igorsterner/acs-benchmark

This is a benchmark of minimal pairs of code-switching sentences. Each pair contains one observed code-switching sentence and one manipulated variant. Automatic tokenization and token-based language identification is also provided.

The data creation is described in the following publication:

@inproceedings{sterner-2025-acs,
    author = {Igor Sterner and Simone Teufel},
    title = {Minimal Pair-Based Evaluation of Code-Switching},
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
}