# Few-Shot and Zero-Shot Learning for Historical Text Normalization

## TL;DR

This paper systematically evaluates 63 multi-task learning configurations for historical text normalization, demonstrating significant improvements in low-resource settings and showing zero-shot learning often outperforms simple baselines.

## Contribution

It provides a comprehensive analysis of multi-task learning architectures and their effectiveness in historical text normalization across multiple languages and datasets.

## Key findings

- Multi-task learning improves normalization with limited data.
- Zero-shot learning often outperforms identity baseline.
- Limited benefits observed when training data is abundant.

## Abstract

Historical text normalization often relies on small training datasets. Recent work has shown that multi-task learning can lead to significant improvements by exploiting synergies with related datasets, but there has been no systematic study of different multi-task learning architectures. This paper evaluates 63~multi-task learning configurations for sequence-to-sequence-based historical text normalization across ten datasets from eight languages, using autoencoding, grapheme-to-phoneme mapping, and lemmatization as auxiliary tasks. We observe consistent, significant improvements across languages when training data for the target task is limited, but minimal or no improvements when training data is abundant. We also show that zero-shot learning outperforms the simple, but relatively strong, identity baseline.