Angle is an important concept in geometry. Young children have difficulty separating angle size from other dimensions such as the length of angle sides, perhaps due to whole-object bias in word learning. The present study used the pre-test–training–post-test design to investigate the effectiveness of two ways of separating angle from angle size in 3–6-year-old Chinese preschoolers. A total of 228 children were given a pre-test and 219 of them failed the crucial test. 168 of the 219 children were present at school during the training phase and were randomly assigned to three groups: the “toma” group (n = 57), which received training to call the whole angle figure as “toma” and angle size as angle size; the “angle/angle size” group (n = 56), which received the training of separating “angle” from “angle size”; and the control group (n = 55), which used “angle size” alone to represent both the overall angle figure and angle size. Results showed that the “toma” group improved significantly more than the other two groups, the latter of which did not differ from each other. These results suggest that it is insufficient to have two separate words/phrases (angle and angle size) for children to learn to differentiate angle from angle size, perhaps due to their shared usage of the word angle. Instead, the use of a novel term is necessary and sufficient to improve learning. Implications for preschool education are discussed.