ABSTRACT
In this chapter, the authors present correspondence analysis (CA) for the analysis of a large number of very short documents which they apply to a free text answers example, analyzed without aggregation. Textual CA can be performed on the source documents or on the aggregate ones, starting respectively from a lexical table (LT) or an aggregate LT. In the CA setting, the differences between these two types of analysis derive mainly from the fact that the source documents are simultaneously more homogeneous and more different to one another than the aggregate documents are, leading to higher cloud inertia. CA shows that they are almost exclusively used by men or women, given their extreme and opposing positions on the second axis. The authors aim to study variability in primary vital concerns, as related to gender and education. They show that the highest values are for the direct analyses.
