Expressing stance in Mandarin conversation: Epistemic and non-epistemic uses of wo juede

This study investigates uses of wo juede 'I feel/think' in Mandarin conversation. My aim is to characterize the ways wo juede, generally working as an epistemic stance marker, is used to facilitate interaction between conversation participants by using data from actual conversation.;Chapter 1 is an introduction. I explain the methodology I adopt -- Interactional Linguistics. Notions in Discourse-functional Linguistics and Conversation Analysis are summarized as these frameworks provide the basis for my quantitative and qualitative analyses respectively.;In Chapter 2, I introduce preliminary ideas about stance and modality, by reviewing recent findings on these subjects. I pay particular attention to studies of complementation, especially when used with the first person subject.;Chapter 3 deals with general patterns of juede 'feel' (including its epistemic use in wo juede 'I feel/think') by comparing its use with that of its synonyms xiang 'think' and renwei 'think'. It is shown that most complement clauses of wo juede are assessments of someone or something other than the speaker. It is also shown that complement clauses of wo juede tend to have contestability features: strong evaluation, epistemic modality, contrast and negation. The positions of wo juede in a clause and in a turn are also examined.;Chapter 4 concerns of wo juede at the turn-initial position. I examine cases of wo juede used in responses to three different types of first pair part: an assessment/judgment, an informing and a question. I argue that wo juede generally works to mitigate disagreement between participants.;Chapter 5 is devoted to analyses of wo juede in the turn-medial position. When wo juede is used in the middle of a turn, wo juede works to attenuate an anticipated trouble in interaction. The specific functions identified are shifting to opinion, distancing from a claim and marking a topic shift.;In Chapter 6, I examine cases in which wo juede is used at the end of a turn. Uses of wo juede at TCU-initial position and TCU-final position are analyzed. I argue that wo juede may be used to promote turn-transition.;In Chapter 7, I investigate non-epistemic uses of wo juede that contribute to affective or evaluative stancetaking.;Chapter 8 is discussion and conclusion. I argue that my findings in Chapters 3 to 7 indicate that grammaticalization of wo juede is position-specific.