Applying theory-building techniques to the design of modelling languages

In their 2004 paper Hevner et. al proposed a set of guidelines for conducting design science research projects in the IS discipline. While useful, these guidelines have a relatively high level of abstraction. However, various IT artifacts such as models, methods, techniques and implementations require IS researchers to apply differing methods in order to construct and evaluate purposeful artifacts respectively. In this paper we discuss a particular class of IT artifacts: conceptual modeling languages. As constituent parts of software development methods, a multitude of such languages has been proposed and discussed. Yet, in the related literature on method design only little guidance is provided on how to derive appropriate conceptual modeling languages from empirical data. We believe that “good methods” need to be rigorously grounded in empirical findings. Taking a look at the related literature on inductive theory building reveals that at there are prominent similarities between the elements that constitute theories and those that constitute conceptual modeling languages: whereas theories comprise of constructs and relationships between these, conceptual modeling languages comprise of language constructs and relationships among these. We draw from the body of literature on grounded theory building and propose a new approach to designing conceptual modeling languages.