This page is under construction!

 

WS 1. What Type of Research Can a Linguist Do with ChatGPT?

This introductory workshop examines how linguistic research can leverage ChatGPT despite fundamental differences between linguistic units and computational units in LLMs. It aims to clarify what constitutes a “word” or “grammar” for ChatGPT versus for a linguist, exploring whether a subword-based AI (operating on tokens) can serve as a useful proxy for tasks like corpus annotation. The core question: Given ChatGPT’s subword architecture and lack of explicit grammar, in what ways can it contribute meaningfully to linguistic research (or does it mislead)?

 

WS 2. Grammar Without Grammar: How ChatGPT Handles Syntax and Morphology

This workshop dives into whether and how ChatGPT exhibits knowledge of grammar (syntax and morphology) despite not being explicitly programmed with grammatical rules. The aim is to investigate if grammatical regularities are emergent from the training data or somehow encoded implicitly in the model’s architecture. Participants will explore where ChatGPT’s ability to handle grammatical agreement, word order, or morphological inflection comes from, and identify its systematic weaknesses. The guiding questions include: Does ChatGPT “know” grammar, or does it just mimic patterns? If it produces grammatically correct sentences, from where does that knowledge arise? By the end, the goal is for participants to understand the difference between true rule-governed grammatical competence and the model’s statistical approximation of it.

 

Information about registration fees 

Registration form for onsite participation

Registration form for online participation

Registration form for 5-minute talks