For many years now, early word learning in children has been an important subject among many researchers. There are many ways in which children learn word-object pairings including using co-occurrences, forwards integration, and backwards integration. This study primarily focuses on backwards integration. Backwards integration entails using learned information to be able to recall a word-object pairing from a previous time. In this thesis, three different studies were conducted with children aged 3-7 years old. In the general task, children were presented with a computerized word-learning task in which they could track word-referent pairings using co-occurrence statistics, forward integration, and backward integration. The goal of Study 1 and Study 2 was to determine the best task design to study backwards integration. The goal of the final study, Study 3, was to provide preliminary data on backwards integration. The overall results indicate that a between subjects design is the most beneficial way to test backwards integration because as a group, children were learning when compared to chance. In addition, the results from Study 3 showed that children were not learning in the task. In general, this suggests that this task may have been very difficult for children to complete. One limitation of Study 3
was that there was a small sample size of only 29 children. In order to account for this, the sample sizes in Study 2 and Study 3 were combined. This combined data did show that children succeeded at the backwards integration condition. It is noteworthy to mention that backwards integration was above chance in Study 2 and in the Study 2 and 3 combination. Therefore, the overall results suggest that children may possibly be able to backwards integrate; however, no evidence of learning in any of the other conditions were present.