UC Santa Cruz

This dissertation proposes a novel approach to tracking not-at-issue contributions to discourse by bringing together several strands of research concerning evidentiality, illocutionary discourse particles, and speaker bias marking, all of which concern phenomena that indicate the relationships between speakers and the information they express in conversation. I argue that the overall discourse effects in each case should be captured in terms of commitments that are conditioned on various evidential bases. These bases derive from the speaker's private beliefs, his interlocutors' discourse commitments, and other contextually-rooted sources. They share the common purpose of publicizing the reliability of the commitments that invoke them.

The body of this dissertation provides arguments for this approach from several empirical domains. Chief among these are the sentence-final discourse particles of Japanese, and biased polar questions in English. For Japanese, I argue that the particles yo, ne, and their combination yone publicize the speaker's beliefs about his relative authority to sponsor the content of the particle-marked utterance, compared to that of his interlocutors. These simple conditions, encoded in the evidential base, interact with the default effects of an utterance to derive the total discourse effects of the particle-marked utterances, including the ways that these particles seem to limit possible felicitous responses, and why they are disallowed with certain sentence forms but not others. For English biased questions, I argue that high negation polar questions (HNPQs) and two kinds of tag questions (TQs) weakly commit the speaker to one or both answers, depending on the question's form. These weak commitments can be based on either a prior or current version of their default evidential base, which contains the totality of their private beliefs and any public contextual evidence. The notion of weak commitment is defined in terms of the evidential base's resistance to change in light of future discourse moves. %With both empirical domains, novel data is brought to bear to contrast the analysis advanced here to previous literature.

The results of this investigation are threefold. First, I argue thanks to novel data that each of the above phenomena---as well as others including English polarity particles and the Singapore Colloquial English (Singlish) particle lah—receive superior empirical coverage under the individual analyses advanced here, compared to previous literature. Second, because the general architecture of each solution is shared, the discourse model advanced here allows for a more parsimonious explanation of the formal pragmatic effects of these phenomena. Third, by highlighting the commonalities among these domains and their relationship to evidentiality, the model allows for greater insight into the way that linguistic discourse is organized, the differences between default and non-default utterances, and more broadly, the distinction between formal pragmatics and general reasoning about language use.