Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Geser–Indonesian–English Parallel Corpus

A manually curated trilingual dataset in Geser (Bahasa Seram dialek Geser), Indonesian, and English — created by native speakers to document and preserve an endangered language of the Maluku archipelago, Indonesia.


Dataset Description

This dataset contains ~7,000 sentence pairs across three languages:

Language Code Notes
Indonesian id Standard Bahasa Indonesia
English en Standard English
Geser geser Dialect of Bahasa Seram, Maluku

The data was entirely handcrafted by native speakers of Geser, making it one of the very few documented corpora for this language. Every sentence was written, reviewed, and validated by fluent speakers — no machine translation or automated alignment was used.


Authors & Origin

This dataset was created by:

  • Rakiba Kelian — Native speaker of Geser (Bahasa Seram, Dialek Geser)
  • Farida Kelian — Native speaker of Geser (Bahasa Seram, Dialek Geser)

Both authors are from the Seram island region of Maluku, Indonesia, and contributed their linguistic knowledge to document this language for future generations.


Language Background

Geser is a dialect of Bahasa Seram, a language spoken in the Seram island area of the Maluku province, eastern Indonesia. It is classified as an endangered language, with the majority of its speakers coming from older generations (born in the 1970s–1980s), and its usage is gradually being eroded by modern societal and technological changes.

  • Language family: Austronesian > Malayo-Polynesian > Central Malayo-Polynesian
  • Region: Seram Island, Maluku, Indonesia
  • Status: Endangered / Severely endangered
  • Speaker community: Small, aging population; intergenerational transmission is declining

Preserving and documenting this language is urgent. This dataset is one of the first structured linguistic resources for Geser made available to the public and research community.


Data Fields

Field Type Description
id_sentence string Sentence in Indonesian (Bahasa Indonesia)
en_sentence string Sentence in English
geser_sentence string Sentence in Geser (Bahasa Seram, Dialek Geser)

Sample Data

id_sentence en_sentence geser_sentence
Saya pergi ke pasar I go to the market (geser translation)
Dia makan nasi She eats rice (geser translation)
Anak itu bermain di pantai The child plays on the beach (geser translation)

Intended Uses

✅ Direct Use

  • Low-resource machine translation (Indonesian ↔ Geser, English ↔ Geser)
  • Language documentation and archival research
  • Linguistic analysis of Geser grammar, phonology, and morphology
  • NLP benchmarking for extremely low-resource language pairs
  • Training multilingual language models with indigenous language data

⚠️ Out-of-scope Use

  • This dataset should not be used to build tools that misrepresent or distort the Geser language
  • Commercial use without attribution to the authors is discouraged

Data Collection & Quality

  • Method: Fully manual — written sentence by sentence by native speakers
  • Verification: Reviewed by the authors themselves as native speakers
  • No machine translation was used at any stage
  • Domain coverage: General/everyday language, covering a wide range of topics and sentence structures

Limitations

  • Dataset reflects the Geser dialect specifically; it may not represent other dialects of Bahasa Seram
  • As the language is endangered, some vocabulary or expressions may not have widely agreed-upon standard orthography
  • ~7,000 sentences is a small corpus by NLP standards; results on downstream tasks may be limited without additional augmentation

Citation

If you use this dataset in your research or applications, please cite and credit the authors:

@dataset{kelian2024geser,
  title     = {Geser–Indonesian–English Parallel Corpus},
  author    = {Rakiba Kelian and Farida Kelian and Abdul Wahid Rukua},
  year      = {2024},
  language  = {Geser, Indonesian, English},
  note      = {Manually created by native speakers of Geser (Bahasa Seram, Dialek Geser)},
}

License

This dataset is released under a Custom Research-Only License.

✅ Permitted

  • Academic and non-commercial research
  • Linguistic documentation and language preservation work
  • NLP experiments and publications (with proper attribution)
  • Educational use

❌ Not Permitted without explicit written permission

  • Any commercial use of this dataset or models trained on it
  • Integration into commercial products, APIs, or services
  • Redistribution for profit

📩 Commercial Use Requests

For commercial licensing inquiries, please contact the authors directly:

  • Rakiba Kelian
  • Farida Kelian
  • Abdul Wahid Rukua

Use of this dataset implies agreement to these terms. The authors retain all rights not explicitly granted above. I create this for my tribe please be aware!


Acknowledgements

This dataset is a contribution to the documentation and revitalization of endangered languages in Indonesia. The authors' dedication to preserving the Geser language through this manual effort is invaluable for linguistics, NLP research, and the Geser-speaking community.

"Setiap bahasa yang punah adalah perpustakaan yang terbakar." "Every language that dies is a library that burns."

Downloads last month
9