If you are searching for how to translate passport numbers, how to translate national ID numbers, or how to handle identity document codes in a certified translation, the first rule is simple: the identifier itself is usually not translated. Passport numbers, national identification numbers, residence permit numbers, document serials, check digits and machine-readable identifiers function as evidence. Their job is to identify one exact document or record, so a translator should protect the string while translating the labels and explanatory language around it.
Identity-document translation matters in immigration files, visa applications, school admissions, employment records, banking compliance, legal proceedings, civil registration, insurance, travel and cross-border administration. A fluent translation can become unusable if a single letter is changed, a zero becomes the letter O, a hyphen disappears where the receiving system expects it, or a translator silently rewrites a local identifier into a target-country format that was never issued.
This guide explains how to translate identity-document labels while preserving passport numbers, national ID numbers, permit numbers, document codes, prefixes, suffixes and machine-readable data. It also shows how to separate transliteration of names from transcription of identifiers, how to handle masked numbers, how to preserve uncertainty in damaged documents, and how to verify that the target version still points to the same real document as the source.
Why identity-document numbers are not ordinary text
An identity number is a data object embedded inside language. The words around it can be translated, but the identifier often belongs to an issuing authority’s system rather than to the grammar of the source language. Its letters may encode document type, place of issue, year, sequence or internal validation. Even when the code looks meaningful, a translator should not decode and rewrite it unless an official receiving requirement explicitly calls for that.
Different identity systems also use different character sets and structures. Some identifiers are entirely numeric. Others combine Latin letters and digits. Some contain visible separators that are merely typographic; others rely on those separators for readability or system entry. A document may display one number visually while also encoding the same identity in a barcode, QR code, chip or machine-readable zone. The translator’s task is not to redesign the system but to preserve the source evidence accurately.
Identity documents frequently contain several numbers at once: passport number, personal number, national ID number, document serial number, file number, visa number, residence permit number, tax identifier or application reference. Treating all of them as “ID number” can destroy distinctions that matter to a receiving authority. Labels and values must remain paired.
The practical standard is traceability. A reviewer should be able to compare source and translation and see immediately that each translated label points to the same unaltered identifier. Where source quality is poor, the translation should preserve uncertainty rather than invent certainty.
A reliable translation method
1. Separate labels from identifiers
Translate labels such as “Passport No.”, “National ID No.”, “Document Number” or “Residence Permit Number” according to the target language. Copy the actual identifier exactly unless an authoritative transliteration rule applies to non-Latin characters inside the identifier itself. This separation prevents translators from treating evidence as prose.
2. Preserve character identity, not visual resemblance
Characters such as 0/O, 1/I/l, 5/S, 8/B and hyphen-like marks can be confused in scans. Compare the source carefully, zoom the image if needed, and check duplicate occurrences elsewhere on the document. Never guess merely because one reading looks more familiar.
3. Keep leading zeros
Leading zeros are part of many identifiers. Spreadsheet software and careless retyping often remove them. Treat identity numbers as text strings rather than mathematical values. A number such as 001274 is not interchangeable with 1274 when it is used as a document identifier.
4. Preserve prefixes and suffixes
Letters before or after the numeric core may encode series, issuing region or document type. Do not drop them as decorative. If the source presents AB1234567, the translation should not become 1234567 unless an official rule states that AB is merely a printed label rather than part of the identifier.
5. Keep formatting unless there is a documented reason to change it
Spaces and separators can improve readability and sometimes reflect official grouping. Preserve the source form in the translation, especially for certified work. If a receiving system requires a separator-free entry, that is a data-entry instruction, not a reason to alter the translated document.
6. Treat machine-readable zones as protected data
Machine-readable zones contain tightly structured characters and filler symbols. They are not prose. Reproduce them only when the translation format requires it, and do so exactly. Do not replace filler characters, reflow lines or insert translated words inside the sequence.
7. Mark illegible characters transparently
If the source is damaged or blurred, use the project’s accepted notation for illegibility rather than inventing a character. A translator can note “[illegible]” or a similarly approved convention in a certification context. The exact notation should follow the receiving authority or translation provider’s practice.
8. Verify every identifier independently
Do a second pass that ignores prose and checks only identifiers. Compare character by character, including leading zeros, prefixes, suffixes, punctuation and spaces. This dedicated QA pass catches mistakes that ordinary reading often misses.
Twenty recurring identity-document translation problems
1. Passport number versus personal number
A passport can display both a passport number and a separate personal or national number. The two strings may sit near each other and look equally official. Translate each label precisely and keep each value attached to the correct field. Do not merge them under one generic “ID number” heading.
For quality assurance, trace every use of the number in visas, endorsements, machine-readable data or accompanying forms. If a later page repeats only one of the identifiers, make sure the target wording still tells the reader which identifier it is.
2. National ID numbers with check digits
Some national identification numbers contain a final check digit or letter calculated from the preceding characters. That last character may look optional, especially when separated by a dash. It is not optional if it belongs to the official identifier. Preserve it exactly and never “correct” it from memory.
A useful QA technique is to compare the identifier anywhere else it appears on the document or supporting records. If an authoritative validation tool is publicly available, it can sometimes confirm whether a copied sequence is structurally plausible, but validation should never replace faithful transcription of the source.
3. Residence permit numbers
Residence cards can include card numbers, permit numbers, personal numbers and application references. Their labels may be abbreviated. Translate the field label according to its institutional meaning and preserve the value exactly. Avoid translating every number as “permit number” simply because the document is a permit.
Where the issuing authority publishes a multilingual specimen, use it to confirm field meaning. Official specimens are especially useful because they show which abbreviations correspond to which administrative concepts.
4. Visa numbers
A visa sticker or electronic visa can contain a visa number, application number, control number and passport number. Keep them distinct. If the source label is missing and the number’s role must be inferred from layout, use official specimen documents rather than guesswork.
Do not convert a visa number into a target-country visa format. Translation is not administrative reissuance. The translated document must continue to identify the original visa.
5. Document serial numbers
A serial number may identify the physical card or booklet rather than the person. That distinction matters when documents are renewed or replaced. Translate “document serial number” accurately so readers do not mistake it for a lifelong personal identifier.
In long files, keep a terminology note that distinguishes document number, serial number and personal number. Consistent labels reduce the risk of later cross-reference errors.
6. Application and file references
Administrative documents often show a case, file or application reference near the identity number. These references belong to a process rather than the person’s identity. Translate the label, protect the string and avoid collapsing it into the national ID field.
If a cover letter instructs the recipient to quote the reference, the translation should preserve exactly which reference is required. A wrong label can cause a valid number to be entered into the wrong field.
7. Leading zeros
Leading zeros are frequently lost when identifiers are copied through spreadsheet or database software. A translator should treat identifier strings as text. The safest workflow is to copy and verify the displayed sequence, not to retype it into a numeric field that may normalize it automatically.
When the target document is produced in a word processor or CMS, check the final rendered output as well. Automatic formatting can sometimes change spacing, convert hyphens or alter digit grouping.
8. Letter O versus zero
The source font may make O and 0 almost identical. Do not decide from appearance alone if other evidence exists. Compare the machine-readable zone, barcode text, duplicate field or issuing-system pattern. If the source remains genuinely unclear, mark uncertainty rather than inventing a confident value.
This principle applies equally to I/1, B/8 and S/5. Identity translation rewards slow visual checking more than clever inference.
9. Non-Latin letters inside identifiers
Some systems may display native-script characters alongside or inside official identifiers. Determine whether the issuing authority has an official Latin representation. Transliteration should follow that official convention where available; otherwise preserve the source form and provide a carefully labelled transliteration rather than silently substituting a look-alike Latin letter.
The key distinction is between translating a field label and converting the script of an identifier. Script conversion must not change identity.
10. Machine-readable passport zones
The machine-readable zone contains document type, issuing state, name data, document number, nationality, date fields, sex markers and check digits in a fixed technical format. If reproduced, it should normally be copied exactly rather than translated.
Do not wrap the line differently, replace filler characters with spaces or change punctuation for visual neatness. The sequence is structured data. Translate any explanatory caption outside the zone instead.
11. Masked identifiers
Source documents sometimes display only the last four digits or replace characters with asterisks. Preserve the masking. A translator should not reconstruct hidden digits from another document unless the assignment explicitly requires a verified cross-document reference and the receiving context permits it.
Masking is part of the evidence because it shows what the source actually disclosed. A full number in the translation can create a false impression that the source contained it.
12. Redacted identifiers
If a number is blacked out or formally redacted, preserve that fact. Use the accepted notation for redaction and do not infer the hidden value. Redaction is an intentional information-control decision, not a gap for the translator to fill.
The translated layout should make the extent of the redaction clear enough that a reader understands which field was withheld.
13. Expired and cancelled documents
An expired or cancelled passport number still identifies a historical document. Preserve it exactly. Do not replace it with a newer passport number from supporting records, even if the newer number belongs to the same person.
Translation should preserve document history. Status words such as expired, cancelled, void or replaced should be translated separately from the identifier.
14. Replacement document numbers
Where a record states that one document replaced another, both identifiers may appear. Keep the temporal relationship explicit. The old number is not an error merely because a new number exists.
Use labels such as previous document number and current document number only if the source supports that distinction. Do not infer which one is current from sequence alone.
15. Identity numbers embedded in sentences
Legal and administrative prose may say that “the holder of ID number X appeared before the authority.” Translate the sentence naturally but isolate the identifier mentally as a protected token. This prevents grammar editing from changing the number.
After drafting, compare the identifier in prose with any table or header version. Repetition creates both an opportunity for verification and a risk of inconsistency.
16. Identity numbers inside tables
Tables can shift columns during translation because target labels expand. The number may remain correct but end up under the wrong heading. Check column association, not only the string itself.
For long tables, use row-by-row QA: person name, document type, document number and status should stay together. Layout is part of factual integrity.
17. Identity document type codes
Systems may use short document-type codes such as P, ID, RP or locally defined abbreviations. Do not expand a code unless its meaning is known. Preserve the code as data and translate the descriptive label around it.
If an official glossary exists, record the expansion once and keep it stable across the document. An invented expansion can misclassify the credential.
18. Country and issuing-authority codes
Identity documents often use abbreviated country or authority codes. These can be technical identifiers rather than ordinary abbreviations. Keep them unchanged unless an official target representation exists.
Translate the country or authority name in prose according to established convention, but do not silently rewrite the code to match a translated name.
19. Handwritten corrections
An older document may show a handwritten correction to an identifier. Preserve the documentary state faithfully: original entry, correction, stamp or annotation as required by the translation format. Do not choose whichever value appears more plausible.
Where certification rules require a note such as “handwritten amendment,” make that editorial status explicit so the reader can distinguish source content from translator commentary.
20. Illegible or damaged identifiers
Water damage, glare, compression artefacts and worn print can make one or more characters unreadable. The correct response is not to fill the gap from pattern knowledge. Mark the illegibility according to accepted practice and, if allowed, request a clearer source outside the translated text.
Accuracy includes knowing when the source cannot support a confident transcription. A transparent uncertainty is better evidence than a polished invention.
Common failure modes
1. Translating the identifier itself
Identifiers are normally protected strings. Translate the field label, not the number or code. Replacing an official series prefix with a target-language abbreviation can create an identifier that never existed.
2. Removing leading zeros
This often happens automatically in spreadsheet software. Treat identifiers as text and compare the final output to the source.
3. Normalising punctuation without evidence
A translator may remove spaces or hyphens for neatness. Certified and official contexts generally benefit from preserving the source presentation unless a receiving requirement states otherwise.
4. Guessing ambiguous characters
Visual similarity is not enough. Use duplicate occurrences, technical zones or official patterns as evidence. If ambiguity remains, mark it.
5. Merging different identifiers
Passport number, personal number and document serial may all be different. Generic labeling can make valid data unusable.
6. Reconstructing masked or redacted values
The translation should reflect what the source shows. Hidden data should remain hidden unless the assignment explicitly changes the evidentiary task.
7. Using a new document number to “correct” an old one
Historical records identify the document that existed at that time. Preserve old and replacement numbers as distinct evidence.
8. Forgetting layout association
A perfect identifier under the wrong translated heading is still wrong. Check labels, rows, columns and footnotes as a system.
Worked practice
Practice 1: Passport number and personal number
The source passport displays two similar alphanumeric strings. One is labelled passport number; one is labelled personal number. Translate both labels separately, copy each value exactly and check the machine-readable zone to confirm which string belongs to the document number.
Practice 2: Leading zero in a national ID
The source ID begins with 00. Preserve both zeros even if software tries to remove them. Store the field as text, not number, and compare the final rendered version character by character.
Practice 3: Blurred final character
The final character could be 8 or B. Compare duplicate occurrences and technical data on the document. If no reliable evidence resolves the ambiguity, mark the character as illegible according to the required translation convention rather than guessing.
Practice 4: Masked residence permit number
The source displays only the last four digits. Preserve the masking and translate the field label. Do not reconstruct the hidden digits from another page unless the translation brief explicitly requires a consolidated verified record.
Practice 5: Replaced passport
A letter refers to an old passport number and then records a replacement passport. Keep both numbers and the chronological relationship. Do not replace the old number throughout the historical letter with the new one.
Practice 6: Identifier in a table
A translated heading becomes longer and pushes columns. After layout, verify that each person’s name, document type and number remain in the same row and under the correct heading. Formatting QA is part of identity QA.
Practice 7: Source acronym and number
The document labels a national identity field with a local acronym. Translate the full concept in the target text and retain the official acronym where it helps traceability. Do not invent a new acronym unless the issuing authority uses one.
Practice 8: Damaged document corner
Two digits are missing because the paper is torn. The translator should record the visible portion and mark the missing part according to accepted notation. Pattern completion is not evidence.
Using official specimens, OCR and AI safely
Official specimen documents are valuable because they identify fields and show standard layouts. They can help distinguish document number from personal number or explain an unfamiliar code. They should not be used to overwrite what the source actually shows.
OCR can speed up transcription but is risky for short identifiers because one incorrect character can invalidate the whole string. Treat OCR output as a draft and compare every identifier visually with the source. The shorter the string, the less redundancy there is to reveal an error.
AI can explain document fields and identify likely code structures, but it should never invent missing identifier characters. Use it for context, terminology and QA questions; use the source document as the authority for the actual value.
How this fits the wider eduKate translation system
Identity-document translation sits where language meets evidence. The broader translation method is developed in Master Art of Translation | The Complete System for Moving Meaning Between Languages. Vocabulary knowledge connects to the Vocabulary Learning Hub, while noun phrases, labels, reference and apposition connect to How English Works. The additional discipline here is evidence preservation: every translated label must still point to the exact same document identifier.
Deep verification: how to prove an identifier survived translation
Identity-document work deserves a verification method that is stricter than ordinary proofreading because identifiers are low-redundancy strings. A prose sentence often remains understandable after a small typo; a passport number may not. The practical solution is to separate linguistic review from data-integrity review. First review the translation for meaning and natural target language. Then perform a second pass in which you temporarily ignore style and inspect only labels, identifiers, dates, names and document relationships. This changes the reviewer’s attention from “Does the sentence read well?” to “Does every target field still point to the same source evidence?”
Build an identifier inventory before final QA
For multi-page files, create a small inventory of every identifier that appears: passport number, national ID number, residence permit number, visa number, application reference, document serial, previous-document number and any machine-readable or barcode-associated text that must be reproduced. Record the source label, the exact string, the page or location where it appears, and the target label chosen for it. This inventory is not another public document; it is a working control that lets the translator detect accidental relabeling and inconsistent transcription across pages.
The inventory is especially valuable when two identifiers differ by only one character or when one person has several documents in the same file. Human memory is poor at comparing short alphanumeric strings after repeated exposure. Externalising the comparison into a checklist reduces that cognitive burden and makes final review more systematic.
Use a three-column comparison
A useful verification layout has three columns: source label, source value and translated label plus reproduced value. Check the label pair first, then the value pair. This matters because an identifier can be copied perfectly but assigned to the wrong concept. For example, a correct document serial placed beside a target label meaning national identification number is still a serious translation error. The three-column method forces the reviewer to validate both semantics and data identity.
Read identifiers in grouped chunks
Long strings are easier to compare when read in stable chunks. Rather than visually scanning AB019274631 as one shape, compare AB / 019 / 274 / 631, while still preserving the original printed grouping in the published translation. Chunking is a review technique, not a licence to add spaces to the identifier. It helps the eye detect transpositions such as 47 becoming 74 or a missing middle digit that might otherwise be overlooked.
Check repetition as evidence, not as permission to overwrite
If the same identifier appears three times, repeated agreement gives useful confirmation. But repetition should not lead the translator to silently overwrite a source discrepancy. If one occurrence genuinely differs, record what the source shows and investigate whether the difference is a source error, an old identifier, a typographical error or a distinct field. Translation should preserve documentary reality; it should not manufacture internal consistency that the source never had.
Distinguish transcription notes from translated content
When a translator needs to mark illegibility, handwriting, erasure, overwriting or a damaged character, the notation should be recognisably editorial rather than pretending to be source text. The exact convention varies by certified-translation practice, court rule, institution or provider, but the principle is stable: a target reader must be able to tell what was printed on the source and what the translator is saying about the source. This protects evidentiary transparency.
Verify names and numbers separately
Names and identifiers often appear together, but they require different treatment. A person’s name may need transliteration, an established romanisation or preservation of diacritics. A document number usually remains unchanged. Do not let the fact that a name is legitimately converted between scripts create a habit of “translating” the adjacent identifier. In final QA, run one pass for personal names and a separate pass for identifier strings.
Protect identifiers during search and replace
Global find-and-replace can be efficient for repeated terminology, but identifiers should be protected from broad text transformations. A replacement designed to change a source abbreviation, punctuation pattern or date style can accidentally alter an alphanumeric identifier that contains the same character sequence. Before bulk operations, exclude known identifiers or rerun the inventory check afterward. Automation is useful only when the evidence remains unchanged.
Check the final file, not only the editor
The translation can be correct in the editing environment and wrong in the exported PDF or published webpage. Fonts may substitute similar-looking characters, line wrapping may separate a prefix from the rest of the number, tables may shift, and bidirectional text may alter visual grouping. Therefore the last data-integrity review should occur in the same format the recipient will receive. Copy a few critical identifiers from the final file and compare them with the source to confirm that export did not introduce changes.
Use check digits as a warning signal, not a rewriting tool
Some official identifiers use check digits or letters. If a reliable validation method says a copied string is structurally invalid, that is a reason to inspect the source again. It is not permission to calculate a different check digit and replace the source. The source document remains the evidentiary authority. A validation failure may indicate a transcription error, a damaged character, an older numbering scheme, a nonstandard document or simply that the assumed validation rule does not apply.
Maintain privacy while verifying identity data
Identity numbers are sensitive information in many contexts. Verification should use only the copies and systems required for the translation task. Avoid moving identifiers into unnecessary notes, public examples or external tools merely for convenience. When building internal QA inventories, keep them within the approved working environment and remove them according to the project’s retention rules. Translation accuracy and privacy are compatible goals: the translator can verify a string rigorously without spreading it further than necessary.
Know when an explanatory note is useful
A target reader may not know whether a source field called “personal number” is a national identity number, a population-register number or another administrative identifier. If the translation brief allows explanatory notes and the concept is important, a concise note can clarify the field’s function without changing the identifier. The note should explain what the source system calls the field, not claim equivalence with a target-country number unless that equivalence is formally established.
Run a cold final read
After intensive work, the translator becomes visually familiar with the document and starts seeing what they expect. A cold read after a short break, or a second-person review where available, improves identifier QA. The reviewer should compare from source to target rather than reading the target alone. Read one field at a time, point to the source value, point to the target value, and confirm exact identity before moving on. This intentionally slow final pass is appropriate because the cost of a single-character error can be much greater than the time saved by rushing.
The deeper lesson is that identity-document translation has two products at once: readable target-language information and a trustworthy bridge back to the source evidence. Good prose serves the first product. Exact transcription, transparent uncertainty and disciplined verification serve the second. Neither can substitute for the other.
FAQ
Should passport numbers be translated?
Normally no. Translate the field label and preserve the passport number exactly as shown.
Should spaces and hyphens be removed?
Not automatically. Preserve the source display unless a specific receiving-system instruction requires another input format.
What if a character is unclear?
Use duplicate occurrences, official structure and technical data as evidence. If uncertainty remains, mark it rather than guess.
Can I replace an old passport number with a current one?
No. The translation should preserve the document history recorded in the source.
What about national ID numbers with leading zeros?
Keep every leading zero. Treat the identifier as text rather than a mathematical number.
Should machine-readable passport lines be translated?
No. If reproduced, they should normally be copied exactly as structured data.
How should redacted numbers be handled?
Preserve the redaction. Do not reconstruct hidden values unless the task explicitly changes the evidentiary purpose.
Can OCR be trusted for passport numbers?
Use OCR only as a draft aid. Every identifier should be checked visually character by character.
Can AI infer a missing digit?
It may guess, but a guess is not evidence. Missing or illegible characters should remain uncertain unless another authoritative source resolves them.
What is the simplest rule for identity-document translation?
Translate the label; preserve the identifier.
Final checklist
- Have I separated every label from its identifier?
- Did I preserve leading zeros, prefixes, suffixes and check digits?
- Have I verified confusing characters such as O/0 and I/1?
- Are passport number, personal number and document serial still distinct?
- Did I preserve masking, redaction and illegibility?
- Are old and replacement document numbers kept in the correct chronology?
- Are machine-readable zones treated as protected data?
- Do tables and layout still pair the correct label with the correct value?
- Have I avoided inventing or normalising identifiers?
- Did I run a dedicated character-by-character QA pass?
Identity-document translation succeeds when the target text remains traceable to the exact evidence in the source. Translate the administrative language, preserve the identifier, maintain uncertainty where the source is uncertain, and verify every character independently. The result should help the target reader understand the document without ever changing which document, person or record the source actually identifies.