Home › Categories › Technology› How Synchronous Are Adjuncts In Translation Data?

How Synchronous Are Adjuncts In Translation Data?.pdf

W14-4019.pdf

Preview of How Synchronous are Adjuncts in Translation Data?

🔗 Source: aclweb.org

📊 Size: 144 KB

👤 Author: Sophie Arnoult ; Khalil Sima'an

⬇️ Downloads: 145

Summary

The argument-adjunct distinction is central to most syntactic and semantic theories, with adjuncts being optional elements that refine the meaning of a phrase. In machine translation, adjuncts are often treated as modifiers applying synchronously in source and target derivations. This paper presents an empirical study of translation equivalence of adjuncts in French-English parallel corpora, varying word alignments to gauge the effect of errors. The results show that 70% of manually identified adjuncts have adjunct translation equivalents in training data, compared to 50% for automatically identified adjuncts. The study also introduces a new view of adjunct translation equivalence, allowing for many-to-many pairings and capturing equivalence despite mismatched parses. The findings suggest that adjunct labelling accuracy is crucial for adjunct alignment and that applications can gain from decreasing dependence on word alignments and idealized experimental conditions.

Download File

Description

The argument-adjunct distinction is central to most syntactic and semantic theories, with adjuncts being optional elements that refine the meaning of a phrase.

Technical Information

File Format: PDF
File Size: 144 KB
Pages: 9
Language: EN
Author: Sophie Arnoult ; Khalil Sima'an
Total Downloads: 145
Last Updated: 2 hours ago

Document Overview

This PDF document about How Synchronous are Adjuncts in Translation Data? provides comprehensive information and guidance. Whether you're a beginner or advanced user, this resource offers valuable insights into How Synchronous are Adjuncts in Translation Data?.