What is romanization?

Romanization, transliteration, transcription and translation

These four words are often used as if they meant the same thing. They do not, and the difference decides which spelling you end up with. Romanization is the umbrella term for any method that ends in Latin letters. Transliteration and transcription are the two ways of getting there, and translation is a different activity altogether.

Most official systems mix the two methods. The Revised Romanization of Korean is mainly a transcription, which is why it writes the sound changes between syllables, yet it also contains a letter-by-letter provision for cases where the Hangul spelling has to stay recoverable. ISO 9 for Cyrillic is a pure transliteration. ELOT 743 for Greek writes most letters one for one but lets a few letter pairs change with the sound that follows, as the Greek section below shows.

The main standards for each script

A standard is only useful when you know who issued it and what it is for. The table lists every system the converter supports, with the body that published it, the year it dates from, and whether the Latin text can be turned back into the original script without loss. Where a single year cannot be given, the table says so instead of guessing.

Reversible has a strict meaning here: one Latin form for each source character, so a program can rebuild the original exactly. That matters for catalogues, databases and anything else that must survive a round trip. The other systems give up that property in exchange for spellings that are easier to read and say aloud, which is why they are the ones you meet on signs, in passports and in place names.

Greek: ELOT 743 and ISO 843

Greek has no page of its own on this site, so this section is its reference. ELOT 743 is the Greek national standard for writing Greek in Latin characters, issued by the Hellenic Organization for Standardization, and it corresponds to the international standard ISO 843. Romanize applies it to modern monotonic Greek: accents are dropped, η and ι are both written i, υ is written y, and θ, χ and ψ become th, ch and ps.

Most letters map one to one. The exceptions are pairs whose sound depends on what follows them. αυ and ευ are written av and ev before a vowel or a voiced consonant, and af and ef before a voiceless consonant or at the end of a word. ντ and γκ are written nt and gk in every position, while μπ is written b at the start and at the end of a word and mp inside one, and γ before γ, ξ or χ becomes n. A diaeresis keeps a vowel apart from the one before it, so αϋ is a followed by y rather than a pair.

Worked Greek examples

Each word below was converted by the same engine as the tool above, and each one is also a case in its test suite. The last column lists the rules the engine reported.

Capitals follow the source. A capital at the start of a word gives a capital followed by lowercase letters, so Θεσσαλονίκη becomes Thessaloniki, while a word written entirely in capitals stays in capitals, as ΧΑΝΙΑ ΚΡΗΤΗΣ becomes CHANIA KRITIS. Polytonic text, with breathings and the iota subscript, is reduced to the monotonic transcription, and the converter adds a note whenever it does so.

Why one name can have several spellings

A name written in Hangul, kana or Cyrillic has one spelling in its own script, but it can have many in Latin letters. Each standard is consistent with itself; the standards simply make different choices. Some aim at readers of English, some at a one-for-one record of the letters, some at a single language community, and some at machine-readable travel documents that use plain Latin letters only.

Hangul and kana have no capital letters, so the converter writes them in lowercase unless names mode is on. Every spelling in this table is checked by the converter's test suite.

Three more things add to the variety. Older documents may use a system that has since been replaced, so two papers for the same person can disagree. Customary spellings live on beside the official ones: the Revised Romanization of the family name 김 is Gim, yet Kim is the customary spelling, so in names mode the converter writes Kim and shows Gim in its explanation. And once a spelling is printed in a passport, keeping it consistent usually matters more than following whatever the current rules say.

If you need a spelling for an official form, copy the one on your existing documents. If there is none yet, use the standard that the issuing authority names; the table of standards above shows which body issued which system.

What Romanize does not do

A converter is only trustworthy when it is clear about where it stops. These are the limits, said plainly.

• It does not translate. 한국어 becomes hangugeo, not the Korean language. The meaning of a word is outside the job of a romanizer.
• It does not read kanji. Japanese text in kanji is left unchanged with a note, because the reading of a kanji depends on the word it is in. Write the word in hiragana or katakana and the converter romanizes it.
• It does not support Chinese, Arabic or Thai yet. Text in those scripts, and in Hebrew or Devanagari, passes through untouched with a message that says so, instead of a guess.
• It does not apply every Korean sound change. A few changes depend on how a compound word is formed or on the word class, such as the extra n sound in some compounds. The converter writes the regular form and adds a note when a word may be one of them.
• It does not recover what a spelling has lost. Only the letter-by-letter systems, ISO 9 for Cyrillic and the transliteration provision of the Revised Romanization, are designed to convert back exactly. A spelling based on pronunciation cannot be turned back into the original script with certainty.

Converters for each script

The tool on this page handles every supported script at once. Each script also has its own page with the full letter tables, the rules worked through with examples, and a comparison of its standards.

Frequently asked questions

Sources

The rules the converter applies, and the facts on this page, come from these documents and conventions. The converter's tests check its output against examples taken from them.

• Revised Romanization of Korean: Ministry of Culture and Tourism Notice No. 2000-8 (2000).
• Revised Romanization, letter-by-letter transliteration: Ministry of Culture and Tourism Notice No. 2000-8 (2000), transliteration provision.
• McCune-Reischauer: George M. McCune and Edwin O. Reischauer (1939).
• Yale romanization of Korean: Samuel E. Martin (1942).
• Modified Hepburn: Kenkyusha Japanese-English dictionary, 3rd edition (1954).
• Traditional Hepburn: Hepburn's dictionary, 3rd edition (1886).
• Passport Hepburn: Ministry of Foreign Affairs of Japan, romanization rules for passport names.
• Kunrei-shiki: Cabinet notification (1954), first issued in 1937; adopted as ISO 3602.
• Nihon-shiki: Tanakadate Aikitsu (1885).
• ISO 9:1995.
• ICAO Doc 9303, 7th edition (2015), Part 3.
• GOST 7.79-2000, System B.
• BGN/PCGN romanization system for Russian (1947).
• Scholarly transliteration, an academic convention in linguistics and Slavic studies.
• Ukrainian national transliteration: Cabinet of Ministers of Ukraine Resolution No. 55 of 27 January 2010.
• Bulgarian Transliteration Act: Transliteration Act, Republic of Bulgaria (2009).
• Serbian Latin alphabet, official alongside Cyrillic.
• ELOT 743, equivalent to ISO 843.