The ability to communicate through spoken language is not only one of the most complex cognitive abilities we possess, but it is also the aspect of the brain that makes us uniquely human. While people communicate almost constantly in daily life—telling stories over dinner, gossiping over coffee, and so much more—the majority of previous work has failed to study language in this same real-world context and instead focuses on the representation of single words and sentences in isolation. Accordingly, and in an effort to expand upon existing models of language neurobiology, this dissertation will present a series of experiments examining how the brain supports everyday pragmatic discourse. In these experiments, I use a patient-lesion model and study non-aphasic patients with behavioral variant frontotemporal dementia (bvFTD)—a rare neurodegenerative disease characterized by social-executive deficits and atrophy in frontal and anterior temporal cortices. In Chapters 2-3, I examine the cognitive and neural substrates of social coordination and referential communication—that is, how a speaker describes an object so that a listener can identify that object. Across both experiments, I find that impaired referential communication in bvFTD is related cognitively to mental flexibility and anatomically to a social-executive network including non-language regions in prefrontal cortex. In Chapters 4-5, I turn from language production to language comprehension, examining the cognitive and neural substrates of indirect reply and indirect request comprehension, respectively. In doing so, I examine how listeners make pragmatic, bridging inferences to derive a speaker’s true, intended meaning. Confirming my previous results, I find that patients with bvFTD struggle to interpret indirect speech due to social-executive deficits and degradation of a multimodal, extra-Sylvian network. The fifth and final experiment is more clinically-motivated, examining the progression of conversation difficulties in bvFTD and identifying potential prognostic markers. Here, I find that patients with poor executive function and focal disease in prefrontal cortex at baseline are likely to experience communication problems later in disease. Altogether, these findings help to define a more comprehensive model of language neurobiology that can account for the complexities of real-world communication from the perspective of both speakers and listeners.