Jongno, Not Jongro: The English on Korean Signs Spells the Sound

Stand in front of the sign for one of central Seoul’s main streets and read it the way part one taught you. 종로. ㅈ-ㅗ-ㅇ, then ㄹ-ㅗ. Jong-ro. You are reading correctly. Every letter is where you think it is.

The roman line under the Hangul says Jongno. Whatever you now type into a map is a spelling that Korean signage does not use.

It will happen again, and it feels like sloppiness — as if whoever made the signs could not spell. It is the opposite. The roman letters on a Korean sign are not a transcription of the Korean letters at all. They are a transcription of the Korean sound, and that is not an editorial habit. It is Article 1 of the rule, before anything else is said.

The rule is one sentence long

The regulation currently in force is a Ministry of Culture, Sports and Tourism notice, and Chapter 1 opens by disposing of the whole question:

국어의 로마자 표기는 국어의 표준 발음법에 따라 적는 것을 원칙으로 한다.

Romanization of Korean is, in principle, written according to the Standard Korean Pronunciation Rules.

Ministry of Culture, Sports and Tourism, Romanization of Korean, Chapter 1, Article 1 문화체육관광부 고시 「국어의 로마자 표기법」 제1장 제1항 18 August 2026

Read that once more with a traveller’s eye. It means the roman line on a Korean sign is doing something no other writing system on your trip does for you: it is telling you how to say the word. Not how it is spelled. How to say it.

Which makes it more useful than you thought, and less trustworthy as a guide to the Hangul than you thought — both at once.

The six names the rule is built around

The regulation does not describe sound changes in the abstract. It lists worked examples, and six of them are place names you may well pass through. This table is those six, with the letters spelled out so you can see exactly where your reading and the sign diverge.

On the sign	Letter by letter	Said aloud	Roman on the sign
백마	baek-ma	[뱅마]	Baengma
신문로	sin-mun-ro	[신문노]	Sinmunno
종로	jong-ro	[종노]	Jongno
왕십리	wang-sip-ri	[왕심니]	Wangsimni
별내	byeol-nae	[별래]	Byeollae
신라	sin-ra	[실라]	Silla

The bracketed forms are the regulation’s own, printed beside each example. Nothing in that column is our reconstruction.

Two things are going on, and one of them is tidier than the other.

The tidy one: a hard stop turns nasal when a nasal follows it. ㄱ becomes ng, ㅂ becomes m. 백마 is baek + ma and comes out Baengma. That is the whole of it, and it never surprises you twice.

The untidy one: ㄹ will not sit quietly beside another consonant. After the ng that ends 종 it gives up and becomes n — 종로 → Jongno. Beside a ㄴ, one of the two has to yield, and the roman spelling tells you which: sometimes the ㄴ becomes l and the spelling doubles it (별내 → Byeollae, 신라 → Silla), sometimes the ㄹ becomes n instead (신문로 → Sinmunno). 왕십리 puts both processes in one word — the ㄹ of 리 turns into n, and the ㅂ of 십 then goes nasal in front of that new n, giving Wangsimni.

Look at those last two again. 신라 and 신문로 have ㄴ and ㄹ meeting in that order in both words, and the official spellings go opposite ways. Both are printed in the same short list, three lines apart. We are not going to hand you a rule that predicts which one you are looking at, because the romanization notice does not contain one — it lists results, not reasons, and working out the reason means going into the pronunciation rules and their exception lists, which is a linguistics errand and not a travel one.

The practical version is shorter and it is the whole message of this article: read the roman line. Do not try to derive it.

That is also why the Silla capital at Gyeongju is never written Sinra anywhere in Korea, in any font, on any brown heritage sign. It cannot be. The rule spells what the mouth does, and no Korean mouth says Sinra.

The sound the rule refuses to write

Here is the seam, and it is worth knowing because it will otherwise look like an error.

Korean also tenses consonants — the k in 압구정 is pronounced hard, closer to kk, and this is completely regular. If the rule really followed pronunciation everywhere, the sign would say something like Apkkujeong. It does not. The regulation stops and says so:

된소리되기는 표기에 반영하지 않는다.

Tensification is not reflected in the spelling.

Ministry of Culture, Sports and Tourism, Romanization of Korean, Chapter 3, Article 1, supplementary note 문화체육관광부 고시 「국어의 로마자 표기법」 제3장 제1항 [붙임] 18 August 2026

So the roman line is a partial pronunciation guide. It will move a consonant across a syllable boundary for you, but it will not tell you how hard to hit it. 압구정 is written Apgujeong; 낙동강 is written Nakdonggang; 팔당 is written Paldang — all with soft letters, all said with hard ones. Saying them the soft way will be understood. It is simply not quite what you will hear back.

We flag this because it is the one place where the article’s own headline claim needs qualifying, and we would rather qualify it ourselves than have you discover it at a ticket window.

Two words with the same consonants, spelled two different ways

This one is a small pleasure. The regulation contains both of these examples, a few articles apart.

독립문, the Independence Gate, is romanized Dongnimmun. Letter by letter it is dok-rip-mun; every sound change is applied and shown.

한복남, a personal name, is romanized Han Boknam. Letter by letter it is han-bok-nam, and the same ㄱ-before-ㄴ sequence that turned 독립 into dongnim sits right there in Boknam — untouched. The rule states plainly that sound changes occurring inside a given name are not reflected.

Same consonants, opposite treatment, because one is a monument and one is a person. Names are exempt. So are administrative suffixes: the hyphen in 해운대 → Hae-undae exists purely so you do not read it as Haeun-dae, and the rule instructs that changes across such a hyphen are not written either.

And then Article 7 gives up gracefully: personal names, company names and organisation names may keep whatever spelling they have been using. That single clause is why Korean surnames and corporate names romanize in ways the rest of the system would never produce, and why you should never try to reverse-engineer the Hangul from a company’s English name. The system has a hole in it, on purpose, and the hole is where all the famous words live.

Why the older signs disagree

The current system replaced an earlier one in 2000, and the regulation as published carries a supplementary provision giving signboards — road signs, advertising, cultural-property information boards — until 31 December 2005 to conform, with printed material such as textbooks given until 28 February 2002.

That is a transition, not a purge. Older spellings survive in book titles, in academic writing, in registered business names, in personal passports, and in the places where nobody has repainted anything. If you see two spellings of the same place, this is usually why, and neither one is a mistake in the sense that someone was careless.

There is a companion problem worth knowing about: even where the romanization is right, the information beside it may be years out of date. We checked a batch of official English pages against their Korean originals and the fares did not match.

What this actually buys you on the street

Three things, in descending order of usefulness.

Say the roman line, not the Hangul you sounded out. If you are asking a taxi driver or a station attendant, the roman spelling is closer to what they expect to hear. Wangsimni, not wang-sip-ri.

Search the roman line, not your reading of the letters. The official spelling is the string that Korean signage, station boards and public listings all carry, so it is the one worth typing. Jongno, not Jongro.

Do not try to write the Hangul back from the English. The transformation destroys information. Several different Korean letters arrive at the same roman one once the sound change has run — the n in Jongno and the n in Sinmunno started life as ㄹ, and nothing in the roman spelling says so. Going the other way is guesswork. This is exactly why the shop-suffix words in part four are worth learning in Hangul rather than in transcription.

What we read was the rule, not the street

We read the regulation. We did not survey signs. Everything above comes from the text of the Romanization of Korean notice and the Standard Pronunciation Rules as published by the National Institute of Korean Language, opened on 18 August 2026. Every example word, every bracketed pronunciation and both quoted sentences appear in those documents. We have not photographed or audited any physical sign, station board or map application to confirm it complies, and the transitional dates above are the deadlines the regulation sets, not a finding that the deadlines were met.

“Closer to what they expect to hear” is a judgement, not a measurement. We have not tested comprehension rates for the two pronunciations, and a Korean listener will usually understand either. Nor have we tested any specific map or transit application to see how it handles a misspelled search — our advice to type the official spelling follows from the fact that it is the spelling on the signage, not from a search we ran.

We did not resolve the 신라 / 신문로 split. We can show you that the official spellings go opposite ways from the same pair of letters, because both are printed in the notice. We cannot tell you which words fall on which side, and we are not guessing. If you need to know, the answer is in the Standard Pronunciation Rules rather than in the romanization notice, and it involves a list of exceptions rather than a pattern.

This article covers the sound changes the romanization rule chose to illustrate. It is not a complete account of Korean phonology, and it is not the Standard Pronunciation Rules in full — that document runs to seven chapters, including vowel length, which no sign will ever show you.

We could not find a photograph for this article. The stock we hold includes one Insadong wall sign carrying 인사동 in Hangul with no roman line at all, which would have illustrated the absence rather than the rule, and it is already carrying part one. Everything else we have is scenery. A diagram of the two routes does the job a street photograph could not do anyway: it shows the step in the middle, and the middle step is the entire point.