This dissertation contains a collection of essays centered on the relationship between theoretical model-building and empirical evidence-gathering in linguistics and related language sciences. The first chapter sets the stage by demonstrating that the subject matter of linguistics is manifold, and contending that discussion of relationships between linguistic models, evidence, and language itself depends on the subject matter at hand. The second chapter defends a restrictive account of scientific evidence. I make use of this account in the third chapter, in which I argue that if my account of scientific evidence is correct, then linguistic intuitions do not generally qualify as scientific evidence. Drawing on both extant and original empirical work on linguistic intuitions, I explore the consequences of this conclusion for scientific practice. In the fourth and fifth chapters I examine two distinct ways in which theoretical models relate to the evidence. Chapter four looks at the way in which empirical evidence can support computer simulations in evolutionary linguistics by informing and constraining them. Chapter five, on the other hand, probes the limits of how models are constrained by the data, taking as a case study empirically-suspect but theoretically-useful intentionalist models of meaning.