The 2019 JIAMCATT (International Annual Meeting on CAT and Terminology Tools) conference was held in Luxembourg from 13 to 15 May, with the topic ‘The shape of things to come: how technology and innovation are transforming the language profession’. Representatives from the Translation Centre attended the event and took part in different working groups.
Machine translation (MT) was, as would be expected nowadays, the main topic of the conference. Several discussions showed that various MT solutions have been implemented in a number of organisations in the past few years. Many efforts have been made to develop not only statistical machine translation (SMT) but also neural machine translation (NMT). The European Commission’s NMT solution, eTranslation, has been up and running for some time already in many EU institutions and other organisations, covering all EU languages. In the United Nations family, the World Intellectual Property Organisation (WIPO) has achieved very good results with ‘WIPO Translate’. Other smaller organisations have also tested and implemented MT solutions in their workflow by relying on market solutions.
Jochem Hummel, the keynote speaker of this year’s JIAMCATT, set the tone for the conference with his presentation entitled “Sunsetting of CAT tools”.
The discussions focused on various hot topics such as the increased cognitive effort required in post‑editing; the elevation of terminology from a concept-based approach to a knowledge-based approach; the new skills required by those involved in the translation process; the inclusion of experts in the translation workflow; user interface re-thinking; and change management.
Spread over 12 knowledge café working groups, participants from UN and EU bodies, universities and NGOs brainstormed around the main challenges of specific topics, and proposed solutions.
Two Centre representatives took part in the working groups on neural machine translation (NMT), which focused on quality, post-editing and its implications, and the future of computer-assisted translation (CAT) and computer-assisted revision (CAR). One of their conclusions was that nowadays practically no translators translate a sentence from scratch without consulting an existing resource such as translation memories, multilingual corpora (internal or publicly available resources), terminology databases (IATE, UNTERM, FAOTERM, etc.) and MT. In terms of training, this means that translators need to learn post‑editing techniques and be aware of the difference between post-editing – when a translator has to edit the output of a machine translation system – and revision – when a translator revises the translation provided by a human translator.
JIAMCATT 2019 also provided the Centre with an opportunity to present its current work on speech recognition technologies. The aim is to automate as much as possible the speech-to-text processes to be applied to transcription and/or subtitling services. Today, more than ever, the future is now – and JIAMCATT was there to remind us.