A personal name is not merely a word that can be converted mechanically from one alphabet into another.

It is also an identity label connected to passports, education records, professional work, publications, financial accounts, family traditions, immigration files, and personal preference.

For that reason, one name may legitimately have several Latin-script forms.

The Korean surname 박 can be systematically romanized as Bak, yet Park is far more common as an established family spelling. The Japanese surname 大野 might be represented as Ōno, Oono, Ohno, or Ono, depending on the system and document. The Arabic name محمد has widely used forms including Muḥammad, Muhammad, Mohammad, Mohammed, and Mohamed. A Chinese name may have a Hanyu Pinyin form, an older Wade–Giles form, a Cantonese-derived spelling, and a separate passport spelling.

These variants do not necessarily indicate error.

They may reflect:

• Different romanization standards
• Different languages or regional pronunciations
• Historical spelling systems
• Passport restrictions
• Family preference
• Professional identity
• Migration history
• Technical removal of diacritics
• Differences in name order
• Established conventional usage

The central rule is therefore simple:

A mechanically generated romanization should not automatically replace the spelling a person officially or consistently uses.

This guide explains how personal names are romanized, why variants arise, which standards take priority in different contexts, and how databases and publications should preserve both linguistic accuracy and personal identity.

Quick Answer

When romanizing a personal name, use the person’s official or preferred Latin-script spelling whenever one exists. Use a formal romanization system only when no established form is available or when a specific institutional standard requires one.

A practical order of priority is:

1. Official spelling on the relevant identity document
2. Spelling explicitly preferred by the person
3. Established professional or public spelling
4. Spelling consistently used by the family or institution
5. Romanization required by a legal, library, academic, or government standard
6. Automatically generated romanization

The generated form may still be valuable for searching, cataloging, comparison, and linguistic explanation. It should normally be stored as an alternative rather than presented as a correction.

The Five Forms a Personal Name May Have

A multilingual identity record may contain several distinct name forms.

These fields may contain different values without contradiction.

For example, a Korean record might contain:

The official public form is not linguistically identical to the mechanical romanizations, but it is the appropriate identity form because it is the spelling established by the person.

Romanization Is Not the Same as Identity

A romanization standard answers a linguistic question:

How should these source characters be represented in Latin script?

An identity document answers a legal or administrative question:

Under which name is this person recorded by the issuing authority?

The answers may differ.

A technical standard may produce one form, while a passport authority permits another form already used in official or social life. A library may generate a standardized access point, while the author publishes under a different spelling. A news organization may use the best-known form rather than a fully marked scholarly transliteration.

A name should therefore be evaluated according to context.

There is no contradiction in using different forms in different fields, provided that the relationships between them are documented.

Personal Names in Passports

International travel documents illustrate why name romanization is both linguistic and technical.

ICAO Doc 9303 permits national-script characters in the passport’s visual inspection zone. When those characters are not Latin-based, a Latin transcription or transliteration must also be provided.

A passport can therefore display:

• A name in the national script
• A Latin-script version visible to people
• A more restricted machine-readable representation

These versions may not look identical.

Visual Inspection Zone

The visual inspection zone, or VIZ, is the part of a passport data page read by a person.

Depending on the issuing state, it may preserve:

• Lowercase and uppercase distinctions
• Spaces
• Selected punctuation
• Diacritics
• The original national script
• A Latin transliteration

ICAO allows issuing states to use diacritical marks in the visual zone.

Machine-Readable Zone

The machine-readable zone, or MRZ, is designed for global automated processing.

ICAO requires names in the MRZ to use a limited set of uppercase OCR-B characters without diacritics. National-script characters cannot appear there. Names may also be truncated when they exceed the available character positions.

Consequently:

• José may become JOSE
• Müller may become MUELLER or another authorized form
• Šime may become an approved ASCII equivalent
• A long multi-component name may be shortened
• Apostrophes and punctuation may disappear or be transformed

The MRZ is therefore a technical representation of identity data, not necessarily the person’s preferred display name.

Why the Passport and Booking Must Match

Airlines, immigration systems, visa systems, and border databases often compare bookings against passport machine-readable data. A difference caused by transliteration, spacing, hyphenation, or name order can therefore create operational problems even when both spellings are linguistically defensible.

Current UK passport guidance, for example, requires applicants generally to align the name used for official purposes across their supporting and foreign documents, while allowing exceptions when legal, cultural, social, or technical restrictions make alignment unreasonable.

For travel, the safest practical rule is:

Enter the name according to the passport instructions or the Latin spelling displayed on the travel document, not according to a newly generated romanization.

There Is No Universal Passport Romanization

ICAO standardizes the structure and machine-readability of travel documents, but issuing states still apply their own naming laws and transliteration policies.

Even within the same script, countries may adopt different tables. Long vowels, diacritics, surname order, apostrophes, and alternative names may be handled differently.

ICAO’s own Arabic guidance notes that a name such as محمد has multiple applied Latin variants, including forms based on different transliteration practices.

This means that an ICAO-compliant document does not guarantee that every country will issue the same Latin spelling for the same original-script name.

Original Name, Official Name and Generated Name

These three concepts should never be merged into a single database field.

Original-script name

The name as written in its native or legally relevant script.

Examples:

• 李小龍
• 박지성
• 大野太郎
• محمد علي
• Олександр Шевченко
• कृष्ण कुमार

Official Latin-script name

The form recorded by the relevant authority or consistently used in official international contexts.

Examples might include:

• Bruce Lee
• Park Ji-sung
• Taro Ohno
• Muhammad Ali
• Oleksandr Shevchenko
• Krishna Kumar

Generated romanization

A form calculated according to a named linguistic system.

Possible generated outputs might include:

• Mandarin Hanyu Pinyin
• Cantonese Jyutping
• Korean Revised Romanization
• Japanese Hepburn
• Cyrillic ISO 9
• Arabic ALA-LC
• Sanskrit IAST

A generated result should be labeled with its system. A field containing only romanized_name is often inadequate because it does not reveal what authority or rules produced the spelling.

Why Personal-Name Variants Arise

Different Romanization Systems

One script may have several competing systems.

Examples include:

A name can therefore receive several technically systematic outputs.

Different Source Languages

The same characters may be read differently in different languages.

A Chinese-character name may have:

• A Mandarin reading
• A Cantonese reading
• A Hokkien reading
• A Japanese reading
• A Korean reading

A Mandarin Pinyin result is not automatically the correct representation of a Cantonese-speaking person’s name.

Historical Systems

A family may have adopted a Latin spelling decades before the current official standard existed.

Changing it could disrupt:

• Family continuity
• Educational records
• Professional recognition
• Published citations
• Property records
• Migration documents
• Brand or company identity

Historical forms should therefore not be treated as mistakes merely because a newer standard would produce another spelling.

Colonial and Migration History

Names may have passed through another administrative or linguistic tradition.

A spelling may reflect:

• French
• Dutch
• Portuguese
• Spanish
• British
• Russian
• Ottoman
• Japanese
• Local Southeast Asian conventions

Such a spelling may be indirect rather than a modern conversion directly from the current source script.

Dialect and Regional Pronunciation

Arabic, Chinese, and other multilingual script traditions have substantial regional variation.

The Arabic letter ج, for example, may be reflected differently when a name follows a standard-language reading, an Egyptian pronunciation, or an established foreign spelling. A Chinese surname may be represented through Mandarin in one context and Cantonese or Hokkien in another.

Technical Restrictions

Older systems frequently lacked support for:

• Unicode
• Diacritics
• Modifier letters
• Long names
• Multiple spaces
• Apostrophes
• Hyphens
• Non-Latin scripts

A marked form such as Muḥammad, Čajkovskij, or Kṛṣṇa may therefore have acquired a simplified form such as Muhammad, Chaykovskiy, or Krishna.

Personal Choice

People sometimes select a spelling because it is:

• Easier to pronounce internationally
• Already used by relatives
• Consistent with a professional identity
• More recognizable
• Easier to type
• Less likely to be misread
• Required by an earlier document

A linguistic standard can explain alternatives, but it does not determine a person’s identity preference.

Name Order Is Part of Identity

Romanization involves more than converting letters. The order of name components also matters.

Many systems assume a simple Western structure:

Given name + surname

But personal names around the world can contain:

• One name only
• Multiple given names
• Multiple surnames
• Patronymics
• Matronymics
• Generational names
• Clan or tribal names
• Religious components
• Prefixes
• Particles
• Titles
• Names that change form by gender or grammar

Unicode CLDR’s person-name model explicitly recognizes that people may have one, two, or many name components and that individual name fields can themselves contain multiple words.

CLDR also distinguishes given-name-first and surname-first formatting and allows a person’s preferred order or the cultural order associated with the name’s locale to determine presentation. Chinese, Japanese, Korean, and several other name traditions are commonly formatted surname first.

Do Not Reverse Names Automatically

A Korean name written Kim Minseo should not automatically become Minseo Kim in every context.

A Chinese name written Wang Wei may be ambiguous unless the system knows which component is the surname.

A Japanese name may appear surname first in Japanese institutional usage but given name first in an older English-language source.

The safest database model stores components separately:

The display order should be generated for the required locale without changing the underlying identity record.

Chinese Personal Names

Chinese names demonstrate the importance of language variety and historical convention.

A name written with Han characters may have:

• A Mandarin Hanyu Pinyin form
• A Wade–Giles form
• A Cantonese-derived form
• A Hokkien or Teochew-derived form
• A family-established diaspora spelling
• A Western given name
• A passport spelling

A well-known traditional form may differ greatly from mechanical Mandarin Pinyin.

For example, systematic conversion should not automatically replace:

• Sun Yat-sen with a generated Mandarin form
• Chiang Kai-shek with a generated Pinyin form
• Bruce Lee with a character-based Mandarin romanization

These names function as established identities.

Chinese name order

Chinese names traditionally place the family name first.

When a person adopts a Western given name, several arrangements are possible:

• Wang Xiaoming
• Xiaoming Wang
• David Wang
• Wang David Xiaoming
• David Xiaoming Wang

A database should not infer the components from position alone.

Which form should be used?

Use:

• The person’s official Latin spelling for identity and current biography
• A systematic Pinyin form when linguistic analysis requires it
• Historical variants for archival discovery
• Regional-language romanization when that reflects the person’s actual name tradition

Japanese Personal Names

Japan’s current public romanization guidance recognizes the importance of individual preference.

The Japanese government’s 2025 explanatory guidance says that the wishes of individuals and organizations should be respected when their names are written in Roman letters. It specifically acknowledges established spellings that may use forms such as oh for a long vowel.

This means that one surname can appear as:

• Ono
• Ōno
• Oono
• Ohno

A generic Hepburn converter cannot determine which spelling the person actually uses.

Japanese passport identity

Japanese passports base the formal surname and given name on the family register. Under defined conditions, a former surname or alternative surname or given name recognized in social life can also appear in brackets in the visual zone, although that alternative is not entered in the MRZ or passport chip.

This creates several potentially distinct fields:

• Family-register name
• Passport Latin spelling
• Alternative name shown visually
• MRZ spelling
• Preferred professional name

A website or database should preserve those distinctions rather than collapsing them.

Korean Personal Names

South Korea’s Revised Romanization rules provide a systematic method for names but also preserve established usage.

The official rules place the family name first, followed by a space and the given name. Given-name syllables are normally joined, although a hyphen is permitted. The rules also state that personal names and company names may continue to be written as they have traditionally been written.

Consequently:

The National Institute of Korean Language has repeatedly clarified that although a mechanical form may follow the standard table, an established personal spelling can continue to be used.

Given-name spacing

A Korean given name usually consists of two syllables that belong to one name.

Possible presentations include:

• Minseo
• Min-seo
• Min Seo

The applicable official, personal, or institutional form should be preserved. Search systems should connect the spacing variants without silently rewriting the display name.

Arabic Personal Names

Arabic personal names present special problems because Arabic writing usually omits short vowels and because names can contain several grammatical and genealogical elements.

A complete name may include:

• Ism: personal given name
• Kunyah: teknonym, often beginning with Abū or Umm
• Nasab: patronymic, often using ibn or bint
• Nisbah: family, tribal, occupational, or geographic association
• Laqab: epithet or honorific

The same Arabic letters can produce multiple Latin forms because of:

• Unwritten vowels
• Dialect differences
• Sun-letter assimilation
• French or English spelling conventions
• Different treatment of ʿayn and hamza
• Joined versus separated name elements

Examples include:

The technically detailed form is useful for scholarship. The official or preferred spelling should control identity use.

Do not normalize every Abd name

A name beginning with عبد may be divided or joined differently:

• Abd al-Rahman
• Abdelrahman
• Abdul Rahman
• Abdurrahman

These may reflect separate regional, grammatical, passport, or family conventions. A database should not assume that one is the canonical form for every bearer.

Cyrillic-Script Personal Names

Cyrillic is used for many languages, each with its own letters and national conventions.

A Russian transliteration table should not automatically be applied to:

• Ukrainian
• Belarusian
• Bulgarian
• Serbian
• Macedonian
• Mongolian
• Kazakh in Cyrillic
• Other non-Slavic Cyrillic languages

A name can also have several forms because ISO 9, ALA-LC, BGN/PCGN, and national passport systems use different mappings.

For example, a Russian name may receive:

• A reversible ISO-style transliteration
• An ALA-LC library form
• A practical English-oriented form
• A passport form
• A historically established spelling

The Ukrainian name Олександр should not automatically be treated as though its source form were Russian Александр. Language identification is part of accurate name romanization.

Indic Personal Names

Indic-script names may have a precise scholarly transliteration and a separate conventional identity spelling.

Examples include:

Library of Congress guidance distinguishes technical South Asian romanization, which uses diacritics, from conventional romanization, which often replaces marked letters with unmarked spellings or letter sequences.

Modern-language pronunciation creates another complication. A strict Devanagari transliteration may preserve inherent vowels that are deleted in Hindi speech and official names.

Thus:

• Scholarly Sanskrit-style form: Bhārata
• Common Hindi-oriented form: Bhārat
• Simplified identity or public form: Bharat

The correct choice depends on the language and person, not only the script.

Diacritics and Identity

Diacritics can preserve important linguistic distinctions:

• ā versus a
• č versus c
• ḥ versus h
• ṣ versus s
• ṭ versus t
• š versus s
• ʿayn versus an omitted symbol

However, official documents and computer systems may remove them.

Current UK passport guidance, for example, uses defined alternative spellings when a name contains a diacritic that cannot be included in the passport system. When no standard alternative is available, supporting evidence from an authority or embassy may be requested.

This creates an important distinction:

• Diacritic-preserving display form
• Official document form
• ASCII or machine-readable form

A database should support all three.

Example:

Removing the diacritic from a search field is reasonable. Deleting the preferred form is not.

Hyphens, Apostrophes and Spaces

Punctuation may be legally or culturally significant.

Examples include:

• O’Connor
• D’Alessandro
• Abd al-Rahman
• Min-seo
• Jean-Pierre
• Nur ad-Din
• De la Cruz
• van den Berg

A system that removes punctuation may produce a technically necessary alias, but it should not automatically redefine the name.

ICAO’s MRZ restrictions mean punctuation is represented differently in machine-readable passport data.

Therefore:

• Passport visual spelling and MRZ spelling may differ
• Search matching should ignore selected punctuation differences
• Display output should preserve the person’s preferred punctuation
• Legal workflows should use the relevant document form

Name Variants Are Not Always Aliases of Equal Status

A strong identity model distinguishes the reason for each variant.

Do not store all of these as an unlabeled list of “other names.”

Labels provide provenance and prevent an obsolete or technical form from replacing the active identity.

How Libraries Handle Romanized Names

Libraries require standardized names so that works by the same person can be discovered together.

ALA-LC Romanization Tables provide approved transliteration schemes for more than 145 languages and dialects written in non-Roman scripts. The Library of Congress maintains language-specific tables and records their approval or revision dates.

A library authority record may therefore contain:

• Authorized access point
• Original-script form
• Alternate romanization
• Earlier form
• Later form
• Pseudonym
• Variant spelling
• Related identity

The catalog form is designed for retrieval and authority control. It is not necessarily the spelling that should appear in a person’s passport, social profile, or biography.

Designing a Name Database

A multilingual name system should avoid one field called full_name.

A more resilient model is:

Not every culture divides names naturally into first_name and last_name. CLDR’s person-name model uses more flexible name parts and locale-sensitive formatting because name structures vary widely.

Never require a surname without necessity

Some people have one name.

A form that requires both “first name” and “last name” may force users to:

• Duplicate their name
• Enter a placeholder
• Invent a surname
• Create a mismatch with their passport

Use fields that allow:

• One name
• Multiple given names
• Multiple surnames
• Optional components
• A complete display name independent of parsing

Store order separately

Do not infer identity order from screen position.

Store components and a preferred order, then format them according to the person’s preference or the target locale.

Store provenance

Every official or generated form should record its source:

• Supplied by person
• Copied from passport
• Derived from library authority record
• Generated by ISO 9
• Generated by Pinyin
• Imported from an older database
• Manually corrected

Without provenance, a system cannot determine which spelling should control a transaction.

Search and Name Matching

Search systems should tolerate reasonable variation without claiming that every similar spelling belongs to the same person.

Useful normalization may include:

• Case folding
• Diacritic-insensitive matching
• Hyphen and space variants
• Apostrophe variants
• Joined and separated name components
• Alternate romanization standards
• Common historical spellings

For example, a search for Muhammad may also consider:

• Mohammed
• Mohammad
• Mohamed
• Muḥammad

A search for Tchaikovsky may consider:

• Chaykovskiy
• Chaikovskii
• Čajkovskij
• Чайковский

However, name matching should also use:

• Date of birth
• Nationality
• Record identifier
• Address
• Organization
• Original-script name
• Other verified attributes

Romanization can merge different source names into the same Latin spelling. Similarity is evidence for retrieval, not proof of identity.

Editorial Policy for Personal Names

A publication can use the following hierarchy.

Contemporary people

Use the spelling the person uses in reliable official or professional sources.

On first mention, optionally add the native-script name:

Park Ji-sung (박지성)

Do not replace the established form with a theoretical mechanical spelling.

Historical people

Use the name most recognizable to the intended audience, while providing systematic or source-language variants when relevant.

Example format:

Laozi, also written Lao-tzu in Wade–Giles sources

Scholars and authors

Use the spelling under which the person publishes, even when a library’s romanization table would produce another form.

Quoted sources

Preserve the spelling used in the title, quotation, or historical document.

Linguistic discussion

Use a declared romanization system and explain differences from the person’s established identity form.

Inconsistent sources

When reliable sources disagree:

1. Prefer the person’s own verified usage.
2. Check official documents or institutional profiles.
3. Preserve important alternatives as aliases.
4. Avoid declaring one form incorrect without evidence.

A Practical Name-Romanization Workflow

Step 1: Preserve the source name

Store the complete original-script name exactly as supplied.

Step 2: Identify the language

Do not rely only on script detection.

Determine whether the name is:

• Mandarin or Cantonese
• Russian or Ukrainian
• Arabic or Persian
• Sanskrit or Hindi
• Japanese or Chinese
• Another relevant language

Step 3: Ask whether an official Latin form exists

Check:

• Passport
• National identity card
• Government record
• Institutional profile
• Author publication
• Person’s own stated usage

Step 4: Record the preferred display form

Allow the individual or authoritative source to determine how the name should be displayed.

Step 5: Generate systematic alternatives

Use named standards only for clearly defined purposes:

• Cataloging
• Linguistic comparison
• Search
• Academic notation
• Data exchange

Step 6: Record name order

Store whether the person prefers:

• Family name first
• Given name first
• A single undivided name
• Another culturally appropriate structure

Step 7: Generate technical forms separately

Create separate fields for:

• MRZ-compatible spelling
• ASCII search form
• URL slug
• Sort key

Do not use these technical values as the preferred display name.

Step 8: Preserve provenance

Record where each form came from and when it was confirmed.

Recommended Output for Romanization.org

A personal-name tool should not display only one result.

A useful result panel could contain:

Original name

The source-script form.

Detected language and script

For example:

• Korean — Hangul
• Japanese — kanji and kana
• Mandarin Chinese — Han characters
• Arabic — Arabic script
• Ukrainian — Cyrillic

Official or preferred Latin form

Displayed when supplied or documented.

Systematic romanizations

A side-by-side comparison:

Name order

Show:

• Family name
• Given name
• Preferred order
• Alternative locale order

Information-loss warnings

Examples:

• Diacritics removed
• Long vowels omitted
• Source characters merged
• Dialect unknown
• Multiple readings possible
• Name order uncertain
• Official spelling may differ
• Result is not passport advice

Identity notice

A clear notice should state:

Generated romanizations are reference forms. Use the spelling on the relevant official document for travel, immigration, banking, and legal transactions.

Common Mistakes

Correcting a person’s name to match a standard

A standard form may be linguistically systematic but socially and legally inappropriate.

Treating the passport MRZ as the preferred name

The MRZ may remove accents, punctuation, spacing, and parts of a long name.

Assuming one script means one language

Arabic script, Cyrillic, Devanagari, and Han characters are each used across multiple languages.

Reversing name order automatically

Family-first order is not an error.

Removing diacritics from the only stored record

Create a simplified search form while preserving the marked display form.

Using first name and last name as universal concepts

Some names do not fit that structure.

Ignoring historical aliases

Old spellings are essential for archives, citations, genealogy, and search.

Assuming similar romanizations prove identity

Different original names can collapse into the same Latin form.

Mixing standards without labels

A name should not combine one standard’s consonants, another’s vowel marks, and a third system’s spacing unless the result is explicitly documented as a house style.

Frequently Asked Questions

Which romanization should be used for a personal name?

Use the person’s official or preferred Latin spelling when available. Use a systematic romanization when a specific academic, library, or technical purpose requires it.

Is the passport spelling always the person’s preferred spelling?

Not necessarily. Passport systems may impose character restrictions, omit diacritics, or use a state-mandated transliteration. It remains the controlling form for travel-document matching.

Can a passport show both native script and Roman letters?

Yes. ICAO permits national characters in the visual zone and requires Latin transcription or transliteration when the national characters are non-Latin.

Why does the MRZ look different from the name above it?

The MRZ uses a restricted uppercase character set, excludes diacritics and national-script characters, replaces punctuation according to defined rules, and may truncate long names.

Should a surname be written first or last?

Follow the person’s preferred or culturally appropriate order. For passports and forms, follow the document’s field structure. CLDR supports both surname-first and given-name-first presentation.

Is Park an incorrect romanization of 박?

No. Bak is the direct result under standard Revised Romanization rules, but South Korean guidance permits established personal-name spellings to continue. Park is a well-established identity form.

Should Japanese Ōno be written Ohno or Ono?

Use the spelling chosen or officially recorded for the person. Japanese policy explicitly recognizes that personal names may retain spellings such as oh for long vowels.

Why are there so many spellings of Muhammad?

Arabic short vowels are often absent from ordinary writing, and different standards, dialects, passport authorities, and European-language traditions represent the name differently. ICAO documentation itself recognizes multiple applied variants.

Can names be romanized automatically?

They can be converted mechanically, but the result may not match the person’s official identity, language variety, established family spelling, or preferred name.

Should diacritics be used in a person’s name?

Use them when they are part of the person’s preferred or authoritative spelling and the platform supports them. Maintain a separate simplified form where technical matching requires one.

How should a website handle a one-word name?

Store it as a complete name without forcing an invented surname. Forms should permit empty or non-applicable name components.

Should old spellings be deleted after a person changes romanization?

No. Mark them as former or historical variants when they remain useful for record matching, citations, or discovery.

Which name should be used in an airline booking?

Use the form required by the airline and matching the relevant passport, usually based on the Latin-script passport name or MRZ instructions.

Can two romanized names be assumed to represent the same person?

No. Romanization variants can suggest a possible match, but identity must be confirmed with additional information.

Final Checklist

Before publishing or storing a romanized personal name, verify:

• The original-script name has been preserved.
• The source language has been identified.
• An official Latin spelling has been checked.
• The person’s preferred form has been considered.
• Name order has not been changed automatically.
• The romanization standard has been named.
• Historical and regional variants have been labeled.
• Diacritics have not been removed from the only stored copy.
• MRZ and display forms are stored separately.
• Search aliases do not overwrite the primary identity.
• Single-name and multi-surname structures are supported.
• The provenance of each spelling has been recorded.
• Generated forms are not presented as legal advice.
• Similar spellings are not treated as proof of identity.

Conclusion

Romanizing a personal name is not simply a process of replacing non-Latin characters with Latin letters.

It involves three overlapping questions:

1. How is the name written?
2. How is the name pronounced or systematically romanized?
3. Which spelling represents the person’s identity in the relevant context?

A technical standard may produce a linguistically precise form. A passport may require a restricted form. A library may maintain a separate authority form. A person may use a historical, family, professional, or regionally based spelling.

All can be legitimate within their proper roles.

The safest approach is to preserve several clearly labeled layers:

• Original-script name
• Official Latin spelling
• Preferred display name
• Systematic romanization
• Machine-readable form
• Historical and searchable aliases

Romanization.org should treat personal names as identity data rather than ordinary vocabulary. Its tools should explain what each form represents, warn users about information loss, respect documented personal usage, and never imply that an automatic conversion overrides the name a person officially or consistently uses.