Linguists should learn to love speech-based deep learning models

TL;DR

This paper advocates for linguists to embrace speech-based deep learning models, emphasizing their importance over text-based models for understanding human language.

Contribution

It introduces a framework connecting deep learning systems with linguistic theories and highlights the need to focus on audio-based models for linguistic research.

Findings

Audio models better capture spoken language nuances

Text-based models miss key linguistic features

Speech models enhance understanding of human language

Abstract

Futrell and Mahowald present a useful framework bridging technology-oriented deep learning systems and explanation-oriented linguistic theories. Unfortunately, the target article's focus on generative text-based LLMs fundamentally limits fruitful interactions with linguistics, as many interesting questions on human language fall outside what is captured by written text. We argue that audio-based deep learning models can and should play a crucial role.