Say cot. Now say caught. If the two came out identical, one of two things is true: you just made a pronunciation mistake, or you sound like roughly half of the United States. With this particular pair, there is no way to tell which from the sound alone.
That makes this a strange article to write. The usual minimal-pair piece shows you two vowels, proves they matter with a table of word pairs, and sends you off to drill until you can keep them apart. For /ɑ/ and /ɔ/, the honest version is different, because tens of millions of Americans, including plenty of news anchors, never make the distinction at all. Linguists call it the cot–caught merger, and it is one of the largest sound changes underway in North American English right now. So before you spend a month of practice on this contrast, it is worth asking what the natives are doing. The answer changes what you should do with your mouth.
Dictionaries list two vowels here: /ɑ/ in cot, stock, Don (jaw open, lips relaxed) and /ɔ/ in caught, stalk, dawn (tongue a little higher, lips slightly rounded). Roughly half of American speakers, and most younger ones, use one vowel for both, and the merged pronunciation is completely standard. So unless you are matching a specific unmerged accent like New York’s or Chicago’s, merge on purpose. Put the practice time into the two things that still carry meaning: opening the jaw on the merged vowel so hot doesn’t sound British, and keeping the whole word class away from /oʊ/, because caught said as coat is a different word.
Two vowels on paper
Open any pronunciation dictionary and the two vowels are right there. The FATHER vowel /ɑ/ is the most open sound in American English: the jaw drops, the tongue lies low and back, and the lips do nothing at all. It is the vowel a doctor asks for (“say ah”), and it carries cot, stock, hot, job, clock, father. The SAW vowel /ɔ/ sits a step away. The tongue rises slightly from /ɑ/, the lips round just a little, and for many speakers the vowel runs a touch longer. On paper it carries caught, stalk, dawn, law, talk, coffee, thought.
A step apart in the mouth, and English has stacked word pairs across the gap:
| /ɑ/ — jaw open | /ɔ/ — slight rounding |
|---|---|
| cot | caught |
| stock | stalk |
| Don | dawn |
| tot | taught |
| knotty | naughty |
| chock | chalk |
| odd | awed |
| wok | walk |
| collar | caller |
Here is what makes this pair unlike every other one on this blog: read those columns to a friend from Seattle, Denver, Boston, or anywhere in Canada, and they will hear you read the same list twice. The gap that table is built on does not exist in their speech. Read it to a friend from Chicago or New York and the two columns are plainly different words. Both friends speak ordinary, standard American English.
The merger map
The technical name is the low back merger: the two vowels at the low back corner of the mouth collapsing into one. What the single vowel sounds like varies a little by region (open, unrounded /ɑ/ across the West; a slightly rounded cousin of it in Canada, Boston, and Pittsburgh), but the system is the same everywhere: for merged speakers, caught simply joins cot. The unrounded version is the one this article teaches: it is the pattern of the West and of General American, and it reads as native everywhere, including the regions that round.
The geography is well mapped. The Atlas of North American English, the big telephone-survey dialect atlas published in 2006, drew the merged zone across nearly all of the American West, Canada, eastern New England around Boston, and western Pennsylvania around Pittsburgh. The distinction held in New York City, Philadelphia, Baltimore, the Inland North cities along the Great Lakes (Chicago, Detroit, Cleveland, Buffalo), and most of the South.
That was the snapshot. The motion since has gone one way. Younger speakers merge at higher rates than their parents in nearly every region surveyed, including inside the historically unmerged zones. Put together, merged speakers come to somewhere around half the country and climbing. This is also why you cannot settle the question by listening to broadcasters: network anchors come from both sides of the map, and nobody in the business considers either side an accent problem.
For a learner, one fact matters more than all the cartography: no American hears a merged speaker and thinks “foreign.” They think Denver, or they think nothing at all. You cannot say that about most pronunciation shortcuts.
So which target do you pick?
Merge, on purpose, unless you have a concrete reason not to.
The case is lopsided. A merged target means one less vowel category to build and zero risk: every caught-class word you say with /ɑ/ lands inside normal American speech. An unmerged target means learning which of several thousand words belongs to which class (the spelling helps, but only mostly), and the reward for all that bookkeeping is sounding like one set of American cities instead of another. Nobody is paying you for that distinction. A half-learned distinction is also worse than none: a speaker who rounds the wrong words sounds off in a way a consistently merged speaker never does.
There are concrete reasons to learn the contrast anyway. Your life is in New York, Philadelphia, Chicago, or the South and you want to match the people around you. Or you act, and the script says Brooklyn. Otherwise, take the simpler target and spend the saved effort where it pays.
One asterisk on “ignore it”: your ear still has to know the distinction exists. A New Yorker’s coffee opens with a long, rounded aw (caw-fee), and their talk and thought carry the same vowel. If your ear has filed all of those under /ɑ/, the words still resolve from context, but speech from the unmerged cities will read as harder to follow than it needs to be. Spend one listening session with a New York or Chicago speaker, notice the rounded vowel, and file it as “same word, different paint.” That is the whole perception assignment.
Jaw open, lips out of it
The merged vowel is simple to describe and easy to underdo. Drop the jaw further than feels reasonable; two fingers should fit between your teeth. Let the tongue lie low and flat, pulled slightly back. And take your lips out of the game entirely: no rounding, no tension. The result is the ah of father, and in the merged target it serves cot and caught, stock and stalk, Don and dawn alike.
Nearly everything that goes wrong here is one of two errors.
The first is rounding. If your first language writes the letter o and says a rounded vowel, your mouth will try to do the same in hot, job, stop. A short, rounded hot (the British /ɒ/) is exactly how British English says the word, so to most American ears the rounding reads as British at best. The fix is mechanical. Stand at a mirror, say hot, and watch your lips: if they pucker even slightly, open the jaw, flatten them, and listen to how much more American the word turns.
The second is not opening enough. A lazy, half-open /ɑ/ drifts toward /ʌ/, the vowel of cup and but, and now cop sounds like cup and stock like stuck. That pair carries meaning constantly. When in doubt, the jaw goes down.
If you do want the unmerged /ɔ/, build it from the /ɑ/ you just made: raise the tongue slightly, add a small amount of lip rounding (still less than for an oh), and let the vowel run a beat longer. That is the generic unmerged target. New York and Philadelphia push the same vowel much further, with heavier rounding and extra length (the caw-fee vowel from earlier), so if those cities are your model, round harder than the generic version. The common miss everywhere else is over-rounding it into a full /o/, which is a different problem, and the subject of the next section.
One boundary worth marking before you drill: none of this applies before R. The vowel in for, north, born is /ɔr/, a separate vowel that stays rounded for every American, merged or not. The merger never touches it. If you generalize the open ah into for, you get far, and that mix-up does get noticed. It is the card vs cord confusion, and it is a real one.
The trap that does matter
Here is the contrast worth your practice time, and it sits one vowel over from the pair in the title.
The border that needs guarding is not /ɑ/ versus /ɔ/. It is both of them against /oʊ/, the vowel of coat: say caught with a pure rounded o and you have said a different word.
Most languages with a written o have one rounded vowel behind it, a clean, pure /o/. That vowel is much closer to the English GO diphthong /oʊ/ than to either vowel in this article. So when a learner reaches for caught or law, the vowel that comes out is often the native /o/, and an American ear files it under the nearest English category: /oʊ/. The sentence then contains a different word. Caught becomes coat and law becomes low. And unlike the cot–caught distinction, which half the country has discarded, this border is load-bearing for everyone. No American merges bought and boat, or walk and woke.
The mouth tells you which side of it you are on. The caught-class vowel, merged or not, lives with the jaw open; /oʊ/ starts half-closed. And /oʊ/ moves — it is a diphthong, gliding toward oo with the lips tightening as it goes — while /ɑ/ and /ɔ/ hold still. If your law ends with your lips in a small circle, it became low. Our low vs law comparison page puts the two side by side with audio.
The spelling is more cooperative than usual, which helps:
| Spelling | Word class | Examples |
|---|---|---|
| o + consonant | /ɑ/, the open ah (carve-out below) | hot, stop, job, clock, stock |
| au, aw | caught-class | cause, law, dawn, sauce, saw |
| alk | caught-class, and the L is silent | talk, walk, chalk, stalk |
| ought, aught | caught-class | thought, bought, caught, taught |
| o + consonant + e, oa, ow | /oʊ/, the one that glides | note, coat, boat, low, show |
A few notes on that table. The caught-class rows take /ɔ/ if you keep the distinction and the open /ɑ/ if you merge; either way they never take /oʊ/. Words in -all (all, tall, call, ball) sit in the caught class too, and unlike talk they keep their L. The carve-out on the first row, for anyone keeping the distinction: o before g, ng, s, th, and f usually crosses to the rounded side, so dog, long, boss, cloth, and off are caught-class words in unmerged American English generally, despite their o-plus-consonant spelling. Merged speakers can ignore that sentence. Treat the last row as a tendency rather than a law (ow also spells the vowel of now). And the aw spelling is a standing temptation to glide: saw is not soh, and for Mandarin speakers aw is also not the ow of pinyin ao (the diphthong of English now). One open vowel, no movement, no lip circle at the end.
Practice phrases
Read these out loud, twice each, with the merged target: every bolded word gets the same open, unrounded vowel, jaw dropped, lips loose. The italicized words are the /oʊ/ neighbors; they glide and close while everything bolded stays open, and those lines are the ones to slow down on. Line 4 runs the names Don and Dawn back to back, and with the merged target they come out identical. That is the point.
- I caught a cold last fall. I caught a cold last fall.
- The dog walked across the lawn. The dog walked across the lawn.
- He bought a boat, not a bottle. He bought a boat, not a bottle.
- Don gets up at dawn. Don gets up at dawn.
- Stop talking and walk. Stop talking and walk.
- She thought the stock talk ran long. She thought the stock talk ran long.
- The law office is low on coffee. The law office is low on coffee.
- I lost my watch in the parking lot. I lost my watch in the parking lot.
- Chalk it up to a long walk. Chalk it up to a long walk.
- It's not wrong, it's just odd. It's not wrong, it's just odd.
If the low in line 7 came out matching everything around it, that is the border failing: low has to glide and close while law stays open. Park on that line until the two words feel like different mouths.
How different first languages handle this
Where you start depends on what your first language put behind the letter o, and behind open vowels generally. None of it is a deficiency; it is just the gap you happen to be closing.
| Your L1 | Where the o-words land now | What to focus on |
|---|---|---|
| Spanish, Greek | One rounded /o/ and an open central /a/; spelling sends hot and caught to the rounded one | Send every o-plus-consonant and aw word to the open ah instead. Save the rounded vowel for coat-class words, and let it glide there. |
| Japanese | あ /a/ is open and central, a usable anchor; お /o/ is pure | Your ah is nearly ready. The work is spelling: au/aw/al/ough words take the open vowel, never お. |
| Korean | aw words tend to land on ㅗ /o/; ㅓ /ʌ/, the nearest native vowel, sits a step too high | Start from ㅓ and drop the jaw further, lips flat, for the whole cot and caught set. ㅗ belongs only to coat-class words, with a glide added. |
| Mandarin Chinese | Pinyin o is rounded; ao /aʊ/ tempts on aw spellings | Saw and law hold one open vowel with no glide and no final lip circle. Open the jaw past your a in ma and keep it still. |
| Hindi, Urdu | आ /ɑː/ matches the merged target closely; the o-class often takes a rounded short vowel | American hot uses your आ, not your o. If you keep Indian English’s two-way contrast, fine, but don’t round the cot class. |
| Arabic | Long ā /aː/ is close to the target; the dialects’ /oː/ (often from an old aw) is the trap | Anchor both word classes on your long ā with the jaw fully open. Don’t let English aw spellings pull in your dialect’s /oː/. |
| Russian | Stressed o is rounded /o/; unstressed o already relaxes toward /a/ | The relaxed vowel is the right instinct; extend it to stressed syllables. Hot and caught take the open unrounded vowel even under stress. |
| French, Italian, Portuguese | /a/, /ɔ/, and /o/ all exist; spelling pulls toward /o/ | You own all the raw material. If you merge, drop the rounding entirely on both classes. If you don’t, your /ɔ/ is close; just keep it off the pure /o/. |
| German | Short o /ɔ/ exists; long /aː/ exists | German short o in hot reads as British. Use your long ah quality instead, jaw fully open, for the whole merged set. |
FAQ
No — roughly half do. The merger covers the West, Canada, eastern New England, western Pennsylvania, and most younger speakers nationwide, while New York, Philadelphia, Baltimore, the Great Lakes cities, and much of the South keep the two vowels apart. Both pronunciations are standard, and Americans generally don’t notice which one a speaker uses.
Not in American English. Merging them is the everyday pronunciation of about half the country, including broadcasters, and it is the simpler target for learners: one open vowel, jaw dropped, lips unrounded, covering both word classes. No listener will register it as a mistake.
New York City, Philadelphia, and Baltimore keep the contrast, with a strongly rounded, often lengthened /ɔ/. So do the Inland North cities along the Great Lakes (Chicago, Detroit, Cleveland, Buffalo) and much of the South, especially among older speakers. Younger speakers in all of these regions are merging at rising rates.
Almost never; context separates the pairs the same way it does for merged Americans. The one genuine collision is the names Don and Dawn, which sound identical for merged speakers. Americans themselves resolve it by spelling (“Dawn with a w”). If that is the worst case, the merger is safe.
For speakers who distinguish them, cot has /ɑ/: jaw fully open, tongue low and back, lips completely relaxed. Caught has /ɔ/: tongue slightly higher, lips lightly rounded, often a touch longer. The two sounds sit a small step apart at the low back corner of the vowel space, which is part of why they merged so easily.
Because the vowel is rounded. British English says hot with a short rounded /ɒ/; American English uses the open, unrounded /ɑ/ (the doctor’s ah), with the jaw dropped and the lips doing nothing. Unround the lips and open the jaw, and the same word lands as American.
Most pronunciation advice asks you to add a distinction. This article is asking you to skip one. Open the jaw, leave the lips out of it, and let cot and caught fall together the way they already have for half the country. The border that deserves the practice time runs one vowel over: caught and coat are different words everywhere in America, and that one the jaw and lips have to get right.
