Tværfagligt valgfag, emne 3: Language, Cognition, and Biases in the Digital Era

Text-generating language models like ChatGPT are omnipresent, can automatically compose, translate and edit texts and are about to fundamentally change Humanities and scientific practice.

This course introduces students to key linguistic theory and linguistic methods underlying large language models (LLMs), in particular cognitive linguistics, and collocation analysis for a better understanding of challenges and opportunities of new language technologies including generative AI: How does the knowledge encoded in LLMs differ from human knowledge? Why do LLMs exhibit biases and how do they relate to human biases?

Participants will get familiar with digital tools most used in Humanities (digital text analyses, translation tools, linguistic corpora) and learn how to use these tools in a meaningful way. Students will conduct case studies (in groups, preferably cross-languages) to uncover collocational biases and stereotypes relating to e.g. race, gender, ethnicity, nationality in digital tools (e.g., Word Thesauruses, translation tools, chatbots) and linguistic corpora (e.g., webcorpora available via SketchEngine).