Adults can use cross-situational information to learn words, but it is unclear how much information they retain about the potential referents that occur with a word on each observation. We tested this question using novel mouse-tracking and finger-tracking paradigms. Adults encountered novel words in ambiguous training trials and were then tested on the words’ referents. In some test trials, participants saw both the target and a high-probability competitor that had repeatedly occurred with the word. Participants’ mouse trajectories were slower, less accurate, and more complex when the competitor was present, indicating participants were aware that both the target and competitor had occurred with the word. This suggests that learners can retain multiple potential referents for a word and mouse tracking provides a promising way of assessing this knowledge. However, this knowledge was not evident in participants’ finger movements, suggesting that finger tracking might not capture real-time competition between referents.