Phonemic vs Phonetic Transcription: Slashes and Brackets

Phonemic transcription records only the sounds that can change the meaning of a word, and is written between slashes: pin is /pɪn/. Phonetic transcription records how the sounds were actually produced, including detail that never changes meaning, and is written between square brackets: the same word is [pʰɪn], with a small raised h marking the puff of air on the p.

The two are not competing systems. They are the same alphabet used at two levels of detail, chosen according to what you are trying to show. This guide covers when each is used and why the distinction exists at all.

Phoneme, phone and allophone

Three terms do all the work here, and they are easier than they look.

A phone is any speech sound, considered simply as a sound. The p you say in pin and the p you say in spin are two different phones, because one comes with a puff of air and the other does not.

A phoneme is a category of sound that the language treats as one unit. English has one p phoneme, /p/, even though it is produced in several different ways, because swapping those ways never produces a different word.

An allophone is one of the ways a phoneme can be produced. Aspirated [pʰ] and plain [p] are two allophones of the English phoneme /p/.

The test that separates phonemes from allophones is the minimal pair. If swapping sound A for sound B produces a different word, A and B belong to different phonemes. Pin and bin are different words, so /p/ and /b/ are separate phonemes. Say pin without the puff of air and you have said pin in a slightly unusual accent, not a different word, so the puff is allophonic.

That test is the whole reason minimal pairs matter for pronunciation practice. They map exactly onto the contrasts your language actually requires you to make.

What the brackets mean

The punctuation is not decorative. It tells a reader which claim is being made.

Mixing them is a genuine error rather than a style choice. Writing [pɪn] for pin claims that the p was produced without aspiration, which for a word said in isolation is almost certainly false. Writing /pʰɪn/ claims that aspiration is contrastive in English, which would mean there is an English word distinguished from pin only by the puff of air. There is not.

The aspiration example

This is the standard illustration because it is easy to feel. Hold your palm a few centimetres from your mouth and say pin, then spin.

On pin you feel a clear puff of air after the p. On spin you feel almost nothing. Phonemically both words contain /p/. Phonetically the first has [pʰ] and the second has plain [p].

The pattern is completely regular: /p/, /t/ and /k/ are aspirated at the start of a stressed syllable and plain after /s/. Because it is predictable from the surroundings, a dictionary does not need to write it down. You will produce it automatically once you know the word.

This also explains a common listening problem. To a speaker of a language where aspiration is contrastive, the unaspirated p in spin can sound like a b. Nothing is wrong with either ear. The two languages simply draw the category boundary in different places.

More English allophones

Aspiration is not the only case. English /t/ is the most variable consonant in the language.

Every one of those is /t/. A British speaker saying water with a full [tʰ] and an American saying it with a tap are pronouncing the same phoneme, and neither has changed the word.

English /l/ splits the same way. Before a vowel it is clear, as in leaf [liːf]. After a vowel it is dark, with the back of the tongue raised, as in feel [fiːɫ]. Both are /l/, and the difference is written with a small tilde through the letter only when it matters.

Vowels have allophones too. A vowel before a nasal consonant picks up nasality from it, so ban is [bæ̃n] narrowly and /bæn/ broadly. A vowel before a voiced consonant is noticeably longer than the same vowel before a voiceless one, which is why bad [bæːd] holds longer than bat [bæt].

None of these change which word you have said. All of them are part of sounding natural.

Complementary distribution: the test that sorts them

There is a second test alongside minimal pairs, and it is what lets a linguist decide that two sounds are allophones rather than phonemes.

Two sounds are in complementary distribution when each one appears only in positions where the other never appears. Aspirated [pʰ] shows up at the start of a stressed syllable. Plain [p] shows up after /s/. Neither ever turns up in the other’s territory, so they can never contrast, so they must belong to the same phoneme.

Compare that with /p/ and /b/. Both appear at the start of a word, both appear between vowels, both appear at the end. They compete for the same positions, and where two sounds compete, a minimal pair usually exists to prove they are separate.

The same reasoning explains clear and dark l. Clear [l] appears before a vowel, dark [ɫ] appears after one. No English word is distinguished by which you use, and no English word lets you choose. They are one phoneme in two costumes.

This is why the phoneme inventory of a language is a discovery rather than a given. You collect the sounds, look at where each one can occur, and group together the ones that never compete.

The same sentence at both levels

Seeing a whole sentence transcribed twice makes the gap concrete. Take Pat stopped the little water bottle, said by a General American speaker at a normal conversational pace.

Phonemically, it is a straightforward list of the contrastive sounds:

/pæt stɒpt ðə ˈlɪtəl ˈwɔːtə ˈbɒtəl/

Narrowly, with the allophones filled in, considerably more is happening:

[pʰæt̚ stɑpt ðə ˈlɪɾɫ̩ ˈwɑɾɚ ˈbɑʔɫ̩]

Four things showed up in the second version that the first left out. The p in Pat is aspirated and the p in stopped is not. The final t in Pat is held without release. Three of the t sounds became taps or a glottal stop. Two of the l sounds became dark and syllabic, carrying a whole syllable on their own with no vowel.

Not one of those changes altered a single word. A listener hearing the second version understands exactly the sentence written in the first. That gap, between what distinguishes words and what merely fills them out, is the entire subject of this page.

Notice also that the narrow version is tied to one accent. A standard British speaker would produce something different again, with full [tʰ] sounds and no rhotic ending on water. The phonemic line covers both.

The diacritics you will meet

Narrow transcription adds small marks to the base symbols. You do not need to produce these, but recognising them helps when reading a phonetics text.

The length mark ː is the one that causes confusion, because it appears in both systems. In British phonemic transcription, /iː/ is a separate phoneme from /ɪ/, so the mark is contrastive and belongs in the slashes. In narrow transcription the same mark can also indicate extra length caused by the following consonant, which is not contrastive. Same symbol, different job, decided by the brackets around it.

Why dictionaries choose phonemic

A dictionary is answering one question: which word should come out of your mouth. The phonemic level answers it completely, and it has three practical advantages.

It is shorter. /ˈwɔːtə/ is easier to read than [ˈwoːʔə], and a dictionary entry has limited space.

It is stable. A narrow transcription describes one production by one speaker. Change the speaker, the speed or the surrounding words and the narrow transcription changes too. The phonemic form does not.

It is accent-tolerant. A single phonemic transcription can be read correctly by speakers with different accents, because each fills in their own allophones automatically. That is why the same entry works for readers who tap their /t/ and readers who do not.

This site uses phonemic transcription throughout for the same reasons, which is why every entry in the consonant guide and the vowel guide appears in slashes.

When narrow transcription is the right tool

There are real jobs that phonemic transcription cannot do.

Describing an accent. The difference between a Scottish and a London pronunciation of butter is entirely allophonic. Both are /ˈbʌtə/. Only narrow transcription can show what actually separates them.

Speech and language therapy. A clinician needs to record what a client produced, not what the target word was. [θ] where /s/ was intended is exactly the information that matters.

Language documentation. When describing a language for the first time, nobody yet knows which differences are contrastive. You transcribe narrowly, then work out the phoneme inventory from the data.

Teaching a specific habit. If a learner is releasing every final stop, telling them to aim for [t̚] is more useful than telling them the word contains /t/, which they already knew.

Three common misunderstandings

“Square brackets mean more correct.” They mean more detailed. A narrow transcription of a mispronounced word is still a mispronounced word, recorded precisely.

“Phonemes are sounds.” Phonemes are categories. /t/ is not a sound you can play; it is a label for a group of related sounds that English treats as one. The things you can actually play are phones.

“Every language has the same phonemes.” The inventory differs, and so do the boundaries. Aspirated and unaspirated p are allophones in English and separate phonemes in Hindi and Thai. English /r/ and /l/ are separate phonemes and are not everywhere. Which differences you can hear easily depends on which ones your first language made you pay attention to.

Practice: which notation?

Decide whether each statement should be written in slashes or brackets.

1. The English word cat contains three sounds.
2. That speaker released the final t with an audible burst.
3. Ship and sheep differ in one sound.
4. In this recording the vowel was nasalised throughout.
5. American English has a tap between vowels in better.
6. The l at the end of feel is darker than the one at the start of leaf.
7. English has 24 consonant phonemes.
8. The child produced an f where the target word had a th.

Question 8 is the one that shows why both systems are needed at once. A record of the form “produced [f] for /θ/” carries two different kinds of claim in a single line: what the language required, and what the speaker did. Neither notation could express that alone.

A short note on where the idea came from

The phoneme was not obvious. For most of the history of writing, people assumed that a letter stood for a sound and that was that. It took until the late nineteenth and early twentieth centuries for linguists working on a range of languages to notice that speakers systematically fail to hear differences their own language does not use.

The International Phonetic Alphabet was created in the 1880s to write sounds precisely, and the phonemic idea arrived alongside it as a way of deciding which of those precise differences a given language actually cares about. The two developed together, which is why one alphabet serves both jobs and only the punctuation tells them apart.

The practical consequence for a learner is the useful part. Your difficulty with an English contrast is not a defect of your ear. It is a filter your first language installed for good reasons, and it can be retrained by giving it evidence, which is what pair practice does.

Where this fits

For everyday use, phonemic transcription is all you need. It is what appears in dictionaries, in learner materials, and in the Studio, and it is enough to say any English word correctly.

Narrow transcription becomes useful the moment you start describing speech rather than prescribing it: comparing accents, recording what someone actually said, or working on a specific physical habit. Learning to read it is worthwhile, but it comes second.

If the symbols themselves are the sticking point rather than the notation around them, how to read IPA takes a transcription apart from the beginning.