Argumentation schemes, fallacies, and evidence in politicians’ argumentative tweets—A coded dataset

This coded database presents a corpus of argumentative tweets published by four politicians (Matteo Salvini, Donald Trump, Jair Bolsonaro, and Joe Biden) within 6 months from their taking office, which corresponds to the official end of their election campaign. The coding is based on a threefold method of analysis based on the instruments of argumentation theory and pragmatics. First, the types of arguments are recognized and classified according to a systematic organization of the argumentation schemes developed in the literature. Second, arguments are evaluated considering the fallacies committed. Third, the uses and misuses of “emotive words” are assessed. Based on this theoretical framework, each tweet is thus attributed three categories of codes: 1) argument types (maximum two, corresponding to the most important ones); 2) fallacies (maximum two types of fallacies, plus a distinct indication of the lack of necessary evidence or false presupposition); and 3) emotive language (maximum three emotive words, plus the most important emotion expressed). A total of 2657 tweets are coded, providing a ground for comparative works and an instrument for training further coding of different corpora.