The following paper deals with the division of pragmatic labor between two types of wh-interrogatives in Marzahn German (MG). Use of the first type, marked by the enclitic particle n ([n-int]), is near obligatory for and confined to canonical, i.e., information-seeking question acts. The second type, lacking n ([∅-int]), has to be employed in non-canonical questions, such as rhetorical ones. This pattern of apparent markedness-reversal challenges the pretense-based approach to exam questions by Plunze and Zimmermann (2006) (Section 2) and plausibilizes an approach to information-seeking questions in terms of social cost in the sense of Levinson (2012) (Section 3.1). Overall empirical evidence, however, favors an account of n-marking as reinforcement of question act defaults in line with Farkas (2022) (Section 3.2). Section 5 offers a formulation of reinforcement in terms of the "table model" of discourse (Farkas 2022), such that the peculiar status of MG [n-int] follows from the prohibition of contextually overriding "basic conventional discourse effects". In the course of the above discussion, we will scrutinize different notions of interrogative sentential force (Sections 1, 2, 5), illustrate the form and workings of several types of non-canonical questions (guess, rhetorical, echo etc.), and analyze question use in the light of institutional settings and interpersonal effects (3.3).
