Morphologically Aware Word-Level Translation

TL;DR

This paper introduces a morphologically aware probabilistic model for bilingual lexicon induction that jointly models lexeme translation and inflectional morphology, significantly improving accuracy over previous methods.

Contribution

It presents a novel structured model that integrates morphological information into bilingual lexicon induction, addressing limitations of existing approaches.

Findings

19% accuracy improvement in supervised setting

16% accuracy improvement in weakly supervised setting

Highlights issues in modern BLI due to ignoring morphology

Abstract

We propose a novel morphologically aware probability model for bilingual lexicon induction, which jointly models lexeme translation and inflectional morphology in a structured way. Our model exploits the basic linguistic intuition that the lexeme is the key lexical unit of meaning, while inflectional morphology provides additional syntactic information. This approach leads to substantial performance improvements - 19% average improvement in accuracy across 6 language pairs over the state of the art in the supervised setting and 16% in the weakly supervised setting. As another contribution, we highlight issues associated with modern BLI that stem from ignoring inflectional morphology, and propose three suggestions for improving the task.