Transcribing and translating First and Second World War soldiers' letters

Can WWI soldiers' letters be transcribed automatically?

Yes: photographed letters from WWI soldiers can be transcribed automatically, pencil included, as long as the photo is sharp. wondarchive is an online service that transcribes, translates and indexes photos of archives taken in the reading room (manuscripts, typescripts, printed material), for historians, doctoral students, archivists and genealogists. A letter recognised as handwritten gets an enhanced review pass, using the most accurate model of wondarchive's AI. The text is displayed beside the photo.

Quality is high on handwriting. It is excellent on the typed or printed parts of wartime correspondence: official notices, letterheads, forms. As with any transcription, a proper name or a date should be checked with a glance at the straightened photo, shown in the left-hand pane.

What documents are found in wartime correspondence, and how are they read?

Wartime correspondence mixes handwritten letters, postcards, envelopes, pre-printed forms filled in by hand, typed administrative letters and sometimes letters in a foreign language. wondarchive reads each item according to its nature: enhanced review for handwriting, layout respected for typescript. The table below shows, for each type of item, what to check.

State as of October 2026.

How do you photograph WWI or WWII letters for transcription?

For automatic transcription, photograph each side of wartime letters sharply, lying flat, under even light and without glare. Pencil graphite shines under a direct lamp; diffuse light makes it more legible. There is no need to rename files or straighten photos: wondarchive turns upright a page rotated by a quarter or a half turn.

1. Keep the order of the correspondence: the envelope, then the letter it contained.
2. Unfold each sheet flat, without forcing fragile folds.
3. Photograph both sides of postcards and forms.
4. For a letter on a double sheet, preferably photograph each page separately, in reading order. An open sheet taken in a single shot is read left then right. Yet on the outer face of a folded letter, the last page sits to the left of the first.
5. Upload the folder as the camera wrote it, subfolders included.

The six shooting rules are detailed in the guide to photographing your archives.

What difficulties specific to soldiers' letters should you check?

In soldiers' letters, the main things to check are personal spellings, military abbreviations, withheld places and incomplete dates. These difficulties stem less from the handwriting than from the context. An automatic transcription of such letters should therefore be proofread first on the names of people, units and places, comparing the text with the photo displayed beside it.

• Personal spellings: a soldier's spelling is checked against the photo. The transcription can be corrected without losing the machine-read version (see correcting the transcription).
• Withheld places: so as not to reveal the unit's position, the address gives a postal sector in WWI, or a Feldpostnummer on the German side in WWII.
• Military abbreviations: an "R.I." for infantry regiment, ranks, company numbers. They are checked at proofreading. In letters to be translated, the collection glossary fixes how they are translated from one letter to the next.
• Partial dates: a "the 12th" with no month or year. wondarchive understands even imprecise dates, but the real date should be cross-checked against the previous letter.
• Regional languages: Occitan, Breton or Alsatian are not in the list of languages cited by wondarchive. Try a few letters on the 100 free pages.

How do you read and translate a German soldier's letters?

wondarchive transcribes in German the letters of German soldiers (Feldpostbriefe), then translates them into your language, page by page. These letters are often written in Kurrent or Sütterlin, the German cursive scripts taught before 1941. The translation takes into account the end of the previous page and the start of the next, which keeps the thread of a multi-sheet letter. Kurrent and Sütterlin are read: try them on the 100 free pages.

The collection glossary fixes the translation of recurring terms, such as unit names or ranks, from one letter to the next. To recognise these scripts and understand why they are so baffling, see the guide reading Kurrent and Sütterlin.

How do you find a name, a date or a place across hundreds of letters?

Once transcribed in wondarchive, hundreds of war letters can be searched in full text. Proper names are indexed with their variants, dates are understood even when imprecise, and the search finds two terms that sit apart from one another. You can thus find a comrade mentioned, a village or a period of leave without rereading the whole bundle.

You can also ask a question in natural language, for example "where was he in March?". The AI answers by citing the folios the answer comes from. A binder can gather the letters of a single correspondent taken across several shelfmarks. The details of searching are presented in searching your archives.

Which sources can you cross-check a soldier's letters against?

The letters of a WWI soldier are first cross-checked against public military sources. The Mémoire des hommes website, run by the French Ministry of the Armed Forces, provides the records of those who died for France and the unit war diaries (journaux des marches et opérations, JMO). Military recruitment registers (registres matricules) are held in the departmental archives. On the German side, the Volksbund Deutsche Kriegsgräberfürsorge records military graves.

If you also upload images of JMO or registres matricules into the same collection as the letters, they are transcribed and searchable along with the rest. You can then compare a letter with the unit's official account.

How do you share the transcribed letters with your family, and what happens to these private letters?

From wondarchive, transcribed letters can be shared in three ways: by sharing a shelfmark, through a stable link to a page, or by exporting to PDF or DOCX. These formats suit a family booklet or sharing among cousins. TEI and CSV exports are for anyone who wants to rework the text. The formats are detailed in exporting and citing.

Images are hosted in France, with OVHcloud, and the originals are stored in Paris. They are not used for any training, and processing complies with the GDPR. A full export remains possible at any time (your data).

Follow-up questions

Are letters written in pencil read well?

Yes, if the photo is sharp and free of glare, because reading depends first on the photo. A very faded pencil is best photographed under diffuse light rather than under a direct lamp.

What about a letter in Occitan, Alsatian or Breton?

These languages are not in the list of languages cited by wondarchive. Try a few pages on the 100 free pages before uploading the whole bundle.

Can you correct the transcription without losing the original?

Yes: the text can be edited, the machine-read version is kept and each correction is dated. The original photo stays intact, never recompressed.

Do you need to sort the letters before uploading them?

No. The pages of an upload can stay "to be sorted", then be moved in batches to their shelfmark, for example one shelfmark per correspondent or per year.

How many letters can you try for free?

The free account includes 100 transcribed and translated pages, with no credit card. A hundred or so multi-sheet letters exceeds that volume. Start with the hardest to read, in pencil or Kurrent, to judge the result.

To try it on your own letters, open a free account, or first browse the demo.