This dissertation investigates the semantics, pragmatics, and syntax of propositional attitude reports; in particular, how assertion and presupposition are reflected in these different parts of the grammar. At the core of the dissertation are factive attitude reports, involving predicates like know, discover, realize, resent, appreciate, and like. Since Stalnaker (1974), factivity is taken to encompass both the discourse status of the embedded proposition p as Common Ground and the projection of the inference that the speaker is committed to p from the scope of operators—in both cases, unlike asserted content. Syntactically, factivity and assertion are argued to provide the semantic-pragmatic underpinnings for a range of complementation patterns (Kiparsky and Kiparsky 1970, Hooper and Thompson 1973, Rizzi 1997, a.o.). The central contributions of the dissertation are: (i) demonstrating what precise dimensions of assertion and presupposition are reflected in the syntax and semantics of clausal embedding, and (ii) decomposing the classically multifaceted notion of factivity into a set of more specific theoretical notions; importantly, dissociating the discourse status of p and the projection-prone speaker commitment inference. We attribute the speaker commitment inference to a lexical presupposition of an evidential modal base that entails p. We argue that this evidential base is always anchored to a Judge, which, depending on the type of factive predicate, is bound by different individuals. In the case of doxastic factives like discover, the judge is bound by the speaker, whereas in the case of emotive factives like appreciate, it is bound by the attitude holder, and for fact that nominals, it is realized as an index on the noun. The discourse status of p, we attribute to a separate dimension of discourse new vs. Given content (in the sense of Schwarzschild 1999), which cross-cuts both factive and non-factive verbs. Among the predicates which treat their complements as Given, we differentiate between the requirement (of response predicates like accept and not say) that p has an antecedent in the discourse, and the requirement (of emotive factives like resent and appreciate) that the situation or individual providing the attitude holder’s evidential basis for p is contextually accessible. We further argue for a fundamental semantic distinction between primarily acquaintance-based predicates —which include both factives (evidentials) like discover and non-factives like fear— and fundamentally doxastic or epistemic predicates, like believe and trust. Making these distinctions allows us to account for a wide range of apparently connected, yet clearly disparate empirical phenomena, some of which represent open problems in the literature and some of which are new observations made in the dissertation. Importantly, we are able to capture: (i) the dissociation of the discourse status of p and the commitment-to-p inference in doxastic factives (Chapters 3 and 5); (ii) a number of asymmetries between doxastic and emotive factives regarding their apparent entailment properties, interactions with operators, and sensitivities to contextual effects (Chapter 5); (iii) variations in entailment and argument-structural patterns across verbs like know and believe (Chapter 4); and (iv) the distribution of a set of proposed syntactic correlates of assertion and presupposition; in particular, V-to-C movement, wh-extraction, and selection for DP vs. CP-complements (Chapters 2 and 3).