I'm trying to create a website to help my partner learn phonetics. She is taking a class as part of her English degree. The issue is that I do not understand how phonetic translation works and resources (words datasets) I've found online differs in their translations and the phones they're using.

Based on the bibliography from her class (in French sorry), the phones they are using come from the Longman pronunciation dictionary and the "Manuel de transcription phonétique de l'anglais" (simply translated: "manual for english phonetic transcription").

Here is a screenshot of the phones considered necessary to translate english, if I only base my knowledge on the "Manuel de transcription ..."

But, will browsing various *.stackexchange question, I stumbled upon two different phone sets, on phoible:

Both inventories are targeting British English, but their phones differs. It kind of confuses me even more...

Furthermore, I feel like her professors use phonetic and phonemic translation interchangeably (or i am missing something), even during the same translation. Here is an exercise they saw in class.

The document is supposed to be a phonetic translation, but Britain kinda matches the wiktionary phonemics [ˈbɹɪt.n̩] instead of phonetics /ˈbɹɪt.ən/.

And what confuses me the most, is that I have the impression the phonetic translation isn't the same from one website to another. She assures me it should, since it's all based on the IPA, but I have seen different results for the same word on LONGMAN DICTIONARY, Cambridge dictionary and Wiktionary.

Am I missing something? If so, could you help me understand all this a little bit better, I would like to help her more

  • 25
    Are you laboring under the assumption that there's only one way to pronounce any given word in English? Feb 26 at 21:30
  • 9
    Side note: We don't call them "phones". They're "phonemes".
    – Jay
    Feb 27 at 2:07
  • 9
    Actually, we do call them "phones", if we're talking about just the sound and not its meaning or how it fits into the language. It's a bit jargonish, but you find this definition both in Merriam-Webster and in various dictionaries in the Oxford family, such as this one. Feb 27 at 8:18
  • 7
    To add to what @the-baby-is-you wrote, there are somewhere between 20 to 40 different dialects of English in Great Britain alone, depending on how one counts. This does not count the over 100 dialects in the US, Canada, India, Australia, South Africa, the Caribbean, and other places. Some dialects are completely non-understandable even to native speakers of English who do not know that particular dialect. Feb 28 at 0:13
  • 12
    I came here expecting a discussion about beer
    – WendyG
    Feb 28 at 12:25

2 Answers 2

  • There are many dialects and accents in Britain. All words have more than one correct pronunciation. Words will be pronounced differently when spoken slowly and deliberately compared to when they are spoken quickly or even slurred.
  • IPA is intended to be a tool that one can use to represent, with various degrees of accuracy, spoken sounds. Different users will use the tool in different ways.
  • A transcription may include or omit some details. For example the "light" and "dark" l sound, or the aspiration of some consonants.
  • There is a lot of variation in vowels. Vowels are often reduced in rapid speech to schwas, or entire syllables can be lost! Vowels exist in a spectrum - there is a smooth variation from any one vowel to any other.
  • For example, "n̩" (syllabic alveolar nasal) vs. "ən" (schwa vowel followed by alveolar nasal) are very similar and represent variation in pronunciation. A particular speaker may produce something more like "n̩" on one day, and more like "ən" the next day (and not even be aware that they have done so.
  • Dictionaries, in particular, will omit details, or even use the "wrong" symbol, as they are meant to be a general guide for non-specialists. (A common example is the use of /r/ instead of /ɹ/ for the "r" sound. But there is a lot of dialect variation in this consonant and using /r/ helps non-specialists to read the transcription.
  • So you should expect lots of variation. Nothing here is specific to English. The same would be true in French - in fact, if you are developing a website you should get it to work for French words first - then you won't be confused by the English words. Moreover, you will know when a transcription represents a "standard" careful pronunciation, and when it represents a Marseilles factory worker speaking quickly, or a drunken Breton fisherman.
  • 9
    Regarding the question's example of "button" transcribed with "n̩" (syllabic alveolar nasal) vs. "ən" (schwa vowel followed by alveolar nasal): these are very similar and it is reasonable that dictionaries might differ here.
    – nschneid
    Feb 26 at 22:04
  • 2
    @nschneid and in particular, the syllabic pronunciation is more common after a dental stop
    – Tristan
    Feb 27 at 11:33
  • 2
    If it helps for French, «pâte» 'dough' is a good example. The conventional transcription is /pɑt/, but most French people don't distinguish it from «patte» 'paw', /pat/, and the actual vowel they produce might be somewhere in the middle like [pät]. The reason it's transcribed differently is historical, and Belgians and Quebeckers still distinguish it. Quebeckers talking casually will even say [pɑɔ̯t]. One complication though is that it's also common to see a "usual" system in French, with backslashes, that doesn't reflect historical changes, so for «pâte» you might see "\pat\ or \pɑt\".
    – wjandrea
    Feb 29 at 3:06
  • 1
    There's certainly debate about how to transcribe French, and the first 2 sources I looked at (Wikipedia easypronunciation.com) differ with regard to nasal vowels. From the question it seems that the OP has no familiarity with the use of IPA for French and thus doesn't realise these issues are common to all languages.
    – Stuart F
    Feb 29 at 16:32
  • This answer does not address her specific examples, which are for English, not French. Also, her question is not about allophones, which are what explain accent differences but not word differences.
    – Lambie
    Feb 29 at 17:45

John Trim's English Pronunciation Illustrated, Cambridge University Press is still the best guide to English pronunciation I know. He shows the difference between sheep and ship, for example, and gives the spellings with pictures.

It is the only book you need to teach someone English Pronunciation or to learn the phonemes and phones of English:

enter image description here

And the big, main difference between AmE and BrE can be taught by the teacher if need be, which is the a in some BrE words versus AmE. Such as kɑː for car in BrE and AmE. Phonemes are not detailed enough for this. That is the main difference. You can hear the difference here with father: hear father in AmE and BrE English has 44 phonemes and this book covers them all.

This book shows contrasting phonemes (sheep/ship or beat/bet) along with pictures and the IPA in the top corner so you learn how to refeer to the sound.

For example:

A tree and three leaves [spelling: ee and ea in these English words]. The vowel sound is i:

He goes through all the 44 phonemes of English. The top left corner has the IPA symbol for the sound being practiced on that page.

It's on YouTube with the illustrations, just enlarge the page!

Trim_English Pronunciation Guide

The Phonetic Manual of English (Manuel phonétique de l'anglais) given in the question explains the levels of representation of English sounds very clearly:

It states first (page 17): [...] A language such as English can be represented in writing in three different ways:
• with letters of the alphabet called "graphemes". Graphemes are written between < > angle brackets like such <b>, <c>, <x>, <a>, and <y>.
• with phonetic symbols that can be written between slashes / / (phonological or phonemic transcription) or between square brackets [ ] (phonetic or allophone transcription). Although the aim of phonological and phonetic transcriptions is above all used to transcribe spoken language (oral), they show, nevertheless, a different positioning for spoken language.

It goes on to say:

The basic unit of phonological transcription is the phoneme, that is a minimal unit differing in sound but not in meaning. A phoneme is a contrastive unit that contributes to distinguishing a series of phones that are not per se at word level. The principal function of a phoneme is to help speakers identify and distinguish words from each other. It is possible to show all the phonemes of a language by using the principle of commutation (substitution) of minimal pairs. A "minimal pair" is a made up of two pairs with only a single difference in sound (phonological element). If the phoneme /k/ for /kɑː/ is substituted for /b/ the sound becomes /bɑː/. The two forms /kɑː/ ~ /bɑː/ show that the /k/ phoneme in initial position as /kɑː/ is different from a /b/ in initial position as /bɑː/. Minimal pairs such as this show the distinguishing status of the two phonemes and lead to interpreting /kɑː/ and /bɑː/ as two different words with different meanings. [as in car and bar, British accent, this is the a sound that differs so much between BrE and AmE]

The manual also describes allophones and their transcription in English very a good deal of detail. [I am not going to translate the entire manual here.]

As for the differences found in the corpora referenced above (Phoible inventory,they are allophonic ones come from this gentleman:

Steven Moran Background I am a scientist trained in computational linguistics, linguistic fieldwork, and language teaching. My research focuses on the evolution of the phonological system in humans, quantitative approaches to linguistic diversity, and aspects of language ontogeny from a cross-linguistic perspective. See my CV and my Google Scholar profile for access to my publications.

Those databases are a compendium of British English on different corpora and show allophonic differences. But the main pages show standard IPA. Those allophonic differences found in the corpora are shown as percentages in the column labeled segments. But he is using IPA and the allophonic differences (accents) from his corpora which obviously had accent differences in them.


Allophone, one of the phonetically distinct variants of a phoneme (q.v.). The occurrence of one allophone rather than another is usually determined by its position in the word (initial, final, medial, etc.) or by its phonetic environment. Speakers of a language often have difficulty in hearing the phonetic differences between allophones of the same phoneme, because these differences do not serve to distinguish one word from another. In English the t sounds in the words “hit,” “tip,” and “little” are allophones; phonemically they are considered to be the same sound although they are different phonetically in terms of aspiration, voicing, and point of articulation.

Allophones have some complicated diacritical marks, all of which are explained in the manual.


AND: A phone is a distinct speech sound. An allophone is a term used to describe the group of phones represented by a single phoneme within a particular language. For example, [tʰ] (aspirated), [t] (unaspirated), and [tʃ] (affricated) are all allophones for for the phoneme /t/.

That is also known as narrow transcription. Here is an explanation of the diacritical marks used for allophones:

allophone diacritics

Here is an example of pit and spit.

‘pit’ / »pIt / [»pHItH], [»pHIt|], [»pHIt] ‘spit’ / »spIt / [»spItH], [»spIt|], [»spIt]

As said elsewhere, there is no change in the meaning of a word via allophones. Allophones merely represent a variation in accent. And are generally not used in introductory levels of phonetics and phonemics.

In sum, IPA for English is not different across its definitions. What differs is the pronunciation of sounds. It is the same thing. What differs is accent which is represented in writing by allophones.

So, for example:

Marry, merry and Mary are often the same in American English, but usually different in British English: /ˈmari/, /ˈmɛri/ & /ˈmɛːri/. Notice, though, the IPA is the same and to marry someone is basically the same in both. The phonetic symbols from IPA are the same in both.

[all bolding mine]

Phonetic sounds are represented between square brackets, [...].
Phonemic sounds are represented between slash marks / /.
Allophones are also in square brackets with diacritical marks.

In that transcription shown above in the question:

The low stress mark ˌ is simply used before a syllable to indicate a lesser degree of stress than the high stress mark ˈ. single quote stress mark. The rest is regular IPA.

All these use IPA as shown in that manual and on Mr. Moran's pages.

ANSWER TO THE QUESTION: IPA, which is used for all languages, is not different "across" its definitions.

AND: "Both inventories are targeting British English, but their phones differs. It kind of confuses me even more..."|

That is not correct. The inventories are from two different corpora showing different accents via allophones. Ig samples were taken in France from the north and south of the country, differences in allophones would be found.

  • 5
    This doesn't really answer the question, which concerns the differences between phonetic and phonemic transcriptions, the fact that different resources adopt different IPA symbols to represent English phonemes, and the inconsistencies in narrow phonetic transcriptions.
    – alphabet
    Feb 29 at 4:56
  • It answers this: I'm trying to create a website to help my partner learn phonetics.. This can teach the phonetics required using standard IPA so teach the partner the 44 phonemes of English. And it does away with the problem of phonetics and phonemics by teaching the sounds via minimal pairs.
    – Lambie
    Feb 29 at 13:27
  • Also, she says: "Here is a screenshot of the phones considered necessary to translate english." The Trim book shows all of those through minimal pairs and the video has actors pronouncing the phonemes through actual words. That is standard IPA in her chart and in the book.
    – Lambie
    Feb 29 at 13:33
  • 1
    There is no standard IPA for English phonemes. That's part of the problem. For instance, some resources will use separate /ʌ/ and /ə/ symbols when transcribing American English; others won't.
    – alphabet
    Feb 29 at 18:04
  • It does now. I explained the difference in the transcriptions she found. Actually, they don't adopt different IPA symbols. English has 44 phonemes and anyone discussing English will use the standard IPA symbols for those phonemes for it. They may or may not get into the nitty gritty of accent variation and allophones.
    – Lambie
    Feb 29 at 18:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .