English

TADA: Task-Agnostic Dialect Adapters for English

Computation and Language 2023-11-15 v1

Abstract

Large Language Models, the dominant starting point for Natural Language Processing (NLP) applications, fail at a higher rate for speakers of English dialects other than Standard American English (SAE). Prior work addresses this using task-specific data or synthetic data augmentation, both of which require intervention for each dialect and task pair. This poses a scalability issue that prevents the broad adoption of robust dialectal English NLP. We introduce a simple yet effective method for task-agnostic dialect adaptation by aligning non-SAE dialects using adapters and composing them with task-specific adapters from SAE. Task-Agnostic Dialect Adapters (TADA) improve dialectal robustness on 4 dialectal variants of the GLUE benchmark without task-specific supervision.

Keywords

Cite

@article{arxiv.2305.16651,
  title  = {TADA: Task-Agnostic Dialect Adapters for English},
  author = {Will Held and Caleb Ziems and Diyi Yang},
  journal= {arXiv preprint arXiv:2305.16651},
  year   = {2023}
}

Comments

5 Pages; ACL Findings Paper 2023

R2 v1 2026-06-28T10:47:09.360Z