The study of language typology has played a critical role in revealing potential constraints on possible linguistic systems. Such constraints, often called “typological universals” have long been used to support a foundational premise of generative linguistics—that languages share a set of underlying commonalities. However, recent research has challenged the idea that typological universals—which (at least on the surface) are often statistical tendencies rather than absolute laws—reflect meaningful biases in the linguistic or cognitive system. In part as a response to these critiques, new behavior methods have been developed to probe the link between recurrent typological patterns and the linguistic (and broader cognitive) system. This article focuses on the novel findings which have resulted from this trend, in particular those which use Artificial Language Learning (ALL) paradigms. This exciting strand of research suggests the viability of experimental methods for investigating constraints on human language, and points to new ways of gaining traction on critical questions in cognitive science.