Asking children to gesture while being taught a concept facilitates their learning. Here, we investigated whether children benefitted equally from producing gestures that reflected speech (speech-gesture matches) versus gestures that complemented speech (speech-gesture mismatches), when learning the concept of palindromes. As in previous studies, we compared the utility of each gesture strategy to a speech alone strategy. Because our task was heavily based on language ability, we also considered children's phonological competency as a predictor of success at posttest. Across conditions, children who had low phonological competence were equally likely to perform well at posttest. However, gesture use was predictive of learning for children with high phonological competence: Those who produced either gesture strategy during training were more likely to learn than children who used a speech alone strategy. These results suggest that educators should be encouraged to use either speech-gesture match or mismatch strategies to aid learners, but that gesture may be especially beneficial to children who possess basic skills related to the new concept, in this case, phonological competency. Results also suggest that there are differences between the cognitive effects of naturally produced speech-gesture matches and mismatches, and those that are scripted and taught to children.