The role of interrogative sentences in political argumentation remains largely unexplored. This study addresses this gap by introducing a new Polish-language dataset featuring diverse examples of interrogative sentences in political discourse (election debates). The dataset serves as a unique resource for theoretical research in Argumentation Mining and Natural Language Inference through the annotation of ⟨IS, C⟩ and ⟨IS, P⟩ pairs, where IS denotes an interrogative sentence, C represents its corresponding conclusion, and P indicates a premise. The annotations primarily capture implicitly expressed argumentative structures and can serve as a benchmark for large language models (LLMs), particularly those trained on Polish-language data. Furthermore, this is the first study in Argumentation Mining where annotators independently verbalize the content of conclusions and premises conveyed through speech acts constructed with interrogative sentences. Our findings reveal that interrogative sentences in political debates most frequently function as implicature (approx. 45%), normative propositions (approx. 31%), statements expressing epistemic states (approx. 20%), and presuppositions (approx. 4%). Semantic similarity analysis confirms that annotators achieve a high level of consistency in identifying and verbalizing the content implied by interrogative sentences. The dataset provides a robust foundation for developing advanced language models and for further research into the role of interrogative sentences in political discourse.

A well-known ambiguity in the term ‘argument’ is that of argument as an inferential structure and argument as a kind of dialogue. In the first sense, an argument is a structure with a conclusion supported by one or more grounds, which may or may not be supported by further grounds. Rules for the construction and criteria for the quality of arguments in this sense are a matter of logic. In the second sense, arguments have been studied as a form of dialogical interaction, in which human or artificial agents aim to resolve a conflict of opinion by verbal means. Rules for conducting such dialogues and criteria for their quality are part of dialogue theory. Usually, formal accounts of argumentation dialogues in logic and artificial intelligence presuppose an argument-based logic. That is, the ways in which dialogue participants support and attack claims are modelled as the construction of explicit arguments and counterarguments (in the inferential sense). However, in this paper formal models of argumentation dialogues are discussed that do not presuppose arguments as inferential structures. The motivation for such models is that there are forms of inference that are not most naturally cast in the form of arguments (such as abduction, statistical reasoning and coherence-based reasoning) but that can still be the subject of argumentative dialogue. Some recent work in artificial intelligence is discussed which embeds non-argumentative inference in an argumentative dialogue system, and some general observations are drawn from this discussion.