Resolve Magazine Fall 2023 >> Making Sense of Machine Learning >> Stories >> Protecting diversity
“Saving and revitalizing endangered languages is important for maintaining cultural diversity,” says Maryam Rahnemoonfar (pictured), an associate professor in the Department of Computer Science and Engineering and the Department of Civil and Environmental Engineering. “Language loss is equal to losing identity, memory, culture, and knowledge, and can even affect the mental health of Indigenous people.”
Rahnemoonfar serves as the faculty mentor for a project in Lehigh’s Mountaintop Summer Experience program that is using machine learning to revive the languages of Native Americans. However, a major hurdle in the project is the lack of resources featuring these languages that are available to feed the deep learning algorithms. Comparable sources in English, Spanish, French, or German are plentiful.
“That scarcity results in very low accuracy when applied to Indigenous languages,” she says. “To tackle this issue, we will collect a large set of text and audio resources in collaboration with local communities, develop novel machine learning techniques that combine deep learning algorithms with rule-based methods that add constraints in the learning process about language morphology, and create reinforcement learning algorithms that integrate the feedback of native speakers.”
While the final product for these tools is still being conceived, Rahnemoonfar hopes the team’s work will enable younger generations to learn and speak their native language and better connect with their elders and their culture.
Main image: Thepalmer/iStock