You looked up the same word in two dictionaries and got two different answers. One wrote pin as /pɪn/. The other wrote it as [pʰɪn]. Same word, same accent, but one uses slashes and the other uses square brackets, and there's an extra little superscript sitting on the p. Did one of them make a mistake?
Neither did. They're answering two different questions. The slashes tell you which sounds matter for meaning. The brackets tell you what your mouth actually does. This is the difference between phonemic and phonetic transcription, also called broad and narrow transcription, and once it clicks, half the confusion around IPA disappears.
Broad and narrow in one sentence
Here it is: slashes /like this/ mark a phonemic, broad transcription; square brackets [like this] mark a phonetic, narrow one.
A phonemic transcription records the sounds that distinguish words in a language, one symbol per phoneme: the smallest unit of sound that can change a word's meaning. In British English teaching it is often called phonemic script. It is the "what" of pronunciation, the contrasts you need to hear and produce to be understood.
A phonetic transcription records what actually comes out of a speaker's mouth, including details that never change meaning but do vary with the position in the word, the accent or the speaker. It is the "how".
Broad and narrow are not two fixed settings but the two ends of a scale. A broad transcription shows only the most important features; a narrow one adds detail "to varying degrees", as Wikipedia's article on phonetic transcription puts it. The same symbols appear in both. The brackets just carry more of them.
Phoneme vs allophone: the core concept
A phoneme is a sound category that the speakers of a language treat as a single basic sound. English speakers think of "the t sound" as one thing. But that single category is produced as several physically different sounds depending on where it sits in a word. Those physical variants are called allophones.
Allophones of the same phoneme never change a word's meaning. Swap one for another and you get an accent or an oddity, never a different word. That's why your brain files them all under one label.
Take English /t/. It is one phoneme, but it surfaces as a whole family of sounds:
| Word | Phonemic (broad) | Phonetic (narrow) | What's happening |
|---|---|---|---|
| top | /tɒp/ | [tʰɒp] | aspirated: a puff of air after the t |
| stop | /stɒp/ | [stɒp] | unaspirated: no puff after the t |
| water (US) | /ˈwɔtɚ/ | [ˈwɔɾɚ] | flapped: a quick tap, almost a d |
| cat (US) | /kæt/ | [kʰæt̚] | unreleased: the t is closed off, never popped |
| button | /ˈbʌtən/ | [ˈbʌʔn̩] | a glottal stop in place of the t |
These five come from the realizations of English /t/ that Wikipedia's allophone article lists; it also gives a nasal flap, [ɾ̃], for American winter. No native speaker hears five different words; they hear "t" every time. The phonemic transcription writes /t/ for all of them because that's the unit that matters for meaning. The phonetic transcription spells out each realization with its own symbol or diacritic.
Aspiration: the [pʰ] you never noticed
Hold your hand in front of your mouth and say pin, then spin.
On pin, you feel a puff of air on your palm. On spin, you don't. That puff is aspiration, and the difference is real and consistent, yet you almost certainly never noticed it, because English doesn't use it to tell words apart.
Phonetically, pin is [pʰɪn]: the superscript ʰ marks aspiration. The p in spin is plain, so the word is [spɪn]. Wikipedia's allophone article says it directly: "[pʰ] as in pin and [p] as in spin are allophones for the phoneme /p/." Phonemically, both words start with the same /p/.
The dictionary that wrote /pɪn/ wasn't being lazy. It was telling you what an English learner strictly needs: this word starts with the /p/ phoneme. The dictionary that wrote [pʰɪn] was being precise about what your mouth does.
Here's the catch for language learners: in some languages, aspiration does change meaning. In Mandarin and Korean, aspirated and unaspirated stops are separate phonemes, so they make separate words. The detail English buries inside one phoneme is a meaning-bearing contrast elsewhere. What counts as "the same sound" depends entirely on the language.
How minimal pairs prove two sounds are phonemes
If allophones don't change meaning, how do linguists decide what counts as a separate phoneme? The test is the minimal pair: two words that differ in exactly one sound and mean different things.
tip and dip differ only in their first sound, /t/ versus /d/, and they're different words. That proves /t/ and /d/ are separate phonemes in English. Compare:
- pin /pɪn/ vs bin /bɪn/ → /p/ and /b/ are separate phonemes
- sit /sɪt/ vs seat /siːt/ → /ɪ/ and /iː/ are separate phonemes
- light /laɪt/ vs right /raɪt/ → /l/ and /r/ are separate phonemes
Now try to find a minimal pair that hinges on aspirated [tʰ] versus unaspirated [t] in English. You can't. No two English words differ only in that. The missing minimal pair is the evidence that [tʰ] and [t] are allophones of one phoneme.
That is why minimal pairs matter so much for pronunciation training: they isolate exactly the contrasts a learner must master. We go deeper in the minimal pairs English learners confuse most.
Clear L, dark L: same phoneme, different mouth
English /l/ is one phoneme with two noticeably different realizations. Say leaf, then feel, slowly.
In leaf the tongue tip touches behind your teeth and the body of the tongue stays neutral: that's "clear" or "light" L. In feel the back of the tongue also rises toward the soft palate, giving a darker, hollower sound: "dark" or velarized L. A narrow transcription writes the light [l] of leaf [ˈliːf] and the dark [ɫ] of feel [ˈfiːɫ], with a tilde through the l.
A broad transcription writes /liːf/ and /fiːl/, because in English the difference is automatic: your tongue picks the variant from the position in the word, and it never changes meaning. For most learners that's a footnote. If you're polishing an accent, though, dark L is exactly the kind of detail that separates "understandable" from "native-sounding", and it lives entirely in the narrow layer.
How narrow can a transcription get?
As narrow as the transcriber needs. Wikipedia's phonetic transcription article uses little to show the range: a broad transcription that suits both British and American dictionaries is /ˈlɪtᵊl/, while a narrow one of a typical American speaker is [ˈɫɪɾɫ̩] and of a speaker from southern England [ˈlɪʔo]. That is the trade-off. The narrower the transcription, the more it describes one accent, or one speaker on one day, rather than the word.
Narrow transcriptions get their extra detail from diacritics and from symbols a broad one would not use. These are the ones you will meet most in English:
| Mark | Name on the IPA chart | Example |
|---|---|---|
| ʰ | aspirated | pin [pʰɪn] |
| ◌̚ | no audible release | US cat [kʰæt̚] |
| ◌̩ | syllabic | button [ˈbʌʔn̩], where the n is a syllable by itself |
| ◌̃ | nasalized | the vowel of US man, [mæ̃n] |
| ɫ | velarized (dark) l | feel [fiːɫ] |
| ɾ | alveolar tap | US water [ˈwɔɾɚ] |
| ʔ | glottal stop | button [ˈbʌʔn̩] |
Why IPA converters give you broad transcription
Every text-to-IPA converter, including the IPAtics IPA translator, starts from spelling. The spelling tells it which word you typed, but not how any particular person says it. So the most a converter can honestly give you is the word's dictionary pronunciation in a reference accent, and that is a broad, phonemic transcription.
That is what IPAtics outputs: one symbol per phoneme, plus stress and length marks where a language's dictionaries normally show them. Its English transcriptions never add aspiration, unreleased stops or dark L, because those depend on the speaker, the speed and the neighbouring words. (Its Mandarin and Korean ones do write ʰ, because there aspiration is a phoneme.) It keeps two narrow habits in English, both American: General American output writes the t of words like water and city as the flap [ɾ], and the t of button, kitten and certain as a glottal stop [ʔ] before a syllabic n, because that is how those words are normally said in that accent.
A "narrow transcription generator" that works from text alone could add more detail by rule, for example aspiration on every voiceless stop at the start of a stressed syllable. But the result would be a prediction for an idealised speaker, not a record of anyone's speech. A genuinely narrow transcription needs a recording and a trained listener, or a speech model built for the job; audio to IPA explains what those can and cannot do.
Why two dictionaries show different IPA for the same word
Now we can answer the question you started with. Three things cause the same word to appear with different IPA across sources.
1. Broad versus narrow. A learner's dictionary gives a broad, phonemic transcription in slashes: the minimum you need. A pronunciation reference or a phonetics paper may give a narrow, phonetic one in brackets: much more detail. pin is /pɪn/ broadly and [pʰɪn] narrowly. Both are correct; they're zoomed to different levels.
2. Different accents. butter is /ˈbʌtər/ in General American and /ˈbʌtə/ in British Received Pronunciation: the American keeps the r, the British drops it. Phonetically, the American t flaps to [ˈbʌɾɚ], so an American-focused source may show the flap while a British one won't. American vs British IPA compares the patterns word by word.
3. Different conventions. Even for the same accent, editors make choices. Wikipedia notes that phonemic symbols "will frequently be chosen to avoid diacritics as much as possible, under a 'one sound one symbol' policy." One dictionary marks length with /iː/, another writes /i/; one gives a final syllable a vowel, another makes the consonant syllabic. These are house styles, not disagreements about the sound.
So when two dictionaries differ, ask: are they zoomed to different levels, describing different accents, or using different house styles? It's almost always one of those three, not an error. What IPA is and why phonetic transcription matters covers the practical side.
When you are working in the other direction, recovering a word from its transcription, both kinds of brackets are packaging you strip before reading the sounds; see how to read a transcription back into a word.
Which one should a learner care about?
For most language learning, you want the phonemic layer. The slashes give you the contrasts that change meaning: the sounds you have to get right to be understood and to understand others. Learn /ɪ/ vs /iː/ so sit and seat don't collapse. Learn /l/ vs /r/ so light and right stay separate. That's where the payoff is.
Phonetic transcription, the brackets, earns its keep in narrower situations:
- Accent fine-tuning. Aspiration, dark L, the flapped t: none of them change meaning, but they make you sound native.
- Speech therapy and accent coaching, where the goal is the exact physical target, not just the contrast.
- Linguistics coursework, where you're documenting how a language or dialect actually behaves.
- A language where an English allophone is a real phoneme, like aspiration in Mandarin or Korean, where ignoring it means saying the wrong word.
A good rule: start phonemic, stay phonemic until you're reliably understood, and drop into phonetic detail only when you're chasing a specific accent or a contrast your target language cares about and English doesn't. If you're still learning the symbols, our beginner's guide to reading IPA covers the chart from zero.
Trying It Yourself
The fastest way to internalize the distinction is to look at broad transcriptions of words you already know and notice which details the slashes leave out.
IPAtics shows the broad IPA of any word you select, in any app, with one hotkey. Or use the IPA translator in your browser without installing anything.
Frequently Asked Questions
What is the difference between phonemic and phonetic transcription?
Phonemic transcription, written in slashes /like this/, records only the sounds that distinguish words in a language: one symbol per phoneme. Phonetic transcription, written in square brackets [like this], records how those sounds are actually pronounced, including details like aspiration or flapping that don't change meaning. Phonemic is "broad", phonetic is "narrow".
What is the difference between broad and narrow transcription?
A broad transcription shows only the features that matter most, usually the phonemes. A narrow transcription adds phonetic detail, such as the aspiration in [pʰ], the unreleased stop in [t̚] or a dark [ɫ]. They are two ends of a scale rather than two fixed systems, and dictionaries normally use the broad end.
What is an allophone?
An allophone is one of the physical variants of a single phoneme that don't change meaning. English /t/ has allophones like aspirated [tʰ] in top, flapped [ɾ] in American water and unreleased [t̚] in cat. Native speakers hear them all as "t" because swapping them never produces a different word.
Can an IPA converter produce a narrow transcription?
Not of a real speaker. A converter works from spelling, so it can only give the dictionary pronunciation of a word in a reference accent, which is a broad transcription. Details like aspiration or dark L can be predicted by rule, but recording them requires the audio itself.
What does the little raised ʰ mean in [pʰ]?
The superscript ʰ marks aspiration, a puff of air released after the consonant. English /p/, /t/ and /k/ are aspirated at the start of stressed syllables, as in [pʰɪn] for pin, but not after /s/, as in [spɪn] for spin. It appears in narrow transcription, not broad.
Does every language have the same phonemes?
No. Each language has its own inventory of phonemes and its own rules for which allophones appear where. A sound that's just an allophone in one language can be a full phoneme in another: aspiration is meaningless for English /p/ but distinguishes words in Mandarin and Korean.
Sources: the definitions of broad and narrow transcription, the little example and the "one sound one symbol" quotation follow Wikipedia's Phonetic transcription article; the allophones of /p/, /t/ and /l/ follow Wikipedia's Allophone article; the diacritic names follow the International Phonetic Association's IPA chart.
Related reading: How to read IPA — a complete beginner's guide · Minimal pairs: the English sounds learners confuse · Audio to IPA: can you transcribe speech to IPA?