6. Strong verbs, old pasts and participles, old plurals

English has been regularising its verbs for a thousand years. EModE stands in the middle of the process: many verbs still had the old vowel-change («strong») forms, some had both, and some strong forms of 1611 have since gone weak (holpen → helped) or changed vowel (spake → spoke). The biblical register freezes a selection of them. Hejná & Walkden (2022, §4.3.2 and §5.3.2) give the background: frequent verbs resist regularisation; holpen for helped is their example of a form a reader meets in EModE.

6.1 The forms that appear in Mova's texts

Counts: tales pool (132 697 sentences) / KJV Torah / JPS Torah, whole-word, case-insensitive, 2026-10-17. The last column is what Morph::lemmatize returns with the tag shown (a Rust probe linked to the en crate of this branch, 2026-10-17); bold marks a wrong lemma.

Sources: forms from the texts; historical status from Abbott 1884 §§339–345 (past indicative in u: «sung» for sang, «begun» for began; participles in -en; participle formation), and from the OED-style glossaries in «wanted» (Wright's Bible Word-Book).

The JPS Torah needs only a few of these: smote, slew, wrought, builded, digged, sod, sodden, waxen, graven, molten, hewn, hid, gotten, abode. The tales add spake, bade, quoth, sate, clomb, durst, trod, holp, stricken. The KJV adds begat, bare, sware, gat, clave, drave, shew.

6.2 How the db annotates them (2026-10-17)

Counted together, the clearly wrong strong-verb lemmas (wrought, smote, abode as a VERB, digg, sod, waxen VERB, hewn, gotten, sodden VERB, graven VERB, molten) are 67 tokens in 62 sentences. The teacher (Sonnet or Opus) wrote most of these lemmas; they pass every gate, because the gates check the lemma of an AUX only.

6.3 Participial adjectives: verb or adjective?

molten image, graven image, hewn stones, sodden flesh, waxen poor, stricken in years, laden with, well stricken in age. UD English treats a participle used attributively as ADJ when it is lexicalised (molten, graven) and as VERB with VerbForm=Part when it keeps verbal force. Whichever the choice, the lemma of the VERB reading must be the verb (melt, grave, hew, seethe, wax, strike, lade) and the ADJ reading keeps the form (molten, graven). The db mixes both (graven ADJ 7, VERB 1 with lemma graven, VERB 2 with lemma grave). A rule by context is cheap: before a noun (amod) → ADJ with lemma = form; after be/been, or with an agent or object → VERB with the verb lemma.

6.4 Old plurals

brethren deserves a decision. It is the old plural of brother, specialised for «kinsmen, members of one people or order». UD lemmatizes inflection, not meaning, so the lemma is brother (EWT has no brethren to compare with). The db's 120 brethren lemmas would become brother; the special meaning belongs in the world layer, not in the lemma.

6.5 Weak verbs with old forms

6.6 Lemma decisions for Mova

1. The lemma is the modern verb: spake → speak, wrought → work, sod/sodden → seethe, digged → dig, builded → build, begat → beget, bare (with a verb tag) → bear, shew → show, durst → dare, clomb → climb, holpen → help, wist/wot → wit (EModE «know»; keep wit, not know: lemmatization is not translation), quoth → quoth (it has no other forms; a few treebanks use quethe).
2. The table must run only with a verb tag (bare, sod, sate, abode, laden, bade, gat are also or mainly adjectives or nouns in modern text), except for forms that are never anything but verbs (spake, begat, sware, clave, holpen, wrought — wrought iron is ADJ, so wrought needs the verb tag).
3. A list of ~50 forms is enough for the biblical register and the tales; it belongs next to IRREGULAR in en/src/morph.rs, or better as a data file (en/data/archaic.tsv) that build.rs compiles in, so that the word list can grow without code changes. Chapter 13 (L4) gives the list.

6.7 A negative control

Any rule for old forms must leave the modern homographs alone. From the db and the tales:

• art the noun: «the art of the perfumer» (db art|NOUN 2, correct), «the art of healing» (round 121 draft, book:pg3027:3607), «taught in art and magic» (book:pg2892:2487) — 437 tales lines with art, most of them nouns.
• bare the adjective: «bare feet» (most of the 137 tales hits).
• sod the noun: «under the sod» (tales).
• abode the noun: «his abode».
• laden the adjective: «laden with gold».
• bad the adjective: never bid.

The unit test archaic_verbs in en/src/morph.rs already has this shape (art as a noun, greatest, rest, digest, interest must be left alone); every new entry should come with its noun or adjective counterpart in the test.

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/languages/en-x-emode/06-strong-verbs/