We’re delighted to invite you to the on-site tutorial at COLING 2025 that will discuss the latest work on bridging the worlds of linguistic theory with Large Language Models: “Bridging Linguistic Theory and AI: Usage-Based Learning in Humans and Machines.”
For More information, visit: https://sites.google.com/view/linguistic-theory-and-ai/
The takeaways of this tutorial, which will be held in-person, will be an overview of the shared and divergent aspects of human and machine usage and data-driven learning, outlined from the theoretical perspective of usage-based psycholinguistic theory, with an emphasis on how this can shed light on the capabilities and limitations of LLMs, including multimodal models. This will serve as the bedrock for guiding participants and the NLP community towards more informed evaluation of large, pre-trained models, as well as energising solutions drawing upon the multi-modal information and linguistic theory that enriches language and many dimensions of interaction.
Background: Unlike our past NLP tools, such as syntactic parsers and automatic semantic role labelling, LLMs lack grounding in linguistic theory. Instead, their development is based on the encoder-decoder architecture, which was originally designed for sequence- to-sequence tasks, specifically translation. This dichotomy impedes methods for evaluating LLMs, as their performance on meta-linguistic tasks, such as semantic role labelling, which previously served as benchmarks for the individual components in an NLP pipeline, are poor predictors of their fluency on downstream applications. However, the fact that LLMs, designed primarily to meet information-theoretic needs, can capture any linguistic information at all is fascinating. Additionally, it offers a novel foundation for exploring what can be achieved through exposure to information alone.
Therefore, it has been compelling to turn to usage-based theories of language, such as Construction Grammar, to establish experimentally validated structures of language that speakers of a given language consistently recognise and are able to generalise over. We can then compare such structures to the linguistic structure that we can probe for within LLMs.
For More information, visit: https://sites.google.com/view/linguistic-theory-and-ai/
We look forward to seeing you at COLING 2025 in January.
On behalf of, Claire Bonial, Harish Tayyar Madabushi, Nikhil Krishnaswamy, James Pustejovsky