Abstract:
Understanding of programmers’ attention provides benefits for developing comprehension models, bugprediction models, for increasing software productivity, and facilitating programming education activities. Here we conduct a gaze-contingent study involving a real-time restriction of the viewing area and compare professionals’ and novices’ verbal feedback after the parafoveal view was restricted during source-code comprehension. Such information provides clues about the differences in parafoveal processing during programming.
We recorded the participants’ verbal feedback and divided their answers into seven topics and types. Then we analysed the differences between the answers given by the experts and the novices. We compared the proportion of utterances used to comment upon a certain topic by each group. This allowed to identify the relative importance of a particular topic. Some topics turned out equally important for both the experts and the novices: 1. analysis of the working process, 2. personal evaluation of the source code, and 3. the use of life-hacks. The experts and the novices used a different proportion of utterances to comment upon the experimental conditions and visualizing, which was unexpected. The restriction of the extrafoveal area evoked a more emotional response from the expert programmers. At the same time, the novices perceived the restriction of semantic information in the extrafoveal area in a less emotional way. We suggest that the explanation is that the experts have increased expectations for the information to be obtained from the attentional objects located in the extrafoveal area. The masking of parafoveal objects makes them less easily available, which runs counter to the experts’ expectations and, therefore, produces a stronger emotional feedback. This illustrates that it is a common thing for expert programmers to use the parafoveal information during source code comprehension.
We recorded the participants’ verbal feedback and divided their answers into seven topics and types. Then we analysed the differences between the answers given by the experts and the novices. We compared the proportion of utterances used to comment upon a certain topic by each group. This allowed to identify the relative importance of a particular topic. Some topics turned out equally important for both the experts and the novices: 1. analysis of the working process, 2. personal evaluation of the source code, and 3. the use of life-hacks. The experts and the novices used a different proportion of utterances to comment upon the experimental conditions and visualizing, which was unexpected. The restriction of the extrafoveal area evoked a more emotional response from the expert programmers. At the same time, the novices perceived the restriction of semantic information in the extrafoveal area in a less emotional way. We suggest that the explanation is that the experts have increased expectations for the information to be obtained from the attentional objects located in the extrafoveal area. The masking of parafoveal objects makes them less easily available, which runs counter to the experts’ expectations and, therefore, produces a stronger emotional feedback. This illustrates that it is a common thing for expert programmers to use the parafoveal information during source code comprehension.