To encode corrections and authorial edits in TEI, you mark what the author added with add, what they struck out with del, and pair the two with subst when one replaces the other. Your own editorial fixes are kept separate, using sic/corr inside a choice. The golden rule for beginners: never silently change the text — record both what the source shows and what you think it should be, and let the reader choose.

Why keep author edits and editor edits apart?

This is the single idea that prevents most confusion. There are two different kinds of change in a manuscript:

• Authorial — the writer revised their own words on the page. Encoded with add, del, subst.
• Editorial — you, the editor, are correcting or normalising. Encoded with sic/corr and orig/reg.

Mixing them is the classic beginner mistake. If you put your spelling fix in a del, you make it look as if the author crossed it out. Keep the families separate and your encoding stays honest about who changed what.

How do you encode a simple deletion and addition?

Here is the smallest worked example: the author wrote "quick", crossed it out, and wrote "swift" above the line.

<p>The <subst>
  <del>quick</del>
  <add place="above">swift</add>
</subst> messenger arrived.</p>

The subst ties them together as one revision. Read diplomatically it shows the deletion and the insertion; read as a final text it gives "The swift messenger arrived." That dual reading is exactly what genetic editing wants.

What if the author just added or just deleted?

Not every change is a substitution. Use the elements on their own:

<!-- a pure addition in the margin -->
<p>She left <add place="margin">in haste</add> before dawn.</p>

<!-- a pure deletion you can still read -->
<p>He paid <del>five</del> ten shillings.</p>

The @place on add (above, below, margin, inline) records where the insertion physically sits — useful evidence for how the page evolved.

How do you handle your own corrections?

When the source has an obvious slip — say "teh" for "the" — record both readings with choice:

<p>I sent <choice>
  <sic>teh</sic>
  <corr>the</corr>
</choice> letter yesterday.</p>

A stylesheet can then show the diplomatic text (sic) for textual scholars or the clean reading (corr) for general readers. The same pattern with orig/reg handles normalising old spelling without destroying it.

Which element do I reach for? A quick map

Situation	Element(s)	Whose change
Author crossed words out	del	author
Author inserted words	add (with @place)	author
Author replaced X with Y	subst wrapping del + add	author
Obvious error you fix	choice with sic + corr	editor
Old spelling you normalise	choice with orig + reg	editor
Author's deletion is illegible	del containing gap	author

Print this table and you can encode most first-project manuscripts confidently.

What about text you simply cannot read?

If the author deleted something and scored it out so heavily you cannot recover the words, do not guess. Mark the loss:

<del><gap reason="illegible" extent="2" unit="words"/></del>

Use gap only for what is genuinely unrecoverable; if you can read even a doubtful word, transcribe it inside unclear instead so the evidence survives.

What should a beginner not worry about yet?

Genetic editing can get deep — revision campaigns, hands, sequencing of changes with @seq. For your first project, ignore all of that. Concentrate on three habits: keep author and editor changes in separate elements, never silently alter the text, and record where additions sit. Those alone produce encoding that is correct and that you can enrich later without rework.

Key Takeaways

• Author changes use add/del/subst; editor changes use sic/corr and orig/reg — never mix them.
• subst pairs a del with the add that replaces it, capturing one revision.
• Record @place on add to show where an insertion sits on the page.
• Use choice so readers can see both the source reading and your correction.
• Never silently change the text — that loses the witness the manuscript provides.
• Use gap only for genuinely illegible text; prefer unclear when you can read a doubtful word.

Frequently Asked Questions

What is the difference between del and subst in TEI?

del marks something the author struck out. subst is a wrapper that pairs a del with the add that replaces it, recording a substitution as one connected change rather than two unrelated edits.

How do I show a word the author added above the line?

Use add with place="above" (or supralinear). The @place attribute records where on the page the addition sits, which matters for genetic readings of how the text grew.

Should I correct an obvious slip silently or mark it?

Record it with sic (what the source says) and corr (your correction), usually grouped in choice. Silent correction loses the witness; explicit choice lets a reader or stylesheet pick either the diplomatic or the edited reading.

What is the difference between an authorial and an editorial change?

Authorial changes (add, del, subst) are the author's own revisions on the page. Editorial changes (sic/corr, orig/reg) are your interventions as editor. Keep them in separate element families so the two are never confused.

How do I encode text the author deleted but I can still read?

Wrap the readable struck text in del; keep the words inside. If it is illegible, use del with a gap and a @reason. Only use gap for what you genuinely cannot recover.

Can I record the order in which changes were made?

Yes, with @change pointing to documented revision campaigns, or with @seq on substitutions. For a first project, recording each change accurately is enough; sequencing is an advanced genetic-editing concern.

Related reading

• Getting started with TEI P5
• Encode a critical apparatus
• Encode a manuscript description
• Encode a genetic edition
• Encode uncertain dates in TEI