1 Alternatives to QUD. Alternatives to questions.  Pavel Ozerov University of Münster Abstract The paper critically examines some central principles of the Question Under Discussion (UD) framework and ultimately explores the concept of ‘question’, central to QUD-models. It demonstrates how fine-grained, interactionally informed analyses of language-specific categories can reveal building blocks of interaction and explain the sources of the observed information- and discourse-structuring interpretations (such as update, contrast and more). Employing data from Anal Naga (Trans-Himalayan, India), it proceeds to a fine-grained analysis of the notion of ‘question’. The decomposition of ‘questions’ into smaller building blocks similarly reveals how diverse categories and discourse processes can trigger the interpretation of an information request. These findings and additional theoretical arguments suggest that QUD-models are problematic for various reasons: such models are non-parsimonious as they add superfluous extra layers to explain the observations; the explanatory apparatus is circular, as the extra layers are derived from within the explananda but regarded as underlying explanatory factors; and the models gloss over the actual factors by channelling them into cover terms prematurely regarded as primitive. Finally, since ‘question’ does not constitute a primitive concept but is a product of diverse discourse processes, discourse cannot be modelled on this foundation. 1 Introduction The Question Under Discussion (QUD) framework has gained ground in the current study of Information Structure (IS) and discourse. It is regarded as paving the way for an integrated pragmatic theory of discourse (Roberts 2012) and bridging IS and discourse structure (Beaver et al. 2017). It is also used for a universal definition of IS-concepts (Riester et al. 2018) and for an analysis of multiple IS- and discourse phenomena (Benz & Jasinskaja 2017; Clifton & Frazier 2018). Although originally this framework limited its study to idealised and largely unnatural tokens of conversation, its implementation has recently been expanded to naturally occurring data (Riester et al. 2018, and contributions to this special issue). However, as this paper aims to show, such attempts remain uninformed by findings of interactional studies which directly examine information- and discourse-structuring processes in natural interaction. This paper demonstrates how fine-grained interactionally-informed analyses of diverse language- specific categories and discourse strategies reveal the broad array of factors that underlie IS and discourse phenomena. These analyses and additional theoretical considerations suggest that the QUD-framework and  The research on the Anal Naga language was supported by ELDP (grant SG0428).