Datasets:

HiTZ
/

EuskanolDS

license: cc-by-nc-4.0
language:
  - eu
  - es
pretty_name: EuskañolDS
size_categories:
  - 10K<n<100K
dataset_info:
  - config_name: gold
    features:
      - name: text
        dtype: string
  - config_name: silver
    features:
      - name: text
        dtype: string
configs:
  - config_name: silver
    data_files:
      - split: train
        path: silver.tsv
  - config_name: gold
    data_files:
      - split: test
        path: gold.tsv

Dataset Card for EuskañolDS

EuskañolDS is a naturally sourced corpus for Basque-Spanish code-switching, created by filtering publicly available corpora in Basque and Spanish.

Dataset Details

Dataset Description

Code-switching (CS) remains a significant challenge in Natural Language Processing (NLP), mainly due a lack of relevant data. In the context of the contact between the Basque and Spanish languages in the north of the Iberian Peninsula, CS frequently occurs in both formal and informal spontaneous interactions. However, resources to analyse this phenomenon and support the development and evaluation of models capable of understanding and generating code-switched language for this language pair are almost non-existent. We introduce a first approach to develop a naturally sourced corpus for Basque-Spanish code-switching. Our methodology consists of identifying CS texts from previously available corpora using language identification models, which are then manually validated to obtain a reliable subset of CS instances.

Language(s) (NLP): Basque (eu), Spanish (es)
License: EuskañolDS is distributed under the same licenses as the corpora it is derived from.

Dataset Sources

Name	Size(Tokens)	Source	Topics
BasqueParl	14M	Parliamentary transcriptions	Political discourse
Heldugazte	37M	Twitter	News, sport, music, nationalist left
Covid-19	57M	Twitter (September 2019 to February 2021)	Covid-19, political issues

Dataset Structure

The dataset has two subsets:

Silver: automatically filtered.
Gold: manually validated.

Splits

Split	Tokens	Instances	Avg. Length
Silver	537,648	20,008	26.87
Gold	36,860	927	39.76

Dataset Instances

Examples from the dataset:

Source	Instance	Translation	Type of CS
HelduGazte	bihar zazpi terditan gora y yo me muerooooooo	tomorrow up at seven thirty and i'm going to die	Intra-sentential
BasqueParl	Por lo tanto, no tengo nada más que añadir. Eta eskerrik asko denoi akordio batera heldu garelako.	Therefore, I don't have anything else to add. And thank you everyone for having reached an agreement.	Inter-sentential
Covid-19	Katxis!Veo a la tropa baja... Eutsi goiari!	Heck! I see the spirits are low... Cheer up!	Emblematic

BibTeX:

@inproceedings{heredia-etal-2025-euskanolds,
    title = "{E}uska{\~n}ol{DS}: A Naturally Sourced Corpus for {B}asque-{S}panish Code-Switching",
    author = "Heredia, Maite  and
      Barnes, Jeremy  and
      Soroa, Aitor",
    editor = "Winata, Genta Indra  and
      Kar, Sudipta  and
      Zhukova, Marina  and
      Solorio, Thamar  and
      Ai, Xi  and
      Hamed, Injy  and
      Ihsani, Mahardika Krisna Krisna  and
      Wijaya, Derry Tanti  and
      Kuwanto, Garry",
      booktitle = "Proceedings of the 7th Workshop on Computational Approaches to Linguistic Code-Switching",
      month = may,
      year = "2025",
      address = "Albuquerque, New Mexico, USA",
      publisher = "Association for Computational Linguistics",
      url = "https://aclanthology.org/2025.calcs-1.1/",
      pages = "1--5",
      ISBN = "979-8-89176-053-0"
}

APA:

Heredia, M., Barnes, J., & Soroa, A. (2025). EuskañolDS: A Naturally Sourced Corpus for Basque-Spanish Code-Switching. In Proceedings of the 7th Workshop on Computational Approaches to Linguistic Code-Switching (pp. 1–5). Association for Computational Linguistics.