Appearance
Turn transcriptions into an edition when there are real editorial questions to resolve — multiple witnesses to collate, abbreviations and uncertainties to adjudicate, persons and places to identify, and a scholarly argument a reader will cite. Stay with a transcription archive when one good witness already answers your users' needs, when you lack the staffing for long-term editorial maintenance, or when the value is access rather than interpretation. The honest test: if you cannot name the editorial decisions your edition would make, you do not yet need an edition.
What actually separates a transcription from an edition?
A transcription answers "what does this witness say?" An edition answers "what is the text, and how do I know?" That second question pulls in collation, an apparatus of variants, resolved abbreviations, emendations with responsibility, annotation, and a documented policy. The transcription is evidence; the edition is the argument.
text
Transcription: reproduce one witness, faithfully
Edition: establish a text + record the reasoning + annotate + citeIf your project never makes an editorial judgement, calling the result an edition oversells it.
When should you turn transcriptions into an edition?
Look for these signals. The more that apply, the stronger the case:
| Signal | Edition warranted? |
|---|---|
| Multiple witnesses disagree | Yes — collation is the point |
| Heavy abbreviation / damage | Yes — readers need resolution |
| Named entities to identify | Yes — annotation adds research value |
| One clean witness, no variants | Often no — a transcription suffices |
| No maintenance commitment | No — do not start what you cannot sustain |
| Audience wants access, not analysis | No — searchable transcriptions win |
How much more work is it, really?
Budget two to five times the plain-transcription effort per page. The multiplier comes from collation of every witness, building the apparatus, writing and applying an editorial policy, and annotating. A single-witness diplomatic transcription with light notes sits at the low end; a multi-witness critical edition with dense commentary sits well above it. Underestimating this is the most common way edition projects stall mid-grant.
What is the minimum that earns the word "edition"?
Add, at least: a written editorial policy; a stable citation scheme; resolved abbreviations and marked uncertainties; and basic authority links for persons, places and dates. In TEI that looks like:
xml
<p>Brother <persName ref="#tho_more">Tho<expan>mas</expan> More</persName>
wrote from <placeName ref="#tower">the Tower</placeName>
on <date when="1535-06">June</date>.</p>The expan, the ref authority links and a documented policy are what lift this above a bare transcription.
Can crowdsourced or HTR output become an edition?
Yes — that is increasingly the normal pipeline. A Transkribus or FromThePage layer gives you a strong draft. The editorial pass then normalises the encoding, reconciles volunteer disagreements, adds apparatus, and links authorities. Treat the machine or crowd output as a high-quality starting point, never as the finished scholarship.
When is staying with transcriptions the right call?
Choose a transcription archive when access is the deliverable: a parish-register corpus where users want to find their ancestors, or a newspaper run where full-text search is the value. Add solid metadata and a citation scheme and you have served your audience without pretending to an editorial apparatus you never built.
Key Takeaways
- An edition adds editorial judgement; a transcription only reproduces a witness.
- Decide based on real editorial questions — variants, abbreviations, identifications.
- Budget two to five times the transcription effort once apparatus and annotation enter.
- The minimum bar: documented policy, citation scheme, resolved uncertainties, authority links.
- Crowdsourced and HTR output make excellent drafts but need an editorial pass.
- If access is the goal, a well-described transcription archive is the honest choice.
- Never start an edition you cannot maintain for the long term.
Frequently Asked Questions
What is the difference between a transcription and an edition?
A transcription reproduces what a witness says; an edition adds editorial judgement — establishing a text, recording variants, resolving abbreviations, annotating and contextualising — so a reader can trust and cite it. The transcription is raw material; the edition is the scholarly argument built on it.
When should I NOT turn transcriptions into an edition?
When a searchable transcription archive already meets your users' needs, when you cannot commit to long-term maintenance, or when you have only one witness and no editorial questions to resolve. In those cases publish clean transcriptions with good metadata instead.
Do I need TEI to make an edition from transcriptions?
Not strictly, but TEI is the practical default because it lets you encode editorial interventions — emendations, variants, abbreviations, persons and places — in a documented, reusable way. Plain text or basic HTML cannot carry that scholarship.
How much extra work is an edition over a transcription?
Plan for two to five times the per-page effort of plain transcription once you add collation, apparatus, annotation and a documented editorial policy. The multiplier rises with the number of witnesses and the density of annotation.
Can crowdsourced transcriptions become an edition?
Yes, and many do, but only after an editorial pass: normalising the encoding, resolving disagreements, adding apparatus and authority links. Treat the crowdsourced layer as a high-quality draft, not as the finished edition.
What is the minimum I should add to call it an edition?
A documented editorial policy, a stable citation scheme, resolved abbreviations and uncertainties, and at least basic annotation linking persons, places and dates. Without editorial intervention you have a transcription with a nicer interface.