This article examines the use of literary fiction as a source of data for analysing linguistic patterns in social contexts. By highlighting two interconnected methodological challenges in sociolinguistic studies of literary fiction, this study aims to foster interdisciplinary dialogue on study designs and methods. Building on previous unresolved discussions, it is argued that, firstly, studies that are limited to identifying a fictional text's mimetic function and drawing real-world-related conclusions from that text risk circular reasoning, thereby contributing little new knowledge. Secondly, the structural unreliability and ambiguity of fictional dialogue often make it problematic to incorporate into sociolinguistic analyses unless its specific conditions are considered. Using scholarly articles as examples, the discussion addresses these challenges to better integrate literary studies and sociolinguistics.