Source Code Comprehension: A Contemporary Definition and Conceptual Model for Empirical Investigation

Be it in debugging, testing, code review or, more recently, pair programming with AI assistance: in all these activities, software engineers need to understand source code. Accordingly, plenty of research is taking place in the field to find out, for example, what makes code easy to understand and which tools can best support developers in their comprehension process. And while any code comprehension researcher certainly has a rough idea of what they mean when they mention a developer having a good understanding of a piece of code, to date, the research community has not managed to define source code comprehension as a concept. Instead, in primary research on code comprehension, an implicit definition by task prevails, i.e., code comprehension is what the experimental tasks measure. This approach has two negative consequences. First, it makes it difficult to conduct secondary research. Currently, each code comprehension primary study uses different comprehension tasks and measures, and thus it is not clear whether different studies intend to measure the same construct. Second, authors of a primary study run into the difficulty of justifying their design decisions without a definition of what they attempt to measure. An operationalization of an insufficiently described construct occurs, which poses a threat to construct validity. The task of defining code comprehension considering the theory of the past fifty years is not an easy one. Nor is it a task that every author of a primary study must accomplish on their own. Therefore, this paper constitutes a reference work that defines source code comprehension and presents a conceptual framework in which researchers can anchor their empirical code comprehension research.