The distribution of discourse relations within and across turns in spontaneous conversation

TL;DR

This study investigates how discourse relations are distributed within and across turns in spontaneous conversation, adapting a written language system and analyzing annotations to understand contextual differences and prediction feasibility.

Contribution

It introduces a method for annotating discourse relations in spontaneous dialogue and compares their distributions across various conversational contexts.

Findings

Different discourse contexts yield distinct distributions of relations.

Single-turn annotations are more uncertain for annotators.

Discourse relations can be predicted from embeddings of discourse units.

Abstract

Time pressure and topic negotiation may impose constraints on how people leverage discourse relations (DRs) in spontaneous conversational contexts. In this work, we adapt a system of DRs for written language to spontaneous dialogue using crowdsourced annotations from novice annotators. We then test whether discourse relations are used differently across several types of multi-utterance contexts. We compare the patterns of DR annotation within and across speakers and within and across turns. Ultimately, we find that different discourse contexts produce distinct distributions of discourse relations, with single-turn annotations creating the most uncertainty for annotators. Additionally, we find that the discourse relation annotations are of sufficient quality to predict from embeddings of discourse units.