Style Analysis of Argumentative Texts by Mining Rhetorical Devices

Khalid Al Khatib, Viorel Morari, Benno Stein


Abstract
Using the appropriate style is key for writing a high-quality text. Reliable computational style analysis is hence essential for the automation of nearly all kinds of text synthesis tasks. Research on style analysis focuses on recognition problems such as authorship identification; the respective technology (e.g., n-gram distribution divergence quantification) showed to be effective for discrimination, but inappropriate for text synthesis since the “essence of a style” remains implicit. This paper contributes right here: it studies the automatic analysis of style at the knowledge-level based on rhetorical devices. To this end, we developed and evaluated a grammar-based approach for identifying 26 syntax-based devices. Then, we employed that approach to distinguish various patterns of style in selected sets of argumentative articles and presidential debates. The patterns reveal several insights into the style used there, while being adequate for integration in text synthesis systems.
Anthology ID:
2020.argmining-1.12
Volume:
Proceedings of the 7th Workshop on Argument Mining
Month:
December
Year:
2020
Address:
Online
Editors:
Elena Cabrio, Serena Villata
Venue:
ArgMining
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
106–116
Language:
URL:
https://aclanthology.org/2020.argmining-1.12
DOI:
Bibkey:
Cite (ACL):
Khalid Al Khatib, Viorel Morari, and Benno Stein. 2020. Style Analysis of Argumentative Texts by Mining Rhetorical Devices. In Proceedings of the 7th Workshop on Argument Mining, pages 106–116, Online. Association for Computational Linguistics.
Cite (Informal):
Style Analysis of Argumentative Texts by Mining Rhetorical Devices (Al Khatib et al., ArgMining 2020)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-2/2020.argmining-1.12.pdf
Data
New York Times Annotated Corpus