Listen to any American say the word button. There’s no T in there. Where the T used to be, there’s a tiny catch in the throat, and then the N takes over. Buh’n. The same thing happens in mountain: moun’n. And in certain: sur’n. And in kitten, written, cotton, and forgotten. Half the T’s in the dictionary aren’t pronounced as T at all in American speech.
If you’ve worked on the flap-T, you’ve already met one half of the American T-system, the half where a T turns into a soft, fast tap that sounds like a D. The other half is this: a tiny stop in the throat called the glottal stop T, and it covers most of what the flap doesn’t.
When a T sits before a syllabic N (the -tn ending in words like button, mountain, certain, kitten, written), Americans replace it with a glottal stop, a brief catch in the throat that takes the place of the T (some speakers retain a faint residual tongue contact; the perceptual effect is the same). The technical symbol is /ʔ/. It’s standard pronunciation across General American, and it pairs with the flap-T to cover the two biggest categories of mid-word T: flap when an unstressed vowel follows, glottal stop when a syllabic N follows. (A third pattern, the NT-cluster deletion in winter → winner, lives in the flap-T article.) Knowing which rule fires where is one of the bigger differences between sounding word-by-word correct and sounding like a native.
What the glottal stop is
The glottal stop is the briefest possible consonant. Your vocal cords close, the airflow stops for a hundredth of a second, and then they release. There’s no tongue movement, no lip movement. The sound is happening in your throat.
Most English speakers produce a glottal stop dozens of times a day without ever naming it.
- The catch in the middle of “uh-oh.”
- The brief stop you make at the start of any vowel-initial word when you say it carefully (“an apple”, “an idea”).
- The little hiccup some speakers use to separate words that would otherwise blur together.
In American English it has a specific structural job. When a T sits before a syllabic N, the T disappears and a glottal stop takes its place. The schwa that would normally connect them drops out, and the N becomes the syllable on its own.
Compare these three versions of the same T:
- Crisp British T in button: /ˈbʌt.ən/, two clean syllables, both T and schwa pronounced.
- Flap-T in butter: /ˈbʌɾɚ/, the T turns into a quick tap.
- Glottal stop T in button the American way: /ˈbʌʔn̩/, the T becomes a catch in the throat, the schwa is gone, and the N carries its own syllable.
The first version preserves the underlying T. The other two are American substitutions. Neither one registers as a substitution to an American ear, even though the mouth is doing something different each time.
Where the glottal stop replaces T
The textbook rule is narrow.
A T becomes a glottal stop when a syllabic N follows.
That covers most of the words where you’ll hear it. The pattern is the -tn ending: T followed by a schwa-and-N that collapses into a single syllabic N. Here are the most common examples:
| Spelled | What Americans say | IPA |
|---|---|---|
| button | buh’n | /ˈbʌʔn̩/ |
| mountain | moun’n | /ˈmaʊnʔn̩/ |
| certain | sur’n | /ˈsɝʔn̩/ |
| kitten | kih’n | /ˈkɪʔn̩/ |
| written | rih’n | /ˈɹɪʔn̩/ |
| cotton | cah’n | /ˈkɑʔn̩/ |
| forgotten | fer-GAH’n | /fɚˈɡɑʔn̩/ |
| curtain | kur’n | /ˈkɝʔn̩/ |
| important | im-POR’n(t) | /ɪmˈpɔɹʔn̩t/ |
Two other environments produce a glottal stop, less consistently:
Before a consonant in the next syllable
Words like atmosphere, outfit, footprint, hotbed (where the T sits at the end of one syllable and another consonant starts the next) can surface with a glottal stop in some speakers, especially in fast speech. The realization varies by speaker and speed. Slow careful speech typically keeps the T as a brief unreleased stop; faster speech often pre-glottalizes it (a quick catch in the throat right before the T) or replaces it entirely with a glottal stop. This pattern is less consistent than the -tn rule and isn’t worth drilling on its own.
Utterance-final and word-final T
At the end of an utterance, a T often surfaces as a glottal stop in normal speech: Wait, That’s it, I can’t, what. The substitution isn’t emphasis-only. Americans glottalize word-final T routinely, especially when nothing follows it. Under explicit emphasis (Wait!) the closure is harder and longer, but the underlying mechanism is the same one you’d hear in casual speech.
The syllabic-N case is the one to learn first. The other two are tendencies. The -tn case is the only one that’s structural.
Glottal stop or flap-T? How to tell
Both the glottal stop T and the flap-T are substitutions for a written T. They live in similar environments (between a vowel and another sound), and learners often confuse them in both directions. The most common over-correction after discovering the flap is to flap everything, including button and mountain. The most common over-correction after discovering the glottal stop is to use it for water and better. Neither sounds right.
The decision rule is small. Look at what follows the T.
If the T is followed by an unstressed vowel (or a syllabic L), use the flap. If the T is followed by a syllabic N, use the glottal stop. (A T that starts a stressed syllable stays a full T regardless of what follows; see Section 4.)
That’s it. Same basic environment (a T in the middle of a word), two outputs, based entirely on the next sound.
| Word | T followed by | Output | Spoken |
|---|---|---|---|
| water | vowel | flap | waa-der |
| butter | vowel | flap | budder |
| city | vowel | flap | siddy |
| little | syllabic L | flap | liddle |
| bottle | syllabic L | flap | boddle |
| button | syllabic N | glottal stop | buh’n |
| mountain | syllabic N | glottal stop | moun’n |
| certain | syllabic N | glottal stop | sur’n |
This is also why button and butter are pronounced quite differently in American English even though their endings differ by only two letters. The vowels are the same. The first consonant is the same. The difference is what comes after the T. A vowel-ish sound triggers the flap; a syllabic N triggers the glottal stop.
The same word can show the rule and its exceptions at once. Important has two T’s. The first is glottalized (im-POR’n(t)) because a syllabic N follows. The second sits at the very end of the word, where it’s often unreleased or also realized as a glottal stop. Both are standard, but neither realization comes from the -tn rule that handled the first T. Same letter, different jobs in one word.
Where the glottal stop does NOT replace T
The most common over-correction is to apply the glottal stop to every T after a vowel. That over-application sounds Cockney or Estuary English, where the T also glottalizes between vowels (better, water) and before a syllabic L (bottle). That broader territory is the flap-T’s job in American English. The -tn case itself, the one this article is about, is now shared between modern American and modern British English. Below are three environments where the American glottal stop does NOT fire, so the T stays a real T or becomes something else.
1. At the start of a stressed syllable
Words like retain, attain, attempt, attack, atomic, Italian, hotel, photographer all keep a crisp aspirated T at the start of their stressed syllable. Re-TAIN, not re-uh-AIN. The clearest evidence is the stress-shift pairs where the same root surfaces with different stress: compare PHOto (flap, stress on the first syllable) with phoTOGrapher (full T, stress on the second), or AT-om with a-TOM-ic. The rule is positional: the T’s environment, not the word’s identity, decides whether it glottalizes.
2. Before a regular vowel (or a syllabic L)
This is the flap-T’s territory. Water is waa-der, not wah-uh-er. The glottal stop never substitutes for the flap. If you find yourself producing a catch in your throat for water, better, or city, you’ve over-corrected.
3. Word-initial T
Two, ten, today, tomorrow always begin with a full aspirated T. American English doesn’t glottalize a word-initial T. (Vowel-initial words like apple or idea often get a small glottal onset before the vowel, but that’s a separate process and doesn’t replace any consonant.)
4. After an N (the NT-cluster)
A T sandwiched between an N and an unstressed vowel (as in winter, center, counter, twenty, plenty, internet) is a third pattern, neither flap nor glottal stop. The T usually disappears entirely: winter sounds like winner, internet like innernet. This is the nasal flap or NT-cluster T-deletion, covered in the flap-T article’s exceptions section. Worth knowing it exists so you don’t try to glottalize winter.
How to make the sound
For most people the glottal stop is already there in the throat. The job is to deploy it on purpose, in the right places.
- Say “uh-oh” slowly. Notice the tiny stop between uh and oh. That’s the glottal stop. It’s the same stop you produce when you start any vowel-initial English word emphatically, the way a singer might attack a note.
- Try saying just the catch on its own: hold your breath for a moment with your mouth open. The held silence is the glottal stop. The release back into a vowel is what makes the closure audible.
- Say kitten with a full T (kit-ten, two clean syllables). Now say it again, but instead of releasing the T into the second syllable, replace the T with the catch from step 1, hold the catch for a fraction of a second, then let your tongue release into the N. Kih’n.
- Move into real words: button, mountain, certain, written, cotton. Each one has the same shape: vowel, glottal stop where the T used to be, syllabic N.
- A common transitional mistake is to say but-uh-n with a real schwa in the middle. The whole point of the glottal stop is that the schwa drops out. The N takes the second syllable on its own.
The motion is smaller than a regular T. You don’t actually need any tongue contact at the alveolar ridge; the stop can happen entirely in your throat. By the time the catch releases, your tongue is already in position for the N.
Practice phrases
Read these out loud, twice each. Don’t rush. The format is spelled sentence → “spoken version, with glottal stops in bold.”
- I lost a button on my coat. I lost a buh'n on my coat.
- The mountain is taller than it looks. The moun'n is taller than it looks.
- I'm certain that's important. I'm sur'n that's im-POR'n(t).
- Have you written it down? Have you rih'n it down?
- The kitten is on the curtain. The kih'n is on the kur'n.
- I've forgotten the cotton shirt. I've fer-GAH'n the cah'n shirt.
- The kitten ate the cotton ball. The kih'n ate the cah'n ball.
- The kitten drank the water. The kih'n drank the waa-der.
- Cotton or button-down? Cah'n or buh'n-down?
- A kitten in Manhattan. A kih'n in man-HA'n (the stressed vowel rhymes with *cat*, not *father*).
If those feel like you’re choking on the word, you’re holding the stop too long. The catch should be brief, the same length as the T it replaces, a few hundredths of a second at most.
Where you’ve already heard it
You’ve heard thousands of glottal stop T’s in American media without ever noticing them. The substitution is so consistent that natives don’t even hear it as a substitution. A few places worth listening for it:
- Any American newscaster reading the word *important*
Anderson Cooper, Lester Holt, Rachel Maddow. All of them glottalize the T in important, mountain, certain whenever those words come up in a script. The substitution is built into the standard register at broadcast formality, though the exact realization slides between a clean glottal stop and a glottalized T depending on tempo and emphasis.
- TV legal dramas
Courtroom scenes lean heavily on certain, important, and mountain. The T disappears every single time.
- NBA play-by-play
Listen for button in a halftime ad about phone interfaces. It will be buh’n every time. Mountain and important land as moun’n and im-POR’n(t) almost every time in the post-game commentary.
- Movie titles
The Mountain Between Us becomes “the moun’n between us.” Manhattan becomes man-HA’n (the stressed vowel is the cat /æ/, not the father /ɑ/). Cotton Club becomes cah’n club.
- Bill Clinton speeches, on the word *important*
A reliable speaker for hearing the glottal stop because important is one of his most-used words; same with certain.
- Animated kids' shows
When a character explains a rule with patient adult speech, button and mountain still glottalize. The substitution holds across conversational and broadcast speeds; only deliberate citation-form speech (a teacher slowly spelling the word out) tends to restore the full T.
Pick any 60-second clip of American speech with a transcript handy. Mark every -tn word. Count how often the speaker produces a real T versus a glottal stop. You’ll find the glottal stop wins almost every time.
How different first languages handle this
Your starting point depends on your first language. Many languages have a glottal stop somewhere in their system, sometimes as a phoneme and sometimes as a transitional sound, and learners with one already in their toolkit have an easier time deploying it on purpose.
| Your L1 | Already have /ʔ/? | What to focus on |
|---|---|---|
| Arabic | ✓ Yes the hamza ء is a phonemic glottal stop, as in سَأَلَ sa’ala “asked” | The sound is identical to the English glottal stop. The new part is applying it before a syllabic N in English. |
| Hebrew | ~ Partial the aleph א was a phoneme historically; in modern Israeli Hebrew it’s largely realized only in careful or liturgical speech | If you produce aleph in careful speech, that’s the same closure you need in English. Otherwise treat it as a sound you have access to but rarely use, and practice deploying it in English -tn words. |
| German | ✓ Yes a glottal stop is the default onset for stressed vowel-initial morphemes (Apfel /ˈʔapfl̩/, Theater /teˈʔaːtɐ/); it’s variable / often omitted before unstressed vowel-initial syllables, more reliable in Northern than Southern varieties | The sound is there. The new part is putting it where the T was, not before a vowel. |
| Danish | ~ Partial the stød is laryngealization (creaky voice) on a vowel, related to but not the same as a true glottal stop closure | The throat-tension instinct is similar, but you’ll need to practice the catch as a discrete closure between syllables, not as a vowel quality. |
| Japanese | ~ Partial a true glottal stop appears at the end of short exclamations like あっ!, distinct from the sokuon’s gemination of following stops | The catch is familiar from short exclamations. Use that same throat-closure before a syllabic N in English. |
| Mandarin Chinese | ~ Partial no phonemic glottal stop, but a light [ʔ] sometimes surfaces as an optional onset on vowel-initial syllables in careful speech (e.g. 安 ān) | The optional light glottal onset some speakers produce on vowel-initial syllables (安, 爱) is closely related to the closure you need in English button. Practice making that catch deliberate and a little firmer. |
| Spanish, Italian, Portuguese | ✗ No no glottal stop; T stays crisp | Build the catch from scratch. The bigger challenge is unlearning the urge to release the T. |
| French | ✗ No no glottal stop; T stays crisp | Build the catch from scratch. The -tn substitution feels unnatural at first because French likes clean consonants. |
| Korean | ~ Partial no isolated phonemic glottal stop, but vowel-initial syllables in phrase- or word-initial position commonly take a glottal onset (e.g. 아 a, 이 i) | The optional glottal onset you may already produce on vowel-initial syllables is close to the gesture you need. Move it from word-onset to mid-word (where the T was) and you’ve got it. |
| Hindi | ✗ No no phonemic glottal stop in standard Hindi | Build from scratch. The -tn substitution is the unfamiliar part. |
For learners coming from languages without the sound, the production itself is easy once you find it: a couple of days of practicing “uh-oh” in isolation will give you the closure. After that the work is deployment: remembering to use it in the right English words. The sound is small; the habit takes longer.
FAQ
No. They sound similar, but the environments are different. Cockney English uses a glottal stop very broadly, including before a syllabic L (so bottle becomes bo’l) and between vowels (so better becomes be’er). American English doesn’t glottalize between vowels or before a syllabic L. Bottle is boddle in American English, never bo’l. Better is bedder, never be’er. The American glottal stop’s structural job is the -tn case (with looser preconsonantal and utterance-final variants noted in Section 2). If you over-apply it to intervocalic positions, you’ll sound British, not American.
Standard. Newscasters, judges, professors, and CEOs all use it. The glottal stop in button, mountain, certain is not a sign of fast or careless speech. It’s part of how General American treats those words at any speed. If anything, refusing to use it sounds non-native.
Older RP pronounced the full T in button (BUT-ən) with both T and schwa. Modern RP also glottalizes /t/ before a syllabic N in many speakers, so button often surfaces as buh’n in younger or less-formal British speech, though older or more conservative RP still preserves the released T and the rate of replacement remains higher in American than in British English. Estuary English and Cockney use the glottal stop more broadly (also between vowels and before a syllabic L). For the -tn environment specifically, the American-vs-British contrast has narrowed substantially.
The glottal stop substitution comes with a structural change. The schwa that would normally connect the T to the N drops out, and the N becomes syllabic. So button isn’t pronounced buh’-uh-n with the catch and a real schwa between it and the N. It’s pronounced buh’n: the T turns into the catch, the schwa drops out, and the N alone carries the second syllable. Two changes happen at the same time, not one after the other.
For words like tonight and attain specifically, nothing changes. Both have T at the start of a stressed syllable followed by a vowel, so the T stays full and aspirated. The structural glottal-stop substitution fires before a syllabic N (the schwa-N collapse). The looser preconsonantal and utterance-final glottal stops covered in Section 2 are tendencies, not substitutions you have to deploy on purpose, and they don’t apply in tonight or attain either.
For learners with a glottal stop already in their first language (Arabic, German, Danish, Japanese exclamation closures), often a few days of focused practice. For learners building the catch from scratch, one to two weeks is typical. The sound itself is small; the work is in remembering which environments call for it.
For most learners, the glottal stop T is a smaller adjustment than the flap-T, and it pays back almost as much clarity. The two rules together (flap when an unstressed vowel follows, glottal stop when a syllabic N follows) handle the two biggest categories of mid-word T in American English. A week of the practice phrases above is usually enough for the substitution to start running on its own.
