Coffee comes out as copy. Light and right share a single hangul spelling, 라이트, and often a single English pronunciation. And bus picks up a second beat English never gave it: beo-seu.
If you grew up speaking Korean, none of those three will surprise you, though you may have stopped hearing them in your own voice. They aren’t carelessness. Korean assembles speech in neat syllable blocks with strict rules about what can sit where, runs one liquid consonant where English runs two, and never needed an /f/, a /v/, or a /z/. English keeps requesting all of it. And there’s a second force pressing the same habits in deeper: thousands of English words already live inside Korean as loanwords, respelled by a national orthography that writes the repairs directly into the dictionary. Strike has been 스트라이크, five beats, since before you could read.
The eight habits below are the distinctively Korean ones. The two TH sounds are missing from Korean too, but that problem is shared with most of the planet, and the TH article handles it; this list is about what marks a speaker as Korean specifically, and which of those marks cost the most.
Korean syllables allow no consonant clusters and only seven word-final sounds, none of them a hiss, so English clusters and endings grow filler vowels: strike becomes seu-teu-ra-i-keu. Korean’s one liquid ㄹ taps between vowels and holds as L at syllable ends, so light and right trade places by position. Missing /f/, /v/, /z/ get patched with ㅍ, ㅂ, ㅈ: coffee leans toward copy, vote toward boat, zip toward jip. And two vowel mergers, ㅐ/ㅔ and the single Korean i and u, blur bed/bad, seat/sit, pool/pull. Delete the filler vowels first; they cost the most. Two assets you already own: the puff on P, T, K, and the cleanly closed word ending.
Why Korean makes American English hard
Four structural facts drive nearly everything on this list.
Korean syllables are built in blocks, and the blocks have rules. A syllable can open with at most one consonant, and it can close with only seven sounds: the closures k, t, p, the nasals m, n, ng, and l. Nothing else gets a seat. No s, no ch, no j, no f, and no consonant ever stands next to another consonant inside the block. English routinely demands both: street opens with three consonants in a row, and buzz ends on a sound Korean has no slot for. When English hands a Korean mouth a shape the block can’t hold, the mouth adds a small vowel to give the extra consonant a block of its own.
The seven legal endings never release. A final k, t, or p in Korean closes and stays closed; 밥 ends with the lips sealed. This one is a gift. American English finals work the same way, and learners from most other languages have to be taught to stop releasing them.
One letter covers both of English’s liquids. ㄹ is a quick tongue-tap between vowels and an L-like hold at the end of a syllable. Both sounds exist in every Korean speaker’s mouth; what doesn’t exist is the right to choose freely. Position decides, and English ignores the position rules.
Two vowel distinctions English leans on aren’t there. ㅐ and ㅔ were once different vowels; for most younger speakers they have collapsed into one, to the point that 개 (dog) and 게 (crab) are homophones in young Seoul speech. And where English splits seat/sit into a tense and a lax vowel, and pool/pull into another such pair, Korean covers each territory with a single vowel. Bed/bad, seat/sit, and pool/pull all land on this.
Konglish hardens all of it into vocabulary; that part of the story gets its own section after the list.
Group A: Four consonant swaps
1. L and R trade places by position
Light and right merge. So do collect and correct, lane and rain. But the Korean version of this famous confusion has its own shape: both English sounds already live in your mouth.
Say 노래 and the ㄹ in the middle is a quick tap of the tongue tip against the ridge behind your teeth; that tap is a close cousin of the flap-T Americans use in water, a sound other learners drill for weeks. Say 달 and the final ㄹ holds against that same ridge like an English L. Korean grants the letter both pronunciations but assigns them by address: tap between vowels, hold at syllable ends. English hands out the same two sounds with no address rules at all, which means an L between vowels (belly) tends to get tapped into something R-like, a word-initial L gets the same tap (lane and rain both land on 레인), and the borrowed spelling system files both English letters under ㄹ, so light and right come out of the dictionary as the same word, 라이트.
The two mouths are easy to separate. For an L, the tongue tip presses the ridge behind the top teeth and stays while the sound flows past its sides. For an American R, the tip touches nothing; the tongue bunches back and the lips round a little. The hard part is the ear, after a lifetime of one category, and ear training has to come first. The L vs R article covers both positions and the listening work.
Drill: say belly slowly and feel whether the tongue tip stays planted on the ridge through the L. If it bounced off, that was ㄹ doing its between-vowels routine. Then berry, tip parked low, touching nothing.
2. F borrows P’s mouth
Coffee drifts toward copy. Fan and pan both come out as pan; Korean spells them with the same letter, and the frying pan in every Korean kitchen is a peu-ra-i-paen. Korea’s favorite word of encouragement, fighting!, is 파이팅.
Korean has no /f/, and the nearest native sound is ㅍ, a P with a strong puff of air. The two are neighbors but use different plumbing: a P seals both lips and pops; an F rests the top teeth on the bottom lip and pushes air through the gap continuously, no seal, no pop. Because the puff on ㅍ is so healthy, the swap is loud and easy for American ears to catch, and F is everywhere in English: first, before, office, afford.
Drill: set your top teeth on your bottom lip and push air until the lip tickles, then hold that friction for a slow beat before the vowel: fffan. If your lips ever touch each other, it became a P.
3. V collapses into B
Video is 비디오. Virus is 바이러스. Vote lands as boat, vest as best.
The story is item 2 with the voice switched on. Korean has no /v/, the patch is ㅂ, and the fix is the same teeth-on-lip posture as F, now with the vocal cords buzzing. You can borrow the F you just built: hold the friction, turn the voice on, and the buzz that results is a V. English pairs the two constantly (five, over, every, never), so the posture needs to become cheap. The V vs W article takes the sound apart in more detail.
Drill: alternate boat, vote, boat, vote with a finger resting on your bottom lip. On vote you should feel the teeth arrive and a full second of buzz before the vowel.
4. Z lands on J
Pizza is 피자. Zero is 제로. Zipper is 지퍼. Say zoo through the Konglish filter and an American hears Joo.
Korean has no /z/. The patch the orthography chose is ㅈ, which is a stop-start consonant: the tongue touches, blocks the air for an instant, then releases into a hiss. English /z/ never touches. It’s an S with the voice on, a continuous buzz with the tongue tip floating close to the ridge but landing nowhere. The swap is worth singling out because it points at Korean specifically: Japanese has a native /z/, and Mandarin speakers trade the buzz for a voiceless ts or s hiss. A J-for-Z is one of the faster ways an experienced listener places an accent as Korean.
Drill: stretch a long ssssss, then switch the voice on mid-stream without moving your tongue: ssszzzz. The moment you feel the tongue tap the ridge, it turned into ㅈ. Then run zip, zone, busy, music, lazy.
Group B: The one vowel Korean inserts everywhere
5. Clusters get a block of their own
Strike, one syllable in English, runs five beats in Konglish: seu-teu-ra-i-keu. Christmas runs five: keu-ri-seu-ma-seu. Program runs four: peu-ro-geu-raem.
A Korean syllable block holds one opening consonant, so a cluster like str has nowhere to live. The repair issues each homeless consonant its own block, and the vowel that fills the empty slot is almost always the same one: ㅡ, the flat eu, lips relaxed, tongue high and back. Unlike Japanese, which rotates its filler vowels, Korean runs nearly the whole repair shop on this single vowel, so once you learn to hear it, you can hear yourself doing it.
The cost is recognition, not politeness. English listeners retrieve words by syllable count and stress shape before they check individual consonants, and a one-syllable word arriving as three is a worse mismatch than a wrong consonant. This is the habit on the list that most directly decides whether you’re understood.
Drill: build the cluster silently. Hiss the s, let it crash into the t, slide into the r, no voice anywhere, then switch the voice on only at the vowel: ssstr-ike. Any voicing before it means a ㅡ slipped in.
6. Word endings grow a tail
Bus is 버스, two beats. Switch is 스위치. Juice is 주스. Cake is 케이크, three beats for a one-syllable word.
Same repair, other end of the word. A hiss, a buzz, or a ch/j sound can’t close a Korean block; there’s no slot for it, so it gets a block of its own and a ㅡ (or after ch and j, an i) to sit on. Stops are subtler. Korean codas hold k, t, p perfectly well, and 캡 (cap) ends as cleanly as English could ask. But the loanword layer often adds the tail anyway (케이크, 테이프), and careful speech releases word-final stops with a small echo: make-uh, like-uh. That echo is the tell. American finals close and stay closed, which Korean already does natively; the unlearning here is trusting the closed ending instead of dressing it up.
Drill: end bus on the hiss and stretch the hiss: busssss. Then end make with the tongue sealed at the K and hold the silence for a beat. Anything voiced after the seal was a tail vowel.
Group C: Three vowel pairs that merge
7. Bed and bad share a vowel
Bed and bad, pen and pan, men and man: each pair collapses into one word somewhere between the two.
Hangul still keeps the two letters apart, and the loanword dictionary dutifully writes bed as 베드 and bad as 배드. But spoken Seoul Korean merged ㅐ and ㅔ within living memory, so the written distinction no longer points at two different mouths. English keeps the pair apart mostly with the jaw: /ɛ/ (bed) sits mid-height, and /æ/ (bad) drops the jaw visibly lower and spreads the lips. American ears treat the two as entirely different vowels and use them to separate everyday words, so the merger costs real vocabulary. The TRAP vowel article is the full treatment of /æ/.
Drill: say bed, then drop your jaw about a finger’s width without changing anything else, and say bad. Alternate pen/pan, men/man in front of a mirror; the two words should look different, not just sound different.
8. Seat, sit, pool, and pull collapse into the tense halves
Seat and sit land on the same vowel. So do pool and pull; Konglish writes the swimming kind as 풀, and pull would come out the same.
Korean covers the seat/sit territory with one vowel, ㅣ, and the pool/pull territory with another, ㅜ. English splits each territory into a tense vowel (seat /iː/, pool /uː/) and a lax one (sit /ɪ/, pull /ʊ/), and the lax halves don’t exist in Korean, so both English vowels in each pair get filed under the tense Korean one. The difference isn’t length on a stopwatch; it’s muscle. The tense vowels hold a tight smile or firm lip rounding; the lax ones let the whole face go slack and the tongue drop slightly. Sit said with a tight smile reads as seat no matter how short you clip it. The ship vs sheep article walks the front pair through in detail.
Drill: say seat and hold the smile. Now let the smile die, let the tongue sag a touch, and say sit. Check in a mirror: if sit is smiling, it’s still seat.
The Konglish filter
The reason these habits survive years of English classes is that, for you, most of them aren’t pronunciation at all. They’re spelling and vocabulary.
Konglish words are real Korean words. 커피, 버스, 컴퓨터, 아이스크림: learned in childhood, retrieved automatically, pronounced flawlessly, in Korean. The national loanword orthography standardizes exactly how an English word gets repaired into legal Korean syllables, which means the filler vowels and consonant swaps on this list aren’t your habits alone; they’re printed in dictionaries and on every storefront. When you reach for an English word mid-sentence, the Korean twin often arrives first, and it arrives with its repairs pre-installed. A few of the twins aren’t even English anymore: a 핸드폰 (“hand phone”) is a cell phone, a 노트북 (“notebook”) is a laptop, and 서비스 (“service”) can mean something a restaurant gives you for free.
The unlearning move is the same one that works for katakana English in Japan: stop treating the twin as a pronunciation guide. 스트라이크 and strike are two words in two languages that happen to share an ancestor. For any English word you first met in hangul, relearn it by ear, syllable count first, before the Konglish version can volunteer. The perception-before-production article makes the longer argument for why the ear has to go first.
What an L1 detector would tell you
Software trained to spot Korean-accented English would flag, in roughly this order: the filler vowels at clusters and word endings, because they touch nearly every long word; the F and Z swaps, because those sounds run all through everyday English (the S in dogs and runs is a /z/); the merged vowel pairs; and the positional L/R trades. It would also notice an even, syllable-by-syllable rhythm where English expects one heavy stress per word, a trait Korean shares with several of its neighbors; the word stress and rhythm articles cover that layer.
That order is close to the right practice order, and the economics are unusually good. Deleting the ㅡ asks for no new sound at all, and it repairs the syllable count of hundreds of everyday words at once. F and V share one posture, teeth on lip, with the voice off and then on; Z is item 4’s buzz, a few days of drilling. The vowel pairs are jaw and face work. L/R is the slow one, an ear project measured in months; start the minimal-pair listening now and let it run in the background.
And two expensive features of American English came free with Korean. The puff that separates pie from buy in American ears is the same breath you’ve put on ㅍ, ㅌ, and ㅋ since childhood; the aspiration article shows where English wants it and where it switches off. And your unreleased word-final stops are exactly what American finals want; most learners have to be talked out of releasing them.
FAQ
Korean has one liquid consonant, ㄹ, with two pronunciations assigned by position: a tap between vowels and an L-like hold at syllable ends. Both English sounds exist in a Korean speaker’s mouth, but the choice between them was never free, so English words that put L between vowels or R at an edge trigger the wrong variant, and loanword spelling files both letters under ㄹ. Retraining is mostly ear work: minimal pairs like light/right and collect/correct, then the two tongue positions (held contact for L, no contact for R).
Korean has no /f/ sound, and its loanword system replaces F with ㅍ, an aspirated P. So coffee leans toward copy and fan toward pan. The fix is mechanical: rest the top teeth on the bottom lip and push air through continuously instead of sealing both lips and popping. The same posture with voicing added produces the English /v/ that Korean also lacks.
A Korean syllable allows no consonant clusters and only seven final sounds, so English consonants that don’t fit get a syllable of their own, filled with the neutral vowel ㅡ (eu). Strike becomes seu-teu-ra-i-keu and bus becomes beo-seu; every eu in those romanizations is a vowel English never asked for. The repair is standardized in Korean loanword spelling, which is why it feels like vocabulary rather than error. Deleting these vowels is the highest-leverage pronunciation fix available to Korean speakers, because English listeners rely on syllable count to recognize words.
The two lists overlap but don’t match. Both languages insert vowels into clusters, merge L and R onto a native liquid, lack a native /v/ and both TH sounds, and carry an even rhythm into English. They diverge on the details: Japanese has a native /z/ where Korean substitutes a J-like ㅈ, Japanese rotates several filler vowels where Korean leans on ㅡ, and Korean’s aspirated consonant row means the puff on P, T, K comes free, which Japanese speakers have to scale up. A practiced ear can usually tell the two accents apart within a sentence.
Konglish is a normal part of Korean vocabulary, and inside Korean it isn’t an error at all. It interferes with English in one specific way: Konglish forms are memorized, automatic, and standardized in spelling, so they surface faster than a from-scratch English pronunciation and carry their repairs with them. You don’t need to avoid Konglish. Stop using it as a pronunciation guide, and relearn high-frequency English words by ear, with their real syllable counts.
The inserted ㅡ vowels, at clusters and at word endings. The fix is a deletion, so it needs no new sound, and it restores the syllable count English listeners use to retrieve words, which makes it the largest intelligibility gain per hour of practice. F and Z come second; they’re frequent, and each is one mouth posture away. The L/R contrast matters but retrains slowly through ear work, so run it in parallel rather than first.
On this list, the one slow project is the ear work of L and R; nearly everything else is quick mechanics. So here’s a self-test that costs five minutes: record yourself saying coffee, copy, fan, pan, zest, jest, seat, sit in shuffled order, wait a day, and label each word blind. The pairs you can’t sort in your own voice are the ones American listeners are sorting for you, by context and guesswork, in every conversation.
