Autonomous recovery from programming errors made by primary school children

Abstract

Programming classes offer unique opportunities to learn both semantic and syntactic precision, even for primary school children without prior knowledge in computer science. In order to make students progress autonomously, programming languages and environments need to be chosen with care to their didactic quality. This paper introduces four classes covering the majority of what we call structural programming errors. These mistakes are either syntactical errors or the result of invocations that do not match the signature of any user-defined command, and therefore prevent the execution of a program. Furthermore, we present a methodology that allows for detecting as many structural programming errors as possible, and show how we integrated this methodology in our Logo programming environment for primary schools. Finally, we reflect on an evaluation we carried out at school that confirms the didactic benefits of the chosen approach.