Multi-Task Neural Models for Translating Between Styles Within and Across Languages

TL;DR

This paper introduces a multi-task neural approach to generate text with varying styles, achieving state-of-the-art results in formality transfer and enabling style-sensitive translation without explicit style annotations.

Contribution

It presents a joint multi-task learning framework for style transfer and style-sensitive translation, advancing the ability to generate stylistically appropriate language.

Findings

Achieved state-of-the-art performance in formality transfer

Enabled style-sensitive translation without explicit style annotations

Demonstrated effectiveness of multi-task learning for style control

Abstract

Generating natural language requires conveying content in an appropriate style. We explore two related tasks on generating text of varying formality: monolingual formality transfer and formality-sensitive machine translation. We propose to solve these tasks jointly using multi-task learning, and show that our models achieve state-of-the-art performance for formality transfer and are able to perform formality-sensitive translation without being explicitly trained on style-annotated translation examples.