Controlling Extra-Textual Attributes about Dialogue Participants -- A Case Study of English-to-Polish Neural Machine Translation

TL;DR

This paper explores how external metadata can improve English-to-Polish neural machine translation of TV dialogue by controlling attributes like gender and number, addressing morphological challenges.

Contribution

It introduces a case study with multiple approaches for attribute control, a new dataset, and a morphological analysis tool for evaluating translation quality.

Findings

Best model improves translation metrics by +5.81 chrF++/+6.03 BLEU

Multiple models achieve competitive performance

Provides a novel attribute-annotated dataset for Polish TV dialogue

Abstract

Unlike English, morphologically rich languages can reveal characteristics of speakers or their conversational partners, such as gender and number, via pronouns, morphological endings of words and syntax. When translating from English to such languages, a machine translation model needs to opt for a certain interpretation of textual context, which may lead to serious translation errors if extra-textual information is unavailable. We investigate this challenge in the English-to-Polish language direction. We focus on the underresearched problem of utilising external metadata in automatic translation of TV dialogue, proposing a case study where a wide range of approaches for controlling attributes in translation is employed in a multi-attribute scenario. The best model achieves an improvement of +5.81 chrF++/+6.03 BLEU, with other models achieving competitive performance. We additionally contribute a novel attribute-annotated dataset of Polish TV dialogue and a morphological analysis script used to evaluate attribute control in models.