Keynote
Multilingual gender-inclusivity in translation and beyond
Abstract
Multilingual capabilities are increasingly available in general-purpose systems, rather than from dedicated MT systems alone. This shift impacts many practical concerns for improving gender inclusivity such as understanding downstream developer usage patterns, improving the validity of upstream evaluations, and scaling to global cultural contexts. It also raises sociotechnical research challenges in creating new kinds of transparently multilingual user experiences, improving controllability of gender-inclusive representations, and enabling new modalities like multilingual image understanding and audio generation. I discuss empirical work to measure potential misgendering harms in PaLM 2, and share experiences from more recent research at Google.
Bio
Kevin Robinson is a Senior Research Engineer at Google, working on developing new techniques for inclusive, controllable, and robust machine learning systems by effectively blending technical and sociocultural perspectives. Kevin has worked on research efforts like PaLM, PaLM-FLAN and PaLM 2, and contributed to products like Bard and Gemini. Kevin has separately co-authored publications on language models related to pre-training data, synthetic data generation, and measuring misgendering harms in translation systems. He is currently focused on measuring cultural and representational harms in ways that incorporate community-informed perspectives. Kevin has also worked as a special education teacher, and a computer science education researcher at MIT focused on bias within CS classrooms.
Keynote
Intersectionality and gender in translation — how ethical must one automatically be?
Abstract
To which extent should ethical considerations inform (automated) inclusive translation processes? This talk will present a reflection on criteria for the minimum requirements of translation ethics that could be applied systematically to any text, from the point of view of intersectional, queer and feminist principles. By critically examining the ethical dimensions of translation through these lenses, this talk will seek to illuminate the path toward more inclusive, equitable, and socially responsible translation practices.
Bio
Begoña Martínez Pagán is a translator, interpreter, and author based at the English Studies Department of the University of Murcia. Her activism, lecturing, and research include intersections of her profession with feminist and LGBTIQ+ literature, inclusive language, human rights, business organization, and open-source software.
PANELISTS
Bio
Helena Moniz is the President of the European Association for Machine Translation and President of the International Association for Machine Translation. She is also the Vice-Coordinator of the Human Language Technologies Lab at INESC-ID, Lisbon. Helena is an Assistant Professor at the School of Arts and Humanities at the University of Lisbon, where she teaches Computational Linguistics, Computer Assisted Translation, and Machine Translation Systems and Post-editing. Since 2015, she is also the PI of a bilateral project with INESC-ID/Unbabel, a translation company combining AI + post-editing, working on scalable Linguistic Quality Assurance processes for crowdsourcing.
She is now in a very exciting project, coordinated by Unbabel, the Center for Responsible AI, within the Portuguese Recovery and Resilience Plan, as Chair of the Ethics Committee, loving the challenge and the outstanding discussions within the entire consortium.
Bio
Paula Manzur holds a Bachelor's degree in Translation and Interpretation Studies and over 14 years of dedicated experience in the Localization industry. Beginning as a professional translator, she has seamlessly transitioned through various roles, navigating and overcoming the diverse tech-related challenges within the Language Industry.
Presently, working as a Localization Process Lead at Booking.com, she leverages her expertise in machine translation application and hands-on experience to design and optimize language technology solutions. Speaker at MT Summit, GALA Connected, and as a panelist at the Phrase (formerly Memsource) MT Workshop Series.