Flexing in 73 Languages: A Single Small Model for Multilingual Inflection

TL;DR

This paper introduces a compact, multilingual inflection model trained on 73 languages that outperforms monolingual models, demonstrating the benefits of a single, robust system for morphological inflection across diverse languages.

Contribution

The authors develop a lightweight, single-model approach for multilingual inflection that handles unseen words and simplifies deployment, outperforming monolingual baselines across many languages.

Findings

Outperforms monolingual models in most languages

Robust to unseen words across languages

Simplifies deployment by using a single model

Abstract

We present a compact, single-model approach to multilingual inflection, the task of generating inflected word forms from base lemmas to express grammatical categories. Our model, trained jointly on data from 73 languages, is lightweight, robust to unseen words, and outperforms monolingual baselines in most languages. This demonstrates the effectiveness of multilingual modeling for inflection and highlights its practical benefits: simplifying deployment by eliminating the need to manage and retrain dozens of separate monolingual models. In addition to the standard SIGMORPHON shared task benchmarks, we evaluate our monolingual and multilingual models on 73 Universal Dependencies (UD) treebanks, extracting lemma-tag-form triples and their frequency counts. To ensure realistic data splits, we introduce a novel frequency-weighted, lemma-disjoint train-dev-test resampling procedure. Our work addresses the lack of an open-source, general-purpose, multilingual morphological inflection system capable of handling unseen words across a wide range of languages, including Czech. All code is publicly released at: https://github.com/tomsouri/multilingual-inflection.