Correspondence Analyses to Understand the Structure of Texts: a First Step towards a Structure Model