Triangulating Experiments in an Industrial Setting to Evaluate Preferred Representational Systems of Software Developers

Abstract

Software development results in historic data. In recent years, researchers have been conducting linguistic analyses of mailing lists to understand the intricacies of software development. A new approach for that is to use Neurolinguistic Theory (NT). NT postulates that each person uses a preferred representational cognitive system (PRS) in each specific context. Thus, although each software developer uses a variety of representational systems to understand software, each has his or her own preferred representational system. In this paper, we analyze a psychometrically-based neurolinguistic method to classify the PRSs of software developers in an industrial setting. Our experimental evaluation of the approach is carried out in three combined experiments: (1) a study assesses the PRSs of industrial developers using a project's mailing lists; (2) a survey of software developers to establish their PRSs; and (3) a controlled experiment that carefully analyzes and visualizes strategies to perform software comprehension activities supported by a visualization tool. The results indicate that our approach can indeed be used to determine software developers' PRS.