Pointing to the weaknesses of elicited data use in cross-cultural research on linguistic politeness, tacitly based on a dyadic interaction model, this corpus-based comparative study of requests and proposals in British, German, and Polish variants of the reality show Big Brother, demonstrates how quantitative inquiry of spontaneous data, not confined to dyadic interactions, requires new categorizations of examined interaction parameters, and offers new insights into language-specific influence of contextual factors upon politeness strategies.