Elamur Research⌕
☰
home / papers / 2026 / Using Vedalakṣaṇa texts for the validation and normalization of Vedic Corpus

Using Vedalakṣaṇa texts for the validation and normalization of Vedic Corpus

Paper · 2026 · Team Svarupa

Summary

Digital Vedic texts available online, such as VedaWeb, the Vedic Heritage Portal, SanskritDocuments, Wikisource, DCS and the Vedic Scriptures portal, vary in quality. Accent (svara) marks are encoded inconsistently, OCR introduces character errors, and mantras can be missing, duplicated or wrongly segmented.

The Svarupa team is building a curated Vedic corpus to fix this. Most text came from downloadable spreadsheets on the Vedic Scriptures portal, supplemented by web scraping and OCR of printed editions. The data was then normalized for Unicode, formatting and mantra boundaries. Sanskrit scholars compared it against one authoritative edition per text, paying special attention to accents.

As a second, tradition-based check, the authors draw on Vedalakṣaṇa texts. These indigenous manuals record phonological, sandhi and accent particulars, and they help choose the correct Padapāṭha split where the continuous Saṁhitā is ambiguous. The authors also propose regrouping lakṣaṇa rules by linguistic function so they map onto computational tasks.

The available text reports no accuracy figures. The paper's value is methodological: it shows that traditional rule systems can provide verifiable checks for Vedic NLP data.

Key points

  • The validated dataset covers several recensions:
    • Ṛgveda Śākala: 10,552 mantras
    • Sāmaveda Kauthuma: 1,875 mantras
    • Taittirīya (Kṛṣṇa Yajurveda): 631 anuvāk
    • Mādhyandina: 1,975 mantras
    • Kāṇva: 2,086 mantras
    • Atharvaveda Śaunaka: 5,987 mantras
  • The reference editions include the VSM (Vaidika Saṃśodhana Maṇḍala) edition for the Ṛgveda and Shadangi (2024) for the Sāmaveda.
  • The errors corrected were mistyped or OCR-garbled syllables, missing or extra words and mantras, and svara mistakes.
  • An illustrative case is Ṛgveda 8.102.17, where the Āvarṇi lakṣaṇa's list of ā-initial words confirms the reading ājananta.
  • Lakṣaṇa rules do not perform segmentation themselves. They only select among candidate splits, so they offer a partial, text-specific evaluation. Cross-śākhā validation is planned as future work.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 2The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 works

Cite

Cite as Team Svarupa. 2026. Using Vedalakṣaṇa texts for the validation and normalization of Vedic Corpus. Proceedings of the 8th International Sanskrit Computational Linguistics Symposium.

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{svarupa2026using,
  title = {{Using Vedalakṣaṇa texts for the validation and normalization of Vedic Corpus}},
  author = {Team Svarupa},
  booktitle = {Proceedings of the 8th International Sanskrit Computational Linguistics Symposium},
  year = {2026},
  url = {https://aclanthology.org/2026.iscls-1.12/}
}

References

23 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. K. V. Abhyankar. 1974. Veda-Padapāṭha-Carcā: A Study of the Linguistic Analysis of the Padapāṭha. Bhandarkar Oriental Research Institute, Pune, India. Detailed analysis of Vedic word segmentation. find
  2. K. Parameswara Aithal. 1991. Veda-lakṣaṇa: Vedic Ancillary Literature (A Descriptive Bibliography). Franz Steiner Verlag, Stuttgart. Also published by Motilal Banarsidass (Delhi, 1993). find
  3. Bhagyashree Shreeram Bhagwat. 2021. Maitrāyaṇī Padapāṭha: A Study. Vaidika Saṃśodhana Maṇḍala, Pune, India. find
  4. R. N. Dandekar and C. G. Kashikar, editors. 1958–1994. Śrautakośa: Encyclopedia of Vedic Sacrificial Ritual. Vaidika Saṃśodhana Maṇḍala, Pune, India. A multi-volume comprehensive record of śrauta rituals. find
  5. Madhav M. Deshpande. 1975. Critical studies in indian grammarians i: The theory of homogeneity [savarnya]. The University of Michigan, 2:1–103. find
  6. T. N. Dharmadhikari. 1989. Yajñāyudhāni: An Album of Implements Used in Vedic Rituals. Vaidika Saṃśodhana Maṇḍala, Pune, India. Visual and descriptive catalog of ritual tools. find
  7. We provide the results of all the analysis at Svarupa-Vedalakṣaṇa, and the validated data is presented on the Svarupa platform. 14 https://svarupa.org/ ↗
  8. T. N. Dharmadhikari. 2008. The Maitrāyaṇī Saṃhitā: Its Ritual and Language. Adarsha Sanskrit Shodha Samstha, Pune, India. Critical study of the Maitrāyaṇī recension of the Yajurveda. find
  9. Detlef Eichler, 2012–2025. Veda-lakṣaṇa: Traditional Mechanism in Vedic Texts. find
  10. Oliver Hellwig, 2010–2024. The Digital Corpus of Sanskrit (DCS). find
  11. Kātyāyana. 1967. Vājasaneyi Prātiśākhya. Chowkhamba Sanskrit Series Office, Varanasi. Ancillary text for the Shukla Yajurveda. find
  12. S. D. Khadilkar. 2003. Kātyāyana Śulba Sūtra: With English Translation, Explanatory Notes, Diagrams and Articles. Vaidika Saṃśodhana Maṇḍala, Pune, India. Crucial for ancient Vedic geometry. find
  13. Sriram Krishnan, Sepuri Gayathri, and Amba Kulkarni. 2025. Challenges in processing Vedic Sanskrit: Towards creating a normalized dataset for the Ṛgveda-saṃhitā. In Amba Kulkarni and Oliver Hellwig, editors, Computational Sanskrit and Digital Humanities - World Sanskrit Conference 2025, pages 131– 1… find
  14. Śrīpād Dāmodar Sātvalekar. 2022. Atharvaveda Saṃhitā. Chaukhamba Sanskrit Pratishthan, Varanasi, India. Standard Sanskrit text edition of the Śaunaka recension. śrīpād Dāmodar Sātvalekar. n.d. Atharvaveda Saṃhitā. Nag Publishers, Delhi, India. Sanskrit text edition of Atharvaveda Saṃhita. find
  15. Śaunaka. 1927. Rigveda Prātiśākhya. Asiatic Society of Bengal, Calcutta. A foundational Vedalakṣaṇa text for the Rigveda. find
  16. Girijaprasad Shadangi. 2016. Samaveda-Padapāṭha. Dr. Girijaprasad Shadangi, Tirupati, India. First edition research manual for Samaveda word segmentation. find
  17. Girijaprasad Shadangi. 2020. Prakrutigana of Samaveda. Girija Prasad Shadangi, Tirupati, India. Essential for musical notation validation. find
  18. Girijaprasad Shadangi. 2024. Sāmavedasaṃhitā: Kauthumaśākhīyam (Volume 23 of Series). Dr. Girijaprasad Shadangi, Tirupati, India. Primary source for Samaveda validation in modern research. find
  19. Girijaprasad Shadangi. n.d. Ṛktantram: Sāmavedīya Prātiśākhyam (Sabhāṣyam). Girija Prasad Shadangi, Tirupati, India. A Pratisakhya of Samaveda with Sanskrit and Hindi commentary. N. S. Sontakke and T. N. Dharmadhikari. 1970–1972. Taittirīya Saṃhitā: With Padapāṭha and the Commentaries of Bhaṭṭa Bhās… find
  20. Gaṅgeśvarānanda Udāsīna. 1994. Yajurvedasaṃhitā. Yogesvara-guru-gangeshava charitable trust, Bombay, India. find
  21. Vaidika Saṃśodhana Maṇḍala. 1933–1951. Ṛgveda-Saṃhitā: With the Commentary of Sāyaṇācārya (Critical Edition in 5 Volumes). Vaidika Saṃśodhana Maṇḍala, Pune, India. The standard ”Poona Edition” of the Rigveda. find
  22. Albrecht Weber, editor. 1972. Śukla Yajurveda: Madhyandina Saṃhitā. Chowkhamba Sanskrit Series Office, Varanasi, India. Edition in both Madhyandina and Kanva recensions. find
  23. William Dwight Whitney. 1871. The Taittirīya Prātiśākhya: With its Commentary, the Tribhāṣyaratna. American Oriental Society, New Haven. Crucial manual for the Krishna Yajurveda. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/svarupa-2026-vedalak-texts-validation-normalization-vedic/