Bilingual Rhetorical Structure Parsing with Large Parallel Annotations

TL;DR

This work tackles cross-lingual discourse parsing by creating a large parallel Russian annotation (RRG) for the English GUM RST corpus and by developing an end-to-end top-down RST parser that achieves state-of-the-art results on both English and Russian datasets. The model combines a ToNy segmentation module and an E-BiLSTM EDU encoder, trained without augmentation and with a Dynamic Weighted Average loss that uses a window to stabilize training. It demonstrates strong zero-shot and bilingual transfer capabilities, showing that even limited second-language data can support effective cross-lingual parsing, and that full parallel training yields the best results for Russian. The accompanying resources and public release underscore the practical potential of cross-lingual RST parsing for multilingual NLP tasks and downstream applications like summarization and analysis.

Abstract

Discourse parsing is a crucial task in natural language processing that aims to reveal the higher-level relations in a text. Despite growing interest in cross-lingual discourse parsing, challenges persist due to limited parallel data and inconsistencies in the Rhetorical Structure Theory (RST) application across languages and corpora. To address this, we introduce a parallel Russian annotation for the large and diverse English GUM RST corpus. Leveraging recent advances, our end-to-end RST parser achieves state-of-the-art results on both English and Russian corpora. It demonstrates effectiveness in both monolingual and bilingual settings, successfully transferring even with limited second-language annotation. To the best of our knowledge, this work is the first to evaluate the potential of cross-lingual end-to-end RST parsing on a manually annotated parallel corpus.

Figures (8)

• Figure 1: N-to-One EDU mapping; news_iodine.
• Figure 2: One-to-N EDU mapping; fiction_wedding.
• Figure 3: Architectural overview of DMRST.
• Figure 4: Impact of second language injection on the end-to-end Full performance.
• Figure 5: Example of N-to-One EDU mapping. From academic_art.