corpus of 15 000 forms showing spelling errors. This corpus, ORTHOTEL, is the result of Minitel users wondering about word spelling. An automatic treatment has been applied to the corpus to separate and analyse erors. Half of the forms of the corpus are rightly spelled. It indicates the users' degree of linguistic insecurity. An automatic text-to-phone system applied on the badly spelled words shows that a great part are homophone to a correct word taken from a reference lexicon of 80 000 canonical forms. An alignment algorithm has classified the orthographic transformations which account for deviations from the reference lexicon .