The detection of hate speech and fake news in political discourse is at the same time a crucial necessity for democratic societies and a challenge for several areas of study. However, most of the studies have focused on what is explicitly stated: false article information, language expressing hatred, derogatory expressions. This paper argues that the explicit dimension of manipulation is only one – and the least problematic – of the risks of political discourse. The language of the unsaid is much more dangerous and incomparably more difficult to detect, hidden in different types of fallacies and inappropriate uses of emotive language. Through coding scheme developed by integrating instruments drawn from argumentation theory and pragmatics, a corpus of argumentative tweets published by 4 politicians (Matteo Salvini, Donald Trump, Jair Bolsonaro, and Joseph Biden) within 6 months from their taking office is analyzed, detecting the types of argument, the fallacies, and the uses and misuses of “emotive words.” This coding results in the argumentation profiles of the speakers, which are compared statistically to show their different implicit strategies and deceptive tactics.

在政治话语中发现仇恨言论和假新闻同时是民主社会的关键必要性，也是几个研究领域的挑战。然而，大多数研究都集中在明确陈述的内容上：虚假文章信息、表达仇恨的语言、贬义表达. 本文认为，操纵的明确维度只是政治话语风险的一个——也是问题最少的——。未说出口的语言更危险，也更难以发现，隐藏在不同类型的谬误和情感语言的不当使用中。通过整合论证理论和语用学工具开发的编码方案，分析了 4 位政治家（Matteo Salvini、Donald Trump、Jair Bolsonaro 和 Joseph Biden）上任后 6 个月内发表的论证推文语料库，检测类型论点，谬误以及“情感词”的使用和误用。这种编码会产生演讲者的论证概况，