← All drafts

Reference library — index#

Working reference shelf for the Early New High German in `the-offer-de.fountain`. The books are too large and too partly-licensed to commit, so only this index is tracked. Rebuild your own copy of the shelf with:

``` gcloud auth login # once, with read access to the bucket python3 resources/sync-books.py # ~1.5 GB into resources/ ```

That pulls the whole shelf from the project's storage bucket. Everyone syncs their own copy into their own checkout; nothing here depends on any one machine's filesystem.

Two shelves, split by the question they answer:

• Diction and idiom — what words, in what constructions? Indexed below.
• Pronunciation — how did it sound? See [`pronunciation-library.md`](pronunciation-library.md), which documents that shelf in full. Summarised at §5 here so this file stays a complete inventory.

Start with [`enhg-translation-guide.md`](enhg-translation-guide.md) if the task is turning an English line into period German. It distils this shelf into a working checklist — which layers of the language the draft renders period and which it deliberately normalizes, the morphological transforms, and the procedure for checking a word.

Paths are relative to this directory: `books-pdf/` for page images, `books-txt/` for extracted text (same slug), `bible/` for per-verse scripture in `Book C:V<TAB>text` form. All three are gitignored and populated by the sync.

Everything with a `.txt` is in `corpusidx`, so `corpusidx search "<query>"` reaches the whole shelf at once and is usually the right first move.

Text extraction is not uniform. Volumes set in Fraktur produce unusable OCR (`frud^t` for Frucht), so many titles below are page images only — read them from the PDF; searching them will silently fail. See Pending work.

1. Luther's own German — the primary model#

Everything else here is a tool for interrogating this material. When a line's diction is in doubt, the question is "how does Luther say it?"

The 1545 Bible, as searchable text#

The largest body of Luther's German in a form you can query line by line. Same tab-separated format as the rest of `bible/`:

``` grep -P '\tWenn wir aber' resources/bible/luther-1545.txt ```

Caveat: this is translation register — elevated, shaped by the Hebrew and Greek. The right authority for vocabulary, spelling and morphology; the wrong one for how a printer talks to an apprentice.

The critical edition (page images)#

Luther's prose voice, which the Bible does not give you.

WA 50 is the closest volume to the film's date. Anything Luther wrote in German in 1539 is in it or next to it.

Sendbrief vom Dolmetschen (WA 30 II) repays reading whole: it is Luther arguing explicitly about how German should sound, and it is the best single statement of the register the German draft is reaching for.

`corpusidx` also carries a separate `luther-wa` corpus of 112 files — Luther's printed German in bulk, searchable, and the fastest way into this material.

The Tischreden — complete run#

Page images only. Bd. 4 is what the German draft's header cites; the rest are here so surrounding entries can be read for idiom and parallel recollections located.

Nr. 4690 no longer needs the page images. It is transcribed, apparatus and all, in [`table-talk-4690.md`](table-talk-4690.md), with a translation made for this project, reading notes on the words the film rests on (exemplaria, sein gnade, rum), and Nr. 5151 as an appendix. That file — not the OCR, which is unusable here — is what drafts should quote from. It was read off the page images at 400 dpi; the recipe for re-rendering them is at the end of it.

This is also the nearest thing to a record of Luther talking, which is what the film needs. Filtered through the note-takers, but the sentence rhythms are closer to speech than anything else in the corpus.

2. Luther's language specifically#

Dietz is the workhorse. A dictionary of Luther's own vocabulary with dated citations, and its text extracted cleanly — so it answers "did Luther use this word, and in what sense?" by search rather than page-turning.

Franke is the systematic grammar of Luther's written language and dates the forms: when `yder` gave way to `yeder`, how long `fride` held out against `friede`. It is the guard against false archaisms, which are the harder error — a modernism reads as a slip, an invented archaism reads as costume.

Franke's `.txt` files exist and are indexed, but they are Fraktur OCR and

largely mush. Treat search hits in Franke as a page-locator only, then read

the PDF. Do not quote from the extracted text.

Stolt is about how Luther's German persuades — its affective and rhetorical shape. Relevant to the refusal speeches, less so to the print shop.

3. Period register — trade, town, and speech#

Luther will not supply the print shop's working vocabulary, and the Bible will not supply artisan speech. These fill those gaps.

Page images only except the Urkundenbuch.

The Ständebuch verses are the most directly useful item here. Sachs wrote a stanza per trade, printers among them, in period German — the diction the workshop scenes need, which no dictionary hands you pre-assembled.

Buchhandel vol. 1 is the one to read; vols. 2–4 run centuries past the film and are here for completeness. Vol. 1 carries the trade's terms of art for contracts, privileges, reprinting and the fairs.

Two caveats, both real:

• Sachs is Nuremberg, not Wittenberg, and writes verse, not speech. Use him as a register model — how an artisan is portrayed as talking — not as a dialect authority for Saxony.
• The Buchhandel volumes are 19th-century scholarship in 19th-century German. Their quoted documents are period; their prose is not. Take terms from the quotations, never from the surrounding narrative.

The Urkundenbuch is Wittenberg's own chancery language — right town, right decades, right formal register for anything contractual. Early documents are Latin; the German ones come later in the volume.

4. Lookup tools, no local copy#

No download exists; these are query interfaces.

Reach for the FWB by reflex. Dietz tells you Luther used a word; the FWB tells you what it meant to everyone else at the time.

5. The pronunciation shelf#

Documented in full in [`pronunciation-library.md`](pronunciation-library.md) — listed here only so this file is a complete inventory. Modern scholarship:

• `ebert-reichmann-solms-wegera-fruehneuhochdeutsche-grammatik-1993` — the standard reference grammar, organized by dialect region; text clean
• `hartweg-wegera-fruehneuhochdeutsch-2005` — the teaching introduction; covers the Lutheridiom and the printers' languages; text clean
• `penzl-fruehneuhochdeutsch-1984` — the method book: how pronunciation is inferred from spelling; text clean
• `mcgee-singing-early-music-1996` — IPA transcriptions written for performers; text clean
• `reichmann-wegera-fruehneuhochdeutsches-lesebuch-1988` — dated, localized period texts; text clean
• `moser-stopp-grammatik-des-fruehneuhochdeutschen-vokalismus-1978` — unstressed syllables only; text clean
• `philipp-einfuehrung-ins-fruehneuhochdeutsche-1980` — general introduction; page images only

Contemporary witnesses (page images — the OCR of these is not trustworthy):

• `vier-seltene-schriften-des-16-jahrhunderts-1882` — Ickelsamer, Jordan, Grüßbeutel; reading primers, so they describe what letters sound like
• `ickelsamer-teutsche-grammatica-1534` — second witness to the above
• `frangk-orthographia-deutsch-1531` — a Wittenberg orthography holding up Luther's printed German as the model; closest source in place and time
• `clajus-grammatica-germanicae-linguae-1578` — built on the 1545 Bible, with a section on correct pronunciation; roman type, so this one is searchable

Note the division of labour: the grammar and reference books settle what the sound was; Frangk and Clajus are the ones written by people who could still hear Luther's German spoken.

Open checks these resources enable#

Two questions surfaced while the shelf was assembled. The second is now settled; the first has been sharpened by the Nr. 4690 transcription.

1. 1 Tim. 6:8. The German draft's closing line reads `Hab ich Nahrung und Kleider, so laß ich mich benügen`. The 1545 wording is `Wenn wir aber Narung vnd Kleider haben / so lasset vns benügen`. The draft matches in vocabulary but not construction, while its header states the clause "follows Luthers 1545 wording." Either the line or the header note should change — it can be a deliberate choice, but it should be one. Sharpened: in Nr. 4690 this clause is Latin — `Habens victum et amictum contentus ero`, the Vulgate. So the German line is not a quotation of the source at all; it is a rendering, and whichever wording it takes is an editorial choice either way. See `table-talk-4690.md`.

2. ~~The Nr. 5151 citation.~~ Settled 7 Aug 2026. Nr. 5151 is in Bd. 4, p. 535 — verified on the page image. Bd. 4's numbering runs past 5188 even though Bd. 5's title page advertises 1540–1544; the volumes are divided by collection, not by year. The citation stands. But its description needs fixing: Nr. 5151's 400 is the Elector's provision, not the printers' offer, so "parallel recollection of the salary" invites exactly the conflation the film cannot afford. See the appendix in `table-talk-4690.md`.

Pending work#

Fraktur re-OCR. Franke, Stolt, Sachs, the Ständebuch, the Buchhandel set, the contemporary witnesses and every WA volume are effectively page-image-only, because the available OCR of Fraktur is unusable. Re-OCR would make roughly two-thirds of this shelf searchable — a change in kind, not degree, since `corpusidx` would then reach it.

`tesseract` and `ocrmypdf` are installed; the German and Fraktur models are not. Installing them needs root:

``` ! pkexec sh -c 'apt-get install -y tesseract-ocr-deu tesseract-ocr-script-frak' ```

Afterwards, re-process into `books-txt/` and run `corpusidx update`.