Human Factors in Neural Machine Translation

As part of the artificial intelligence hype cycle, grand claims regularly appear about new breakthroughs in machine translation (MT) quality. This presentation provides a short history of MT development, during which promises have often been followed by disappointment, and an overview of contemporary MT technology. Recent evaluations show improvements in fluency and accuracy for gist translation between major languages, but any leap forward in productivity for translators working with MT appears to be modest.

This presentation will focus on reasonable quality expectations for NMT when applied to different fields and text types, and report on the latest research into how translators are currently working with NMT. This leads on to the broader repercussions of NMT for work in translation, a discussion of the enormous data requirements for system training, and other applications of neural networks within translation workflows.

Joss Moorkens is an Assistant Professor at the School of Applied Language and Intercultural Studies at Dublin City University and a researcher affiliated with the ADAPT Centre and the Centre for Translation and Textual Studies. He has authored articles and chapters on translation technology, machine translation post-editing, translation evaluation, translator precarity, and translation technology standards. He is co-editor of the book ‘Translation Quality Assessment: From Principles to Practice’ (Springer 2018) and a March 2019 special issue of Machine Translation journal focusing on human factors in neural machine translation.