LianHong/zomi-monolingual-corpus

Zomi Monolingual Corpus v1.0

The Zomi Monolingual Corpus v1.0 contains 363,401 cleaned, deduplicated, reviewed, and permission-approved Zomi sentences. Zomi is represented with the ISO 639-3 language code ctd (Tedim Chin).

Quick start

from datasets import load_dataset

dataset = load_dataset("LianHong/zomi-monolingual-corpus", split="train")
print(dataset.num_rows)  # 363401
print(dataset[0]["zomi_text"])

Data fields

Field	Type	Description
sentence_id	string	Permanent ZOMI-###### identifier
zomi_text	string	One Zomi sentence
topic	string	Controlled topic label
source_type	string	Source medium/type
source_name	string	Source name
permission_status	string	Approved for every v1.0 row
region	string	Associated region(s), or Unknown
spelling_standard	string	Spelling annotation
reviewer_1	string	Biak Hong for every v1.0 row
reviewer_2	string	Blank in v1.0
review_status	string	approved for every v1.0 row
notes	string	Provenance and extraction notes

Data splits

Only a train split is provided. This reflects the nature of a general monolingual corpus; users should construct and document task-specific validation and test sets without silently changing permanent sentence IDs.

Sources

• Zomi Daily: 354,042 retained sentences.
• Agape Digest Magazine: 9,359 retained sentences.

Written permission for the requested public corpus uses was confirmed received by corpus owner Biak Hong on 2026-08-26. Permission evidence is retained privately by the maintainer.

Curation and review

The corpus was screened for duplicate IDs/text, English/non-Zomi rows, fragments, headings, navigation and metadata text, captions, lists, and direct contact/privacy risks. The excluded audit archive is not distributed. Biak Hong confirmed all retained rows reviewed and good. Final release verification found 363,401 unique IDs and zero audit errors.

Limitations

• This is monolingual data and does not provide Zomi-English alignments.
• Two publishers dominate the source distribution.
• Region is Unknown for 267,456 rows.
• Spelling-standard metadata is incomplete.
• A single primary reviewer is recorded in v1.0.
• Source and topical bias should be measured for every downstream task.

Responsible use

Preserve sentence IDs in derived data, document filters and splits, and involve qualified Zomi reviewers in evaluation. Do not use the corpus for surveillance, sensitive-trait inference, targeting, or attempts to reconstruct excluded privacy-risk rows.

License and citation

CC BY 4.0. See LICENSE and ATTRIBUTION.md in this dataset repository.

Hong, B. (2026). Zomi Monolingual Corpus v1.0 (Version 1.0.0) [Dataset]. Zenodo. https://doi.org/10.5281/zenodo.22132443

Integrity

Canonical CSV SHA-256:

7c84b835e45fdb26aef8822367dbc698988c5a28bdfd7d992cc50bed011353fd