A multi-register sentence corpus for the Hmar language (hmr, ISO 639-3), containing 260,492 train sentences and 5,316 evaluation sentences (~4.15 million words).

Maintained by the Hmar Heritage Foundation (hmarheritage.pages.dev).

Overview

Language: Hmar (hmr, ISO 639-3, Glottolog: hmar1241)
Family: Zo Languages
Volume: 260,492 train sentences | 5,316 evaluation sentences (265,808 total, ~4.15M words)
Validation: 100% verified via hmaraniam zero-dependency language detector
Format: JSONL / Hugging Face Datasets
License: MIT
DOI: 10.57967/hf/10421

Register Distribution

Linguistic Register	Verified Sentences	% of Corpus	Description
Classical & Canonical	49,339	18.6%	Formal historical and theological literature
Literary Prose & Educational	46,327	17.4%	Cultural essays, school reader texts, historical chronicles
Conversational & Orthography	27,784	10.5%	Natural community dialogues and spelling discussions
Contemporary Web & Commentary	141,238	53.1%	Modern digital public discourse and social commentary
Poetry & Cultural Lyrics	1,120	0.4%	Traditional and contemporary lyrical compositions
Total	265,808	100.0%	~4.15 Million Words

Dataset Structure

{"text": "Nangni sininrim le phur rik phur po pohai, ka kuomah hung unla, keiin inchâwltir ka ti cheu."}

Usage

from datasets import load_dataset

dataset = load_dataset("hmar-heritage-org/sentences")
print(f"Train sentences: {len(dataset['train']):,}")
print(f"Test sentences:  {len(dataset['test']):,}")

Source & Provenance

Curated, segmented, normalized, and deduplicated across classical literature, cultural chronicles, community dialogues, and contemporary web registers for low-resource NLP research and language modeling. Language identification and syntactic verification conducted using the Foundation's hmaraniam engine. Emoji boundaries and Unicode NFC orthography are fully standardized.

Limitations & Sociolinguistic Overlap

While all sentences undergo strict filtering via the Foundation's hmaraniam detector:

Cross-Dialectal Overlap: Due to the close phylogenetic kinship and mutual intelligibility between Hmar (hmr) and sibling Zo varieties—particularly Mizo (lus)—sentences lacking dialect-exclusive markers can overlap across varieties, or feature colloquial particle shifts (such as Mizo "A ṭha khawp mai" vs. canonical Hmar "A ṭha khawp el").
Digital Code-Switching: Community discussions from web and social media sources naturally reflect modern bilingualism and code-switching across Northeast regional languages.
Dialect Disambiguation: Sentences containing distinctive morphology and vocabulary leverage hmaraniam's frequency-weighted exclusive particle and lexicon tables to enforce high Hmar dialectal purity.

Citation

@dataset{hmar_heritage_sentences_2026,
  author    = {{Hmar Heritage Foundation}},
  title     = {sentences: A Sentence Corpus for Hmar},
  year      = {2026},
  publisher = {Hugging Face},
  doi       = {10.57967/hf/10421},
  url       = {https://huggingface.co/datasets/hmar-heritage-org/sentences},
  keywords  = {Hmar, hmr, hmar1241, Zo Languages},
  note      = {Language: Hmar (ISO 639-3: hmr, Glottolog: hmar1241). Language family: Zo Languages.}
}