Table of Contents
Name transliteration screening is the part of sanctions work that deals with names written in another script. A sanctioned person may be recorded in Arabic, Russian, or Chinese. Your customer data holds a Latin-alphabet version of that same name. Transliteration is the bridge between the two. Get the bridge wrong and a real hit slips through, because the spelling in your file never lines up with the spelling on the list.
One foreign name can turn into a dozen Latin spellings. There is no single correct way to write most of them. This guide explains why that happens, how it differs by script, which standards exist, and what to do about it in a screening program. It focuses on transliteration itself. For the scoring side of near-matches, see our companion guide below.
What transliteration means in screening
Transliteration converts a name from one writing system into another. It maps letters or sounds, not meaning. “Translation” turns words into a new language. Transliteration keeps the name and just changes the script it is written in. When a Russian name in Cyrillic becomes a Latin spelling, that is transliteration.
This matters because sanctions lists and your records often hold the same person in two different scripts. Government lists like the OFAC Specially Designated Nationals list publish names in Latin script, and many entries carry the native-script version and several aliases too. Your job is to make the person in your file match the person on the list, even when the spellings differ.
Why one name produces many spellings
Several forces pull the same name into different Latin forms. Knowing them helps you predict the variants you need to catch.
- Different source languages. A name transliterated for a French passport looks different from the same name transliterated for an English one.
- Missing vowels. Some scripts leave vowels out, so a reader has to add them back and different people add different ones.
- Sounds with no exact match. A sound in Arabic or Russian may have no clean Latin letter, so writers pick the nearest one.
- Personal choice. People pick their own spelling when they migrate, and that spelling can differ from any official rule.
- Deliberate change. Bad actors alter a spelling on purpose to dodge a screening system.
The result is that Mohammed, Muhammad, Mohamad, and Mohamed can all point to the same person. A screening tool that treats them as four different people is not doing its job.
How the challenge changes by script
Each writing system breaks in its own way. A tool that handles one well can fail on another.
Arabic
Arabic is an abjad. Short vowels are often left out of the written form. A reader supplies them, and choices vary. The definite article “al” gets attached, dropped, or hyphenated in different records. One Arabic name can appear as Abdul Rahman, Abdulrahman, or Abd al-Rahman.
Cyrillic
Russian, Ukrainian, and other Cyrillic names carry sounds that Latin letters handle in more than one way. The same name can become Aleksandr or Alexander, Petrov or Petroff, Yury or Iurii. Cyrillic also has letters that look like Latin ones but are not, so raw data needs cleaning before any match.
Chinese
Chinese uses characters that stand for whole syllables and meanings. Turning them into Latin form loses information. Two romanization systems compete: Pinyin, used on the mainland, and Wade-Giles, older and still seen in some regions. The family name comes first in Chinese order, which flips how a name looks once written out.
Other scripts
Korean, Japanese, Thai, and Persian each add their own quirks. Japanese uses more than one script at once. Korean has competing romanization rules. The lesson holds across all of them: a name that starts in a non-Latin script rarely has one Latin spelling.
Transliteration standards you should know
Formal standards exist to make transliteration consistent. Governments and libraries use them. They do not stop the problem, because the real world does not follow one rule, but knowing the main ones helps you understand your data.
| Standard | Used for | Common context |
|---|---|---|
| Pinyin | Chinese to Latin | Mainland China, most modern records |
| Wade-Giles | Chinese to Latin | Older and regional records |
| BGN/PCGN | Many scripts to Latin | US and UK government use |
| ISO 9 | Cyrillic to Latin | Library and academic use |
| ICAO Doc 9303 | Names in passports | Machine-readable travel documents |
| ALA-LC | Many scripts to Latin | Library cataloguing |
The trouble is that different bodies use different standards. A passport may follow ICAO rules. A bank may follow its own. Your customer might spell their name a third way. You cannot assume any one standard produced the name in front of you.
How sanctions lists handle transliteration
Good news: the lists do part of the work for you. OFAC and other bodies build alternate spellings into their entries so you are not left guessing every variant.
Aliases and AKAs
An SDN entry can list many “also known as” names. These cover spelling variants, transliterations, and known aliases. OFAC marks some as strong and some as weak. A weak alias is a partial or common name that will trigger more false hits, so it needs extra care in review. You can see the full entries on the OFAC SDN human-readable lists.
Native-script fields
Many list entries also carry the name in its original script. If your system can screen in the original script, you skip transliteration for those records and match script to script. That is often more reliable than matching two guessed Latin spellings against each other.
Secondary identifiers
List entries include date of birth, place of birth, passport numbers, and nationality where known. These do not depend on spelling. Use them to confirm or clear a name hit when the spelling alone is unclear.
Handling transliteration in your program
You cannot control how names arrive. You can build a process that copes with them. These steps make transliteration a managed risk instead of a blind spot.
- Normalize your data first. Strip accents, fix casing, and standardize spacing before any match runs.
- Keep the original script where you have it. Do not throw away native-script data during onboarding.
- Screen native script to native script when both sides carry it. Match Latin to Latin only when you must.
- Test known transliterations. Run names like Mohammed, Muhammad, and Mohamad through your tool and confirm they all alert.
- Lower the threshold for high-risk scripts. Arabic and Cyrillic names need a more sensitive setting than plain English ones.
- Lean on secondary identifiers. Use date of birth and nationality to settle unclear name hits.
- Ask your vendor direct questions. Find out which scripts and standards their engine supports before you trust it.
Transliteration is one edge case among several that break weak matching. It works alongside spelling variants, nicknames, and reordered names. Our guide on fuzzy name matching in sanctions screening covers the scoring methods and thresholds that decide when a near-match becomes an alert. Read the two together. This page explains why the spellings differ. That one explains how a system measures the gap.
Deliberate spelling changes are also a warning sign. When a party alters a name to slip past screening, treat it as a red flag, not a clerical quirk. See red flags of sanctions evasion for the wider pattern.
Fitting it into the wider screening process
Transliteration handling is one setting inside a full program. It only helps when the rest of the process is sound. You need a clear workflow that says when you screen, who reviews alerts, and how you record each decision. Start with our denied party screening hub for the full picture, and use the step-by-step process for denied party screenings to structure reviews.
Names transliterated from Arabic, Russian, or Chinese also feed straight into the false positive problem, because common transliterated names match many list entries at once. Our note on how to resolve a false positive walks through clearing those hits. And because regulators expect proof, keep an audit trail as covered in record keeping for denied party screening. Every cleared transliteration hit should carry a documented reason.
Frequently Asked Questions
What is the difference between transliteration and translation in screening?
Translation changes words into another language. Transliteration keeps the same name and rewrites it in another script. Screening cares about transliteration, because a sanctioned person keeps the same name across scripts. You are matching the person, not translating a meaning.
Why does one foreign name have several English spellings?
Some sounds have no exact Latin letter, and some scripts leave vowels out. Different people and different standards fill the gaps in different ways. That is why one Arabic or Russian name can appear as several English spellings that all point to the same person.
Should I screen names in their original script?
Yes, when both your data and the list carry the original script. Matching script to script avoids the guesswork of two separate transliterations. Keep native-script data during onboarding so you have the option. Fall back to Latin matching only when the original script is missing.
Do sanctions lists include alternate spellings?
Yes. OFAC and other bodies list “also known as” names that cover spelling variants and transliterations. Many entries also carry the native script and identifiers like date of birth. Use those extra fields to confirm or clear a hit when the spelling alone is unclear.
Key Takeaways
Transliteration is why the same sanctioned name can look different in your records than on the list. One foreign name yields many Latin spellings, and the pattern shifts by script. Arabic drops vowels, Cyrillic sounds map more than one way, and Chinese has competing romanization systems. Standards exist but nobody follows just one. Handle it by normalizing your data, keeping the original script, screening script to script when you can, and leaning on identifiers like date of birth. Test known variants, lower thresholds for high-risk scripts, and document every cleared hit. If you want help building or reviewing a screening process that copes with transliterated names, contact us.