Language is at the epicenter of our existence, allowing us to communicate and think about complex ideas, feelings, and anything else we want. While language is often regarded as an isolated, unique phenomenon when looking across taxa, it is important to realize that this apparent detachment from the rest of cognition is illusory – our human language abilities are in fact highly dependent upon more basic cognitive processes. Much of this dissertation focuses on the ways in which the process of statistical learning both allows for and constrains language learning and processing. Statistical learning can be thought of as a cognitive process by which learners implicitly form associations between stimuli by tracking and storing the underlying statistical relationships between such elements. To provide insight into the relationship between statistical learning and language, two studies are reported here. I first demonstrate the reliability of paradigms that are frequently used to test this construct, while also showing how individual differences in statistical learning are correlated with biases in language processing. In the second study, I characterize how constraints on the input available to learners can affect their ability to acquire statistically learned grammatical regularities. I also establish that such knowledge is retained over time, by examining performance at a follow-up session two-weeks after training. The third study puts long-held assumptions about the modularity of the brain’s language network to the test by examining neuroplasticity in adult patients with brain tumors. The results of this study show that the right frontal lobe is capable of maintaining language function when there is damage to the left frontal lobe. Together, the findings reported within offer evidence for a language system that is highly sensitive to the distributional properties of the input, and is characterized by processes of entrenchment and plasticity.