Can Transformers Process Recursive Nested Constructions, Like Humans?

Yair Lakretz, Théo Desbordes, Dieuwke Hupkes, Stanislas Dehaene


Abstract
Recursive processing is considered a hallmark of human linguistic abilities. A recent study evaluated recursive processing in recurrent neural language models (RNN-LMs) and showed that such models perform below chance level on embedded dependencies within nested constructions – a prototypical example of recursion in natural language. Here, we study if state-of-the-art Transformer LMs do any better. We test eight different Transformer LMs on two different types of nested constructions, which differ in whether the embedded (inner) dependency is short or long range. We find that Transformers achieve near-perfect performance on short-range embedded dependencies, significantly better than previous results reported for RNN-LMs and humans. However, on long-range embedded dependencies, Transformers’ performance sharply drops below chance level. Remarkably, the addition of only three words to the embedded dependency caused Transformers to fall from near-perfect to below-chance performance. Taken together, our results reveal how brittle syntactic processing is in Transformers, compared to humans.
Anthology ID:
2022.coling-1.285
Volume:
Proceedings of the 29th International Conference on Computational Linguistics
Month:
October
Year:
2022
Address:
Gyeongju, Republic of Korea
Venue:
COLING
SIG:
Publisher:
International Committee on Computational Linguistics
Note:
Pages:
3226–3232
Language:
URL:
https://aclanthology.org/2022.coling-1.285
DOI:
Bibkey:
Cite (ACL):
Yair Lakretz, Théo Desbordes, Dieuwke Hupkes, and Stanislas Dehaene. 2022. Can Transformers Process Recursive Nested Constructions, Like Humans?. In Proceedings of the 29th International Conference on Computational Linguistics, pages 3226–3232, Gyeongju, Republic of Korea. International Committee on Computational Linguistics.
Cite (Informal):
Can Transformers Process Recursive Nested Constructions, Like Humans? (Lakretz et al., COLING 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.coling-1.285.pdf