In this dissertation, I explored the semantic processing of a construction that has been largely uninvestigated, namely, the nested structure of two epistemic modals in a single clause, as illustrated in the sentence “He may certainly have forgotten” (Lyon, 1977). There are two different theoretical approaches that may account for the processing of this structure. The formal semantics account (Lyons, 1977; Potsdam, 1998; Moss, 2015) claims that the meaning of the inner modal should be interpreted within the scope of the outer modal. Based on this account, if the first and second modals switch their positions, a change in meaning should be expected. This account is also referred to as the “scope account”. In contrast, a good-enough processing account (Ferreira & Lowder, 2016) predicts that the scope of nested modals may not be thoroughly processed, and thus, the order of the modals may not change interlocutors’ interpretations of the nested expression. Using a combination of Bayesian modeling and judgment tasks, in six experiments I evaluated these two theoretical perspectives, and the result suggested a holistic processing mechanism in line with the good-enough processing framework. This dissertation consists of five chapters. The first chapter introduces the research questions and outlines the structure of this dissertation. The second chapter provides an overview of the research background where my dissertation project is situated. The third chapter reports six experiments I conducted examining how interlocutors process nested epistemic expressions in casual conversations, focusing on the extent to which the prediction of the scope account matches the patterns observed in the experiments. The fourth chapter discusses the major findings that have been consistently replicated in the six experiments, and proposes a possible account for the cognitive mechanism underlying the processing of nested epistemic expressions. The last chapter concluded this dissertation, summarizing the answers to the research questions.