De l'arabe standard vers l'arabe dialectal : projection de corpus et ressources linguistiques en vue du traitement automatique de l'oral dans les médias tunisiens

In this work, we focus on the problems of the automatic treatment of oral spoken in the Tunisian media. This oral is marked by the use of code-switching between the Modern Standard Arabic (MSA) and the Tunisian dialect (TD). Our goal is to build useful resources to learn language models that can be used in automatic speech recognition applications. As it is a variant of MSA, we describe in this paper an adjustment process of the MSA resources to the TD. A first evaluation in terms of lexical coverage and perplexity is presented. MOTS-CLÉS : corpus oral, dialecte tunisien, modèle de langue, ressources.