Project NUMTEST: Assessing basic number competence without language

Although numerical skills are essential in modern societies, 5- 7 % of the population suffer from mathematical learning disabilities. Due to the hierarchical nature of mathematical knowledge, screening during the earliest stages of learning is essential to intervene efficiently. While different screening tools exist, they rely on verbal instructions and task content, hampering their usefulness in linguistically heterogeneous young school populations. We developed a computerized task that requires subjects to encode both auditory and visual numerical information to successfully respond to the task’s demands. For task instruction, participants were presented a video showing a person correctly solving three easy items of the task, before moving on to a practice session of three different items. If any item was solved incorrectly, the entire practice session was repeated for the participant. We administered the task to a sample of first grade students and collected participant’s performance in standardized addition, subtraction and number comparison (1 & 2-digit) tasks. A multivariate analysis with practice repetition as between-subject factor on the four standardized control measures revealed that participants that repeated the practice session scored significantly lower in three out of four dependent measures. These results suggest that the immediate understanding (i.e. practice items solved correctly on the first try) of our task, requiring deductive reasoning and an abstract, format-independent representation of quantity, is able to differentiate between high and low performers on standardized measures of basic math competence non-verbally at an early stage of learning. Implications and limitations will be discussed.