Stephen Doherty is a lecturer in the School of Humanities & Languages at the University of New South Wales, Sydney, Australia. He conducts research on translation and cognition, including translation technologies and translation process studies.
He currently lectures in Translation Technology, Media Translation, Translation Theory, and is an active trainer and consultant in the international translation technology industry, including, LSPs and government users (e.g. National Accreditation Authority for Translators and Interpreters, New South Wales Ministry of Health).He has several years of experience working with translation technologies, with extensive experience in teaching translation technologies, and a variety of research and teaching roles at Dublin City University and the Centre for Next Generation Localisation.Under the recent European Commission project, QTLaunchPad, he has collaborated closely with international organizations to improve translation quality assessment and the usage of technologies, including: GALA, TAUS, and FIT.He is an active author of industry publications and academic research and holds numerous scientific committee and reviewer positions (including Australian Institute of Interpreters and Translators, Association for Machine Translation in the Americas, AMTA Post-Editing Workshop series, the European Association for Machine Translation, and several international journals).

Stephen Doherty

will present…

Improving Translator Competencies by Teaching Statistical Machine Translation: Evidence and Experiences from University, LSP, Public Service, and Community Training Programmes



 Translator training and professional development programmes have been slow to integrate machine translation (MT) into their curricula, both in educational institutions as well as in industry training programmes and especially in relation to statistical machine translation (SMT). Given the growing online prevalence of SMT and the success of cloud-based SMT solutions, translators and their educators are in an ideal position to better engage with contemporary SMT in order to understand what such technologies can and cannot do and the value of the human translator in an increasing technology-embedded workplace.

Technological inadequacies, both perceived and real, are often cited as barriers for translators, especially freelancers, to explore SMT in their work. Student translators often feel that technology is not important to their future careers, despite the overwhelming evidence that technological expertise have long since become an proven requirement (Gaspari, Almaghout & Doherty, 2015). Additionally, professional translators report that the absence of contemporary training programmes and materials makes training inaccessible, where most MT literature and tools having a high technological entry point as they are not aimed at translators themselves, and technical support is unavailable, especially for freelancers and community (public service) translators.

In order to bridge these gaps, an open syllabus for teaching SMT to translators was developed and validated in accordance with psychometric standards and defined learning outcomes (Doherty & Kenny, 2014). This work was the first to showcase the additional educational value for translators in learning how SMT works and how it could be used in their work. The findings have shown that such directed training significantly increases student translators’ technological self-efficacy (an established psychological construct of self-performance that is readily transferrable to other technologies), improved linguistic awareness of both languages, proficiency in effectively using translation quality assessment models (e.g. accuracy, fluency, diagnostic evaluation, LISA QA, MQM), and informed post-editing of SMT output.

In expanding this training to other contexts, we report on recent applications of teaching SMT to translators in additional contexts of post-graduate training programmes, small LSPs, government users, and open professional workshops for translators in the community (public service translation). Findings from these applications are similar in terms of the aforementioned educational benefits, but such diverse training brings with it a host of new challenges, chiefly: working with cohorts of 25+ language pairs, languages with limited online resources, different levels of linguistic knowledge and translation experience, and economic pressures.

In closing, we detail the lessons learned from these experiences and propose how SMT training materials can be created and shared to overcome these challenges and better cater for the changing diversity in professional training needs, different modes of delivery, and the need for evidence-based and informed teaching practices.



Gaspari, F., Almaghout, H. & Doherty, S. (2015). A survey of machine translation competences: insights for translation technology educators and practitioners. Perspectives: Studies in Translatology, (ahead of print). DOI:10.1080/0907676X.2014.979842

Doherty, S. & Kenny, D. (2014). The design and evaluation of a Statistical Machine Translation syllabus for translation students. The Interpreter and Translator Trainer, 8(2), 295-315. DOI: 10.1080/1750399X.2014.937571