Influence of semantic alignment in word-problem solving: An eye-tracking study in adults [Author Accepted Manuscript]

Word-problem solving presents a significant challenge for many children and adults, as it requires both linguistic and numerical abilities. According to the SEmantic COngruence model (Gros et al., 2020a), individuals’ world knowledge influences their ability to solve mathematical word problems. Specifically, the semantic relation between two quantities can prompt solvers to infer a particular mathematical structure, which may affect their problem-solving success. The current study investigated the cognitive load associated with semantic alignment during adults’ word-problem solving using eye tracking. Participants solved additive compare problems in which semantic alignment was manipulated. They encountered unaligned problems involving asymmetrically related quantities (e.g., tulips and vases), aligned problems with symmetrically related quantities (e.g., tulips and roses), and control problems using the same quantities (e.g., tulips and tulips). We recorded two outcomes: (1) the number of revisits on the referent and the compared quantities, and (2) the total number of saccades. Results showed that the number of revisits and saccades was higher for unaligned problems. This suggests that unaligned additive word problems demand greater cognitive effort, likely because adults must reread the referent and compared quantities to understand and inhibit their relation, which could lead them to create an inadequate mathematical structure, impeding word-problem solving. These findings demonstrate that semantic alignment influences adults’ word-problem solving. The theoretical and practical implications of this influence are discussed.