Cyrillic, diacritics and unusual names in documents

Why Cyrillic, diacritics and unusual names break retyping in client documents, how scripts and transliteration differ, and how to verify every name once.

Names in client documents must be copied exactly as printed, character by character, from the document that governs the matter. That rule sounds obvious and is broken daily, because names arrive in different scripts, with diacritics the keyboard does not offer, and in several transliterations that are all, in some sense, correct. This post is about why names break retyping and how to make them safe.

Why do names break retyping?

A name is the one value in a document with no internal check. A date can be implausible, an amount can fail arithmetic, but a name is right only if it matches the source, and the typist often has no feel for what the source should say. Unfamiliar names are the worst case: the typist has never seen the spelling before, so nothing looks wrong.

  • Autocorrect quietly normalises: a phone or word processor rewrites an unusual name into a familiar word, and the change looks intentional.
  • Diacritics get dropped or approximated: Đ becomes D, ž becomes z, ë becomes e, because the keyboard layout in front of the typist has no key for the real character.
  • Doubled and swapped letters hide in long unfamiliar surnames the way a wrong digit hides in a long number.
  • The same person appears under two spellings in two source documents, and the typist picks one without noticing there was a choice.

Scripts are not fonts

A Cyrillic name is not a Latin name displayed differently. It is a different set of characters. Several Cyrillic letters look identical to Latin ones on screen and on paper: А, Е, О, Р, С, Х, В, М, Н, Т and К have Latin lookalikes, but they are different characters with different codes. A name typed with a mix of the two scripts looks perfect and is wrong, and the error surfaces later, when a registry search, an alphabetical sort or an exact-match lookup fails to find a person who is plainly in the system.

The practical rule: never retype a name across scripts by eye, and never fix a name by swapping in the lookalike letter from the other script. Copy from a digital source where one exists, and when the source is paper, treat script as part of the value. If the governing document carries the name in Cyrillic, the filled document carries it in Cyrillic, unless the receiving institution explicitly requires a transliteration.

One name, several correct spellings

Transliteration is the second trap. When a name written in Cyrillic or another non-Latin script appears in a passport, the machine readable zone at the bottom carries a Latin transliteration produced under published tables: ICAO Doc 9303 Part 3 (icao.int) specifies transliteration for the machine readable zone, while the visual zone may keep the national script. The same ICAO documentation allows for national differences in how governments transliterate certain Cyrillic characters. The consequence for your office is concrete: one person can legitimately appear under more than one Latin spelling across their own documents.

  • Decide which document governs the spelling for this matter, and say so in the file. Often it is the passport or national ID, but the answer can differ by institution.
  • Copy the governing spelling exactly, including its transliteration choices, even where you would transliterate differently.
  • When two source documents disagree, record both spellings and resolve the discrepancy explicitly instead of silently choosing.
  • Never construct your own transliteration for an official document. Someone else’s table already governs, and yours will not match it.

How do you verify a name?

Verify a name the way you verify a long number: once, character by character, against the governing document, at the point where it enters your document. Read it forwards in pieces of two or three characters rather than as a word, because reading it as a word invites your eye to see the name it expects. Check the diacritics as separate items, since they are the most likely loss.

  • Check script first: is the whole name in one script, the script the receiving institution expects?
  • Check diacritics: every mark present, none approximated.
  • Check against the governing document, not against another retyped copy of the name.
  • Do this once, at entry. A name verified at entry does not need re-verification every time the document is reopened.
Run Done removes the retyping step where names most often break. You upload the Word document you already use, add the client’s materials as photos, scans, documents or pasted text, and the software reads them and enters the values in the right places, taking each name from the materials you supplied rather than from anyone’s keyboard. A name the materials do not contain leaves its field empty; nothing is invented. You then review every entered value against the source, correct anything, and download the finished Word file. Control never leaves the office.

The short version

Names have no internal check, so they must be copied, not retyped. Treat script as part of the value and never mix Cyrillic and Latin lookalikes. Expect one person to carry several legitimate Latin spellings across documents, pick the governing one deliberately, and verify each name once, character by character, diacritics included, at the moment it enters your document.

See it on your own documents

Your first 50 fills are free. Upload a document you already use and the materials you have, and download the finished file.