Data and Models for Understanding andGenerating Stylistic Variation in Language.

Stylistic choices are a fundamental aspect of verbal communication both from the language understanding and from the language production points of view. The goal of this thesis is twofold. From one side the focus is on the study of Natural Language Processing techniques that are able to capture stylistic variations in texts. On the other side, the thesis investigates how those techniques can be applied to assess the capability of the Natural Language Generation System to reproduce such stylistic variations in generated texts.