Korean Double Batchim Rules: Why 읽다 Is [익따] but 읽어 Is [일거]

You learned 읽다 (to read) from a flashcard. The card said ik-tta, you drilled it, you own it. Then a friend holds up a paperback and asks 이 책 읽어 봤어? and what reaches your ear is [일거 봐써]. Il-geo. The k you expected has turned into a g and moved to the next syllable, and there's an l you never pronounced. You don't recognize your own vocabulary.

That word ends in a 겹받침 (gyeop-batchim), a double batchim: two different consonants packed into the bottom of one syllable block. Korean has eleven of them (ㄳ ㄵ ㄶ ㄺ ㄻ ㄼ ㄽ ㄾ ㄿ ㅀ ㅄ), and a Korean mouth never pronounces both at the end of a syllable. It keeps one, and which one it keeps depends on the sound that comes next. Don't confuse them with ㄲ and ㅆ, which are 쌍받침 (ssang-batchim), the same letter doubled; those behave like ordinary single finals (Encyclopedia of Korean Culture).

If single batchim still trip you up, start with the seven final sounds and liaison, because double batchim sit on top of them. This post covers the rest: the one question that predicts every double batchim, which letter survives, the ㅎ pair behind 괜찮아, the 밟다 exception, and a drill for turning what you hear back into a word you can look up.

One verb, four endings: what 읽다 sounds like in real sentences

Here is 읽다 in four ordinary forms. The National Institute of Korean Language gives 읽다 as [익따] and 읽고 as [일꼬] (NIKL Q&A 315279), and the other two follow from the same standard rules:

The block 읽 comes out as ik, il, il-g and ing. You can't store it as one sound. What you can store is a question to ask every time you meet a double batchim: what comes right after it?

• A consonant: you hear one of the two letters.
• A vowel that belongs to an ending or particle (-어, -이, -을): you hear both. The second letter slides onto the vowel. (The two ㅎ clusters are the exception, covered below.)
• A separate word starting with a vowel: the cluster shrinks to one sound first, and only that sound slides. 닭 앞에 (in front of the chicken) is [다가페], not [달가페]. That example is printed word for word in rule 15 of the Standard Pronunciation Rules.

Which of the two letters you hear before a consonant

값 (price) is said [갑], and 닭 (chicken) is said [닥]. In the first, the first letter wins. In the second, the second letter wins. The standard splits the eleven clusters into these two camps, plus a pair of ㅎ clusters that get their own section.

First letter survives: ㄳ, ㄵ, ㄼ, ㄽ, ㄾ, ㅄ

몫 [목], 앉다 [안따], 넓다 [널따], 여덟 [여덜], 핥다 [할따], 없다 [업따]. Rule 10 of the standard lists these, and its commentary says this simplification holds regardless of region, social group or word, with one troublemaker: ㄼ.

Look at 앉다 again. It's [안따], with a tense ㄸ. When a verb stem ends in ㄵ or ㄻ, the next consonant tenses (rule 24), even though the ㅈ itself goes silent, so [안다] is wrong. Stems ending in ㄼ and ㄾ tense the next consonant too: 넓게 [널께], 핥다 [할따]. Nouns don't. 여덟도 (eight, too) is [여덜도], with a plain ㄷ.

Second letter survives: ㄺ, ㄻ, ㄿ

닭 [닥], 흙 [흑], 삶 [삼], 젊다 [점따], 읊다 [읍따]. Only three clusters keep the second letter, so memorize ㄺ, ㄻ and ㄿ and default everything else to its first letter. Don't try to sort them by whether they start with ㄹ. ㄼ, ㄽ, ㄾ and ㅀ start with ㄹ too, and they keep it.

The exceptions that break the split

By the ㄼ rule, 밟다 (to step on) should be [발따], like 넓다 [널따]. It isn't. Rule 10 makes 밟- keep the ㅂ before any consonant: 밟다 [밥따], 밟고 [밥꼬], 밟는 [밤는].

넓다 has its own small leak: in a couple of compounds, 넓죽하다 [넙쭈카다] and 넓둥글다 [넙뚱글다], it flips to [넙]. You'll also hear [넙따] for plain 넓다 from some speakers. Sujin Oh and Harim Kwon's 2025 study of this cluster describes ㄼ as the most variably pronounced cluster in Korean. The standard is still [널따].

ㄺ before ㄱ is the other exception. When a verb or adjective stem ends in ㄺ and the next syllable starts with ㄱ, the ㄹ survives instead of the ㄱ: 읽고 [일꼬], 맑게 [말께]. Nouns don't get this treatment. 닭고기 (chicken meat) is [닥꼬기], and 흙과 (soil and) is [흑꽈].

Before a vowel, both letters come back

없어 (there isn't) is [업써]. That second syllable is the ㅅ from the bottom of 없, moved over onto 어 and tensed. This is rule 14: when an ending or particle starting with a vowel follows, the first letter stays put and the second slides across. The same rule gives you:

• 앉아 [안자] (sit)
• 닭을 [달글] (the chicken, object)
• 젊어 [절머] (young)
• 값이 [갑씨] (the price, subject)

A sliding ㅅ always comes out as ㅆ. So 돈 없어 (I've got no money) is [돈 업써], and you'll hear it in every drama where somebody is broke.

That's why 읽다 and 읽어 feel like different words. In 읽다, the ㄹ disappears before the ㄷ. In 읽어, it's back, and the ㄱ has moved into the next syllable. Your ear is searching for ik and gets il-geo.

In casual speech you'll hear a third version. The National Institute itself notes that 닭이 is often said [다기] and 흙이 [흐기] in real conversation, with the whole cluster flattened before the particle; it just doesn't accept those as standard (NIKL Q&A 318120). If you hear [다기] at a fried-chicken place, the speaker hasn't made a mistake you need to fix. Say [달기] yourself and understand both.

Why 괜찮아 sounds like [괜차나]: the ㅎ clusters ㄶ and ㅀ

괜찮아 is [괜차나]. 많이 is [마니]. 싫어 is [시러]. These three are probably the double batchim you hear most in K-dramas, and none of them sounds the way it's spelled, because the ㅎ half of the cluster does something different depending on what follows (rule 12).

Before a vowel, the ㅎ vanishes and the ㄴ or ㄹ slides over:

Before ㄱ, ㄷ or ㅈ, the ㅎ doesn't vanish. It merges with that consonant and makes it aspirated: 많다 [만타], 않던 [안턴], 싫고 [실코], 닳지 [달치]. Before ㅅ it tenses the ㅅ instead: 많소 [만쏘]. Before ㄴ it's silent again, 않는 [안는], and in ㅀ the leftover ㄹ then turns the ㄴ into a second ㄹ: 뚫는 (piercing) is [뚤른].

So one adjective gives you 많다 [만타] and 많아 [마나], with a puff of air in one and nothing at all in the other. If they sounded like two different words to you, that's the ㅎ. And once you can hear 괜찮아요 as [괜차나요], the next problem is working out which of its five meanings you got.

A listening drill that runs the rule backwards

[업써요] is 없어요, which is 없다. Hear the sound, work out which written cluster could have produced it, and land on the dictionary form. Four more:

• [일거] → 읽어 → 읽다
• [안자] → 앉아 → 앉다
• [시러] → 싫어 → 싫다
• [밥찌] → 밟지 → 밟다

Then say pairs out loud, the consonant form and the vowel form back to back: 읽다 [익따] / 읽어 [일거], 앉다 [안따] / 앉아 [안자], 없다 [업따] / 없어 [업써], 많다 [만타] / 많아 [마나]. The goal is to stop hearing them as two words.

One honest limit. This drill only works for verbs you already know. If [일거] is a word you've never met, no rule will reverse-engineer it for you, and you'll need the subtitles like everyone else. What the drill fixes is the more frustrating case: a word you studied, walking past you in a sentence, unrecognized.

Double batchim rules at a glance

Dictionaries mark long vowels in some of these (없다 appears as [업ː따]). The standard's own commentary on rule 1 admits that length often isn't kept apart in real speech, so don't lose sleep over it.

The next time [일거] goes past you in a drama, you won't be hunting for a word that starts il. The ㄹ came back because a vowel followed it, and the word you're looking for is 읽다, the one on your flashcard.