How Often Is This Word Actually Used?
A dictionary is deliberately democratic: a word attested once in a single late commentary gets an entry formatted exactly like ca or bhū. That is what makes dictionaries complete — and what makes them silent about usage. This page adds the missing signal: corpus attestation counts from Oliver Hellwig's Digital Corpus of Sanskrit (DCS, CC BY), a lemmatized, human-validated corpus. The feed used here counts 4,550,704 tokens over 59,282 lemmas.
Sanskrit is extremely top-heavy
A small core of lemmas does most of the work in real texts:
| Most frequent… | Tokens covered | Share of the counted corpus |
|---|---|---|
| 100 lemmas | 1,605,836 | 35.3% |
| 500 lemmas | 2,534,202 | 55.7% |
| 1,000 lemmas | 3,022,015 | 66.4% |
| 2,000 lemmas | 3,503,508 | 77.0% |
Of 4,550,704 tokens counted over 59,282 lemmas (Digital Corpus of Sanskrit, whole-corpus counts).
Two thousand lemmas — roughly one percent of Monier-Williams' 185,803 entries — account for over three quarters of all corpus tokens. For a learner this is the strongest prioritization signal there is: the MW reading guide teaches you how to read an entry; this layer tells you which entries you will actually meet.
The most frequent lemmas
| Rank | Lemma | SLP1 key | POS | Tokens | Per 10,000 tokens |
|---|---|---|---|---|---|
| 1 | ca | ca | ind | 155,088 | 340.8 |
| 2 | tad | tad | pron | 151,248 | 332.4 |
| 3 | na | na | ind | 53,981 | 118.6 |
| 4 | mad | mad | pron | 49,367 | 108.5 |
| 5 | eva | eva | ind | 45,884 | 100.8 |
| 6 | yad | yad | pron | 45,466 | 99.9 |
| 7 | iti | iti | ind | 44,379 | 97.5 |
| 8 | tvad | tvad | pron | 37,263 | 81.9 |
| 9 | idam | idam | pron | 36,787 | 80.8 |
| 10 | tu | tu | ind | 35,004 | 76.9 |
| 11 | api | api | ind | 34,022 | 74.8 |
| 12 | kṛ | kf | 2.Ā. | 33,251 | 73.1 |
| 13 | bhū | BU | 1.Ā. | 32,612 | 71.7 |
| 14 | sarva | sarva | pron | 31,377 | 68.9 |
| 15 | vac | vac | 2.P. | 30,970 | 68.1 |
| 16 | mahat | mahat | adj | 24,957 | 54.8 |
| 17 | as | as | 2.P. | 24,567 | 54.0 |
| 18 | etad | etad | pron | 24,239 | 53.3 |
| 19 | tathā | taTA | ind | 22,741 | 50.0 |
| 20 | tatas | tatas | ind | 21,960 | 48.3 |
| 21 | hi | hi | ind | 19,452 | 42.7 |
| 22 | vā | vA | ind | 19,249 | 42.3 |
| 23 | rājan | rAjan | m | 17,082 | 37.5 |
| 24 | iva | iva | ind | 16,240 | 35.7 |
| 25 | gam | gam | 6.Ā. | 15,950 | 35.0 |
POS tags are the DCS grammar codes (ind indeclinable, pron pronoun, m/f/n
noun by gender, adj adjective, 1.P.-style codes for verb roots by class and pada).
Every lemma is keyed in SLP1, so it joins directly
against Cologne headword keys.
Most headwords are corpus-rare
Only 59,282 lemmas carry a whole-corpus count (83,277 appear in the source layer at all) — against 185,803 MW entries. The gap is not an error: large dictionaries carry enormous tails of words inherited from the indigenous lexicographic tradition (MW's "L." = lexicographers marker; see the SKD and VCP kośa pages), attested rarely or only in lexica. When a dictionary entry cites only lexicographers and the corpus count is zero, you are looking at a word of the dictionary tradition rather than of surviving usage — a distinction no dictionary layout shows you, but the corpus layer does.
Words with a history
The feed carries a per-period vector for each lemma (DCS's chronological buckets —
numbered slots ending at an approximate year, e.g. 1 -800 ≈ texts up to 800 BCE,
5 1200 ≈ up to 1200 CE; the labels 3200 and 4700 are slots 3 and 4 — up to
200 CE and 700 CE — with the space lost in the upstream export; plus the undatable
genre buckets 9 Vedic, 11 Epic, 12 Classic). Frequency profiles differ
dramatically:
| Lemma | Rank | 9 Vedic | 1 -800 | 2 -300 | 3200 | 4700 | 5 1200 | 6 1700 | 7 1900 | 11 Epic | 12 Classic |
|---|---|---|---|---|---|---|---|---|---|---|---|
| ūti | 1667 | 49% | 49% | 0% | 0% | · | 0% | · | · | · | 1% |
| rayi | 1874 | 49% | 47% | 2% | · | 0% | 0% | · | · | 0% | 0% |
| vṛṣan | 1258 | 48% | 48% | 1% | 1% | 0% | 0% | · | · | 1% | 0% |
| vaiśampāyana | 559 | · | · | · | 49% | 0% | 1% | · | · | 49% | 1% |
| vaidehī | 1691 | 0% | · | · | 49% | 1% | 0% | 0% | · | 49% | 1% |
| pārada | 788 | · | · | · | 0% | 0% | 18% | 22% | 8% | 1% | 50% |
| gandhaka | 569 | · | · | · | 0% | · | 20% | 28% | 2% | 0% | 51% |
| abhraka | 1199 | · | · | · | · | · | 22% | 26% | 2% | · | 50% |
Share of each lemma's period-datable tokens per DCS period bucket (rows sum to 100%; · = no tokens in that bucket).
- ūti "help, favour", rayi "wealth", vṛṣan "bull, mighty" — Vedic workhorses that almost vanish afterwards. For these, Grassmann (a dedicated Rigveda dictionary with complete attestation) serves you better than any general-purpose dictionary.
- Vaiśampāyana, Vaidehī — epic names concentrated in the
11 Epicbucket: Mahābhārata/Rāmāyaṇa reading is PWG and MW territory, both of which cite the epics massively (see what each dictionary quotes). - pārada "mercury", gandhaka "sulphur", abhraka "mica" — alchemical (rasaśāstra) vocabulary whose counts sit in the late buckets. Late technical Sanskrit is real Sanskrit too, and its vocabulary barely overlaps the poetic core.
The same lemma-level profiles power the corpus-era view in the dictionary landscape page's date reasoning: which dictionary to open depends on when your text was written.
The same words in the accented Rigveda
Classical learners meet Sanskrit stripped of its pitch accent; the encoding guide explains udātta notation abstractly. This layer shows real attested accented word-forms for lemmas you may already know from the frequency work above — drawn from the VedaWeb bulk export of the Zurich Rigveda annotation corpus (Casaretto et al. 2025, CC BY 4.0), whose word-split carries an udātta mark on every accented syllable:
| Measure | Value |
|---|---|
| Rigveda stanzas in the accented word-split | 10,552 |
| Word tokens carrying a lemma annotation | 164,758 |
| Top-2,000 DCS lemmas attested in the Rigveda slice | 845 of 2,000 |
| Share of all DCS tokens those lemmas carry | 45.3% |
Join: VedaWeb lemma keys normalised and matched against the SLP1-keyed top-2,000 DCS lemmas — see the trust block for what the join misses.
Under half of today's most frequent Classical vocabulary is already attested in the Rigveda slice (845 of 2,000 joined lemmas carrying 45.3% of all DCS tokens) — and where a lemma does appear, seeing its accent live teaches what no paradigm table quite can:
| DCS rank | Lemma | SLP1 key | RV tokens | Attested accented forms |
|---|---|---|---|---|
| 1 | ca | ca | 1,093 | ca · ca + |
| 3 | na | na | 2,099 | ná |
| 4 | mad | mad | 236 | mátsi · mádema · ámadan |
| 5 | eva | eva | 209 | evá · évāḥ · évān |
| 7 | iti | iti | 83 | íti |
| 10 | tu | tu | 58 | tú · tú + |
| 11 | api | api | 90 | ápi |
| 12 | kṛ | kf | 1,213 | kr̥dhi · kr̥tam · kr̥ṇván |
| 13 | bhū | BU | 862 | bhava · bodhi · bhuvat |
| 14 | sarva | sarva | 68 | sárvā · sárve · sárvam |
| 15 | vac | vac | 213 | voce · óktā · vocaḥ |
| 19 | tathā | taTA | 15 | táthā |
| 20 | tatas | tatas | 23 | tátas |
| 21 | hi | hi | 824 | hí · hitám · hinóti |
| 22 | vā | vA | 345 | vā · vāti · vātu |
| 23 | rājan | rAjan | 247 | rā́jā · rājan · rā́jan |
| 24 | iva | iva | 1,024 | iva |
| 25 | gam | gam | 632 | gata · gahi · gamat |
| 26 | deva | deva | 1,646 | deva · déva · devám |
| 27 | dṛś | dfS | 125 | dr̥śé · dárśam · ádarśi |
Forms are verbatim from the Casaretto et al. accented word-split (acute ́ = udātta). Up to three shortest distinct forms per lemma.
A few of those with room to breathe:
- kṛ
kf— DCS rank 12, 1,213 RV tokens, e.g. kr̥dhi - agni
agni— DCS rank 45, 1,724 RV tokens, e.g. ágne - vāc
vAc— DCS rank 166, 131 RV tokens, e.g. vā́k - brāhmaṇa
brAhmaRa— DCS rank 87, 15 RV tokens, e.g. brāhmaṇáḥ
Even a single glance shows placement variety that printed paradigms flatten: vā́k has its
udātta on the stem vowel while the locative vācí shifts it rightward, and agni's vocative,
theme-vowel and accusative shapes carry the mark in three different places (ágne, agne,
agním).
- Evidence:
src/data/corpus-frequency.json(top 2,000 lemmas by corpus rank + whole-corpus stats), regenerated byscripts/build-corpus-frequency.mjsfromkosha/data/frequency/lemma_frequency.tsv(83,277 lemmas), itself built from the VisualDCS M9 archive of DCS data. Upstream: Digital Corpus of Sanskrit (Oliver Hellwig), CC BY. The Vedic accents section addssrc/data/vedic-accent.json, regenerated byscripts/build-vedic-accent.mjs(SHA-256 edge contract over the upstream bytes,--checkmode) from the landed VedaWeb bulk export — the one-time acquisition in VisualDCS/non-derived/vedaweb (H096, 08-07-2026); this site never queriesvedaweb.uni-koeln.dedirectly. Every table and percentage on this page is computed from the committed feeds at build time — nothing is hand-typed. - Limitations: the frequency tables see the top-2,000 slice, not all 83,277 lemmas. DCS is genre-skewed (its later periods are dominated by śāstric and alchemical texts, which is why the "late-heavy" examples are rasaśāstra terms — a corpus fact, not a claim about all late Sanskrit). Period bucket boundaries are DCS-coded and approximate (the upstream QL layer README flags the label-to-boundary mapping as not verified cell-by-cell). Counts are lemmatized tokens from DCS's analyses — a different corpus would give different numbers. The Vedic join is a lower bound: RV lemma citation forms differ from Classical ones for a real tail of stems (e.g. the pronouns cited as sá- ~ tá- where the corpus layer keys tad), so lemmas can be RV-attested yet missing from the join; accent marks are stripped only for keying and always verbatim on display.
- Rights: DCS is CC BY; the feed is a derived aggregation with attribution
(evidence record §1 in
NON_COLOGNE_SOURCES.md). The vedic-accent feed: Casaretto et al. (2025)/VedaWeb, Universität zu Köln, CC BY 4.0 (rights confirmed per H359; evidence record §3 in the same file). - Owner repos: DCS (upstream) → VisualDCS →
kosha (frequency layer) → this repo (rendering);
VedaWeb bulk export → VisualDCS
non-derived/vedaweb→ this repo. - Built by: Fable 5 (
claude-fable-5), 07-07-2026, handoff H280; Vedic-accent section by OxAlpha (opencode/x-preview-f-free), 26-08-2026.
See also
- How machines read Sanskrit — the morphology & segmentation layer
- The dictionary landscape — era × novelty × size for all 43 dictionaries
- What each dictionary quotes — the citation-frequency layer
- Reading Monier-Williams — entry structure, symbols, dhātu-tracing