Ad Fontesto the sources

Word Atlas

Distribution graphs for nouns and verbs across the entire Greek and Latin corpora · Day 55

For every noun and verb in the library, a bar graph of its occurrences per work — raw counts, with a switch to occurrences per 1,000 words of each work (AlHaTorah's permille idea). 24,462 Greek lemmas (1,379,105 occurrences) and 15,873 Latin lemmas (956,543 occurrences) across 380 works.

Search by lemma

Browse by letter

Greek

Α (3,783) Β (576) Γ (524) Δ (1,402) Ε (3,368) Ζ (161) Η (213) Θ (521) Ι (637) Κ (2,134) Λ (536) Μ (1,223) Ν (417) Ξ (75) Ο (901) Π (2,770) Ρ (265) Σ (2,254) Τ (730) Υ (570) Φ (639) Χ (522) Ψ (123) Ω (118)

Latin

A (2,003) B (504) C (1,920) D (886) E (829) F (477) G (371) H (423) I (1,151) J (60) K (3) L (509) M (862) N (338) O (486) P (1,496) Q (70) R (634) S (1,588) T (631) U (519) X (6) Z (107)

Most frequent greek lemmas

Most frequent latin lemmas

Method

Two tiers

EXACT — the biblical texts (Septuagint Greek, Vulgate Latin) are counted from the curated per-token lemma and part-of-speech morphology already in the LXX & Vulgate reader: Greek lemmas from openscriptures/Stanza, Latin lemmas from lascivaroma. 58 Greek + 48 Latin biblical books.

AUTOMATIC — the classics, Fathers and Second Temple works are counted through the Days 28–29 per-work form-to-lemma maps (LXX reader morphology + Perseus Ancient Greek Dependency Treebank v2.1 + Lingua Philonica lemmatizer for Greek; Whitaker's Words data for Latin). Each token is assigned once, to its highest corpus-frequency candidate; a form shared with a stoplisted function word is assigned to the function word and drops out, so it never inflates a noun or verb (this is what keeps est from counting as “to eat” and sum from counting as “to take”). 225 Greek + 49 Latin works. Automatic counts are never presented as exact.

Accuracy

Hand-checked sample accuracy: 92.5% (555/600 tokens, token-weighted). By stratum:

Stratified hand check: 150 tokens per stratum (Greek prose/verse, Latin prose/verse) drawn from the tokens the Atlas counts; verdict per distinct (form, lemma, part of speech), applied token-weighted. A token counts as correct only when both the lemma and the part of speech are right in context.

What is counted

Only nouns and verbs. Everything else is out, and the closed function-word classes are removed first by the published stoplist below — articles, conjunctions, particles, prepositions, pronouns — plus the auxiliaries εἰμί / sum (grammatically verbs, statistically noise). Counts only: no per-occurrence positions are stored in v1. Part of speech comes from the dictionary headword (LSJ / Lewis & Short, checked against the treebank and Whitaker paradigms); proper names count as nouns. Dialogue speaker labels (ΣΩ., ΕΛΕΝΗ:) are not words of the text and are not counted.

Known limits: a lemma whose forms are shared with a function word can be under- or over-counted where the dictionary headword is the verb (Latin eo covers the adverb too); Greek lemmas that differ only by accent share one key (the reader's convention); participles tagged as adjectives in the Latin reader morphology are not counted as verbs in the EXACT tier, while the AUTOMATIC tier counts them as verbs. 180,255 Greek biblical tokens had no lemma in the reader morphology and are not counted; the Vulgate has none missing.

Greek stoplist (76 lemmas)

αλλα, αλληλων, αλλοσ, αμα, αμφι, αν, ανα, αντι, απο, αυτοσ, γαρ, γε, δε, δη, δια, εαν, εγω, ει, ειμι, εισ, ειτε, εκ, εκαστοσ, εκεινοσ, εν, ενθα, επει, επι, εωσ, η, ημεισ, ινα, καθωσ, και, κατα, μεν, μετα, μεχρι, μη, μηδε, μηδεισ, μην, μητε, ο, οπου, οσ, οστις, οταν, οτε, οτι, ου, ουδε, ουδεισ, ουν, ουτε, ουτοσ, παρα, περ, περι, πλην, που, πριν, προσ, πωσ, συ, συν, τε, τις, τοι, τοτε, υμεισ, υπερ, υπο, χωρισ, ωσ, ωστε

Latin stoplist (102 lemmas)

a, ab, ac, ad, aliquis, alius, alter, an, ante, apud, atque, aut, autem, circa, citra, contra, coram, cum, cur, de, donec, dum, ecquis, ego, enim, erga, ergo, et, etiam, ex, extra, hic, iam, idem, igitur, ille, in, infra, inter, ipse, is, iste, itaque, iuxta, licet, nam, ne, nec, nemo, neu, neuter, nisi, nos, nullus, ob, penes, per, post, prae, pro, prope, quam, quamquam, quamuis, quando, qui, quia, quicumque, quidam, quilibet, quis, quispiam, quisque, quo, quod, quomodo, quoque, sed, si, sine, solus, sub, subter, sum, super, supra, tamen, tenus, trans, tu, tum, tunc, ubi, uel, uero, ullus, ultra, unde, uos, usque, ut, uter

Works covered

Method note: ambiguous forms were assigned 673,942 times in the automatic tier ({'single': 3568985, 'prior': 111611, 'stoplist': 97921, 'tie': 23} Greek, {'single': 1030902, 'stoplist': 158242, 'prior': 292680, 'tie': 13465} Latin). Counts architecture: word-atlas-counts.json holds per-lemma per-work counts and absorbs further works as they are lemmatized.

Lexicon · Concordance · Home