The use of technologies in Humanities opens new research op-portunities as it allows the access to vast amounts of data such as textualcorpora. As, in the Digital Humanities domain, a considerable amount ofthe research is done on digitised corpora, Natural Language Processing toolscan be of much help in their exploitation for they help extracting linguisticinformation. We present a series of experiments in which we propose texttransformations to generate vocabulary learning exercises based on NaturalLanguage Processing. We describe the corpus, databases and tools we haveemployed in our approach and we offer an overview of a multilingual languageprocessing pipeline. Then we present the experiments and their output. Wefinally discuss the strengths and shortcomings of our approach