ZennyKenny/modern-russian-with-prereform-equivalent

Dataset Card for Russian Modern–Pre-Reform Orthography Parallel Corpus

Dataset Summary

This dataset contains 50,000 parallel text examples of Russian phrases and sentences, aligned between modern Russian orthography (post-1918 spelling reform) and pre-reform Russian orthography (as used before the 1918 spelling changes). It is intended for use in training and evaluating models for Russian orthography conversion, historical text processing, and related NLP tasks.

The dataset includes a mix of:

• Short phrases (e.g., "Мир и порядок" → "Мiръ и порядокъ")
• Complete sentences with varying syntactic structures, time expressions, and subordinate clauses

Supported Tasks and Leaderboards

• Spelling conversion (modern ↔ pre-reform)
• Historical text normalization
• Parallel corpus training for machine translation models between orthographic standards

Languages

• Russian (ru)

Dataset Structure

Columns:

• russian_modern — modern Russian text (UTF-8)
• russian_prereform — pre-reform Russian text (UTF-8)

Number of rows: 50,000

Format: CSV (UTF-8). Can be converted to Parquet.

Data Creation

Modern Russian phrases were programmatically generated from a template-based system with a vocabulary of people, actions, objects, places, and time expressions, combined with pre-written natural sentences.
Pre-reform orthography was produced using a heuristic conversion algorithm, applying common 19th–early 20th century spelling rules:

• Final hard sign (ъ) after hard consonants
• Prepositions в, с → въ, съ before consonants
• і before vowels/й
• Common lexical substitutions with ѣ (e.g., вѣтер, дѣло, лѣс, тѣнь)

Limitations

• The orthography converter is heuristic and may not match all authentic pre-1918 lexical exceptions.
• Phrases are synthetic, not sourced from historical corpora.
• Not suitable for linguistic studies of authentic pre-reform usage without further manual validation.

License

Apache 2.0

Citation

If you use this dataset, please cite as:

@misc{russian_modern_prereform_2025,
  title = {Russian Modern–Pre-Reform Orthography Parallel Corpus},
  year = {2025},
  author = {Kenneth Hamilton},
  howpublished = {\url{https://huggingface.co/datasets/}},
  note = {Parallel corpus of modern and pre-1918 Russian orthography}
}