Cross-Situational Statistical Learning (CSSL) is an example of statistical learning where learners gradually learn word-object-associations based on the co-occurrence of word and referent. In this paradigm, participants are exposed to multiple situations where a word and a referent always co-occur while competitors vary (Yu & Smith, 2007). Even as every situation by itself is ambiguous, the correct mappings can be extracted by combining information across time. For some theories, during this process, multiple word-object-mappings are acquired and compete during the acquisition (McMurray et al., 2012). Currently, it is not well-understood how CSSL is affected by multiple simultaneously acquired overlapping mappings. Thus, we will investigate online competition between referents via eye-tracking during CSSL. Participants will acquire 1:1 (one word to one object) and 1:2 (one word to two objects, interlingual homographs) mappings in Experiment 1. We will track their eye movements, and we will analyse them using a four points logistic. This will allow us to estimate how competition impacts specific timing components of referent activation. We predict that 1:2 mappings will require more competition; therefore, looks to their targets will have a delayed fixation and a shallower slope. In Experiment 2, participants will acquire only 1:2 mappings. The experiment aims to examine the competition between the two existing meanings. If participants maintain multiple mappings, they will look to the second referent more than at baseline. We also predict that online competition will progressively increase.