Abstract: Low-resource languages present a challenge for data-hungry approaches to machine translation, speech recognition, and other technologies that promise to open the way for universal participation in the global information society. In this talk I will present a new perspective on the language technology for all (LT4All) agenda, beginning with the structure of the world's linguistic diversity and the actual linguistic challenges on the ground. I will draw on experiences working in societies where there is no clear case for the popular practice of replicating human capabilities in translation or speech recognition, but where there are myriad other opportunities for language technologies. I will describe new ways of working with local communities, oral languages, and human-curated linguistic resources. The result is a radically inclusive approach to language technology which embraces and sustains linguistic diversity.
Bio: Steven Bird has spent much of his career pursuing scalable computational methods for capturing, enriching, and analysing data from endangered languages, drawing on fieldwork in West Africa, South America, and Melanesia. Over the past 5 years he has shifted to working from the ground up with remote Aboriginal communities in Australia, supporting language learning and development in an Aboriginal ranger program, school, and arts centre. He is a co-developer of the Natural Language Toolkit (NLTK), co-founder of the Open Language Archives Community (OLAC), founder of the ACL Anthology, and director of the Aikuma Project. He has held academic appointments at the universities of Edinburgh, Pennsylvania, UC Berkeley, and Melbourne, and is now professor at Charles Darwin University, in Darwin, Australia.