Incorporating Human Translator Style into English-Turkish Literary Machine Translation

TL;DR

This paper enhances English-Turkish literary machine translation by incorporating individual translator styles, demonstrating that stylistic features can be effectively modeled and reproduced in machine-generated translations.

Contribution

It introduces a method to adapt machine translation models to specific human translator styles using manual fine-tuning and stylistic evaluation techniques.

Findings

Human translator style can be effectively recreated in machine translations.

Model adaptation improves stylistic consistency in literary translation.

Data augmentation and alignment methods influence translation quality.

Abstract

Although machine translation systems are mostly designed to serve in the general domain, there is a growing tendency to adapt these systems to other domains like literary translation. In this paper, we focus on English-Turkish literary translation and develop machine translation models that take into account the stylistic features of translators. We fine-tune a pre-trained machine translation model by the manually-aligned works of a particular translator. We make a detailed analysis of the effects of manual and automatic alignments, data augmentation methods, and corpus size on the translations. We propose an approach based on stylistic features to evaluate the style of a translator in the output translations. We show that the human translator style can be highly recreated in the target machine translations by adapting the models to the style of the translator.