Aarhus Universitets segl

How do we make society understand what translation automation is all about? By developing a standard for levels of translation automation?

Aktivitet: Tale eller præsentation - typerForedrag og mundtlige bidrag

Se relationer på Aarhus Universitet

Tina Paulsen Christensen - Foredragsholder

Kristine Bundgaard - Foredragsholder

Anne Schjoldager - Foredragsholder

Helle Dam Jensen - Foredragsholder

Translation technology that attempts to automate the translation process partially or completely is being developed at high speed these years. Thus, in the translation industry, the control of the translation process is being transferred from translators to computers to an increasing extent (European Language
Industry Survey, 2020), and in society at large, free online machine translation (MT) systems are now an integrated part of many people’s digital practices. In this light, we might need a standard that can help us grasp and explain the idea of translation automation (TA). Such a standard would be useful for language service providers, software developers, users of TA technologies as well as researchers.
We argue that in order to make society understand what TA is all about and what translation systems can and cannot do, we should start comparing TA to an example of automation that most people can relate to, namely driverless cars. In driving automation (DA), the decreasing degree of human control over cars is described using a taxonomy outlined by the Society of Automotive Engineers (SAE, 2018).
The taxonomy divides DA into six levels ranging from no automation (level 0) to full automation (level5). The DA taxonomy has been disseminated widely and has thus made the concept of DA more understandable across disciplines and in society at large. As emphasised by Pym (2019, 7), society needs knowledge about TA and in particular about MT, because virtually everyone is using it. In Christensen et al. (2021), we suggested to use the SAE taxonomy as a useful framework for describing different levels of TA and translation features, and we
adapted the DA taxonomy to the field of translation. The proposed taxonomy operates with six TA levels. Briefly explained, the taxonomy describes whether it is the translator or the system that translates by means of source text analysis and target text production and checks for and corrects errors and inadequacies, whether the translator or the system responds to system failures, and whether or not the performance of the system is limited to a certain domain. The level is determined by the TA features – for instance translation memory (TM), MT, a concordance feature or a termbase – that are engaged at a given instance of operation of a TA system. Thus, for instance, a computer assisted translation (CAT) tool equipped with MT only operates at a certain level as long as this feature is turned on. If translators decide to deactivate the MT feature and translate segments themselves, this would decrease the level of automation. Adopting the idea of features as a defining principle hopefully makes the taxonomy flexible enough to accommodate future TA developments. This addresses recent criticism of existing spectrums, such as that of Hutchins and
Somers (1992), which is seen as outdated, because it does not reflect the way existing tools, such as MT and TM, are incorporated into each other. Hence, the TA taxonomy may also help overcome the terminological confusion regarding concepts such as CAT, MT, and TM (Bundgaard 2017; Vieira 2019; Zetsche 2019).
The adaption of the SAE taxonomy turned out to be a complex endeavour. Naturally, our taxonomy is simply a first step towards a standardization of TA levels, which must be further discussed. In our presentation, we will reflect on the advantages of developing standardized levels of TA and discuss the usefulness of the taxonomy and the challenges we faced when developing it. Drawing on recently conducted focus-group interviews, we will integrate feedback on the taxonomy received from translators working in the European Commission Directorate-General for Translation as well as from Danish freelance and in-house translators.
4 jul. 2022

Begivenhed (Konference)

TitelNeTTT: new trends in translation and technology

ID: 276233112