Previous research has shown that individuals prefer performing mathematical operations that are aligned with the relationship between objects in a problem, a phenomenon that Bassok calls semantic alignment. For example, objects sharing a taxonomic relationship (tulips and daisies) tend to elicit addition. Our study tests the extent to which these processes occur automatically, through the use of a speeded processing task. Participants are presented with simple mathematical problems, either addition or division, in which the operands bare verbal labels of objects which share a semantic relationship, either taxonomic or thematic. We examine the latencies and error rates for answer verification. Our prediction is that semantically aligned problems (e.g. 12 tulips + 4 daisies) will show lower latencies and error rates than semantically misaligned problems (e.g. 12 tulips + 4 vases). .