Quality in Human and Machine Translation: An Interdisciplinary Survey

Quality in Human and Machine Translation: An Interdisciplinary Survey Bettina Hiebl and Dagmar Gromann University of Vienna, Austria {bettina.hiebl, dagmar.gromann}@univie.ac.at Abstract Quality assurance is a central component of human and machine translation. In translation studies, translation quality fo- cuses on human evaluation and dimen- sions, such as purpose, comprehensibil- ity, target audience among many more. Within the field of machine translation, more operationalized definitions of quality lead to automated metrics relying on ref- erence translations or quality estimation. A joint approach to defining and assessing translation quality holds the promise to be mutually beneficial. To contribute towards that objective, this systematic survey pro- vides an interdisciplinary analysis of the concept of translation quality from both perspectives. Thereby, it seeks to inspire cross-fertilization between both fields and further development of an interdisciplinary concept of translation quality. 1 Introduction Translation quality has been a source of debate in translation studies for decades (Koby et al., 2014), since it is considered highly subjective and de- pendent on how translation and quality are de- fined. One common denominator is the central role played by accuracy and fluency (Koby et al., 2014; Castilho et al., 2018), a view shared by the field of machine translation (Yuan and Sharoff, 2020; Koehn and Monz, 2006). An accurate semantic correspondence between source and translation as well as an adequate degree of fluency in the latter © 2023 The authors. This article is licensed under a Creative Commons 4.0 licence, no derivative works, attribution, CC- BY-ND. are expected. Aside from these shared notions, ap- proaches to define, assess and measure translation quality differ substantially in the field of transla- tion studies and machine translation. This interdis- ciplinary survey analyzes literature on translation quality from both perspectives. The idea to join the theoretical basis of trans- lation studies with the operationalized quality definitions of machine translation is not new ( ˇCulo, 2014). However, existing surveys on the topic focus either on machine translation (Rivera- Trigueros, 2022; Han et al., 2021), post-editing (Koponen, 2016) or the perspective of translation studies (Koby et al., 2014). From a theoretical per- spective, Castilho et al. (2018) present key qual- ity theories from both fields and argue that the line between human and machine translation is in- creasingly blurring, especially in post-editing. The Multidimensional Quality Metric (MQM) (Lom- mel et al., 2014) proposes a comprehensive catalog of quality issues, which can be used to calculate a score for evaluating translations. Inspired by the PRISMA method (Page et al., 2021) and guidelines by Kitchenham (2004), this paper presents a systematic literature review on translation quality in the field of translation stud- ies and machine translation. Resulting publica- tions are deduplicated and ranked by a keyword rating method that takes the number of occurrences across platforms and keywords into account. The resulting top 41 publications are presented based on the authors’ fields and translation quality per- spective. Thereby, the present survey contributes an overview of types of translation quality per field and interdisciplinary publications in the re- …

Content truncated to 500 words