Statistical learning—sensitivity to distributional patterns in the world—has been implicated in many aspects of behavior and cognition, particularly within the study of language. In language learning, learners engage with their environment, incorporating cues from different sources. However, statistical learning experiments are generally based on passive exposure to simple input streams where many of the cues and features that are part of real-world language learning are stripped away. Little is therefore known about the time-course of statistical learning, including what information learners use and when. This dissertation investigates the statistical learning of linguistic structure under more complex circumstances, involving the simultaneous learning of word-referent mappings and language-like syntactic regularities when provided with different kinds of feedback. The first paper demonstrates that feedback not only enhances statistical learning but may even be necessary to resolve the degree of ambiguity and complexity that learners face when learning grammatical structure that better approximates some of the syntactic complexities in natural language. The second paper utilizes eye tracking to examine how feedback affects learning, uncovering not only what information learners use (e.g., syntactic, semantic, or interactional cues), but also how the use of such information changes across the learning trajectory. The final paper asks how the learners’ ability to pick up on and use case marking cues to disambiguate the meaning of sentences in a verb-initial artificial flexible-word-order language is impacted by feedback, their first language, and case marker saliency. Together, the studies in this dissertation reveal novel insights into statistical learning and its constraints, highlighting the fundamental role of cues in language learning.