LINGUISTIC ASPECTS OF AUTOMATIC TEXT ALIGNMENT IN PARALLEL CORPORA

Abstract

Parallel corpora play a crucial role in multilingual natural language processing, machine translation, and contrastive linguistics. A fundamental task in constructing parallel corpora is automatic text alignment which refers to linking corresponding textual units (sentences or paragraphs) across different languages. This article explores the linguistic aspects influencing alignment accuracy, including syntactic structure, word order, phraseology, and translation strategies. We also examine common alignment techniques and assess their linguistic robustness using case studies from English-Russian corpora. The findings show that integrating linguistic features significantly improves alignment precision, especially in complex or free word order languages