ChangeMyView Through Concessions: Do Concessions Increase Persuasion?

TL;DR

This study empirically investigates whether concessions increase persuasion in online discourse, finding that argumentative concessions are used equally in persuasive and non-persuasive comments, challenging traditional assumptions.

Contribution

The paper introduces a typology and annotation method for identifying argumentative concessions and presents a self-training approach for automatic detection in large datasets.

Findings

Concessions are equally common in persuasive and non-persuasive comments.

Self-training method achieved moderate F1 scores for identifying concessions.

Results challenge the assumption that concessions directly increase persuasion.

Abstract

In discourse studies concessions are considered among those argumentative strategies that increase persuasion. We aim to empirically test this hypothesis by calculating the distribution of argumentative concessions in persuasive vs. non-persuasive comments from the ChangeMyView subreddit. This constitutes a challenging task since concessions are not always part of an argument. Drawing from a theoretically-informed typology of concessions, we conduct an annotation task to label a set of polysemous lexical markers as introducing an argumentative concession or not and we observe their distribution in threads that achieved and did not achieve persuasion. For the annotation, we used both expert and novice annotators. With the ultimate goal of conducting the study on large datasets, we present a self-training method to automatically identify argumentative concessions using linguistically motivated features. We achieve a moderate F1 of 57.4% on the development set and 46.0% on the test set via the self-training method. These results are comparable to state of the art results on similar tasks of identifying explicit discourse connective types from the Penn Discourse Treebank. Our findings from the manual labeling and the classification experiments indicate that the type of argumentative concessions we investigated is almost equally likely to be used in winning and losing arguments from the ChangeMyView dataset. While this result seems to contradict theoretical assumptions, we provide some reasons for this discrepancy related to the ChangeMyView subreddit.