Combining cognitive, semiotic and discourse analysis to explore the power of notations in visual programming

Using game design and programming to foster computational thinking acquisition has proved to be a successful strategy in recent years. In previous research with AgentSheets, we concluded that the semiotic richness of this visual programming environment, specifically designed to support computational thinking acquisition, could be explored more extensively to the benefit of learners. In particular, we realized that there are some additional representations of AgentSheets games and simulations that are not presented as programming tools in the interface, and yet they communicate new relevant meanings to the users. This paper reports on research where we artificially introduced such representations in a small follow-up experiment with selected participants from our previous research experiment. Our goal was to investigate the impact of such additional representations on program comprehension and modification tasks. To this end we contrasted empirical evidence of their performance in the two tasks with their verbal account of experience with AgentSheets. We used a combination of discourse analysis and inspections using Semiotic Engineering methods and the Cognitive Dimensions of Notations framework. Our findings go in two directions. First, we observed that additional representations have allowed participants to expand and correct previous learning. Therefore such representations can support new teaching strategies in computational thinking acquisition programs with AgentSheets. Second, we learned that the combination of methods we used to analyze empirical data - discourse analysis with semiotic and cognitive inspection techniques - can be used systematically in other research contexts, holding the promise of insightful results.