Performing mental arithmetic tasks in real classrooms: Does performance depend on the type of background noise and task difficulty?

International audienceA clear and effortless oral communication underlies a successful teaching-learning process. Too long reverberation times and/or high levels of background noise are known to impact on student?s speech reception and in turn on their academic attainments. The sound environment affects verbal and non-verbal tasks in a different way, and the effect of background noise is further mediated by its spectral and temporal characteristics, by the student?s abilities and by the task characteristics. Whereas a plurality of studies dealt with verbal tasks in the school setting, only a few examined the impact of a realistic sound environment on arithmetic, in spite of its relevance in school education. In this study, performance in a mental calculation task was explored, evaluating both response accuracy and response time (RT) the latter informative on the allocation of mental resources (effort) during the task. Normal hearing children (N=159, 11 to 13 years) were aurally presented with mental calculations (addition and subtraction), with two levels of difficulty (low vs high). Three listening conditions were selected: quiet (signal-to-noise ratio SNR >15 dB), traffic noise (SNR=0 dB), babble noise plus typical classroom sound events (classrooms noise, SNR=0dB). The task was presented in the children?s classroom (reverberation time: 0.68 s), via loudspeakers and tablets in one hour sessions. A significant interaction between children?s age and listening condition was found for accuracy alone, indicating a developmental effect in how children coped with noisy conditions. In addition, the results showed for both dependent variables a significant interaction between listening condition and task difficulty (low difficulty: most detrimental impact of classroom noise; high difficulty: no difference between listening conditions). This might suggest that different processing strategies were implemented for task-irrelevant stimuli, depending on the task load