Speaker
Dr
Juri Opitz
(Universität Zürich)
Description
Large Language Models (LLMs) are impressively fluent, and seem to function without any modules specially designed to capture grammar or semantic coherence. This can lead us to question even more the (once broadly presumed) relevance of linguistic expertise in Natural Language Processing (NLP). In this talk, I will explore six key areas where linguistics, in overt and less overt ways, continues to play a crucial role in NLP, encapsulated by the acronym RELIES: Resources, Evaluation, Low-resource settings, Interpretability, Explanation, and the Study of language. On a macro level, these facets show how linguistic insights remain vital for advancing and validating NLP technologies.
