Search

This dissertation is a case study of two grammatical categories, evidentiality and mood. I argue that evidentiality and mood are grammatical expressions of epistemic modality and have an epistemic modal component as part of their meanings. While the empirical foundation for this work is data from Bulgarian, my analysis has a number of empirical and theoretical consequences for the previous work on evidentiality and mood in the formal semantics literature. Evidentiality is traditionally analyzed as a grammatical category that encodes information sources (Aikhenvald 2004). I show that the Bulgarian evidential has richer meaning: not only does it express information source, but also it has a temporal and a modal component. With respect to the information source, the Bulgarian evidential is compatible with a variety of evidential meanings, i.e. direct, inferential, and reportative, as long as the speaker has concrete perceivable evidence (as opposed to evidence based on a mental activity). With respect to epistemic commitment, the construction has different felicity conditions depending on the context: the speaker must be committed to the truth of the proposition in the scope of the evidential in a direct/inferential evidential context, but not in a reportative context. Finally, the distribution of the evidential is sensitive to the temporal relations specified in the context. In the previous literature, the Bulgarian evidential is analyzed as encoding indirect sources of information; no mention is made of its temporal meaning (Izvorski 1997, Sauerland and Schenner 2007). I propose a uniform semantic analysis of the Bulgarian evidential, which incorporates both a temporal and a modal component, and accounts for the full range of evidential meanings. The central aspect of the analysis is the assumption that the proposition in the scope of the evidential is evaluated with respect to different sets of worlds, depending on the discourse context: the belief worlds of the speaker in inferential/direct evidential contexts, and the belief worlds of the original reporter in reportative contexts. My analysis of mood explains the distribution of the subjunctive and the indicative in Bulgarian as being dependent on the epistemic commitment of the attitude holder (cf. Giannakidou 1998). Previous analyses attribute mood distribution to semantic properties of the selecting verb alone (cf. Farkas 1992, Villalta 2008). These analyses cannot be extended to Bulgarian, where the distribution of mood is sensitive not only to the semantics of matrix verbs, but also to context. This is particularly clear in cases when the same verb can select both the subjunctive and the indicative, and the choice of mood correlates with the attitude holder’s epistemic commitment. The indicative is selected iff the attitude holder is strongly committed to the truth/falsity of the proposition expressed by the complement clause. The subjunctive is selected iff the attitude holder has a weaker epistemic commitment. My formal analysis uses the tools from the analysis of modals (Kratzer 1979) and specifies how the meaning of the matrix propositional attitude verb interacts with the meaning of mood in the embedded clause. This is the first formal semantic analysis of mood in Bulgarian.