Reference shelf assembled to answer one question: **how did German sound in and around Wittenberg in 1539?** Luther's variety is East Central German (Ostmitteldeutsch) of the Saxon chancery, inside the Early New High German (Frühneuhochdeutsch) period, conventionally 1350–1650.
Everything listed here is on disk. PDFs live in `resources/books-pdf/`, extracted plain text in `resources/books-txt/`, same slug in both. All of it is indexed in `corpusidx`, so `corpusidx search "<query>"` reaches it.
This file is a working note, not a draft — it sits in `resources/` and `build.py` only publishes root-level Markdown, so it will not appear on the site.
Ed. Oskar Reichmann & Klaus-Peter Wegera. Sammlung kurzer Grammatiken germanischer Dialekte A/12. Tübingen: Niemeyer, 1993. 572 pp. `ebert-reichmann-solms-wegera-fruehneuhochdeutsche-grammatik-1993`
The current standard reference grammar. Part 1, Schreibung und Lautung (Reichmann & Wegera), is the phonology. Its decisive advantage for us: it is organized **by dialect region**, so East Central German can be isolated rather than averaged in with Upper German. Start here for any specific sound.
Text layer is clean and searchable.
2nd, revised ed. Germanistische Arbeitshefte 33. Tübingen: Niemeyer, 2005. 208 pp. `hartweg-wegera-fruehneuhochdeutsch-2005`
The teaching introduction — read this before the reference grammar. Carries dedicated treatment of the Lutheridiom and of the printers' languages (Druckersprachen), which makes it doubly relevant: the film is about Wittenberg printers, and the Druckersprache question is literally what Lufft's delegation trades in. 220 occurrences of "Luther."
Text layer is clean.
Germanistische Lehrbuchsammlung 9. Bern/New York: Peter Lang, 1984. 212 pp. `penzl-fruehneuhochdeutsch-1984`
The method book, and the one to read if we want to know how much confidence any claim here can carry. Penzl works the question of how you infer actual pronunciation from spelling variation plus the testimony of contemporary grammarians. He names as his evidence base exactly the primary sources listed below — "Ickelsamer, Johann Kolroß, Fabian Frangk, Peter Jordan (1533), J.H. Meichßner, Ortholph…" — so this volume and the facsimile section of this shelf are designed to be read against each other.
Text layer is clean in the body; the scanned title page carries a previous owner's handwriting, which garbles the first screen of extracted text.
Music: Scholarship and Performance. Bloomington: Indiana University Press, 1996. 324 pp. `mcgee-singing-early-music-1996`
The practical entry point. Chapter 16, "Late Medieval German and Early New High German" (Peter Frenzel), gives IPA transcriptions of sample texts. Written for performers who have to say the lines rather than for philologists, which is our actual problem. The print edition ships with a CD of the sounds spoken; we have the book, not the audio.
Text layer is clean.
Tübingen: Niemeyer, 1988. 264 pp. `reichmann-wegera-fruehneuhochdeutsches-lesebuch-1988`
Companion reader of dated, localized ENHG texts. This is the check: it lets a claim from the grammar be tested against actual documents from the right region and decade instead of taken on authority.
Text layer is clean.
Bearb. Hugo Stopp. Heidelberg: Carl Winter, 1978. 310 pp. `moser-stopp-grammatik-des-fruehneuhochdeutschen-vokalismus-1978`
Note the scope. This is not Virgil Moser's 1929 Frühneuhochdeutsche Grammatik
and its Lautlehre volumes, which is what one actually wants for stressed vowels. This
is one volume of the successor Moser/Stopp project, and it covers vowels in unstressed
syllables only. Genuinely useful for the unstressed-`-e` question that governs ENHG
rhythm — which matters for line readings — but it does not cover the stressed-vowel
system. That gap is open. Moser 1929 was not obtainable; it would have to come from a
library.
Text layer is clean.
Read these as page images, not as text. All but Clajus are Fraktur facsimiles and the
OCR is rough to useless — the Ickelsamer reprint renders its own title as "Ceütſche
Grammatica." Grep will not work on them. The extracted `.txt` files exist but should not
be trusted for these four.
Ed. Heinrich Fechner, with an essay on Ickelsamer by F. L. K. Weigand. Berlin: Wiegandt und Grieben, 1882. `vier-seltene-schriften-des-16-jahrhunderts-1882`
The most valuable item in this section. Contains four works, all reading primers:
1. Valentin Ickelsamer, Ein Teütsche Grammatica — the first edition 2. Ickelsamer, Die rechte weis auffs kürtzist lesen zu lernen (1534 edition) 3. Peter Jordan, Leyenschül (1533) 4. Jacob Grüßbeutel, Eyn Besonder fast nützlich stymen büchlein (1534)
Why primers rather than grammars: a book teaching a child to read has to describe what each letter sounds like. That descriptive layer is the closest thing to a contemporary phonetic record we have, and it is precisely the evidence Penzl reasons from.
Ed. Ludwig Kohler, 1881. `ickelsamer-teutsche-grammatica-1534`
The first book to call itself a German grammar, by a schoolmaster, grown out of his 1527 reading primer — so it is largely about how letters sound. Superseded for our purposes by the Fechner volume above, which carries the first edition; kept as a second witness.
36 pp. Imprint reads Wittemberg. `frangk-orthographia-deutsch-1531`
A Wittenberg orthography that explicitly holds up Luther's printed German as the model — the closest source geographically and chronologically to our scene. Short.
Edition/year unverified: the title-page date could not be read off the facsimile, and the
OCR yields no year. Conventionally dated 1531, which is what the slug asserts. Confirm
against the page image before citing a date in any draft.
Ed. Friedrich Weidling, in Ältere deutsche Grammatiken in Neudrucken, 1894. Original 1578. `clajus-grammatica-germanicae-linguae-1578`
Built directly on Luther's 1545 Bible, and its four-part scheme includes a Prosodia section on correct pronunciation — the closest thing to a rulebook for "Luther's German" written by someone who could still hear it spoken. Two generations later than our scene, so its testimony is about where the language landed, not where it stood in 1539.
Roman type, so unlike the others this one OCR'd well and is genuinely searchable.
`corpusidx` also carries a `luther-wa` corpus — 112 files of the Weimarer Ausgabe. Luther's own printed German in bulk, searchable alongside everything above.
The honest constraint: **there is no recording, and no consensus reconstruction with the authority that Original Pronunciation Shakespeare carries.** What these books support is a defensible flavor, not a phoneme-by-phoneme reconstruction.
Suggested order of work:
1. Frenzel's chapter in McGee — get a speakable baseline with IPA. 2. Hartweg & Wegera on the Lutheridiom — narrow it to the right region and register. 3. Ebert et al., Schreibung und Lautung — settle specific sounds, East Central German column. 4. Penzl — calibrate how much confidence each decision deserves. 5. Spot-check against the Lesebuch and the primers before committing.
Then pick a small number of features, apply them consistently across `the-offer-de.fountain`, and document the choice. Per `AGENTS.md`, any reconstruction decision that goes beyond what the sources support is a dramatic choice and gets marked as one.