A Targeted Assessment of Incremental Processing in Neural Language Models and Humans
Authors
Authors
- Roger Levy
- Ethan Wilcox
- Pranali Vani
Authors
- Roger Levy
- Ethan Wilcox
- Pranali Vani
Published on
10/10/2021
We present a targeted, scaled-up comparison of incremental processing in humans and neural language models by collecting by-word reaction time data for sixteen different syntactic test suites across a range of structural phenomena. Human reaction time data comes from a novel online experimental paradigm called the Interpolated Maze task. We compare human reaction times to by-word probabilities for four contemporary language models, with different architectures and trained on a range of data set sizes. We find that across many phenomena, both humans and language models show increased processing difficulty in ungrammatical sentence regions with human and model ‘accuracy’ scores (a la ` Marvin and Linzen (2018)) about equal. However, although language model outputs match humans in direction, we show that models systematically under-predict the difference in magnitude of incremental processing difficulty between grammatical and ungrammatical sentences. Specifically, when models encounter syntactic violations they fail to accurately predict the longer reaction times observed in the human data. These results call into question whether contemporary language models are approaching human-like performance for sensitivity to syntactic violations.
Please cite our work using the BibTeX below.
@inproceedings{wilcox-etal-2021-targeted,
title = "A Targeted Assessment of Incremental Processing in Neural Language Models and Humans",
author = "Wilcox, Ethan and
Vani, Pranali and
Levy, Roger",
booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
month = aug,
year = "2021",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2021.acl-long.76",
doi = "10.18653/v1/2021.acl-long.76",
pages = "939--952",
}