Datasets:

Modalities:
Text
Formats:
parquet
License:
fdelucaf's picture
Update language code
ce2f3cc verified
|
Raw
History Blame Contribute Delete
11.2 kB
metadata
language:
  - ca
  - oc
multilinguality:
  - multilingual
pretty_name: Catalan-Aranese Parallel Corpus
size_categories:
  - 100M<n<1B
task_categories:
  - translation
license: cc-by-4.0

Dataset Card for Catalan-Aranese Parallel Corpus

Dataset Description

Dataset Summary

A bilingual parallel corpus for the low-resource language pair Catalan-Aranese. Built by aggregating and filtering multiple public sources, along with data obtained through direct data sharing with external partners, it provides sentence-level alignments for training Machine Translation systems. The dataset includes both authentically parallel data as well as synthetic Catalan translations generated from Aranese monolingual data using SalamandraTA 7B Instruct.

Supported Tasks and Leaderboards

The dataset is primarily designed for Machine Translation between Catalan and Aranese. Typical uses include supervised MT training, fine-tuning multilingual models, and data augmentation.

Languages

The dataset contains one parallel language pair: Catalan–Aranese (ca-oc_arn), totaling 539,110 sentence pairs.

Language pair Codes Size (sentences)
Catalan-Aranese ca-oc_arn 539,110

The Aranese language

Aranese is a variant of the Occitan language spoken in the Aran Valley, in the province of Lerida, Spain. The Occitan language belongs to the Romance or Neo-Latin language group and consists of six dialect groups: Vivaro-Alpine, Provençal, Limousin, Auvergnat, Languedocien and Gascon. Aranese is a variant of the Gascon dialect.

According to the 1978 Statute of Catalonia, Aranese is subject to teaching and protection. The Law on Administrative Autonomy of the Aran Valley establishes that Aranese is a co-official language in the Aran Valley, along with Catalan and Spanish. In accordance with these regulations, Aranese is taught at all levels of compulsory education and is also used for communication among the public administrations and with the general public.

Dataset Structure

Data Instances

The dataset is provided in parquet format. Each row contains a parallel sentence pair with the following structure:

{
  "l1_sentence": "Example sentence in first language",
  "l2_sentence": "Example sentence in second language",
  "l1": "ca",
  "l2": "oc_arn"
}

Data Fields

  • l1_sentence: The sentence in the first language (string)
  • l2_sentence: The parallel sentence in the second language (string)
  • l1: ISO 639-1 code for Catalan (string)
  • l2: specific language code for Aranese (string)

Data Splits

The dataset contains a single split: train.

Dataset Creation

Curation Rationale

As an extremely low-resource language, Aranese lacks official representation in the ISO 639 standard for language name codes, where only the generic code for Occitan (OC) is available. While some systems do provide specific codes for Aranese, such as Glottolog ("aran1260") and IETF ("oc-aranes"), in the NLP and digital AI resource landscape, the generic OC code is predominantly used. This creates a significant challenge: the vast majority of publicly available resources (datasets and language models) fail to distinguish between Occitan variants, resulting in data that mixes different varieties and consequently exhibits poor linguistic quality and specificity. Similarly, machine translation models often produce outputs that conflate various Occitan variants.

With this dataset and other resources we are releasing, we aim to promote deeper research into these linguistic variants and contribute to improving the quality of machine translation systems. By providing textual data resources specifically focused on the Aranese variant of Occitan, we seek to enable more precise and linguistically accurate NLP applications. For this purpose, we have adopted a specific code ("oc_arn") to label our data, explicitly distinguishing Aranese from other Occitan varieties.

This dataset is therefore aimed at promoting the development of Machine Translation between Catalan and Aranese, supporting research in bilingual and multilingual NLP with proper linguistic granularity, and facilitating the development of translation systems that respect and preserve the unique characteristics of low-resource language varieties.

Source Data

Initial Data Collection and Normalization

The corpus is a combination of authentic Catalan-Aranese parallel data and synthetic Catalan translations generated from Aranese monolingual data. Data was collected via direct data sharing agreements between the BSC and other parties, as well as from public web-based sources.

Bilingual source datasets:

  • DOGC: Parallel text extracted from Diari Oficial de la Generalitat de Catalunya
  • JS Translations: Aranese-Catalan translations produced by a professional translator and obtained through direct data sharing
  • Parallel text collected from several public web-based sources

Monolingual source datasets:

Synthetic Data Generation:

For monolingual Aranese data, synthetic Catalan parallel data was created by translating using SalamandraTA 7B Instruct.

The following monolingual datasets were translated to provide the Catalan side:

Dataset Size (sentences)
PILAR & IEA 392,300
Public web sources 48,536
Escaletas TV3 2,944
Total 443,780

Data Filtering and Normalization:

The data underwent minimal filtering due to data scarcity:

  • Normalization: Text was minimally normalized using Bifixer to ensure consistency and quality.
  • Deduplication: The filtered datasets were deduplicated to remove redundant sentence pairs.

The filtered and normalized datasets were then concatenated to form the final corpus.

Who are the source language producers?

Annotations

Annotation process

The dataset does not contain any manual annotations beyond the parallel alignments, which were either preserved from source datasets or validated through automated alignment scoring.

Who are the annotators?

[N/A]

Personal and Sensitive Information

Given that this dataset is derived from pre-existing datasets that may contain crawled data, and that no specific anonymisation process has been applied, personal and sensitive information may be present in the data. This needs to be considered when using the data for training models.

Considerations for Using the Data

Social Impact of Dataset

By providing this resource specifically focused on the Aranese variant of Occitan, we aim to address a critical gap in NLP resources for extremely low-resource languages. The conflation of linguistic variants under generic language codes (such as using OC for all Occitan varieties) has historically resulted in lower-quality NLP tools that fail to respect the unique characteristics of individual language varieties. This has a direct impact on speaker communities, as translation systems and language technologies that mix variants can produce outputs that are linguistically inaccurate or culturally inappropriate.

Furthermore, by making high-quality Aranese data publicly available, we enable researchers and developers to create technologies that better serve minority language communities, respecting their linguistic identity and contributing to the vitality and continued use of Aranese in digital contexts.

Discussion of Biases

No specific bias mitigation strategies were applied to this dataset beyond deduplication and minimal quality filtering. Inherent biases may exist within the data, reflecting the biases present in the source datasets, which include web-crawled content, subtitles, news articles, and other user-generated or institutionally produced text. Users should be aware that the datasets contains synthetically generated Catalan text, which may reflect biases present in the translation model used.

Other Known Limitations

The dataset contains predominantly data from the administrative and legal domains, as well as news articles. Application of this dataset in other domains such as biomedical, technical, or other specialized fields would be of limited use. Additionally, the synthetic Catalan data may not achieve the same quality or naturalness as naturally parallel data.

Additional Information

Dataset Curators

Language Technologies Unit at the Barcelona Supercomputing Center (langtech@bsc.es).

Funding

This work has been promoted and financed by the Government of Catalonia through the Aina Project.

This work is funded by the Ministerio para la Transformación Digital y de la Función Pública - Funded by EU – NextGenerationEU within the framework of ILENIA Project with reference 2022/TL22/00215337.

Acknowledgements

We gratefully acknowledge the following individuals and organizations for their valuable contribution of data to this corpus:

Licensing Information

This work is licensed under a Creative Commons Attribution 4.0 International licence.

Citation Information

[N/A]

Contributions

[N/A]