loading the word pool…
What is in here
Three segments, counted in their own units. They are never added into a single total, because a kośa stem, a padapāṭha occurrence and a generated form are not the same kind of thing.
Segment overlap
How many words carry which combination of segment tags.
Vaidika texts
The render gate
Some keys in the raw union are not padas — a locus or a catalogue numeral that leaked into a headword — and a few cannot be written in Devanagari without loss, which is what an unresolved hyphen-compound looks like. They are withheld rather than coerced into something printable, and counted here so the shortfall is visible.
Sources
Every title here is read from the kośa's own TEI teiHeader, never
hand-typed. Where a header is empty the site prints
title unavailable rather than guessing.
Method & limits
A reference work has to be checkable. This page says what each segment means, where its authority comes from, and what is not yet true.
The three segments
- R · Rūḍha — conventional vocabulary, non-compositional by definition, so it must be enumerated rather than generated. Authority is kośa attestation, cited to the printed page where the digitisation preserves one.
- V · Vaidika — words of the Veda as the padapāṭha segments them, with svara intact as VijayaDV codepoints. Authority is textual occurrence. These are inflected forms, not lemmas.
- Y · Yaugika — forms produced by Pāṇinian derivation, each carrying its full prakriyā chain. Authority is the sūtra chain itself, which is printed on every card so it can be checked.
Known limits
How to read an entry
Sources appear in the order the project owner has set, which lives in an editable file rather than in the code: Śrauta padārtha first, then Veda references, Amarakośa, Vācaspatyam, Śabdakalpadrumaḥ, then the remaining kośas, with Monier-Williams last and de-emphasised. A source with no entry for the word is skipped silently. You can override the order for your browser session — "my order" or "my top 3 only" — with no account.
Each source is named in full and paired with the citation we hold for it, normally its printed page. Definition text is not published yet. What is published is attestation: which kośa records the word, and where. Serving the definitions needs a database rather than static files, and that decision has not been taken.
Typing a word
Devanagari and IAST are detected automatically. SLP1 (kfzRa)
and Harvard-Kyoto (kRSNa) can be chosen explicitly. A trailing
visarga or anusvāra is matched loosely, because kośas cite stems and
printed editions cite nominatives.
Not built yet
The design carries more than this. These parts are named here so nobody mistakes an absent feature for an empty one.
- Stem analysis by the other three schools — Nirukta (Yāska), Sthaulāṣṭhīvi, Śākapūṇi. School A (Vyākaraṇa) is what ships.
- The grammarian-philosophy layer: Bhartṛhari, Nāgeśa, Kāśikā, Mahābhāṣya, Siddhāntakaumudī, attached at the category level.
- Usage and vivakṣā flags, and the guided new-word coinage engine.
- Domain lenses. Śrautapadārthanirvacanam is extracted and staged as the pilot, but is not published: its licence is unverified and its headwords still need a scholar's pass.
- Reverse lookup by meaning, and synonym clusters across kośas. Entry text is not published here — only the fact of attestation — so meanings are not searchable yet.