A dictionary for a wider readership.
Bhai Kahan Singh Nabha’s Gur Shabad Ratnakar Mahan Kosh, first published in 1930, is a foundational reference for Sikh vocabulary, history, literature and religious terminology. This project brings its entries into a structured multilingual dataset and reading experience.
The dataset.
| Measure | Published scope |
|---|---|
| Dictionary entries | 64,218 |
| Languages | 28, with one JSON file per language |
| Original Gurmukhi bodies available | 60,196 entries · 93.7% |
| Reuse | Dataset packaging: CC BY 4.0 · original 1930 source text: public domain |
Precision across languages.
Mahan Kosh has been one of this project’s most demanding multilingual datasets. Aligning entries, comparing source text and retaining multiple meanings require care, especially where Panjabi, Sanskrit and Persian terminology meet historical and religious context.
How the dataset is checked
English and Hindi serve as reference layers alongside the original Gurmukhi; all language editions are subject to repeated audits. The project’s current workflow uses multiple passes to check entry alignment, compare meanings with the source, resolve terminology in context, and identify duplicated words or phrases. Consistency checks across related entries help catch translations that read fluently but lose a distinction in the original.
Source note: the published repository describes the remaining 25 translated editions as predominantly machine-translated, with approximately 97–100% machine-translation coverage. Its coverage and quality notes remain available with the dataset. For close study, compare the original entry and other scholarly sources.
Where an original entry body is missing, the project records the gap rather than inventing source text. A multilingual entry is a route into the source, not a replacement for it.
Development and seva.
Sukhjeet Singh Sodhi carried out the project’s development and execution, leading to mahankosh.site. Gurpreet Singh sponsored the seva. The underlying work remains credited to Bhai Kahan Singh Nabha; repository contributor credits and licence notices accompany the open dataset.
Open the dataset on GitHub ↗redroyals/mahan-kosh-multilingual · source, coverage notes and licence
Gurpreet Singh · Seva sponsorgithub.com/gurpreet-fe ↗