Discourse constraints on prosodic marking in lexical replacement repair

•I investigate prosodic marking in lexical replacement repair.•I use discourse analysis to identify contexts in which marking is favoured.•I highlight the relevance of discourse coherence and epistemics for marking.•I explore the empirical grounds for proposals regarding the function of marking.

This paper reports on an investigation of instances of lexical replacement repair, in which a speaker replaces one lexical choice with another, sampled from Dutch spontaneous interaction. The study is driven by the question as to what motivates a speaker to produce a particular instance of self-repair with or without ‘prosodic marking’ — with or without notable prosodic prominence — and the notion that a close consideration of the discourse context in which the repair is embedded, and its function in that context, is paramount in addressing this question. The study explores the empirical grounds for two proposals regarding the function of prosodic marking: one in which marking is a response to the speaker's embarrassment or unease at the error or infelicity, and one in which marking is done for the listener's benefit, to highlight particularly important information. This paper describes three discourse contexts in which prosodic marking is notably common, and argues that both proposals find some support in these contexts. The analysis suggests that speakers’ decisions for or against prosodic marking are based at least on considerations of epistemic authority, precision and exaggeration, and discourse coherence.