A language and task-based taxonomy of programming environments

Standard desktop computer applications are becoming increasingly sophisticated, allowing competent users to gain greater levels of efficiency through powerful capabilities. Examples are wide ranging - from simple tailoring of Web-browsers so that frequently used links appear on the toolbar, to writing code to customize the behaviour of a word processor. These interface features promise benefits of end-user programming to those who can access them. To help people access these features, many programming environments have been developed to make programming more accessible. Despite this plethora of developmental research, there has been surprisingly little analytical work in characterising and comparing the capabilities of the environments. This paper presents a methodological analysis of programming environments - particularly educational ones - to shine light on how different languages are used to support three fundamental programming activities. The analysis reveals how programming environments use different languages for the three tasks. We discover that programming environments can be more usable by displaying multiple languages for program representation, but only displaying multiple representations risks creating three cognitive gulfs. To avoid these gulfs, our taxonomy describes why every representation must be editable. Our taxonomy also describes why all languages should be animated when a program is executed. We intend this taxonomy and analysis to be useful for anyone who is building or evaluating a programming environment.