This article proposes a reflection on a specific way of envisioning and valorising the scholarly contribution of scientific software, namely by making explicit the model of data analysis that underlies it. It seeks to illustrate this way of studying a software construct by applying it to a particular text analysis program. Fundamental aspects of this program's design (input and output, data structures, process model, and user interface) are reviewed and discussed from the point of view of their implications in terms of theoretical commitments to a specific conception of text and text analysis. The conclusions of this case study notably emphasise the central role of user modelling in the assessment of scientific software's epistemological contribution as well as the necessity of extending the proposed approach to a broader range of software applications.