Usages linguistiques des éléments supplémentaires dans l'Analyse factorielle des correspondances

Description

This contribution shows the interest of supplementary variables with Correspondence Analysis (CA). From a CA vector space crossing the main morpho-syntactic categories and the French Presidents of the fifth Republic, we project the lemma "indiquer" (to indicate) as a supplementary variable. Will it be located in the "verb subspace" of the graph? And if not, what should we conclude from this counterintuitive positioning? Beyond this example, it is the linguistic homogeneity of the rows of the contingency table (words, lemmas, grammatical categories, etc.) that we question, by projecting, in more or less interpretable ways, other linguistic elements into additional elements. The contribution ends with an open discussion on the complementarity of CA and deep neural networks.

Abstract

International audience

Additional details