In the vast literature on evidentiality, empirical and theoretical focus has mostly been on propositional evidentiality, i.e. evidentials that scope over propositions. In this work, we undertake a crosslinguistic comparative study of propositional and nominal evidentiality, i.e. evidentials that scope over nominals, and are fused with the determiner/demonstrative systems or with nominal tense markers. I demonstrate that there are cohesive parallels in how flavors of both propositional and nonpropositional evidentiality interact with verbal and nominal tense and aspect. I use tools from modal logic to show that we can (i) unify the subdomains of evidentiality using modal accessibility relations while also preserving important distinctions between them, (ii) use the same tools to compositionally capture the interaction between evidentials and tense and aspect, and (iii) have the representation of an agent’s certainty of belief be reflected in quantificational force. Concretely, directly encoding the subtype of evidence in the semantics, I argue that three distinct evidential flavors embody three distinct spatio-temporal modal accessibility relations: direct (sensory) evidentials are temporally-sensitive historical necessity relations (yielding the factive nature of perception); inferential evidentials of pure reasoning are epistemic accessibility relations; inferential evidentials of results are a combination of the above two.

Chapter 6 modifies the view of Chapter 5 in response to a variety of objections. This modified view can account for difficulties arising from the relation between will-claims and might-claims.