Institutionality of discourse, as one of the key concepts of modern discourse analysis and an integral property of a number of discourse types, often poses a research problem of practical and/or methodological character. The dynamics of discourse and its social nature do not always fit into the framework of institutionality, extending it or even transgressing it. Apart from general methodological problems, the researcher often encounters cases of deliberate violation of written and unwritten rules of communication. Such examples, however, not only represent an issue, but also grant the possibility of looking at the problem of institutionality from a different angle. The article considers the problems of the methodology of institutional discourse associated, on the one hand, with the uncertainty (or an insufficiently clear definition) of a number of concepts and types of discourse, and on the other hand, with shifts of status and roles of communicants. It might seem that such shifts make the principle of institutionality rather volatile. Our study shows, however, that functionally all the institutional properties of discourse stand and even serve as a tool for manipulation. We consider examples of such violations that present a methodological issue and hamper practical discourse research. It is shown that a formal classification of discourse types via corresponding social institutions, despite its simplicity, tends to produce a number of issues including the overlapping of several types of discourse in one discursive act or difficulties in attributing it to any particular type due to changes in the communicative tactics of participants.