MLV 2026 New Testament + MLVDC — USFM 3.0 / USX 3.1 export
Built 2026-11-11 from: MLV_RL.bbli (MLV2026-RL), 01_NKGT6.bbli (K6 interlinear), MLVDC.lexi (MLVDC v4)
All files UTF-8. Every USFM file parses with zero errors (usfm-grammar 3); every USX file was generated from it.

FOLDERS
  usfm_english/     00FRTMLV.usfm (module "Read Me First" info) + 27 NT books, Strong's-tagged MLV English
  usx_english/      same, USX (XML)
  usfm_greek/       27 books, RP Byzantine Greek from K6: every word has Strong's, morphology, K6 gloss
  usx_greek/        same, USX
  usfm_dictionary/  90GLO  MLV English Concordance  (6,570 entries: the English headwords and the
                           English dictionary articles such as Baptism, Church, "She ~")
                    91XXA  MLV Greek Lexicon        (5,825 G-number entries)
                    92XXB  Extras                   (6 "_" front sections + 321 synonym essays)
  usx_dictionary/   same three, USX
  json/             the same three dictionary sets as structured JSON (parsed fields, Greek words, refs,
                    plus the original HTML for each entry). Easiest form for an app developer to import.
  _alignment.tsv    every MLV English word: which Greek word/Strong's/morph it was matched to
  _unmatched.tsv    every K6 Greek word that got no English word (QC list)
  _qc_tag_vs_MLVDC_concordance.tsv   the 596 content words whose tag differs from the MLVDC concordance

ENGLISH TEXT MAPPING (e-Sword -> USFM)
  <i> supplied words     -> \add ...\add*   (never Strong's-tagged)
  <red> words of Jesus   -> \wj ...\wj*
  * and ° marks          -> kept as literal characters touching the word, inside the \w tag
  <br /><br />           -> new paragraph \p
  Verse 0 book intros    -> \is Introduction / \ip paragraphs
  blue {Mat 1:1-17 & ...} section/parallel lines standing between paragraphs -> \r line
  all other blue {...} notes ({Psa 2:7}, {i.e., ...}, NOTES:, {T}, {F}) -> \f footnotes
  {T} footnotes read "{T} Wording not found in the majority of manuscripts (see the Definitions section)."
  Strong's tag example:  \w loved*|strong="G25" x-morph="rmac:V-AAI-3S"\w*

HOW THE ENGLISH STRONG'S TAGS WERE MADE
  K6 has Greek-order words with glosses. The MLV English is not word-aligned anywhere, so each
  K6 gloss alternative (e.g. "in-order-that|that*") was matched to the MLV words of the same verse.
  Articles (G3588) are assigned last, to the "the" directly before their own noun's English word.
  If no gloss matched, the English word was left UNTAGGED, never guessed.
    Greek words matched to English:  120,221 of 140,266 (85.7%); excluding the article: 91.6%
    Untagged non-italic English words are mostly helper words with no single Greek source.
  Independent checks:
    Content words vs. the MLVDC concordance's own G-numbers: 99.3% agree (86,750 / 596)
    Article vs. following noun case-number-gender agreement: 96.2%
    Text fidelity: all 7,960 verses match the source word-for-word after tag removal.
  Multi-word renderings get the same number on each word ("of Jesus" -> both G2424), as in KJV+ style.

CHARACTER SUBSTITUTIONS (USFM/USX files only; JSON keeps originals)
  "|" in plain text  -> "｜" (U+FF5C)   (bare | is reserved in USFM)
  "<=" in K6 x-gloss -> "⇐", other "=" -> "＝"   (= is not allowed inside USFM attribute values)

DATA ISSUES FOUND IN THE SOURCES (not changed)
  K6 Rom 9:29 ends with a stray fragment "<num>G302</num><sup>] we would have been similar ¶ </sup>"
      (skipped; K6 has 140,266 real words, the raw <num> count is 140,267)
  MLVDC G1490d (ἴδεν) Supplement contains a working note: "...Manuscript source not independently
      confirmed; verify against your own K6/RP2018 apparatus for the exact witness."
  1Jn 5:21 end note has an unbalanced "}" (kept as text in the footnote)

NOT INCLUDED
  Old Testament rows of the .bbli shell (NT only).  GK45 Greek-word topics / RMAC entries (not in this MLVDC).
