The evolution of digital technologies has profoundly transformed the way we exchange political ideas. The rise of social networks, blogs, and online discussion forums has provided an increasingly accessible platform for political organizations to communicate and interact with voters. In order to explore words in digital transformation, and the use of tools and/or approaches for studying lexicon and phraseology in evolving discursive domains, we chose to focus our study on words in the political domain, in the context of the 2017 and 2022 French political elections. Thus, in the interval of two presidential campaigns, the political context has changed enormously (evolution of the French political landscape, recomposition of parties and the electorate), and corpus analysis technologies have also undergone a great evolution. The analysis of digital political discourse has become increasingly important due to the growing importance of the Internet and social media in public debate and opinion formation. New challenges in digital political discourse analysis include the sheer volume of data (which is often noisy and contains redundant or irrelevant information), polarization, misinformation, and the difficulty of distinguishing between facts, opinions, and rumors, especially in short messages. This paper will therefore address both the methodological and technological transformations, as well as the discursive and argumentative transformations, of the analysis of political words in an electoral context, through the presentation of different projects and initiatives that have marked the scientific landscape during the campaigns. Finally, to deepen this inventory, and to address the issue of lexicon and political discourse, we will focus on the theme of the analysis of the candidates' style, by highlighting the way in which the use of deep learning and textual statistics can help to better understand the evolution of political discourse, and to measure the contribution of recent technologies and tools mobilizing Artificial Intelligence.