Datasets:
smyth-grammar
Herbert Weir Smyth, A Greek Grammar for Colleges (New York: American Book
Company, 1920 — public domain). PROSE-witness t0 source repo for Greek
morphology and syntax, the sibling of allen-greenough-grammar (Latin) and
gesenius-kautzsch-grammar (Hebrew) in the distributional-grammar programme:
one row per numbered Smyth paragraph (§1–§3048, complete, plus the 213
"D"-suffixed dialect paragraphs), Greek examples in polytonic Unicode, Smyth's
own cross-references as bare section numbers.
evidence_status=grammar-author-claim — every row is Smyth's own 1920
grammatical analysis, not an independently verified linguistic fact.
License and Attribution
| Content | License | Source |
|---|---|---|
| Underlying grammar text (1920) | Public Domain (US, pre-1929 publication) | Smyth, A Greek Grammar for Colleges, American Book Co., 1920 |
| Digitization lineage | Perseus Project, Tufts University (TEI XML, text 1999.04.0007, NEH-funded; prepared under Lisa Cerrato, William Merrill, Elli Mylonas, David Smith; principal Gregory Crane) → Alpheios Project alpheios-project/grammar-smyth src/smyth_eng_u.xml @ d7796df9 (2020-05-13; Unicode Greek, "with corrections, formatting and indexes added by Alpheios"; README asserts CC BY-NC-SA 3.0 US over the encoding) |
https://github.com/alpheios-project/grammar-smyth |
Topic index (topic_index) |
Alpheios' own 86-row navigation index (index/alph-index-smyth), resolved here to Smyth's headings/section numbers |
same |
Ruling on record (owner, 2026-08-23): this repo ships Smyth's PD prose,
section numbering and Greek examples — a faithful reproduction of a
public-domain work — and is therefore carded public-domain with attribution
(doctrine §2.5: "pre-1929 imprint or faithful PD reproduction → k:
public-domain, attribution only"). The digitizers' own contributions (TEI
markup, Beta Code transliteration, <ref> target ids, Alpheios indexes and
formatting) are not redistributed: the builder flattens the markup to plain
text, converts Beta Code back to Unicode Greek, and carries only Smyth's own
section numbers and the encoders' LSJ entry keys as join keys. Owner
follow-up ruling 2026-08-23: the Alpheios index and corrections may be
leveraged — a navigation index over a PD text is factual. We gratefully
acknowledge the Perseus Project / Tufts University and the Alpheios Project
for the digitization and corrections. (The neelsmith/smyth Beta Code
derivative was evaluated first and set aside: it lacks the verb appendix and
carries Greek in untagged spans — see VERIFY.md.) This supersedes the "smyth-grammar BLOCKED" line of
grammar/plans/DISTRIBUTIONAL_GRAMMAR_PLAN.v2.md §2/§8.
Configs
sections (3,261 rows)
| column | meaning |
|---|---|
section_id |
Smyth's paragraph number as printed ("792", "3D", "30 D. 1") — the join key |
section_num |
integer part of section_id |
is_dialect_note |
true for the 213 "D" (dialect) paragraphs |
section_label |
display form § N |
outline_part / outline_chapter / outline_section / outline_subsection / outline_subsub |
Smyth's outline hierarchy (titles; prefixed so they are never read as Bible addresses) |
head |
paragraph heading where the source has one |
prose_text |
the paragraph's full text, Greek in Unicode; nested dialect paragraphs excluded (they are their own rows) |
greek_examples |
JSON list of the Greek examples in the paragraph, in order |
glosses |
JSON list of the English glosses that follow Greek examples |
cross_refs |
JSON list of bare section numbers Smyth cross-references |
citations |
JSON list of classical citations (author/work/locus strings) |
parent_section_id |
for nested (dialect) paragraphs, the enclosing paragraph |
source_url, source_sha256, doc_order |
provenance (pinned source commit + file sha) and document order |
toc (665 rows)
Smyth's outline: toc_order, level (part/chapter/section/subsection/subsub),
n, title, ancestor outline_* titles, source_url.
verb_list (761 rows)
Smyth's Appendix "List of Verbs" (Part V): lemma, poetic_only (Smyth's *
= Epic/poetic), letter_page, entry_text (full entry), principal_parts
(the forms Smyth sets in bold), greek_forms, glosses, cross_refs,
lsj_entry_key (Perseus LSJ entry key in Beta Code, 761/761 — join surface to
the estate's LSJ), provenance.
topic_index (86 rows)
Alpheios' topic keys (alph-case-dative, alph-declension-1st, …) resolved
to resolved_heading + first_section_id (81/86 resolve; 5 point at pages
without a Smyth heading and carry the raw anchor only).
Build
huggingface/candidates/build/smyth-grammar/build.py — deterministic from
the cached TEI (cache/smyth_eng_u.xml, sha256 3d5ba15f…) and index CSV;
--verify rebuilds and diffs every column. Text is NFC-normalised; fullwidth
parentheses (U+FF08/FF09, a Perseus artifact) are normalised to ASCII.
- Downloads last month
- 37