Learning variable length units for SMT between related languages via Byte Pair Encoding

TL;DR

This paper investigates Byte Pair Encoding (BPE) units as a flexible, data-driven approach for statistical machine translation between related languages, outperforming linguistically motivated orthographic syllables across various writing systems.

Contribution

It demonstrates that BPE units, learned from data, are more effective than orthographic syllables for translation, especially in non-vowel writing systems, across multiple language families.

Findings

BPE units improve BLEU scores by up to 11% over orthographic syllables.

BPE outperforms other units in non-vowel writing systems.

Extensive experiments validate BPE's effectiveness across diverse languages.

Abstract

We explore the use of segments learnt using Byte Pair Encoding (referred to as BPE units) as basic units for statistical machine translation between related languages and compare it with orthographic syllables, which are currently the best performing basic units for this translation task. BPE identifies the most frequent character sequences as basic units, while orthographic syllables are linguistically motivated pseudo-syllables. We show that BPE units modestly outperform orthographic syllables as units of translation, showing up to 11% increase in BLEU score. While orthographic syllables can be used only for languages whose writing systems use vowel representations, BPE is writing system independent and we show that BPE outperforms other units for non-vowel writing systems too. Our results are supported by extensive experimentation spanning multiple language families and writing systems.