The Comparison of Translationese in Machine Translation and Human Transation in terms of Translation Relations

Fan Zhou

2024-03-28

Abstract:This study explores the distinctions between neural machine translation (NMT) and human translation (HT) through the lens of translation relations. It benchmarks HT to assess the translation techniques produced by an NMT system and aims to address three key research questions: the differences in overall translation relations between NMT and HT, how each utilizes non-literal translation techniques, and the variations in factors influencing their use of specific non-literal techniques. The research employs two parallel corpora, each spanning nine genres with the same source texts with one translated by NMT and the other by humans. Translation relations in these corpora are manually annotated on aligned pairs, enabling a comparative analysis that draws on linguistic insights, including semantic and syntactic nuances such as hypernyms and alterations in part-of-speech tagging. The results indicate that NMT relies on literal translation significantly more than HT across genres. While NMT performs comparably to HT in employing syntactic non-literal translation techniques, it falls behind in semantic-level performance.

Computation and Language

What problem does this paper attempt to address?

The paper primarily explores the differences between machine translation (especially neural machine translation, NMT) and human translation (HT) in terms of translation relations, using this as a metric for measuring translation quality. Specifically, the paper attempts to address the following issues:

1. **Differences in overall translation relations between NMT and HT**: By comparing the performance of NMT systems (represented by GoogleNMT) and human translation in different types of translation relations, the paper investigates the differences between the two.

2. **Differences in the application of non-literal translation techniques**: The study examines the differences between NMT and HT in the use of non-literal translation techniques (such as metonymy, transposition, etc.).

3. **Factors influencing the use of specific non-literal translation techniques**: The analysis identifies factors that lead to differences between NMT and HT in the choice of certain non-literal translation techniques.

The translation relations mentioned in the paper refer to a classification system of translation techniques, covering various strategies and techniques from literal to non-literal translation. These techniques are used to evaluate translation quality, especially when the translated text is too close to the source language, making it unnatural or difficult to understand (the so-called "translationese" or "translationese style"). By comparing the performance of NMT and HT in handling different types of translation relations, the paper aims to reveal areas where NMT can be improved to better approach or even achieve the quality level of human translation.