Reference
Phonemes vs allophones
Why English speakers cannot hear the difference between two Korean stops, and why Korean speakers cannot hear English b and v. The distinction is the single most useful idea in learner phonetics.
Say pin. Now say spin. Hold a sheet of paper in front of your mouth and do it again.
On pin, the paper moves. On spin, it does not.
Those are two physically different sounds — one aspirated, one not, [pʰ] and [p] — and if you are an English speaker you have produced this difference reliably, every day, for your entire life, without ever once noticing it.
The definitions
A phoneme is a sound category that can change meaning in a given language. Swap it for another and you get a different word.
An allophone is one of the physical ways a phoneme gets realised. Which allophone appears is predictable from context, and swapping one for another never changes a word — it just sounds slightly odd, or not even that.
English /p/ is one phoneme with (at least) two allophones:
/p/ ← one phoneme
╱ ╲
[pʰ] [p] ← two allophones
│ │
word- after
initial /s/
"pin" "spin"You cannot find an English pair of words distinguished only by aspiration. There is no word [pɪn] meaning something different from [pʰɪn]. So English files both under one category, and native speakers are blind to the difference.
Where it gets consequential
In Korean, that same difference is phonemic. Korean distinguishes plain /p/, aspirated /pʰ/ and tense /p͈/ as three separate categories, and they distinguish real words. To a Korean speaker these are as obviously different as /p/ and /b/ are to you.
To an English speaker they are one sound, three times.
This is the whole point, and it generalises:
The distinctions your language treats as allophonic are exactly the distinctions you will be unable to hear in another language.
It runs in both directions. English /b/ and /v/ are separate phonemes; Spanish has one phoneme /b/ realised as [b] after a pause or nasal and as [β] between vowels. So Spanish speakers hear English berry and very as one word — the same blindness, aimed the other way.
More examples
| Language | Phonemic there | Allophonic or absent elsewhere |
|---|---|---|
| Korean | /p pʰ p͈/ three-way | One category in English |
| Hindi | /t̪ t̪ʰ ʈ ʈʰ/ dental vs retroflex | One /t/ in English |
| Arabic | /s/ vs /sˤ/ emphatic | No emphatics in English |
| Finnish | /tuli/ vs /tuːli/ length | English length is allophonic |
| Mandarin | tone | English pitch is sentence-level |
| English | /b/ vs /v/ | Allophones of one phoneme in Spanish |
| English | /iː/ vs /ɪ/ | One /i/ in Spanish, Italian, Greek |
| Japanese | mora timing | Not contrastive in English |
Read that table as a prediction machine. Given a learner’s first language and a target language, you can say in advance which contrasts will be difficult, without ever having heard them speak. This is the basis of the hard-sound lists on our language guides.
How to tell which is which
Two tests, and they are the actual methodology.
Test 1 — find a minimal pair. Two words differing in exactly one sound, with different meanings. pin / bin proves /p/ and /b/ are separate English phonemes. If a minimal pair exists, the sounds are contrastive.
Test 2 — check for complementary distribution. If two sounds never occur in the same environment — one always here, the other always there — they are allophones of one phoneme. English [pʰ] appears at the start of stressed syllables; [p] appears after /s/. They never compete, so no word can ever be distinguished by them.
Complementary distribution is the signature of allophony. Contrast is the signature of phonemes.
Slashes and brackets
The notation encodes exactly this distinction, which is why IPA transcriptions come in two flavours.
/slashes/ — phonemic. The abstract categories of the language, with predictable detail stripped out. pin is /pɪn/. This is what dictionaries print and what learners want.
[brackets] — phonetic. What was actually produced. pin is [pʰɪn]. Used when the detail matters — describing an accent, diagnosing a learner error, or documenting a language.
Both are correct at their own level of description. A phonemic transcription is not sloppier; it is answering a different question.
Why teachers should care
Three practical consequences.
1. It predicts errors before you hear them. A student’s first language tells you which contrasts will be collapsed. You do not have to discover this by trial and error over six weeks.
2. It explains why repetition fails. When a student cannot reproduce a distinction, the usual instinct is to model it again. But if the contrast is allophonic in their language, they are not hearing two sounds and missing one — they are hearing one sound twice, and no amount of modelling changes that. Perception has to be trained first, with minimal pairs and forced choice.
3. It tells you what to ignore. A learner who produces English /p/ without aspiration sounds slightly foreign and is perfectly intelligible, because aspiration is not contrastive in English. A learner who merges /iː/ and /ɪ/ will be misunderstood constantly, because that contrast carries hundreds of real word distinctions. Same effort, wildly different payoff — and the phoneme/allophone distinction is what tells you which is which.
That last point is the whole argument for teaching by functional load rather than by exoticism. The phoneme inventory of the target language is the map of what actually matters.
Frequently asked questions
What is the difference between a phoneme and an allophone?
A phoneme is a sound category that can change meaning in a language. An allophone is one of the physical realisations of that category — a variant that occurs in a predictable context and never changes meaning. English /p/ is one phoneme with at least two allophones, aspirated and unaspirated.
How do you tell if two sounds are separate phonemes?
Look for a minimal pair — two words differing only in those sounds, with different meanings. If one exists, they are separate phonemes in that language. If the two sounds never appear in the same environment, they are allophones of one phoneme.
Why does this matter for language learning?
Because your phoneme inventory determines what you can hear. A distinction that is allophonic in your language is invisible to you, and a distinction that is phonemic in the target language but allophonic in yours is exactly where your errors will be.
What is the difference between slashes and square brackets?
Slashes mark phonemic transcription — the abstract categories, with predictable detail stripped out. Square brackets mark phonetic transcription — what was actually produced, including allophonic detail.