This paper starts from the observation that in discourse analysis, unlike in other fields of linguistics, very few studies combine corpus-based and experimental data. It seeks to understand the reasons behind this methodological monism by taking into account the specificities of the discourse level and the benefits and drawbacks of each method. In the first part of the paper, challenges facing the combination of corpora and experimentation are developed. In the second part, a selection of mixed-method discourse-analytic studies is reviewed to illustrate the different ways in which corpora and experiments can interact, how much convergence (or lack thereof) can be found across production and comprehension paradigms, and the technical and methodological solutions that researchers have found to make the data as comparable as possible. This structured literature review concludes with the recommendation that discourse analysis can benefit from more mixed-method approaches, provided that the methods are planned with the respective affordances of each method in mind.